AGN Feedback at the Parsec Scale Feng Yuan Shanghai Astronomical Observatory, CAS with: F. G. Xie (SHAO) J. P. Ostriker (Princeton University) M. Li (SHAO)
OUTLINE Intermittent activity of compact radio sources Outburst: 10^4 years Quiescent: 10^5 years previous interpretation & its problem thermal instability of radiation-dominated thin disk Explaining the intermittent activity with Global Compton scattering feedback mechanism in hot accretion flows What is global Compton scattering? When L > 0.02 L_Edd: no steady solutions; BH activity oscillates Estimations of durations of active and inactive phases
AGN feedback: an important process in galaxy formation & evolution M BH correlation suppression of BH mass & star formation in elliptical galaxies Great progress made; still many details need further exploration (Ostriker 2010) seeking direct observational evidence Feedback often causes intermittent activity of AGNs Investigating feedback at various scales
Observational evidence (I): Relics and new jets
Observational evidence (II): double-double radio sources Courtesy: A. Siemiginowska
Population problem of compact young radio sources Many compact young (10^3 year) radio sources found If the total activity lasts for 10^8 yr, the number of sources with the ages < 10^3 yr should be ~ 10^5 times lower than the number of sources older than 10^3 yr But the population studies show far too many compact young sources: why?
Interpretation: intermittent activity Courtesy: A. Siemiginowska
Compact radio sources: Age 1. Kinematic age 2. Synchrotron age Typical age: <10^4 yr Czerny et al. 2009
Compact radio sources: Luminosity Typical bolometric L: 0.1L_Edd or 0.02 L_Edd (preferred) Czerny et al. 2009
Existing models for intermittent activity Galaxy merger: 10^8 year Ionization instability: 10^8 year Thermal instability of radiation-pressure dominated thin disk (Czerny et al. 2009) Limit-cycle behavior intermittent activity But two questions: Can jets be formed in standard thin disks? Is the radiation-dominated thin disk unstable?
Accretion equations ie i i ie e e k s out out q q dr d p dr d v q q q dr d p dr d v p r j r v dr dp r dr dv v R R M v RH M ) (1 ) ( 1 4 2 2 2 2 2.. Mass Conservation: Momentum Conservation: The electrons energy conservation: The ions energy conservation:
Two kinds of accretion solutions (I): the standard thin disk Shakura & Sunyaev 1973; Pringle 1981 Cool: ~10 6 K; Geometrically thin Low radial velocity Optically thick Spectrum: black body High efficiency because: Radial velocity is low and radiation is strong, so gas has enough time to radiate away its thermal energy Thus all gravitation energy is radiated away
Two types of Accretion Solutions (II): Hot accretion Hot: virial; Geometrically thick High radial velocity Optically thin Spectrum: complicated (synchrotron+bremsstrahlung+comptonization) Low efficiency because Narayan & Yi 1994; 1995 Radial velocity is high and radiation is weak So gas has no enough time to radiate away its thermal energy most of the gravitational energy is stored in the gas these energy disappear into the black hole horizon!
Accretion Solutions: summary Luminous Hot Accretion Flows Yuan 2003 Yuan 2001
Hard & soft states: spectrum Zdziarski. & Gierliński 2004
Jet can only be formed in hard states (hot accretion flows) soft/high state: Standard thin disk No radio emission without jets Low/hard state: Hot accretion flow Strong radio emission with jets Esin, McClintock & Narayan 1997
Thermal stability of Radiationdominated standard thin disks It has been thought radiationdominated thin disk (L>0.2) is thermally unstable (e.g., Piran 1978; Janiuk et al. 2002) However: Observations: Gierlinski & Done (2004): a sample of soft state BHXBs; 0.01< L/L_Edd<0.5; no variability quite stable Possible exception: GRS1915+105: L too high? Confirmed by 3D MHD Numerical Simulations (Hirose, Krolik & Blaes 2009) M Stable or not??
