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

4.1. Embedded Disks, Spiral Structure, and Star Formation

Central to our image of bulges as elliptical galaxies living in the middle of a disk is their morphological resemblance to ellipticals. Central to our growing awareness that something different is going on in late-type galaxies is the observation that their high-surface-brightness centers look nothing like ellipticals. Instead, they are dominated by young stars and by disk structure. This is especially true in barred and oval galaxies, that is, in the objects in which secular evolution should be most rapid.

What are we looking for?

A clear statement that classical bulges are equivalent to ellipticals is Sandage & Bedke's (1994) description of E/S0 galaxies: "On short-exposure plates showing only the central regions, no evidence of a disk ... is seen; the morphologies of the central regions are pure E." The section on elliptical galaxies contains this caution: "The presence or absence of dust is not used as a classification criterion. Some E galaxies ... have dust patches and remain classified as E types." The same is true for bulges; e.g., S0 galaxies range from dustless (S0₁) to dusty (S0₃), but all have bulges. We need to be careful that what we identify as pseudobulges are not just dust features or the outer disk extending inside a classical bulge all the way to the center. On the other hand, part of the definition of an elliptical, hence also of a bulge, is that "There is no recent star formation, inferred from the absence of luminous blue and red supergiants". Even of Sab galaxies, Sandage & Bedke say that "the central bulge is ... nearly always devoid of recently formed stars." Of course, old bulges must have contained young stars in the past; these definitions - and the Hubble sequence - are understood to apply to present-day galaxies and long after major mergers are completed. But ubiquitous ongoing star formation is a pseudobulge signature.

Turning to pseudobulges, Kormendy (1993) noted that the prototypical oval galaxy NGC 4736 has a disk-like "bulge": "The central brightness profile ... is an r^1/4 law that reaches the high central surface brightness characteristic of a bulge (Boroson 1981). However, the r^1/4-law component shows a nuclear bar and spiral structure to within a few arcsec of the center. Bars are disk phenomena. More importantly, it is not possible to make spiral structure in a bulge. Thus the morphology already shows that the r^1/4-law profile belongs to the disk." This conclusion is consistent with dynamical evidence shown in Figure 17 and and with the nuclear star formation ring shown in Figure 8.

Sandage (1961) comments similarly and presciently about flocculent spirals, including NGC 4736, in his description of NGC 5055: "The curious and significant feature of [these galaxies] is the sharp discontinuity of surface brightness of the spiral pattern between the inner and the outer regions [close to the center]. The spiral structure is as pronounced and well defined in the bright region as in the outer parts. The important point here is that, if the inner arms were to coalesce and to lose their identity as spiral arms, the region would look amorphous, would have a high surface brightness, and would resemble the central regions of NGC 2841 [a classical bulge] ... and all members of the E and S0 classes." The observation that the spiral structure is as pronounced in the bright region as in the outer parts has important implications. If a high-surface-brightness classical bulge were projected in front of the (relatively faint) inward extrapolation of the outer disk, it would dilute the spiral structure. This is not seen. Therefore it is the high-surface-brightness component that contains the spiral structure. Again, this is a pseudobulge signature.

HST spatial resolution reveals disk structure in the "bulge" regions of surprisingly many galaxies. Carollo and collaborators have carried out a snapshot survey of 75, S0 - Sc galaxies with WFPC2 and the F606W filter approximating V band (Carollo et al. 1997, 1998; Carollo & Stiavelli 1998; Carollo 1999) and of 78 galaxies with NICMOS F160W approximating H band (Carollo et al. 2001, 2002; Seigar et al. 2002). Figures 10 - 13 show pseudobulges from these papers. What is remarkable about these generally Sb and Sbc galaxies is how often the central structure looks like a smaller version of a normal, late-type disk.

Figure 10. NGC 1353 pseudobulge (top image: 18" × 18" zoom, and middle: full WFPC2 F606W image taken with HST by Carollo et al. 1998). The bottom panel is a 2MASS (Jarrett et al. 2003) JHK composite image with a field of view of 4'.4 × 4'.4. The plots show surface photometry with the HST profile shifted to the K-band zeropoint. The lines show a decomposition of the major-axis profile into a Sérsic (1968) function and an exponential disk. The outer part of the pseudobulge has the same apparent flattening as the disk. This nuclear disk produces much of the rapid upturn in surface brightness toward the center.

