SAGE and GALLEX: The Gallium Gallery of Solar Neutrinos

The Sun’s total neutrino flux at Earth is dominated — by a factor of 60 — by the lowest-energy component, the pp neutrinos produced in the initial proton-proton fusion reaction. Yet no experiment before 1991 had ever directly detected one. The Homestake chlorine experiment of Ray Davis had a threshold of 0.814 MeV, roughly double the 0.420 MeV endpoint of the pp spectrum. Kamiokande and Super-Kamiokande, using Cherenkov detection on electron scattering, needed several MeV to get a clean signal. The entire Standard Solar Model prediction for the Sun’s total luminosity depended on neutrinos that no detector could see.

Two radiochemical experiments — SAGE (the Soviet-American Gallium Experiment) at Baksan in the North Caucasus, and GALLEX (the Gallium Experiment) in Hall A of the Gran Sasso Laboratory in Italy — changed that. Beginning in 1990 and 1991 respectively, they used the weak-neutral-current reaction ${\nu }_e+{}^{71}\text{Ga}\to {}^{71}\text{Ge}+e^{-}$ with a threshold of only 0.233 MeV, low enough to capture essentially the full solar neutrino spectrum including the pp component. After two decades of operation, their measurements bounded the total solar neutrino flux at Earth to within 5% precision — and, as a remarkable bonus, inherited a persistent calibration anomaly that has resisted resolution for over thirty years.

The radiochemical technique

A gallium experiment is, physically, a barrel of a gallium compound (typically gallium trichloride or metallic gallium) buried deep underground. Solar neutrinos interact with the gallium atoms to produce radioactive germanium-71. The germanium is extracted from the gallium by chemical procedures, purified, and counted for its decay activity. Each decay event corresponds to one neutrino interaction.

This sounds straightforward but is technically extraordinary. ${}^{71}$Ge has a half-life of 11.4 days. Over a typical month-long run, a 30-ton gallium target would produce roughly a dozen germanium atoms from solar neutrino capture, against a background of whatever other activities (cosmic ray spallation, natural radioactivity, chemical contamination) happen to leave similar signatures. Extracting and counting exactly those germanium atoms required chemical purification procedures unprecedented in radiochemistry.

The SAGE group at Baksan used metallic gallium as the target — 50 tons in the Soviet era, rising eventually to 60 tons. GALLEX used 30 tons of gallium as GaCl₃ in hydrochloric acid solution; when the experiment transitioned to GNO (Gallium Neutrino Observatory, 1998–2003), the target remained essentially the same.

Both experiments reported their results as event rates in “SNU” units — Solar Neutrino Units, defined as $10^{-36}$ captures per target atom per second. The Standard Solar Model prediction is approximately 128 SNU; the measured rates were about 70 SNU. The deficit — about half the predicted rate — was consistent with the deficits at Homestake (about a third of predicted) and Super-Kamiokande (about half), though at a different energy window.

Confirming the solar neutrino problem

By 2000, the solar neutrino problem was confirmed across the energy spectrum:

Experiment	Target	Threshold	Ratio (measured / SSM)
Homestake (chlorine)	³⁷Cl	0.814 MeV	~0.33
Kamiokande / Super-K	H₂O	5–10 MeV	~0.47
SAGE + GALLEX	⁷¹Ga	0.233 MeV	~0.55

These three experiments, with three different targets, three different thresholds, and three different technologies, all agreed: solar neutrinos arrive at Earth at a fraction of the predicted rate, with the deficit growing at lower energies. This energy-dependent pattern was the pre-SNO evidence that implicated neutrino oscillation as the most likely explanation.

SNO’s 2001 neutral-current measurement confirmed the oscillation interpretation by measuring the flavor-summed total flux — which matched SSM. But the gallium experiments held a unique position: they were the only ones that accessed the pp-component spectrum, the dominant flux. Their deficit was the deficit that mattered for the energy budget of the Sun.

