Gas Fraction and Star Formation Efficiency at z < 1.0

After new observations of 39 galaxies at z ~ 0.6-1.0 obtained at the IRAM 30m telescope, we present our full CO line survey covering the redshift range 0.2 < z < 1. Our aim is to determine the driving factors accounting for the steep decline of the star formation rate during this epoch. We study both the gas fraction, defined as M_gas/(M_gas+M_star), and the star formation efficiency (SFE) defined by the ratio between far-infrared luminosity and molecular gas mass (L_FIR/M_H2), i.e. a measure for the inverse of the gas depletion time. The sources are selected to be ultra-luminous infrared galaxies (ULIRGs), with L_FIR larger than 10^12 Lo, experiencing starbursts. Adopting a standard ULIRG CO-to-H2 conversion factor, their molecular gas depletion time is less than 100 Myr. Our full survey has now filled the gap of CO observations in the 0.2<z<1 range covering almost half of the cosmic history. The detection rate in the 0.6 < z < 1 interval is 38% (15 galaxies out of 39), compared to 60% for the 0.2<z<0.6 interval. The average CO luminosity is L'_CO = 1.8 x 10^10 K km/s pc^2, corresponding to an average H2 mass of 1.45 x 10^10 Mo. From the observation of 7 galaxies in both CO(2-1) and CO(4-3), a high gas excitation has been derived; together with the dust mass estimation, this supports the choice of our low ULIRG conversion factor between CO luminosity and H2 for our sample sources. We find that the gas fraction in our sample increases strongly with redshift, and is the most determining factor in the cosmic star formation evolution. The SFE evolution in comparison is shallower, and we conclude that it plays a less significant role in setting the cosmic star formation evolution.