21 cm Cosmology - Lunar University Network forAA48CH05-Morales ARI 23 July 2010 15:26 Reionization...

21 cm CosmologyMiguel F. Morales

Boulder, October 5th, 2010

Tuesday, November 23, 2010

See invited ARAA review

AA48CH05-Morales ARI 23 July 2010 15:26

Reionization and Cosmologywith 21-cm FluctuationsMiguel F. Morales1 and J. Stuart B. Wyithe2

1Department of Physics, University of Washington, Seattle, Washington 98195;email: [email protected] of Physics, University of Melbourne, Parkville, 3052 Victoria, Australia;email: [email protected]

Annu. Rev. Astron. Astrophys. 2010. 48:127–71

First published online as a Review in Advance onApril 20, 2010

The Annual Review of Astronomy and Astrophysics isonline at astro.annualreviews.org

This article’s doi:10.1146/annurev-astro-081309-130936

Copyright c! 2010 by Annual Reviews.All rights reserved

0066-4146/10/0922-0127$20.00

Key Wordscosmic reionization, dark energy, galaxy formation, observationalcosmology

AbstractMeasurement of the spatial distribution of neutral hydrogen via the red-shifted 21-cm line promises to revolutionize our knowledge of the epochof reionization and the first galaxies, and may provide a powerful new toolfor observational cosmology from redshifts 1 < z < 4. In this review wediscuss recent advances in our theoretical understanding of the epoch ofreionization (EoR), the application of 21-cm tomography to cosmology andmeasurements of the dark energy equation of state after reionization, andthe instrumentation and observational techniques shared by 21-cm EoR andpostreionization cosmology machines. We place particular emphasis on theexpected signal and observational capabilities of first generation 21-cm fluc-tuation instruments.

127

Ann

u. R

ev. A

stro

. Ast

roph

ys. 2

010.

48:1

27-1

71. D

ownl

oade

d fr

om w

ww

.ann

ualre

view

s.org

by D

r. M

igue

l Mor

ales

on

09/0

9/10

. For

per

sona

l use

onl

y.


The cosmological HI signal


0 (now)~6CMBBigBang

EOR

redshift

How did galaxies form?

?Tuesday, November 23, 2010

Short history of hydrogen

0 (now)~6CMBBigBang

redshift

EOREOREOR

Ionized Hydgrogen

Neutral Hydgrogen


Dark energy with HI

Wyithe & Loeb (2007)

4 Wyithe & Loeb

567 4 3100.1

1

10

0.001

0.01

0.1

1

z

<T

>

(m

K)

567 4 31010–6

10–5

10–4

10–3

10–2

10–1

100<

FH

I>V

,

<F

HI>

M

zz

!1

2

(10

-12

s-1

)

567 4 3

0.1

1

567 4 3

1

10

100

z

"

(ph

ys

ica

l M

pc

)

!c=5 f*esc=0 .0048

!c=20 f*esc=0 .0038

(1-Q

i)

Figure 1. Models for the reionization of the IGM and the subsequent post-overlap evolution of the ionizing radiation field. In each paneltwo cases are shown, corresponding to di!erent values for the critical overdensity prior to the overlap epoch ("c = 5, thin dark lines;and "c = 20, thick grey lines). We show the cases for the mean IGM with ! = 0. Upper Left Panel: The ionization rate as a functionof redshift. The observational points are from Bolton et al. (2007b). Upper Right Panel: The volume (lower curves) and mass (uppercurves) averaged fractions of neutral gas in the universe. Also shown (dotted lines) is the fraction of the IGM yet to overlap (1!Qi). Theobservational points for the volume averaged neutral fraction are from Bolton et al. (2007b), while the observed mass-fractions are fromthe damped Ly" measurements of Prochaska et al. (2005). Lower Left Panel: The mean-free-path for ionizing photons computed usingthe formalism in § 2. The data points are based on Storrie-Lombardi et al. (1994). Lower Right Panel: The evolution of the mean 21cmbrightness temperature (in mK) with redshift (solid lines). For comparison, the fraction of IGM yet to overlap (1! Qi) is overplotted.

may then be written as

dn!

dt= f",escN!

dFdt

(z)!b, (7)

where !b is the co-moving mass-density of baryons.

