Orbital migration and heating history of the Galactic disc: a transition between the bimodal discs

TL;DR

The paper develops a chrono-chemo-dynamical framework to reconstruct the Milky Way's disc evolution over 13 Gyr by jointly fitting metallicity evolution, radial migration, and heating to the observed age-metallicity-kinematics distribution of LAMOST subgiants. It models birth metallicity gradients, angular momentum diffusion, vertical and radial heating, and disc growth non-parametrically via cubic splines and constrains them with a likelihood pipeline validated on mock data. The results show an anisotropic diffusion in action space with RMS ratios $\delta J_R:\delta J_z:\delta L_z \approx 4:1:50$ and reveal a transition in migration efficiency around 8–10 Gyr ago, coinciding with the onset of the low-$\alpha$ disc and likely linked to bar formation and gas-content evolution. The metallicity-gradient history and inside-out growth imply a Sun birth radius of $\sim5.7$ kpc and a high-$\alpha$ old disc that is already radially mixed, while the low-$\alpha$ disc shows three migration regimes consistent with bar-driven and spiral-arm processes. These findings constrain bimodal-disc formation mechanisms and connect to high-redshift bimodal structures and ISM metallicity gradients.

Abstract

Stellar orbits in the Galactic disc evolve from their birth to the current shape through both radial migration and dynamical heating. The history of their secular evolution is imprinted in the current kinematics and age-metallicity distribution. We construct a chrono-chemo-dynamical model of the disc, incorporating inside-out growth, metallicity evolution, radial migration, and heating to fit the observed age-metallicity-kinematics distribution of LAMOST subgiant stars in both the low and high-$α$ disc. By modelling all distribution parameters with spline fitting, we present the first non-parametric stellar migration and heating history of the Galaxy. We determine the heating-to-migration ratio, the ratio of the root-mean-square changes in radial/vertical and azimuthal actions, to be $\approx0.075$ for radial to azimuthal actions and $\approx0.015$ for vertical to azimuthal actions, implying a highly anisotropic diffusion in the action space. Furthermore, we identify a transition in radial migration efficiency coinciding with the transition moment of the bimodal disc, for which the radial migration was more efficient for the high-$α$ disc than for the low-$α$ disc. This transition may be attributed to two correlated scenarios: 1) a bar formation epoch accompanied by violent outward migration, and 2) a drop in the gas mass fraction in the disc when the low-$α$ disc began to form. These findings offer further constraints on the formation mechanisms of bimodal discs, favouring the downsizing scenario. We also briefly discuss the connection between our results and recent high-redshift observations. In addition to the secular evolution history, our model maps the Milky Way ISM metallicity gradient at different lookback times, which we find has only varied a little (in the range of $-0.07~\rm to~-0.10~dex/kpc$) since disc formation.

Figures (15)

• Figure 1: The column-normalised age-metallicity distribution of the LAMOST subgiant stars we used in this work Xiang_Rix2025. The distribution of the age and metallicity are shown individually on the side panels next to the axes.
• Figure 2: Results of the mock sample test of the model. The ground truth is shown in the red dashed line, and the recovered parameters are shown in the blue solid line and the blue shaded region. The black dots are the fitted knots in the cubic spline. The background yellow region denotes $1-\sigma$ of the Gaussian prior of these fitted parameters. Each different panel is for a different free parameter in the model, for which the top left panel is the initial scale length of the disc when the star was forming, top middle panel is for the angular momentum diffusion coefficient, top right panel is for the initial radial metallicity gradient when the star was forming (or the metallicity gradient of the Galactic ISM), bottom panels are the radial and vertical velocity dispersion.
• Figure 3: The parameters of the best fit model. Left: the angular momentum diffusion coefficient for stars at different ages. The results at different guiding radii are coloured differently. Middle: the radial velocity dispersion of the stars at different ages, coloured according to the guiding radius. Right: the vertical velocity dispersion of the stars at different ages, coloured according to the guiding radius. Vertical dotted lines label $\rm age = 9.5~Gyr$, around the transition time of $\sigma_{L_z}$.
• Figure 4: The parameters of the best fit model. Left: The initial metallicity gradient at birth (or ISM metallicity gradient) at different lookback times. The blue line and blue bands are the fitting results within $1\sigma$ uncertainty. The black dot-dashed line is the fitting from Frankel2020 and the black dashed line is the inferred ISM metallicity in Lu2024, obtained using a different method. Right: the initial scale length of the stars at birth. The blue line and blue bands are the fitting results within $1\sigma$ uncertainty. The black dotted line is the model results in Frankel2020 using low-$\alpha$ red clump stars, and the black dashed line is the present scale length of the disc fitted in Funakoshi2025.
• Figure 5: The best fit metallicity evolution model of the Milky Way. The coloured lines show the time evolution of ISM metallicity at different radii. The blue dot labels the Sun, implying that the Sun was born $\sim5.7$ kpc and migrated to its present location.