Solving Crystal Structure of Molecular Materials from Powder Diffraction Data using Monte Carlo and Maximum Entropy Methods. M. Tremayne, W.Dong, C.J.Gilmore, Department of Chemistry University of Glasgow, Glasgow G12 8QQ, Scotland, UK and B.M.Kariuki and K.D.M. Harris, Department of Chemistry, University of Birmingham, Birmingham, B15 2TT, UK

The majority of crystal structures that have been determined from X-ray powder diffraction are framework and non-molecular solids, and much less has been achieved for molecular crystals, where the effective resolution of the data can lie between 1.5-2(, and where conventional direct methods usually fail. We present here the results of two methods that overcome these problems: (1) The maximum entropy-likelihood Method (Gilmore, C.J., Henderson, K. & Bricogne, G. (1991). Acta Cryst. A47, 830-841) (2) The Monte Carlo Method (Tremayne, M., Kariuki, B.M. & Harris, K.D.M., Angew. Chemie (1997) In press.)

In (1) a phasing tree is built in which the nodes represent possible phase choices, and in which likelihood is used to assess the correct phases; at less than atomic resolution, the resulting maps are investigated visually, and molecular models, are fitted to the electron density. In (2) Monte Carlo methods are used to position and modify the conformations of molecules in the unit cell (using the program OCTOPUS), and to use the powder diffraction pattern as a measure of correctness in deriving initial atomic coordinates. This method is limited by the number of degrees of freedom in the geometry of the molecule. It is quite possible to combine these methods to give a third option: (1) is used to generate maps from which initial models are derived that are then modified by (2) before Rietveld refinement. Examples will be presented of all three methods solving organic molecules.

Research funded by the EPSRC.