The Coevolution of SMBHs

1
The Coevolution of SMBHs Galaxies Since the
First Billion Years (z4.5) as Seen Through
Grav. Lensing
2
Overview

• Motivation
• How we study the MBH-Mbulge scaling relation at z
  gt 1
• Gravitational lensing (of quasar host galaxies)
• Results
• Future study
• Conclusion

3
Local SMBH and Galaxy Correlations
Mbulge / MBH ? 800
MBH-s relation
(Gebhardt et al. 2000, Ferrarese Merritt
2000, Tremaine et al. 2002)
(e.g. Kormendy Richstone 1995, Magorrian et
al. 1998, Haering Rix 2004)
Inactive galaxies
Dwarf AGNs
Inactive galaxies
Greene Ho (2005) Barth et al. (2004, 2005),
Marconi Hunt 2003
4
Theoretical Picture
di Matteo, Robertson, Granato, Hopkins, Fontanot,
Li, etc.
Image courtesy of Phil Hopkins
5
Quenching of Star Formation (by feedback or
otherwise) is Useful for Understanding Galaxy
Evolution

• The origin of the MBH-Mbulge scaling relations
  (Silk Rees 1998, Fabian 1999, Di Matteo et al.
  2005)
• The existence of massive (1011 M?),
  post-starburst galaxies at even z?6 (e.g.
  Mobasher, Dickinson, Ferguson, Giavalisco,
  Wiklind, et al. 2005)

6
No SF Quenching in Standard Model
Cattaneo, Dekel, Devriendt, Guiderdoni, Blaizot
2006
7
SF Quenching in the New Model
Cattaneo, Dekel, Devriendt, Guiderdoni, Blaizot
2006
8
Do AGNs Shut off SFR?
AGN feedback may truncate star formation (Silk
Rees 1998, Fabian 1999, Di Matteo, Hernquist,
Springel 2005)
SF rate as inferred via OII ?3727, Correcting
for AGN contam.
For a given gas content, AGN hosts have lower SFR
compared to (weakly) star-burst galaxies and
(U)LIRGS.
(Ho 2005)
9
Theoretical Perspectives Scenario 1

• SPH merger sims of disks virial mass, fgas,
  EOS ISM, surf. mass density, dark matter
  concentration, z (Robertson et al. 2006)
• Finding BH vs. Bulge relations dont evolve
  much

MBH vs. ?
MBul vs. MBH
10
Theoretical Perspectives Scenario 2
Croton (2006)

• Millennium ?-CDM sims
• BHs grow by gas accretion, BH mergers
• Bulges grow in situ SFdisk disruption via major
  mergers

3x
(See also Malbon et al. 2006)
11
Theoretical Perspectives Scenario 3
Fontanot et al. (2006)

• MORGANA semi-analytic models
• Incorporates
• Mergers trees, and different contribns from
  stellar AGN driven winds

12
What do Observations say about the MBH/Mbulge
Ratio at z gt 1 relative to z 0?
Larger in the past (9) The Same (2?) Smaller
(1)
Galaxy luminosity density argument (Hopkins et
al. 2006)
CO linewidths of QSO host (Walter et al. 2004,
Shields et al. 2006) OIII proxy for ?
(Shields et al. 2003) Groundbased AO imaging of
z?2 QSOs (Croom et al. 2004) Star formation rate
vs. stellar mass density evolution (Merloni et
al. 2003) High-z EXOs (Koekemoer et al.
2005) M (radio gal) vs. MBH(QSOs) at z gt 1
(McLure et al. 2006)
X-ray detected sub-mm gal (Borys et al. 2005)
Groundbased AO imaging of z?2 QSOs (Kuhlbrodt et
al. 2006)
HST imaging of quasar hosts at z gt 1 (Peng et al.
2006a, Peng et al. 2006b)
13
How does the MBH/Mbulge ratio change as we look
to high redshift (z gt 1, MBH ? 108 M?)?
???z?MBH/Mbulge(z) relative to today
14
How do we study BH-Bulge relation at z gt 1?
15
Step 1 Estimate MBH in broad-line AGNs By using
the virial technique

