The natural emergence of the correlation between H2 and star formation rate surface densities in galaxy simulations

TL;DR

This paper uses high-resolution galaxy simulations with advanced chemistry and radiative transfer models to explore the natural emergence of the correlation between H2 surface density and star formation rate, highlighting the importance of shielding and radiative effects.

Contribution

It introduces and compares two sub-grid models for H2 formation in galaxy simulations, demonstrating their effectiveness and computational efficiency in reproducing observed correlations.

Findings

Both sub-grid models match radiative transfer results well.

H2-based star formation relations are highly sensitive to radiative flux and shielding.

Total gas Kennicutt-Schmidt relation remains relatively unaffected.

Abstract

In this study, we present a suite of high-resolution numerical simulations of an isolated galaxy to test a sub-grid framework to consistently follow the formation and dissociation of H$_2$ with non-equilibrium chemistry. The latter is solved via the package KROME, coupled to the mesh-less hydrodynamic code GIZMO. We include the effect of star formation (SF), modelled with a physically motivated prescription independent of H$_2$, supernova feedback and mass losses from low-mass stars, extragalactic and local stellar radiation, and dust and H$_2$ shielding, to investigate the emergence of the observed correlation between H$_2$ and SF rate surface densities. We present two different sub-grid models and compare them with on-the-fly radiative transfer (RT) calculations, to assess the main differences and limits of the different approaches. We also discuss a sub-grid clumping factor model to enhance the H$_2$ formation, consistent with our SF prescription, which is crucial, at the achieved resolution, to reproduce the correlation with H$_2$. We find that both sub-grid models perform very well relative to the RT simulation, giving comparable results, with moderate differences, but at much lower computational cost. We also find that, while the Kennicutt-Schmidt relation for the total gas is not strongly affected by the different ingredients included in the simulations, the H$_2$-based counterpart is much more sensitive, because of the crucial role played by the dissociating radiative flux and the gas shielding.