CNTs & The Standards Gap
Edition #009: CNT’s are not well defined or classified, and who is going to change that?
Welcome back y’all. Last edition was Where the Tonnage Goes, with batteries eating 80% of global CNT volume and set to eat more. If you missed Edition #008, catch up before reading this one, it sets the commercial scene.
A few of you wrote in asking variants of the same question. How do I tell if a CNT producer is actually shipping real material, not just a pitch deck? When a buyer specs “CNT” and gets delivered a pallet of grey powder, what did they actually get? Who polices any of this?
Good questions, and the honest answer is that for most buyers and most investors, you can’t really tell from the outside. Not without lab kit you don’t have.
So this edition I sat down with Terrance Barkan, Executive Director of the Advanced Carbons Council, which does most of the actual policing here. The ACC started in 2013 as the Graphene Council, then broadened to cover all engineered and reclaimed carbons about eighteen months ago because customers were already combining graphene, CNTs, synthetic graphite, biochar, and recycled carbon fibre into hybrid products, and there was nobody connecting the dots across the whole chain from raw petroleum or biomass through to end application.
What we spent most of the call on, and the most under-reported story in nanocarbons right now, is standards: classification, third-party validation, and the unglamorous infrastructure that sits between a producer’s claims and a buyer’s trust in those claims. For CNT as a commercial material, this is the rate-limiting step. The fix is arriving, from work the ACC has been doing for a decade.
Today’s edition: why is there is only a vague definition of carbon nanotubes, what does the graphene experience teach us, where does the regulatory patchwork actually matter, and who’s doing something about it?
——— ——— ———
5 Things You’ll Learn
1. What exactly is a CNT? Existing ISO and ANSI standards cover only some variants and leave genuine ambiguity for buyers. Specify “MWCNT” in a purchase order today and you might get materials ranging tenfold in price and performance.
2. ISO/TS 9651 for graphene just happened. The ACC assembled more than 100 subject matter experts into a volunteer task force, produced a five-part Graphene Classification Framework, then licensed the whole thing to ISO. The technical specification landed in 2025. A separate terminology standard (ISO/TS 80004-13) is in parallel flight under ISO/TC 229. CNT is one step behind on the same playbook.
3. Regulation splits across three axes. Jurisdiction (REACH, TSCA, METI, or nothing at all), role in the value chain (producer, distributor, end-user), and form factor (raw material, component, finished article). The same CNT material is regulated differently depending on which combination applies.
4. Toxicology is real but overstated. The asbestos-morphology concern is legitimate for one specific short-rigid-fibre variant of inhalable length, and has been over-generalised to the whole class. Long CNTs physically can’t stay straight at nanoscale aspect ratios, so they curl up and are biologically inert. Risk equals toxicity multiplied by exposure, and exposure is the controllable variable.
5. CNTs are already a success. Roughly 30 years into the standard 30-year adoption curve, with 80% of volume going into batteries, CNTs have found their commercial sweet spot. Barkan’s view of the end state is the same as polymers: used everywhere, in every industry, and no consumer can name a producer.
——— ——— ———
I. What Are You Actually Buying?
There’s a number Barkan dropped early in the call worth parking. The ACC actively tracks over 200 graphene companies and over 125 carbon nanotube companies. That 125 is the current set of serious outfits the Council keeps files on, not every firm that has ever claimed to make CNTs, and the Council validates whether each one is actually commercially active, meaning shipping product against real purchase orders rather than running on a website and a press release.
Worth anchoring that 125 against the hard graphene data Barkan and co-authors published in Nature Reviews Physics in late 2024. Global graphene production now runs at roughly 23,000 metric tonnes per year. Over 150,000 graphene-related patents have been filed since 2004, at a rate of around 18,000 new patents annually over the past five years. The commercialising company count peaked at about 250 globally and sits at 194 today after some winnowing, which is what you would expect as weaker producers drop out and the stronger ones consolidate. Of that 194, about 36% focus purely on producing GRMs as raw materials, about 38% develop GRM-integrated intermediates, and the rest do both.
