Phase Equilibria and Diffusivities of HFC-32 and HFC-125 in Ionic Liquids for the Separation of R-410A

Abstract

Current legislation calling for the phase out of hydrofluorocarbon (HFC) refrigerants is driving a global market shift that has prompted industry and research institutions to investigate new refrigerant mixtures and sustainable separation techniques for recycling refrigerants. The recent American Innovation and Manufacturing (AIM) Act of 2020 requires an 85% phase down of HFC production over the next 15 years. To achieve this goal, azeotropic refrigerant mixtures, such as R-410A composed of 50 wt % HFC-32 (difluoromethane, CH 2 F 2 ) and 50 wt % HFC-125 (pentafluoroethane, CHF 2 CF 3 ), will have to be separated to recycle the lower global warming HFC-32 component. The present work investigates the solubility of HFC-32 and HFC-125 in six ionic liquids (ILs) with halogen anions for the purpose of developing the thermophysical property data required for designing extractive distillation recycling processes and understanding the choice of cation and anion type. A gravimetric microbalance was used to collect isothermal vapor liquid equilibrium data for each of the ILs at 298.15 K and pressures from 0.05 to 1.0 MPa. The Peng−Robinson equation of state was used to model the solubility of the HFCs in the ILs. The solubility of HFC-32 in the ILs showed small differences, while the solubility of HFC-125 had significant variations with respect to the anion type and the cation alkyl chain length. Fick's law was applied to calculate diffusion coefficients for each HFC/IL system. HFC-32 has a greater diffusivity than HFC-125 based on smaller molecular size. The 1-n-hexyl-3-methylimidazolium chloride and the trihexyl(tetradecyl)phosphonium chloride ILs have the highest HFC-125/HFC-32 selectivity at 298.15 K. Based on both the mass uptake and selectivity ratio, these two ILs are potential entrainers for the separation of R-410A using extractive distillation.

What data are most valuable to screen ionic liquid entrainers for hydrofluorocarbon refrigerant reuse and recycling?

“…As expected, we find that the predictions utilizing M T , which has no κ temperature dependence and was calibrated only with experimental ternary data, has a 1% in-sample MAPE. In contrast, we find that the predictions of M B and M B,K (both models only calibrated with experimental binary solubility data) have an out-of-sample MAPE of 10.1% and 10.2%, compared to the ternary experimental compositions in Baca et al 56 The error is consistent with the difference observed between the data sets in Figure 3. We emphasize these measurements we obtained with different experimental methods and equipment (as described in the references), so a modest difference is to be expected.…”

“…49,50 Regression of thermodynamic models Because equations of state (EoS) are continuous and differentiable functions, they are wellsuited for computer-aided process design and optimization. [51][52][53] EoS models, including van der Waals, 54,55 Peng-Robinson, 56 Redlich-Kong, 18,19 and soft-SAFT EoS, 57 have been shown to reliably predict thermodynamic properties of HFC and IL mixtures. Yet, many of these models include parameters which must be calibrated from data before they are used in process design.…”

“…These parameters are often fitted using a minimizing least-squares approach 58,59 involving the pressure of the system. 18,19,[54][55][56] However, different objective functions, such as those involving liquid or vapor compositions, can be utilized in practice and may provide different parameter values. 60 There is a rich history of nonlinear parameter estimation within the chemical engineering discipline and, more specifically, thermodynamics community.…”

“…Based on previous studies from Morais et al 55 and Baca et al, 28,56 three types of data are considered in this work:…”

“…, and [hmim][FAP] collected with a gravimetric microbalance. 17,19,55,56 2. HFC-32/HFC-125 binary equilibrium data collected with gas chromatograph equipped with a flame ionization detector.…”