Supernova neutrinos
A collapsing star releases about 3 × 10⁴⁶ joules in a few seconds, almost all of it as neutrinos rather than light. Two dozen of these neutrinos, caught in 1987, opened neutrino astronomy beyond the Sun, and detectors around the world now wait for the next burst.
Beginner
The death of a giant star
Stars shine by fusing light elements into heavier ones in their cores. A star at least about eight times heavier than the Sun keeps doing this until its core is made of iron. Iron cannot release more energy by fusion, so the core stops producing heat. When the iron core grows to about 1.4 times the mass of the Sun, it can no longer support its own weight and collapses in a fraction of a second. Matter equal to the Sun's mass is squeezed into a ball only a few tens of kilometres across, a newborn neutron star.
Where the energy goes
The collapse releases an enormous amount of energy, more than the Sun will emit in its entire life. Surprisingly, only about one percent of it goes into the explosion that blows the star apart, and only a tiny fraction becomes the light that astronomers see. About 99 percent leaves the star as neutrinos, around 10⁵⁸ of them in roughly ten seconds8. In the extremely dense core even neutrinos cannot escape at once. They bounce around like light in thick fog, slowly working their way out from a glowing surface called the neutrinosphere.
Neutrinos may also help the explosion itself. The collapse launches a shock wave outward, but the shock quickly stalls. Many scientists think that a small share of the outgoing neutrinos heats the material behind the shock enough to restart it, like wind filling a sail.
Supernova 1987A
On 23 February 1987 a star exploded in the Large Magellanic Cloud, a small galaxy next to our own, about 168,000 light-years away. A few hours before anyone saw the new light in the sky, three underground detectors had recorded a short burst of neutrinos within about 13 seconds: 11 in Kamiokande-II in Japan1, 8 in IMB in the United States2 and 5 at Baksan in Russia3. Only two dozen particles were caught, but they confirmed the basic picture of how a supernova works. The neutrinos arrived before the light because they leave the collapsing core at once, while light has to wait until the shock wave reaches the star's surface. Masatoshi Koshiba, who led Kamiokande, shared the 2002 Nobel Prize in Physics partly for this observation.
Waiting for the next one
In our own galaxy, a star collapses roughly one and a half to two times per century6, but the last one seen was long before neutrino detectors existed. Many are hidden behind dust, yet their neutrinos would pass straight through. Today detectors around the world are linked in the SuperNova Early Warning System5. If several of them see a burst at the same time, an automatic alert tells astronomers to point their telescopes, because the light may follow minutes to hours later. A supernova in the middle of our galaxy would produce thousands of signals in Super-Kamiokande and many more in the new detectors being built.
A faint glow from all past supernovae
Every collapsed star in the history of the universe has left its neutrinos behind. Together they form a faint, steady background, the diffuse supernova neutrino background. It has not yet been detected clearly. To catch it, Super-Kamiokande added the element gadolinium to its water in 2020, which makes the rare signals easier to recognise7.
Quiz
Test your knowledge
4 questions on this level · Print as worksheet
Intermediate
Core collapse
In a star of more than about eight solar masses, successive burning stages end with an iron core supported by electron degeneracy pressure. Electron captures on nuclei and photodisintegration of iron remove that support, and once the core exceeds roughly the Chandrasekhar mass of about 1.4 solar masses it collapses within a few hundred milliseconds. At nuclear density the inner core stiffens and bounces, launching a shock wave. The shock loses energy dissociating infalling iron and stalls at a radius of about 100–200 km.
The gravitational binding energy of the resulting neutron star, about 3 × 10⁵³ erg (3 × 10⁴⁶ J), must be released. Roughly 99 % leaves as neutrinos and antineutrinos of all flavours, about 1 % becomes kinetic energy of the ejecta and about 0.01 % visible light8. In 1966 Stirling Colgate and Richard White first proposed that neutrinos drive the explosion; in 1985 Hans Bethe and James Wilson showed that neutrino heating behind the stalled shock can revive it after a delay of a few hundred milliseconds4. Modern three-dimensional simulations support this delayed neutrino-heating mechanism, aided by convection and shock instabilities.
