CERN's ATLAS Recreates Cosmic Ray Showers With First Proton Oxygen Collisions

Tens of kilometres above the surface of the Earth, high-energy particles arriving from deep space constantly slam into the atmosphere, setting off cascades of secondary particles that rain down toward the ground. Physicists call these cascades cosmic-ray air showers, and they are one of the few ways to study particles accelerated far beyond anything human machines can produce. Approximately one such particle passes through a human head every second, yet the cosmic rays that produce these showers are still not fully understood.

Cosmic rays were discovered more than a century ago by the physicist Victor Hess, who flew hot-air balloons to show that ionising radiation increases with altitude, a result that earned him a share of the 1936 Nobel Prize in Physics. Because most cosmic rays carry an electric charge, magnetic fields bend their paths as they travel, so astronomers cannot trace their arrival directions backward the way they can with light. They must instead combine arrival directions, energies and composition, and that work depends on understanding exactly what happens in the first collision with the atmosphere.

Today astrophysicists study these showers with arrays of hundreds of specialised cameras spread over miles, such as the Telescope Array in Utah, and then rely on computer simulations to interpret the data. Those simulations depend on properties of the strong force that are difficult to model accurately, and they disagree with one another. That gap is the reason a new measurement from the Large Hadron Collider matters.

The ATLAS Collaboration at CERN reported on 14 September 2026 in Physical Review Letters the first measurement of proton-oxygen collisions, recreating in the laboratory the atmospheric collisions that cosmic rays trigger high above Earth. The analysis covered about 246 million collisions and 5.11 billion particle tracks. CERN had introduced oxygen beams into the LHC for the first time in 2025, and the pioneering proton-oxygen collisions took place on 1 July 2025, recreating cosmic rainstorms in the laboratory for a few days.

Key Facts

In the experiment, a proton beam acted as the incoming cosmic ray while an oxygen beam played the role of Earth's atmosphere, which is composed primarily of nitrogen and oxygen. The LHC accelerated oxygen nuclei to 3.4 teraelectronvolts per nucleon while sending 6.8 TeV protons in the opposite direction, so that their collisions produced a centre-of-mass energy of 9.62 TeV per nucleon pair. ATLAS recorded data corresponding to 634 inverse microbarns of integrated luminosity.

phys.org reported on September 14, 2026 that the ATLAS Collaboration analysed the world's first laboratory collisions of oxygen atoms with protons, and that the article describing the work was authored by lead researcher Jesse Liu, an assistant professor of physics at New York University. ATLAS itself is the size of a football field and is run by thousands of scientists, and its high-speed silicon camera can take more than 200 million photos a day. The team measured how many particles the collisions created and at what energies they flew out, achieving more than 10 times the precision of computer-model predictions. Liu and physicist Lydia Beresford had made the case for reconfiguring existing CERN instruments for this purpose.

The Brighter Side of News reported on September 15, 2026 that no major cosmic-ray interaction model successfully reproduced all the observed particle behaviour. The article reported a fiducial proton-oxygen cross section of 396 millibarns and an inferred inelastic proton-air cross section of 406 millibarns, compatible with earlier cosmic-ray measurements. Comparing detailed particle production against seven models, the team found that none reproduced all measurements. The models disagreed on particle multiplicity, with some predictions differing by roughly an order of magnitude, while the Angantyr model inside the PYTHIA framework came closest on transverse-momentum and pseudorapidity distributions.

The laboratory collision corresponds to a cosmic-ray proton of about 49 petaelectronvolts, or 49 quadrillion electronvolts, striking the atmosphere. CERN reported on April 21, 2026 that the new ATLAS measurements achieve a precision level of a few percent, significantly improving knowledge of proton-oxygen collisions, and that an event display from the July 2025 run shows nineteen charged-particle tracks.

Liu said that the work renews links between particle physics and high-energy astrophysics and will help determine how many high-energy cosmic rays are made of hydrogen versus heavier atoms. He added that the team now has a better understanding of how particles from objects more exotic than the Sun interact with our atmosphere. Cigdem Issever of Humboldt University of Berlin said that the previous models for this type of interaction are actually very inaccurate.

Analysis

What this really means is that the Large Hadron Collider has quietly become an astrophysics instrument. ATLAS was built primarily to study fundamental interactions, including the Higgs boson, but its silicon tracking detector provides an exceptionally detailed picture of the debris produced when a proton meets an oxygen nucleus. That is the same first step that begins every cosmic-ray air shower above the atmosphere, reproduced under controlled conditions with a known beam energy and a known collision point.

The precision is the headline. Daily Galaxy reported on September 16, 2026 that the resulting measurements reached precision more than 10 times greater than existing model predictions, and that silicon tracking gave an unusually detailed view of how these particle cascades begin. When seven widely used simulations are compared against a single data set and none of them reproduces everything, the shortfall is not a small technical quibble. It is a direct signal that the modelling of the strong force inside air-shower codes is incomplete in ways that matter for real observatories.

That matters because small differences at the first interaction can propagate through a developing shower and change what ground-based observatories such as the Telescope Array in Utah eventually record. A model that miscalculates particle multiplicity, or the energies and angles at which particles emerge, will misread the energy and the composition of the primary cosmic ray. The ATLAS data give theorists a controlled experimental anchor for the hadronic-interaction models used to interpret air showers, which is precisely what the field has been missing.

The bigger picture here is that particle physics and high-energy astrophysics are being reconnected through a shared measurement. For decades the two communities worked with different tools and different uncertainties, one at colliders and one at observatories. A proton-oxygen run at the LHC provides the missing bridge, because it supplies a benchmark that is far more precise than anything the atmosphere alone can offer.

Why It Matters

Improved simulations could help determine whether incoming cosmic rays are protons, helium or iron nuclei, and where they originate, including the transition between Milky Way and extragalactic sources. That composition question is central to the field, because candidate acceleration sites include supernova remnants and active galactic nuclei with supermassive black holes. Distinguishing among them requires knowing what the primary particles are, not only how much energy they carry.

Cosmic rays remain mysterious even though they were discovered over a century ago. The CERN account notes that they are still not fully understood despite the decades of balloon flights and ground arrays that followed Victor Hess. The new measurement does not solve the mystery on its own, but it removes a long-standing source of systematic error in the models used to attack it.

There is also an observational payoff. If the interpretation of air showers becomes more reliable, then the arrays of hundreds of specialised cameras spread over miles, including the Telescope Array, can be used with greater confidence to infer what arrived at the top of the atmosphere and from which direction.

Next Up

The immediate next step is theoretical. Because the ATLAS measurements reach a precision level of a few percent, theorists can use this input to refine their models of cosmic-ray interactions and to identify which parts of the strong-force description need the most attention. The Angantyr model inside PYTHIA, which came closest on transverse-momentum and pseudorapidity distributions, offers a natural starting point for that refinement.

The broader goal is to shed light on the mysterious high-energy particles arriving from the cosmos, and to decide how many of them are hydrogen and how many are heavier atoms. The publication, listed as G. Aad et al in Physical Review Letters with the DOI 10.1103/f3nk-5lt9, is the first proton-oxygen collision measurement of its kind, and it establishes the LHC as a controlled laboratory for the study of cosmic rainstorms.