Connection between AGN and Star Formation activities, or not? Y.Sophia Dai ( 戴昱 ) National Astronomical Observatories of China (NAOC) daysophia@gmail.com
Close Correlations between SMBH and hosts M SMBH σ Relation M SMBH - L bulge log M SMBH (M ) 6 7 8 9 Tremaine+02 Host velocity dispersion (kms -1 ) log M SMBH (M ) 7 8 9 Marconi & Hunt 03 9 10 11 12 log L bul (L )
Consistent Mass ratios between SMBH and hosts galaxy: log (M * /M ) ~ 2.6-3.6 M - M bulge M - M * M SMBH (M ) 7 8 9 McLure & Dunlop 02 9 10 11 12 log M bul (M ) Log(M bul /M ) ~ 2.3 +/- 0.5 (Magorrian et al. 1998) Log(M bul /M ) ~ 2.9 +/- 0.5 (Mclure & Dunlop 2002) Log(M bul /M ) ~ 2.8 +/- 0.4 (Macorni & Hunt 2003) Reines & Volonteri 15 Log(M * /M ) ~ 2.9 +/- 0.4 (Macorni & Hunt 2003) Log(M * /M ) ~ 2.6 +/- 0.5 (Kormendy & Ho 2013) Log(M * /M ) ~ 3.6 +/- 0.5 (Reines & Volonteri 2015)
Merger-driven Model Cold-Stream Accretion 320 kpc Dekel+09 e.g. Springel et al. 2005, Keres et al. 2005, Dekel & Birnboim et al. 2006, Dekel et al. 2009, Bournaud et al. 2011, Di Matteo et al. 2012 Image credit: Hopkins+06; NASA Feedback often needed e.g. Lilly et al. 2013; Lapi et al. 2014; Aversa et al. 2015; Mancuso et al. 2016
AGN-Host Correlation part 1: L SF vs L AGN
Typical Sample Selection AGN growth à SMBH accretion/blr ß X-ray/optical Galaxy growth à SF ß (far)-infrared/ionized lines IR + Xray (majority): Lutz + 2010; Shao + 2010; Rosario + 2012; Santini + 2012 ; Page + 2012; Mullaney + 2012, Barger + 2015; Shimizu + 2015; Xu + 2015; Stanley + 2015; Shimizu + 2017; Dai + 2018a IR + optical: e.g. Netzer 2009; Matsuoka & Woo 2015; Rosario + 2013b; Harris + 2016; Pitchford + 2016, Cowley+18 Comment: AGN: not the same region SF: not the same material (gas vs cold dust)
Sample Selection SMBH - SF X-ray Lx > 10 42 erg/s Absorption corrected far-ir S 250 > 3σ w/z IR-bright AGN XMM-Newton: ~11 deg 2, 10 ks Herschel HerMES DR2 & DR3: ~19 deg 2, 1σ = 2.2 mjy Dai et al. 2018, MNRAS, 478, 4238
Suppressed (e.g. AGN feedback)? Page+12 Barger et al. 2015, ApJ, 801,87
Irrelevant? Harrison+12 Stanley+15 Barger+, in prep
Enhancing (Bi-model) & 3 scenarios? Lutz+10 Harris+16 Chen+15 Pitchford+16 See also: Lutz+10, Shao+10, Satini+12, Rosario+12, Rovilos +12, Povic+16; Harris+16; Pitchford+16
Enhancing (overall linear correlation)? Xu+15 Azadi+15 Shimizu+17 Matsuoka & Woo+15 See also: Satini+12, Rosario+13, Rovilos +12, Xu+15, Chen+15, Povic+16, Shimizu+17, Suh+17, Dai+18a, Matsuoka & Woo 2015, ApJ, 807, 28
Enhancing (linear correlation)? 48 47 Main sample, 0.2 < z < 2.5 slope+/-error [ 0.62+/- 0.05] (no bin) [ 0.59+/- 0.17] (L AGN bin) [ 1.11+/- 0.19] (L IR bin) 4500 log L IR, SF [erg/s] 46 45 450 45 SFR [M O /yr] 44 X-ray detected AGNs, 0.4 < z < 2.6, Symeonidis+ 2011 X-ray and MIR detected AGNs, 0.2 < z < 0.8, Chen+2013 MIR detected AGNs (type 1 + 2), 0.8 < z < 1.7, Chen+2015 Xray obscured AGNs (type 2), 0.2 < z < 1, Azadi+2015 43 0.45 42 43 44 45 46 47 48 log L AGN [erg/s] 4.5 Dai et al. 2018, MNRAS, 478, 4238
AGN-Host Correlation part 2 BHAR vs SFR
A constant log(sfr/bhar) ratio at ~ 3 log (SFR / BHAR) 4 3 2 1 6 y = ( 3.15+/- 0.07) + ( 0.11+/- 0.06)x (b) combined sample? y = ( 2.89+/- 0.05) + ( 0.13+/- 0.04)x 0 1 2 3 z y = ( 3.79+/- 0.65) + (-0.06+/- 0.08)x (d) combined sample y = ( 3.18+/- 0.49) + (-0.01+/- 0.06)x 5 6 7 8 9 log (M.) [M O ] Work in progress 4 2.9 1.9 0.9-0.1 ) log (L IR, SF / L X 2-10 kev Source of large scatter (0.5-0.6 dex): - Instantaneous rates - Different galaxy morphology (bulge, elliptical, disk) log(sfr=bhar) 4 2 3 2 1 redshift - Uncertainties in SFR, BHAR, and mass estimates y = (2.50+/-0.11) + (0.03+/-0.01) x 5 6 7 8 9 10 11 12 logm [Msun] (Dai+18a, Calhau+17, Mullaney+12, Silverman+09)
A constant SFR/BHAR ratio at ~ 3? Log (M * / M )~ 2.6+/-0.4 Kormendy & Ho 2013 Magorrian et al. 1998 Log (M * / M )~ 3.6+/-0.5 Reines & Volonteri 2015 BAT Seyferts Shimizu+15
Cosmic Evolution of Dusty AGN mass and Eddington Ratio 11 10 SDSS (color selected) MMT (MIR selected) log (M BH ) (M sun ) log M BH 9 8 7 SDSS-s SDSS-g MMT-s MMT-g Downsizing! 0 log ER log(l bol /L Edd ) -1-2 Almost Constant -3 Labita+09 0 1 2 3 4 z z (Dai et al. 2014, ApJ, 791, 113)
Is BHAR/SFR ratio really flat (M *, z)? Mullaney et al. 2012, ApJL, 753, 2 Yang et al. 2017, ApJ, 842, 72 Dai et al. 2018, MNRAS, 478, 4238 6 Work in progress 4 See also: Xue+10, Shimizu+17, Suh+17, Stanley+17 log(sfr=bhar) 4 2 3 2 1 redshift 5 6 7 8 9 10 11 12 logm [Msun] Cowley et al. 2018, MNRAS,473, 3710 Open question: What is BHAR truly related to? Gas, stellar mass, metallicity? Is it z/sample dependent?
