types, Central engine, Interrelations
Adapted from The Astronomy and Astophysics Encyclopedia and T.M. Heckman, J.H. Krolik, P. Barthel, R. Antonucci, B.M. Peterson et al.
Unlike the nucleus of a cell, the nucleus of a galaxy is not a precisely defined entity or distinct subcomponent. Rather, it is simply the central-most part of a galaxy. The region referred to as "the nucleus" is roughly the innermost 1% of a galaxy. Most galaxies are rather symmetric in form with the density of stars decreasing smoothly from the center (nucleus) outward. Thus, the nucleus is not only the center of the galaxy, it is also the region of highest density. As such, the nucleus can also be thought of as the "bottom" of the galaxy: gas clouds or stars that move too slowly within the galaxy can be pulled inward by gravity toward the nucleus. This may have some interesting consequences, as described below.
The nuclei of galaxies have been extensively investigated by astronomers for at least two reasons. The first, more prosaic reason is that the nucleus is usually the brightest part of a galaxy (because the density of stars is highest there). This means that the nucleus is the most easily studied part of a galaxy. The second and more exciting reason is that the nuclei of galaxies are often the sites of qualitatively unusual energetic phenomena that are observed nowhere else. These are the so-called active nuclei.
Galactic nuclei (like galaxies themselves) are composed of stars and interstellar matter (mostly gas, plus small dust grains). To explain active galactic nuclei, some additional object must be present. Because the fundamental nature of this object remains mysterious, it is often referred to by deliberately vague and fanciful terms like "the monster" or "central engine."
1. TYPES OF ACTIVE OR UNUSUAL NUCLEI
As already noted, an active nucleus is one in which processes are
observed that cannot be readily explained by the mere presence of
normal stars and interstellar gas clouds. By this definition, a
starburst nucleus is not a truly active nucleus, but we will discuss
such objects in this section because they are rare and can be highly
Quasars were originally defined to be star-like (quasistellar) objects
with large redshifts. Today they are believed by the great majority of
astronomers to be the highly powerful nuclei of distant active
galaxies. Quasars share many properties in common with Seyfert nuclei
(strong, broad emission lines and powerful ultraviolet and x-ray
emission). The subclass of "radio-loud" quasars, which are strong
radio emitters, is closely related to radio galaxies in their observed
Quasars comprise the most luminous subclass of AGNs, with nuclear magnitudes MB < -21.5 + 5 log h0. A small minority (5-10%) of these sources are the strong radio sources that originally defined the quasar class. Quasars are distinguished from Seyfert galaxies in that in general they are spatially unresolved on the Palomar Sky Survey photographs, which means in practice that they have angular sizes smaller than 7 arcsec. Many of these sources, however, are surrounded by a low surface brightness halo (sometimes called ``quasar fuzz''), which does indeed appear to be starlight from the host galaxy, and a few sources have other peculiar morphological features, such as optical jets (e.g., 3C 273). Quasar spectra are remarkably similar to those of Seyfert galaxies, except that (a) stellar absorption features are very weak, if detectable at all, and (b) the narrow lines are generally weaker relative to the broad lines than is the case in Seyfert galaxies. A ``typical'' QSO spectrum, constructed by averaging observations of a large number of QSOs, is shown below.
Figure 2.2. A mean QSO spectrum formed by averaging spectra of over 700 QSOs from the Large Bright Quasar Survey (Francis et al. 1991). Prominent emission lines are indicated. Data courtesy of P. J. Francis and C. B. Foltz.
Seyfert galaxies are usually spiral galaxies whose nuclei are
exceptionally bright. A few percent of spiral galaxies contain a
Seyfert nucleus. The spectrum of the nucleus shows Doppler-broadened
emission lines whose widths are similar to those in LINERs but whose
strengths are much greater. The gas is in a highly ionized state,
requiring the presence of a source of photons of much greater energies
than can be produced by ordinary stars. Direct evidence for this
ionization source is provided by the strong ultraviolet and/or x-ray
continuum emission observed from Seyfert nuclei.
Seyfert galaxies are lower-luminosity active galactic nuclei (than QSRS), with MB > -21.51 + 5 log h0 for the active nucleus the generally accepted criterion, due originally to Schmidt & Green (1983), for distinguishing Seyfert galaxies from quasars. A Seyfert galaxy has a quasar-like nucleus, but the host galaxy is clearly detectable. The original definition of the class (Seyfert 1943) was primarily morphological, i.e., these are galaxies with high surface brightness nuclei, and subsequent spectroscopy revealed unusual emission-line characteristics. Observed directly through a large telescope, a Seyfert galaxy looks like a normal distant spiral galaxy with a star superimposed on the center. The definition has evolved so that Seyfert galaxies are now identified spectroscopically by the presence of strong, high-ionization emission lines. Morphological studies indicate that most if not all Seyferts occur in spiral galaxies
Khachikian & Weedman (1974) were the first to realize that there are two distinct subclasses of Seyfert galaxies which are distinguished by the presence or absence of broad bases on the permitted emission lines. Type 1 Seyfert galaxies have two sets of emission lines, superposed on one another. One set of lines is characteristic of a low-density (electron density ne 103-106 cm-3) ionized gas with widths corresponding to velocities of several hundred kilometers per second (i.e., somewhat broader than emission lines in non-AGNs), and are referred to as the ``narrow lines''. A second set of ``broad lines'' are also seen, but in the permitted lines only. These lines have widths of up to 104 km s-1; the absence of broad forbidden-line emission indicates that the broad-line gas is of high density (ne 109 cm-3 or higher) so the non-electric-dipole transitions are collisionally suppressed. Type 2 Seyfert galaxies differ from Seyfert 1 galaxies in that only the narrow lines are present in type 2 spectra.
Samples of both a Seyfert 1 optical spectrum and a Seyfert 2 spectrum are shown below.
