The team reports the first working optical clock that is steered directly by a nuclear transition. They lock a continuous-wave (CW) laser to the 148 nanometre nuclear transition of thorium-229 and use fast feedback from continuous absorption measurements to keep the laser on resonance. The thorium nuclei are embedded as dopants in a small, room-temperature calcium fluoride crystal, so the device operates as a compact, solid-state clock.

In plain terms, the experiment combines a very stable infrared laser that is converted to vacuum-ultraviolet (VUV) light at 148 nm, a small crystal that absorbs that light when the thorium nuclei are on resonance, and a detector that counts the absorbed photons. The team modulates the probe frequency and reads an ‘‘error’’ signal from the absorption difference to steer the laser back to the nuclear line. A frequency comb and a fiber link to a Yb+ (ytterbium ion) single-ion clock are used to compare and diagnose the nuclear clock.

The present device shows a simple, shot-noise-limited scaling of the fractional frequency instability equal to 3×10^−12 times the square root of the averaging time in seconds. That behavior means the clock’s stability improves with longer averaging. With continuous operation the instability approaches 10^−15 after one day of averaging. In their setup the team reports practical detection numbers: about 65 picowatts of VUV power at the detector, a 10% detector efficiency, and roughly 0.75% absorption on the investigated nuclear line, giving a signal-to-noise ratio near 17 after one second of measurement.

This demonstration matters because the thorium-229 nuclear transition is expected to be intrinsically less sensitive to environmental disturbances than ordinary atomic (electron-shell) transitions. Embedding many nuclei in a solid host can give a large signal and make a practical, robust clock that works at room temperature. The nuclear transition is also unusually sensitive to some proposed changes in fundamental constants. The authors used their clock to search for periodic fluctuations and slow drifts of the nuclear transition on time scales between 20 seconds and one day. Those measurements set limits on some models of ultralight dark matter; the reported limits compete with the best atomic clocks for couplings to photons and improve on previous bounds for couplings to the strong nuclear force and to quarks.