The most accurate clock yet runs on a single lutetium ion
A Singapore clock built on one lutetium ion now keeps time to the 19th decimal place — and clocks like it may soon redefine the second.
The most accurate clock yet reported is a lutetium-ion clock at Singapore’s Centre for Quantum Technologies, at the National University of Singapore. Results published in Nature on 23 September 2026 gave it the lowest uncertainty reported for any optical atomic clock, beating recent record holders:
| Clock | Lab | Year | Uncertainty |
|---|---|---|---|
| Lutetium ion | Centre for Quantum Technologies | 2026 | 1 × 10⁻¹⁹ |
| Aluminum ion | NIST | July 2025 | 5.5 × 10⁻¹⁹ |
| Strontium lattice | JILA, Boulder | 2024 | 8.1 × 10⁻¹⁹ |
Two copies of the lutetium clock agreed with each other to 5.7 × 10⁻¹⁹. NIST’s aluminum result was itself 41% better than the JILA strontium clock. All are optical clocks: instead of shining microwaves on caesium atoms, the standard behind today’s official second, they lock a laser to an atomic transition that oscillates far faster, dividing time more finely. NIST says such clocks could potentially lose only one second every 30 billion years, more than twice the age of the universe.
That figure is a translation of uncertainty, not a stopwatch result. No optical clock runs for eons; until its 2025 upgrade, NIST’s aluminum ion had to be reloaded about every 30 minutes, because hydrogen seeping from the steel vacuum chamber kept colliding with it. Rebuilding the chamber in titanium cut that hydrogen 150-fold and stretched runs to days. The number describes how precisely the tick rate is known.
Lutetium’s advantage is that its clock transition barely responds to changes in temperature or magnetic field. The harder problem is checking a claim like this: at the 10⁻¹⁹ level, clocks feel gravity slowing time over millimetres of height, and Earth’s gravity isn’t mapped well enough yet to compare clocks in different labs.
Next in line is the second itself. Under the international metrology roadmap, the 2026 General Conference on Weights and Measures reviews progress, and a decision on an optical definition is not expected before 2030.
Sources & references
5 referencesWell-established. Corroborated by 4 independent sources.



