PES membranes are workhorses of modern water and wastewater treatment, deployed worldwide to strip out organic, inorganic, and biological pollutants from drinking water and industrial effluents. Yet their performance is fundamentally constrained by an inconvenient chemical truth: the polymer backbone is inherently hydrophobic. This hydrophobicity restricts water transport through the membrane, promotes rapid fouling by proteins and natural organic matter, and undermines the removal of hydrophilic organic contaminants such as pharmaceutical residues. The research team, based at the Institute for Nanotechnology and Water Sustainability at the University of South Africa in Johannesburg, set out to determine whether simple, inexpensive, and environmentally benign additives could overcome these limitations by reshaping the membrane’s structure during fabrication rather than by grafting complex functional groups onto its surface after the fact.

The researchers fabricated six flat-sheet PES membranes using the phase inversion method, dissolving PES in N-methyl-2-pyrrolidone and casting the solutions onto glass plates before immersing them in a water bath to trigger phase separation. The additives were incorporated at controlled weight percentages: PEG-400 at 20 wt%, glycerol at 2 wt%, and starch at 1 wt% dispersed in deionized water. The casting solutions were stirred at 300 rpm and 40 °C for six hours, degassed overnight, and spread to a uniform thickness of 200 micrometers. After immersion in the coagulation bath, the membranes were thoroughly washed to remove residual solvent and loosely bound additives, then stored in water until characterization. This fabrication approach allowed the team to isolate the individual and combined effects of each additive on the membrane’s final morphology, surface chemistry, and transport properties.

Characterization revealed that each additive left a unique structural fingerprint on the membrane. Scanning electron microscopy showed that all membranes retained the asymmetric structure typical of phase-inverted PES — a dense thin skin layer supported by a porous finger-like sub-layer — but the additive dramatically altered the dimensions and character of these features. Glycerol, acting as both a pore former and a plasticizer, produced significantly thinner membranes with elongated, relatively uniform finger-like macrovoids, reflecting accelerated demixing during phase inversion. PEG-400, a low-molecular-weight and highly water-soluble pore former, leached rapidly into the coagulation bath, leaving behind the most porous structure of all membranes at 94.95% bulk porosity, with the thinnest and most discontinuous skin layer and the highest surface roughness at 209.6 nanometers. Starch, by contrast, increased the viscosity of the casting solution, slowing solvent–non-solvent exchange and delaying demixing. This produced a tighter, more uniform skin layer with shorter, sponge-like meso-pores and the smoothest surface of any membrane at 29.9 nanometers roughness.

Surface chemistry measurements confirmed that the additives successfully enriched the membrane interface with oxygen-containing functional groups. Energy-dispersive X-ray spectroscopy showed that pristine PES contained 83.95 wt% carbon and 11.66 wt% sulfur, consistent with the polymer backbone, while modified membranes exhibited dramatically lower carbon (39.75–44.36 wt%) and sulfur (7.98–9.21 wt%) content, confirming surface migration of the hydrophilic additives during phase inversion. Fourier-transform infrared spectroscopy detected broad O-H stretching bands between 3200 and 3600 cm⁻¹ in all membranes, along with C-O and C-O-C stretching peaks near 1147–1151 cm⁻¹ and 1104–1105 cm⁻¹, which enhance surface water affinity. Water contact angle measurements quantified the hydrophilicity gains: pristine PES showed the highest angle at 68.68°, glycerol reduced it to 60.31°, PEG to 56°, and starch achieved the lowest at 43.65°, attributed to its abundant multivalent hydroxyl groups and controlled demixing kinetics. Notably, the ternary PEG/glycerol/starch membrane exhibited a contact angle of 52.90°, higher than both the starch-only and PEG/glycerol membranes, providing quantitative evidence that the three additives competed for surface sites rather than cooperatively enriching the interface.

