Why put a neutrino experiment under rock, sea or ice?

The surface is crowded with cosmic-ray particles

A rare neutrino signal has to be picked out of a far more frequent stream of cosmic-ray particles. Cosmic rays striking the atmosphere make showers of muons and other secondary particles. A muon can enter a detector and produce light or ionization, or create radioactive isotopes and neutrons nearby.

Rock or water above the experiment stops many of these particles before they arrive. How many muons still get through depends on the column density of material along their path, their direction, the altitude and the incoming energy spectrum. Two sites at the same vertical depth can therefore shield differently.

That is why depth is often given as a water-equivalent column density, not just metres of rock. The number compares shielding only; two experiments at the same depth can still have different backgrounds. Which leftover particles and energies matter depends on what the experiment sets out to measure.

Sources: [1] · [2] · [3]

A quiet mountain can still hold radioactive dust

Some backgrounds start inside the shielding: uranium and thorium decay chains, radon, and radioactivity in the detector's own materials. Going deeper does little about contamination already next to the sensitive volume. Clean handling, low-radioactivity components and material screening are each a separate part of planning a site.

SNOLAB runs its underground space as one large clean room. People and equipment coming in from the mine are cleaned before they enter. The laboratory's official facility description gives 2,070 metres of rock overhead. The entry procedure is there to protect the measurement environment.

At Gran Sasso, Borexino could isolate low-energy solar events only with a highly purified scintillator and tight control of radioactivity inside the detector. Shielding outside and purification inside deal with different backgrounds. Even with atmospheric muons strongly suppressed, trace contamination can still limit a low-energy spectrum.

Sources: [1] · [4] · [5]

Sea and ice can become the instrument

High-energy neutrinos interact so rarely that a telescope for them needs an enormous volume. Sensors placed in natural water or ice use it as both the target and the medium that carries the light, so no tank of that size has to be built. A sparse array reconstructs charged-particle tracks from the times and amounts of light each sensor detects.

The water or ice itself has to be measured. In ice, scattering and absorption change with depth. Sea water has its own optical properties, plus background light from radioactive decays and bioluminescence. Currents also move the detector lines, so their positions must be tracked along with the clock and light calibration.

Once IceCube's sensors froze into the ice, their positions stayed fixed, but most of the buried hardware can no longer be reached. Deep-sea detectors face hard problems with deployment, pressure, connections and maintenance. These choices shape the calibration program, operating reliability and how well events can be reconstructed.

Sources: [3] · [6] · [7]

Depth and event selection work together

Every cut that removes background can also remove real neutrinos or add dead time. A veto layer flags particles coming in from outside. Keeping only events in an inner volume cuts contamination near the detector's edges. Delayed signals and links to a muon that passed just before give further ways to reject background.

A telescope that keeps only upward-going tracks lets the Earth block ordinary atmospheric muons from that direction. Atmospheric neutrinos still pass through the Earth, so they remain a background in searches for cosmic sources. At high enough energy the Earth absorbs neutrinos too, and that absorption has to be modeled.

Downward-going events have no single direction cut; analyses combine containment, veto information, energy and reconstruction quality. The useful exposure therefore depends on direction and energy. A deeper site makes selection easier, but the analysis still needs a model of which events its selection accepts.

Sources: [3] · [8] · [9]

The scientific goal decides the site

A solar experiment cares most about a stable, low background at low energy. A cosmic-neutrino telescope cares most about large volume and wide directional coverage. A short-baseline reactor experiment must also sit close to an intense source; it sometimes accepts shallow shielding and makes up for it with a segmented detector and selection of correlated events.

The best site is not necessarily the deepest or most remote. Power, ventilation, transport, data links and access for calibration decide how much useful running time an experiment can sustain. The site has to let a calibrated instrument study the source and energy range it was built for.

A facility's location says where it is, not where the expertise and work come from. Local operators, international engineers and analysis teams each maintain a different part of the same measurement. Seen this way, choosing a location is both a scientific design decision and a lasting commitment to operate there.

Sources: [1] · [6] · [10] · [3]

Try it in the Lab

Primary sources & revision

1. SNOLAB · Underground facilities
2. INFN · Gran Sasso laboratory overview
3. IceCube Collaboration · Instrumentation and Online Systems (2017; arXiv v3)
4. Borexino Collaboration · Detector at Gran Sasso (2009)
5. Benziger et al. · Borexino scintillator purification system (2008)
6. KM3NeT Collaboration · Letter of Intent for KM3NeT 2.0 (2016)
7. IceCube · Observatory overview
8. IceCube Collaboration · Evidence for High-Energy Extraterrestrial Neutrinos (2013)
9. IceCube Collaboration · Evidence for neutrino emission from NGC 1068 (2022; arXiv v2, 2024)
10. PROSPECT Collaboration · Experimental science and detector design

First published 2026-10-11; last revised 2026-10-11. Original explanatory text and diagrams by Sandbox Physics. Illustrations are schematic; no experimental event records are reproduced here. This is an independent educational publication, not an official Nobel or experiment collaboration publication.