ARlogo Annu. Rev. Astron. Astrophys. 2004. 42: 603-683
Copyright © 2004 by Annual Reviews. All rights reserved

Next Contents


This paper reviews internal processes of secular evolution in disk galaxies. We concentrate on one important consequence: the buildup of dense central components that look like classical, i.e., merger-built bulges but that were made slowly by disks out of disk material. These are called pseudobulges. Our discussion updates a review by Kormendy (1993).

The relative importance of the different physical processes of galaxy evolution (Figure 1) changes as the Universe expands. Rapid processes that happen in discrete events are giving way to slow, ongoing processes.

Figure 1

Figure 1. Morphological box (Zwicky 1957) of processes of galactic evolution updated from Kormendy (1982a). Processes are divided vertically into fast (top) and slow (bottom). Fast evolution happens on a free-fall ("dynamical") timescale, tdyn ~ (Grho)-1/2, where rho is the density of the object produced and G is the gravitational constant. Slow means many galaxy rotation periods. Processes are divided horizontally into ones that happen purely internally in one galaxy (left) and ones that are driven by environmental effects such as galaxy interactions (right). The processes at center are aspects of all types of galaxy evolution. This paper is about the internal and slow processes at lower left.

At early times, galactic evolution was dominated by a combination of dissipative collapse (Eggen, Lynden-Bell & Sandage 1962; Sandage 1990) and mergers (Toomre 1977a) of galaxies that virialized out of the density fluctuations of cold dark matter. These are the top processes shown in Figure 1. The evolution timescale was short, tdyn ~ (1 / Grho)1/2, where rho is the mean density and G is the gravitational constant. The processes were violent. Many present-day galaxies owe their properties to this violence. Because mergers scramble disks and induce dissipation and starbursts, they are thought to make classical bulges and elliptical galaxies. We do not review classical bulges other than to contrast them to pseudobulges. Most work on galaxy evolution in the past 25 years has concentrated on hierarchical clustering and mergers. As the Universe expands, and as galaxy clusters virialize and acquire large internal velocities, mergers get less common (Toomre 1977a, Conselice et al. 2003).

In the distant future, internal secular processes will become dominant. These are defined to be slow processes, i.e., ones that have timescales much longer than tdyn. To be interesting, they must operate over long times. Some secular processes, such as disk heating via stellar enounters with molecular clouds, are well known (Spitzer & Schwarzschild 1951, 1953). But star-star relaxation is too slow to be important almost everywhere in almost every galaxy. Therefore, relevant secular processes generally involve the interactions of individual stars or gas clouds with collective phenomena such as bars, oval distortions, spiral structure, and triaxial dark matter halos. Also important are the interactions of these collective phenomena with each other. Given that hierarchical clustering continues today, has there been time for slow processes to be important? For secular evolution to have a significant effect, a galaxy must be free of major mergers for a long time, because merger violence erases the signature of secular processes. Hierarchical clustering results in so many mergers that one might guess that secular processes are relatively unimportant. A clue that this is frequently not the case is provided by galaxies with superthin - and fragile - disks but apparently no bulges (e.g., Matthews, Gallagher & van Driel 1999b; Freeman 2000; van der Kruit et al. 2001). They show that many galaxies have suffered no major merger violence since the onset of star formation in the disk (Tóth & Ostriker 1992). Therefore there has been time for secular evolution to be important in at least some galaxies. Given that mergers make bulges and ellipticals, these tend to be late-type galaxies. Actually, because the latest-type galaxies are (pseudo)bulgeless, secular evolution is likely to be most important in intermediate-late-type galaxies, i.e., Sbcs. But even some S0 and Sa galaxies contain pseudobulges. Secular processes have received less attention than galaxy mergers.

Still, this subject has made rapid progress. Many reviews discuss secular evolution in barred and oval galaxies (Kormendy 1979a, b; 1981, 1982a; Norman 1984; Combes 1991, 1998, 2000, 2001; Martinet 1995; Pfenniger 1996a, b, 2000; Sellwood & Debattista 1996; Buta 1995, 1999, 2000; Athanassoula 2002; and especially Sellwood & Wilkinson 1993 and Buta & Combes 1996). Discussions of pseudobulge formation include, besides the above, Pfenniger & Norman (1990); Courteau (1996b); Carollo, Ferguson & Wyse (1999); Balcells et al. (2003); and especially Kormendy (1993) and Carollo (2003).

Theory and observations point to a variety of processes that redistribute energy in disks. In Section 2, we review evidence that bars and ovals rearrange disk gas into outer rings, inner rings, and central mass concentrations. The resulting star formation produces a central stellar subsystem that has the high density and steep density gradient of a bulge but that was not formed by galaxy mergers.

Secular evolution is not confined to barred and oval galaxies. Bars can self-destruct by building up the central mass concentration, so secular evolution may have happened even if no bar is seen today. Global spiral structure also makes galaxies evolve, albeit more slowly than do bars.

