The Explosion That May Have Detonated Twice

On August 18, 2025, two telescopes on opposite sides of a continent felt the faintest possible tug. The Laser Interferometer Gravitational-Wave Observatory, the same machine that opened the era of gravitational astronomy, registered a tremor so weak it barely cleared the threshold of belief. It looked like the death rattle of something very small. And in the patch of sky the tremor pointed to, 1.3 billion light-years away, a new point of light flickered into view and then began, almost at once, to fade. For three days it behaved exactly like the most famous explosion in modern astronomy. Then it stopped behaving like anything astronomers had ever confirmed.

A combination nobody had seen

The event carries two names, because two different instruments saw two different things and astronomers are not yet certain they belong to the same object. The gravitational-wave trigger is catalogued as S250818k. The point of light is AT2025ulz, first flagged by the Zwicky Transient Facility at Palomar Observatory as ZTF25abjmnps. Separately, each is unremarkable. Sub-threshold gravitational-wave triggers wash through the LIGO data stream constantly, most of them noise. Fading transients appear in survey images every night. What made this pair worth a paper in The Astrophysical Journal Letters, published in December 2025 and led by Mansi Kasliwal of Caltech, was that they arrived at the same place and the same time, and that the gravitational signal carried a number that should not exist.

The merging objects, taken together, had a chirp mass of roughly 0.87 times the mass of the Sun. From that signal the team concluded, at greater than 99 percent confidence, that at least one of the two objects weighed less than a single solar mass. That is a problem. Neutron stars, the collapsed cores left behind by dying massive stars, are not supposed to be that light. The lightest ones astronomers have weighed cluster around 1.2 solar masses, and theory has long held that a neutron star much below that simply cannot form by the ordinary route. A star massive enough to leave a neutron star behind leaves behind a heavy one.

At first, for about three days, the eruption looked just like the first kilonova in 2017. But then it started to look more like a supernova.

So the signal posed a paradox before anyone even looked at the light. A merger of two objects, at least one of them too small to be a normal neutron star, in a galaxy where something had also recently brightened. To understand why a small team of astronomers spent months chasing it, you first have to understand what a kilonova is, and what happened the last time the universe sent one clearly enough to read.

The night the gold was made

On August 17, 2017, LIGO and its European partner Virgo caught a gravitational-wave signal unlike the black-hole mergers they had detected until then. This one was a long, rising whine lasting roughly a hundred seconds, the signature of two neutron stars spiraling together. It was named GW170817. Just 1.7 seconds after the merger, NASA's Fermi telescope recorded a short burst of gamma rays from the same direction. Within hours, observatories around the world swung toward a sky region about 130 million light-years away and found a new point of light in the galaxy NGC 4993. It was catalogued AT2017gfo, and it was the first kilonova ever caught in the act.

A kilonova is what happens when two neutron stars collide. Each star is a city-sized ball of matter so dense that a sugar-cube volume of it would weigh as much as a mountain. When they merge, a fraction of that matter is flung outward at a sizable fraction of the speed of light. In that expanding cloud, free neutrons slam into atomic nuclei faster than the nuclei can radioactively decay, a runaway process called rapid neutron capture, or the r-process. It builds the heaviest elements in the periodic table, the ones ordinary stars cannot make: gold, platinum, the lanthanides, the actinides. The cloud glows as those freshly minted radioactive isotopes decay, and that glow is the kilonova.

AT2017gfo confirmed, for the first time with direct evidence, that neutron-star mergers are a dominant cosmic forge for the heaviest elements. The brightness of its glow implied an ejecta mass of roughly 0.03 to 0.06 times the mass of the Sun, split into a fast, hot, blue component rich in lighter r-process nuclei and a slower, cooler, red component loaded with lanthanides and heavier material. The gold in a wedding ring and the platinum in a catalytic converter trace their atoms to collisions like this one. AT2017gfo set the template. Every kilonova candidate since has been measured against its light curve, and for three days in 2025, AT2025ulz matched it.

What would make a kilonova super

The prefix "super" here is not about brightness in the way "superluminous supernova" is. It refers to a hypothesized origin story, one that nests a kilonova inside a larger explosion. The idea, developed in part by Brian Metzger of Columbia University, a long-time theorist of kilonovae, begins with a single massive, rapidly rotating star reaching the end of its life.

In the standard picture, such a star collapses into one neutron star or one black hole, and that is the end. But if the collapsing core is spinning fast enough, the physics can branch. Theory allows the infalling material to fragment, splitting through fission or breaking apart in a whirling accretion disk into two separate, unusually lightweight neutron stars rather than one ordinary-mass remnant. These would be the "forbidden" sub-solar neutron stars, objects that cannot form on their own but might be born in pairs inside the chaos of a collapse. If those two fragments then spiral together and merge within the still-expanding debris of the supernova that created them, the result would be a kilonova going off inside a supernova. A superkilonova.

