Wednesday, 18 November 2026
NEXA NEWSHD
Science & Technology

Diamond Maser Amplifier Boosts Faint Quantum Signals 1,000-Fold at Near Absolute Zero

Researchers in Japan have built an amplifier from a diamond that can boost the faintest microwave signals in a quantum computer more than 1,000-fold while adding almost no noise of its own, according to a study published on Monday 16 November 2026 in the journal Nature Communications. The device revives a technology that predates the laser by six years, and it works only at temperatures a fraction of a degree above absolute zero — conditions that were once the maser’s fatal weakness and are now, in the age of quantum computing, exactly the point.

The amplifier was built by a team at the Okinawa Institute of Science and Technology (OIST), led by Dr Yuimaru Kubo, head of the institute’s Hybrid Quantum Device Team, with first author Morihiro Ohta of OIST’s Experimental Quantum Information Physics Unit. At its heart sits a small pink diamond, held inside a copper microwave resonator and cooled to 13 millikelvin — thirteen thousandths of a degree above absolute zero — inside a dilution refrigerator. The diamond contains nitrogen impurities known as P1 centres, and the electron spins of those impurities do the amplifying. When the team pumped the spins with a weak microwave signal, the device produced a power gain of more than 30 decibels, which is a thousand-fold increase, while adding only 2.55 quanta of noise. The lowest noise physically permitted by quantum mechanics is 0.5 quanta, so the device comes strikingly close to the absolute limit.

“Despite being invented before optical lasers, masers — microwave lasers — have been left far behind because they often require cryogenic conditions to function, which made them impractical until very recently,” Kubo said, according to the institute’s account of the work. “But now, with quantum computers operating at temperatures very close to absolute zero, we have given them new life as very promising candidates for amplifying quantum signals with very little noise.”

The result matters because reading a quantum computer is one of the hardest parts of running one. The signals that carry information about a qubit’s state are vanishingly weak microwaves, and every amplifier in the chain adds its own hiss. Too much added noise, and the measurement is wrong. Too little gain, and the signal is lost. Today’s best amplifiers are superb at keeping noise down, but they have weaknesses of their own: they saturate when asked to read many qubits at once, they are knocked out by magnetic fields, and they need complicated wiring and power. The OIST device addresses all three problems at once, which is why the result has drawn attention well beyond the specialist community — even though, as the researchers themselves acknowledge, the device is still a laboratory demonstration with a narrow bandwidth and a long road to practical use.

How the diamond maser works

A maser and a laser work on the same principle, stimulated emission. In both, atoms or spins are first pushed into an excited, high-energy state — a condition called population inversion, because more of them sit in the upper state than in the lower one, which is the opposite of the natural arrangement. When a photon of the right energy arrives, it stimulates one of the excited spins to drop to its lower state, releasing a second photon identical to the first. One photon in, two photons out: that is amplification.

In the OIST device, the active material is a diamond crystal containing about 16 parts per million of P1 centres — single nitrogen atoms sitting in place of carbon atoms in the diamond lattice, each carrying an unpaired electron with a spin of one half. The diamond sits inside a loop-gap microwave resonator, a copper structure tuned to a frequency of about 6.595 gigahertz, and the whole assembly is cooled to 13 millikelvin on the mixing-chamber plate of a dilution refrigerator.

The team created the population inversion with a microwave pump tone at 6.501 gigahertz, applied at a power of 100 nanowatts or less. That is an extraordinarily small amount of power — the researchers report that they were surprised by how much amplification they achieved with so little pump power. The pumping works through a four-spin cross-relaxation process: the pump excites spins in the central transition, and those spins exchange energy with spins in two neighbouring transitions, so that one set of spins is pushed up while another is pulled down. The trick that makes the inversion last is an imbalance in how fast the two sets of spins relax back to their ground states. One set relaxes in about 1,030 seconds, the other in about 1,650 seconds, and a second, faster-relaxing spin system in the diamond — possibly neutral nitrogen-vacancy centres or nickel-related centres, though the team says the microscopic origin remains uncertain — drains the lower set still faster. Because the lower state keeps emptying, the upper state keeps accumulating, and the inversion is sustained.

