Unlocking the potential of polymeric desalination membranes by understanding molecular-level interactions and transport mechanisms

Abstract

Polyamide reverse osmosis (PA-RO) membranes achieve remarkably high water permeability and salt rejection, making them a key technology for addressing water shortages through processes including seawater desalination and wastewater reuse. However, current state-of-the-art membranes suffer from challenges related to inadequate selectivity, fouling, and a poor ability of existing models to predict performance. In this Perspective, we assert that a molecular understanding of the mechanisms that govern selectivity and transport of PA-RO and other polymer membranes is crucial to both guide future membrane development efforts and improve the predictive capability of transport models. We summarize the current understanding of ion, water, and polymer interactions in PA-RO membranes, drawing insights from nanofiltration and ion exchange membranes. Building on this knowledge, we explore how these interactions impact the transport properties of membranes, highlighting assumptions of transport models that warrant further investigation to improve predictive capabilities and elucidate underlying transport mechanisms. We then underscore recent advances in in situ characterization techniques that allow for direct measurements of previously difficult-to-obtain information on hydrated polymer membrane properties, hydrated ion properties, and ion-water-membrane interactions as well as powerful computational and electrochemical methods that facilitate systematic studies of transport phenomena.

Fig. 1. (a) Summary of the properties that give rise to interactions and govern membrane transport and performance: membrane properties such as morphology and chemistry; ion properties such as size, shape, and charge; and solution properties like composition, temperature, pH, and pressure. (b) Phenomena at many scales govern transport. Nanoscale interactions occur between ions and polymer groups while mesoscale gradients develop across the membrane thickness. Macroscale conditions like module geometry and flow parameters also affect interactions/gradients. Specifically, nanoscale interactions provide insight into the molecular level mechanisms behind (c) membrane function including surface ion partitioning and ion diffusion through the membrane interior. These transport phenomena dictate membrane permeability and selectivity performance.

Fig. 2. (a) Comparison of membrane properties for reverse osmosis (RO), nanofiltration (NF), and ion exchange (IX) membranes. Units for fixed charge density are moles of charge per liter of free water in the membrane. (b) Chemistry and morphology of RO, NF, and IX membranes. RO and NF chemistry is based on interfacial polymerization of trimesoyl chloride (TMC) and m-phenylenediamine (MPD) for RO or piperazine (PIP) for NF. Here only the fully crosslinked chemistry is shown; in actual membranes, unreacted portions lead to carboxylic acids, carboxylate, and amino groups in place of some amide groups. IX chemistry of a traditional Nafion membrane is shown. The morphology of RO, NF and IX membranes vary on the nm length scale. Adapted from ref. and .

Fig. 3. Kosmotrope to chaotrope classifications for anions (top) and cations (bottom). These are ordered in accordance with the Hofmeister series.

Fig. 4. Interactions that an ion experiences as it moves from the feed solution into the (negatively charged) membrane. In the bulk solution, (a) dissociation of cation–anion pairs pose an energetic barrier to partitioning into the membrane, while (b) an applied hydraulic pressure is a driving force for water and ion transport. At the surface, (c) anion repulsion from the negatively charged membrane and (e) partial ion dehydration present energy barriers to membrane entry, while (d) attractive cation–membrane interactions partially offset these repulsive forces. In the membrane interior, (f) reorganization of an ion's hydration shell as it traverses the heterogeneous void network and (g) interactions with charged membrane groups present resistances to diffusive transport. Finally, the (h) concentration and (i) potential gradients created between the feed and permeate side of the membrane create driving forces for ion transport while a (j) dielectric gradient from the bulk solution through the membrane poses an additional transport energy barrier.

Fig. 5. An illustration of the assumptions used to calculate the dielectric constant of water in the membrane. (a) Straight, cylindrical pores with one layer of oriented water molecules on pore walls are assumed and the remainder of water is assumed to behave like the bulk solution. The effective dielectric constant of the water in the pores is the weighted average of the dielectric constants of the water at the pore wall and interior. (b) A more realistic picture of water in the dynamically fluctuating voids where water molecule orientation spatially and temporally fluctuates.

Fig. 6. Average surface interactions at (a) a charged, hydrophilic surface where electrostatic forces dominate and attract small, strongly hydrated ions and (b) a neutral, hydrophobic surface where hydrophobic forces, a combination of entropic forces and van der Waals interactions, dominate to stabilize large, weakly hydrated ions at the surface.

Fig. 7. Summary of local nanoscale interactions between ions and membrane groups. (a) Hydrogen bonding occurs between hydrogens and electronegative atoms, illustrated with a urea compound. (b) Electrostatic interactions occur between ions and carboxylate, carboxylic acid, amide, and amino functional groups on the membrane. (c) Cation bridging occurs when a cation forms crosslinking bonds between carboxyl groups of organic matter and the membrane. Finally, (d) pi bonding occurs between aromatic rings on the membrane and cations, anions, and aromatic rings of organic matter.

Fig. 8. (a) Illustration of the treatment of ion transport by the Donnan steric pore model with dielectric exclusion (top) and solution–diffusion model (bottom) compared to (b) a more realistic model of ion transport with more accurately determined parameters for ion size, void size distribution, and dielectric constant that accounts for interactions and resistances that an ion experiences at the surface and interior regions of the membrane.