• MBH???f VBLR2 RBLR/G
• (Kaspi et al. 2000, 2005, Vestergaard Peterson
  2006)
• RBLR ? LAGNa
• MBH f La VFWHM2
• Fudge factor f MBH norm. to MBH-s (Onken et
  al. 2004).
• MBH scatter of a factor of 4

Radius (light days)
Bentz et al. (2006)
16
Step 2 Measure the luminosity of AGN Host
Galaxies to infer bulge mass
GALFIT (Peng et. al. 2002)
Non-lensed
Deblend AGN/host w./ 2-Dimensional Parametric
image fitting
Lensed
LENSFIT (Peng et al. 2006)
17
Traditional (Axisymmetric) Parametric Fitting of
Low z Quasar Host Galaxies(i.e. fitting light
distribution with parametric ellipsoid functions)
18
How to See Quasar Host Galaxies, Normally.
19
2-D Fitting Doesnt Have to Be UnrealisticGALFIT
3.0 (Peng et al. 2006, in prep.)
Data
GALFIT Analytic Model
Residuals
Bottom line asymmetries and spirals are 2nd
order effects that dont affect total luminosity,
avg. size much, but important for decomposition
20
High z (? 2-3) Quasar Hosts are Really Hard to
See!
From Ridgway et al. (2001) also see Kukula et
al. 2001, Yun et al. 1996
Quasar subtracted images, HST/NICMOS H-band
(restframe V)
Deep images 4-7 orbits each (30 total)
21
GRAVITATIONALLENSING
22
your favorite mountain
23
your favorite building
24
your favorite person
25
The Castle on the Mall in Washington, D.C.
26
with a Saturn-mass black hole over the Mall
27
What Lensing does to a Resolved Source
(Courtesy of Jim Lovell)
28
Principles of Lensing

• Consider the paths of light in the presence of
  masses, which curve space according to GR
• Assumptions
• FRW smooth metric, plus distortions
• All local perturbations are weak
• Weak field ? ltlt c2
• Fermats Principle
• Images formed at local stationary points (min,
  max,
• saddle) of the light travel time (delay)
  surface
• Light travel time delay has two components
• geometric delays and relativistic time dilation

29
The Lens Equation

• From geometry
• And spherically symmetric lenses give typical
  image separations of

b (true) source position q(seeming) image
position a(scaled) deflection
30
Fermats Principle

• Relativistic time dilation leads to an effective
  index of refraction
• neff 1 2?/c2
• Images formed if
• is obeyed
• Images are formed in pairs, ? odd number of
  images (but central highly demagnified)

Blandford and Narayan (1986)
Geometric only Weak lens
Time delay on the sky
Increase to 3-image Increase to 5-image
31
Image Formation Locations
Arrival Time


Angular Distance from Lens (center)
32
Quasar Hosts from z 0.2 to z 3

• z ? 1 Boyce et al. (1998), McLeod McLeod
  (2001), McLure et al. (1999), Dunlop et al.
  (2003), Sanchez GEMS (2004), etc.
• Massive, early-type hosts, dominated by old
  stars, but with some young post-AGB stars, Re
  10s kpc.
• Hosts of RLQs RQQs are indistinguishable at
  z0.2.
• z gt 1 Hutchings (2000), Ridgway et al. (2001),
  Kukula et al. (2001), Jahnke GEMS (2004),
  Hutchings et al. (2003), Falomo et al. (2003)
• RQQ hosts
• Consistent with Ly Break (2-4 times brighter
  than L), Re lt 8 kpc.
• RLQ hosts Kukula et al. (2001), Falomo et al.
  (2003)
• 1-2 mag brighter than RQQs, Re gt 10 kpc (similar
  to RGs).
• Fairly consistent with passively evolving E
  models.

33
CASTLES Project (CfA-Arizona-HST Lensing Survey)

• HST V, I, H survey of strong lenses of quasars.
• Shallow, mostly single orbit images in each
  filter.
• 80 systems (? half w./ HST H-band analyzed).

34
(No Transcript)
35
Road Map how to get BH Bulge mass
GALFIT (Peng et. al. 2002)
Non-lensed
Deblend AGN/host w./ 2-Dimensional Parametric
image fitting
Lensed
LENSFIT (Peng et al. 2006)
Bulge mass Inferred from host luminosity
Black Hole mass Virial technique of Type 1 AGN
using C IV (z gt 1.5), Mg II (0.8 lt z lt 1.5), H?.
36
LENSFIT A New, Parametric, Way to Solve theLens
Equation While Image Fitting
Peng et al. (2006)
N number of comps. (light deflector),
unrestricted.