In 2014, Borexino achieved a direct measurement of the pp spectrum itself, confirming the gallium-era inference that the pp flux was roughly 55% of the SSM prediction — exactly what the MSW-LMA oscillation parameters would predict for the crossover between vacuum-dominated and matter-dominated oscillation regimes.

Source calibrations and the anomaly

A fundamental question for any radiochemical experiment is: do you believe your extraction efficiency? Germanium-71 is produced as individual atoms embedded in tens of tons of gallium; extracting 90%+ of these atoms, reliably, across multiple chemical steps is non-trivial to verify from first principles. Both SAGE and GALLEX addressed this by performing source calibrations — running the experiment with an intense artificial neutrino source of known activity and measuring the response.

The sources were:

• ⁵¹Cr (half-life 28 days), producing neutrinos at 0.747 MeV (90%) and 0.427 MeV (10%)
• ³⁷Ar (half-life 35 days), producing neutrinos at 0.811 MeV (90%) and 0.750 MeV (10%)

These sources could be prepared with known activity through well-characterised electron-capture chains, and placed at the center of the target tank to produce a known neutrino flux with a known spectrum. The expected capture rate could then be compared against the measured rate.

Both SAGE and GALLEX found capture rates systematically below expectation — by roughly 80-90% of the expected rate, consistent across multiple source runs. This “gallium anomaly” has been reproduced in subsequent campaigns, most recently the BEST experiment at Baksan (2022), which used a ⁵¹Cr source to achieve a 4σ statistical detection of the deficit.

Three competing explanations

The anomaly has three plausible explanations, none yet definitive.

1. Sterile neutrinos. If an additional, weakly-coupled neutrino state exists with $\Delta m^2\sim 1$ eV², some of the source neutrinos would disappear into it before reaching the gallium target even over the short (~metre) baseline of a source experiment. This interpretation would connect the gallium anomaly to the LSND/MiniBooNE and reactor short-baseline anomalies — a unifying picture. Cosmological constraints, however, strongly disfavor the required sterile parameter space.

2. Cross-section systematic. The ⁷¹Ga → ⁷¹Ge cross-section at the relevant energies depends on nuclear matrix elements that are not perfectly known from first principles. Two excited states of ⁷¹Ge at 175 and 500 keV could be accessible in the capture reaction; their contribution to the total cross-section has been re-evaluated several times, with each evaluation giving slightly different values. A systematic error in this matrix element — at the 20% level — would explain the entire anomaly without requiring new physics.

3. Extraction efficiency underestimate. The chemical extraction procedure might be less efficient than believed. This seems unlikely given the multiple independent cross-checks, but has not been fully excluded.

As of 2026, the community consensus is leaning toward the cross-section explanation, but no consensus has been fully reached. A definitive test would require a dedicated experiment using a different target (e.g., indium or selenium) that would be immune to the ⁷¹Ga-specific systematic. Several such proposals exist but none are funded.

Legacy

SAGE, GALLEX, and GNO established:

1. Direct detection of the solar pp component — closing the solar neutrino flux accounting
2. Energy-dependent oscillation signature — the gallium rate was closer to SSM than the chlorine rate, hinting at a shape-changing effect across the spectrum that SNO and Super-K later confirmed
3. Radiochemical calibration methodology — the source-based calibration approach is now standard for any similar experiment
4. The gallium anomaly — the longest-standing unexplained anomaly in neutrino physics, alive for 30+ years and recently reconfirmed

The experiments are largely historical artifacts now. GNO ended in 2003. GALLEX’s tanks at Gran Sasso were removed in 2010. SAGE remains at reduced capacity at Baksan, used primarily for sterile-neutrino calibration campaigns rather than solar monitoring.

But their contribution to the field outlasts their operational era. The existence of the gallium anomaly continues to drive short-baseline sterile-neutrino searches, and their pp-flux measurements form part of the precision baseline for every future solar-neutrino experiment. The programme was, in the pattern of many foundational experiments, modest in scale, labor-intensive, and enormously productive.