Figure 2. The optical spectrum of the Seyfert 1 galaxy NGC 1275. The prominent broad and narrow emission lines are labeled, as are strong absorption features of the host galaxy spectrum. The vertical scale is expanded in the lower panel to show the weaker features. The full width at half maximum (FWHM) of the broad components is about 5900 km s-1, and the width of the narrow components is about 400 km s-1. The strong rise shortward of 4000 Å is the long-wavelength end of the ``small blue bump'' feature which is a blend of Balmer continuum and FeII line emission. This spectrum is the mean of several observations made during 1993 with the 3-m Shane Telescope and Kast spectrograph at the Lick Observatory. Data courtesy of A. V. Filippenko.
Figure 3. The optical spectrum of the Seyfert 2 galaxy NGC 1667 is shown, with important emission lines identified (Ho, Filippenko, and Sargent 1993). Some strong absorption lines that arise in the host galaxy rather than the AGN itself are also identified. This spectrum can be compared with the spectrum shown in Figure 1. The units are: Wavelength (Å) for the x-axis and F (ergs s-1 cm-2 Å-1) for the y-axis. Data courtesy of A. V. Filippenko.
In addition to the strong emission lines, weak absorption lines due the late-type giant stars in the host galaxy are also observed in both type 1 and type 2 Seyfert spectra; the absorption lines are relatively weak because the starlight is diluted by the non-stellar ``featureless continuum''. Indeed, the AGN continuum is usually so weak in Seyfert 2 galaxies that it is very difficult to isolate it from the stellar continuum unambiguously.
The narrow-line spectra are clearly distinguishable from the HII-region spectra seen in some normal galaxies, as the Seyfert spectra show a wide range in ionization level, which is typical of a gas ionized by a source where the input continuum spectrum falls off slowly (relative to a Wien law) at ionizing wavelengths. A common, but sometimes misleading spectroscopic criterion for distinguishing Seyfert galaxies from HII-region galaxies is that the flux ratio [OIII] 5007 / H > 3.
The origin of the differences between Seyferts of type 1 and type 2 is not known. There are a few clear examples where galaxies have been identified as type 2 Seyferts because the broad components of the lines have proven to be very hard to detect. One school of thought holds that all Seyfert 2s are intrinsically Seyfert 1s where we are unable to see the broad components of the lines from our particular vantage point. It is not clear, however, that this hypothesis can explain all of the observed differences between the two subclasses.<>Osterbrock (1981) has introduced the notation Seyfert 1.5, 1.8, and 1.9, where the subclasses are based purely on the appearance of the optical spectrum, with numerically larger subclasses having weaker broad-line components relative to the narrow lines. In Seyfert 1.9 galaxies, for example, the broad component is detected only in the H line, and not in the higher-order Balmer lines. In Seyfert 1.8 galaxies, the broad components are very weak, but detectable at H as well as H. In Seyfert 1.5 galaxies, the strengths of the broad and narrow components in H are comparable. Caution must be exercised in using this subclassification of Seyfert 1 spectra; whereas the statistics of occurrence might conceivably have some bearing on unification issues, there are some cases in which broad emission-line variability is so pronounced that the subclassification changes with time. Indeed, there have been cases reported where the broad lines in Seyfert 1 galaxies have nearly completely disappeared (e.g., Penston & Perez 1984) when the nucleus has faded to a very faint state. Based on casual inspection, the source would have been classified as a Seyfert 2 galaxy. However, close examination of the spectra seems to indicate that the broad lines never completely disappear.
Blazars are members of the family of active galactic nuclei and quasars, defined specifically by their strong optical Polarization and variability. These unique defining properties seemed mysterious and even paradoxical in the l960s and 1970s, but now there is a growing consensus that their behavior and their role among quasars is qualitatively understood. Many of the modern ideas started to emerge during an important meeting in 1978 (the Pittsburg Confernce on BL Lac Objects). This is a good place to pick up the historical thread.
BL Lac objects have historically been defined as point-like sources of optical radiation that show little or no line emission, and strong and variable brightness and polarization. Pittsburgh meeting participants made it clear that some nearby objects exhibit all of these properties, along with narrow emission lines of considerable strength. Because they did not seem fundamentally different from the original BL Lacs, they were generally accepted as members of the class.
This was especially reasonable in light of the fact that the narrow emission line equivalent widths (strengths of the lines compared with that of the continuum) vary inversely with continuum flux. Without this unification, an object's class would sometimes be a function of time!
Optically violently variable (OVV) quasars presented a similar situation. They were defined as broad emission line objects which otherwise showed the characteristics of BL Lacs. In fact, in their contributions to the Proceedings, Joseph Miller and collaborators showed that very high signal-to-noise ratio spectroscopy of known BL Lacs sometimes reveals broad emission lines. Furthermore, some OVVs clearly look like BL Lacs when their continua are in high brightness states. These facts are closely related. Since 1978, several of the BL Lacs discussed by the meeting participants have shown broad emission lines when observed carefully in low states. Therefore, it is no longer possible to distinguish BL Lacs and OVVs in a rigorous way and the two classes were merged under the name blazars. (Of course, this does not imply that all blazars are intrinsically exactly the same.)
The old-fashioned view of blazars was that their high polarization and tiny sizes(from variability arguments) meant that they were bare quasars, with the fundamental energy generation process being observed directly, perhaps within a few gravitational radii of supermassive black holes. In this picture, ordinary quasars are surrounded by gas that reprocesses and depolarizes the radiation and damps out the variability.
Roger Blandford and Martin Rees presented a very different idea at Pittsburgh, an idea which has since had many successes and which prevails among most researchers today. The high polarization and "power law" spectra could naturally be produced by synchrotron radiation, as in the Crab nebula. However, Blandford and Rees pointed out the very severe constraints on any such model that result from the rapid optical variability and high observed luminosities. The variability seems to require that even the luminous sources are intrinsically tiny (light-days or less). However, the polarization requires that both the optical depth to electron scattering and the optical depth to the synchrotron self-absorption process must be low; the reason is that both of these processes destroy polarization. A source satisfying all of these constraints basically cannot be as luminous as those observed!
