Galaxy Mergers and Merger Rates
Galaxies must grow with time through both discrete galaxy mergers and smooth gas accretion. When and how this growth occurs remains an outstanding observational question. The smooth accretion of gas and dark matter onto distant galaxies is extremely challenging to observe, and complex baryonic physics makes it difficult to infer a galaxy’s past assembly history. In contrast, counting galaxy mergers is relatively straightforward. By comparing the frequency of galaxy mergers to the mass growth in galaxies, one can place robust constraints on the importance of discrete galaxy mergers in galaxy assembly throughout cosmic time. The mass accretion rate via mergers is likely to be be a strong function of galaxy mass, merger ratio, environment, and redshift (Stewart et al. 2009); these dependencies can test both the cosmological model and the galaxy-halo connection. In addition to contributing to the overall buildup of galaxy mass, the violent processes associated with mergers are expected to significantly influence the star formation histories, structures, and central black hole growth of galaxies. However, other physical mechanisms may influence galaxy evolution in similar ways, so direct observations of galaxy mergers are needed to answer the following questions:
• What fraction of the global star formation density is driven by mergers and interactions? Is the frequency of galaxy mergers consistent with the “tightness” of the star formation per unit mass vs. stellar mass relation throughout cosmic time?
• Are typical red spheroids and bulges formed by major mergers, or by secular evolution? Do z > 1 compact galaxies grow in size by (minor) mergers?
• Are today’s most massive ellipticals formed via dissipationless mergers? If so, when?
• Do gas-rich mergers fuel active galactic nuclei? Which forms first, the bulge or super massive black hole?
In a CDM model, the rate at which dark matter halos merge is one of the fundamental processes in structure formation. Numerical simulations predict that this rate evolves with redshift as (1+z)m, with 1.0 < m < 3.5 (Gottl¨ober et al. 2001; Berrier et al. 2006; Fakhouri & Ma 2008; Stewart et al. 2009). It is difficult to directly compare the predicted dark matter halo merger rate with the observed galaxy merger rate due to the uncertainty in the halo occupation number. However, if this comparison is done self-consistently, measuring the merger frequency as a function of cosmic epoch can place powerful constraints on models of structure formation in the Universe. Numerous observational studies over the past two decades have focused on measuring the galaxy merger rate, yielding highly discrepant values of m, ranging from no evolution (m ~ 0) to strong evolution (m ~ 5) (Zepf & Koo 1989; Carlberg et al. 1994, 2000; Patton et al. 2000, 2002; Bundy et al. 2004; Lin et al. 2004; Bridge & Carlberg 2007; Lotz et al. 2008a). As a consequence, the importance of galaxy mergers to galaxy assembly, star formation, bulge formation, and supermassive black hole growth is strongly debated. These observational discrepancies may stem from small sample sizes, improperly accounting for the timescales over which different techniques are sensitive, and the difficulty in tying together surveys at high and low redshift with different selection biases.
The galaxy merger rate is traditionally estimated by measuring the frequency of galaxies residing in close pairs, or those with morphological distortions associated with interactions (e.g., double nuclei, tidal tails, stellar bridges). The detection of distortions is done either by visual analysis and classification (Le F`evre et al. 2000; Bridge et al. 2009 in preparation) or through the use of quantitative measures (Abraham et al. 1996; Conselice 2003; Lotz et al. 2008b). A key uncertainty in calculating the galaxy merger rate is the timescale associated with identifying a galaxy merger. The merger of two comparable mass galaxies may take 1–2 Gyr to complete, but the appearance of the merger changes with the merger stage, thus a given merger indicator (i.e., a close companion or double nucleus) may only be apparent for a fraction of this time (Lotz et al. 2008b). Galaxies at z < 1 with clear signatures of merger activity are relatively rare (<10-15% of L galaxies at z < 1, <5% at z = 0), although the fraction of galaxies which could be considered to be ‘merging’ may be significantly higher.
