Quantum Materials

Quantum materials are solids whose useful behaviour comes from electrons acting collectively — through correlation, coherence or topology — rather than from the independent-electron band picture that explains silicon. The category covers three families that share almost no chemistry: topological matter, where a bulk invariant forces protected conducting states at the boundary; moiré systems, where stacking two atomic layers at a small relative twist creates a long-wavelength superlattice with nearly flat electronic bands; and strongly correlated systems, where interactions dominate and the theory is still unsettled after four decades.

Established The physics is real, reproducible, and almost entirely cold. Nearly every flagship effect in this brief is measured below one kelvin, several below one hundred millikelvin. That is not a detail to be engineered away later; it is the central fact that determines which of these materials can ever become a product and which will remain instruments for discovering physics.

Established Exactly one quantum material is load-bearing in world infrastructure, and it is not in a computer. Since the 2019 redefinition of the SI, the ohm and the ampere are realised from fixed values of the elementary charge and the Planck constant, and the practical resistance standard is the quantized Hall resistance — the von Klitzing constant, close to 25,813 ohms. Graphene devices now realise it at around 4 K and a few tesla rather than in gallium arsenide near 1.5 K and ten tesla, which means the single deployed quantum material is a two-dimensional one and its application is metrology rather than electronics.

This brief is organised around what actually limits the field: reproducible synthesis, disorder, device yield, cryogenics, and the absence of a function anyone would buy. It owns the materials question. The economics of superconducting wire belong to high temperature superconductors and the energy cost of running cold machines belongs to ultra efficient computing energy systems.

It also carries, as a section of its own, the field’s two recent integrity failures — the LK-99 episode and the retracted room-temperature superconductivity claims from Ranga Dias’s group. They belong here not as scandal but as the clearest available evidence of how fast this field corrects itself, and of exactly where that correction is slow.

Established A note on sourcing. This brief was commissioned in September 2026 from the Institute’s research base. Reading-list entries without links are cited from the bibliographic record rather than re-fetched, and claims are dated no later than early 2026 unless carried by a linked source.

Established Topological surface states are spectroscopically confirmed and, in transport, a long disappointment. Angle-resolved photoemission has seen the single Dirac cone on the surface of bismuth selenide and bismuth telluride since 2008 and 2009, and the observation is not contested. The difficulty is that the bulk of a real crystal is never insulating: selenium vacancies and other native defects dope it into a conductor, so the protected surface channel is measured against a parallel bulk channel that usually dominates. Twenty years after the prediction, the quantized transport signature remains a cryogenic laboratory result rather than a device property.

Established The quantum anomalous Hall effect works, at temperatures that make the point. It was first measured in 2013 in magnetically doped bismuth antimony telluride at around 30 millikelvin, and raised in 2020 to roughly 1 kelvin in the intrinsic magnetic topological insulator manganese bismuth telluride. A dissipationless quantized edge channel with no external magnetic field is exactly the phenomenon a low-power interconnect would want. Its operating temperature has improved by a factor of about thirty in a decade and needs a further factor of three hundred.

Established Moiré systems are the field’s most reproducible surprise. Twisted bilayer graphene near 1.1 degrees showed correlated insulating states and superconductivity below about 1.7 K in 2018, and has since been replicated in laboratories worldwide; twisted trilayer raised the transition to roughly 3 K. In 2023 three groups independently reported the fractional quantum anomalous Hall effect in twisted molybdenum ditelluride — fractionally charged states at zero external magnetic field, the condition previously thought to require a large field. That is a genuinely new phase of matter, found by stacking two flakes at an angle, and it was replicated within months.

Frontier Altermagnetism is a new magnetic class whose poster material may not be a member. The proposal, formalised in 2022, is a third category alongside ferromagnets and antiferromagnets: collinear magnetic order with zero net moment but a spin splitting that depends on momentum, which would give antiferromagnet-like robustness with ferromagnet-like spintronic functionality. Photoemission has confirmed the predicted spin-split bands in manganese telluride and chromium antimonide. Ruthenium dioxide, the material behind a large fraction of the early experimental literature, is contested: muon spin rotation and nuclear magnetic resonance measurements have found no long-range magnetic order in it, which if correct removes the premise of the transport papers built on it. The class is probably real; a substantial part of its evidence base is disputed.

