Disentangling Milky Way halo populations at low metallicities using [Al/Fe]

TL;DR

This study demonstrates that aluminium abundances, properly measured and corrected for NLTE effects, effectively distinguish accreted from in-situ Milky Way halo populations, even at low metallicities where kinematic classifications are ambiguous. Using high-quality UVES spectra of the Nissen & Schuster sample, the authors show a robust separation in $[Al/Fe]$ with a practical threshold around $-0.3$, complementing traditional $\alpha$-element trends. They carefully address uncertainties, NLTE corrections, and blends, and discuss the nucleosynthetic origins and environment dependence of aluminium enrichment. The work highlights the value of combining chemical tagging with kinematics to unravel the Galaxy’s assembly history, particularly for the Gaia-Sausage-Enceladus remnants in the metal-poor regime.

Abstract

Differentiating between in-situ and accreted populations in the Milky Way halo is a challenging task. Various kinematic spaces are often used to identify distinct accreted populations from the in-situ Milky Way halo. However, this approach has limitations, especially at low orbital energies. To overcome this ambiguity, elemental abundances are typically used to distinguish between the populations. Yet, for many elemental abundance ratios, it remains difficult to make this distinction at low metallicities. Aluminium abundances, on the other hand, have been empirically found to be an effective discriminator, allowing for the separation of accreted and in-situ populations in the Milky Way halo even at low metallicities and low orbital energies. We aim to test the discriminating power of [Al/Fe] using a well-studied sample of high-velocity stars in the solar vicinity with high-quality spectra. With these stars, we explore the [Al/Fe] ability to separate the in-situ from accreted stars and test its limitations. We derived aluminium abundances from the Al I 3944 and 3961 {\rm Å} lines for 45 stars observed in two ESO programmes, along with 11 stars with archival spectra. Aluminium abundances were determined using 1D LTE and 1D NLTE spectral synthesis and line profile fitting. We confirm that the low-$α$ population systematically has lower [Al/Fe] compared to high-$α$ stars. Aluminium abundances, when carefully measured and NLTE effects taken into account, are effective tracers of the chemical history of halo stars. They provide an independent constraint on origin, complementing $α$-element abundances trends, and help us to disentangle subpopulations within the accreted halo, especially in the metal-poor regime.

Figures (8)

• Figure 1: Normalised spectra for seven stars spanning a range of metallicities, from $\rm [Fe/H]=-1.41$ at the top to $\rm [Fe/H]=-0.66$ in the bottom panel. The Al I 3944 and 3961 Å lines used to derive Al abundance are highlighted in red. The position of the Ca II H & K lines and H$_\epsilon$ are also indicated.
• Figure 2: Synthetic NLTE spectra showing the shape of the aluminium lines 3944 Å and 3961 Å and their blends as indicated. For the synthetic spectra, the star HD 177095 at $\rm[Fe/H]=-0.74$ serves as an example for the NLTE calculations. Dashed lines indicate a spectrum with no aluminium present, while the red and blue lines show spectra with [Al/Fe]=$-0.5$ and $0.0$, respectively.
• Figure 3: Relation between [Al/Fe] and [Fe/H] for the full stellar sample, shown in LTE in the top panel and in NLTE in the bottom panel. The abundances are derived from the Al I 3961 Å line.
• Figure 4: Top panel: Relation between [Al/Fe] and [Fe/H] for the stellar sample. The elemental abundances derived under NLTE assumptions for the Al I 3961 Å line. The data are colour-coded by high-Al and low-Al, with symbols following the classification of Nissen10. Thick-disc stars are represented by crosses, high-$\alpha$ stars are shown as diamonds, and low-$\alpha$ stars as circles. Bottom panel: Distribution of the stellar sample in the [Mg/Fe]$_{\rm NLTE}$ versus [Fe/H]$_{\rm LTE}$, according to the low-Al and high-Al classification used in this paper, adopting the same symbols as the top panel.
• Figure 5: Relation between $\sqrt{J_R}$ and angular momentum, $L_z$, for the stellar sample. Panel (a): Low-Al stars. Panel (b): High-Al stars. The low-$\alpha$ stars are shown as filled circles, the high-$\alpha$ stars as diamonds, and the thick-disc stars are displayed as crosses in both panels. The black box marks the Feuillet22 selection scheme for the Gaia-Sausage-Enceladus. APOGEE DR17 data are plotted in the background.