Disorder in crystal structures : new approaches in finding the best model

A new approach has been implemented in the refinement program CRYSTALS that enables the crystallographer to treat structures with disordered parts in an intuitive way, improving the refined models and saving a considerable amount of time. Roughly half of all crystal structure analyses suffer from some kind of problems. Many of them are problems involving disorder, which keep the crystallographer busy, often for many hours and days. Many disordered structures can be described rationally as the molecule taking advantage of one or more of its degrees of freedom. However, an equivalent easy approach is not provided by the traditional way to build crystallographic models for refinement.
New scripts in CRYSTALS now help the scientist to deal with disorder with more ease. The scripting environment in CRYSTALS provides a mechanism for formalising procedures which have to be repeated frequently and partially automating them. The new scripts help the crystallographer to first regularize the geometry of structural fragments that have become distorted during initial refinement of the starting model. After this step, disorder models with two components are set up by applying a non-crystallographic symmetry operator to the prototype regularised component in order to create a duplicate in the alternative position. Restraints to keep the geometrical features of the disordered fragments within defined boundaries can be created in an automated way. Finally, the user is assisted in the different steps of refining this two-component disorder model before integrating it into the final refinement of the whole structure.
On the way through the refinement of the disorder the new scripts keep the model consistent, even in the case where extensions of the disorder assembly or modifications to it are necessary. This feature too results in saving time. Searching errors in long hand edited instruction lists used to be tedious and time consuming.
While first tests show promising results the current development is aimed to extend the number of cases this approach can be adapted to. Disorder models with more than two components, assisted model building for disordered solvent molecules, inclusion of non-atomic electron density descriptors in the scripts or the development of validation criteria and tools for the way disorder has been refined are possible fields for future work on the base of this concept.