The Most Precise Clocks Ever Built

An optical lattice clock built by the National Institute of Standards and Technology (NIST) and JILA, using strontium atoms, is precise enough that it would neither gain nor lose a second over roughly 15 billion years, longer than the current age of the universe, according to research from the team led by physicist Jun Ye. That is not a rounding error or a marketing figure. It is a directly measured performance number, and it makes optical lattice clocks the most precise timekeeping devices humans have ever built, edging out even the cesium atomic clocks that currently define the official length of a second worldwide.

Getting there took decades of work on laser cooling, atom trapping, and a technology called the optical frequency comb, and the story of how physicists built a clock this good is also the story of why the world's official definition of time is now on track to change. This piece walks through what "precision" actually means for a clock, how optical lattice clocks trap and measure atoms, how they stack up against cesium fountains and ordinary quartz watches, and why shaving still more zeroes off a clock's error rate has real consequences for GPS, satellite navigation, and tests of Einstein's general relativity.

What "precision" actually means for a clock

Precision — how consistently a clock ticks at the same rate over time, expressed as a fractional error such as 1 part in 10^18. Every clock, from a sundial to a strontium lattice clock, works by counting repetitions of something periodic. A pendulum clock counts swings. A quartz watch counts vibrations of a crystal driven by an applied voltage. An atomic clock counts oscillations of light absorbed or emitted when an electron inside an atom jumps between two specific, fixed energy levels.

The reason atoms make dramatically better timekeepers than any mechanical part is that atoms of a given element and isotope are physically identical everywhere in the universe. A cesium-133 atom in a lab in Colorado has the exact same internal structure as a cesium-133 atom anywhere else, so the frequency of light it absorbs during a specific transition is a fixed, universal number, not something that depends on manufacturing tolerances, temperature drift, or metal fatigue. A pendulum's swing rate, by contrast, depends on its exact length, the local strength of gravity, air resistance, and how worn its pivot is. Two "identical" pendulum clocks built in the same factory will still drift apart over time in ways two cesium atoms never will.

This distinction matters because "precise" and "accurate" are not the same thing, even though people use them interchangeably. A clock can tick with extraordinary consistency (precise) while still running at a rate that's slightly off from the internationally agreed second (inaccurate), the way a stopwatch could click at a perfectly steady interval that happens to be 1.001 seconds instead of exactly one second. The clocks in this article are calibrated against international standards specifically so both properties hold at once: they tick consistently, and that consistent tick matches the internationally defined second as closely as physics currently allows.

Cesium atomic clocks became the practical foundation for modern timekeeping because a specific microwave-frequency transition in cesium-133 is stable, reproducible in labs worldwide, and was well enough characterized by the 1960s to standardize. In 1967, the 13th General Conference on Weights and Measures formally redefined the SI second as 9,192,631,770 cycles of the radiation corresponding to a specific hyperfine transition in cesium-133, replacing an older definition tied to Earth's rotation, which turned out to be a worse timekeeper than a metal atom.

How cesium fountain clocks work, and where they run out of room

A cesium fountain clock, the technology behind NIST's official U.S. time standard, works by launching a cloud of laser-cooled cesium atoms upward through a microwave cavity, letting gravity pull them back down through the same cavity, and measuring how strongly the atoms respond to a microwave frequency tuned near their natural transition frequency. NIST's current primary standard, NIST-F2, reaches an uncertainty low enough that it would neither gain nor lose a second over roughly 300 million years, a figure NIST itself has published for the fountain-clock generation of atomic frequency standards.

The fountain design was a major leap over earlier cesium beam clocks because the slow, gentle arc of laser-cooled atoms through the microwave cavity gives far more time to interrogate the atoms precisely, and laser cooling itself, recognized with the 1997 Nobel Prize in Physics awarded to Steven Chu, Claude Cohen-Tannoudji, and William Phillips, slows atoms from hundreds of meters per second down to a crawl by hitting them with carefully tuned laser light from multiple directions.

