Cosmic evolution of the star formation efficiency in Milky Way-like galaxies

TL;DR

This study uses high-resolution cosmological simulations to analyze how the star formation efficiency in Milky Way-like galaxies evolves over cosmic time, revealing a nearly universal efficiency despite changing ISM conditions.

Contribution

It provides the first detailed analysis of the evolution of star formation efficiency and ISM turbulence in a Milky Way-like galaxy using high-resolution zoom-in simulations.

Findings

Star formation efficiency remains around 1% over cosmic time.

ISM velocity dispersion varies significantly during starbursts and mergers.

Theoretical and observational estimates of efficiency differ, affecting model constraints.

Abstract

Current star formation models are based on the local structure of the interstellar medium (ISM), yet the details on how the small-scale physics propagates up to global galactic-scale properties are still under debate. To investigate this we use {\small VINTERGATAN}, a high-resolution (20 pc) cosmological zoom-in simulation of a Milky Way-like galaxy. We study how the velocity dispersion and density structure of the ISM on 50-100 pc scales evolve with redshift, and quantify their impact on the star formation efficiency per free-fall timescale, $\epsilon_{\rm ff}$. During starbursts the ISM can reach velocity dispersions as high as $\sim 50$ km s$^{-1}$ for the densest and coldest gas, most noticeable during the last major merger event ($1.3 < z < 1.5$). After a merger-dominated phase ($1<z<5$), {\small VINTERGATAN} transitions into evolving secularly, with the cold neutral ISM typically featuring velocity dispersion levels of $\sim 10$ km s$^{-1}$. Despite strongly evolving density and turbulence distributions over cosmic time, $\epsilon_{\rm ff}$ at the resolution limit is found to change by only a factor of a few: from median efficiencies of 0.8\% at $z>1$ to 0.3\% at $z<1$. The mass-weighted average shows a universal $\langle \epsilon_{\rm ff} \rangle \approx 1\%$, caused by an almost invariant virial parameter distribution in star forming clouds. Changes in their density and turbulence levels are coupled so the kinetic-to-gravitational energy ratio remains close to constant. Finally, we show that a \textit{theoretically} motivated instantaneous $\epsilon_{\rm ff}$ is intrinsically different to its \textit{observational} estimates adopting tracers of star formation e.g. H$\alpha$. Since the physics underlying star formation can be lost on short ($\sim$ 10 Myr) timescales, caution must be taken when constraining star formation models from observational estimates of $\epsilon_{\rm ff}$.