Super-Kamiokande: How 50,000 Tonnes of Water Proved Neutrinos Have Mass

A Tank of Ultrapure Water Under a Mountain

Super-Kamiokande sits in an active zinc mine beneath Mount Ikeno, in Hida, Gifu Prefecture. The rock above it is a kilometre thick, equivalent to about 2,700 metres of water in shielding power. That overburden is the entire reason the site was chosen: it absorbs almost all of the cosmic-ray muons that would otherwise flood the detector, cutting the rate by a factor of roughly a hundred thousand.

The detector itself is a stainless steel cylinder 39.3 metres in diameter and 41.4 metres tall, filled with 50,000 tonnes of water. The word ultrapure is doing real work here. The water is continuously recirculated and filtered to remove dissolved gases, ions and, above all, traces of radon and uranium, because natural radioactivity in the water would produce a constant background indistinguishable from a low-energy signal.

The tank is optically divided in two. An inner detector holds the fiducial volume used for physics, watched by roughly 11,100 photomultiplier tubes of 50 centimetres diameter - about 40 percent of the inner wall is photocathode. Surrounding it, an outer detector layer with 1,885 smaller tubes acts as a veto: anything that lights up the outer layer first came from outside, and is therefore not a neutrino interaction worth keeping.

The facility is operated by the Institute for Cosmic Ray Research at the University of Tokyo. Its predecessor, Kamiokande, occupied the same mine and detected a dozen neutrinos from supernova 1987A - a result that brought Masatoshi Koshiba a share of the 2002 Nobel Prize and established that supernova neutrinos could be caught at all.

Why Water, and Why It Has to Be This Clean

Water is an unglamorous detector medium with excellent properties. It is cheap enough to buy 50,000 tonnes of, transparent to blue light over tens of metres once purified, dense enough to give neutrinos something to hit, and free of the intrinsic radioactivity that plagues many solids. It is also, unlike the Antarctic ice used by IceCube, completely uniform - there are no dust layers to map and no frozen-in optical structure to model.

The trade-off runs the other way too. Water scatters light less than ice but absorbs it more, which limits how far apart the sensors can be. That is one reason water detectors are measured in tens of thousands of tonnes while ice detectors are measured in cubic kilometres: the two media push the engineering in different directions, and the physics each one can reach follows from that.

Purity is not a one-time task. The water is cycled through a filtration plant continuously, and even so, the collaboration has to model how radon seeps in from the surrounding rock and the tank walls. The upper region of the detector is measurably less clean than the centre, which is why the fiducial volume used for the most sensitive analyses is smaller than the tank itself - about 22,500 tonnes rather than 50,000.

The result of all that care is a medium in which a single charged particle produces a signal that a human could, in principle, see. Photographs of the drained tank, with technicians in a rubber dinghy inspecting the golden wall of photomultipliers, are among the most recognisable images in physics precisely because the scale is so tangible.

Reading a Ring of Light

When a neutrino interacts with a nucleus in the water, it produces a charged particle - typically an electron or a muon - moving faster than light moves through water. That particle emits a cone of Cherenkov radiation, and where the cone meets the wall of the tank, the photomultiplier tubes record a ring.

The ring is the measurement. Its centre gives the direction of the particle. Its total brightness gives the energy. And, crucially, its sharpness identifies which particle made it. A muon is heavy and ploughs through the water in a straight line, producing a clean, sharply defined ring. An electron is light, scatters repeatedly and triggers a small electromagnetic shower, so its ring comes out fuzzy at the edges.

That single distinction is what made the 1998 discovery possible. Being able to tell an electron event from a muon event, reliably and event by event, is what allowed the collaboration to count muon neutrinos separately from electron neutrinos - and to notice that one population was disappearing while the other was not.

The technique has limits worth knowing. Super-Kamiokande cannot see neutral particles, cannot determine the sign of the charge, and its energy threshold sits in the range of a few MeV, set by the radioactive background rather than by the optics. Those constraints shape which questions it can answer, and they are the reason different neutrino observatories are built around different technologies.

The Instrument in Numbers

The figures below describe the detector in its current configuration. It has been running, with two significant interruptions, since 1 April 1996 - approaching three decades of continuous operation.

• Location: Kamioka mine, Hida, Gifu Prefecture, Japan, 1,000 metres underground (about 2,700 metres water equivalent)
• Tank: stainless steel cylinder, 39.3 metres diameter by 41.4 metres height
• Medium: 50,000 tonnes of ultrapure water; roughly 22,500 tonnes used as fiducial volume
• Inner detector: approximately 11,100 photomultiplier tubes of 50 centimetres diameter, about 40 percent photocathode coverage
• Outer detector: 1,885 photomultiplier tubes of 20 centimetres, used as a cosmic-ray veto
• Operating since: 1 April 1996, run by the Institute for Cosmic Ray Research, University of Tokyo
• Energy range: from a few MeV for solar neutrinos up to the multi-GeV atmospheric range
• Since 2020: loaded with gadolinium sulfate to make neutron capture visible

1998: The Result That Changed the Standard Model

Cosmic rays striking the upper atmosphere produce showers containing muon and electron neutrinos in a ratio that theory predicts quite precisely - roughly two muon neutrinos for every electron neutrino. Super-Kamiokande counted them, and found a deficit of muon neutrinos. That alone was not new; earlier experiments had hinted at it.

What Super-Kamiokande added was direction. Because it could reconstruct where each neutrino came from, it could compare neutrinos produced overhead - which had travelled about 15 kilometres - with those produced on the far side of the planet, which had travelled roughly 13,000 kilometres. The overhead population was as expected. The population that had crossed the Earth was depleted by about half.