But cesium fountains have a physical ceiling. The transition they measure sits in the microwave part of the electromagnetic spectrum, oscillating around 9.2 billion times per second. That sounds fast, but it is roughly one hundred thousand times slower than the optical-frequency transitions that newer clocks exploit. Frequency is the fundamental resolution limit of any oscillator-based clock, the same way a ruler with millimeter markings can't distinguish two lengths that differ by a tenth of a millimeter. A cesium fountain, no matter how cleverly engineered, cannot subdivide a second more finely than its 9.2-gigahertz "tick rate" ultimately allows. To get meaningfully better, physicists needed an atomic transition that oscillates at a much higher frequency, and that meant leaving the microwave spectrum behind for visible and near-visible light.

Inside an optical lattice clock

Optical lattice clock — an atomic clock that measures a light-frequency (rather than microwave-frequency) transition in atoms held nearly motionless by intersecting laser beams. Optical lattice clocks, the strontium clocks built by Jun Ye's group at JILA (a joint institute of NIST and the University of Colorado Boulder) being the best known example, exploit an optical transition that oscillates hundreds of thousands of times faster than cesium's microwave transition. Because the clock is effectively counting oscillations to measure elapsed time, a much higher oscillation frequency lets it subdivide a second far more finely, the same logic that lets a ruler etched with millimeter lines measure more precisely than one etched only with centimeter lines.

Why atoms have to be trapped and nearly frozen in place

Any atom drifting through free space experiences a Doppler shift, the same physical effect that raises the pitch of an ambulance siren as it approaches and lowers it as it recedes, applied here to the frequency of light the atom absorbs rather than sound. An atom moving even slowly relative to the measurement laser will absorb light at a slightly shifted frequency, which corrupts the clock's precision if left uncorrected. The fix that gives optical lattice clocks their name is trapping thousands of atoms inside a lattice formed by intersecting laser beams, an arrangement that holds the atoms nearly motionless in place like eggs in an egg carton, and dramatically suppresses this motion-induced Doppler error. The atoms are laser-cooled first, down to temperatures within a few millionths of a degree of absolute zero, before being loaded into the lattice, which further minimizes the residual jostling that would otherwise smear out the measurement.

The optical frequency comb, the tool that makes any of this measurable

Even after trapping atoms and driving a clean optical transition, physicists faced a second, more subtle problem: nobody had a way to directly count oscillations happening hundreds of trillions of times per second, or to compare one optical frequency to another, or to a microwave cesium standard, with the precision the clock itself demanded. The solution was the optical frequency comb, a device that generates a spectrum of evenly spaced laser frequencies resembling the teeth of a comb, which acts as a precise ruler connecting optical frequencies to countable microwave frequencies. The 2005 Nobel Prize in Physics went to John L. Hall and Theodor W. Hänsch specifically for developing the optical frequency comb technique, work that made it practical to measure optical clock frequencies with the accuracy needed to actually realize their theoretical advantage over cesium. Without the frequency comb, an optical lattice clock's raw transition would be effectively impossible to compare against anything else with useful precision, a landmark measurement technology sitting quietly underneath every headline optical-clock result since.

Getting the uncertainty down to parts in 10^18

Modern strontium lattice clocks at JILA have pushed systematic uncertainty into the range of roughly 1 to 2 parts in 10^18, and more recent work from the Ye group and University of Delaware has reported systematic uncertainty as low as about 8 x 10^-19, among the lowest uncertainties reported for any clock to date (arXiv:2509.13991). At that level of uncertainty, the clock's tick rate is known and reproducible to a level equivalent to measuring the age of the universe, roughly 13.8 billion years, to within a fraction of a second. Reaching this level required identifying and correcting for an enormous list of tiny systematic error sources, including stray electric and magnetic fields, blackbody radiation from the surrounding vacuum chamber, and even the gravitational pull of nearby laboratory equipment, all of which can shift an atomic transition frequency by amounts that would have been irrelevant to a cesium clock but matter enormously once uncertainty drops this far.

