Thermodynamically consistent modeling of ion exchange membranes in multi-ionic environments

This paper presents a thermodynamically consistent model for ion exchange membranes in multi-ionic environments that integrates mass-action site occupation with mean-field electrostatic interactions to accurately reproduce both static and dynamic membrane properties, thereby providing a robust foundation for theory-driven membrane optimization.

The Gist

Imagine a membrane as a highly selective bouncer at an exclusive club. This club is made of a special polymer chain with "sticky spots" (fixed charges) that love to grab onto certain guests (ions) while trying to keep others out.

For a long time, scientists have tried to write the rulebook for how this bouncer works. Some rulebooks say the bouncer is a simple gatekeeper who just checks size (Steric Donnan Dielectric model). Others say the bouncer is a strict manager who only lets a few specific guests sit right next to the sticky spots, while the rest wander the hallway (Donnan-Manning model). Yet another rulebook suggests that at high concentrations, the sticky spots get so crowded that guests have to "hop" from one spot to another like a game of musical chairs (low-T* model).

The problem is that these rulebooks often break down when the club gets crowded, when the mix of guests changes, or when the conditions get tough.

The New "Universal Rulebook"
The authors of this paper, Noah Lettner, Felix K. Schwab, and Birger Horstmann, have created a new, thermodynamically consistent rulebook. Think of this as a master guide that combines the best parts of the old rulebooks into one coherent story.

Here is how their new model works, using simple analogies:

1. The Two States of a Guest

In their model, every ion (guest) in the membrane is in one of two states:

• The "Condensed" State (The VIP Lounge): These ions are stuck right next to the sticky spots on the polymer chain. They are "condensed" because the electrostatic attraction is so strong they can't easily leave.
• The "Uncondensed" State (The Main Hall): These ions are free to roam the water-filled pores of the membrane, moving around more freely.

The magic of this model is that it treats the transition between these two states like a chemical reaction. Just as you can predict how many people will sit at a table based on how hungry they are and how many tables are available, this model uses "mass-action" math to predict exactly how many ions will be in the VIP lounge versus the main hall based on the concentration of salt outside.

2. The "Crowded Room" Effect (Mean-Field Interactions)

The old models often assumed that the sticky spots acted alone. This new model realizes that the sticky spots are neighbors. If one sticky spot is holding a guest, it might make it harder (or easier) for the next sticky spot to hold a guest.

The authors added a "neighborly interaction" term to their math. Imagine a crowded dance floor: if everyone is dancing close together, your ability to move depends on what your neighbors are doing. This helps the model predict what happens when the membrane gets very crowded with ions, a situation where older models often fail.

3. Testing the Bouncer

The team tested their new rulebook against real-world data using a commercial membrane called CR61. They checked three things:

• Partitioning (Who gets in?): How many ions of different types end up inside the membrane compared to the water outside?
• The Donnan Potential (The Voltage): The electrical pressure difference created by the ions.
• Permeability (How fast they move): How quickly ions can travel through the membrane.

The Results:

• The new model matched the experimental data almost as well as the best existing models.
• Crucially, it did this without needing to constantly tweak its numbers (fitting parameters) every time the salt concentration changed.
• While the "low-T*" model (the hopping one) worked well for simple, low-concentration salt, it broke down when the salt got too concentrated because it assumed a limit on how many guests could sit at a table.
• The "SDE" model (the size-checker) worked okay but required the scientists to manually adjust its settings for every new scenario, making it less reliable for prediction.

4. Why This Matters for the Future

The paper highlights that this new model is particularly good at handling salt mixtures. Imagine a battery where you have both single-charged ions (like Sodium) and double-charged ions (like Magnesium) trying to get through the membrane at the same time.

Old models struggle here because they treat these ions as if they are in separate, simple worlds. The new model naturally handles the competition between them. It predicts that if you have a lot of Magnesium, it might push the Sodium out of the "VIP lounge," changing how the whole system behaves.

In Summary:
The authors didn't invent a new type of physics; they just built a better, more consistent instruction manual for how ions behave in membranes. By combining the idea of "sticky spots" (condensation) with the idea of "crowded neighbors" (mean-field interactions), they created a tool that works reliably whether the membrane is in a dilute solution or a highly concentrated one, and whether it's dealing with one type of salt or a complex mixture. This makes it a powerful tool for designing better membranes for things like water desalination and batteries, without needing to guess and check every time.

Technical Summary

Technical Summary: Thermodynamically consistent modeling of ion exchange membranes in multi-ionic environments

Problem Statement
Ion exchange membranes (IEMs) are critical components in technologies ranging from water desalination to aqueous batteries. However, existing modeling frameworks often struggle to simultaneously capture static partitioning and dynamic transport properties across a wide range of conditions, particularly in multi-ionic environments and at elevated concentrations. Current models, such as the Steric Donnan Dielectric (SDE) model, the Donnan-Manning (D-M) model, and the low-Reduced Temperature (low-T*) model, each emphasize different physical mechanisms but suffer from specific limitations. The SDE model relies on fitting parameters that lose physical meaning under varying conditions; the low-T* model is restricted to highly charged, low-dielectric membranes where ions are fully condensed; and the D-M model, while robust, relies on limiting laws and empirical extensions that may not hold for complex salt mixtures. There is a need for a framework that is thermodynamically consistent, structurally applicable to multi-ionic systems, and capable of linking microscopic interactions to macroscopic transport without arbitrary parameter adjustments.

