Tao Wang (Saclay, France) - Towards a complete census of star formation in massive galaxies at z > 3

Our current understanding on galaxies at z > 3 is largely based on UV-selected samples, mostly Lyman-break galaxies (LBGs). A critical question is then whether we are missing a substantial population of dusty/UV-faint galaxies. In this talk I will address this question by exploring the number density and star formation properties of a sample of z > 3 massive dusty galaxies in the GOODS field. The sample are selected by their extreme red colors in the WFC3/F160W and IRAC bands, including those galaxies with only IRAC detections and no F160W detections (i.e., H-droputs). These galaxies have typical stellar mass M* ~ 10^10.5 Msun, and contribute ~60% to the stellar mass function at M* > 10^10.5 Msun at z = 4-5 while LBGs only contribute 10%. Based on a stacking analysis across Spitzer, Herschel, and radio wavelengths, we show that these galaxies have typical infrared luminosity of LIR ~ 1.3 * 1e12 Lsun albeit not detected at MIPS 24um due to their high redshift. Although the number density is small (0.25 arcmin^-2), they contribute ~ 0.01 Msun yr^{-1} Mpc^{-3} to the cosmic star formation density at z~ 4-5, which was not included in previous estimates based on LBGs. We conclude that these red galaxies are representative of massive galaxy populations at z > 3, and are key to our understanding on the earliest stages of massive galaxy formation. I will also show some preliminary results on how these galaxies lead to a discovery of the most densest structure in the early Universe.