The origin of SMBHs that power the luminous quasars at high redshifts remains unknown. Spectroscopic observations revealed that BHs with mass greater than 10 billion solar masses were already in place when the age of the universe was less than one billion years (for a review, see Fan, Carilli & Keating 2006). Potentially, the existence of such early SMBHs might pose a challenge to the current cosmological standard model which is based on bottom-up, hierarchical structure formation. The observed SMBHs have likely grown from some smaller seed BHs that were formed earlier, in the progenitors of the luminous quasar host. The first galaxies were plausible sites for seed BH formation, but their own structure and evolution was likely affected by the presence of such early BHs as well. We thus have to tackle a complex, feedback-regulated problem, where our current knowledge is patchy at best.
It is instructive to consider a schematic representation of possible SMBH formation pathways inside the first galaxies (see Figure 9). Figure 9 is reproduced from Regan & Haehnelt (2009b), who in turn adopt the well-known flow-chart towards SMBH formation introduced by Rees (1984). The key bifurcation concerns whether the gas inside the first galaxy, here taken to be an atomic cooling halo, can cool below ~ 104 K or not. Such cooling depends on the presence of either H2 or heavy-element coolants. To prevent molecular hydrogen from forming, the presence of an extremely strong LW radiation background, capable of photo-dissociating H2 even in the presence of self shielding would need to be invoked (Bromm & Loeb 2003b; Wise, Turk & Abel 2008; Dijkstra et al. 2008; O'Shea & Norman 2008; Regan & Haehnelt 2009a; Shang, Bryan & Haiman 2010). To maintain metal-free conditions in the first galaxies, star formation and SN activity in the progenitor minihalos would have to be suppressed, which may be possible in a subset of cases, in ~ 10-20% of atomic cooling halos collapsing at z 10 (Johnson et al. 2008). Below we discuss some of the SMBH formation pathways in greater detail.
Figure 9. Pathways towards the first supermassive black holes. Here, possible SMBH formation channels in high-redshift atomic cooling halos are shown. The main bifurcation arises from whether the gas inside the first galaxy can cool below ~ 104 K, via H2 or metal cooling, or not. If the gas can cool, star formation will ensue. SMBH formation would then have to rely on stellar-dynamical processes of catastrophic runaway collisions. In the opposite case, the path towards a SMBH involves gas-dynamical processes, possibly resulting in the intermediate stage of a supermassive star (or quasi-star). Such a star would rapidly turn into a SMBH. Adopted from Regan & Haehnelt (2009b).
5.1. Formation Models
Devising viable models for SMBH formation has been a long-standing challenge in astrophysics (Rees 1984). The requirements on such models are even more stringent in the high-redshift case, where any formation channel has to operate on rapid timescales. There are currently two main ideas, one based on (Pop III) stellar seeds, some of them may grow via gas accretion and BH mergers, and one on the direct collapse of massive primordial gas clouds. Both classes of models face challenges, leaving still open the possibility for alternative, more exotic pathways towards SMBH formation.
5.1.1. POPULATION III STELLAR REMNANTS A popular model assumes that the remnant BHs of Pop III stars seeded the growth of SMBHs (Madau & Rees 2001; Li et al. 2007; Volonteri & Rees 2006; Tanaka & Haiman 2009). In this case, the initial seed mass would be of order 100 M. Given efficient, Eddington-limited accretion, even such low-mass seeds could readily grow to the SMBHs inferred to power the high-z SDSS quasars in the roughly 500 Myr between seed formation and z ~ 6 (Haiman & Loeb 2001). Recent studies suggest, however, that the gas accretion onto early BHs is inefficient until the BHs are incorporated into larger mass halos. One impeding effect is that the gas is already evacuated by photoionization heating from the progenitor massive star (Kitayama et al. 2004; Whalen et al. 2004; Alvarez, Bromm & Shapiro 2006; Abel, Wise & Bryan 2007). After the progenitor star has died and directly collapsed into in intermediate mass BH, it thus finds itself in a very low-density region. Accretion rates are then negligible for at least the free-fall time of the dark matter host systems (Johnson & Bromm 2007; Pelupessy, Di Matteo & Ciardi 2007; Alvarez, Wise & Abel 2009). In addition, the radiative feedback from the accreting BH can reduce the cooling of the surrounding gas, e.g., by photo-dissociating H2, thus further reducing accretion. Even if the gas supply in the vicinity of the remnant BH has been replenished, accretion likely continues to be severely suppressed compared to the Eddington rate. This is because of radiation pressure on the high-density infalling gas (Milosavljevic et al. 2009a, b). As a result, an episodic, quasi-periodic accretion flow is established, with a time-average significantly below the Bondi-Hoyle and Eddington rates (see Figure 10).
