The core problem the team addresses is kinematic. Conventional direct detection experiments, such as XENON, LUX-ZEPLIN (LZ) and PandaX, watch for the tiny nuclear recoils produced when a dark matter particle from the Galactic halo smashes into an atomic nucleus in a detector. But halo dark matter moves at only about 220 kilometres per second, roughly a thousandth the speed of light. A particle weighing less than a gigaelectronvolt (GeV), far lighter than a proton, simply cannot deliver enough momentum to produce a recoil above the keV-scale thresholds of these instruments. The result is a vast, theoretically motivated wilderness of parameter space, the entire sub-GeV regime, where ordinary direct detection is essentially blind.

The cosmic-ray boosted dark matter framework, or CRDM, offers a way through. Relativistic cosmic rays pervade the Galaxy, streaming at energies vastly exceeding those of the sluggish halo particles. When a cosmic ray elastically scatters off a dark matter particle, it can transfer an enormous fraction of its kinetic energy, kicking the dark matter up to velocities approaching a tenth the speed of light. Such a boosted particle carries enough punch to deposit a detectable keV-scale recoil even in a large liquid xenon detector buried more than a kilometre underground. The idea, first formalised by Bringmann and Pospelov and by Ema and colleagues in 2019, has since matured into a serious probe, and the LZ collaboration itself has applied it to derive model-independent bounds on the dark matter-nucleon scattering cross section.

What the new study adds is particle-physics structure. Rather than treating the interaction as a featureless constant, the authors embed the scattering in four benchmark simplified models, in which a Dirac fermion dark matter particle communicates with nucleons through a mediator: a scalar, a vector, a pseudoscalar or an axial-vector boson. Scalar and vector exchange produces spin-independent scattering, which benefits from coherent enhancement across all the nucleons in a nucleus, scaling with the square of the atomic mass. Pseudoscalar and axial-vector exchange produces spin-dependent scattering, which lacks that enhancement and is best probed by detectors containing hydrogen, such as the Borexino neutrino experiment in Italy. For each model the team wrote down the full interaction Lagrangians, the differential cross sections including hadronic and nuclear form factors, and the resulting flux of boosted dark matter arriving at Earth.

The calculation chains together several layers of astrophysics and particle physics. Starting from the local interstellar spectra of cosmic-ray nuclei up to nickel (atomic number 28), the authors integrate over the scattering kinematics to obtain the differential flux of boosted dark matter, weighted by the local dark matter density of about 0.3 GeV per cubic centimetre and an effective propagation distance that encodes the Galactic dark matter distribution under a Navarro-Frenk-White halo profile. They then track the boosted particles as they descend through the atmosphere and kilometres of rock, losing energy through further elastic scattering, before computing the event rate they would generate in the detector target. Crucially, the mediator propagator factor, one over the square of the mediator mass plus the momentum transfer squared, is kept in full rather than approximated, which turns out to matter enormously for light mediators.

Using published exclusion curves from LZ, XENON1T, XENONnT, MiniBooNE and Borexino, the team derived constraints on both the scattering cross section and the underlying coupling product, the combination of the dark matter and nucleon couplings normalised by four pi, for mediator masses spanning from a millionth of a GeV up to 1 GeV. In the model-independent, constant cross section case, the interplay between detector depth and sensitivity produces a distinctive band of exclusion. Cross sections that are too small yield no detectable excess, while cross sections that are too large cause the boosted dark matter to scatter out in the Earth’s crust before it ever reaches the detector. Because LZ and XENON sit roughly 1400 metres underground while MiniBooNE lies just below the surface, the experiments are complementary: the shallow detector catches the very large cross sections that deep experiments cannot see, stitching together a much wider excluded region.

The most striking theoretical result concerns the momentum dependence of light mediators. When the mediator mass falls below roughly 10 MeV, the propagator forces the differential cross section to scale approximately as the inverse fourth power of the momentum transfer. Scattering is then overwhelmingly dominated by small momentum transfers, pushing the recoil spectrum to energies right at the threshold of detection. In this regime LZ, with its exceptionally low energy threshold and the coherence boost of spin-independent interactions, delivers the strongest constraints. For heavier mediators the low-momentum enhancement disappears, sensitivity becomes statistics-driven, and high-exposure neutrino experiments such as Borexino become competitive. The authors also show that the common contact-interaction approximation, which assumes the amplitude scales as the inverse fourth power of the mediator mass, breaks down badly once the mediator mass drops below the characteristic momentum transfer of the recoil, rendering dashed-line approximations unreliable exactly where the new physics is most interesting.

Perhaps the most elegant feature of the results is a characteristic turnover in the exclusion limits for mediator masses around ten to the minus two to ten to the minus three GeV. This kink marks the transition between two regimes: above it, the mediator mass term dominates the propagator and the constraints track the mass; below it, the momentum transfer takes over and the limits flatten and shift. The team compared their dark matter bounds with independent constraints on light mediators from meson decays, stellar cooling, and the trapping and cooling limits from supernova SN1987A, noting that recasting those astrophysical and collider bounds onto the dark matter coupling product requires additional assumptions that they adopt consistently from earlier work.

The broader significance is twofold. Practically, the analysis extends the reach of existing detectors, machines built primarily to hunt for WIMPs, deep into the sub-GeV window without pouring a single extra litre of xenon. Conceptually, it demonstrates that momentum dependence is not a technical footnote but a decisive factor in interpreting any direct detection limit involving light mediators, and that the uncertainties in the Galactic dark matter distribution, which make the constraints scale with the square root of the effective propagation distance, must be handled honestly. As boosted dark matter searches mature, the crossover regions identified here, where detector thresholds, overburden attenuation and mediator dynamics all compete, are precisely where the next generation of experiments should focus if the universe’s most abundant matter particle is hiding below a GeV.

Subject of Research: Cosmic-ray boosted sub-GeV dark matter detection constraints via light mediator models

Article Title: New constraints on cosmic-ray boosted sub-Gev dark matter via light mediators

Article References: Yu, Y., Wang, G.-S., Zhang, B., Tang, T.-P., Su, B.-Y., & Feng, L. (2026). New constraints on cosmic-ray boosted sub-Gev dark matter via light mediators. The European Physical Journal C, 86(9), Article 1081. https://doi.org/10.1140/epjc/s10052-026-16368-7

Image Credits: AI Generated

DOI: 10.1140/epjc/s10052-026-16368-7

Keywords: dark matter, cosmic rays, sub-GeV dark matter, light mediators, direct detection, LZ experiment, XENON, Borexino, spin-independent scattering, spin-dependent scattering, particle astrophysics, mediator propagator