The Milky Way's Earliest Stars May Be Buried in Its Heart

Old stars have been found in the Milky Way's stellar halo, in ancient globular clusters, and in dwarf galaxies far from the bright disk. But the Dorcha simulation suite now argues that the remnants of the Galaxy's earliest star-forming building blocks should be concentrated much closer in: in the innermost regions of the bulge and inner halo.

A dark-matter map of ancient star-forming sites

Dorcha is Gaelic for dark, pronounced roughly DUR-uh-khuh. The suite consists of 20 high-resolution, dark-matter-only cosmological zoom-in simulations that follow 25 Milky Way-analogue haloes. Fifteen of those haloes are isolated; the other ten are in five pairs, similar to the Milky Way and Andromeda (M31). These are gravity-only simulations: they follow dark matter, not gas, stars, or supernovae.

The researchers use merger trees to identify the most bound 10 percent of particles in high-redshift progenitor haloes at redshifts z ≥ 5. Higher redshift means earlier cosmic time, so these are among the small structures that could have hosted the first episodes of star formation. Deeply bound particles trace the cores of those early haloes, where gas could cool and collapse.

By tagging those particles and following them to the present, the team can ask a simple question: where would the material associated with the Milky Way's first stars be today?

The oldest material ends up at the center

The answer is strikingly central. Across the 25 simulated haloes, 90 to 100 percent of this ancient material ends up within about 15 h⁻¹ kpc of the halo center, roughly a tenth of the halo's virial radius. For the oldest tagged material, from z ≥ 15, the probability distributions peak at 100 percent. Material tagged at lower redshifts, between z = 5 and z = 10, is somewhat less concentrated, with peaks around 70 to 80 percent, but the trend is the same.

The simulations also reveal a distinctive kinematic fingerprint. The ancient material has a steep density profile, falling off roughly as r⁻⁴ with increasing radius. Its velocity dispersion—the spread of velocities around the local mean—is low. Its orbits are preferentially radial: the material plunges in and out of the center more than it circles. And the mean radial velocity is near zero, indicating that it has had time to relax into dynamical equilibrium.

In phase space, the combination of position and velocity, this ancient component appears mixed, relaxed, and cold compared with the bulk of the halo. Some fine structure survives as caustics and ridges from past mergers, but the high-redshift material itself has settled deep in the potential well.

Mergers and neighbors barely change the picture

The result is surprisingly robust. The team split the haloes into early-forming and late-forming groups, and into isolated and paired systems. If recent mergers or a massive neighbor like Andromeda were scrambling the oldest material, these categories should look different. They mostly do not.

Mass density profiles remain consistent, and kinematic differences are minor. The clearest variation is in orbital anisotropy, with late-forming haloes—mostly in pairs—showing somewhat less radial orbits in their outskirts. That points to environment rather than formation time. The broader lesson is that once ancient material sinks into the central region of a Milky Way-scale halo, it is hard for later mergers to erase its dynamical signature.

This matters for Galactic archaeology, the field that reads the Milky Way's assembly history from the ages, chemistry, and motions of its stars. Surveys such as Gaia, GALAH, APOGEE, 4MOST, and LSST are mapping large numbers of stars. The Dorcha simulations suggest that targeted searches for the oldest populations should focus on the crowded, dust-obscured inner Galaxy.

What this means for the first stars

The paper argues that the fossil record of the first generations of stars, including metal-free Population III stars and their extremely metal-poor descendants, should be sought in the innermost regions of the Milky Way. There, the simulations predict, old material retains a distinctive combination of low velocity dispersion, radial orbits, and central concentration.

Observations already offer a hint. The Pristine Inner Galaxy Survey found metal-poor stars in the Galactic bulge with kinematic features that qualitatively match the simulations, including lower rotation for lower-metallicity stars. But the match is not proof. It is a physically motivated baseline for interpreting what surveys see.

The authors also stress a chemical caveat. Low metallicity does not uniquely identify ancient stars, nor do all ancient stars have low metallicity. Metal-poor stars can form later from gas enriched by earlier generations. Age and chemistry overlap, but they are not interchangeable labels.

What the simulations cannot yet say

These are dark-matter-only simulations. They do not include gas cooling, star formation, stellar feedback, metal enrichment, or the gravitational pull of a real galactic disk and bar. The tagged particles mark potential sites of early star formation, not actual stars. The suite also does not distinguish between long-lived Population III stars and extremely metal-poor stars that formed later.

The team plans to address those gaps with the Solas suite, a set of hydrodynamical simulations that will include ordinary matter and galaxy formation physics. Until then, Dorcha provides a theoretical guide rather than a direct detection. The work is posted as a preprint, and its predictions await observational tests.

Even with those caveats, the implication is that the search may need to look inward. The Milky Way's oldest stories may not be written only in its distant halo. They may lie beneath the dust and glare of its heart, waiting for surveys patient enough to read them.