Why do galaxies form stars inefficiently?

Although the evolution of gas in galaxies is relatively rapid, galaxies form stars surprisingly slowly. The characteristic timescale on which typical star-forming galaxies convert their gas into stars is several billion years. Such gas depletion times are orders of magnitude longer than any physical timescale of processes related to star formation. The origin of this global inefficiency of star formation was a long-standing puzzle.

Connection between star formation rates on large scales and gas cycling on small scales

Using the insights from a suite of idealized galaxy simulations, my collaborators and I formulated a physical framework that explains why galaxies form stars inefficiently. In our framework, galactic gas rapidly cycles between dense star-forming regions and diffuse interstellar medium under the influence of stellar feedback (e.g., explosions of supernovae), turbulence, passing spiral arms, and other dynamical processes (see the video below). When gas becomes star-forming, feedback and dynamical processes quickly disperse the gas, rendering it non-star-forming. As a result, during each cycle gas spends only a small fraction of time in the star-forming state and converts only a small fraction of mass into stars. This explains why global star formation is slow despite rapid gas evolution: gas has to go through a large number of cycles to be converted into stars.

This video shows the evolution of 3 representative gas tracers in the interstellar medium (ISM) of a simulated galaxy. Left panels show spatial positions of tracers; right panel shows their positions in the plane of gas density and total, thermal+turbulent, velocity dispersion. Gas rapidly cycles between dense actively star-forming state (below the dotted line) and diffuse non-star-forming ISM (above the dotted line). Gas spends only a small fraction of time in the star-forming state which makes global star formation inefficient.

For details see our paper:
The Physical Origin of Long Gas Depletion Times in Galaxies

How star formation is (self-)regulated in galaxy formation simulations

Our framework explains the counter-intuitive behavior of star formation in galaxy simulations. Galaxy simulations cannot resolve the cites where individual stars form and therefore star formation and stellar feedback are modeled using “subgrid” prescriptions with tunable parameters. The effect of these parameters on the global star formation of simulated galaxies is rather nontrivial. For example, when stellar feedback is efficient, total star formation rate (SFR) becomes insensitive to the star formation efficiency assumed locally in each computational cell. This phenomenon is usually described as “self-regulation” (e.g., Dobbs et al. 2011, Hopkins et al. 2018). In this regime, total SFR depends on the strength of stellar feedback, e.g., the amount of energy and momentum injected by each supernova. When feedback is weak, global SFR becomes dependent on local star formation efficiency (e.g., Agertz & Kravtsov 2015). As the figure below shows, our framework explains all parts of this puzzling behavior both qualitatively and quantitatively.

Comparison of our simulation results (points) with the predictions of our analytical model (lines). Two panels show gas depletion time, i.e. (total gas mass)/(total SFR), and star-forming gas mass fraction, i.e. (star-forming gas mass)/(total gas mass), as a function of star formation efficiency assumed locally. At high efficiencies, global depletion time becomes independent of efficiency but scales almost linearly with the feedback strength. In this regime star formation is self-regulated by stellar feedback. At low efficiencies, global depletion time scales with local star formation efficiency. In this regime star formation is regulated by dynamical processes not directly related to star formation, such as gas compression and expansion in spiral arms.

For details see our paper:
How Galaxies Form Stars: The Connection between Local and Global Star Formation in Galaxy Simulations

What sets the slope of molecular Kennicutt-Schmidt relation?

Our framework also elucidates the origin of the linear correlation between SFR and molecular gas surface densities, also known as the molecular Kennicutt-Schmidt relation. Such linear correlation is observed in star-forming (non-starburst) galaxies when SFR and molecular gas densities are averaged on kiloparsec and larger scales. A mere correlation between SFR and molecular gas is not surprising because they both trace dense gas in the interstellar medium. What is surprising is that this correlation is linear: many popular models of galactic star formation predict a much steeper relation. Our simulation results and theoretical framework show that a linear relation emerges when stellar feedback efficiently regulates the evolution of star-forming and molecular gas.

Thick lines show the molecular Kennicutt-Schmidt relation (KSR) in our isolated galaxy simulations. Different colors correspond to simulations with different slopes of the star formation relation adopted at the resolution scale (shown by dashed lines). Three panels show simulations with different efficiency of stellar feedback: no feedback at all (left), fiducial feedback (middle), and strong feedback due to very high local star formation efficiency (right). In case of no feedback, the slope of molecular KSR becomes sensitive to the value of the local slope. When feedback becomes very efficient, any such dependence disappears and molecular KSR becomes close to linear (i.e., molecular gas depletion time becomes constant). This insensitivity is another manifestation of self-regulation discussed above.

For details see our paper:
What Sets the Slope of the Molecular Kennicutt-Schmidt Relation?