Deep Eutectic Solvents Mixed with Fluorinated Refrigerants for Absorption Refrigeration: A Molecular Simulation Study.

Molecular simulations were performed to evaluate mixtures of fluorinated refrigerants with deep eutectic solvents (DESs), for potential use in single-effect absorption refrigeration cycles that use low quality waste heat sources at temperatures of ∼80 °C. The refrigerants considered were the hydrofluorocarbon R245fa and the hydrofluoroolefins R1234zeE and HFO1336mzzE, whereas the DESs evaluated were 1:2 molar mixtures of choline chloride with either ethylene glycol (ethaline) or levulinic acid (levuline) as hydrogen bond donors (HBDs). Assuming the same cycle operating conditions, the waste heat cycle efficiency η was computed for all working fluid mixtures from molecular simulation results of the mixture densities and Henry's law constants of the refrigerants in the DESs, coupled with phase equilibrium calculations and the enthalpies of the pure refrigerants. The largest efficiency was obtained for the mixture R245fa-ethaline (η = 6.82), followed by R245fa-levuline (η = 4.64) and HFO1336mzzE-levuline (η = 2.10). These modest efficiency values could be further increased by tailoring the cycle operating conditions to each particular refrigerant-DES system, as well as optimizing our choice of working fluid mixtures, neither of which we attempted in this study. Strong interactions were observed between the chlorine anions and some of the hydrogen atoms of the refrigerants, but in general the cation-refrigerant and HBD-refrigerant interactions are weaker compared to the refrigerant-refrigerant interactions. Refrigerant molecules have the largest diffusivities and make the cations, anions and HBD to move faster compared to systems of DESs without refrigerant; in general, species in refrigerant-ethaline mixtures have larger diffusivities compared to those for refrigerant-levuline mixtures. We also computed waste heat cycle efficiencies for the same R134a-DES mixtures studied in our previous work, finding significant differences between the efficiencies determined from molecular simulation data and those determined before using the COSMO-RS approach using two standard parametrizations. This observation suggests that further work is needed to improve the accuracy of the COSMO-RS predictions for these systems.