Physicists at CERN have recreated, for the first time, the collisions that trigger cosmic-ray air showers in Earth's atmosphere, and their results show that none of the computer models used to simulate these showers gets every detail right. The measurements, made by the ATLAS experiment at the Large Hadron Collider, are more than ten times more precise than the differences between the models, offering a path to improving the simulations that underpin cosmic-ray astronomy.
Cosmic rays are energetic particles, mostly protons, that arrive from outer space and strike the atmosphere at nearly the speed of light, setting off cascades of secondary particles. «Cosmic-ray air showers are sprays of particles raining down from high in the sky,» said Jesse Liu of New York University, lead author of the study. The most energetic cosmic rays carry a hundred billion-billion electron volts or more, far beyond the reach of any human-made accelerator. At such energies, only about one particle hits each square kilometer of Earth's surface per century, so scientists rely on detecting the extensive air showers they produce.
Interpreting those showers depends on computer simulations, but the strong nuclear force that governs the underlying collisions is notoriously difficult to calculate. The models therefore rely on data from accelerators, yet they disagree widely on how showers form. Until recently, the LHC's beams consisted only of protons or heavy nuclei such as lead, not the light nitrogen and oxygen nuclei found in air. That changed in July 2025, when the LHC was reconfigured to collide protons with oxygen nuclei, mimicking cosmic rays hitting the atmosphere. The ATLAS detector recorded the charged particles produced.
The team measured how many particles are created, their energies and angles, and the cross section, which indicates how often collisions occur. The measured cross section lies at the low end of model predictions, agreeing with only two of the seven models tested. The result also allowed the physicists to infer the cross section for protons colliding with air, a quantity previously measured at such high energies only by cosmic-ray observatories, and it agrees with earlier air-shower measurements. «Our data show that no model correctly describes the number of particles created in these collisions,» Liu said. Some models misjudge the frequency of rare, particle-rich collisions by a factor of ten. A model called Angantyr best reproduces the particles' energies and directions, but no model describes all results consistently.
Karl-Heinz Kampert of the University of Wuppertal, who was not involved in the study, called the cross section «a key parameter» because it determines the average depth at which cosmic rays first interact in the atmosphere, affecting essentially all air-shower measurements needed to infer the mass of the incoming particle. He also noted a blind spot: multipurpose detectors like ATLAS and CMS are optimized to track particles flying out at relatively large angles to the beam, while most of the collision energy escapes close to the beam pipe at very small angles. Specialized experiments such as LHCf are designed to fill that gap.
The next step, Liu said, is to apply the data to improve the computer models of cosmic-ray showers. Physicists will then check whether this eases the «muon puzzle» — facilities such as the Pierre Auger Observatory in Argentina see more muons in air showers than simulations predict. «Our proton–oxygen data are an important step toward resolving these puzzles and help decipher cosmic mysteries, namely what high-energy cosmic rays are made of and where they come from,» Liu added. Kampert said model builders have been eagerly awaiting the data and have started improving their models. Because Auger results suggest the most energetic cosmic rays are mostly heavier nuclei such as carbon, nitrogen, oxygen and silicon, he called data from oxygen–oxygen or nitrogen–nitrogen collisions «another important step forward.»
