Multiscale mechanistic design of hydrophobic natural DES for PFAS extraction: from quantum chemistry to high-throughput screening

Abstract

Per- and polyfluoroalkyl substances (PFAS) are persistent, toxic pollutants that pose serious challenges for water remediation due to their resistance to degradation and the limitations of existing treatment technologies. In this study, we explore a nature-inspired strategy for PFAS extraction from water using hydrophobic natural deep eutectic solvents (HNADES), applying a multiscale in silico approach. Through quantum-chemical methods rooted in Density Functional Theory (DFT), COnductor-like Screening MOdel for Real Solvents (COSMO-RS) thermodynamic modeling and classical molecular dynamics (MD) simulations, we demonstrate that the selected HNADES exhibits strong affinity for three representative PFAS (PFOA, PFOS and HFPO-DA). This affinity arises from robust hydrogen bonding between PFAS acidic proton of the terminal functional group and the carbonyl group of decanoic acid, supported by favorable van der Waals interactions. The extraction process is further stabilized by internal hydrogen bonding within the HNADES, preserving its structural integrity. These molecular-level interactions translate into high PFAS solubilities and activity coefficients approaching zero, effectively enabling the disruption of PFAS hydration shells and facilitating their migration across the (HNADES)-(PFAS + water) interface into the organic phase. A strong correlation was observed between molecular descriptors and mixture behavior, laying the groundwork for predictive models of extraction efficiency. Building on these insights, we conducted an extended virtual screening of 1746 PFAS against 2589 HNADES combinations, identifying promising candidates for selective, scalable and environmentally friendly remediation technologies, with the most hydrophobic HNADES—primarily those containing linalool, verbenone and cineole—solubilizing the widest range of PFAS.