Generalizable Synthesis of Highly Fluorinated Ionic Liquids

Abstract

The unique chemistry of fluorocarbons (in particular, their weak intermolecular interactions and high degree of intrinsic free volume) makes them promising building blocks for ionic liquids with high gas capacities. Here, we report a generalizable method for the synthesis of fluorinated ionic liquids, which relies on the evolution of gaseous byproducts to drive product formation. This synthetic strategy overcomes solubility challenges that can hinder the synthesis of highly fluorinated ionic liquids via conventional methods and enables a systematic investigation of the effect of fluorination on ionic liquid viscosity and gas solubility.

Transferability and Accuracy of Ionic Liquid Simulations with Equivariant Machine Learning Interatomic Potentials

“…In the Supporting Information, we provide the experimental details for the synthesis and characterization of [F-OMIM] + [C 4 F 9 CO 2 ] − (Section 1) , and additional details for the computational methods employed that are briefly described below (Sections 2–5).…”

“…To work toward this goal, we synthesize and characterize (details in Supporting Information Section 1) a novel IL [F-OMIM] + [C 4 F 9 CO 2 ] − , the structure of which is shown in Figure a; that is of interest for its O 2 solubility. , As can be seen in Figure b, this IL has a highly heterogeneous, lamellar-like structure, which creates voids for O 2 to dissolve into . In ref , the solubility of O 2 was correlated to thermal expansion and isothermal compressibility of the ILs, both of which are related to the formation of voids in the IL .…”

“…Of the routes tested, we find that at least 2 compositions are required, and the best approach is to train on "high-entropy" Li-salt molar fractions, e.g., 0.4–0.6, as these compositions ensure interactions are well sampled. Second, we synthesize and experimentally characterize a novel IL, [F-OMIM] + [C 4 F 9 CO 2 ] − , which is similar to other ILs of interest for their high O 2 solubility. , We investigate how to train an accurate Allegro model with ∼200 DFT frames, capable of capturing experimental densities, thermal expansion, and isothermal compressibility of this novel [F-OMIM] + [C 4 F 9 CO 2 ] − , with the main limitation being the approximations of our chosen DFT settings. For [F-OMIM] + [C 4 F 9 CO 2 ] − we also investigate the role of how the structures are generated and perform a hyperparameter search to optimize the model.…”