Paper detail

The cosmological consequence of an obscured AGN population on the radiation efficiency

In light of recent indications for a large population of obscured active galactic nuclei (AGNs), we revisit the mean radiation efficiency from accretion onto supermassive black holes (SMBHs), epsilon, applying a bayesian approach. We use the integrated comoving energy density emitted by AGNs and compare it to the mass density of local SMBHs. When considering only optically-selected unobscured AGNs, we derive log_10[epsilon]=-1.77+0.16-0.11 or epsilon=1.7+0.8-0.4%. Using the AGNs selected using hard X-rays, which include unabsorbed and Compton-thin absorbed AGNs, we find log_10[epsilon]=-1.20+0.15-0.10 or epsilon=6.4+2.6-1.3%. Using optically-selected AGNs, and correcting for the obscured population, we inferr log_10[epsilon]=-1.17+0.11-0.08 or epsilon=6.7+1.9-1.1%, which we consider our best estimate. We also repeat this calculation for intrinsically luminous AGNs (M_B<-23, quasars), comparing to the SMBH mass density in local elliptical galaxies, and find log_10[epsilon]=-1.27+0.15-0.11 or epsilon=5.4+2.2-1.2%. We discuss how our results can be used to estimate the mean spin of accreting SMBHs, both assuming the standard thin-disk model of accretion onto black holes and also comparing to more recent simulations. Our results can rule out maximally rotating SMBHs (<a>=0.998 Gm_bh/c^2) at the >=98% confidence level, as well as high rotation values (<a> >=0.9 Gm_bh/c^2) with >=92% confidence. Our preferred values of <a> are ~0.25-0.60 Gm_bh/c^2, although even these might be overestimated. Hence, we find that on average, SMBHs are not rapidly spinning during accretion. Finally, using an independent measurement of Eddington ratios, we estimate the SMBH e-folding time for the brightest AGNs (quasars) to be <tau> =100+151-60 Myr.