3 RESULTS

Figure 1 shows example models for the reionization of theIGM and the subsequent post-overlap evolution of the ion-izing radiation field. Two cases are shown, correspondingto two di!erent values for the critical overdensity prior tothe overlap epoch ("c = 5 and "c = 20), with values off",esc = 0.0048 and f",esc = 0.0038 respectively. These casesdo not represent the best fit to the data, but rather bracketthe range of the overlap epoch redshifts (6 ! z ! 7) forwhich our model is consistent with observations at z ! 6(without invoking an additional population of more massivestars at high redshift; e.g. Wyithe & Loeb 2003). The ex-amples have di!erent values of "c, and thus illustrate themild dependence of our results on this unknown parame-ter. We note that the values of f",esc required for our modelto reproduce existing observations are in excellent agree-ment with external considerations. In particular our valueof f",esc ! a few "10!3 corresponds to product of recent

estimates for the escape fraction (a few "10!2; Gnedin,Kravtsov, & Chen 2007), with estimates of the average star-formation rate (! 10!1 from the ratio between the mass den-sity in stars and baryons; Fukugita, Hogan & Peebles 1998).

In the top left panel of Figure 1 we show the evolutionof the ionization rate. The observational points are fromthe simulations of Bolton et al. (2007b; based on the obser-vations of Fan et al. 2006). In the top-right panel we plotthe corresponding volume (lower curves) and mass (uppercurves) averaged fractions of neutral gas in the universe.Also shown (dotted lines) is the fraction of the IGM yetto overlap (1 # Qi). The redshift where these curves dropto zero is normally quoted as the redshift of reionization,and these curves correspond approximately to the standardsemi-analytic calculation (e.g. Haiman & Loeb 1997). How-ever our formalism allows for the calculation of both massand volume averaged neutral fractions to lower redshifts.The observational points for the volume averaged neutralfraction are from Bolton et al. (2007b), while the observedmass-fractions are from the damped Ly" measurements ofProchaska et al. (2005), and therefore represent lower limitson the total HI content of the IGM. Both curves show excel-lent agreement with these quantities, despite their di!eringby 3 orders of magnitude. In the lower left panel we plot theevolution of the ionizing photon mean-free-path. The data

c" 0000 RAS, MNRAS 000, 000–000

<FH1>V related to Lyman-α absorption, ~10-4

<FH1>M related to H1 emission, ~10-2


HI during EoR

100

200

300

400

500

7

7.58

8.59

Mpc/h

z Zahn Furlanetto, Sokasian, Hernquist (2004)


Statistical EoR detection

f

iy

ImageCube

ix

v

f

Visibilities

u

v

h

FourierRepresentation

u

FT SkyCoordinates

FTFrequency

Morales & Hewitt (2004)


Spherical symmetry

Morales & Hewitt (2004)

v

u

h


EoR power spectrum

k [Mpc-1 ]

(k3 P

/ 2π

2 )1/2 [M

pc-3

/2 J

y H

z]

10-3 10-2 10-1 100 10110-1

100

101

102

103

Furlanetto, Zaldarriaga, Hernquist (2004a,b)Bowman, Morales & Hewitt (2005)

Kaplinghat (2005)z = 8, 360 hours of integration


McQuinn

Power spectrum dynamics


HI power spectra evolution3

Fig. 1.— Redshift evolution of the 21 cm power spectrum inour fiducial model. The redshifts and volume-averaged ionizationfractions shown are (!xi", z) = (0.02, 11.46); (0.15, 8.76); (0.21,8.34); (0.54, 7.32); (0.82, 6.90); and (0.96, 6.77). The MWA probeswavenumbers between roughly k # 0.1$ 1h Mpc!1. At the lowestionization fraction shown (!xi" = 0.02), the 21 cm power spectrumtraces the density power spectrum. There is a brief equilibrationphase (see text) where the 21 cm power spectrum amplitude dropson large scales near !xi" = 0.15. The amplitude on MWA scalessubsequently increases until !xi" # 0.5, and then falls o! at higherionization fractions. As the ionization fraction grows, after a briefsteepening during the equilibration phase, the slope of the powerspectrum on MWA scales flattens.

ble models, the bias of the ionizing sources is a weakfunction of redshift (Furlanetto et al. 2006b). The 21cm power spectrum depends somewhat more strongly onredshift at fixed ionization fraction than the ionizationpower spectrum, since the 21 cm field explicitly involvesthe density field, which evolves in time. Nevertheless,this dependence is relatively weak during most of theEoR when the ionization fluctuations are large (§2.3),and one can think of our predictions in a given model asroughly invariant with redshift for a given ionization frac-tion. Moreover, we aim to extract information regardingthe filling factor of HII regions from the observations, soit will be convenient to consider the signal as a functionof this quantity. Of course, as we noted earlier, the un-certain mapping between ionization fraction and redshiftcan strongly impact the detectability of the signal (§3).

