New benchmarks for direct detection of freeze-in dark matter in vector portal models

Abstract

We investigate the freeze-in of MeV-scale fermionic dark matter (DM) that couples to the Standard Model via a new vector mediator to assess the potential that future direct detection experiments have to observe new physics in either the DM or neutrino sectors. We study the minimal kinetic mixing dark photon of a secluded $U(1)_D$ as well as gauge bosons of the anomaly-free $U(1)_{L_i-L_j}$, with $i,j=e,μ,τ$, and $U(1)_{B-L}$ gauge extensions, exploring the impact of low reheating temperatures on the DM production rates. For the ultralight dark photon scenario, we show that current experimental constraints from electron recoil data in DAMIC-M and PandaX-4T can be avoided if the DM fermion is only a subcomponent (smaller than 40%) of the total cold DM and that future detectors can be sensitive to a DM fraction below 1% for masses above 1 MeV. For a massive dark photon, there are allowed regions of the parameter space with masses in the range 50 MeV $\lesssim m_{\rm DM}\lesssim$ 500 MeV that can be within the reach of direct detection experiments through nuclear recoils if freeze-in occurred at a low reheating temperature. Finally, the case of $U(1)_{L_i-L_j}$ and $U(1)_{B-L}$ is particularly interesting since the discovery of new physics can come from either the DM or the neutrino sector, which features new interactions. We find that freeze-in at low reheating temperatures can reproduce the observed abundance in large parts of the parameter space up to gauge couplings of $g_X\sim10^{-2}$ for MeV DM. Most notably, direct detection experiments will be sensitive to considerable parts of this parameter space in nuclear recoils for 50 MeV $\lesssim m_{\rm DM}\lesssim$ 500 MeV. Additionally, the enhanced signal from solar neutrino coherent scattering is observable in these scenarios, which can serve as a further handle to identify the underlying particle physics model.

Figures (9)

• Figure 1: Contribution to the DM yield as function of the DM mass in the minimal $U(1)_{D}$ (left) and $L_\mu-L_\tau$ (right) models for several self-annihilation channel: quarks (red), SM plasmon/$A^\prime$ (magenta), $\nu_{\mu,\tau}$ (blue), $\mu$ (green), $\tau$ (yellow), electron (black). Top, middle and bottom panels correspond to a reheating temperature of 10 MeV, 1 GeV and 100 GeV, respectively. In the left (right) panel $\kappa=2\times10^{-11}$ ($g_{\mu\tau}^2/e=2\times10^{-11}$) has been fixed and in both panels $m_{A^\prime}=1.5\,m_\chi$.
• Figure 2: Constraints from electron recoil data of direct detection experiments on freeze-in DM in the ultra-light dark photon scenario. Left: Constraints on $\xi \sigma_{\chi e}$ as a function of the DM mass, $m_\chi$. The black solid line represents the freeze-in solution in a standard cosmology with high reheating temperature $T_{\rm rh}$ where $\chi$ accounts for all of the DM, $\Omega_\chi h^2=0.12$, whereas the brown, green and orange lines correspond to an under-abundance of $\xi=0.45$, $0.025$, and $0.001$, respectively. The black dashed, dotted and dot-dashed lines are the freeze-in solutions where $\chi$ constitutes all the DM in non-standard cosmologies with low reheating temperatures $T_{\rm rh}$ of $10^{-2}$ GeV, $10^{-1}$ GeV, and $1$ GeV, respectively. The blue shaded areas correspond to the regions excluded by the DAMIC-M DAMIC-M:2025luv and PandaX-4T PandaX:2025rrz electron recoil data. The blue dotted, dot-dashed and dashed lines represent the future sensitivity of SuperCDMS, TESSERACT and OSCURA SuperCDMS:2022kse2020Snowmass2021LetterOIOscura:2023qik, respectively. The region below the blue hatched line corresponds to the neutrino floor for electron recoils in a Si target Carew:2023qrj. Right: Same experimental constraints and future sensitivities as on the left, but shown on the under-abundance fraction, $\xi$, as a function of the DM mass, $m_\chi$.
• Figure 3: Black lines correspond to the regions in the $(m_{\chi},\epsilon_D)$ plane where the correct relic density is obtained via freeze-in for a reheating temperature of $T_{\mathrm{rh}}=10^2$ GeV (solid), $1$ GeV (dot-dashed), $10^{-1}$ GeV (dotted), and $10^{-2}$ GeV (dashed). We have chosen a benchmark value of $g_D=0.1$ and four different values for the mediator and DM mass ratio, $m_{A'}/m_\chi=1.5$, $1/2$, $1/10$, and $1/20$. The shaded blue areas are constrained via electron recoils in DAMIC-M DAMIC-M:2025luv, DarkSide-50 DarkSide:2022knj, PandaX-4T PandaX:2025rrz and XENON1T from solar-reflected DM Emken_2024, while the green areas are constrained via nuclear recoils in CRESST CRESST:2020wtj, DarkSide-50 DarkSide-50:2025lns, LZ LZ:2025igz and PandaX-4T PandaX:2025rrz. The green solid lines show the projected sensitivities (from nuclear recoils) for SuperCDMS SNOLAB and DarkSide-20k SuperCDMS:2022kseDarkSide-50:2025lns. The hatched green (blue) lines represents the DM discovery limit in nuclear (electron) recoils in a Si target.
• Figure 4: Parameter space relevant for freeze-in production of a vector-like fermion at low reheating temperature within $U(1)_{L_\mu-L_\tau}$. Results are shown for four different mediator-to-DM mass ratios of $m_{A'}/m_\chi = 3/2$ (top left), 1/2 (top right), 1/10 (bottom left), and 1/20 (bottom right). For each configurations, we show the relic density line for the observed DM abundance ($\Omega h =0.12$) for a reheating temperature of $T_R=10$ MeV (black dashed), 100 MeV (black dotted), 1 GeV (black dot-dashed), and 100 GeV (black solid). The dark red line illustrates the correct DM relic abundance from freeze-out production, as an indicator of when the DM thermalises with the SM plasma.
• Figure 5: DM discovery limits in a $U(1)_{L_\mu-L_\tau}$ model with $m_{A'}/m_\chi=1/10$ for a direct detection experiment employing Si with a nominal exposure of 1 tonne-years for varying energy thresholds $E_{\rm th}$ of 1 eV (solid), 10 eV (dashed) and 100 eV (dotted). While the mass reach of the DM discovery limit deteriorates significantly with increasing thresholds, the BSM neutrino discovery limit (dot-dashed) is essentially unaffected (at these low thresholds). The stars illustrate the three benchmark points defined on the right, for which we show the resulting DM and neutrino spectra in \ref{['fig:specs']}.