Automating the determination of pore size distribution in liquid separation membranes via solute retention experiments

Abstract

Membranes used for molecular separation in liquid media require precise determination of their pore size distribution (PSD). Analytical methods, such as scanning electron microscopy and atomic force microscopy, are limited to surface pore-size measurements and encounter problems in characterizing membranes with complex anisotropic morphologies. Meanwhile, theoretical calculations for PSD determination involve intricate analyses that depend on solute and solvent properties as well as various mathematical models. To address these challenges, we developed PoreInsight, an open-source Python package that automates PSD determination via solute retention experiments. While PSD determination via solute retention is primarily applicable to pure size-exclusion membranes, including those used in microfiltration and ultrafiltration, it also provides a valuable framework for membranes like those in nanofiltration and reverse osmosis, where size exclusion is not the sole transport mechanism. PoreInsight employs a systematic approach that integrates solute and solvent properties to estimate key PSD parameters — mean pore size and standard deviation — by fitting experimental retention data to sigmoid functions or assuming a log-normal probability density function. This automated method enhances the reliability and reproducibility of PSD characterization. The PoreInsight code, documentation, working demo, and examples are available online at www.OSNdatabase.com. We evaluated the accuracy of various mathematical models and analyzed the effects of various parameters, such as the number of retention points, experimental retention errors, variation in solute radii, and molar volume equivalent estimation, on PSD determination. A good approximation of PSD depends on careful focus on experimental conditions, including precise determination of solvated solute radii, accurate retention data measurements, and careful selection of appropriate fitting models.