Now all of the constraints would be greatly alleviated if we made one assumption: Suppose the synchrotron sources are not stationary, but are moving in bulk at relativistic speeds toward Earth. (This idea is called the beam model.) Then two things happen. Because the radiation is "beamed forward" by special relativistic aberration, the observed fluxes are greatly boosted. Therefore, the luminosities in the rest frames are much less than was otherwise thought. Also, with the emitting volume moving toward Earth and nearly keeping up with its own past images, the rapid observed variability is partially an illusion. The variations have been compressed in time. In the rest frames they are substantially slower, so the sources can be rather larger than in a stationary model.
After Blandford and Rees' paper was written, the variability constraints became even stronger. Papers by Chris Impey and collaborators and by P. A. Holmes and collaborators reported studies of variability in the infrared. This is where blazars put out most of their energy. Now, independent of the emission mechanism, radiation from black hole accretion is generally not expected to vary on time scales shorter than the travel time of light across the event horizon of a maximally accreting black hole. Yet infrared monitoring showed such enormous apparent luminosities and such rapid variability that even this conservative expectation was violated in at least five cases!
The assumption that all blazars are moving relativistically toward Earth may seem ad hoc or even crazy. In fact, it is very reasonable. Blazars invariably have very bright compact radio cores, and these cores often show very strong evidence for such a scenario. It was well known since the work of Fred Hoyle, Ceoffrey Burbidge, and Wallace Sargent in the 1960s that a stationary synchrotron model for compact radio sources was not tenable. Radio variability seems to require extremely compact sources, and from these sizes and the observed radio fluxes, the surface brightnesses can be calculated. These turned out to be far above the "Compton limit" at 1012 K in brightness temperature. A stationary synchrotron source must emit fantastically large and observationally excluded inverse-Compton x-ray emission in order to have such a high brightness temperature. Therefore, relativistic motion in the line of sight had already been invoked. The idea was confirmed when superluminal(apparent faster-than -light) motion of milliarcsecond-scale radio jets was discovered.
The "time compression" of the observed variability was also verified by James Condon and B. Dennisoh. They showed that if the sources were really as small as naively expected from the radio variability data, they should have such small angular sires that they should show interstellar scintillation (twinkling), and they do not!
Blandford and Rees supplied an astrophysical context for synchrotron sources undergoing relativistic bulk motion in the line of sight. They suggested that the sources were simply the bases of the jets of normal double radio galaxies and quasars that happened to point in our direction. After all, some of these objects must be oriented in that way. Beaming of radiation by the aberration effect referred to earlier boosts the radio core fluxes in such objects, so they should be greatly over-represented in flux-limited surveys.
M. ORR and I. Browne adopted a simplified version of Blandford and Rees' idea. They postulated that all blazars, other core-dominant radio sources, and normal double sources all have similar relativistic bulk speeds, that the motions are along straight lines, and that the jets are linear in shape (rather than, say, conical). They concluded that such a simple model was consistent with a variety of source count data. Finally, Orr and Browne gave the name unified scheme to the hypothesis that flat-spectrum core-dominant sources are just normal doubles seen end-on. (The flat-spectrum core-dominant sources are just a slightly larger superset of blazars.)
The hypothesis that blazars are double radio sources seen along their jet (symmetry)axes obviously predicts that the double lobes should be seen projected as halos on the strong radio cores. It was just becoming possible in the early 1980s to achieve the required dynamic range in interferometer maps that was needed to detect such halos. (Remember that the blazar radio cores are tremendously strong. )Several groups discovered significant diffuse radio emission around many sources; this includes work by R. T. Schilizzi and A. G. de Bruyn with the Westerbork telescope, and Wardle and collaborators and James Ulvestad and collaborators with the NRAO Very Large Array.
R.Antonicci and Ulvestad carried out an exhaustive blazar mapping program with the VLA and discovered substantial diffuse radio emission in almost all cases. The emission had qualitatively the right power, morphology, and projected linear size for the unified scheme. They critically examined various counter arguments in the literature, and then showed that if the beam model is qualitatively correct, the unified scheme must be, too. Suppose the beam model is correct and the core radio flux is beamed into a small solid angle that includes the direction to Earth. Suppose also that the large diffuse sources discovered in association with blazars emit isotropically. (This is very likely for the large, diffuse, and often two-sided halos.) Some blazars have sufficient flux in the diffuse radio halos alone to qualify for inclusion in the flux-limited radio catalogs.
Therefore, under our two hypotheses, blazars not directed at Earth would still be in the catalogs, but classified as something else. The only candidates are the normal double sources. In fact, statistically, many or most normal doubles would have to be misdirected blazars!
Two exciting recent developments need to be mentioned. First, according to the unified scheme, normal double quasars should show much lower speeds in their cores than blazars do (although they should still be superluminal). Sensitive, very long baseline interferometry experiments are now being carried out, and the speeds are, in fact, coming in at 1-5c rather than the 5-l0c typical of blazers.
