|Annu. Rev. Astron. Astrophys. 1989. 27:
Copyright © 1989 by . All rights reserved
Two kinds of structure are commonly seen at the centers of early-type galaxies. When observed with sufficient resolution, the steep brightness profile of an elliptical usually flattens into a nearly constant surface brightness core. In addition, a nucleus is sometimes seen inside the core. By this, we mean a dynamically distinct cluster of stars that is much smaller and denser than the core. Cores are reviewed by Kormendy (K82, 1984, K87, 1987c) and Lauer (1988b); here we summarize their properties in Section 2.1. Nuclei are less well studied and understood; we review them in Section 2.2.
2.1. Summary of Core Properties
Reliable core photometry became available only when problems of seeing and photographic photometry were resolved. All photographic core photometry proves to be unreliable. CCDs do better: Profiles derived by different authors routinely agree to 0.1 mag arcsec-2 (Lauer 1985a, K87). Seeing affected early work on cores so strongly that it was not clear whether most galaxies have cores at all (Schweizer 1979, 1981b). Seeing corrections could be derived if one assumed that galaxies have cores, but the derived core parameters were model dependent (K82, Kormendy 1984). Then CCD observations by Lauer (1985a, b) found nearly isothermal cores in about a dozen ellipticals, increasing the sample of resolved cores by a factor of four. More recently, the Canada-France-Hawaii Telescope (CFHT) has provided seeing a factor of about two better than available previously; nearly isothermal cores are now resolved in almost all bright, nearby ellipticals (Kormendy 1985a, 1987c, K87).
CCD data are accurate enough for seeing corrections based on deconvolutions. Simple techniques are used by Djorgovski (1983) and Lauer (1985a, b), and more powerful techniques are discussed by Bendinelli et al. (1982, 1984a, 1985, 1986, 1988, and references therein). The latter techniques have a mixed record, successfully revealing a stellar nucleus in M81 but finding a similar nucleus in M32 that is not confirmed by higher resolution observations (Section 2.2). Their weakness is that they magnify nonrandom errors, such as kinks in the observed profile. Lauer's simpler technique mines less resolution from the data but finds fewer spurious features. Better still is resolution good enough to require little correction. By the time this article appears, the Space Telescope may provide a long-awaited factor-of-five additional improvement in resolution.
Core profile shapes have been examined systematically by Lauer (1985b). His CCD data show that virtually all cores have slightly non-isothermal brightness profiles. Kormendy's (1985a, K87) high-resolution CCD photometry shows further that core profile shape correlates weakly with galaxy luminosity. A few galaxies, including some first-ranked galaxies in clusters, have isothermal profiles. Fainter galaxies have profiles that do not flatten completely into a core. The faintest galaxy with a well-resolved core is M31; it is even less isothermal than the ellipticals (Kent 1983). Such profiles have been interpreted as evidence for central black holes (Young et al. 1978, 1979) or anisotropy (Kormendy 1985a).
Some cores also show kinematic evidence for anisotropy (K87). For example, the core of NGC 1600 falls far below the "oblate line" describing isotropic oblate rotators in the Vmax / - diagram (Illingworth 1977). Like the overall shapes of ellipticals, the E3 shape of the core of NGC 1600 must be maintained by anisotropy. Dynamical modeling is required to explore the orbital distribution functions implied by these observations.
Considerable effort has also gone into the measurement of characteristic parameters of cores, i.e. central surface brightnesses µ0, core radii rc at which the surface brightness has fallen by a factor of two, and central velocity dispersions . Structural scaling laws revealed by these data are discussed in Section 8.
Dense nuclear star clusters superposed on much larger cores have been recognized in at least six bright galaxies. The best example is in M31 (Light et al. 1974); this has rc 0".4 and µOV 12.4 mag arcsec-2, compared with rc = 17" and µOV = 15.7 mag arcsec-2 for the bulge. Tremaine & Ostriker (1982) have shown that the nucleus and bulge of M31 are dynamically independent. We refer to these central star clusters as nuclei and distinguish them from bulges with cuspy brightness profiles (e.g. a pure power law like that in M32; Tonry 1984b). We also distinguish nuclei from nonthermal point sources such as those in Seyfert galaxies, quasars, and M87.
