Topology of Disordered 3D Graphene Networks
Summary¶
Disordered carbons are frequently conceptualized as three-dimensional networks of graphene fragments, but the prevailing local curvature motif—fullerene-like positive Gaussian curvature from pentagons, schwarzite-like negative curvature from heptagons/octagons, or ribbon-like nearly flat segments—has remained controversial. This Physical Review Letters article analyzes nanoporous and glassy carbon atomistic models that were previously shown to reproduce a broad suite of experimental observables (imaging, diffraction, porosity, mechanics, and thermal transport in the cited prior work). Rather than relying only on ring statistics, the authors apply a surface mesh construction to classify topology and global curvature contributions, then connect mesh-level features to local bonding environments including sp- and sp\(^3\)-rich defects.
Methods¶
4 — PRL post-processing of imported atomistic models (not a new MD “production” study in this Letter). The Physical Review Letters reuses disordered-carbon atomistic networks (~32,000 atoms, 0.5–1.5 g cm⁻³) made earlier with self-assembly / annealed molecular dynamics and already benchmarked against HRTEM, diffraction, porosimetry, mechanics, and transport in cited prior work. This Letter then applies a triangular surface mesh to each structure (after Franzblau-style ring detection to nonagons in the main text) and maps angular defects to signed Gaussian curvature patches on the mesh, separately treating sp- and sp³-tagged perimeter/tetrahedral sites off the sp² surface graph. ReaxFF-specific force-field development in this file: N/A; the interatomic model lives in the earlier MD publications cited as the source of the networks. A fresh blueprint-style LAMMPS table (timestep, NVT, thermostat) for this PRL is N/A because no new time-marching trajectory is reported here—only geometric post-processing on static frames imported from those studies. Shear / shock / E-field in a pristine MD sense: N/A for this Letter’s evidence chain. DFT in this PRL as a primary result: N/A—the mesh analysis is classical geometry on classical carbon models**.
Findings¶
1 — Outcomes & mechanisms (mesh-level topology). Negative Gaussian curvature (saddle-like patches) dominates the global curvature budget in the analyzed disordered 3D graphene networks, whereas fullerene-like positive curvature and near-flat ribbon segments are present but subdominant in the mesh histograms reported in the PRL (see main-text figures). sp-rich and sp³-rich defect atoms track to line-like and screw-like junctions that terminate free edges and stitch low-curvature ribbons into 3D porous morphology. 2 — Comparisons. The work contrasts with classical ring-counting / Euler-style arguments alone by showing that all three curvature classes coexist in the same self-consistent disordered carbon morphology at laboratory-relevant densities (per the PRL and the cited model-vs-experiment agreement in earlier work). 3 — Sensitivity & levers. Curvature partitions at fixed mass density are expected to depend on the generative (anneal) protocol (temperature path, quench rate, and similar kinetic levers in the source MD); this Letter flags that pathway as a sensitivity axis for comparing future models. 4 — Authored limitations / outlook are tied to which 3D graphene-like networks are admitted into the test set; if a synthesis pathway favors one topological element over another, the curvature mix shifts (see PRL ** discussion and our ## Limitations). 5 — Corpus / KB honesty — the PRL reuses imported configurations; do not treat this wiki as a source of independent synthetic carbon dynamics without consulting the generator papers cited in the PRL bibliography.
Limitations¶
Conclusions are conditioned on the specific generative MD protocol used to build the model carbons; different formation pathways could yield different curvature partitions even if bulk densities match.
Reproducibility notes¶
The Letter explicitly ties conclusions to large disordered networks (~32,000 atoms) spanning 0.5–1.5 g cm\(^{-3}\); smaller soot-like clusters may not exhibit the same curvature statistics. When coupling these diagnostics to ReaxFF pyrolysis trajectories, treat mesh extraction as a post-processing step that depends on robust surface identification from noisy atomistic configurations—implementation details (smoothing, mesh density) can shift measured curvature histograms. If reproducing figures, track the mesh resolution and any smoothing radius applied before computing Gaussian curvature histograms, because discretization can artificially broaden peaks.
Relevance to group¶
Provides structural diagnostics for disordered carbon morphologies that may complement reactive MD studies of carbonization or pyrolysis.