Bayesian optimization enhanced neural networks for predicting metal-organic framework morphology: A ZIF-8 synthesis case study

Abstract

Zeolitic imidazolate framework-8 (ZIF-8) is a metal–organic framework (MOF) for diverse applications, including drug delivery and gas storage, where its morphology such as size significantly affects controlled drug release and gas absorption. Traditional trial-and-error methods for optimizing ZIF-8 morphology are time consuming and inefficient. To improve this, we integrate machine learning with experimental studies by developing a neural network model with hyperparameters optimized through Bayesian optimization. This model predicts how different experimental conditions such as precursor concentrations affect ZIF-8 morphology, thus setting the foundation for synthesis optimization that can improve morphology control.