Day 9: A Clock You Can Use to Measure a Mountain

Yesterday I said cesium was about to be deposed. Today’s the deposition.

The successor is the optical lattice clock. It is so precise that, it can detect the altitude difference between two desks in the same office.

That last sentence sounds crazy but Its what optical clocks do.

Here’s how.

More ticks per second

The trick of the optical lattice clock is simple in concept: use a finer ruler.

A cesium atom oscillates at about 9 × 10⁹ Hz (~9 GHz, microwave). That’s how many “ticks” per second the clock has to count to keep time.

A strontium-87 atom has a clock transition at about 4 × 10¹⁴ Hz (~430 THz, visible light). That’s 50,000 times more ticks per second.

Same second. More tick marks on the ruler.

Faster clock makes better time.

Faster clocks are better than slower ones.

Faster clocks make for better timekeeping.

See the difference and why we need to upgrade the second?

Today’s best optical clocks drift about one second over the age of the universe. 13.8 billion years. Our best cesium clocks are nowhere near that, drifting the same second in roughly 300 million years instead. About a hundred times worse.

The magic wavelength trick

So if optical is obviously better, why didn’t we do this first?

Because it’s hard. Two reasons, really.

First, you can’t just point an optical clock at a free-floating atom. Atoms in flight have all sorts of velocity, Doppler-shifting the frequency you measure. You have to hold them still. But the only way to hold an atom still is to grip it with something, and gripping it with a laser changes the energy levels you’re trying to measure. Catch-22.

Second, even if you could pin the atom, you couldn’t count that fast. Microwave at 9 GHz fits inside what electronics can divide down and tick off. Visible light at 400+ THz does not. We had no way to count optical-frequency oscillations until the optical frequency comb showed up in 1999, work that won Hänsch and Hall the 2005 Nobel.

In 1967 when the CGPM picked cesium, none of this existed. The laser was seven years old. Laser cooling wouldn’t be demonstrated until the late 1970s. Frequency combs were 30+ years away. Cesium at 9 GHz wasn’t the best clock you could imagine. It was the best clock you could build.

The fix came in 2003 from a Japanese physicist named Hidetoshi Katori. He proposed trapping the atoms in a standing-wave laser pattern, a 3D optical lattice, like an egg crate made of light, and tuning the lattice laser to a specific frequency called the magic wavelength.

At the magic wavelength, the lattice light affects both energy levels of the clock transition by exactly the same amount. The trap is invisible to the clock. The atoms are held still, but the energy levels they emit at are unperturbed. You get to measure the atomic transition cleanly while the atom sits frozen in midair.

Optical lattice clocks are a beautiful piece of physics. Do not ask me about the math behind them.

Which atom?

Two main candidates have been thinking about for the title of “next SI second”:

Strontium-87. Run in optical lattices by labs around the world: NIST/JILA in Boulder, NPL in the UK, RIKEN in Japan, SYRTE in Paris. About 1 part in 10¹⁸ uncertainty. Currently the favorite.
Ytterbium-171. Comparable accuracy, different sensitivity profile (different ways the clock can go wrong, which is good for cross-checking against strontium).

There are also trapped-ion optical clocks, which hold a single ion in an electric field instead of a lattice of neutral atoms. The aluminum-ion logic clock at NIST has hit about 1 part in 10¹⁹. That’s one second of drift in 300 billion years. The amount of drift is older than the universe itself. That sounds pretty good to me.

What this gets you

Optical clocks are sensitive enough to see general relativity at human scale.

What does that mean? It means you can put two clocks on different shelves and watch one tick slower than the other.

Einstein said clocks deeper in a gravitational field run slower than clocks higher up. Near Earth, the effect is about 1 part in 10¹⁶ per meter of altitude. A clock on the floor runs slower than a clock on the table. For most of human history this was a theoretical curiosity. Now it’s a measurement.

In 2022, a JILA strontium clock measured a 1 millimeter height difference as a frequency shift. One millimeter. The thickness of a credit card. The clock could tell which side of the card it was sitting on, from gravitational time dilation alone.

This opened a new field: relativistic geodesy. Put an optical clock at two locations, compare frequencies, and you’ve directly measured the gravitational potential difference between them. Detect underground oil, magma movement, ice sheet melt, anything that shifts mass around the planet.

Cesium can’t do this.

The next “second”

The BIPM is planning to redefine the SI second around an optical transition by 2030. Same dance as 1955: measure the new number against the current cesium standard, freeze it at that precision, declare it the new second.

Four layers of backwards-compatibility instead of three. The fingerprints accumulate.

Where this goes

We’ve now spent three days on how a second is built. The next question is the one the calendar dodges: which second?

You need a reference point. A zero. An origin.

Tomorrow we’ll talk about where you start counting from, and why astronomers picked noon on January 1, 2000.

Sources

I’d appreciate a follow. You can subscribe with your email below. The emails go out once a week, or you can find me on Mastodon at @[email protected].

Jun 1, 2026 / Science / Time / Physics / 30daysoftime / Metrology