The first nuclear clocks are ticking in Vienna and Beijing

For seventy years, the best clocks on Earth have kept time by watching electrons. As of last week, two laboratories keep time by watching an atomic nucleus instead. On 7 October, Nature published two independent papers describing working thorium-229 nuclear clocks: one from TU Wien and Germany's national metrology institute PTB, the other from a Tsinghua University team in Beijing. Physicists have chased this device since it was first proposed in 2003. It now exists, twice.

What a nuclear clock actually is

Every precision clock works the same way. You find something that oscillates at a fixed, natural frequency, lock an oscillator to it, and count. In an optical atomic clock that reference is an electron jumping between energy levels in an atom such as strontium or ytterbium, and a laser is tuned to exactly the frequency of light that drives the jump.

A nuclear clock swaps the electron for the nucleus. The appeal is size: according to TU Wien, atomic nuclei are more than 10,000 times smaller than atoms and react far more weakly to stray electric and magnetic fields, the disturbances that limit electron-based clocks. A reference that the outside world barely touches is, in principle, a better reference.

The catch is energy. Flipping a nucleus between states normally takes far more energy than any laser delivers. Thorium-229 is the known exception. Its nucleus has an excited state sitting only about 8.4 electron volts above the ground state, which corresponds to vacuum-ultraviolet light at roughly 148 nanometers. That is hard to make, but it is laser territory. Pinning that transition down took about two decades; the Vienna and PTB groups first excited it with a laser in April 2024, and later that year a TU Wien and JILA collaboration measured it against a strontium atomic clock.

Vienna: the clock that steers itself

The TU Wien and PTB paper, "A thorium-229 optical nuclear clock with feedback loop" by L. Toscani de Col and colleagues, closes the loop that earlier experiments left open. Previously, the thorium crystal was read out against a conventional atomic clock. Now the nuclei discipline the laser directly: they absorb light only when the frequency is exactly right, so when the laser drifts, absorption drops, and the system pulls the frequency back automatically. PTB built the laser system; Thorsten Schumm's group at TU Wien grew the thorium-doped crystals.

According to TU Wien's press release, the clock ran for more than 24 hours without intervention and reached a precision of about 10⁻¹⁵ over one day, which the university equates to one second of error in roughly 30 million years. Coverage varies here; the Boston Globe framed the Vienna clock as losing a second every few million years, so treat the headline conversion loosely and the fractional figure as the real number.

Schumm was candid in the release that the result is not yet at the level of the world's best optical atomic clocks, while calling it a strong showing for a first prototype. The team also flagged a practical problem, as Interesting Engineering reported: the signal shifted slightly depending on where the laser passed through the crystal, so the clock kept slightly different time from day to day. Better crystals and stronger lasers are the stated fixes.

Beijing: less thorium, steadier ticks

The Tsinghua team, led by Ding Shiqian, attacked the hardest engineering problem head on. No one had built a continuous-wave laser at 148 nanometers, so the group made one, generating the ultraviolet light in heated cadmium vapour, and aimed it into a millimeter-scale calcium fluoride crystal doped with thorium-229. Chinese state broadcaster CGTN put the thorium used at 1.4 micrograms, about 200 times less than the European setup.

On stability, the Chinese clock comes out ahead. The South China Morning Post reported it as roughly six times as stable as Vienna's; CGTN described its short-term stability as nearly an order of magnitude better. The precise ratio depends on the averaging time compared, and the Nature papers are the place to settle it.

The more important Tsinghua result may be reproducibility. Two separately produced crystals gave nearly identical frequencies, agreeing to about three parts in 10 trillion, according to Interesting Engineering, and the value matched earlier JILA measurements. That matters because solid-state clocks live inside a crystal, and if each crystal nudged the nuclear frequency unpredictably, every device would need its own calibration. "There's still much room for improvement," Ding told the Boston Globe.

What it is good for, and what it is not

Today's best atomic clocks remain about 10,000 times more reliable than either nuclear prototype, the Boston Globe reported. Nobody is swapping out the clocks behind GPS or the definition of the second this year. Thorium-229 is also scarce, obtained from the decay of weapons-grade uranium stock, which limits how many crystal designs teams can try.

The near-term value is scientific. Because the thorium transition depends on a delicate balance between the strong nuclear force and electromagnetism, comparing a nuclear clock with an atomic clock is an unusually sensitive test of whether fundamental constants drift. The Vienna team already ran that comparison against a ytterbium-ion clock to hunt for ultralight dark matter. It found no signal, but according to the team's preprint, its limits compete with the best atomic clocks for dark matter coupling to photons and go beyond previous measurements for coupling to the strong force and quarks. A null result from a first prototype that already sets new bounds is a strong signal about what a mature version could do.

Longer term, the pitch is robustness. A reference that shrugs off external fields could support compact, portable clocks for satellite navigation, deep-space missions, and geodesy, where timing differences reveal gravity variations underground. UCLA physicist Eric Hudson, who was not involved, told the Globe the field is "moving at breakneck speed."

The race is now an engineering problem

The scientific question, whether a nucleus can be a clock, is answered. What remains is the unglamorous stack: more powerful 148 nm lasers, cleaner and more uniform crystals that remove the position-dependent drift, and enough thorium to iterate. Vienna has the self-stabilizing architecture and the dark matter physics; Beijing has the laser and the reproducibility. Schumm predicted in 2024 that nuclear clocks could overtake the best atomic clocks within a few years, and that timeline is now testable. The number to watch is not seconds per million years. It is the day a nuclear clock's uncertainty dips below an optical lattice clock's, and the first time two nuclear clocks in different cities are compared against each other.