The Twilight Zoneof Reionization
The Twilight Zoneof Reionization
Steve Furlanetto
Yale University
March 13, 2006
Steve Furlanetto
Yale University
March 13, 2006Collaborators: F. Briggs, L. Hernquist, A. Lidz, A. Loeb,
M. McQuinn, S.P. Oh, J. Pritchard, A. Sokasian, O. Zahn, M. Zaldarriaga
OutlineOutline
Reionization on a Global Level Assumptions Feedback
Inhomogeneous Reionization Early Phases Late Phases
Observational Prospects
Reionization on a Global Level Assumptions Feedback
Inhomogeneous Reionization Early Phases Late Phases
Observational Prospects
Simple Reionization Models: Ingredients
Simple Reionization Models: Ingredients
Source Term: Identify sources Assign f*
Assign IMF Assign fesc
Sink Term: ne nH C
Source Term: Identify sources Assign f*
Assign IMF Assign fesc
Sink Term: ne nH C
Sokasian et al. (2003)
Simple Reionization Models: Ingredients
Simple Reionization Models: Ingredients
Source Term: Identify sources Assign f*
Assign IMF Assign fesc
Sink Term: ne nH C
Doesn’t fit WMAP+SDSS
Source Term: Identify sources Assign f*
Assign IMF Assign fesc
Sink Term: ne nH C
Doesn’t fit WMAP+SDSS
Reionization Models: Feedback I
Reionization Models: Feedback I
Any or all parameters may evolve! Photoheating Metallicity H2 cooling Feedback on clumping
Double reionization difficult to arrange (SF, AL 2005)
Any or all parameters may evolve! Photoheating Metallicity H2 cooling Feedback on clumping
Double reionization difficult to arrange (SF, AL 2005)
Reionization Models:Feedback II
Reionization Models:Feedback II
Pop III/Pop II transition IGM Enrichment Clustering ISM Enrichment Gradual?
See Cen’s talk later on
Pop III/Pop II transition IGM Enrichment Clustering ISM Enrichment Gradual?
See Cen’s talk later on
SF, AL (2005)
The Global 21 cm SignalThe Global 21 cm Signal
SF (in prep)
Pop II Stars Pop III Stars
Inhomogeneous ReionizationInhomogeneous Reionization
z=18.3z=18.3
13 Mpc comoving13 Mpc comoving
=0.1 MHz=0.1 MHz
SF, AS, LH (2004)SF, AS, LH (2004)
z=16.1z=16.1
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
z=14.5z=14.5
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
z=13.2z=13.2
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
z=12.1z=12.1
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
z=11.2z=11.2
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
z=10.4z=10.4
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
z=9.8z=9.8
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
z=9.2z=9.2
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
z=8.7z=8.7
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
z=8.3z=8.3
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
z=7.9z=7.9
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
z=7.5z=7.5
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
z=9.2z=9.2
=0.1 MHz=0.1 MHz
13 Mpc comoving13 Mpc comoving
SF, AS, LH (2004)SF, AS, LH (2004)
Inhomogeneous ReionizationInhomogeneous Reionization
Photon CountingPhoton Counting
Simple ansatz:
mion = mgal
= f* fesc N/b / (1+nrec)
Then condition for a region to be fully ionized is
fcoll > -1
Simple ansatz:
mion = mgal
= f* fesc N/b / (1+nrec)
Then condition for a region to be fully ionized is
fcoll > -1
Neutral IGM
Ionized IGM
Galaxy
Photon CountingPhoton Counting
Simple ansatz:
mion = mgal
= f* fesc N/b / (1+nrec)
Then condition for a region to be fully ionized is
fcoll > -1
Simple ansatz:
mion = mgal
= f* fesc N/b / (1+nrec)
Then condition for a region to be fully ionized is
fcoll > -1
Neutral IGM
Ionized IGM
Galaxy
Photon CountingPhoton Counting
Simple ansatz:
mion = mgal
= f* fesc N/b / (1+nrec)
Then condition for a region to be fully ionized is
fcoll > -1
Simple ansatz:
mion = mgal
= f* fesc N/b / (1+nrec)
Then condition for a region to be fully ionized is
fcoll > -1
Neutral IGM
Ionized IGM?