The second recent development also seems to be a great success for the beam model and the unified scheme. Luminous double radio sources have two lobes that are generally fairly similar in flux, but jets that are very dissimilar in flux. This is at first sight unexpected because the jets appear to be the source of energy feeding the lobes. In the beam model, the jet radiation asymmetry is nicely explained as the result of beaming of the radiation from the jet closer to the line of sight toward us and beaming of the far jet radiation away from us. This does not require the axis to be very close to the line of sight as the blazar phenomenology does. Now the exciting new development is that Robert Laing and collaborators have discovered that in almost every case, one of the radio lobes is depolarized by passage through a magnetoionic medium (or"Faraday screen"), so that the depolarized lobe would be past the screen and the polarized lobe would be on its near side. (The Faraday screen would then probably be associated with the host galaxy.) The near side determined in this way is essentially always the side with the strong radio jet! This seems to mean that the jet radiation is beamed forward. Other interpretations are still possible but most researchers feel that the discovery of Laing and collaborators is a tremendous boost for the beam model.Finally, there is evidence that the normal double quasars and broadline radio galaxies that lie very close to the sky plane are observed and classified as narrow-line radio galaxies, at least in some cases. The optical continuum sources and broad emission line regions are apparently obscured by opaque tori composed of dust clouds. The evidence comes from optical spectropolarimetry and from some statistical tests which seem to show that too few objects classified as quasars lie very close to the sky plane. These arguments are summarized and the appropriate references given in a recent paper by the present author, which discusses orientation effects in radio-quiet objects as well.
LINERs are the most common type of active nucleus, and may in fact be
present at a very low level in the nucleus of every early-type galaxy.
Their spectra are characterized by weak emission lines that have been
significantly broadened by the Doppler effect, indicating high speed
gas motions (typically a few hundred to a few thousand kilometers per
second). LINERs usually contain a compact source of radio synchrotron
emission that is qualitatively similar to (but much weaker than) the
radio sources seen in radio galaxies and quasars.
A very low nuclear-luminosity class of low-ionization nuclear emission-line region galaxies (LINERs) was identified by Heckman (1980). Spectroscopically, they resemble Seyfert 2 galaxies, except that the low-ionization lines, e.g., [O I] 6300 and [N II] 6548, 6583, are relatively strong. LINERs are very common, and might be present at detectable levels in nearly half of all spiral galaxies (Ho, Filippenko, and Sargent 1994). A sample LINER spectrum is shown in the Figure.
Figure 1. The optical spectrum of the LINER NGC 1052 is shown, with important emission lines identified (Ho, Filippenko, and Sargent 1993). Some strong absorption lines that arise in the host galaxy rather than the AGN itself are also identified. Important differences between Seyfert 2s and LINERs are apparent: the [O III] 5007 / H flux ratio is much larger in Seyfert 2s (in NGC 1667, the weak H line is obscured by blending with the stellar H absorption line) than in LINERs, and low-ionization lines ([N II] 6716, 6731, [S II] 6548, 6853, [O II] 3727, and [O I] 6300) are all relatively prominent in LINER spectra. Data courtesy of A. V. Filippenko. The y-axis units are F (ergs s-1 cm-2 Å-1).
The [O III] / H flux ratio is often used to distinguish Seyfert galaxies from other types of emission-line galaxies. The criterion that the flux ratio [O III] / H > 3 in AGNs is not a robust indicator, however, because this flux ratio is also typical of low-metallicity HII regions. Indeed, LINER, Seyfert-galaxy, and HII-region spectra cannot be unambiguously distinguished from one other on the basis of any single flux ratio from any pair of lines. However, Baldwin, Phillips, and Terlevich (1981) have shown that various types of objects with superficially similar emission-line spectra (i.e., characteristic of a 104 K gas) can be distinguished by considering the intensity ratios of two pairs of lines; the relative strengths of various lines are a function of the shape of the ionizing continuum, and they therefore can be used to distinguish between, for example, blackbody and power-law ionizing spectra. Figure 2.3 is an example of a ``BPT'' (for Baldwin, Phillips, and Terlevich) diagram which demonstrates how LINERs can be distinguished from normal HII regions and normal AGNs (Seyferts and QSOs) on the basis of the [O III] 5007 / H, [N II] 6583 / H, and [S II] 6716, 6731 / H flux ratios. Here it is seen that the Seyfert 2s have high values of each ratio. H II regions define a locus of lower values which does not overlap with the region of parameter space occupied by the Seyferts. The LINERs can be distinguished from the Seyfert 2s by their low values of [O III] 5007 / H relative to [N II] 6583 / H, and from the H II regions by their larger values of [N II] 6583 / H.
Some models indicate that the emission-line spectra of LINERS are consistent with photoionization by a Seyfert-like continuum which is very dilute. The presence of strong [O I] 6300 is especially indicative of a power-law ionizing spectrum, because the ionization potential of O0 is nearly identical to that of H0; the [O I] line, which is collisionally excited, will only occur in a zone which has a sufficiently high electron density and temperature to excite the upper level. With a stellar input spectrum, these conditions only occur within the H+ Strömgren sphere, where the O0 abundance is negligible. However, a gas ionized by a relatively flat power-law spectrum has an extended partially ionized zone where the [O I] emission arises.
The relationship between LINERs and AGNs is not
Some, but by no means all, LINERs appear to be simply very
low-luminosity Seyfert galaxies. LINER-type spectra can also be
produced in cooling flows, in starburst-driven winds, and in
shock-heated gas (Heckman 1987, Filippenko 1992).
Starburst nuclei can rival Seyfert nuclei or even
some of the less
powerful quasars in terms of their total power output. However, unlike
Seyferts or quasars, the properties of starburst nuclei can be
adequately explained by young stars (albeit a highly unusual
such stars). Starburst nuclei often radiate most strongly in the
infrared portion of the electromagnetic spectrum. This infrared
emission comes from dust grains that have been heated to temperatures
of several tens to several hundreds of degrees Kelvin by the
ultraviolet light produced by the hot young stars. The presence of the
dust is not surprising, because dust is found to be closely associated
with cool, dense molecular clouds of the kind that are apparently
present with great abundance in starburst nuclei.
A large number of the Seyfert galaxies
presently known were identified
in the low-dispersion spectroscopic survey of Markarian and his
colleagues at the Byurakan Observatory in Armenia (see Lipovetsky,
Markarian, and Stepanian 1987). The Byurakan survey was done with an
objective prism on a 1-m telescope, at a dispersion of ~ 1800 Å
These spectra were used to identify UV-excess objects. Huchra (1977)
estimates that about 11% of the objects in the Markarian catalogs are
Seyfert galaxies, ~ 2% are Galactic stars (mostly very hot white
dwarfs), another ~ 2% are QSOs and BL Lac objects, and the rest are
galaxies which are rather blue for their morphological type (blue
compact dwarf galaxies and starburst galaxies).