Frontier The Majorana programme is the largest gap between claim and independently confirmed result anywhere in this field. A 2018 Nature paper reporting quantized Majorana conductance in a semiconductor-superconductor nanowire was retracted in 2021 after outside physicists obtained the underlying data and showed that the published traces had been selected. In February 2025 a Nature paper from the same industrial group reported interferometric single-shot parity measurement in indium arsenide and aluminium hybrid devices, alongside an announced eight-qubit processor. The paper was published with an editorial note, and the referee reports released with it stated that the data presented do not establish the presence of Majorana zero modes; named physicists disputed the screening protocol publicly at the March 2025 American Physical Society meeting. No group outside the company has reproduced the measurement in a device it fabricated itself. The honest statement is that a well-funded programme has produced a parity measurement whose topological interpretation is unconfirmed.

Established The strongly correlated family produced a genuinely new superconductor, and it replicated. Cuprate superconductivity has resisted a settled theory since 1986. The nickelates are the first new high-transition-temperature family since: ambient-pressure onset above 40 K in bilayer nickelate films was reported in 2025, and bulk superconductivity above 90 K in samarium-doped bilayer nickelates followed in the same year, with review literature consolidating the picture in 2026. Separately the compressed hydrides remain the highest-temperature superconductors ever measured — hydrogen sulphide near 200 K and lanthanum hydride near 250 K, at pressures of one to two million atmospheres — and a direct measurement of the superconducting gap in hydrogen sulphide in December 2025 confirmed the mechanism rather than merely the resistive transition.

Established The limit on quantum computing hardware is a materials limit, and the field has been quietly winning there. Superconducting qubit coherence is set by two-level-system defects at metal surfaces, substrate interfaces and native oxides. Replacing niobium films with tantalum, and improving surface preparation, moved relaxation times from tens of microseconds into the hundreds. The 2024 demonstration that a surface code below threshold suppresses errors exponentially with code distance rests on that materials work, and alternative architectures using bias-preserving bosonic qubits are explicit attempts to trade materials quality for hardware efficiency. This is the clearest case in the brief of quantum materials research delivering a measurable engineering result.

Established Two-dimensional materials have a twenty-year device record and no logic product. Graphene was isolated in 2004 and recognised with a Nobel Prize in 2010; a European flagship programme spent on the order of a billion euros over ten years to 2023. The commercial outcome is real and unglamorous: composites, coatings, thermal films, sensors. No transistor built from a two-dimensional channel is in any commercial logic product, and industry roadmaps place such channels in the 2030s. The most concrete recent demonstration is a 2025 report of a RISC-V microprocessor built from a few thousand molybdenum disulphide transistors running at kilohertz clock speeds — a genuine achievement that is roughly nine orders of magnitude behind silicon in device count.

Established The field’s most striking supply fact is that its standard dielectric comes from one laboratory. Essentially all of the high-quality hexagonal boron nitride used to encapsulate two-dimensional devices worldwide is grown by a single small group in Japan using a high-pressure, high-temperature method, and that group is consequently listed as co-authors on thousands of papers. A discipline whose flagship results all depend on crystals from one source carries a reproducibility risk that has never been measured, because there is no second source to compare against.

Frontier Are the industrial topological devices topological? This is the field’s decisive open question and the one with the most capital behind it. The measurement reported is a parity readout; the claim attached is a topological qubit. The gap between those is a demonstration that the degeneracy is protected and that exchange is non-Abelian, which no published experiment has shown. Both outcomes are informative and the field has no mechanism to reach either without a second fabrication line.

Frontier Can moiré fractional states be braided? Fractional Chern insulators at zero magnetic field host anyons in principle, and the device geometry is far more flexible than a fractional quantum Hall bar in a large magnet. Whether interferometry of sufficient coherence can be built in a stacked flake, at the required temperatures, is unresolved. If it can, the shortest route to demonstrated non-Abelian statistics runs through moiré systems rather than through nanowires.

