18 SEP 2026 — Physicists in Switzerland have built a cold, bright beam of an atom that does not exist in nature, in order to drop it and watch what gravity does.
The beam is muonium, and the paper describing it appeared in Nature Physics on 14 September. The measurement it is built for has not been made yet.
What they built
Muonium is an exotic atom: a positively charged antimuon with an ordinary electron bound to it. It behaves chemically much like hydrogen and survives for microseconds, which is long enough to steer and far too short for most of the techniques that work on ordinary atoms.
The team at the Paul Scherrer Institute in Villigen and ETH Zurich made their beam by firing antimuons into superfluid helium. The particles slow down, pick up electrons to form muonium, and are then expelled vertically as the chemical potential of the helium converts into kinetic energy. The atoms leave at a mean speed of roughly 2,180 metres per second, travelling in nearly parallel directions with a narrow spread.
Lead author Jesse Zhang describes the mechanism without ceremony: "So we're using the chemical potential as an atomic cannon."
Why a second generation matters
Ordinary matter is built from first-generation particles. The muon is a second-generation particle — the same charge as an electron, around two hundred times the mass, and no known role in everyday matter.
The equivalence principle holds that gravitational and inertial mass are the same thing, so everything falls at the same rate regardless of composition. It has been tested to extraordinary precision on ordinary matter. It has never been tested on a second-generation particle, because nobody could hold one still enough for long enough.
Anna Soter, who leads the work, states the goal simply: "We want to measure the gravitational interaction of the muon." If the muon falls differently, the discrepancy would point to a force nobody has catalogued.
How the measurement works
The measurement will use interferometry. The beam passes through a grating arrangement that produces an interference pattern, and gravity acting on the atoms during their flight shifts that pattern by a small amount. Measuring the shift measures the acceleration.
The beam's properties are therefore the result, not a preliminary step. An interference measurement needs atoms that are slow, numerous and travelling in nearly the same direction. Muonium's microsecond lifetime means each of those properties has to be achieved immediately, not by cooling over time.
What would count as a result
Nothing has been measured yet. Method testing is planned for this year; the gravity measurement itself is two to three years away.
When it arrives, the interesting outcome is agreement. A muon that falls exactly like ordinary matter confirms the equivalence principle in a regime where it had never been checked, which is a real result and will be reported as a non-event. A disagreement would be far louder and would require the kind of scrutiny that a single 2.6-sigma event in a dark matter detector received earlier this month — which is to say, treated as a hint and not a discovery until it survives replication.
What to watch
Watch for the method test first. If the interferometer can resolve a known acceleration at this beam intensity, the gravity run becomes credible, and that should be reportable well before the measurement itself.
The beam is an instrument in its own right, too. A high-intensity muonium source is also useful for laser spectroscopy, which tests quantum electrodynamics in a system with no nuclear structure to complicate it. The paper names both uses, and the spectroscopy results may well arrive first.
Watch the language on this one. Some coverage has described the beam as antimatter. Muonium contains an antimuon, but it is not an antimatter atom in the sense of antihydrogen, and the experiment is not an antimatter gravity test of the kind CERN has been running. The distinction will matter when results appear and someone reports them as confirming or contradicting work that was measuring something else.
View author profile → · Editorial policy
