Scientists used high-energy neutrinos from the inner Milky Way to search for very light dark matter. The idea is simple: fast-moving cosmic rays hitting low-mass dark matter can break particles apart in a process called deep inelastic scattering. That produces short‑lived mesons, which then decay and release neutrinos. The IceCube and ANTARES neutrino observations of the Galactic plane and the so‑called Galactic Ridge make this neutrino signal measurable in principle.

The authors calculated how many neutrinos would come from collisions between cosmic‑ray nuclei (mainly protons and helium) and sub‑GeV dark matter in the Galactic halo. They used realistic maps of how cosmic rays are distributed across the Galaxy, and a standard dark matter density profile known as Navarro–Frenk–White with a scale radius of 20 kpc and a local density of 0.4 GeV per cubic centimetre. The predicted neutrino spectra were produced with particle‑physics simulation tools (MadGraph5_aMC@NLO for the hard collisions and Pythia8 for the subsequent particle showers and decays).

To compare prediction and data, the team focused on the Galactic Ridge region (Galactic longitude |l|<30° and latitude |b|<2°), where ANTARES reports neutrino limits based on 13 years of data (2007–2020). ANTARES characterizes the Ridge emission in six energy bins and allows for a wide range of spectral shapes. The authors required that the neutrinos from cosmic‑ray–dark‑matter scattering not exceed the 99% confidence‑level upper limits reported by ANTARES in any energy bin. They used a benchmark particle physics model with a Dirac dark matter particle interacting via a vector mediator to map those flux limits into bounds on the dark‑matter–nucleon cross section.

The result is a new set of upper limits on the dark‑matter–nucleon interaction strength that reach down to keV‑scale dark matter masses. In other words, neutrino telescopes looking at the Galactic Center can probe dark matter far lighter than what many ground experiments can detect. The authors emphasize that future instruments, like IceCube‑Gen2 and the KM3NeT detector, should considerably improve the sensitivity of this method.