Informatic Calibration of a Materials Properties Database for Predictive Assessment of Mechanically Activated Disordering Potential for Small Molecule Organic Solids

Abstract

The potential for small molecule organic crystalline materials to become disordered as a result of high shear mechanical processing was investigated. A data-driven model was generated from a database of critical materials properties, which were expected to correlate with the potential of a small molecule organic crystalline solid to become fully disordered by the application of mechanical energy. The model was compared with a previously published disordering model based on fundamental thermodynamic relationships. Samples of 23 crystalline solids were subjected to extensive comminution under controlled temperature conditions; powder X-ray diffraction (PXRD) and differential scanning calorimetry (DSC) were used to confirm disordering. Logistic regression was used to investigate the significance of each materials property with respect to the prediction of disordering potential. Seven materials properties (glass transition temperature, melting temperature, heat of fusion, crystallographic density, Young's modulus, molar volume and attachment energy) were identified as having a significant correlation with the potential for material disordering. Stepwise multivariate logistic regression was used to further assess the correlation between disordering potential and each of the seven properties. A linear probability model based on two materials properties (glass transition temperature and molar volume) was developed for the prediction of disordering potential.