2023A&A...677A..74D

Context. Metal-poor stars are key for studying the formation and evolution of the Galaxy. Evidence of the early mergers that built up the Galaxy remains in the distributions of abundances, kinematics, and orbital parameters of its stars. Several substructures resulting from these mergers have been tentatively identified in the literature.
Aims. We conducted a global analysis of the chemodynamic properties of metal-poor stars. Our aim is to identify signs of accreted and in situ stars in different regions of the parameter space and investigate their differences and similarities.
Methods. We selected a sample of about 6600 metal-poor stars with [Fe/H] ≤ -0.8 from DR3 of the GALAH survey. We used unsupervised machine learning to separate stars in a parameter space made of two normalised orbital actions, plus [Fe/H] and [Mg/Fe], without additional a priori cuts on stellar properties.
Results. We divided the halo stars into four main groups. All groups exhibit a significant fraction of in situ contamination. Accreted stars of these groups have very similar chemical properties, except the group of stars with very retrograde orbits. This points to at most two main sources of accreted stars in the current sample, the major one related to Gaia-Enceladus and another possibly related to Thamnos and/or Sequoia. Gaia-Enceladus stars are r-process enriched at low metallicities, but a contribution of the s-process appears with increasing metallicity. A flat trend of [Eu/Mg] as a function of [Fe/H] suggests that only core-collapse supernovae contributed to r-process elements in Gaia-Enceladus.
Conclusions. To better characterise accreted stars in the low-metallicity regime, high precision abundances and guidance from chemical evolution models are needed. It is possible that the in situ contamination in samples of accreted stars has been underestimated. This can have important consequences for attempts to estimate the properties of the original systems.