18 areas where 2D materials are used or promised, sorted by how close each is to something you can buy. What ships today is mostly additives and films inside other products. The devices that made graphene famous are still years away, and some may never arrive.
This page is written for everyone. Each material links to its catalogue page, which has notes for theoreticians; the theory track covers the open questions behind these applications.
This page is written for everyone. The experiment track covers how the materials behind these applications are grown, transferred and characterised.
Where an area shows a readiness level, it comes from the engineering track, which also gives the reasoning, the blocking issue and who is working on it.
When each use is expected to reach products, in years from now: a dot for what is sold today, a bar for the window the rest are expected in, grey where no date can be given. Select one for the details.
5 areas
Shipping now
Sold today, usually as an additive or a film inside someone else’s product.
Composites and coatings
The one place 2D materials are already sold by the tonne
Stronger, stiffer or more conductive plastics, rubbers, concrete and paints, made by mixing in a small amount of graphene powder rather than redesigning the product.
Where it stands today
Commercial and growing, but the improvements are incremental percentages rather than transformations. This is where most of the world’s graphene production is actually consumed.
Read moreShow less
How it works
Graphene powder is not a single perfect sheet. It is a grey dust made of millions of tiny , each a few atomic layers thick and a few thousandths of a millimetre wide. Stirred into a resin or a cement mix at well under one percent by weight, the flakes get in the way of cracks trying to spread and they bridge gaps between the other ingredients. The result is a material that is tougher, or conducts more heat and electricity, without changing the factory process. Almost all of the value here comes from the flakes being thin and stiff, not from the exotic quantum behaviour that made graphene famous.
Real example
HEAD has sold tennis racquets marketed with graphene since 2013, and Vittoria markets graphene-enhanced bicycle wheels and tyres. Neither is a product: both use small additions of graphene-related flakes.
The catchMarketing rarely says how much graphene is in the product, and a 2018 study that tested material from 60 producers worldwide concluded that most were really producing graphite microplatelets rather than graphene. The physics of a perfect single sheet does not carry over to a bucket of flakes.
Thermal management in phones and chips
Moving heat sideways out of a device with no room for a fan
Thin films that spread heat away from a hot processor across the body of a phone, laptop or LED module, so the chip does not have to slow itself down to stay cool.
Where it stands today
In shipping consumer hardware, but most so-called graphene are highly oriented graphite films, closer to compressed pencil lead than to single-layer graphene.
Read moreShow less
How it works
Heat travels through a solid as vibrations of its atoms. In graphite and graphene the atoms are bonded very stiffly within each layer but only weakly between layers, so vibrations run along a sheet extremely fast and pass between stacked sheets slowly. That makes a stack of aligned sheets an excellent sideways heat pipe and a poor vertical one, which is exactly what you want for pulling heat from under a chip and spreading it across a large area. Commercial films are compressed graphite or reduced graphene oxide sheets, tens of micrometres thick, laminated with adhesive.
Real example
Huawei’s Mate 20 X (2018) was marketed with a ‘graphene film’ cooling system working alongside a vapour chamber; the film supplier was reported to be China’s The Sixth Element Materials.
The catchThe eye-catching figure for graphene, a thermal conductivity of several thousand watts per metre per kelvin, was measured on a single suspended sheet in a laboratory. Real films land far below that, which is still better than copper sideways, but the headline number is not what is in your phone.
Sensors and medical diagnostics
The first commercial devices where the 2D material really is the active part
Magnetic field that work over a wide temperature range, down to cryogenic temperatures, and chemical and biological sensors that respond to very small amounts of a target molecule.
Where it stands today
Magnetic sensing is a real, small, shipping business used in scientific and industrial instruments. Chemical and medical sensing has many demonstrations, a few small products, and a persistent problem with drift and selectivity.
Read moreShow less
How it works
Graphene is all surface. Every one of its atoms is exposed, so anything that lands on it, or any magnetic field passing through it, immediately changes how easily current flows. In a Hall sensor a magnetic field pushes the moving charges to one side of the sheet and you measure the voltage that builds up across it. Because graphene has few charge carriers and they move very fast, that voltage is unusually large, which makes the sensor sensitive and stable over a wide temperature range. In a chemical sensor the same all-surface property means a handful of adsorbed molecules produces a measurable signal.