Once the inversion is established, an incoming microwave signal at the resonator frequency stimulates the excited spins to release their energy, and the signal emerges amplified. The researchers measured the gain with a vector network analyser and characterised the noise with a calibrated, temperature-variable noise source. Above a pump power of minus 58 decibel-milliwatts at the device input, the inverted spin population became positive — the signature of population inversion — and at a pump power of minus 39 decibel-milliwatts the gain exceeded 30 decibels.

The device is also tunable. By varying the external static magnetic field — about 232 millitesla, aligned with the diamond’s cubic crystal axis — the team could shift the gain profile by about 10 megahertz. That matters because a useful amplifier must be able to match the frequency of the qubit it is listening to, and different qubits operate at different frequencies.

The numbers that matter

The headline figures, as reported in the paper and in the institute’s account, are these. The maximum power gain exceeds 30 decibels, which means the output signal carries more than a thousand times the power of the input. The added noise is approximately 2.55 quanta, against a quantum limit of 0.5 quanta. The minimum noise temperature reported for the device is 0.86 kelvin. The maximum 1-decibel output compression point — the input power at which the gain starts to sag by one decibel, a standard measure of how strong a signal an amplifier can handle before distorting — is minus 63 decibel-milliwatts at 6.595 gigahertz, which is substantially higher than that of a conventional Josephson parametric amplifier. The gain bandwidth is about 70 kilohertz, giving a gain-bandwidth product of about 2.8 megahertz. The pump power needed is 100 nanowatts or less. The operating temperature is 13 millikelvin.

Two of those numbers deserve emphasis, because they are the ones that distinguish this device from the amplifiers quantum computers use today. The first is the noise: 2.55 quanta is close enough to the quantum limit that the maser adds only about three times less noise than the transistor amplifier that follows it in the measurement chain, according to the paper’s measurements. The second is the saturation power. Conventional Josephson parametric amplifiers — the current standard — are easily overwhelmed when they are asked to read several qubits simultaneously, because a strong signal saturates them and the gain collapses. The diamond maser, by contrast, keeps amplifying at input powers that would saturate its superconducting rivals. The paper also notes that the maser works in a magnetic field, which matters because some quantum computing platforms, notably semiconductor spin qubits, require magnetic fields that can disable superconducting amplifiers entirely.

There is one number that cuts the other way, and the researchers do not hide it: the bandwidth. At about 70 kilohertz, the maser’s gain covers a very narrow slice of frequency. A quantum processor that needs to read many qubits, each at its own frequency, would need either many such amplifiers or a much wider one. The paper suggests two routes to a wider device — multi-mode resonator designs, and a travelling-wave geometry in which the signal passes along a long amplifying structure instead of bouncing inside a single cavity — and the team’s Proof of Concept programme at OIST is already working on the travelling-wave version, aiming for a power gain of more than 100 with a bandwidth of about a gigahertz. For now, though, the narrow bandwidth is the device’s clearest limitation.

Why today’s amplifiers hit a wall

To understand why a 70-year-old idea is suddenly interesting again, it helps to understand what reading a quantum computer actually involves. A superconducting qubit stores information in the state of a tiny electrical circuit cooled to a few thousandths of a degree above absolute zero. To find out what state the qubit is in, the machine sends a faint microwave pulse to a resonator coupled to the qubit, and the pulse that comes back carries the answer encoded in its phase. The returning signal is so weak that it must be amplified many times before ordinary electronics can measure it, and the first amplifier in the chain matters most: any noise it adds is amplified along with the signal by everything downstream.

For the past two decades, that first amplifier has almost always been a superconducting parametric amplifier based on Josephson junctions — tiny superconducting devices whose inductance changes with the current flowing through them. These amplifiers can operate at, or very near, the quantum limit of added noise, and they have enabled high-fidelity readout of superconducting qubits, semiconductor qubits, weak spin echoes in magnetic resonance, and even the measurement of mechanical motion in quantum optomechanics. They are also used in particle-physics experiments, including searches for the dark-matter candidate particle known as the axion. They are, by any measure, a triumph of engineering.

