Mining the Oldest Stars with Gaia: Putting Metal-Poor Star Classifications to the Test

Understanding how the Milky Way formed requires finding and studying its oldest stars, which are also its most metal-poor, meaning they contain very few elements heavier than helium. In this paper, first author Riley Thai and collaborators evaluate how well modern classification methods based on Gaia low-resolution XP spectra can identify these rare stars. The motivation is clear: extremely metal-poor (EMP; [Fe/H] ≤ −3.0) and very metal-poor (VMP; [Fe/H] ≤ −2.0) stars preserve chemical and dynamical information from the early universe, but they are difficult to find. The authors focus on testing the reliability of metallicity estimates from the large Gaia XP catalog presented by Andrae et al. (2023), using detailed follow-up observations to see how accurate those estimates really are.

Target Selection and Observations

The study begins with target selection and observations. From the Andrae et al. catalog, the authors selected 90 candidate metal-poor red giant branch stars using cuts on brightness, surface gravity, sky position, and Gaia XP metallicity estimates ([M/H]XP < −2.5). High-resolution spectra were obtained with the FEROS spectrograph at La Silla Observatory. After inspecting the data, 15 stars were identified as OBA-type contaminants, hot stars whose reddened spectra can mimic metal-poor stars, leaving a final sample of 75 genuine metal-poor candidates. This step highlights a known challenge: along the Galactic plane, dust extinction increases the risk of contamination, while stars at higher Galactic latitude are much cleaner selections.

Deriving Stellar Parameters

Next, the authors determine stellar parameters such as effective temperature, surface gravity, metallicity, and microturbulence. Temperatures were estimated from Gaia colors corrected for dust extinction, while surface gravities were derived using standard stellar physics relations combined with Gaia distances. In regions of high extinction, the commonly used two-dimensional dust maps produced unrealistic results, so the team employed three-dimensional dust maps or isochrone fitting to correct the problem. Metallicity ([Fe/H]) was then measured from iron absorption lines in the high-resolution spectra. These careful steps ensure that the derived parameters are physically consistent and suitable for precise chemical analysis.

Chemical Abundance Analysis

With reliable stellar parameters in hand, the authors perform a detailed chemical abundance analysis. Using the MOOG code and one-dimensional local thermodynamic equilibrium models, they measure abundances for up to 22 elements per star, including light elements, iron-peak elements, and neutron-capture elements. The results show that the majority of stars follow the expected abundance patterns of Milky Way halo stars. Importantly, the sample includes 13 extremely metal-poor stars and 62 very metal-poor stars, with the most metal-poor star reaching [Fe/H] = −3.43. The team also identifies chemically unusual objects, such as a carbon-enhanced metal-poor (CEMP) star and stars with unusual magnesium abundances, which provide clues about early star formation and nucleosynthesis.

Stellar Orbits and Galactic Context

The paper then examines the stars’ kinematics to understand their orbits within the Galaxy. By combining Gaia astrometry with measured radial velocities, the authors calculate full three-dimensional orbits in a realistic Milky Way potential. Most stars show orbital properties typical of the stellar halo, including high eccentricities and, in many cases, retrograde motion. This consistency between chemistry and dynamics supports the idea that these stars are ancient halo members, some of which may have originated in small galaxies later accreted by the Milky Way.

Validating Gaia XP Metallicity Estimates

A key goal of the study is to test the accuracy of Gaia XP metallicity estimates. Comparing their high-resolution [Fe/H] measurements with the original XGBoost-based [M/H] estimates from Andrae et al., the authors find strong agreement down to [Fe/H] ≈ −3.0, with only modest scatter. The Gaia XP estimates do not systematically label stars as more metal-poor than they truly are, which is crucial for efficient target selection. The authors also compare several other XP-based catalogs and find that multiple methods can robustly classify metal-poor stars, provided extinction and selection effects are handled carefully.

Summary and Implications

In summary, Thai et al. demonstrate that Gaia XP spectra are a powerful tool for identifying the most metal-poor stars across the Galaxy when combined with thoughtful selection criteria and follow-up observations. Their work validates current XP-based metallicity catalogs, reveals new EMP and VMP stars, and highlights where selection cuts may miss the rarest objects. For students and researchers alike, this study shows how large surveys and detailed spectroscopy work together to uncover the Milky Way’s earliest history, one ancient star at a time.

Source: Thai