Galaxy
Photon CountingPhoton Counting
Simple ansatz:
mion = mgal
= f* fesc N/b / (1+nrec) Then condition for a
region to be fully ionized is
fcoll > -1
Can construct an analog of Press-Schechter mass function = mass function of ionized regions
Simple ansatz:
mion = mgal
= f* fesc N/b / (1+nrec) Then condition for a
region to be fully ionized is
fcoll > -1
Can construct an analog of Press-Schechter mass function = mass function of ionized regions
Neutral IGM
Ionized IGM
Galaxy
SF, MZ, LH (2004a)SF, MZ, LH (2004a)
=40xH=0.96
xH=0.70
xH=0.25
Bubble SizesBubble Sizes
Bubbles are BIG!!! Many times the size of
each galaxy’s HII region
2 Mpc = 1 arcmin Much larger than
simulation boxes
Bubbles are BIG!!! Many times the size of
each galaxy’s HII region
2 Mpc = 1 arcmin Much larger than
simulation boxes
Typical galaxy bubble
SF, MZ, LH (2004a)SF, MZ, LH (2004a)
=40xH=0.96
xH=0.70xH=0.25
Bubble SizesBubble Sizes
Bubbles are BIG!!! Have characteristic
size Scale at which typical
density fluctuation is enough to ionize region
Galaxy bias gives a boost!
Bubbles are BIG!!! Have characteristic
size Scale at which typical
density fluctuation is enough to ionize region
Galaxy bias gives a boost!
€
σ(R) ≈ δx
The Characteristic Bubble SizeThe Characteristic Bubble Size
Bubbles are BIG!!! Have characteristic
size Depends primarily on
the bias of ionizing sources
Bubbles are BIG!!! Have characteristic
size Depends primarily on
the bias of ionizing sources
xH=0.84
xH=0.025
SF, MM, LH (2005)SF, MM, LH (2005)
xH=0.35
€
fcoll ≈ ζ f coll (1+ bδ) >1
SF, MM, LH (2005)SF, MM, LH (2005)
Bubbles: Redshift DependenceBubbles: Redshift Dependence
Bubbles are BIG!!! Have characteristic
size Sizes independent of
z (for a fixed xH)
Bubbles are BIG!!! Have characteristic
size Sizes independent of
z (for a fixed xH)
xH=0.84
xH=0.025
xH=0.35
SF, MM, LH (2005)SF, MM, LH (2005)
BubblesBubbles
Bubbles are BIG!!! Have characteristic
size Sizes independent of
z (for a fixed xH) It works! See
McQuinn talk and poster
Bubbles are BIG!!! Have characteristic
size Sizes independent of
z (for a fixed xH) It works! See
McQuinn talk and poster
xH=0.84
xH=0.025
xH=0.35
SF, MM, LH (2005)SF, MM, LH (2005)
A Curious Result…A Curious Result…
FZH04 bubbles grow to be infinitely large!
What do we mean by a “bubble”? Full extent of ionized
gas? (Wyithe & Loeb 2004)
Mean free path of ionizing photon? (SF, SPO 2005)
FZH04 bubbles grow to be infinitely large!
What do we mean by a “bubble”? Full extent of ionized
gas? (Wyithe & Loeb 2004)
Mean free path of ionizing photon? (SF, SPO 2005)
xH=0.84
xH=0.025
xH=0.35
Much Ado About ClumpingMuch Ado About Clumping
For bubble to grow, ionizing photons must reach bubble wall
For bubble to grow, ionizing photons must reach bubble wall
Neutral IGM
Ionized IGM
Much Ado About ClumpingMuch Ado About Clumping
Mean free path must exceed Rbub larger bubbles must ionize blobs more deeply
Mean free path must exceed Rbub larger bubbles must ionize blobs more deeply
Neutral IGM
Ionized IGM
Much Ado About ClumpingMuch Ado About Clumping
Outskirts of blobs contain densest ionized gas recombination rate increases with mean free path
Outskirts of blobs contain densest ionized gas recombination rate increases with mean free path
Neutral IGM
Ionized IGM
Much Ado About ClumpingMuch Ado About Clumping
Growing bubble thus requires ion rate > recombination rate (see also Miralda-Escude et al. 2000)
Clumping factor is model-dependent!!!
