The advantages of X-ray selection of AGN include
High contrast between the AGN and the stellar light (see Figure 2).
Penetrating power of X-rays. Even column densities of 3 × 1023 cm-2 (corresponding to AV ~ 150 mags) do not reduce the flux at E > 5 keV significantly (see Figure 3).
Great sensitivity of Chandra and XMM-Newton. Sources in the luminosity range 1042 - 1046 ergs s-1 can be detected out to z ~ 3, independent of the nature of the host galaxy (e.g., Steffen et al. 2003).
Accurate positions from Chandra. Unique identifications can be made with counterparts in other wavelength bands.
A relatively large fraction of the bolometric energy (3 - 20%) is radiated in the classical X-ray bands (Ho 1999).
Figure 2. (Left panel) X-ray (ROSAT High-Resolution Imager) and (Right panel) optical (SDSS) images of the nearby Seyfert galaxy NGC4051. Notice in the X-ray image that essentially the only observed emission is due to the AGN, while the optical image is dominated by starlight. This is one of the original objects identified by Seyfert (1943).
In contrast to the optical, where stellar light is a major contributor, or the UV, where light from young massive stars often dominates, or the IR, where dust reradiation from massive stars dominates, or the radio, where emission from HII regions, young supernovae, and other indicators of rapid star formation are often very important, there are very few sources of radiation that can confuse the issue in the hard X-ray band.
Point-like X-ray emission is easy to recognize as being caused by low-luminosity AGN. Using surveys of the low-redshift universe as a guideline, if the total integrated X-ray luminosity of a small (< 2 kpc in size) object is greater than 1042 ergs s-1, then the object is almost certainly an AGN. In the low-redshift universe, there are no galaxies with a total (non-AGN) luminosity exceeding this level. Thus, even without detailed X-ray spectra or imaging, the identification of the nature of the source is clear.
X-rays are also rather penetrating. Column densities corresponding to AV = 5 (NH ~ 1022 atoms cm-2) only reduce fluxes by ~ 3 in the Chandra and XMM-Newton soft X-ray bands. One can see in the 2 - 10 keV X-ray surveys that approximately half of the brightest objects are highly reddened in the optical and often invisible in the UV. At z ~ 10 the absorber has to be Compton-thick (AV ~ 2000!) to "kill" the X-ray flux (see Fig. 3). Thus, there are no dark ages for very high-redshift AGN in the X-ray band caused by the Gunn-Peterson effect. X-rays have a "reverse" Ly forest effect - redshifting reduces the effects of absorption. Thus, for a fixed flux and column density, high-redshift quasars are easier to detect. This effect is similar, but of smaller magnitude, to that seen for the submillimeter sources.
Figure 3. X-ray spectra of two AGN at z ~ 10, one with no absorption, and the other with a line of sight column density of ~ 1024 atoms cm-2 and pure photoelectric absorption. Note that at energies greater than 2 keV (observer's frame), there is no reduction in the source flux. The effects at lower energies are much larger.
From a more physics oriented point of view (Mushotzky, Done, & Pounds 1993), the X-ray emission originates from very close to the central black hole, often shows large amplitude rapid variability, and is characterized by a non-thermal spectrum. Thus, the X-ray properties are directly connected to the black hole nature of the AGN and are not due to reprocessing of the radiation
The fundamental properties of black holes should not be functions of metallicity or environment but only of mass, accretion rate, and black hole spin. Since the X-ray flux originates from very close to the event horizon, the X-ray properties of high-redshift "primordial" black holes should be very similar to that of lower redshift objects. This allows a reasonable calculation of their observable properties at high redshifts (Haiman & Loeb 1999; see also Haiman & Quataert, this volume).