Real example
Paragraf, a University of Cambridge -out, sells graphene sensors grown on without a , including a cryogenic version in volume production for high-field and quantum-computing measurements.
The catchSensitivity is the easy half. A sensor that responds to everything tells you nothing, and getting a graphene device to give the same answer tomorrow as it did today, in humid air, is where most diagnostic projects run into trouble.
Batteries
A useful additive, not the battery breakthrough that keeps being announced
Faster charging and longer life in otherwise conventional cells, mainly by improving how electricity moves through the electrode powder.
Where it stands today
Graphene are used in commercial cell manufacturing. The recurring headline of a graphene battery with several times the capacity is not on sale, and there is no commercial cell in which graphene is the material that stores the charge.
Read moreShow less
How it works
A battery electrode is a compressed powder of active particles that store lithium, glued together and wired up by a conductive filler, normally carbon black. Swapping some of that filler for graphene flakes gives the same electrical connection with less added weight, because flat flakes touch more particles than round soot does. Graphene is also used as a flexible wrapper around silicon particles, which store far more lithium than graphite but swell and crack while doing so. In both roles graphene is the supporting cast; the lithium chemistry does the storing.
Real example
In China, graphene-containing conductive pastes are an established product category for lithium-ion electrodes. No commercial cell has been independently shown to store more energy because of graphene.
The catchConsumer products labelled graphene battery usually mean a normal lithium-ion cell with a graphene-containing additive, or simply a graphene heat spreader taped to the pack. Energy density is set by chemistry, and adding a conductor does not change the chemistry.
Supercapacitors
Storing much less energy than a battery, but delivering it in seconds
Devices that charge in seconds, survive a million cycles and work in the cold, used for regenerative braking in trams and buses, engine starting, and holding wind turbine blades steady.
Where it stands today
Commercial using carbon materials with graphene-like structure are on sale and in service. They store roughly one-twentieth to one-fiftieth of the energy of a lithium-ion cell of the same weight, and that gap has not closed.
Read moreShow less
How it works
A battery stores charge by changing the chemistry of its electrodes, which is slow and eventually wears them out. A supercapacitor stores charge physically, as ions parked on an electrode surface, which is fast and almost endlessly repeatable. The amount you can store is therefore set by how much surface you have, and a gram of graphene has a theoretical surface area of about 2630 square metres, roughly ten tennis courts. In practice the sheets stick back together into stacks, so real electrodes get nowhere near that number.
Real example
Skeleton Technologies (Estonia and Germany) sells supercapacitors built on its proprietary ‘Curved Graphene’ – a powder of crumpled graphene-like sheets, produced at Bitterfeld-Wolfen in Germany – for grid, transport and data-centre power systems.
The catchPapers routinely report huge capacitance per gram, measured on electrode coatings far thinner and lighter than any commercial electrode. Numbers quoted per gram of active material tell you almost nothing about the energy of a packaged, sealed, commercially useful cell.
3 areas
In 2–5 years
Working prototypes exist; cost, scale and qualification decide whether they become products.
Electromagnetic shielding
Blocking radio noise with something thinner and lighter than metal foil
Thin, flexible, sprayable shields that stop radio signals leaking out of one circuit and into another, for aerospace, satellites and dense electronics where every gram counts.
Where it stands today
Laboratory performance is excellent and has been reproduced by many groups. The move to products is early: the material is not yet made at industrial scale and the films degrade in humid air.
Read moreShow less
How it works
Radio waves are reflected by anything that conducts electricity, which is why metal boxes block them. A conductive coating does the same job if it conducts well enough, and inside a stack of flakes every internal interface reflects the wave again, so it rattles around and loses energy rather than passing through. , a family of made by chemically a layered ceramic, stand out here because they conduct like a metal and can be sprayed from water as an . Films a few tens of micrometres thick block the overwhelming majority of incoming radio energy.
Real example
Laboratory benchmark rather than a product: a 45-micrometre MXene film blocked 92 decibels of radio energy (Shahzad et al., Science, 2016). No confirmed commercial product yet.
The catch papers quote decibels, a scale where 20 decibels means 99 percent blocked and 60 decibels means 99.9999 percent. Very high numbers look dramatic but often far exceed what any application needs, while the real obstacles, cost and shelf life, get no headline.
