Mechanistic Insights into PFAS Rejection in Nanofiltration and Reverse Osmosis from Data-Driven Analysis

Area of Science:

• Environmental Science
• Chemical Engineering
• Materials Science

Background:

• Per- and polyfluoroalkyl substances (PFAS) exhibit complex transport behaviors in membrane systems.
• Characterizing PFAS removal universally is challenging due to their structural diversity and varied interactions.

Purpose of the Study:

• To develop a unified, data-driven framework for understanding PFAS removal in membrane systems.
• To model the influence of solute, membrane, and solution properties on PFAS rejection using machine learning.
• To identify critical data gaps and guide future membrane design and treatment strategies.

Main Methods:

• Compiled a comprehensive dataset of 2353 literature data points on PFAS rejection by nanofiltration and reverse osmosis membranes.
• Utilized machine learning to model the relationships between experimental system descriptors and PFAS removal.
• Analyzed the impact of 13 system descriptors, including membrane properties, PFAS characteristics, and feedwater composition.

Main Results:

• Membrane water permeance and PFAS molecular volume were the strongest predictors of rejection, highlighting steric exclusion.
• Background ions and dissolved organic matter showed condition-dependent, nonmonotonic effects on PFAS removal.
• Low concentrations of organic matter and ions enhanced rejection via complexation, while high concentrations decreased it through charge shielding and concentration polarization.

Conclusions:

• The study provides a unified framework for interpreting inconsistent findings in PFAS membrane removal literature.
• Steric exclusion is a primary mechanism, but solute-membrane-solution interactions are complex and condition-specific.
• Results offer insights for optimizing membrane processes for effective PFAS remediation.