Aluminum Siting in Zeolite RTH From a Combined Machine Learning - NMR Approach

Determining the distribution of aluminum in zeolite frameworks remains a significant challenge, due to the limited sensitivity of conventional characterization techniques and the complexity of possible Al configurations. In this work, we have developed a procedure which combines experimental two-dimensional (2D) dipolar-based 27Al–29Si D-HMQC and MQMAS solid-state NMR spectroscopy with machine learning (ML) techniques to achieve a detailed and quantitative assignment of Al environments in zeolites. We applied this approach to investigate the Al distribution in RTH-type zeolites (Si/Al = 16 and 28), which contain four crystallographically distinct T-sites. A reactive ML potential was used to model framework dynamics under realistic conditions, including temperature and hydration, while kernel ridge regression enabled the accurate prediction of isotropic NMR chemical shifts. Comparison between theoretical and experimental spectra reveals that Al preferentially occupies the T2 (39 %) and T1 (27 %) sites, with lower populations in T3 (18 %) and T4 (16 %). The excellent agreement between computed and observed NMR data demonstrates that this ML- augmented spectroscopic approach is a powerful tool for quantitative elucidation of Al distributions in structurally complex zeolites via solid state NMR.