Frontier Do the nickelates have room to run? Bulk transition temperatures above 90 K at ambient pressure make this the first credible challenger to the cuprates, and the open questions are whether the mechanism is the same, whether the onset-to-zero-resistance gap closes in bulk samples, and whether any of it can be made as a conductor rather than as a crystal. The last question is answered in high temperature superconductors, and the answer there is that nothing on that path is near a wire.

Established Device yield is the binding constraint and the literature does not report it. A magic-angle device is assembled by hand: flakes are exfoliated, identified, picked up with a polymer stamp, rotated to a target angle, and encapsulated. Angle relaxation during assembly, bubbles at interfaces and strain all vary between attempts, and a working device is selected from several that are not. Published papers describe the device that worked. The absence of yield statistics is not a minor reporting gap; it is the reason nobody can say whether a negative replication is a failed experiment or a failed fabrication.

Established Cryogenics is a hard power floor, not an inconvenience. A dilution refrigerator delivers on the order of tens of microwatts of cooling at 20 millikelvin and draws kilowatts from the wall to do it, almost all of it in the pulse-tube compressors that pre-cool the system. The efficiency implications for computing are worked out in ultra efficient computing energy systems; the materials implication is narrower and sharper. Any quantum-material function that requires millikelvin temperatures must beat its classical competitor by enough to pay a fixed overhead measured in kilowatts, whatever the chip does.

Frontier The helium-3 that makes those temperatures possible is a weapons by-product. It is produced almost entirely by tritium decay in national stockpiles, the global supply is allocated administratively, and demand from quantum research and from neutron detection has repeatedly exceeded it. Closed-cycle designs recover the inventory of a running machine, which bounds the exposure but does not remove it, and no substitute route to sub-100 millikelvin cooling at comparable cost exists.

Established Wafer-scale growth has not reached device-grade quality. Chemical vapour deposition produces continuous two-dimensional films over wafers, with grain boundaries, point defects and substrate interactions that degrade mobility by one to two orders of magnitude against exfoliated flakes. Every flagship result in this brief was measured on a flake. Every product would have to be made on a wafer. Nobody has closed that gap for any correlated or topological phase.

Established The field waits on crystal growth more than on theory. The single most valuable unfunded programme in quantum materials would be a second and third independent source of device-grade hexagonal boron nitride, together with published cross-comparisons of devices built from each. This is not a discovery; it is a facility, and it would immediately convert an unmeasured systematic into a measured one.

Established It waits on automated assembly. Robotic exfoliation, optical identification and stacking systems have been demonstrated and are not yet standard. The argument for them is not labour cost but statistics: an automated line that makes a hundred nominally identical devices makes yield and disorder measurable for the first time.

Frontier It waits on characterisation that scales. Local probes that map twist angle, strain and carrier density across a device exist and are slow. Coupling them to automated synthesis in a closed loop, of the kind described in artificial scientists, is the plausible route from one device per week to a designed materials search, and no group has demonstrated the full loop on a correlated system.

This section ranks the tests that would most change the brief’s assessment, most decisive first.

Frontier The decisive experiment is an independent laboratory, with no commercial stake in the outcome, reproducing the topological parity measurement in a device it fabricated itself. Confirmation would establish that a protected qubit degree of freedom exists in a real material and would justify the largest single bet in quantum materials. Failure would show that a decade of nanowire results describe a non-topological state, which is also worth knowing and is the outcome several named physicists expect. Nobody has announced funding for such a replication, and it would require a second fabrication line of comparable sophistication, which is the reason this dispute has lasted years while a room-temperature superconductivity claim was settled in weeks.

Frontier Second: an interference experiment demonstrating non-Abelian exchange in a moiré fractional Chern insulator. The states exist at zero magnetic field, were replicated by three groups within months, and are made in laboratories that do not need an industrial fabrication line. If non-Abelian statistics are demonstrated anywhere in the next decade, this brief expects it here rather than in nanowires.

Established Third, and requiring no new physics: a published yield study. One laboratory fabricating fifty magic-angle devices under fixed protocol and reporting the full distribution of twist angle, correlated-gap size and superconducting transition — including the failures — would tell the field more about the reliability of its own literature than any new phase. The experiment is cheap, unglamorous, and nobody has been funded to do it.

