Observing the First Galaxies and the Reionization Epoch
Steve Furlanetto
UCLA
February 5, 2008
Outline
Introduction: Observing Reionization
Galaxy Surveys
Current observations of LAEs
The Clustering Signature
The 21 cm Transition as a Cosmological Probe
Basic Physics
The Mean 21 cm Background
Measurements and Challenges
The Pre-reionization IGM
Reionization
Conclusion
A Brief History of the Universe
Big Bang
Last
Scattering
Last scattering: z=1089, t=379,000 yr
Dark Ages
Today: z=0, t=13.7
Gyr
First
Galaxies
Reionization
Reionization: z=6-20, t=0.2-1 Gyr
Galaxies,
Clusters, etc.
First galaxies: ?
G. Djorgovski
Reionization
First stars and galaxies produce ionizing photons
Ionized bubbles grow and merge
Affects all baryons in the universe
Phase transition
Mesinger & Furlanetto
Reionization
First stars and galaxies produce ionizing photons
Ionized bubbles grow and merge
Affects all baryons in the universe
Phase transition
Mesinger & Furlanetto
Reionization
First stars and galaxies produce ionizing photons
Ionized bubbles grow and merge
Affects all baryons in the universe
Phase transition
Mesinger & Furlanetto
Reionization
First stars and galaxies produce ionizing photons
Ionized bubbles grow and merge
Affects all baryons in the universe
Phase transition
Mesinger & Furlanetto
Reionization:
Observational Constraints
Quasars/GRBs
CMB optical depth
Ly
-selected galaxies
Furlanetto, Oh, & Briggs (2006)
Ly
Emitters and HII Regions
Total optical depth in
Ly
transition:
GP
3x10
5 x
HI
1
z
7
3 / 2
Damping wings are strong
IGM HI
x
H
=0
LAEs During Reionization x
H
=0.26
x
H
=0.51
x
H
=0.77
Mesinger & Furlanetto (2007)
z=9, R=125 observation, with M>1.7x10
10 Msun
Galaxies in small bubbles (underdense regions) masked out by absorption
A Declining Number Density?
Largest survey to date with Subaru
Apparent decline at bright end
Disputed by Dawson et al. (2007)
Kashikawa et al. (2006)
A Declining Number Density?
Similar behavior to z=7
One (!) detection
L>10 43 erg/s detection threshold
Iye et al. (2006)
An Increasing Number Density?
Stark et al. (2007) found 6 candidate LAEs behind massive clusters
Search along lensing caustics z=9-10
L~10 41 -10 42 erg/s
Most obvious interlopers ruled out
Stark et al. (2007), z=9
Kashikawa et al. (2006)
An Increasing Number Density?
Solid curves show mass functions with absorption
Four scenarios for luminosities
(right to left):
Same as z=6 LAEs
Same as z=6 LAEs, but Pop III
All baryons form Pop II stars, simultaneously
All baryons form Pop III stars, simultaneously
Reasonable scenarios require fully ionized!
Mesinger & Furlanetto (2008)
LAE Clustering
During Reionization
Nearly randomly distributed galaxy population
Small bubble: too much extinction, disappears
Large bubble: galaxies visible to survey
LAE Clustering
During Reionization
Small bubble: too much extinction, disappears
Large bubble: galaxies visible to survey
Absorption selects large bubbles, which tend to surround clumps of galaxies
LAE Clustering
During Reionization
Small bubble: too much extinction, disappears
Large bubble: galaxies visible to survey
Absorption selects large bubbles, which tend to surround clumps of galaxies
Enhanced Clustering
During Reionization
Shows enhanced probability to have
N>1 galaxies in an occupied cell
Measuring requires deep survey over
~10 6 -10 7 Mpc 3
Mesinger & Furlanetto (2008)
The Future of LAE Surveys
Advantages:
Familiar strategies
Study galaxies as well
Disadvantages:
Uncertainties about galaxy formation
Need large volume, deep surveys
Indirect information about IGM
The Spin-Flip Transition
Proton and electron both have spin magnetic fields
Produces 21 cm radiation
( n
=1420 MHz)
Extremely weak transition
Mean lifetime ~10 7 yr
Optical depth ~1% in fully neutral IGM
The 21 cm Transition
Map emission (or absorption) from IGM gas
Requires no background sources
Spectral line: measure entire history
Direct measurement of
IGM properties
No saturation!
