Introduction - Discovery of Active Galactic Nuclei

By  Bill Keel (Alabama) (edited by Xue-Bing Wu)

Five and a half decades after the recognition of the class of Seyfert galaxies, active galactic nuclei of all kinds still present major puzzles to astrophysicists. They clearly embody some of the most extreme conditions to be found anywhere, and include the most powerful individual objects that we've found throughout the Universe.

The recognition of active galactic nuclei may be conveniently traced to three discoveries, which taught us distinct aspects of the phenomenon as well as how they might be linked.

The class of Seyfert galaxies was first recognized by Carl Seyfert in a 1943 paper, which discussed the set of (mostly spiral) galaxies whose spectra showed unusually broadened emission lines from bright, starlike nuclei. In retrospect, these were hints that large masses might be involved, to produce such high gas velocities without spraying the material right out of the galaxy, and that the phenomenon was concentrated in a small volume (giving the starlike appearance of the nuclei). While Seyfert nuclei had been occasionally observed earlier - in fact, NGC 1068 was among the first few galaxies whose redshift was measured - this was the first definition of a class of similar objects. Seyfert nuclei were divided into two classes (1 and 2, with the panache for which astronomers are so famous) by Ed Khachikian and Daniel Weedman, based on whether all their emission lines had similar amounts of broadening. In the type 1 nuclei, certain emission lines were much broader - the ones that could originate at the highest gas density. In type 2 objects, all lines have broadly similar widths. Both kinds have very similar sets of emission lines seen, implying the simultaneous existence of atoms in states normally associated with a huge range of density, temperature, and incident radiation.

Meanwhile, the opening of new spectral windows beyond visible light showed us other kinds of active nuclei. New developments in radio astronomy started to allow accurate measurements of just where radio sources are in the sky, so that in the 1950s optical astronomers could start asking what visible-light objects (if any) were producing the mysterious radio emission. A common pattern for radio sources away from the Milky Way was to see a pair of radio sources with a galaxy in between - what came to be called radio galaxies. Many of these were normal-looking elliptical galaxies, but a few showed interested, or just plain odd, features. M87 showed a jet of emitting matter shooting thousands of light-years from its core, one which had been discovered in the optical range as early as 1918 by astronomers at Lick Observatory. Centaurus A looked like en elliptical galaxy cut in two by a thick, irregular absorption lane of dust and gas. And Cygnus A, at the then-remarkable redshift of z=0.056, was a small fuzzy image with two main lumps. Colliding galaxy, splitting galaxy, Centaurus A clone seen far away? Speculation ran wild. As radio-astronomical techniques improved, not only could vast numbers of radio galaxies be found, but their structures could be mapped in exquisite detail. Interferometer arrays such as the Westerbork Synthesis Telescope (WSRT) and Very Large Array (VLA) revealed that many showed jets of radio-emitting material tracing from the twin lobes of emission back to a tiny nuclear source; M87 was only the tip of the iceberg. Whatever produced the radio emission had to have a long memory, preserving its direction over millions of years. Optical telescopes showed a variety of spectra for radio galaxies - some looked much like type 1 Seyferts, some type 2's, some showed only a few weak emission lines, and many showed no spectroscopic peculiarities at all - only the combined spectra of normal old stars.

The discovery of quasars has been recounted often. A few strong radio sources stubbornly resisted identification with any obvious visible-light counterpart until positional accuracies from radio observations reached only a few seconds of arc. Some radio sources appeared to be nothing more than galactic stars, but their spectra were very peculiar, with strong, broad emission features at wavelengths that didn't match any plausible features expected from stars - young, old, or exploding. It took some time for Maarten Schmidt at Palomar Observatory to show that these were indeed familiar spectral features, but redshifted to an unprecedented degree. The name quasi-stellar radio source (soon shortened to quasar) was coined for these enigmatic objects. As it turned out, many similar objects are not strong radio sources, and these are distinguished as quasi-stellar objects (QSOs), though both are often lumped together as quasars. Quasars taught us more new aspects of the phenomenon of active nuclei. To be so bright at the large distances implied by their redshifts, they had to be much more luminous than any ordinary galaxies - hundreds of times brighter. Yet they must be tiny, with most of the light coming from a region no larger than our own solar system. This was found from the fact that quasars vary (in both visible light and radio core output) in timescales so short that the object responsible (which cannot be any larger than the distance light travels in this timescale) must by only a light-day or so in size.

