The HELLAS2XMM survey - IV. Optical identifications and the evolution of the accretion luminosity in the Universe

We present results from the photometric and spectroscopic identification of 122 X-ray sources recently discovered by XMM-Newton in the 2-10 keV band (the HELLAS2XMM 1dF sample). Their flux cover the range 8 X 10(-15)-4 X 10(-13) erg cm(-2) s(-1) and the total area surveyed is 0.9 square degrees. One of the most interesting results (which is found also in deeper sourveys) is that about 20% of the hard X-ray selected sources have an X-ray to optical flux ratio (X/O) ten times or more higher than that of optically selected AGN. Unlike the faint sources found in the ultra-deep Chandra and XMM-Newton surveys, which reach X-ray (and optical) fluxes more than one order of magnitude lower than the HELLAS2XMM survey sources, many of the extreme X/O sources in our sample have R less than or similar to 25 and are therefore accessible to optical spectroscopy. We report the identification of 13 sources with X/O greater than or similar to 10 (to be compared with 9 sources known from the deeper, pencil-beam surveys). Eight of them are narrow line QSO (seemingly the extension to very high luminosity of the type 2 Seyfert galaxies), four are broad line QSO. The results from our survey are also used to make reliable predictions about the luminosity of the sources not yet spectroscopically identified, both in our sample and in deeper Chandra and XMM-Newton samples. We then use a combined sample of 317 hard X-ray selected sources (HELLAS2XMM 1dF, Chandra Deep Field North 1Msec, Chandra SSA13 and XMM-Newton Lockman Hole flux limited samples), 221 with measured redshifts, to evaluate the cosmological evolution of the hard X-ray source's number and luminosity densities. Looking backward in time, the low luminosity sources (log L2-10 keV = 43-44 erg s(-1)) increase in number at a much slower rate than the very high luminosity sources (log L2-10 keV > 44.5 erg s(-1)), reaching a maximum around z = 1 and then levelling off beyond z = 2. This translates into an accretion driven luminosity density which is dominated by sources with log L2-10 keV < 44.5 erg s(-1) up to at least z = 1, while the contribution of the same sources and of those with log L2-10 keV > 44.5 erg s(-1) appear, with yet rather large uncertainties, to be comparable between z = 2 and 4.