Leftover Gas Disks May Divert Neutron Star Mergers into Fast Blue Transients

A disk of leftover gas could redirect the evolution of a neutron star binary before the second neutron star even forms. In a preprint posted on November 5, 2026, Aldana Grichener and colleagues model how these disks can turn some prospective double neutron star mergers into earlier collisions between a neutron star and a stellar core. They propose that those collisions could power luminous fast blue optical transients, or LFBOTs.

The study connects these outcomes through simulated binary evolution, rather than demonstrating that an observed transient followed this route.

How leftover gas changes a binary

During common-envelope evolution, a neutron star and its giant companion’s core orbit inside the giant’s envelope. If the binary survives and sheds that envelope, some gas may remain gravitationally bound. With sufficient angular momentum, it can settle into a circumbinary disk: a disk surrounding both objects rather than either one individually.

The disk exchanges mass and angular momentum with the pair. Depending on the interaction, the orbit can shrink or widen. In the authors’ models, substantial shrinkage generally dominates. A tighter orbit can help a future double neutron star system merge, but too much contraction brings the existing neutron star into the exposed stellar core before that core can explode and form another neutron star.

The researchers incorporated disk formation and evolution into the COMPAS population-synthesis code, using results from two-dimensional hydrodynamical simulations. Their grid contains 55 models, each following 10 million massive binaries, with variations in disk properties and uncertain stellar-evolution prescriptions.

Fewer mergers overall, but new mergers too

In the reference disk model, 5% of the giant’s envelope remains in a disk with an assumed characteristic lifetime of 100,000 years. Including disk interactions lowers the predicted local double neutron star merger rate from about 23 to 21 events per cubic gigaparsec per year, an approximately 8% reduction. Across the explored models, the reduction reaches 28%.

The net change hides substantial reshuffling. Some binaries lose their chance to become merging neutron star pairs, mainly through early neutron star–core collisions. Others become future mergers only because disk-driven contraction makes them tighter or helps them survive the second stellar explosion.

The decrease is therefore not universal. With one alternative prescription for supernova remnants and natal kicks, disks increase the merger rate because the newly enabled mergers outweigh the losses. Merger rates alone cannot uniquely identify disk effects.

A possible source of fast blue flashes

LFBOTs brighten over a few days and reach luminosities comparable to supernovae. The paper considers whether rapid accretion after a neutron star disrupts a stellar core could power such emission through outflows, with surrounding gas supplied by the core, disk and disk winds.

The reference model produces candidate neutron star–core mergers at about 11 events per cubic gigaparsec per year. That is approximately 15% of one observational LFBOT rate estimate quoted in the paper. However, the observed rate is highly uncertain, and the calculation does not establish that every candidate merger produces an LFBOT.

The predicted locations offer another test. About 99% of modeled candidates travel less than 100 parsecs relative to their local stellar population after the first supernova. This is not their distance from the galaxy’s center. The channel favors association with relatively young populations, but not necessarily ongoing star formation: the original star-forming environment may have dispersed before the transient occurs. Events far from any young population would be harder to explain without additional motion absent from these simulations.

Heavier survivors and missing early mergers

Gas accretion also changes the surviving binaries. In the reference model, their mean chirp mass rises from about 1.20 to 1.28 solar masses. The fraction with chirp masses resembling GW190425 increases to roughly 10%, compared with less than 1% in most models without disks.

That makes disk-fed growth a possible explanation for GW190425, not a demonstrated one. The substantial accretion is difficult to reconcile with measured masses of Galactic double neutron stars. Neither reference model fully reproduces those component masses.

Disk interactions also reduce mergers with delays below about 40 million years by 40% in the reference comparison. This would worsen the challenge of explaining early heavy-element enrichment solely through double neutron star mergers. Whether the alternative neutron star–core collisions themselves contribute substantially to r-process production remains unresolved here.

What limits the predictions

Important uncertainties include disk mass, thickness, viscosity and accretion efficiency. The underlying disk simulations omit self-gravity, and the population models do not self-consistently include disk torques in mass-transfer stability or expansion of stripped stars after renewed accretion. A realistic cosmic metallicity and star-formation history is also beyond the calculation. The preprint establishes a plausible, testable connection—not observational confirmation of the proposed pathway.