Corrosion barriers
Slowing rust, when the coating is formulated correctly
Paints and primers that slow the corrosion of steel in marine and industrial settings, potentially using less zinc and fewer additives.
Where it stands today
Graphene-containing coatings are marketed, from industrial primers to consumer car-care products. Independent long-term field data comparing them fairly against good conventional coatings remains thin.
Read moreShow less
How it works
Rust needs water, oxygen and dissolved salts to reach the metal. Flat, chemically inert flakes dispersed in a paint force those molecules to travel a long winding path around each flake instead of diffusing straight through, which can slow the attack considerably. There is a catch: graphene also conducts electricity, and a conductive layer in direct contact with steel can act as one half of a tiny battery and make the corrosion faster, not slower. Working formulations keep the flakes well dispersed and electrically isolated from the metal underneath.
Real example
Consumer ‘graphene’ car-care coatings are widely marketed, usually without a disclosed graphene content. The warning result is Schriver et al. (ACS Nano, 2013), who found that bare graphene films made copper corrode faster over long periods – their title called graphene ‘worse than nothing’ as a long-term barrier.
The catchThe galvanic corrosion problem is real and was documented early. A badly formulated graphene coating accelerates rusting, which is the exact opposite of the sales pitch, and salt-spray test results are easy to present selectively.
Photodetectors and infrared imaging
Seeing wavelengths that silicon simply cannot see
Cameras that see short-wave infrared light, which passes through haze, plastic and silicon wafers, at a fraction of the cost of the exotic used today.
Where it stands today
Prototype image sensors have been built directly on silicon readout chips, and spin-off companies are pursuing products. Whether the technology can undercut established infrared sensors on price at volume, and match them on noise and stability, is still undecided.
Read moreShow less
How it works
A silicon camera sensor is blind to infrared beyond about 1.1 micrometres because that light does not carry enough energy to knock an electron loose in silicon. The standard alternative, indium gallium arsenide, works well but is expensive and hard to make in large arrays. One 2D approach uses a layer of light-absorbing sitting on a graphene sheet: the dots catch the infrared , and graphene, which conducts extraordinarily well, amplifies the resulting tiny charge into a readable signal. The whole stack can be deposited directly onto a conventional silicon readout chip.
Real example
Demonstration: a graphene-quantum-dot image sensor built directly on a readout chip (Goossens et al., Nature , 2017, ICFO Barcelona). No confirmed volume product yet.
The catchBare graphene absorbs only about 2.3 percent of the light that hits it, so on its own it is a poor detector. Nearly every impressive graphene is really a hybrid, and the quantum dots doing the absorbing bring their own stability and toxicity problems.
8 areas
5–15 years away
Demonstrated in laboratories and sometimes in trials, but a product still needs problems solved that are not yet solved.
Biomedical implants and neural interfaces
An electrode that talks to nerve cells without being made of metal
Brain and nerve interfaces that record and stimulate with much smaller electrodes than metal allows, and that do not blur medical scans.
Where it stands today
In early human clinical testing rather than routine use. First-in-human studies of a graphene cortical interface began in 2024; results so far concern safety and signal quality, not treatment of disease.
Read moreShow less
How it works
An implanted electrode has to move charge into salty tissue without corroding or triggering chemical reactions. Metal electrodes do this through a small effective surface, so making them smaller quickly makes them useless. Graphene is all surface and chemically inert, so a graphene electrode can move much more charge for its size, letting you pack many more recording sites into the same footprint. Being carbon and only atoms thick, it also produces far less distortion in an MRI scan than a metal wire does.
Real example
INBRAIN Neuroelectronics (Barcelona) carried out the first human procedure with a graphene cortical interface in September 2024 at Salford Royal Hospital, Manchester, placing it temporarily on the brain during tumour surgery. Enrolment in the first-in-human safety study was completed in 2026.
The catchThe bottleneck for neural implants has never been the electrode material alone. It is whether the device still works after ten years inside a warm salty body that is actively trying to wall it off, and no 2D material has been in a human that long.
Water filtration and desalination
A sieve with holes the size of a single molecule
that separate salt, dyes or specific ions from water while letting water pass faster, which could make desalination plants smaller and industrial separations cheaper.
Where it stands today
Laboratory membranes work impressively at the scale of a coin. Commercial reverse osmosis is a mature, heavily optimised technology, and no 2D membrane has yet been shown to beat it over the size and lifetime a plant requires.
