First Nuclear Clock Ticks With Thorium-229 in a Crystal at Room Temperature

Physicists have long dreamed of building a clock that keeps time using the heartbeat of an atomic nucleus rather than the electron shell of an atom. That dream has now taken a decisive step toward reality. A team led by researchers at TU Wien, working with partners including the Physikalisch-Technische Bundesanstalt and the Austrian metrology institute BEV, has realized a thorium-229 optical nuclear clock in which a continuous-wave laser is locked directly to a nuclear transition using a rapid feedback loop. The device, described in Nature, operates at room temperature inside a millimetre-sized calcium fluoride crystal and represents the first stand-alone nuclear clock in which the interrogating laser is steered by the nuclei themselves rather than by an external frequency reference.

The key to the achievement is one of the strangest energy levels known in nuclear physics. The thorium-229 isotope possesses an exceptionally low-lying isomeric state, first proposed in the 1970s and only recently pinned down at an energy of about 8.3 electronvolts, corresponding to a wavelength of roughly 148 nanometres in the vacuum ultraviolet. Because this energy is so low, the transition can be driven with laser light, something impossible for essentially any other nuclear transition, which typically requires energies thousands of times higher. The idea of exploiting this quirk for timekeeping was proposed in 2003 by Ekkehard Peik and Christian Tamm, who envisioned a nuclear clock with a potential resonance quality factor on the order of 10 to the 19th power, far beyond what ordinary atomic clocks can offer.

The path to that vision spanned half a century of incremental discoveries. Gamma spectroscopy experiments gradually narrowed the transition energy from a conjectured value below 100 electronvolts to around 8 electronvolts. In 2016, the existence of the isomer was proven through its internal conversion decay channel, and in 2022 researchers achieved the first optical detection of the radiative nuclear decay, refining the energy to 8.338 electronvolts. Resonant laser excitation of the nucleus followed in 2024 and 2025 in thorium-doped calcium fluoride and other hosts, and the nuclear quadrupole structure of the transition in the crystal was subsequently resolved. The final missing ingredient was a continuous-wave laser at 148 nanometres, developed in 2025, which enabled absorption spectroscopy of the nuclear transition and, with it, the possibility of true clock operation.

Absorption spectroscopy is the technical heart of the new clock. Rather than waiting for excited nuclei to fluoresce, a slow process limited by the roughly ten-minute lifetime of the isomeric state, the team shines the vacuum ultraviolet laser through the crystal and measures how much light the nuclei absorb. A photomultiplier tube with a cesium iodide photocathode, operated in photon-counting mode behind the crystal, records the transmitted photons. This approach delivers three orders of magnitude more signal photons per second than fluorescence detection and allows the transition to be probed continuously. In the experiment, 65 picowatts of vacuum ultraviolet power reached the detector, and the thorium nuclei absorbed about 0.75 percent of the light on the investigated transition, yielding a signal-to-noise ratio of approximately 17 after one second of measurement.

The clock architecture combines two control loops. A high-finesse cavity-stabilized external-cavity diode laser operating at 1,187 nanometres provides short-term stability, while the nuclear resonance provides the long-term reference. The 1,187-nanometre seed laser is frequency-quadrupled, with the final doubling stage performed in a randomly quasi-phase-matched strontium tetraborate crystal to generate the 148-nanometre radiation. By modulating the offset lock frequency between two points on either side of the absorption line, the researchers generate an error signal with a zero crossing at the line centre. Each clock cycle, the measured deviation is inverted and applied in a single adjustment step to an electro-optical modulator, steering the laser back onto the nuclear resonance and closing the feedback loop. Depending on the chosen integration time, feedback steps occur every 20 seconds or every 30 minutes.

The performance, quantified by the Allan deviation, shows a shot-noise-limited fractional frequency instability of 3 times 10 to the minus 12 divided by the square root of the averaging time in seconds, approaching instabilities of 10 to the minus 15 over one day of continuous operation. The clock ran unattended for a full day, unperturbed by crystal temperature fluctuations of about 0.1 kelvin, consistent with the known temperature sensitivity of the transition of 1.8 kilohertz per kelvin near room temperature. The nuclear transition frequency itself is 2.0204 times 10 to the 15th hertz, with a measured linewidth of approximately 100 kilohertz. To benchmark the device, the team compared it with a ytterbium-ion single-ion clock at BEV via a Doppler-compensated fibre link and a pair of frequency combs, allowing the ratio of the nuclear and atomic transition frequencies to be monitored continuously.

