In October, a team in Vienna built the first self-stabilizing clock that ticks on the nucleus of an atom instead of its electrons. It is not the world's best clock. It may turn out to be the world's most sensitive instrument for catching nature changing its own rules.
| Feature: The Nucleus Ticks | 3 |
| Breakthroughs | 4 |
| So What? + Field Notes | 5 |
| Against the Grain | 6 |
| Signals | 7 |
| By the Numbers | 8 |
| The Long View + Sources | 9 |
| Next Issue | 10 |
Every great instrument is first sold as an improvement and later used as a discovery. The telescope was marketed to sailors. The cesium clock was built to keep better time and ended up running GPS. So when a lab announces a clock that is, today, less precise than the leaders, the useful question is not "how good is it?" but "what is it unusually sensitive to?"
This week's answer is surprising. A nucleus is more than ten thousand times smaller than an atom and is shielded from the stray fields that plague electrons. But the physics inside it is a tug-of-war between two of nature's strongest forces, so if the "constants" of nature ever drift even slightly, the nucleus should feel it first, and loudest.
That is this issue's lens: a good sensor is a question you can ask the universe. Read the feature, then the contrarian column, and decide for yourself whether Vienna built a clock or a lie detector for physics.
— The Editors
Each Friday: one deep dive, one contrarian read, three signals from other fields, and the numbers worth remembering. Every figure is traced to a source on page 9. Where a result is a prototype, a simulation or a mouse study, we say so.
Vienna has built the first nuclear clock that corrects itself. Today it is a prototype. Tomorrow's physics may run on it.
Every clock you have ever trusted is a metronome for electrons. Your phone's quartz hums at 32,768 hertz; the atomic clocks behind GPS count the microwave flutter of cesium's outer electrons; the newest optical clocks trap single ions or atoms and tap their electrons with laser light. In every case the pendulum is an electron, and electrons are fragile: nudge them with a stray magnetic field, a warm wall, a passing electric charge, and the beat wobbles. The history of timekeeping is the history of shielding electrons from the universe.
Why not tick on something the universe can barely disturb? The nucleus is the obvious candidate: it sits at the heart of the atom, more than 10,000 times smaller than the atom itself, buried under its own electron cloud. The trouble is energy. Nuclei normally need X-rays or gamma rays to change state, light for which we have no precise, stable lasers.
One nucleus is the exception. Thorium-229 has two energy states separated by an unusually small gap, about 8.4 electron-volts, which corresponds to ultraviolet light near 148 nanometers. That is reachable with a laser. Physicists suspected this for decades, and for decades nobody could find the exact frequency.
Finding it was like hunting for one radio station across a band nobody could scan. In April 2024 the Vienna group, working with Germany's PTB, finally excited the thorium nucleus with a laser. That was the first act. A "clock" needs more than one flash of light: it needs a laser locked to the nucleus, and a way to keep it locked.
By fall 2024, a team at JILA in Colorado led by Jun Ye, with collaborators including TU Wien, used a frequency comb to drive the transition inside a thorium-doped calcium fluoride crystal and compared it to a strontium clock. The ratio came out as 4.707072615078, measured to twelve digits. A nuclear clock was no longer a theory. It was an instrument with a number.
Until this month, the thorium nucleus needed a babysitter: an ordinary optical atomic clock to keep the laser steady. That is like testing a new watch by staring at an old one. The new paper, A thorium-229 optical nuclear clock with feedback loop, published in Nature by Thorsten Schumm's group at TU Wien, removes the crutch.
The principle is elegant. The nuclei absorb laser light only at one exact frequency. If the laser drifts, absorption drops. The system notices the dip and nudges the laser back. The nucleus itself is the reference, and the feedback is the clock.
The prototype reached a relative precision of about 10⁻¹⁵, equivalent to roughly one second of error over 30 million years. The authors are careful: that is an extrapolation from measured precision, not a 30-million-year test. It ran unattended for more than 24 hours.
Conventional optical clocks laser-cool and trap atoms in ultra-high-vacuum apparatus the size of a room. The thorium clock uses nuclei embedded in a crystal made at TU Wien. Billions of nuclei in a solid can be interrogated at once. That is a very different engineering path, one that points toward something compact.
Theory suggests the thorium transition is far more sensitive to changes in the fine-structure constant, the number that sets the strength of electromagnetism, than any atomic clock. One calculation puts the enhancement factor near 8,200, compared with about 0.1 to 6 for atomic clocks: roughly three orders of magnitude.
Nobody is buying a nuclear clock this year. Precision is below the best optical clocks, the laser light sits in the vacuum ultraviolet where beams are weak and awkward, and the crystals are still being perfected. Schumm's own words: a "fantastic result" for a first prototype.
