Star Formation Threshold Density

Bryanne McDonough
  • 12 Jun

Hello Dylan and all,

I’ve been trying to confirm the threshold density for star formation for use in a post-processed dust model. In Pillepich et al. (2018) they say that the threshold density is n_H~0.1 cm^-3 and in Vogelsberger et al. (2013) they give it as rho_SFR = 0.13 cm^-3. I assume the latter is also a hydrogen number density, but I would like to avoid making assumptions where possible. If I understand correctly, the threshold density for star formation is not a value that is directly fed into the model, but rather is computed internally as a mass density based on the input supernovae fraction and temperature, star formation timescale, and the cooling function at some reference temperature and density. The cooling function makes it non-trivial to verify the TNG threshold density with independent calculations.

So instead, I looked at the densities of all gas cells with non-zero SFRs in the z=0 snapshot. (I started looking in only a few chunks, but after not finding any near the stated threshold, I verified by looking through all the chunks.) In both TNG100 and 300, the lowest density of any star-forming cell is about 0.00755 in code units, or 5.11e-25 g/cm^3. Assuming a total hydrogen composition gives a number density of 0.305cm^-3, and taking a hydrogen number density (rho*X/m_H) gives 0.232 cm^-3.

Is there any reason why none of the gas cells in the entire z=0 snapshot would be close to the threshold density? Am I misunderstanding something? And is there another way to verify the threshold density used in the simulations? Thanks!

Dylan Nelson
  • 13 Jun

Some debug output of use:

USE_SFR: A0=573   PhysDensThresh=0.000754654 (int units) 0.232138 h^2 cm^-3   expected fraction of cold gas at threshold=0.904362   tcool=4.82656e-10   dens=277.475   egyhot=2104.87
USE_SFR: run-away sets in for dens=0.172583   dynamic range for quiescent star formation=228.692                   USE_SFR: isotherm sheet central density=0.0481551   z0=0.0519156
USE_SFR: SNII energy=272312 [internal units] = 5.41628e+48 [erg/M_sun] = 0.00541628 [1e51 erg/Msun]

where you can find the parameter meanings, and full model description, back in Springel & Hernquist (2003), although I am not sure why the exact value is useful?

Bryanne McDonough
  • 16 Jun

Thank you, this is helpful! As usual, it was little h that caused my confusion. 0.232138 h^2 cm^-3 = 0.1065 cm^-3, or the ~0.1 value given by Pillepich et al. (2018). I assume the difference between that value and the one given for Illustris in Vogelsberger et al. (2013) is mainly due to the updated minimum mass for SNII changing the supernovae mass fraction (i.e., Beta in Springel & Hernquist 2003). Regardless, it is helpful for me to have the value in internal units so I can avoid unit conversions and little h mistakes!

The density threshold comes into the calculation for neutral hydrogen abundance in star-forming cells. And I imagine it would be important in comparing star formation models in different sims.

  • Page 1 of 1