Supernova 1987A: The First Extragalactic Neutrino Burst

At 07:35:41 UT on 23 February 1987, a burst of neutrinos from a supernova 168,000 light-years away swept through the Earth. Three underground detectors on three continents — Kamiokande-II in Japan, IMB in Ohio, and Baksan in the Soviet Caucasus — recorded a combined total of 24 events within a 13-second window. Three hours later, optical astronomers in Chile identified the collapsing blue supergiant Sanduleak −69 202a in the Large Magellanic Cloud as the source.

The event, designated SN 1987A, became the first — and so far only — supernova with confirmed neutrino detection. It transformed a theoretical picture of core collapse, built from decades of simulation and a handful of observational constraints, into an observed phenomenon. It remains the single most important event in the history of neutrino astronomy.

The progenitor

Sanduleak −69 202a was a B3 I blue supergiant in the Tarantula Nebula of the Large Magellanic Cloud, a satellite galaxy of the Milky Way. It was about 20 solar masses, 40,000 K at the surface, and roughly $6\times 10^4$ times as luminous as the Sun. Pre-supernova photographs — a rare prize in supernova astronomy — show the star clearly in archival plates. Its identification as the progenitor was made within two days of the optical discovery.

The star had reached the end of its nuclear fuel sequence. After silicon burning finished, the iron core — no longer able to extract energy by fusion — exceeded the Chandrasekhar limit and began to collapse. Electron capture on nuclei reduced the pressure; neutron degeneracy eventually halted the collapse at roughly 10% of the original iron-core radius. The rebound launched a shock that, assisted by neutrino heating, eventually ejected the stellar envelope.

What the detectors saw

Three underground neutrino detectors were live that night:

Kamiokande-II in the Kamioka mine, Japan — a 2.14-kiloton water Cherenkov detector optimized for proton-decay searches. It recorded 11 events clustered within a 12.4-second window at 07:35:35 UT.

IMB (Irvine-Michigan-Brookhaven) — a 6.8-kiloton water Cherenkov detector in the Morton Salt Mine in Ohio. It recorded 8 events within 5.6 seconds, beginning at the same absolute time as the Kamiokande burst to within measurement uncertainty.

Baksan (Soviet Baksan Neutrino Observatory) — a 200-ton liquid scintillator telescope in a tunnel under Mount Andyrchy. It recorded 5 events within 9.1 seconds, with the first event slightly preceding the water Cherenkov detectors.

A fourth detector — the 90-ton liquid scintillator at the LSD site in Mont Blanc — reported a burst of five events five hours earlier than the three confirmed detectors. The LSD events remain unexplained and have generally been interpreted as a statistical fluctuation or background, though a small theoretical community still speculates about a two-stage collapse.

The 24 confirmed events were almost all inverse-beta-decay reactions of electron antineutrinos on free protons in water or hydrocarbon: ${\bar{\nu }}_e+p\to e^{+}+n$ The positrons produced Cherenkov or scintillation light that the detectors registered. Electron-flavor neutrinos, and ${\nu }_{\mu }/{\nu }_{\tau }$ of either helicity, would also have interacted via elastic scattering and neutral currents, but with much smaller cross-sections — their contribution to the 24 events is estimated at one or two.

The 13-second window

The burst lasted about 13 seconds — roughly the timescale expected from core-collapse simulations for the neutronization, accretion, and cooling phases of a newborn proto-neutron star. The time structure of the events showed a tendency for higher-energy events to occur earlier, consistent with spectral hardening in the accretion phase followed by spectral softening during cooling.

Total energy carried by the neutrino burst, integrated across all six flavors (three neutrinos and three antineutrinos), was estimated from the 24 detected events at $(3\pm 1)\times 10^{53}$ erg — about 99% of the gravitational binding energy released by the collapsing core, and roughly two orders of magnitude larger than the kinetic energy of the ejected envelope. The prediction had been that neutrinos would carry away essentially all the gravitational energy of the collapse. SN 1987A confirmed it.

Constraints on neutrino properties

Because the neutrinos arrived in a short burst from a known distance, SN 1987A also provided a time-of-flight test of neutrino propagation. Two constraints followed immediately.

Mass. If neutrinos have mass $m_{\nu }$, a neutrino of energy $E$ arrives with a time delay relative to light of $\Delta t\approx \frac{D}{2c}{\left(\frac{m_{\nu }{c}^2}{E}\right)}^2$ With $D=168,000$ light-years and $E\sim 20$ MeV, the 13-second burst width places an upper limit on the electron-antineutrino mass of roughly 5.7 eV. Laboratory experiments — first the early Mainz and Troitsk tritium-endpoint measurements, now KATRIN — have since improved this limit by more than an order of magnitude, but the SN 1987A constraint remains a celebrated early bound.

Lifetime. The observed flux and energy spectrum constrain the lifetime of a decaying electron antineutrino to be at least $10^{15}$ seconds at $E=10$ MeV — a strong argument that the neutrino is, within experimental limits, a stable particle on cosmological timescales.

A third constraint, on the neutrino’s electric charge, was derived from the arrival-time dispersion and excluded $|q_{\nu }/e|$ above about $10^{-17}$.

The shadow burst

The neutrino arrival preceded the optical discovery by about three hours. This delay was entirely expected: the shock wave launched by the core bounce takes hours to traverse the stellar envelope before emerging as a luminous supernova. Neutrinos, on the other hand, escape from the proto-neutron star surface on a timescale of seconds, once the density drops below the neutrinosphere threshold. The three-hour offset is a direct measurement of the progenitor’s envelope opacity and mass.

This observation is the reason the field talks about SNEWS — the Supernova Early Warning System. Modern neutrino detectors worldwide (Super-Kamiokande, IceCube, LVD, Borexino until 2021, KamLAND, and soon DUNE, JUNO, and Hyper-Kamiokande) share a real-time burst-alert coordination system. A Galactic core-collapse supernova today would deliver tens of thousands of events across those detectors, with alert latency under a minute — hours before the optical flash.

What remains unresolved

One anomaly has stood for nearly forty years: the compact remnant. Core-collapse simulations predict that the iron core leaves behind either a neutron star or, for sufficiently massive progenitors, a black hole. No compact remnant has yet been conclusively detected in the center of the SN 1987A debris field. A candidate point source (HST-1E) was reported in JWST imaging in 2024 and remains under analysis. If it is a neutron star, it would be the first direct post-collapse observation of a stellar remnant whose birth was witnessed in neutrinos.

A second open question concerns the neutronization burst: the brief, intense ${\nu }_e$ pulse from electron capture in the first ten milliseconds of collapse. SN 1987A’s detectors were sensitive almost exclusively to ${\bar{\nu }}_e$ and could not separately identify the neutronization burst. A future galactic supernova, observed by DUNE’s liquid-argon detector, would.

Legacy

SN 1987A is the founding event of multi-messenger astronomy. It established that information carried by neutrinos can precede, complement, and in some circumstances supplant information carried by light. Its 24 events are still more cited, per event, than any other astrophysical detection in history.

The next galactic supernova will deliver three orders of magnitude more events. It may happen tomorrow; it may happen in a century. When it does, the community is ready.