2.4. AGN Selection Completeness

Are there significant numbers of luminous AGN that are not found even in the deepest X-ray surveys? This could be the case if there is a population of AGN that is luminous at non-X-ray wavelengths but is X-ray weak, perhaps due to an intrinsic inability to produce X-rays. For example, such AGN could lack accretion-disk coronae, the structures putatively responsible for creating much of the emission above 0.5-1 keV. Such intrinsically X-ray weak AGN appear to be rare, in line with the dogma that X-ray emission is a universal property of AGN, although some may exist (e.g., Brandt, Laor & Wills 2000; Risaliti et al. 2003; Brandt, Schneider & Vignali 2004a; Leighly, Halpern & Jenkins 2004).

Alternatively, strong X-ray absorption could render a luminous AGN X-ray weak, even if it were intrinsically producing X-rays at a nominal level. Such absorption is seen in "Compton-thick" AGN, which comprise 40% or more of AGN in the local universe (e.g., Risaliti, Maiolino & Salvati 1999; Comastri 2004). Compton-thick AGN are absorbed by column densities of N_H >> 1.5 × 10²⁴ cm^-2, so that the optical depth to electron scattering is >> 1 (for comparison, the column density through a person's chest is N_H 10²⁴ cm^-2). Within such thick absorption, direct line-of-sight X-rays are effectively destroyed via the combination of Compton scattering and photoelectric absorption, even at high energies of 10-200 keV. Compton-thick AGN are thus only visible via indirect X-rays which are 50-150 times weaker (e.g., Comastri 2004); these reach an observer via a less-obscured path by "reflecting" off neutral material and "scattering" off ionized material. ¹

Some well-known, nearby Compton-thick AGN are shown in Figure 6 along the left-hand side of the plot. They are plotted at the level of their observable X-ray luminosities; their absorption-corrected luminosities would be 50-150 times higher. Note that NGC 1068, NGC 4945, and the Circinus Galaxy would become undetectable in the 2.0 Ms CDF-N if placed at z 0.1-0.5, while NGC 6240 and Mrk 231 would become undetectable at z 2. It thus appears likely that many Compton-thick AGN remain undetected in even the deepest X-ray surveys to date. Only a small number of Compton-thick AGN have been isolated among the currently detected X-ray sources in deep surveys. However, since Compton-thick AGN have a diversity of complex X-ray spectral shapes and are expected to be X-ray faint, additional ones could be residing among the currently detected sources with limited photon statistics. More refined searches for Compton-thick AGN among the detected X-ray sources are underway, utilizing characteristic X-ray signatures (e.g., strong iron K lines) and new multiwavelength data (e.g., Spitzer measurements of re-radiated infrared "waste heat").

Figure 6. Rest-frame 0.5-8 keV luminosity versus redshift for CDF-N (triangles) and CDF-S (squares) extragalactic sources with spectroscopic redshifts. The dotted curve shows the 2.0 Ms sensitivity limit near the center of the CDF-N. Sources with I = 15-20, I = 20-22, I = 22-23, and I > 23 are shown as violet, blue, green, and red symbols, respectively (symbol sizes also increase with I-band magnitude). Note the systematic progression of I-band magnitudes; the apparent lack of sources with z 1.5-3 and 0.5-8 keV luminosity 1043 erg s-1 is due to spectroscopic incompleteness at faint I-band magnitudes. Characteristic 0.5-8 keV luminosities for some well-known local AGN are shown along the left-hand axis for comparison purposes. Those in light gray suffer from Compton-thick absorption and are shown at the location of their observable X-ray luminosities; their absorption-corrected luminosities would be 50-150 times higher. Note that typical Seyfert 1 galaxies, such as NGC 3783 and NGC 5548, could have been detected to z 5-10 in the Chandra Deep Fields. Note also that many local Compton-thick AGN would remain undetected at z 0.2-1.

If there is indeed a large population of luminous AGN that has eluded detection in the deepest X-ray surveys, this population also appears to have mostly eluded AGN searches at other wavelengths. There are only a few secure AGN in the Chandra Deep Fields, for example, that have not been detected in X-rays (see Section 4 of Bauer et al. 2004). These include an optically selected, narrow-line AGN at z = 2.445 (Hunt et al. 2004), a radio-bright ( 6 mJy at 1.4 GHz) wide-angle-tail source at z 1-2 (e.g., Snellen & Best 2001), and perhaps a couple of AGN selected optically in the COMBO-17 survey (Wolf et al. 2003). ² Sarajedini, Gilliland & Kasm (2003) have also reported some galaxies with optically variable nuclei in the HDF-N that have not been detected in X-rays. While some of these may be nuclear supernovae or statistical outliers, some are plausibly low-luminosity AGN with X-ray fluxes that lie below the current X-ray sensitivity limit.

Another way to assess AGN selection completeness in deep X-ray surveys is to consider "book-keeping" arguments: can the observed sources explain the total CXRB intensity, and can all the observed accretion account for the local density of SMBH? The answer to the first question is "no" according to the 5-10 keV number counts and source-stacking analyses, as described in Section 1.3. Worsley et al. (2004) propose that the sky density of obscured, X-ray undetected AGN may be 2000-3000 deg^-2 or higher. The answer to the second question is plausibly "yes" but with some uncertainty; this is discussed further in Section 3.1.

¹ In some "translucent" cases, where the column density is only a few times 10²⁴ cm^-2, direct "transmission" X-rays from a Compton-thick AGN may become visible above rest-frame energies of 10 keV. Back.

² While the wide-angle-tail radio source is not included in the CDF-N X-ray source catalogs of Alexander et al. (2003b), manual analysis of the Chandra data at the AGN position indicates a likely detection. Back.