MOFSimplify, machine learning models with extracted stability data of three thousand metal–organic frameworks

Abstract

We report a workflow and the output of a natural language processing (NLP)-based procedure to mine the extant metal–organic framework (MOF) literature describing structurally characterized MOFs and their solvent removal and thermal stabilities. We obtain over 2,000 solvent removal stability measures from text mining and 3,000 thermal decomposition temperatures from thermogravimetric analysis data. We assess the validity of our NLP methods and the accuracy of our extracted data by comparing to a hand-labeled subset. Machine learning (ML, i.e. artificial neural network) models trained on this data using graph- and pore-geometry-based representations enable prediction of stability on new MOFs with quantified uncertainty. Our web interface, MOFSimplify, provides users access to our curated data and enables them to harness that data for predictions on new MOFs. MOFSimplify also encourages community feedback on existing data and on ML model predictions for community-based active learning for improved MOF stability models.

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

“…These firstgeneration ML models were naturally constrained in their performance, in part by the representations or ML models themselves but primarily by the amount of data that could be easily extracted from the literature. Toward that end, we have provided our ML models and extracted properties from the data sets in a user-friendly web interface, which welcomes both feedback and the incorporation of additional community data 75 . We expect our approach, including natural extensions to other properties, to accelerate the time to discovery of stable, practical MOF materials by both computational and experimental researchers.…”

“…However, expert review of the literature could always be expected to improve the quality of our data sets and models. Toward that end, we have provided our ML models and extracted properties from the data sets in a user-friendly web interface, which welcomes both feedback and the incorporation of additional community data . The web interface also supports the uploading of new MOFs not in our data set, allowing researchers to calculate both the RAC features and ML-predicted properties on these new materials.…”

A State-of-the-Art of Metal-Organic Frameworks for Chromium Photoreduction vs. Photocatalytic Water Remediation

“…Here is where machine learning could play a disruptive role during the coming years in order to identify, screen, classify, and correlate MOFs’ potentials on the basis of geometric, chemical, topological, energetic, and performance-based descriptors [ 316 , 317 ]. In fact, machine learning is already having a deep impact on unraveling synthesis paths and engineering strategies of MOFs for gas adsorption and separation purposes [ 318 , 319 , 320 , 321 ]. As far as the investigations of MOFs for photo-oxidative and photoreductive processes expand, it is more likely that machine learning could be applied to unravel the underpinning chemical–physical features that make the MOFs feasible for this application.…”

MOFTransformer: A Multi-modal Pre-training Transformer for Universal Transfer Learning in Metal-Organic Frameworks

“…Although both the local descriptors (e.g. CGCNN, 19,21 chemical descriptors, 18 RACs, 22,23 and building-block embedding. 11,24,25 ) and the global features (e.g., geometric features calculated by ZEO++, 26 the histograms of energy-grids.…”