1. Cooling Flows & Galaxy Formation James Binney Oxford University

2. Outline Cooling flows – historical introduction Cooling flows – historical introduction Current issues in CF dynamics Current issues in CF dynamics Much work from Henrik Omma’s (05) thesis Much work from Henrik Omma’s (05) thesis SEQUEL TOMORROW SEQUEL TOMORROW Implications for galaxy formation and BH growth Implications for galaxy formation and BH growth

3. “Cooling flows” Potentials of E galaxies & galaxy clusters filled with gas @ T vir (10 6 – 10 8 K) Potentials of E galaxies & galaxy clusters filled with gas @ T vir (10 6 – 10 8 K) Detected in Xrays since early 1970s (forman et al 72; Mitchell et al 76) Detected in Xrays since early 1970s (forman et al 72; Mitchell et al 76) First model (Cowie & B 1977) involved mass-conserving flow to centre First model (Cowie & B 1977) involved mass-conserving flow to centre Predicted j X (R) inconsistent with Einstein images Predicted j X (R) inconsistent with Einstein images Stewart et al 84

4. Distributed mass drop-out Consistency with measured j X (r) obtained by assuming ICM multiphase (Nulsen 86) Consistency with measured j X (r) obtained by assuming ICM multiphase (Nulsen 86) Field instability analysis implied runaway cooling of overdense regions (t cool / 1/  ) Field instability analysis implied runaway cooling of overdense regions (t cool / 1/  ) Cooler regions radiate all E while at r À 0 Cooler regions radiate all E while at r À 0 Predicts that there should be (a) cold gas and (b) line radiation from T<10 6 K throughout inner cluster Predicts that there should be (a) cold gas and (b) line radiation from T<10 6 K throughout inner cluster

5. G modes Malagoli et al (87): overdense regions just crests of gravity waves Malagoli et al (87): overdense regions just crests of gravity waves In half a Brunt-Vaisala period they’ll be underdensities. In half a Brunt-Vaisala period they’ll be underdensities. Oscillations weakly overstable ( Balbus & Soker 89 ) but in reality probably damped. Oscillations weakly overstable ( Balbus & Soker 89 ) but in reality probably damped. Conclude: over timescale <t cool heating must balance radiative losses Conclude: over timescale <t cool heating must balance radiative losses Systems neither cooling nor flowing! Systems neither cooling nor flowing!

6. AGN Heating AGN natural heaters AGN natural heaters Cooling first becomes catastrophic @ centre Cooling first becomes catastrophic @ centre Where there’s a massive BH Where there’s a massive BH Accretion onto BH will be sensitive to local  gas Accretion onto BH will be sensitive to local  gas BH could heat through (a) Compton scattering (Ciotti & Ostriker 97, 01) or (b) jets BH could heat through (a) Compton scattering (Ciotti & Ostriker 97, 01) or (b) jets With point-like heat source expect generation of adiabatic core With point-like heat source expect generation of adiabatic core In Tabor & Binney (93) growing core matched to CF envelope In Tabor & Binney (93) growing core matched to CF envelope In Binney & Tabor (95) jets episodically heat gas in distributed fashion In Binney & Tabor (95) jets episodically heat gas in distributed fashion

7. 1993 - 2001 Distributed mass dropout still regarded as established fact in mainstream (Fabian 94) Distributed mass dropout still regarded as established fact in mainstream (Fabian 94) Conflicts with observation finessed with epicycles: Conflicts with observation finessed with epicycles: Internal absorption (Allen & Fabian 97) Internal absorption (Allen & Fabian 97) Magnetic locking (Tribble 89, Balbus 91) Magnetic locking (Tribble 89, Balbus 91) Abundance anomalies (Fabian et al 01) Abundance anomalies (Fabian et al 01) Conduction from large to small r (Bertschinger & Meiksin 86, Narayan & Medvedev 01) Conduction from large to small r (Bertschinger & Meiksin 86, Narayan & Medvedev 01)

9. 2001 – Chandra & XMM-Newton XMM doesn’t see lines of <10 6 K gas XMM doesn’t see lines of <10 6 K gas XMM shows that deficit of photons at <1keV not due to internal absorption XMM shows that deficit of photons at <1keV not due to internal absorption But associated with “floor” T ' T vir /3 But associated with “floor” T ' T vir /3 Chandra shows that radio plasma has displaced thermal plasma Chandra shows that radio plasma has displaced thermal plasma (Bohringer et al 02) (Peterson et al 02)