Read moreShow less
How it works
Desalination today mostly works by forcing seawater at high pressure through a polymer film that water molecules can wriggle through and salt ions cannot. Most of that pressure is needed just to overcome the osmotic pressure of the salt, which no membrane can avoid: a membrane three times more permeable would save only about 15 percent of the energy in a seawater plant, though it could cut the number of pressure vessels almost in half (Cohen-Tanugi and colleagues, 2014). Two 2D approaches exist: punch precise holes in a single graphene sheet, or stack graphene oxide flakes so water slips through the gaps between them while larger ions are blocked. The stacked version is far easier to make at any useful size.
Real example
No confirmed commercial desalination product yet. The key laboratory result: graphene-oxide membranes with physically restrained swelling sieved common salt out of water (Abraham et al., Nature Nanotechnology, 2017).
The catchA membrane is only as good as its worst hole. One defect across a square metre lets salt through and ruins the separation, so the real challenge is not making a great small membrane but making a merely good enormous one. Graphene oxide stacks also tend to swell in water, which widens the gaps and undoes the selectivity.
Flexible and wearable electronics
Circuits that bend, because a layer this thin barely strains when it does
Electronics that fold, stretch or sit on skin: health patches, rollable screens, sensors woven into clothing.
Where it stands today
Physics demonstrated many times over; products scarce. Indium tin oxide, the incumbent transparent conductor, is brittle but cheap and well understood, and silver-nanowire and metal-mesh films have taken much of the flexible market that graphene was expected to win.
Read moreShow less
How it works
Bend a material and its outer face stretches while its inner face compresses. The thinner the layer, the less stretch it experiences for the same bend, which is why a sheet one atom thick can be wrapped around a hair without cracking. A flawless graphene flake can in principle be stretched by around twenty percent before it fails, an enormous figure for a stiff material; real films with crack far sooner. Combine a transparent, conducting, bendable sheet with a bendable plastic backing and you have the basic ingredients for a circuit that survives being folded.
Real example
No confirmed mainstream product. The best-known demonstration is still the 30-inch roll-to-roll graphene film built into a working touch panel (Bae et al., Nature Nanotechnology, 2010).
The catchBendability was never the hard part. Making transparent graphene electrodes cheaply, transferring them onto plastic without tearing, and getting low enough to compete with a metal mesh has defeated a decade of effort and a lot of funding.
Displays and LEDs
The application graphene was supposed to own, and does not
Transparent electrodes for screens, and light-emitting structures built by stacking different 2D crystals a few atoms apart.
Where it stands today
Not in any commercial display. The incumbent materials are cheap and adequate, and the 2D versions have never reached the uniformity, brightness or lifetime that a mass-produced screen demands.
Read moreShow less
How it works
Every pixel in a screen needs a transparent electrical contact on top so light can escape while current gets in. A single graphene sheet absorbs about 2.3 percent of visible light and conducts well, which sounds ideal. Separately, some 2D semiconductors emit light efficiently when they are exactly one layer thick, and a sandwich of a light-emitter between two insulating boron nitride layers and two graphene contacts makes a working LED thinner than a virus. Both ideas have been demonstrated repeatedly in laboratories.
Real example
No confirmed commercial example yet.
The catchThis application has been announced as imminent since around 2010. Displays are an extremely cost-driven industry, and a marginally better electrode that costs more per square metre never wins.
Hydrogen production and catalysis
Cheap catalysts in place of platinum; refineries have used MoS2 for decades
Cheaper for splitting water into hydrogen and oxygen with electricity, reducing dependence on platinum-group metals.
Where it stands today
A very large, very old industrial use of layered molybdenum disulfide already exists for removing sulfur from crude oil, running for decades at refinery scale. For water splitting, 2D catalysts remain in the research and pilot stage and have not displaced platinum in commercial electrolysers.
Read moreShow less
How it works
Splitting water needs a catalyst that binds hydrogen atoms just tightly enough: too weak and nothing sticks, too strong and nothing leaves. Platinum happens to sit at that sweet spot, which is why it is used and why it is expensive. In molybdenum disulfide the flat top surface is nearly inert, but the exposed edges of each sheet bind hydrogen almost as well as platinum does. So the whole game is making material with as much edge and as little flat surface as possible, by shredding it, curling it, or drilling defects into it.