• The simplest model has a minimum of 22 free
  parameters (no maximum), all simultaneously
  adjusted to reduce pixel ?2.
• But, no need to fear!
• - Objs. well resolved
• - (x,y) accurate
• - Shapes are different
• ? Most params have small covariance, and covar.
  quantifiable!

Light profiles (analogous to GALFIT)
Deflection models
Foreground Galaxy
37
PG 1115080
Lenses 1 SIE 2 SIS
Host n 4 re? 2 kpc H 20.4 ? MV -22.5
MBH 1 x 109 M? (expect re10 kpc if host fully
formed, passively evolving)
38
HE 1104-1805
Host n 1.5 re?2.3 kpc H 20.6 ? MV -23.5
MBH 2 x 109 M? (expect re15 kpc if host fully
formed, passively evolving)
39
RXJ 09110551
Host re?1-2 kpc H 22.2 ? MV -22.4
MBH 8 x 108 M? (expect re?10 kpc if host is
fully formed, passively evolving)
40
B 1600434
Edge-on spiral galaxy lens face on barred
spiral external perturber (SIE ?).
Host n 1.6 re ??3 kpc H 21.3 ? MV -22.1
MBH 1 x 108 M?
(expect re?3 kpc if host fully evol.)
41
BRI 0952-0115
Highest redshift host in lensed sample
Host re ??? kpc H 22.3 ? MV ? -25
MBH 1 x 109 M? (expect re?10 kpc if host fully
evol.)
42
CTQ 414
2-comp. Bulge/disk decomp. of host
2
Cautionary tale early/late type class depends
on S/N and resolution.
Diff. in lum 30
43
MG 11310456
Double nuclei source!
Host n 1.76 Re ??7.6 kpc H 18.8
44
MG 04140534
Object X, SIE Perturber contributes no light to
model!
x
Host n 4 Re ??6 kpc H 18 ? MV -26.7
45
B 2045265
Host is a dwarf if _at_ z1.28. AGN magnified 240
times!
Host n 1.4 Re ??0.6 kpc H 22.9 ? MV
-19.9 (-17 by z0)
46
B 1938666
Lopsided ring ? asymmetric host. Modeled w/ 2
comps.
Host n 2.4 Re ??6 kpc H 20.6 MV ? -22.5
47
MG 2016112
The weird critical lines are caused by large
external perturbn (2 SIE).
Host n ? 4 Re ??1 kpc H 22.5 ? MV ? -23
48
LBQS 1009-0252
Host n 3.4 Re ? 3 kpc H 21.1 ? MV -23.9
49
SBS 1520530
Lens is edge on spiral.
Host n 0.5 Re ? 3 kpc H 21.3 ? MV -22
50
Color of MG 04140435 (z2.64) Original Data
51
Color of MG 04140435 Host Galaxy Residual
52
Color of MG 04140435Overall Residual
53
Color of MG 04140435 (Gravitational Lens)
Peng et al. (2007)
UV flux ? SFR ? 5 M?/yr
Rest
54
The Lensed Quasar Sample

• Total gt 90 lensed quasars
• Analyzed 35 w./ HST H-band data (blue circles)
• This study has 51 objects (both lensed and non-)