CNT sits roughly a decade ahead of graphene on the same curve. Read the two numbers side by side and you’re watching the same adoption curve play out ten years apart.
For a CNT buyer in 2026, only a fraction of that 125 is actually shipping battery-grade or specialty-grade material at commercial volumes. The rest are academic spinouts, pre-production, or strategic pivots away. From the outside it’s hard to tell which subset, and that’s the core of the classification problem. When a battery OEM’s procurement team specifies “multi-walled carbon nanotube”, what have they actually specified? MWCNTs run from roughly €40-100 per kilo at the commodity end up to several thousand per kilo at the specialty end, with performance variance across that range wider than the pricing suggests. Without a shared classification framework and a shared technical data sheet format, the buyer is trusting the seller’s characterisation of their own product. A bit like asking a fund manager how good their own fund is.
Barkan makes a related point in the Nature Reviews Physics paper, worth quoting directly: “Many large-volume applications of graphene are treated as trade secrets, often not appearing in patent reviews or advertisements. This makes it easy to underestimate the actual level of graphene use.” The same applies to CNT, probably more so. When battery OEMs treat their additive suppliers as confidential and composite manufacturers protect formulations as IP, the public market size understates the installed material base. Another reason to be sceptical of any tonnage forecast that leans heavily on patent filings or public company disclosures.
Barkan’s observation on the investor side was sharper than I expected, and it matches my own experience. The VC community by and large doesn’t have the technical sophistication to test whether a CNT startup’s claimed material matches its spec, and nor would you expect them to. Characterising a CNT properly needs Raman spectroscopy, transmission electron microscopy (TEM), scanning electron microscopy (SEM), X-ray photoelectron spectroscopy (XPS), thermogravimetric analysis (TGA) and a set of electrical transport measurements, and someone who knows which subset matters for the specific application. That’s a capability most seed-stage investors don’t have in-house, and it’s expensive to outsource deal-by-deal. The industry default has been to trust producer-supplied technical data, which is the same trust problem the buyer has, one rung down the chain.
So the question the ACC has been circling since day one is whether you can build a framework that lets a buyer know what they’re getting, a seller prove what they made, and a third party audit the claim independently. For graphene that framework now exists, which is where the interview got properly interesting.
——— ——— ———
II. The Graphene Warning
Barkan’s framing is that the rate-limiting step for new materials going into real industry is trust, rarely science or scaling. I agree. A buyer at a large industrial OEM needs three things before they’ll spec a new material into a product: the material is what the data sheet says, they can get it reliably, and it’s cleared to ship into the markets they sell to. Everything else is downstream.
So the Council built the graphene playbook first, on the basis that if you can’t solve the trust problem for a single material, you won’t solve it for any of the related ones. The mechanism was a Graphene Classification Framework produced by a volunteer task force of about 100 subject matter experts working together over several years. The framework has five parts: which physical and chemical characteristics of the material must be measured, the specific techniques used to measure each, the expected ranges of values, a neutral nomenclature to describe the material without vendor jargon, and a standardised technical data sheet for both sides to use.
The Council then licensed the whole framework to ISO, which turned it into ISO/TS 9651, published in 2025. For graphene, there’s now a canonical international definition of what the material is, how to measure it, and how to describe it on a data sheet.
On the CNT side, the same playbook is one step behind. There’s an ISO terminology standard in place (ISO/TS 80004-3:2010) but it’s skeletal: it defines a CNT as a “nanotube composed of carbon” with a note that they “usually consist of curved graphene layers”. Commercially, that doesn’t tell a buyer what they’re getting. Existing ANSI and ISO standards on multi-walled CNTs specifically cover the parameter space incompletely and rest heavily on older source data. The ACC is now working with ANSI in the US and with ISO internationally on a revision, using the lessons from the graphene process.
