A New Muonium Beam Brings a Gravity Test Closer

Physicists have produced a muonium beam for a future gravity test. The experiment establishes a source of exotic atoms, not a measurement of their free fall.

What the new beam experiment measured

J. Zhang and colleagues report the result in Nature Physics, published on 14 September 2026. Muonium emerges from superfluid helium. Detectors register positrons from subsequent muon decays.

Comparing these detection times with simulations gives a longitudinal speed of 2,180 metres per second. Statistical uncertainties span −130 to +160 metres per second at 90% confidence. These describe uncertainty in the fitted speed, not the beam’s velocity spread.

The superthermal model, with a narrow spread around directed motion, fits the timing distributions better than the thermal alternative. Percent-level gravity sensitivity remains a projection.

Why muonium gives gravity a different test

Muonium contains an electron bound to a positive antimuon. Its constituents carry opposite electric charges, leaving the atom neutral. The antimuon belongs to the second particle generation in the Standard Model.

That composition matters. Hydrogen contains a proton, with quarks and strong-interaction structure. Muonium contains two leptons. It provides a different physical system for testing whether gravity accelerates all freely falling bodies equally.

The MAGE collaboration’s 2018 proposal developed this motivation and an interferometer design. Its projected sensitivities were design estimates. The proposal did not report a gravitational acceleration.

A measurement would concern the whole atom. Calling it a direct experiment on an isolated antimuon would erase the electron and binding energy. Their contributions matter when translating an atomic result into constraints on a particular theory.

Gravity has only microseconds to act

The positive muon’s mean lifetime is about 2.2 microseconds. The MuLan experiment measured it to one part per million. This is a mean lifetime, not a deadline shared by every atom.

Consider a simple calculation. Assume ordinary terrestrial gravity, negligible initial vertical velocity and a flight time equal to that lifetime:

Using g = 9.81 metres per second squared gives a fall of roughly 24 picometres. This estimate illustrates the scale; it is not a measured displacement.

Waiting twice as long quadruples the displacement. But exponential decay reduces the surviving population from about 37% to 14%. These survival fractions follow from exp(−t/τ), before detector losses. Longer observation improves the gravitational signal while removing atoms from the sample.

Interference can reveal a tiny displacement

An atom interferometer converts small changes in motion into shifts of an interference pattern. The MAGE design uses three gratings. Diffraction creates paths whose amplitudes recombine, and gravity changes the pattern’s position relative to the apparatus.

This uses the same amplitude-addition rule behind the double-slit experiment. Detecting many atoms reveals a pattern that no single detection can establish.

The proposed apparatus must also distinguish gravity from instrumental motion. Grating alignment, mechanical stability and calibration therefore belong in the measurement, alongside atom counts. An intense source cannot substitute for control of those effects.

Antihydrogen has already passed a free-fall test

In 2023, the ALPHA collaboration observed antihydrogen responding to Earth’s gravity. Its best-fit downward acceleration was:

The first uncertainty combines statistical and systematic contributions. The second comes from simulation. The result agrees with ordinary downward acceleration within its uncertainties. It does not establish that antihydrogen falls exactly 25% more slowly than hydrogen.

Antihydrogen contains an antiproton and a positron. Muonium contains an antimuon and an electron. Both are neutral, but they test different constituents.

Calling a future muonium experiment the first test of antimatter falling would therefore misstate the history. The more specific opportunity concerns a neutral atom containing a second-generation antilepton.

Precision and particle coverage answer different questions

Ordinary matter already supports much tighter equivalence-principle tests. The MICROSCOPE mission compared titanium and platinum test masses. Its final result found no difference at a sensitivity of a few parts in 10¹⁵.

That result constrains composition-dependent acceleration for those materials. A much less precise measurement with muonium would explore different particle content. Whether it adds a stronger constraint depends on the proposed interaction and its couplings.

Precision alone cannot rank every gravity experiment. An extremely accurate comparison of two materials and a first measurement with an exotic atom serve different purposes. Neither makes the other unnecessary.

Spectroscopy offers another route to useful physics

Muonium also provides a clean system for atomic spectroscopy. Without an extended proton, its energy levels avoid proton-size uncertainties. Quantum electrodynamics still requires recoil, radiative and other corrections.

A 2023 calculation by Irene Cortinovis and colleagues examined the 1S–2S transition. It identified the electron-to-muon mass ratio as an important source of uncertainty. Better spectroscopy can improve that ratio and test bound-state QED.

This offers a useful editorial test for the next result: ask which observable improved. More atoms, a narrower spectral line and a smaller uncertainty on a fundamental constant are related milestones. They are not interchangeable claims.

The gravity question remains open for muonium. A convincing departure from ordinary free fall would require calibrated controls and independent confirmation. Agreement would extend the tested domain of a principle that already works extraordinarily well. The next decisive number is an acceleration, not a headline about overturning Einstein.