The Stellar Chemical Abundances of Simulated Massive Galaxies at $z = 2$

We analyze the stellar abundances of massive galaxies ($\log M_\ast/M_\odot>10.5$) at $z=2$ in the IllustrisTNG simulation with the goal of guiding the interpretation of current and future observations, particularly from the James Webb Space Telescope. We find that the effective size, $R_e$, of galaxies strongly affects the abundance measurements: both [Mg/H] and [Fe/H] are anti-correlated with $R_e$, while the relative abundance [Mg/Fe] slightly increases with $R_e$. The $α$ enhancement as tracked by [Mg/Fe] traces the formation timescale of a galaxy weakly, and mostly depends on $R_e$. Aperture effects are important: measuring the stellar abundances within 1~kpc instead of within $R_e$ can make a large difference. These results are all due to a nearly universal, steeply declining stellar abundance profile that does not scale with galaxy size -- small galaxies appear metal-rich because their stars live in the inner part of the profile where abundances are high. The slope of this profile is mostly set by the gas-phase abundance profile and not substantially modified by stellar age gradients. The gas-phase abundance profile, in turn, is determined by the strong radial dependence of the gas fraction and star formation efficiency. We develop a simple model to describe the chemical enrichment, in which each radial bin of a galaxy is treated as an independent closed-box system. This model reproduces the gas-phase abundance profile of simulated galaxies, but not the detailed distribution of their stellar abundances, for which gas and/or metal transport are likely needed.