2007ApJ...661..660K

We discovered a significant anticorrelation between the mass of a supermassive black hole (SMBH), M_BH, and the luminosity ratio of infrared to active galactic nuclei (AGN) Eddington luminosity, L_IR/L_Edd, over 4 orders of magnitude for ultraluminous infrared galaxies with type 1 Seyfert nuclei (type 1 ULIRGs) and nearby QSOs. This anticorrelation (M_BHvs. L_IR/L_Edd) can be interpreted as the anticorrelation between the mass of a SMBH and the rate of mass accretion onto a SMBH normalized by the AGN Eddington rate, M{dot}_BH/M{dot}_Edd. In other words, the mass accretion rate M{dot}_BH is not proportional to that of the central BH mass. Thus, this anticorrelation indicates that BH growth is determined by the external mass supply process, and not the AGN Eddington-limited mechanism. Moreover, we found an interesting tendency for type 1 ULIRGs to favor a super-Eddington accretion flow, whereas QSOs tended to show a sub-Eddington flow. On the basis of our findings, we suggest that a central SMBH grows by changing its mass accretion rate from super-Eddington to sub-Eddington. According to a coevolution scenario of ULIRGs and QSOs based on the radiation drag process, it has been predicted that a self-gravitating massive torus, whose mass is larger than a central BH, exists in the early phase of BH growth (type 1 ULIRG phase) but not in the final phase of BH growth (QSO phase). At the same time, if one considers the mass accretion rate onto a central SMBH via a turbulent viscosity, the anticorrelation (M_BH vs. L_IR/L_Edd) is well explained by the positive correlation between the mass accretion rate M{dot}_BH and the mass ratio of a massive torus to a SMBH.