Machine Learning to Enhance PFAS Degradation in Flow Reactor

The “forever chemicals”, per- and poly-fluoroalkyl substances (PFAS, >4700 species), have been widely used in the textile industry, water-proof coatings, food packages, and fire-fighting foams. PFAS contamination has grown to be a worldwide problem, garnering public, scientific, and regulatory concern due to their persistence, bioaccumulation, mobility, and high toxicity to living organisms. Driven by the “Zero-Pollution Ambition”, both EU and US have established near-zero regulations for PFAS. This synergy project aims to integrate machine learning-based algorithms and automated flow reactors to achieve complete degradation of PFAS while preserving energy consumption. By identifying limiting factors (e.g., flow rate, sulfite dosage, and UV intensity), we will develop a self-optimizer of PFAS degradation as the function of the identified factors through machine learning integrated with the knowledge of PFAS degradation. This will not only allow in depth understanding of mechanisms of PFAS degradation, but also provide a digital platform for process tunability and optimization.