LAUSR

Abstract In this work, we adapt the COSMO-RS and -SAC methods to solve computer-aided mixture design (CAMxD) problems. Popular methods in CAMxD require the use of binary interaction parameters to calculate mixture thermodynamics, and this necessity places inherent limitations on the possible chemical search space. Our COSMO-based approach is free of binary interaction parameters and requires only molecular volumes and molecule-specific charge density distributions called sigma profiles for the estimation of solution properties. Additionally, this methodology enables the integration of highly accurate molecular information from ab initio quantum chemistry calculations into mixture design problems. To address the search problem, we project molecular identities and mole fractions on the space of each mixture component's sigma moments, which are analogous to statistical moments for sigma profiles. This approach exploits a natural problem decomposition and capitalizes on fast methods for pure compound design and mixture fraction design. We apply the methodology to two case studies: the design of a liquid–liquid extraction solvent and a reaction rates optimization solvent.