Refrigerants

Overview

Refrigerants are the working fluids at the heart of air conditioning, refrigeration, and heat-pump systems, a large and growing global industry. Because these systems consume an enormous amount of electricity worldwide, even modest gains in refrigerant efficiency translate into substantial energy savings. At the same time, many widely used refrigerants are potent greenhouse gases, and regulations are driving a transition away from high global warming potential hydrofluorocarbons (HFCs) toward lower-impact alternatives. Designing the next generation of refrigerants means finding fluids that are simultaneously efficient, safe, chemically stable, and environmentally benign, a difficult multi-property balancing act.

Our group has a long history of using molecular simulation to study the phase equilibria and thermophysical properties of refrigerants. This work began with an interest in replacing traditional HFCs with hydrofluoroolefins (HFOs), and has since grown into a broader effort spanning property prediction, the separation and recycling of refrigerant mixtures, and the computational discovery of entirely new refrigerant molecules. We are now part of the NSF Engineering Research Center EARTH (Environmentally Applied Refrigerant Technology Hub), a multi-university center focused on developing the refrigerants and refrigerant technologies of the future.

Force Fields and Property Prediction for HFOs and HFCs

Reliable simulation of a refrigerant starts with an accurate molecular model. An early step in our refrigerant work was building a force field for fluorinated propenes (in collaboration with Prof. Gabriele Raabe) and using it to predict the vapor-liquid equilibrium of HFO-1234yf, an alternative refrigerant of strong industrial interest. More recently, we have used machine learning and formal optimization (in collaboration with Prof. Alex Dowling) to develop force fields for hydrofluorocarbons that are both accurate and transferable across compounds and state points, rather than fitted to a single fluid. A related strand of work assesses and ranks existing interatomic potentials for common refrigerants such as R-32 and R-125 against a broad set of thermophysical and transport properties, so that users can choose a model knowing where it is reliable and where it is not. This emphasis on knowing the limits of a model, not just reporting agreement, connects directly to the group's broader work on reliable property calculation (see Molecular Simulations: Methods and Software Development).

Raabe & Maginn, J. Phys. Chem. Lett. 2010, 1, 93; Raabe & Maginn, J. Phys. Chem. B 2010, 114, 10133; Wang et al., J. Chem. Theory Comput. 2023, 19, 4546; Befort et al., Comput. Aided Chem. Eng. 2022, 49, 1249; Agbodekhe et al., J. Chem. Eng. Data 2024, 69, 427; Carlozo et al., Digital Discovery 2025, 5, 1650 (https://doi.org/10.1039/D5DD00537J)

Separation and Recycling of Refrigerant Mixtures

Many commercial refrigerants are azeotropic or near-azeotropic mixtures, which makes them difficult to separate and therefore difficult to recycle or reuse. R-410A, a mixture of HFC-32 and HFC-125, is a leading example. In close collaboration with the experimental groups of Mark Shiflett and Aaron Scurto, we have studied the use of ionic liquids as separation agents for these mixtures, combining molecular dynamics and Monte Carlo simulation with experimental phase-equilibrium and diffusivity measurements. The simulations predict how each fluorocarbon dissolves and diffuses in candidate ionic liquids and help explain the molecular structure and dynamics of the resulting mixtures, while free-energy and replica-exchange methods are used to compute the underlying sorption isotherms. We have also examined adsorption-based separation, studying how HFC-32 and HFC-125 and their mixtures adsorb in the zeolite silicalite-1. Together, this body of work supports the recovery and reuse of refrigerants, an increasingly important consideration as older high-global-warming-potential fluids are phased out.

Morais et al., Ind. Eng. Chem. Res. 2020, 59, 18222; Baca et al., ACS Sustainable Chem. Eng. 2022, 10, 816; Wang et al., J. Phys. Chem. B 2022, 126, 8309; Befort et al., Fluid Phase Equilib. 2023, 572, 113833; Wang et al., J. Chem. Theory Comput. 2023, 19, 3324; Al-Barghouti et al., J. Phys. Chem. B 2025, 129, 7311; Marin-Rimoldi et al., J. Chem. Phys. 2024, 161, 074701; Baca et al., Chem. Rev. 2024, 124, 5167.

Discovering New Refrigerants

The set of refrigerants in use today is small, and the space of well-known candidate molecules has been heavily searched. Yet the space of chemically plausible molecules that have never been made or tested is vast. A central goal of our refrigerant research, pursued within the EARTH ERC, is to explore that larger space computationally: generating candidate molecules, screening them with machine learning and molecular simulation for the right combination of efficiency, stability, safety, and low environmental impact, and passing the most promising few to experimental collaborators for synthesis and testing. The simulation methods and machine-learning property models that make this possible are described more fully on the Thermophysical Property Prediction and Machine Learning page.

Funding

This work is supported by the National Science Foundation (NSF), including the NSF Engineering Research Center EARTH.