Probing Sub-GeV Dark Matter with Conventional Detectors

The direct detection of dark matter particles with mass below the GeV scale is hampered by soft nuclear recoil energies and finite detector thresholds. For a given maximum relative velocity, the kinematics of elastic dark matter nucleus scattering sets a principal limit on detectability. Here, we propose to bypass the kinematic limitations by considering the inelastic channel of photon emission from bremsstrahlung in the nuclear recoil. Our proposed method allows us to set the first limits on dark matter below 500 MeV in the plane of dark matter mass and cross section with nucleons. In situations where a dark-matter–electron coupling is suppressed, bremsstrahlung may constitute the only path to probe low-mass dark matter awaiting new detector technologies with lowered recoil energy thresholds.

Figure 1

Photon emission resulting from DM-nucleus scattering. The thick line represents the nucleus in the atomic initial (final) state $i$ ($f$) with intermediate state $n$, represented by the thin solid line.

Figure 2

Elastic ($dR/d{E}_R$) and photon-emission ($dR/d\omega$) rates in xenon. The ionization threshold is 12 eV. The dotted line is derived from the naive cross section (3).

Figure 3

$(m_{\chi },{\sigma }_n)$ plane of DM mass and DM-nucleon cross section. Regions labeled CRESST-II and CDMSlite were previously excluded from elastic DM-nucleus scattering [31, 32]; regions labeled XENON10 and XENON100 show newly derived constraints based on (1b). Above the line “m.f.p.” limits are invalidated as DM scatters before reaching the detector. The region XQC is excluded from rocket-based X-ray calorimetry data [33]. Projections for CRESST-III and for a dedicated liquid Xe experiment, labeled LXE, are also shown; see main text. The monojet constraint CDF is model dependent; CMB constraints can be evaded [34] and are not shown.

Figure 4

Comparison of electron recoil (blue) vs photon emission (orange) events in a leptophobic model per logarithmic energy interval, $dR/d({\log }_{10}{E}_{R,e})$ vs $dR/d({\log }_{10}\omega )$, respectively, assuming that ${\sigma }_e={\alpha }^2/(16{\pi }^2){\sigma }_n$. The dotted (orange) line is the naïve rate. Thin gray lines break down the ionization contribution from the respective atomic shells (following previous calculations [3, 4].)