10. Bubble Models ( Churazov et al 2001; Quilis et al 2001; Brueggen & Kaiser 2001,2002; Brueggen et al 2002) Start with elliptical high-T cavity Start with elliptical high-T cavity Watch it rise Watch it rise Cavity can’t be in pressure equlibrium with surroundings Cavity can’t be in pressure equlibrium with surroundings The flow field around cavity dynamically important The flow field around cavity dynamically important Need for jet simulations Need for jet simulations Churazov et al 01

11. Injection Models (Quillis et al 01, Brueggen & Kaiser 02) Add thermal energy at some fixed off- centre location Add thermal energy at some fixed off- centre location Poor representation of effects of moving jet hot-spot Poor representation of effects of moving jet hot-spot Brueggen & Kaiser 02

12. Jet Simulations Early simulations 2D (Reynolds et al 02, Vernaleo 06) Early simulations 2D (Reynolds et al 02, Vernaleo 06) Or on non-refined grids (Basson & Alexander 03) Or on non-refined grids (Basson & Alexander 03) Usually there’s a spherical boundary around the origin with free-flow condition Usually there’s a spherical boundary around the origin with free-flow condition Omma et al (04) eliminated this boundary and had novel scheme for firing jets Omma et al (04) eliminated this boundary and had novel scheme for firing jets

13. Omma’s Simulations Simulations on 3d hydro with adaptive grids (Bryan’s code ENZO) Simulations on 3d hydro with adaptive grids (Bryan’s code ENZO) Entropy (no cooling) Entropy (no cooling) Entropy Density (no cooling) Density (no cooling) Density Entropy (cooling) Entropy (cooling) Entropy Density (cooling) Density (cooling) Density Key processes: Key processes: 1) Uplift 1) Uplift 2) Mixing 2) Mixing 3) Excitation of non-linear gravity waves 3) Excitation of non-linear gravity waves

14. Outward increasing entropy Omma thesis 05 Donahue 04

15. Current Issues 1) enough E? Probably Quasar mode & Radio-galaxy mode depending on whether accreting cold or hot gas (Binney 04, Croton et al 06) Probably Quasar mode & Radio-galaxy mode depending on whether accreting cold or hot gas (Binney 04, Croton et al 06) In RG mode L ¿ L Edd and ~all output mechanical (Virgo A prime example) In RG mode L ¿ L Edd and ~all output mechanical (Virgo A prime example)

16. In M87 Chandra resolves r Bondi Chandra resolves r Bondi M Bondi = 0.1 M ¯ /yr (Di Matteo et al 03) M Bondi = 0.1 M ¯ /yr (Di Matteo et al 03) So L = 5 £ 10 44 erg/s if 0.1mc 2 released So L = 5 £ 10 44 erg/s if 0.1mc 2 released L X (<20kpc) = 10 43 erg/s (Nulsen & Boehringer) L X (<20kpc) = 10 43 erg/s (Nulsen & Boehringer) L X (AGN) < 5x10 40 erg/s L X (AGN) < 5x10 40 erg/s L Mech (jet) = 10 43 – 10 44 erg/s (Reynolds et al 96; Owen et al 00) L Mech (jet) = 10 43 – 10 44 erg/s (Reynolds et al 96; Owen et al 00) So BH accreting at near M Bondi & heating on kpc scales with high efficiency (Binney & Tabor 95) So BH accreting at near M Bondi & heating on kpc scales with high efficiency (Binney & Tabor 95)

17. Current Issues 2) the duty cycle AGN known to be unsteady AGN known to be unsteady Energy dissipated @ centre only if jet channels have quiet time (or jets precess) (Omma & Binney 04, Vernaleo & Reynolds 06) Energy dissipated @ centre only if jet channels have quiet time (or jets precess) (Omma & Binney 04, Vernaleo & Reynolds 06) Sometimes two generations of bubbles (Birzan et al 04) Sometimes two generations of bubbles (Birzan et al 04) Suggests inter-outburst time ~ rise time ~100Myr Suggests inter-outburst time ~ rise time ~100Myr E of outburst > 2.5PV of bubble E of outburst > 2.5PV of bubble Suggests L mech ~ L X Suggests L mech ~ L X Actually L mech may be significantly larger Actually L mech may be significantly larger

18. Define cavities by  <  0 /4 Define cavities by  <  0 /4 Evaluate PV Evaluate PV Peaks at only 10% of actual input Peaks at only 10% of actual input Omma 05

19. Current Issues 3) does mixing destroy Z gradients? Follow tracer dye from (a) r<5kpc, and (b) 5<r<77 kpc Follow tracer dye from (a) r<5kpc, and (b) 5<r<77 kpc Omma 05

