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

4. ACQUISITIONS & ACCRETIONS

Galaxies can acquire material from passing companions via mass transfer or accrete much less-massive neighbors, such as satellites, suffering only relatively minor damage. Although individual events like these add only a modest amount of mass to large galaxies, their consequences can have a significant effect on observable properties of galaxies and provide indications as to the frequency and import of galaxy collisions.

Shells & Other Fine Structures

Beginning with Arp's (1966) classic study of peculiar galaxies, ``fine structures,'' such as ``shells,'' ``ripples,'', ``plumes,'' boxy isophotes, and ``X-features'' have been detected in an ever-larger fraction of S0's and ellipticals (Malin & Carter 1980, 1983; Schweizer 1980). Indeed, in the most recent survey by Seitzer & Schweizer (1990), 32% of their S0's and 56% of their E's possess shells and ~ 10% and ~ 30% of their entire sample have X's and boxy isophotes, respectively. The actual percentage of all galaxies exhibiting such fine structure is rather uncertain, owing to the difficulty of determining if the faint features detected are bona fide. This complication is especially acute for later-type disks, which often possess diffuse spiral patterns that mimic, e.g. shells. However, there are observational indications that shells exist in at least some Sa and Sb galaxies (e.g. Schweizer & Seitzer 1988).

Simulations employing restricted methods support the view that the accretion of material by large galaxies provides a natural explanation for the origin of fine structures. Shells can form either by the ``phase-wrapping'' of debris on nearly radial orbits (Quinn 1984) or by the ``spatial-wrapping'' of matter in thin disks (Hernquist & Quinn 1988). Contrary to Quinn's (1984) hypothesis, the shell-forming material need neither reside initially in a thin disk nor be on precisely radial orbits. Shells can be produced during the accretion of spheroidal companions by larger galaxies, provided that the total phase-space volumes they occupy are sufficiently different (Dupraz & Combes 1986; Hernquist & Quinn 1988), and in non-radial collisions through mass transfer, eliminating concerns that an improbable encounter geometry would be needed to produce them (Toomre 1983; Hernquist & Quinn 1988).

These various studies have also shown that other types of fine structure develop in encounters which simultaneously produce shells. Plumes consisting of debris on unbound or weakly-bound orbits occur frequently in shell-forming collisions (Hernquist & Quinn 1988). Moreover, material accreted by non-spherical potentials can produce X-structures and features that appear boxy in projection (Hernquist & Quinn 1989).

In spite of the general acceptance of the accretion model, a number of difficulties remain. The numerical studies mostly ignore self-gravity and the consequences of dynamical friction. While the self-gravity of the victim does not appear crucial to the issue of whether or not fine structures will form (Barnes 1989; Heisler & White 1990; Salmon et al. 1990), it is impossible to predict, e.g. the radial distribution of shells formed by a close collision without including dynamical friction (Dupraz & Combes 1987). Indeed, if the accretion scenario is correct, it appears that dynamical friction probably played a significant role in structuring some of these objects since many contain extremely tightly bound shells. Self-consistent calculations by Heisler & White (1990) indicate that the disruption of satellites can indeed inject material onto tightly bound orbits, populating inner shells. However, even these models are not sufficiently detailed to explain objects such as NGC 3923, which contains more than 20 shells distributed regularly around the galaxy.

As noted by Thomson & Wright (1990), the restricted simulations are deficient in other regards. In some galaxies, the intensity of shells increasingly far from the galaxy drops similarly to that of the elliptical, suggesting that these shells may be of internal origin. Accordingly, Thomson & Wright have put forward an alternate model in which shells are induced in a thin disk component of an elliptical by a tidal encounter with another galaxy. The existence of such disks in ellipticals is problematic, but the inability of restricted calculations like those of Hernquist & Quinn (1988) to explain the radial luminosity of shell systems re-emphasizes the need for more physical models.

