Failed Supernova: When a Giant Star Vanishes Into a Black Hole

A collapse that never rebounds

Every core-collapse event begins the same way. A star above roughly 8 solar masses fuses its way to an inert iron core — iron is the end of the line, because fusing it consumes energy rather than releasing it. Once that core exceeds an effective Chandrasekhar mass (~1.4–2 M☉), electron degeneracy pressure fails and it implodes. In a fraction of a second the core shrinks from a few thousand kilometres to a proto-neutron star of ~30–50 km, with infall speeds reaching ~0.1–0.25 c. When the centre hits nuclear density (~2.7×10¹⁴ g/cm³) it stiffens abruptly and rebounds, launching a shock wave outward.

In a successful supernova, that shock plows outward and blows the star apart. But the shock almost always stalls within milliseconds, at a radius of ~100–200 km. Two things kill it: incoming iron is photodissociated into free nucleons, draining ~8.8 MeV per nucleon, and neutrinos stream out of the shocked gas, bleeding away pressure. The shock becomes an accretion shock, standing still while matter rains through it. Whether the star lives or dies now depends on a race: can neutrinos streaming off the proto-neutron star reheat the region behind the shock and revive it within a few hundred milliseconds to a second?

In a failed supernova, they cannot. The revival is too weak, fallback continues at ~0.1–1 M☉/s, and the proto-neutron star grows heavier until it crosses the Tolman–Oppenheimer–Volkoff limit (~2.2–2.5 M☉, the maximum a neutron star can support). At that instant neutron degeneracy pressure loses, and the remnant collapses through its own event horizon. The star does not explode — it swallows itself, and a black hole is born from the inside out.

The neutrino ghost and the star that vanishes

The collapse is not silent — it is just invisible. The gravitational binding energy released is enormous, ~3×10⁵³ erg (a hundred times what the Sun will radiate in its entire life), but roughly 99% of it escapes as neutrinos, which pass through the star and out into space without lighting it up. Almost none of that energy is deposited in the envelope, so there is no bright optical flash.

Yet the star cannot vanish with no trace, and here the physics is beautifully subtle. As the proto-neutron star radiates neutrinos over ~1–3 seconds before it collapses, it loses ~0.2–0.5 M☉ of gravitational mass (energy has weight). The core's pull on the overlying star suddenly weakens. For a compact star nothing much happens — but for a bloated red supergiant, whose tenuous hydrogen envelope is only weakly bound, this abrupt loss of gravity launches a gentle pressure wave (~10⁴⁷–10⁴⁸ erg) outward. As modelled by Nadezhin (1980) and Lovegrove & Woosley (2013), that wave can unbind ~0.1–0.5 M☉ of the outer envelope at only ~50–100 km/s.

The result is a faint, cool, months-long transient: a peak luminosity near ~10⁶ L☉ (a thousand times fainter than a real supernova), a red temperature of ~3,000–4,000 K powered by hydrogen recombination, lasting roughly a year before fading. Piro (2013) noted the wave may also break out through the surface as a brief precursor flash. Then nothing. Where a supergiant blazed, only a fading infrared point remains — the signature of a little leftover gas falling onto a new black hole.

The red supergiant problem

The first hint that some stars die quietly came not from watching an explosion but from cataloguing the ones that should have exploded. Because supernovae recur in nearby galaxies, astronomers can dig into pre-explosion archival images and identify the actual star that blew up. Stephen Smartt and collaborators did this systematically for Type II-P supernovae (the ordinary hydrogen-rich kind from red supergiants) and found something odd: the progenitors all had masses below ~16–18 M☉.

The trouble is that red supergiants are observed up to ~25–30 M☉. So where are the supernovae from the heaviest supergiants? Statistically, a couple dozen higher-mass progenitors should have turned up by now — and none had. This deficit is the red supergiant problem. The most economical explanation is direct collapse: stars above ~17–18 M☉ tend to fail, imploding to black holes without a bright supernova, so they never appear in the progenitor tally at all.

The problem is not fully settled. Davies & Beasor (2018, 2020) argued that better dust corrections and bolometric estimates push the upper mass limit higher, softening or even erasing the deficit. Others note that heavy mass loss and circumstellar dust can hide the most massive progenitors. So the red supergiant problem is real evidence for failed supernovae — but the exact mass threshold, and how sharp it is, remains debated.

