Carbon Nanotubes

A graphene sheet, rolled into a straw

Take a single sheet of graphene — a flat honeycomb of carbon atoms, each one bonded to three neighbors by strong sp² bonds, exactly one atom thick. Now roll it into a seamless cylinder. That cylinder is a carbon nanotube. The diameter is typically 0.4 to 3 nm for a single-walled tube, and the length can run from tens of nanometers to, in record cases, half a meter — an aspect ratio of 10⁸, the most extreme of any known material.

Every carbon sits in the same chemical environment it had in flat graphene: three σ-bonds to neighbors and one electron in a p-orbital perpendicular to the surface. Those perpendicular p-orbitals overlap into a delocalized π-system that runs the length of the tube. So chemically nothing dramatic happened when we rolled the sheet — the bonds are the same. What changed is the boundary condition: an electron traveling around the circumference must come back to itself in phase. That single quantization rule is where all the electronic magic comes from.

The crucial degree of freedom is which direction you roll. You can roll along the line of hexagons, across them, or at any angle in between. The roll-up direction is captured by a vector drawn on the flat sheet from one atom to the atom it gets glued to.

The chiral vector (n,m)

Graphene's lattice has two primitive vectors, a₁ and a₂, each 0.246 nm long and 60° apart. The roll-up direction — the chiral vector — is any integer combination of them:

C_h = n·a₁ + m·a₂          (n, m integers, n ≥ |m|)

The pair of integers (n, m) completely specifies the tube. The circumference is just the length of that vector, and the diameter follows immediately:

|C_h| = a·√(n² + n·m + m²),   a = 0.246 nm
d     = |C_h| / π = (a/π)·√(n² + n·m + m²)

Three named families fall out of the geometry:

• Armchair — (n, n). The cut edge looks like a row of armchairs. Always metallic.
• Zigzag — (n, 0). The edge is a zigzag. Metallic only when n is a multiple of 3.
• Chiral — everything else, (n, m) with n ≠ m ≠ 0. The hexagons spiral around the tube like the stripes on a candy cane. Metallic when (n − m) is a multiple of 3.

The wrapping angle θ — the chiral angle — runs from 0° (zigzag) to 30° (armchair):

cos θ = (2n + m) / (2·√(n² + n·m + m²))

Why one rule decides everything: (n − m) mod 3

Flat graphene is a famous oddity: a semimetal, or zero-gap semiconductor. Its filled valence band and empty conduction band do not overlap and do not leave a gap — they just barely touch at six special points in momentum space, the corners of the hexagonal Brillouin zone, called the K points. Electrons exactly at a K point can move with no energy cost, which is why graphene conducts.

When you roll the sheet into a tube, the electron's momentum around the circumference can no longer take any value — it is quantized, because the wavefunction must close on itself after one loop. The allowed momenta become a set of parallel lines slicing through graphene's 2D band structure. The tube is metallic if and only if one of those allowed lines lands exactly on a K point. Work the honeycomb geometry through and the condition reduces to a clean arithmetic test:

(n − m) mod 3 = 0   →  metallic
(n − m) mod 3 ≠ 0   →  semiconductor

Because n − m lands on 0, 1, or 2 (mod 3) with roughly equal frequency across random tubes, about one-third of all nanotubes are metallic and two-thirds are semiconducting. That statistic is the single most consequential fact in nanotube electronics: you cannot grow a batch of pure semiconductors; nature hands you a 1:2 mix and you must sort it.

(A footnote chemists like: curvature reintroduces a tiny "secondary gap" of a few meV in the (n,m) tubes that the simple rule calls metallic but where n ≠ m. True armchair tubes, (n,n), stay rigorously gapless because of their symmetry. For room-temperature purposes the simple rule is what matters.)

Sizing the band gap: 0.7 eV·nm over diameter

For the semiconducting tubes, the band gap is not fixed — it shrinks as the tube gets fatter, because a larger circumference packs the allowed momentum lines closer together, nudging them nearer the K point:

E_g ≈ 2·a_cc·γ₀ / d ≈ 0.7 eV·nm / d

  a_cc = 0.142 nm  (C–C bond length)
  γ₀   ≈ 2.7–3.0 eV (graphene nearest-neighbor hopping)
  d    = tube diameter in nm

So the band gap is a tunable knob set entirely by diameter — something silicon, with its fixed 1.1 eV gap, simply cannot offer:

Tube	Type	Diameter	Band gap E_g
(5,0) zigzag	semiconductor	0.39 nm	~1.8 eV
(10,0) zigzag	semiconductor	0.78 nm	~0.9 eV
(13,0) zigzag	semiconductor	1.02 nm	~0.7 eV
(17,0) zigzag	semiconductor	1.33 nm	~0.5 eV
(9,0) zigzag	metallic (9 = 3×3)	0.70 nm	0 (tiny curvature gap)
(10,10) armchair	metallic	1.36 nm	0 exactly

Compare a 1 nm semiconducting tube's ~0.8 eV gap to silicon's 1.1 eV: the same order of magnitude, which is precisely why people want to build transistors out of these. A semiconducting CNT field-effect transistor switches with the gap doing the same job the silicon gap does — but in a channel less than a nanometer wide.

