Testing Feedback from Star Clusters in Simulations of the Milky Way Formation

Abstract: We present a suite of galaxy formation simulations that directly model star cluster formation and disruption. Starting from a model previously developed by our group, here we introduce several improvements to the prescriptions for cluster formation and feedback, then test these updates using a large suite of cosmological simulations of Milky Way mass galaxies. We perform a differential analysis with the goal of understanding how each of the updates affects star cluster populations. Two key parameters are the momentum boost of supernova feedback $f_{\mathrm{boost}}$ and star formation efficiency per freefall time $\epsilon_{\mathrm{ff}}$. We find that $f_{\mathrm{boost}}$ has a strong influence on the galactic star formation rate, with higher values leading to less star formation. The efficiency $\epsilon_{\mathrm{ff}}$ does not have a significant impact on the global star formation rate, but dramatically changes cluster properties, with increasing $\epsilon_{\mathrm{ff}}$ leading to a higher maximum cluster mass, shorter age spread of stars within clusters, and higher integrated star formation efficiencies. We also explore the redshift evolution of the observable cluster mass function, finding that most massive clusters have formed at high redshift $z>4$. Extrapolation of cluster disruption to $z=0$ produces good agreement with both the Galactic globular cluster mass function and age-metallicity relation. Our results emphasize the importance of using small-scale properties of galaxies to calibrate subgrid models of star cluster formation and feedback.