It should also be noted that boxy or X-shaped structures need not arise through accretion events; these features simply reflect phase-mixed populations of stars in certain kinds of orbits which may have been created by a variety of processes (Binney & Petrou 1985). An example is provided by the recent work of Merritt & Hernquist (1991) who show that dynamical ``bending'' instabilities produce remnants which are photometrically similar to ellipticals having boxy isophotes and weak X-structures. These models do not display shells, however, so a strong correlation between all these various features would support the view that they are of external origin.

Polar Rings

Many S0 galaxies possess rings of gas-rich material that appear to be kinematically distinct from the galaxy proper (e.g. Schweizer et al. 1983; Athanassoula & Bosma 1985). The recent survey by Whitmore et al. (1990) implies that ~ 0.5% of all S0's have polar rings; i.e., rings of gas, dust, and young stars with axes aligned perpendicular to the major axis of the disk. The actual fraction of S0's with such features may be even larger when observational selection effects and ring lifetimes are taken into account.

Searches for similar features in elliptical galaxies have met with mixed success. Obscuring dust lanes (e.g. Bertola 1987) and diffuse, extended disks and rings of gas are seen in many ellipticals (e.g. van Gorkom et al. 1986; Lake et al. 1987; Kim et al. 1988; Schweizer et al. 1989), but few ellipticals with rings as substantial as those of classical polar rings in S0's are known. A notable exception is the galaxy AM2020-5050 (Whitmore et al. 1987), which is classified as Hubble type E4. However, this object rotates rapidly and appears somewhat intermediate between an elliptical and a S0 galaxy.

Since these rings often appear kinematically distinct from their hosts, it is generally assumed that they are of external origin. Two scenarios have been proposed for the formation of polar rings: accretion of gas-rich dwarfs (e.g. Athanassoula & Bosma 1985) and capture of material from a passing galaxy by mass transfer. An example of the latter is NGC 3808, where a stream of material can be seen linking a large disk galaxy to an S0 of comparable luminosity as shown in Figure 1. Unfortunately, neither scenario has been tested in detail. Theoretical studies of polar rings have instead focussed mainly on determining the equilibrium states accessible to material on closed orbits in axisymmetric and triaxial potentials. These calculations have shown that the ``settling time,'' which is the time for gas in an axisymmetric or triaxial potential to settle into steady state, is small compared to the age of the universe (Tohline et al. 1982; Steiman-Cameron & Durisen 1982, 1988; Habe & Ikeuchi 1985, 1988; Varnas et al. 1987), and so there is sufficient time for accreted gas to form polar rings.

However, all these studies are limited by their neglect of self-gravity and time-dependent effects and their use of unrealistic initial conditions. The first two difficulties will likely be overcome by Lagrangian codes, such as TREESPH (Hernquist & Katz 1989) which explicitly include gravitational forces between mass elements. Some preliminary steps in this direction have already been taken by Rix & Katz (1991). The role of initial conditions in ring-forming events is not likely to be elucidated without resort to calculations which treat the dynamics of galaxy collisions and mergers in detail, including dynamical friction and dissipation. Such models may eventually determine why rings are most common in S0 galaxies, largely avoiding other galactic types, and also explain why fine structures such as shells should be more common than rings if all result from accretion of external matter.

Sinking Satellites

As emphasized by Ostriker & Tremaine (1975) and Tremaine (1981), dynamical friction will cause the orbits of satellites around massive galaxies to decay. Eventually, the victims will be destroyed tidally and/or cannibalized by the primary galaxy. Using Chandrasekhar's treatment of dynamical friction, Tremaine showed that galaxies like the Milky Way have probably accreted a non-negligible amount of mass in the form of discrete objects over a Hubble time. Although it is difficult to make precise estimates of this effect, owing to the absence of a complete theory of galaxy formation and large-scale structure, Tremaine's simple argument implies that galaxies do not age peacefully in isolation, even if they do not experience a strong encounter with a comparably-massed neighbor.

