The ALICE experiment, designed for the LHC heavy-ion programme, is particularly well-suited for the detection and study of very high-energy cosmic events. The apparatus is located in a cavern 52 m underground, with 28 m of overburden rock, offering excellent conditions for the detection of muons produced by the interaction of cosmic rays in the upper atmosphere.

During pauses in the LHC operation (no beam circulating) between 2010 and 2013, the experiment collected cosmic-ray data for 30 days of effective time. Specific triggers were constructed from the information delivered by three detectors: ACORDE (A COsmic Ray DEtector), TOF (Time-Of-Flight) and SPD (Silicon Pixel Detector). The tracks of muons crossing the ALICE apparatus were reconstructed from the signals recorded by the TPC (time projection chamber). The unique ability of the TPC to track events with a large number of muons, unimaginable with standard cosmic-ray apparatus, has opened up the opportunity of studying the muon multiplicity distribution (MMD), and in particular rare events with extremely high muon density.

Atmospheric muons are created in extensive air showers that originate from the interaction of primary cosmic rays with nuclei in the upper atmosphere. The MMD has been measured by several experiments in the past, in particular by the ALEPH and DELPHI detectors at LEP. Neither of these two experiments was able to identify the origin of the high-multiplicity events observed. In particular, ALEPH concluded that the bulk of the data can be described using standard hadronic production mechanisms, but not the highest-multiplicity events, for which the measured rate exceeds the model predictions by over an order of magnitude, even when assuming that the primary cosmic rays are solely composed of iron nuclei.

The MMD measured from the data set collected by ALICE exhibits a smooth distribution up to a muon multiplicity of around 70. At larger multiplicities, five events have been detected with more than 100 muons, confirming the detection of similar events by ALEPH and DELPHI. The event with the highest multiplicity (276 muons) shown in the figure corresponds to a density of around 18 muons/m2.

These particular events triggered the question whether the data can be explained by a standard cosmic-ray composition, with usual hadronic-interaction models, or whether more exotic mechanisms are required. To answer these questions, as a first step, the MMD has been reproduced at low-to-intermediate multiplicities using the standard event generator CORSIKA, associated with QGSJET as the hadronic-interaction model. CORSIKA simulates the development of extensive air showers following the collision of a cosmic ray with the nuclei in the atmosphere. The shower particles are tracked through the atmosphere until they reach the ground.

These simulations successfully described the magnitude and shape of the measured MMD in the low-to-intermediate multiplicity, so the same model was then used to explore the origin of the five high-multiplicity events. This investigation revealed that these rare events can only be produced by primary cosmic rays with energies higher than 10,000 TeV. More importantly, the observed detection rate of one event every 6.2 days can be reproduced quite well by the simulations, assuming that all cosmic rays were due to iron nuclei (heavy composition). For proton nuclei (light composition) the expected rate would be of one event every 11.6 days.

Hence, for the first time, the rate of these rare events has been satisfactorily reproduced using conventional hadronic-interaction models. However, the large error in the measured rate (50%) prevents us from drawing a firm conclusion on the exact composition of these events, with heavy nuclei being, on average, the most likely candidates. This conclusion is in agreement with the deduced energy of these primaries being higher than 10,000 TeV, a range in which the heavy component of cosmic rays prevails.

The data collected this year will be extremely valuable in performing more detailed analysis using all of the measured variables and different hadronic-interaction models, and will therefore allow further progress in comprehending the origin of high-muon-multiplicity events.