27Al NMR chemical shifts in zeolite MFI via machine learning acceleration of structure sampling and shift prediction

Zeolites, such as MFI, are versatile microporous aluminosilicate materials that are widely used in catalysis and adsorption processes. The location of the aluminium within the zeolite framework is one of the important determinants of performance in industrial applications, and is typically probed by 27Al NMR spectroscopy. However, interpretation of 27Al NMR spectra is challenging, while first-principles computational modelling struggles to achieve the timescales and model complexity needed to provide reliable assignments. In this study, we deploy advanced machine learning-based methods to predict 27Al chemical shifts, complemented by molecular dynamics simulations with neural network potentials to achieve significant speed-up compared to traditional density functional theory (DFT) approaches, while maintaining high accuracy. This allows us to comprehensively explore various conditions relevant to catalysis, including water loading, temperature, and the relative positions of aluminium (pairs). We demonstrate that both water content and temperature significantly affect the chemical shift and do so in a non-trivial way that is highly T-site dependent, highlighting a need for adoption of realistic, case-specific models. Notably, we are able, based on quantitative agreement with relevant experimental data, to assign experimental NMR peaks to specific T-sites, even in such a complex zeolite as MFI. These findings provide a testament to the capabilities of machine learning approaches in providing reliable predictions of important spectroscopic observables for complex industrially relevant materials under realistic conditions.