Prof. Dr. Dirk Zahn
Molecular simulations are becoming increasingly prominent for the in-silico design of crystalline materials with desired properties. A particularly vivid aspect of this trend is given by ab-initio based prediction of molecular crystal polymorphs and their solubility in different solvent environments. This is propelled by the pharmaceutical industry, which aims at the rapid development of large-scale syntheses routes along with suitable formulation of pharmaceutical compounds as well-defined crystals. To avoid under- or overdosing of drugs, it is crucial to control not only the amount of drugs in the pill but also their solid-state structure (crystal polymorphism), grain size and the overall dissolution kinetics. For this reason, modelling-based polymorph prediction is typically initiated already at early stages of drug development when all synthesized material is dedicated to clinical trials and thus unavailable for crystal precipitation and pill formulation studies.
In the present project, we aim at a) enabling the prediction of crystal polymorph stability and b) the prediction of drug solubility in common, non-toxic solvents. The former is based on computing the free enthalpy of vaporization of molecular crystals, whereas the latter requires computing the free energy of solute insertion into a solvent model.
For this, our recently developed molecular dynamics (MD) scheme for thermodynamic integration via hyperspace coordinates shall be implemented in the massive parallel MD package LAMMPS. The theory is fully developed, however, so far we only have a single-node DLPOLY implementation thereof. For broader application in both academic and industrial research, we argue that state-of-the-art HPC is inevitable. As a long-term perspective, we desire to extend the LAMMPS package to enable crystal stability and solubility calculations to an unrestricted community of users encompassing mineralogy, chemistry, and pharmaceutical engineering.