Star Cluster Formation in Cosmological Simulations

Abstract

Using cosmological hydrodynamic simulations, researchers are making rapid progress in reproducing different types of galaxies and various global scaling relations. One important step has been to recognize the importance of feedback mechanisms that suppress the excess of star formation activities. Although many feedback processes have been explored, the star formation prescription has remained unchanged for over two decades.

To model star formation in a more realistic way, in my thesis, I develop a new implementation in cosmological simulations, continuous cluster formation (CCF) , by considering star clusters as a unit of star formation, inspired by observations that most stars form in clusters. In CCF, a cluster particle grows its mass through gas accretion within a star-forming sphere. The accretion is terminated by its own feedback, thus the final mass is set self-consistently. I also introduce the initial bound fraction, fi, to estimate the mass fraction that is remind bound to the cluster when it emerges from the giant molecular clouds (GMCs). I implement CCF in the Adaptive Refinement Tree code and perform a series of simulations of Milky Way-sized galaxies.

I find that the global star formation history (SFH) of the main galaxy is sensitive to the feedback parameter. Varying the star formation efficiency per free-fall time epsff, on the other hand, has no systematic effect on SFH. However, epsff has a dramatic effect on the properties of modeled star clusters, which can be used to calibrate the star formation and feedback models on a scale that is compatible with the size of GMCs.

I find that the cluster initial mass function is best described by the Schechter function. The cutoff mass scales with the star formation rate of the host galaxies, suggesting that cluster formation depends strongly on galactic environments. I find fi increases strongly with cluster masses, irrespective to the global galactic environment. However, fi is very sensitive to the choice of epsff: the higher the epsff, the larger the fi. This trend also leads to a positive correlation between the maximum cluster mass and epsff. I measure the integrated cluster formation efficiency and find it correlates with star formation rate surface density. Moreover, I find a clear trend that cluster formation timescale is shorter with higher epsff. Future observations of this timescale in the nearby star formation regions can be used as another powerful diagnostic to constrain epsff.

Based on CCF, I implement a new algorithm to model the tidal disruption of clusters along their orbits around the galaxies. I find that various disruption processes significantly changes the shape of the mass function from Schechter-like to log-normal, which suggests young massive clusters formed at high-z are promising candidates of the progenitor of globular clusters (GCs). To better understanding the formation and evolution of GCs, I construct a semi-analytical model onto the halo merger trees in Millennium-II simulations. This model successfully reproduces the observed multi-modal metallicity distribution of GCs in a wide range of host galaxy masses in the Virgo cluster.