The neutrino signal
At densities above about 10¹¹–10¹² g/cm³ neutrinos are trapped; their mean free path is shorter than the core. They diffuse out over seconds and decouple at the neutrinosphere, with mean energies of about 10–15 MeV. The signal has three phases:
- Neutronization burst: a flash of electron neutrinos lasting about 10 ms when the shock breaks through the neutrinosphere and electron capture on free protons accelerates.
- Accretion phase: a few hundred milliseconds of bright emission in all flavours while matter falls onto the proto-neutron star.
- Cooling phase: about ten seconds of declining emission as the hot proto-neutron star contracts and radiates its energy.
SN 1987A
At 07:35 UT on 23 February 1987, about two to three hours before the first light of supernova SN 1987A reached Earth, a burst of antineutrinos arrived from the Large Magellanic Cloud, about 50 kpc away. The water Cherenkov detectors Kamiokande-II and IMB recorded 111 and 82 events, mostly by inverse beta decay, ν̄e + p → n + e⁺; the Baksan scintillation telescope recorded 53. The events, spread over about 13 seconds, gave a total energy and a neutrino temperature consistent with the theory of a cooling neutron star. The arrival-time spread also placed upper limits on the neutrino mass of order 10 eV and constrained many hypothetical particles that would have drained energy from the core.
SNEWS
Because a galactic supernova may be obscured by dust or happen while telescopes look elsewhere, the SuperNova Early Warning System (SNEWS) has operated automatically since 2005. It issues an alert when several independent detectors, among them Super-Kamiokande, IceCube, KamLAND and LVD, see a burst within a short time window, keeping false alarms rare5. SNEWS 2.0 adds triangulation using arrival-time differences and pre-supernova alerts from neutrinos emitted in the last hours of silicon burning, detectable from stars within about 1 kpc, such as Betelgeuse.
The diffuse background
All core collapses since the first stars have filled the universe with the diffuse supernova neutrino background (DSNB). Its ν̄e component should be detectable above about 10 MeV, where reactor antineutrinos fade. To suppress backgrounds, Super-Kamiokande dissolved gadolinium in its water, 0.01 % by mass from 2020 and 0.03 % from 2022. Gadolinium captures the neutron from inverse beta decay and emits a gamma cascade of about 8 MeV, giving a delayed coincidence tag. An analysis of 956 days of gadolinium data showed only a mild excess of about 1.2σ over background7; a discovery is expected after more years of data and with JUNO.
A galactic supernova
A collapse at 10 kpc, typical for the Milky Way, would yield about 4,000–8,000 inverse beta decay events in Super-Kamiokande, tens of thousands in Hyper-Kamiokande, a rise of the photon count across all IceCube sensors, and an electron-neutrino signal in DUNE sensitive to the neutronization burst5. The galactic rate is about 1.6 per century6.
Quiz
Test your knowledge
4 questions on this level · Print as worksheet
Expert
Energetics and emission
The energy budget is set by the binding energy of a cold neutron star, Eb ≈ (3/5) G M² / R ≈ 3 × 10⁵³ erg for M ≈ 1.4 M☉ and R ≈ 12 km. Neutrinos carry ≳ 99 % of it; the explosion energy of ≈ 10⁵¹ erg is a percent-level by-product8. In the diffusion regime the luminosity is roughly equipartitioned among νe, ν̄e and the four heavy-lepton species νx, with hierarchical mean energies ⟨Eνe⟩ < ⟨Eν̄e⟩ ≲ ⟨Eνx⟩, typically 10–16 MeV, because νx interact only via neutral currents and decouple deeper. Spectra are well described by quasi-thermal distributions with a pinching parameter α ≈ 2–4.
