|Annu. Rev. Astron. Astrophys. 1997. 35:
Copyright © 1997 by . All rights reserved
DESTRUCTION OF DISKS BY MERGERS The current paradigm of structure formation in the universe is the hierarchical clustering of dominant dissipationless dark matter; galaxies as we see them form by the dissipation of gas into the potential wells of the dark matter, with subsequent star formation (e.g. Silk & Wyse 1993). The first objects to collapse under self-gravity are the highest density perturbations on scales which are characteristic of dwarf galaxies, and globular clusters, though globular clusters seem, on chemical evolution grounds, not to be the first objects to have formed. Large galaxies form by the merging of many smaller systems. The merging rate of the dissipationless dark haloes is reasonably straightforward to calculate (e.g. Lacey & Cole 1993). Unfortunately, many badly understood parameters are involved in the physics of gaseous heating/cooling and star formation, which determine how the baryonic components evolve. In the absence of understanding, the naive separation of different stellar components of galaxies is achieved by the following prescription (Baugh et al 1996, Kauffmann 1996): Star formation occurs in disks, which are destroyed during a merger with a significantly larger companion, with "significant" meaning a free parameter to be set by comparison with observations. In such a merger, all the extant "disk" stars are reassigned to the "bulge," the cold gas present is assumed to be driven to the center and fuel a burst of star formation, and a new disk is assumed to grow through accretion of intergalactic gas. Ellipticals are simply bare bulges, which are more likely in environments that prevent the subsequent reaccretion of a new disk - environments such as clusters of galaxies (e.g. Gunn & Gott 1972). One consequence (see Kauffmann 1996) of this prescription is that late-type spirals, which have a large disk-to-bulge ratio, should have older bulges than do early-type spirals, since to have a larger disk the galaxy must have been undisturbed and able to accrete gas for a longer time. This does not appear compatible with the observations discussed above. Bulge formation is highly likely to be more complex than this simple prescription.
ACCRETION OF DENSE STELLAR SATELLITES The central regions of galaxies are obvious repositories of accreted systems, as they are the bottom of the local potential well, provided that the accreted systems are sufficiently dense to survive tidal disruption while sinking to the center (e.g. Tremaine et al 1975). Should the accreted systems be predominately gaseous, then the situation is simply that described by Eggen et al (1962), with the chemical evolution modified to include late continuing infall. [It is worth noting that late infall of gas narrows resulting chemical abundance distribution functions (e.g. Edmunds 1990), and at least the Milky Way bulge has an observed very broad distribution.] We now consider models of bulge formation by accretion of small stellar systems.
As discussed above, the mean metallicity of the Galactic bulge is now reasonably well established at [Fe/H] ~ -0.3 dex (McWilliam & Rich 1994, Ibata & Gilmore 1995b, McWilliam 1997), with a significant spread below -1 dex and above solar. Thus, satellite galaxies that could have contributed significantly to the bulge are restricted to those of high metallicity. Given the fairly well-established correlation between mean metallicity and galaxy luminosity/velocity dispersion (e.g. Bender et al 1993, Lee et al 1993, Zaritsky et al 1994), only galaxies of luminosity comparable to the bulge can have been responsible. That is, one is immediately forced to a degenerate model, in which most of the stellar population of the bulge was accreted in one or a few mergers of objects like the Magellanic Clouds or the most luminous dwarf spheroidals (dSph). Because the metallicity distribution of the bulge is very broad, significantly broader than that of the solar neighborhood, a compromise model is viable, in which only the metal-poor tail of the bulge abundance distribution function has been augmented by accretion of lower luminosity satellite galaxies. Quantification of this statement awaits more robust measurement of the tails of the bulge metallicity distribution function and of appropriate element ratios.
Limits on the fraction of the bulge that has been accreted can be derived from stellar population analyses, following the approach utilized by Unavane et al (1996) concerning the merger history of the Galactic halo.
The Sagittarius dSph galaxy was discovered (Ibata et al 1994) through spectroscopy of a sample of stars selected purely on the basis of color and magnitude to contain predominantly K giants in the Galactic bulge. After rejection of foreground dwarf stars, the radial velocities isolated the Sagittarius dwarf galaxy member stars from the foreground bulge giants. The technique (serendipity) used to discover the Sagittarius dSph allows a real comparison between its stellar population and that of the bulge. Not only the radial velocities distinguish the dwarf galaxy, but also its stellar population - as seen in Figure 8 (taken from Ibata et al 1994), all giant stars redder than BJ - R 2.25 have kinematics that place them in the low velocity-dispersion component, i.e. in the Sagittarius dwarf. This is a real quantifiable difference between the bulge field population and this, the most metal-rich of the Galactic satellite dSph galaxies.
Figure 8. Heliocentric radial velocities of the sample of stars observed by Ibata et al (1994), towards = -5° b = -12°, -15°, and -20°. The stars with velocities less than about 120 km/s are predominately bulge K giants. Those with velocities between about 120 and 180 km/s are members of the Sagittarius (Sgr) dSph galaxy, which was discovered from this figure. Note the real difference between the color distributions of bulge and Sgr members. Thus, the bulge cannot be built up by merger of several galaxies like the Sgr dwarf.
Furthermore, the carbon star population of the bulge can be compared with those of typical extant satellites. In this case, there is a clear discrepancy between the bulge and the Magellanic Clouds and dSph (Azzopardi & Lequeux 1992), in that the bulge has a significantly lower frequency of carbon stars.
Thus, although accretion may have played a role in the evolution of the bulge of the Milky Way, satellite galaxies like those we see around us now cannot have dominated. However, accretion is the best explanation for at least one external bulge - that of the apparently normal Sb galaxy NGC 7331, which is counter-rotating with respect to its disk (Prada et al 1996). It should also be noted that for S0 galaxies - those disk galaxies that at least in some models have suffered the most merging - Kuijken et al (1996) have completed a survey for counter-rotating components in the disks and found that only 1% of S0 galaxies contain a significant population of counter-rotating disk stars. This is a surprisingly low fraction and suggests some caution prior to adopting late merger models as a common origin of early-type systems.