The Milky Way stellar halo is twisted and doubly broken: insights from DESI DR2 Milky Way Survey observation

TL;DR

This work uses DESI DR2 halo K giants to map the Milky Way's stellar halo out to ~200 kpc, revealing a twisted, tri-axial structure with two significant density breaks linked to Gaia-Sausage/Enceladus and the Large Magellanic Cloud. By applying forward modeling that corrects for angular and radial incompleteness, the authors fit a triple-power-law density profile to a rotating ellipsoid, finding an inner oblate, disk-aligned halo and an outer prolate, disk-perpendicular halo, with pronounced sky-area–dependent anisotropies including HAC-N/S, VOD, and the LMC wake in Pisces and the northern Galactic cap. They further show that more metal-poor halo stars are more extended and that the halo’s orientation and shape vary with radius, consistent with a complex assembly history in a hierarchical universe. The results underscore the importance of substructure and satellite perturbations in shaping the MW halo and demonstrate the power of DESI’s deep, wide-field spectroscopic data for constraining galaxy formation in the Local Group.

Abstract

Using K giants from the second data release (DR2) of the Dark Energy Spectroscopic Instrument (DESI) Milky Way (MW) Survey, we measure the shape, orientation, radial profile, and density anisotropies of the MW stellar halo over 8 kpc$<r_\mathrm{GC}<200$ kpc. We identify a triaxial stellar halo (axes ratio $10:8:7$), 43 degrees tilted from the disk, showing two break radii at $\sim16$ kpc and $\sim76$ kpc, likely associated with Gaia-Sausage/Enceladus (GSE) and Large Magellanic Cloud (LMC), respectively. The inner stellar halo ($<30$ kpc) is oblate and aligned with the disk, whereas the outer stellar halo becomes prolate and perpendicular to the disk, consistent with the Vast Polar Structure of MW satellites. The twisted halo may arise from the disk-halo angular momentum shift triggered by the infall of a massive satellite. The anisotropic density distribution of the stellar halo is also measured, with successful re-identification of the Hercules-Aquila Cloud South/North (HAC-N/-S) and Virgo overdensities (VOD). Break radii are found at 15/30 kpc for VOD/HAC-N(-S). We identify the LMC transient density wake with a break radius at 60 kpc in the Pisces overdensity region. We also find new observational evidence of the LMC collective density wake, by showing a break radius at $\sim$100 kpc in the northern Galactic cap with a clear density peak at 90 kpc. In the end, we found that more metal-poor halo stars are more radially extended. Our results provide important clues to the assembly and evolution of the MW stellar halo under the standard cosmic structure formation framework.

Figures (15)

• Figure 1: DESI DR2 giants in [Fe/H]-[Mg/Fe] space from the rvj pipeline. The giants are selected from the MAIN-BRIGHT program. Halo stars are defined by [Fe/H] < -0.85 (vertical red dashed line).
• Figure 2: Angular selection function of MAIN-BLUE subsample as a function of sky position. Here, the angular selection function is defined as the completeness fraction of DESI observed stars with respect to the targets. Brighter colors indicate higher completeness. MAIN-RED subsample shares a similar angular selection function with MAIN-BLUE. We do not repeatedly present the angular selection function of MAIN-RED subsample.
• Figure 3: Posterior contours for different combinations of nine parameters of the triple power-law model. Red and blue dashed lines indicate 50th, 16th, and 84th percentiles. The meanings of different model parameters can be found in Section \ref{['sec:parameterized_model']}. The yellow, light blue, and dark blue contours represent the $30\%$, $1\sigma$, and $2\sigma$ regions of the MCMC post-burn distributions, respectively. The black line in the upper right panel shows the selection effect free best-fit model radial density profile (renormalized to unity) with the red shaded region representing the 1-$\sigma$ model uncertainty, and two vertical black dashed lines mark two break radii.
• Figure 4: Best-fit model parameters and uncertainties for the double power-law model (red), the triple power-law model in this work (black), and the triple power-law model (green) in Han_stellar_halo_density_profile. The errorbars represent the 1-$\sigma$ uncertainties of three models. The meanings of different model parameters can be found in Section \ref{['sec:parameterized_model']}. Three models share similar flattening parameters and yaw angle (the difference is within 1-$\sigma$ uncertainty), while they show a greater discrepancy in radial density profile parameters.
• Figure 5: The upper panel shows in green dots the observed stellar density multiplied by $r_q^2$ as a function of flattened radius, $r_{q}$, compared to the best-fit triple power-law model (red solid line) and double power-law model (blue solid line). Errorbars represent the 1-$\sigma$ uncertainties computed from 100 bootstrap subsamples of our K giants. Here, the best-fit triple power-law model in this work (red solid line), that in Han_stellar_halo_density_profile (orange solid line), and the double power-law model (blue solid line) have been convolved with the angular and radial selection functions to have a fair direct comparison with the data (see Section \ref{['sec:selection function']}). Two vertical black dotted lines in both upper and bottom panels represent the two break radii, $r_{b,1}$ and $r_{b,2}$, respectively. Two horizontal dashed gray lines in the bottom panels represent 10 % regions of the model-predicted density over observed density, with the black dashed horizontal line marking $y=1$. Both triple power-law and double power-law models fit the stellar halo within 70 kpc well, but the double power-law model shows a worse match of the outer stellar halo beyond $\sim$ 70 $\mathrm{kpc}$.