1. Black hole accretion history of active galactic nuclei 曹新伍 中国科学院上海天文台

2. Outline 1. Introduction to accretion history of bright AGNs 2. Inefficient accretion history of faint AGNs 2.1. Different accretion modes 2.2. Constraints from the hard X-ray background 2.3. Constraints from the Eddington ratio distribution 2.4. Confront with the numerical simulation 3. Summary

3. Quasar space density as a function of redshift. 1. Introduction to accretion history of bright AGNs

4. Luminosity function describes the space number density of AGNs with luminosity at redshift The black hole mass density accreted in bright AGN phases is where is the AGN luminosity function, is radiation efficiency.

5. Local black hole mass density mainly comes from the accretion during bright AGN phases. The local black hole mass density can be estimated from the SDSS data by using relation (Yu & Tremaine 2002):

6. Results for hard X-ray LF.

7. Lifetime of bright AGNs: different investigations gives different values: years. The Hubble timescale years, The Salpeter timescale: years. (mass-doubling timescale for accreting at Eddington rate) Observations show an upper limit on massive back hole mass solar masses. Why the lifetime is so short compared with ? what halted the black hole accretion?

8. The scenario for AGN formation and evolution 1. Birth of AGNs Mergers between galaxies trigger nuclear gas flows to feed the black hole, and trigger nuclear starbursts. The nucleus is obscured by dense gas. 2. Bright AGN phase accreting at around or slightly less than the Eddington rate. The gas becomes transparent and the nucleus can be seen. 3. Death of bright AGNs radiation of AGNs expels gases to quench both accretion and star formation. 4. Faint AGN phases accreting at very low rates

9. 2.1. Different accretion modes Slim disk: optically thick, disk-thickness: Standard thin disk: optically thick, disk-thickness: Radiatively inefficient accretion flow (RIAF): optically thin, hot, disk-thickness: faint due to low radiative efficiency 2. Inefficient accretion history of faint AGNs Accretion mode transition occurs while

10. Spectra of slim disks (Wang et al. 1999) Spectra of RIAFs (Manmoto 2000) Spectra of thin disks (Laor & Netzer1989)

11. 2.2. Constraints from the hard X-ray background Observed X-ray background (Comastri 2004)

12. Synthesis models for X-ray background The cosmological XRB is mostly contributed by AGNs. A typical synthesis model consists of 1. A template X-ray spectrum for AGNs: a power-law spectrum + an exponential cutoff at several hundred keV; for example, 2. A soft/hard X-ray luminosity function for AGNs

13. Blue dashed line: observed XRB Thick solid black line: synthesis model (taken from Ueda et al. 2003) bright AGNs+Compton-thick type 2 AGNs Synthesis model based on hard X-ray luminosity function given by Ueda et al.

14. The contribution to HXRB is dominantly from bright AGNs, but HXRB (especially in ) can constrain accretion history of faint AGNs X-ray spectra of RIAFs

15. The total monochromatic X-ray luminosity of faint AGNs in the co-moving volume is ( is the faint AGN lifetime) where n f (M bh, z) is the black hole mass function for faint AGNs and Lx (M bh, E) is the X-ray luminosity of an RIAF accreting at the critical rate. The source number density where l X (E, t) is the faint-AGN light curve for a black hole with 10 8 solar mass.

16. where ( bright AGNs) described by X-ray LF, is for faint AGNs). In order to calculate the contribution to XRB from RIAFs in faint AGNs, one need to know the space number density of faint AGNs, most of which have not been observed in any waveband except those in the local universe. Every bright AGN will finally be switched to a faint AGN, so the density of faint AGNs is ( is the bright AGN lifetime) where f bh > 1 is the ratio of average black hole masses of faint AGNs to bright AGNs. The black hole mass density of faint AGNs is

17. The RIAF timescale is defined by Comparison with observed XRB can set constraints on The contribution from the RIAFs in all faint AGNs to the cosmological XRB can be calculated as

18. Cao, 2005, ApJ, 631, L101 Blue: 0.05 Red: 0.01 Green: 0.005

19. The number counts in unit logrithm of Eddington ratio is where is mainly determined by time-dependent accretion rate In principle, we can derive the accretion history from the comparison with the observed Eddington ratio distributions. 2.3. Constraints from the Eddington ratio distributions

20. Eddington ratio distributions Blue: Ho 2002; red: corrected distribution by Hopkins et al. 2005

21. Cartoon cross-section of a RIAF+standard thin disk system

22. Standard disk+RIAF, Solid: ; dashed:.

23. Blue lines: for Ho (2002)’s sample; red lines: for the sample corrected by Hopkins et al (2005). Solid lines: ; dotted lines:.

24. 2.4. Confront with the numerical simulation (Taken from Di Matteo et al., 2005, Nature, 433, 604) T~1.7Gyr, AGN radiation is peaked at around Eddington luminosity.

25. Taken from Di Matteo et al., 2005, Nature, 433, 604

26. 1. Bright AGN lifetime is years, comparable with the Salpeter timescale, which may be governed by the feedback from AGNs. 2. Accretion rate declines rapidly (compared with bright AGN lifetime) from to which provides evidence that the gas near the black hole is blown away by AGN radiation. 3. Black hole growth is not important in faint AGN phases. 3. Summary

27. The scenario for AGN formation and evolution 1. Birth of AGNs Mergers between galaxies trigger nuclear gas flows to feed the black hole, and trigger nuclear starbursts. The nucleus is obscured by dense gas. 2. Bright AGN phase accreting at around or slightly less than the Eddington rate. The gas becomes transparent and the nucleus can be seen. 3. Death of bright AGNs radiation of AGNs expels gases to quench both accretion and star formation. 4. Faint AGN phases accreting at very low rates