Improved Heavy Dark Matter Annihilation Search from Dwarf Galaxies with HAWC

Understanding dark matter’s elusive nature is crucial for the framework of particle physics and expanding the Standard Model. This analysis utilizes the High Altitude Water Cherenkov (HAWC) gamma ray Observatory to indirectly search for dark matter (DM) by studying gamma ray emission from dwarf spheroidal galaxies (dSphs). Selected for their high ratio of dark matter to baryonic matter, dSphs are useful for this type of search owing to the low background emission. In comparison to previous HAWC studies, we significantly improve our sensitivity to DM from dSphs due to improvements to our event reconstruction and reduced hadronic contamination. We expanded the number of dSphs studied, DM annihilation channels into the Standard Model (SM), and the amount of data collected on each previously studied dSph. We searched for DM signals in each dSph using the latest version of the algorithms used to reconstruct data from the primary detector of the HAWC instrument. We report that we do not detect evidence of DM from dSphs, so we place upper limits for the velocity-weighted DM annihilation cross-section ($\langleσv \rangle$) on the order of $10^{-23}~\text{cm}^3\text{s}^{-1}$ for a DM mass range of $1-10^4$ TeV.

💡 Research Summary

This paper presents a comprehensive indirect search for heavy weakly interacting massive particle (WIMP) dark matter (DM) using the High‑Altitude Water Cherenkov (HAWC) gamma‑ray observatory. The authors focus on 17 dwarf spheroidal galaxies (dSphs)—Boötes I, Canes Ventriculi I & II, Coma Berenices, Draco I & II, Hercules, Leo I, Leo II, Leo IV, Leo V, Pisces II, Segue 1, Sextans, Ursa Major I & II, and Willman 1—selected for their high dark‑matter‑to‑baryon ratios and low astrophysical gamma‑ray backgrounds.

Key methodological advances over previous HAWC dSph analyses include: (1) adoption of the new “Pass 5” event reconstruction, which improves angular resolution by >30 % and reduces hadronic contamination; (2) implementation of a neural‑network (NN) based energy estimator that yields ~1.5× better energy resolution than the traditional f_hit method; and (3) deployment of a machine‑learning (ML) optimization suite that refines gamma‑hadron separation cuts, enhancing background rejection by ~20 %. These upgrades, combined with a dramatically larger data set (3070 days of livetime versus ~500 days in earlier works), increase the statistical power of the analysis.

The particle‑physics modeling of DM annihilation spectra uses the HDMSpectra (HDMS) library, which incorporates electroweak corrections and higher‑order loop effects. This is particularly important for the heavy mass regime (up to 10 PeV) where standard PPPC4DMID spectra become unreliable. For the γγ, WW, and ZZ channels the authors convolve the intrinsic line‑like features with a Gaussian kernel (σ = 5 % mχ) to account for HAWC’s energy resolution.

Astrophysical J‑factors are derived from two recent catalogs (Pace & Strigari and Geringer‑Sameth), and their uncertainties are treated as nuisance parameters in a Bayesian likelihood framework. Background estimation follows HAWC’s standard direct‑integration method, with additional validation using HEALpix randomization; the dSphs lie well away from the Galactic plane, minimizing diffuse Galactic emission contamination.

Statistical inference is performed with the HAL and 3ML packages, employing a profile‑likelihood approach that marginalizes over systematic uncertainties (energy scale, J‑factor, background modeling). No significant excess is observed in any of the 17 dSphs across all 11 Standard Model final states (e⁺e⁻, μ⁺μ⁻, τ⁺τ⁻, b b̄, t t̄, W⁺W⁻, Z⁰Z⁰, u ū, d d̄, ν_eν̄_e, γγ). Consequently, the authors set 95 % confidence upper limits on the velocity‑averaged annihilation cross‑section ⟨σv⟩ of order 10⁻²³ cm³ s⁻¹ for DM masses between 1 TeV and 10 PeV.

These limits improve upon previous HAWC dwarf analyses (which reached ≈10⁻²² cm³ s⁻¹) and fill the sensitivity gap between GeV‑scale constraints from Fermi‑LAT (≈10⁻²⁴ cm³ s⁻¹) and ultra‑high‑energy neutrino limits. The work demonstrates that the combination of advanced reconstruction, machine‑learning‑driven event selection, and extended exposure can substantially enhance the reach of ground‑based water‑Cherenkov detectors in the heavy‑WIMP regime. Future prospects include further data accumulation, refined ML classifiers, and potential joint analyses with other TeV‑scale instruments, which could push the limits toward the canonical thermal relic cross‑section for multi‑TeV dark matter.