The inferred evolution of the cold gas properties of CANDELS galaxies at 0.5 < z < 3.0

TL;DR

This study estimates cold gas, atomic hydrogen, and molecular gas in 24,000 galaxies from the CANDELS survey across redshifts 0.5 to 3.0, revealing how gas properties evolve with stellar mass and cosmic time using an inversion of star formation laws.

Contribution

It introduces a novel method to infer cold gas content in galaxies over a wide redshift range, extending scaling relations beyond current direct measurement capabilities.

Findings

Cold gas mass correlates with stellar mass at all redshifts.

Molecular fractions increase with stellar mass and look-back time.

Cold gas fraction decreases over time regardless of stellar mass.

Abstract

We derive the total cold gas, atomic hydrogen, and molecular gas masses of approximately 24 000 galaxies covering four decades in stellar mass at redshifts 0.5 < z < 3.0, taken from the CANDELS survey. Our inferences are based on the inversion of a molecular hydrogen based star formation law, coupled with a prescription to separate atomic and molecular gas. We find that: 1) there is an increasing trend between the inferred cold gas (HI and H2), HI, and H2 mass and the stellar mass of galaxies down to stellar masses of 10^8 Msun already in place at z = 3; 2) the molecular fractions of cold gas increase with increasing stellar mass and look-back time; 3) there is hardly any evolution in the mean HI content of galaxies at fixed stellar mass; 4) the cold gas fraction and relative amount of molecular hydrogen in galaxies decrease at a relatively constant rate with time, independent of stellar mass; 5) there is a large population of low-stellar mass galaxies dominated by atomic gas. These galaxies are very gas rich, but only a minor fraction of their gas is molecular; 6) the ratio between star-formation rate (SFR) and inferred total cold gas mass (HI + H2) of galaxies (i.e., star-formation efficiency; SFE) increases with star-formation at fixed stellar masses. Due to its simplicity, the presented approach is valuable to assess the impact of selection biases on small samples of directly-observed gas masses and to extend scaling relations down to stellar mass ranges and redshifts that are currently difficult to probe with direct measurements of gas content.