Two separate research groups — one at the Vienna Center for Quantum Science and Technology in Austria, one at Tsinghua University in Beijing — have built the world’s first working nuclear clocks, publishing their peer-reviewed results this week in Nature. Neither team knew the other would arrive at the same milestone simultaneously, and neither relied on the other’s methods. The fact that they succeeded in parallel — using different laser designs, different crystal preparations, and different experimental focuses — tells physicists that the approach is robust and not a laboratory fluke.

That alone would be enough to call this the most significant advance in precision timekeeping since the first optical atomic clock. What makes this week’s papers remarkable beyond the clocks themselves is what the Vienna team did immediately after building theirs: they ran the first dark matter search ever conducted using a nuclear frequency reference, setting new limits on how dark matter might couple to the strong nuclear force — a domain that conventional atomic clocks, which probe only electromagnetic interactions, cannot reach at all.

What Makes Nuclear Clocks Different from the Atomic Clocks Running Every GPS Satellite

Every smartphone’s GPS chip, every data center’s network synchronization, every financial exchange’s trade-timestamp system depends on an atomic clock somewhere. Atomic clocks have kept time since the 1950s by locking a laser or microwave oscillator to the energy transition of an electron — most commonly in cesium-133, whose hyperfine transition defines the official SI second. The best optical atomic clocks today (using strontium or ytterbium ions) achieve stabilities approaching one second of error in 55 billion years of operation.

Nuclear clocks take the same principle one level deeper. Rather than probing the electrons orbiting a nucleus, they interrogate transitions between quantum energy states inside the nucleus itself — between protons and neutrons. The nucleus is shielded from its environment by the surrounding electron cloud, which means nuclear transitions are far less susceptible to stray electric fields, temperature fluctuations, and other perturbations that limit how precisely any clock can be measured. In principle, a nuclear clock could eventually surpass the best atomic clocks by a significant margin.

The challenge is that nuclear transitions almost always require gamma rays or X-rays — energies far beyond the reach of any laser. Of all 3,300-plus known nuclides in the nuclear chart, exactly one is currently useful for nuclear timekeeping: thorium-229.

Thorium-229: A Cosmic Accident That Took 50 Years to Exploit

Thorium-229 occupies a unique position in the nuclear landscape because of what physicists describe as an accidental cancellation. Every atomic nucleus is a battlefield between two enormous competing forces: the Coulomb repulsion that pushes the positively charged protons apart, and the strong nuclear force that holds protons and neutrons together. In most nuclei, these forces are both immense but unequal by a large margin, producing transition energies in the keV–MeV range — inaccessible to lasers. In thorium-229, these two enormous quantities cancel almost perfectly, leaving a residual transition energy near 8.4 electron volts. That energy sits squarely in the vacuum-ultraviolet range of the electromagnetic spectrum — challenging to work with, but reachable by laboratory laser technology.

This quirk was first noticed in 1976 during gamma-ray spectroscopy of uranium-233’s alpha decay. The nuclear clock concept waited until 2003, when physicists Ekkehard Peik and Christian Tamm at Germany’s Physikalisch-Technische Bundesanstalt (PTB) published the first detailed engineering proposal for a clock based on this transition. The decades between that theoretical proposal and this week’s working instruments were spent painstakingly characterizing the transition — locating its frequency precisely enough to target it with a laser.

The critical last step came in September 2024, when physicist Chuankun Zhang and JILA and NIST Fellow Jun Ye’s group at JILA in Boulder, Colorado, measured the thorium-229 nuclear transition frequency with 12 significant digits using an optical frequency comb linked to a strontium atomic clock. That measurement gave the Vienna and Beijing teams the precise target they needed.

How the Two Clocks Work: Why One Is Already More Stable

Both clocks embed thorium-229 nuclei inside crystals of calcium fluoride (CaF2), a material transparent to the vacuum-ultraviolet (VUV) light needed to drive the nuclear transition. The crystals hold up to roughly a quadrillion (10¹⁵) thorium nuclei simultaneously — a massive signal advantage over single-atom approaches — at room temperature on a tabletop.

The similarity ends there.

The Vienna clock, led by Luca Toscani de Col and senior physicist Thorsten Schumm at TU Wien — with key contributions from colleagues at PTB and Austria’s Federal Office of Metrology (BEV) — implements what the team calls “a thorium-229 optical nuclear clock with feedback loop.” A continuous-wave laser, tuned to the 148-nanometer nuclear transition, continuously interrogates the crystal. When thorium nuclei absorb VUV photons, they shift energy states; a detector monitors the transmitted laser intensity. The feedback loop reads the transmitted signal and continuously corrects the laser frequency to stay locked on the nuclear resonance — with no external frequency reference required in real time. This makes the Vienna clock the first nuclear clock that “operates as a stand-alone device,” in the team’s words.

