AbstractsPhysics

Being able to predict solubility has been a long standing goal for the pharmaceutical industry. The aim of this project was to evaluate whether melting temperatures and solubilities of drug molecules could be predicted with high accuracy. Crystals (polymorphs, co-crystals and solvates) of indomethacin and carbamazepine were produced experimentally and characterized by X-ray diffraction, differential scanning calorimetry (melting temperature) and thermal gravimetric analysis. Dissolution rates in water were determined by absorbance measurements. Surface morphology and size distribution were investigated by scanning electron microscopy and laser diffraction measurements. To assess differences in accuracy between simulation methods, simulations were performed on a benchmark set (C19RT) consisting of 19 organic molecular crystals. Crystalline geometries and lattice energies of the benchmark set were evaluated. The Grimme (D3) dispersion method was found to give the highest accuracy. A correlation between sublimation free energy and melting temperature was found. Entropy simulations were performed to obtain Gibbs’ free energies. Correlations linking entropy and zero point energy to the unit cell volume were found. Solvation energies were calculated by simulating hydration free energies utilizing a thermodynamic cycle. Solubilities were then calculated from the solvation free energies of the benchmark set. For the simulations of indomethacin and carbamazepine crystals, the correlations found for entropy and zero point energy were used. Melting temperatures were predicted from simulated sublimation free energies using the previously found correlation and compared to generated experimental data. A weak correlation was found. Calculated solubilities of the indomethacin and carbamazepine crystals did not fully agree with experimental data, indicating the need for development of more accurate simulation methods.