55
Non-circles non-lenses
Peng et al. (2006)
Confused sources
56
Non-circles non-lenses
Peng et al. (2006)
Confused sources
57
Radio Loud vs. Radio Quiet
RQQ? optically selected lenses (probably
RQQs).
Peng et al. (2006)
No differences seen between RLQ and RQQ hosts.
Possible explanation Kukula et al. select from
gt Jy radio sources.
58
Road Map how to get BH Bulge mass
GALFIT (Peng et. al. 2002)
Non-lensed
Deblend AGN/host w./ 2-Dimensional Parametric
image fitting
Lensed
LENSFIT (Peng et al. 2006)
Bulge mass Inferred from host luminosity
Black Hole mass Virial technique of Type 1 AGN
using C IV (z gt 1.5), Mg II (0.8 lt z lt 1.5), H?.
59
Galaxy Bulge SMBH Correlations
MBH - s
MBH ? LBul
MBH vs. s
MBH vs. LBul
? evolutionary connection between galaxy bulge
and SMBH
60
RESULTS
61
Black Hole - Bulge Correlation _at_ z ? 2
Peng et al. (2006, ApJ, 649, 616)
62
Black Hole - Bulge Correlation _at_ z ? 2
Peng et al. (2006)
63
Black Hole - Bulge Correlation _at_ z 1
Peng et al. (2006)
64
Black Hole / Bulge Mass Ratio at z ? 2
R MBH/Mbulge
R was 4 times larger in the past than today
R was independent of host luminosity
Peng et al. (2006)
65
Black Hole / Bulge Mass Ratio at z ? 1 (MBH ? 108
M?)
???z?MBH/Mbulge(z) relative to today
Peng et al. (2006)
66
New Gas Depletion Tree-SPH Model
Hopkins et al. (2006)
67
Is it possible to make up the mass deficit in
quasar hosts since z?2 (without merging)?

• Back of the envelope estimate
• - z2 to z1 ?
• ?t 2.5 Gyrs
• - Mass deficit ?(z) 3 to 4 ? specific SFR
  of
• ?? (?(z)-1)/?t Gy-1
• (3-1)/2.5 to (4-1)/2.5
• 0.8 to 1.2 Gy-1
• Typical host mass for MBH109 M?
• Mbulge(z2) ? 800 MBH(z0) / ?
  
• 2x1011 M?

68
The Sizes of Quasar Hosts
Median re?2 kpc
No objs larger than 10 kpc
GEMS AGNs
Vast majority smaller than 4 kpc
69
Size vs. Bulge Mass (?800 x MBH / ?)
Assume local relation of Mbulge 800 MBH
zgt2 quasar hosts appear too small by 2-5 times
relative to E/S0 today. Did not correct for
systematic errors of 30-50 due to PSF systematic
issues, color gradient.
Restframe UV
70
Predicted Observed Size Evolution
Massive galaxies could be up to five times
smaller at high redshifts than now, because
they are more likely to be formed during a
gas-rich major merger. -- Sadegh Khochfar
Joseph Silk (2006)
Lensed QSO Hosts
71
Revisiting Shields et al. (2002) MBH vs. OIII

• Q Shields et al. (2003) find no evolution in the
  MBH-s relationship (where OIII was used as
  substitute for s). What does this say about the
  MBH/Mbulge ratio?
• How much evolution do we expect to see?
• 2x1012 Msol gt ??450 km/s
• 5x1011 Msol gt ??300 km/s
• expect a 0.18 dex change in ?OIII since z3

72
?? vs. M evolution in merger simulations
Robertson et al. (2006) GADGET merger
simulations For a constant ?, M is lower in
the past (evol of Faber-Jackson
relation). Unevolving MBH-?(z) (Shields et al.
2002) implies ?(z)5 by z3
Disperson km s-1
5x
M Msol
73
How Do Galaxy Mergers Affect the MBH-MBulge
Relation?
74
Theoretical Perspective
Croton (2006)

• Millennium ?-CDM sims
• BHs grow by gas accretion, BH mergers
• Bulges grow in situ SFdisk disruption via major
  mergers

3x
(See also Malbon et al. 2006)
75
How Galaxy Mergers Affect the Black Hole - Bulge
Correlation
Peng (2007)
76
How Galaxy Mergers Affect the Black Hole - Bulge
Correlation
Peng 2007
77
Future What Can Greatly Weaken the Conclusion
that ??zgt2)?4?

• Black hole masses over-estimated by a factor of
  2-3 (evolution of the virial relation?, Dep. on
  L/Ledd?)? Will estimate MBH using H??linewidth.
  But fundamentally, the limitation on the
  normalization is the small size of the
  reverberation mapped sample and redshift regime.
• Dust cant be ruled out (but, locally, at least,
  star formation wins over dust extinction.) Will
  do rest-frame IR imaging of lensed hosts,
  resolved IFU kinematics of lensed hosts at z gt 1.

