The Solar Neutrino Problem

Summary

From 1968 a detector in a South Dakota gold mine caught only a third of the neutrinos the Sun was predicted to emit, and for decades nobody knew why. In 2001 and 2002 the Sudbury Neutrino Observatory showed that the missing neutrinos had changed type.

The Sun was producing what the solar model said, and the experiment was measuring correctly. The error was an assumption of particle physics: that a neutrino cannot change from one type into another. It can, which means neutrinos have mass. Two Nobel prizes followed.

What is documented

Why neutrinos. The Sun shines by nuclear fusion in its core, and the reactions release neutrinos, particles that pass almost untouched through the Sun and the Earth. A detector that catches a few of them sees the core directly. In 1964 the astrophysicist John Bahcall and the chemist Raymond Davis Jr. proposed, in back-to-back papers in Physical Review Letters, to count them with chlorine: a chlorine-37 atom that captures an electron neutrino becomes an atom of radioactive argon-37. Bruno Pontecorvo had suggested the chlorine method in 1946.

Homestake. Davis's detector became operational in 1967, 4,850 feet (1,478 metres) underground in the Homestake gold mine in South Dakota. It was a tank of 100,000 gallons (378,000 litres, about 615 tonnes) of perchloroethylene, the dry-cleaning fluid. It could register neutrinos above 0.814 MeV. Over the whole experiment about 2,200 argon atoms were counted.

The first result, 1968. Davis, Don Harmer and Kenneth Hoffman reported an upper limit of 3 solar neutrino units (one unit is 10⁻³⁶ captures per target atom per second). In the same issue of the journal, Bahcall, Neta Bahcall and Giora Shaviv gave the solar model's prediction: 7.5 ± 3 units.

The final result. From 108 runs beginning in 1970, the Homestake team reported an average of 2.56 ± 0.16 (statistical) ± 0.16 (systematic) units, published in 1998. The standard solar model of 2001 predicted 7.6, with an uncertainty of about 1.1 to 1.3. The measurement was about one third of the prediction, 34 per cent.

Other methods, same shortfall. Kamiokande-II, a water detector in the Japanese Alps, recorded the flashes from electrons struck by neutrinos and could tell their direction, which showed for the first time that the neutrinos came from the Sun. From 450 days of data between January 1987 and May 1988 it measured 0.46 ± 0.13 ± 0.08 of the predicted flux of high-energy neutrinos (46 per cent). The gallium experiments, which could register the low-energy neutrinos from the Sun's main fusion reaction, followed: SAGE, with 30 tonnes of gallium at the Baksan Neutrino Observatory in Russia (first result 1991), and GALLEX, with 30.3 tonnes at Gran Sasso in Italy (first result 1992). By 2009 the three gallium experiments, SAGE, GALLEX and its successor GNO, had together measured 66.1 ± 3.1 units, about half the 132 units the solar model predicted in 1991.

Three possibilities. Either the experiments were wrong, or the solar model was wrong, or something happened to the neutrinos after they were made. The third was set out for solar neutrinos by Vladimir Gribov and Bruno Pontecorvo in 1969: neutrinos of one type might turn into another. Lincoln Wolfenstein (1978) and Stanislav Mikheyev and Alexei Smirnov (1985) showed that dense matter, such as the inside of the Sun, alters such changes and can greatly amplify them.

Super-Kamiokande, 1998. A detector holding 50,000 tonnes of water, 1,000 metres underground in a zinc mine in Japan, began work in 1996. In 1998 Takaaki Kajita presented evidence that neutrinos made in the atmosphere switch identity on their way to it.

