Tracing the accretion history of the Milky Way with halo stars

Abstract

The ΛCDM model predicts a hierarchical formation for galaxies. Many substructures were found in the Galactic halo, identified as clumps in kinematic spaces. If they also feature different chemical properties, they are then associated to independent merger debris. Our aim is to explore to what extent different clumps in the energy-angular momentum space (E −L_z) with different metallicity distribution functions (MDF) trace the accretion history of the Galaxy. By means of N-body simulations, we confirm that accreted stars from a massive merger redistribute in a wide range of E and L_z, due to dynamical friction, thus not being associated to a single clump. Since satellite stars with different metallicities can be deposited in different regions of the E −L_z space, a single accretion can manifest with different MDFs. This method can bias the merger tree of the Galaxy towards increasing the number of accretions and decreasing the masses of the progenitor galaxies.

1. Introduction

This work overall aims at shedding light on the formation and evolution of galaxies in the standard cosmological scenario (i.e. Lambda Cold Dark Matter, ΛCDM). The aforementioned model predicts a hierarchical mass assembly for galaxies (e.g., White & Rees 1978), meaning that the massive galaxies that we observe today have grown their mass by merging with other galaxies, starting from the smallest ones. Assuming this framework, the study of the galactic merging history is thus key to unravelling the mechanisms of galaxy formation and evolution.

In particular, in this work we aim to find methods to reconstruct the hierarchical build-up of galaxies by focusing on our own galaxy, the Milky Way (MW), through the kinematic and chemical abundance distributions of its halo stars. Galactic Archaeology, i.e. the reconstruction of the history and formation of the MW components from the observations of today-stellar populations, can now rely on an unprecedented amount of data, driven by the ESA astrometric mission Gaia and the complementary spectroscopic surveys (e.g., APOGEE, GALAH, Gaia-ESO and LAMOST). Understanding how to interpret these new and upcoming data is crucial to unveil the properties of the building blocks that made up our Galaxy.

Galaxies grow their stellar mass both by forming new stars, which represent the in-situ population, and by accreting nearby satellite galaxies, which bring their own stars that form the so-called ex-situ populations. The latter ones provide information on the number and mass of accreted satellites, and thus on the build-up of the MW. Stars belonging to satellite galaxies in the process of being accreted are still observable in the Galactic halo, which then serves as a fossil record of the recent accretion history of the Galaxy, since these satellites are not totally disrupted yet, but rather form stellar streams that are identifiable in the plane of the sky (e.g., the Sagittarius Stream).

The reconstruction of the past accretions of our Galaxy is not trivial, since after a few Gyr the accreted stars associated with those events have lost their spatial coherence and are nowadays completely mixed with the ones formed in situ. Nevertheless, several studies have shown that stars accreted a long time ago, in the first billion years after the formation of our Galaxy, could still be detected as coherent structures in kinematic spaces of integrals of motions, which are supposed to be conserved during merger events, such as orbital energy E and angular momentum w.r.t. the galactic disc L_z (e.g., Helmi & de Zeeuw 2000). These results, however, have been based on a number of assumptions, as discussed in Jean-Baptiste et al. (2017): when these assumptions are removed, accreted stars do not generally conserve their initial clumping in the kinematic spaces, provided that the merger mass ratio is high enough (≥ 1 : 10). In situ stars can also occupy regions of the kinematic spaces where accreted stars are deposited, making the use of these spaces alone to reconstruct the merger history of our Galaxy questionable.

The complementary chemical information can come to aid, and in particular metallicity distribution functions (hereafter MDF) of stars located in different regions of kinematic spaces: clumps featuring different MDFs may be indicative of independent accretion events. The discovery of the merger event with the Sequoia dwarf galaxy (M_* ∼ 5 × 10⁷M_⊙, see Myeong et al. 2019) is an example of this procedure. Sequoia was proposed to have provided the bulk of the high-energy retrograde stars in the stellar halo, while only the null component (L_z ∼ 0) would be related to the Gaia-Sausage-Enceladus (hereafter GSE, M_* ∼ 6 × 10⁸M_⊙, Belokurov et al. 2018; Haywood et al. 2018; Helmi et al. 2018), namely the last significant merger event experienced by the MW ∼10 Gyr ago. This distinction was confirmed by the fact that these two structures have different MDFs: while the GSE MDF peaks at [Fe/H] = −1.3, the Sequoia MDF peaks at [Fe/H] = −1.6, being more metal-poor (Myeong et al. 2019). This kind of study has been applied to other regions of kinematic spaces (see, for example, Naidu et al. 2020; Ruiz-Lara et al. 2022), making use of the same overall approach: once different MDFs have been identified, the average metallicity value for each identified region of the kinematic spaces is calculated and these different average metallicities are associated with different stellar masses of the progenitor satellites by making use of mass-metallicity relations.

The aim of this study is to explore to what extent we can couple kinematic characteristics and metallicities of stars in the Galactic halo to reconstruct the accretion history of the Milky Way (MW). In particular, we want to understand whether different clumps in the E −L_z space with different MDFs should be associated with distinct merger debris.

2. Method

We analysed the outcomes of dissipationless, high resolution N-body simulations, which modelled a Milky Way-type galaxy accreting a satellite with mass ratio 1:10. These simulations are fully self-consistent, since both the satellite and the Milky Way-type galaxy are modelled as a collection of particles which can react to the interaction, get kinematically heated, experience tidal effects and dynamical friction and finally exchange energies and angular momenta.

