A Benchmark Framework for the Fast and Reliable Evaluation of Zeolite Surface Stability

Accurately identifying the external surface structure and crystal morphology of zeolites is essential for establishing structure–performance relationships. Surface free energy serves as a quantitative descriptor of surface stability, which is a prerequisite for theoretical investigations of surface properties. However, a standard computational protocol that effectively balances accuracy and efficiency has yet to be established. To address this, we propose an efficient computational workflow developed through systematic benchmarking. This workflow integrates two key components: an analytical model developed using the SISSO machine learning approach to replace expensive frequency calculations and a configuration screening process based on machine learning potentials to account for the vast number of possible surface terminations. Using this workflow, we efficiently calculate the surface free energies of MFI and FAU zeolites and explore their morphological evolution under dynamic conditions. This study establishes a general and efficient theoretical framework for investigating the thermodynamic properties of zeolite surfaces.