The Milky Way Lights Up in Neutrinos: IceCube's 2023 Galactic Plane Detection

The Milky Way’s stars, gas, and dust dominate the visible-light sky. They also dominate the radio sky, the infrared sky, the X-ray sky, and the gamma-ray sky. Each electromagnetic band shows a different aspect of the Galaxy: hot stars in the visible, cold dust in the infrared, hot plasma in the X-ray, accelerated charged particles via inverse-Compton in the gamma-ray.

For decades, neutrino astronomers have asked whether the Galaxy would also be visible in neutrinos. The expectation: cosmic-ray protons accelerated within the Galaxy collide with interstellar gas, producing pions. The pions decay to neutrinos and gamma rays. The integrated emission from this process should produce a diffuse signal along the Galactic plane, peaking at energies in the few-TeV to 100-TeV range. The flux is small but non-zero.

In June 2023, the IceCube collaboration published an article in Science (Vol. 380, Issue 6651, pp. 1338-1343) reporting the first detection of this Galactic-plane neutrino emission. The significance was 4.5$\sigma$ — formally not yet a discovery in the strict sense (5$\sigma$), but well above the typical “evidence” threshold and consistent with the predicted spectrum and angular distribution.

For the first time, our own galaxy has been imaged in neutrinos. This is not the first detection of neutrinos from cosmic sources (that was the diffuse extragalactic flux, established by IceCube around 2013) nor the first identification of a specific astrophysical source (TXS 0506+056 in 2018, NGC 1068 in 2022). It is, however, the first detection of a diffuse Galactic source — the integrated emission from cosmic-ray accelerators distributed throughout our galaxy.

This post is about how the detection was made, what it implies for cosmic-ray physics, and what it means for the broader neutrino-astronomy programme.

Why expect Galactic neutrinos

Cosmic rays — high-energy charged particles, predominantly protons — have been observed at Earth for over a century. Their origin remains imperfectly understood. The standard picture: most Galactic cosmic rays originate from supernova remnants, where shock-front diffusive acceleration produces a power-law spectrum extending to energies of approximately $10^{15}$ eV (the “knee” of the cosmic-ray spectrum, where the slope steepens). Above the knee, the origin is more uncertain, with possibilities including extreme galactic events (e.g., supernova interactions with massive stellar winds) or extragalactic sources.

For Galactic cosmic rays at energies of $10^{12}$–$10^{14}$ eV, the propagation through the interstellar medium is well-studied via gamma-ray observations. The interactions are dominantly hadronic: $p+p\to {\pi }^{\pm }+X\to {\mu }^{\pm }+\nu \to e^{\pm }+2\nu$ or $p+p\to {\pi }^0+X\to 2\gamma$

The pion-decay channel produces both gamma rays and neutrinos. The ratio of gamma-ray to neutrino flux depends on the relative branching of charged versus neutral pion production and the kinematics of the decay chains. Theoretical predictions, including more sophisticated treatments of the cosmic-ray spectrum and propagation, give an expected diffuse Galactic-plane neutrino flux of order $10^{-7}$–$10^{-8}$ TeV/cm²/s/sr, integrated over a band of width approximately ±10° around the Galactic plane.

This flux is small but, with a kilometre-cubic detector and approximately a decade of exposure, in principle detectable.

Detection challenges

Two main challenges had previously prevented detection:

Atmospheric background. Cosmic-ray induced atmospheric neutrinos at TeV energies are produced predominantly above the Earth’s atmosphere — including in the direction of the Galactic plane as seen from the South Pole (IceCube’s location) and the Mediterranean (KM3NeT). The atmospheric background, integrated over the same angular region, is much larger than the expected Galactic-plane signal. Distinguishing the signal from this background requires careful angular and energy modeling.

Angular resolution. Cascade events (neutral-current and electron-neutrino charged-current) have angular resolution of order 5-10°. Track events (muon-neutrino charged-current) have much better angular resolution (~0.5°) but are concentrated in different directions on the sky depending on the analysis. The typical IceCube map at PeV-TeV energies has angular resolution similar to or worse than the apparent width of the Galactic plane (~10°).

Both challenges have been addressed in recent years through improved reconstruction algorithms, particularly machine-learning-based techniques that more efficiently use the spatial and temporal information of light-scattering events in the ice.

The 2023 analysis

The IceCube 2023 analysis used 60,000+ cascade events from approximately 10 years of detector running. The events were processed with a deep-learning reconstruction (DNN) that improves the angular resolution by approximately a factor of 2-3 compared to traditional reconstructions. After applying quality cuts and energy thresholds, the resulting sample is composed primarily of astrophysical neutrinos plus some atmospheric background.

