First nuclear clocks run on thorium-229 crystals in two independent labs

The two clocks share a design but put their effort in different places. The PTB-TU Wien team loaded more thorium-229 into its crystals, and the Tsinghua team drove its crystals with more laser power ^[9]. Either change tends to raise the signal-to-noise ratio ^[9]. On the reported numbers, the Tsinghua device comes out ahead. Its fractional frequency instability is 5 x 10^-13 at one second of averaging, against 3 x 10^-12 for PTB-TU Wien, and both fall with the square root of the averaging time ^[10]. Tsinghua's figure is six times lower ^[14]. Because noise averages down with the square root of time, the PTB-TU Wien clock needs 36 times as long to reach the same level ^[15].

The thing this doesn't tell you is how accurate either clock is. Instability describes how fast the noise averages away. Accuracy, meaning how far the reading sits from the true transition frequency, is a separate measurement, and the Physics World account does not report one for either device.

The idea behind the two Nature papers ^[1] dates to the late 1990s ^[3]. The case for it was that a clock based on a nucleus is less prone to disturbance from electromagnetic fields, and that its transition sits at ultraviolet frequencies, above those of optical clocks, which are themselves more precise than microwave clocks because they tick more often ^[2]. The obstacle was the light. The nuclear transition is so narrow that driving it takes highly stable lasers with extremely small bandwidths ^[4]. Ding's group had already built the first continuous-wave, narrow-linewidth UV laser at 148.4 nm ^[6]. The Peik and Schumm group had been first to excite the transition with a laser and first to embed thorium-229 in calcium fluoride, which made a solid-state device with no trapped atoms or ions ^[5].

Tsinghua's paper was mostly about the machine: high-power vacuum-ultraviolet interrogation, clock performance, reproducibility between crystals and metrological consistency ^[11]. The result I would weight most is the least showy one. Clock frequencies measured in two independently fabricated crystals agreed with each other and with earlier frequency-comb measurements at JILA ^[12]. Had the frequency moved from crystal to crystal, each new device would need its own calibration, and a crystal clock would be hard to use as a shared reference.

The PTB-TU Wien dark matter search put a nuclear clock to work on a physics question. It also constrained drifts in the nuclear transition energy, and its limits came out comparable to those from today's best atomic clocks ^[13]. I think that earns the thorium crystal the description of a working platform for fundamental-physics tests: two labs run the design, and one has already published a search with it ^[8][13]. The original argument for nuclear clocks was that they could be more precise than atomic ones ^[2]. On the figures reported so far, neither clock has shown that yet ^[10][13].