A method for analysis of chiral molecules for the purpose of property prediction of EMs

Abstract. Nearly every molecular crystal structure solved by X-ray diffraction has been deposited in the Cambridge Structural Database (CSD). The CSD therefore enables large scale data analysis and machine learning (ML) based on real, isolated materials. This database is especially useful for energetic material research because each molecular structure includes crystal densities, one of the most important characteristics for understanding the performance of energetic materials. CSD allows for the export of large numbers of structures as simplified-molecular-input line-entry-system (SMILES) strings, but three-dimensional information such as chiral centers are generally lost during this translation. This work describes a method for the translation of CSD entries to stereospecific chiral strings that we anticipate will be useful for the development of better ML models of EM properties.