|
Annu. Rev. Astron. Astrophys. 1992. 30:
705-742
Copyright © 1992 by Annual Reviews. All
rights reserved |
6. ACTIVE GALAXIES
Observationally, it has long been known that there may be a link
between interactions and unusual forms of energy generation in
galaxies.
Baade & Minkowski (1954) argued
that the radio source in
Cygnus A consists of ``. . . two galaxies in actual collision.''
However, it was widely believed that objects like Cygnus A were
anomalous and that most galactic activity was produced by explosions
or other violent events within isolated galaxies. Only during the
past decade have observations challenged this long-standing prejudice
(for a discussion, see
Balick & Heckman 1982).
Attempts to establish a definitive causal relationship between
galactic activity and interactions have evoked considerable
controversy. Statistical analyses are typically plagued by the lack
of reliable ``control samples'' against which the active galaxies
should be compared. This problem is especially acute for
cosmologically distant sources since it is not easy to identify the
morphological types of the galaxies (e.g.
Stockton 1990). Detailed
studies of individual objects, on the other hand, provide direct
evidence as to the import of interactions in select cases, but provide
little basis for generalization. Even so, such ``case'' studies seem
more persuasive than statistical analyses in view of
the difficulties associated with establishing definitive control
samples.
Global Starbursts
In a classic paper,
Larson & Tinsley (1978)
demonstrated that the
peculiar galaxies in the Arp (1966) atlas tend to be bluer, on
average, than their isolated counterparts. The best fits to their
data invoke intense bursts of star formation throughout the peculiar
galaxies to account for their anomalous colors. These findings have
been corroborated by many subsequents studies, including those of
Joseph et al. (1984),
Kennicutt et al. (1987),
Bushouse (1987),
Telesco et al. (1988),
and Laurikainen & Moles (1989).
There is even
evidence for induced star formation between colliding galaxies
(e.g.
Thronson et al. 1989).
However, it should be emphasized that not all interacting pairs display
this behavior
(Keel et al. 1985;
Bushouse 1986;
Kennicutt et al. 1987;
Bushouse et al. 1988;
Solomon & Sage 1988).
Thus, the observational link between encounters and enhanced
star formation seems quite compelling. A likely explanation is that
interactions accelerate star formation as gas is compressed in shocks
and cloud-cloud collisions (e.g.
Young et al. 1986).
However, the
severity of these effects undoubtedly depends on the orientations of
the disks and the orbital geometry, so a starburst is not guaranteed
in all collisions.
Theoretical work on this issue remains ambiguous. Using numerical
simulation,
Noguchi & Ishibashi (1986) and
Olson & Kwan (1990)
argue that star formation is enhanced during galaxy interactions
through increased collisions between molecular clouds in disks; a
finding disputed by
Mihos et al. (1991).
In reality, the connection
between cloud-cloud collisions and star formation is poorly
understood. While the numerical simulations do demonstrate that
large-scale ``shocks'' develop in the gas as it is compressed by
tidal perturbations, no existing method possesses the dynamic range
needed to translate the measured effect into a reliable star formation
rate.
Infrared-Luminous Galaxies
The most extreme examples of starbursting objects are those where the
peculiar emission is generated mainly in the nucleus (for a summary of
the observations, see
Soifer et al.
1987). Spectacular examples
of this phenomenon were revealed by the IRAS survey: Sources having
infrared luminosities up to 1013 L , the brightest of which
invariably possess features typical of merging galaxies (e.g.
Soifer et al. 1984a,
b;
Allen et al. 1985;
Joseph & Wright 1985;
Sanders et al. 1986;
Armus et al. 1987;
Sanders et al. 1988a,
b;
Kleinmann et al. 1988).
CO observations of these objects indicate that they
often contain large quantities of gas in their nuclei. For example,
Mrk 231,
Arp 220,
and NGC 3256 appear to contain nuclear gas masses in
excess of 1010 M , confined to their
inner few hundred parsecs (e.g.
Scoville et al. 1986;
Sargent et al. 1987;
Sargent et al. 1989).