Where each material sits on the standards arc:
ACC-developed framework. Graphene has the Graphene Classification Framework (five parts, 100+ SMEs). CNT is in revision, following the graphene template.
ISO characterisation standard. Graphene has ISO/TS 9651, published 2025. CNT has older ANSI / ISO standards still in revision.
ISO terminology standard. Graphene has ISO/TS 80004-13 under ISO/TC 229. CNT has ISO/TS 80004-3:2010, a minimal placeholder.
ACC-tracked producers. Graphene peaked at 250, sits at 194 today. CNT at 125 and rising.
Annual production. Graphene at roughly 23,000 metric tonnes. CNT at roughly 35,000 metric tonnes (2022 baseline, rising).
Three lessons from the graphene slog are worth pulling out, because they tell you what to expect on the CNT side.
First, the hardest part of a classification effort isn’t the measurement science. Most of the individual techniques (Raman, BET, TGA, conductivity) are mature. The hardest part is the political work of getting a hundred people who work for competing producers to agree on which techniques go in the canonical set, which values count as acceptable ranges, and which vendor terms are banned as marketing jargon. Consensus takes years even when the science is settled, and that’s the part the CNT revision will spend most of its time on.
Second, a published standard only solves half the problem. Classification tells you what the material should be and how to measure it, but it doesn’t enforce anything. Somebody still has to check that the producer’s output actually matches the classification (an audit job), and somebody still has to get the material through every jurisdiction’s regulatory regime (a lawyers-and-toxicologists job).
Third, standardisation plays out over years and keeps evolving after publication. Graphene has ISO/TS 9651 for characterisation, the ACC-led framework we’ve just described. In parallel, ISO/TC 229 has been working on a terminology standard, ISO/TS 80004-13, which defines “graphene-related two-dimensional materials” (GR2Ms) as a category. A 2026 correspondence in Carbon by Clifford and co-authors specifically defends the GR2M definition against alternatives proposed by Gulumian and Fadeel. Translation: even after the standards land, the scientific community keeps arguing about the words. CNT should expect the same, and the first document is rarely the last word.
Barkan’s co-authored Nature Reviews Physics paper also shows that by 2022, corporations filed 76% of all graphene patents, up from earlier years. Within that corporate share, traditional technology and industrial companies (not graphene-native startups) now account for ~94% of filings, up from 84% in 2019. The biggest holders of graphene IP today are mainstream industrials pulling graphene into their own product roadmaps rather than specialist firms trying to sell graphene to them. That’s the adoption signal you want. For CNT, the equivalent test is whether Tier-1 battery OEMs, coatings majors, and composite manufacturers are filing CNT-related patents under their own names. Increasingly they are, though the public data is thinner because the CNT IP ecosystem is a decade behind graphene’s.
——— ——— ———
III. The Regulatory Patchwork
One interview moment I really wanted to pin down: how does the regulatory stack actually look for a Western producer shipping battery-grade CNT globally? The journalism on this is vague, the compliance cost is not, and the answer depends on three axes.
Jurisdiction. Every advanced economy has a chemicals regime that classifies nanomaterials as chemicals. Europe has REACH, the US has TSCA under the EPA, Japan and Korea have METI and KCS equivalents, and Canada and Australia run their own variants. A producer has to clear the registration process for each market they sell into, and the data requirements aren’t harmonised across them. Several Middle Eastern markets don’t regulate nanomaterials at the same level, which gives freedom to operate and complicates the global picture rather than simplifying it.
Role in the value chain. Obligations differ depending on whether a company produces the material, imports or distributes it, or incorporates it into a product as an end-user. Producers carry the heaviest registration burden. Distributors and end-users have lighter, overlapping duties. Confusing the categories is how early-stage companies typically get into compliance trouble.
