3D Magnetohydrodynamic Simulations of Turbulence in the Interstellar Medium with Stellar Feedback

Abstract

The energy and momentum feedback from Type II Supernovae (SN) and other young stars are known to drive turbulence in the Interstellar Medium (ISM) influencing the conditions that control future star formation rate. Through observations from surveys, it has been confirmed that the turbulent pressure depends on the star formation rate per unit area (ΣSFR). However, we would like to understand what sets the velocity scale of this turbulence, since a low velocity dispersion and high density as well as a high velocity dispersion and low density could result in the same turbulent pressure. Simple theoretical considerations suggest that the space-time correlation of supernova events is related to the velocity dispersion, with a higher effective mass clustering producing a higher velocity dispersion. In this thesis, we present 3D numerical magnetohydrodynamic (MHD) simulations of the ISM using the TIGRESS framework built upon the Athena MHD code to solve the equations to evolve the ISM. We test several modifications to the standard TIGRESS implementation to explore effects on the resulting velocity dispersion. First, we explode SN only when the parent cluster particles exceed a minimum mass. Second, we randomize the positions of the SN explosion relative to the parent cluster particle based on a gaussian distribution. Third, we vary the simulation box size and resolution. For either extreme in the mass threshold and randomized position we observe higher star formation rates but modest increases in some models for velocity dispersion. Increasing box size increases the velocity dispersion. The weak dependence of velocity dispersion on mass of individual cluster particles, combined with the box size dependence, and with visual evidence of particle clustering, all suggest that the effective clustering is on larger scales independent of individual clusters. SN spatial and temporal correlations reveal that most models reach a similar effective cluster mass. The similar effective value of large-scale clustering, of order 10^5M⊙, explains the velocity dispersion results.