quasars probing quasars: shedding (quasar) light on high redshift galaxies
DESCRIPTION
Quasars Probing Quasars: Shedding (Quasar) Light on High Redshift Galaxies. Joseph F. Hennawi UC Berkeley. Ohio State February 20, 2007. Suspects. Xavier Prochaska (UCSC). Scott Burles (MIT). Juna Kollmeier (Carnegie) & Zheng Zheng (IAS). Outline. Motivation Finding close quasar pairs - PowerPoint PPT PresentationTRANSCRIPT
Quasars Probing Quasars: Quasars Probing Quasars: Shedding (Quasar) Light on Shedding (Quasar) Light on
High Redshift GalaxiesHigh Redshift Galaxies
Joseph F. HennawiUC Berkeley
Ohio StateFebruary 20, 2007
Suspects
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
Xavier Prochaska(UCSC)
Scott Burles(MIT)
Juna Kollmeier (Carnegie) & Zheng Zheng (IAS)
OutlineOutline
• Motivation
• Finding close quasar pairs
• IGM Primer
• Quasar-Absorber Clustering
• Fluorescent Ly Emission
Bottom Line: The physical problem of a quasar illuminating an optically thick cloud of HI is very simple compared to other problems in galaxy formation.
MotivationMotivation
A Simple ObservationA Simple Observation
Spectrum from Wallace Sargent
Quasars Evolution for PoetsQuasars Evolution for Poets
nQSO(> L) :
tQSO
tH
Ω4π
⎛⎝⎜
⎞⎠⎟nRelics(> MBH )
Com
ovin
g N
um
ber
Den
sity
L*(
z)/L
*(0)
Dramatic evolution of number density/ luminosity
look back time
Boyle et al. (2001)
Richards
et al. (2006)Tremaine et al. (2002)
z (redshift)
nQSO(> L) :
tQSO
tH
Ω4π
⎛⎝⎜
⎞⎠⎟nHosts
Quasar Evolution for PunditsQuasar Evolution for Pundits
BLAGN Steffen et al. (2003)
unidentified
non-BLAGN
The AGN unified model breaks down at high luminosities.
“Almost all luminous quasars are unobscured . . . ”
Barger et al. (2005)
AGN unified model
106 M
3105 M
105 M Engargiola et al. (2002)
HI in High Redshift Galaxies?HI in High Redshift Galaxies?
Image credit: Fabian Walter
Radial CO and HI profiles for 7 nearby galaxies
(Wong & Blitz 2002).
M33 HI/H/Optical M33 HI/CO
• The HI is much more extended than the stars and molecular gas.
• Until SKA, no way to image HI at high redshift.
• HI is what simulations of galaxy formation might predict (reliably).
The Power of Large SurveysThe Power of Large SurveysApache Point Observatory (APO) • Spectroscopic QSO survey
– 5000 deg2
– 45,000 z < 2.2 ; i < 19.1– 5,000 z > 3; i < 20.2– Precise (u,g,r, i, z) photometry
• Photometric QSO sample– 8000 deg2
– 500,000 z < 3 ; i < 21.0– 20,000 z > 3 ; i < 21.0 – Richards et al. 2004; Hennawi et al. 2006
SDSS 2.5m
ARC 3.5m
Jim Gunn
Follow up QSO pair confirmation
from ARC 3.5m and MMT 6.5m
MMT 6.5m
= 3.7”
2’55”
ExcludedArea
Finding Quasar PairsFinding Quasar Pairs
SDSS quasar @ z =3.13
4.02.0
3.0
2.03.0
3.0
2.04.0
low-zQSOs
Cosmology with Quasar PairsCosmology with Quasar PairsClose Quasar Pair Survey
• Discovered > 100 sub-Mpc pairs (z > 2)
• Factor 25 increase in number known
• Moderate & Echelle Resolution Spectra
• Near-IR Foreground QSO Redshifts
• 45 Keck & Gemni nights. 8 MMT nights
= 13.8”, z = 3.00; Beam =79 kpc/h
Spectra from Keck ESI
Keck Gemini-N
Science• Dark energy at z > 2 from AP test
• Small scale structure of Ly forest
• Thermal history of the Universe
• Topology of metal enrichment from
• Transverse proximity effects
Gemini-S MMT
Collaborators: Jason Prochaska, Crystal Martin, Sara Ellison, George Djorgovski, Scott Burles, Michael Strauss
Ly Forest Correlations
CIV Metal Line Correlations
Nor
mal
ized
Flu
x
IGM PrimerIGM Primer
Quasar Absorption LinesQuasar Absorption Lines
DLA (HST/STIS)
Moller et al. (2003)
LLS
Nobody et al. (200?)