2.2. Simulated Power Spectrum in our Fiducial Model

In Figure 1, we show the spherically-averaged 21 cmpower spectrum from our fiducial model for a range ofredshifts and ionization fractions. At early times and lowionization fraction (z = 11.46, !xi" = 0.02), the 21 cmpower spectrum simply traces the density power spec-trum, except on very small scales where early HII re-gions have some impact. At slightly later times, there isa brief phase (at !xi" = 0.15, 0.21, z = 8.76, 8.34) wherethe large-scale 21 cm power spectrum falls below the den-

sity power spectrum, and steepens in slope. This occursbecause the large-scale overdense regions initially containmore neutral hydrogen than underdense ones, and con-sequently appear brighter in 21 cm. On the other hand,the overdense regions ionize first, and quickly transitionto being dimmer in 21 cm than underdense ones, whichremain neutral. This transition leads to a brief ‘equili-bration’ phase where overdense and underdense regionshave similar brightness temperatures, and the large-scale21 cm power spectrum is low as a result (Furlanetto etal. 2004, Wyithe & Morales 2007). At these early stagesof reionization, however, our calculations may be inac-curate since we neglect the impact of spin temperaturefluctuations (Pritchard & Furlanetto 2007).

After the brief equilibration phase, the HII regionsquickly grow, boosting the large-scale power spectrumand suppressing the power on small scales. As we demon-strate in the next section, the wavenumbers relevant forthe MWA are k # 0.1 $ 1h Mpc!1. On these scales, theamplitude of the 21 cm power spectrum rises rapidly fromwhen the filling factor of ionized regions is !xi" = 0.21to when !xi" = 0.54, and then drops o! at higher ion-ized fractions, falling rather quickly when !xi" ! 0.8. Inconjunction with the increased power, the slope flattensand is close to k3P21(k) % const. for !xi" ! 0.6.

Provided that there are substantial ionization fluctu-ations on the scales that the MWA is sensitive to, it isunsurprising that the 21 cm power spectrum amplitudepeaks around the epoch in which the IGM is & 50%ionized. Recall that the variance in the ionization fieldaveraged on small scales, !2

x = !x2i " $ !xi"2, must re-

duce to !2x # !xi" $ !xi"2 in the limit that each pixel

is either completely neutral or completely ionized. Inthis limit, the variance in the ionization field peaks when!xi" = 0.5, close to the ionization fraction at which the21 cm power spectrum reaches its maximum on MWAscales. Of course the 21 cm power spectrum is morecomplicated, since it depends on the cross-correlation be-tween ionization and overdensity and higher order con-tributions (Lidz et al. 2007b), and since a large portionof the ionization variance comes from scales that are notprobed by the MWA. Nevertheless, our simple argumentmotivates why the maximum in fluctuation amplitudeoccurs near !xi" # 0.5.

In summary, although the MWA may be limited tomeasuring the power spectrum over # a decade in scale,our model 21 cm power spectra evolve considerably withredshift and ionization fraction over this range. Hence,sensitivity only to a decade in scale may still be quitevaluable, provided the MWA can make measurementsover a number of redshift intervals.

2.3. Model Dependence

How robust to model uncertainties is the rise and fall in21 cm power spectrum amplitude, and the flattening inpower spectrum slope, at increasing ionization fraction?In order to check this, we examine the amplitude andslope of the 21 cm power spectrum for two other casesthat roughly bracket model uncertainties. McQuinn etal. (2007a) investigated the many uncertain physical pa-rameters that can impact reionization and found thatthe nature of the ionizing sources and the abundanceof mini-halos have the strongest influence on the ioniza-tion power spectrum. If the sources are rare but very

LidzTuesday, November 23, 2010

Galactic emission (polarized and Faraday rotated)

Bright point sources

Faint point sources

Instrumental contamination

Radio recombination lines

RFI

Mode mixing

…

Why is this hard? Foregrounds

!Tuesday, November 23, 2010

Foreground symmetryx

y

z FT

x

y

z FT


k-space measurement!

!

