In this paper, we ran a suite of isolated dwarf galaxies using the hydrodynamical code SPHGal (Hu et al. 2017). Here I summarise some the main results. For more information, please refer to the papers and references therein.
In this study, we focussed on two main things. The first was varying the star formation efficiency, that is, the fraction of star-forming gas that is turned into stars in a given time step. In hydrodynamical simulations, the value used is typically around 2%. This number is based on observations that star formation seems to be relatively inefficient in turning gas into stars. For simulations where star particles represent populations of stars, this is appropriate as an average value for a star forming region. In our simulations however, we resolve individual massive stars, and therefore we wanted to explore whether this star formation efficiency parameter still gave us results consistent with observations. We used several different observational constraints for this, such as the cluster formation efficiency, cluster mass function, and the size and density properties of the clusters themselves.
I will skip the methodology to keep this brief, and just summarise what we found, but first I will just clarify exactly what is meant by the star formation efficiency and when we form stars... A gas particle becomes star forming once it becomes dense and/or cool enough to be below 8 Jeans Masses. Gas particles below this threshold form stars with the star formation efficiency. With a star formation efficiency of 2%, this means at a given timestep, 2% of gas particles below 8 Jeans masses will be turned into stars. We also have another threshold however, which is at 0.5 Jeans masses. When a gas particle becomes dense and cool enough to be below 0.5 Jeans masses, it is instantaneously turned into a star (with a 100% efficiency).
So with that cleared up, here are the main results:
Lower star formation efficiencies result in higher cluster formation efficiencies
With lower star formation efficiencies, there is less star formation happening in this regime between 8 and 0.5 Jeans masses and a higher fraction happening at higher densities. The densities of star formation as well as the densities at which supernovae explode can be seen in Figure 1. Gas is allowed to collapse more before star formation happens, resulting in more compact and denser systems. When star formation does happen, the initial stellar feedback (photoionsation) is less effective at disrupting the system due to the gas already being so dense. By the time that supernovae go off some Myrs later, the star cluster is already very tightly bound and difficult to disrupt. For this reason, high star formation efficiencies result in many, tightly bound structures which do not disrupt.
Figure 1: (Fig. 6 in the paper) Ambient density distributions of SNe explosions within the first 400 Myr (bold lines)
compared to the star formation densities (faded lines).
Conversely, higher star formation efficiencies result in more star formation at lower densities. The photoionisation and later the supernovae are more effective at disrupting the star forming systems (and the surroundings) due to lower gas densities. For this reason, we see evidence of cluster disruption in these models, but likely because the star clusters are born with very low surface densities to begin with. These low surface densities are lower than expected from observations.
Figure 2: (Fig. 12 in the paper) Average cluster formation efficiency Γ for each model for young
FoF groups (crosses) and young bound clusters (circles) as a function of average star formation rate surface density, ΣSFR.