Lyz = 2.96
Lyman Limitz = 2.96
QSO z = 3.0 LLS
Lyz = 2.58
DLA
• Ly Forest– Optically thin diffuse IGM / ~ 1-10; 1014 < NHI < 1017.2
– well studied for R > 1 Mpc/h
• Lyman Limit Systems (LLSs)– Optically thick 912 > 1
– 1017.2 < NHI < 1020.3
– almost totally unexplored
• Damped Ly Systems (DLAs)– NHI > 1020.3 comparable to disks
– sub-L galaxies?
– Dominate HI content of Universe
Self Shielding: A Local ExampleSelf Shielding: A Local Example
Sharp edges of galaxy disks set by ionization equilibrium with the UV background. HI is ‘self-shielded’ from extragalactic UV photons.
Braun & Thilker (2004)M31 (Andromeda) M33 VLA 21cm map
DLA
Ly forest
LLS
What if the MBH = 3107 M black hole at Andromeda’s center started accreting at the Eddington limit? What would M33 look like then?
bump due
to M33
Average HI of Andromeda
Neutral Gas
Isolated QSO
Proximity EffectsProximity Effects
• Proximity Effect Decrease in Ly forest absorption due to large ionizing flux near a quasar
• Transverse Proximity Effect Decrease in absorption in background QSO spectrum due to transverse ionizing flux of a foreground quasar– Geometry of quasar radiation field (obscuration?)
– Quasar lifetime/variability
– Measure distribution of HI in quasar environments
Are there similar effects for optically thick absorbers?
Ionized Gas
Projected QSO Pair
nQSO :
tQSO
tH
Ω4π
⎛⎝⎜
⎞⎠⎟nHosts
Fluorescent LyFluorescent Ly Emission Emission
• In ionization equilibrium ~ 60% of recombinations yield a Ly photon
• Since 1216 > 104 912 , Ly photons must ‘scatter’ out of the cloud
• Photons only escape from tails of velocity distribution where Ly is small
• LLSs ‘reflect’ ~ 60% of UV radiation in a fluorescent double peaked line
Zheng & Miralda-Escude (2002)
912 ~ 1 in self shielding skin
Shielded HI
UV Background
x =δυ /υσ / c
= 0e−(x2 /2)
Only Ly photons in tail can escape
P(v)
v dist of cloud
Imaging Optically Thick AbsorbersImaging Optically Thick Absorbers
Cantalupo et al. (2005)
Column Density Ly Surface Brightness
• Expected surface brightness:
• Still not detected. Even after 60h integrations on 10m telescopes!
or
Sounds pretty hard!
SBLy =3.7 ×10−20 J −22
912
4⎛
⎝⎜⎞
⎠⎟1+ z4
⎛⎝⎜
⎞⎠⎟
−4
ergs cm-2s-1W" μLyα = 30 mag/W"
Help From a Nearby QuasarHelp From a Nearby Quasar
Adelberger et al. (2006)
DLAtrough
2-d Spectrum of Background Quasar
Spatial Along Slit (”)W
avel
engt
h
extended emission
r = 15.7!
