Dynamic Star Formation

Relevant Papers

On a galactic scale, star formation is known to be a slow process, converting only ~2% of gas mass into stars in the disk dynamical time. Stars form in giant molecular clouds (GMC)---clouds primarily composed of hydrogen molecules with heights comparable to the galactic molecular disk scale height. Whether or not the star formation is also slow on the scales of GMCs and smaller is debated. In a series of papers, we argue that the star formation is a dynamic process on GMC scales: star formation rates accelerate in time.


Using ENZO, I simulated star formation in a self-gravitating cloud (16 x 16 x 16 pc) with continuously driven supersonic turbulence (Mach number = 9).

Top: projected density showing filaments and dense clumps in which star particles (black circles) emerge. See Lee, Chang, & Murray (2015) for FLASH simulations run by a co-author Philip Chang.

We show that the total stellar mass grows quadratically with time and so the star formation rate grows linearly with time. We stress the importance of gas self-gravity; the star formation rate appears constant with time when we consider only gas-on-star and star-on-star gravitational interaction.

Image: Time evolution of the stellar mass fraction for two different unigrid resolutions (256^3 and 512^3). When the star particles dominate the collapse dynamics, we observe the stellar mass to grow quadratically with time.


In Lee, Murray & Rahman 2012, we compiled 280 star forming complexes (SFCs) in the Milky Way. These SFCs are massive star clusters containing at least a single O star. They are identified by large scale (50--100 pc) bubbles seen in GLIMPSE and MSX 8 micron images of extended free-free emission sources detected in the WMAP.

Cross-correlating the SFCs with an all-sky catalog of Milky Way GMCs built by Miville-Deschênes, Murray, & Lee (2017), we find that the star formation rate per free-fall time (SFRff) of GMCs harbouring SFCs span approximately 4 orders of magnitude. GMCs convert as low as 0.01% of their gas mass and as high as 100% of their gas mass into stars over one cloud free-fall time.

We show in Lee, Miville-Deschênes, & Murray (2016) that the variations in cloud properties cannot explain the large scatter. Instead, the large spread in the SFRs of these actively star-forming clouds is an evidence for the time-varying star formation rate.

Top: comparison of the observed and model-inferred SFRff vs. total (gas + stellar) mass of the Milky Way giant molecular clouds. The data (left panel) show 4 orders of magnitude scatter in SFRff. Turbulence-regulated models of star formation (Krumholz & McKee 2005 in the middle panel and Hennebelle & Chabrier 2011 in the right panel) can account for only an order of magnitude scatter in SFRff at best. In these models, only the variations in the cloud velocity dispersion, virial parameter, and the driving scale of turbulence effect any variation in SFRff.