One limitation emerged from the solid-state nature of the clock. Reproducibility between runs on different days was limited to about 5 times 10 to the minus 13, and measurements at four different positions on the crystal revealed line-centre deviations of up to 1.7 kilohertz, well above the 0.13-kilohertz statistical spread. Because the laser beam must be realigned before each run, it probes slightly different local regions of the crystal, and the team conjectures that local strains from structural inhomogeneity cause these shifts. Remeasuring the same position gave reproducible frequencies. Future improvements could constrain the optical path to a fixed volume or improve doping homogeneity through optimized crystal growth, potentially restoring full reproducibility.

Even at this early stage, the nuclear clock delivers competitive results in the search for dark matter. Many theories predict that dark matter may consist of ultralight scalar bosons whose feeble couplings to photons, gluons or quarks would cause oscillations in fundamental constants such as the fine structure constant, the QCD scale parameter and quark masses. The thorium-229 transition is extraordinarily sensitive to such variations because its anomalously low energy arises from a near-cancellation of mega-electronvolt-scale electromagnetic and strong-force contributions, giving it sensitivity factors predicted to exceed those of the best atomic clocks by orders of magnitude. Analysing roughly 23 hours of clock data with the Lomb-Scargle method, the team found no oscillation amplitudes above a 5 percent detection threshold established through Monte Carlo simulations, and used the resulting upper limits to constrain dark matter couplings.

The constraints are striking. For couplings to photons, the nuclear clock already rivals the best atomic clock comparisons, and for couplings to the strong force, probed through variations of the QCD scale and quark masses, it reaches a factor of 10 to 1,000 deeper into parameter space than previous experiments. The fast feedback of the absorption scheme also allowed direct probing of boson masses that previously could only be restricted through line-shape analyses. A linear fit of the frequency ratio data over the run gave a drift of (2 plus or minus 4) times 10 to the minus 14 per day, consistent with zero, corresponding to drift limits on the fundamental constants scaled by the thorium sensitivity factors.

The outlook for the technology is ambitious. Increasing the vacuum ultraviolet laser power through enhancement cavities, four-wave mixing in cadmium vapour or patterned nonlinear crystals could push the instability toward 10 to the minus 15 per square root second, and longer crystals or vacuum ultraviolet cavities could raise the signal-to-noise ratio by an order of magnitude. Alternative host crystals such as thorium fluoride, where thorium is part of the stoichiometric structure rather than a dopant, promise narrower linewidths and better reproducibility, while spinless solids could suppress magnetic broadening from neighbouring fluorine nuclei. With a kilohertz linewidth and modest laser power, the researchers estimate that a solid-state nuclear clock could reach instabilities near 10 to the minus 16 per square root second, matching state-of-the-art optical atomic clocks while retaining the compact form factor of a crystal at room temperature, and delivering a four-order-of-magnitude boost in sensitivity to variations of the fundamental constants that may reveal new physics beyond the Standard Model.

Subject of Research: A solid-state optical nuclear clock based on the thorium-229 nuclear transition with laser feedback stabilization

Article Title: A thorium-229 optical nuclear clock with feedback loop

Article References: Toscani De Col, L., Riebner, T., Morawetz, I., Schneider, F., Sempelmann, N., Schlachet-Lépinay, J., Schaden, F., Bartokos, M., Kazakov, G. A., Beeks, K., Gerstenecker, B., Pimon, M., Lahs, S., Hellerschmied, A., Lercher, T., Denker, H., Premper, J., Niessner, A., Matus, M., … Schumm, T. (2026). A thorium-229 optical nuclear clock with feedback loop. Nature. https://doi.org/10.1038/s41586-026-11084-4

Image Credits: AI Generated

DOI: 10.1038/s41586-026-11084-4

Keywords: thorium-229, nuclear clock, optical clock, vacuum ultraviolet laser, absorption spectroscopy, calcium fluoride crystal, dark matter, fundamental constants, frequency metrology, isomeric transition, feedback loop, quantum metrology