A NEW MONTE CARLO APPROACH TO STRUCTURE SOLUTION FROM POWDER DATA. Yuri G. Andreev, Philip Lightfoot and Peter G. Bruce, School of Chemistry, University of St Andrews, St Andrews, Fife, KY16 9ST, UK

Unlike Monte Carlo procedures for structure solution reported thus far we report a new approach which not only allows random movements of individual atoms and rigid-body fragments comprising an asymmetric unit but offers a means of changing internal configuration of stereochemically constrained constituent groups. The last option can be crucial when flexible atomic assemblies like chains, sheets and other moieties are considered.

A way of describing the flexible groups mentioned above is proposed which uses stereochemical descriptors such as bond lengths, bond and torsion angles along with positional and orientational parameters of the group as a whole. An appropriate Monte Carlo algorithm combined with simulated annealing for the solution of this type of structure is discussed in detail. Similar to the previuosly reported algorithm¹, the Pearson-distribution [[chi]]² criterion of powder-profile fit is used as an analogue of the 'energy' function.

The new technique is illustrated by solving the previously unknown structure of poly(ethylene oxide)₃:LiN(CF₃SO₂)₂ comprising 25 non-hydrogen atoms in the asymmetric unit partitioned into an imide moiety, poly(ethylene oxide) chain and individual Li atom.

¹Newsam, J. M., Deem, M. W. and Freeman, C. M. NIST Spec. Publ 846., Proc. Int. Conf. Accuracy in Powder Diffraction II, held at NIST, Gaithersburg, MD. May 26-29, 1992.