(191d) Accelerated Prediction of Metal–Organic Framework Electronic Properties Via High-Throughput Quantum-Chemical Calculations and Machine Learning

Metal–organic frameworks (MOFs) are a widely investigated class of extended solids with tunable structures that make it possible to impart specific chemical functionality for a given application. However, the enormous number of possible MOFs that can be synthesized makes it difficult to determine which materials would be most promising, especially for applications governed by electronic structure properties such as photocatalysis and novel (opto)electronic devices. Here, we present the Quantum MOF (QMOF) Database¹ — a publicly available database of geometric, energetic, and electronic properties for 20,000+ MOFs and coordination polymers derived from high-throughput density functional theory calculations. Throughout this talk, we showcase how this new database of predicted properties can aid both theorists and experimentalists alike in the search for MOFs with targeted electronic properties. Furthermore, we highlight how the QMOF Database can be used to train machine learning models that accurately predict the properties of new MOFs at a small fraction of the computational cost of conventional quantum-mechanical calculations. A top-performing graph neural network model is then used to rapidly identify MOFs with unusually low band gaps, a challenging task given the electronically insulating nature of most MOF structures. Recent work that uses this dataset to uncover fundamental errors in density functional theory will also be discussed.² We conclude with a demonstration of how to access and explore the QMOF Database, including via a new web application on the Materials Project (https://materialsproject.org/mofs), and how anyone can run similar high-throughput quantum-mechanical calculations through the newly developed Quantum Accelerator (QuAcc) code (https://github.com/arosen93/quacc).

1. A.S. Rosen, S.M. Iyer, D. Ray, Z. Yao, A. Aspuru-Guzik, L. Gagliardi, J.M. Notestein, R.Q. Snurr, Machine Learning the Quantum-Chemical Properties of Metal-Organic Frameworks for Accelerated Materials Discovery, Matter, 4, 1578–1597 (2021).

2. A.S. Rosen, V. Fung, P. Huck, C.T. O’Donnell, M.K. Horton, D.G. Truhlar, K.A. Persson, J.M. Notestein, R.Q. Snurr, High-Throughput Predictions of Metal–Organic Framework Electronic Properties: Theoretical Challenges, Graph Neural Networks, and Data Exploration, ChemRxiv (2022). DOI: https://doi.org/10.26434/chemrxiv-2021-6cs91.