A new wrinkle in the AGN picture was added in the late 1970s, with the identification of a few enigmatic objects from variable-star catalogs as highly variable nuclei of distant galaxies. Named after their prototype, such BL Lacertae objects are frustrating for the optical astronomer because their spectra are almost perfectly featureless - the nucleus produces a smooth rainbow of radiation, which can be bright enough to swamp the surrounding galaxies and has no telltale emission or absorption lines to measure its redshift. Redshifts have been measured, either from the surrounding galaxy or by waiting for the object to appear unusually dim so it doesn't drown out the emission lines from surrounding gas. BL Lac objects are most notable for being strongly and rapidly variable at all wavelengths, in both intensity and polarization. Their properties are usually thought to reflect our viewing the jet of a radio galaxy almost along its own axis, so our view is dominated by Doppler-boosted radiation from the jet rather than the more usual view of the nucleus and its surroundings. Some quasars with unusually weak emission lines share some of these variability properties as well, so they and BL Lac objects may be lumped together as blazars.

In the 1980s, it became clear that nuclear activity can extend to lower levels, and appear in more galaxies, than a census of Seyfert nuclei would show. Many galaxies were found to show gaseous emission from their nuclei which could not be explained in detail by young stars, and this was dubbed the class of Low-Ionization Nuclear Emission-line Regions (LINERs) by Tim Heckman and Bruce Balick. There remains debate as to how many kinds of objects fall in this category, but spectroscopy, X-ray and ultraviolet observations show that some of these are indeed lower-power versions of Seyfert nuclei, complete with X-ray source, blue continuum, and variability. The phenomenon of nuclear activity now appears to range over 6 orders of magnitude in luminosity, and to show up in a significant fraction of bright galaxies.

Optical spectra of various kinds of active galactic nuclei

Many of the distinctions among the various flavors of AGN rely on spectroscopic clues, shown here in a montage of optical spectra of some examples. They have all been shifted to their emitted wavelength scales for ease of comparison. Seyfert and radio galaxies come in flavors with all emission lines about the same width (Seyfert 2, narrow-line radio galaxy or NLRG) and with certain emission lines much broader (Seyfert 1, broad-line radio galaxy or BLRG). These pairs are similar in optical spectrum, except that BLRGs may have emission lines that are broader and contain more profile structure than found in Seyfert 1 nuclei. Quasars, represented here by a composite produced from many individual objects, have a family resemblance to Seyfert 1 nuclei, and in most cases, the bumps of Fe II emission are even more prominent in quasars, rippling the spectrum between the string individual lines. BL Lacertae objects have virtually featureless spectra, making even their redshifts difficult to measure unless the surrounding galaxy can be detected, or emission lines show up when the nucleus is temporarily much fainter than usual. At lower activity levels, many galaxies contain nuclear emission regions known as LINERs (Low-Ionization Nuclear Emission-Line Regions), which are in at least some cases a lower-luminosity version of the processes seen in more traditional active nuclei. For example, NGC 4579, shown here, has a very faint Seyfert 1-like broad component to its H-alpha emission, and a modestly bright ultraviolet central source. Finally, a normal galaxy spectrum (of an early-type spiral, NGC 3368) is shown for comparison. Most of its spectrum shows the combined absorption features from the atmospheres of individual stars, with weak emission lines from gas in star-forming regions ionized by hot young stars.

The QSO and BL Lacertae object spectra have good data only in the bluer range, so that they are plotted only from 3500-6000 Angstroms, rather than 3500-7000 as for the other kinds of object.

Variability of Active Galactic Nuclei in various spectral domains

Quasars, BL Lacertae objects, and Seyfert galaxies show prominent variability at many wavelengths, which has been used to provide clues to the sizes and structure of the region producing the radiation. This graph compares three objects showing characteristic kinds of variability. The most complete variation history shown is for BL Lacertae itself from a long-running monitoring program at the University of Michigan Radio Observatory, observed at a frequency of 8 GHz (wavelength 3.8 cm). Averaging all measurements within each five-day interval gives 792 data points over a 28-year period. This shows why the radio behavior of such violently-variable objects is often spoken of as a series of flares, sometimes overlapping in time.

A recent discovery has been that the class of Seyfert nuclei known as narrow-line Seyfert 1 are continually and strongly variable in the X-ray range, as seen in the ROSAT data series of the Seyfert IRAS 13224-3809. These objects were originally defined from their optical spectra, as having the broad lines of type 1 Seyferts but with some of the narrowest widths ever seen for these lines. They may be a special subset of Seyferts, or perhaps ordinary Seyferts seen from a particular direction such as along the pole of a disk-shaped gas structure. For these data, statistical error bars are indicated on each measurement to indicate how significant these huge changes are.