Explosion mechanism
In the delayed neutrino-driven mechanism4, a gain region forms behind the stalled shock where net heating by νe n → e⁻ p and ν̄e p → e⁺ n exceeds cooling. Explosion requires the heating timescale to become shorter than the advection time through the gain layer, a condition strongly aided by multidimensional effects: neutrino-driven convection and the standing accretion shock instability (SASI). Self-consistent three-dimensional models with energy-dependent Boltzmann or moment transport now produce explosions for many progenitors, though explosion energies and the boundary between successful explosions and black-hole formation remain under study8. The neutrino signal carries direct imprints: SASI induces quasi-periodic luminosity modulations with frequencies of order 100 Hz, and black-hole formation would cut the signal off abruptly.
SN 1987A in numbers
Kamiokande-II (2.14 kt fiducial) recorded 11 events above threshold in 13 s at 07:35:35 UT1; a twelfth event appeared in later analyses. IMB (≈ 6.8 kt, higher threshold) recorded 8 events in 6 s2; the Baksan scintillation telescope recorded 5 events in 9 s3, with an absolute clock uncertainty of tens of seconds. Joint likelihood analyses infer Eb ≈ (2–5) × 10⁵³ erg and ⟨Eν̄e⟩ ≈ 10–15 MeV, in agreement with standard proto-neutron-star cooling. The burst duration also excluded efficient cooling by additional light particles; the resulting stellar-energy-loss argument constrains axions, sterile neutrinos and other light states. Time-of-flight dispersion gave mν̄e bounds of order 10 eV, now superseded by laboratory limits. Reported events at the Mont Blanc LSD detector about 4.7 h earlier remain unexplained and are generally not considered part of the burst.
Detection channels for a galactic burst
For a collapse at 10 kpc the expected event yields are5:
- Super-Kamiokande: about 4,000–8,000 inverse beta decays, 120–250 elastic scatterings on electrons (pointing to a few degrees) and about 80–250 interactions on oxygen; gadolinium neutron tagging separates the channels.
- Hyper-Kamiokande: of order 10⁴–10⁵ inverse beta decays and about 1,000 elastic scatterings.
- IceCube: 10⁵–10⁶ additional photoelectrons as a coherent rate increase across all 5,160 sensors, giving the light curve with millisecond resolution but no event-by-event energy.
- DUNE: νe charged-current capture on ⁴⁰Ar, the main sensitivity to the neutronization burst.
- JUNO and other scintillator detectors: inverse beta decay plus elastic scattering on protons, which measures the νx spectrum.
Flavour conversion
Supernova neutrinos undergo MSW resonances in the stellar envelope; the neutronization burst should be strongly suppressed in the νe channel for normal ordering and less so for inverted ordering, giving an ordering-dependent signature. Close to the neutrinosphere the neutrino density is so high that neutrino–neutrino forward scattering produces collective oscillations, including fast flavour instabilities driven by angular crossings. Their effect on spectra and on the explosion is an active research field; in the outer layers they can swap spectra between flavours over certain energy ranges. Earth-matter effects for detectors at different locations add further information.
SNEWS and multimessenger strategy
SNEWS has run as an automated coincidence network since 20055. SNEWS 2.0 adds a public alert stream, triangulation from arrival-time differences between detectors, pre-supernova alerts (KamLAND estimates a 3σ detection about 48 h before the collapse of a 25 M☉ star at 690 pc) and a combined pointing of about 5° from DUNE and about 12° from liquid scintillator detectors. Precise bounce timing would narrow gravitational-wave searches, and the neutrinos would precede shock breakout by minutes for stripped progenitors and by hours for supergiants. The expected rate is 1.63 ± 0.46 core collapses per century in the Milky Way, so the probability of at least one within a 20-year detector lifetime is about 30 %6.