20. Effect on Z gradient Boehringer et al 04 Omma thesis 05

21. Current Issues 4) fixing the radial density profile For steady state, E(r) must match j X (r) For steady state, E(r) must match j X (r) Why do clusters have similar j x profiles? Why do clusters have similar j x profiles? Effervescent heating? (Roychowdhury et al 05) Effervescent heating? (Roychowdhury et al 05) Damped sound waves? (Fabian et al 04, Ruszkowski et al 05) Damped sound waves? (Fabian et al 04, Ruszkowski et al 05) Other physics? (Vernaleo & Reynolds 06) Other physics? (Vernaleo & Reynolds 06)

22. Omma & Binney 04 A more powerful jet disrupts further out A more powerful jet disrupts further out A more concentrated profile disrupts jet further in A more concentrated profile disrupts jet further in Later jet ignition → bigger outburst Later jet ignition → bigger outburst Later ignition → more centrally concentrated density profile Later ignition → more centrally concentrated density profile So later ignition ! strong, centrally concentrated heating So later ignition ! strong, centrally concentrated heating

23. Simulations Start from present configuration of Hydra (David et al 2000) Start from present configuration of Hydra (David et al 2000) Wait (i) 262 Myr (ii) 300 Myr Wait (i) 262 Myr (ii) 300 Myr In (ii) extra 4x10 59 erg lost to radiation, so add 8x10 59 erg rather than 4x10 59 erg as in (i) In (ii) extra 4x10 59 erg lost to radiation, so add 8x10 59 erg rather than 4x10 59 erg as in (i) E Bondi =5(M/10 9 M ¯ ) 2 £ 10 59 erg in 262Myr; E Bondi =7(M/10 9 M ¯ ) 2 £ 10 59 erg in 300Myr E Bondi =5(M/10 9 M ¯ ) 2 £ 10 59 erg in 262Myr; E Bondi =7(M/10 9 M ¯ ) 2 £ 10 59 erg in 300Myr

24. Outbursts have undone 300 Myr of cooling Outbursts have undone 300 Myr of cooling System with later ignition ends less centrally concentrated System with later ignition ends less centrally concentrated Implies that systems can oscillate around an attracting profile Implies that systems can oscillate around an attracting profile

25. Current Issues 5) Shocks Unsharp-masked X-ray images show ripples (Fabian et al 03, 06; Forman et al 03) Unsharp-masked X-ray images show ripples (Fabian et al 03, 06; Forman et al 03) Are these sound waves / weak shocks? Are these sound waves / weak shocks? Expected T variations not seen (Fabian et al 06) Expected T variations not seen (Fabian et al 06) Or Gravity waves? Or Gravity waves?

26. Conclusions “Cooling flows” thermostated by AGN “Cooling flows” thermostated by AGN This was predicted in early 90s This was predicted in early 90s AGN are in “radio mode” and have high mechanical efficiency AGN are in “radio mode” and have high mechanical efficiency They heat episodically via jets (non-adiabatic) They heat episodically via jets (non-adiabatic) Central gas density regulates energy production Central gas density regulates energy production Profile of heat generation regulated by density profile of gas via radius of jet disruption Profile of heat generation regulated by density profile of gas via radius of jet disruption Nature of small-scale structures still unclear Nature of small-scale structures still unclear

27. Part II: Connection to Galaxy Formation (Binney 04; Dekel 04)

28. CDM Clustering Small-scale cosmic web of DM develops around z~30 Small-scale cosmic web of DM develops around z~30 Subsequently larger-scale webs form from collapsed structures from earlier webs Subsequently larger-scale webs form from collapsed structures from earlier webs Gradually accumulate superposition of halos with ~power-law mass function Gradually accumulate superposition of halos with ~power-law mass function Mass function unlike galaxy L function Mass function unlike galaxy L function

29. Galaxy Formation Low M galaxies suppressed by photoionization, evaporation & SN feedback (Efstathiou 92; Dekel & Silk 86; Dekel 04) Low M galaxies suppressed by photoionization, evaporation & SN feedback (Efstathiou 92; Dekel & Silk 86; Dekel 04) Infalling gas shocks Infalling gas shocks Accretion shock near centre if t cool <t free-fall Accretion shock near centre if t cool <t free-fall Condition holds for most mass in halos with M<10 12 M ¯ (Dekel & Birnboim 03, 06) Condition holds for most mass in halos with M<10 12 M ¯ (Dekel & Birnboim 03, 06)

