A subgrid model for chemical enrichment in cosmological simulations

Abstract

We present the modules for stellar nucleosynthesis, stellar mass loss, and turbulent diffusion of the new COLIBRE subgrid model for cosmological hydrodynamical simulations of galaxy formation. COLIBRE models the thermal evolution of the multi-phase interstellar medium, dust grains, star formation, and stellar and AGN feedback. This work focuses on the model for chemical enrichment. We track the evolution of 12 chemical elements produced by a broad range of nucleosynthetic channels, including core-collapse supernovae and stellar winds, Type Ia supernovae, and asymptotic giant branch (AGB) stars. Enrichment from $s$- and $r$-process elements is modelled via contributions from AGB stars, neutron star mergers, common envelope supernovae, and collapsars. We present an updated compilation of stellar yields taken from the literature, which we release alongside this work. Small-scale element mixing is implemented through a turbulent diffusion process. While diffusion has only a minimal impact on basic integrated galaxy properties, it does reduce the slope of the gas-phase metallicity-mass relation compared with simulations that do not include it. The distribution of element ratios of individual stellar particles is sensitive to diffusion, but only at low metallicities ($Z \lesssim 10^{-1}\,\rm{Z}_\odot$). The model is tested using redshift $z=0$ results from a set of cosmological simulations, mostly of (25 Mpc)$^3$ volumes, demonstrating generally good agreement with Milky Way stellar abundance trends from the APOGEE survey. The model also reproduces the alpha-element enhancement relations observed in galaxies from SDSS, ATLAS-3D, and the Local Group.

Figures (12)

• Figure 1: Fraction of mass returned to the ISM in a Hubble time per unit stellar mass formed at Solar metallicity (Z$_{\odot}=0.0134$, Asplund09). The figure compares the mass returned by AGB (left panel), CCSN and winds from their progenitors (middle panel), and SNIa (right panel) in the COLIBRE, EAGLE, TNG and Illustris galaxy formation models.
• Figure 2: Element abundance ratios from simulated star particles with Galactocentric radii smaller than 9 kpc and azimutal distances under 2 kpc. The element abundances of these star particles, characterized by a median metallicity of [Fe/H]=-0.23, are compared to observed abundances in the Milky Way. Top panels: Pie charts that show the percentage of each element that was produced by AGB stellar enrichment (light blue), CCSN (dark blue), and SNIa (yellow). Middle and bottom panels: element abundance ratios for Milky Way-mass galaxies from the simulation (L025m6/Default diffusion). Each panel corresponds to a different element: carbon, nitrogen, and magnesium (middle panels), as well as oxygen, silicon, and neon (bottom panels), all plotted as a function of iron over hydrogen. Solid lines represent the simulated median relations, while shaded regions mark the 16-84th, 5-95th, and 1-99th percentiles. The contour plots show weighted Milky Way stellar abundances from the APOGEE DR14 data release.
• Figure 3: $\alpha$-enhancement relations. Top panel: [Mg/Fe] vs. [Fe/H] relation for all stars within Milky Way-like galaxies in the COLIBRE simulations with different box sizes and/or resolutions, as indicated in the legend. The solid colored lines represent the median relation, with shaded regions showing the 16th to 84th percentiles. Black contours correspond to the weighted stellar distribution from the APOGEE survey. Bottom panel: Total stellar [Mg/Fe] from centrals galaxies as a function of the galaxies' stellar mass. As in the top panel, solid lines show the median relations from different cosmological simulations. Grey symbols correspond to observations from Gallazzi21 (stars symbols), and Romero23, including data from massive early-type galaxies from the ATLAS-3D survey (circles) and dwarf galaxies from the Local Group (triangles).
• Figure 4: Analysis of metal-poor stars in the Milky Way. Left panel: [Mg/Fe] vs. [Fe/H] for star particles in Milky Way-like galaxies. Solid lines represent the median values, and the shaded regions show the 16th to 84th percentiles from simulations with different diffusion coefficients: high ($C_{\rm{diffusion}}=0.1$, orange), default ($C_{\rm{diffusion}}=0.01$, light green), low ($C_{\rm{diffusion}}=0.001$, light blue), and no diffusion ($C_{\rm{diffusion}}=0$, dark blue). The diffusion treatment significantly impacts the abundance patterns of metal-poor stars ($[\rm{Fe}/\rm{H}]<{-}1$). Observational data from the Milky Way are taken from Venn04, Yong13, and the Pristine Survey (Venn20Lucchesi22). Right panel: Comparison to the average [Mg/Fe] as a function of [Fe/H] for Milky Way stars. Note that the axis range differs from the left panel. The observational data points show the mean values in different metallicity bins, with error bars corresponding to the 95% confidence intervals. The coloured lines indicate the mean [Mg/Fe] of star particles from simulations. Solid lines correspond to L025m6 simulations, and dashed lines to L025m7 simulations. Metal diffusion mostly impacts the chemical abundances of metal-poor stars.
• Figure 5: Total stellar metallicity (left panel) and total stellar iron-to-hydrogen ratios (right panel) as a function of stellar mass. The data in the panels is shown relative to solar assuming $Z_{\odot}=0.0134$ and $\log_{10}(n_{\rm{Fe}}/n_{\rm{H}})_{\odot}+12=7.5$ (Asplund09). Curves show the median mass-weighted mean metallicities for the L025m6 simulations at $z=0$ with different diffusion coefficients, and the shaded regions show the 16th-84th percentiles. The panels show observations reported by Yates21, Kudritzki16, Zahid17, Panter08, and Gallazzi05 (all renormalized to solar abundances values from Asplund et al. 2009 for consistency).