Real example
Industrial precedent: cobalt- and nickel-promoted molybdenum disulfide catalysts have run refinery desulfurisation for decades. For hydrogen, the key laboratory result is that MoS2 edges are the active sites (Jaramillo et al., Science, 2007); no commercial electrolyser uses a 2D catalyst yet.
The catchPapers usually compare against platinum on one measure, the voltage needed to start the reaction, and stop the test after a few hours. Commercial electrolysers run for tens of thousands of hours at high current, and almost no 2D catalyst has published data on that timescale.
Solar cells
A supporting layer in someone else’s solar cell
Transparent electrodes and charge-collecting layers that could make emerging cheaper, more stable, and printable on flexible sheets.
Where it stands today
Panel-scale demonstrations exist from research consortia. Silicon photovoltaics has become so cheap that the commercial space for an added, more expensive layer is very small.
Read moreShow less
How it works
A solar cell needs to absorb light, separate the resulting positive and negative charges, and get each one out of a different side. 2D materials are mostly proposed for the getting-them-out part, as transparent contacts or as very thin layers that let one type of charge through and block the other. In perovskite solar cells, which are cheap and efficient but degrade, graphene-family layers have been used to slow that degradation. The light absorption itself is still done by silicon or perovskite, not by the 2D material.
Real example
A stand-alone solar farm of nine graphene-perovskite panels – 4.5 square metres in total, peak output about 250 watts – ran in Heraklion, Crete, and kept about 80 percent of its initial power after eight months (Nature Energy, 2022). Commercial silicon panels carry 25-year warranties.
The catchRecord efficiencies for graphene-containing cells are usually set on devices smaller than a fingernail. Solar economics are set by cost per watt on a panel of about two square metres, and small-area records have a long history of not surviving the scale-up.
Electronics beyond silicon
The reason chipmakers care, and the slowest thing on this list
that keep working when the channel is a single atomic layer thick, which is where silicon starts to fail, and stacked layers of logic built vertically.
Where it stands today
Serious industrial research. Major chipmakers and research institutes have grown 2D semiconductors on 300 millimetre wafers and built test transistors, and 2D channels appear on the semiconductor industry’s long-range roadmap. In 2025 a Chinese group reported a working microprocessor of several thousand transistors built from molybdenum disulfide, which is roughly the complexity of a commercial chip from 1975.
Read moreShow less
How it works
A transistor is a switch: a gate electrode controls whether current flows through a thin channel. Making everything smaller has made chips better for fifty years, but as the silicon channel gets thinner than a few nanometres its surfaces start scattering the electrons and it conducts badly. A 2D semiconductor such as molybdenum disulfide is naturally about 0.65 nanometres thick with no rough surfaces and no , so it still conducts respectably at a thickness where silicon does not. That is a specific, narrow and valuable advantage.
Real example
WUJI, a 32-bit RISC-V microprocessor built by Fudan University from 5,900 molybdenum disulfide transistors (Nature, April 2025). The previous record for a 2D logic chip was 115 transistors.
The catchA modern processor has tens of billions of transistors and every one must work. Going from thousands to billions is not a matter of trying harder; it requires defect densities and that no 2D material has yet come close to, and industry roadmaps put the earliest realistic insertion in the 2030s.
Quantum technology and single-photon sources
Making light one particle at a time, from a defect in a crystal
Sources that emit exactly one photon on demand for quantum communication and quantum sensing, and hosts for quantum bits that can be stacked and stuck together by hand.
Where it stands today
Active, credible physics research with commercial materials suppliers but no commercial device. The central problem is control: nobody can reliably place an emitter where they want it or make two of them identical.
Read moreShow less
How it works
Quantum cryptography needs light emitted strictly one particle at a time, because two photons let an eavesdropper steal one unnoticed. Certain atomic-scale imperfections in hexagonal boron nitride, an insulating 2D crystal, behave like isolated artificial atoms: excite one and it emits a single photon, then must be re-excited before it can emit again. Because the crystal is only atoms thick, that photon escapes easily instead of being trapped inside a bulk material. These emitters are unusually bright and, unlike many rivals, work at room temperature.
Real example
No confirmed commercial example yet.
