Diffuse astrophysical neutrinos from dark matter around blazars

TL;DR

The paper investigates a sub-GeV dark matter (DM) scenario in which inelastic scatterings between jet protons in blazars and DM surrounding the supermassive black hole produce high-energy neutrinos. By modeling blazar jets with one-zone lepto-hadronic spectra and DM spikes around SMBHs, the authors compute neutrino fluxes for four DM–quark mediators ($V$, $V'$, $\phi$, $a$) and: (i) the TXS 0506+056 neutrino, and (ii) a stacked ensemble of 324 blazars, with the DM flux boosted to Earth and redshifted. They find that DM–p DIS can saturate the IceCube diffuse high-energy neutrino flux and also reproduce the TXS 0506+056 signal within current constraints, across mediator types, opening a testable link between sub-GeV DM searches and astrophysical neutrinos. The work outlines multiple observational and laboratory tests, including sub-GeV DM probes, correlations between IceCube events and blazars, and complementary DM-induced photon signals, with a companion paper exploring blazar-boosted DM signals in more detail.

Abstract

Neutrinos from blazars can originate from inelastic scatterings between protons within their jets and sub-GeV dark matter (DM) around them, explaining IceCube detections of neutrinos from TXS 0506+056 that are otherwise challenging for models of its jet. In this paper we calculate such DM-induced high-energy neutrinos, from TXS 0506+056 as well as from a stacked blazar sample, in the four cases where DM-quark interactions are mediated by a new massive vector, axial, scalar, and pseudoscalar particle. Intriguingly, we find that this mechanism can saturate the diffuse astrophysical neutrino flux observed by IceCube at high energies. Our mechanism will be tested by additional blazar observations and by various searches for sub-GeV DM.

Figures (3)

• Figure 1: Neutrino fluxes from blazar TXS 0506+056 induced by deep inelastic scatterings between protons within its jet and dark matter around it, using as input the jet parameters as fitted in Keivani:2018rnh. The four coloured lines correspond to the four mediators (vector, axial, scalar and pseudoscalar) of DM-quark interactions defined in Sec. \ref{['sec:toymodels']}, for a common choice of coupling $g_{\chi Y}g_{q Y} = 3.5\times 10^{-2} \;(6.5\times 10^{-3})$, DM mass $m_\mathsmaller{\mathrm{DM}}=100$ MeV (1 MeV), mediator mass $m_Y=5$ GeV (where $Y=V,V',\phi,a$) and line-of-sight integral $\Sigma_{\text{DM}} = 1.5 \times 10^{26} \;\mathrm{GeV}\;\mathrm{cm}^{-2}$ (i.e. our conservative benchmark BMCII defined in Sec. \ref{['sec:LOS']}). The yellow band is the diffuse astrophysical $\nu_\mu + \bar{\nu}_\mu$ flux measured by IceCube (95% C.L.) Abbasi:2021qfz, the blue star is the best-fit flux from the TXS 0506+056 neutrino and the blue line its uncertainty IceCube:2018cha. In grey the limits set by 9 year of IceCube data from the null detection of extremely-high-energy (EHE) neutrinos IceCube:2018fhm, rescaled to 6 months.
• Figure 2: Neutrino flux from $p-\mathrm{DM}$ deep inelastic scatterings around 324 blazars that have been fitted with single-zone lepto-hadronic jet models in Rodrigues:2023vbv. The lines correspond to spin-1 DM mediators with mass $m_Y= 5 \;\mathrm{GeV}$, for $m_\mathsmaller{\mathrm{DM}} = 10 \;\mathrm{MeV}$, $g_{\chi Y}g_{q Y} = 1.5\times 10^{-2}$, $R_\mathsmaller{\mathrm{min}} = 10^4 R_S$. The case of spin-0 mediators is a simple rescaling of the spin-1, as in Fig. \ref{['fig:flux_comparison']}. Blazars in the sample are separated into FSRQ (green) and BL Lac (purple) objects, the cumulative flux of all of them except the two outliers (see text) is shown as a thick blue line. In black is the neutrino flux computed in Rodrigues:2023vbv as a result of $p-\gamma$ interactions, with no DM. The gray shaded area is excluded by the null detection of extremely-high-energy (EHE) neutrinos in 9 year of IceCube data IceCube:2018fhm, the yellow band is as in Fig. \ref{['fig:flux_comparison']}.
• Figure 3: DM parameter space of Eqs. (\ref{['eq:scalar']})--(\ref{['eq:axial']}) for mediator mass $m_Y=500$ MeV (left column) and $m_Y=5$ GeV (right column), where $\sigma_\mathsmaller{\mathrm{NR}} = (g_{\chi Y}^2 g_{N Y}^2/\pi)(\mu_{\chi N}^2/m_Y^4)$ and $g_{N Y}$ are functions of $g_{q Y}$ that we take from DeMarchi:2025uoo. High-energy diffuse neutrinos Abbasi:2021qfz and the IC-170922A neutrino from TXS 0506+056 IceCube:2018dnn are explained, respectively, along the green and blue lines, where their width accounts for uncertainty on the energy $E_\nu$ of the TXS 0506+056 neutrino and on the diffuse neutrino flux, which itself is shown as a shaded band in Fig.\ref{['fig:flux_comparison']}, \ref{['fig:neutrino_flux_stacking']}. Continuous (dashed) lines correspond to BMCI (BMCII) for the DM LOS integral, see Sec. \ref{['sec:LOS']}. The parameter space on the left of the 'BBN' dashed line is excluded by big bang nucleosynthesis Sabti:2019mhn, unless the universe is reheated below the QCD scale Berlin:2018ztp. Direct detection (DD) of halo DM SuperCDMS:2023sqlCRESST:2019jnqDarkSide:2018bpjLZ:2024zvoNEWS-G:2024jmsXENON:2023cxcXENON:2025vwd excludes the gray regions in the top-right of each plot. Note that DM-nucleon scatterings are spin-independent (SI) if induced by $V$ and $\phi$, spin-dependent if by $V'$, and momentum and spin-dependent if by $a$, so that DD limits in the $a$ case come from loop-induced SI scatterings Abe:2018emu. The other gray regions are excluded as follows. $V$: by monojet at Tevatron and LHC Shoemaker:2011vi; $\phi$: by DM searches from proton beam dump at MiniBooNE MiniBooNE:2017nqe recast for a scalar mediator decaying invisibly Batell:2018fqo, by monojet at LHC ATLAS:2020wzf recast in Ema:2020ulo and by $K\to\pi$ invisible decays Cox:2024rew which we do not shade as it can be evaded by specific combinations of $g_{u\phi}$ and $g_{d\phi}$Pascoli:atmospheric; $V'$: by monojet at Tevatron and LHC Shoemaker:2011vi; $a$: by monojet at LHC ATLAS:2020wzfEma:2020ulo, invisible meson decays ParticleDataGroup:2024cfk and CR upscattering (CRDM) Ema:2020ulo. All lines and limits depend on $g_{\chi Y} g_{q Y}$, except monojet limits if $m_\mathsmaller{\mathrm{DM}} < m_Y$, in which case we assume $g_{\chi Y} = 1$. See Sec. \ref{['sec:tests']} for more details.