Accelerating Discovery Through Integration: A DFT validated Machine Learning Framework for Screening MOF Photocatalysts

Abstract

The discovery of Metal-Organic Framework (MOF) photocatalysts for CO2 reduction is hindered by the computational cost of quantum chemical screenings. To overcome this barrier, we introduce a Machine Learning (ML) accelerated workflow that integrates the speed of ML with the accuracy of Density Functional Theory (DFT). While a DFT-based screening of over 20,000 MOFs identified 105 promising candidates in nearly a month, a ML-driven approach using the Molecular Graph Transformer (MGT) required only 4.5 hours. We show that a naive candidate selection strategy using ML predictions yield a low candidate recovery rate of 20%; however, by implementing a selection strategy informed by the model's error, this rate can be increased to 70%. Although ML models can exhibit inconsistencies, particularly with chemically complex systems, their significant speed advantage makes them ideal for an integrated strategy. We therefore propose a work ow where ML acts as a high-speed primary filter to identify a manageable set of candidates for subsequent high-fidelity DFT calculations, effectively combining the strengths of both methods to accelerate materials discovery.