261009 – Thorium 229 Clocks

by | Oct 9, 2026 | Physics | 0 comments

The Nucleus Starts Keeping Time

· @Thomas Lee Abshier, ND

What happened

On 7 October 2026, Nature published two papers describing the first clocks that keep time by the rhythm of an atomic nucleus rather than the electrons around it. One came from Thorsten Schumm’s group at TU Wien in Vienna, working with Germany’s national metrology institute (PTB); the lead author is Luca Toscani De Col. The other came from a large team led by Tsinghua University in Beijing, with Beichen Huang as lead author and Shiqian Ding as senior author (Science Media Centre Spain; arXiv 2606.08870).

Both clocks use the same atom, thorium-229, embedded in a small calcium fluoride crystal. Both lock a vacuum-ultraviolet laser at about 148 nanometres to a transition inside the thorium nucleus. Both work. Neither is yet as good as the best conventional clocks.

The report from The Independent, syndicated on MSN, gets the headline right. A genuinely new kind of clock now exists. But the article is thin, it never mentions that half of the achievement is Chinese, and several of its sentences blur the difference between what these clocks can do today and what physicists hope they will do in ten years. This essay fills in the details, corrects the errors, and explains why the work matters.

How a clock that counts light works

Every clock is an oscillator plus a counter. A pendulum swings about once a second; a quartz crystal vibrates 32,768 times a second. An atomic clock replaces the swinging part with the frequency of light that an atom absorbs when one of its electrons jumps between two energy levels. Because every atom of a given isotope is identical, that frequency is a natural constant that anyone, anywhere, can reproduce.

The clock does not simply watch the atom. A laser (or, in older clocks, a microwave source) is tuned near the atom’s resonance. A detector reports whether the atoms absorbed the light. Electronics nudge the laser back toward the exact resonance whenever it drifts. The atom is the referee; the laser is the pendulum; the feedback loop is what turns a measurement into a clock. That distinction, between measuring a resonance once and locking a laser to it continuously, is the whole difference between the 2024 milestone and the 2026 one.

The first caesium atomic clock was built by Louis Essen and Jack Parry at Britain’s National Physical Laboratory in 1955, so the article’s “more than 70 years” is correct. Since 1967 the SI second has been defined by the caesium microwave transition. Today’s best optical clocks, using strontium or ytterbium atoms or aluminium ions, run at frequencies some 50,000 times higher and reach systematic uncertainties below one part in 10^18. JILA’s strontium clock, for example, reported a total uncertainty of 8 × 10^-19 (Zhang et al., Nature 633, 2024). That is the bar any newcomer must eventually clear.

A nuclear clock keeps the same architecture but changes the referee. Instead of an electron jumping between shells, the protons and neutrons inside the nucleus rearrange into a slightly higher-energy configuration, called an isomer, and then relax back. The nucleus is roughly 100,000 times smaller across than the atom, and it is shielded by its own electron cloud. In principle, stray electric and magnetic fields that push electron energy levels around should barely touch it.

Why only thorium-229 will do

Nuclear energy levels are normally separated by thousands to millions of electron-volts. Those are gamma-ray energies; no laser can drive them with the precision a clock needs. Thorium-229 is the single known exception. Its first excited state sits only about 8.4 electron-volts above the ground state, which corresponds to vacuum-ultraviolet light near 148 nanometres. That is hard to make, but it is laser light.

The low energy is an accident of near-cancellation. The strong nuclear force and the electromagnetic repulsion between protons each contribute hundreds of thousands of electron-volts to the difference between the two states, and they almost exactly offset. José Crespo López-Urrutia of the Max Planck Institute for Nuclear Physics compares it to putting a different oil tanker on each pan of a scale and finding they balance to within a kilogram (SMC Spain). That accident is the source of both the clock’s promise and its scientific value, as the implications section explains.

