Black Hole Accretion and Wind Feng Yuan Shanghai Astronomical Observatory, Chinese Academy of Sciences
Accretion Regimes Hyper-accretion, slim disk, ADAF (Abramowicz et al. 1988) TDEs, ULXs, SS433 Thin accretion disk: radiatively efficient (Shakura-Sunyaev 1973) Typical QSOs, Seyferts XRBs in thermal soft state Hot Accretion, ADAF, radiatively inefficient (Narayan & Yi 94; Yuan & Narayan 2014, ARA&A) LLAGN, BL Lac objects, Sgr A*, M87 XRBs in hard state, quiescent state
OUTLINE Some observations on outflow (wind) Theoretical study of wind from hot accretion flow Existence of wind Various properties of wind (vs. jet) Mechanism of wind production Application of accretion wind: the Fermi bubbles Summary
Outflow (wind): why interesting? Accretion physics Outflow: exists or not? important ingredient of accretion physics Crucial to explain many observations of AGNs & BH binaries AGNs feedback Outflow plays an important role in AGN feedback: momentum Need to constrain their properties: mass flux, velocity, density
AGNs Some observations on wind Frequently detected in luminous AGNs: BAL quasar But also in LLAGN; e.g. NGC 4395: L bol = 10 3 L Edd (Crenshaw & Kraemer 2012) Wind and jet coexist in radio loud galaxies (Tombesi et al. 2014) Black hole X-ray binaries Usually in soft state but also in hard state (Diaz Trigo; Neilson & Homan 2012) More difficult to detect wind from hot accretion flow than form thin disk.
Accretion rate decreases inward Stone, Pringle & Begelman 1999; Stone & Pringle 2001; Hawley & Balbus 2002; Machida et al 2003; Pen et al. 2003; Igumenshchev, Narayan & Abramowicz 2003; Pang et al. 2010; Yuan & Bu 2010; Yuan, Wu & Bu 2012; Li, Ostriker & Sunyaev 2013 Inflow rate Outflow rate M r = M(r out )(r/r out ) 0.5 0.8 Net rate Stone, Pringle & Begelman 1999
Density profile flattens If Mdot is constant, then: ρ r 3/2 When Mdot decreases inward, we have ρ r 0.85 Yuan, Wu & Bu 2012
Confirmed by Observations of Sgr A* & NGC 3115 Aitken et al. 2001; Bower et al. 2003, 2005; Yuan, Quataert & Narayan 2003 Chandra observations + Bondi theory give the Bondi rate: (consistent with numerical simulation of Cuadra et al. 2006) High linear polarization at radio waveband requires innermost region accretion rate (rotation measure requirement): 7 9 1 (10 10 ) M yr So Mdot must decrease inward Density profile consistent with numerical simulation Sgr A*: NGC 3115: 10 5 M yr 1
Question: Why does Mdot decrease inward? Two models have been proposed.
Model One: ADIOS (Adiabatic inflow-outflow solution) Blandford & Begelman 1999; 2004; Begelman 2012 Assumption: Mass loss in outflow Mdot decreases Self-similar solutions (ADIOS) constructed based on this assumption Origin of outflow: positive Bernoulli? no Blandford & Begelman 2004
Model Two: CDAF (Convection-dominated accretion flow) Narayan, Igumenshchev & Abramowicz 2000; Quataert & Gruzinov 2000 Hot hydro accretion flow is convectively unstable Because entropy increases inward (Narayan & Yi 1994) It is true even when radiation is strong (Yuan & Bu 2010) A convective envelope solution is found Gas then circulates in convective eddies Mdot decreases Debate on MHD flows: applicable to MHD flow or not? No (Hawley, Balbus, Stone): dynamics of MHD flow is controlled by magnetic field & MRI Yes (Narayan, Abramowicz, Quataert): there is a convection component in the instability criteria
Which one is correct, ADIOS or CDAF? Yuan et al. (2012, 2015) Three arguments (analysis based on MHD simulation data): If convective turbulence, we expect: inflow & outflow properties roughly same; different! Analyze the convective stability of MHD accretion flow stable! Study the trajectory of ``test particles : wind really exists or convection? strong wind exists! Conclusion: significant outflow exists!
