Application of design of experiment (DOE) to polymorph screening and subsequent data analysis

Abstract

Carbamazepine was used as a well characterised model compound to assess three polymorph screening methodologies. It has been shown that parallel crystallisation followed by in situ analysis by XRPD is an excellent way of generating and analysing new solid forms of a substance in situ. Further methodologies utilising aspects of design of experiment (DOE) show that not only can new forms of a substance be found but also that analysis of the input variables against output response (i.e. polymorphic form) allows one to predict the outcome of a crystallisation experiment. Using these methodologies three of the four known anhydrous forms of carbamazepine were found, as were the known dihydrate, acetone and 1,4-dioxane solvates. For a structured carbamazepine polymorph screen, relationships between input variables and output responses were investigated using the software packages Curvaceous Visual Explorer, INForm and FormRules, showing in the case of carbamazepine, the dielectric constant of the solvent seems to play a large role in determining what polymorphic form is obtained from a crystallisation. It is possible to then predict the polymorphic form from a given set of input conditions. It is also possible to optimise the outcome of an experiment i.e. maximise the chances of obtaining the desired form in a crystallisation experiment.