1. Super-Critical Accretion Flow? Shin Mineshige (Kyoto Univ.) with K. Ohsuga, K. Vierdayanti, K. Ebisawa, T. Kawaguchi AGN conference @ Xian (10/16/2006) 1.General introduction 2.Slim disk model (simplified model) 3.Observational tests - case of ULXs 4.2D radiation-hydrodynamical (RHD) simulations

2. 1. Introduction (background) It is widely believed that the luminosity (L) of any accreting objects cannot exceed the Eddington luminosity (L E ). ・ What is the Eddington luminosity? ・ What do we know from observations ?

3. Spherical accretion system cannot shine at L > L E. What is the Eddington luminosity? What is the Eddington luminosity? radiation pressure accreting gas accreting gas Gravity > rad. pressure ⇒ GM/r 2 >κF/c ⇒ L < L E =4πCGM/κ ( ∵ F=L/4πr 2 )

4. Super-Eddington flux (F > L E /4πr 2 ) is possible because of radiation anisotropy (!?) Disk accretion may achieve L>L E Disk accretion may achieve L>L E radiation pressure accreting gas accreting gas

5. BH Low- energy photons trapped photons Low- energy photons High- energy photons radiative diffusion & accretion (c) K. Ohsuga What is Photon trapping? Begelman (1978), Ohsuga et al. (2002) When photon diffusion time (Hτ/c) exceeds accretion time, photons are trapped within flow. r trap ~ (Mc 2 /L E ) r s.

6. Ultra Luminous X-ray sources (ULXs) Ultra Luminous X-ray sources (ULXs) Makishima et al. (2000), van der Karel (2003) Bright (~10 40 erg s -1 ) compact X-ray sources Successively found in off-center regions of nearby galaxies. If L 100 M sun. L E ~ 10 38 (M/M sun ) erg s -1 Two possibilities Sub-critical accretion onto intermediate-mass BHs (M>100M sun ). Super-critical accretion onto stellar-mass BHs (M~3-30M sun ). IC342 galaxy

7. Narrow-line Seyfert 1 galaxies (NLS1s) What are NLS1s? Narrow “broad lines” (< 2000 km s -1 ) Seyfert 1 type X-ray features Extreme soft excess & variability Seem to contain less massive BHs Good analogy with stellar-mass BHs in their very high (large L ) state. High T bb ( ∝ M BH -1/4 ) ⇒ large soft excess Small (GM BH /R BLR ) 1/2 ⇒ narrow line width Boller et al. (NewA 44, 2000) NLS1s = high L/L E system

8. 2. Slim disk model Slim disk model was proposed for describing high luminosity flows as an extension of the standard disk. ・ What is distinct from the standard disk model? ・ What are its observational signatures?

9. Basics This occurs within trapping radius r trap ~ (Mc 2 /L E ) r s Model Similar to the standard disk model; radially one-zone model but with radiation entropy advection Slim disk model Slim disk model Abramowicz et al. (1988); Watarai et al. (2000) log L/L E L=L E log m ≡ log M/(L E /c 2 ).. accretion energy trapped photons. partly

10. Slim-disk calculations Slim-disk calculations Mineshige, Manmoto et al. (2002) Low M r in ～ 3r S ;T eff ∝ r -3/4 High M r in ～ r S ; T eff ∝ r -1/2 3 rS3 rS. M/(L E /c 2 )=1,10,10 2,10 3 M BH =10 5 M sun.. Slim-disk signatures 1.small innermost radius 2.flatter temp. profile Wang & Zhou (1999), Watarai & Fukue (1999)

11. Disk spectra = multi-color blackbody spectra Temperature profiles affect spectra T ∝ r -p ⇒ F ∝ ν 3-(2/p) ・ standard disk (p =3/4) ⇒ F ∝ ν 1/3 ・ slim disk (p =1/2) ⇒ F ∝ ν -1 Spectral properties Spectral properties (e.g. Kato et al. 1998) ν 1/3 ν -1 hνhν FνFν hν hν FνFν (small r)

12. 3. Observational tests: Case of ULXs We examined the XMM/Newton data of several ULXs which were suggested to be candidates of intermediate-mass black holes. ・ How to test the theory? ・ What did we find?

