Skip to content

Multiscale Modeling of Structure, Transport and Reactivity in Alkaline Fuel Cell Membranes: Combined Coarse-Grained, Atomistic and Reactive Molecular Dynamics Simulations

Summary

This open-access Polymers article develops a three-tier modeling chain for hydrated anion-exchange membranes (AEMs) based on functionalized poly(phenylene oxide): a high-resolution coarse-grained model for morphology, an atomistic polarizable (APPLE&P) model for local hydration, ion–water structure, and vehicular transport, and ReaxFF for reactive questions such as hydroxide transport mechanisms (including Grotthuss-like behavior as framed in the paper) and chemical degradation pathways. The study argues that no single model can span this whole property space, but sequential mapping between resolutions enables practical materials-by-design guidance for AEM chemistry, because morphology sets the water network that both polarizable MD and reactive MD must inherit. From a fuel-cell perspective, hydroxide mobility and chemical stability under alkaline stress are coupled: vehicular transport baselines from APPLE&P help bracket when explicit bond rearrangement must be turned on with ReaxFF.

Methods

Coarse-grained morphology: A coarse-grained representation equilibrates hydrated anion-exchange membranes built from functionalized poly(phenylene oxide) backbones; equilibrated CG structures are backmapped to atomistic coordinates for later stages (Polymers Methods).

Polarizable atomistic MD (APPLE&P): APPLE&P runs on the backmapped cells supply morphology, hydration structure, ion–water correlations, and vehicular OH⁻ transport baselines while bonds remain fixed (non-reactive mode).

ReaxFF reactive MD: ReaxFF trajectories—typically on configurations mapped from APPLE&P—treat bond-making and bond-breaking, including Grotthuss-like OH⁻ shuttling hypotheses and chemical degradation channels invisible to the polarizable nonreactive model. LAMMPS integration settings (timestep, QEq, cutoffs) appear in the open-access PDF tables.

Literature scope: The article is a primary modeling paper, not a methods-free review; background citations frame AEM physics for the chosen ionomer chemistry.

Atomistic protocol spine (CG → APPLE&P → ReaxFF): LAMMPS drives APPLE&P and ReaxFF on hydrated ionomer cells on the order of 10⁴–10⁵ atoms after backmapping, with three-dimensional periodic boundary conditions. NVT sampling uses the thermostat brands and timestep values tabulated per stage; temperature setpoints near ambient (300 K class) appear in the article unless a subsection states otherwise. Equilibration and multi-ns production lengths are reported separately for CG, APPLE&P, and ReaxFF segments. Barostat / pressure: N/A — the summarized workflow emphasizes constant-volume NVT segments; confirm the PDF for any optional NPT swelling studies. Electric field: N/A — no applied field in the quoted transport analysis. Replica / enhanced sampling: N/A — no umbrella sampling or metadynamics in the excerpted multiscale chain.

2 — Force-field training: N/A as a headline contribution — the article uses literature ReaxFF and APPLE&P parameterizations for the membrane chemistry; any QM training citations belong to the original force-field papers referenced in Polymers Methods, not a new global refit documented as the core result here.

Findings

  • Mechanism / outcomes: Multiscale coupling is presented as necessary to capture morphology, ion correlations, and bond-breaking chemistry in the same material class; ReaxFF resolves decomposition and Grotthuss-like OH⁻ events where APPLE&P stays nonreactive.
  • Comparisons: CG morphologies are compared to backmapped atomistic snapshots before reactive segments; literature AEM datasets contextualize the chosen functionalization level.
  • Sensitivity: Hydration level, temperature, and functionalization density shift channel connectivity that feeds both vehicular and reactive transport conclusions.
  • Limitations / outlook: Mapping uncertainty between resolutions remains an authored caveat; conductivity targets require cross-checking against experiment.
  • Corpus honesty: Workflow labels follow the open-access PDF at pdf_path; confirm QEq and cutoff settings when cloning LAMMPS inputs.

Limitations

  • Mapping fidelity between CG, polarizable atomistic, and ReaxFF representations introduces uncertainty; sensitive observables need cross-checks.
  • Parameter sets and water models strongly affect ionic conductivity and mechanistic attribution.
  • AEM morphologies evolve with hydration history; the sequential workflow assumes representative backmapped snapshots capture the ionomer heterogeneity relevant to both vehicular and reactive segments without large hysteresis errors.

Relevance to group

Adri C. T. van Duin coauthorship anchors the ReaxFF segment of a large collaborative membrane modeling effort.

Citations and evidence anchors

  • DOI: https://doi.org/10.3390/polym10111289 (papers/Dong_Polymers_2018.pdf).

Reader notes (navigation)