30. Lumpy Accretion Extended Press-Schechter predicts lumpy accretion (mergers/cannibalism) Extended Press-Schechter predicts lumpy accretion (mergers/cannibalism) Accretion shock unhelpful concept for lumpy accretion Accretion shock unhelpful concept for lumpy accretion So without SN heating all gas cold So without SN heating all gas cold

31. SN Heating After starbursts SN heat much gas to ~10 7 K After starbursts SN heat much gas to ~10 7 K Flows out of halos with v c <100 km s -1 (Larson 74, Dekel & Silk 86) Flows out of halos with v c <100 km s -1 (Larson 74, Dekel & Silk 86) In larger halos SN-heated gas accumulates In larger halos SN-heated gas accumulates As infall continues, central density rises As infall continues, central density rises Cannot be stabilized by SN heating Cannot be stabilized by SN heating

32. AGN Heating t cool = 3/2  m p kT/n  shortest @ centre t cool = 3/2  m p kT/n  shortest @ centre BH accretion rate rises with n 0 BH accretion rate rises with n 0 Mechanical L stabilizes hot gas Mechanical L stabilizes hot gas In absence of cold infall hot gas cannot cool In absence of cold infall hot gas cannot cool

33. Cold Infall Cold infall widely observed: Cold infall widely observed: Magellanic stream Magellanic stream Perseus filaments Perseus filaments At hot/cold interface At hot/cold interface (a) ablation by conduction/mixing (small blobs) (a) ablation by conduction/mixing (small blobs) (b) condensation and star formation (larger blobs) (b) condensation and star formation (larger blobs) Conduction more important at high T (Nipoti & B 04) Conduction more important at high T (Nipoti & B 04) Conselice et al 01

34. Connection with BH growth BH growth known to take place in bursts: BH growth known to take place in bursts: Yu & Tremaine (2002) find (i) AGN have radiated in optical/UV as much E as released by all nuclear BHs; (ii) L~L Edd and ε>0.1 needed to produce observed quasars from observed BHs Yu & Tremaine (2002) find (i) AGN have radiated in optical/UV as much E as released by all nuclear BHs; (ii) L~L Edd and ε>0.1 needed to produce observed quasars from observed BHs @L Edd M~exp(t/t Salpeter ); t Salpeter ~25 Myr @L Edd M~exp(t/t Salpeter ); t Salpeter ~25 Myr So M from 10 3 M ¯ To 10 9 M ¯ with 14t S ~0.4Gyr and 10Gyr at <0.05L Edd So M from 10 3 M ¯ To 10 9 M ¯ with 14t S ~0.4Gyr and 10Gyr at <0.05L Edd Magorrian relation M~M bulge, high α/Fe of bulges, high ages of bulges all imply L Edd (quasar) phase associated rapid star formation Magorrian relation M~M bulge, high α/Fe of bulges, high ages of bulges all imply L Edd (quasar) phase associated rapid star formation Conjecture this is when there is cold gas @ centre Conjecture this is when there is cold gas @ centre Episodes end when well deep enough to trap 10 7 K gas; then Mdot 0.002 to 0.02 M ¯ /yr to offset 10 43 – 10 44 erg/s of L X Episodes end when well deep enough to trap 10 7 K gas; then Mdot 0.002 to 0.02 M ¯ /yr to offset 10 43 – 10 44 erg/s of L X

35. Semianalytic GF (Croton et al 06 & Cattaneo et al 06) From model of DM clustering take merger history of halo population From model of DM clustering take merger history of halo population Fraction 0.17 or 0.14 of M in baryons Fraction 0.17 or 0.14 of M in baryons Primary halos have hot gas, cold gas, stars Primary halos have hot gas, cold gas, stars Secondary halos has stars & cold gas Secondary halos has stars & cold gas They spiral in by dynamical friction They spiral in by dynamical friction Bulges form in (a) merger-driven starbursts and (b) disk instabilities Bulges form in (a) merger-driven starbursts and (b) disk instabilities SNe expel gas SNe expel gas

36. Cattaneo et al (06) Standard models: Standard models: Gas shock heated & arranged in singular isothermal sphere Gas shock heated & arranged in singular isothermal sphere Cools to exponential disk Cools to exponential disk dot M * =M cool /(  t dyn ) dot M * =M cool /(  t dyn ) ½ dot M wind v e 2 =  SN  sn E SN dot M * ½ dot M wind v e 2 =  SN  sn E SN dot M * Makes too many bright blue galaxies Makes too many bright blue galaxies Makes luminous galaxies too late Makes luminous galaxies too late Lack of COMBO-17 red galaxies Lack of COMBO-17 red galaxies