!"#$%"&'()*'+,-./

+,01'23$1'&3

$

"

45&!)4'65(78)5*

&9'((&

#$%&'%$()*+


Mode mixing

Frontier of foreground subtraction is interactions between calibration and foregrounds

Need measurement fidelity of 10-4 – 10-6

Effectively a product of the calibration errors and foreground uncertainty


Examples

Chromatic array beam (PSF) & residual source flux, residual frequency ripple

Polarized foreground & polarization mis-calibration, flux leakage from Q & U → I

Antenna beam dependence & point sources, decorrelation of visibilities at different frequencies

Freq

uenc

y

Position

Mis-subtractedsource

Representativeline of sight



!

!"#$%"&'()*'+,-./

+,01'23$1'&3

$

"

45&!)4'65(78)5*

&9'((&

#$%&'%$()*+


Bright source location error

Datta et al. (2010)


Foreground subtraction

for 300 hours of integration with the MWA (Bowman et al. 2009) Datta et al. (2010)


Confusion level sources

"

!"

!"

!"!#

!"!!

!""

!"!!

!""

k!

[cM

pc"

1]

6%&&'.50:5.(8/*&.8),&12.*3!*(4356$37.89

&

Bowman et al. (2008)



!

!"#$%"&'()*'+,-./

+,01'23$1'&3

$

"

45&!)4'65(78)5*

&9'((&

#$%&'%$()*+


MWA sensitivity

for 300 hours of integration with the MWA (Bowman et al. 2009)


0

1

2

3

4

5

6

7

8x 108

P [J

y2 Hz2 ]

0

1

2

3

4

5

6

7

8x 108

10-3 10-2 10-1 100 10110-1

100

101

102

103

k [Mpc-1 ]

(k3 P

/ 2π

t2 )1/2 [M

pc-3

/2 J

y H

z]

10-3 10-2 10-1 100 10110-1

100

101

102

103

MWA power spectrum sensitivity

z = 8, 360 hours of Furlanetto, Zaldarriaga, Hernquist (2004a,b)

Bowman, Morales & Hewitt (2005)Kaplinghat (2005)


k-space Fisher matrixes

Tuning parameter sensitivity

Visbal, in prep.


Opportunities for lunar 21 cmLow RFI

No ionosphere

Slow rotation rate

Dark ages (redshift > 40)

Very large arrays, novel hardware (e.g. MOFF?)

Challenge of very fast ground-based development


See invited ARAA review

AA48CH05-Morales ARI 23 July 2010 15:26

Reionization and Cosmologywith 21-cm FluctuationsMiguel F. Morales1 and J. Stuart B. Wyithe2

1Department of Physics, University of Washington, Seattle, Washington 98195;email: [email protected] of Physics, University of Melbourne, Parkville, 3052 Victoria, Australia;email: [email protected]

Annu. Rev. Astron. Astrophys. 2010. 48:127–71

First published online as a Review in Advance onApril 20, 2010

The Annual Review of Astronomy and Astrophysics isonline at astro.annualreviews.org

This article’s doi:10.1146/annurev-astro-081309-130936

Copyright c! 2010 by Annual Reviews.All rights reserved

0066-4146/10/0922-0127$20.00

Key Wordscosmic reionization, dark energy, galaxy formation, observationalcosmology

AbstractMeasurement of the spatial distribution of neutral hydrogen via the red-shifted 21-cm line promises to revolutionize our knowledge of the epochof reionization and the first galaxies, and may provide a powerful new toolfor observational cosmology from redshifts 1 < z < 4. In this review wediscuss recent advances in our theoretical understanding of the epoch ofreionization (EoR), the application of 21-cm tomography to cosmology andmeasurements of the dark energy equation of state after reionization, andthe instrumentation and observational techniques shared by 21-cm EoR andpostreionization cosmology machines. We place particular emphasis on theexpected signal and observational capabilities of first generation 21-cm fluc-tuation instruments.

127

Ann

u. R

ev. A

stro

. Ast

roph

ys. 2

010.

48:1

27-1

71. D

ownl

oade

d fr

om w

ww

.ann

ualre

view

s.org

by D

r. M

igue

l Mor

ales

on

09/0

9/10

. For

per

sona

l use

onl

y.


21 cm Cosmology - Lunar University Network forAA48CH05-Morales ARI 23 July 2010 15:26 Reionization...

Documents

Transcript of 21 cm Cosmology - Lunar University Network forAA48CH05-Morales ARI 23 July 2010 15:26 Reionization...