Carbon nanotubes are rolled-up graphene cylinders whose chirality sets whether they conduct like a metal or a semiconductor, and after three decades they are shipping as bulk additives, composites and thermal materials while the high-value prize, replacing silicon as a logic channel, remains blocked by contacts, doping and purity.
Summary
A carbon nanotube (CNT) is a seamless cylinder of graphitic carbon, either single-walled (SWCNT) or multi-walled. The electronic structure follows from folding the graphene band structure onto the tube circumference: the resulting chirality and diameter decide whether a given tube is metallic or semiconducting and, if semiconducting, what its bandgap is. Transport in a clean SWCNT approaches the ballistic limit, and the wrap-around geometry gives close to ideal electrostatic control of a channel, which is the basis of the argument that SWCNTs are the natural successor channel material once silicon CMOS runs into direct source-to-drain tunnelling, an unscalable subthreshold swing and the dark-silicon thermal ceiling.
The same object has three quite different commercial lives. First, as a bulk material: dispersed powder used as a conductive or structural additive, and vertically aligned CNT (VACNT) forests grown and transferred onto prepregs, veils and films for composites, catalysts, thermal interfaces, electrodes and sensors ref. Second, as a functional single object: chirality-sorted tubes as near-infrared emitters, strain transducers or quantum-confinement templates. Third, as an electronic channel in transistors, which is where nearly all the unsolved problems sit.
The parameters that decide outcomes are consistent across the literature. Chirality and diameter set the band structure and must be controlled or sorted after growth, typically today by density gradient ultracentrifugation in the lab. Wall number, intertube spacing and diameter govern the mechanics of assembled architectures such as graphene-CNT stacks and aerogels. Environment matters as much as the tube: the same nanotube shows higher field-effect mobility where it lies on hexagonal boron nitride than where it lies on SiO2, measured within a single tube to remove chirality and defect variability. Stable doping, especially n-type, is still hard because of environmental sensitivity and competing extrinsic effects, which is why encapsulating charge-transfer molecules inside the tube cavity is being pursued as a route that leaves the outer surface free for processing. And the contact remains a first-order problem: Schottky barrier height, transfer length and contact resistance in low-dimensional channels are set by metal-induced gap states whose scaling depends on dimensionality.
Composites, coatings and windings (web-enriched 12 Oct 2026)
Research for CNT Dispatch #013, brief at projects/cnt-dispatch/2026-10-12-edition-013-composites-brief.md.
- Tonnage is batteries. Cnano, the largest MWCNT maker, took c.97% of FY2025 revenue (c.RMB 1.26bn of 1.3bn) from battery slurry; its shipped c.75,000 t is slurry tonnes, not dry CNT ref. IDTechEx sizes the CNT market at c.US$540m (2025) rising past US$1.25bn by 2035 on batteries ref; Grand View’s US$3.71bn with “polymers” at 69.7% conflicts with producer data and likely counts compound value ref. No producer publishes a dry-tonnage split by application.
- Polymers and coatings use the electrical property, not the strength. Production cases are conductive fuel lines (Hyperion fibrils from the 1990s ref; Arkema MWCNT at 2-3 wt% ref) and electrostatically paintable parts (LG Chem CNT fenders on three Mitsubishi models, Jan 2023, c.4 kg saved per car ref, vendor PR). Conductive primers, tyres and EV battery shielding remain validation, prototype or patent stage.
- EU REACH caps OCSiAl’s SWCNT at 100 t/yr in Europe (no US cap) ref (vendor).
- Structural never arrived as promised. Single-shell strength of 11-63 GPa ref does not survive into bulk; NASA’s 2024 review says CNT-in-resin behaves like very short chopped fibre and the payoff was electrical ref. Lab fibre records (ECUST 12.5 GPa, Jul 2026 ref) beat T1100G carbon fibre on paper with no production path shown.
- CNT motor windings are demonstrators. MTC (UK) showed a “non-metallic motor” with CNT-yarn windings c.Sep 2024, no performance data published ref; KIST’s 2025 metal-free motor reached 3,420 rpm vs 18,120 rpm for a copper equivalent ref. Winding fibre conductivity sits at c.7.7-16 MS/m vs copper c.59 MS/m ref.
Evidence base
Open questions
- Does the metal-induced-gap-state scaling law in imply a contact resistance floor for one-dimensional channels that is compatible with high-density logic, or one that rules it out?
- Can endohedral encapsulation deliver n-type doping that survives ambient processing, and does any of the three filling routes compared in scale beyond gram-scale sorting by ultracentrifugation?
- Is the hBN mobility advantage over SiO2 reported in large enough to force an expensive substrate change, or can it be recovered by encapsulation or passivation on standard oxides?
- In thermal interface materials, do CNT-based products win design-ins against graphene-enhanced and metal-based alternatives, or remain one of several coexisting options as implied by ref?
Assessment drafted 2026-08-31 from up to 18 KB sources using the technology-scorecard framework; scores are a draft read pending review.