1. History of Cosmological Reionization
Renyue Cen
Princeton University Observatory
@End of Dark Ages Workshop (STScI)
March 14, 2006
History of cosmic structure formation
Observational constraints on reionization
Reionization process calculations
Numerical radiative transfer simulations
Detecting first galaxies
Conclusions

2. The Standard Cosmological Model
Adiabatic, Gaussian, scale-free density perturbation
--- Baryons 5%
--- Cold Dark Matter 23%
--- Dark Energy (cosmological constant?) 72%
n=0.99, s8 = 0.9, Wbh2=0.024,
Wxh2 = 0.126, H0 = 72, L0= 0.71
(subject to adjustments in 2 days)
Spergel etal (2003)
consistent with:
Inflation
Light element
nucleosynthesis
q0 from SNe Ia
H0 (HST key
project, SNe Ia)
Age of the universe
(stellar evolution)
WHIM

3. Cosmic Timeline in Standard Model
0.0003
Time 
13 Gyr
 Redshift
106K
z=1100
30 – 15
10-6
6 - 1
1 - 0
Recom-
bination
Real
Dark
Ages
Pop III Stars
Galaxies
Quasars
1st Reion
2nd gen
Galaxies
Quasars
Final Reion
Lya forest
Majority of
Quasars
Ellipticals
Majority of
Galaxy
Clusters
LSS
3000 K
10000 K
Temp
100K
Hierarchical structure formation        …………………………………..       Log(Mnl)

4. Naïve implication: te=0.03-0.04
Observ. Constraints on Reionization
1: SDSS QSOs: neutral hydrogen fraction changes from 10-4 to >10-2 from z=5.8 to 6.3
Cen & McDonald (2002)
Fan et al (2002)
SDSS
z=6.28
Put Fan fig6 here
Naïve implication: te=

5. White, Becker, Fan, Strauss (2003)
More new z>6 quasars
White, Becker, Fan, Strauss (2003)

6. (assuming x=nHI/nHtot=0)
2: WMAP (1st Yr): te=
Kogut et al. 2003
Bennett et al (2003)
What does it mean?
zri=
(assuming x=nHI/nHtot=0)

7. Hui & Haiman 2003; Theuns et al 2002
3: Lya forest: zri < 9-10
Hui & Haiman 2003; Theuns et al 2002
Hui & Haiman (2002)

8. One viable pre-WMAP physical model:
Solution: Prolonged Reionization
Process
One viable pre-WMAP physical model:
Universe Was Reionized Twice!
(Cen 2003a; Wyithe & Loeb 2003)

9. What could reionize the universe early: More ionizing photons wanted
Quasar space density
Star formation rate
(Haiman, Abel & Madau 2001)
log [d* /dt / M yr-1 Mpc-3]
log [n(z) / n(peak)]
redshift 2 1 ? ? -1 Z Z

10. IMF for Population III (First) Stars
Recent theoretical works suggest a new picture for Pop III IMF
(Nakamura & Umemura 2001, 2002; