Populating H$_2$ and CO in galaxy simulation with dust evolution

TL;DR

This study uses hydrodynamic galaxy simulations with dust evolution to investigate how H$_2$ and CO trace star formation across different metallicities, revealing limitations at low metallicity and conditions for a tight star formation law.

Contribution

It introduces a consistent dust evolution model in galaxy simulations to analyze the relationship between molecular tracers and star formation, especially at low metallicity.

Findings

H$_2$ is a poor star formation tracer at metallicities below 0.4 Z$_\odot$

A tight star formation law emerges at metallicities above 0.8 Z$_\odot$ for both H$_2$ and CO

Higher spatial resolution improves CO abundance estimates, less so for H$_2$ at higher metallicities

Abstract

There are two major theoretical issues for the star formation law (the relation between the surface densities of molecular gas and star formation rate on a galaxy scale): (i) At low metallicity, it is not obvious that star-forming regions are rich in H$_2$ because the H$_2$ formation rate depends on the dust abundance; and (ii) whether or not CO really traces H$_2$ is uncertain, especially at low metallicity. To clarify these issues, we use a hydrodynamic simulation of an isolated disc galaxy with a spatial resolution of a few tens parsecs. The evolution of dust abundance and grain size distribution is treated consistently with the metal enrichment and the physical state of the interstellar medium. We compute the H$_2$ and CO abundances using a subgrid post-processing model based on the dust abundance and the dissociating radiation field calculated in the simulation. We find that when the metallicity is $\lesssim 0.4$ Z$_\odot$ ($t<1$ Gyr), H$_2$ is not a good tracer of star formation rate because H$_2$-rich regions are limited to dense compact regions. At $Z\gtrsim 0.8$ Z$_\odot$, a tight star formation law is established for both H$_2$ and CO. At old ($t \sim 10$ Gyr) ages, we also find that adopting the so-called MRN grain size distribution with an appropriate dust-to-metal ratio over the entire disc gives reasonable estimates for the H$_2$ and CO abundances. For CO, improving the spatial resolution of the simulation is important while the H$_2$ abundance is not sensitive to sub-resolution structures at $Z\gtrsim 0.4$ Z$_\odot$.