Data-Driven Insights into Resin Screening for Targeted Per- and Polyfluoroalkyl Substances Removal Using Machine...

Area of Science:

• Environmental Chemistry
• Water Treatment Technologies
• Computational Chemistry

Background:

• Per- and polyfluoroalkyl substances (PFASs) are persistent environmental contaminants found in diverse water matrices.
• Effective removal of a wide range of PFASs, including long- and short-chain variants, remains a significant challenge.
• Traditional methods for resin screening and operational optimization are often time-consuming and resource-intensive.

Purpose of the Study:

• To develop and validate machine learning (ML) models for predicting PFAS removal efficiency.
• To identify key factors influencing PFAS removal by resins across different water conditions.
• To utilize ML for inverse screening of resins and optimizing operational parameters for targeted PFAS removal.

Main Methods:

• Development of ML models correlating resin properties, operational conditions, and water matrix characteristics with PFAS removal efficiency.
• Validation of ML model performance using independent test datasets and experimental verification.
• Comprehensive investigation of feature importance and interaction effects on PFAS removal.
• Application of ML models for inverse screening and optimization of resin selection and operating conditions.

Main Results:

• ML models accurately predict PFAS removal efficiency across diverse water matrices.
• Key resin properties, operational conditions, and water matrix features influencing PFAS removal were identified.
• The ML-guided approach successfully identified optimal resins and conditions, achieving high removal efficiencies (e.g., 86.56% for PFBS, 83.73% for GenX).
• Experimental validation confirmed the efficacy of the ML-driven strategy.

Conclusions:

• Machine learning methodologies offer a powerful tool for efficient resin screening and operational optimization in PFAS removal.
• The developed ML models enable targeted removal of structurally diverse PFAS compounds across various water matrices.
• This approach significantly accelerates the development of effective water treatment strategies for persistent contaminants like PFASs.