Methodology
The authors propose a new framework, termed the Interaction-Occupation Model (IOM), derived from fundamental thermodynamics. The methodology involves:

1. Thermodynamic Formulation: The model distinguishes between two states for ions within the membrane: condensed (dominated by electrostatic attraction to fixed charges) and uncondensed (governed by random thermal motion). The free energy density of the membrane is constructed to include:

   • Entropic contributions from ideal mixing of uncondensed ions.
   • Entropic contributions from occupation statistics of condensed ions (mass-action site occupation).
   • Mean-field nearest-neighbor interaction terms to account for electrostatic interactions along the polymer backbone.
   • Electrostatic and excess chemical potential contributions.

2. Derivation of Equilibrium: By minimizing the free energy density under constraints of charge neutrality and site normalization, the authors derive a set of coupled equations. This results in an implicit condensation equation that determines the occupation fractions of fixed sites. Unlike the D-M model, which uses a piecewise approximation based on a critical Manning parameter, the IOM uses mass-action association constants ($K_\alpha$) and interaction energies ($U$) to describe the equilibrium between states.

3. Kinetics: Transport is modeled using the Nernst-Planck equation. The effective diffusion coefficient is derived as a weighted average of the mobilities of condensed and uncondensed ions, ensuring thermodynamic consistency (non-negative entropy production).

4. Parameterization: The model parameters (fixed site concentration, association constants, interaction energy, exclusion factors) are estimated using theoretical principles (e.g., Bjerrum theory for association, Born model for dielectric exclusion) and experimental data for a commercial CR61 membrane. The authors explicitly address the "excess exclusion factor" ($S_{ex}$), treating it as a concentration-independent estimate to avoid the uncertainties associated with defining pore geometry and permittivity variations.

Key Contributions

• Unified Framework: The IOM unifies the mass-action perspective of the low-T* model with the mean-field electrostatic approach of the D-M model. It treats condensed and uncondensed ions as co-existing in dynamic equilibrium, avoiding the strict "all-or-nothing" condensation assumption of the low-T* model.
• Thermodynamic Consistency: The derivation ensures that static partitioning and dynamic transport coefficients are intrinsically linked through the same thermodynamic potential, eliminating the need for independent fitting of transport and partitioning parameters.
• Multi-Ionic Applicability: The model is structurally designed to handle salt mixtures without requiring empirical extensions (such as ad-hoc rules for condensed ion fractions in the D-M model). It naturally accounts for competition between different counterions for fixed sites.
• Parameter Interpretability: The model parameters (e.g., association constants, interaction energy) are derived from or closely related to physical quantities, offering greater interpretability compared to the purely empirical parameters often required by the SDE model.

Results
The authors validated the IOM against experimental data for a CR61 membrane in single-salt (NaCl, CaCl$_2$, MgCl$_2$) environments and compared it with the SDE, D-M, and low-T* models.

• Static Properties: The IOM accurately reproduced experimental co- and counterion concentrations (partitioning) and Donnan potentials across a wide range of bulk salt concentrations. It performed comparably to the D-M model and outperformed the low-T* model at high concentrations, where the latter fails due to saturation of active sites.
• Dynamic Properties: The model predicted membrane permeability with high accuracy, matching experimental diffusion cell data for NaCl and MgCl$_2$. Notably, the IOM achieved this using a single set of parameters for both salts, whereas the SDE and low-T* models required refitting of parameters (e.g., hindrance factors or association constants) to match different salt datasets.
• Salt Mixtures: In a theoretical analysis of NaCl/MgCl$_2$ mixtures, the IOM predicted charge reversal and variations in effective membrane charge ($q_{eff}$) as a function of salt composition. In contrast, the standard D-M model predicted a constant $q_{eff}$ determined solely by the highest valence counterion. The IOM's predictions aligned with physical expectations of competitive binding and charge inversion.

Significance and Claims
The paper claims that the IOM provides a thermodynamically consistent basis for membrane modeling that bridges the gap between microscopic interaction mechanisms and macroscopic transport. Its primary significance lies in its internal consistency: it links static and dynamic properties through a unified framework, reducing the need for empirical corrections when moving from single-salt to multi-ionic environments or from dilute to concentrated regimes.

The authors argue that while the D-M model remains a strong benchmark, the IOM offers a more coherent approach for salt mixtures at elevated concentrations, a scenario common in applications like aqueous batteries where species crossover and degradation are critical. By recovering limiting laws (such as Manning condensation) in the low-salt limit while remaining structurally valid at high concentrations, the model is presented as a promising tool for theory-driven membrane optimization and the tailored design of IEMs for diverse technologies. The paper concludes that systematic experimental data for mixed electrolytes is still needed to fully validate these predictions, but the framework is ready to support such investigations.