Frontier Fourth: settling whether ruthenium dioxide orders magnetically. A third independent technique on well-characterised single crystals, with the sample provenance published, would either restore or invalidate a substantial experimental literature. This is a measurement, not a programme.

Established Fifth, already running: the nickelate replication race is a natural experiment on how fast this field converges. A new superconducting family reported in 2025 has been pursued by many groups at once, in the open, with ambient-pressure results consolidated in review literature within a year. That base rate is the correct benchmark against which to judge any claim that the field is slow to check itself.

Frontier Every engineering requirement in this field is a variance requirement. The performance targets are already met in the best devices; what is missing is the ability to make the best device twice. That reframes the engineering programme away from better materials and toward process control, metrology and statistics — the ordinary content of semiconductor manufacturing, applied to a research practice that currently resembles glassblowing.

Established Transfer is the process step that must industrialise. Moving a monolayer from its growth substrate to its device substrate without tearing it, trapping contaminants or shifting its alignment is the step that breaks at wafer scale, and it has no equivalent in silicon processing, so there is no inherited toolset. Damage-free transfer at wafer scale is the single capability that would move two-dimensional electronics from demonstration to pilot line.

Speculative Room-temperature topological electronics remains an engineering fiction. The energy scales that protect topological states in known materials correspond to temperatures far below ambient, and no candidate material has been identified whose gap would survive room temperature with the transport signature intact. This is a statement about the materials that exist, not a prohibition, and the honest version is that the search has not produced a candidate in twenty years.

Ultra efficient computing energy systems is the nearest neighbour and owns the question this brief deliberately stops short of: whether any cold computing substrate can beat warm silicon once the refrigerator is counted. Read together, the two make an argument neither makes alone — that the materials are not the limiting factor and the thermodynamics of keeping them cold is.

High temperature superconductors is the sibling materials brief and the division is by unit of account: this brief owns the physics of new superconducting families, that one owns what a conductor costs per kilo-ampere-metre and why nobody is winding nickelates. Superconducting infrastructure owns what has actually been installed.

Fault-tolerant quantum computing is adjacent in the strongest sense and is not on this map as a link target from here; the relation worth recording is that its error rates are set by interface chemistry, so progress in that field is partly a quantum-materials result reported under another name. Artificial scientists is adjacent as the plausible answer to this field’s throughput problem.

Established The most consequential results in this field are now produced inside companies, and the verification machinery is not built for that. An industrial laboratory can fund a fabrication line no university can match, which is why the topological-qubit claim exists at all; it also means the only group capable of checking the claim is the group making it. Peer review can assess whether an analysis supports a conclusion. It cannot supply a second device.

Established Publishing referee reports alongside a disputed paper is a genuine institutional innovation and it worked. Readers of the February 2025 topological-qubit paper could see, in the journal’s own record, that the referees did not accept the strongest reading of the data. That is a better outcome than either silent acceptance or rejection, and it converted a dispute that would once have played out in corridors into a documented disagreement any reader can evaluate.

Speculative National quantum programmes have funded devices rather than materials, and may be correcting. The large public initiatives launched from 2018 onward allocated most money to computing, communication and sensing demonstrations. The materials base underneath them — crystal growth, characterisation facilities, yield metrology — is the part with the weakest constituency, and shifting even a small fraction of that funding would be the highest-leverage policy change available in this field.

Established This field ran two public integrity tests within three years, and it passed both — at very different speeds. They are treated here as substance rather than as scandal, because how fast a field detects a wrong claim is a measurable property of that field and one of the few available.

Established LK-99 was corrected in about three weeks. In late July 2023 a Korean group posted preprints claiming ambient-pressure, room-temperature superconductivity in a lead apatite compound, with a video of partial levitation. Because the synthesis was cheap and the recipe public, laboratories on several continents attempted it within days. The levitation was shown to be diamagnetic and ferromagnetic response in an inhomogeneous sample rather than the Meissner effect; the resistivity drop near 104 degrees Celsius was traced to a first-order structural transition in a copper sulphide impurity. By mid-August 2023 the replication verdict was reported as settled, and a national verification committee reached the same conclusion formally by the end of that year. Nothing in that sequence required an institution to act. It required a published recipe and a large number of people willing to spend a day on it.

