Understanding Water Diffusivity in Ultrathin Polyamide Membranes Starting from the Molecular Level

Abstract

Accurate relationships between water diffusivity and chemical–morphological characteristics of membranes are mandatory for membrane material optimizations. To meet this challenge, careful post-processing of previous extensive Molecular Dynamics (MD) simulations was performed in this work to obtain the distribution of water molecules in nanometer-thick polyamide active layers, both at equilibrium and during water permeation. The polyamide active layer was modeled as a linear, uncross-linked TMC–MPD network, representing the low-crosslink-density limit of the FT-30 chemistry. This allowed us to gain insights into molecular-level structures and macroscopic transport patterns. The water volume fractions and Fick diffusivities, extracted from the aforementioned MD simulations, were also used to evaluate the thermodynamic factor accounting for the water activity gradient and the Maxwell–Stefan diffusivity, directly related to the chemical structure of the ultrathin polyamide active layer. Furthermore, three effective-medium-based approaches (Mackie–Meares, Bruggeman, and Maxwell models) were used to predict the water diffusion coefficients using, as for MS coefficients, input derived from MD simulations. Hence the results were compared for a rigorous assessment of the predictive capability of these models. Using the equilibrium membrane hydration, the analysis shows that Bruggeman and Maxwell models provide more accurate predictions compared to the Mackie–Meares model; indeed, the former provide diffusion coefficients in the range of experimental values available in the literature. On the contrary, the Mackie–Meares model yields less accurate values, even compared to diffusivities obtained using the previous molecular dynamics simulations. Ultimately, the results demonstrate that the effective medium-based approaches lose accuracy when transient hydration is used.