Infrared luminosity functions from the Chandra Deep Field South : the Spitzer view on the history of dusty star formation at 0<z<1

E. Le Floc'h, C. Papovich, H. Dole, E. Bell, G. Lagache, G. Rieke, E. Egami, P. Perez-Gonzalez, A. Alonso-Herrero, M. Rieke, M. Blaylock, C. Engelbracht, K. Gordon, D. Hines, K. Misselt, J. Morrison, J. Mould

Published 2005-06-20Version 1

We analyze a sample of ~2600 MIPS/Spitzer 24mic sources brighter than ~80muJy and located in the Chandra Deep Field South to characterize the evolution of the comoving infrared (IR) energy density of the Universe up to z~1. Using published ancillary optical data we first obtain a nearly complete redshift determination for the 24mic objects associated with R<24 counterparts at z<1. We find that the 24mic population at 0.5<z<1 is dominated by ``Luminous Infrared Galaxies'' (i.e., 10^11 L_sol < L_IR < 10^12 L_sol), the counterparts of which appear to be also luminous at optical wavelengths and tend to be more massive than the majority of optically-selected galaxies. We finally derive 15mic and total IR luminosity functions (LFs) up to z~1. In agreement with the previous results from ISO and SCUBA and as expected from the MIPS source number counts, we find very strong evolution of the contribution of the IR-selected population with lookback time. Pure evolution in density is firmly excluded by the data, but we find considerable degeneracy between strict evolution in luminosity and a combination of increases in both density and luminosity (L*_IR prop. to (1+z)^{3.2_{-0.2}^{+0.7}}, Phi*_IR prop. to (1+z)^{0.7_{-0.6}^{+0.2}}). Our results imply that the comoving IR energy density of the Universe evolves as (1+z)^(3.9+/-0.4) up to z~1 and that galaxies luminous in the infrared (i.e., L_IR > 10^11 L_IR) are responsible for 70+/-15% of this energy density at z~1. Taking into account the contribution of the UV luminosity evolving as (1+z)^~2.5, we infer that these IR-luminous sources dominate the star-forming activity beyond z~0.7. The uncertainties affecting these conclusions are largely dominated by the errors in the k-corrections used to convert 24mic fluxes into luminosities.