There seems little doubt that mergers are somehow responsible for the
activity in at least the brightest IRAS objects (e.g.
Sanders et al. 1988a,
b;
Sanders 1992). This hypothesis
is supported by
numerical experiments which indicate that mergers between
comparable-mass disks can create substantial nuclear gas concentrations
(Negroponte & White 1983;
Noguchi 1991;
Barnes & Hernquist 1991). Gas
accumulates in the center of each galaxy prior to the merger and
eventually the two gas concentrations sink to the center of the ensuing
remnant by dynamical friction. An example may be
Arp 220, which
appears to contain a double-Seyfert nucleus
(Norris 1988;
Diamond et al. 1989;
Graham et al. 1990; but see
Smith et al. 1988). There
are indications that many and perhaps even all of the ultraluminous
IRAS objects possess double nuclei and that the intensity of the
emission is correlated with the proximity of the nuclei
(Sanders 1992). Indeed the
apparent dearth of single-nucleus luminous
infrared galaxies is somewhat unexpected - numerical experiments
indicate that the double-nucleus phase is rather short-lived (e.g.
Barnes & Hernquist 1991) - and
there is no obvious mechanism that
would shut off the IR emission once the nuclei have merged.
Seyfert Galaxies
The most common AGN in the local universe are those which were first
studied in detail by Seyfert
(1943; for a discussion, see
Osterbrock 1991). Unlike radio
ellipticals, Seyferts are associated with
late galaxy types but, like radio galaxies, are more common in the
field than in rich clusters (e.g.
Dressler et al. 1985). A
variety of
studies hint that Seyferts tend to be found in galaxies interacting
with nearby companions (e.g.
Adams 1977;
Dahari 1984;
Kennicutt & Keel 1984;
Keel et al. 1985;
MacKenty 1989), a proposal which has
evoked controversy owing to difficulties with control samples
(Fuentes-Williams & Stocke
1988). Nevertheless, there now
appears to be a general consensus that at least some types of Seyfert
activity are correlated with galaxy collisions
(Osterbrock 1991).
There are also many indications that mergers, in addition to transient
encounters, are related to Seyfert activity. Some Seyferts display
multiple nuclei and tidal tails, characteristic of merger events
(Fricke & Kollatschny 1989;
Kollatschny & Fricke 1989). It is
probably also significant that a large fraction of Seyferts in
MacKenty's (1990) sample are
``amorphous'' or otherwise disturbed.
Numerical simulations by Hernquist
(1989a,
b;
1991a) demonstrate that
the accretion of small satellites by gas-rich disks lead to rapid
nuclear inflows of gas and leave remnants that have distorted, but
mostly featureless disks. As in mergers of comparable-mass galaxies
containing gas, these inflows can be driven by large-scale stellar
bars, excited by the decaying satellite. Moreover, this effect can
operate even if stellar bars are not generated during a merger
(Hernquist 1989a,
b). In some cases, the tidal
field of the satellite
``squeezes'' gas orbits in the disk, leading to rapid dissipation as
streamlines intersect. If the gas is self-gravitating it can
fragment, and the blobs of gas left over will sink to the center of
the disk by shedding angular momentum to surrounding stars via
dynamical friction. It is natural to suppose that events such as
these can simultaneously trigger Seyfert activity and leave remnants
with global morphologies similar to those of the galaxies in
MacKenty's sample. However, the fate of the central gas in the models
is problematic and there is little hard observational evidence to
support an evolutionary connection between starbursts and Seyfert
activity (e.g.
De Robertis & Shaw 1988).
A variety of models suggest that transient encounters can also drive
gas to the centers of galaxies (e.g.
Byrd et al. 1987;
Noguchi 1987;
1988;
Lin et al. 1988).
However, these calculations either ignore the
effects of dissipation or are not fully self-consistent. Judging
from the simulations of
Barnes (1988) and
Barnes & Hernquist (1991)
it is not inconceivable that the violent interactions needed to
provoke nuclear inflows of gas would lead to mergers and complete
destruction of the disks involved. The remnants of such events do not
resemble Seyferts but are instead similar to the elliptical and
peculiar galaxies which harbor starbursts and radio sources. These
issues will not be resolved until detailed parameter surveys are
available which include the full self-gravity of all components. It
should also be noted that not all Seyfert activity is triggered by
interactions and that some other fueling mechanism, such as
bar-induced inflow
(Simkin et al. 1980),
may be more widespread than encounters of galaxies.
