COSMO-Descriptor Based Computer-Aided Ionic Liquid Design For Separation Processes

Ionic liquids (ILs) are novel solvents for separation processes with their unique properties. Since the separation performance strongly relies on the structure of IL molecule, a rational IL design for specific mixtures is highly desired.To enable the computer-aided ionic liquid design (CAILD), a modified group contribution method, namely COSMO-GC-IL method, is proposed to predict ?-profiles and cavity volumes (Pcosmo) of ILs quickly. In this method, the interaction between groups is entirely transformed into the stretch and translation as modification on COSMO descriptors, which are derived from ?-profiles of groups. Therefore, isomeric problems can be solved through distinct ?-profiles. Activity coefficients of system is then calculated by COSMO-SAC model which can be utilized to quantify the property of the whole separation process.A mixed simulated annealing-genetic algorithm is developed and integrated into CAILD program combing the merits of simulated annealing with genetic algorithm, which is able to solve MINLP problems with reliable convergence performance. In this algorithm, ILs are represented by rooted trees, where the type and relative position of groups can be characterized and stored. The tree representation facilitates the operations of mutation and crossover as well. The CAILD program is applied to design IL structures for extractive separations, such as separations of benzene-cyclohexane, heptane-thiophene and vitamin E-deodorizer distillate. The optimized ILs all exhibit higher separation performance than those from literature.