Deep learning and big data mining for Metal-Organic frameworks with high performance for simultaneous desulfurization and carbon capture.

Carbon capture and desulfurization of flue gases are crucial for the achievement of carbon neutrality and sustainable development. In this work, the "one-step" adsorption technology with high-performance metal-organic frameworks (MOFs) was proposed to simultaneously capture the SO2 and CO2. Four machine learning algorithms were used to predict the performance indicators (NCO2+SO2, SCO2+SO2/N2, and TSN) of MOFs, with Multi-Layer Perceptron Regression (MLPR) showing better performance (R2 = 0.93). To address sparse data of MOF chemical descriptors, we introduced the Deep Factorization Machines (DeepFM) model, outperforming MLPR with a higher R2 of 0.95. Then, sensitivity analysis was employed to find that the adsorption heat and porosity were the key factors for SO2 and CO2 capture performance of MOF, while the influence of open alkali metal sites also stood out. Furthermore, we established a kinetic model to batch simulate the breakthrough curves of TOP 1000 MOFs to investigate their dynamic adsorption separation performance for SO2/CO2/N2. The TOP 20 MOFs screened by the dynamic performance highly overlap with those screened by the static performance, with 76 % containing open alkali metal sites. This integrated approach of computational screening, machine learning, and dynamic analysis significantly advances the development of efficient MOF adsorbents for flue gas treatment.

MOF-GRU: A MOFid-Aided Deep Learning Model for Predicting the Gas Separation Performance of Metal-Organic Frameworks.

The remarkable versatility of metal-organic frameworks (MOFs) stems from their rich chemical information, leading to numerous successful applications. However, identifying optimal MOFs for specific tasks necessitates a thorough assessment of their chemical attributes. Conventional machine learning approaches for MOF prediction have relied on intricate chemical and structural details, hampering rapid evaluations. Drawing inspiration from recent advancements exemplified by Snurr et al., wherein a text string was used to represent a MOF (MOFid), we introduce a MOFid-aided deep learning model, named the MOF-GRU model. This model, founded on natural language processing principles and utilizing the gated recurrent unit architecture, leverages the serialized text string representation of metal-organic frameworks (MOFs) to forecast gas separation performance. Through a focused study on CH4/N2 separation, we substantiate the efficacy of this approach. Comparative assessments against traditional machine learning techniques underscore our model's superior predictive accuracy and its capacity to handle extensive data sets adeptly. The MOF-GRU model remarkably uncovers latent structure-performance relationships with only MOF sequences, obviating the necessity for intricate three-dimensional (3D) structural information. Overall, this model's judicious design empowers efficient data utilization, thereby hastening the discovery of high-performance materials tailored for gas separation applications.

Interpretable Machine-Learning and Big Data Mining to Predict Gas Diffusivity in Metal-Organic Frameworks.

For gas separation and catalysis by metal-organic frameworks (MOFs), gas diffusion has a substantial impact on the process' overall rate, so it is necessary to determine the molecular diffusion behavior within the MOFs. In this study, an interpretable machine learing (ML) model, light gradient boosting machine (LGBM), is trained to predict the molecular diffusivity and selectivity of 9 gases (Kr, Xe, CH4 , N2 , H2 S, O2 , CO2 , H2 , and He). For these 9 gases, LGBM displays high accuracy (average R2 = 0.962) and superior extrapolation for the diffusivity of C2 H6 . And this model calculation is five orders of magnitude faster than molecular dynamics (MD) simulations. Subsequently, using the trained LGBM model, an interactive desktop application is developed that can help researchers quickly and accurately calculate the diffusion of molecules in porous crystal materials. Finally, the authors find the difference in the molecular polarizability (ΔPol) is the key factor governing the diffusion selectivity by combining the trained LGBM model with the Shapley additive explanation (SHAP). By the calculation of interpretable ML, the optimal MOFs are selected for separating binary gas mixtures and CO2 methanation. This work provides a new direction for exploring the structure-property relationships of MOFs and realizing the rapid calculation of molecular diffusivity.

Large-Scale Screening and Machine Learning for Metal-Organic Framework Membranes to Capture CO2 from Flue Gas.

To combat global warming, as an energy-saving technology, membrane separation can be applied to capture CO2 from flue gas. Metal-organic frameworks (MOFs) with characteristics like high porosity have great potential as membrane materials for gas mixture separation. In this work, through a combination of grand canonical Monte Carlo and molecular dynamics simulations, the permeability of three gases (CO2, N2, and O2) was calculated and estimated in 6013 computation-ready experimental MOF membranes (CoRE-MOFMs). Then, the relationship between structural descriptors and permeance performance, and the importance of available permeance area to permeance performance of gas molecules with smaller kinetic diameters were found by univariate analysis. Furthermore, comparing the prediction accuracy of seven classification machine learning algorithms, XGBoost was selected to analyze the order of importance of six structural descriptors to permeance performance, through which the conclusion of the univariate analysis was demonstrated one more time. Finally, seven promising CoRE-MOFMs were selected, and their structural characteristics were analyzed. This work provides explicit directions and powerful guidelines to experimenters to accelerate the research on membrane separation for the purification of flue gas.