Doubled Peaked Resonant Profile?
Background QSO spectrum
Transverse flux = 5700 UVB!
f/g QSO
R = 384 kpc
11 kpc
4 kpc
Why Did Chuck Get So Lucky?Why Did Chuck Get So Lucky?
f/g QSO
R||
b/g QSO
R = 280 kpc/h
DLA must be in this
region to see emission
• Surface brightness consistent with expectation for R|| = 0
• R|| constrained to be very small, otherwise fluorescence would be way too dim.
If we assume emission was detected at (S/N) = 10, then (S/N) > 1 requires:
R|| < R [(S/N) -1]1/2 = 830 kpc/h or dz < 0.004
Since dN/dz(DLAs) = 0.2, then the probability PChuck = 1/1000!
I should spend less time at Keck, and more time in Vegas $$
Chuck Steidel
Perhaps DLAs are strongly clustered around quasars?
Quasar-Absorber Clustering
Quasar-Absorber Clustering
Quasars Probing QuasarsQuasars Probing Quasars
Hennawi, Prochaska, et al. (2007)
Transverse ClusteringTransverse Clustering
• 29 new QSO-LLSs with R < 2 Mpc/h
• High covering factor for R < 100 kpc/h
• For T(r) = (r/rT)-, = 1.6, and NHI > 1019
cm-2, rT = 9 1.7 (2.9 QSO-LBG)
Hennawi, Prochaska et al. (2007); Hennawi & Prochaska (2007)
Chuck’s object
= Keck = Gemini = SDSS
= has absorber = no absorber
En
han
cem
ent
over
UV
Bz
(re
dsh
ift)
= 2.0 = 1.6
QSO-LBG
Proximate DLAs: LOS clusteringProximate DLAs: LOS clustering
• Found 12 PDLAs out of ~ 2000 z < 2.7 quasars
Prochaska, Hennawi, & Herbert-Fort (2007)
dN
dz(< 3000 km/s) =(1.4 ±0.3)
dNdz
• Transverse clustering strength at z = 2.5 predicts that nearly every QSO
should have an absorber with NHI > 1019 cm-2 along the LOS??
• Rapid redshift evolution of QSO clustering compared to paucity of
proximate DLAs implies that photoevaporation has to be occurring.
PhotoevaporationPhotoevaporation
f/g QSO
b/g QSO
R
QSO is to DLA . . . as . . . O-star is to interstellar cloud
Γ =nphotons
nH
= 2.6 ×10−4 S56RMpc-2 n−1
H, -1
Hennawi & Prochaska (2007)
δ =500ΓNH
1020.3cm-2
⎛⎝⎜
⎞⎠⎟
−1
< 1
Otherwise it is photoevaporatedBertoldi (1989), Bertodi & McKee (1989)
Cloud survives provided
r = 17r = 19r = 21
nH = 0.1
Proximity Effects: SummaryProximity Effects: Summary
• There is a LOS proximity effect but not a transverse one.
• Photoevaporation plausible for absorbers near quasars.
• Our measured T(r) gives, PChuck = 1/65.
• Fluorescent emission proves Chuck’s DLA was illuminated.
• Clustering anisotropy suggests transverse systems are not.
• Two possible sources of clustering anisotropy:
– QSO ionizing photons are obscured (beamed?)
– QSOs vary significantly on timescales shorter than crossing time:
tcross ~ 4 105 yr @ = 20” (120 kpc/h).
Current limit: tQSO > 104 yr
Proximity Effects: Open QuestionsProximity Effects: Open Questions
• Can we measure the average opening angle?
– Yes, but must model photoevaporation assuming an
absorber density profile.
– Much easier for optically thin transverse effect (coming
soon).
• Does high transverse covering factor conflict with
obscured fractions (~ 10%) of luminous QSOs?
• Why did Chuck’s DLA survive whereas others are
photoevaporated?