No single study conducted so far has been able to uniformly map the galaxy merger rate from z = 0 to z 2, as current studies must trade off between depth and volume. An additional limiting factor is the observed wavelength range, as galaxy morphology and pair luminosity ratios are often a strong function of rest-frame wavelength. Very few merger studies have been done with SDSS (optimized for the z < 0.2 Universe). SDSS fiber collisions and low precision photometric redshifts prevent accurate pair studies, while the the relatively shallow imaging and moderate ( 1.4") seeing reduce the sensitivity to morphological distortions. Deeper spectroscopic and imaging studies probe the z ~ 0.2 − 1 Universe, but do not have the volume to also constrain the low redshift Universe or the depth at near-infrared (rest-frame optical) wavelengths to constrain the z > 1 Universe. Ultra-deep Hubble Space Telescope studies (GOODS, UDF) can detect L* mergers at z > 1, but have very small volumes and are subject to strong cosmic variance effects. The CFHTLS-Deep survey is well matched to the proposed LSST depths, wavelengths, and spatial resolution, but, with an area 5000 times smaller, is also subject to cosmic variance (Bridge et al., 2009).
Unlike the current studies, LSST has the depth, volume, and wavelength coverage needed to perform a uniform study of L mergers out to z ~ 2, and a statistical study of bright galaxy mergers out to z ~ 5. The wide area coverage of LSST will be critical for addressing the effects of cosmic variance on measures of the merger rate, which can vary by a factor of two or more even on projected scales of a square degree (Bridge et al., 2009). A variety of approaches will be used to identify mergers in the LSST data:
• Short-lived strong morphological disturbances, such as strong asymmetries and double nuclei, which occur during the close encounter and final merger stages and are apparent for only a few 100 Myr. These will be most easily found in z . 0.2 galaxies, where the LSST 0.7" spatial resolution corresponds to 1–2 kpc. Lopsidedness in galaxy surface brightness profiles can provide statistical constraints on minor mergers and requires similar spatial resolution.
• Longer-lived but lower surface brightness extended tidal tails, which occur for ~0.5 Gyr after the initial encounter and for up to 1 Gyr after the merger event. These tails are the longest-lived merger signatures for disk-galaxy major mergers, and should be easily detected at z ~ 1 in the full depth LSST r i z images (see Figure). Scaling from the CFHTLS-Deep survey, LSST should detect on the order of 15 million galaxies undergoing a strong tidal interaction.
• Residual fine structures (faint asymmetries, shells, and dust features) detected in smooth model subtracted images. These post-merger residual structures are visible for both gas-rich and gas-poor merger remnants, and contribute < 1–5% of the total galaxy light, with surface brightnesses 28 mag arcsec−2.
• The statistical excess of galaxy pairs with projected separations small enough to give a high probability for merging within a few hundred Myr. With 0.700 seeing, galaxies with projected separations > 10 kpc will be detectable to z ~ 5.
LSST’s six-band photometry will result in photometric redshift accuracies of about ~0.03(1+z). This is comparable to or better than those used in other studies for the identification of close galaxy pairs, and will allow for the selection of merging galaxies with a wide range in color. With LSST’s high quality photometric redshifts and large number statistics, it will be possible to accurately measure the galaxy pair fraction to high precision (although the identification of any given pair will be uncertain). Current surveys detect only about 50-70 red galaxy pairs per square degree for 0.1 < z < 1.0. With LSST, we should be able to observe more than a million “dry” mergers out to z ~ 1.0.
One of the advantages of the LSST survey for studying the evolving merger rate is the dense sampling of parameter space. A large number of merger parameters — galaxy masses and mass ratio, gas fractions, environment — are important for understanding the complex role of mergers in galaxy evolution. For example, mergers between gas-poor, early-type galaxies in rich environments have been invoked to explain the stellar mass build-up of today’s most massive ellipticals (e.g., van Dokkum 2005; Bell et al. 2007. Each of the approaches given above will yield independent estimates of the galaxy merger rate as a function of redshift, stellar mass, color, and environment. However, each technique probes different stages of the merger process, and is sensitive to different merger parameters (i.e., gas fraction, mass ratio). Therefore, the comparison of the large merger samples selected in different ways can constrain how the merger sequence and parameter spaces are populated.
Finally, the cadence of the LSST observations will open several exciting new avenues. It will be possible to identify optically variable AGN in mergers and constrain the SMBH growth as a function of merger stage, mass, and redshift. With millions of galaxy mergers with high star formation rates, we will detect a significant number of supernovae over the ten-year LSST survey. We will be able to determine the rate of SN I and II in mergers, and obtain independent constraints on the merger star formation rates and initial stellar mass functions.