Machine Learning-Assisted High-Throughput Screening of Metal–Organic Frameworks for CO<sub>2</sub> Separation from CO<sub>2</sub>-Rich Natural Gas
Yinjie Zhou, Sibei Ji, Songyang He, Wei Fan, Liang Zan, Li Zhou, Xu Ji, Ge He
Abstract
Under the appeal of carbon peaking and carbon neutrality goals, it is highly advisable to develop green chemical technologies. Based on this, it is even more attractive to synthesize methanol with the H 2 generated from water electrolysis by offshore wind power and the CO 2 separated from offshore CO 2 -rich natural gas. Therefore, the separation and adsorption of CO 2 -rich natural gas in this context is of great socioeconomic significance. However, the conventional high-throughput screening methods for metal–organic frameworks (MOFs) in separating natural gas components and CO 2 suffer from great challenges such as high model complexity and long computation time. To address the aforementioned problems, a machine learning-assisted modeling and screening strategy is proposed herein for the rapid and efficient separation of CO 2 from the actual natural gas of six components (N 2, CO 2, CH 4, C 2 H 6, C 3 H 8, and H 2 S). First, structural analysis is used to eliminate the MOFs that cannot adsorb CO 2 from the Computation-Ready Experimental Metal–Organic Frameworks (CoRE-MOFs) database. Six structural and 17 chemical descriptors of the remaining MOFs were calculated. Grand Canonical Monte Carlo (GCMC) simulations were applied to evaluate the separation performance metrics of the randomly selected training and testing MOF samples. By combining 23 descriptors and separation performance metrics, a Random Forest (RF) regression model was obtained with R 2 exceeding 0.92 on the test samples, which was employed to predict the separation performance of the remaining MOFs. As a result, 10 MOF candidates with the best CO 2 separation performance were obtained. Furthermore, a structure–property relationship of MOFs with satisfactory regenerability was conducted. Three design strategies were proposed to guide the development of high-performance novel MOFs for CO 2 separation. This study offers a high-throughput screening framework for MOFs to facilitate the separation of CO 2 from a CO 2 -rich natural gas.