37. New Models Sharp transition: cold infall ! virialization @ M crit =M shock £ Min(1,10 1.3(z-z c ) Sharp transition: cold infall ! virialization @ M crit =M shock £ Min(1,10 1.3(z-z c ) At M>M crit reheat cold gas At M>M crit reheat cold gas Now dot M * =(1+z)  M cold /(  t dyn ) Now dot M * =(1+z)  M cold /(  t dyn ) Find  ' 0.6, M shock ' 2 £ 10 12 M ¯, z c ' 3.2 Find  ' 0.6, M shock ' 2 £ 10 12 M ¯, z c ' 3.2

38. New Models Good agreement global SFR Good agreement global SFR

39. Croton et al (06) Gas shock heated to T vir & cools to disk Gas shock heated to T vir & cools to disk Either immediately (r cool >r vir ) or at rate 4  (r cool )r cool 2 dr cool /dt Either immediately (r cool >r vir ) or at rate 4  (r cool )r cool 2 dr cool /dt In disk steady SF at rate / (m-m crit )/t dyn In disk steady SF at rate / (m-m crit )/t dyn SN inject energy  E SN /  m * to mass 3.5  m * SN inject energy  E SN /  m * to mass 3.5  m * When in hot halo gas has energy 3.5 £ 0.5  m * V c 2 When in hot halo gas has energy 3.5 £ 0.5  m * V c 2 Surplus E used to eject gas from halo Surplus E used to eject gas from halo

40. AGN Feedback Croton et al follow m BH (t) Croton et al follow m BH (t) Mergers drive quasar mode:  m BH =f m cold /[1+(280/V vir ) 2 ] with f(m sat /m host ) Mergers drive quasar mode:  m BH =f m cold /[1+(280/V vir ) 2 ] with f(m sat /m host ) No feedback No feedback In radio mode dm BH /dt / m BH f hot V vir 3 In radio mode dm BH /dt / m BH f hot V vir 3 L BH =  c 2 (dm BH /dt) offsets radiative cooling L BH =  c 2 (dm BH /dt) offsets radiative cooling

41. Croton et al Results Feedback suppresses cooling at large V vir and low z Feedback suppresses cooling at large V vir and low z Eliminates very luminous galaxies Eliminates very luminous galaxies Establishes red/blue dichotomy Establishes red/blue dichotomy Croton et al

42. Conclusions Now clear that AGN heating important for GF Now clear that AGN heating important for GF Distinguish quasar & RG modes Distinguish quasar & RG modes RG mode when dense atmosphere @ T vir RG mode when dense atmosphere @ T vir RG mode only in massive halos RG mode only in massive halos BHs grow principally from cold gas simultaneously with rapid SF in bulge BHs grow principally from cold gas simultaneously with rapid SF in bulge Gas at T vir never forms stars – galaxies don’t form from cooling gas Gas at T vir never forms stars – galaxies don’t form from cooling gas Gravitational heating certainly unimportant at M<2 £ 10 12 M ¯ Gravitational heating certainly unimportant at M<2 £ 10 12 M ¯ SN heating vital SN heating vital Role of thermal conductivity/ablation to be clarified Role of thermal conductivity/ablation to be clarified

43. Heating CFs by BHs In absence of heating n(0) → ∞ in t<t cool (0) In absence of heating n(0) → ∞ in t<t cool (0) Such a cooling catastrophe must provoke a response from the central BH Such a cooling catastrophe must provoke a response from the central BH

44. Bondi accretion Area of sonic flow Area of sonic flow Particle density Particle density Accretion rate Accretion rate Luminosity Luminosity So balance possible with E α ∫ dt L X So balance possible with E α ∫ dt L X

45. Characteristic M*=3x10 10 M. (Kauffmann et al 03) At M>M* dSB/dM=0; at M 0 At M>M* dSB/dM=0; at M 0 At M>M* galaxies old; at M M* galaxies old; at M<M* younger At M>M* light centrally concentrated At M>M* light centrally concentrated

46. Theory of Galaxy Formation Standard picture: gas heated to T vir on falling into Φ (Rees & Ostriker 1977; White & Rees 1978) Standard picture: gas heated to T vir on falling into Φ (Rees & Ostriker 1977; White & Rees 1978) Actually fraction f enters at T<<T vir (Binney 1977; Katz et al 2003; Birnboim & Dekel 2003) Actually fraction f enters at T<<T vir (Binney 1977; Katz et al 2003; Birnboim & Dekel 2003) f~1 on galaxy scales M* and below f~1 on galaxy scales M* and below Katz et al 02 Birnboim & Dekel 04