Annu. Rev. Astron. Astrophys. 1992. 30: 705-742 Copyright © 1992 by Annual Reviews. All rights reserved

7. COSMOLOGICAL CONNECTIONS

With the discovery that collisions and the peculiar behavior they induce generally represent but short episodes in the life of a galaxy came the recognition that for each spectacularly interacting system we see there must be many which have gone through such phases in the past (e.g. TT). This realization encourages a search for evidence that galactic interactions have been at work in creating some of the structures we observe today; a search which might logically begin by examining how the scars of violent encounters fade with time.

Aging Merger Remnants

Intermediate-age merger remnants are expected to be plentiful, and the most youthful examples not difficult to find; Toomre (1977) listed a half-dozen ongoing mergers with but a single conspicuous center each. One of the best-studied of these is NGC 7252 (Schweizer 1982). This galaxy, shown in the last panel of Figure 1, sports a pair of well-defined tidal tails which extend from a single, messy-looking ellipsoidal body. Photometrically, it exhibits an r^1/4-law luminosity profile with numerous loops, ripples, and plumes superimposed (e.g. Schweizer 1990). Strong Balmer lines indicate the presence of a substantial number of A-type stars in the main body of the system. Raw materials for star formation are present in the form of a central counter-rotating disk of ionized gas (Schweizer 1982) and in a substantial amount of molecular material (Dupraz et al. 1990).

NGC 7252 is naturally explained as the merger of two comparable disk galaxies. Age estimates based on the kinematics of the tails and on the colors of the main galaxy indicate that this merger took place ~ 10⁹ yr ago (Schweizer 1982). Self-consistent simulations, while not specifically intended to model NGC 7252, show that pairs of merging disk/halo galaxies may come to resemble this system. By the stage roughly corresponding to the present epoch, material from the tidal tails is falling back into the galaxy on two fairly cold and organized streams; spatial and phase-wrapping of this material can produce shells, loops, and plumes (Barnes 1992) much like seen in NGC 7252. Likewise, models including gas dynamics indicate that counter-rotating or otherwise distinctive gas kinematics should be fairly common in mergers of comparable disk galaxies (Hernquist & Barnes 1991).

Further examples are not wanting; NGC 3921 (Schweizer 1978; 1990), Mrk 231 (Hamilton & Keel 1987), and ESO 341-IG04 (Bergvall et al. 1989) all exhibit extended tails, r^1/4 luminosity profiles, shells and other fine structures, and Balmer absorption spectra; these systems seem similar in age and gas content to NGC 7252. On the other hand, NGC 6776 (Sansom et al. 1988) features but one well-defined tidal tail and seems to be poor in gas, dust, and young stars; this system may well have originated in a merger between an elliptical and a smaller, early-type disk galaxy.

Older merger remnants apparently lurk among those elliptical and S0 galaxies with fine structures - such as ripples, plumes, boxy isophotes, or X-shapes. Schweizer et al. (1990) find that such galaxies tend to have stronger H absorption and weaker CN and Mg indices than average galaxies of the same luminosity; they propose that the galaxies responsible for these correlations are merger remnants with ages greater than ~ 10⁹ yr but considerably less than 10¹⁰ yr. Boxy elliptical galaxies are also louder than average in the radio and X-ray bands (e.g. Bender et al. 1989), and Bender (1988) and Nieto & Bender (1989) have suggested that these galaxies are merger remnants with intermediate ages. The increasingly large fraction of elliptical galaxies which exhibit ever-fainter scars of past interactions lends much support to the idea that merger remnants blend into the population of normal ellipticals as they age.

