Skip to content

Multi-step mechanism of carbonization in templated polyacrylonitrile derived fibers: ReaxFF model uncovers origins of graphite alignment

Reactive molecular dynamics with ReaxFF is used to interpret how in situ templating by double-walled carbon nanotubes versus graphitic nanoplatelets steers polyacrylonitrile (PAN) carbonization and graphitic alignment in nanofibers, with discussion tied to experimental electron-diffraction–style observables in the publication.

Summary

Carbon fiber performance depends on graphitic order formed during multi-step carbonization of PAN-based precursors. Saha et al. simulate templated carbonization in models inspired by electrospun fibers that incorporate low loadings of double-walled CNTs or graphitic nanoplatelets. Depending on carbonization temperature, the dominant template–medium interaction shifts between physisorption (physical templating) and chemisorption (chemical templating). Strong template coupling yields aligned graphitic structures; in their simulations, nanotube templates produce more robust alignment than graphite nanoparticle templates. The study emphasizes quantitative comparison between atomistic structure (e.g., ring-density–related measures) and diffraction-related data for the experimental fiber systems.

Methods

1 — MD application (ReaxFF carbonization in LAMMPS). Saha et al. simulate templated PAN-derived carbonization using ReaxFF as implemented in LAMMPS (Carbon 94 (2015) Methods). Starting chemistry is a stabilized PAN precursor (SPP) represented as C\(_{14}\)H\(_8\)N\(_4\), with 500 SPP molecules arranged around either a hydrogen-terminated (5,5)@(10,10) double-walled CNT (26 Å length, ~6.2 wt.% filler) or an AB-stacked double-layer graphene (DLG) template (26 Å × 24 Å, ~4.6 wt.%), giving 13 690 atoms (SPP/DWCNT) or 13 540 atoms (SPP/DLG). Periodic boundary conditions are applied in all three directions; the simulation cell is sized for initial composite densities of ~1.41 g cm⁻³ (SPP/DWCNT) and ~1.40 g cm⁻³ (SPP/DLG). Equilibration runs ~160 ps at 300 K; three independent trajectories are branched from the last 40 ps of equilibration for statistics. Carbonization uses NVT MD with heating rate 10 K ps⁻¹ in three stages: heat to 2200 K then anneal 2 ns; heat to 2500 K then anneal 500 ps; heat to 2800 K then anneal 750 ps. Nosé–Hoover thermostat with \(T_\text{damp}\) = 25 fs; time step 0.25 fs. Volatile gases (N\(_2\), H\(_2\), HCN, NH\(_3\)) are removed every 100 ps during annealing to mimic experimental byproduct purge. Total NVT annealing time available for analysis is 3.25 ns (three averaged runs). N/A — barostat / external stress control: protocol is constant-volume NVT with no stated NPT stage during the reported annealing ladder. N/A — electric field: not part of the described setup.

2 — Force-field training. N/A — new parameterization in this article: the authors use published ReaxFF parameters for C/H/N/O systems validated against DFT/DFTB in prior work, as summarized in their Methods section.

3 — Static QM / DFT. N/A — production QM: DFT appears as prior validation literature for the ReaxFF choice, not as the dynamics engine.

4 — Review / non-simulation framing. N/A: primary templating / carbonization study.

Findings

Outcomes and mechanisms. Depending on temperature and template chemistry, the authors distinguish physisorption-driven physical templating versus chemisorption-driven chemical templating; strong template–medium coupling promotes aligned graphitic motifs in the fiber medium. DWCNT templates produce more robust alignment than DLG (graphitic nanoplatelet) templates under the simulated carbonization ladder, consistent with the narrower high-ordering region at the nanotube walls versus the broader, more fluctuating graphene stacks in their ring-population maps.

Comparisons. The manuscript links atomistic ring statistics, radial distribution functions, hybridization, and interface density profiles to electron diffraction–style observables for the experimental electrospun fibers discussed in the article.

Sensitivity and design levers. Template topology (DWCNT vs DLG), distance from the filler surface (they emphasize strong templating within ~4 Å on the simulated timescales), and the multi-stage temperature program control whether physical stacking or chemical coupling dominates and how six-membered carbon rings accumulate near the filler.

Limitations and outlook (as authored). The authors note that real carbonization occurs over hours with lower temperatures than the accelerated MD program; volatile removal and temperature ramps are emulations, not identical reactor protocols.

Corpus / PDF honesty. The stable paper_id 2017kowalik-venue-bez-tytu is a legacy manifest slug; the ingested PDF is Saha et al., Carbon 2015 (DOI above). extraction_quality: partial in front matter reflects short local extracts; tables, SI parts, and quantitative plots should be read from the PDF/SI when extending beyond this summary.

Limitations

ReaxFF remains an empirical reactive model: transferability across full industrial carbonization schedules and exact quantitative agreement with experiment are limited by force-field and sampling constraints, as in any large-scale pyrolysis simulation. Repository automation maps this stable paper_id to normalized/papers/2017kowalik-venue-bez-tytu.json and the repo-relative pdf_path. Where extraction_quality is partial, the tracked PDF and DOI remain the quantitative authority over short local extracts.

Relevance to group

Demonstrates ReaxFF application to carbonization / polymer pyrolysis and nanocarbon templating, adjacent to broader reactive MD on carbon materials and interfaces in the corpus.

Citations and evidence anchors

Reader notes (navigation)

The stable paper_id paper:2017kowalik-venue-bez-tytu reflects a legacy manifest slug; the ingested PDF is Saha et al., Carbon 2015, not a 2017 “bez tytułu” placeholder. Maintainer corpus records document legacy slug-to-PDF pairings; the bibliography and DOI above identify Saha et al., Carbon (2015).