Comparing clock precision across generations

Clock type	Approximate precision	Physical basis
Optical lattice clock (strontium, JILA/NIST)	~1 second error per ~15 billion years; best systematic uncertainty near 8 x 10^-19 (arXiv:2509.13991)	Optical-frequency transition in laser-cooled atoms trapped in an optical lattice
Cesium fountain clock (NIST-F2, current U.S. primary standard)	~1 second error per ~300 million years (NIST)	Microwave-frequency transition (9,192,631,770 Hz) in cesium-133, the current SI second definition
Aluminum-ion optical clock (NIST)	Comparable systematic uncertainty to strontium lattice clocks, in the low parts-in-10^18 range	Optical-frequency transition in a single trapped, laser-cooled aluminum ion
Quartz crystal clock (typical consumer device)	~1 second error per several months to a few years	Mechanical vibration of a quartz crystal driven by an applied voltage
Pendulum clock (mechanical)	~1 second error per several days to weeks	Mechanical pendulum swing period, sensitive to temperature and air resistance

Each generation of clock technology traded a lower measurement frequency for a higher one, and each jump in frequency bought roughly several more orders of magnitude of precision.

Why cesium still officially defines the second, and why that is about to change

Despite everything above, the SI second is still officially defined by that 1967 cesium-133 microwave transition, not by any optical clock, because a formal redefinition of an SI base unit requires overwhelming international consensus and redundant, independently verified measurements, not just a single impressive result from one lab. That consensus process has been moving forward. In 2022, the 27th General Conference on Weights and Measures (CGPM), the international body responsible for the SI system, adopted a resolution laying out a roadmap toward redefining the second using optical clocks, targeting the 2030 CGPM meeting, contingent on technical milestones including having at least three independent optical clock systems, in different institutions, each demonstrating systematic uncertainty better than 2 x 10^-18.

That threshold is no longer theoretical. By 2025, multiple national metrology institutes, including labs in the United States and China, had independently reported optical clocks meeting or beating that 2 x 10^-18 uncertainty bar, bringing the number of qualifying systems to at least three, which is the specific condition the 2022 CGPM roadmap set as a prerequisite for moving forward with redefinition. The International Bureau of Weights and Measures (BIPM), the Paris-based organization that maintains the SI system, has been coordinating comparisons between these clocks across national labs specifically to build the redundant, cross-checked evidence base that a unit redefinition requires. If the 2030 timeline holds, the second's official definition will shift from a microwave cesium transition to an optical transition for the first time since the SI second was defined at all, a change driven directly by the fact that optical clocks have become measurably more stable and precise than the standard they would replace.

This is a slower, more conservative process than the underlying physics might suggest is necessary, and that conservatism is deliberate. Every downstream measurement in physics, engineering, and commerce that depends on the second, from GPS timing to financial transaction timestamps, needs the definition to change exactly once, cleanly, with full international agreement, rather than drifting through years of competing regional standards.

Why this level of precision actually changes things you use

GPS satellites carry their own atomic clocks (rubidium and cesium standards, historically, with more precise systems studied for future generations), and even tiny timing errors translate directly into position errors on the ground, because GPS receivers calculate location by measuring how long radio signals took to arrive from multiple satellites and multiplying by the speed of light. A clock error of just one microsecond corresponds to a position error of roughly 300 meters (about 1,000 feet), which is why the GPS system's design already has to account for general-relativistic time dilation caused by the satellites' altitude and orbital velocity, an effect on the order of tens of microseconds per day that would make GPS useless within minutes if left uncorrected.

Optical lattice clocks have also been used to directly measure gravitational time dilation, a core prediction of Einstein's general relativity stating that clocks closer to a massive body run measurably slower than clocks farther away, over remarkably small height differences. A landmark 2010 experiment by Chou, Hume, Rosenband, and Wineland at NIST, using aluminum-ion optical clocks, measured a detectable difference in clock rate between two clocks separated by just 33 centimeters (about 13 inches) in elevation, published in Science. Later work with strontium lattice clocks pushed this "chronometric leveling" technique to even smaller height differences, effectively turning an atomic clock into an instrument that can, in principle, measure changes in Earth's gravitational field, and by extension changes in elevation, at the centimeter scale.

Beyond navigation and relativity tests, ultra-precise clocks are used to search for tiny hypothetical variations in fundamental constants of physics, such as the fine-structure constant, over time. If a constant like this were found to drift by even a minuscule amount across years of comparison between independent optical clocks, it would be evidence for physics beyond the standard model, a search that is only possible because these clocks are stable enough to detect changes many orders of magnitude smaller than anything achievable with cesium-era instruments. None of these applications, GPS, relativity tests, or fundamental-constant searches, were the original motivation for building better clocks, but each has become dependent on the precision gains that optical lattice clocks made possible.

