High precision X-ray log N - log S distributions: implications for the obscured AGN population

Context. Our knowledge of the properties of AGN, especially those of optical type-2 objects, is very incomplete. Because extragalactic source count distributions are dependent on the cosmological and statistical properties of AGN, they provide a direct method of investigating the underlying source populations. Aims. We aim to constrain the extragalactic source count distributions over a broad range of X-ray fluxes and in various energy bands to test whether the predictions from X-ray background synthesis models agree with the observational constraints provided by our measurements. Methods. We have used 1129 XMM-Newton observations at vertical bar b vertical bar > 20 degrees. covering a total sky area of 132.3 deg(2) to compile the largest complete samples of X-ray selected objects to date both in the 0.5-1 keV, 1-2 keV, 2-4.5 keV, 4.5-10 keV bands employed in standard XMM-Newton data processing and in the 0.5-2 keV and 2-10 keV energy bands more usually considered in source count studies. Our survey includes in excess of 30 000 sources and spans fluxes from similar to 10(-15) to 10(-12) erg cm(-2) s(-1) below 2 keV and from similar to 10(-14) to 10(-12) erg cm(-2) s(-1) above 2 keV where the bulk of the cosmic X-ray background energy density is produced. Results. The very large sample size we obtained means our results are not limited by cosmic variance or low counting statistics. A break in the source count distributions was detected in all energy bands except the 4.5-10 keV band. We find that an analytical model comprising 2 power-law components cannot adequately describe the curvature seen in the source count distributions. The shape of the log N(> S) -log S is strongly dependent on the energy band with a general steepening apparent as we move to higher energies. This is due to the fact that non-AGN populations, comprised mainly of stars and clusters of galaxies, contribute up to 30% of the source population at energies < 2 keV and at fluxes >= 10(-13) erg cm(-2) s(-1), and these populations of objects have significantly flatter source count distributions than AGN. We find a substantial increase in the relative fraction of hard X-ray sources at higher energies, from >= 55% below 2 keV to = 77% above 2 keV. However, the majority of sources detected above 4.5 keV still have significant flux below 2 keV. Comparison with predictions from the synthesis models suggest that the models might be overpredicting the number of faint absorbed AGN, which would call for fine adjustment of some model parameters such as the obscured to unobscured AGN ratio and/or the distribution of column densities at intermediate obscuration.