The Milky Way in context: The formation of galactic discs and chemical sequences from a cosmological perspective

TL;DR

This work uses the Auriga cosmological simulations to study the formation of chemical sequences in Milky Way–mass discs, focusing on the origin of bi-modal $[Fe/H]$–$[Mg/Fe]$ distributions and the role of mergers such as Gaia-Enceladus/Sausage. It finds a broad diversity of chemical morphologies across discs, with bimodality not universally tied to GSE-like mergers; instead, transitions between sequences are driven by modulations in the star formation rate and the geometric/dynamical state of the gas disc. Merger gas dilution from individual events rarely produces the large, long-lived metallicity resets required to explain the Solar neighborhood overlap, and most low-$ ext{α}$ stars form from smooth CGM accretion and outer disc gas, while high-$ ext{α}$ stars originate earlier in a thicker, more turbulent phase. Radial migration plays a secondary role, and the slope of the low-$ ext{α}$ sequence encodes long-term SFR trends and disc thickness, linking chemical evolution to the evolving structure of the gas disc. These results imply that the Milky Way’s chemical dichotomy arises from a combination of CGM fueling, SFR history, and gas-disc geometry, with mergers acting as a secondary trigger rather than the primary cause, a finding with broad implications for interpreting alpha sequences in other disc galaxies.

Abstract

We study the formation of chemical sequences in the stellar disc of Milky Way (MW)-mass galaxies in a full cosmological context with the auriga simulations. We focus on the conditions giving rise to bi-modal $α$-chemistry in the MW disc and the potential influence of mergers (e.g. Gaia-Enceladus, GSE). We find a wide diversity of chemical sequences, without correlation between the emergence of dichotomous $α$-chemistry and GSE-like mergers. The transition between multiple $α$-sequences is sequential, and is mediated by modulations in the star formation rate (SFR). In some cases, this can be caused by the starburst and subsequent quiescence induced by mergers. In others, by exhaustion or violent disruption of the gas disc. Realisations with singular sequences either lack significant modulations in their SFR, or form too late to have a significant high-$α$ sequence. The metallicity overlap between the high-$α$ and low-$α$ sequences (as seen in the Solar neighbourhood of the MW) arises from accretion of metal-poor gas from the circum-galactic medium. This depends on gas disc thickness, with thinner discs losing their metal-poor extremities. Gas donation from singular gas-rich merger events are incapable of driving long-lived metal dilution ($Δ\text{[Fe/H]} \gtrsim 0.3$), and we rule-out this scenario for the low-$α$ sequence in the MW. Finally, the shape of $α$-sequences in the [Fe/H] versus [Mg/Fe] plane is related to long-term SFR trends, with a downward slanted locus (as is observed in the low-$α$ sequence of the MW) owing to a sustained or declining SFR.

Figures (16)

• Figure 1: Examples of chemical sequences in four auriga galaxies (where the simulation name is indicated in the top-left corner of the left-most panels), over a range of radial bins. Stars with halo-like kinematics have been excluded, following the methods described in Section \ref{['sec:methods']}. The coloured 2-D histograms indicates the normalised mass-weighted density of star particles in units of M$_{\odot}\,\rm{dex}^{-1}\,\rm{dex}^{-1}$. Whereas stars in the MW are typically depicted with a logarithmic weighting, we opt for a linear weighting to better emphasise the main overdensities. The histogram colour represents the mean orbital circularity within each pixel, as indicated by the colourbar in the top-right, where red colouration corresponds to thin-disc kinematics. Each panel includes grey 1-D histograms that show the [Mg/Fe] abundance ratio distributions, where we overplot Gaussian profile fits to the high- (blue) and low-$\alpha$ (red) selections. These selections are shown with a black line.
• Figure 2: A measure of the bi-modality strength across the auriga suite, as defined by a modified Bhattacharyya distance in Equation \ref{['equ:B_D']}. Stars are split into a range of radial bins, as given in the left legend. The ordering of simulations along the $x$-axis is determined by their average value of $D_{\rm Bmod}$. The coloured markers along the top $x$-axis denote the type of chemical distribution as judged by eye (see the right legend), with an additional $\mathord{\text{✚}}$ symbol to highlight galaxies that have a stellar halo debris feature (as found in the grey-dashed box of figure 3 in fattahi2019, galaxies that have a stellar halo with a component of velocity anisotropy $\beta>0.8$ and contribution fraction $>0.5$). These features were linked to specific massive mergers with radial infalls orkney2023. There is no strong correlation between $D_{\rm Bmod}$ and the presence of anisotropic halo debris.
• Figure 3: A comparison of four salient properties for the high- and low-$\alpha$ selections described in Section \ref{['sec:bimods']} (for stars in the radial range $0<R_{\rm G}/\rm{kpc}<20$). In order; a) The median azimuthal velocity, b) The standard deviation of the velocity in the $z$-direction, c) the summed stellar mass, d) the median of the stellar age. The values for each auriga simulation are provided by unique markers as indicated in the legend. In black, we show expected values for the MW, where kinematic properties are taken from anguiano2020, and the high/low-$\alpha$ mass ratio value is taken from khoperskov2024.
• Figure 4: Included are a selection of auriga realisations with bi-modal (column 1), uni-modal (column 2) and smeared chemistry (column 3). The entire suite can be seen in Appendix \ref{['AppendixSFH']}. Each panel shows the star formation history in units of $\rm{M}_{\odot}\,\rm{yr}^{-1}$. The blue and red histograms are based on the high-$\alpha$ and low-$\alpha$ selections defined in \ref{['sec:bimods']}. These selections are subdivided into stacked histograms corresponding to distinct radial bands (based on the stellar positions at $z=0$), as described in the left legend. The total overlapping region is highlighted with a green outline, which shows the contribution of stars that formed coevally. A light-grey histogram shows the additional contribution of stars on halo-like orbits. The pie charts on the right-hand-sides of each panel show the total mass ratios within each radial band. The coeval fraction is once again represented with a green line, but here it is calculated per radial band.
• Figure 5: Each row shows the in-situ SFR of Au-18 for stars in different radial regions (as indicated in the top-left corners). The left column is for stars at their $z=0$ positions, whereas the right column is for positions closest to the snapshot that the star was born in (approximately the birth radius). In all panels, the histograms are subdivided into stars of different metallicities, as described in the colourbar. Vertical black dashed lines mark the pericentres of important mergers. Additional plots for the full suite are included https://drive.google.com/drive/folders/1QFg0YvuOLpgyXGf-nUKu0tqos9BnZFSl?usp=sharing.