The road from suspicion to working clock took fifty years:

Year Milestone
2026 (Oct) Vienna–PTB and Tsinghua nuclear clocks published in Nature; first closed-loop operation
2026 (Jun) Both teams post preprints; Vienna clock also used for a dark-matter search (arXiv 2606.04997)
2024 (Sep) JILA (Boulder) measures the thorium-to-strontium frequency ratio to 12 digits; often called the first nuclear-clock prototype (LMU)
2024 (Apr) PTB and TU Wien first excite the nucleus with a tabletop laser in a CaF2 crystal (Physics World)
2023 ISOLDE at CERN observes the isomer’s VUV photon in a crystal, pinning the energy down (LMU)
2016 LMU Munich makes the first direct detection of the isomer (Physics World)
2003 Ekkehard Peik and Christian Tamm (PTB) propose a thorium-229 nuclear clock
1976 Gamma-ray spectroscopy first hints at an unusually low-lying excited state

The thorium itself comes from an unlikely source. Thorium-229 is a decay product of uranium-233, and the crystals used in Europe were made with material from a US stockpile associated with nuclear weapons and disarmament programs (Physics World). The isotope is radioactive, with a half-life of roughly 7,900 years, and it is scarce. Supply is one of the practical constraints Schumm names for the coming years (Vienna.at).

What the two teams actually built

Both groups closed the loop: a continuous-wave laser at 148 nm shines through a thorium-doped crystal, a detector measures how much light the nuclei absorb, and fast feedback keeps the laser pinned to the nuclear resonance. They took different routes to enough signal.

Vienna (TU Wien / PTB) Beijing (Tsinghua)
Lead author Luca Toscani De Col Beichen Huang
Strategy Crystal with about 200 times more thorium nuclei Brighter laser: 10 µW VUV from four-wave mixing in cadmium vapour
Short-term instability (preprint) 3 × 10^-12 per √(τ/s), nearing 10^-15 over one day 2 × 10^-12 per √(τ/s)
Distinctive result Ran as a stand-alone clock against a ytterbium-ion atomic clock over optical fibre; used for a dark-matter search Two independently grown crystals agreed to about 1 part in 10^13
Crystal temperature Room temperature Locked to a nuclear line with weak temperature sensitivity

Sources: arXiv 2606.04997, arXiv 2606.08870, Interesting Engineering, ScienceAlert.

Press reports of the published versions describe the Beijing clock as about six times steadier than Vienna’s in the short term (SCMP); Dolores del Campo of Spain’s national metrology centre gives the same figure, while noting that Vienna’s comparison against an established ytterbium clock is its own strength (SMC Spain). The preprint numbers above differ by less than that, so the published papers presumably report improved Beijing figures; readers should treat the factor of six as the reported headline rather than a figure verified against the final text.

The Beijing result on two crystals deserves more attention than it has received. A clock standard is only useful if a second copy, built separately, ticks at the same rate. Agreement between independently grown crystals is early evidence that a solid-state nuclear clock can be reproduced, which is the property that would eventually let one lab’s clock be compared with another’s.

The honest scorecard: both clocks are roughly ten thousand times less stable than the best optical atomic clocks today. The authors and outside experts expect improvements of around four orders of magnitude from better lasers, purer and more transparent crystals, and tighter temperature control (SMC Spain).

Checking the article, claim by claim

Most of the article is accurate in outline. Its problems are overstatement, garbled physics in two places, and what it leaves out.

Claim in the article Verdict What is actually the case
“The first ever working nuclear clocks” True, with a nuance JILA’s 2024 work was already called a nuclear-clock prototype. What is new in 2026 is closed-loop operation: the nucleus continuously steers the laser.
Scientists “have come up with a new and even more precise technology” Misleading Not yet. The same article later admits the clocks are less stable than the best atomic clocks; today they trail by about four orders of magnitude. Greater precision is a hope, not a result.
Atomic clocks have been used “for more than 70 years” True The first caesium clock ran in 1955.
Atomic clocks showed “that time behaves differently in different parts of space” Garbled The real result is gravitational time dilation: clocks run measurably slower lower in a gravitational field, now detectable over height differences of a millimetre. GPS must correct for it.
Nuclear clocks “could be used … to detect the presence of a form of dark matter” Overstated The Vienna search found nothing. What the clock can do is rule out ranges of one hypothetical class, ultralight scalar dark matter. Most dark-matter candidates would leave no trace in it.
Nuclei are “about 100,000 times smaller than atoms” and so less sensitive to electric fields Mostly true The size ratio is right. But in a crystal the nucleus still feels the surrounding electric field gradient, which splits the line into several components, and the frequency shifts with temperature. Vienna also found the signal varied with where the laser passed through the crystal.
A 2003 proposal chose thorium-229 “because it is in a low energy state” Garbled Every nucleus has a ground state. What matters is that thorium-229’s excited state lies unusually close to its ground state, about 8.4 eV instead of the usual thousands or millions.
Better performance “might involve … much more powerful lasers” Partly true Laser power is one factor. Crystal quality, thorium density and temperature control matter as much; the two teams improved different ones.
The two papers are titled as listed Unconfirmed The Vienna title matches its preprint. The Beijing preprint was titled “A nuclear clock based on 229Th”; the published title may have changed.

