Constraints on galactic outflows from the metallicity-stellar mass-SFR relation of EAGLE simulation and SDSS galaxies

Abstract

Stellar feedback-driven outflows regulate the stellar formation and chemical enrichment of galaxies, yet the underlying dependence of mass outflow rate on galaxy properties remains largely unknown. We develop a simple yet comprehensive non-equilibrium chemical evolution model~(NE-CEM) to constrain the mass-loading factor $\hat{I}\cdot$ of outflows using the metallicity-stellar mass-SFR relation observed by SDSS at $z=0$. Our NE-CEM predicts the chemical enrichment by explicitly tracking both the histories of star formation and mass-loading. After exploring the EAGLE simulation, we discover a compact yet flexible model that accurately describes the average star formation histories of galaxies. Applying a novel method of chemically measuring $\hat{I}\cdot$ to EAGLE, we find $\hat{I}\cdot$ can be parametrised by its dependence on stellar mass and specific SFR as $\log \hat{I}\cdot \propto M_{\ast }^{\hat{I}}\pm s{\mathrm{SFR}}^2$, with $\hat{I}\pm =-0.12$ and ${\hat{I}}^2=0.32$ in EAGLE. Our chemically-inferred $\hat{I}\cdot$ agrees remarkably well with the kinematic measurements by Mitchell et al. After extensive tests with EAGLE, we apply an NE-CEM Bayesian analysis to the SDSS data, yielding a tight constraint of $\log (\hat{I}\cdot /0.631)=0.731\pm 0.002\times (M_{\ast }/10^{9.5}{M}_{\odot }{)}^{-0.222\pm 0.004}(s\mathrm{SFR}/10^{-9.5}y{r}^{-1}{)}^{0.078\pm 0.003}$, in good agreement with the down-the-barrel measurements. Our best-fitting NE-CEM not only accurately describes the metallicity-stellar mass-SFR relation at $z=0$, but also successfully reproduce the so-called "fundamental metallicity relation'' at higher redshifts. Our results reveal that different galaxies form stars and enrich their gas in a non-equilibrium but strikingly coherent fashion across cosmic time.

22 pages, 17 figures, with key results in Fig. 13, 15, & 16. Comments welcome! Fixed a typo in the ηparameterization and relevant annotations in the figures. Conclusion unchanged