Several Seyfert galaxies have been the subjects of intense spectroscopic campaigns to enable reverberation mapping of their nuclei. This approach uses the notion that any change in the ionizing continuum source will be reflected in the emission-line gas, after a delay that corresponds closely to the extra light-travel time from the nucleus to each gas parcel and then on to us compared to that directly from the nucleus. The data here are from a series of IUE observations of NGC 5548, (DATES) with the continuum measured at three ultraviolet wavelengths compared to three strong emission lines which have very different ionization levels and so might be expected to arise in quite different regions. The continuum is most variable in the deep ultraviolet, as can be seen by comparing the three continuum bands. The Lyman alpha and C IV emission lines follow the continuum changes closely, indicating that they come from areas a few light-days in extent, while Mg II hardly changes at all, coming from a region at least light-months in radius. More sophisticated modelling suggests that the shapes as well as sizes of these regions differ as well.


ASCA - Advanced Satellite for Astrophysics and Cosmology, a Japanese X-ray astronomy satellite optimized for X-ray spectroscopy. It was launched on February 20, 1993.

Blazar - broader term including BL Lacertae objects and those quasars which share their characteristics of unusually weak spectral features, plus strong and rapid variability.

BL Lacertae object - a variety of active galactic nucleus with a nearly featureless spectrum and rapid strong variability. These may be other kinds of radio-loud AGN seen nearly along their jets, so that the Doppler-boosted radiation from the jet overwhelms everything else.

Cerenkov radiation - radiation produced when a particle enters a medium travelling faster than the speed of light in that material. It appears as very short pulses in a cone around the particle's direction of travel. This radiation is used to detect very-high-energy gamma rays, which produce pairs of subatomic particles when they interact with material in the Earth's atmosphere. These particles then produce Cerenkov radiation as they dive deeper into the atmosphere, and it is these flashes that can be detected by large mirror arrays on the ground.

CGRO - Compton Gamma-Ray Observatory, the second of NASA's "Great Observatories". It carries the Energetic Gamma-Ray Experiment Telescope (EGRET), Oriented Scintillation Spectrometer Experiment (OSSE), Compton Telescope (COMPTEL), and Burst and Transient Source Experiment (BATSE). CGRO was deployed in low Earth orbit by the space shuttle Atlantis on April 5, 1991 during STS-37, and remains operational as of 1998.

Doppler boosting - an effect of relativity which enhances the radiation from material that is moving toward us at nearly the speed of light, and hides material moving away from us at such speeds. For example, if there are many quasars with radio jets moving at relativistic speeds, the strongest radio sources will be those that are most nearly pointed in our direction. This is important in understanding BL Lacertae objects and superluminal radio sources.

EGRET - Energetic Gamma-Ray Experiment Telescope on the Compton Gamma-Ray Observatory (CGRO). This instrument has carried out an all-sky survey in many energy bands, along with dedicated pointed observations of particularly interesting or variable objects.

electron volt (eV) - the amount of energy an electron gains when accelerated across a difference in electric potential of one volt. This is a convenient and customary energy unit in high-energy astrophysics. A hydrogen atom can be ionized by absorbing a photon of energy 13.6 eV, or wavelength 912 Angstroms. X-rays are generally taken to have energies from about 0.5-10 thousand eV (kilo-eV or keV), while gamma rays have energies in the million-eV (MeV), billion-eV (giga- or GeV) range or even higher.

EUVE - Extreme Ultraviolet Explorer. This mission conducted the first deep all-sky survey in the extreme ultraviolet region, below the Lyman limit and into the soft X-ray region. This survey detected many active nuclei in directions where foreground absorption from neutral hydrogen in the Milky Way is unusually low. The mission, operated by the Center for EUV Astrophysics in Berkeley, California, has also carried out spectroscopic measurements of the brighter sources.

HST - Hubble Space Telescope. Deployed into low Earth orbit by the space shuttle Atlantis during STS-31 (1990), since refurbished and given instrument upgrades during STS-61 (December 1993) and STS-82 (February 1997). The current complement of science instruments includes the Wide Field Planetary Camera 2 (WFPC2), Faint-Object Camera (FOC), Near-Infrared Camera and Multi-Object Spectrometer (NICMOS) and Space Telescope Imaging Spectrograph (STIS); the Faint-Object Spectrograph (FOS) and Goddard High-Resolution Spectrograph (GHRS) were replaced by these latter two instruments during the 1997 refurbishment mission. Built in cooperation with the European Space Agency (ESA), this is the first of NASA's planned four "Great Observatories".

