Super-Kamiokande and the 1998 Discovery of Oscillations

On the afternoon of 5 June 1998 at the Neutrino ‘98 conference in Takayama, Japan, a 39-year-old experimentalist named Takaaki Kajita presented 30 minutes of plots to a room of physicists. By the end of the talk the audience had been shown evidence that neutrinos have mass — the first direct demonstration of a property that the Standard Model, as written for thirty years, had assumed was zero.

This is the story of the Super-Kamiokande atmospheric oscillation result, the measurement that anchors every subsequent discussion of neutrino mass, oscillation, and CP violation.

The detector

Super-Kamiokande sits 1,000 metres underground in the Mozumi mine in Gifu Prefecture, central Japan. It is a cylindrical tank 40 metres in diameter and 40 metres tall, filled with 50,000 tonnes of ultra-pure water. Facing inward from the tank walls are 11,146 photomultiplier tubes, each 50 cm in diameter, giving about 40% photocathode coverage.

A charged particle traversing the water above Cherenkov threshold produces a cone of blue light. The cone intersects the PMT array as a ring; the shape and time profile of the ring reconstruct the particle’s direction, energy, and flavor. Electrons produce fuzzy rings — they trigger electromagnetic showers that smear the cone pattern — while muons produce sharp, crisp rings because they travel coherently as single particles.

Super-Kamiokande began taking data in April 1996, succeeding the earlier 3-kiloton Kamiokande detector in the same mine.

Atmospheric neutrinos

Cosmic rays — mostly high-energy protons from outside the solar system — strike the upper atmosphere and produce pions through nuclear interactions. Each pion decays through a two-step chain: ${\pi }^{+}\to {\mu }^{+}+{\nu }_{\mu },{\mu }^{+}\to e^{+}+{\nu }_e+{\bar{\nu }}_{\mu }$ yielding two muon-flavor neutrinos and one electron-flavor neutrino. The expected ratio of ${\nu }_{\mu }+{\bar{\nu }}_{\mu }$ to ${\nu }_e+{\bar{\nu }}_e$ events at detection is therefore about 2 to 1, with small corrections for the energy-dependent branching and for the fact that at high energies muons reach the ground before decaying.

This ratio, and the absolute flux, are well calculated. They depend on well-measured cosmic-ray composition and on well-measured pion production cross-sections. The theoretical uncertainty is about 10%.

Kamiokande, Super-K’s predecessor, had reported in the late 1980s that the measured $\mu$-like to $e$-like ratio was below expectation — a deficit of muon-flavor neutrinos, or an excess of electron-flavor events, or some combination. The “atmospheric anomaly” was real but limited in statistical power. Super-K, with fifty times the target mass, could settle the question.

The zenith-angle key

The decisive observable was the zenith angle of the incoming neutrino.

Neutrinos reaching a detector in Japan come from a wide range of distances. Downward-going neutrinos — those produced in the atmosphere directly above the detector — have travelled only about 15 km before reaching the PMTs. Upward-going neutrinos — produced in the atmosphere on the other side of the Earth and penetrating the entire planet — have travelled about 12,800 km. The ratio of upward to downward fluxes should be essentially 1:1, modulo small geomagnetic and path-length corrections.

If neutrinos oscillate, the oscillation probability depends on the ratio $L/E$: longer baselines give more oscillation. A zenith-angle-dependent deficit, growing with baseline, would be the signature of oscillation.

Super-K analyzed 535 days of data from April 1996 to April 1998. The up-down ratio for electron-like events was consistent with unity — no zenith-dependent effect. The up-down ratio for muon-like events was significantly below one for upward-going, with the deficit growing monotonically with zenith angle.

The deficit as a function of $L/E$ matched a two-flavor oscillation fit with near-maximal mixing and $\Delta m^2\approx 2\times 10^{-3}$ eV². The interpretation as ${\nu }_{\mu }\to {\nu }_{\tau }$ oscillation was consistent with the electron non-deficit (if ${\nu }_e$ had been produced it would have contributed to the $e$-like sample, which showed no excess).

The Takayama announcement

Kajita presented the result at Neutrino ‘98 in Takayama, a small mountain town hosting one of the standing international neutrino-physics conference series. The talk was detailed — zenith distributions, energy dependence, fit contours, systematic checks — but the conclusion was simple. Atmospheric muon neutrinos were disappearing in a way that required oscillation, and oscillation required at least one non-zero mass difference.

The reception was striking. Within hours, the Super-K spokespersons were briefing journalists. The New York Times carried the story on the front page two days later. By the time the formal publication appeared in Physical Review Letters in August 1998, the physics community had already accepted the conclusion.

What it meant

The Standard Model, as originally written, contained massless neutrinos. The atmospheric oscillation discovery made that choice untenable. Oscillation requires at least two mass eigenstates to be non-zero and non-degenerate. Exactly how non-zero, and in what ordering, was not yet known — but the minimal Standard Model was now incomplete.

The next three years clarified the picture:

• 1999–2001: SNO’s heavy-water solar-neutrino measurement confirmed a second, independent oscillation channel
• 2003: KamLAND’s reactor result confirmed the solar-sector oscillation parameters with a terrestrial source
• 2012: Daya Bay measured the last mixing angle ${\theta }_{13}$, enabling CP-violation searches

Super-K itself has continued to operate through the 2020s, steadily improving its atmospheric measurements and contributing to long-baseline accelerator experiments (K2K, T2K) as the far detector.

The Nobel

Takaaki Kajita shared the 2015 Nobel Prize in Physics with Arthur McDonald, the director of SNO. The citation — “for the discovery of neutrino oscillations, which shows that neutrinos have mass” — honoured both complementary discoveries: atmospheric at Super-K and solar at SNO.

Kajita was trained under Masatoshi Koshiba at Kamiokande, the 3-kiloton predecessor whose 1987A detection and subsequent real-time solar neutrino measurements laid the intellectual foundation for Super-K. Koshiba himself received the 2002 Nobel Prize — the same field, the same site, a generation earlier.

The legacy

Super-Kamiokande is still operating, now in its sixth data-taking configuration (SK-VII), with gadolinium dissolved in the water to enable neutron-tagging for diffuse supernova-neutrino sensitivity. Its successor, Hyper-Kamiokande, is under construction nearby with a 260-kiloton design mass — eight times Super-K’s fiducial volume. First data is expected in 2027.

The 1998 Takayama result, twenty-six years on, is still the single most-cited neutrino-physics paper of the modern era. Every subsequent discussion of neutrino mass, of the mass ordering, of CP violation, of the seesaw mechanism, of leptogenesis — all begin with the experimental fact that Super-K established.