Which stars fail: compactness and islands of explodability

Mass alone does not decide a star's fate. What matters is the compactness of the core at collapse — essentially how much mass is packed into the innermost region. O'Connor & Ott (2011) captured this with a single number, the compactness parameter ξ, roughly the mass interior to a given radius divided by that radius (evaluated at the inner ~2.5 M☉). A high compactness means the mass profile is steep and dense: fallback is fast, the accretion ram pressure is crushing, and neutrino heating struggles to revive the shock — so the star tends to collapse to a black hole. A low compactness means a gentler, more diffuse core that explodes more easily.

Crucially, compactness does not rise smoothly with a star's birth mass. Late-stage nuclear burning (carbon and oxygen shell flashes) makes it swing up and down, producing "islands of explodability" — interleaved mass ranges that explode and mass ranges that fail, rather than a single clean cutoff. Detailed studies (Sukhbold et al. 2016; Ertl et al. 2016, using a two-parameter explosion criterion) find that many stars in the ~20–30 M☉ range, and pockets even below that, are prone to failure. This explains why the transition is fuzzy: two stars of nearly identical mass can meet opposite ends depending on the fine structure of their final burning stages.

Catching a star in the act of disappearing: N6946-BH1

If failed supernovae are real, the smoking gun is not a new light in the sky but the absence of one — a star that was there and then is not. Christopher Kochanek and collaborators launched exactly this hunt in 2008: a survey with the Large Binocular Telescope monitoring roughly a million red supergiants across ~two dozen galaxies within ~10 Mpc, watching year after year for a supergiant that simply switches off.

In 2015, Gerke, Kochanek & Stanek reported the first strong candidate, and follow-up by Adams et al. (2017) made it famous: N6946-BH1, a ~25 M☉ red supergiant in the star-forming galaxy NGC 6946 (the "Fireworks Galaxy," ~6.7–7.7 Mpc away). In 2009 the star brightened modestly, reaching ~10⁶ L☉ for several months — exactly the faint, cool transient the unnova models predict — and then faded away. By 2015, Hubble found no optical counterpart where a luminous supergiant had once shone; only a dim, fading infrared source remained, consistent with a trickle of fallback onto a newborn black hole. It is the leading candidate failed supernova known.

It is not airtight. Later JWST imaging has complicated the picture, resolving the position into what may be more than one source and probing whether the "vanished" star is instead dust-obscured or the remnant of a stellar merger. The disappearance interpretation remains the best explanation but is still being tested — which is precisely why deeper, longer disappearance surveys matter.

How else we could catch one — and the mass gap it explains

Optical disappearance is only one channel. A failed supernova in our own Galaxy would announce itself through neutrinos: detectors like Super-Kamiokande, IceCube, and DUNE would record a burst that is hotter than a normal supernova's (because the collapsing core keeps heating up) and that ends with a sharp, abrupt cutoff — the instant the proto-neutron star crosses its horizon and the neutrino source blinks out. That sudden truncation is a near-unique fingerprint of black-hole formation. Failed collapses are also thought to contribute a hard tail to the diffuse supernova neutrino background. Gravitational waves from the turbulent, sloshing pre-collapse core (driven by the standing accretion shock instability and neutrino-driven convection) offer a further, if fainter, signal.

Failed supernovae also relieve pressure on a puzzle from the gravitational-wave era. LIGO and Virgo routinely find black holes far heavier than the ~5–15 M☉ predicted if every massive star exploded and shed most of its mass. Direct collapse solves this: a star that fails to explode keeps nearly its entire helium core (and sometimes more), forming a heavier black hole. The pattern of which stars fail helps sculpt the observed black-hole mass gap — the scarcity of compact objects between the heaviest neutron stars (~2.5 M☉) and the lightest black holes (~5 M☉). In short, the stars that quietly vanish may be the very ones building the black-hole population we now hear merging across the cosmos.

Outcomes of iron-core collapse in a massive star

Outcome	Shock revived?	Optical display	Remnant	Example / progenitor
Classic Type II-P supernova	Yes — neutrino heating relaunches it	Bright: peak ~10⁸–10⁹ L☉ for months	Neutron star	SN 1987A (~18 M☉ blue supergiant)
Fallback / weak supernova	Partially — a feeble explosion	Dim, low-⁵⁶Ni supernova	Black hole by late fallback	Some faint Type II / 'gap' transients
Failed supernova (unnova)	No — shock stalls and is smothered	Faint ~1-yr glow, then the star vanishes	Black hole by direct collapse	N6946-BH1 (~25 M☉ red supergiant)
Electron-capture supernova	Yes — soft explosion of an O–Ne core	Faint, fast supernova	Neutron star	~8–10 M☉ super-AGB stars
Pair-instability supernova	N/A — thermonuclear disruption, no core collapse	Extremely bright, long-lived	Nothing left (full disruption)	~140–260 M☉ metal-poor stars