2. AGN - CENTRAL ENGINE
Objects with a great variety of names - QSOs (or quasars), blazars, Seyfert galaxies, radio galaxies, and sometimes liners(low ionization nuclear emission line galaxies) - are all grouped into the category active galactic nuclei (AGN) because they share a basic set of common properties: very small spatial extent (on the galactic scale), luminosity comparable to or greater than that of an entire galaxy, and substantial power radiated in frequency bands where stars emit very little if at all. In addition to this set subscribed to by all AGN, many show evidence for bulk motion at relativistic speeds. Somewhere inside each object there must be a system responsible for the tremendous amounts of energy released; because they share so many basic characteristics, it is generally thought that in each of the different varieties of active galaxy this "central engine" is built according to basic design that is common to all. The "specifications" for this central engine are exactly this list of common properties, and we begin by briefly elaborating on them. At present, observations only give upper limits on the sizes of these objects. Atmospheric "seeing" limits angular resolution of ground-based telescopes to -1 arcsecond, corresponding to -100 parsec(pc) in even the nearest AGN. Some AGN are strongly variable; in these, causality causality limits the size to the distance light can travel in a characteristic variability time. This limit is often considerably less than 1 parsec, but systematic studies of AGN variability are still in their infancy. Active galactic nuclei can be found over a very wide range of luminosity. The all-time record is ~ 1048 erg s-1, or more than 104 times brighter than an average galaxy, but luminosities this large are quite rare. At redshifts around 2, AGN with luminosities ~ 1046 erg s-1 existed in ~ 1% of galaxies , whereas at the present epoch a few percent of all galaxies contain AGN with luminosities ~ 1044 erg s-1. It is possible tat somewhat weaker AGN are still more common.
Perhaps most remarkable of all, whereas stars emit nearly all of their power in a frequency band a mere factor of 3 wide, and the range of stellar temperatures broadens that range for a galaxy by no more than another factor of 3, most AGN produce roughly equal amounts of power per logarithmic frequency band all the way from the mid-infrared to hard x-rays - a span of 107 in frequency. The exceptions (radio galaxies) produce such a large ratio of very low frequency (radio) power to optical that they, too, could hardly be stars.
Whatever constitutes the central engine, it almost certainly must have a very large mass. Two arguments lead to this conclusion. First, because the force due to radiation pressure falls off with distance from the source in exactly the same inverse square fashion as gravity, there is a critical luminosity to mass ratio beyond which a self-gravitating and radiating structure cannot exist. This is called the Eddington luminosity, and is 4 × 104 in units of solar luminosities per solar mass. From this argument we infer that the central engines of active galaxies must have a mass at least ~ 106(L / 1044 erg s-1) M.
Second, the total active lifetime of an AGN must be at least ~ 108 yr. This is the minimum mean active lifetime derived from the observed frequency of AGN if all galaxies are occasionally active. It is possible that only a few percent of all galaxies have ever been active, but in that case the observed frequency of AGN means that they must have been active throughout the lifetime of the universe, ~ 1-2 × 1010 yr. Thus the minimum total energy released by an average AGN is ~ 1060 erg. It is possible to estimate the minimum accumulated mass in the central engine by supposing that this energy was derived from processing some sort of "Fuel. "Chemical fuels release ~ 10-9 of their rest-mass energy when burnt; nuclear reactions release ~ 10-3. Only the conversion of gravitational potential energy into heat when matter falls into a relativistically deep potential well produces energy with efficiency approaching unity in rest-mass units: Accretion onto a neutron star releases a fraction 0.1-0. 2, accretion onto a maximally rotating black hole can release up to 0.29. Even with these high efficiencies, the minimum accumulated mass for a typical AGN is still ~ 107 M, and if only a small fraction of galaxies ever become active, the minimum accumulated mass could be much greater than this.
It is the difficulty in understanding how such large masses are brought so close to the centers of galaxies that has led most astronomers to believe that the basic power source for an AGN is accretion into a relativistically deep potential, probably a massive black hole. Although a dense cluster of neutron stars cannot absolutely be ruled out, it seems less likely: If typical AGN are less than few light-days across(as the variability would in some cases suggest), the cluster would have to be so dense that stellar collisions would cause collapse to a black hole in less than the minimum active lifetime of ~ 108 yr.
The principal hurdle in bringing so much mass so close to the center of a galaxy is dumping the matter's angular momentum. Average stars in galaxies have 105 times more angular momentum than the maximum permitted for accretion onto a central black hole, and it is hard to identify mechanisms that efficiently remove angular momentum from material orbiting in galaxies. By contrast, energy can be lost comparatively easily by radiation. For this reason, it is generally thought that material approaches the central black whole along trajectories in which the energy is the minimum consistent with an orbit of that angular momentum. These trajectories taken together form a flat disk. At large distances from the center, it is possible that global disturbances in the gravitational field of the host galaxy remove angular momentum from the accreting matter; at small distances, friction between material on neighboring orbits may cause a slow outward transport of angular momentum and an associated slow sifting inward of the matter in the disk.
Energy can be released and transformed into radiation in a variety of ways. Within the disk, the same friction causing the angular momentum transport also causes local heating. The energy source, of course, is the slow fall of material in the gravitational field of the central object. This heat can then be lost by thermal radiation. Because the greatest amount of gravitational potential energy is the lost in the innermost rings of the disk, this inner region dominates the total power radiated, and its typical temperature (~ 105 K)forces most of the photons radiated by the disk to emerge in the ultraviolet.
