Left: The observed N_H distribution (hatched blue) of the sources detected in the deep X-ray survey of the Extended Groth Strip. This is compared with the prediction of the X-ray background population synthesis models (red histogram) adopting a fixed obscured AGN fraction of 80% (parameterised by R=4). Clearly, this model does not agree with the data. Recent survey, including ours, suggest that a luminosity dependent N_H distribution is needed to explain the observations. Right: This figure demonstrates the previous point by showing the obscured AGN fraction observed in different surveys, both shallow (probing luminous QSOs) and ultra-deep (comprising mostly of Seyfert luminosity AGN). It is clear that models adopting a fixed fraction of obscured AGN (parameterized by R in this plot), cannot simultaneously reproduce the observations over the flux range shown here. A variable fraction is required that increases from about ~20% at bright fluxes (luminosities) close to 80% at faint fluxes (luminosities).

The launch of the XMM-Newton and the Chandra X-ray observatories with imaging capabilities at hard energies (2-10keV) has given new impetus in studies of the nature of the sources that make up the diffuse X-Ray Background (XRB). In the pre-Chandra and XMM-Newton era models have been developed that re-constructed the XRB by assuming a mix of obscured and unobscured AGN. The main prediction of these models was a dominant population of heavily obscured systems. This however, has not been confirmed by recent XMM-Newton and Chandra surveys, suggesting that revision of the model assumptions is required (see figure).

These surveys indicate that the fraction of obscured sources (N_H>10²²cm^-2) is not constant, as the earlier versions of the XRB models assume (see figure), but depends on X-ray luminosity: more powerful AGN comprise a smaller fraction of obscured sources. For example wide angle shallow surveys, probing mostly luminous QSOs (see figure; Georgantopoulos et al. 2004; Akylas et al. 2006), estimate an obscured AGN fraction of ~20-30%. On the contrary, ultra-deep pencil-beam surveys, comprising many moderate luminosity AGN, find this fraction to be close to 80% (see figure; Perola et al. 2004; Treister et al. 2005), similar to the value determined in the local Universe.

We have recently completed a deep Chandra survey in the Extended Groth Strip region (Nandra et al. 2005). This consists of 8x200ks Chandra pointings covering a total area of 0.5deg² and is intermediate to ultra-deep and shallow samples, in terms of both area and flux limit. Initial results confirm the rapid change of obscured AGN fraction with luminosity (of flux; see figure). The advantage of our sample is that it probes the flux regime where the rapid change in the obscured AGN fraction occurs, providing the opportunity to explore this in detail. Additionally, the X-ray observations are part of the All-wavelength Extended Groth strip International Survey (AEGIS), which provides multiwavelength data, essential to understand the nature of the sources responsible for the XRB. These include deep optical spectroscopy from the DEEP2, optical photometry from the CFHTLS, near, mid and far-IR data.