A Walk on the Retrograde Side (WRS) project. II. Chemistry to disentangle in situ and accreted components in Thamnos

TL;DR

The study provides the first large-sample, high-resolution chemical characterization of Thamnos within the WRS framework and directly compares it to GSE. It finds that, while Thamnos is generally metal-poor, many higher-metallicity stars show in situ Milky Way chemistry, indicating substantial contamination and a complex chemodynamical mixture; a two-component MDF suggests a faint accreted, metal-poor tail consistent with a true Thamnos progenitor alongside dominant in situ/GSE contamination. GCE modelling supports a metal-poor, rapidly formed Thamnos with strong winds and early star formation, coupled with significant in situ contribution, while GSE-like chemistry is better matched by a relatively more efficient, extended SFH. Overall, the results reveal a heterogeneous retrograde halo structure where chemical tagging is crucial for disentangling accreted and in situ components, with implications for reconstructing the MW's assembly history; forthcoming Gaia DR4 and 4MOST/WEAVE data will further enhance such chemodynamical mapping.

Abstract

We present the results of the first systematic and dedicated high-resolution chemical analysis of the Thamnos substructure, a candidate relic of the process of hierarchical merger of the Milky Way. The analysis was perfomed in comparison with the Gaia-Sausage-Enceladus (GSE) remnant, within the fully self-consistent and homogeneous framework established by the 'A Walk on the Retrograde Side' (WRS) project. We analysed high-resolution and high signal-to-noise ratio spectra obtained with UVES at VLT for 212 red giant branch stars classified as candidate members of Thamnos and GSE, based on selections in the space of the integrals of motion. We derived precise abundances for 16 atomic species. Compared to GSE, stars attributed to the Thamnos substructure are, on average, more metal-poor, yet most of them show higher [X/Fe] abundance ratios in several elements, such as Na, Mg, Al, Ca, Cu, Zn, as well as lower [Eu/Fe]. The majority of candidate Thamnos stars show chemical signatures more consistent with the in situ Milky Way halo rather than a typical low-mass accreted dwarf galaxy. Our findings are further supported by comparisons with tailored galactic chemical evolution models, which fall short in reproducing the observed enhancement in the $α$-elements, but are able to fit the more metal-poor component present in the Thamnos substructure. These results confirm a high level of contamination in the Thamnos substructure from the in situ population and to a lesser degree from GSE, while still leaving room for a genuine accreted population from a small disrupted dwarf galaxy.

Figures (11)

• Figure 1: CMD of stars dynamically associated with GSE and Thamnos (filled green and purple symbols, respectively). In the background a density map of nearby MW halo stars ($d < 2$ kpc) is plotted.
• Figure 2: UVES spectra of four target stars with similar atmospheric parameters and different metallicity. The spectra have been arbitrarily shifted vertically for plotting purposes.
• Figure 3: Distribution of the observed stars in $E$ - $L_{\mathrm{z}}$ - $L_{\perp}$ spaces. Stars associated with Thamnos and GSE by dodd23 are plotted in purple and green, respectively. Stars from ceccarelli2024 are reported in background as grey points. The colour coding of the border of these points reflect the association provided in ceccarelli2024: dark green for GSE, brown for Sequoia, blue for Antaeus, magenta for ED-2, orange for ED-3, and grey for stars that are not associated with any substructure.
• Figure 4: Metallicity distributions of the GSE (top panel) and Thamnos (bottom panel) coming from neutral Fe lines. The median metallicity of the distributions and the standard deviation are also reported in each panel. For GSE we report the MDF derived in this work (green) alongside the stacked MDF using also stars from ceccarelli2024 (dark green). We note that the shift in metallicity we find between the two samples in GSE might be due to a gradient in the progenitor reflected by the position of its stars in the $E$ - $L_{\mathrm{z}}$ plane (see discussion in the text).
• Figure 5: Abundance ratios of the light elements Na and Al for the selected stars corrected for NLTE effects. The colour coding is the same as in Fig. \ref{['fig:op']}. In the lower left corner of each panel, we report typical uncertainties for the abundance ratios. Literature abundances for MW stars are taken from edvardsson93fulbright2000stephens2002gratton03reddy2003reddy06barklem05bensby05bensby14roederer2014, reggiani2017, and ceccarelli2024. Grey points with dark green borders are GSE stars from ceccarelli2024.