IceCubevs.KM3NeT
IceCube vs. KM3NeT: how do the two neutrino telescopes differ?
In short
Both detect Cherenkov light from neutrino interactions in about a cubic kilometre of transparent medium. IceCube uses Antarctic ice and has been complete since 2010; KM3NeT uses Mediterranean sea water, is still under construction and sees the southern sky through the Earth.
| Criterion | IceCube | KM3NeT |
|---|---|---|
| Location | South Pole, Antarctica | Mediterranean Sea: ARCA off Sicily, ORCA off Toulon |
| Medium | Glacial ice | Deep sea water |
| Depth | 1450–2450 m below the ice surface | About 3500 m (ARCA) and 2450 m (ORCA) |
| Size | About 1 km³ | ARCA planned at about 1 km³; ORCA smaller and denser |
| Optical modules | 5160, one 10-inch photomultiplier each | 31 3-inch photomultipliers per module |
| Status | Complete since 2010 | Under construction, already taking data |
| Best up-going view | Northern sky | Southern sky, including the Galactic Centre |
| Low-energy part | DeepCore | ORCA |
| Landmark result | Astrophysical flux (2013), TXS 0506+056, NGC 1068 | KM3-230213A, about 220 PeV (2025) |
Same principle, different media
IceCube and KM3NeT work on the same principle. A neutrino occasionally interacts in or near the detector, and the charged particles it produces emit Cherenkov light that is recorded by a three-dimensional grid of optical modules. IceCube uses the clear glacial ice beneath the South Pole: 5160 modules on 86 strings fill about one cubic kilometre between 1450 and 2450 m depth, and the array has been complete since 2010. KM3NeT instead anchors vertical detection units to the floor of the Mediterranean Sea. Each of its modules holds 31 small photomultipliers rather than one large tube, which adds directional information within a single module.
The media behave differently. Deep ice Cherenkov detectors benefit from long absorption lengths but suffer strong light scattering, while sea water scatters much less, which sharpens the reconstruction of directions, especially for cascade-like events. In return, the sea brings a steady background from potassium-40 decays and bioluminescence, and the detectors must be deployed and connected by ship and remotely operated vehicle.
Different skies and energy ranges
A neutrino telescope sees cosmic sources most cleanly through the Earth, which filters out atmospheric muons. From the South Pole, this up-going view covers the northern sky; from the Mediterranean, it covers most of the southern sky, including the Galactic Centre. The two instruments therefore complement each other. Each also has a dense low-energy part: IceCube's DeepCore and KM3NeT's ORCA site off Toulon measure atmospheric oscillations and contribute to the mass ordering, while ARCA off Sicily is optimised for TeV to PeV astrophysical neutrinos.
Results so far
IceCube established the diffuse astrophysical neutrino flux in 2013, linked a neutrino to the blazar TXS 0506+056 in 2017–2018, found evidence for emission from the galaxy NGC 1068 in 2022 and observed the plane of the Milky Way in neutrinos in 2023. KM3NeT is still being built while taking data. Even so, the partially completed ARCA recorded KM3-230213A, published in 2025, with an estimated energy of about 220 PeV, the most energetic neutrino observed so far. How it fits with IceCube's non-detection of comparable events is still being studied. IceCube plans to grow further with IceCube-Gen2.
Read on
Related terms
- IceCube Neutrino Observatory
- KM3NeT
- Ice / Deep-water Cherenkov telescope
- DeepCore & IceCube upgrade
- KM3-230213A
- IceCube-Gen2
Last updated: · Editorial standard
Cite this page
Text
Neutrino Atlas (2026). “IceCube vs. KM3NeT: how do the two neutrino telescopes differ?”. Retrieved 22 November 2026, from https://neutrino-atlas.com/comparisons/icecube-vs-km3net/BibTeX
@misc{neutrinoatlas2026mparisonsicecubevskm3net,
title = {IceCube vs. KM3NeT: how do the two neutrino telescopes differ?},
author = {{Neutrino Atlas}},
year = {2026},
url = {https://neutrino-atlas.com/comparisons/icecube-vs-km3net/},
note = {Accessed 2026-11-22},
language = {english}
}