These processes are manifestations of very general dynamical principles. Disks spread in radius - the inner parts shrink and the outer parts expand - because this lowers the total energy for fixed total angular momentum (Lynden-Bell & Kalnajs 1972, Tremaine 1989). Two-dimensional spreading by angular momentum transport is as fundamental for rotation-dominated disks as is three-dimensional spreading by energy transport in the core collapse of ellipsoidal systems dominated by random motions. The reason is the same, too. Self-gravitating systems have negative specific heats, so increasing the central density by flinging away the periphery lowers the total energy (Lynden-Bell & Wood 1968, Binney & Tremaine 1987). What makes evolution important in some systems and not in others? The determining factor is whether any evolution process is fast enough. Core collapse requires short relaxation times. Galaxy disks have long relaxation times, so their evolution is interesting only if they have an alternative to relaxation. Nonaxisymmetries provide the engine for rapid evolution.

We argue here that pseudobulges are one result of this evolution. Of key importance is the observation that they retain some memory of their disky origin. We review this subject in detail (Section 4), because it is central to any conclusion that evolution has happened and because it is the only way that we can recognize pseudobulges.

Next we review observations of gas content and star-formation rates. From these, we estimate pseudobulge growth times. These prove to be in good agreement with stellar population ages. So the picture of pseudobulge growth from rearranged disk gas is internally consistent.

Our purpose is to connect up a large number of apparently disparate results into a well-developed and (as we hope to show) a well-supported paradigm. Still, many questions remain unanswered. In particular, we need a better understanding of the relative importance of mergers and secular evolution as a function of galaxy type and luminosity. We hope that this review will provide a concrete context that will allow efficient progress in this subject.

1.1. What is a Bulge? Classical and Physical Morphology

What do we mean by a bulge? The answer shows why, for some galaxies, we use the term pseudobulge.

Renzini (1999) clearly states the canonical interpretation of Hubble-Sandage-de Vaucouleurs classifications: "It appears legitimate to look at bulges as ellipticals that happen to have a prominent disk around them [and] ellipticals as bulges that for some reason have missed the opportunity to acquire or maintain a prominent disk." We adopt this point of view. However, as observations improve, we discover more and more features that make it difficult to interpret every example of what we used to call a bulge as an elliptical living in the middle of a disk. This leads authors to agonize: "Are bulges of early-type and late-type spirals different? Are their formation scenarios different? Can we talk about bulges in the same way for different types of galaxies?" (Fathi & Peletier 2003).

We conclude that early- and late-type galaxies generally do make their dense central components in different ways. This is not recognized in classical morphology, because it defines classification bins - deliberately and with good reason - without physical interpretation. Sandage & Bedke (1994) describe how, in the early stages of investigating a subject, a classifier should look for "natural groups" (Morgan 1951) of objects with similar features. Sandage emphasizes that it is important not to be led astray by preconceptions: "The extreme empiricist claims that no whiff of theory may be allowed into the initial classification procedures, decisions, and actions." Nevertheless, some choice of which features to consider as important and which to view as secondary must be made. After all, the goal is to understand the physics, and the exercise is useful only if classification bins at least partly isolate unique physics or order galaxies by physically relevant parameters. The Hubble-Sandage-de Vaucouleurs classification scheme has done these things remarkably well.

However, it is reasonable to expect that improved understanding of galaxies will show that the classification missed some of the physics. Also, some features of galaxies could not be observed well enough in the photographic era to be included. These include high-surface-brightness disky substructures in galaxy centers. Consistent with physical morphology as discussed in Kormendy (1982a), we wish to distinguish components in galaxies that have different origins.

At the level of detail that we nowadays try to understand, the time has passed when we can make effective progress by defining morphological bins with no guidance from a theory. Disks, bulges, and bars were different enough that we could do this. Afterward, robust conclusions could be reached, e.g., about the relative timescales of collapse and star formation (Eggen, Lynden-Bell & Sandage 1962). But even inner rings and spiral arms - which are not subtle - do not scream the appropriate message, which is that spiral arms are details that would disappear quickly and without a trace if the driving mechanism switched off, whereas we will see that rings are a permanent rearranging of disk material. Inner rings are, in this sense, more fundamental than spiral arms. Years ago, people commonly reacted badly to a classification as complicated as (R)SB(r)b (de Vaucouleurs et al. 1991). The reason, we believe, was that the phenomenology alone did not sell itself. People did not see why this level of detail was important. Now, we will show that every letter in the above classification has a clear-cut meaning in terms of formation physics. This is the goal of physical morphology.

We adopt the view that bulges are ellipticals living in the middle of disks. Ellipticals form via mergers (Toomre 1977a, Schweizer 1990). Therefore, we do not use the term bulge for every central component that is in excess of the inward extrapolation of an exponential fitted to the disk brightness profile. If the evidence suggests that such a component formed by secular processes, we call it a pseudobulge. In practice, we cannot be certain about formation mechanisms. Therefore, if the component in question is very E-like, we call it a bulge, and if it is disk-like, we call it a pseudobulge. Intermediate cases are discussed in Sections 4, 7, and 9.1.

Finally, we comment on one of the biggest problems in this subject. It is exceedingly easy to get lost in the details. Many authors interpret observations or simulations in much more detail than we do here. For example, it is common for observers to distinguish nuclei, nuclear bars, nuclear disks, nuclear spiral structure, exponential bulges, boxy bulges, and star-formation rings. We discuss all these features, because they are central to the developing picture of what secular evolution can accomplish. But we consider them all to be features of pseudobulges, because the evidence is that they are all built by secular evolution out of disk material. In the same way, global spiral structure, flocculent spiral structure, and the absence of spiral structure in S0 galaxies are all features of disks.

Next Contents