Figure 10. Accretion onto the first black holes. Gas number densities n (top row), and neutral fraction H (bottom row) in the vicinity of the accreting black hole. Shown is the situation during an accretion minimum (left column), and during a maximum (right column). At maximum, central densities are high, and the H II region grows in response. The structure near the vertical axis (dashed lines) is a numerical artifact. The resulting hydrodynamics is complex, exhibiting overlapping in- and outflows that establish an episodic pattern of accretion and radiation-pressure feedback. Adopted from Milosavljevic, Couch & Bromm (2009b).
This early bottleneck for growing the seeds to SMBHs poses a serious challenge to the Pop III stellar remnant scenario. However, it is important to note that the emergence of SMBHs should not be too common, to be compatible with the abundance of observed luminous quasars (Tanaka & Haiman 2009). It is not necessary that a particular process is able to feed all seed BHs efficiently, although there must be at least one physical mechanism that enables the early formation of SMBHs perhaps under some extraordinary conditions. Models that invoke special conditions such as super-Eddington growth in accretion disks might therefore be acceptable solutions to the early bottleneck problem.
5.1.2. DIRECT COLLAPSE The early bottleneck to growth described above arises because of the negative feedback from star formation. In principle, the same is true for the rapid collapse of more massive clouds (Loeb & Rasio 1994; Eisenstein & Loeb 1995). However, there is again an intriguing possibility in atomic cooling halos. If H2 and metal cooling were suppressed, atomic hydrogen cooling could still allow the gas to collapse into the halo with Tvir ~ 104 K. But due to the absence of lower temperature coolants, the collapse would proceed isothermally without any sub-fragmentation, and therefore without star formation. Recently, the atomic cooling halo pathway has received considerable attentions, both from the simulation side (Bromm & Loeb 2003b; Wise, Turk & Abel 2008; Regan & Haehnelt 2009a; Johnson et al. 2010; Latif, Zaroubi & Spaans 2011; Shang, Bryan & Haiman 2010), and with analytical work (Begelman, Volonteri & Rees 2006; Lodato & Natarajan 2006, 2007; Spaans & Silk 2006). The key question is whether the gas can indeed remain free of H2 molecules (Dijkstra et al. 2008; Ahn et al. 2009), and of metals (Johnson, Greif & Bromm 2008; Omukai, Schneider & Haiman 2008) Again, it is important to remember that such a mechanism, where already more massive seed BHs with 104 M form via direct collapse of a primordial gas cloud, needs to successfully operate only in a few, rare cases. Indeed, if every atomic cooling halo were to produce a massive seed BH in its center at z 10, we would exceed the locally measured total BH mass density (e.g., Yu & Tremaine 2002). Fragmentation may also be suppressed by the strong turbulence in inflows with high Mach number, where gas temperatures are significantly below the virial temperature (Begelman & Shlosman 2009). This scenario still needs to be tested, however, with realistic simulations. Recently, a qualitatively different variant of massive seed BH formation during direct collapse has been suggested (Mayer et al. 2010). In this model, two very massive ( ~ 1013 M) halos merge at high redshifts, triggering massive inflows into the center of the ensuing potential well on such a rapid timescale that negative feedback from star formation has no opportunity to interfere with the BH assembly process. It is not entirely clear, however, whether such a set-up will occur in a realistic cosmological setting.
5.1.3. OTHER MODELS Overall, there appears to remain a large uncertainty in these models. The Pop III seed model requires a number of optimistic assumptions on the efficiency of gas accretion and multiple BH mergers, whereas the rapid collapse model critically relies on the assumption that a massive BH does indeed form in a hot, dense gas cloud. Alternative models for SMBH formation have also been proposed recently. Primordial stars powered by dark matter annihilation (Spolyar et al. 2008; Iocco et al. 2008; Umeda et al. 2009) are suggested to have long lifetimes, because they do not consume hydrogen by nuclear burning. If such objects continued to accrete the surrounding gas, they could grow to become more massive than 105 M. Such very massive "dark stars" can be as luminous as ~ 1010 L, in principle detectable with JWST (Freese et al. 2010), and they can also collapse to massive BHs at their death.
5.2. SMBH-First Galaxy Coevolution
It is well-known that in the local Universe, there is a tight correlation between the bulge properties of a galaxy and the mass of its central BH (Gebhardt et al. 2000; Ferrarese & Merritt 2000). Whether or not the same relationship holds in the young Universe is an intriguing question. Volonteri & Natarajan (2009) argue that a similar relationship can be quickly established, and that it would be mainly driven by accretion onto BHs after major mergers of the host galaxies. Coevolution of the first galaxies and early BHs might be a key in shaping the high-redshift galaxies, as has been advocated for somewhat lower-redshift galaxies (Di Matteo, Springel & Hernquist 2005). The detailed study of the star-formation history of z > 6 galaxies might provide clues as to whether star formation was episodic, both within themselves and in their progenitor systems (e.g., Labbe et al. 2010).