[Paper Review] Dark Matter's secret liaisons: phenomenology of a dark U(1) sector with bound states
The paper analyzes a dark U(1) sector with bound-state formation, computing relic density and indirect-detection signals, and deriving constraints from cosmology and astrophysical searches.
Dark matter (DM) charged under a dark U(1) force appears in many extensions of the Standard Model, and has been invoked to explain anomalies in cosmic-ray data, as well as a self-interacting DM candidate. In this paper, we perform a comprehensive phenomenological analysis of such a model, assuming that the DM abundance arises from the thermal freeze-out of the dark interactions. We include, for the first time, bound-state effects both in the DM production and in the indirect detection signals, and quantify their importance for Fermi, AMS, and CMB experiments. We find that DM in the mass range 1 GeV to 100 TeV, annihilating into dark photons of MeV to GeV mass, is in conflict with observations. Instead, DM annihilation into heavier dark photons is viable. We point out that the late decays of multi-GeV dark photons can produce significant entropy and thus dilute the DM density. This can lower considerably the dark coupling needed to obtain the DM abundance, and in turn relax the existing constraints.
Motivation & Objective
- Motivate and study dark matter charged under a new dark U(1) with a massive dark photon as mediator.
- Compute the relic abundance including bound-state formation and decay effects in the dark sector.
- Assess constraints from beam dumps, supernovae, direct detection, and BBN.
- Investigate indirect detection signals from gamma rays, antiprotons, and CMB anisotropies.
- Explore how late decays of multi-GeV dark photons can dilute DM and relax constraints.
Proposed method
- Specify a Dirac fermion DM X coupled to a dark photon V_D with kinetic mixing to hypercharge.
- Compute annihilation and bound-state formation cross-sections including a ground-state BSF channel and its decay (to 2 or 3 V_D).
- Describe dark photon decays to SM states via kinetic mixing and determine corresponding branching ratios.
- Relate the dark-sector coupling alpha_D to M_DM and m_VD by requiring thermal freeze-out yields the observed DM density.
- Incorporate bound-state effects into relic-density calculations and propagate into constraints from cosmology and indirect searches.
- Use velocity distributions for Milky Way and dwarf galaxies to evaluate cross-sections and signal spectra.

Experimental results
Research questions
- RQ1How does bound-state formation affect the thermal relic abundance of DM in a dark U(1) model?
- RQ2What regions of (M_DM, m_VD, alpha_D, epsilon) are compatible with cosmological and astrophysical constraints when BSF is included?
- RQ3Under what conditions do bound-state decays and dark-photon decays produce observable indirect-detection signals?
- RQ4Can late decays of heavier dark photons dilute the DM density and relax coupling/constraints?
- RQ5Which indirect-detection channels (gamma rays, antiprotons, CMB) provide the strongest constraints for this scenario?
Key findings
- Including bound-state formation reduces the required dark coupling to achieve the observed DM abundance.
- DM with mass between 1 GeV and 100 TeV annihilating into MeV–GeV dark photons is in conflict with observations, while annihilation into heavier dark photons remains viable.
- Late decays of multi-GeV dark photons can inject entropy and dilute the DM density, weakening constraints on the model.
- Indirect searches with Fermi, AMS-02, and Planck CMB data provide significant constraints on the parameter space.
- The analysis highlights resonances in annihilation and BSF cross-sections, and the importance of the Coulomb regime vs. Yukawa effects for cross-sections.
- BSF is significant in dwarfs and the Milky Way, but negligible for CMB-era DM velocities.

Better researchstarts right now
From reading papers to final review, dramatically reduce your research time.
No credit card · Free plan available
This review was created by AI and reviewed by human editors.