Form factor. This was the cleanest illustration in the whole interview. Take the same CNT material. If you disperse it into a liquid paint and ship the paint into the United States, the paint is a chemical formulation and each component must be registered under TSCA, including the CNT. If you incorporate that same CNT into a pre-preg composite, cure it, and ship the finished composite part into the US, that part is treated as an “article” under TSCA and is largely exempt from the chemical registration requirement. You can import the finished article without the embedded CNT going through full registration.
How the same MWCNT material gets treated under TSCA, by form factor:
Raw CNT powder or dispersion sold to a processor. Intermediate chemical; full chemical registration required.
Dispersed in a liquid paint, sold as paint. Each component must be registered, including the CNT.
Incorporated into a cured pre-preg composite, sold as part. Treated as an “article”; largely exempt from chemical registration.
Same molecule, treated differently depending on what shape it’s sold in. This has real commercial consequences, because a producer’s decision to sell dispersions versus finished composite parts turns out to be a regulatory choice as much as a margin one.
Now the toxicology question, because it keeps coming up whenever someone writes about CNTs and most people are loose with it. The legitimate historical concern was that certain CNT morphologies share a fibrous structure with asbestos. When high aspect-ratio rigid fibres of the wrong length are inhaled, the lung’s clearance mechanisms struggle. That concern drove two decades of toxicology research, which produced a more nuanced picture than the original alarm.
What the research actually showed is that toxicity is parameter-dependent, and the parameters don’t vary independently. A CNT that’s nanometres wide and millimetres long can’t physically stay straight or rigid, which makes long CNTs biologically inert: they curl up in the body and get cleared. The dangerous morphology is the opposite, short rigid fibres of inhalable length, stiff enough to lodge in tissue. The regulatory regime was built on that short-rigid-fibre variant, and most commercial CNT today inherits its shadow despite being structurally different. The ACC’s CNT task force is working on parameter-aware standards rather than one-size-fits-all.
The practical point is that risk equals toxicity multiplied by exposure. A toxic material with effectively zero exposure is zero risk. For CNT production and handling, proper practice (gloves, fume hoods, dispersion-phase handling rather than dry powder) takes occupational exposure near zero. For most commercial end-products, the CNT is dispersed in a matrix and bonded to active material, so consumer exposure at the use phase is minimal. For battery-grade MWCNT in dispersions going into sealed cells, the toxicology question is effectively answered.
What remains genuinely sticky is the fragmentation itself. A producer today has to navigate five or six jurisdictions with different timelines, different data requirements, and different interpretations of the same material. This is where the ACC’s standards work eventually pays the biggest commercial dividend, because a harmonised classification lets regulators share data rather than re-testing from scratch each time.
——— ——— ———
IV. The Task Force, or the Dow Jones of Nanocarbons
The most interesting part of what the ACC does happens after the standards are written. In the interview I compared it to Dow Jones, which audits the financials of companies listed on it rather than just setting the listing rules. Barkan agreed with the analogy, and it’s a useful way to frame the Council’s role.
The Council runs six standards and classification task forces, covering graphene, carbon nanotubes, carbon nanofibres, recycled carbon fibre, biochar, and MXenes. Graphene is the furthest along, having produced ISO/TS 9651. CNT is one step behind, in revision at ANSI and ISO. The others sit at various points earlier, with biochar and MXenes at the start of their commercial arcs and therefore at the start of their standards arcs.
ACC runs six standards and classification task forces:
Graphene. ISO/TS 9651 published 2025. Flagship output.
Carbon Nanotubes. MWCNT revision underway at ANSI and ISO.
Carbon Nanofibres. Earlier stage, structural overlap with CNT.
Recycled Carbon Fibre. Earlier stage, upcycling driver for nano-enhancement.
Biochar. Early in commercial arc, early in standards arc.
MXenes. Newest of the six, earliest arc.