The catchThe emitters are still not fully understood at the atomic level, and different laboratories report different behaviour from nominally identical samples. Room temperature operation is a real advantage, but reproducibility, not temperature, is what blocks products.
2 areas
Speculative
Striking physics with no clear path to a product yet.
Neuromorphic computing
Memory that computes, built on components that behave like synapses
Chips that process data where it is stored instead of shuttling it back and forth, which is where most of the energy in AI hardware currently goes.
Where it stands today
Individual devices and small arrays in laboratories. Competing technologies based on metal oxides are years ahead and already in limited commercial production, so 2D versions must beat a moving target.
Read moreShow less
How it works
In a conventional computer, memory and processing are separate and moving data between them costs far more energy than the arithmetic. A memristor is a device whose electrical resistance depends on the current that has flowed through it before, so it remembers, and an array of them can perform the multiply-and-add operations at the heart of neural networks directly in memory. 2D materials make interesting memristors because switching can happen within a layer only atoms thick, which promises low switching energy and dense stacking. In some devices, switching has been traced to a single atom-sized defect.
Real example
No confirmed commercial example yet.
The catchAnalogue in-memory computing has a general problem that has nothing to do with 2D materials: device-to-device variation and drift make the arithmetic imprecise. Choosing an exotic material adds variation rather than removing it.
Spintronics and valleytronics
Beautiful physics, and no product yet
In principle, information carried by an electron’s magnetic orientation or by which of two energy it occupies, rather than by counting charges, potentially with far lower energy per operation.
Where it stands today
Fundamental research. Spin and valley information in these materials survives for microseconds at best and often only at cryogenic temperatures, and no device architecture has shown a clear advantage over conventional electronics.
Read moreShow less
How it works
Electrons carry a property called spin, which behaves like a tiny compass needle that can point up or down. If you could store and move information as spin directions instead of pushing charge around, you would not have to charge and discharge wires, which is what burns energy in a normal chip. In some 2D semiconductors the electrons also have a second label, called a valley, that can be set with left- or right-circular polarised light. In 2017 magnetism itself was found to survive down to a single atomic layer, which made 2D magnetic components conceivable at all.
Real example
No confirmed commercial example yet.
The catch in bulk materials already succeeded commercially, in the read heads of hard drives, and that took about a decade from discovery to product. The 2D version is nowhere near that stage, and press coverage often blurs the two.
The materials behind them
Which material families the 18 application areas above rely on, and how soon each use is expected in products.
Shipping nowIn 2–5 years5–15 years awaySpeculative
Each bar is a material family and each segment an application area on this page that relies on it, shaded by when that application is expected in products; select a segment to go to the application. A segment is shaded by the area’s horizon, not the material’s: 4 families are named in areas that already ship, but what those areas sell today is made from graphene only. The count is of application areas, not of products or papers.
Why the future needs them
7 pressures in today’s technology that 2D materials could help relieve.
Which application areas each pressure pulls forward, as read from its text below. The other 11 of the 18 areas on this page are not named by any of the 7.
01
Making transistors smaller has stopped being easy
For fifty years, chips improved mostly because each generation of transistors was smaller than the last.
Read moreShow less
That is now over as a simple exercise. Below a few nanometres of thickness, the silicon channel that carries current starts to conduct badly, because its own surfaces scatter the electrons. Chipmakers have responded by changing shape rather than size, wrapping the gate around the channel on three sides and then on all four, but each change buys less than the last. A 2D semiconductor is naturally under a nanometre thick with atomically smooth faces, so it does not suffer that particular collapse. This is the one argument for 2D materials that the semiconductor industry itself makes, in its own long-range roadmap, and it is why companies with no interest in wonder materials fund this work.
Data centres consumed on the order of 400 terawatt-hours of electricity in 2024, roughly one and a half percent of world demand, and the International Energy Agency projects that roughly doubling by 2030 as AI workloads grow.
Read moreShow less
Most of that energy does not go into arithmetic. It goes into moving data between memory and processor, and into removing the resulting heat. That makes two 2D properties directly relevant: extremely good in-plane heat spreading, and the possibility of putting logic and memory in the same physical place. Neither is a solution on its own, and efficiency gains have historically been eaten by increased use. But the pressure is real, it is measured in gigawatts, and it is what makes chipmakers willing to consider a material that is not silicon.