T b
23 x
HI
(1
)
1
z
10
1/ 2
T
S
T bkgd
T
S
H ( z ) /(1
z )
v r
/
r
mK
SF, AS, LH (2004)
The Spin Temperature
CMB photons drive toward invisibility:
T
S
=T
CMB
Collisions couple T
S
to T
K
Dominated by electron exchange in H-H collisions in neutral medium (Zygelman 2005)
Dominated by H-e collisions in partially ionized medium (Furlanetto & Furlanetto
2006), with some contribution from H-p collisions (Furlanetto & Furlanetto 2007)
The Global Signal:
The Dark Ages
Straightforward physics
Expanding gas
Recombination
Compton scattering
SF, PO, FB (2006)
1
S
1/2
0
S
1/2
The Wouthuysen-Field
Mechanism I
Selection Rules:
D F=0,1 (except F=0 F=0)
2
P
3/2
1
P
3/2
1
P
1/2
0
P
1/2
Mechanism is effective with
~0.1 Ly
photon/baryon
The Wouthuysen-Field
Mechanism II
Ly
Ly
Ly
Ly
Relevant photons are continuum photons that redshift into the Ly
resonance
…
The Global Signal:
First Light
First stars (quasars?) flood Universe with photons
W-F effect
Trigger absorption in cold IGM feedback
Pop III Stars
Pop II Stars
SF (2006)
The First Sources of Light:
X-ray Heating
X-rays are highly penetrating in IGM
Mean free path >Mpc
Deposit energy as heat, ionization
Produced by…
Supernovae
Stellar mass black holes
Quasars
Very massive stars
The Global Signal:
First Light
First stars (quasars?) flood Universe with photons
W-F effect
Heating
Ionization
Timing depends on f
*
, f esc
, f
X
, stellar population feedback
Pop III Stars
Pop II Stars
SF (2006)
21 cm Observations
Experiments
Global Signal: CoRE-
ATNF, EDGES
Fluctuations: 21CMA,
LOFAR, MWA, GMRT,
PAPER, SKA
Imaging: SKA
MWA
Terrestrial Interference
Mileura spectrum, 15 sec integrations
Two types:
Fixed site (low frequency filling factor)
Aircraft/meteor trail reflections (low duty cycle)
Basic strategy: excise contaminated channels
Bowman et al. (2007)
Ionospheric Distortions
Refraction in ionosphere distorts wavefronts
Analog of optical seeing layer
Solved on software level with calibration sources
Challenge: wide-field imaging
W. Cotton
Astronomical Foregrounds
Map at 150 MHz
Contours are in Kelvin
Landecker et al. (1969)
The Synchrotron Foregrounds
A single synchrotron electron produces a broad but smooth spectrum
B
Intensity
Frequency e path
The Synchrotron Foregrounds
A single synchrotron electron produces a broad but smooth spectrum
Intensity
Electron velocity scales the spectrum
Frequency
The Synchrotron Foregrounds
Synchrotron spectrum mirrors distribution of
Intensity fast electrons
Typically near powerlaw, with ~K/MHz gradient
Frequency
Measuring the Global Signal?
Signal gradient is few mK/MHz
Foregrounds vary as
(near) power law
Synchrotron, free-free
Gradient is few K/MHz
CoRE-ATNF, EDGES experiments are trying
Distinctive shape may help
SF (2006)
Foregrounds on Small Scales
0.5 MHz
Foreground Removal
Total Signal ~ 400 K
Removal algorithms fairly well-developed
Zaldarriaga et al.