78
The Next Step in Quasar Host Galaxy Studies
?

• IFU is key
• Use spatial decomp. to separate host from
  quasar.
• Integrate up flux at each ? slice to increase
  S/N for host.
• ? high S/N of host SED

79
Future Studying High-z Quasar Host gas/stellar
kinematics
Nirspec IFU

• FOV 3x3 perfect for grav. lenses
• Sampling 0.1? 1 kpc _at_ z gt 1. But, lens boost
  by x10.
• R100, 1000, 2700 ? resolved spatial
  kinematics.
• Low sky background for NIR spectroscopy

Arribas, ferruit, Jakobsen
80
Future Host gas/stellar kinematics and IFU
Arribas , ferruit, Jakobsen
81
Future
82
Conclusion

• z?2 hosts have a MBH vs. Lbulge relation almost
  like the one at z0.
• The MBH vs. MBulge appears 3-6 times higher at z
  gt 2 than at z0, so galaxies may gain mass by a
  factor of ?3-6 since z?2. This may be
  understood by physical models from Croton 2006,
  Fontanot et al. 2006.
• The re of hosts are 1/2 to 1/5 the size expected
  of fully formed, passively evolving E/S0s, so
  they likely have grown since z?2 (by an amount in
  good agreement with theory and observations of
  normal galaxies).
• Systematics issues (dust, BH mass normalization,
  normalization dependence on z) remain to be
  reduced.

83
3 different ways of asking the same question
MBH f vline2 LQSO?
, where, RBLR LQSO?

• Can ? (MBH/Mbulge) be biased high due to a
  Malmquist bias on LQSO which selects high MBH (of
  gt 108 M?)?
• Is MBH estimate biased from its true value for
  AGNs in high states compared to in low states?
• Do high luminosity AGNs obey the same virial
  relation as low luminosity AGN?

84
Do high luminosity AGNs obey the same virial
relation as LLAGN?
MBH f vline2 RBLR
, where, RBLR LQSO?

• LLAGNs lie on the same L-R relation as QSOs, and
  vline?follows (GM/r)0.5.
• For Malmquist bias in LQSO to cause a bias of
  ?????????would have to be systematically
  overestimated by a factor of 4 (0.6 dex) from its
  actual value at high LQSO. But, MBH, normalized
  to MBH-? (below), does not allow this and
  uncertainty in f is only a factor of 2 (Onken et
  al. 2006).
• What makes MBH difficult to measure in LLAGNs is
  that the line width is harder to measure and the
  vline may be biased by low S/N of the spectra
  used to estimate vline. But this is an argument
  to obtain high S/N spectra, not that there is a
  Malmquist bias for selecting MBH due to LQSO.

85
Can ? (MBH) be biased high due to Malmquist bias
on LQSO?
MBH f vline2 LQSO?
????? from LQSO-RBLR

• For ? (MBH/Mbulge) to be biased 4 times away
  from some nominal value, either MBH or Mbulge has
  to be biased away from its nominal value by the
  same amount.
• So the question really is, is MBH estimate
  biased from its true nominal value for AGNs in
  high states compared to in low states?, and NOT
  whether high LQSO? high MBH (even though, on
  average it does, but with a huge scatter due to
  the fact MBH?vline2).

86
Does the Virial Relation Change with z?
MBH f vline2 LQSO0.5
i.e. does f, hence the BLR geometry, changes with
z?

• What we do know
• BLR has short dynamical times of few years,
  (re-) adjustment to perturbations is
  instantaneous compared to cosmological time
  scales.
• Quasar spectra look nearly identical at high z
  vs. at low z, implying that BLR structure is
  probably not different in a crazy way.

87
Does vline ? (MBH/r) 0.5 ?
Reverberation mapping Broad emission lines,
photoionized by the central AGN, respond to
changes in brightness of the AGN. High
ionization lines closer to BH, shorter response
time (?tlag), larger doppler widths, such that
vline (MBH / r) 0.5 ? (MBH / c ?tlag)0.5
Peterson Wandel (1999, 2000), Onken Peterson
(2002)
88
Are Quasar CIV Profiles Problematic?
(Courtesy of Marianne Vestergaard)
15
(FWHM)
(Richards et al. 2002)
89
Can CIV-based MBH be over-estimated?
Baskin Laor (2006)

• Concern CIV emission line may be dominated by
  jet-wind, rather than Keplerian motion, line
  asymmetries, Eddington ratio, etc.. But
• Baskin Laor (2005) show that C IV
  under-estimate MBH by gt 2x.
• Collin et al. (2006) show that FWHM gt 4000 km/s
  over-estimate MBH by
• 30-50.

