Framework Flexibility on Water Sorption and Transport in ZIF-90: Insights from Neural-Network Molecular Dynamics

Abstract

Metal–organic frameworks (MOFs) are promising materials for thermochemical energy storage, yet the role of the flexibility of the framework in governing water adsorption and transport is not adequately understood at the atomistic level. In this work, we employ quantum assisted neural network-based molecular dynamics simulations (DeepMD) to systematically investigate the water uptake, adsorption enthalpies, diffusion coefficients, and structural correlations of a hydrophilic MOF (ZIF-90) with both rigid and flexible frameworks. The simulated results showed that the flexible ZIF-90 exhibits a nearly constant heat of adsorption between –46 and –64 kJ/mol in a wide water loading range (4, 8, 16, 25 and 40 per unit cell), in good agreement with the experimental findings, whereas the rigid framework shows a sharp variation. Similarly, diffusion analysis consistently reveals higher water mobility in the flexible framework, particularly at low and intermediate loadings. The radial distribution functions for key atom pairs (Zn–Ow, O–Hw, O–Ow) show excellent agreement with previous experimental and computational reports, which further validates our simulations. Overall, these findings provide molecular-level insights into the interplay between framework flexibility, adsorption energetics, and transport properties, offering guidance for the rational design of porous materials for thermal energy storage applications.