Diffuse supernova neutrino background
The DSNB integrates the cosmic core-collapse rate RCC(z) over redshifted neutrino spectra, including failed explosions forming black holes, which produce harder spectra. The ν̄e window lies between reactor antineutrinos (below about 10 MeV) and atmospheric backgrounds (above about 30 MeV). With gadolinium (0.01 % by mass from 2020, 0.03 % from 2022), Super-Kamiokande analysed 956.2 days above 9.3 MeV neutrino energy and found a spectral fit about 1.2σ from the background-only hypothesis, comparable to the pure-water phases7. JUNO and Hyper-Kamiokande with gadolinium are expected to establish the signal and constrain the fraction of failed supernovae.
Quiz
Test your knowledge
4 questions on this level · Print as worksheet
Open questions
- No neutrino burst has been detected from a supernova in the Milky Way; the next one would provide tens of thousands of events.
- The diffuse supernova neutrino background has not yet been detected at significance.
- How collective flavour conversion near the neutrinosphere affects the explosion and the observed spectra.
Terms to explore
- Supernova neutrinos
- SN 1987A
- SN1987A neutrino burst
- Core-collapse supernova mechanism
- Neutrinosphere
- Neutronization burst
- Collective neutrino oscillations
- Diffuse supernova neutrino background
- SNEWS (SuperNova Early Warning System)
- Supernova early warning
- Gadolinium loading
- Inverse beta decay
People
- Masatoshi Koshiba
- Stirling Auchincloss Colgate
- Richard H. White
- Hans Bethe
- James Ricker Wilson
- Hans-Thomas Janka
- Kate Scholberg
- Georg G. Raffelt
On the timeline
- 1966Neutrinos proposed to drive supernova explosions
- 1985Delayed neutrino-heating supernova mechanism
- 1983Large water Cherenkov detectors begin operation
- 1987Neutrinos from supernova SN 1987A
- 2002Nobel Prize for neutrino astrophysics
- 2020Gadolinium is added to Super-Kamiokande
References
- K. Hirata et al. (Kamiokande-II Collaboration), Observation of a neutrino burst from the supernova SN1987A, Phys. Rev. Lett. 58, 1490 (1987), doi:10.1103/PhysRevLett.58.1490
- R. M. Bionta et al. (IMB Collaboration), Observation of a neutrino burst in coincidence with supernova 1987A in the Large Magellanic Cloud, Phys. Rev. Lett. 58, 1494 (1987), doi:10.1103/PhysRevLett.58.1494
- E. N. Alexeyev et al., Detection of the neutrino signal from SN 1987A in the LMC using the INR Baksan underground scintillation telescope, Phys. Lett. B 205, 209 (1988), doi:10.1016/0370-2693(88)91651-6
- H. A. Bethe, J. R. Wilson, Revival of a stalled supernova shock by neutrino heating, Astrophys. J. 295, 14 (1985), doi:10.1086/163343
- S. Al Kharusi et al., SNEWS 2.0: a next-generation supernova early warning system for multi-messenger astronomy, New J. Phys. 23, 031201 (2021), doi:10.1088/1367-2630/abde33
- K. Rozwadowska, F. Vissani, E. Cappellaro, On the rate of core collapse supernovae in the Milky Way, New Astron. 83, 101498 (2021), doi:10.1016/j.newast.2020.101498
- Super-Kamiokande Collaboration, Search for diffuse supernova neutrino background with 956.2 days of Super-Kamiokande gadolinium dataset, arXiv:2511.02222 (2025), doi:10.3847/1538-4357/ae6c20
- H.-T. Janka, Explosion mechanisms of core-collapse supernovae, Annu. Rev. Nucl. Part. Sci. 62, 407 (2012), doi:10.1146/annurev-nucl-102711-094901
Last updated: · Editorial standard
Cite this page
Text
Neutrino Atlas (2026). “Supernova neutrinos”. Retrieved 22 November 2026, from https://neutrino-atlas.com/learn/supernova-neutrinos/BibTeX
@misc{neutrinoatlas2026learnsupernovaneutrinos,
title = {Supernova neutrinos},
author = {{Neutrino Atlas}},
year = {2026},
url = {https://neutrino-atlas.com/learn/supernova-neutrinos/},
note = {Accessed 2026-11-22},
language = {english}
}