Then there’s the Verified Producers Programme, which is the part I think matters commercially. It goes well beyond standards-setting. ACC auditors physically visit production facilities. They watch raw materials come in, observe the full process from intake to output, check quality control and health and safety practice, sample the output stream, and characterise it themselves against ISO/TS 9651 or the equivalent CNT standard. They also validate the producer’s capacity claims, verifying that a reactor the producer says makes X tonnes per year actually makes X tonnes per year. The producer ends up with a technical data sheet based on ACC’s own third-party testing, rather than one produced in-house.
That’s an audit activity, closer in shape to financial audit than to standards-setting. A third party with its own access rights and its own testing methodology sits between the producer and the buyer, underwriting the truthfulness of claims that both sides have commercial reasons to want believed.
For a battery OEM trying to source Western-origin MWCNT at spec, a verified producer list does a chunk of your due diligence for you. For a VC sizing up a CNT startup, the ACC-issued TDS is arguably more useful than the company’s own pitch deck, because it’s produced by a party that didn’t write the deck and isn’t paid by the company.
On the investor point specifically, I asked Barkan what question a pre-seed investor could ask a CNT startup that would filter out the worst pitches. His answer, paraphrased: ask for the technical data sheet, and ask whether it was produced by third-party testing against an ISO or ANSI standard. If the company can’t put an independent TDS in front of you, everything they say about performance is self-reported.
Useful filter. I’ll be using it.
——— ——— ———
From Classification to Infrastructure
So what’s the story?
The CNT industry has been through the hype cycle graphene went through a decade later. It went into the desert at the bottom of the BASF-era over-build. It’s now 30 years into the standard 30-year adoption curve, at the point where the material finds its commercial sweet spot. For CNT that sweet spot has become batteries, 80% of volume in a single vertical. Graphene is roughly a decade behind on the same curve, at 23,000 tonnes per year and 194 active commercialising companies, still searching its sweet spot across 40-45 verticals.
Barkan’s framing of the end state, which I’ll steal because it’s the right one, is plastic. Polymers are a platform technology that built modern manufacturing. They show up in every product in every industry. No consumer can name a producer. The big polymer companies sit in industrial parks nobody visits, making boring-sounding compounds with names nobody knows. That’s success for a materials platform, because the material has become infrastructure.
Nanocarbons are on the same trajectory. The invisible additives, conductive networks and thermal layers that nobody writes breathless press releases about but without which the next decade’s batteries, chips, composites and coatings don’t function.
Barkan’s view in Nature Reviews Physics gets closer to the real destination than most market forecasts do. A material’s principal applications, he argues, are the ones it creates rather than the ones where it offers only incremental improvements. For CNT that means the application that matters most in 2035 probably isn’t the one getting most of the press coverage in 2026. Watch the applications where CNT enables something that simply didn’t exist before: DexMat-style conductive fibres, Carbice-style vertically aligned thermal interfaces, and whatever comes next in silicon anodes and solid-state. The commodity uses follow behind, but the category-creating ones are what actually define the platform.
The path there runs through standards. Classification, third-party validation, harmonised regulation, and an audited supplier list that buyers can actually trust. The ACC’s view, and mine after this conversation, is that the commercial bottleneck for nanocarbons in 2026 is a trust problem rather than a scientific one. You build trust slowly, through the unglamorous work of measuring the same thing the same way and then going on-site to check.
For CNT buyers, the practical takeaway is to start asking suppliers for third-party technical data sheets produced against ISO or ANSI standards. For investors, Barkan’s filter (an independent TDS or a self-reported one) sorts most decks without a technical deep-dive. Everyone else should watch the standards story, because it’s slow and procedural and consensus-driven, and because good procedural stories don’t photograph well. It’s also what determines whether the tonnage numbers in Edition #008 actually get delivered.
Further reading: Barkan, Ratwani, Johnson, Thodkar & Hill, “Mapping the landscape for graphene commercialization”, Nature Reviews Physics 6, 646-647 (2024). And the 2026 correspondence on GR2M nomenclature in Carbon, responding to Gulumian and Fadeel.
Thanks, Lawrence x