Growing bubble thus requires ion rate > recombination rate (see also Miralda-Escude et al. 2000)
Clumping factor is model-dependent!!!
Neutral IGM
Ionized IGM
€
dfcoll
dt> AuC(R,δ)
SF, SPO (2005)SF, SPO (2005)
xH=0.49
xH=0.32
xH=0.08
Bubbles and RecombinationsBubbles and Recombinations
Recombinations impose saturation radius Rmax
Rmax limit depends on… Density structure of
IGM Emissivity (rate of
collapse)
Recombinations impose saturation radius Rmax
Rmax limit depends on… Density structure of
IGM Emissivity (rate of
collapse)
xH=0.16
Overlap and Phase TransitionsOverlap and Phase Transitions
In simulations, reionization appears to be an extremely rapid global phase transition
In simulations, reionization appears to be an extremely rapid global phase transition
Gnedin (2000)Gnedin (2000)
The Hidden Meaning of OverlapThe Hidden Meaning of Overlap
Gnedin (2000)Gnedin (2000)
Box Size
SF, SPO (2005)SF, SPO (2005)
Rmax
Without recombinations
Fuzzy OverlapFuzzy Overlap
For any point, overlap is complete when bubble growth saturates
Gives reionization an intrinsic width!!! Constrains density
structure Quasars show z~0.3
For any point, overlap is complete when bubble growth saturates
Gives reionization an intrinsic width!!! Constrains density
structure Quasars show z~0.3
SF, SPO (2005)SF, SPO (2005)
Much Ado About ClumpingMuch Ado About Clumping
Assuming uniform ionizing flux: C>30 (Gnedin & Ostriker 1997)
Assuming voids ionized first: thin lines (MHR00)
Assuming uniform ionizing flux: C>30 (Gnedin & Ostriker 1997)
Assuming voids ionized first: thin lines (MHR00)
SF, SPO (2005)SF, SPO (2005)
Much Ado About ClumpingMuch Ado About Clumping
Assuming ionizing sources are clustered: thick lines Spatially variable Depends on P() AND
bubble model!!!
Assuming ionizing sources are clustered: thick lines Spatially variable Depends on P() AND
bubble model!!!
SF, SPO (2005)SF, SPO (2005)
Reionization ObservablesReionization Observables
The 21 cm Sky CMB Temperature Anisotropies Ly Emitters Quasar (or GRB) Spectra
The 21 cm Sky CMB Temperature Anisotropies Ly Emitters Quasar (or GRB) Spectra
The 21 cm Power SpectrumThe 21 cm Power Spectrum
Model allows us to compute statistical properties of signal
Rich set of information from bubble distribution (timing, feedback, sources, etc.)
Full 3D dataset
Model allows us to compute statistical properties of signal
Rich set of information from bubble distribution (timing, feedback, sources, etc.)
Full 3D dataset
xi=0.78
z=10
xi=0.13
xi=0.36xi=0.48
xi=0.59
xi=0.69
Ly Emitters and HII RegionsLy Emitters and HII Regions
Total optical depth in Ly transition:
Damping wings are strong
See many later talks!
Total optical depth in Ly transition:
Damping wings are strong
See many later talks!
