Although the tailored metal active sites and porous architectures of MOFs hold great promise for outstanding challenges in catalysis, a lack of understanding of how to improve their stability limits their use in practice. I will describe our recent efforts to overcome this limitation. I will describe how we have extracted thousands of published experimental reports of the key aspects of MOF stability necessary for their practical application in catalysis: the ability to withstand high temperatures without degrading and the capacity to be activated by removal of solvent molecules. Natural language processing and image analysis was used to obtain thousands of quantitative measures of stability. I will describe 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, it is possible to make predictions of stability orders of magnitude faster than conventional physics-based 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 3<i>d</i>-transition-metal-containing MOFs that are frequently targeted for catalytic applications. Time-permitting, I will also describe how we are using these tools to discover new catalysts and the public web interface we have developed to encourage the community to engage with and improve upon our data set.

Using machine learning and data mining to leverage community knowledge for the engineering of materials and catalysts

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.