In a 24-hour comparison against an established ytterbium-ion atomic clock, the Vienna clock showed one-second error per 30 million years of operation — already a remarkable figure, though still roughly 1,800 times less stable than the best atomic clocks.

The Beijing clock, led by Beichen Huang and Shiqian Ding at Tsinghua University, took a different path. The team used a substantially more powerful laser and generated VUV light by heating cadmium vapor to approximately 600 degrees Celsius (about 1,112 degrees Fahrenheit) as part of their frequency conversion chain. Their crystal contained fewer thorium nuclei than Vienna’s, but the stronger laser more than compensated. Initial measurements suggest the Beijing clock is approximately six times more stable than Vienna’s current result.

Crucially, the Beijing team also demonstrated something Vienna did not yet test: reproducibility across independently grown crystals. Two separate CaF2 crystals, grown at different times, produced nearly identical clock frequencies. This matters enormously for practical deployment — it means future nuclear clocks may function as interchangeable standards, not bespoke laboratory curiosities requiring individual calibration.

“The two teams worked independently and reached operating thorium-229 nuclear clocks at the same time, using different experimental approaches,” said Ding. “I think this is very encouraging because it shows that the concept is robust and not dependent on one particular technical implementation.”

What Nuclear Clocks Can Detect That Atomic Clocks Cannot: The Strong Force Frontier

The Th-229 transition energy of 8.4 eV results from the near-cancellation of the Coulomb and strong-force contributions to the nuclear energy levels. This means the transition is extraordinarily sensitive to any change in the ratio between these two forces — far more sensitive than electron-shell transitions in atomic clocks, which probe primarily the electromagnetic force. NIST has described this explicitly: nuclear clocks probe not just the electromagnetic force, as other atomic clocks do, but also the strong nuclear force, which holds protons and neutrons together inside the atomic nucleus.

This dual sensitivity opens a physics search space that atomic clocks cannot access. One of the most consequential open questions in fundamental physics is whether the universe’s constants — the fine-structure constant α governing electromagnetism, the QCD coupling constant governing the strong force, and others — are truly constant, or whether they vary over time or in response to as-yet-unknown fields. Many theoretical models of dark matter predict that dark matter could be an oscillating field that slightly perturbs these constants with a characteristic period.

The Vienna team did not let their clock idle. They used it to search for exactly this kind of signal — periodic fluctuations or slow drifts in the nuclear transition energy, on timescales between 20 seconds and one day, which would indicate coupling of ultralight dark matter to the nucleus. No signal was detected. But the constraints the search produced already compete with the best atomic clocks on dark matter coupling to photons (the electromagnetic channel), and go beyond any previous measurement when it comes to dark matter coupling to the strong force and quarks. Atomic clocks cannot probe that strong-force channel at all.

“A nuclear clock was something that physicists dreamt of for almost 50 years,” said Schumm. “In my team, we have been working toward this goal since 2008.”

How Does a Nuclear Clock Hunt for Dark Matter?

For readers unfamiliar with how a timekeeping device becomes a particle physics instrument: the principle is the same in both uses. A clock is, at its core, a device that counts very precisely how fast something oscillates. If dark matter is an oscillating field — as “ultralight dark matter” models propose — that field would cause the nuclear transition frequency to oscillate slightly at the characteristic frequency of the dark matter particle. Running a nuclear clock for extended periods and looking for such oscillations in the output is equivalent to running a precision spectroscopy experiment on the universe’s background structure.

The nuclear clock’s advantage here is not just precision — it is what physical interaction it is sensitive to. The Th-229 nucleus is particularly sensitive to changes in the strong nuclear force coupling constant, because the 8.4 eV transition energy results from a near-cancellation between the Coulomb and strong-force energies. Any dark matter field that couples to the strong force would produce a shift in that 8.4 eV energy — and the nuclear clock, locked to that frequency, would detect it as a drift in the clock’s output. The electromagnetic channel that atomic clocks probe is entirely separate. The Vienna team’s null result is therefore not a disappointment — it is the first data point in a new experimental program, ruling out certain dark matter models and establishing the nuclear clock as a working physics instrument.

What Still Stands Between Today’s Clocks and the Theoretical Target

Schumm was candid about how far the technology has to go. The clock, he said, remains “far from its target performance.”

The primary bottleneck is the observed linewidth of the nuclear transition signal inside the crystal. The natural linewidth of the Th-229 nuclear transition — how narrow the signal would be for a perfectly isolated nucleus — is well below 1 hertz. Inside a crystal, however, the observed linewidth is approximately 30 kilohertz wide — 30,000 times wider. This broadening comes from the crystal lattice environment: no two thorium-229 nuclei in the crystal experience exactly the same local electric and magnetic field from the surrounding calcium and fluorine atoms, so each resonates at a slightly different frequency, smearing the signal. A higher concentration of thorium dopants makes the problem worse, because the thorium atoms distort the surrounding crystal structure.