QSOs & Quasars
There is considerable indirect evidence to support the conjecture that
quasar activity is triggered by galaxy collisions. Studies of the
morphological structure of quasar hosts by several groups (e.g.
Gehren et al. 1984;
Hutchings et al. 1984;
Malkan et al. 1984;
Smith et al. 1986)
indicate that a large fraction of these
galaxies are disturbed. Among low-redshift quasars, 70% or more have
nearby companions
(French & Gunn 1983;
Heckman et al. 1984;
Vader et al. 1987;
Hutchings et al. 1989),
some possess features reminiscent of tidal
tails
(Stockton 1978;
MacKenty & Stockton 1984;
Stockton & Ridgway 1991), and
still others appear to be linked to neighboring galaxies by
bridges or display evidence for multiple nuclei
(Stockton & MacKenty 1987;
Hutchings et al. 1988;
Block & Stockton 1991). In addition,
there are compelling arguments indicating that many of the more
luminous IRAS galaxies are, in fact, ``buried quasars,'' or systems
which will eventually evolve into quasars (e.g.
Sanders et al. 1990).
As noted above, the brightest of these systems are invariably
associated with mergers.
In general, theoretical studies on the relevance of mergers to quasar
activity mainly comprise wishful thinking. While calculations like
those by
Negroponte & White (1983),
Noguchi (1988), and
Barnes & Hernquist (1991)
demonstrate that tidal forces can drive gas into the
nuclei of interacting galaxies, they do not predict the ultimate fate
of the gas and so cannot determine if an AGN will develop. Moreover,
it is not clear if the galaxy models used by these workers, which are
analogues of present-day disks, are even reasonable caricatures of
quasar hosts. Nevertheless, if collisions between galaxies and an
abundant supply of gas are necessary ingredients for forming quasars,
then it is expected that AGN should be more abundant at high redshifts
than in the local universe since interactions were more frequent then
and, presumably, more free gas was available than today (e.g.
Yee & Green 1984,
1987;
Roos 1985).
Radio Galaxies
The most convincing observations linking interactions with radio
activity in ellipticals are studies by
Heckman et al. (1986) and
Vader et al. (1989)
implying that many radio galaxies are the products of
recent mergers. These objects display structural irregularities often
associated with merger candidates, including dust lanes, tails,
bridges, shells, and double nuclei (see, also,
Schweizer 1980;
van Albada et al. 1982;
de Ruiter et al. 1988;
Smith & Heckman 1989).
Others display stellar disks that may be debris from a merger (e.g.
Gonzalez-Serrano & Perez-Fournon
1989). Moreover, some show
evidence of recent or ongoing star formation and unusually high levels
of infrared emission, suggesting that they are relatively young. It
is quite natural, therefore, to speculate that these galaxies formed
by mergers of disk progenitors which also triggered the observed
activity.
The numerical studies reviewed in Section 5
indicate that merger remnants
are morphologically similar to elliptical galaxies and that if the
galaxies involved are gas-rich a substantial fraction of the gas may
be fall into the nuclei of merger remnants. In this regard, it is
interesting to note that the kinematics of radio ellipticals are quite
similar to those of normal ones
(Heckman et al. 1985;
Smith et al. 1990)
and that many merger candidates exhibit r1/4 law luminosity
profiles (e.g.
Schweizer 1982;
Wright et al. 1990).
The calculations
lack sufficient dynamic range to determine the evolution of the gas on
scales much smaller than 100 pc, but a variety of
phenomenological arguments suggest that the gas should continue to
fragment and contract on smaller scales and may eventually form a
black hole or be accreted by an existing one
(Begelman et al. 1984;
Norman & Scoville 1988;
Scoville & Norman 1988;
Shlosman et al. 1989).