Models of young planet-forming disks struggle to explain low-mass disks

What makes a young disk change?

Planets form from gas and dust surrounding newborn stars. Understanding how those disks change helps establish how much material is available for building planets, and when.

Fabiola Antonietta Gerosa and colleagues investigate whether disk self-gravity and turbulent viscosity can explain observations across early evolutionary stages. Self-gravity is the attraction of disk material to other disk material; turbulent viscosity describes how turbulent motions redistribute angular momentum, allowing some material to move inward while the disk spreads outward.

Their study is available as an arXiv preprint, submitted on 9 November 2026. Its accompanying record states that it has been accepted for publication in Monthly Notices of the Royal Astronomical Society. This is a numerical modeling study compared with existing observations, not a new telescope survey.

Following gas and dust from a collapsing cloud

Using the FEOSAD hydrodynamics code, the team ran 12 main simulations beginning with collapsing cloud cores between 0.15 and 2.65 times the Sun’s mass. Eight additional runs explored changes in cloud rotation, turbulent viscosity and the collision speed at which dust fragments rather than sticks together.

The calculations follow gas and dust in a two-dimensional, thin-disk approximation for up to one million years. Dust can grow, fragment and drift toward the central star. The models do not include magnetic fields, interactions with neighboring stars or a mechanism for making planetesimals—the larger solid bodies that can become planetary building blocks.

To compare calculations with telescope surveys, the researchers converted simulated dust distributions into predicted emission. They applied survey sensitivity limits and weighted the models to reflect the stellar masses represented in the comparison samples. For several size comparisons, they measured the radius enclosing 90% of the simulated dust emission, matching the observational definition.

Evolutionary stages were assigned using the envelope of material remaining around each modeled system, rather than a single age schedule. This matters because the modeled systems evolve at different speeds: the early, largely envelope-free stage begins at about 130,000 years for a core initially containing 0.40 solar masses, but around 750,000 years for a 2.40-solar-mass core.

Right brightness early, a mismatch later

The models broadly reproduce the millimeter brightness of the youngest, deeply embedded Class 0 disks in Perseus. That agreement is consistent with gravity and viscosity being important early transport processes, although it does not establish that they are the only ones operating.

The agreement deteriorates later. In Perseus, the median observed millimeter emission is roughly three times lower in Class I than in Class 0. The simulated emission changes much less, leaving the modeled Stage I median about four times brighter than the observed Class I median.

These are comparisons between populations, not direct measurements of individual disks’ mass-loss rates. Fainter millimeter emission can reflect dust drifting toward the star, but it can also reflect grains growing into larger bodies that emit less efficiently at those wavelengths.

A comparison with ten young disks in Ophiuchus from the AGE-PRO survey reveals a related problem. Using the study’s adopted mass estimates, the simulated median gas mass and the dust mass inferred from simulated emission both exceed the corresponding observed medians by more than a factor of ten. The models can reproduce the high-mass end, but not the low-mass disks or the full observed spread—nearly four orders of magnitude in gas mass.

Meanwhile, the modeled dust-emission radii agree reasonably well with those Ophiuchus disks. Any proposed explanation must therefore account for the mass distributions without spoiling that size agreement.

How much material does the emission hide?

Disk masses inferred from telescope data depend on assumptions. Dense dust can hide material behind an opaque emitting surface, while grain composition and porosity change how efficiently dust absorbs and emits radiation.

A substantial observational caveat appears in the paper’s Ophiuchus comparison. The authors cite a multiwavelength analysis by Pulgarés and colleagues that uses the Ricci dust-opacity prescription and estimates a median dust mass roughly seven times the observational value adopted here. Higher observed dust masses could therefore reduce the dust-mass discrepancy; its size is not securely established by the simpler estimate alone.

That factor of seven is distinct from a separate result within the simulations: their actual dust masses are about twice those inferred from their synthetic millimeter emission under an optically thin assumption. Both sides of the comparison can hide dust, but not necessarily by the same amount.

Neither dust correction resolves the separate gas-mass discrepancy. Observational gas estimates have their own uncertainties, including molecular abundances and whether the measured emission is optically thin.

The authors also tested three dust-opacity prescriptions for their synthetic emission. These changed predicted brightness, especially at longer wavelengths and later stages, but did not eliminate the disagreement with the Perseus Class I and II samples.

Age and selection further complicate comparisons. The simulations stop at one million years, whereas the bright Class II disks selected from exoALMA are estimated to be 2–10 million years old. Agreement with that unusually bright sample is better, but cannot establish that the models explain typical older disks.

Missing physics, not an identified culprit

Within the limited parameter variations tested, changing rotation, viscosity or dust fragmentation did not solve the population-wide mismatch.

The authors discuss magnetic winds, accretion outbursts, radiation-driven mass loss and stellar encounters as possible additions. Removing gas could also accelerate inward dust drift. Encounters might affect some disks while leaving others massive, potentially broadening the distribution.

None of those explanations was demonstrated by these simulations. The result is a constraint on the tested models: gravity and viscosity reproduce several early disk properties, but do not explain the full observed range of inferred masses, particularly the low-mass disks. Better dust-mass estimates could narrow part of that gap without settling the separate gas problem.