13. P-free disk model P-free disk model (Mineshige et al. 1994) Fitting with superposition of blackbody (B ν ) spectra: Fitting parameters: T in = temp.of innermost region ( ～ max. temp.) r in = size of the region emitting with B ν (T in ) p = temperature profile ⇒ Good fits to the Galactic BHCs with p=0.75

14. Fitting with disk blackbody (p=0.75) + power-law We fit XMM-Newton data of several ULXs ⇒ low T in ～ 0.2 keV and photon index ofΓ=1.9 If we set r in ~ 3 r S, BH mass is M BH ~ 300 M sun. Spectral fitting 1. Conventional model Spectral fitting 1. Conventional model (Roberts et al. 2005) NGC 5204 X-1 log hν log conts

15. Spectral decomposition of DBB & PL components DBB comp. is entirely dominated by PL comp. Can we trust values derived from the minor component? Problem with DBB+PL fitting log hν log conts NGC 5204 X-1

16. P-free model fitting, assuming T ∝ r -p We fit the same ULX data but with p-free model only ⇒ high T in ~ 2.5 keV and p =0.50 ±0.03 (no PL comp. ) M BH ~ 12 M sun & L/L E ~ 1, supporting slim disk model. Spectral fitting 2. p-free model Spectral fitting 2. p-free model (Vierdayanti et al. 2006, PASJ in press) NGC 5204 X-1 log hν log conts

17. Why can both models give good fits? Because the spectral shapes are similar below ~10 keV. Both show f ν ∝ ν -1 in 0.1~10keV range. power-law (Γ=2) p-free with p=0.5

18. log kT (keV) log L x Spectral fitting: Summary Spectral fitting: Summary (Vierdayanti et al. 2006, PASJ in press; astro-ph/0609017) No evidence of IMBHs so far P-free model fitting gives M BH <30M sun. Low-temperature results should be re-examined!! The same test is needed for NLS1s

19. 4. Radiation Hydrodynamical simulations The slim-disk model applies only to the flow with L ~ L E. For even higher L, we need radiation-hydrodynamical (RHD) simulations. ・ What are the properties of the simulated flow? ・ What can we understand them?

20. Our RHD simulations Our RHD simulations Ohsuga, Mori, Nakamoto, S.M. (2005, ApJ 628, 368) First simulations of super-critical accretion flows in quasi-steady regimes. (cf. Eggum et al. 1987; Kley 1989; Okuda 2002; …) Matter with 0.45 Keplerian A.M. is continuously added through the outer boundary → disk-outflow structure Flux-limited diffusion adopted. Mass input rate: 1000 (L E /c 2 ) → luminosity of ~3 L E Injection BH r/rs r/rs z/rs z/rs 500 Initially empty disk

21. (c) K. Ohsuga Overview of 2D super-critical flow Overview of 2D super-critical flow BH r/rs r/rs z/rs z/rs density contours & velocity fields Ohsuga et al. (2005, ApJ 628,368) gas energy radiation energy density density disk flow outflow Case of M =10 M sun & M =1000 L E /c 2 ・

22. Why is accretion possible? Why is accretion possible? Ohsuga & S.M. (2006, submitted) Radiation energy density is high; E rad ≫ E Edd ≡ L E /4πr 2 c. Then why does the radiation not prevent accretion? Note radiation energy flux is F rad ∝ (κρ) -1 ∇ E rad. → High ρ and flat E rad profile result in weak P rad force. Low ρ and strong P rad yield super-Eddington flux.

23. Gas & radiation flow: summary Gas & radiation flow: summary Ohsuga & S.M. (2006, submitted) In the disk region: large gas density flat E rad profile (+ photon trapping) → weak P rad force → slow accretion In the outflow region: low gas & rad. density → strong P rad force → high-vel. outflow

24. The observed luminosity is sensitive to the viewing- angle; Maximum L ~ 12 L E !! luminosity BH  Density contours 12 8 4 4  D 2 F(  )/L E our simulations (c) K. Ohsuga Significant radiation anisotropy Significant radiation anisotropy ⇒ mild beaming Ohsuga et al. (2005, ApJ 628,368) 0 viewing angle

25. What are the causes of beaming? Photon energy increases as θ decreases, why? - Because we see photons from deeper, hot region. - Because of (relativistic) Doppler effect. Photon number increases as θ decreases, why? - Because of anisotropic gas distribution. - Because I ν /ν 3 is Lorentz invariant & hν increases. radiation gas Heinzeller, S.M. & Ohsuga (2006, MNRAS, astro-ph/0608263)

26. Further issues (1) Line spectra of super-critical flow X-ray absorption in quasars? (Pounds et al. 2003) Outflow of 0.05-0.2c in X-ray spec. Mass outflow rate ~ the Eddington rate Can be explained by P rad driven wind. Same for BAL quasars?

27. Further issues (2) Interaction with magnetic fields Magnetic fields are essential ingredients in disks Photon-bubble instability (Begelman 2002) Magnetic tower jet → global RHD+MHD simulations necessary (Kato, S.M., & Shibata 2004)

28. Conclusions Near- and super-critical accretion flows seem to occur in many systems (ULXs, NLS1s…?). Slim disk model predicts flatter temperature profile. Spectral fitting with p-free model proves the presence of supercritical accretion in ULXs. How about NLS1s? 2D RHD simulations of super-critical flow show super- Eddington luminosity, significant radiation anisotropy (beaming), photon trapping, high-speed outflow, etc. L can be ~10 L E !! Issues: line spectra, magnetic fields, jet formation,…