Air-shower light hints at a changing cosmic-ray mix

Reading the shower, not the original particle

What kinds of particles make up cosmic rays at different energies? A new arXiv preprint by M. Yu. Ternovoy and colleagues approaches that question through extensive air showers: cascades of secondary particles produced when an incoming cosmic ray interacts with the atmosphere.

The researchers estimate where each shower reaches its maximum development, a quantity called Xmax. This is an atmospheric depth, measured in grams per square centimetre, rather than simply a height above the ground. They then use simulations to translate the average depth into an estimate of the incoming particles’ mass composition.

Their preliminary results suggest a changing mixture: heavier particles become more prominent up to about 30–50 petaelectronvolts, or PeV, followed by a shift toward lighter particles at higher energies. One PeV is a million billion electronvolts.

Two ways to read Cherenkov light

TAIGA-HiSCORE sits in Russia’s Tunka Valley. Its 120 optical stations cover approximately one square kilometre and record Cherenkov light from atmospheric showers. The analysis uses about 900 observing hours collected across four seasons between 2021 and 2025.

The team reconstructs shower depth in two ways. The first examines how sharply the light density falls with distance from the shower axis. Called the P-method, it uses the ratio of light densities at 80 and 160 metres, obtained from a fitted spatial distribution. This analysis includes 820,854 events between 1 and 30 PeV.

The second examines pulse duration. Its key measurement is the full width at half maximum: how long a pulse remains above half its peak strength. The researchers use stations between 270 and 330 metres from the shower axis to estimate the characteristic pulse width at 300 metres. This timing method includes 35,453 events between 10 and 250 PeV.

Neither measurement directly supplies a shower depth. The conversion relies on simulated showers, including the distortions introduced by detector electronics. The team generated approximately 4,000 showers initiated by protons, helium, nitrogen and iron nuclei, using two models of high-energy particle interactions.

A heavier mix, then a lighter one

To describe composition, the researchers calculate the mean logarithmic mass, written as ⟨ln A⟩. Here, A is the nucleus’s mass number. Larger values indicate a heavier mixture, but this average does not uniquely identify the proportions of individual elements.

The conversion compares the measured average shower depth with simulated proton and iron expectations, using the QGSJET-II-04 interaction model as the main reference.

The inferred composition becomes heavier between the cosmic-ray “knee” region and roughly 30–50 PeV. Around that upper range, the mean logarithmic mass approaches 2.6, a value characteristic of the carbon–nitrogen–oxygen group. That does not establish that these elements dominate: different mixtures can produce similar averages.

Above approximately 50 PeV, the timing measurements instead suggest a lighter composition. The authors report similar energy-dependent behaviour in published LHAASO measurements and reanalysed KASCADE data. These are comparisons presented in the preprint, not independent verification of its results.

Why the conclusion remains preliminary

The two reconstruction methods yield shower depths differing by approximately 10 grams per square centimetre in their overlapping energy region. The authors are investigating that difference with more detailed simulations of the array geometry and detector saturation near the shower axis.

Saturation also limits the spatial method above 30 PeV: close to the axis, recorded pulse amplitudes can no longer reliably represent the light signal.

A separate uncertainty enters when shower depth is converted into mass. At 50 PeV, the paper reports that switching from QGSJET-II-04 to Sibyll 2.3d raises the inferred mean logarithmic mass by about 0.35; using EPOS-LHC raises it by about 0.08. The same measured shower depth therefore does not imply a model-independent composition.

The study offers evidence for a turnover in the inferred cosmic-ray mixture, rather than a settled inventory of incoming nuclei. The authors explicitly describe the measurements as preliminary and subject to revision.