Extending the Power of Global Optimisation Methods in Direct Space Structure Determination from Powder Diffraction Data

Abstract

Despite substantial advances in the use of direct space approaches for structure determination of inorganic and organic crystal structures, some problems remain essentially intractable. For example, very few molecular structure solutions with three molecules in the asymmetric unit or with greater than fifteen fully flexible torsion angles have been reported. Indeed, these levels of complexity represent the current limits of applicability of the technique. Here, three new directions that should increase the scope of structure determination from powder diffraction data are discussed. Firstly, the use of a maximum-likelihood analysis to permit structure determination in the case of incomplete molecular descriptions. Secondly, the incorporation of prior chemical information to reduce the extent (or dimensionality) of the torsion angle search space and finally, the use of global optimisation methods other than the widely investigated simulated annealing (SA) and genetic algorithm techniques.