The analysis divided the sky into pixels and looked for excess emission along the Galactic plane. Two parameterisations were used:

• A specific spectral and angular model (e.g., the KRA gamma-ray model for cosmic-ray-induced emission)
• A model-independent search for excess emission within a few degrees of the Galactic plane

The latter analysis found an excess at 4.5$\sigma$ above the no-galactic-emission hypothesis. The angular distribution was consistent with the expected concentration toward the Galactic centre. The energy spectrum, while not precisely measured, is consistent with the expected $E^{-2.7}$ power law from cosmic-ray-induced emission.

This is the first detection of a diffuse Galactic neutrino source.

What this confirms

The result confirms several aspects of cosmic-ray physics:

Cosmic-ray protons are accelerated within the Galaxy to energies sufficient to produce neutrinos via interactions with interstellar gas. This is consistent with the supernova-remnant picture but does not specifically identify supernova remnants as the source — only that hadronic acceleration occurs throughout the Galaxy at the relevant energies.

The hadronic interaction model is correct. The neutrino flux is consistent with the gamma-ray flux from the same regions, scaled by the appropriate hadronic-versus-electromagnetic branching ratios. This is a non-trivial check: the gamma rays could be produced by either hadronic or leptonic processes (the latter not producing neutrinos). The observed neutrino flux confirms a substantial hadronic component.

The Galactic-plane cosmic-ray distribution is consistent with conventional models of cosmic-ray transport. The angular distribution of neutrino emission is what is expected from cosmic rays diffusing through the Galaxy with the inferred density profile.

What it does not yet tell us

The detection is at 4.5$\sigma$ — substantial but not yet a precision measurement. Several questions remain:

Source population identification. The diffuse signal is integrated over the entire Galactic plane. Specific source classes (supernova remnants, pulsar wind nebulae, gamma-ray binaries) cannot be individually identified from the diffuse emission alone. Whether the dominant contribution is from supernova remnants or from a different class of accelerators is still open.

Energy spectrum precision. The 2023 analysis was sensitivity-limited, not precision-limited. The energy spectrum is constrained only at the order-of-magnitude level. Future analyses with more events and better-calibrated detector response will refine this.

Comparison with extragalactic flux. The Galactic-plane flux constitutes only a small fraction of the total astrophysical neutrino flux observed by IceCube. The bulk is extragalactic, dominated by sources beyond the Milky Way. The relative contributions of Galactic versus extragalactic sources at different energies remains to be quantified.

What’s next

Several developments will refine the picture in the coming years:

IceCube continued running. By 2030, the integrated exposure will more than double. The Galactic-plane signal significance should approach 7-8$\sigma$, and the spectral measurement should reach precision sufficient to test specific source models.

KM3NeT (Mediterranean) is approaching full deployment. Once operational, it will provide independent measurements of the Galactic-plane signal from the Northern hemisphere (where the South-Pole-based IceCube has limited coverage). The combination of IceCube and KM3NeT will give the first all-sky neutrino map with precision sufficient for source-population studies.

IceCube-Gen2 (planned expansion) will increase the instrumented volume by a factor of 8 by approximately 2032. This will dramatically improve the sensitivity to point sources within the Galactic plane and to spectral features in the diffuse emission.

Multi-messenger correlation. The Galactic-plane neutrino emission can be correlated with gamma-ray observations from existing facilities (HESS, MAGIC, VERITAS, the upcoming CTA) and with ultra-high-energy cosmic-ray observations (Auger, Telescope Array). Such correlations will test the hadronic-source-emission framework with multiple independent observables.

By 2035, the diffuse Galactic neutrino emission should be a well-characterised observable, with the source-population contributions identified and the cosmic-ray transport model precisely constrained. The 2023 detection is the foundation for this longer programme.

A new branch of astronomy

For the past century, astronomy has been done primarily with electromagnetic radiation — visible light, radio, infrared, X-ray, gamma-ray. Each band reveals different physics. The neutrino sky, by contrast, is sensitive to specific processes (hadronic interactions of accelerated particles) that are difficult to isolate in electromagnetic observations.

The Galactic-plane neutrino detection is, in this sense, a new branch of astronomy. We can now image our own galaxy in a fundamentally new way. The view is dim — the integrated TeV-PeV neutrino emission from the entire Milky Way produces only a few hundred IceCube events per year — but it is a real image.

The image confirms what cosmic-ray physicists have inferred for decades: the Galaxy is filled with accelerated charged particles, interacting with the interstellar gas, producing neutrinos. This is the integrated signature of cosmic-ray-source physics, observed for the first time.

For the broader high-energy astrophysics community, the result is a confirmation of the hadronic origin of a substantial fraction of the high-energy cosmic-ray flux. The identification of specific sources (TXS 0506+056, NGC 1068) plus the diffuse Galactic-plane detection together establish that the universe at PeV energies is dominated by cosmic-ray-induced hadronic processes — exactly as the standard picture has predicted.

The 2023 detection is, then, the moment when neutrino astronomy moved from a niche speciality to a fully realized observational science. The Milky Way has been seen in neutrinos. The methodology is established. The next decade is about deeper, sharper images.