HUT - Hopkins Ultraviolet Telescope, flown on the Astro-1 and Astro-2 missions of the space shuttle. Designed to measure the spectra of celestial objects in the deep ultraviolet range from 912-1800 Angstroms, especially in the rich spectral region just shortward of 1150 Angstroms where the Hubble Space Telescope's mirror coatings and detectors become ineffective.

ISO - Infrared Space Observatory, a mission of the European Space Agency carrying a cryogenically cooled 60 cm telescope and suite of deep-infrared cameras and spectrometer. It operated from November 1996 to April 1998, when the liquid-helium coolant was exhausted (almost a year after the primary mission specification).

IUE - International Ultraviolet Explorer, a satellite observatory sponsored jointly by NASA, the UK Science Research Council, and the European Space Agency, and launched into geosynchronous orbit in 1978. Designed strictly to measure the ultraviolet spectra of celestial objects, it operated until 1996. One of the most successful of space astronomy, its interactive operation introduced a generation of astronomy to space-based measurements.

LINER - Low-Ionization Nuclear Emission-line Region, gaseous regions common in the centers of many kinds of galaxies. Some of these have been shown to be low-luminosity active galactic nuclei, perhaps an extension of Seyfert activity to the lowest levels and implying that the whole phenomenon of nuclear activity occurs in a significant fraction of bright galaxies.

MERLIN - Multi-Element Radio-Linked Interferometer, a dedicated network of radio telescopes in the United Kingdom for long-baseline interferometry. While there are fewer elements than the VLA so the level of detail is typically smaller, its resolution is substantially greater than that of the VLA because of the long baselines; it has been used to bridge the gap between Westerbork or VLA measurements and maps using intercontinental baselines.

Quasar - from QUAsiStellar Radio Source, an object at large redshift (z>0.1) showing strong broad emission lines. Variability shows that the energy must arise in a tiny region, although some quasars have hundreds of time the energy output of normal galaxies. Their radio structures often include jets and lobes similar to what we see from radio galaxies.

Quasistellar object (QSO) - an object with optical properties as described for quasars, but not necessarily a strong radio source. Only about 10% of QSOs are radio-loud. "Quasar" is often used more loosely to include QSOs.

radio galaxy - a galaxy showing unusually strong radio emission, too intense to be produced by the normal processes of starbirth and stardeath. This may come only from the nucleus, or from a pair of more or less symmetric lobes stretching as far as a million light-years. Many show emission from jets connecting the nucleus to these lobes. Optical spectra of radio galaxies may show nothing unusual, but in many instances show strong emission lines, either narrow (NLRG, like type 2 Seyferts) or incl uding broad lines of certain species (BLRG, like quasars and type 1 Seyferts).

ROSAT - ROentgen SATellit, a Germany/UK/USA satellite dedicated to X-ray astronomy. It as launched June 1, 1990 on Delta II booster and remains in service as of 1998. It carries three instruments, one of which was used in a dedicated two-year survey of the entire celestial sphere. A second, the HRI (High-Resolution Imager), delivered the most detailed X-ray images of deep-sky sources to date.

Seyfert galaxy - galaxy, usually a spiral or disturbed system, whose nucleus shows strong emission lines which are too broad and of ionization too high to be produced by the galaxy's stellar population. Often, we see a bright starlike nucleus associated with this. In type 1 Seyferts, some of the emission lines, those that an be produced at high densities, are still broader, while in type 2 nuclei, all the linewidths are comparable. Seyfert nuclei are strong X-ray sources, and many show significant radio emission.

superluminal sources - radio sources which show internal motions (for example, increasing separation between the core and a knot in the jet) which appears faster than the speed of light in our frame of reference. The data are consistent with this being a transformation effect from seeing jets moving almost directly toward us, so that the emitting material almost catches up with its own radiation. This has the effect of compressing the scale of time that we measure for it, and so increasing the observed speed.

UIT - Ultraviolet Imaging Telescope, flown on the Astro-1 and Astro-2 missions of the space shuttle. Used an image intensifier to photograph numerous celestial objects at wavelengths 1200-2800 Angstroms, all of which are absorbed by the Earth's atmosphere.

unified scheme - one of two pictures in which different kinds of active galactic nuclei have been suspected to be the same, only viewed from different directions relative to their axis. For Seyfert galaxies, the notion is that some Seyfert 2 nuclei are actually Seyfert 1 objects with our view of the innermost region blocked by a thick disk or torus of dust and gas. For BL Lacertae objects, quasars, and radio galaxies, the scheme is that all are quite similar - some radio galaxies are quasars with our view of the central engine blocked, and BL Lacertae objects are the ones in which we happen to be looking right down the jet. These schemes have broad support from several kinds of evidence, but some nagging details mean that at least some parts need to be added to the theories.

