Polymeric membrane design is a multidimensional process involving selection of membrane materials and optimization of fabrication conditions from an infinite candidate space. It is impossible to explore the entire space by trial-and-error experimentation. Here, we present a membrane design strategy utilizing machine learning-based Bayesian optimization to precisely identify the optimal combinations of unexplored monomers and their fabrication conditions from an infinite space. We developed ML models to accurately predict water permeability and salt rejection from membrane monomer types (represented by the Morgan fingerprint) and fabrication conditions. We applied Bayesian optimization on the built ML model to inversely identify sets of monomer/fabrication condition combinations with the potential to break the upper bound for water/salt selectivity and permeability. We fabricated eight membranes under the identified combinations and found that they exceeded the present upper bound. Our findings demonstrate that ML-based Bayesian optimization represents a paradigm shift for next-generation separation membrane design.

Two-dimensional (2D) material-based membranes hold great promise in wastewater treatment. However, it remains challenging to achieve highly efficient and precise small molecule/ion separation with pure 2D material-fabricated lamellar membranes. In this work, laminated graphene oxide (GO)–cellulose nanocrystal (CNC) hybrid membranes (GO/CNC) were fabricated by taking advantages of the unique structures and synergistic effects generated from these two materials. The characterization results in physiochemical properties, and the structure of the as-synthesized hybrid membranes displayed enhanced membrane surface hydrophilicity, enhanced crumpling surface structure, and slightly enlarged interlayer-spacing with the incorporation of CNCs. Water permeability increases by two to four times with the addition of different CNC weight ratios in comparison to a pristine GO membrane. The optimal GO/CNC membrane achieved efficient rejection toward three typical antibiotics at 74.8, 90.9, and 97.2% for sulfamethoxazole (SMX), levofloxacin (Levo), and norfloxacin (Nor), respectively, while allowing a high passage of desirable nutrients such as NO3 – and H2PO4 –. It was found that SMX removal is primarily governed by electrostatic repulsion, while adsorption plays a crucial role in removing Levo and Nor. Moreover, the density functional theory calculations confirmed the increased antibiotic removal in the presence of an organic foulant, humic acid. Such a 2D material-based hybrid membrane offers a new strategy to develop fit-for purpose membranes for resource recovery and water separation.

A green synthesis method was used to prepare GO–nZVI nanohybrids to provide an adsorbent with high adsorption efficiency that can be removed from aqueous solutions easily by magnetic separation.