Optical atomic clocks: among the most precise timekeepers yet built

Optical atomic clocks are among the most precise timekeepers built so far. The international body that maintains the SI, the BIPM, says several optical frequency standards have reached systematic uncertainties two orders of magnitude lower than caesium fountain clocks, with correspondingly lower frequency instabilities (BIPM FAQ). One 2024 laboratory clock was reported by its authors to have the lowest uncertainty of any clock at that time. Even so, optical clocks still exist mainly in specialist laboratories.

They keep time by counting the oscillations of light that atoms or single ions absorb. Those oscillations are far faster than the microwave signal used by the caesium clocks that currently define the second, which is the SI unit of time. This matters for measuring time and frequency, for measuring small differences in height on the ground, and for physics experiments that test whether fundamental quantities stay constant. The BIPM is working towards a redefinition of the second, and optical transitions are at the centre of the options under consideration.

How an optical clock works

A laser is locked to a very narrow transition inside an atom or ion, so that the laser's frequency matches the transition. Counting the laser's cycles then gives the time. A 2015 review of the field by Ludlow and colleagues notes that optical frequency references have certain advantages over their microwave predecessors, and that caesium-133 has served as the primary standard of time and frequency since 1967 (Ludlow et al., 2015; open-access version).

There are two main designs. Lattice clocks trap many neutral atoms in an optical lattice, a standing wave of laser light. Ion clocks trap one or a few charged atoms. The review covers both kinds. The 2019 aluminium clock described below is an ion design.

What has been demonstrated in the laboratory

These results are laboratory measurements. Each one comes with its own conditions, and none of them is a field or operational result.

• Single aluminium ion (2019). Brewer and colleagues reported an aluminium-27 ion clock in Physical Review Letters with a systematic uncertainty of 9.4 × 10⁻¹⁹. A magnesium-25 ion is trapped alongside the aluminium ion and used for sympathetic cooling and state readout (Brewer et al., 2019). The journal published an erratum in 2023. We could not read what it corrects, so the 2019 figures should be read with that in mind.

• Strontium lattice clock (2024). Aeppli and colleagues reported a clock using a dilute ensemble of fermionic strontium atoms in a shallow, vertically oriented, one-dimensional optical lattice. The abstract gives a total systematic uncertainty of 8.1 × 10⁻¹⁹, which the authors describe as the lowest uncertainty of any clock to date, and says all other systematic effects have uncertainties below 1 × 10⁻¹⁹ (Aeppli et al., 2024). This is the authors' evaluation of their own error budget, not a measure of how the clock performs over months in ordinary use. The paper was published in Physical Review Letters as volume 133, article 023401.

• Link to a nuclear clock (2024). Zhang and colleagues used a vacuum ultraviolet frequency comb, stabilised to the JILA strontium-87 clock, to excite the thorium-229 nuclear transition in a calcium fluoride crystal and directly measure the frequency ratio between the nuclear transition and that clock. The work was published in Nature (Zhang et al., 2024). The abstract does not give a numerical value for the ratio.

What is reported but not yet settled

Some claims come from a preprint and from a standards body, not from peer-reviewed measurements. A 2024 arXiv preprint from the International Clock and Oscillator Networking (ICON) Collaboration describes bringing transportable optical clocks from Japan and Europe together for international comparisons. The authors state that the comparisons need neither a high-performance frequency link nor information on the geopotential difference between the sites, and that the clocks were reproducible enough after transport to determine geopotential height offsets at the level of 4 cm (ICON Collaboration, 2024). The arXiv record gives no journal reference, and we did not find a peer-reviewed version.

A BIPM news item from February 2024 states that optical frequency standards are systematically achieving uncertainties below those of caesium fountain primary standards. It names ytterbium-171 and strontium-87 as candidate species, already reported to the BIPM as secondary representations of the second. It says there is growing confidence in the time and frequency community that a new definition will be adopted within the next decade or soon thereafter, and that adoption by the General Conference on Weights and Measures (CGPM) is anticipated in 2030 or later, depending on whether mandatory criteria are met (BIPM, 2024). These are expectations, not decisions.

The BIPM's frequently asked questions page (version 2.2, last updated on 13 October 2026) describes the options for a new definition. Option 1 fixes the frequency of a single optical transition. Option 2 creates a new defining constant from a weighted geometric average of the frequencies of several optical transitions. Option 3, fixing another fundamental constant such as the electron mass, is described as less favoured, because no such constant is currently known accurately enough. The page says caesium fountains would become a secondary representation of the second. On timing, it says the 2025 roadmap update keeps 2030 as the decisive milestone: the 2026 CGPM reviews progress and a draft resolution for continued work towards a proposal at the 2030 CGPM; the Consultative Committee for Time and Frequency (CCTF) evaluates the criteria in 2028 and 2029 and settles on one proposal; and the 2030 CGPM decides, provided there is consensus and the mandatory criteria are met (BIPM FAQ). The FAQ refers to eight mandatory criteria; this article does not set them out. We did not find a published outcome of the 2026 CGPM, so the choice of definition is not yet decided.

What is speculative

The following uses are proposed by researchers but have not been shown in routine practice:

• Using transportable optical clocks for geodetic sensing, precise navigation and future timing networks. The ICON authors list these as areas their campaign could affect, alongside tests of general relativity (ICON Collaboration, 2024).
• Detecting a change over time in the relative strengths of the fundamental forces. Ludlow and colleagues write that comparing clocks based on different atoms or molecules may one day make such effects observable (Ludlow et al., 2015).

Limits and open problems

The uncertainty figures above describe the error budgets of particular laboratory clocks. They do not show how well a clock keeps time in a field setting over long periods. The ICON abstract does not give an uncertainty budget for the transported clocks, so the 4 cm result cannot yet be checked against one.

Portability, power use, size and cost are also open questions. The sources we checked do not give figures for these, so this article does not quantify them. The BIPM FAQ notes that satellite time transfer is not well matched to optical clocks, and that intercontinental links remain an active area of research (BIPM FAQ). The choice of definition depends on international agreement that has not yet been reached. The thorium nuclear clock is still a research measurement.

What to watch next

• The outcome of the 2026 CGPM and its draft resolution on continued work towards a 2030 proposal.
• The CCTF evaluations in 2028 and 2029, and which redefinition option it proposes.
• A journal version of the ICON transportable clock comparison, with a full uncertainty budget for the transported clocks.
• A published numerical value for the thorium-229 to strontium-87 frequency ratio in an accessible summary.
• Field trials that report clock performance over long periods, outside a single laboratory.