The second major advance in our picture of bulge formation involves the observation that many high-z disks are very gas-rich and dominated by 108 – 109 M⊙, kpc-size star-forming clumps (Elmegreen et al. 2005, 2007, 2009a, b; Bournaud et al. 2007; Genzel et al. 2006, 2008, 2011; Förster Schreiber et al. 2009, 2011a, b; Tacconi et al. 2010). These galaxies evidently accrete cold gas so rapidly that they become violently unstable. Bulgeless disks tend to have small epicyclic frequencies κ. If the surface density Σ rapidly grows large and is dominated by gas with low velocity dispersion σ, then the Toomre (1964) instability parameter Q = 0.30 σ κ/ G Σ ≲ 1 (G = graviational constant). The observed clumps are interpreted to be the result. Theory and simulations suggest that the clumps sink rapidly toward the center by dynamical friction. They also dump large amounts of additional cold gas toward the center via tidal torques. The result is violent relaxation plus a starburst that produces a classical bulge. Many papers discuss this evolution (e.g., Dekel, Sari, & Ceverino 2009; Ceverino, Dekel, & Bournaud 2010; Cacciato, Dekel, & Genel 2012; Forbes et al. 2014; Ceverino et al. 2015). Bournaud (2015) reviews this subject in the present book. I include it here for two reasons, it is a major advance, so it deserves emphasis in this concluding chapter, and I want to add two science points:
Figure 1 illustrates my first point: Evolution by clump sinking, inward gas transport, violent relaxation, and starbursts proceeds much as it does in our picture of wet major mergers. That is, in practice (if not in its beginnings), classical bulge formation from clump instabilities is a variant of our standard picture of bulge formation in wet major mergers. The process starts differently than galaxy mergers – what merges here are not finished galaxies but rather are clumps that formed quickly and temporarily in unstable disks. Nevertheless, what follows – although two- and not three-dimensional – is otherwise closely similar to a wet merger with gas inflow and a starburst. That is, it is a slower, gentler version of Arp 220.
Figure 1. Mergers of clumpy initial conditions make Sérsic (1968) function remnants with indices n ∼ 2–4. A remarkably early illustration is the n-body simulation of van Albada (1982), whose initial conditions (grayscale densities) resemble the clumpy high-z galaxy UDF 1666 studied by Bournaud et al. (2007). Van Albada's initial conditions were parameterized by the ratio of twice the total kinetic energy to the negative of the potential energy. In equilibrium, 2T / W = 1. For smaller values, gentle collapses (2T / W = 0.5) make Sérsic profiles with n < 4. Violent collapses (2T / W ≲ 0.2) make n ≳ 4. Clump sinking in high-z disks is inherently gentle. The hint is that the clumps merge to make classical bulges with n < 4. This figure is from Kormendy (2012).
Early models by Elmegreen et al. (2008) confirm that gas-rich galaxy disks violently form clumps like those observed. The clumps quickly sink, merge, and make a high-Sérsic-index, vertically thick bulge. It rotates slowly, and rotation velocities decrease with increasing distance above and below the disk plane. These are properties of classical bulges, and Elmegreen and collaborators conclude that this process indeed makes classical (not pseudo) bulges. Many of the later papers summarized above and reviewed by Bournaud (2015) reach similar conclusions.
However, Bournaud (2015) goes on to review more recent simulations that– among other improvements – include strong feeback from young stars. The results complicate the above picture. For example, Genel et al. (2012) find that “galactic winds are critical for [clump] evolution. The giant clumps we obtain are short-lived and are disrupted by wind-driven mass loss. They do not virialize or migrate to the galaxy centers as suggested in recent work neglecting strong winds.” Other simulations produce pseudobulge-like, small Sérsic indices. Some results are inherently robust, such as the conclusion that gas-rich, violently unstable disks at high z gradually evolve into gas-poor, secularly evolving disks at lower redshifts (Cacciato, Dekel, & Genel 2012; cf. Ceverino, Dekel, & Bournaud 2010). However, the conclusions from the models are substantially more uncertain than the inferences from the observations. This is part of a problem that I emphasize in the next section:
Simulations of baryonic galaxy evolution inside CDM halos formed via n-body simulations of cosmological hierarchical clustering are making rapid progress as the baryonic physics gets implemented in better detail. But these simulations still show clearcut signs of missing important physics. In contrast, practitioners of this art who carefully put great effort into improving the physics tend to be overconfident about its results. We are – I will suggest – still in a situation where robust observational conclusions that are theoretically squishy are more trustworthy than conclusions based on state-of-the-art simulations, at least when baryonic physics is involved.
Another caveat is the observation that the clumps in high-z disks are much less obvious in the inferred mass distributions than they are in rest-frame optical or blue light (Wuyts et al. 2012). Frontier observations have opened up a popular new window on the formation of classical bulges, but its importance is not entirely clear.
In the present subject of bulge formation, it seems provisionally plausible that formation via high-z disk instabilites and consequent clump sinking represents a significant new channel in the formation of classical bulges. Meanwhile, a large body of work from the 1980s and 1990s continues to tell us that major galaxy mergers make classical bulges, too. Can we distinguish the results of the two processes? We do not yet know, but my second point is that Figure 1 provides a hint: Although results are still vulnerable to unknown details in (for example) feedback, it seems likely that the classical bulges produced by sinking clumps have Sérsic indices that are systematically smaller than those made by major galaxy mergers. This is one aspect of many that deserves further work. See also point (8) in Section 8.