Optical Clock Ratios Reach Redefinition-Level Uncertainty

Optical clocks are now accurate enough that the limiting question is often not whether one clock can be built, but whether independent clocks can be compared fast and cleanly enough to expose hidden systematics. That matters for any future redefinition of the SI second: the case depends on repeated frequency-ratio measurements between different atomic species, not only absolute evaluations of a single standard. This paper reports high-precision ratios among $^{27}\mathrm{Al}^+$, $^{171}\mathrm{Yb}$, and $^{87}\mathrm{Sr}$ clocks, with total fractional uncertainties at or below $3.2\times10^{-18}$.

Core Contribution

The central contribution is an experimental architecture for comparing several optical clocks against a common ultrastable optical reference. The authors distribute light derived from a cryogenic single-crystal silicon cavity to all clocks, including across a 3.6 km phase-stabilized fiber link between institutions. That common reference reduces the instability that otherwise slows ratio measurements and makes it harder to separate white averaging noise from drift, temperature effects, or species-specific systematics.

The scientific result is also deliberately comparative. The paper does not only report internally precise ratios; it compares them with previous measurements and with recommended values. The reported discrepancies are small in absolute terms but meaningful at this level: about $1\times10^{-16}$ in some $^{87}\mathrm{Sr}$ ratios and $1.6\times10^{-17}$ for the Al$^+$/Yb ratio relative to the authors’ previous measurements. The authors use those shifts to argue that repeated high-precision campaigns by different laboratories remain necessary before clock ratios can be treated as settled constants.

Technical Approach

The method combines three ingredients: mature optical clocks, a shared reference laser, and real-time environmental correction. The Al$^+$ ion clock, Yb lattice clock, and Sr lattice clock are compared through optical frequency combs and phase-stabilized links rather than through independent local oscillators at each site. The common cryogenic-cavity reference improves the short-term comparison stability by a factor of 2 to 3 over the previous system, and the optical lattice clock ratio reaches a fractional instability of $1.3\times10^{-16}$ at 1 second.

One practical detail is the treatment of blackbody radiation shifts, especially for Yb. The paper validates a real-time Yb BBR correction by examining the Yb/Sr ratio on January 24, 2025. Figure 6 shows the corrected ratio against the 2021 CIPM recommended ratio, comparing a 311 s rolling real-time correction with a correction based only on the average temperature over the full run. The inset makes the scale clear: omitting the temperature correction exposes the room-temperature Yb BBR shift, while the real-time correction refines the residual variation during the measurement.

Results and Analysis

The headline number is the uncertainty floor: all reported ratios among Al$^+$, Yb, and Sr reach fractional uncertainties no larger than $3.2\times10^{-18}$. That meets a milestone criterion relevant to redefining the second, but the paper’s more useful lesson is about measurement repeatability. At $10^{-18}$-level uncertainty, a $10^{-16}$-level shift in a Sr ratio is not a rounding error; it is a sign that long-term reproducibility, laboratory-to-laboratory agreement, and systematic-error accounting remain active constraints.

The stability improvement is the strongest support for the experimental design. A 2–3× gain in comparison stability means fewer hours are needed to reach a target statistical uncertainty, which makes campaigns less vulnerable to slow non-white processes. The $1.3\times10^{-16}$ at 1 second lattice-clock ratio instability is therefore not just a technical benchmark. It changes what the experiment can diagnose: faster averaging gives the authors more leverage to identify drift, thermal transients, and hidden offsets before they are averaged into a final ratio.

Caveats

The evidence is strong for the reported apparatus and clock ensemble, but it does not close the broader metrological problem. The discrepancies with previous measurements are part of the result, not a nuisance term. They show that even carefully evaluated optical standards can disagree at levels relevant to future recommended frequencies. The paper supports the case for optical-ratio networks built around shared ultrastable references; it also supports a conservative reading of the field’s readiness, where repeated independent comparisons remain the deciding evidence.