(691d) Using Machine Learning Interatomic Potentials to Understand Water Structure in Zeolites for Sustainable Fuels

Water is essential in various industrial processes, particularly in the production of fuel and chemicals utilizing ethanol and other bio-based feedstocks. Zeolites are porous catalysts used extensively due to their shape-selective adsorption and confinement interactions. Zeolite effectiveness is significantly impacted by the presence of water, which may reduce product selectivity and reactivity depending on the polarity of transition states. Using advanced computational techniques including ab-initio molecular dynamics (AIMD) and machine learning force fields (MLFF), we examined the stability of various water loadings in zeolite AFI with different Si-Al ratios and Al distributions. This analysis enables us to identify the water distribution in the pore, revealing the formation of water clusters in acid zeolites. This further assists in uncovering the intricate molecular details of complex solvated reactions such as ethanol to olefins for sustainable aviation fuels.

Our analysis reveals that the behavior of water in zeolites is influenced by the Si/Al ratio and the distribution of Al in the framework. We find the average adsorption energy per water molecule is distinct for each Al distribution, as determined by both MD and DFT calculations. Interestingly, we observe that the relationship between the average adsorption energy and the Al distribution differs between MD and DFT, due to the kinetics involved. Further, we found that the peaks in the vibrational density of states (vDOS) of water in Al-defected AFI are shifted, particularly at the saturation water loading. Our radial distribution function (RDF) analysis and heat maps reveal that, with increasing water loading, water molecules tend to distribute evenly throughout the entire pore in zeolite AFI. Our results provide crucial insights into the molecular-level behavior of water in zeolites which we are now applying to understand ethanol to olefins kinetics when water is present.