There is a range of black hole masses that, according to stellar physics, should be nearly empty. Stars above a certain size are expected to blow themselves apart entirely in what astronomers call a pair-instability explosion, leaving no remnant at all. That expectation carves out a forbidden band, the pair-instability mass gap. And yet gravitational-wave detectors keep finding black holes with masses above roughly 40 times the Sun's, reaching into that gap and possibly past it.
A team led by Max Briel, working with twelve collaborators, set out to test whether the most ordinary explanation could account for these heavyweights: two massive stars orbiting each other, one collapsing into a black hole, and then that black hole feeding off its still-living companion until both end as black holes that spiral together and merge. No dense star cluster required, no exotic assembly. Just a binary running its course.
The question the team focused on is how greedily a black hole can eat. Standard theory says there is a speed limit, the Eddington limit, set by the radiation pressure of infalling gas pushing back against the inflow. Feed a black hole faster and the light it produces should blow the rest of the meal away. But detailed simulations of the magnetized, general-relativistic flows around real black holes have suggested this limit can be beaten, sometimes badly. So the authors ran their populations three ways: strictly Eddington-limited, at rates informed by those general-relativistic magnetohydrodynamic simulations, and fully conservative, meaning the black hole keeps every gram its companion hands over.
Three appetites, three different universes
The simulations used POSYDON, a code that models binary star evolution in detail rather than through simplified analytic recipes. On top of the three accretion settings, the team varied the natal kick, the velocity jolt a newborn black hole receives when it forms, including a version where very massive black holes get strong kicks. Kicks matter because they can tilt an orbit, and a tilted orbit leaves a fingerprint in the spins that detectors can read.
The first result is a clean negative. Super-Eddington accretion does not suppress high-mass mergers. One reasonable worry had been that a rapidly gorging black hole would drain and destabilize the binary, choking off the merger channel. It does not.
The second result is the sharper one. Fully conservative accretion, the greediest setting, produces more mergers, but it produces them with a signature that does not match the sky. The simulated population piles up at an effective spin of 0.6, a measure of how much of the pair's spin is aligned with its orbit, and it does so regardless of how hard the black holes get kicked. It also produces a narrow spike in mass ratio near 0.5, meaning the lighter black hole tends to be about half the mass of the heavier one. Observations point somewhere else entirely: effective spins clustered around zero and a mass-ratio distribution that looks flat, with no such spike. On the strength of that mismatch, the authors disfavor fully conservative accretion as the main way high-mass binary black holes are made.
The two more restrained prescriptions fare better. Both the Eddington-limited and the general-relativistic-simulation-informed versions reproduce the observed masses and mass ratios reasonably well. But they have their own gap: to produce systems with negative effective spin, where a black hole spins against the orbit, they need natal kicks. Modest kicks help, and with them these models can account for part of the high-mass population. Not all of it. The authors are explicit that something else is still needed to explain both the large fraction of negative effective-spin systems and the high spins measured for secondary black holes.
Why it matters
Gravitational-wave astronomy has moved past the era of counting detections one at a time. There are now enough mergers to speak of a population with a shape, and that shape is a constraint on physics that cannot be tested any other way. Nobody can put a black hole in a laboratory and measure how fast it eats.
What this work demonstrates is the leverage that comes from looking at three properties together rather than separately. Mass alone can be matched by several stories. Mass ratio alone can too. Line up primary mass, mass ratio, and effective spin at once and most of those stories fall apart, leaving a much narrower set of assumptions about accretion and kicks standing. That joint analysis is the method as much as the result.
The honest conclusion is a partial one, and the authors present it that way. Isolated binary evolution with sensible accretion physics explains some of the heavy black holes, and the greediest version of the story appears to be wrong. Something beyond a single pair of stars, whether mergers of black holes in crowded clusters or another route, is likely doing the rest of the work. The paper is a preprint submitted to The Astrophysical Journal and has not yet been through peer review.