Enhancing Ballistic Transport and C₃–C₄ Alcohol Dehydration through Machine Learning-Designed Cationic Graphene Oxide Membranes

Abstract

Machine learning (ML) plays a pivotal role in material design and performance prediction. However, research in ML related to fabricating two-dimensional (2D) graphene oxide (GO) membranes remains limited, facing challenges due to inherent structural variations and the need for precise modifications. Inspired by biological cells, this study highlights the importance of incorporating cations into GO membranes to enhance ballistic transport and alcohol dehydration performance. Through the exploration of different cations, it is identified that the Ca 2+ -GO membrane not only stabilizes the membrane structure by hydrogen bonding interactions, but also maximizes the watercapture ability of GO membranes by electrostatic attractions. For the first time, the CatBoost algorithm is employed in conjunction with Monte Carlo-molecular dynamics simulations to quantitatively assess the correlation and feature importance of operating temperature, chemical group, cationic loadings, cationic size, and its charges with membrane performance. A backpropagation ML algorithm is then developed to generate the post-training response for performance prediction with an accuracy above 0.96. Optimal Ca 2+ -GO performance is predicted at 32.1 mg•g −1 cationic loading, with water separation factors of 5922 and 46,369 for alcohol (C 3 −C 4 ) dehydration, respectively, and water permeance ranging from 48.5 to 123.6 GPU, nearly 10 times higher than commercial membranes. This theoretical study pioneers an accurate ML algorithm to fabricate the cationic GO membranes, serving as a blueprint for developing high-performance 2D membranes for alcohol dehydration.