VLA - Very Large Array, a radio telescope consisting of 27 linked antennas. Operated by the U.S. National Radio Astronomy Observatory, and located about 100 km west of Socorro, New Mexico.

VLBA - Very Long Baseline Array, a deicated array of radio telescopes for very-long-baseline interferometry (VLBI) operated by NRAO. It stretches from the island of Hawaii to the U.S. Virgin Islands, and provides much more extensive opportunities for VLBI measurements than ad hoc collections of miscellaneous antennas.

VLBI - Very Long Baseline Interferometry. A technique for combining signals received by physically separated radio telescopes to yield some of the results of observations with a single dish as large as the separation between dishes. This can yield resolutions of order one thousandth of an arcsecond (milliarcsecond), and even better when one of the antennas involved is in space and thus yields baselines larger than the Earth's diameter.

The Electromagnetic Spectrum

Perhaps more than any other kind of objects in the Universe, active galactic nuclei have required observations at many wavelengths to piece together our present understanding. While divisions of the continuous electromagnetic spectrum are rather arbitrary, and driven as much by our particular means of detection as by more basic distinctions such as how they are produced, the following slices are generally recognized:
  • Radio - the lowest-frequency domain that we've needed to name. It extends from wavelengths of a kilometer or so, the longest that will propagate through the interstellar medium, down to about a millimeter (where we generally start speaking of the millimeter, microwave, or even far-infrared). Detection of radio radiation is often done using wave techniques rather than photon-counting, because of the low photon energies, and this offers distinct advantages for such applications as interferometery which astronomers working in the infrared and optical regimes view with some envy. From active nuclei, we often detect synchrotron radiation in this range - radiation produced as energetic charged particles (mostly electrons) produce when they are deflected by magnetic fields.
  • Infrared - A very broad slice of the spectrum, reaching from about 1 micron on the short side to hundreds of microns (depending on whether one wants to separate a distinct microwave portion of the spectrum). Starlight, line emission, and especially the thermal emission from dust grains warmed by various processes can be detected in the infrared. An important feature is that infrared radiation is much less subject to dimming by dust grains than optical or ultraviolet radiation, so we can see deeper into dusty regions at these wavelengths.
  • Optical - The original domain of astronomy (and all the sciences), defined at first by the wavelengths that the human eye responds to. This extends from about 4000-6800 Angstroms, often extended in astronomy to include the entire window around this band which the atmosphere passes with little attenuation (3000 Angstroms to about 10,000 Angstroms = 1 micron). Familiar technologies (lenses, aluminized mirrors, photographic emulsions, charge-coupled devices) are at their best here. Stars are obviously important sources of optical radiation, as are ionized gas and, in some nuclei and jets, synchrotron radiation from particles carrying very large energies.
  • Ultraviolet - There is a naturally imposed bound for distant objects at 912 Angstroms (13.6 eV), the energy needed for a photon to be absorbed by a hydrogen atom and liberate its lone electron in the process. There is so much neutral hydrogen in our galaxy that we cannot see very far away at wavelengths shortward of this until about 100 Angstroms, when the absorption by hydrogen (now assisted by other, trace elements) has dropped enough for the interstellar gas to become transparent once again. Aside from starlight and synchrotron radiation, the UV contains several very important strong spectral lines from abundant atoms.
  • Extreme ultraviolet - loosely, the range between the trough produced by hydrogen absorption starting at 912 A and the X-ray regime. By rough convention, taken to be 10-500 Angstroms wavelength. Observations in this range require some of the same techniques of grazing-incidence mirrors and photon detection as do X-ray measurements.
  • X-rays - In ordinary use in physics, loosely associated with processes in atomic nuclei rather than the surrounding electrons. However, some of the innermost electrons in elements such as oxygen and especially iron can produce features in the soft X-ray region (say at energies below a few keV). The very property which makes X-rays so useful in ordinary life - their penetrating ability through many kinds of matter - creates difficulty in trying to collect them for analysis. X-rays will travel right through ordinary mirrors, even metal ones. Except for devices working in a single, very narrow band of energy, X-rays can be focussed into images only by mirrors which make a very shallow angle with the X-rays' direction of travel - so-called grazing incidence mirrors.

  • Gamma rays - more or less, any radiation more energetic than X-rays. Energies as high as 1 TeV (tera-electron volt, trillion eV) have been recorded from cosmic sources. Gamma-ray detection generally has very poor directional sensitivity, so detail discrimination is poor compared to other wavelength regimes. These photons trace the most energetic processes in the Universe.