NGC 1353 (Figure 10) is one of the clearest examples. The top-right panel shows the central 18" × 18" of the PC image (Carollo et al. 1997, 1998). The middle panel is the full WFPC2 field of view, and the bottom image is the 2MASS JHK-band composite. The images show, as Carollo and collaborators concluded, that the central structure in NGC 1353 is a disk with similar flattening and orientation as the outer disk. To make this quantitative, we measured the surface brightness, ellipticity, and position angle profiles in the PC and 2MASS images using the PROFILE tool in the image processing system VISTA (Lauer 1985). The left panels show that the apparent flattening at 2" r 4" is the same as that of the main disk at large radii. The position angle is the same, too. So the part of the galaxy shown in the top-right panel really is a disk. The brightness profile shows that this nuclear disk is responsible for much of the central rise in surface brightness above the inward extrapolation of an exponential fitted to the outer disk. Presented only with the brightness profile or with the bottom two panels of images, we would identify the central rise in surface brightness as a bulge. Given Figure 10, we identify it as a pseudobulge.

We have decomposed the major-axis profile into an exponential outer disk plus a Sérsic (1968) function, I(r) e^{-K[(r / r_e)1/n-1]}. Here n = 1 for an exponential, n = 4 for a de Vaucouleurs (1948) r^1/4 law, and K(n) is chosen so that radius r_e contains half of the light in the Sérsic component. In Section 4.2, we will discuss evidence that "bulges" in late-type galaxies are generally best described by Sérsic functions with n ~ 1. That is, they are nearly exponential. This behavior is characteristic of many pseudobulges. Here we note that NGC 1353 is an example. The best fit gives n = 1.3 ± 0.3.

The 2MASS image and the and PA profiles show that NGC 1353 contains a weak bar with a projected radius of ~ 15" and an approximately NS orientation. This is one example among many of the association between pseudobulges and nonaxisymmetric features that can drive secular evolution. In visible light, the galaxy is classified SBb by de Vaucouleurs et al. (1991) and Sbc by Sandage & Bedke (1994).

Figure 11 shows another example. NGC 5377 is classified SBa or Sa by Sandage and (R)SBa by de Vaucouleurs, and it easily satisfies the photometric criteria for recognizing an oval outer disk. It is one of the earliest-type galaxies discussed in this paper. An Sa should be dominated by a bulge. Indeed, the brightness profile at r 1" and at about 6" to 10" is bulge-like. But the galaxy also contains a high-surface-brightness embedded nuclear disk that is seen as the shelf in the brightness profile at r 1" to 3". Again, this has approximately the same apparent flattening and position angle as the outer disk. If a bulge is defined to be the extra light at small radii above the inward extrapolation of the outer disk profile, then that definition clearly includes the nuclear disk. We prefer not to adopt this definition but rather to identify NGC 5377 as a galaxy with a substantial pseudobulge component. Whether this is embedded in a classical bulge or whether the whole of the central rise in surface brightness is a pseudobulge, we cannot determine from the available data.

Figure 11. NGC 5377 pseudobulge (top image: 18" × 18" zoom, and middle: full WFPC2 F606W image taken with HST by Carollo et al. 1998). At the bottom is a 7' × 7', r-band image of the outer ring (Frei et al. 1996). The plots show surface photometry of the HST, r-band, and 2MASS JHK composite images, all shifted to the 2MASS, K-band zeropoint. The two shelves in the brightness profile are the nuclear disk and inner oval. The nuclear disk has the same apparent flattening and orientation as the outer ring. It may be embedded in a less obviously disky bulge, but it produces a rapid upturn in surface brightness toward the center.

Figure 12 shows a third case study, NGC 6384. Its bar is subtle; the galaxy is classified Sb by Sandage and SABbc by de Vaucouleurs. But it is clearly visible in the WFPC2 image (middle panel). Sandage and Bedke (1994) note that "There is a smooth inner bulge ..." and the Carollo et al. (1998) image (top panel in Figure 12) confirms that the central brightness distribution is smooth enough - ignoring dust - that one would ordinarily identify this as a classical bulge. However, photometry of the PC image shows that the PA and apparent flattening are essentially the same at 2" r 12" as in the outer disk. This "bulge" is quite flat. Also, it is quite different from a de Vaucouleurs r^1/4 law. Carollo et al. (1998) conclude that it is exponential. We get n = 2.2 ± 0.2, but this does not take into account the light in the bar. If bar stars that pass through the outer bulge were subtracted from the profile, then n would get smaller. So the flatness of the central component is enough to identify this as a pseudobulge, and its small value of n contributes to the identification of exponential profiles as a pseudobulge characteristic (Section 4.2). NGC 6384 demonstrates that pseudobulges can be subtle enough so that photometry, and not just morphology, is needed to recognize them.

