Reveal the major factors controlling quinolone adsorption on mesoporous carbon: Batch experiment, DFT calculation, MD simulation, and machine learning modeling

Abstract

Mesoporous carbon materials (MCs) exhibit excellent adsorption capacity for various pollutants, yet the adsorption capacity varies significantly through the type of pollutants and the property of MCs, and the factors controlling the adsorption are unclear. Herein, we investigated the adsorption behaviors of 15 quinolones (QNs) onto four kinds of MCs and further explored the major factors driving their adsorption using density functional theory (DFT) calculation, molecular dynamics (MD) simulation, and machine learning modeling. The adsorption data of QNs on MCs conformed to both linear and Freundlich isotherms. DFT calculation revealed the stronger π-π interactions than hydrogen bonding between QNs and MCs, while redundancy analysis unveiled the contribution of hydrophobic interactions (logKow) to QNs adsorption. MD simulations displayed the shoulder peaks near the pore surface and peaks in the pore space of MCs for QNs molecules, indicating the multilayer adsorption and weak pore filling effects of QNs onto MCs. Eight factors were selected through the machine learning modeling, with the importance order of hydrogen bond accepting ability (B) > binding energy of hydrogen bonding > logKow > O-C% > BET surface area > pore volume > hydrogen bond donating ability (A) > binding energy of π-π interaction. This study provides new insights into the driving factors on the adsorption of QNs on MCs, and offers a novel perspective for revealing the adsorption mechanisms of pollutants onto various adsorbents.