Nonthermal mechanisms are possible also. Indeed, to explain the very broad range of photon energies seen, they are probably required. These generally involve populations of relativistic electrons(and sometimes positrons) in which the numbers of particles with a particular energy are proportional to a power of the energy. Relativistic particles are often distributed in energy in this way because the only characteristic energy scale relevant to them is the particle rest mass. Relativistic electrons can create new photons by the synchrotron mechanism and they can also multiply the energy of already-existing photons by large factors as a result of inverse compton scattering. Many suggestions have been made about how to produce such large quantities of very energetic electrons, but no consensus currently exists. Some, but not all, of these suggestions depend on accretion into a relativistic gravitational potential.
Electromagnetic effects are also likely to play a role in the energy release: The characteristic scale of magnetic field strength near the edge of the black hole is ~ 104(L / LE)(L / 1046 erg s-1)-1/2 G, though electric fields should (in most places) be efficiently shorted out by the high densities of ionized plasma. Magnetic fields threading the surface of the black hole and coupled to the external plasma can allow the rotational energy in the black hole to be tapped, and energy stored in the field itself can be released nonthermally if regions with oppositely directed field can be brought together to reconnect.
Our present understanding of the generation of bulk relativistic motions is even cruder. Possibilities include acceleration of plasma on magnetic field lines attached to a rotating black hole, hydrodynamic acceleration inside funnels formed by general relativistic dynamical effects along the rotation axis of a black hole, or acceleration by radiation pressure, but it is quite possible that the correct answer is something else altogether.
In sum, measured against the specifications for a
massive black holes do better than any other model yet proposed. They
are certainly sufficiently compact, having radii ~
10-5(M / 108
very large luminosities can be produced with a minimum mass in fuel
consumed; the high energies afforded by the depth of their potential
wells help in the creation of relativistic particles that can radiate
over a broad range of frequencies; and they potentially provide sites
for the bulk acceleration of matter to relativistic speeds. However,
there are few statements that can be made on this subject with great
Currently, the most popular theory holds that the monster that powers active galactic nuclei is a supermassive black hole, a region of high density within which the escape velocity exceeds the speed of light. The mass of the black hole must be at least several million times the mass of the sun for a Seyfert nucleus, and several billion times the mass of the sun for a powerful quasar. Energy would be produced by the supermassive black hole as its powerful gravitational field compresses and heats infalling gas, causing the gas to emit highly energetic photons before it falls into the hole and vanishes. Recall that galactic nuclei are at the bottom of the galaxy, a favorable location for accreting material to "feed" the monster. Although there are a variety of plausible lines of indirect evidence that favor this model, the case is by no means clear.
The difficulty in proving this model rests largely with the fact that powerful active nuclei are located so far away that the direct presence of the supermassive black hole cannot be unambiguously detected. However, there are at least two pieces of evidence that suggest that dormant supermassive black holes may reside at the centers of many nuclei that are not presently in a highly active state. The first is the LINER phenomenon, which may be the result of a "starved monster" - a supermassive black hole that is producing very little energy because it is receiving only a slow trickle of food in the form of infalling gas. The second is the fact that the most powerful active nuclei were evidently much more common in the distant past than they are today. Many more galactic nuclei may contain the essential equipment for producing a quasar (i.e., supermassive black hole) than is evident from the scarcity of highly active nuclei in the present universe.
Motivated by such ideas, astronomers have recently searched the nuclei of nearby galaxies (including the Milky Way) for the presence of a "dead quasar" (a supermassive black hole that is presently producing little or no light). The basic technique is to measure the motions of stars in the nucleus, and then to use Newton's laws of motion and law of gravity to "weigh" the mass contained in the center of the nucleus. The presence of a supermassive black hole would be revealed by stellar velocities that increase rapidly toward the center of the nucleus (because of the strong gravitational field of the black hole) and by a calculated mass that is far in excess of the mass that could be contained in normal stars. The interpretation of such observations is a subtle and difficult task. Nevertheless, there is now good enough evidence to strongly suggest that the nuclei of several of the nearest galaxies may well contain supermassive black holes with masses that are several millions to several tens of millions times that of the sun.
Discovering and investigating the various properties of QSOs and active galaxies such as radio galaxies, Seyfert galaxies, and BL Lacertae objects objects over the past three decades, astronomers noted that different classes of objects share properties, even in a quantitative sense. This led them to propose interrelating schemes for seemingly different objects. Some of these schemes had to be abandoned; other ones survived, sometimes after minor or major revision. This entry deals with such interrelating and/or unifying schemes. Because knowledge of the similarities and dissimilarities of the various types of QSOs and active galaxies is essential to understanding possible interrelations, I will first briefly introduce the subject and review the basic, relevant properties of the different classes of objects.
Normal galaxies such as our Milky Way galaxy or the Andromeda galaxy (M31) emit the combined radiation of some hundred billion stars as the bulk of their radiation. The evolution of such normal galaxies is a gradual one, governed by the evolution of these very many stars, which are these very many stars, which are born, go through various phases of stella stellar life, and finally die. Unless major disturbances from outside take place, the structure and radiation of a normal galaxy will not change during a time span of a hundred million years. Active galaxies produce more than just the radiation of stars. Radio and x-ray radiation from nonthermal processes, but also optical line radiation originating in hot gas clouds are among the signs of this activity. The activity usually originates in the centers of these galaxies: hence the acronym AGN, for active galactic nuclei. Species in the active galaxy "zoo" include Seyfert galaxies, radio galaxies, BL Lacertae(BL Lac) objects, and QSOs. The QSO or quasar class can be subdivided into weak radio sources and strong radio sources(the latter group is often referred to as quasistellar radio sources, or QSRs). Taken with more or less effort, optical images of BL Lac objects, Seyfert galaxies, and (nearby) radio galaxies generally show the underlying galaxy associated with the active nucleus. Although the actual galaxy is not observed in most quasars, all but a few astronomers nowadays classify these objects as distant and ultraluminous AGN.
