C. Heath TURNER1, Gabriel BARBOSA1, Liu XIAOYANG1
1The University of Alabama, Tuscaloosa, United States
Ionic liquids (ILs) may provide unique advantages for various gas capture and separation applications. While there are many challenges associated with industrial applications of ILs, there are several polymer and composite materials that can be synthesized based on the IL functionalities, and these materials can circumvent many of the cost, transport, and other inherent limitations. In order to accelerate commercial adoption, a lot of experimental and simulation work has been performed to characterize and predict the thermophysical and gas absorption properties of different ILs, but the vast majority of this work has focused on common monovalent ILs. Here we investigate the gas (O2, NH3, CH4, H2S, SO2, Ar, H2, N2, CO2) absorption properties of multivalent ILs, which may be beneficial to several different applications. The properties of these ILs are not well understood, and furthermore, they fall outside of previous machine learning datasets, limiting the predictability of their properties.
We use a combination of quantum chemical (QC) and molecular simulations (Monte Carlo and molecular dynamics) to understand the gas absorption and selectivity performance of a series of different multivalent ILs. The gas absorption behavior can be interpreted using our recently-developed QC-level descriptors, such as the ionic polarity index (IPI) and the solvation affinity index (SAI). These QC-level predictions are fast and reliable, based on benchmarking against molecular simulation data, as well as experimental performance. This QC-level screening framework is now being extended to encompass ILs outside of the original training set.
One of the more difficult properties to calculate in molecular simulations is gas transport in ILs, due to the inherently slow liquid dynamics. Even long trajectories (exceeding a microsecond) can be insufficient to properly capture the gas diffusion behavior. We have recently developed scaling approaches, based on the IL surface properties that allow us to extrapolate the results from high-temperature simulations down to the relevant temperatures of interest. This scaling approach has been tested for several different systems, and it provides an efficient route for estimating gas diffusion behavior in viscous liquids that would otherwise be extremely difficult to capture in a molecular simulation.