Sites of Interactions

Evidence that galactic mergers play a significant cosmological role comes from a simple demographic argument presented by TT and refined by Toomre (1977). Out of the ~ 4000 galaxies in the NGC catalog, there are at least a dozen which are either close pairs of disk galaxies strongly interacting or recent merger remnants with prominent double tails. Adopting a nominal age of 5 x 10⁸ yr for these objects, Toomre (1977) concluded ``we should expect to find roughly 250 old relics of mergers among the NGC systems alone, provided that the present rate of those intense encounters is at all typical of the 10-15 billion years that galaxies have existed.'' In fact, the present merger rate probably underestimates the average rate over the past ~ 10¹⁰ yr. The pairs we find in violent interactions and mergers today presumably spent most of the last 10¹⁰ yr loitering near apogalacticon, and have only recently fallen back together (TT). Almost any reasonable assumption for the distribution of binding energies for an ensemble of such pairs yields a merger rate which declines with time (e.g. Toomre 1977).

Besides the relatively isolated pairs which appear to account for most of the objects listed by Toomre (1977), violently interacting galaxies are also found in other settings. Galaxies in compact groups frequently exhibit strong tidal distortions (e.g. Rose 1979; Hickson 1982) and kinematic peculiarities (Rubin et al. 1991). Numerical simulations indicate that such groups experience on the order of one merger per crossing time as a result of low-velocity encounters (e.g. Barnes 1989). This view gains further support from observations indicating that ~ 6% of compact group members have colors characteristic of recent merger remnants (Zepf & Whitmore 1991). Such mergers might not be as easily recognized as those involving isolated pairs since the tidal forces of other group members tend to shred extended tidal tails. It remains difficult to estimate how many merger remnants are being produced in compact groups, largely because characteristic lifetimes are not well known for many of these systems (e.g. White 1990).

Rich clusters provide another possible setting for interactions and mergers. Initially, attention focused on ``cannibalism'' as a mechanism for forming the extremely luminous galaxies found at the centers of rich regular clusters (Lecar 1975; Ostriker & Tremaine 1975; White 1976). In these early models, dynamical friction was invoked to bring massive galaxies into the core of the cluster, where they would merge to form a D or cD galaxy. However, more detailed studies showed that the victim galaxies would be shorn of much of their halo mass by the cluster's tidal field, and therefore would not spiral in rapidly enough to contribute much more than ~ 10% of the luminosity of the central giant (Merritt 1985 and references therein; see also Malumuth & Richstone 1984). This result has been substantiated by observational studies (Tonry 1984; Merritt 1984; Lauer 1986, 1988) which show that most of the ``secondary'' nuclei in cD galaxies are merely passing through with velocities typical of the cluster as a whole. Such high-speed collisions also occur elsewhere in rich clusters and their effects can be studied using analytic approximations (e.g. Gerhard & Fall 1983); in general, however, it seems unlikely that these collisions cause significant damage to a large number of galaxies or contribute a substantial amount of stripped luminosity to a cluster-wide background.

Galactic interactions may have played a more important role during the formation of rich clusters. Within the context of ``hierarchical clustering'' models for the growth of large-scale structure (e.g. White & Rees 1978; Peebles 1980), rich clusters are expected to form by the amalgamation of smaller ones. Before collapsing, such a system probably resembles a hierarchical federation of compact groups. Such systems are favorable sites for violent interactions since most of the galaxies reside in pockets of substructure with relatively high densities and low velocity dispersions, and a glance at a photograph of the Hercules cluster reveals many interacting galaxies. A toy model for the dynamical evolution of a hierarchy of 128 core/halo galaxies illustrates how such a system passes through stages resembling irregular clusters before relaxing to form a regular, centrally concentrated cluster with a substantial population of merger remnants (Barnes 1991). It remains to be shown that more realistic initial conditions can produce enough mergers to account for the elliptical populations of rich regular clusters without also depositing more than 10-15 L in a central star-pile (Tremaine 1990).

