Determination of crystal structures from poor-quality data using Patterson methods

Abstract

It is found to be possible to solve the structures of fairly simple materials from very poor-quality diffraction data by the use of reciprocal-space Patterson methods. Data sets assessed include those from a high-resolution neutron powder diffractometer, data sets with inaccurate randomized |Fh| values, very small data sets (as few as ten reflections) and data sets with no estimates of |Fh| values given. While refinement is not always feasible from such data it is found possible to obtain atomic positions and consequent structural information with reasonable accuracy. Reasons for using Patterson rather than direct methods in such cases are discussed.