Next Contents Previous

4.2. The source redshift distribution

Most of the sources identified in the CDFN, CDFS and Lockman Hole are found to be at z < 1 (see Fig. 10 of Barger et al. 2002 for the CDFN; Fig. 9 of Tozzi et al. 2001b for the CDFS; Fig. 1 of Mainieri et al. 2002b for the Lockman Hole). Synthesis models of the X-ray background predict that at these faint fluxes the X-ray source redshift distribution is peaked at z ~ 1.3 - 1.5, in contrast with the observed data. A direct comparison between the published distributions and the model predictions is however complicated by the following effects:

1) optical identifications are still far from complete. In general, the still unidentified sources are expected to be at higher redshift than those already observed.

2) ~ 15 - 20% of the sources populating the deep surveys are non-active galaxies at z < 1.

3) given the narrow field of view of the deep surveys (0.1-0.2 deg2) the redshift distribution could be dominated by the presence of large scale structures. This is indeed the case in the CDFS (Gilli et al. 2003), where narrow spikes in the source redshift distribution have been found at z = 0.67 and z = 0.73, and also in the CDFN, where redshift structures have been found at z = 0.84 and z = 1.02 (Barger et al. 2002).

Nevertheless, the spectroscopic completeness can be significantly increased by selecting sources with an X-ray flux well above the survey limit in those subregions with higher optical coverage. Also, if only bright and hard X-ray selected sources are considered, the percentage of non-AGN objects drops dramatically.

In Fig. 4 (top) it is shown the redshift distribution of the 93 sources with f2-10 > 5 × 10-15 erg cm-2 s-1 in the central regions of the CDFS, CDFN, Lockman Hole (Mainieri et al. 2002a), Lynx field (Stern et al. 2002) and SSA13 field (Mushotzky et al. 2000). The spectroscopic completeness of this combined sample is ~ 80%. Sources belonging to the large scale structures observed in the CDFS and CDFN have been excluded. The redshift distribution predicted by model B of GSH01 at the same limiting flux and normalized to 116 objects (93/116 = 80%) is also shown in Fig. 4 (top). An excess of sources at z < 1 with respect to the model predictions is still found in this clean sample.

Figure 4a
Figure 4b

Figure 4. Top: Redshift distribution for the 93 X-ray sources with f2-10 > 5 × 10-15 erg cm-2 s-1 in the central regions of the CDFS, CDFN, Lockman Hole, Lynx field, and SSA13 field (see text; only spectroscopic redshifts are considered). With this selection the achieved spectroscopic completeness is ~ 80%. Sources belonging to large scale structures in the CDFS and CDFN are excluded. The data (shaded area) are compared with the predictions of model B by GSH01 at the same limiting flux (dotted line), normalized to 116 sources (93/116 = 80%). A clear excess of sources is observed at z < 1 with respect to the model predictions.
Bottom: Ratio between AGN with logNH > 22 and logNH < 22 as a function of redshift for the sources with f2-10 > 5 × 10-16 erg cm-2 s-1 detected in the inner regions of the CDFS and CDFN (the spectroscopic completeness of this sample is ~ 60%). Redshift bins have been chosen to contain approximately the same number of sources. The predictions for the Franceschini et al. (2002) model and the Gandhi & Fabian (2002) model at comparable fluxes (f2-10 = 8 × 10-16 and 5 × 10-16 erg cm-2 s-1 , respectively) are also plotted as a solid and dashed line, respectively. The shaded area shows the possible range covered by the data under the two extreme assumptions that the unidentified sources (40% of the sample) are either all obscured or all unobscured (see text). Both models seem to predict too many obscured AGN at z < 1. For comparison the prediction by model B of GSH01 at f2-10 > 5 × 10-16 erg cm-2 s-1 is plotted as a dotted line.

Next Contents Previous