The Chandra Deep Field is the region of the sky with the highest concentration of X-ray data available: 4Ms of Chandra and 3Ms of XMM data, allowing excellent quality spectra to be extracted even for faint sources. We take advantage of this in order
to compile a sample of heavily obscured Active Galactic Nuclei (AGN) using X-ray spectroscopy. We select our sample among the 176 brightest XMM sources, searching for either a) flat X-ray spectra (Photon index<1.4 at the 90% confidence level) suggestive of a reflection dominated continuum or b) an absorption turn-over suggestive of a column density higher than ~10^{24} cm-2. We find a sample of nine candidate heavily obscured sources satisfying the above criteria. Four of these show statistically significant FeKalpha lines with large equivalent widths (three out of four have EW consistent with 1 keV) suggesting that these are the most secure Compton-thick AGN candidates. Two of these sources are transmission dominated while the other two are most probably reflection dominated Compton-thick AGN. Although this sample of four sources is by no means statistically complete, it represents the best example of Compton-thick sources found at moderate-to-high redshift with three sources at z=1.2-1.5 and one source at z=3.7. Using Spitzer and Herschel observations, we estimate with good accuracy the X-ray to mid-IR (12 micron) luminosity ratio of our sources. These are well below the average AGN relation, independently suggesting that these four sources are heavily obscured.
As the mid-IR luminosity represents a good isotropic proxy of the AGN power, a low X-ray to mid-IR luminosity ratio is often claimed to be a reliable indicator for selecting Compton-thick (CT) AGN. We assess the efficiency of this diagnostic by exami
ning the 12mu IRAS AGN sample for which high signal-to-noise XMM observations have been recently become available. We find that the vast majority (10/11) of the AGN that have been classified as CT on the basis the X-ray spectroscopy by Brightman & Nandra present a low Lx/L6 luminosity ratio, i.e. lower than a few percent of the average AGN ratio which is typical of reflection-dominated CT sources. At low Lx/L6 ratios we also find a comparable number of AGN, most of which are heavily absorbed but not CT. This implies that although most Compton-thick AGN present low Lx/L6 ratios, at least in the local, Universe, the opposite is not necessarily true. Next, we extend our analysis to higher redshifts. We perform the same analysis in the CDFS where excellent quality chandra (4 Ms) and xmm (3 Ms) X-ray spectra are available. We derive accurate X-ray luminosities for chandra sources using X-ray spectral fits, as well as 6mu luminosities from SED fits. We find 8 AGN with low Lx/L6 ratios in total, after excluding one source where the 6mu emission primarily comes from star-formation. One of these sources has been already demonstrated to host a CT nucleus, while for another one at a redshift of z=1.22 we argue it is most likely CT on the basis of its combined chandra and xmm spectrum. We find a large number of non CT contaminant with low Lx/L6 ratios. The above suggest that a low Lx/L6 ratio alone cannot ascertain the presence of a CT AGN, albeit the majority of low Lx/L6 AGN are heavily obscured. The two most reliable CT AGN in the high redshift Universe have high Lx/L6 ratios, showing that this method cannot provide complete CT AGN samples.
Spitzer/IRS has revealed many sources with very deep Si features at 9.7micron (tau>1). We set out to investigate whether a strong Si absorption feature is a good indicator for the presence of a heavily obscured AGN. We compile X-ray spectroscopic obs
ervations available in the literature on the optically-thick,tau(9.7)>1 sources from the IRAS Seyfert sample. We find that the majority of the high-tau optically confirmed Seyferts (6/9) in this sample are probably CT. Thus we provide direct evidence for a connection between mid-IR optically-thick galaxies and CT AGN, with the success rate being close to 70% in the local Universe. This is at least comparable, if not better, than other rates obtained with photometric information in the mid to far-IR, or even mid-IR to Xray. However, this technique cannot provide complete CT AGN samples,ie there are many CT AGN which do not show significant Si absorption, with the most notable example being N1068. Having assessed the validity of the high 9.7micron technique locally, we attempt to construct a sample of candidate CT AGN at higher redshifts. We compile a sample of 7 high-tau sources in the GOODS and 5 in the Spitzer FLS. All these have been selected to have no PAH features EW(6.2)<0.3 in order to maximize the probability that they are AGN. 6 out of 7 sources in the GOODS have been detected in X-rays, while for the five FLS sources only X-ray flux upper limits are available. The high X-ray luminosities of the detected GOODS sources corroborates that these are AGN. For FLS, ancillary optical spectroscopy reveals hidden nuclei in two more sources. SED fitting can support the presence of an AGN in the vast majority of sources. We cannot derive useful X-ray spectroscopy constraints on whether these are CT. However, the low LX/L6 ratios, suggest that at least 4 out of the 6 detected sources in GOODS may be associated with CT AGN.
