Globular Clusters May Preserve Clues to Mixing in a Milky Way Progenitor

Globular clusters are dense groups of stars that formed early in the history of galaxies. Many show little variation in iron from star to star. Yet some metal-poor clusters have appreciable differences in elements associated with the rapid neutron-capture process, or r-process, which produces many heavy elements. The study uses europium as a tracer of r-process material. The puzzle is how stars in one cluster could inherit different europium abundances despite having similar iron abundances.

One possibility is that a cluster’s stars formed from gas that was already chemically uneven. Rare enrichment events can deposit r-process material into a galaxy, and turbulent motions redistribute it. If some chemical structure survives before stars form, the stars could preserve a record of their natal environment. Lailani Kenoly and colleagues test whether this explanation is plausible in simulations; they do not establish it as the only explanation for the observations.

Following rare elements through turbulent gas

The researchers analyzed a three-dimensional simulation of a high-star-formation-rate Milky Way progenitor. The modeled patch of galactic disk spans one kiloparsec and is followed for 120 million years, with a finest spatial resolution of about four parsecs. Frequent core-collapse supernovae supply iron and drive turbulence. Rarer events add r-process material: in the model, their rate is one-hundredth of the core-collapse supernova rate, and each injects one-hundredth of a solar mass of r-process ejecta. These are simulation assumptions, not measurements of event rates or ejecta masses. The calculation does not identify the rare event as a particular kind of supernova or a compact-object merger; it tracks the material after it enters the disk.

To examine gas that might trace star-forming environments, the team selected cooler, denser simulation cells and sampled spherical regions around them at different sizes. They measured how abundances differed between regions and how much variation remained within each region. This distinction matters: a galaxy-wide abundance spread does not directly reveal the variation in one future cluster’s birthplace.

The selected gas was not a resolved model of molecular clouds. The simulation also did not follow a cluster’s collapse, fragmentation, or star formation. Its results describe chemical structure before those stages, not the final abundance spread among cluster stars.

Patchy enrichment, changing with time and scale

In the simulation, rare r-process enrichment produces an uneven europium-to-iron pattern: some locations are more enriched than others. Turbulent transport redistributes the ejecta, while inhomogeneities remain across a range of scales. As chemical evolution proceeds, abundance variance in the selected cooler, denser gas decreases.

The scale used to sample the gas also matters. Smaller regions show greater differences in their characteristic abundances from one region to another. Larger regions average over more of the abundance structure and have more similar characteristic abundances. This is a real-space averaging result, not a direct measurement of a particular turbulent wavelength or scale.

The team compared the simulations with published stellar abundance measurements, focusing its main observational comparison on clusters analyzed on a homogeneous abundance scale. Several of the most metal-poor clusters show substantial star-to-star r-process variation, and the sample as a whole suggests smaller variations at higher iron abundance. The simulations provide substantial pre-collapse variation at low metallicity and show a broad decline as the gas becomes more enriched. This resembles the ensemble trend, but the comparison is qualitative—not a point-by-point match to individual clusters.

A possible fossil, not a settled explanation

The results support the possibility that at least some cluster stars inherited r-process differences from their natal gas. But simulated gas variations are larger than the observed stellar variations. The authors note that cloud assembly, collapse, turbulence, and feedback from young stars could further redistribute the material before stars form. Their calculation does not model these stages or determine how much of the initial pattern survives.

The observational trend is not equally secure in every cluster. The paper describes evidence for an intrinsic spread in M5 as tentative and says NGC 2298 alone does not strongly establish the broader trend. The metallicity pattern is an ensemble suggestion, not a rule demonstrated independently in every system.

The study does not identify the source of the r-process material, prove that pre-existing gas structure caused the stellar differences, or rule out enrichment associated with forming clusters. It shows that a modeled turbulent disk can retain chemical structure across finite regions, and that the available variance declines with enrichment in a way broadly consistent with the cluster sample. The work is an arXiv preprint, so it should be read as research that may not have undergone journal peer review. Simulations that follow cloud collapse and star formation, together with more precise observations, will help test how much of this proposed chemical fossil can survive in stars.