|Annu. Rev. Astron. Astrophys. 1996. 34:
Copyright © 1996 by Annual Reviews Inc. All rights reserved
6.1 Formation of Ellipticals
In disk-disk collisions of galaxies, dynamical friction and subsequent relaxation may produce a mass distribution similar to that in classic elliptical galaxies. From the relative numbers of mergers and ellipticals in the New General Catalogue Toomre (1977) estimated that a large fraction of ellipticals could be formed via merging. The first direct observational evidence for the transition from a disk-disk merger toward an elliptical was presented in the optical study of NGC 7252 by Schweizer (1982). The brightness distribution over most of the main body of this galaxy is closely approximated by a de Vacouleurs (r-1/4) profile. However, NGC 7252 still contains large amounts of interstellar gas and exhibits a pair of prominent tidal tails (see Figure 9b); neither property is typical of ellipticals.
Near-infrared images are less effected by dust extinction and also provide a better probe of the older stellar population, which contains most of the disk mass and therefore determines the gravitational potential. K-band images of six mergers by Wright et al. (1990) showed that the infrared radial brightness profiles for two LIGs - Arp 220 and NGC 2623 - follow an r-1/4 law over most of the observable disks. Among eight merger remnants, Stanford & Bushouse (1991) found K-band brightness profiles for four objects that were well fitted by an r-1/4 law over most of the observable disks. Kim (1995) finds a similar proportion (~ 50%) of ULIGs whose K-band profiles are well fit by a r-1/4 law.
More recently, Kormendy & Sanders (1992) have proposed that ULIGs are elliptical galaxies forming by merger-induced dissipative collapse. The extremely large central gas densities (~ 102-103 M pc-3) observed in many nearby ULIGs (see Section 4.6.2) and the large stellar velocity dispersions found in the nuclei of Arp 220 and NGC 6240 (Doyon et al. 1994) are comparable to the stellar densities and velocity dispersions respectively, in the central compact cores of ellipticals.
Despite the K-band and CO evidence that LIGs may be forming ellipticals, we still need to account for two important additional properties of ellipticals: 1. the large population of globular clusters in the extended halos of elliptical galaxies, which cannot be accounted for by the sum of globulars in two preexisting spirals (van den Bergh 1990), and 2. the need to remove the large amounts of cold gas and dust present in infrared-luminous mergers in order to approximate the relative gas-poor properties of ellipticals. These two issues are addressed below.
6.2 Formation of Star Clusters
Populations of bright blue pointlike objects have recently been discovered with HST in the galaxy at the center of the Perseus cluster, NGC 1275 (Holtzman et al. 1992), and in the disk-disk mergers NGC 7252 (Whitmore et al. 1993), ``The Antennae'' (Whitmore & Schweizer 1995), and Arp 299 (Meurer et al. 1995, Vacca 1996). These objects have Mv ~ -11 to -16, and it has been proposed that they are young clusters that are, or may evolve into, globular clusters. If this hypothesis is true, then the number of globular clusters would indeed increase during the merger of gas-rich spirals, thus weakening one of the main arguments against ellipticals being formed through disk mergers.
The interpretation of these star clusters as protoglobulars has been questioned by van den Bergh (1995a, b), who argues that mergers simply increase the rate of normal star and cluster formation and do not promote a specific population of very massive clusters that will evolve into globulars. In fact, the luminosity function of the blue clusters in The Antennae has a power-law shape as do open clusters in the Milky Way, rather than the Gaussian shape of the luminosity function of globular clusters (Whitmore & Schweizer 1995, van den Bergh 1995b; but see Meurer 1995). Furthermore, the light radii of the young clusters in The Antennae seem to be larger than in typical globular clusters by a factor of about three. Future observations of these objects with corrected HST optics may provide better images that can prove if these blue clusters indeed have the star densities typical of globulars.