Two interpretations for the stability(i): Time-lag model Hirose, Krolik & Blaes 2009, ApJ r P causality Fluctuations in thermal energy are correlated to fluctuations in turbulent magnetic and kinetic energies, but with a time lag
Two interpretations for the stability (II): magnetic pressure model Zheng, Yuan, Gu & Lu 2011, ApJ Previous analysis neglected the role of magnetic pressure Assume, when T (thus H) increases, B decreases then we have: B H const. BH R Result: The critical Mdot of instability increases! Advantage: can explain why GRS 1915+105 is unstable
We propose: Global Compton heating feedback as an interpretation
Two effects of Compton scattering in accretion flows Consider collision between photons and electrons in hot accretion flow, two effects: Momentum U Radiation force: c T c Balance with grav. force Eddington luminosity Energy For photons: Compton up-scattering or Comptonization, which is the mechanism of producing X-ray emission in BH systems For electrons: they can obtain or loss energy due to the scattering with photons (e.g., Compton radiative cooling)
We will focus on electrons and non-local scattering (because hot accretion flow is optically thin in radial direction) Assume the electrons have T e and the photon energy is Є, after each scattering on average the electron will obtain energy: Thompson limit:
The spectrum received at radius r It is difficult to directly calculate the radiative transfer when scattering is important. So we use two-stream approximation, calculate the vertical radiative transfer in a zone around r. The spectrum before Comptonization is: The spectrum after Comptonization is calculated based on Coppi & Blandford (1990)
The spectrum received at radius r When calculating the radiative transfer from dr to r, we neglect for simplicity the scattering. Then from the region inside of r: From the region outside of r:
The Compton heating/cooling rate The number of scattering at radius r with unit length and es optical depth is : unit length in r So the heating/cooling rate (per unit volume of the accretion flow) at radius r is:
When Compton heating/cooling important? We compare Compton heating/cooling with viscous heating Yuan, Xie & Ostriker 2009 Result: Cooling is important when Mdot>0.01 Heating is important when Mdot>0.2 (function of r!)
Get the self-consistent solutions procedure: using the iteration method guess the value of Compton heating/cooling at each radius, solve the global solution, compare the obtained Compton heating/cooling with the guessed value to see whether they are identical. If not, use the new value of Compton heating and get the new solution until they are identical.
When Mdot is large: oscillation When L >~0.02 L_Edd, Compton heating is so strong that electrons at r_virial~10^5r_s will be heated above T_virial Thus gas will not be captured by BH, no steady hot solution exists! r virial ~ 10 5 r s (L / 2% L Edd 1/ 2 Accretion resumes after cooled down oscillation of the activity of BH )
Oscillation scenario: general picture Yuan & Li 2011 Active phase Inactive phase
Active phase Duration of active phase: accretion timescale at r_virial for L ~ 2% L Edd, r virial 5 ~ 10 r s So: why more luminous sources tend to be younger: r virial ~ 10 5 r s (L / 2% L Edd ) 1/ 2
Inactive phase What is the spatial range of heated gas during the active phase? The energy equation of electrons: The solution is: From: We get the range of heated gas:
Inactive phase Properties of heated gas: temperature: T= T_x ~ 10^9K Density=? From pressure balance with ISM: n_inact T_x = n_ism T_ISM (T_ISM~10^7 K) (how to know n_ism? L ~ 2%L_Edd Mdot n_ism) Duration of inactive phase Cooling timescale: for T_x & n_ism, t_cool~10^5 yr accretion time at 10^6r_s: >> 10^5 yr We should choose the shorter one
Summary The global Compton scattering feedback can explain: L~0.02 L_Edd More luminous sources are younger Duration of active phase: 3 10^4 yr Duration of inactive phase: 10^5 yr
Discussions: Other related works & questions Other works of feedback: scales & mechanisms Proga et al.: also small scale; momentum feedback Ciotti & Ostriker et al.: larger scales Di Matteo et al. : mainly energy feedback Murray, Quataert & Thompson 2005 Springel et al..croton et al... Questions: Momentum & energy feedback: when & which one is dominant? Small & large scales: is there any characteristic scale? Or they play their roles at various scales? If so, pc-scale feedback in this work should be taken into account
Thank you!