Hunting the first black holes.

Deep X-ray surveys have shown that the cosmic X-ray background (XRB) is largely due to the accretion onto supermassive black holes, integrated over the cosmic time. The Röntgensatellit (ROSAT), Chandra and XMM-Newton satellites have resolved more than 80% of the 0.1-10 keV X-ray background into discrete sources. Optical spectroscopic identifications are ca. 90% and 60% complete, for the deepest ROSAT and Chandra/XMM-Newton surveys, respectively, and show that the sources producing the bulk of the X-ray background are a mixture of obscured (type-1) and unobscured (type-2) active galactic nuclei (AGN), as predicted by the XRB population-synthesis models, following the unified AGN scenarios. The characteristic hard spectrum of the XRB can be explained if most of the AGN are heavily absorbed and, in particular, a class of highly luminous type-2 AGN, so called QSO-2s, exist. The Chandra and XMM-Newton satellites have recently detected several examples of QSO-2s. The space density of the X-ray selected AGN, as determined from ROSAT surveys, does not seem to decline as rapidly as that of optically selected QSOs; however, the statistics of the high-redshift samples is still rather poor. The new Chandra and XMM-Newton surveys at significantly fainter fluxes are starting to provide additional constraints here, but the preliminary observed redshift distribution peaks at much lower redshifts (z = 0.5-1) than the predictions based on the ROSAT data. Nevertheless, there is an indication of a decline of the space density of X-ray-selected AGN at redshifts above four.