The raw materials question is now industrial policy
Graphite is the largest ingredient by mass in a lithium-ion battery anode, and China dominates the mining and especially the processing of it.
Read moreShow less
Both the European Union and the United States have classified graphite as a , and China introduced export licence controls on graphite items at the end of 2023. The EU’s Critical Raw Materials Act, adopted in 2024, sets targets for extracting, processing and recycling more critical materials within Europe and for reducing reliance on any single supplier. This matters for 2D materials in two directions. It raises the value of getting more performance from less material, and it forces a closer look at the 2D materials themselves: tungsten and boron are also on the EU critical list, so a 2D material is not automatically a supply-chain escape route.
Electrification needs batteries that survive being charged fast
Electric vehicle sales have moved from a niche to a substantial share of the global car market, and annual battery demand has grown with them towards a terawatt-hour.
Read moreShow less
Alongside price, the persistent consumer objections include charging time and cold-weather performance. Both are limited by how quickly ions and electrons can move inside an electrode without tearing it apart. This is unglamorous territory where a conductive additive that is thin, flat and light helps, and where graphene is already used commercially. It will not double anyone’s range. It is a few percent here and a few percent there, on a product manufactured by the hundred million, which is a perfectly respectable way for a material to matter.
Industrial equipment, farmland, packaging and human bodies are all being instrumented.
Read moreShow less
What those settings demand is not maximum sensitivity but sensors that are small, cheap, low power, flexible and able to survive their environment. A material that is entirely surface is an inherently good transducer, because whatever you are trying to detect touches the whole thing rather than the outside of a lump. This is the application family where 2D materials have most convincingly crossed into products, in magnetic field sensing, and where the near-term commercial case is strongest. The obstacle is not physics but stability and calibration, which is a slower and less exciting problem to solve.
Quantum devices lose their quantum behaviour when the surrounding material jostles them, through stray electric fields, nuclear spins or vibrations.
Read moreShow less
Progress therefore depends less on clever design than on cleaner materials. Hexagonal boron nitride has become the standard insulator for high-quality 2D devices for precisely this reason: it is flat, chemically inert, and provides a much quieter environment than a conventional oxide. Certain defects inside it also emit single photons at room temperature, which is useful for quantum communication. None of this has produced a commercial quantum device, and 2D materials are currently a supporting material in quantum technology rather than the platform anyone expects to win.
A modern AI accelerator can draw around a kilowatt, concentrated on a piece of silicon a few centimetres across.
Read moreShow less
That is roughly half the power of an electric kettle, concentrated on an area not much bigger than a postage stamp, and it is the reason data centres are switching from air to liquid cooling. Stacking chips vertically, which is the main remaining route to more performance per package, makes it worse by burying heat sources under other heat sources. Materials that move heat sideways very efficiently at very small thickness are therefore no longer a nice extra. This is the least glamorous 2D application and, measured in units already shipped, one of the most successful.
The 8 hurdles below, placed where each one bites on the way from a laboratory result to a product. Of these, 6 bite before a factory is involved at all. Select one for why it is hard and why it is still unsolved.
Making electrical contact to something one atom thick
To use a 2D material you must get current into it from a metal wire. But a metal deposited onto an atomically thin sheet does not simply touch it; the two materials interact, the metal can damage the layer as it condenses, and an energy barrier forms at the junction that resists current flow.
Read moreShow less
The resulting contact resistance is often larger than the resistance of the device itself, which means the measurement tells you more about the contacts than about the material.
Why it is still unsolvedThere is no bulk to hide the junction in. Every approach that improves the electrical quality, such as semi-metal contacts made with bismuth or antimony, or transferring pre-made metal electrodes rather than evaporating them, is either incompatible with high-volume manufacturing or works for one material and not the next. Industry roadmaps repeatedly name contact resistance as one of the largest obstacles to 2D transistors.
Defects and grain boundaries
A film grown over a large area does not form as one crystal. It nucleates in many places at once and those patches grow until they collide, leaving boundaries where the atomic rows do not line up.
Read moreShow less
Electrons scatter at these boundaries, and in a semiconductor a single missing sulfur atom can trap a charge and shift how the device behaves. A material that is entirely surface has nowhere to put a defect where it does not matter.