Following the selection criteria of their discoverer Carl Sevfert, Seyfert galaxies are characterized by having small, bright nuclei (optical) and strong emission lines in their optical spectrum. Such emission lines are emitted by hot, ionized gas clouds, which in turn require the presence of an intense flux of ionizing ultraviolet photons originating in the galactic nucleus. Broad emission lines originate in chaotic moving dense, hot gas (the broad-line region, BLR), located within a few light-years from the nucleus of the galaxy. Narrow emission lines are produced in a more tenuous hot gas further out; this narrow-line region (NLR) stretches out to distances of several thousand light-years, and, in Seyfert galaxies that are not too far away, the NLR can sometimes be resolved with optical telescopes. Early in the study of Seyfert galaxies, it appeared that two subclasses in the population could be separated, based on the relative width or strength of narrow emission lines with respect to the broad emission lines. These lines are of comparable strength in Seyfert type 1 galaxies, whereas the broad lines are considerably less luminous than the narrow lines in type 2 Seyferts. Most Seyfert galaxies produce stronger radio emission than normal galaxies; however, the radio luminosity of a typical Seyfert galaxy is in the range of 0. 1-1% of an average radio galaxy or QSR. The radio emission generally emanates from within the dimensions of the optical galaxy.
Radio galaxies display strong radio emission, most of which originates in two giant radio lobes that straddle the associated optical galaxy. The radio sources can be of gargantuan dimension; radio galaxy 3C 236 is the largest known object in the universe, having a (projected!) linear size of 13 million light-years. The associated galaxies are of (giant) elliptical type in radio galaxies of low and moderate luminosity, but they often have peculiar optical appearances in the most luminous cases. Most notable in the latter cases are spatial elongations of hot emission line gas in the direction of the extended radio emission and the presence of morphological peculiarities such as faint tails and wisps. Luminous radio galaxies often display strong narrow emission lines. Based on the radio luminosity, the radio galaxy population can be divided into into two subclasses: Above a certain luminosity the radio sources have a linear, edge-brightened, simple double-lobed morphology, whereas most nearby radio galaxies that are below this "break luminosity"tend to have more or less complex, edge-darkened morphologies. Whereas(two-sided, at both sides of the nucleus) radio jets are generally observed in the latter group, the luminous radio galaxies seldom display jets.
BL Lacertae or BL Lac objects are elliptical galaxies with very bright, variable nuclei, displaying strong, variable radio emission, the morphology of which is dominated by a compact radio core. Emission lines are not seen; the redshift, and thereby distance of a BL Lac object, is inferred from stellar absorption features in the faint extended optical emission of the elliptical galaxy itself.
An important subpopulation of quasistellar objects (QSOs or quasars) is formed by the quasistellar radio sources (QSRs), the radio-loud QSOs. Historically, quasars have been characterized by a stellar appearance and the presence of strong, broad, redshifted emission lines. The optical spectra also display narrow, redshifted emission lines. These redshifts, resulting from the general expansion of the universe, indicate that quasars are the most distant objects in the observable universe. QSRs are powerful radio sources, all exceeding the previously mentioned break point in radio luminosity. Broadly speaking, they have either a compact core-halo radio structure(with dominant, variable core emission) or an extended double-lobed morphology(with dominant lobe emission). QSRs of the latter morphology resemble the luminous radio galaxies, although the QSRs are smaller. A further difference with powerful radio galaxies is that many QSRs display radio jets, but always at one side of the nucleus only. Many of die compact QSRs vary rapidly in their optical light; such optically violent variable(OVV)QSRs are usually combined with the BL Lac objects into the so-called blazar class. Not all radio-quiet QSOs are radio silent; sensitive radio telescopes detect weak, compact radio emission in a considerable fraction of the QSO population. The radio luminosity of these QSOs is two to four orders of magnitude weaker than in typical, otherwise similar QSRs. The fraction of QSRs among QSOs is about ten percent.
It will be clear that characteristic AGN properties, such as a bright nucleus, powerful radio lobes, or strong emission lines, are shared by members of different types. With growing databases, the notion developed that the various AGN types could be interrelated.
The qualitative similarities between the Seyfert 1 class and the QSOs and the quantitative continuity in their properties were already recognized in the early 1970s. Observations of radio jets were reported from 1977 onward, and the identification of BL Lacs with AGN in which jets were pointing at us was subsequently suggested in 1978. This suggestion attempted for the first time to attribute widely differing properties of active galaxies to the effects of their orientation. Several more were to follow, which will be discussed, in roughly chronological order.
Using the radio astronomical very long baseline interferometry (VLBI) technique, by the end of the 1970s three QSRs and one Seyfert type 1 galaxy had been found to display superluminal velocities in their nuclear radio jets. The accepted explanation for this phenomenon is based on relativistic motion in a radio jet pointing nearly at the observer. Matter moving at nearly the speed of light in a direction close to the line of sight will almost overtake its own radiation; time intervals will be compressed, creating the illusion of transverse speeds in excess of the speed of light. This relativistic beaming model is attractive, because it also explains the apparent one-sidedness of the radio jets, as well as the observed core dominance in superluminal and core-halo QSRs in general, as a result of Doppler boosting of relativistically approaching material. Expanding on this picture, a proposal that the radio-loud QSRs and the radio-quiet QSOs could be interrelated through orientation, in the sense that QSRs are QSOs with jets pointing in our direction, was put forward. This unified scheme had to be dismissed a few years later, after close examination of the weak radio emission of the QSOs. A subsequent proposal, however, unifying the compact, core-halo QSRs and the extended, double-lobed QSRs through orientation, has found rather widespread approval. Defining the source axis as the line connecting the two radio lobes (and passing through the nucleus), the important parameter in this scheme is the angle 0 of this source axis with respect to the line of sight, toward the observer. One prediction of this scheme, namely that the lobes of a QSR should be seen in projection on the strong, boosted core, when seen end-on(at small angle 0), has been successful. Furthermore, the relative numbers of core-halo and double-lobed QSRs in radio source catalogs appear reasonably consistent with this scheme. The model implicitly assumes that the axes of quasistellar radio sources are randomly oriented in space. This assumption may not be valid, as we will see later on. The alternative to this QSR unified scheme would be the picture where intrinsically small (young?) QSRs have stronger and more variable radio core emission. To many astronomers these and related lines of thought appear somewhat contrived, however. A combination of the effects of orientation and source intrinsic effects, such as individual source evolution, is likely of course. The relative importance of these two effects will no doubt be determined in the next few years.