Besides M31, nuclei are found in M81 (Kormendy 1985a, Bendinelli et al. 1986) and in other nearby bulges (Kormendy 1985a). Their detection is limited by poor resolution (Lauer 1988b). Nuclei are also seen inmany dwarf spheroidal galaxies (Reaves 1977, 1983, Romanishin et al. 1977, Caldwell 1983, 1987, Binggeli et al. 1984, 1985, Ichikawa et al. 1986, van den Bergh 1986, Caldwell & Bothun 1987) and in many disk galaxies that are late enough in type so that they do not contain bulges (e.g; M33; Gallagher et al. 1982, and references therein).
Nuclei are rarely seen in ellipticals. For example, in M87, the nuclear spectrum shows no stellar absorption lines when the spectrum of the underlying core is subtracted (Dressler 1988, Kormendy, 1989). A small central excess of brightness above, an isothermal core in NGC 3379 (de Vaucouleurs & Capaccioli 1979, K82, Nieto & Vidal 1984) is not due to a stellar nucleus either, but only to a nonisothermal core exactly like those in other ellipticals (Bendinelli et al. 1984b, Kormendy 1985a). Nuclei suspected to exist in M32 and NGC 4649 (Bendinelli et al. 1982, Lauer 1988b) are not confirmed at better resolution (J. Kormendy, in preparation). Nuclei should be relatively easy to see in bright ellipticals because they have large, well-resolved cores. Their apparent scarcity could be due to the existence of a maximum luminosity for nuclei. Also, few ellipticals are as close to us as the bulges that are known to contain nuclei.
Since nuclei are poorly resolved, little further is known about them. Stellar kinematic data are, available in M31 (K87, Dressler & Richstone 1988, Kormendy 1988b), NGC 3115 (J. Kormendy & D.O. Richstone, in preparation), and NGC 4594 (Kormendy 1988c): Rapid rotation and velocity dispersions of ~ 100 km s-1 (after bulge subtraction) indicate that all three nuclei are disks. (Dressler & Richstone did not come to this conclusion, but they did not subtract the bulge spectrum; then detection of the cold component is difficult.) In NGC 3115 and NGC 4594, the disk structure is also seen in the isophotes.
The available data suggest that disklike nuclei are built out of gas that has fallen into the center (van den Bergh 1976, K82, Kormendy 1982b, 1988b, c, Gallagher et al. 1982, Kormendy & Illingworth 1983). This idea is a natural consequence of the hypothesis that black holes are fueled by infalling gas. If gas can reach the black hole, it may form stars along the way when the density gets high enough in the gravitational funnel. This may even be a necessary step in the formation of nuclear black holes, since collapse times of cores in giant ellipticals are long, whereas nuclei can evolve more rapidly (Kormendy 1988a, b, c). Further discussion is given in Shlosman & Begelman (1987) and in Duschl (1988a, b).
Nuclei may originate in other ways, too. Globular clusters sink toward the center by dynamical friction and may form nuclei (Tremaine et al. 1975). A large galaxy can accrete a small one with a compact core (Section 3). And black holes may produce central density cusps. Accreted nuclei should be distinguishable from black hole cusps: In general they should have smaller and a different rotation axis than the rest of the galaxy. However, accreted nuclei and ones grown by gas infall and star formation may be difficult to distinguish.
There are indications that nuclei in dwarf spheroidal and disk galaxies are similar to those seen in bulges. The nucleus of M33 is interpreted by Gallagher et al. (1982) as a composite-age stellar population, consistent with late infall of gas and subsequent star formation. Spectra of nuclei in dwarf spheroidal galaxies suggest that they are 5 Gyr old but sometimes contain a contribution from younger (A-F) stars; this is also consistent with secondary formation (Bothun et al. 1985, Caldwell & Bothun 1987, Bothun & Mould 1988). It is also possible that some "nuclei" in dwarf galaxies are really very low-luminosity bulges, since small bulges have small rc and high µ0 (see K87 for a review).