and in fluxes, a set of detectors with sufficient complementarity and some redundancy is required to untangle the different aspects of neutrino physics and of solar physics.
In this remarkable experiment , neutrinos are detected using a large tank containing 615 tons of perchloroethylene, C2Cl4, which is a commercially available cleaning fluid. The target isotope, 37Cl, can capture a neutrino, producing a radioactive isotope 37Ar by the following process:
.ve + 37Cl → 37Ar + e−. (1)
This reaction can occur for electron-type neutrinos with energy greater than 0.8 MeV. The neutrino reactions are registered by counting the radioactive nuclei, 37Ar. In order to avoid extraneous (background) interactions caused by cosmic rays incident on the surface of the earth, the experiment is conducted 1500 meters underground in the Homestake gold mine in Lead, South Dakota. After about two months in the solar neutrino “sunshine,” the buildup of 37Ar atoms from neutrino capture by 37Cl is approximately balanced by the loss due to decay. At this point, the standard solar model  predicts that only 54 37Ar atoms are present in the 615 tons of C2Cl4. It is an experimental challenge to extract and count this minuscule number of atoms in such a large target mass. Over the past two decades, more than 100 extractions of 37Ar have been performed. After correction for detection efficiency the number of 37Ar atoms in the detector at extraction is observed to be only 17, corresponding to a solar-neutrino-induced production rate of 0.48 atoms per day, far fewer than the 1.5 atoms per day expected on the basis of the standard model. In terms of the solar neutrino unit, SNU (pronounced “snew,” 1 SNU = 10−36 interactions per target atom per second), the observations yield 2.55 ± 0.25 SNU, about one third of the 8 ± 1 SNU predicted by the standard model. For two decades, this discrepancy between the measured chlorine event rate and the event rate predicted by the standard model constituted the well-known solar neutrino problem.
Like a supersonic aircraft creating a sonic boom, a charged particle moving through matter faster than the speed of light gives off a conical shock wave of light called Cerenkov radiation. Sensitive photomultiplier tubes can record the flashes of light that betray the presence and direction of these high-speed particles. With the aid of special large photomultiplier tubes, experimenters working in a mine in the Japanese alps and using a 3000-ton detector of ultrapure water (680-ton fiducial volume), called Kamiokande, were able to detect electrons that had been struck by neutrinos from the sun [4,5]. The reaction that was observed is
ν + e− → ν + e−. (2)
At the energies of interest for solar neutrino detection, reaction (2) is about six times more likely for νe than for νμ or ντ. Despite a significant background from the radioactivity of the rocks in the mine, from radioactivity in the water, and from cosmic-ray-induced events, the neutrinos were unambiguously detected because the electrons struck by neutrinos recoiled preferentially in the direction from the sun to the earth. Kamiokande, the first “active” (electronic rather than radiochemical) solar neutrino experiment, showed two very important things: first, the detected neutrinos do indeed come from the sun, and, second, the observed flux above an electron-detection threshold of about 8 MeV is about half as large as predicted by the standard solar model. The discrepancy between prediction and observation is less severe at energies above 8 MeV (the Kamiokande experiment) than it is for all energies above 0.8 MeV (the chlorine experiment).
The predicted counting rate for any particular solar neutrino experiment depends on calculations that are usually based on a standard solar model. Solar models are constrained by measurements of the solar luminosity, age, chemical composition, and thousands of seismological frequencies. Nevertheless, the predicted neutrino emission rates cannot be tested independent of solar neutrino experiments. Changes in the input data for the standard solar model could in principle be made just so as to fit any one of the existing experiments. However, it appears essentially