Stirring Things Up: Bar-induced substructures in the stellar halo of a cosmological Milky Way analogue

TL;DR

This study addresses how a galactic bar reshapes the stellar halo by creating substructures in the integrals of motion space, specifically $E-L_z$, through bar-driven resonances. Using a high-resolution cosmological zoom-in simulation from the Auriga Superstars suite (Halo 18), the authors compute orbital frequencies via FFT on snipshots, derive $E$, $L_z$, and axisymmetric actions, and identify resonant families such as corotation ($r_{\Omega}=0$) and the 1:1 resonance ($r_{\Omega}=-1$). They find a prominent ridge in $E-L_z$ formed mainly by resonant trapping at CR and the 1:1 resonance, with the retrograde 1:1 component aligning along lines of constant Jacobi energy $E_J=E-\Omega_{bar}L_z$ and slopes set by the bar pattern speed $\Omega_{bar}$; the ridge gathers stars from multiple accreted progenitors, particularly M3 and M4, and exhibits higher metallicity than surrounding halo stars, due to metallicity gradients in the progenitors and resonant scattering. The work demonstrates that internal bar dynamics can generate chemically distinct, dynamically coherent structures in the stellar halo, complicating merger-based attributions in $E-L_z$ and chemical spaces, and it also suggests that the ridge could serve as an independent probe of the bar pattern speed in the Milky Way.

Abstract

The stellar halo of the Milky Way contains the remnants of past accretion events, which could be detectable as substructures in the classical integrals of motion space, such as energy and angular momentum (E-Lz). However, our galaxy also contains a non-axisymmetric stellar bar, which traps stars in resonant orbits, leading to substructures in phase-space. Using a high-resolution magneto-hydrodynamic cosmological zoom-in simulation of a Milky Way analogue, we explore the connection between the bar and the accreted stellar halo. We find that the bar induces prominent substructures, or "ridges", in E-Lz, caused by the resonances. The most pronounced of these is caused by the corotation and the retrograde 1:1 resonances, with weaker ridges visible due to the prograde 1:1 and outer Lindblad resonance. The ridges are present across much of the stellar halo, with variations in radius due to the morphology of different orbital families. We explore the scattering of orbits at the resonances, finding that stars trapped at the 1:1 retrograde resonance become more circularised and have more negative angular momentum. Additionally, stars can move between the corotation and retrograde 1:1 families, thus alternating between prograde and retrograde motion. Due to these scatterings and the pre-existing metallicity gradients in the accreted population, the bar-induced substructures have distinct metallicities compared to stars in the surrounding phase-space. Our results suggest the need for caution when searching the Milky Way stellar halo for accreted substructures in both integral of motions and chemical spaces, since these can be induced by internal perturbations.

Figures (13)

• Figure 1: Logarithmic stellar surface density projection in x-y and x-z of the simulated halo, Auriga 18. We define a solar neighbourhood in this simulation following observations, where the sun is located at x = -8 kpc, 30 degrees behind the bar. The circles around the centre of the galaxy and the sun indicate the two regions studied, each with a radius of 4 kpc.
• Figure 2: This figure shows all stars in the simulation in the $E-L_z$ plane in two regions of the simulation, which are used throughout the paper, a 4 kpc sphere around the centre of the halo (top) and a 4 kpc sphere around a "solar neighbourhood" (bottom) as illustrated in Fig. \ref{['fig:sdens']}. The leftmost panels show the density of stars in $E-L_z$ space with an unsharp mask applied, the panels in the second column show a logarithmic density projection of the same selection of stars. In the panels in the third column, the bins in $E-L_z$ space are coloured by the mean metallicity of stars inside the bin, and in the rightmost panels the bins are coloured by the mean ages of the stars. In all panels a ridge feature stands out starting at $\mathrm{E \approx -1.5 \times 10^5 \; km^2 \; s^{-2}}$ and $\mathrm{L_z \approx -1 \times 10^3 \; kpc \; km \; s^{-1}}$ (also indicated by arrows in the leftmost panels). Stars inside the ridge tend to have lower ages and higher metallicities than surrounding stars in the same region in $E-L_z$ space.
• Figure 3: The logarithmic density distribution of accreted stars in $E-L_z$ space is shown for a central region (top) and a "solar neighbourhood" (bottom), as shown in Fig. \ref{['fig:sdens']}. The plots on the left show these distributions without contours to clearly show the presence of a ridge overdensity, in both spatial regions. Additionally, a line of constant Jacobi energy is plotted in dark red. On the right the same plots are shown with added contours (90th percentile) which show the distribution of stars in three different resonances (corotation - purple, OLR - yellow, 1:1 - red). These colours are used to denote the different resonances throughout the paper. The histogram insets in the two right hand panels show the distribution of stars according to their $\mathrm{r_{\Omega}}$ resonance value; the colours indicate our selection of the resonances, identical to the colours of the contours. We can see that the ridge overdensity is traced mainly by stars in corotation and 1:1 resonance.
• Figure 4: The four panels show stars from the four main accreted progenitors in the simulation coloured by their logarithmic density in $E-L_z$ space. Shown are all stars from these progenitors within a radius of 20 kpc, not just the central or "solar neighbourhood" regions shown previously. Differently coloured contours (90th percentile) show the distribution of resonant stars for all accreted stars (CR: purple, OLR: yellow, 1:1: red). The mergers are in order of their infall times (left to right, top to bottom). Mergers M1 and M2 merged several Gyrs before bar formation, mergers M3 and M4 merged around the same time shortly before the bar is formed. M4 dominates in terms of its contribution in the region where the ridge is found. However, it is important to note that all four mergers are present inside the ridge.
• Figure 5: The face-on (top panels) and edge on (bottom panels) density distribution of all stars within 20 kpc in the retrograde 1:1 resonance (left), corotation resonance (middle) and OLR (right). Overplotted are examples of orbits from each family, in the reference frame rotating with the bar, with colours corresponding to lookback time. In all panels the bar is located along the x-axis. We see that the 1:1 resonant stars are concentrated in the central regions, while the corotation stars have a peak in density at larger radii.