2012A&A...547A..50Z

Lyman break galaxies (LBGs) are widely thought to be prototypical young galaxies in the early universe, particularly representative of those undergoing massive events of star formation. Therefore, LBGs should produce significant amounts of X-ray emission. We aim to trace the X-ray luminosity of LBGs across cosmic time and from that derive constraints on their star formation history.We utilize the newly released 4Ms mosaic obtained with the Chandra X-ray Observatory, the deepest X-ray image to date, alongside with the superb spectroscopic data sets available in the CDF-S survey region to construct large but nearly uncontaminated samples of LBGs across a wide range of redshift (0.5<z<4.5) which can be used as input samples for stacking experiments. This approach allows us to trace the X-ray emission of LBGs to even lower, previously unreachable, flux density limits (∼10^–18mW/m²) and therefore to larger redshifts.We reliably detect soft-band X-ray emission from all our input redshift bins except for the highest redshift (z∼4) one. From that we derive rest-frame 2-10keV luminosities and infer star formation rates and stellar masses. We find that star formation in LBGs peaks at a redshift of z_peak≃3.5 and then decreases quickly. We also see a characteristic peak in the specific star formation rate (sSFR= SFR/M_*) at this redshift. Furthermore, we calculate the contribution of LBGs to the total cosmic star formation rate density (SFRD) and find that the contribution of LBGs is negligible. Therefore, we conclude that most of the star formation in the early universe takes place in lower luminosity galaxies as suggested by hierarchical structure formation models.