Moreover, these simulations differ from each other in the initial inclination of the satellite orbital plane with respect to the main galaxy disk, in order to check for the dependence of the results on the orbital parameters.

Finally, we have assigned chemical abundances to the star particles of the simulation a posteriori, with different initial conditions for the abundance gradients for both galaxies, driven by observational evidence in the Milky Way: a gradient along the z axis (i.e., the thin disk more metal-rich and less α-enriched than the thick one) and an additional radial metallicity gradient in the thin disk (i.e., the inner parts more metal-rich than the outskirts). All parameters are reported in Mori et al. (2024).

3. Results

The main results obtained from the analysis (reported in Mori et al. 2024) are summarised as follows. We confirm that stars accreted from massive merger events spread throughout the E-L_z space. In Fig. 1, we show the final distribution of in-situ, accreted and all stars in the simulation (left, middle, and right panel, respectively) in the E-L_z space, colour-coded by mean metallicity. Therefore, accreted stars are not associated with a single region, as E and L_z are not generally conserved due to dynamical friction - at least when a massive merger is considered (≤1:10 mass ratio). This is a crucial mass ratio, as it is the one that GSE was supposed to have and as MW-type galaxies are expected to accrete (Fragkoudi et al. 2020).

Figure 1.

Final distribution in the E − L_z space of in situ (left panel), satellite (middle panel), and total (right panel) star particles in the simulation, colour-coded by the mean metallicity. The scales of the colour bars are different depending on which galaxy is considered, in order to make their specific metallicity patterns more evident. We also recall that in these plots stars on prograde orbits (which rotate as the disc) have negative L_z, while stars on retrograde orbits have positive L_z.

In particular, stars deposited at higher energies have on average different metallicities than those of stars that end up being more gravitationally bound to the MW. Since satellite stars with different metallicities can be found in different regions of the E − L_z space, this implies that a single accretion of ∼1:10 can manifest with different MDFs throughout the E − L_z space. This can be appreciated from Fig. 2, which shows some examples of MDFs of the accreted, in-situ and all stars in the simulation (blue, red, and black histograms, respectively) across the E − L_z space. The accreted (blue) MDFs feature different shapes, mean and peak(s) values, depending on the region considered, even if they all result from the same accretion event. Groups of stars with different E, L_z, and metallicities may be interpreted as originating from different satellite galaxies, but our analysis shows that these interpretations are not physically motivated, since different MDFs do not necessarily trace different accretion events. Tracing the Galactic accretion history through mass-metallicity relations using the MDF mean values of single clumps can then be very misleading. Indeed, coupling kinematic information with the MDFs to reconstruct the accretion history of the MW can bias the reconstructed merger tree towards increasing the number of past accretions and decreasing the masses of the progenitor galaxies, as shown by Table 1, where MDF mean values are reported for the different kinematic regions in the E-L_z space.

Figure 2. MDFs of different regions of the E −L_z, highlighted by the coloured rectangles.

The red, blue, and black histograms show the MW-type galaxy, satellite, and total MDF, respectively. They are all normalised. The solid lines show the mean values.

Table 1.

Mean values of the MDF of all the stars in every cell of the E − L_z space (left panel of Figure 2).

↓ E; Lz →	(−50, −30)	(−30, −10)	(−10, 10)	(10, 30)	(30, 50)	(50, 70)
(0,5)	-1.45	-1.33	-1.27	-1.31	-1.31	-1.38
(-5,0)	-0.28	-0.74	-1.11	-1.12	-1.19	-1.28
(-10,-5)	-0.09	-0.12	-0.69	-1.07	-1.22	-
(-15,-10)	-	-0.08	-0.22	-0.95	-	-
(-20,-15)	-	-	-0.21	-	-	-

When the analysis is restricted to a solar-like volume, we can no longer capture the very metal-poor accreted stars found at very high energies. This implies that the MDF of the GSE may still exhibit biases. Nonetheless, we found trends in the MDFs varying E and L_z values. In particular, we showed that the mean and the peak(s) of the global (black) MDFs are found at lower metallicities the higher the E and L_z. This is a consequence of the increasing importance of the more metal-poor satellite relative number in stellar density, when considering more unbound and retrograde orbits. From the metallicity gradient in the E − L_z space as a function of energy, one can retrieve information about the initial conditions of the radial metallicity gradient in the original satellite galaxy.

4. Conclusions

A number of possible accretion events have been proposed in the literature by making use of this chemo-dynamic approach, and our analysis reinforces previous suggestions that at least some of these accretion events are disputable. Indeed, we have shown how a single massive accretion can produce MDFs whose characteristics change across the E − L_z plane.

Ultimately, overall the merger tree of the MW is still far from being robustly established, and - even more importantly - we still need to establish sound methods of discriminating between accretion scenarios. Although various abundance ratios have already been used to characterise different accretion events (e.g., Matsuno et al. 2019; Monty et al. 2020; Matsuno et al. 2022b, a; Horta et al. 2023; Ceccarelli et al. 2024), our study emphasises the importance of going not only beyond [Fe/H] but also a revision of the kinematic identification of merger debris.

This is the object of an ongoing work (Mori et al. 2025, in prep). Preliminary results show that some of the kinematic clumps known in literature are actually contaminated by stars from both the last significant merger event of the MW (i.e., GSE) and from in-situ kinematically-heated disc stars. This kind of approach will be vital for a more realistic reconstruction of the MW merger history in light of coming data from WEAVE, 4MOST, SDSSV, and other spectroscopic surveys.