Accelerating the Discovery of Light Dark Matter

TL;DR

Sub-GeV thermal DM annihilating through SM mixing is analyzed by defining a vector-portal framework and a relic-density target $y \equiv \epsilon^2 \alpha_D \left(\frac{m_{\rm DM}}{m_{A'}}\right)^4$ that governs early-universe annihilation for $m_{A'} \gg m_{\rm DM}$. The authors classify annihilation mechanisms, compare constraints across direct detection, collider, fixed-target, and astrophysical data, and identify a small set of flagship experiments capable of decisively testing the thermal light-DM paradigm. They find that Dirac-fermion DM is excluded by CMB constraints in the $s$-channel case, while scalar, pseudo-Dirac, and asymmetric DM remain viable and within reach of planned experiments across the MeV–GeV range. The work provides a concrete experimental roadmap—combining direct electron scattering, mono-photon B-factory searches, and fixed-target missing-momentum experiments—to either confirm or rule out sub-GeV thermal DM coupled via the vector portal.

Abstract

We analyze the present status of sub-GeV thermal dark matter annihilating through Standard Model mixing and identify a small set of future experiments that can decisively test these scenarios.

Figures (2)

• Figure 1: Constraints and projections for representative vector-portal DM scenarios. For definiteness, we evaluate all constraints for $m_{\rm DM} /m_{A^\prime} = 1/3$ and (except for the LSND$\times$SIDM bound -- see below), $\alpha_D = 0.5$, near the perturbativity limit. The relic density, CMB, and direct detection contours scale roughly as $\epsilon^2 \alpha_D (m_{\rm DM}/m_{A^\prime})^4$ (plotted on the $y$-axis), and so are insensitive to separate factors in the above. For other constraints, this choice is conservative, in that smaller choices of $\alpha_D$ and/or $m_{\rm DM} /m_{A^\prime}$ would shift the shaded regions downward (see text); arrows denote the shift in sensitivity for $\alpha_D \rightarrow 0.05$. We illustrate these constraints for (left) pseudo-Dirac/inelastic fermion thermal-relic DM, with splitting $\delta \gtrsim100 \, \keV$, (center) asymmetric Dirac fermion DM, and (right) scalar elastic-scattering thermal relic DM. Dirac fermion thermal-relic DM is fully excluded by the CMB constraint and inelastic or asymmetric scalar DM is quite similar to the right figure, but with CMB and direct detection constraints weakened. CMB, self-interaction (SIDM), and direct detection constraints all depend on the $\chi(\varphi)$ abundance, and are computed assuming the full DM abundance, not the thermal abundance expected for given masses and couplings. In all plots, gray shaded regions (color online) represent traditional DM constraints (e.g. direct detection), while non-traditional accelerator probes are shaded beige. We note that pseudo-Dirac limits are modified (and new dedicated searches are possible OurFuturePaper) if $\delta$ is large enough that $\chi_+$ can decay on detector length-scales.
• Figure 2: Same as Fig \ref{['fig:money']}, but assuming $m_{A^\prime} = 1.1 m_{\chi}$, where the relic abundance is still achieved through $s-$channel annihilation, but $A^\prime$ decays visibly. Dashed blue contours show projected sensitivities from Essig:2013lka for future $A^\prime \to \ell^+\ell^-$ searches in the next few years. Depending on the $\chi$ and $A^\prime$ masses, models below the relic density line may still achieve viable thermal abundances through other processes --- see JoshAndRafaelle.