It is well established that characterizing alternate crystal structures (polymorphs) in pharmaceuticals is an important part of the development process. Previously unknown polymorphs have caused market recalls in pharmaceuticals already in use by the general population. Knowledge of the phase diagram of small molecule pharmaceuticals can direct the storage protocols for these drugs. Here, we present a new method to predict phase diagrams in small molecule compounds. In this method, multistate reweighting, using the multistate Bennett acceptance ratio (MBAR) is combined with iterative sampling to predict the most stable polymorph within a range of temperatures and pressures. Sampling is performed at an initial set of temperatures and pressures, and MBAR is used in combination with a known reference free energy to compute the relative free energy of all polymorphs at all thermodynamic states. The lowest free energy polymorph corresponds to the most stable crystal polymorph at that state. Linear interpolation is used to estimate coexistence points between sampled or unsampled states. Additional sampling is conducted iteratively around the predicted coexistence points of the phase diagram until a desired reduction of uncertainty is achieved.<br/>The initial test systems were the well studied Lennard-Jones system and benzene, which has three experimentally observed polymorphs. The behavior observed in the phase diagrams, as predicted in this method, is consistent with experimental results for benzene and previous simulations in the Lennard-Jones potential. This method has several advantages over previous phase diagram prediction methods. No a priori knowledge of a coexistence point is required and the uncertainty in the coexistence line is not a function of the distance from the reference point. This method can be automated and applied to a wide variety of systems. In the future this method will be applied to more complex molecules of commercial interest and those with liquid-solid coexistence.

Phase diagram prediction using multistate reweighting combined with iterative sampling

251st National Meeting (2016)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.