Using Machine Learning and Data Mining to Leverage Community Knowledge for the Engineering of Stable Metal–Organic Frameworks

Abstract

Although the tailored metal active sites and porous architectures of MOFs hold great promise for engineering challenges ranging from gas separations to catalysis, a lack of understanding of how to improve their stability limits their use in practice. To overcome this limitation, we extract thousands of published reports of the key aspects of MOF stability necessary for their practical application: the ability to withstand high temperatures without degrading and the capacity to be activated by removal of solvent molecules. From nearly 4,000 manuscripts, we use natural language processing and automated image analysis to obtain over 2,000 solvent-removal stability measures and 3,000 thermal degradation temperatures. We analyze the relationships between stability properties and the chemical and geometric structures in this set to identify limits of prior heuristics derived from smaller sets of MOFs. By training predictive machine learning (ML, i.e., Gaussian process and artificial neural network) models to encode the structure-property relationships with graph-and pore-structure-based representations, we are able to make predictions of stability orders of magnitude faster than conventional physicsbased modeling or experiment. Interpretation of important features in ML models provides insights that we use to identify strategies to engineer increased stability into typically unstable 3d-containing MOFs that are frequently targeted for catalytic applications. We expect our approach to accelerate the time to discovery of stable, practical MOF materials for a wide range of applications.

Recent progress, mechanisms, and perspectives for crystal and interface chemistry applying to the Zn metal anodes in aqueous zinc‐ion batteries

“…Researchers should provide predictive models through machine learning (i.e., deep learning and artificial neural network) to achieve a flexible and robust strategy model mainly based on crystal plane matching and reversible SEI. 121,125 These well-trained models predict two orders of magnitude faster than traditional screening techniques via experiments. 4.…”

“…The analyses of the relationship between the prone problems and the protection strategy show that the development of one or more heuristic models will achieve a wider improvement for the research in the future. Researchers should provide predictive models through machine learning (i.e., deep learning and artificial neural network) to achieve a flexible and robust strategy model mainly based on crystal plane matching and reversible SEI 121,125 . These well‐trained models predict two orders of magnitude faster than traditional screening techniques via experiments. In the case of practical or industrial application, things get complicated because the current studies remained in the stage of the basic academy.…”

High-throughput computational screening of nanoporous materials in targeted applications

“…Later, Nandy et al performed a meta-analysis of thousands of articles associated to the CoRE MOF 2019 database to extract their experimental solvent-removal stability and thermal decomposition temperature. 150 These data were then leveraged in the training of multiple ML models to predict stability; such predictions can be very useful to gauge the relative stability of each material and to restrict screening to only structures considered experimentally stable. Other types of materials have been explored, Turcani et al published 60 000 organic cage structures and used machine learning to predict their stability based on the shape persistence metric.…”

Connecting the Dots for Fundamental Understanding of Structure-Photophysics-Property Relationships of COFs, MOFs, and Perovskites using a Multiparticle Holstein Formalism

“…Skillful exploration of data through data-mining and big data approaches as well as electronic structure modeling have greatly benefited the fields of MOFs and perovskites. 103,[110][111][112][113][114][115][116][117][118][119] Machine learning and high throughput techniques have been routinely employed to screen the chemical space of hybrid materials with the goal of proposing new MOF and perovskite structures that are suitable for different applications, including but not limited to catalysis, [120][121][122][123] gas storage and separation, 124 and optoelectronics. [125][126][127][128] While materials modeling based on data-driven approaches has achieved some remarkable success in predicting electronic properties (e.g., band gaps, charge densities, electronic couplings, oscillator strengths, etc.…”