|Annu. Rev. Astron. Astrophys. 2009. 47:
Copyright © 2009 by Annual Reviews. All rights reserved
Galaxy mergers are easily visible in about 1%-2% of all luminous galaxies - frequent enough to suggest that they could be important to galaxy evolution, but also infrequent enough that real statistical samples of mergers are only becoming available today. Figure 16 shows a subset of a sample of merging galaxies selected from SDSS DR6 by eye (Christina Ignarra 2008, private communication). A wide variety of merger types and tidal features are evident, including "dry mergers" between pairs of red galaxies, minor mergers, merger-driven starbursts, and large tidal tails. Although Struck (1999) present a detailed classification of merging systems that attempts to characterize their physical nature, it has not been applied to any of the newly available large samples.
Figure 16. SDSS images of merging galaxies, selected by eye from the SDSS DR6 (Christina Ignarra 2008, private communication). The images are sorted by absolute magnitude in the horizontal direction, ranging between Mr - 5 log10 h ~ -18.5 and -22 from left to right, and g - r color in the vertical direction, ranging between 0.2 and 0.9 mag from the bottom to the top. Thus, the brightest, reddest mergers are in the upper-right.
There are two general ways of counting "mergers": searching for signs in the images, and counting close pairs. For example, in Figure 16, we have simply searched for merger signs in the galaxies by eye. More objective samples have been created by measuring asymmetry and/or using unsharp-masking techniques (De Propris et al. 2007, McIntosh et al. 2008). Though such perturbation-based samples are pure in the sense of essentially containing only real physical pairs, they are biased because they require signs of interaction to be detectable. An alternative technique is to search for close pairs within about 50 kpc or so. Close pair samples are closer to unbiased but even when chosen from redshift surveys include non-physical pairs (Barton, Geller & Kenyon 2000, Smith et al. 2007). Although the distance to the nearest neighbor and the degree of perturbation are correlated, when De Propris et al. (2007) performed a careful comparison of the two approaches, they found generally non-overlapping samples. This result may indicate that close pair samples reveal the pre-merger population whereas searching for perturbations reveals a later stage.
Using large samples from the SDSS, several groups have studied the relationship between mergers and star-formation. As the results of Barton, Geller & Kenyon (2000) and others had previously indicated (Kennicutt et al. 1987, Kennicutt et al. 1998), close pairs of galaxies show a factor of 1.5-2 enhancement in their star-formation rate relative to a control sample. Equal mass mergers show the clearest signatures (Li et al. 2008, Ellison et al. 2008). Barton et al. 2007 use a carefully calibrated isolation criterion to select pairs that are not part of larger groups, finding a clearer signature of star-formation enhancement for such isolated pairs.
Although close neighbors are associated with true starbursts only rarely, those rare cases still account for about 40% of the existing starbursts in the Universe. Theory suggests that major gas-rich mergers would create such starbursts - a consequence of the large amount of gas that is driven to the center of the merging system (e.g., Mihos, Richstone & Bothun 1992, Cox et al. 2006). For these reasons, the most luminous infrared galaxies, whose luminosity is powered by dust-obscured star formation, are often associated with major mergers (Sanders & Mirabel 1996).
A class of mergers that do not have associated star-formation are the so-called "dry mergers" - red, gas-poor galaxies merging with other red, gas-poor galaxies. A classic example of such a merger is the infall through dynamical friction of an elliptical galaxy in a cluster onto the central system, building up its stellar mass and perhaps helping to create a cD envelope (Section 5.5). Indeed, surveys at higher redshift indicate that these sorts of mergers play some role for galaxies on the red sequence (e.g., Bell et al. 2006, Masjedi, Hogg & Blanton 2008). However, at least occasionally mergers of red galaxies turn out to not be quite "dry," and indeed have substantial gas content (Donovan, Hibbard & van Gorkom 2007).
A very rare but oft-studied subset of galaxies that may be related to mergers are the "post-starbursts." Such galaxies can be identified by their lack of ionized gas producing H or other emission lines, indicating no O and B stars, but strong Balmer lines in their spectra, indicating the presence of A stars. They are often referred to as "K+A" or "E+A" galaxies because of their distinct spectral characteristics (Dressler & Gunn 1983, Zabludoff et al. 1996). Because of the life-times of A stars, they must have ended star-formation within the last Gyr or so, and may be undergoing a transformation. It is unknown whether such transformations result from ram pressure stripping events, mergers, or something else. Indeed, post-starburst galaxies may have more than one formation mechanism.
In Figure 17, we show one method for selecting such galaxies, used by Quintero et al. 2004. They fit the full SDSS spectrum to a sum of an A star template and an old galaxy template, which results in an arbitrarily normalized ratio of A stars to old stars, denoted A / K. By comparing this ratio to H equivalent width they can identify galaxies with no recent star-formation but a significant contribution of A stars to the integrated spectrum. Indeed, they observe a spur of such galaxies.
Figure 17. Distribution of A / K and H equivalent width for SDSS galaxies, as determined by Quintero et al. (2004). The greyscale and contours indicate the density of points in this plane. Outliers are shown individually. In blue are shown several toy models: the line labeled "14" corresponds to a constant star-formation rate model over 14 Gyrs; the line labeled "10" corresponds to cutting off that star-formation abruptly at 10 Gyrs; the line labeled "3" corresponds to a cutoff at 3 Gyrs. In each of the latter two cases, the abrubt cutoff results in a post-starburst spectrum. The dashed red lines indicate the criteria for selection that Quintero et al. (2004) use.
This method of selection differs from previous methods in two ways. First, they use the H emission line rather than [O II] 3727 (e.g., Blake et al. 2004) or a combination of multiple lines (e.g., Poggianti et al. 2004). As Yan et al. (2006) show, using [O II] 3727 alone can easily exclude half or more of the K+A sample, which commonly have weak AGN (Yang et al. 2006). Any method using H is preferable in this respect. Second, they use full spectral fits rather than Balmer line equivalent widths (e.g., Balogh et al. 2005). This selection alters slightly the "purity" of the sample, as any sufficiently blue continuum will lead to a K+A classification, regardless of A star content.
These galaxies tend to be high in surface brightness and to be highly concentrated, yet blue (Norton et al. 2001, Quintero et al. 2004), with a strong possibility that as their stellar population fades they will become consistent with the red sequence, and become elliptical galaxies. HST imaging and optical spectroscopy by Yang et al. (2008) suggests that currently they are discrepant from the fundamental plane of ellipticals - that is, they are consistent with having the lower mass-to-light ratios appropriate to their young stellar populations, and may fade onto the fundamental plane. Most appear considerably more disturbed than the typical elliptical (Zabludoff et al. 1996), though those disturbances may disappear over time.
Interestingly, there is very little evidence that these galaxies occur more frequently in any particular environment, generally following the environmental trends of late-type galaxies (Quintero et al. 2004, Blake et al. 2004, Hogg et al. 2006, Yan et al. 2009). Poggianti et al. (2004) claim that in the Coma cluster there is a population of low luminosity (MV > -18.5) K+A galaxies associated with dense areas in the hot intracluster medium. As Yan et al. (2009) point out, some K+As may be explained by interaction with the ICM, but most cannot; the distribution of most K+As across environment is more consistent with a merger scenario (Zabludoff et al. 1996).