Characterizing Transport Properties of Surface Charged Nanofiltration Membranes via Model-Based Data Analytics.

This study presents an integrated framework for characterizing the transport properties of surface-charged nanofiltration (NF) membranes through dynamic diafiltration experiments and model-based data analytics. By incorporating startup dynamics, time corrections, and the influence of water flux on solute permeation, the framework effectively captures key transport properties governing membrane performance. A comparison of lag and overflow experimental modes highlights the superior precision of the lag startup mode, improving parameter estimates by 8%, 138%, and 83% as quantified by A-, D-, and E-optimality, respectively. Diafiltration experiments in the diluting regime further enhance model discrimination, enabling the identification of dominant transport mechanisms. This work marks a step toward developing self-driving laboratories (SDLs) that leverage model-based design of experiments (MBDoE) to expedite the characterization and development of high-performance NF membranes. The framework provides critical insights into transport phenomena, enabling the inverse design of membranes tailored to the demands of modern separation systems.