The perils of large 'small' molecules - successful refinement of metallosupramolecular grid-like assemblies

Abstract

With recent advances in diffractometers and as computers grow ever more powerful, it is possible to collect and solve several data sets every day. Thus the technique is now being used as alternative form of characterisation. This also has meant that the size of crystals required to obtain data has decreased, and the size of the molecules involved has increased, which can lead to a severe shortfall in data to parameter ratio. A recent area which has gained from these advancements is that of molecular grid and tape structures, although the successful solution and refinement may often be problematic due to the generally large asymmetric units. As chemists design and prepare these larger structures, the composition is generally known, but the geometry is still vital to determine. Most programs for full structure determinations were designed for small molecules with up to 100 non-hydrogen atoms. Although they can cope with larger numbers, quite a few crash once more than 500 atoms become involved. This does cause problems not only in attempts to solve structures, but also (especially for the chemist) when trying to visualise them later. Through our work with the UK National Service it has been necessary to address some of the problems described. For example the application of focussing mirrors to a rotating anode generator significantly increases the number of observed reflections for small crystals and for those of the above type of compound which are often weakly diffracting. This increase in quantity of experimental data coupled with the large number of variables provides a real test for the software. Generally, of the initial solution refinement programs, it is often only Shelxs-97 which produces a reasonable solution, although the assignment of the atoms is tedious. Most of the 'brute-force' programs give a nonsensical solution, if at all. For refinements Shelxh-97 does work, although sometimes the source code needs to be tweaked to cope with the largest structures (4000 parameters might be ample for most systems but 8000 is often a better limit). This poster will present successful structure determination of a number of these cases and the pitfalls and method of treatment will be outlined.

More information

Identifiers

ORCID for P.N. Horton: orcid.org/0000-0001-8886-2016

Catalogue record

Export record