Abstract

We explore the redshift evolution of the specific star formation rate (SSFR) for galaxies of different stellar mass by drawing on a deep 3.6 μm selected sample of >105 galaxies in the 2 deg2 COSMOS field. The average star formation rate (SFR) for subsets of these galaxies is estimated with stacked 1.4 GHz radio continuum emission. We separately consider the total sample and a subset of galaxies that shows evidence for substantive recent star formation in the rest-frame optical spectral energy distributions. At redshifts 0.2 2, at least for high-mass (M* ≳ 4 × 1010 M☉) systems where our conclusions are most robust. Our data show that there is a tight correlation with power-law dependence, SSFR ∝ M*β, between SSFR and stellar mass at all epochs. The relation tends to flatten below M* ≈ 1010 M☉ if quiescent galaxies are included; if they are excluded from the analysis a shallow index βSFG ≈ −0.4 fits the correlation. On average, higher mass objects always have lower SSFRs, also among SF galaxies. At z>1.5 there is tentative evidence for an upper threshold in SSFR that an average galaxy cannot exceed, possibly due to gravitationally limited molecular gas accretion. It is suggested by a flattening of the SSFR–M* relation (also for SF sources), but affects massive (>1010 M☉) galaxies only at the highest redshifts. Since z = 1.5 there thus is no direct evidence that galaxies of higher mass experience a more rapid waning of their SSFR than lower mass SF systems. In this sense, the data rule out any strong "downsizing" in the SSFR. We combine our results with recent measurements of the galaxy (stellar) mass function in order to determine the characteristic mass of an SF galaxy: we find that since z ∼ 3 the majority of all new stars were always formed in galaxies of M* = 1010.6±0.4 M☉. In this sense, too, there is no "downsizing." Finally, our analysis constitutes the most extensive SFR density determination with a single technique out to z = 3. Recent Herschel results are consistent with our results, but rely on far smaller samples.