Boulder team reports optical clock comparisons hitting 3.2×10⁻¹⁸ uncertainty, a milestone toward redefining the second

A Boulder-based team says it has measured frequency ratios among three of the world’s leading optical atomic clocks with uncertainties no higher than 3.2 parts in 1 quintillion, passing a benchmark that international metrology groups have identified on the road toward a possible redefinition of the second. The result marks a clear technical milestone, but the same study also found small differences from the group’s earlier measurements, underscoring that caution is still required.

The findings appear in an arXiv preprint dated July 20 from the Boulder Atomic Clock Optical Network, or BACON, collaboration, led by Alexander Aeppli and involving JILA, the National Institute of Standards and Technology, and the University of Colorado Boulder. The experiment directly compared three optical clocks in Boulder: an aluminum-ion, or Al+, clock and a ytterbium, or Yb, optical lattice clock at NIST, and a strontium, or Sr, optical lattice clock at JILA. The comparison campaign ran from Jan. 16 to March 21, 2025. BACON reported fractional standard uncertainties of 2.2 × 10^-18 for Al+/Sr, 3.2 × 10^-18 for Al+/Yb and 3.1 × 10^-18 for Yb/Sr.

The main technical change was to have all three clocks share the same ultrastable laser reference. In the preprint, the authors wrote: “The optical local oscillators of all three clocks are phase-locked to the JILA cryogenic silicon laser reference cavity (Si cavity) via a 3.6-km optical fiber network.” In practice, that meant a highly stable optical signal from a cryogenic single-crystal silicon cavity was sent between JILA and NIST over a phase-stabilized fiber link. According to the authors, the common reference improved comparison stability by a factor of 2 to 3 over earlier systems. One lattice-clock comparison reached a fractional instability of 1.3 × 10^-16 at 1 second, showing how quickly the setup could average down noise.

That matters because the SI second is still defined using a microwave transition in cesium-133, fixed at 9,192,631,770 cycles per second. Optical clocks run at much higher frequencies, which is why they can achieve much lower fractional uncertainty and are widely viewed as the leading candidates for a future definition of the second. BACON said in the paper’s abstract, “With total fractional uncertainties at or below 3.2 × 10^-18, these measurements meet an important milestone criterion for redefinition of the second in the International System of Units.” International bodies including the Consultative Committee for Time and Frequency, the International Committee for Weights and Measures and the General Conference on Weights and Measures have laid out a roadmap for any change, but meeting the uncertainty target is necessary, not sufficient.

The caution comes from the comparisons themselves. BACON’s earlier 2021 work reported ratio uncertainties in roughly the 6 × 10^-18 to 8 × 10^-18 range, so the new study is a substantial improvement. But the new Sr-related ratios differ from the 2021 results by about 1 × 10^-16 fractionally, and Al+/Yb differs by 1.6 × 10^-17. The paper also says the Yb/Sr comparison showed more day-to-day scatter than expected. Those are not deal-breakers, but they are exactly the kind of discrepancies that keep standards bodies from moving too quickly. The authors say they show why repeated, high-precision comparisons by different laboratories are still needed before any redefinition of the second.

Even with that caveat, the work strengthens the technical case for optical time standards. It shows that three different clock platforms, compared directly across two Boulder institutions, can now reach a level long cited as a prerequisite for a future optical second. If that performance can be reproduced across more labs, it could support more precise timekeeping and navigation, improved geodesy and sharper tests of fundamental physics.