But they have three weaknesses that grow more serious as quantum processors grow larger. First, they saturate: when several qubits are read at once, the combined signal can overwhelm them. Second, they are sensitive to magnetic fields, which rules them out for spin-qubit platforms that need a field to operate. Third, the more capable variants — travelling-wave parametric amplifiers and kinetic-inductance amplifiers — need more electrical power and more wiring, which adds heat and complexity inside a refrigerator that must stay colder than outer space. As the paper puts it, balancing low noise, high input-power capacity, simple wiring and magnetic-field compatibility “remains an important challenge for quantum computing”. The diamond maser is an attempt to solve all four at once, with a device that has no superconducting junctions, needs almost no pump power, tolerates a magnetic field, and handles strong signals.

A brief history: from 1954 to a pink diamond in Okinawa

The maser is older than the laser, and older than the transistor radio in most people’s homes. It was first demonstrated in 1954 by Charles Townes and his colleagues at Columbia University, using ammonia molecules, six years before Theodore Maiman demonstrated the first laser in 1960. Townes shared the 1964 Nobel Prize in Physics for the work. Early masers were used as exquisitely precise clocks, as amplifiers in radio astronomy, and as the low-noise front ends of deep-space communications — NASA’s deep-space network relied on masers for decades, because a signal from a spacecraft at the edge of the solar system arrives so faintly that only the quietest possible amplifier can hear it.

The reason masers faded from view is the same reason they are interesting again: they need cold. A maser works by maintaining a population inversion, and at room temperature thermal energy constantly scrambles the spins back toward equilibrium. Optical pumping — using light to push spins into the excited state — revived diamond masers at room temperature in recent years. A team at Imperial College London demonstrated a diamond maser oscillator, and engineers at the University of New South Wales built a room-temperature diamond maser amplifier using a purple diamond, reporting in December 2024 that it could boost weak microwave signals by a factor of up to a thousand, with a commercial device possibly two to three years away. Those devices use nitrogen-vacancy centres, a different kind of defect from the P1 centres in the OIST device, and they are pumped with light rather than microwaves.

The OIST approach is different in two important ways. First, it works at millikelvin temperatures rather than room temperature, which is where quantum computers live. Second, it is pumped with microwaves rather than light. Optical pumping is hard to integrate with microwave quantum devices: it requires precise alignment of the magnetic field along one of the diamond’s crystal axes, and it means bringing light into a refrigerator designed to keep light out. Microwave pumping, by contrast, uses the same kind of signals the quantum computer already handles. The paper describes the result as the first non-superconducting amplifier to reach near-quantum-limited noise at millikelvin temperatures, and notes that the lowest noise temperature previously reported for any maser — 1.43 kelvin, in a ruby maser operated at 1.6 kelvin with about 200 milliwatts of pump power — was achieved with roughly two million times more pump power than the OIST device uses.

The work did not appear from nowhere. A preprint of the paper was posted in May 2025, and the research was supported by OIST’s own Proof of Concept programme, by Japan’s JST Moonshot research and development programme and by JST-PRESTO. The author list spans OIST’s Experimental Quantum Information Physics Unit and its Science and Technology Group, the Quantum Dynamics Unit, and collaborators at Japan’s National Institutes for Quantum Science and Technology, Sumitomo Electric Industries and the University of Tsukuba. The paper was published in Nature Communications on 16 November 2026, and the institute says the device is now undergoing further development toward commercialisation.

What it could be used for

The most immediate application is the one the device was built for: reading out quantum computers. A low-noise, high-saturation amplifier that works in a magnetic field could sit at the front of the measurement chain in a superconducting processor, reading many qubits at once without saturating, or in a semiconductor spin-qubit processor, where the magnetic field would disable a conventional amplifier. As quantum processors scale from hundreds of qubits toward thousands, the readout chain becomes a bottleneck: more qubits means more signals to amplify, and more amplifiers means more heat, more wiring and more noise. A device that amplifies a thousand-fold on 100 nanowatts of pump power, with almost no added noise, is an attractive answer to that problem — provided its bandwidth can be widened.

Beyond computing, the same qualities matter anywhere a faint microwave signal must be heard clearly. In spectroscopy and nuclear magnetic resonance — the technology behind MRI scanners — a quieter first amplifier means weaker signals can be detected, which can mean faster scans or clearer images. In fundamental physics, quantum-limited amplifiers are already used in the search for axions, hypothetical particles that could make up dark matter; an axion converting into a microwave photon inside a magnetic field would produce a signal so faint that only a near-quantum-limited amplifier could hope to hear it, and the magnetic field required for the search is precisely what disables superconducting amplifiers. The paper also notes that masers could eventually serve as alternatives to the transistor amplifiers used at higher temperatures, if their bandwidth can be broadened.

