Student Project 2

The use of pattern recognition techniques in the analysis of structural databases

Description

The Cambridge Structural Database (CSD) contains an enormous amount of structural information resulting from single-crystal diffraction analyses. Among the 250,000 compounds currently present in the database there are many identical and isomorphous entries. However, it is plausible to expect that the CSD contains certain more hidden structural principles, in analogy with the systematics found for the relatively simple inorganic compounds.
The purpose of this project is to find isostructurality at a more general level in the CSD by using pattern recognition methods applied to powder diffraction patterns. Initial results show that there are many related structures in the CSD. This was found via sophisticated PXRD pattern comparison. The first results also show that patterns that are found to be similar show unit cell parameters that, after scaling, are very closely related. This means that the CSD could also be used to extract structural information for experimentally obtained PXRD patterns.

Keywords

Crystal engineering, isostructurality, database analysis, pattern matching, programming

Research program

Chemical Crystallography: Pattern Matching in Powder Diffraction

Contact person

When you are interested in this student project please contact:

  R. de Gelder    Phone 52842    Room N4018

References

1. A Generalized Expression for the Similarity of Spectra: Application to Powder Diffraction Pattern Classification.
   R. de Gelder, R. Wehrens and J.A. Hageman.
   Journal of Computational Chemistry (2001), 22, 3, 273-289.

2. Isostructurality at different scales as seen in the Cambridge Structural Database.
   R. de Gelder, J.M.M. Smits.
   Poster presented at the Lunteren Conference "Kristal en Structuuronderzoek", 22 and 23 april 2002.