Computer-Aided Molecular Design by Combining Genetic Algorithms and COSMO-RS

Increasing demand for tailor-made chemicals gives rise to challenging molecular design tasks. Previous molecular design approaches have relied on simplified thermodynamic models to be computationally tractable. In contrast, quantum mechanics offers the most comprehensive molecular picture but a direct integration into computer-aided molecular design (CAMD) is challenging. In this work, we therefore aim at integrating quantum-level information into molecular design while still allowing for efficient computations. For this purpose, a framework for optimization-based molecular design is introduced based on property predictions by COSMO-RS and a genetic algorithm for molecular design. The resulting framework is applied to a case study for solvent design in liquid-liquid extraction.