Molecular Interactions and Layer Stacking Dictate Covalent Organic Framework Effective Pore Size
Scope
Reactive (ReaxFF) molecular dynamics of an imine-linked carboxylated COF (C-COF) membrane, compared with experiments, to relate effective pore size and solvation to organic-solvent nanofiltration performance.
Summary¶
Covalent organic frameworks (COFs) offer designer pores for molecular separations, yet crystallographic pore dimensions can mislead when solvation, framework flexibility, and layer stacking renormalize what solutes experience in the liquid phase. This ACS Applied Materials & Interfaces study couples ReaxFF reactive molecular dynamics with experiments on an imine-linked carboxylated two-dimensional COF (C-COF) membrane platform used for organic solvent nanofiltration (OSN). The authors relate effective pore size, layer registry, and interfacial interactions to measured permeance and selectivity, arguing that transport metrics must be interpreted with explicit solid–liquid structure rather than nominal pore metrics alone. Adri C. T. van Duin co-authors, placing the work in the group’s reactive FF portfolio for soft porous materials.
Methods¶
Experiments (C-COF membrane, OSN). The imine-linked carboxylated 2D C-COF is synthesized and processed into a membrane platform; organic solvent nanofiltration (OSN) tests report permeance and selectivity vs solute probes and feeds as in ACS Appl. Mater. Interfaces 13, 42164–42175 (DOI 10.1021/acsami.1c10866). Characterization (PXRD, microscopy, porometry-class data as figured) frames the real membrane morphology (mosaicity, defects, layer stacking).
MD application (ReaxFF; LAMMPS in practice). The ACS Appl. Mater. Interfaces Methods describe ReaxFF in periodic 3D C–COF+solvent supercells (in-plane xy and z-periodic pore channels from the article; exact stoichiometry in Table/SI). Equilibration: NPT at 300 K and 0 atm to relax lattice+pore void (0.25 fs time step, 200 ps); Berendsen thermostat (relaxation ~100 fs) and Berendsen barostat (relaxation ~1000 fs). Production for diffusion: NVT 200 ps at the same T (Berendsen thermostat, ~1000 fs coupling in the “weak” NVT block as written). E-field, shear, impact, metadynamics — N/A in the methods as summarized.
FF reparameterization. N/A (literature ReaxFF for the C–COF chemistry, per article).
Standalone DFT benchmark. N/A for the main MD line (DFT is cited only in the general ReaxFF-validation literature sense).
Findings¶
Outcomes and mechanisms. Mosaic C-COF membranes with contiguous 1D pore percolation can deliver strong OSN; poor in-plane registry or defects that chop channels hurt transport even when crystal chemistry is the same. Simulations (ReaxFF) and permeation data converge on the idea that solvated solute hindrance and carboxylate ionization (set by pH/base) tune exclusion and affinity, not the naked X-ray pore metric alone. N/A here to re-tabulate every D_solv or J value—see PDF.
Comparisons. The work is positioned as a case for experimental + atomistic agreement on effective pore throat; N/A third-party industrial dataset in the cited text.
Sensitivity. Solvent polarity/H-bonding and solute diameter (with shell) trend with rejection/J; N/A in this note to sweep E-field.
Limitations¶
Results are tied to the specific C-COF chemistry and solvents studied; generalization to other link chemistries or feed mixtures requires additional modeling and experiment.
Relevance to group¶
Demonstrates ReaxFF for porous organic frameworks and membrane separations with experimental co-validation.