Galaxies in the first two billion years: Earlier formation and quenching with surface-density modulated star formation and feedback

Abstract

JWST has revealed massive, UV-bright, compact, and in some cases quenched galaxies in the first two billion years of cosmic history, challenging current galaxy formation models. We develop an extended L-Galaxies semi-analytic model, run on the MillenniumTNG dark-matter-only simulation, and follow galaxy evolution over $0\le z\lesssim 15$. The model ties the star formation efficiency and the coupling of stellar feedback energy to the local cold gas surface density, making star formation efficient and feedback ineffective in dense gas. We include dissipative gas-rich mergers and disc instabilities in stellar and gaseous discs. All parameters are calibrated simultaneously against stellar mass functions and quenched fractions at $z\simeq 0-4$ and UV luminosity functions at $z\simeq 11-12$ with a new parallelised Markov chain Monte Carlo framework. The model reproduces the stellar mass function up to $z\approx 11$ and increases the abundance of massive and UV-bright galaxies at $z\gtrsim 9$ by up to two orders of magnitude over the legacy version. Although the AGN feedback prescription is unchanged, the enhanced bulge growth and black hole fuelling make massive quenched galaxies at $z\simeq 3-8$ two orders of magnitude more abundant, while dissipative mergers produce compact remnants with UV size-luminosity relations matching observations at $z\simeq 10-13$. At low redshift, galaxy sizes, cold-gas content, metallicities, radial profiles, and morphologies remain consistent with observations. Our results indicate that the tension between models and JWST observations at cosmic dawn can be substantially reduced by regulating star formation and feedback through the local gas surface density, without a modified initial mass function, non-standard cosmology, or new AGN physics.

23 pages main text, 13 figures