A spectroscopic measurement of galaxy formation time-scales with the Redshift One LDSS3 Emission line Survey

We present measurements of the specific star formation rate (SSFR)-stellar mass relation for star-forming galaxies. Our deep spectroscopic samples are based on the Redshift One LDSS3 Emission line Survey (ROLES) and European Southern Observatory (ESO) public spectroscopy at z = 1, and on the Sloan Digital Sky Survey (SDSS) at z = 0.1. These data sets cover an equally deep mass range of 8.5 log(M * /M ) 11 at both epochs. We find that the SSFR-mass relation evolves in a way which is remarkably independent of stellar mass, as we previously found for the SFR density (SFRD)-mass relation. However, we see a subtle upturn in SSFR-mass for the lowest mass galaxies (which may at least partly be driven by mass-incompleteness in the K-selected sample). This upturn is suggestive of greater evolution for lower mass galaxies, which may be explained by less massive galaxies forming their stars later and on longer time-scales than higher mass galaxies, as implied by the 'cosmic downsizing' scenario. Parametrizing the e-folding time-scale and formation redshift as simple functions of baryonic mass gives best-fitting parametrizations of (M b ) M -1.01 b and 1 + z f (M b ) M 0.30 b . This subtle upturn is also seen in the SFRD as a function of stellar mass. At higher masses, such as those probed by previous surveys, the evolution in SSFR-mass is almost independent of stellar mass. At higher masses log(M * /M ) > 10 the shapes of the cumulative cosmic SFRDs are very similar at both z = 0.1 and 1.0, both showing 70 per cent of the total SFRD above a mass of log(M * /M ) > 10. Mass functions are constructed for star-forming galaxies and found to evolve by only <35 per cent between z = 1 and 0.1 over the whole mass range. The evolution is such that the mass function decreases with increasing cosmic time, confirming that galaxies are leaving the star-forming sequence/blue cloud. The observational results are extended to z 2 by adding two recent Lyman break galaxy samples, and data at these three epochs (z = 0.1, 1, 2) are compared with the GALFORM semi-analytic model of galaxy formation. GALFORM predicts an overall SFRD as a function of stellar mass in reasonable agreement with the observations. The star formation time-scales inferred from 1/SSFR also give reasonable overall agreement, with the agreement becoming worse at the lowest and highest masses. The models do not reproduce the SSFR upturn seen in our data at low masses, where the effects of extinction and active galactic nuclei feedback should be minimal and the comparison should be most robust.