The Redshift One LDSS-3 Emission line Survey (ROLES): survey method and z ∼ 1 mass-dependent star formation rate density

Motivated by suggestions of 'cosmic downsizing', in which the dominant contribution to the cosmic star formation rate density (SFRD) proceeds from higher to lower mass galaxies with increasing cosmic time, we describe the design and implementation of the Redshift One LDSS3 Emission line Survey (ROLES). This survey is designed to probe low-mass, z similar to 1 galaxies directly for the first time with spectroscopy. ROLES is a K-selected (22.5 < K-AB < 24.0) survey for dwarf galaxies [8.5 less than or similar to log(M-*/M-circle dot) less than or similar to 9.5] at 0.89 < z < 1.15 drawn from two extremely deep fields [Great Observatories Origins Deep Survey-S (GOODS-S) and MS1054-Faint Infra-Red Extragalactic Survey]. Using the [O ii]lambda 3727 emission line, we obtain redshifts and star formation rates (SFRs) for star-forming galaxies down to a limit of similar to 0.3 M-circle dot yr-1. We present the [O ii] luminosity function measured in ROLES and find a faint-end slope of alpha(faint) similar to -1.5, similar to that measured at z similar to 0.1 in the Sloan Digital Sky Survey. By combining ROLES with higher mass surveys (Gemini Deep Deep Survey and European Southern Observatory GOOD-S public spectroscopy) we measure the SFRD as a function of stellar mass using [O ii] (with and without various empirical corrections) and using spectral energy distribution fitting to obtain the SFR from the rest-frame UV luminosity for galaxies with spectroscopic redshifts. Our best estimate of the corrected [O ii] SFRD and UV SFRD both independently show that the SFRD evolves equally for galaxies of all masses between z similar to 1 and z similar to 0.1. The exact evolution in normalization depends on the indicator used, with the [O ii]-based estimate showing a change of a factor of approximate to 2.6 and the UV-based estimate a factor of approximate to 6. We discuss possible reasons for the discrepancy in normalization between the indicators, but note that the magnitude of this uncertainty is comparable to the discrepancy between indicators seen in other z similar to 1 works. Our result that the shape of the SFRD as a function of stellar mass (and hence the mass range of galaxies dominating the SFRD) does not evolve between z similar to 1 and z similar to 0.1 is robust to the choice of indicator.