Sensitivity to Axion-like Particle dark matter with very-high-energy gamma-ray observations of Active Galactic Nuclei located behind Galaxy Clusters

Axion-Like-Particles (ALPs) are hypothetical pseudo-scalar particles actively searched as light dark matter candidates. The coupling of ALPs to photons can give rise to distinctive spectral features in the observed gamma-ray spectrum of astrophysical sources. We perform a forecast study on the sensitivity to ALP-photon interactions using stacked mock observations of selected active galactic nuclei (AGNs) located behind galaxy clusters (GC). The ALP-photon conversion in the magnetic fields of galaxy clusters give rise to absorption-like features in AGN spectra that are subject to large variance in their prediction for individual sources. We consider here a stacking analysis of multiple AGN-cluster pairs, which yields a more controlled prediction of the expected ALP-induced spectral patterns in the observed gamma-ray spectra. Using realistic mock observations of selected Fermi-LAT AGNs by ongoing Imaging Atmospheric Cherenkov Telescopes such as H.E.S.S., MAGIC and VERITAS, we provide a careful assessment of the expected sensitivity of a combined statistical analysis of many AGN-GC pairs, together with the impact of modelling and instrumental uncertainties. The sensitivity reaches ALP-photon couplings down to 6$\times$10$^{-13}$ GeV$^{-1}$ for an ALP mass of 3$\times$10$^{-8}$ eV, and is currently statistically dominated indicating further improvements from more observations. Such a stacking analysis approach enables exploration of the yet-uncharted ALP dark matter parameter space in the 10$^{-8}$ - 10$^{-7}$ eV mass range.

💡 Research Summary

This paper investigates the potential of very‑high‑energy (VHE) gamma‑ray observations of active galactic nuclei (AGN) that lie behind galaxy clusters (GC) to probe axion‑like particles (ALPs) as dark‑matter candidates. ALPs couple to photons via a two‑photon vertex; in the presence of an external magnetic field, photons can oscillate into ALPs and back. When VHE photons from a distant AGN traverse the µG‑level magnetic field of a foreground galaxy cluster, the conversion probability becomes energy‑dependent and imprints characteristic absorption‑like features on the observed spectrum. For a single AGN‑GC pair, the detailed shape of these features is highly sensitive to the poorly known small‑scale properties of the cluster magnetic field (domain orientation, coherence length), leading to large variance and weak constraints.

To overcome this limitation, the authors propose a stacking analysis of many AGN‑GC pairs. By averaging over many independent magnetic‑field realizations, the stochastic oscillations smooth out, yielding a predictable, smooth spectral imprint that can be searched for with higher statistical power.

Source selection – The authors cross‑matched the high‑latitude (|b|>10°) fourth Fermi‑LAT AGN catalog (4LAC‑DR3‑h) with Sunyaev‑Zeldovich, optical, and X‑ray cluster catalogs. They required the AGN redshift to be equal or larger than the cluster redshift and the line‑of‑sight impact parameter to be ≤ 500 kpc. This produced 29 AGN‑GC pairs; after restricting to sources listed in the 3FHL catalog (detected above 10 GeV) and visible to current imaging atmospheric Cherenkov telescopes (IACTs), a final sample of 16 AGN was obtained. Visibility calculations for H.E.S.S., MAGIC, and VERITAS yielded 41 independent mock observation sets (some AGN are observable by more than one instrument).

Magnetic‑field modeling – Each cluster is modeled as a turbulent field with an average strength of order µG, a coherence (domain) length of ~10 kpc, and a Kolmogorov‑type power spectrum. The authors generate 10³ random realizations of domain orientations and lengths to capture the intrinsic variance. For each realization they solve the photon‑ALP propagation equations (using the transfer‑matrix formalism) to obtain the energy‑dependent conversion probability P_{γ→a}(E).

Gamma‑ray propagation – Intrinsic AGN spectra are taken as power‑laws (or log‑parabolas) fitted to Fermi‑LAT data. Attenuation by the extragalactic background light (EBL) is applied using a standard model (e.g., Franceschini). The cluster‑induced ALP‑photon conversion is then applied to produce the expected observed VHE flux.

Mock observations – Using instrument response functions (IRFs) for H.E.S.S., MAGIC, and VERITAS at representative zenith angles (20°–30°), the authors simulate 50 h of exposure per source. Poisson fluctuations and systematic uncertainties (energy scale, IRF variations) are added to generate realistic mock spectra.

Statistical stacking – For each source a log‑likelihood L_i(g_{aγ}, m_a) is computed. The combined likelihood L = Σ_i L_i is used to construct a profile‑likelihood ratio test statistic TS = –2 ln(L_null/L_alt). By scanning over the ALP parameter space and evaluating the 95 % confidence level (CL) exclusion, the authors derive sensitivity curves.

Results – In the mass range around m_a ≈ 3 × 10⁻⁸ eV, the stacked analysis reaches a coupling sensitivity of g_{aγ} ≈ 6 × 10⁻¹³ GeV⁻¹. This improves upon single‑source IACT limits by roughly one to two orders of magnitude. The sensitivity is presently statistics‑dominated; increasing the total exposure or the number of AGN‑GC pairs would linearly improve the reach. Systematic studies show that variations in the EBL model affect the limits by less than 10 %, whereas uncertainties in the cluster magnetic‑field parameters (strength, coherence length) dominate the systematic error budget.

Conclusions and outlook – The 10⁻⁸–10⁻⁷ eV ALP mass window is largely inaccessible to laboratory haloscopes or helioscopes, making astrophysical searches crucial. The stacking approach demonstrated here provides the first realistic forecast for probing this region with current IACTs. The authors argue that future facilities such as the Cherenkov Telescope Array (CTA), with greater sensitivity and the ability to monitor many more AGN‑GC pairs, could push the coupling reach down to g_{aγ} ~ 10⁻¹³ GeV⁻¹. Improved measurements of cluster magnetic fields (e.g., via Faraday rotation) and refined EBL models would further reduce systematic uncertainties, enhancing the robustness of the ALP search.