Discovery of Innovative Polymers for Next-Generation Gas-Separation Membranes using Interpretable Machine Learning

Abstract

Polymer membranes perform innumerable separations with far-reaching environmental implications. Despite decades of research on membrane technologies, design of new membrane materials remains a largely Edisonian process. To address this shortcoming, we demonstrate a generalizable, accurate machine-learning (ML) implementation for the discovery of innovative polymers with ideal separation performance. Specifically, multitask ML models are trained on available experimental data to link polymer chemistry to gas permeabilities of He, H2, O2, N2, CO2, and CH4. We interpret the ML models and extract chemical heuristics for membrane design, through Shapley Additive exPlanations (SHAP) analysis. We then screen over nine million hypothetical polymers through our models and identify thousands of candidates that lie well above current performance upper bounds. Notably, we discover hundreds of never-before-seen ultrapermeable polymer membranes with O2 and CO2 permeability greater than 104 and 105 Barrer, respectively, orders of magnitude higher than currently available polymeric membranes. These hypothetical polymers are capable of overcoming undesirable trade-off relationship between permeability and selectivity, thus significantly expanding the currently limited library of polymer membranes for highly efficient gas separations. High-fidelity molecular dynamics simulations confirm the ML-predicted gas permeabilities of the promising candidates, which suggests that many can be translated to reality.

Predicting Extraction Selectivity of Acetic Acid in Pervaporation by Machine Learning Models with Data Leakage Management

“…To evaluate whether the CBR model (depth = 6, learning rate = 0.1, iterations = 300, L2 leaf regularization = 5) appropriately represents known separation mechanisms of acetic acid separation from water in PV and to interpret the relative importance of key variables for future material and system design, we analyzed Shapley values for each input variable in the CBR model, which quantitatively defines the contribution of each variable to the prediction parameter (i.e., separation factor). For each variable, relatively low values (less than the mean) are represented in blue, and relatively high values (greater than the mean) are represented in red. For example, for experimental temperatures, red and blue colors represent high and low temperatures, respectively.…”

Emerging Trends in Machine Learning: A Polymer Perspective

“…In the context of polymers, SHAP has been used to investigate the contributions of various features. ,,, For example, it has been used to determine which functional groups and polymer properties are most predictive of membrane permeability . It has also been used to look at the effect of monomer type and degree of polymerization on protein stability for polymer–protein hybrids.…”

“…34,79,106,107 For example, it has been used to determine which functional groups and polymer properties are most predictive of membrane permeability. 106 It has also been used to look at the effect of monomer type and degree of polymerization on protein stability for polymer−protein hybrids. In this example, they also used active learning and probed how the SHAP values changed as a function of the iteration.…”