90
How Good is the SIE Assumption for the Lens Mass
Profile Slope?
Treu (2006), Treu et al. (2005)
? measured within re/8 ? provides a slope of the
mass density profile. SIE mass distribution is a
pretty accurate representation on average.
91
Is There a Bias Due to High Quasar/Host Contrast?
???z?MBH/Mbulge(z) relative to today
Peng et al. (2006)
92
Highest Contrast Objects
1
1
1
SBS 0909
B 1422
FBQ 0951
93
How Accurate Are SIE Mass Profiles?
Treu (2006)

• ?????is typically measured within
• re/8
• ????matches ???really well -- a conspiracy
  between the DM halo and the stellar mass profile

94
Do zgt2 Quasar Hosts Have M/L as High as
Ellipticals Today?
Jahnke GEMS (2004) restframe UV colors show the
host galaxies are blue ? possibly lower M/L than
E/S0s today
95
Conclusions

• Most quasar hosts at z gt 1 have Re lt 6 kpc, even
  ones well resolved by gravitational lens
  stretching.
• RQQ host galaxies at z gt 1 show a mix of
  exponential and de Vaucouleurs profiles, based on
  SIE lens model.
• RLQ and RQQ host luminosities in lensed sample
  are similar.
• z2 hosts almost follow the MBH vs. R-band
  luminosity of z?0.
• many (RLQ RQQ) may not be fully evolved early
  types.
• their MBH vs. MBulge is 3-6 times higher than at
  z0, so have to gain mass by a factor of ?3-6.
• Duty-cycle of AGNs is about 0.01 or ?107 years.

96
What about the luminosity density reasoning in
Hopkins et al. (2006)?

• ?(z) jz0 (LgtL0) / jz (L gt L0 / ?)

• There is no real disagreement at z lt 1.5. Their
  models allow ? 2 of evolution.
• At z gt 1.5, significant morphological mix of red
  gals (Toft et al. 2006). Only need 30-50
  contam. B/D decomp, Sersic selection.
• Steep part of the luminosity functions have large
  uncert..
• Slope of MBH/Mbulge may change at high z.
• MBH over-estimated by factor of 2?

97
On-going

• Constrain host galaxy mass and evolution
• Color of CASTLES lensed hosts galaxies to
  measure SFR.
• Obtain high res. sub-mm imaging of hosts to
  compare SFR predictions with optical/NIR colors.
• Environmental studies of hosts to constrain
  merger rate.
• Constrain morphology evolution
• Analyze 30 H-band (post-NCS) images of lensed
  quasars.
• Obtain deeper H-band/optical images to get
  better morphological constraints, color.
• Analyze deep images of 5 lensed hosts with new,
  deep NICMOS images.
• Environmental studies to constrain mass
  evolution using morphology density relation.
• Try other lens deflection models.
• Black hole to bulge relationship of lensed
  quasar hosts.
• Low luminosity AGNs black hole -- bulge
  relation (GO-10149, PI. Peng)

98
Host Galaxy Profile Index
Peng et al. (2006)
The host galaxies have a range of concentrations,
from exponential to de Vaucouleurs... A mixed
population of morphologies.
99
Critical Curves Caustics
Lens plane
Source plane
Critical curves
Caustics
Number of images
Critical curves have divergent magnification and
map into caustics, pairs of images appear across
caustics.
100
Resolved Source Disk
Here, a QSO in a starburst disk (i 60º) is
lensed. Red is receding gas and blue is
approaching gas.
101
z?2 Rest-frame UV Images
Jahnke GEMS (2004)
ACS Composite
F606W (v) 1 orbit F850LP (z) 1 orbit
102
Radio Loud vs. Radio Quiet
Differences in BH Mass? (from SDSS FIRST data)
Log (L5GHz/Lopt)
6000 Objects
McLure Jarvis (2004)