Data‐Driven Strategies for Accelerated Structural Exploration of High‐Performance 2D Carbon‐Based Seawater Desalination Membranes

Abstract

The insufficiency of freshwater supplies has posed a serious threat to sustainable socioeconomic growth, and seawater desalination is considered to be the most promising solution to alleviate such pressure. Currently, 2D carbon membranes are identified as deserving candidates due to their high permeability and multiple tunable properties. However, they remain challenging to systematically uncover the potential relationships between structures and properties in various 2D carbon materials. For this, a machine learning (ML) model based on feature datasets of 2D carbon materials effecting desalination properties is trained. The results suggest that structures with a maximum pore size of 10–12 atoms and atomic densities between 0.28 and 0.41 are more likely to achieve high properties. Cml‐MOR based on MOR‐type mordenite zeolite for validation is selected. Further, Cml‐MOR is demonstrated to feature remarkable salt ion adsorption. The effective water flux of Cml‐MOR is 113.51 L cm−2 day−1 MPa−1, and the salt rejection at 110 MPa can reach 98.9%. This work is expected to apply this efficient method to investigate the structure and properties of 2D carbon membranes with great structural diversity; this will attract more people to focus on them and explore their important potential for practical applications.