Performance testing revealed that structural changes translated directly into transport behavior, but not always in predictable ways. The PEG/glycerol membrane achieved the highest pure water permeability at 1238 liters per square meter per hour at 5 bar, owing to its largest mean pore radius of 7.5 nanometers and the most open macrovoid structure. According to the Hagen-Poiseuille relationship, flux through pores scales with the fourth power of pore radius, so the 3.5-fold larger skin pores of the PEG/glycerol membrane compared to PEG alone resulted in approximately 150-fold higher flow per pore, more than compensating for its lower bulk porosity of 67.14%. PEG alone, despite achieving the highest bulk porosity at 94.95%, exhibited only moderate flux of 838.1 LMH because its mean pore radius was the smallest at 2.2 nanometers — individual PEG molecules templated the skin-layer pores during leaching. This demonstrated that skin-layer pore size, not bulk porosity, is the rate-determining factor for water permeability. The ternary membrane achieved the second-highest flux at 1133 LMH through a combination of 74.12% porosity and 6.4 nanometer pores, though antagonistic interactions with starch reduced its surface hydrophilicity.

Contaminant rejection experiments using ciprofloxacin (CIP), rhodamine B (RhB), and natural organic matter (NOM) revealed that rejection mechanisms depended on the interplay between solute physicochemical properties and membrane structure. CIP, a zwitterionic antibiotic with a molecular weight of 331 g/mol and molecular dimensions of approximately 1.0–1.3 nanometers, is smaller than the pore radii of all membranes tested, ruling out size exclusion as a dominant rejection mechanism. Retention instead relied on hydrogen bonding between CIP’s polar functional groups and oxygen-containing sites on the membrane surface. The starch-modified membrane achieved the highest CIP rejection at 21.5%, attributed to its high density of O-H groups providing multiple hydrogen-bonding sites. RhB, a larger cationic dye at 479 g/mol with dimensions of 1.5–1.8 nanometers, could be rejected through a combination of steric resistance and electrostatic attraction to the negatively charged membrane surfaces, which all membranes exhibited at the experimental pH of 7 since their points of zero charge ranged from 6.59 to 6.94. The starch membrane again achieved the highest RhB rejection at 22.9%, while the PEG membrane reached 21.2% through its small 2.2-nanometer pores providing greater steric resistance to the larger RhB molecule.

Real water experiments using lake water samples from South Africa revealed that NOM fouling affected contaminant rejection in fundamentally different ways depending on the solute’s charge and size. RhB rejection increased dramatically after fouling for all membranes, reaching 45–69%, far exceeding both deionized water and unfouled real water values. The PEG/glycerol/starch membrane showed the largest net gain, increasing from 10.9% to 69.4%, while the PEG membrane achieved the highest absolute rejection after fouling at 69.0%. This enhancement was explained by RhB’s cationic charge creating strong electrostatic attraction to the anionic NOM fouling layer, combined with physical entrapment within the accumulated organic matrix. CIP rejection, in contrast, generally declined after fouling. The starch membrane exhibited the largest CIP losses, decreasing from 21.5% to 5.5% in real water due to competition for hydrogen-bonding sites by NOM, and to 3.0% after fouling as accumulated NOM blocked the hydroxyl groups responsible for CIP binding. This revealed the fundamental vulnerability of adsorption-based rejection mechanisms: although they achieve high clean-water rejection, they are unsustainable in complex real-water matrices where competing NOM largely occupies active sites. The PEG membrane uniquely showed improved CIP rejection in real water, increasing from 13.5% to 24.5%, attributed to NOM components constricting its already small 2.2-nanometer pores and creating a dynamic filtration layer that enhanced size exclusion.