The current clocks operate at roughly one part in 10¹⁵ stability over 24 hours — extraordinary by everyday standards, but still short of the 10¹⁷ to 10¹⁸ stability that today’s best atomic clocks achieve, and far short of the 10¹⁹ to 10²⁰ theoretical target for optimized nuclear clocks. To close that gap, researchers will need to either engineer crystals with dramatically reduced inhomogeneous broadening, develop much more powerful and narrow VUV laser sources, or migrate to a trapped-ion nuclear clock configuration — in which individual thorium ions are laser-cooled and suspended in electromagnetic traps, eliminating the crystal environment entirely. NIST and several European groups are actively pursuing this direction.

What Comes Next: Competition, Collaboration, and a New Timekeeping Era

Despite finishing their Nature papers simultaneously, both teams have indicated they plan to share findings and collaborate on improving the technology. “What is really nice here,” Schumm noted, “is that the Vienna clock has slightly better thorium crystals — higher concentration, better optical properties — while the Beijing team has a stronger laser system.” The complementary strengths suggest the two approaches may inform each other directly.

The practical applications of mature nuclear clocks are broad. Precision GPS, already accurate to a few meters, could eventually improve significantly with more stable clocks in satellites. Financial market timing systems that timestamp trades to microseconds would gain additional headroom. Deep-space navigation — where spacecraft travel so far that signals to and from Earth-based clocks introduce meaningful delays — could benefit from onboard nuclear clocks that drift far less than current hydrogen masers.

But the more immediate significance is scientific. For nearly 50 years, the nuclear clock was a theoretical dream whose prerequisites — knowing thorium-229’s transition frequency precisely enough to target it with a laser — were only met in 2024. Within two years of that breakthrough, two independent teams have closed the loop. The clock works. And it has already been used to look at the universe’s dark side.

Frequently Asked Questions

What is a nuclear clock, and how does it differ from an atomic clock?

An atomic clock keeps time by locking an oscillator to the frequency at which electrons in specific atoms transition between energy states. A nuclear clock does the same thing but uses transitions between energy levels inside the atomic nucleus — between protons and neutrons — rather than in the electron shell. The nucleus is shielded from its environment by the surrounding electrons, making nuclear transitions less sensitive to external disturbances, which is why nuclear clocks are expected to eventually outperform atomic clocks in long-term stability. The current prototypes are not yet more accurate than the best atomic clocks, but they have demonstrated basic operation and already run competitive physics experiments.

Why can’t atomic clocks search for the same kind of dark matter a nuclear clock can detect?

Atomic clocks probe transitions driven by the electromagnetic force — the interaction that governs how electrons bind to nuclei. Any dark matter field that couples only to the strong nuclear force (the force binding protons and neutrons together) would produce no signal in an atomic clock, because atomic clock transitions are insensitive to changes in the strong-force coupling constant. The thorium-229 nuclear transition is uniquely sensitive to both forces, because its 8.4-electron-volt transition energy results from a near-perfect cancellation between Coulomb (electromagnetic) and strong-force contributions to the nuclear energy. A dark matter field that perturbs the strong force would shift this delicate balance and show up as a drift in the nuclear clock’s output — something no existing atomic clock can detect. The Vienna team’s dark matter search is the first time this strong-force channel has been probed with a working clock-quality instrument.

How precise are the current nuclear clocks compared to the best atomic clocks?

The Vienna clock showed stability equivalent to roughly one second of error in 30 million years over a 24-hour test. The Beijing clock is approximately six times more stable by current measurements. The best atomic optical clocks currently achieve stability equivalent to one second of error in about 55 billion years — meaning nuclear clocks still have roughly three to four orders of magnitude of precision to gain before reaching their theoretical target of one part in 10¹⁹ or better. The primary bottleneck is the crystal lattice environment, which broadens the observed nuclear signal from its natural sub-1-hertz linewidth to approximately 30 kilohertz — a gap that future crystal engineering and more powerful VUV lasers are expected to close.

When could nuclear clocks reach practical deployment in GPS or financial infrastructure?

Neither research team has specified a timeline for practical deployment, and the current clocks remain laboratory instruments. Schumm said explicitly that the technology is “far from its target performance.” Practical deployment would require a sustained improvement in linewidth (from the current ~30 kilohertz toward sub-hertz), improvements in VUV laser power and stability, and probably a demonstration that performance holds over months or years rather than days. NIST noted in August 2026 that “a nuclear clock that outperforms the best optical clocks remains years away.” The more immediate application of nuclear clocks will likely be in fundamental physics experiments — dark matter searches, tests of whether fundamental constants are truly constant — before any role in commercial timing infrastructure.