Figure 12. NGC 6384 pseudobulge (top image: 18" × 18" zoom, and middle: full WFPC2 F606W image taken with HST by Carollo et al. 1998). At the bottom is the B-band image from the Carnegie Atlas of Galaxies (Sandage & Bedke 1994). The top, middle, and bottom panels are shown with logarithmic, square root, and linear stretches. The plots show surface photometry of the HST and other images identified in the key, all shifted to the R-band zeropoint. The decomposition into a Sérsic function bulge and exponential disk is done over a radius range that omits the region 12" < r < 40" affected by the bar.

Further examples from Carollo et al. (1997, 1998) of disky centers in Sa - Sbc galaxies are shown in Figure 13. They look like miniature late-type galaxies. But they occur where the surface brightness rises rapidly above the inward extrapolation of the outer disk profile. This is not obvious in Figure 13 because we use a logarithmic intensity stretch so that we can show the structure over a large range in surface brightness. Spiral structure is a sure sign of a disk. Carollo et al. (1997) conclude that these observations "support scenarios in which a fraction of bulges forms relatively late, in dissipative accretion events driven by the disk."

Figure 13. Sbc galaxies whose "bulges" have disk-like properties. Each panel shows an 18" × 18" region centered on the galaxy nucleus and extracted from HST WFPC2 F606W images taken and kindly provided by Carollo et al. (1998). North is up and east is at left. Displayed intensity is proportional to the logarithm of the galaxy surface brightness. Hubble types are from Sandage & Bedke (1994) except for NGC 4030; its type is from the RC3 and was checked using high-quality images posted on NED.

The statistics of the Carollo sample suggest that pseudobulges are surprisingly common. In the following summary, we distinguish classical bulges that are well described by r^1/4 laws from pseudobulges that show at least one of the following characteristics: they are flat or are dominated by disk morphology such as spiral structure; they are vigorously forming stars; or they have surface brightness profiles that are best described by Sérsic functions with n 2. In a few cases, observing n 1 caused us to reclassify a "regular bulge" in Carollo et al. (1997, 1998) as a pseudobulge. Also, we use the mean of the classifications given in the RC3 and in the UGC/ESO-LV (Nilson 1973; Lauberts & Valentijn 1989). Then in the above sample of 75 galaxies, classical bulges are seen in 69 % of 13 S0 - Sa galaxies, 50 % of 10 Sab galaxies, 22 % of 23 Sb galaxies, 11 % of 19 Sbc galaxies, and 0 % of 10 Sc and later-type galaxies. Most of the rest are pseudobulges or have a substantial pseudobulge component added to a classical bulge. In some cases, there is only a compact nuclear star cluster added to a late-type disk; it is not clear whether the same secular evolution processes make these (see Section 4.9). Distinguishing classical bulges from pseudobulges is still an uncertain process. Even the morphological types are sometimes inconsistent between the RC3 and the UGC by several Hubble stages. However, it is unlikely that the conclusions implied by the above statistics are seriously wrong. As noted by Carollo et al. (1997, 1998), most early-type galaxies appear to contain classical bulges; these become uncommon at types Sb and later, and essentially no Sc or later-type galaxy has a classical bulge. Kormendy (1993) reached similar conclusions.

So an HST V-band survey shows that disky bulges are more common than ground-based data suggested. Clearly it is desirable to check this result. An H-band HST NICMOS survey by Carollo et al. (2002) and by Seigar et al. (2002) complements the V-band survey in several ways. The images are less affected by dust. Classification of nuclear disks is easier. The infrared images are less sensitive to star formation, but Carollo et al. (2002) compensate by including V - K color images. The infrared survey confirms the V-band results. Additional imaging studies that reveal central disk structures, dust, or star formation in disk galaxies include Elmegreen et al. (1998); van den Bosch, Jaffe, & van der Marel (1998); Peletier et al. (1999); Erwin & Sparke (1999, 2002, 2003); Ravindranath et al. (2001); Rest et al. (2001); Hughes et al. (2003); Martini et al. (2003a); Fathi & Peletier (2003), Erwin et al. 2003, and Erwin (2004).

We do not mean to imply that a "bulge" is always either purely classical or purely pseudo. We cannot tell from available data how much of a classical bulge underlies the pseudobulge component in S0 - Sbc galaxies. Indications (e.g., Figure 11) are that the classical bulge component in many early-type galaxies is significant even when an embedded disky structure is recognized. If our formation picture is correct, then there is every reason to expect that secular evolution often adds disky material to a classical bulge that formed in a prior merger. The relative importance of mergers and secular evolution needs further investigation.

At a more subtle level, some galaxies that are dominated by classical bulges contain nuclear disks that contribute a negligible fraction of the galaxy luminosity. These may be cases in which secular evolution produced only a trace effect. Alternatively, they may be later-type examples of the embedded disks seen in elliptical galaxies. If so, they cannot be a result of disk-driven secular evolution. They are discussed in Section 8.3.