Introducing artificial MOFs for improved machine learning predictions: Identification of top-performing materials for methane storage

Insights into hydrogen and methane storage capacities: Grand canonical Monte Carlo simulations of SIGSUA

In the pursuit of sustainable energy solutions, the development of materials with efficient hydrogen and methane storage capacities is imperative, particularly for advancing hydrogen-powered vehicles. Metal-organic frameworks (MOFs) have emerged as promising candidates to meet the stringent targets set by the Department of Energy for both hydrogen and methane storage. This study employs Grand Canonical Monte Carlo simulations to investigate the usable hydrogen and methane gravimetric and volumetric storage capacities of the recently synthesized SIGSUA. A comparative analysis encompasses the selected MOFs with similar metal compositions, those with comparable density and average pore radius, and classical benchmarks, such as IRMOF-15 and IRMOF-20, all evaluated at room temperature and moderate pressures ranging from 25 to 35 MPa. The results reveal that SIGSUA demonstrates noteworthy gravimetric and volumetric storage capacities for both hydrogen and methane, rivaling or surpassing those of the selected MOFs for analysis. These findings underscore the potential of SIGSUA in advancing clean energy storage technologies.

Rapid and Accurate Screening of the COF Space for Natural Gas Purification: COFInformatics

In this work, we introduced COFInformatics, a computational approach merging molecular simulations and machine learning (ML) algorithms, to evaluate all synthesized and hypothetical covalent organic frameworks (COFs) for the CO2/CH4 mixture separation under four different adsorption-based processes: pressure swing adsorption (PSA), vacuum swing adsorption (VSA), temperature swing adsorption (TSA), and pressure-temperature swing adsorption (PTSA). We first extracted structural, chemical, energy-based, and graph-based molecular fingerprint features of every single COF structure in the very large COF space, consisting of nearly 70,000 materials, and then performed grand canonical Monte Carlo simulations to calculate the CO2/CH4 mixture adsorption properties of 7540 COFs. These features and simulation results were used to develop ML models that accurately and rapidly predict CO2/CH4 mixture adsorption and separation properties of all 68,614 COFs. The most efficient separation process and the best adsorbent candidates among the entire COF spectrum were identified and analyzed in detail to reveal the most important molecular features that lead to high-performance adsorbents. Our results showed that (i) many hypoCOFs outperform synthesized COFs by achieving higher CO2/CH4 selectivities; (ii) the top COF adsorbents consist of narrow pores and linkers comprising aromatic, triazine, and halogen groups; and (iii) PTSA is the most efficient process to use COF adsorbents for natural gas purification. We believe that COFInformatics promises to expedite the evaluation of COF adsorbents for CO2/CH4 separation, thereby circumventing the extensive, time- and resource-intensive molecular simulations.

Accelerating In Silico Discovery of Metal-Organic Frameworks for Ethane/Ethylene and Propane/Propylene Separation: A Synergistic Approach Integrating Molecular Simulation, Machine Learning, and Active Learning

Cryogenic distillation, a currently employed method for C2H4/C2H6 and C3H6/C3H8 mixture separation, is energy-intensive, prompting the research toward alternative technologies, including adsorbent-based separation. In this work, we combine machine learning (ML) technique with high-throughput screening to screen ∼23,000 hypothetical metal-organic frameworks (MOFs) for paraffin (C2H6 and C3H8) selective adsorbent separation. First, structure-based prescreening was employed to remove MOFs with undesired geometric properties. Further, a random forest model built upon the multicomponent grand canonical Monte Carlo (m-GCMC) simulation data of training set MOFs was found to be the most successful in learning the relationship between MOF features and olefin/paraffin mixture separation. Using this technique, the separation performance of the remaining (test set) MOFs was predicted, and the top-performing MOFs were identified. We also employed active learning (AL) to evaluate its effectiveness in improving the prediction of olefin/paraffin selectivity. AL was discovered to be ∼29 times more efficient than the best-supervised ML model, as it was able to identify the top materials in limited training data and at a fraction of computational cost and time as compared to ML techniques. Among the top selected materials, framework chemistry was found to be the most important parameter. Nickel and copper (as a metal node) in a tfzd and hms topological arrangement respectively, were discovered to be a prevalent attribute in high-performing MOFs, further demonstrating the prominent significance of framework chemistry. Additionally, the top MOFs discovered were studied in detail and further compared to the previously reported MOFs. These MOFs show the highest selectivity for C2H4/C2H6 and C3H6/C3H8 mixture separation, as reported until date. The hierarchical strategy devised in this study will facilitate the quick screening of MOFs across multiple databases toward industrially significant separation processes by leveraging molecular simulations and AL.

Leveraging Machine Learning for Metal-Organic Frameworks: A Perspective

Metal-organic frameworks (MOFs) have attracted tremendous interest because of their tunable structures, functionalities, and physiochemical properties. The nearly infinite combinations of metal nodes and organic linkers have led to the synthesis of over 100,000 experimental MOFs and the construction of millions of hypothetical counterparts. It is intractable to identify the best candidates in the immense chemical space of MOFs for applications via conventional trial-to-error experiments or brute-force simulations. Over the past several years, machine learning (ML) has substantially transformed the way of MOF discovery, design, and synthesis. Driven by the abundant data from experiments or simulations, ML can not only efficiently and accurately predict MOF properties but also quantitatively derive structure-property relationships for rational design and screening. In this Perspective, we summarize recent achievements in leveraging ML for MOFs from the aspects of data acquisition, featurization, model training, and applications. Then, current challenges and new opportunities are discussed for the future exploration of ML to accelerate the development of new MOFs in this vibrant field.