2.4. Infrared Selection
The use of IR techniques to measure AGN continua started in the 1970s with the advent of the first sensitive IR detectors (Low & Kleinmann 1968). However, the IR "colors" of Seyfert galaxies are only subtly different than those of normal galaxies (Kuraszkiewicz et al. 2003), and the equivalent widths of the IR lines are not sufficient to use as a finding mechanism. Thus, IR color surveys can have a large fraction of "false" AGN, unless great care is taken.
The first large-scale attempt to find AGN in the IR was based on Infrared Astronomical Satellite (IRAS) data. De Grijp, Lub, & Miley (1987) showed that AGN had systematically different 60 µm / 25 µm colors than normal galaxies. An alternative approach (Spinoglio & Malkan 1989) is to obtain optical spectra of every IR-selected galaxy. This is a follow-up of the idea of Huchra & Burg (1992) to obtain optical spectra of every optically-selected galaxy, but it is not really a survey technique. The latest use of the IR to find active galaxies is with the Two Micron All Sky Survey (2MASS; Cutri et al. 2002). In this survey, ~ 60% of the objects with J - K > 2 are found to have the optical properties of AGN. This selection criterion is bootstrapped by using the near-IR colors of known radio and optically-selected AGN (Elvis et al. 1994), and thus will tend to find objects with similar properties. The large space density of these IR-selected objects makes them a major contributor to the AGN population.
One of the main potential advantages of the IR is that, at least for known AGN, the mid-IR (3 - 10 µm) seems to be a "pivot point" in the SED. That is, objects of the same bolometric luminosity have rather similar mid-IR luminosities but very variable radio, optical, UV, and soft X-ray luminosities. Exactly how to use this information in a survey via color-color plots has been investigated by Andreani et al. (2003) and potentially seems to be quite powerful. However, to avoid large numbers of false detections, IR selection requires very broadband, precision IR photometry from 4 - 100 µm. Because of the large ratio of normal to "active" galaxies, there can be "leakage" in the selection criteria, unless photometric errors are small.
It has been demonstrated (de Grijp et al. 1987) that most optically-selected AGN have a "mid-IR excess" and are "warmer" compared to normal galaxies. This also seems to be true for radio-selected AGN. The effect is very subtle, however, with AGN tending to have a "hotter" mid-IR effective dust temperature. However, based on Infrared Space Observatory (ISO) colors of AGN (Haas et al. 2003), it is not clear how complete this technique is, since the mid-IR slopes span a range from 0.9 - 2.2, and the relative brightness of the mid-IR AGN component to the stellar continuum and star-heated dust spans a wide range. At long IR wavelengths, the contribution from normal stars fades, and there is emission from AGN and star formation-heated dust.
There is a long-running discussion in the IR community about how to separate out dusty AGN from starburst galaxies based on optical and IR data (see Veilleux 2002 for an excellent summary of the present situation). In this reviewer's opinion, it is basically impossible without IR spectroscopy. In the mid-IR, the presence and absence of IR emission features associated with polycyclic aromatic hydrocarbons (PAHs) and high-ionization IR lines (Laurent et al. 2000) provide strong AGN diagnostics. However, there have not been any wide-angle IR spectroscopic surveys designed to find active galaxies. This will change with the Spitzer Space Telescope. Because of a lack of data, it is not known if there are mid-IR "quiet" AGN.
Infrared observations of X-ray-selected (Kuraszkiewicz et al. 2003) and optically-selected (Haas et al. 2003) AGN samples show a very wide range of IR to X-ray SEDs and considerable variation in the IR colors of AGN, making the use of IR colors somewhat problematic in an AGN survey. The IR spectral parameters do not correlate with the optical spectral slope, nor with the IR luminosity, nor with the mid-IR-to-far-IR luminosity ratio, nor with inclination-dependent extinction effects in the picture of a dusty torus.