Established The Dias case took four years and needed the journals to act against the authors. A 2020 Nature paper claiming room-temperature superconductivity in carbonaceous sulphur hydride was retracted in 2022 over the objection of its lead author, after outside physicists obtained the underlying magnetic susceptibility data and showed that the published background subtraction was not reproducible from it. A Physical Review Letters paper was retracted the same year following an investigation commissioned by the journal’s editors, and a second Nature paper, claiming near-ambient superconductivity in nitrogen-doped lutetium hydride, was retracted in November 2023 at the request of co-authors. The University of Rochester cleared the researcher in earlier inquiries and only its fourth investigation, conducted with outside experts, found data fabrication and falsification; his employment ended, as reported in late 2024.

Established The correcting was done by individuals, then journals, then finally institutions — in that order, and the order is the finding. Named physicists who requested raw data and re-analysed it moved first and at their own cost. Journals moved second and, in one case, hired an external analyst rather than relying on the authors. The employing university moved last and only after three inadequate inquiries. Any reform proposal that starts with institutions is optimising the slowest link in the chain.

Established What survived the retractions is the most important part of the story. High-pressure hydride superconductivity was not the fraud. Hydrogen sulphide near 200 K and lanthanum hydride near 250 K were reported by other groups, have been reproduced, and were strengthened in December 2025 by a direct measurement of the superconducting gap rather than of resistance alone. Error correction removed the false claims and left the true ones standing, which is exactly what the machinery is for and is the opposite of the lesson usually drawn.

Frontier The machinery works where replication is cheap and fails where it is expensive. LK-99 took weeks because a thousand laboratories could try it. The hydride claims took years because diamond-anvil measurements are hard and few groups compete. The topological-qubit question has run for the better part of a decade because almost nobody on Earth can make the device. That is a structural prediction, not a moral one: this field’s error correction is fast in inverse proportion to the capital cost of the experiment, and it is slowest exactly where the capital is concentrated.

Frontier The costs were borne unevenly. Junior co-authors who did honest work carry retracted papers on their records; a research direction absorbed reputational damage it did not earn; and a public that was told twice in three years that room-temperature superconductivity had arrived is entitled to some scepticism next time. The appropriate institutional response is mandatory deposition of raw measurement data and independent statistical screening of headline claims, both of which would have shortened the second case substantially and neither of which is standard.

Established The confirmed civilizational contribution of quantum materials to date is a unit of measurement. The quantized Hall resistance underpins the realisation of the ohm and, through it, the calibration chain for essentially all electrical measurement. That is a real and permanent contribution and it is much smaller than the rhetoric surrounding the field.

Speculative The large claim — that topological protection makes robust room-temperature electronics possible — has no candidate material. It is coherent physics with no identified realisation after twenty years of searching, and it should be held as a hypothesis about materials that might exist rather than as a technology roadmap.

Handwave Claims that a room-temperature superconductor would transform civilisation overstate the case in a specific and checkable way. The material has been adequate for fifteen years at liquid-nitrogen temperatures and the grid has installed on the order of ten kilometres of cable, for reasons of cost, standards and institutional risk that superconducting infrastructure documents in detail. Temperature is not the binding constraint it is assumed to be.

These horizons track the two variables this brief argues are decisive — reproducibility of synthesis and the operating temperature of useful effects — rather than the discovery rate, which is high and largely uninformative about applications.

