The cosmological properties of AGN in the XMM-Newton Hard Bright Survey

Aims. We investigate here the X-ray luminosity function (XLF) of absorbed (N-H between 4 x 10(21) and 10(24) cm(-2)) and unabsorbed (N-H 4 x 10(21) cm(-2)) AGN, the fraction of absorbed AGN as a function of L-X (and z), the intrinsic N-H distribution of the AGN population, and the XLF of Compton thick (N-H > 10(24) cm(-2)) AGN. Methods. To carry out this investigation, we have used the XMM-Newton Hard Bright Serendipitous Sample (HBSS), a complete sample of bright X-ray sources (f(x) greater than or similar to 7 x 10(-14) erg cm(-2) s(-1)) at high galactic latitude (vertical bar b vertical bar > 20 degrees) selected in the 4.5-7.5 keV energy band. The HBSS sample is now almost completely identified (97% spectroscopic identifications) and it can be safely used for a statistical investigation. The HBSS contains 62 AGN out of which 40 are unabsorbed (or marginally absorbed; N-H < 4 x 10(21) cm(-2)) and 22 are absorbed (N-H between 4 x 10(21) and similar to 10(24) cm(-2)). Results. Absorbed and unabsorbed AGN are characterised by two different XLF with the absorbed AGN population being described by a steeper XLF, if compared with the unabsorbed ones, at all luminosities. The intrinsic fraction F of absorbed AGN (i.e., the fraction of sources with N-H between 4 x 10(21) and 10(24) cm(-2) divided the sources with N-H below 10(24) cm(-2), corrected for the bias due to the photoelectric absorption) with L2-10 (keV) greater than or similar to 3 x 10(42) erg s(-1) is 0.57 +/- 0.11; we find that F decreases with the intrinsic luminosity, and probably, increases with the redshift. Our data are consistent with a flat Log N-H distribution for N-H between 10(20) and 10(24) cm(-2). Finally, by comparing the results obtained here with those obtained using an optically-selected sample of AGN we derive, in an indirect way, the XLF of Compton thick AGN; the latter is well described by a XLF similar, in shape, to that of absorbed AGN, but having a normalization of about a factor of 2 above. The density ratio between Compton thick AGN (N-H >= 10(24) cm(-2)) and Compton thin AGN (N-H <= 10(24) cm(-2)) decreases from 1.08 +/- 0.44 at similar to 10(43) erg s(-1) to 0.57 +/- 0.22 at similar to 10(44) erg s(-1) to 0.23 +/- 0.15 at similar to 10(45) erg s(-1). Conclusions. The results presented here on the anti-correlation between F and -L-x are fully consistent with the hypothesis of a reduction of the covering factor of the gas as a function of the luminosity and are clearly inconsistent with the simplest unified scheme of AGN. These results strongly support the recently proposed radiation-limited clumpy dust torus model although alternative physical models are also consistent with the observations.