ReaxFF Reactive Force Field Study of Polymerization of a Polymer Matrix in a Carbon Nanotube-Composite System
ReaxFF carbon parameters are retrained against PCFF-IFF / DFT-D2 benchmarks for flattened CNT geometries, then used with accelerated (bond-boost) ReaxFF MD to follow epoxy-amine curing near circular vs flattened nanotubes.
Summary¶
The work addresses epoxy (diglycidyl ether of bisphenol F, Bis-F) + curing agent (diethyltoluenediamine, DEDTA) polymerization in the presence of armchair CNTs spanning multiple diameters. ReaxFF C–C graphitic parameters are adjusted so flattened CNT (flCNT) energetics and geometries better track reference data; bond-boost MD promotes cross-linking events within tractable times. Binding energies and polymer alignment differ between curved vs flat flCNT regions, with implications for interfacial load transfer.
Methods¶
- Reference data: DFT-D2 and PCFF-IFF datasets for open vs flattened CNT energies across diameters; train-set augmentation where collapsed flCNT minima appear during optimization.
- ReaxFF training: Monte-Carlo/ReaxFF training workflow (as described) updating carbon parameters relevant to π-bonding and bending; comparison of threshold diameters where flattening becomes favorable.
- Production MD: LAMMPS-style ReaxFF runs reported with timestep 0.25 fs and overall segments up to 500 ps (order 2×10\(^6\) steps) including heating to 500 K (see article for staged protocol); bond-boost constraints to accelerate targeted reaction channels with energy-based rejection of unphysical barrier crossings. Thermostat: Nosé–Hoover in the Ar-bombardment and pressure-serviced segments described in the article; Berendsen/Nosé details for curing stages follow the primary text. Barostat: anisotropic NPT-like control with ~250 fs damping (~1000 timesteps) for graphene + Ar studies; N/A — barostat not used in pure NVT epoxy curing windows if so stated.
- Systems: Isolated (n,n) armchair tubes including (15,15) among others for binding and polymerization comparisons. PBC in 3D for the isolated nanotube + resin supercells. NPT / barostat: N/A in the protocol summary above; constant-volume ReaxFF stages are reported for curing (NVT-class dynamics per article). Electric field: N/A beyond intramolecular bias; replica / umbrella: N/A.
Findings¶
- Optimized parameters reduce over-flattening of large-diameter tubes versus ReaxFF C-2013, improving agreement with PCFF-IFF/DFT-D2 trends for flCNT width and relative energies.
- DEDTA binds more strongly to flat flCNT regions than to curved segments (reported binding energies \(\sim\)15 vs \(\sim\)24 kcal/mol with consistent PCFF-IFF agreement for the gap).
- Dimer/crosslink production is enhanced near flCNTs, suggesting better coating and load transfer than for circular CNTs under the simulated conditions.
- π–π stacking between aromatic moieties and CNT walls yields high alignment of polymer fragments along the tube axis.
The J. Phys. Chem. C abstract emphasizes that flat flCNT regions are more favorable adsorption sites than curved sidewalls (higher binding energies there), that higher dimer generation near flCNTs yields more effective nanotube coating and higher load transfer than circular CNTs under their protocol, and that the simulations are presented as an atomistic route to observe epoxy–amine polymerization together with CNT interactions.
Limitations¶
Bond-boost acceleration changes effective kinetics; isolated-tube geometry omits entangled CNT–CNT networks. Parameter transfer beyond the Bis-F/DEDTA chemistry should be validated.
Relevance to group¶
van Duin-group ReaxFF on CNT–epoxy interfaces with explicit parameter refinement and accelerated reactive MD.