Neutrino flavor changes could reshape tests of dark matter interactions

Neutrinos come in three flavors: electron, muon and tau. These are not identical to the states with definite masses that govern their propagation. The mismatch allows neutrinos to change flavor as they travel. A new theoretical study asks how that journey would change if dark matter interacted preferentially with just one flavor.

In their arXiv preprint, submitted on 6 November 2026, Ki-Young Choi and Ankhbayar Tugssaikhan combine two consequences of such an interaction: altered neutrino mixing and loss of neutrinos through scattering. This article is based on the supplied complete paper text and original arXiv metadata. The work presents model calculations and conditional constraints, not evidence that neutrino–dark matter interactions have been detected.

Two effects from the same dark environment

The researchers use the blazar TXS 0506+056 as a benchmark. They assume neutrinos are emitted inside a dense dark matter spike surrounding its supermassive black hole. Their adopted emission radius is 0.023 parsecs, where the modeled dark matter density is about 640 billion GeV per cubic centimeter. This density profile is an input to the calculation, not a measurement established by the study.

Dark matter can affect neutrinos in two distinct ways. Coherent forward scattering changes the effective conditions for flavor mixing without removing neutrinos from the beam. The authors describe this through a potential proportional to the local dark matter density, with a strength coefficient called A. Separately, scattering can attenuate the observable flux.

The analysis considers interactions with electron, muon or tau flavor individually. It mainly evaluates neutrinos at 100 TeV and assumes normal neutrino mass ordering. It also compares two production patterns: pion decay producing an initial electron-to-muon-to-tau ratio of 1:2:0, and a muon-damped source producing only muon neutrinos in the adopted model.

Why stronger interactions need not erase the whole beam

For the assumed density profile, the researchers find that conditions change slowly enough for propagation to be adiabatic. Each propagation state follows its continuously evolving counterpart as the dark matter density falls, rather than jumping between states.

When the potential is negligible, all three propagation states contain some of the interacting flavor. Scattering therefore attenuates all three, though at different rates. Comparing the predicted flavor fractions with IceCube’s published contours gives cross-section-to-dark-matter-mass limits around 10⁻²⁹–10⁻²⁸ square centimeters per GeV, depending on the scenario and confidence contour.

A sufficiently strong potential changes this picture. Inside the dense spike, it concentrates the interacting flavor almost entirely into one propagation state. The other two then contain very little of that flavor and can escape the inner region with little attenuation.

This produces a survival plateau: removing the strongly interacting state alone need not extinguish most of the initial flux. Farther out, mixing changes again, and the surviving states regain components of the interacting flavor. Larger scattering cross sections can eventually attenuate them there.

For the pion-decay benchmark at A = 10⁻³⁰ cubic centimeters, the cross-section-to-mass values needed to remove 90% of the total flux rise to roughly 1.4–5.6 × 10⁻²⁵ square centimeters per GeV. The authors report that the flux-based bound can weaken by up to four orders of magnitude. That does not mean flavor-composition limits weaken by the same amount.

Flavor still provides a separate test

The potential can change the flavor mix even without appreciable attenuation. Neutrinos and antineutrinos may experience potentials of the same or opposite signs, connecting their source states to different mass states at Earth. The calculation sums their contributions when comparing with IceCube.

The source composition matters sharply. For a muon-damped source interacting only through muon flavor, a sufficiently large potential already moves the predicted composition outside IceCube’s 95% contour before scattering is added. In this case, concentrating the initially populated flavor into one state does not protect the flux: that state is precisely the one strongly attenuated.

Conditional constraints, not universal exclusions

Several assumptions limit the interpretation. The dark matter spike and emission radius set how much material neutrinos encounter. The study compares a source benchmark with IceCube’s all-sky flavor contours rather than performing a full source-specific likelihood analysis. It uses a representative energy instead of a continuous spectrum and assumes scattered neutrinos leave the observed flux; regeneration at lower energies could weaken attenuation limits.

The authors also restrict A to at most 10⁻³⁰ cubic centimeters to keep local dark matter effects on mixing small at their benchmark energy. This is an analysis choice motivated by consistency, not a newly measured universal bound. The central result is that flavor mixing and flux loss must be considered together when testing these dark matter interaction scenarios.