Considerable progress in understanding the interrelation between the classes of Seyfert galaxies was made using the technique of spectro-polarimetry, that is, spectroscopy of the polarized light component. This technique has revealed several cases of type 2 Seyfert galaxies with obscured type l regions; the polarized light spectrum was found to display a strong optical continuum and broad emission lines, which are typical type 1 characteristics. The angle of polarization in general, and the case of the prototypical Seyfert 2 galaxy NGC 1068 in particular, where nuclear light reflected off a dust cloud was actually measured, strongly argue for the presence of an obscuring dust torus(doughnut) around a bright continuum and broad emission line nucleus. Seen from above or below, this nucleus is directly visible; seen from the side, the nucleus indirectly visible through reflection (causing the polarization) by by gas and dust above and below the torus. Evidence for dust obscuration through correlations with x-ray emission has also been reported, but it is not yet clear how general these phenomena in Seyfert galaxies are. However, there is no doubt that anisotropic radiation exists, because an optical image of the proposed bi-conical radiation field escaping from the presumed torus has already been obtained for one Seyfert 2 galaxy.
As described previously, the unification of compact core-halo QSRs and extended double-lobed QSRs via the effects of jet flow speeds of near the speed of light and orientation was quite successful in explaining a number of observed properties. Continued VLBI monitoring of moving radio components in QSR nuclei, however, revealed many more cases of superluminal motion, not only in compact but also in large double-lobed QSRs. Using sophisticated radio imaging techniques, it was found that radio jets in QSRs generally occur at one side of the nucleus only. In the framework of relativistic beaming, this would imply that many if if not all QSRs are oriented more or less toward us. Stated other wise: QSRs whose relativistic beams are perpendicular to our line of sight in effect hide or masquerade themselves as others sorts of objects. Based on the facts that powerful radio galaxies exist everywhere in the (history of the) universe where QSRs exist, and that the former are on average a factor of 2 larger in projected size and have comparable narrow emission line luminosities (hence comparable level of activity in the NLR), unification of these objects was proposed. As in the Seyfert case, a dusty torus perpendicular to the radio axis could hide the QSR broad emission line region as well as the bright continuum radiation, when observed from the side (edge-on). such a configuration would would naturally explain the extension of the emission line gas as discovered in several powerful distant radio galaxies, as due to cones of of ionizing radiation emitted by a hidden energy source and escaping along the radio axis. Also, the strong excesses as fear-infrared wavelengths reported in these radio galaxies could well be due to reradiation by obscuring dust. Many observational facts are consistent with this unified model, but the case has to be fully established. The alternative picture would be to identify radio galaxy with a burned-out QSR. A QSR in which the violent nuclear activity has (temporarily) stopped would look like a powerful radio galaxy, but the questions as to the existence of truly randomly oriented QSRs or the origin of the apparent radio morphological asymmetries remain. As mentioned already, the relative importance of orientation and object evolution will be a subject of detailed study in the coming years.
Although additional effects of orientation are not ruled out, the class of radio-quiet QSOs is most likely evolutionary linked to another, recently recognized, class of active galaxies, namely powerful infrared galaxies. These objects were discovered by the infrared astronomy satellite lRAS and produce more than 1012 L in the infrared part of the electromagnetic spectrum. These luminosities are comparable to QSO luminosities. Because these infrared galaxies also display QSO spectral characteristics and their optical morphologies resemble those of some nearby QSOs, identification with dust-enshrouded QSOs was proposed. Once all the dust has been blasted away, a brilliant QSO should appear. Equally important is the fact that optical images of these QSOs-in-the-making suggest galaxy interaction as the ultimate origin of the nuclear activity.
In a way that is complementary to the unification of radio-loud QSRs with powerful radio galaxies, the association of BL Lac objects with favorably oriented radio galaxies of low and moderate luminosity has been suggested. BL Lac objects occur rather close to our galaxy. Unifying them with the powerful radio galaxies which are rare in the local universe is therefore not possible. Lower luminosity radio galaxies do exist in larger numbers locally, and because their radio lobe luminosities are comparable to the luminosity of the halo emission in BL Lac objects, unification of these classes of objects is likely. The absence of emission lines in BL Lacs is no surprise in this picture, because the lower luminosity radio galaxies also lack those lines. Moreover, the optical emission of BL Lac objects is strongly dominated by the (polarized)non-thermal jet component. Note that the OVV QSRs, which together with the BL Lacs make up the blazar class, do have emission lines. Compared to other QSRs, the effects of beaming are most pronounced in OVV QSRs, which are therefore regarded as QSRs with jets closest to the sight line.
For the previous interrelating schemes to hold, it is imperative that the properties of the host galaxies as well as their local environments are undistinguishable. Investigations to test the various predictions concerning host galaxies and environments are currently in progress.
Important steps have been made in the past 10 years to explain the observed AGN diversity as a result of the combined effects of beaming, projection, and anisotropic radiation/obscuration in a small number of intrinsically different, evolving classes of objects. The effects of orientation and dust obscuration appear far more important than previously thought. The puzzle is not yet solved, but many pieces of the picture seem to be falling into place.
Compiled by G.T.Petrov, 2004