Distance was the variable. Neutrinos were not being absorbed by the Earth; they were changing identity along the way, and the longer they travelled the more of them had changed. This is neutrino oscillation, and quantum mechanics only permits it if the different neutrino types have different masses - which requires that at least two of them have mass at all.

The Standard Model as written did not allow that. The 1998 result was therefore the first laboratory evidence of physics beyond it, and it remains one of the very few such results. Takaaki Kajita received half of the 2015 Nobel Prize in Physics for it; the other half went to Arthur McDonald of the Sudbury Neutrino Observatory, whose measurement of solar neutrinos closed the argument from the other side.

November 2001: Losing Two-Thirds of the Detector in Seconds

On 12 November 2001, while the tank was being refilled after maintenance, one photomultiplier tube at the bottom imploded. Each tube is an evacuated glass bulb half a metre across, and at depth the surrounding water pressure is considerable. The implosion sent a shock wave through the water that shattered its neighbours, whose implosions shattered theirs.

The chain reaction took a matter of seconds and destroyed roughly 6,600 of the approximately 11,100 inner tubes - close to two-thirds of the detector. Nobody was hurt. The cause was not a design error in any single component but a failure to anticipate that the components could destroy each other collectively.

The collaboration rebuilt. The surviving tubes were redistributed to give uniform but thinner coverage, and Super-Kamiokande-II resumed running in 2002 with about half its former photocathode area - degraded, but taking data. Full coverage was restored in 2006. Every tube is now enclosed in a fibre-reinforced acrylic shell designed to contain an implosion rather than propagate it.

The episode is worth recording because it is a rare, well-documented case of a cascading failure in a scientific instrument, and because the response - rebuild at reduced capability rather than shut down and wait - kept a decade of physics on schedule.

What Else Super-Kamiokande Does

It is the far detector for T2K, an experiment that fires a beam of muon neutrinos from the J-PARC accelerator complex 295 kilometres away on the east coast of Japan. Because the beam's composition and timing are known exactly, comparing what arrives with what was sent gives a controlled oscillation measurement - and T2K has used it to search for differences between neutrinos and antineutrinos, one of the routes towards understanding why the universe contains matter rather than nothing.

It sets the world's most stringent limits on proton decay. Grand unified theories predict that the proton is not perfectly stable; Super-Kamiokande has watched a vast number of protons for decades and seen none decay, pushing the partial lifetime for the favoured channel beyond 10^34 years. That is a null result, and it has ruled out entire families of theoretical models.

It measures solar neutrinos in real time and with direction, confirming that they come from the Sun and tracking the flux continuously. And it stands ready for the next galactic supernova as part of the international early warning network, where its ability to reconstruct direction could tell astronomers where to point before the light arrives.

Since 2020 the water has been loaded with dissolved gadolinium, first at 0.01 percent and later increased. Gadolinium captures free neutrons and emits a distinctive gamma cascade, which lets the detector tag the neutron that accompanies an electron antineutrino interaction. That upgrade sharpens the search for the diffuse supernova neutrino background - the faint accumulated glow of every core-collapse supernova in the history of the universe, which has never been detected.

What Super-Kamiokande Does Not Tell Us

The 1998 result established that neutrinos have mass. It did not establish how much. Oscillation experiments are sensitive to the differences between the squared masses, not to the masses themselves, so they cannot say whether the lightest neutrino weighs something or nothing. Answering that requires a different kind of experiment altogether, which is why direct-measurement efforts exist alongside the oscillation programme.

Nor does the detector's scale imply that neutrino interactions are abundant. The opposite inference is the correct one: 50,000 tonnes of water, watched continuously for nearly thirty years, is what it takes to accumulate a statistically useful sample. The instrument is a monument to how weakly these particles couple to matter.

That distinction matters when Super-Kamiokande is cited in discussions of energy. The Nobel-winning result is about the properties of the neutrino - that it has mass and changes type - and not about extracting work from it. Neutrinovoltaic research asks a different question: whether the persistent flux of ambient radiation and thermal fluctuations at a material surface can drive a measurable current in an engineered multilayer. Super-Kamiokande neither supports nor refutes that; it is simply not the experiment that addresses it.

What it does supply is the foundation the whole field stands on. Before 1998 the neutrino was a particle with no mass and no internal structure to speak of. After 1998 it became the one confirmed crack in the Standard Model, and the reason a great deal of subsequent physics got funded.

Hyper-Kamiokande and What Comes Next

The successor is under construction in the same mountain range. Hyper-Kamiokande will hold roughly 260,000 tonnes of water, about five times Super-Kamiokande's mass, with a fiducial volume around eight times larger, and is expected to begin operation towards the end of this decade.

The primary target is leptonic CP violation - a possible difference in how neutrinos and antineutrinos oscillate. If that difference exists and is large enough to measure, it would be a serious candidate explanation for the matter-antimatter asymmetry of the universe, and it is the kind of measurement that needs a far larger sample than T2K can accumulate.

Hyper-Kamiokande will also extend the proton decay search by roughly an order of magnitude, and would collect tens of thousands of neutrinos from a galactic supernova rather than the dozen Kamiokande recorded in 1987 - enough to watch the collapse of a stellar core unfold in real time.

Super-Kamiokande is expected to keep running alongside it for a period. Between the two, plus the different technologies used by IceCube and the other large detectors, the field is arranged so that no single instrument's systematic errors can quietly determine the answer. That redundancy is deliberate, and it is a large part of why the 1998 result has held up for more than a quarter of a century.