If these forbidden stars pair up and merge by emitting gravitational waves, it is possible that such an event would be accompanied by a supernova rather than be seen as a bare kilonova.

The scenario makes a specific, testable prediction about what an observer would see. Early on, before the supernova debris brightens, the kilonova should dominate: a fast, red, rapidly fading glow, the signature of r-process material, exactly like AT2017gfo. Then, as the surrounding supernova ejecta catches up and the radioactive nickel inside it heats up, the source should brighten again, turn bluer, and reveal the ordinary fingerprints of an exploding star, including hydrogen, which a bare neutron-star merger has none of. The kilonova would be the opening act, the supernova the encore. That two-part performance is precisely what AT2025ulz appeared to deliver.

Reading the 2025 candidate

For the first three days, the transient faded fast and stayed red, hugging the curve that AT2017gfo traced in 2017. The team had reason to be excited. Then, around the fourth day, the object reversed course. It stopped fading. It grew brighter, shifted toward blue, and its spectrum sprouted hydrogen and helium, the marks of a stripped-envelope supernova of the type astronomers classify as Type IIb. A bare kilonova does not do this. A superkilonova, in which a supernova surrounds the merger, would do exactly this.

So the case rests on a chain of coincidences that line up suggestively. A gravitational-wave trigger indicating at least one sub-solar object, the kind the superkilonova model needs. A transient at the same position and time. An early phase that mimicked the only confirmed kilonova on record. A late phase that mimicked a supernova. Each link, on its own, has an innocent explanation. Together, they sketch the picture the theory predicted before anyone went looking.

The honest reading is that this picture is one of at least two. The same observations are consistent with a more mundane account: an ordinary Type IIb supernova that happened, by chance, to sit near a meaningless gravitational-wave glitch and to fade in a way that briefly resembled a kilonova during its first days. Supernovae are common. Sub-threshold gravitational-wave triggers are common. Surveys cover enough sky that the two will occasionally overlap with no physical connection at all. The authors are explicit that they cannot statistically rule out such a chance coincidence.

The distinction matters because the two interpretations make opposite demands on physics. A lone Type IIb supernova is a textbook event, the explosion of a massive star that lost most of its hydrogen envelope before dying. It asks nothing new of the universe. A superkilonova asks for a great deal: a collapse that splits its core into two lightweight neutron stars, a merger that follows almost immediately, and a kilonova bright enough to be seen through the surrounding supernova for those first three days. The burden of proof scales with the strangeness of the claim, and the team carries that burden openly rather than waving it away.

What the team will and will not claim

The restraint in the paper is the most scientifically interesting thing about it. The team does not announce a discovery. It announces a candidate and lays out, in the same breath, the reasons to doubt it. The gravitational-wave trigger was below the usual threshold of confidence. The spatial and temporal coincidence is compelling but not unique. The hydrogen that appeared late could mean a superkilonova revealing its supernova shell, or it could simply mean a supernova that was never anything else.

Kasliwal summarized the position plainly: the team does not know with certainty that it found a superkilonova, but the event is eye-opening regardless. That hedge is doing real work. If the superkilonova interpretation holds, it would be the first direct evidence of an entirely new channel for forming neutron stars, lightweight ones born in pairs from a single collapse, and a new way for the universe to forge its heaviest elements. If a sub-solar object were confirmed by other means, an even stranger possibility opens: that one of the merging bodies was not a neutron star at all but a primordial black hole, a relic from the first instants after the Big Bang, masquerading by its mass as something stellar.

None of those conclusions can be drawn from a single candidate. What the event does is sharpen the question and tell observers what to chase. The next sub-solar gravitational-wave trigger that coincides with a fading red transient will be examined against this template, the way AT2025ulz was examined against AT2017gfo. A second example, cleaner than the first, would move the superkilonova from a story the physics allows to a thing the universe actually does.

The instruments that make the question answerable

That a tremor barely above noise could trigger a worldwide hunt at all is a measure of how far the field has come in a decade. Multi-messenger astronomy, the practice of combining a gravitational-wave signal with the light from the same event, did not exist as a working method before GW170817. Now a sub-threshold flicker in the LIGO data can send the Zwicky Transient Facility scanning the relevant sky within hours, and a network of telescopes can follow a candidate's color and spectrum night after night, watching for the precise sequence a theory demands.

As the detectors grow more sensitive in their coming observing runs, the catalog of marginal triggers will swell, and so will the number of coincidences worth chasing. Most will be noise. A few may be the lightweight, forbidden mergers the superkilonova model needs. The work ahead is not to declare AT2025ulz a superkilonova, but to find the next ones and decide, with statistics rather than a single suggestive case, whether this two-part explosion is a genuine feature of the cosmos or a coincidence that fooled a careful team for three days.

For three days a faint tremor and a fading light agreed on a story that physics had only ever told on paper. Then the light changed its mind, and the question it left behind is no longer whether the universe can stage a double explosion, but how soon the next one will arrive to settle the argument.