(2004), Morales &
Hewitt (2004), Santos et al. (2005),
McQuinn et al. (2007)
T b
Cleaned Signal ~ 10 mK
Frequency
Foreground Noise
Thermal noise is NOT smooth: varies between each channel
For first generation instruments, 1000 hr observations still have S/N<1 per pixel
Imaging is not possible until SKA!
The Murchison Widefield Array
Low Frequency
Demonstrator under construction (fully funded, first light
~2008)
Bowman et al. (2007)
Located on sheep ranch in Western
Australia
The Murchison Widefield Array
Bowman et al. (2007)
Bowtie antennae grouped in tiles of 16
Broad frequency response
Large field of view
Murchison Widefield Array:
Low Frequency Demonstrator
Instrument characteristics
Radio-quiet site
500 16-element antennae in
1.5 km distribution
7000 m 2 total collecting area
Full cross-correlation of all 500 antennae
80-300 MHz
32 MHz instantaneous bandwidth at 8 kHz resolution
20-30 degree field of view Bowman et al. (2007)
Error Estimates: z=8
MWA
Foreground limit
(Mpc -1 )
SKA
Survey parameters
z=8
T sys
=440 K t int
=1000 hr
B=6 MHz
No systematics!
MWA (solid black)
A eff
=7000 m 2
1.5 km core
SKA (dotted blue)
A eff
=1 km 2
5 km core
LOFAR very close to MWA
Error Estimates: z=12
Foreground limit
MWA
SKA
Survey parameters
z=12
T sys
=1000 K t int
=1000 hr
B=6 MHz
No systematics!
MWA (solid black)
A eff
=9000 m 2
1.5 km core
SKA (dotted blue)
A eff
=1 km 2
5 km core
(Mpc -1 )
That’s a whole lotta trouble…
So what good is it, really?
The Global Signal
Four Phases
Dark Ages
First Stars
First Black Holes
Reionization
Reionization BHs Stars Dark
Ages
SF (2006)
Ly
Fluctuations
Ly
photons decrease T
S near sources (Barkana &
Loeb 2004)
Clustering
1/r 2 flux
Strong absorption near dense gas, weak absorption in voids
Cold, Absorbing
Cold, invisible
Ly
Fluctuations
Ly
photons decrease T
S sources near
Clustering
1/r 2 flux
Strong absorption near dense gas, weak absorption in voids
Eventually saturates when IGM coupled everywhere
Cold, Absorbing
The Pre-Reionization Era
Net
X-ray
Ly
Pritchard & Furlanetto (2007)
Thick lines: Pop II model, z r
=7
Thin lines: Pop III model, z r
=7
Dashed: Ly
fluctuations
Dotted: Heating fluctuations
Solid: Net signal
X-ray Fluctuations
X-ray photons increase T
K near sources (Pritchard &
Furlanetto 2007)
Clustering
1/r 2 flux
Hot IGM near dense gas, cool IGM near voids
Hot
Cool
+
X-ray and Ly
Fluctuations
=
Hot, emitting
Invisible
The Pre-Reionization Era
Net
X-ray
Ly
Pritchard & Furlanetto (2007)
Thick lines: Pop II model, z r
=7
Thin lines: Pop III model, z r
=7
Dashed: Ly
fluctuations
Dotted: Heating fluctuations
Solid: Net signal
+
X-ray Fluctuations
=
Hot, emitting
Cold, absorbing
The Pre-Reionization Era
Net
X-ray
Ly
Pritchard & Furlanetto (2007)
Thick lines: Pop II model, z r
=7
Thin lines: Pop III model, z r
=7
Dashed: Ly
fluctuations
Dotted: Heating fluctuations
Solid: Net signal
+
X-ray Fluctuations
=
Hot, emitting
The Pre-Reionization Era
Net
X-ray
Ly
Pritchard & Furlanetto (2007)
Thick lines: Pop II model, z r
=7
Thin lines: Pop III model, z r
=7
Dashed: Ly
fluctuations
Dotted: Heating fluctuations
Solid: Net signal
21 cm Observations:
Reionization
100 Mpc comoving
Mesinger & Furlanetto
Reionization “Simulations”
100 Mpc comoving
Implement in numerical simulation boxes
Step 1: Generate initial conditions
Step 2: Identify galaxies
Mesinger & Furlanetto
Biased Galaxy Formation:
Peaks and Patches
Galaxies form at peaks in the density field
Threshold decreases with time
More galaxies
Bigger galaxies
Identifying Galaxies
Filter density field to find peaks
Use excursion-set barrier to find masses
Adjust locations using
Zeldovich approximation
Similar to “peak-patch” method (Bond & Myers
1996), PINOCCHIO,
PTHALOS
Mesinger & Furlanetto (2007)
Identifying Galaxies
Excellent statistical agreement
Large-scale structure
Poisson noise
Accurate one-to-one for large galaxies
(Bond & Myers 1996)
Mesinger & Furlanetto (2007)
Reionization “Simulations”
100 Mpc comoving
Implement in numerical simulation boxes
Step 1: Generate initial conditions
Step 2: Identify galaxies
Mesinger & Furlanetto
Photon Counting
Assume galaxies have fixed ionizing efficiency
Isolated galaxies generate HII regions
Clustered galaxies work together
Galaxy
Ionized IGM
Neutral IGM
Photon Counting
Assume galaxies have fixed ionizing efficiency
Isolated galaxies generate HII regions
Clustered galaxies work together
Reionization “Simulations” z=9.75, x i
=0.2
Implement in numerical simulation boxes
Step 1: Generate initial conditions
Step 2: Identify galaxies
Step 3: Paint on bubbles, working from outside in
100 Mpc comoving
Mesinger & Furlanetto
Reionization “Simulations” z=8.75, x i
=0.4
Implement in numerical simulation boxes
Step 1: Generate initial conditions
Step 2: Identify galaxies
Step 3: Paint on bubbles, working from outside in
100 Mpc comoving
Mesinger & Furlanetto
Reionization “Simulations” z=8, x i
=0.6
Implement in numerical simulation boxes
Step 1: Generate initial conditions
Step 2: Identify galaxies
Step 3: Paint on bubbles, working from outside in
100 Mpc comoving
Mesinger & Furlanetto
Reionization “Simulations” z=7.25, x i
=0.8
Implement in numerical simulation boxes
Step 1: Generate initial conditions
Step 2: Identify galaxies
Step 3: Paint on bubbles, working from outside in
Five hours on desktop!
100 Mpc comoving
Mesinger & Furlanetto
Success!
Filter A on halos Filter B on halos from N-body simulation from N-body simulation
Radiative
Transfer
Simulation
Excellent match for large scale features
Map details depend on filtering algorithm
Zahn et al. (2007), Mesinger & Furlanetto (2007)
The 21 cm Power Spectrum
MWA
SKA
MWA
SKA
(Mpc -1 )
Mesinger & Furlanetto (2007)
21 cm-Galaxy
Cross-Correlation z=8, x i
=0.6
Mesinger & Furlanetto
21 cm-Galaxy
Cross-Correlation
Key advantages
Unambiguous confirmation of cosmological signal
Vastly reduces difficulty of foreground cleaning
Only emission from sources in survey slice contributes
Increase sensitivity and dynamic range
Helps with angular structure
Science!
The 21 cm-Galaxy
Cross-Correlation
Can be done with LAEs or
LBGs
Significant advantages in 21 cm data analysis (SF & AL
2007)
Challenge: wide-field near-IR surveys
JWST?
JDEM?
Ground-based cameras?
Lidz, Zahn, & Furlanetto (in prep)
Conclusions
LAE searches beginning to pay off
Strange star formation?
Robust signatures will be in clustering
The 21 cm transition offers great possibilities
Pre-reionization: properties of first sources, cosmology
Reionization: morphology and growth of bubbles
Experimental challenges still large
Good synergy with other probes of high-z universe!
Cross-correlation, Ly
searches, quasars…
See our Physics Reports review (Furlanetto,
Oh, & Briggs 2006, astro-ph/0608032) for more information on 21 cm possibilities!