Next Contents Previous

2.1. Techniques for Finding Quasars and AGNs

2.1.1. Quasars

Historically, quasars were first discovered (Hazard, Mackey, & Shimmins 1963; Schmidt 1963) via the optical identification of radio sources, a technique that was both effective and efficient because normal stars and galaxies are much weaker sources of radio emission. It was soon realized (Sandage 1965) that the bulk of the quasar population was radio quiet and could be identified through the excess UV (UVX) radiation that quasars demonstrated relative to normal stars 1. We now know that the UVX technique is effective for redshifts up to about 2.2, the point at which Lyalpha emission shifts into the observed B band and quasars begin to lose their characteristic UV excess. We also know now that only about 10% of quasars and AGNs in the early samples were strong radio emitters, or radio-loud objects.

At higher redshifts, different techniques must be used to find quasars. The problem becomes difficult for two reasons: (1) at redshifts around 3, the optical/UV SEDs of quasars are hard to distinguish from stars, and (2) the space density of quasars at z > 3 declines rapidly with increasing redshift.

The slitless-spectrum technique pioneered by Smith (1975) and developed by Osmer & Smith (1976) provided a color-independent method of finding high-redshift quasars through the direct detection of their strong, broad emission lines, in particular Lyalpha, on low-dispersion objective-prism photographs. The technique was then applied to large telescopes through the use of a transmission grating/prism combination (grism) as the dispersing device (Hoag 1976). However, it was also realized that the slitless-spectrum technique was subject to an important selection effect in that it favored the detection of quasars with strong emission lines 2.

Schmidt, Schneider, & Gunn (1986) made an important advance on this problem by using a digital detector with a grism at the Hale 5-m telescope and by developing and applying a numerical selection algorithm for identifying emission-line objects whose properties and efficiencies could be quantified. The effectiveness of their approach is well demonstrated in Figure 2 of their paper, which shows the grism spectra for a variety of high-redshift quasars from their survey.

The advent of rapid plate-scanning machines such as COSMOS and APM enabled the extension to higher redshift of color-based techniques for discovering quasars. Warren et al. (1987) found the first quasar with z = 4 in this way. The machines and multi-color techniques made it possible to use more sophisticated combinations of colors to separate quasars from stars and to provide quantitative estimates of the selection efficiency as a function of redshift and apparent magnitude, which were crucial for determining the luminosity function. Subsequently, the Sloan Digital Sky Survey (SDSS; York et al. 2000) combined the multicolor technique with a dedicated survey telescope and the largest digital camera built until that time to open a new frontier in extragalactic research by undertaking a digital survey of 10,000 deg2 in five filters; the initial results are described in more detail below.

2.1.2. AGNs

We now know that the discovery of AGNs preceded quasars by 20 years (Seyfert 1943), although the connection and understanding was not achieved until the mid-1970s. Seyfert's classic paper described the properties of nearby galaxies with unusually bright nuclei, which also had unusual emission-line spectra, in particular, broad lines and a wide range of ionization. Seyfert galaxies and related AGNs such as LINERs (low-ionization nuclear emission-line regions), which are less luminous than MB = - 23 mag, constitute the bulk of the AGN population. Their most prominent members can be discovered through imaging and spectroscopic surveys following in the footsteps of Seyfert. However, the discovery of lower-luminosity, more elusive members of the class requires much more care, as the work of Ho, Filippenko, & Sargent (1995) has shown. They examined carefully all galaxies within a magnitude-limited survey with high-quality, narrow-slit spectra for evidence of an active nucleus.

Most recently, Heckman (2003) and Hao & Strauss (2003) have demonstrated that careful application of stellar population synthesis modeling to SDSS galaxy spectra can pull out otherwise unrecognizable emission-line and AGN signatures through the careful subtraction of the young stellar and nebular emission population. Their work indicates that the presence of weak AGN activity is much more common than originally thought and is found in the majority of early- and middle-type galaxies.

2.1.3. Radio and X-Ray Techniques

The discovery and identification of quasars and AGNs by radio and X-ray techniques is perhaps the most straightforward of all, because normal stars and galaxies are weak emitters in these wavelengths. One requires sufficient sensitivity to compact sources and positional accuracies of ~ 1" on the sky. Objects in radio and/or X-ray catalogs are then matched to optical catalogs for identifications and follow-up optical spectra with a large telescope are used to confirm the identification and establish the redshift of the object. Radio and hard X-ray sources offer the important advantage that they are not affected by dust obscuration that may occur along the line of sight to the AGN. If spectral information is available in the 1 keV range, then estimates of the column density of any absorbing gas along the line of sight may be made.

Until recently, radio surveys were hampered by the fact that, as mentioned previously, only about 10% of AGNs are radio loud and thus radio surveys included only a small fraction of the total population. However, with the advent of deep, wide-area surveys such as FIRST (Becker, White, & Helfand 1995), important new opportunities have arisen. FIRST, which reaches to milli-Jansky flux limits, is sufficiently sensitive to detect radio-quiet quasars. When used in conjunction with the multi-color imaging data of the SDSS, it has enabled the discovery of new classes of AGNs (e.g., reddened broad absorption-line quasars that are radio sources) and added a new perspective on the issue of dust obscuration.

The combination of FIRST and SDSS data overcomes another problem with earlier radio surveys, namely, the difficulty of achieving effective redshift preselection for candidate objects. The difficulty was that follow-up spectroscopy of a large number of candidates had to be carried out to find high-redshift or rare types of quasars and AGNs. However, the multicolor SDSS data now can be used to pre-sort candidate objects into the desired groups for follow-up work.

X-ray data have been important to the study of quasars and AGNs since their first detections in X-rays because it was realized that the emission likely originated from very close to the central black hole. Indeed, it can be argued that X-ray emission is the defining characteristic of AGNs (e.g., Elvis et al. 1978). However, the point was somewhat moot at the time because of the lack of sensitivity of the original X-ray observatories. Now, following the work with ROSAT and the initial results from Chandra and XMM-Newton, the tables are turned -- the deepest X-ray surveys are picking up objects not previously noted in optical surveys.

2.1.4. Summary

Discovery and survey techniques for quasars and AGNs at X-ray, UV/optical, and radio wavelengths are now sufficiently well developed, quantified, and sensitive that we have the main tools in hand to settle many of the most fundamental observational questions about the evolution of the AGN population. The combinations of multiwavelength data that are now possible add even more opportunities for research on the nature of AGNs.

1 For the record, most stars are UV faint and quasars have relatively flat optical/UV SEDs in nu fnu space. Back.

2 Of course, all observational techniques for discovering quasars and AGNs are subject to selection effects. This has been a long-standing problem in the determination of their luminosity function and its evolution. Nonetheless, it appears that the slitless-spectrum technique indeed discovers the bulk of the high-z population, although it obviously misses objects with weak or no emission lines. Back.

Next Contents Previous