Title: Investigating and Modeling Pure and Binary Adsorption for Separating Azeotropic Fluorocarbon Refrigerant Mixtures

Mitigating global climate change is a crucial challenge facing 21^st century scientists. To reduce carbon emissions, regulations restricting the use and production of greenhouse gases (GHGs) are increasing, especially for hydrofluorocarbons (HFCs), which have high global warming potentials (GWPs) and comprise most of the global fluorocarbon refrigerant supply. As HFCs are replaced with next‑generation refrigerants in existing heating and cooling infrastructure, the high‑GWP HFCs must be reclaimed and recycled to prevent further release and environmental damage; however, most HFC refrigerants are azeotropic mixtures that are impossible to completely recycle without first being separated.

The present study investigates the use of adsorbents for separating azeotropic fluorocarbon mixtures, with a focus on refrigerant R‑410A, a 50/50 wt% mixture of difluoromethane (HFC‑32) and pentafluoroethane (HFC‑125). Over 15 adsorbents were examined, including basic, acidic, and siliceous zeolites, and three activated carbons. Pure and binary adsorption data were measured, and pure adsorption enthalpies were predicted to assess and compare adsorption behavior. The basic zeolites exhibited stronger affinity and selectivity toward HFC‑32, which was attributed to more hydrogen bonding. The non-basic zeolites and activated carbons exhibited stronger HFC-125 affinity and selectivity due to increased van der Waals interactions and molecular confinement. Conclusions were further supported through dynamic breakthrough experiments.

Adsorbed phase activity coefficients for HFC-32 and HFC-125 were determined and revealed highly nonideal adsorption behavior with zeolites 5A, silicalite, and H‑ZSM‑5, and an activated carbon. After comparing activity coefficients with Ideal Adsorbed Solution Theory (IAST) predictions and recent simulations, the nonideality was attributed to adsorption site re‑partitioning and preferential siting. A new adsorbed phase activity coefficient model was developed, consistent end-behavior was confirmed, and the model was fit to HFC-32 and HFC-125 activity coefficients. The regressed parameters were used in Real Adsorbed Solution Theory (RAST) to predict binary adsorption of HFC-125/HFC-32 with %AARDs in composition < 5%.

Open system analyses (OSAs) were performed for silicalite and 5A after conducting breakthrough experiments to assess R‑410A separation performance. After one cycle with silicalite, 53.55% of the HFC‑32 was recovered from R‑410A with 100% purity. Likewise, 52.11% of the HFC‑125 was recovered with 99.94% purity using 5A. R‑410A was processed with a productivity of 0.556 kg/kg‑hr and 0.343 kg/kg‑hr for silicalite and 5A, respectively. Adsorbed phase compositions were determined from the breakthrough experiments and agreed with binary adsorption data.