What people get wrong about ultra-precise clocks

A common misconception is that clock precision beyond, say, a millisecond is a purely academic curiosity with no practical application outside a physics lab. In reality, satellite navigation, synchronized financial trading networks that timestamp transactions to sub-microsecond resolution, and telecommunications infrastructure that relies on precisely synchronized network clocks all depend on extremely precise time standards, and historically, improvements in clock precision have translated into improvements in these applied technologies within years, not decades. The gap between "record-setting lab clock" and "technology quietly running in the background of daily life" has consistently proven shorter than people assume.

Another frequent mistake, discussed above but worth repeating because it trips up even careful readers, is conflating precision with accuracy. A clock can be extremely precise, ticking with remarkable internal consistency, while still being inaccurate if it has never been calibrated against an international time standard. The most precise clocks in the world are useful specifically because national metrology labs like NIST and BIPM continuously compare them against each other and against the official SI second, ensuring that precision and accuracy track together rather than diverging.

A third misconception is assuming that once a clock reaches parts-in-10^18 precision, further improvement stops mattering because "no one needs more precision than that." In practice, each further gain in clock stability has historically unlocked a new class of measurement, from testing general relativity at the centimeter scale to searching for drifting fundamental constants, applications that were not even conceivable when cesium fountain clocks were the state of the art. There is no obvious point at which clock precision stops being scientifically useful, which is part of why national labs keep funding the next generation of optical clock research even after crossing thresholds that would have seemed unimaginably good a generation earlier.

The takeaway

Optical lattice clocks, led by strontium clocks developed at JILA and NIST, achieve their record precision by trapping thousands of laser-cooled atoms nearly motionless in a lattice of laser light and counting an extremely high-frequency optical transition instead of a lower-frequency microwave one, pushing systematic uncertainty down toward 8 x 10^-19, roughly one second of error over 15 billion years. That leap in precision is not a record for its own sake. It is already reshaping GPS accuracy, enabling direct tabletop tests of general relativity over centimeter height differences, and pushing the International Bureau of Weights and Measures toward a formal 2030 redefinition of the SI second itself, away from cesium and toward optical atomic transitions, the biggest change to the definition of time since 1967.

---

Read next:

• The Coldest Laboratory Temperatures Ever Reached
• The Most Powerful Particle Accelerators
• The Fastest Human-Made Objects Ever
• How Pulsars Keep Such Accurate Time

FAQ

Q: What is the most precise clock ever built?

An optical lattice clock developed by JILA and NIST using strontium atoms, which loses or gains less than a second over roughly 15 billion years. The most recent generation of these clocks has demonstrated systematic uncertainty near 8 x 10^-19, among the lowest ever reported for any clock (arXiv:2509.13991).

Q: How does an optical lattice clock work?

It traps thousands of laser-cooled atoms nearly motionless inside a lattice formed by intersecting laser beams, then measures an extremely high-frequency optical transition between two atomic energy levels, counting these oscillations far more finely than a microwave-based cesium clock can.

Q: Why are optical lattice clocks more precise than cesium atomic clocks?

Optical transitions oscillate at frequencies roughly a hundred thousand times higher than the microwave transition cesium clocks use, and a higher oscillation frequency allows much finer subdivision of a second, the same way finer ruler markings allow more precise length measurements.

Q: Will the official definition of a second change because of these clocks?

Likely yes. In 2022, the General Conference on Weights and Measures set a roadmap toward redefining the SI second using optical clocks by 2030, contingent on at least three independent optical clock systems demonstrating uncertainty below 2 x 10^-18, a bar multiple labs had already met by 2025.

Q: Do these ultra-precise clocks have any practical use, or are they purely experimental?

They have direct practical applications already. GPS timing accuracy, tests of Einstein's gravitational time dilation over centimeter-scale height differences, and searches for drift in fundamental physical constants all rely on clock precision that only became achievable with optical lattice and optical ion clocks.