Fouling resistance testing using bovine serum albumin as a model protein foulant, combined with time-course flux decline during real water filtration, demonstrated that fouling behavior depended directly on the dominant rejection mechanism. Glycerol provided the best overall antifouling performance, with the slowest real water flux decline at 41.7% over 180 minutes, the lowest total BSA fouling at 63.5%, the highest flux recovery ratio at 62.2%, and low irreversible fouling at 37.8%. These results were attributed to glycerol’s hydration layer repelling both NOM and protein foulants from the membrane surface. The PEG membrane exhibited paradoxical behavior: the lowest operational flux decline in real water at 20.5% over 180 minutes, but the worst irreversible fouling in BSA studies at 70.5% and the lowest flux recovery ratio at 29.5%. Its very small 2.2-nanometer pores caused permanent pore blockage that minimally reduced flux but could not be reversed, demonstrating that size-exclusion rejection inherently leads to irreversible fouling. The starch membrane showed substantial real water flux decline at 35.3% and high total BSA fouling at 89.2%, but balanced reversible and irreversible fouling components, consistent with its hydrogen-bonding-based rejection mechanism producing interactions that were sufficiently reversible for physical cleaning. The ternary membrane exhibited the highest reversible BSA fouling at 51.2% and the second-highest flux recovery ratio at 59.3%, suggesting that its irregular macrovoid structure created a surface where foulants attached more loosely rather than permanently.

Antimicrobial evaluation against Staphylococcus aureus revealed that only the glycerol-containing membranes — glycerol alone and PEG/glycerol — produced small zones of inhibition at 6.18 and 6.92 millimeters respectively, while no antimicrobial activity was noted for the other membranes. The absence of activity in the ternary membrane suggested that starch may hinder glycerol surface enrichment, reducing the effective surface concentration available for antibacterial action. Bacterial adhesion quantification over 24 hours confirmed that all modified membranes reduced surface colonization compared to pristine PES, which accumulated 1.12 × 10⁷ CFU/cm². The glycerol and PEG/glycerol membranes showed the lowest viable counts at 1.6 × 10⁶ and 2.35 × 10⁶ CFU/cm² respectively, suggesting that glycerol’s hydroxyl groups promoted a hydration layer that reduced direct bacteria–membrane interactions. However, the relatively high bacterial accumulation on the PEG membrane at 5.75 × 10⁶ CFU/cm², despite its hydrophilic modification, demonstrated that hydrophilicity alone was insufficient to prevent long-term colonization, and that surface roughness and topography also governed bacterial retention by providing additional contact area and protected attachment sites.

The study’s most consequential finding is that no single membrane simultaneously optimized flux, rejection, and fouling resistance, and that the ternary combination of all three additives did not provide an overall performance advantage under the conditions investigated. The starch-modified membrane achieved the highest overall NOM rejection at 67% average while having relatively low flux at 257 LMH, attributed to its high density of surface hydroxyl groups forming hydrogen bonds with polar organic contaminants, a thin highly hydrophilic skin minimizing resistance to water flow, and a sponge-like sub-layer increasing tortuosity and providing additional adsorption sites. The PEG/glycerol membrane confirmed the well-established permeability–selectivity trade-off, with high flux of 648 LMH at 3 bar accompanied by only moderate NOM rejection averaging 53%. The PEG membrane achieved high flux of 467 LMH and high rejection of 60% average through its unique structure: an extremely thin skin with very small pores from PEG leaching, combined with massive macrovoids in the sub-layer that reduced hydraulic resistance. These results demonstrate that hydrophilic additive modification of PES membranes is an effective strategy for tuning membrane structure and performance, but that bulk porosity and mean pore radius are independently tunable through strategic additive selection, and that combining multiple additives does not guarantee superior structural or performance properties. The choice of additive, and the decision to combine additives, must be strategically aligned with the target application — whether the priority is contaminant removal, water permeability, or fouling resistance — rather than assumed to be universally beneficial.

Subject of Research: Non-polymeric solvents as membrane additives for tailoring membrane structure and performance

Article Title: Non-polymeric solvents as membrane additives for tailoring membrane structure and performance

Article References: Phillip, N., Muleja, A. A., Motsa, M. M., Mamba, B. B., Nkambule, T. T. I., & Tshangana, S. C. (2026). Non-polymeric solvents as membrane additives for tailoring membrane structure and performance. Frontiers in Environmental Chemistry, 7, Article 1949697. https://doi.org/10.3389/fenvc.2026.1949697

Image Credits: AI Generated

DOI: 10.3389/fenvc.2026.1949697

Keywords: Non-polymeric, solvents, membrane, additives, tailoring, structure, performance, scientific research