(419a) Application of Quantum-Based Descriptors to the Molecular Design of Ionic Liquids

This project incorporates quantum-based descriptors into a computational molecular design framework for the design of novel ionic liquids. Ionic liquids can be tuned for various applications by varying the structure of the cation and anion. However, many feasible cation/anion combinations have never been synthesized, so experimental data is sparse. Therefore, accurate models for property prediction are a fundamental requirement for computational molecular design. While group contribution methods are effective for some properties, other properties require quantum-level information about the molecule. In this project, the conductor-like screening model (COSMO-SAC) is used to calculate the molecular surface charge distribution (sigma profile) from molecular structure information. The surface charge distribution can then be incorporated into thermodynamic property prediction algorithms. These predictions are then applied to optimization-based design algorithms for novel molecules. These optimization problems are black-box MINLPs, which can be solved to near-optimality by stochastic methods, including using the Tabu search algorithm applied in this work. Examples are shown presenting successful designs for ionic liquids in azeotropic separation processes.