Fluorescent Ly Emission
Fluorescent Ly Emission
Transverse Fluorescence?Transverse Fluorescence?
background QSO spectrum
2-d spectrum
f/g QSO z = 2.29
PSF subtracted 2-d spectrum
(Data-Model)/Noise
Hennawi, Prochaska, & Burles (2007)
b/g QSO z = 3.13 Implied transverse ionizing flux
gUV = 6370 UVB!
Near-IR Quasar RedshiftsNear-IR Quasar Redshifts
Transverse Fluorescence?Transverse Fluorescence?
Background QSO spectrum
2-d spectrum
f/g QSO z = 2.27
PSF subtracted 2-d spectrum
(Data-Model)/Noise
Hennawi, Prochaska, & Burles (2007)
b/g QSO z = 2.35 Implied transverse ionizing flux
gUV = 7870 UVB!
metals at this z
LyLy Emission from DLAs Emission from DLAs
Could the proximate DLA emission be fluorescence excited by the quasar ionizing flux?
Moller et al. (2004)
HST STIS Image
2-d Spectrum
QSO zQSO zDLAf Ly
(10-17 erg s-1 cm-2)
L Ly
(1042 erg s-1)
PKS 0458-02 2.286 2.0395 5.4 0.17
PC0953+4749 4.457 3.407 0.7 0.77
Q 2206-1958 2.559 1.9205 26 14
DMS 2247-0209 4.36 4.097 0.5 0.9
PHL 1222 1.922 1.9342 90 25
B 0405-331 2.57 2.570 ??? ???
PSK 0528-250 2.77 2.8115 7.4 0.49
SDSSJ 1240+1455 3.107 3.1078 43 39
Q2059-360 3.10 3.0830 20 18
Intervening DLAs
Proximate DLAs
Fluorescent PhasesFluorescent Phases
R
f/g QSOTransverse
b/g QSO
Absorber
Full Moon? Absorber
f/g QSO
Absorber
Proximate b/g QSO
A Fluorescing PDLA?A Fluorescing PDLA?
• Ly brighter than 95% of LBGs --- unlikely to be star formation.
• Detection of N(N+4) > 1014.4 cm-2 consistent with hard QSO spectrum and requires R|| < 700 kpc.
• Large fLy = 4.310-16 erg s-1 cm-2 suggests R|| ~ 300 kpc.
• If emission is Ly from QSO halo, then we can image DLA in silhouette.
Hennawi, Kollmeier, Prochaska, & Zheng (2007)
R||
DLA
b/g QSO
New Probes of HI in High-z GalaxiesNew Probes of HI in High-z Galaxies
• These observables are predictable given a model for HI distribution in high-z galaxies.
• The physics of self-shielding and resonant line radiative transfer are straightforward compared to other problems in galaxy formation.
Hennawi, Kollmeier, Prochaska, & Zheng (2007)
Statistics of PDLAs Fluorescent Ly Emission
Photo-evaporation of DLAs
Ly Emissivity Map Aperture Spectra
Hennawi, Prochaska,
& Herbert-Fort (2007)
Column distribution near QSOs
SummarySummary
• With projected QSO pairs, QSO environments can be studied down to ~ 20 kpc where ionizing fluxes are as large as 104 times the UVB.
• Clustering pattern of absorbers around QSOs is highly anisotropic.
• Rapid redshift evolution of QSO clustering compared to paucity of proximate DLAs implies that photoevaporation has to be occuring.
• Physical arguments indicate that DLAs within 1 Mpc of a luminous quasar can be photoevaporated.
• QSO-LLS pairs provide new laboratories to study Ly fluorescence.
• Null detections of fluorescence and clustering anisotropy suggest that quasar emission is either anisotropic or variable on timescales < 105 yr.
• Photoevaporation and fluorescent emission provide new physical constraints on the distribution of HI in high-z proto-galaxies. The input physics is relatively simple and it can be easily modeled.