There is also a wider significance that has nothing to do with any single application. The history of technology is full of ideas that were right too early — abandoned because the world they needed did not exist yet, and revived when it finally did. The maser is a clean example. It was invented before the laser, outperformed everything else as a low-noise amplifier, and then spent half a century as a specialist tool, because the world ran at room temperature and the maser did not. Quantum computing changed the temperature of the world. Dilution refrigerators that hold a processor at 13 millikelvin are now standard equipment in quantum laboratories, and in that world the maser’s demand for cold is no longer a drawback. It is, as the paper observes, simply the operating condition.

What the device cannot do yet

Honesty about the limits matters as much as enthusiasm about the result, and the researchers are candid about them. The bandwidth is the main one: about 70 kilohertz is far too narrow to read a whole processor. The gain-bandwidth product of about 2.8 megahertz is a fundamental property of the single-cavity design, and widening it will require a different geometry. The team is working on a travelling-wave version, but it does not exist yet.

The second limit is less visible but just as important: the mechanism that makes the device work is not fully understood. The fast-relaxing spin system that drains the lower state and sustains the inversion is, by the authors’ own account, of uncertain microscopic origin. They suggest it may be neutral nitrogen-vacancy centres or nickel-related centres in the diamond, but they do not know. That does not make the result unreliable — the amplification was measured directly, repeatedly, and the noise was characterised with a calibrated source — but it does mean the device cannot yet be engineered with the confidence that comes from a complete theory. Reproducing it in other diamonds, with other impurity concentrations, may require first understanding what exactly is doing the relaxing.

Third, the device is a single amplifier, not a system. A quantum computer needs a readout chain: the maser, followed by transistor amplifiers, followed by room-temperature electronics, all wired together inside a refrigerator. Integrating a diamond maser into that chain — with its magnetic field, its resonator, its pump tone — is engineering work that has not yet been done. And fourth, the result has been peer-reviewed and published, but it has not yet been independently replicated. The history of quantum devices includes results that proved hard to reproduce, and the test of this one will be whether other laboratories can build the same amplifier and get the same numbers.

None of that diminishes the achievement. A thousand-fold gain with near-quantum-limited noise, on a hundred nanowatts of pump power, in a magnetic field, from a device with no superconducting junctions, is a genuine advance — and it is the first time anyone has done it at millikelvin temperatures without superconductors. But the distance between a laboratory demonstration and a component in a working quantum computer is measured in years, and the bandwidth problem is the distance.

Analysis: Why It Matters

The maser was never wrong; the world was just too warm

The most interesting thing about this result is not the numbers, impressive as they are. It is the shape of the story. The maser was invented in 1954, won a Nobel Prize, and then spent seventy years as a footnote — too cold for a warm world. What changed was not the maser. What changed was the world. Quantum computers run at thirteen thousandths of a degree above absolute zero, and at that temperature the maser’s supposed weakness disappears entirely. This is a pattern worth noticing, because it recurs throughout the history of technology: an idea is judged against the conditions of its own time, shelved, and then revived when the conditions change. The lesson for anyone watching quantum computing is that the winning technologies may not all be new. Some of them are old ideas whose moment has finally arrived, and the laboratories that remember them have an advantage over the ones that only chase the new.

Readout is the quiet crisis in quantum computing

Public discussion of quantum computing is dominated by qubit counts, and that is understandable — the number of qubits is easy to count and easy to compare. But the machines that actually work are limited as much by their readout as by their qubits. Every qubit must be measured, every measurement must be amplified, and every amplifier adds noise, heat and wiring. A processor with a thousand qubits needs a thousand readout chains, or a smaller number of chains that can each handle many qubits at once. The second option is the only one that scales, and it is exactly what today’s amplifiers struggle with: they saturate. The diamond maser’s real promise is not that it is quieter than the best superconducting amplifier — it is roughly comparable — but that it stays quiet while handling signals strong enough to saturate its rivals. If the bandwidth problem can be solved, this is the device that lets one amplifier listen to many qubits at once, and that is a scaling argument, not just a performance argument.

