Indirect searches for realistic sub-GeV Dark Matter models

Indirect searches for Dark Matter (DM) particles with mass in the MeV – GeV scale have received significant attention lately. Pair-annihilations of such DM particles in the Galaxy can give rise to (at the same time) MeV to GeV $γ$-rays via prompt emission, sub-GeV $e^\pm$ in cosmic-rays, as well as a broad photon spectrum ranging from $X$-rays to soft $γ$-rays, produced by the DM induced $e^\pm$ via inverse Compton scattering, bremsstrahlung and in-flight annihilation processes (collectively called `secondary emissions’). We focus on two representative realistic sub-GeV DM models, namely, the vector-portal kinetic-mixing model and the higgs-portal model, and perform a detailed study of the indirect detection constraints from existing $X$-rays, $γ$-rays and cosmic-ray observations, based on all of the above-mentioned signals. We also estimate the future prospects from the upcoming MeV photon telescope COSI, including all possible types of prompt and secondary emission signals. We compare our results with the constraints and (or) projections from cosmological and terrestrial observations. We find that, for both the sub-GeV DM models, the current observations constrain the annihilation cross-section at the level of $\langle σv \rangle \lesssim 10^{-27} {\rm cm}^3/{\rm s}$, or lower for some specific mass ranges or under optimistic assumptions. Moreover, new unconstrained DM parameter space can be probed at the upcoming instruments like COSI, thanks to the inclusion of secondary photons which in many cases provide the dominant signal.

💡 Research Summary

This paper presents a comprehensive study of indirect detection constraints on sub‑GeV dark‑matter (DM) candidates, focusing on two well‑motivated “portal” models: a vector‑portal model with a dark photon that kinetically mixes with the Standard‑Model photon, and a Higgs‑portal model with a scalar mediator that mixes with the Higgs boson. The authors fix the mediator masses to representative values (mV = 3 mDM for the kinetic‑mixing case and mS = mDM/2 for the Higgs‑portal case) so that the dominant annihilation channels are, respectively, direct annihilation through an off‑shell vector (DM DM → V* → ff̄) and secluded annihilation via pair production of scalars (DM DM → SS → SM SM).

Using the state‑of‑the‑art tools HAZMA, HAZMA2 and HERWIG4DM, the authors compute the full spectra of photons and electrons/positrons produced in DM annihilations for masses ranging from 1 MeV to 1 GeV. They emphasize that at these low energies the final states are dominated by light hadronic resonances (π, η, ρ, ω, ϕ, …) rather than quarks, and that the spectra depend essentially only on the DM mass for the chosen mediator‑mass relations.

The paper separates the observable signals into three classes: (1) prompt γ‑rays emitted directly in the annihilation, (2) secondary photons generated when the DM‑induced e± propagate through the interstellar medium and undergo inverse‑Compton scattering (ICS), bremsstrahlung, and in‑flight annihilation, and (3) the cosmic‑ray e± flux itself. A semi‑analytic treatment of the e± propagation is presented, followed by a full numerical solution of the diffusion‑loss equation to obtain the steady‑state e± distribution. The authors show that secondary photons often dominate over the prompt component, especially for DM masses above a few hundred MeV where muon and pion decay channels become important.

For the cosmic‑ray channel, they compare the predicted DM‑induced e± flux with Voyager‑1 measurements (which probe the low‑energy, solar‑modulation‑free regime) and with AMS‑02 data at higher energies. By constructing likelihoods for each dataset, they derive 95 % confidence upper limits on the thermally averaged annihilation cross section ⟨σv⟩. Current observations constrain ⟨σv⟩ to be below roughly 10⁻²⁷ cm³ s⁻¹ across most of the sub‑GeV mass range, with tighter bounds in specific mass windows where the photon or e± spectra are particularly distinctive.

The authors then project the sensitivity of the upcoming Compton Spectrometer and Imager (COSI), a MeV‑range γ‑ray telescope. Simulations indicate that COSI could improve existing limits by one to two orders of magnitude, especially by exploiting the secondary photon component (ICS and bremsstrahlung) that peaks in the 0.2–5 MeV band where COSI has optimal performance. This would open up previously unconstrained parameter space, notably for kinetic‑mixing couplings y ≈ 10⁻⁹–10⁻⁸ and DM masses between 30 MeV and 500 MeV.

Finally, the paper compares these indirect‑detection bounds with complementary constraints from the Cosmic Microwave Background (Planck), Big‑Bang Nucleosynthesis, terrestrial experiments (beam‑dump, fixed‑target, and direct‑detection searches), and the supernova SN 1987A neutrino and γ‑ray limits. The indirect searches, particularly when secondary emissions are included, provide the strongest constraints in the intermediate mass regime (∼100 MeV) and are competitive with, or surpass, cosmological and laboratory limits. The authors conclude that a full, model‑specific treatment of both prompt and secondary signals is essential for robustly probing sub‑GeV dark‑matter scenarios, and that upcoming MeV‑scale missions like COSI will play a pivotal role in closing the remaining viable windows.