Hybrid radio and particle detection of air showers: potential for ultra-high-energy photon identification

Technical Summary: Hybrid Radio and Particle Detection of Air Showers for Ultra-High-Energy Photon Identification

Problem Statement
The identification of ultra-high-energy (UHE) photons (energies $> 10^{17}$ eV) is a critical challenge in cosmic-ray physics, as these particles serve as messengers to pinpoint sources of UHE cosmic rays and constrain their origins. However, the search is hindered by a dominant background of hadronic cosmic rays (primarily protons). Traditional discrimination methods rely on differences in shower development, such as the muon content (proton-induced showers are muon-rich, while photon-induced showers are muon-poor) and the depth of maximum development ($X_{max}$).

Existing searches have largely focused on near-vertical showers using particle detectors (scintillators or water-Cherenkov tanks) often complemented by fluorescence or underground muon detectors. However, for very inclined showers (zenith angles $\theta > 60^\circ$), the electromagnetic and hadronic components are absorbed by the atmosphere before reaching the ground. Consequently, ground-based particle detectors only register the surviving muonic component. While radio detection has advanced significantly for these inclined showers, distinguishing photon-induced events from proton-induced events remains difficult because both produce radio signals, and the particle signal for photons is often too sparse to trigger standard particle detectors efficiently.

Methodology
This work proposes a hybrid detection strategy combining radio antennas and surface scintillation detectors to discriminate between photon- and proton-induced inclined air showers. The study utilizes a virtual experimental setup based on the GRANDProto300 prototype (a 300-antenna radio array in the Gobi desert) complemented by Telescope Array (TA)-like scintillation detectors placed at each antenna position.

The methodology involves the following steps:

1. Simulation: Extensive air showers initiated by protons and photons were simulated using CORSIKA (v7.7550) with CoREAS for radio emission. The simulations covered energies from $10^{17}$ to $10^{18.5}$ eV and zenith angles from $70^\circ$ to $85^\circ$. The analysis focused on the specific physics of very inclined showers, where geomagnetic emission and geosynchrotron radiation play significant roles, and where the particle footprint is spread over large areas ($\sim 30$ km).
2. Triggering Criteria: Events were required to trigger at least 4 radio antennas (peak-to-peak amplitude $> 75$ $\mu$V/m in the 50–200 MHz range) and, for the hybrid analysis, at least 3 scintillators (energy deposit $> 3$ MIPs).
3. Discrimination Observables: Two key observables were extracted for triggered events:
   • $RMS_{radio,tot}$: The total root mean square of the radio signal traces across all triggered antennas.
   • $E_{dep,tot}$: The total energy deposit recorded in the scintillation detectors.
     Photon-induced showers are expected to have higher radio signals (due to a more developed electromagnetic component) but significantly lower particle energy deposits (due to a lack of muons) compared to proton-induced showers of the same energy.
4. Classification: A logistic regression classifier was trained in the $\log_{10}(E_{dep,tot}) - \log_{10}(RMS_{radio,tot})$ space. To account for the steeply falling cosmic-ray spectrum and the rarity of photons, proton events were weighted by a factor of 10. The analysis was performed in 16 bins of energy and zenith angle to optimize performance across different shower geometries.
5. Optimization: The classification threshold ($p_{\gamma,cut}$) was optimized to minimize background contamination (protons misidentified as photons) while maintaining acceptable signal efficiency. The study also investigated the impact of reconstruction resolution (energy and zenith angle) by testing configurations with and without binning and different weighting schemes.

Key Contributions and Results

• Hybrid Trigger Efficiency: The study demonstrates that while radio trigger efficiency is near 100% for both protons and photons in the considered range, the conditional trigger efficiency in particles (given a radio trigger) is significantly higher for protons. This difference arises because proton showers produce a muon-rich component that survives to the ground, whereas photon showers often fail to trigger the scintillators due to low particle density.
• Discrimination Performance: The hybrid method achieves effective separation between the two primary types.
  • For energies above $10^{17.8}$ eV and zenith angles below $82^\circ$, the background contamination (proton leakage) can be reduced to below $10^{-4}$ (0.01%).
  • The maximal background contamination observed is 2.6% for the lowest energies and highest zenith angles, where particle footprints are sparsest.
  • An optimal classification threshold of $p_{\gamma,cut} = 0.9$ was identified, balancing a background contamination of $\sim 0.12\%$ with a signal efficiency ranging from 0% to 78% depending on energy and angle.
• Flux Limits: Applying this method to a hypothetical 1-year operation of a GRANDProto300 hybrid array, the authors derived 95% confidence level upper limits on the integral photon flux.
  • In the energy range of 0.3 to 3 EeV, the projected limits range from 1.2 to 2.6 km$^{-2}$ sr$^{-1}$ yr$^{-1}$.
  • These limits are competitive with existing constraints from the Pierre Auger Observatory, KASCADE-Grande, and Telescope Array, particularly in the high-inclination regime ($\theta > 70^\circ$) where other experiments have limited sensitivity.
  • Expanding the array to the full GRANDProto300 size (approx. 3 times the antennas of the infill) would improve these limits by a factor of $\sim 3$.

Significance and Claims
The paper claims that hybrid radio and particle detection offers a scalable and competitive method for UHE photon identification, specifically targeting the "uncharted territory" of very inclined showers ($\theta > 70^\circ$). By leveraging the distinct muon-poor nature of photon showers against the radio-detectable electromagnetic component, this approach provides a complementary detection channel to existing vertical-shower searches.

The authors modestly note that the study is limited by the absence of real data for validation and relies on simulations. They highlight that future improvements could involve:

• Incorporating additional observables, such as the number of activated scintillators ($N_{scint}$), which shows a clear distinction between protons and photons.
• Utilizing multivariate analysis (MVA) or machine learning techniques.
• Refining reconstruction methods for energy and zenith angle, which currently rely on true simulation values rather than reconstructed estimates.
• Extending the particle footprint analysis to larger distances, which is currently limited by dethinning algorithms optimized for less inclined showers.

Ultimately, the work suggests that planned large-scale radio observatories like GRAND, when equipped with surface particle detectors, could significantly advance the search for UHE photons and provide new constraints on the origin of cosmic rays.