5.2. Theoretical aspects
The unification of the physics implied in all of these properties and phenomena is an appealing challenge. Gas-rich mergers (e.g. two disks), besides heating and thickening the preexisting disk stellar population, removes angular momentum from gas and allows the gas to concentrate further in the center. The low angular momentum gas, which dissipates probably in a turbulent regime, will form a compact disk, where stars are born in a high density environment, rich in dust and molecules. An intense burst of star formation takes place driven by a radiation pressure (Eddington) self-regulation, while the bulk of UV radiation is converted to FIR by dust (Firmani & Tutukov 1994). This phase may be identified as an ULIG. In the central region gas reaches very high density, probably due to a gravothermal catastrophe. This condition favors the growth of the preexisting black hole through an Eddington (Silk & Rees 1998; Fabian 1999; Nulsen & Fabian 1999; Blanford 1999, see also Merrifield et al. 2000) or even super-Eddington regime (Begelman 2001). The accretion, hidden by gas and dust, swallows only a minor fraction of the infalling gas. When the SMBH reaches a critical mass, the gas outflow becomes catastrophic and most of the gas is expelled. Probably, a small fraction of gas is retained in an accretion disk which continues to feed the SMBH during a few millions years (Burkert & Silk 2000). The expulsion of gas ends the FIR emission phase and allows the AGN to appear in all its brightness. The QSO phase lasts the time sufficient to exhaust any residual gas feeding. After this ultra-high luminous phase finishes, the SMBH enters in a dormant phase in the heart of the stellar spheroid, which will appear as a disky elliptical galaxy with a power law inner density profile.
This scenario gives some idea about the role of merging on the formation of spheroids. However, other astrophysical processes may influence the further evolution of spheroids. Gravitation and cooling push gas to inflow. As shown by Ciotti et al. (1991), the gas heating due to SN Ia is enough to refund the energy lost by cooling and to produce supersonic galactic winds, at least in the early phases after the SF burst. As the SN Ia rate decreases with time, the wind changes to an outflow regime. Thereafter, if the spheroid is sufficiently massive, a cooling catastrophe in the inner region converts outflow into inflow. Once inflow is established, the fate of gas in the inner region is uncertain. Low mass SF driven by thermal instabilities can drop out gas from the flow (Mathews & Brighenti 1999a, 1999b), otherwise infalling gas starts to feed the central SMBH, which enters into an AGN phase (Ciotti & Ostriker 1997, 2001). In the latter case, the gas energy balance is strongly influenced by the AGN energy injection. Gas inflow at some time is suddenly converted to an outflow, while SMBH returns to its dormant phase. At this point the cycle outflow-inflow-AGN repeats again. This conjecture, based on cooling flow, identifies an AGN driving mechanism alternative to the one based on major mergers. The cooling-flow driving mechanism for AGN finds some support in the observations of Schade et al. (2000) and Dunlop et al. (2001), who did not see strong evidence of major merging in their sample of AGN host spheroids. Several aspects of this process have to be studied taking into account, besides the stellar mass loss, the gas accretion predicted by the hierarchical scenario.
Several pieces of evidence support the idea that ULIGs represent the connection between mergers and the dust enshrouded formation of spheroids and AGNs. Even if ULIGs at present are rare systems and their contribution to the cosmic energy budget is negligible, at high redshift this contribution grows and becomes dominant (similar to AGNs) (Sanders and Mirabel 1996; Krishna & Biermann 1998; Lilly et al. 1999; Dunlop et al. 2001; Granato et al. 2001). Most ULIGs show evidence of strongly interacting disk galaxies. The FIR luminosity increases with decreasing projected nuclear separation. A double nucleus is present in the majority of "cool" systems (f25 / f60 < 0.2), while a single nucleus is more frequent in "warm" systems. The fact that ULIGs represent ellipticals in formation is supported by the evidence that these objects fall near the FP of disky ellipticals (Genzel et al. 2001).
The sequence: merger between disk galaxies "cool" ULIG "warm" ULIG "IR excess" QSO QSO, is supported by the following arguments (Sanders 2001; Kim et al. 2002; Veilleux et al. 2002). A gradual transition from FIR to UV dominated spectral energy distribution (from "cool" to "warm") is accompanied by the appearance of a "big blue bump" (characteristic of optically selected QSOs) and nuclear superwinds. QSOs with FIR excesses (in transition from "warm" ULIGs) appear related to "disturbed" hosts, while E-like ULIGs and AGNs seem more likely to lie in hosts of similar masses. Hard X-rays will provide more information about the SMBH growth in the ULIG nuclei (Fabian 1999; Wilman et al. 2000).
This scenario is consistent with a simple demographic argument. From HST, Lilly et al. (1996) estimated a local blue luminosity density of 107 L / Mpc-3 Assuming < M / LB > 14 M / L (see Dwek et al. 1998), the luminous matter density is then 1.4 × 108 M Mpc-3. If about half of the stars are assumed to be in spheroids, their mass density is then 7 × 107 M / Mpc-3 and the mass density in SMBHs will be 4 × 105 M Mpc-3, where the ratio MSMBH / MSPH = 0.006 has been assumed (Magorrian et al. 1998). The efficiency of mass converted in FIR during a star burst (Firmani & Tutukov 1994) is 5.3 × 10-4, then the FIR energy density produced by the star bursts of the spheroids is 4 × 104 M Mpc-3. Assuming an efficiency of 0.1 for the energy produced by the SMBH accretion, the energy density derived from this process is 4 × 104 M Mpc-3, an amount similar to the one derived from spheroids. The sum of both agrees with the total FIR energy density of 105 M Mpc-3 obtained by Dwek et al. (1998) in the DIRBE experiment.
A further argument supporting this scenario is given by Granato et al (2001). According to them, the spheroid formation rate, derived from the observed QSO luminosity function, leads to the observed spheroid number density at present. The anti-hierarchical baryonic collapse scenario of Granato et al. is consistent with the FIR source counts, the nature of Ly Break Galaxies (LBGs) and the chemical evolution of spheroids and QSOs. The SF in spheroids, correlated with the QSO luminosity function, shows that its bulk of activity happens at z 2.
The previous scenario is based on some important elements. (i) Gas infall has to reach very high density at the radius of BH influence in order to produce a sufficient BH growth. (ii) The SMBH is formed by a self-regulation mechanism that determines its final mass and induces a natural expulsion of gas which reduces suddenly the optical depth. (iii) The bulk of SF and AGN activity is strongly enshrouded by gas and dust.
If a collision occurs between two gas-poor stellar disks the remnant is not sufficiently dense to be identified with a real spheroid (Hernquist 1992). The presence of bulges in the progenitors plays an important role leading to a central dense spheroid comparable to the observed bulges (Hernquist 1993). The two SMBHs of the progenitors will coalesce spiralling-in towards the center by dynamical friction and gravitational wave emission, creating here a shallow stellar core similar to the ones observed in boxy elliptical galaxies (Makino & Ebisuzaki 1996; Quinlan 1997; Quinlan & Hernquist 1997; Milosavljevic et al. 2002).