• 10 yr: Frontier The topological-qubit question is resolved one way or the other, most likely by an independent fabrication line rather than by more data from the original one. Automated stacking makes device yield a published quantity. Nickelate superconductivity is either understood or has stalled at the crystal stage. Expect no two-dimensional logic product.
• 25 yr: Speculative Wafer-scale two-dimensional channels appear in niche commercial devices — sensors, photonics, possibly memory selectors — rather than in mainstream logic. Quantum anomalous Hall operation reaches liquid-nitrogen temperatures or is abandoned as a device concept. Quantum materials remain a supplier of fabrication quality to quantum computing rather than a product category.
• 50 yr: Speculative Either a designed-materials pipeline — closed-loop synthesis, characterisation and theory — makes new correlated phases findable on demand, in which case the field’s rate changes qualitatively, or it remains a discovery science dependent on skilled hands and rare crystals. Nothing in the present record settles which.
• 100 / 250+ yr: Handwave Room-temperature topological electronics and ambient-pressure room-temperature superconductivity are coherent hypotheses with no candidate material, no theoretical route at ambient pressure, and therefore no forecastable date. This brief declines to supply one, and notes that the two most confident public predictions of an imminent room-temperature superconductor in the last five years were both wrong within weeks.

• Depends on Nothing on this map blocks the physics. The results in this brief are being produced now, in quantity, and the blockage is reproducibility and temperature rather than a pending discovery. The nearest dependency is the cryogenic energy accounting in ultra efficient computing energy systems, which determines whether any millikelvin function can pay for itself.
• Requires (not on this map) A second and third independent source of device-grade hexagonal boron nitride, so that the field’s largest unmeasured systematic becomes measurable. Damage-free transfer of single-crystal two-dimensional films at wafer scale, the process step with no inherited silicon toolset. A signature of non-Abelian exchange that does not depend on trusting one group’s device screening. Dilution refrigeration with milliwatt-class cooling at the lowest stage, which sets how many channels any cold device can have. A publication norm requiring yield and failure distributions, which costs nothing and would change how the literature reads. And a buyer for a function that only works below one kelvin, which outside metrology and quantum computing does not currently exist.
• Enables Higher-coherence qubits through interface chemistry, which is the input fault-tolerant quantum computing consumes; low-power interconnects if quantized edge transport ever warms; and superconducting families that high temperature superconductors would have to turn into wire before any of it reaches a grid.
• Adjacent Superconducting infrastructure, which documents what actually gets installed; artificial scientists, the plausible answer to a throughput problem measured in devices per week; and scientific revolutions, against which this field’s error-correction record is a useful case study.

Handwave “A room-temperature superconductor was found and suppressed.” The two prominent claims of the 2020s were both examined in the open and both failed for stated, published, reproducible reasons — an impurity phase transition in one case, non-reproducible background subtraction in the other. The raw data in the second case was obtained by outside critics and analysed publicly. Suppression is the one hypothesis the record actively rules out.

Frontier “LK-99 and the retractions prove the field is broken.” They demonstrate the opposite, and the timescales are the evidence: three weeks to a replication verdict on a cheap synthesis, and retraction of three papers including two against author wishes. What the record does show is a genuine weakness, which is that correction speed tracks the cost of replication, so the claims hardest to check are the ones the machinery handles worst.

Frontier “Topological qubits have been built.” A parity measurement has been reported and a processor announced. The journal published referee reports stating the data do not establish Majorana zero modes, no independent fabrication has reproduced it, and the demonstration that would settle it — non-Abelian exchange — has not been performed. This is the brief’s decisive experiment precisely because both outcomes remain open.

Established “Graphene failed.” It failed at the thing it was sold for, namely replacing the silicon transistor, because a gapless semimetal makes a poor switch and the gap-opening strategies cost the mobility that motivated it. It succeeded quietly elsewhere: composites, thermal management, coatings, sensors, and the metrological resistance standard. Judging a material by its most-hyped application is how a useful material acquires a reputation for failure.

Handwave “Moiré superconductivity is high-temperature superconductivity.” Magic-angle graphene superconducts below roughly 1.7 K and twisted tungsten diselenide below a few hundred millikelvin. The interest is that the phase diagram is tunable and resembles the cuprates, which makes these systems excellent models of a hard problem. They are three orders of magnitude colder than the materials already in use.

Frontier “Altermagnets are an established new class of matter.” The theory is clean, the symmetry classification is sound, and photoemission confirms the predicted band structure in at least two materials. The complication is that a large share of the early experimental literature used a material whose magnetic order is itself now disputed. Both halves of that sentence are true and the second is rarely reported.