Ballistic conduction and the quantum of conductance

A metallic nanotube is not just "a good conductor" — it can be a ballistic conductor, meaning electrons travel its length without scattering, like marbles down a frictionless pipe. There is no Ohmic resistance growing with length; instead the resistance is quantized. A single perfect metallic SWNT has exactly two conducting channels, giving a theoretical conductance:

G = 2 · G₀ = 2 · (2e²/h) = 4e²/h ≈ 155 µS
  → resistance ≈ 6.45 kΩ   (the quantum lower bound, R = h/4e²)

That 6.45 kΩ is a floor set by quantum mechanics, not by the carbon. A metallic tube can carry current densities above 10⁹ A/cm² — about 1000× what copper survives before it melts from its own heat (~10⁶ A/cm²). The same all-carbon bonding gives it a thermal conductivity near 3500 W/(m·K) along the axis, beating diamond and copper, so it dumps heat as fast as it carries current.

Strength: the sp² bond, stretched into a cable

The electronic story shares a root with the mechanical one: the carbon–carbon σ-bond. Pulling a nanotube along its axis pulls directly on those bonds, the same bonds that make diamond the hardest natural material.

Property	Single-walled CNT	High-strength steel	Kevlar
Tensile strength	~100 GPa	~2 GPa	~3.6 GPa
Young's modulus	~1 TPa	~0.2 TPa	~0.13 TPa
Density	~1.3 g/cm³	~7.9 g/cm³	~1.4 g/cm³
Specific strength	~77 GPa·cm³/g	~0.25 GPa·cm³/g	~2.6 GPa·cm³/g
Electrical type	metal or semiconductor	metal	insulator
Thermal conductivity	~3500 W/(m·K)	~50 W/(m·K)	~0.04 W/(m·K)

A single tube reaches ~100 GPa and a 1 TPa modulus — roughly 50× the strength of steel at one-sixth the density. The honest caveat: macroscopic CNT fibers fall well short, hitting only 5–10 GPa, because the tubes slide against one another rather than failing the bonds. The carbon is not the bottleneck; the load transfer between tubes is.

How they are grown and sorted

The dominant production route is chemical vapor deposition (CVD): a hydrocarbon feedstock (methane, ethylene, or ethanol) flows over nanoparticles of a transition-metal catalyst (Fe, Co, or Ni) at 600–1000 °C. The carbon dissolves into the hot catalyst particle, supersaturates, and precipitates as a tube whose diameter is templated by the particle size:

CH₄  --(Fe/Co/Ni catalyst, 700–1000 °C)-->  C(nanotube) + 2 H₂

Earlier routes — Iijima's original 1991 arc discharge between graphite electrodes, and pulsed laser ablation of a graphite/metal target — give high-quality tubes but in small amounts. CVD scales to kilograms and is how essentially all commercial nanotubes are made today.

Whatever the method, growth yields the statistical 1:2 metallic:semiconducting mix of chiralities. Sorting afterward is its own discipline:

• Density-gradient ultracentrifugation separates tubes by buoyant density, which correlates with diameter and electronic type.
• Gel chromatography exploits how strongly metallic vs semiconducting tubes adsorb to a gel column.
• Conjugated-polymer wrapping uses polymers (e.g. polyfluorenes) that selectively coat semiconducting tubes, lifting them into solution and leaving metallics behind.

The best of these now deliver 99.9% semiconducting purity — the threshold transistor arrays need, since even a single metallic tube can short a device.

Where carbon nanotubes show up

• Transistors. In 2019 a 16-bit microprocessor (RV16X-NANO) built entirely from CNT field-effect transistors was demonstrated — a proof that semiconducting-tube logic can work at silicon's scale. CNTs promise lower power at sub-nanometer channel widths where silicon's leakage explodes.
• Conductive composites. A few weight-percent of nanotubes turns an insulating polymer or epoxy conductive, drains static, and stiffens it — used in car body panels, aircraft parts, and battery electrodes.
• Li-ion battery anodes. CNTs are added as a conductive network that wires up silicon or graphite particles, improving rate and cycle life.
• Transparent conductive films. SWNT networks substitute for brittle, indium-scarce ITO in touchscreens and flexible displays.
• Field emission and sensors. The atomically sharp tube tips emit electrons at low voltage (displays, X-ray sources); single-molecule adsorption shifts a semiconducting tube's conductance enough to detect individual gas molecules.

Common misconceptions and pitfalls

• "Nanotubes are just tiny wires." Two-thirds of them are semiconductors, not wires. Treating a CNT sample as uniformly conductive ignores the chirality lottery that defines the material.
• "Chirality affects strength." It barely does — armchair, zigzag, and chiral tubes of the same diameter have nearly identical ~100 GPa strength. Chirality governs electronics, not mechanics. Don't conflate the two.
• "CNT fibers are as strong as a single tube." Single tubes hit ~100 GPa; real spun fibers reach only 5–10 GPa because the inter-tube van der Waals coupling lets tubes slide. The bond is not the weak link — the interface is.
• "Bigger band gap means better transistor." Band gap shrinks with diameter, so a too-fat tube barely switches (tiny gap, leaky off-state) while a too-thin tube has high contact resistance. There is a sweet spot near 1–1.5 nm.
• "Carbon nanotubes are the same as graphene or graphite." Same atoms, different boundary conditions. Graphene is a flat zero-gap sheet; a tube imposes circumferential quantization that splits its fate into metal or semiconductor; graphite is stacked sheets held by weak van der Waals forces.
• "Multi-walled tubes follow the (n−m) rule." Each shell has its own chirality, so a MWNT is an average over many — it behaves like a metallic-ish conductor overall, and the clean single-tube rule doesn't apply to the stack.