METALLICITY-DEPENDENT QUENCHING OF STAR FORMATION AT HIGH REDSHIFT IN SMALL GALAXIES

The star formation rates (SFRs) of low-metallicity galaxies depend sensitively on the gas metallicity, because metals are crucial to mediating the transition from intermediate-temperature atomic gas to cold molecular gas, a necessary precursor to star formation. We study the impact of this effect on the star formation history of galaxies. We incorporate metallicity-dependent star formation and metal enrichment in a simple model that follows the evolution of a halo main progenitor. Our model shows that including the effect of metallicity leads to suppression of star formation at redshift z > 2 in dark halos with masses less than or similar to 10(11)M(circle dot), with the suppression becoming near total for halos below similar to 10(9.5)-10(10)M(circle dot). We find that at high redshift, until z similar to 2, the SFR cannot catch up with the gas inflow rate (IR), because the SFR is limited by the free-fall time, and because it is suppressed further by a lack of metals in small halos. As a result, in each galaxy the SFR is growing in time faster than the IR, and the integrated cosmic SFR density is rising with time. The suppressed in situ SFR at high-z makes the growth of stellar mass dominated by ex situ SFR, meaning stars formed in lower mass progenitor galaxies and then accreted, which implies that the specific SFR (sSFR) remains constant with time. The intensely accreted gas at high-z is accumulating as an atomic gas reservoir. This provides additional fuel for star formation in 10(10)-10(12)M(circle dot) halos at z similar to 1-3, which allows the SFR to exceed the instantaneous IR, and may enable an even higher outflow rate. At z < 1, following the natural decline in IR with time due to the universal expansion, the SFR and sSFR are expected to drop. We specify the expected dependence of sSFR and metallicity on stellar mass and redshift. At a given z, and below a critical mass, these relations are predicted to be flat and rising, respectively. Our model predictions qualitatively match some of the puzzling features in the observed star formation history.