Not all supernovae go quietly into the cosmic night some fight to the very end. That is how Alexander Gagliano, the lead author of a new paper in The Astrophysical Journal, characterized SN 2023zkd, a supernova unlike anything observed previously. Discovered in July 2023, the event seems to be the first well-documented instance of a massive star exploding as it was trying to engulf a black hole companion.
The drama played out 730 million light-years away, in a binary system where a black hole and a massive star were in a decaying orbit. As their distance narrowed, the black hole’s monstrous gravity stripped gas and dust from the star, creating a hot accretion disk. The star’s distorted core suffered gravitational stress to a breaking point, sending out a supernova before the star was able to completely engulf its companion. Our analysis shows that the blast was sparked by a catastrophic encounter with a black hole companion, and is the strongest evidence to date that such close interactions can actually detonate a star, said Gagliano.
The early light curve appeared to be suspiciously normal one bright peak, after which the supernova faded out. But 240 days afterward, the supernova re-brightened, getting almost back to its original light. Archival data showed an even more peculiar turn: the system had been gradually brightening for over four years prior to the explosion. Long-term pre-supernova activity like this is extremely uncommon and indicates persistent instability in the progenitor system.
In detailed modeling, the early peak was seen to be driven by the blast wave impacting low-density gas that envelops the system, and the second peak resulted from a slower, longer-duration collision with a dense, disk-shaped cloud of material ejected in the final few years of the star. The geometry and chemistry of this circumstellar material helium-rich in the outer, high-velocity layers, hydrogen-rich in the inner, slower layers are consistent with predictions for a massive star under the severe tidal stressing of a compact companion. 5–6 solar masses’ worth of material is estimated to have existed, sufficient to fuel the long-term re-brightening.
The find was enabled by a new machine learning algorithm added to the Young Supernova Experiment’s survey pipeline. Built to sweep across incoming data streams in real-time, the system highlighted SN 2023zkd hours after its emergence. Our machine learning system flagged SN 2023zkd months before its most unusual behavior, which gave us ample time to secure the critical observations needed to unravel this extraordinary explosion, Gagliano described.
The speed of the algorithm was essential. Time-domain astronomy is based on fast response: the time window in which to obtain the most informative phases of a transient can be merely hours once it has been discovered. The Young Supernova Experiment, sponsored by the University of California, Santa Cruz, utilizes the Pan-STARRS1 and Pan-STARRS2 telescopes to survey 4% of the night sky every three days, finding thousands of new transients per year. For SN 2023zkd, the early warning prompted a world-wide observing campaign, tapping into NASA’s Neil Gehrels Swift Observatory, the Zwicky Transient Facility, the Asteroid Terrestrial-impact Last Alert System, and large ground-based observatories such as Magellan, MMT, and Las Cumbres.
From a physics point of view, the system’s destiny was sealed. In one possibility, the star’s gravity pulled the black hole in, but the black hole’s dense mass having an event horizon a few kilometers in extent imposed such strong tidal forces that the star’s core imploded. In the other, the black hole ripped the star apart whole, with the debris that resulted crashing into nearby gas to generate the light seen. In both cases, the end product was a single, heavier black hole.
The consequences extend past this one incident. V. Ashley Villar, a co-author and assistant professor at the Center for Astrophysics | Harvard & Smithsonian, said, 2023zkd shows some of the clearest signs we have seen of a massive star interacting with a companion in the years before explosion. We think this might be part of a whole class of hidden explosions that AI will help us discover.
With the Vera C. Rubin Observatory preparing to scan the entire southern sky every few nights, astronomers expect the detection rate of such anomalies to surge. Rubin’s 3,200-megapixel camera will capture an area equivalent to 40 full moons in each 15-second exposure, generating 500 petabytes of data over a decade. AI-driven anomaly detection will be essential to sift through millions of nightly alerts, flagging the rarest and most scientifically valuable events for immediate follow-up.
To astrophysicists, SN 2023zkd is than an oddity; it is a demonstration of principle that violent binary interactions can initiate supernovae, and that machine learning can catch them in the process. As evermore surveys go on line, the previously occult terminal behavior of supermassive stars can sometime in the near future be made manifest in unparalleled detail.
