New model estimates how aging stars replenish a nearby dwarf galaxy

Stars do not keep all their material for life. In their later stages, some swell into cool giants and shed gas and dust through stellar winds. A new study introduces SEDust, an automated tool for estimating this material loss, and applies it to NGC 6822, also known as Barnard’s Galaxy.

In their study, available as an arXiv preprint and accepted for publication in The Astrophysical Journal, Hamidreza Mahani and colleagues estimate that the galaxy’s sampled population of aging giant stars releases 0.00370 ± 0.00116 solar masses of gas and dust per year. Under their preferred assumptions, that amounts to about 37% of the material being converted into new stars annually. However, uncertainties in the modeling and the conversion from dust to gas make that comparison less settled than the headline number alone suggests.

Reading the light of dusty stars

NGC 6822 is a nearby dwarf irregular galaxy, about 490,000 parsecs away, with relatively low abundances of elements heavier than helium. The researchers focused on asymptotic giant branch, or AGB, stars: evolved stars whose extended atmospheres can support dust formation and substantial mass loss.

Their catalog contains 1,050 oxygen-rich and 560 carbon-rich AGB stars, distinguished by their chemical composition. It includes 59 long-period variables—large-amplitude pulsating stars with periods longer than 100 days. These variables are a subset of the 1,610 stars, not an additional population.

The team combined near- and mid-infrared measurements from UKIRT and Spitzer with optical observations and selected James Webb Space Telescope measurements. Dust absorbs some starlight and emits energy at longer wavelengths. Measuring brightness across multiple wavelength bands therefore helps reveal both the star and its surrounding dusty envelope.

SEDust compares these measurements with predictions from DUSTY, a program that calculates how radiation interacts with dust. This distribution of brightness across wavelengths is called a spectral energy distribution. The pipeline selects the closest match by minimizing an error-weighted measure of disagreement between observations and models.

The researchers constructed separate libraries of 25,200 models for carbon-rich and oxygen-rich stars. They varied stellar temperature, the temperature at the inner dust boundary, and optical depth—a measure of how strongly the envelope blocks radiation. The models assume spherical envelopes and calculate wind structure and final outflow speed using DUSTY’s radiation-driven wind solution.

A recycling rate with sizable uncertainty

Within the SEDust analysis, oxygen-rich stars contribute approximately 73% of the sampled population’s total mass loss. The models also show a general increase in mass loss with stellar luminosity, although individual stars scatter substantially around that trend.

For the pulsating stars, redder near-infrared colors generally accompany thicker dust envelopes. But color becomes a less sensitive guide at the reddest end: additional dust need not produce a proportional color change. The researchers caution that relations calibrated on these strongly pulsating stars do not necessarily describe the broader AGB population.

Two alternative approaches illustrate the uncertainty in the total. A method calibrated on the long-period variables gives about 0.00162 solar masses per year for the full catalog, while the DESK fitting package gives about 0.00887. These are roughly 2.3 times lower and 2.4 times higher than SEDust’s estimate. The authors treat that spread as a method-dependent uncertainty, not something covered by SEDust’s quoted error bar. They associate much of the discrepancy with differences in model grids and parameter sampling, particularly for carbon-rich stars.

The statistical error assessment also has limits. The team repeatedly perturbed the measurements of one well-observed star and refitted it, then applied representative fractional uncertainties across the catalog. Combined with an adopted 30% modeling uncertainty, these yield typical mass-loss errors of about 31% for carbon-rich models and 41% for oxygen-rich models. The authors describe these as best-case estimates, rather than rigorous errors for every star. Crowding and difficulty detecting faint or heavily obscured stars also make the catalog incomplete.

What the numbers can—and cannot—show

Assuming a gas-to-dust mass ratio of 200, the study estimates dust injection at about 0.000019 solar masses per year. A separate JWST-based analysis discussed in the paper obtained a global estimate ten times lower; the cause remains unresolved.

The gas-to-dust ratio is crucial. With the preferred value, AGB recycling falls short of the cited star-formation rate of about 0.01 solar masses per year. But an alternative ratio near 1,000, inferred from galaxy-wide interstellar measurements, would imply enough return to rival or exceed it. Whether that ratio applies to individual stellar envelopes is uncertain.

Nor does extending today’s dust output over billions of years establish a historical dust shortfall: star formation has varied, and dust is destroyed. SEDust provides a reproducible estimate of present-day stellar recycling, not a definitive inventory of the galaxy’s fuel supply or surviving dust.