(201b) Multivariate Characterization, Modeling, and Design of Ionic Liquids
Product design is now considered as an emerging paradigm in the field of chemical engineering because it requires a different set of tools and skill sets from other problems traditionally encountered in the field. Design of environmentally benign solvents and alternative media for extraction and purification are new challenges within product design. One class of novel compounds being studied for such application is ionic liquids (ILs). Ionic liquids that have tailored structures with an array of unique functional properties can have important applications in areas such as CO2 capture and sequestration, sulfur removal from fuels, energy storage, biomass pre-treatment, and chemical separations. Through variation of both cation and anion, particular ionic liquids with tunable physical properties can be tailored.
However, it is estimated that over 1014 unique cation/anion combination are possible for use as room temperature ionic liquids, the majority of which have never been synthesized [[i]]. However, the traditional experimental trial-and-error approach of searching through this large molecular space is unrealistic as it is both time and labor intensive. Thus, it is essential to develop a logical and systematic approach of selectively choosing a given ionic pair that matches a set of desired physico-chemical property targets. The product design framework developed in this work seeks to address this issue by systematically selecting and designing ionic liquid molecules that contain the correct combination of cation and anion that match a set of target physico-chemical properties of interest.
To meet this objective, infrared (IR) spectra of a representative training set of ionic liquids that contain large quantities of descriptor data involving information on molecular architecture is generated using density function theory (DFT). DFT based simulation make the characterization techniques independent of the availability of experimental spectroscopic data. To identify systematic patterns and important features of the molecular architecture in such multivariate data, multivariate statistical techniques such as principal component analysis (PCA) and partial least square (PLS) are used. Appropriate latent variable property models are developed to capture the underlying relationships between physical-chemical properties and/or molecular architecture. Since, conventional regression based GCM does not have property parameters for groups/fragments of ionic molecules, the characterization based group contribution method (cGCM) is utilized for property estimations. The reverse design of potential IL molecules is accomplished by exhaustively generating combinatorial structures from the set of basic groups, which represent the chemical make-up of the training set fragments, until the resulting properties match the target property values. In exhaustive searches, selection from among numerous permutations of anion, cation, and alkyl chain attached to cation groups is performed [[ii]].