A carbon nanotube kinks

Compress a carbon nanotube from the ends and it does not stay a cylinder. It kinks. That silhouette is one of the most recognizable pictures in nanoscale mechanics. We ran it on QuickMDSim.

The science this is built on

Continuum shell theory said a thin cylinder under axial load should flatten and then kink, at strains of a few percent. Yakobson, Brabec, and Bernholc (PRL 1996) showed that a Tersoff carbon nanotube does the same thing atom by atom — not a continuum cartoon, the actual lattice. Iijima, Brabec, Maiti, and Bernholc watched the same flexibility in TEM and in MD the same year. That is the calculation this demo is meant to make easy.

The potential is Tersoff carbon (Tersoff, PRB 1989), the same C–C parameters as the graphene-fracture demo. MANYBODY is in the 22 Jul 2025 image. Upload SiC.tersoff as a project file; it is not baked into the container.

How it ran on QuickMDSim

A (10,10) armchair tube — 1280 atoms, 78.7 Å long, radius 6.78 Å — submitted from app.quickmdsim.com. Job 2579cb45-2e19-4fd9-9a98-41c339660923 is genuine mpirun lmp in a Cloudflare container:

• Bottom rings frozen. Top rings driven at −0.40 Å/ps for 24 ps
• pair_style tersoff + SiC.tersoff C
• Multi-file project: input.lammps + data.lammps + potential
• 37 seconds wall time on one vCPU

What the trajectory actually shows

Job-done is not the result. From dump.buckle.lammpstrj:

• Radial envelope rmax = max √(x²+y²) sits at 7.1 Å through 3.5% strain — still a cylinder, plus the 0.22 Å seed used to pick a buckle plane.
• Between 3.7% and 4.2% strain, rmax jumps 7.4 → 10.6 Å. Thermo pressure collapses in the same window. That is the instability, not a slow squash.
• After the jump the mid-span stays at 10–11 Å. The caps stay at ~7.3 Å. The extra radius is in the middle of the tube, not the open ends flaring.

That is the Yakobson picture at educational scale: a Tersoff lattice that refuses to stay a cylinder past a few percent of end shortening.

What this workflow needs from the product

1. Potentials are project files

Tersoff needs SiC.tersoff next to the input. QuickMDSim stages every project file into the workdir. Same path as EAM aluminum.

2. The 3D player reads a normal dump

The input writes dump.qmd3d.lammpstrj (id type x y z). The player on this page is that file. Open the same dump in Outputs after you run the starter. No special LAMMPS command — a dump custom is enough.

3. Side-view stills still matter

A thin tube is a silhouette problem. The blog movie is a fixed-camera side view of the same dump, bonds drawn, amber on atoms that left the cylinder. The 3D player and the movie are two looks at one trajectory.

Demo files

The project lives in the repo at demos/cnt-buckle/:

• input.lammps — minimize, freeze ends, drive, dumps
• data.lammps — (10,10) coordinates
• SiC.tersoff — Tersoff C
• README.md — geometry and how to check the kink

Regenerate the tube with python3 scripts/gen_cnt_buckle.py. There is also a one-click starter in the app — CNT buckle.

How to try it

1. Open app.quickmdsim.com
2. Pick the CNT buckle starter
3. Press Run (~45 s)
4. Open dump.qmd3d.lammpstrj in the 3D player, or download dump.buckle.lammpstrj and color by √(x²+y²)

Credit

Educational tube, sized for a free-tier run. Cite Yakobson et al. (1996) for the instability, Iijima et al. (1996) for the flexibility, and Tersoff (1989) for the potential if you take this into research. This is not a reproduction of any one paper’s production trajectory.