Strong new limits on light dark matter from neutrino experiments

The nondetection of GeV-scale weakly interacting massive particles (WIMPs) has led to increased interest in more general candidates, including sub-GeV dark matter. Direct-detection experiments, despite their high sensitivity to WIMPs, are largely blind to sub-GeV dark matter. Recent work has shown that cosmic-ray elastic scattering with sub-GeV dark matter would both alter the observed cosmic ray spectra and produce a flux of relativistic dark matter, which would be detectable with traditional dark matter experiments as well as larger, higher-threshold detectors for neutrinos. Using data, detectors, and analysis techniques not previously considered, we substantially increase the regions of parameter space excluded by neutrino experiments for both dark matter–nucleon and dark matter–electron elastic scattering. We also show how to further improve sensitivity to light dark matter.

Figure 1

Schematic diagram of CR-DM scattering. A CR with velocity much higher than that of the DM ($\sim {10}^{-3}c$) scatters with DM, transferring some of the CR’s kinetic energy and boosting the DM to higher velocity.

Figure 2

Left: Flux of DM upscattered by CR nuclei (protons and helium), assuming a DM-nucleon elastic scattering cross section of ${10}^{-30}{\mathrm{cm}}^2$, and DM masses as labeled. Right: Flux of DM upscattered by CR electrons, assuming a DM-electron elastic scattering cross section of ${10}^{-30}{\mathrm{cm}}^2$, and the same DM masses.

Figure 3

Daya Bay background-dominated data [70] compared to predicted DM-induced spectra. Data from a combined 3.6 kton-day exposure of two Daya Bay antineutrino detectors (black points). Also shown are ballistic (blue) and numerically (black) computed spectra for a mass of 1 MeV. Dotted lines are the two calculations of the spectrum at a cross section of $5\times {10}^{-28}{\mathrm{cm}}^2$ (denoted “High $\sigma$”); solid lines are the spectrum at $5\times {10}^{-29}{\mathrm{cm}}^2$ (denoted “Low $\sigma$”). These cross sections are barely excluded in the ballistic case, as the DM-induced spectra are barely above the data, but much more strongly excluded in the numerical case.

Figure 4

Daya Bay background-dominated data [70] compared to predicted DM-induced spectra. Data from a combined 3.6 kton-day exposure of two Daya Bay ADs (black points). Also shown are two predicted DM-induced spectra for a mass of 1 MeV and cross sections of $6\times {10}^{-29}{\mathrm{cm}}^2$ and $7\times {10}^{-28}{\mathrm{cm}}^2$, corresponding, respectively, to the floor and ceiling of the Daya Bay exclusion region in Fig. 6.

Figure 5

KamLAND measured solar neutrino data [86] compared to predicted DM-induced spectra. Data from a 123 kton-day KamLAND observation (black points). In the 13.5–20 MeV range, KamLAND observed only one event. Also shown are two predicted DM-induced spectra for a mass of 1 MeV and cross sections of $3\times {10}^{-31}{\mathrm{cm}}^2$ and $1.3\times {10}^{-28}{\mathrm{cm}}^2$, corresponding, respectively, to the floor and ceiling of the KamLAND exclusion region in Fig. 6.

Figure 6

Our exclusion regions calculated from KamLAND and Daya Bay data (red, labeled), compared to previous results from Ref. [38] (purple), as well as existing limits from direct detection [87, 88] (gray), cosmology [9, 11, 12, 13] (blue), and CRs [16] (teal). DM lighter than $\sim 1\mathrm{keV}$, denoted by the vertical dashed line, cannot have been in thermal contact with ordinary matter in the early universe, due to effects on structure formation (see Ref. [89] and references therein).

Figure 7

Projected sensitivity regions for PROSPECT (red, dotted line) and JUNO (red, dashed line), compared to existing exclusion regions. See text for the difference between the two PROSPECT projections.

Figure 8

Super-K measured data from a DSNB search, compared to predicted DM-induced spectra. Data from a 1497 day Super-K observation, from 16 to 88 MeV [96] (black points). Also shown are three predicted DM-induced recoil spectra for a mass of 1 MeV. The increasingly large cross sections correspond to our limit, the middle of the new region we exclude, and the previous limit from Ref. [39]. The wiggles in the spectra at low energy are due to signal detection efficiency, as is the vertical blue line, below which we take efficiency to be zero (see text).

Figure 9

Our extension of the exclusion region calculated from Super-K data (solid red line) and our projection for Hyper-K (dashed red line) compared to the limit from Ref. [39] (blue line), as well as previous limits from direct detection (DD) [98, 99, 100, 101] and solar reflection (dark grey region from Ref. [40]; Ref. [39] argued that the ceiling of their region should be higher, and this extension is shown by the light gray box above it). The limit based on CR spectra [16] constrains cross sections higher than shown on this plot. DM lighter than $\sim 1\mathrm{keV}$, denoted by the vertical dashed line, cannot have been in thermal contact with ordinary matter in the early universe, due to effects on structure formation (see Ref. [89] and references therein).