IGM HI
€
τGP ≈ 3x105 xHI
1+ z
7
⎛
⎝ ⎜
⎞
⎠ ⎟3 / 2
Clustering on Large ScalesClustering on Large Scales
Large scales: Galaxies in separate
bubbles depends on clustering of bubbles
Large bubbles are rare density peaks: highly clustered
Large scales: Galaxies in separate
bubbles depends on clustering of bubbles
Large bubbles are rare density peaks: highly clustered
Clustering on Large ScalesClustering on Large Scales
Large scales: Galaxies in separate
bubbles depends on clustering of bubbles
Large bubbles are rare density peaks: highly clustered
Large scales: Galaxies in separate
bubbles depends on clustering of bubbles
Large bubbles are rare density peaks: highly clustered
Clustering on Small ScalesClustering on Small Scales
Nearly randomly distributed galaxy population
Small bubble: too much extinction, disappears
Large bubble: galaxies visible to survey
Nearly randomly distributed galaxy population
Small bubble: too much extinction, disappears
Large bubble: galaxies visible to survey
Clustering on Small ScalesClustering on Small Scales
Small bubble: too much extinction, disappears
Large bubble: galaxies visible to survey
Absorption selects large bubbles, which tend to surround clumps of galaxies
Small bubble: too much extinction, disappears
Large bubble: galaxies visible to survey
Absorption selects large bubbles, which tend to surround clumps of galaxies
Clustering on Small ScalesClustering on Small Scales
Small bubble: too much extinction, disappears
Large bubble: galaxies visible to survey
Absorption selects large bubbles, which tend to surround clumps of galaxies
Small bubble: too much extinction, disappears
Large bubble: galaxies visible to survey
Absorption selects large bubbles, which tend to surround clumps of galaxies
The Evolving Correlation Function
The Evolving Correlation Function
Top panel: Small scale bias bsm
Middle panel: Large scale bias b(infinity)
Bottom panel: Ratio of the two
Crossover scale is Rchar
Top panel: Small scale bias bsm
Middle panel: Large scale bias b(infinity)
Bottom panel: Ratio of the two
Crossover scale is Rchar
SF, MZ, LH (2005)
Secondary CMB AnisotropiesSecondary CMB Anisotropies
Nonlinear kinetic Sunyaev-Zeldovich and “Patchy Reionization” signals
Especially large for extended reionization
Nonlinear kinetic Sunyaev-Zeldovich and “Patchy Reionization” signals
Especially large for extended reionization
McQuinn et al. (2005)
Total
104103
Patchy
Quasar SpectraQuasar Spectra
SDSS J1030 (z=6.28) No flux for z=6.2-5.98
SDSS J1148 (z=6.42) Residual Flux! (White
et al. 2005, Oh & Furlanetto 2005)
A signature of reionization? (Wyithe & Loeb 2005, Fan et al. 2006)
SDSS J1030 (z=6.28) No flux for z=6.2-5.98
SDSS J1148 (z=6.42) Residual Flux! (White
et al. 2005, Oh & Furlanetto 2005)
A signature of reionization? (Wyithe & Loeb 2005, Fan et al. 2006)White et al. (2003)
Quasar SpectraQuasar Spectra
But complications! Aliasing (Kaiser &
Peacock 1991)
But complications! Aliasing (Kaiser &
Peacock 1991)
High-k mode
Line of sight
Quasar SpectraQuasar Spectra
But complications! Aliasing (Kaiser &
Peacock 1991) Transmission bias
because only see through rare voids
But complications! Aliasing (Kaiser &
Peacock 1991) Transmission bias
because only see through rare voids
Quasar SpectraQuasar Spectra
Observed variance slightly more than expected from uniform ionizing background Structure in intrinsic
quasar spectra is likely another significant contributor
Difficult but possible!
Observed variance slightly more than expected from uniform ionizing background Structure in intrinsic
quasar spectra is likely another significant contributor
Difficult but possible!Lidz, Oh, & Furlanetto (2006)
Smoothing length=40 Mpc/h
ConclusionsConclusions
Models of global reionization history subject to uncertainties about parameters Feedback especially difficult!
Inhomogeneous Reionization Early phases: photon counting Late phases: recombinations
A number of observational opportunities ahead!
Models of global reionization history subject to uncertainties about parameters Feedback especially difficult!
Inhomogeneous Reionization Early phases: photon counting Late phases: recombinations
A number of observational opportunities ahead!