Constraints on Solar Reflected Dark Matter from a combined analysis of XENON1T and XENONnT data

TL;DR

This study probes sub-GeV dark matter via solar reflection, where DM–electon scattering in the Sun boosts DM kinetic energy and enhances detectability in liquid xenon detectors. By combining XENON1T S2-only and XENONnT low-energy ER data, the authors derive 90% CL limits on the DM–electron cross section across multiple mass ranges, with a minimum cross section of $3.41\times10^{-39}\ \mathrm{cm^2}$ at $m_χ=0.3\ \text{MeV}/c^2$ for a heavy mediator. The approach leverages Monte Carlo solar flux calculations (DaMaSCUS-SUN), specialized event selections to lower thresholds, and robust statistical methods (Yellin optimal interval and Feldman–Cousins likelihood) to constrain solar-reflected DM scenarios. This work extends sensitivity to keV–MeV DM masses and demonstrates the complementary power of combining S2-only and low-energy ER analyses for sub-GeV DM. The findings have implications for light DM models and motivate further improvements in flux modeling and background suppression in upcoming data taking.

Abstract

We report on a search for sub-GeV dark matter upscattered via the solar reflection mechanism in the heavy mediator scenario. Under the Standard Halo Model, keV to MeV dark matter produces nuclear recoils with energies below the detection threshold of liquid xenon time projection chambers. We enhance sensitivity to low-mass dark matter by considering dark matter-electron scattering, employing dedicated event selections to reduce the detection threshold, and exploiting the additional kinetic energy imparted to the dark matter particle by solar upscattering. Using XENON1T ionization-only and XENONnT low-energy electronic recoil datasets, we exclude previously unconstrained DM-electron scattering cross section for masses between $4.6\, \text{keV/}c^2$ and $20\, \text{keV/}c^2$, and between $0.2\, \text{MeV/}c^2$ and $2\, \text{MeV/}c^2$, reaching a minimum of $3.41\times10^{-39}\, \text{cm}^2$ for a mass of $0.3\, \text{MeV/}c^2$ at 90\% confidence level.

Figures (4)

• Figure 1: SRDM differential flux in a terrestrial detector as a function of SRDM speed, computed via Monte Carlo simulations DaMaSCUSsun. The shape of the differential flux depends on the DM mass (indicated by colors) and also the DM-electron cross section (indicated by line styles). The differential fluxes are computed to the lowest DM speed that can produce detectable signals in the detector for the given SRDM mass.
• Figure 2: Top: SRDM differential rates as a function of S2 area in the XENON1T detector evaluated at the 90% C.L. upper limits on the DM-electron cross section for the various SRDM masses (colors). Short vertical gray ticks in the narrow top panel indicate events in the XENON1T S2-only science search dataset. Green line (shaded band) shows the event-selection efficiencies ($1\sigma$ uncertainty). Purple line (shaded band) shows the effective remaining exposure ($1\sigma$ uncertainty) after selections. Bottom: SRDM differential rates as a function of reconstructed energy in the XENONnT detector evaluated at the 90% C.L. confidence interval on the DM-electron cross section for the various SRDM masses (colors). Short vertical gray ticks in the narrow top panel indicate events in the XENONnT low-energy ER dataset. Dashed vertical line indicates the 1 keV threshold. Green line (shaded band) shows the combined detection and event-selection efficiencies ($1\sigma$ uncertainty) of the dataset.
• Figure 3: Event expectation for the various background components in the XENONnT low-energy ER dataset. The spectrum of a $4.6\, \text{keV/}c^2$ SRDM signal with DM-electron cross section of $6.09\times10^{-36}\, \text{cm}^{2}$ (unconstrained 90% C.L. confidence interval) is also shown for reference.
• Figure 4: The dark red line shows the 90% C.L. upper limits on the DM-electron cross section using the XENON1T S2-only dataset. Due to the lack of a full background model for this dataset, Yellin’s optimal interval method is used PhysRevD.66.032005 and no sensitivity band can be constructed. The red line shows the 90% C.L. confidence intervals using the XENONnT low-energy ER dataset after applying the power constraint. Thin solid black line shows the 90% C.L. confidence intervals computed using only the XENONnT low-energy ER dataset without power constraint, while the dashed black line, along with the green and yellow bands, indicate the median expected sensitivity and the corresponding $1\sigma$ and $2\sigma$ bands respectively. We also show results from XENONnT Few Electrons (FE) analysis XENON:2024znc, PandaX PandaX:2024syk, CDEX10 PhysRevLett.132.171001, and theoretical recasts of limits using XENON1T datasets Emken:2021lgcPhysRevD.104.103026 in dotted. The gray shaded region at low DM mass indicates the parameter space excluded by stellar cooling constraints from red giants srdm_rg_Chang_2021.