Composition studies based on the muon content of air showers measured by underground muon detectors

Ultra-high-energy cosmic rays (UHECRs) are among the most energetic particles observed in nature, yet their origin and mass composition remain poorly understood. At energies greater than approximately 1015 eV, direct detection becomes impractical, and primary particles are studied through the extensive air showers (EAS) they generate in the atmosphere. The muonic component of these showers provides a particularly sensitive probe of both primary mass and hadronic interactions at energies beyond the reach of terrestrial accelerators.
This thesis investigates the muon lateral distribution (MLD) and muon-related observables measured with the Underground Muon Detector (UMD) of the Pierre Auger Observatory
to enhance composition sensitivity and assess the consistency between measured data and simulations of EAS based on different high-energy hadronic interaction models.
The shape of the MLD is first characterised by direct comparisons between measured radial muon densities and full detector-level Monte Carlo (MC) simulations of the EAS,
constructed as weighted mixtures of primary nuclei (proton, helium, nitrogen, and iron) with fractions defined by composition models. ... mehrFor the first time, complete air-shower and
detector simulations are compared with UMD data, including the radial dependence of the muon distribution. A bin-by-bin analysis framework is developed, in which the comparison
between measurements and simulations is performed independently in each energy bin, to test whether the discrepancy can be described by a single energy-dependent multiplicative
scale factor applied to a reference mixed-composition model (AugerMix), constructed as a weighted combination of primary nuclei based on composition fractions inferred from
measurements of the depth of the shower maximum Xmax. The study is based on an extended UMD data set comprising data collected in 2018–2021 (Phase 1), together with data recorded in 2023–2024 (Phase 2) following a large-scale upgrade of the detector system, including improved front-end electronics. It includes, for the first time in this context, measurements obtained with the upgraded Phase 2 electronics. The comparison reveals a systematic energy-dependent muon deficit in simulations relative to data. This conclusion is found to be robust in all contemporary high-energy hadronic interaction models, and the radial dependence of the discrepancy remains consistent, within systematic uncertainties, with a global rescaling of the simulated muon densities. This supports the interpretation of the data-simulation mismatch as a predominantly normalisation effect and provides a consistent basis for the development and interpretation of Vbmax observable in subsequent correlation analysis.
Using reconstructed muon densities, mass-sensitive observables based on the spatial structure of the muon distribution are developed and optimised through merit-factor studies.
The statistical correlation between one of the optimised observables Vbmax and the depth of the shower maximum, Xmax, is investigated as a composition probe, exploiting the robustness of correlation-based methods against calibration and scaling uncertainties. The sensitivity of this approach is quantified for both fluorescence-detector measurements and high-statistics surface-detector scenarios, including realistic detector-resolution effects. Among the tested estimators, Kendall’s correlation coefficient provides the highest stability and discrimination power. The correlation-based framework is shown to constrain both the mean and the dispersion of the primary mass distribution, enabling discrimination between pure and mixed composition scenarios. Using the measured cosmic ray (CR) flux reported by the Pierre Auger Observatory and the statistical properties of Kendall’s coefficient, the minimum number of events required to reject the hypothesis of pure composition is derived. To further strengthen the reliability of the muon-based observables and to reduce potential biases associated with detector response, improved reconstruction techniques for the MLDFs are developed, based on both integrated signal measurements (continuous detector response) and hybrid combinations with particle-counting modes. Their performance is evaluated with MC simulations and compared with existing reconstruction approaches.
Although these improved reconstruction methods are not propagated in the primary composition analysis presented in this thesis, they provide a foundation for future high-precision muon measurements and refined correlation-based studies. Together, these studies establish a comprehensive framework linking muon measurements, reconstruction methodology, and correlation-based observables to constrain the mass composition of UHECRs and to test hadronic interaction models at ultra-high energies.