6.3 Formation of Dwarf Galaxies
Collisions between giant disk galaxies may trigger the formation of dwarf galaxies. This idea, which was first proposed by Zwicky (1956) and later by Schweizer (1978), has received recent observational support (Mirabel et al. 1991, 1992; Elmegreen et al. 1993; Duc & Mirabel 1994). Renewed interest in this phenomenon arose from the inspection of the optical images of ULIGs, which frequently exhibit patches of optically emitting material along the tidal tails (see Figures 8a, b, c). These objects appear to become bluer near the tips of the tails at the position of massive clouds of H I. These condensations have a wide range of absolute magnitudes, MV ~ -14 to -19.2, and H I masses, M(H I) ~ 5 x 108 to 6 x 1099 M. Mirabel et al. (1995) have shown that objects resembling irregular dwarfs, blue compacts, and irregulars of Magellanic type are formed in the tails. These small galaxies of tidal origin are likely to become detached systems, namely, isolated dwarf galaxies. Because the matter out of which they are formed has been removed from the outer parts of giant disk galaxies, the tidal dwarfs we observe forming today have a metallicity of about one third solar (Duc 1995).
It is interesting that in these recycled galaxies of tidal origin there is - as in globular clusters - no compelling evidence for dark matter (Mirabel et al. 1995). The true fraction of dwarf galaxies that may have been formed by processes similar to the tidal interactions we observe today between giant spiral galaxies will require more extensive observations of interacting systems. A recent step forward is the statistical finding that perhaps as much as one half of the dwarf population in groups is the product of interactions among the parent galaxies (Hunsberger et al. 1996).
6.4 Enrichment of the Intergalactic Medium
X-Ray and optical evidence for galactic superwinds has been presented for both ULIGs and PRGs (Heckman et al. 1990, 1996; Veilleux et al. 1995). In these objects, it has been proposed that the combined kinetic energy from supernovae and stellar winds from powerful nuclear starbursts drive large-scale outflows that shock, heat, and accelerate the circumnuclear ambient gas. The morphology, kinematics, and physical properties of the optically emitting gas tend to support this model. Continuum-subtracted narrow-band images show emission-line nebulosity extending over tens of kiloparsecs; a good example of this phenomenon is the enormous H bubble found in Arp 220 (Armus et al. 1987). Further kinematic signatures of outflows along a galaxy's minor axis are provided by observed double emission-line profiles with line splittings of 200-600 km s-1; a spectacular example is the 1500 km s-1 splitting found in the nuclear superbubble of NGC 3079 (Veilleux et al. 1994). Additional evidence for mass outflows in LIGs includes the statistical difference found between the H I 21-cm line and optical emission line redshifts. This difference is likely due to outflow motions of the optical line-emitting gas (Mirabel & Sanders 1988).
Galactic superwinds may play an important role in the metal enrichment of the intergalactic medium. Heckman et al. (1990) calculate a mass loss rate, dM/dt = 4 (Lfir, 11 M year-1), where Lfir, 11 is the far-infrared luminosity in units of 1011 L, implying that a galaxy like Arp 220 can be expected to inject ~ 5 x 108 M of metals and ~ 1058 ergs over an estimated lifetime of ~ 107 years. Assuming no cosmological evolution and using a luminosity function for IRAS galaxies similar to that given in Figure 1, Heckman et al. (1990) derived a total mass-injection rate of ~ 2 x 107 (M Mpc-3), 25% of which is in metals. If cosmological evolution is included (see Section 3.1), then the injected mass and energy can increase by an order of magnitude.
The superwinds observed in nearby ULIGs may provide local examples that can be used to understand the X-ray iron abundance of ~ 1/3 solar in the intracluster medium (Arnaud et al. 1992). Recent ASCA observations of large amounts of silicon and oxygen in four nearby clusters indicate that type II supernovae are responsible for the metal enrichment (Loewenstein & Mushotzky 1996). In turn, the metals found in nearby clusters may represent the fossil records of ancient starbursts (Elbaz et al. 1992, Terlevich & Boyle 1993). More specifically, the iron-elliptical correlation observed in clusters may suggest that giant ellipticals indeed went through an ULIG-superwind phase.
On the question of the impact that these galactic superwinds may have on the interstellar gas content of the host galaxy, it is important to know the fraction of the total interstellar gas that these winds can entrain and expel into the intergalactic medium. Heckman et al. (1990) suggest that the maximum amount of ambient material ejected by the wind is Mej,max = 1010 (Ewind / 1058) M, which implies that when the entrained material has velocities above the escape velocity of the host galaxy, the entire interstellar medium could be blown away. Since the observed velocities of the line-emitting gas in LIGs are close to the escape velocities, it appears possible that a large fraction of the interstellar gas may be driven out of the host galaxy, and that in some cases the entire neutral intersteller medium may have been lost, as appears to be the case in most giant ellipticals.