So where does the value show up first? Not in your phone. Follow the path of the cesium clock: first national metrology labs, then telecom networks and navigation, then everyday infrastructure. The near-term customers are research institutions testing fundamental physics, whose appetite for sensitivity is limitless.
The mid-term prize is a solid-state reference: robust, potentially portable, and perhaps less demanding than laser-cooled atoms. Whoever masters thorium crystal growth and deep-UV lasers owns the supply chain: crystals, lasers, locking electronics. These are components, and component businesses are where early hardware revolutions usually pay.
The honest timeline: papers and pilots in the next years; instruments for specialist labs after; mass markets only if the engineering collapses in cost the way atomic clocks eventually did.
1 · Shine. A laser tuned near 148 nm hits thorium-229 nuclei embedded in a crystal.
2 · Listen. The nuclei absorb light only at one exact frequency. Drift off it and absorption falls.
3 · Correct. Electronics read the dip and retune the laser. The locked laser is the clock tick.
Isomer. A long-lived excited state of a nucleus.
Frequency comb. A laser whose light is a ruler of evenly spaced colors, used to compare clocks.
Fine-structure constant. A pure number (about 1/137) setting the strength of electromagnetism.
Relative precision. Fractional error: 10⁻¹⁵ means one part in a quadrillion.
The consensus asks: will nuclear clocks beat atomic clocks at telling time? The better question is what a nuclear clock can tell us that no clock ever has.
Treat the thorium nucleus as a detector that happens to emit a tick. Its transition is a balance between the strong nuclear force and electromagnetism. If the fine-structure constant, or the coupling to hypothetical dark-matter fields, shifts even slightly, the balance shifts by far more than in an electron transition. That enhancement of order 10³ means a modest clock can outperform a magnificent atomic clock at the one job that matters: detecting new physics.
The business logic flips. Don't sell time; sell sensitivity. The first customers are the dark-matter and fundamental-constants community, with grants that don't care about cost per unit time. The ticks are merely the readout.
It is not the best clock. At 10⁻¹⁵ the prototype trails the leading optical clocks, which the metrology body BIPM describes as 10 to 100 times more accurate than cesium and already the frontrunners for redefining the second at the 2030 general conference.
Sensitivity cuts both ways. A clock that responds strongly to new physics responds strongly to crystal defects, temperature and radiation damage. Separating a drifting constant from a drifting crystal is the hard part.
The enhancement is theoretical. The ~8,200 factor is a calculation; direct measurement demands precision the prototype hasn't reached.
Infrastructure is thin. Deep-UV lasers at 148 nm are rare and weak, and thorium-229 is scarce.
So the bet is real but long. If it pays, it won't look like a better wristwatch. It will look like a null result, or a drift, that rewrites a textbook.
Princeton Plasma Physics Laboratory theorists argue fusion ignition may be easier to reach by heating plasma first and densifying it later, finding a low-energy "Cordey saddle" near Q=5 in an idealized plasma. The catch: it is theory only, and tungsten contamination at 1 part in 10,000 roughly doubles the required pressure.
USC researchers found a chemical cocktail that lets immature immune progenitor cells (GMPs) multiply for long periods, then engineered them with CAR receptors. In mice they settled in bone marrow and slowed blood and solid tumors; an "off-the-shelf" path is suggested. Preclinical, and partly industry-funded.
At the Indian Institute of Science, machine learning expanded a CO2-to-methanol reaction network on copper from 152 to 9,389 reactions, predicting about 40 times more conversion than the small model and revealing hydrogen moving as intact H2 in some steps. One catalyst, one reaction: a map, not yet a plant.
Relative precision of the first self-stabilizing nuclear clock
How much smaller a nucleus is than its atom
Wavelength of the thorium-229 transition (~8.4 eV)
Predicted enhancement of sensitivity to the fine-structure constant (atomic clocks: 0.1 to 6)
Clocks feeding the world's UTC time scale
Year the General Conference on Weights and Measures may redefine the second
More CO2 conversion predicted by IISc's 9,389-reaction model
A fusion theory that begins with heat instead of density, immune cells that can be grown without exhausting themselves, a chemistry map expanded sixty-fold: this week's headlines share a quiet habit. Each changes the order of operations, not the ingredients. Heat then compress. Multiply then specialize. Map then predict.
The nuclear clock does the same to metrology. We spent a century improving the pendulum, and the most interesting move may be to pick a different one, one so isolated that what shakes it is news. The question worth carrying out of this issue: what in your own field is measured by a fragile pendulum when a more sensitive one exists?
What does your industry measure with a fragile instrument, and what would it learn from a more sensitive one?
Researched, written & designed with Claude. Typeset in Poppins & Lora on the Anthropic palette.