On this page
Objective
Search for proton decay in 8 kilotons of water instrumented with a Cherenkov photomultiplier array.
Method
Rectangular water tank 22 × 17 × 18 m holding 8,000 tons of purified water, 600 m underground in a salt mine. 2,048 5-inch photomultipliers providing 1% photo-coverage. Operated alongside Kamiokande as a complementary large-volume water Cherenkov detector in the 1980s.
Key results
- No proton decay observed, setting limits competitive with Kamiokande on simple decay modes.
- 1987: detection of 8 neutrinos from Supernova 1987A — the largest single-detector supernova neutrino sample of the era.
- Atmospheric neutrino measurements contributing to the late-1980s atmospheric anomaly that Super-Kamiokande would later resolve.
- Reines's group at UC Irvine was a founding partner, linking the 1980s water-Cherenkov programs back to the 1956 discovery.
Significance
Together with Kamiokande-II, IMB established the water-Cherenkov technique as a viable method for detecting MeV-range astrophysical neutrinos. The joint 1987A detection opened the field of extragalactic neutrino astronomy and validated the theoretical core-collapse neutrino mechanism.
Design and operation
The Irvine-Michigan-Brookhaven detector (IMB) was commissioned in 1982 in a deep salt mine near Cleveland, Ohio. Its 8-kiloton water volume was the largest water Cherenkov detector at the time and remained the benchmark until Super-Kamiokande came online in 1996. Compared to Kamiokande’s 3 kt volume, IMB had advantages in total mass but a much lower photo-coverage (1% vs. 20%), limiting its effective energy threshold.
The collaboration brought together groups from UC Irvine (including Frederick Reines, co-discoverer of the neutrino), the University of Michigan, and Brookhaven National Laboratory. The deep-underground setting and rectangular tank geometry traded construction simplicity for the imaging performance that Kamiokande’s cylindrical / densely-instrumented design offered.
Proton decay search
Like Kamiokande, IMB was designed to search for proton decay. Its larger target mass allowed competitive limits on the simplest SU(5)-predicted channels, particularly . Both detectors progressively excluded simple grand-unified predictions.
Supernova 1987A
On 23 February 1987, IMB recorded 8 electron-antineutrino events in a ~6-second window coincident with the Kamiokande-II burst and the Baksan signal. The IMB events are the highest-energy of the galactic burst (typical 20–40 MeV) — IMB’s threshold was higher than Kamiokande’s because of its lower photo-coverage. The energy and timing distribution matched the expected thermal core-collapse spectrum.
The combined IMB + Kamiokande + Baksan dataset of 24 events remains the only neutrino burst ever detected from a core-collapse supernova. The analysis established upper bounds on neutrino mass (few tens of eV), on neutrino magnetic moment, and on several exotic-physics proposals, and validated the standard picture of stellar core collapse.
The atmospheric anomaly
IMB and Kamiokande both contributed to the late-1980s “atmospheric anomaly” — the observed ratio of muon-like to electron-like atmospheric neutrino events was below theoretical expectations. The smaller detectors lacked the statistical power to resolve whether this was oscillation, systematic, or background. Super-Kamiokande’s much larger mass, beginning in 1996, settled the question in 1998.
Decommissioning and legacy
IMB was decommissioned in 1991. The Morton salt mine continued commercial operation; the experimental hall was not reused for subsequent physics. The IMB collaboration’s experience fed directly into Super-Kamiokande design discussions and into the Sudbury Neutrino Observatory in Canada, where several IMB collaborators relocated.
Frederick Reines, a founding IMB collaborator, received the 1995 Nobel Prize in Physics shortly before his death. The IMB supernova detection was part of the evidence cited in the 2002 Nobel Prize to Masatoshi Koshiba — recognizing both Kamiokande’s and, by association, the broader water-Cherenkov program.