Caveats in AGN-Host Correlation Analysis
Caveat 1: Malmquist Bias 0.09 0.14 0.19 0.42 Dai et al. 2018, MNRAS, 478, 4238
Caveat 2 & 3: Selection Bias + Binning Bias Luminosity (mass)-dependent correlation? Sample dependent correlation? Possible danger with stacking (Dai+2015, Azadi+17) Binned by SFR or AGN (Volonteri+2015, Dai+2018a)
Caveat 4: Uncertainties in SFR & BHAR estimates + SED based AGN modelling 10 16 10 14 IR FIR Elvis+94 Richards+06 Netzer+07 Mullaney+11 Dai+12 Dale+14 Luminosity (a.u.) 10 12 10 10 10 8 0.1 1.0 10.0 100.0 1000.0 10000.0 λ (µ m) General consistency (~ 0.2 dex) over the total IR range (8-1000 µm) Larger deviation (~ 0.6 dex) in the FIR range (30-1000 µm)
Typical AGN-SF decomposition Methods Cowley et al. (2018) using CIGALE (Burgarella et al. 2005; Noll et al. 2009) Rosario et al. (2018, MNRAS, 473, 5658) using SSP (Bruzual & Charlot, 2003 GALAXEV) + AGN template + dust (Dale & Helou 2002) Disadvantages: 1. Template dependent (applicability + uncertainties) 2. Not full use of known information (e.g. X-ray) SSP: single stellar population models
F ) (arbitrary units) E94 SED 3-step AGN IR decomposition: 1 & 2 6 log( LX à L6 from pure AGN 4 A chosen AGN template FIR detected(lower limit) T ~ 30K 2 FIR undetected 0 Stern15 0.1 1 rest 10 100 [µm] x log(νlν(6μm)/1041 erg s 1 LIR (8-1000), LFIR(30-1000), LAGN from pure AGN
3-step AGN decomposition: 3 T-α β fit of the observed IR SED Total L IR & L FIR α β L IR (AGN) & L FIR (AGN) T L IR (SF) & L FIR (SF) SFR BHAR Lx
SFR before and after AGN removal Red: L IR (8-1000) Blue: L FIR (30-1000) 1.0 SFR_true / SFR_uncorr SFR_true / SFR_tot 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 ~20% AGN dominant 1e00 1e01 1e02 1e03 1e04 1 10 SFR_tot 100 1000 SFR_uncorr [M /yr] AGN contribution Ø IR: 23% (11%) Ø FIR: 11% (4%) Ø 70um: ~30% (Dai+2018a, see also, Shimizu+2017)
Summary Using SFR vs BHAR is a common, and useful way to study the AGN-galaxy connection SFR BHAR correlation is confirmed by various observational studies ( L AGN -L IR,SF ), consistent with the scenario of a common mass supply for SMBH and host galaxy A nearly constant ratio of log(sfr/bhar) ~ 2.5-3.0 is observed, agreeing with the local M * /M ratio, indicating homogeneous evolution across z (also M *?) Bear in mind the following caveats, which can potentially mask out intrinsic AGN-SF correlations: 1. Selection Bias (Malmquistbias always there) 2. AGN population (mix of different populations, caution the use of stacking) 3. Binning method (variability, choice of free parameter) 4. Various SFR & BHAR estimators (large scatter + uncertainty, ~ 0.5 dex)
Some Open Questions How to explain the flat(increasing) BHAR/SFR ratios? Is it fundamental or random? Is it mass/luminosity/z/sample dependent? How do we go from the Mdot ratios to the accumulated Mass ratios? What is driving the observed luminosity correlations? Is it common gas supply (merging vs cold flow)-mass, metallicity; stellar mass; selfregulation? What is the physical process that funnels gas/mass from the galaxy to the nucleus region? And vice versa? Can the BHAR/SFR ratio tell us the fraction? In practice, how to define a clean sample, of similar physical properties, at similar evolutionary stages? Using which one can define the intrinsic AGN SED(s)?