The most metal-poor tail of the Galactic halo: Hypothesis for its origin from precise spectral analysis

Context . The origin of the Galactic halo is one of the fundamental topics linking the study of galaxy formation and evolution to cosmology. Aims . Our goal is to derive precise and accurate stellar parameters, Mg abundances, and ages for a sample of metal-poor stars with [Fe/H] < − 2 dex from high signal-to-noise and high spectral resolution archival spectra. Methods . We derived effective temperatures from H α profiles using three-dimensional non-local thermodynamic equilibrium (3D NLTE) models, and surface gravities and ages from isochrone fitting based on Gaia data. Iron abundances were derived in one-dimensional (1D) NLTE, while Mg abundances were derived in 1D LTE, 1D NLTE, 3D LTE, and 3D NLTE to show the increasing level of accuracy. Results . The sample stars show a tight trend in the [Mg/Fe] versus [Fe/H] plane with a knee located at [Fe/H] ≈ − 2.8 dex, which indicates a low level of stochasticity at the sampled metallicities in this kind of population. Their location in the Lindblad diagram confirms that they belong to the Galactic halo, but does not show a distinct clustering that might be expected for a merger with a single low-mass galaxy. Comparison with chemical evolution models is also not fully definitive on whether the sample stars were born in situ or in accreted low-mass galaxy mergers. Conclusions . Overall, we find two plausible explanations for the chemical sequence traced by the stars in the [Mg/Fe] versus [Fe/H] plane. One is that the sample stars originated in the already formed Milky Way, which at that time (12.5 Gyr ago) was already the main galaxy of its Local Group surroundings. Another explanation is that the sample stars originated in several small galaxies with similar properties, which later merged with the Galaxy. Only accurate spectroscopic analysis such as that done here can reveal trustworthy chemical diagrams required to observe the traces of the Galaxy evolution. Increasing the sample size and the analysis of other elements are required to differentiate between the two hypotheses.