Chemistry Reference
In-Depth Information
the phase problem was eventually solved by a strain of human ingenuity called
“direct methods” [
2
,
3
]. These require little more information than just the chemical
composition of the molecule forming the crystal - even a reasonable approximation
will do - and the space-group symmetry.
When properly diffracting single crystals are available, technical problems of
crystal structure solution and refinement by computer are today completely solved
and the software is stored in standard packages that are used thousands of times all
year round [
4
,
5
]; they sometimes come along with data-collection instruments with
nicely composed commercial advertisements. In considering these tools, it may be
worth mentioning that one single paper published in
Acta Crystallographica
with
the recommended quotation for one such package [
6
] has - just by itself - increased
the impact factor of that journal (normally in the 2.0-2.5 range) to an impressive
49.9 (indeed impressive, for better or for worse). There are of course still some
tough nuts to crack, like solving structures on data from very small or recalcitrant
crystals, or for complicated blends of chiral entities [
7
].
Quite often, no single crystals can be obtained for a given substance, and only
powder samples are available. Ten years ago the chances of obtaining a definite
structural determination from such samples were still very low, but things are
quickly changing [
8
].
The International Union of Crystallography supports an electronic newsletter
1
with news and updates on crystallographic computing. Treatises and primers in
X-ray diffraction data handling are available [
9
-
11
] (in order of decreasing length
and detail). For small- or medium-size molecules, in ordinary routine application
the overall solar time from the moment a crystal is handed over to the crystallogra-
pher to the moment a structural picture becomes available is presently of the order
of a few hours. Computer demands in protein crystallography are of course one
order of magnitude larger, parallel to the orders of magnitude of the larger number
of diffracted intensities.
2
There is no good science without good double-checking. Computational and
practical X-ray crystallography has benefited - indeed its very existence has been
made possible - by the availability of an enormously effective tool for checking and
improving crystal data treatment: Anthony Spek's structure validation package,
PLATON
[
12
]. Thanks to this implement, X-ray crystallography appears as the first
(and so far only) discipline that has been able to spot, unequivocally and quantita-
tively, a scientific fraud [
13
].
Nowadays computers are so absurdly fast that the phase problem can be solved
by recursive computation: the newly proposed charge-flipping algorithm [
14
] per-
forms in absence of
any
information on the target crystal structure; not even the
molecular composition or the crystal symmetry is needed. The procedure starts with
1
2
See for the latest example,
http//nobelprize.org/nobel_prizes/chemistry/laureates/2009/
, for a
25-year long struggle for the determination of the structure of the ribosome, involving heavy
computational effort.
