SSP Project Summary:
[EPCC home] [SSP home] [2001 projects] [2000 projects] [1999 projects] [1998 projects] [1997 projects] [1996 projects] [1995 projects] [1994 projects] [1993 projects]

Crystal Structure Determination from Powder Diffraction

Student

Sam Carr, University of Oxford

Supervisors

Robert Hammond, Dept of Mechanical & Chemical Engineering, Heriot-Watt University

Lorna Smith, EPCC

In the past crystal structure determination from powder diffraction has mostly involved highly symmetric inorganic systems and has had limited application for organic molecules. This is primarily due to the fact that whereas single crystal reflection intensities are resolved in three dimensions in reciprocal space, powder diffraction patterns contain the same intensity information projected in two dimensions. The problems which arise include peak overlap (either symmetric or accidental), large background intensities which cannot be defined easily and prefered orientation.

Advances in the hardware and software have however enabled the structures of more complex systems such as 6,13-dichlorotriphendioxazine (Fagan, Hammond et al., 1995) and phthalocyanine (Hammond, Roberts et al., 1996) to be solved.

The methodology which has been developed for solving crystal structures from powder diffraction can be summarised as follows:

1) Collection of high resolution powder diffraction data.

2) Determination of the unit cell parameters and space group of the crystal using a variety of programs.

3) Determination of the molecular configuration of the molecules present within the asymmetric unit of the crystal lattice.

4) Determination of trial crystal structures for Rietveld refinement.

5) Rietveld refinement (Rietveld, 1967) of the trial crystal structure.

The determination of trial crystal structures for Rietveld refinement has in the past led to serious problems. Rietveld refinement requires a good realistic starting structure and hence considerable effort has been applied to obtaining these starting structures.

A systematic search algorithm (SNIFFER97) (Hammond, Roberts et al., 1997) was developed as part of a collaborative research program with Zeneca Specialities, Blackely. This systematically searches all the possible packing configurations of the asymmetric unit within the specified unit cell and assesses the lattice energy, powder diffraction pattern and close contacts of each structure. It permits the prediction of structural models for Rietveld refinement. This systematic search algorithm deals with only one molecule within the asymmetric unit, a limitation that forced the development of an independent algorithm (compack1) to work in conjunction with the systematic search algorithm (Smith, 1997).

The method is however limited in accuracy due to computational limitations. Simple SNIFFER97 runs can involve searching millions of configurations, a number that increases dramatically if torsional effects are considered within the molecule and / or two molecules are present in the asymmetric unit.

This has led to larger step sizes being utilised and changes in the molecular structure being ignored. Some examples where the limitations have created a problem include SALOL (a pharmaceutical drug) and rubidium dodecyl sulfate. Parallelisation of SNIFFER97 (and compack1) would allow structures of complex systems to be solved which have not previously been solved before. Hence this project would involve the parallelisation of SNIFFER97 (using MPI) and if time allows of compack1.

The first task attempted by SNIFFER97 is a distance check on each lattice configurations to obtain realistic structures. The distance check on each configuration is independent of every other configuration and hence the program is inherently parallel and a parallel implementation should be relatively straight forward.

Initially, a simple system already determined using the serial version of the SNIFFER97 code will be utilised to ensure the parallel version can successfully replicate the results (e.g. sodium dodecyl sulfate). Following this a system which has previously failed to determine any successful results due to time constraints will be considered, this will most likely be SALOL.

If time allows compack1 will be parallelised and a system which has been determined using the serial code version will be utilised to ensure reproducibility e.g. rubidium dodecyl sulfate. This system may then be studied using a shorter step size and considering torsion effects as this was known to be a limitation of the serial determination. Finally, (again if time allows) a system which has not previously been studied, for instance that of catechol/urea complex will be examined.

Visualisation of these systems has previously been carried out using the molecular simulation package Cerius2 (Molecular Simulations Limited). This however is an expensive package, and would not be purchasable. Should time constraint require, Heriot-Watt have two licences and the student could uses one of these across the network. However if time allows and interesting addition to this project would be for the student to attempt the visualisation using AVS.

References

Fagan, P.G., Hammond, R.B., Roberts, K.J., Docherty, R., Chlorlton, A.P., Jones, W., Potts, G.D., Chemistry of Materials, 7, 12, 2322, 1995.

Hammond, R.B., Roberts, K.J., Docherty, R., Edmondson, M. and Gairns, R., J. Chem. Soc., Perkin Trans., 2, 1527, 1996.

Hammond, R.B., Roberts, K.J., Docherty, R., Edmondson, M., J. Phys. Chem. B., 101, 33, 6532, 1997.

Rietveld, H.M., Acta Cryst., 2, 22, 151, 1967.

Smith, L.A., PhD Thesis, Strathclyde University, 1997.

The final report for this project is available here.
Webpage maintained by mario@epcc.ed.ac.uk