The largest omission is the second team. The article never says that one of the two clocks was built in Beijing. Crespo López-Urrutia notes that Tsinghua caught up with the pioneering groups in Braunschweig, Vienna and Boulder in a very short time (SMC Spain). The article also omits the US contribution: JILA in Boulder supplied the 2024 frequency measurement on which both clocks build, and the thorium in the European crystals came from American stockpiles.

What it means

A sharper test of whether the constants are constant

This is the real prize, and the article barely touches it. Recall the two oil tankers. The thorium transition frequency is the tiny difference between two enormous energies, one set mainly by the strong force and one by electromagnetism. If either force varied even slightly over time, or from place to place, the small difference would shift by a far larger fraction. Estimates put the thorium clock’s sensitivity to a change in the fine-structure constant, the number that sets the strength of electromagnetism, at several thousand times that of ordinary atomic clocks; Crespo López-Urrutia puts it at almost ten thousand for certain dark-matter couplings (SMC Spain).

The method is to compare a nuclear clock with an atomic clock, year after year, and watch the ratio. Physics as currently understood says the ratio is fixed forever. A drift would mean that something assumed to be a constant of nature is not. JILA has already begun keeping such a “time record” of the thorium frequency against its strontium clock (APS DAMOP 2025).

Dark matter: narrowing, not finding

Some theories propose that dark matter is an extremely light field filling space and oscillating slowly. If such a field touched ordinary matter, it would make the constants wobble at the field’s frequency. The Vienna team looked for wobbles between 20 seconds and one day and found none. Its limits compete with the best atomic clocks for coupling to photons and go beyond earlier measurements for coupling to quarks and the strong force (arXiv 2606.04997). That is useful science. It is not a detection, and it says nothing about the heavier dark-matter particles that most searches target.

Practical clocks: plausible, but a decade or more away

Because the thorium sits in a solid crystal rather than a trapped, laser-cooled cloud of atoms, nuclear clocks could in principle become compact and rugged; Schumm’s group even speaks of a chip-scale device (Vienna.at). Applications would include timekeeping for navigation where satellite signals are jammed or unavailable, synchronising communication networks, and measuring small differences in gravity for surveying. Some press releases go further and talk of predicting earthquakes (IOM3). That is speculative. Optical clocks already sense centimetre-scale height differences, and no clock has yet predicted an earthquake.

The definition of the second

The international metrology community expects to redefine the second around 2030 using optical atomic clocks. Del Campo judges that nuclear clocks are unlikely to be ready for that revision, though they may matter in later ones (SMC Spain).

A race

After the 2024 results, both the United States and China launched larger nuclear-clock programs (Vienna.at). Within two years a Chinese team matched Europe and arguably passed it on short-term stability. Precision timing underpins navigation, finance, telecommunications and defence, so leadership in the next generation of clocks is not merely academic. Schumm calls it a fierce but friendly competition (SCMP).

The measure of a constant

Every clock rests on an assumption that something in nature repeats exactly. Pendulum clocks assumed gravity and the length of the rod stayed fixed. Atomic clocks assume that the electromagnetic force, which sets electron energy levels, never changes. A nuclear clock lets us test that assumption against a different force, the strong force, rather than taking it on faith. Two clocks that rest on different forces, compared over years, become an experiment on the stability of the laws themselves.

That is why this result deserves attention beyond the headline. The new clocks are not yet better timekeepers. They are a new kind of instrument, the first that lets a laser control and read a nucleus continuously, and their first job will be to ask whether the numbers physics treats as absolute really are. A null answer would strengthen the case that they are. Any other answer would be the most important physics result in a generation.

Sources

Submit a Comment