SNO, 2001 and 2002. The Sudbury Neutrino Observatory held 1,000 tonnes of heavy water, 2 kilometres underground in a nickel mine near Sudbury, Ontario, and began work in 1999. Heavy water let it count high-energy neutrinos from the Sun in two ways: with a reaction that only electron neutrinos can cause, and with one that all three types cause equally. In 2001 its electron-neutrino count, set against Super-Kamiokande's solar measurement, gave evidence at 3.3 standard deviations that some of the Sun's neutrinos were arriving as other types. In 2002 the all-types count gave a total of 5.09 million per square centimetre per second, consistent with the solar models. The electron-neutrino flux was 1.76 million, about a third of that. The flux of muon and tau neutrinos came out at 3.41 million, 5.3 standard deviations above zero.

Confirmation. In December 2002 the KamLAND experiment in Japan reported that antineutrinos from nuclear reactors were also disappearing, at 61 per cent of the expected rate, and that of all the proposed solutions to the solar neutrino problem only one, known as the large mixing angle solution, survived.

The prizes. The 2002 Nobel Prize in Physics went half to Davis and Masatoshi Koshiba of Kamiokande "for pioneering contributions to astrophysics, in particular for the detection of cosmic neutrinos", and half to Riccardo Giacconi. The 2015 prize went to Kajita and Arthur McDonald of SNO "for the discovery of neutrino oscillations, which shows that neutrinos have mass".

Since. In 2018 the Borexino detector at Gran Sasso measured the survival of electron neutrinos at different energies. The JUNO detector in China, 20,000 tonnes of liquid scintillator 52.5 kilometres from nuclear reactors, began taking data in August 2025. From its first 59.1 days it measured the mixing angle and mass difference that govern the conversion of solar neutrinos, sin²θ₁₂ = 0.3092 ± 0.0087 and Δm²₂₁ = (7.50 ± 0.12) × 10⁻⁵ eV², 1.6 times more precisely than all earlier measurements combined. The result was published in Nature in 2026.

Leading explanations

Neutrino flavour change, driven by matter inside the Sun. Established. Neutrinos come in three types: electron, muon and tau. Because they have mass, a neutrino made as one type travels as a mixture and can be detected as another. For the higher-energy neutrinos that the chlorine, water and heavy-water detectors mainly counted, most of the change happens inside the Sun, where dense matter drives them into a state that is about one-third electron type, a fraction set by the mixing angle (sin²θ₁₂, about 0.31). Low-energy neutrinos change mainly by oscillation in empty space, and more than half of them still arrive as electron neutrinos. That is why the gallium experiments, which counted low-energy neutrinos, saw a smaller shortfall than the chlorine tank. SNO counted the converted neutrinos directly, KamLAND saw the same mixing in reactor antineutrinos, and Borexino measured the change across the range of energies.

The solar model was wrong. Rejected. On this reading the model overestimated the Sun's output of neutrinos. SNO's all-types count matched the model's total. The Sun was making the neutrinos it was supposed to make.

The Homestake experiment was wrong. Rejected. The chlorine measurement was the first of its kind and rested on counting a few argon atoms extracted from a vast tank. Every later detector, using water, gallium or heavy water, found a shortfall, and SNO's count of electron neutrinos came out at about a third of the total, as Davis's had.

Current status

Later explained. The shortfall is accounted for by neutrino flavour change, and the total flux of solar neutrinos agrees with the solar model. The solar neutrino problem became a tool: the same data now measure neutrino properties and the Sun's interior.

What remains open is about neutrinos, not the Sun. Oscillations measure differences between neutrino masses, not the masses themselves. The KATRIN experiment's direct measurement, published in 2025, puts the mass below 0.45 electronvolts (90 per cent confidence). JUNO's main aim is to determine the order of the three masses. One residue from the gallium technique is also unresolved: gallium detectors exposed to artificial neutrino sources have recorded fewer captures than expected, the so-called gallium anomaly, and the BEST experiment reported shortfalls of 20 to 24 per cent in 2022. A 2025 KATRIN analysis excluded much of the range of extra, "sterile" neutrinos proposed to explain it. None of this reopens the solar neutrino problem.

Last reviewed: September 2026. Records are provisional. Where the evidence changes, the entry changes. Found an error? Tell us.