Gravitational Waves Reveal a Pair-Instability Mass Gap Near 45 Solar Masses

Somewhere in the growing catalog of gravitational-wave detections, a boundary appears. Black holes just below about 45 times the Sun's mass tend to carry small effective spins. Just above that line, the spins are larger and seemingly random. And the total merger rate drops steeply nearby—the paper calls this feature the 'cliff.'

The boundary is not a wall in space. It is a statistical edge in the mass spectrum of black holes, and it may be a signature of how the most massive stars die. In a paper posted to arXiv on September 4, 2025, Fabio Antonini and colleagues analyze the fourth LIGO-Virgo-KAGRA transient catalog, known as GWTC-4, to test whether the so-called pair-instability mass gap is imprinted on the black holes that gravitational waves reveal. The paper is an arXiv preprint.

Why Some Black Holes Should Not Exist

In the cores of extremely massive stars, electron-positron pair production can occur. That process saps pressure, the core contracts, and thermonuclear burning can run away. For stars in a certain mass range, theory predicts pulsational pair-instability eruptions or total explosions that suppress direct black hole formation. The expected result is a gap in the birth masses of black holes, beginning around 40 to 50 solar masses.

Yet gravitational waves have shown black holes heavier than that. How? One route is hierarchical merging in dense star clusters: two black holes merge, the remnant stays in the cluster, finds another partner, and merges again. General relativity predicts that such second-generation black holes can acquire large spins, and dynamical encounters should randomize their orientations. The result would be a high-mass population with large, isotropic effective spins. That parameter, chi_eff, is a mass-weighted measure of how much the black holes' component spins are aligned with their orbital motion.

Reading Spin as a Fingerprint

Antonini and colleagues selected 153 binary black hole mergers from GWTC-4 with low false alarm rates, excluding events likely to involve neutron stars. They modeled how the distribution of chi_eff changes with primary black hole mass. Instead of assuming a fixed shape, they let flexible Gaussian processes infer the mass spectrum and spin distribution, while accounting for the fact that detectors are more sensitive to some masses than others.

Two model variants in the analysis converge on a transition near 45 solar masses. The parametric fit gives a lower edge of 45.3 with uncertainties of plus 6.5 and minus 4.8 solar masses. A more flexible non-parametric fit places the transition at 47.5 with uncertainties of plus 12.2 and minus 8.6. Below the transition, the spin distribution is a narrow Gaussian centered near 0.03, consistent with first-generation black holes born from ordinary stellar collapse. Above it, the spins broaden and become consistent with isotropic orientations. The low-spin population effectively vanishes above the boundary, while the high-spin population spans the full mass range and occupies the gap. The preference for two distinct spin populations over a single shared distribution is strong, with a Bayes factor greater than 10,000.

Another clue is the merger rate itself. The team reports a drop of nearly two orders of magnitude around 40 solar masses, which they call the cliff, followed by a shallower plateau. In cluster models, that shape arises naturally when direct collapse stops producing black holes above the pair-instability limit but repeated mergers continue to build them.

The Gap Is Still There, but Something Is Filling It

The findings support the idea that pair instability shapes the black hole mass spectrum. They also suggest the black holes seen above the gap are not spawned directly at those masses. They are more likely the products of earlier mergers, which can happen in globular clusters, nuclear star clusters, or disks around active galactic nuclei. The paper notes that if all mergers above 45 solar masses come from globular clusters, models suggest those clusters formed with densities around 100,000 solar masses per cubic parsec. That is an inference from simulations, not a direct measurement.

A Nuclear Reaction Written in Black Hole Masses

The exact lower edge of the pair-instability gap depends on how much carbon versus oxygen is left in a massive star's core before it collapses. That balance is governed by the reaction carbon-12 plus helium-4 producing oxygen-16, written 12C(alpha,gamma)16O. If the reaction is faster, stars make more oxygen and the gap begins at lower masses. If slower, carbon lingers and the gap shifts higher.

Assuming the measured transition is the lower edge of the gap, the team translates its mass estimate into an astrophysical S-factor at 300 keV: 242.5 with uncertainties of plus 310.4 and minus 101.5 keV b. The S-factor is a standard way to express the probability of a nuclear reaction at low energies. The estimate is consistent, within its broad uncertainties, with recent nuclear physics determinations. The authors argue that direct nuclear experiments have large uncertainties and that this gravitational-wave-based estimate can provide substantially tighter bounds, making it a new kind of constraint: gravitational-wave data can speak to reaction rates inside stars.

What Could Still Move the Edge

The interpretation rests on two important assumptions. First, that the transition mass is the lower edge of the pair-instability gap. Second, that the population above it consists entirely of second- or higher-generation mergers. Stellar evolution is uncertain. Metal-poor stars that keep a hydrogen-rich envelope, weakened stellar winds, rotation, convection, and failed supernovae can shift the gap or allow some massive black holes to form directly. Primordial black holes are another possible contributor. The analysis also finds a possible isotropic population near 14 solar masses, but the paper says the feature is not statistically required. The lower bound of the high-spin distribution is less constrained than its upper bound.

This analysis shows black hole mergers are becoming a probe of stellar death, cluster dynamics, and nuclear physics at once. The transition near 45 solar masses is not a physical wall. It is a change in the population, a fingerprint left by stars that may have destroyed themselves before they could become black holes, and by their remnants that found each other and merged again. As LIGO-Virgo-KAGRA continues to collect events, that fingerprint should grow sharper, turning the mass spectrum of black holes into a new window on the carbon and oxygen forged in massive stars.