Mini-quenching of $z=4-8$ galaxies by bursty star formation

The recent reported discovery of a low-mass $z=5.2$ and an intermediate-mass $z=7.3$ quenched galaxy with JWST/NIRSpec is the first evidence of halted star formation above $z\approx 5$. Here we show how bursty star formation at $z=4-8$ gives rise to temporarily quenched, or mini-quenched galaxies in the mass range $M_{\star} = 10^7-10^9 \ M_{\odot}$ using four models of galaxy formation: the periodic box simulation IllustrisTNG, the zoom-in simulations VELA and FirstLight and an empirical halo model. The main causes for mini-quenching are stellar feedback, lack of gas accretion onto galaxies and galaxy-galaxy interactions. The abundance of (mini-)quenched galaxies agrees across the models: the population first appears below $z\approx 8$, after which their proportion increases with cosmic time, from $\sim 0.5-1.0$% at $z=7$ to $\sim 2-4$% at $z=4$, corresponding to comoving number densities of $\sim 10^{-5}$ Mpc$^{-3}$ and $\sim 10^{-3}$ Mpc$^{-3}$, respectively. These numbers are consistent with star formation rate duty cycles inferred for VELA and FirstLight galaxies. Their star formation histories (SFHs) suggest that mini-quenching at $z=4-8$ is short-lived with a duration of $\sim 20-40$ Myr, which is close to the free-fall timescale of the inner halo. However, mock spectral energy distributions of mini-quenched galaxies in IllustrisTNG and VELA do not match JADES-GS-z7-01-QU photometry, unless their SFHs are artificially altered to be more bursty on timescales of $\sim 40$ Myr. Studying mini-quenched galaxies might aid in calibrating sub-grid models governing galaxy formation, as these may not generate sufficient burstiness at high redshift to explain the SFH inferred for JADES-GS-z7-01-QU.