These considerations suggest a plausible explanation for the overall correlation between spatial density and types of galaxies described by Dressler (1980). In this view, merger remnants are formed in regions of intermediate density which are undergoing gravitational collapse, and are ``caught up by the subsequent growth of larger-scale structure'' (e.g. Aarseth & Fall 1980; Barnes 1989). Thus the products of interactions are found in the regions of higher density than the regions where interactions are now taking place. If so then rich clusters are rubble-heaps containing nearly a Hubble time worth of galactic collisions.

Mergers and Morphology

The above arguments imply that at least a fair fraction of elliptical galaxies acquired their present forms as a result of mergers. But merging is only one process shaping galaxies, and the origin of ellipticals is only a part of the larger problem of the formation and evolution of all galaxies. For example, elliptical galaxies share many properties with the bulges of disk galaxies; it seems unlikely that this is a coincidence. Indeed, the success of numerical simulations in producing elliptical-like merger remnants can be partly attributed to the presence of fairly substantial bulges in the victim disk galaxy models. Are bulges merely ellipticals which have subsequently acquired disks, or are the central parts of ellipticals instead merged bulges dressed in the remains of their attendant disks?

Purely collisionless mergers are constrained by conservation of phase-space density; if they are to produce remnants with small cores, the victims must have had small cores to begin with. But mergers of gas-rich galaxies could develop cores of much higher phase-space densities through dissipative processes. This is indeed what appears to be happening in many starburst galaxies, and the central concentrations of gas found in these galaxies have characteristic masses and scales consistent with the cores of ellipticals. The simulations indicate that galactic collisions can ``bring deep into a galaxy a fairly sudden supply . . . of interstellar material'' (TT). Core formation from this material has many features in common with ``dissipative collapse'' pictures for galaxy formation (e.g. Kormendy 1989), but two new wrinkles need to be stressed. First, merging galaxies in our neighborhood offer a glimpse of the formation of spheroidal systems in general. Second, the large collapse factors invoked in the dissipative picture may require strong gravitational torques exerted during a merger-like process to get rid of excess angular momentum of the gas.

If this approach to galaxy formation is fruitful, then models of galactic mergers including gas dissipation and star formation might be expected to reproduce the color and metallicity profiles of elliptical galaxies (e.g. Franx & Illingworth 1990) as well as their kinematic properties. Although such models will probably have a number of free parameters reflecting our relative ignorance about the process of star formation, it need not follow that their realization would teach us nothing about events leading up to the formation of real galaxies. On the other hand, real galaxies also have features which may prove difficult to account for in any numerical model feasible with present-generation computers.

Recent observations suggest that nearby galaxies may harbor black holes with masses of 10^6.5 to 10⁹ M (e.g. Kormendy & Richstone 1992). The formation of such massive black holes is attended by a substantial release of energy which would presumably manifest as an active galactic nucleus. As noted above, there is considerable circumstantial evidence linking nuclear activity to violent interactions and mergers. If massive black holes turn out to be common features of galactic spheroids then it seems plausible that spheroidal systems and their central holes were formed as a result of mergers between gas-rich galaxies. This hypothesis has the potential to link galactic activity at redshifts of z ~ 2 with the formation of bulges and elliptical galaxies (e.g. Roos 1985; Carlberg 1990).

Perhaps one of the greatest puzzles associated with a ``dissipative merger'' picture for spheroid formation is the origin of globular clusters. van den Bergh (1990) has argued that the number of globular clusters in a galaxy is well-correlated not with total luminosity but rather with the luminosity of the spheroidal component alone. Mergers of disk galaxies, however, are expected to produce remnants with fewer globular clusters per unit luminosity than typical ellipticals. In fact, the specific frequence of globular clusters in elliptical galaxies seems to be somewhat environment dependent, and field ellipticals are perhaps no richer in globulars than typical disk galaxy merger remnants. But still unexplained are the tremendous numbers of globular clusters in galaxies like M 87 (e.g. Harris 1988); if such galaxies formed as the result of mergers, their progenitors may well have systematically differed from present-epoch disk galaxies in other respects besides globular cluster content (e.g. Efstathiou 1990).