An MHD accretion flow is convectively stable The Hoiland criteria: Here: Result: Most of the region is convectively stable! Yuan, Bu & Wu 2012
Trajectory approach Yuan et al. 2015 Trajectory is different from streamline since accretion flow is never steady but turbulent Trajectory is much more time-consuming to obtain than streamline Trajectory can loyally reflect the motion of fluid elements
Yuan et al. 2015, ApJ Trajectory of virtual ``test particles BZ jet Disk-jet ( jet exists even for a=0 black hole!) wind
Disk-jet vs. BZ jet Yuan & Narayan 2014; Yuan et al. 2015 Matter dominated (not Poynting-flux dominated) Powered by the rotation of disk (not BH spin) so present even a=0 Power typically < 0.1 Mc 2 (but may still > L_disk) For non-mad, power of disk jet > power of BZ jet, even though BH spin is large! Powered by the magnetic tower mechanism (gradient of the pressure of toroidal magnetic field; Lynden-Bell 2006) (not BP mechanism) Question: Disk jet or BZ jet: which one is more relevant to observation??
Various properties of winds Angular distribution Mass flux M wind = M BH r M wind = M BH r Terminal poloidal speed v term r ~0.3v k (r) Energy flux r 20r s, a = 0 r 10r s, a = 0.9 E wind = 1 1000 M BHc 2 (Consistent with AGN feedback simulations)
Whether or where does the wind stop to be produced? M wind = M BH r r 20r s how large r can be? We run simulation, considering accretion flow in such a potential: ψ = ψ BH + ψ star ; ψ BH = GM BH /(r r s ) ψ star = σ 2 ln r + C Result: in the region where the star cluster potential dominates, no winds are produced. This is roughly the Bondi radius Reason: the flatter potential? unclear
Mechanism of outflow production Driving forces: Centrifugal force Yuan et al. 2015 Gradient of gas & magnetic pressure Comparison with Blandford & Payne (1982) BP82: large-scale field + only centrifugal We don t have large-scale poloidal field, we have both centrifugal & magnetic force (since v B!) Z/Rg 120 100 80 60 40 20 Jet Corona Wind Accretion Flow R/Rg F total F gravity 1 - r 1 2 - B r 8p P gas F centrifugal 0 20 40 60 80 100 120 140
Outflow confirmed by new observations 3Ms observation to the quiescent state of Sgr A* by Chandra H-like Fe Kα line profile fitting flat density profile outflow Wang et al. 2013, Science
Ongoing work: Does the same mechanism work for a thin disk?
An application of the wind theory: Formation of the Fermi bubbles
The Fermi Bubbles Su et al. 2010 Two giant γ-ray bubbles discovered by Fermi-LAT Morphology, Surface brightness, Spectrum
Jet Model Guo & Mathews 2012; Guo et al. 2012; Yang et al. 2012, 2013 Jet lasted for ~0.3 Myr, quenched 1-2 Myr ago Jet interacts with ISM and form the bubbles Assumptions: Jet direction Galactic plane Jet velocity low: 0.1c Jet mass loss rate large: Guo et al. 2012: 0.3 M Edd (one jet) Guo & Mathews 2012, Yang et al. 2012, 2013: 3 or 170 M Edd (one jet) Guo et al. 2012
Wind model: Inflated by accretion wind Mou et al. 2014, 2015 Past activity of Sgr A* Independent constrain on Mdot: 10^4 times higher Still a RIAF Numerical simulation model Accretion rate: from other independent constrain Properties of outflow: determined by MHD simulation of accretion flow 3D + two fluid + MHD
Results (II): γ-ray radiation Mou et al. 2015 n e Permeated Layer e 2 Left: density and velocity. Right: CR energy density and magnetic field.
Results: γ-ray radiation and others Total γ-rays: Other features: Surface brightness Temperature of shocked ISM (observed: 0.3 kev): Models Jet Radiationdriven wind Temperature (NPS) Velocity of bubble edge --- consistent with observations 7% SF-driven Wind 18% Hot Accretion Wind > 5 kev ~ 1 kev N 0.4-1 kev Age 1-2 Myr 6 Myr 10 2-10 3 Myr 7-12 Myr
Implication: other bubbles by winds or jet? The success of the wind model suggest us to consider the formation of X-ray cavities and bubbles in galaxy clusters and other galaxies such as M87. M87 Perseus cluster core 20 kpc M~1.2 (Graham+ 2008) Two Bubbles Owen et al. 2000
Wind observations: Summary winds may exist in hot accretion flow Wind coexists with jet Wind must exist in hot accretion flow: trajectory Wind production: magnetic + thermal mechanism Properties of wind: mass flux, angular distribution, terminal velocity, energy and momentum fluxes (winds vs. jet)
END