Why it is still unsolvedGrowing fewer, larger grains means growing more slowly, which conflicts directly with the economics of making anything cheaply. in particular form during growth and can be partly patched afterwards, but no repair method is complete and repairs tend to introduce their own contamination.
Making the same thing across a whole wafer
A chip factory works on 300 millimetre wafers and needs every device on that wafer to behave nearly identically.
Read moreShow less
Growth of a 2D layer is exquisitely sensitive to temperature, gas flow and the exact surface it lands on, so thickness, orientation and drift across the wafer. Research groups report the best flake they measured; a factory has to care about the worst one.
Why it is still unsolvedThe film is one layer thick, so there is no averaging. In a conventional semiconductor process a few nanometres of variation in a hundred-nanometre film is tolerable; here the entire film is less than a nanometre, and a single extra layer in one region is a hundred percent thickness error. Wafer-scale growth of has been demonstrated but uniformity remains well short of production requirements.
Air, water and time
Several of the most promising 2D materials react with their surroundings.
Read moreShow less
Black phosphorus degrades within hours in ambient air. MXene films lose conductivity over weeks to months as they oxidise, faster when humid. Even graphene, which is chemically robust, is contaminated by whatever it has touched, and a monolayer of adsorbed water changes its electrical behaviour measurably.
Why it is still unsolvedThe usual fix is , sealing the material between protective layers, which works well in the laboratory using hand-stacked boron nitride. Doing it at production scale, cheaply, without trapping contamination at the interface, and while still allowing electrical and chemical access to the layer for sensors, is an unsolved engineering problem rather than a scientific one.
Cost, and what you are actually paying for
Graphene spans an enormous price range. Bulk nanoplatelets for composites sell for tens to low hundreds of dollars per kilogram.
Read moreShow less
Electronics-grade film on a wafer costs orders of magnitude more per unit area, and the highest-quality boron nitride crystals, still grown mainly by a small number of specialist laboratories, are effectively priced by the crystal.
Why it is still unsolvedThe two ends of that range are different products entirely, and improving the cheap one does not help the expensive one. Cheap material is not clean enough for electronics; clean material cannot be made in tonnes. There is no single production route that scales across the gap, and most business plans have foundered in the middle of it.
Fitting into a chip factory that already works
A semiconductor fab is one of the most controlled and expensive environments humans build, and it is built around silicon.
Read moreShow less
A new material must survive the temperatures of subsequent processing steps, must not contaminate shared equipment, and must be depositable on the wafer rather than grown elsewhere and transferred by hand. Best-quality 2D layers are usually grown at temperatures above what finished circuitry can tolerate, then moved onto the target wafer in a wet transfer process that tears, wrinkles and dirties them.
Why it is still unsolvedTransfer is a laboratory technique with no volume-manufacturing equivalent yet – wafer-to-wafer transfer by is still being developed – and direct growth at fab-compatible temperatures gives worse material. Meanwhile a fab represents tens of billions of dollars of investment, so the burden of proof on any new material is enormous: it must be better by a wide margin, not a small one, before anyone will disturb a working process.
Measuring what you actually made
Characterising a material one atom thick is difficult. The standard quick check, , tells you a lot about graphene but is easy to over-interpret, and on a powder it cannot reliably tell genuine few-layer graphene from finely ground graphite.
Read moreShow less
Thickness, layer count, defect density, doping and all shift the same measurements, so several different physical situations can produce similar-looking data.
Why it is still unsolvedDefinitive answers require on small areas, which is slow, expensive and destructive, and cannot be applied to every batch. The gap between what is quick to measure and what actually matters is why suppliers could sell graphite as graphene for years without being caught, and why international standards for characterisation only arrived in the 2020s.
Results that do not reproduce between laboratories
Two groups following the same published recipe frequently get different results, because the outcome depends on details that never make it into the methods section: the exact tape, the humidity that day, the previous history of the growth furnace, how long the sample sat in air before measurement.
Read moreShow less
In device work, contact metal deposition rate and cleaning can matter more than the material itself.
Why it is still unsolvedThe field’s sensitivity to its own environment is not a discipline problem but a consequence of everything being surface. Reporting conventions make it worse: papers show the best device rather than the distribution, so a reader cannot tell whether a result came from one flake in fifty or fifty out of fifty. Until yield statistics are reported as a matter of routine, reproducibility will remain the field’s quiet structural weakness.