Diversifying Databases of Metal Organic Frameworks for High-Throughput Computational Screening

Abstract

By combining metal nodes and organic linkers, an infinite number of metal organic frameworks (MOFs) can be designed in silico. Therefore, when making new databases of such hypothetical MOFs, we need to ensure that they not only contribute toward the growth of the count of structures but also add different chemistries to the existing databases. In this study, we designed a database of ∼20,000 hypothetical MOFs, which are diverse in terms of their chemical design space—metal nodes, organic linkers, functional groups, and pore geometries. Using machine learning techniques, we visualized and quantified the diversity of these structures. We find that on adding the structures of our database, the overall diversity metrics of hypothetical databases improve, especially in terms of the chemistry of metal nodes. We then assessed the usefulness of diverse structures by evaluating their performance, using grand-canonical Monte Carlo simulations, in two important environmental applications—post-combustion carbon capture and hydrogen storage. We find that many of these structures perform better than widely used benchmark materials such as Zeolite-13X (for post-combustion carbon capture) and MOF-5 (for hydrogen storage). All the structures developed in this study, and their properties, are provided on the Materials Cloud to encourage further use of these materials for other applications.

The HEALED SBU Library of Chemically Realistic Building Blocks for Construction of Hypothetical Metal–Organic Frameworks

“…A representative set of 554 inorganic SBUs and 568 organic SBUs were selected from popular topological hMOF construction algorithms and tools, such as pormake, ToBaCCo, and TOBASCCO, and several hMOF databases. ,,− Further, additional inorganic SBUs were mined from the CSD-MOF and CoRE-2019 , experimental databases that were determined to possess diverse metal chemistry based on their calculated revised autocorrelation descriptors . To account for the widespread issues in the experimental MOFs from which these SBUs are typically mined, manual inspection of each inorganic and organic SBU was required to ensure their chemical accuracy.…”

ARC–MOF: A Diverse Database of Metal-Organic Frameworks with DFT-Derived Partial Atomic Charges and Descriptors for Machine Learning

“…We now examine both the chemical and geometric diversity of ARC–MOF, with a comparison to other databases of MOFs. Figure shows the distribution of common geometric descriptors within ARC–MOF in comparison to the CoRE 2019, CSD–MOF, QMOF, and Majumdar databases. The distributions plotted are for the entire databases.…”

“…Other top-down and bottom-up databases of hMOFs have also been developed, including databases where structures were created by functionalizing the pores of a parent MOF . While hMOF databases allow for access to a wider combination of organic and inorganic SBUs, they frequently possess dramatically poorer diversity with respect to the inorganic SBUs when compared to experimental databases. , Depending on the property of interest, using such databases could introduce undesired biases which may reduce efficiency of screening and/or transferability of ML models. Thus far, few studies have focused on correcting this lack of diversity in hMOF databases, with a notable exception being the work by Majumdar et al, in which new hMOFs were generated with underrepresented inorganic SBUs to expand the metal chemistry diversity of the existing hMOF space.…”

Mapping the Porous and Chemical Structure–Function Relationships of Trace CH3I Capture by Metal–Organic Frameworks using Machine Learning

“…One of the most appealing advantages of evaluating MOFs as adsorbents for challenging separations is that they have been realized, experimentally or hypothetically, with a vast variety of pore geometries and chemical functionalities, with the combination of the two further enriching the chemical diversity of MOFs. Significant amounts of computational efforts have been devoted to curating structural databases of experimentally synthesized − or computationally generated MOFs. − These large databases have been screened extensively for functional properties − and have been used to quantify and understand the chemical space of MOFs. , Both molecular simulations, including GCMC and molecular dynamics, and DFT calculations have been extensively used to predict the performances of MOFs for various materials functions, such as gas storage and catalysis. − To probe underlying structure–function correlations in the data generated by large-scale property predictions for MOFs, the MOF structures need to be encoded numerically, often in the form of engineered, domain-knowledge-based descriptors. For materials functions dependent on the material being porous ( e.g.…”