Discovery of Sustainable Refrigerants through Physics-Informed RL Fine-Tuning of Sequence Models
This work addresses the problem of finding environmentally friendly refrigerants for air-conditioning systems, representing a domain-specific incremental advance by combining existing methods with physics constraints.
The paper tackles the scarcity of reliable data for discovering sustainable refrigerants by introducing Refgen, a generative pipeline that integrates machine learning with physics-grounded inductive biases, resulting in de novo discovery of molecules balancing efficiency, safety, and environmental impact.
Most refrigerants currently used in air-conditioning systems, such as hydrofluorocarbons, are potent greenhouse gases and are being phased down. Large-scale molecular screening has been applied to the search for alternatives, but in practice only about 300 refrigerants are known, and only a few additional candidates have been suggested without experimental validation. This scarcity of reliable data limits the effectiveness of purely data-driven methods. We present Refgen, a generative pipeline that integrates machine learning with physics-grounded inductive biases. Alongside fine-tuning for valid molecular generation, Refgen incorporates predictive models for critical properties, equations of state, thermochemical polynomials, and full vapor compression cycle simulations. These models enable reinforcement learning fine-tuning under thermodynamic constraints, enforcing consistency and guiding discovery toward molecules that balance efficiency, safety, and environmental impact. By embedding physics into the learning process, Refgen leverages scarce data effectively and enables de novo refrigerant discovery beyond the known set of compounds.