The magnetic field is the detail that matters most

It is easy to overlook, among the gain and noise figures, that this amplifier works in a magnetic field of 232 millitesla. That single fact may matter more than any other number in the paper. Superconducting amplifiers fail in magnetic fields, and several of the most promising routes to a quantum computer — notably semiconductor spin qubits, which encode information in the spin of a single electron — require a magnetic field to operate at all. Until now, those platforms have had to work around the readout problem, accepting noisier amplifiers or elaborate shielding. A near-quantum-limited amplifier that not only tolerates a magnetic field but needs one is a genuine unlock for those platforms. It is also the reason the device could matter for dark-matter searches, which use strong magnetic fields to coax axions into becoming photons. The field compatibility is not a bonus feature. For some users, it is the whole point.

A hundred nanowatts is a statement about scale

Consider the pump power: 100 nanowatts or less. The previous record for a low-noise maser, a ruby device from an earlier era, needed about 200 milliwatts — two million times more. That difference is not a curiosity; it is what makes the device usable inside a dilution refrigerator, where every milliwatt of heat is a problem and the cooling power at 13 millikelvin is measured in microwatts. A quantum processor that needs hundreds of amplifiers cannot afford hundreds of milliwatts of pump power. It can afford hundreds of nanowatts. The low pump power is also what makes the device simple: no optical pumping, no precise alignment of a laser, no light leaking into a dark refrigerator. Microwaves in, microwaves out, and the diamond does the rest. Simplicity is a feature that tends to be undervalued in research announcements and overvalued in engineering, and this device has it.

The honest limits are part of the story, not a footnote to it

It would be easy to present this as a breakthrough that solves quantum readout. It is not that, and the researchers do not claim it is. The bandwidth is 70 kilohertz. The mechanism behind the fast-relaxing spins is unknown. The device has not been replicated independently, and it has not been integrated into a real readout chain. Those are not reasons to dismiss the work — a thousand-fold gain at near the quantum limit is real, measured, and peer-reviewed — but they are reasons to keep the excitement proportionate. The history of quantum technology is littered with laboratory results that never survived the journey to engineering, and the ones that did survive it were the ones whose limitations were stated plainly at the start. The OIST team has stated theirs plainly. The travelling-wave version they are building is the test of whether this idea scales, and the next two years will tell us whether the diamond maser becomes a component or remains a demonstration. Either way, the revival of the maser is a reminder that in science, as in everything else, timing is a technology.

What to Watch Next

  1. The travelling-wave maser: OIST’s Proof of Concept programme is developing a travelling-wave version of the device, aiming for a power gain above 100 with a bandwidth of about a gigahertz. If it works, the bandwidth objection disappears — and the device becomes a realistic candidate for multi-qubit readout.
  2. Independent replication: the test of any quantum-device result is whether another laboratory can build it and get the same numbers. Watch for groups at other institutes reporting diamond maser amplifiers of their own, and for the gain and noise figures they achieve.
  3. Identifying the fast-relaxing spins: the team does not yet know what microscopic defect drains the lower spin state and sustains the inversion. Identifying it — neutral nitrogen-vacancy centres, nickel-related centres, or something else — will determine how reproducibly the device can be made.
  4. Integration into a real readout chain: the next step is putting a diamond maser in front of actual qubits, inside a working dilution refrigerator, and measuring qubit readout fidelity with it in the chain. That experiment has not yet been reported.
  5. Spin-qubit adoption: semiconductor spin-qubit platforms, which need magnetic fields, are the natural first users. Watch for announcements from spin-qubit groups testing maser-based readout.
  6. Commercialisation: OIST says the device is undergoing further development toward commercialisation. The first commercial product, if it comes, will likely be a laboratory amplifier for research rather than a component in a quantum computer.
  7. The room-temperature rival: the University of New South Wales room-temperature diamond maser, reported in December 2024, is on a parallel track. The two approaches — cold and quiet versus warm and convenient — will compete for different applications, and their relative progress is worth following.

Sources

About the Author — Abdul Mannan

Leave a Reply

Your email address will not be published. Required fields are marked *