In the limit of weak encounters, perturbation theory can be used to compute torques on satellites and predict their decay paths (Tremaine & Weinberg 1984; Weinberg 1986; Statler 1988). More generally, there is little alternative to numerical simulation, particularly if the details of the tidal disruption of the companion and/or the self-consistent response of the primary to large perturbations are of interest. Numerical studies have included the orbital decay of satellites around spherical galaxies (Lin & Tremaine 1983; White 1983a) and disks (Quinn & Goodman 1986; Quinn et al. 1991). The results are supported by quasi-analytic calculations, implying that simulations yield reliable decay rates (Hernquist & Weinberg 1989). However, existing studies have typically approximated the self-consistent response of the primary, complicating their interpretation. For example, while Quinn et al. (1991) include the self-gravity of primary disks, they ignore the self-consistent response of the halo. As emphasized by Barnes (1988), orbital decay in composite systems involves a complex interplay amongst the various components. Fully self-consistent calculations by Hernquist (1992) support this claim and show that the torque on a satellite near a disk is derived in roughly equal measure from the disk and halo; provided, of course, that galaxies possess dark matter halos similar to those inferred in external spirals.

Technical issues aside, it seems likely that a variety of apparently disparate phenomena may be blamed on satellite accretions. Some ellipticals contain cores which are kinematically distinct from the surrounding galaxies (e.g. Efstathiou et al. 1982; Franx & Illingworth 1988; Franx et al. 1989; Jedrzejewski & Schechter 1988; Wagner et al. 1988; Bender 1990a). One possible explanation for these peculiar velocity fields was first proposed by Kormendy (1984) who argued that the central rotation of NGC 5813 is a consequence of the accretion of a spinning dwarf galaxy. This idea was tested by Balcells & Quinn (1990) using N-body simulations. In their models, angular momentum is transferred from the orbit of the dwarf to the primary by gravitational torques during a merger. If the satellite's orbit is retrograde, the angular momentum deposited can partly cancel the primary's original rotation and even reverse it near the center where the rotation curve is rising. Following mergers from retrograde orbits, therefore, the central regions of the primary can display distinct velocity fields from the outlying regions. Although this specific scenario is not the only way to produce kinematically decoupled cores, the general idea that angular momentum deposited by satellite accretions can ``organize'' the shapes and velocity fields of elliptical galaxies seems promising (e.g. Quinn et al. 1990).

Satellite mergers may also alter structure of disk galaxies (for a review, see Hernquist 1991a). The self-consistent models of Quinn et al. (1991) indicate that decaying satellites can excite transient warps in the outer parts of disks that may be sufficiently long-lived to explain some stellar warps in external spirals. Moreover, dynamical heating by these satellite accretions produce disturbed, featureless disks possessing little or no large-scale spiral structure (Hernquist 1989a, b), reminiscent of amorphous galaxies (Sandage & Brucato 1979). Tidal debris from cannibalized satellites may account for a number of peculiar features in our Galaxy, including its two-component system of globular clusters (e.g. Freeman 1990), retrograde halo stars, and the Magellanic Stream (e.g. Lin & Lynden-Bell 1982).

More generally, the consequences of satellite accretions are important for theories of the formation of galaxies and large-scale structure. Schweizer (1990) has proposed that the bulges of spiral galaxies may be relics of past accretion events. A similar suggestion has been made for thick disk components of external, edge-on systems (for a discussion, see Quinn et al. 1991). Such processes may be responsible for many aspects of galactic structure that were once attributed to events at the time galaxies were born. If so, then it is impossible to relate the observed properties of galaxies to cosmological models without considering late and on-going evolutionary effects. Indeed, as noted recently by Toth & Ostriker (1992) the existence of cold, thin disk components in spiral galaxies limits the rate of satellite accretion in cosmological models. Whether or not these constraints can be applied to test actual scenarios for the formation of large-scale structures remains to be seen.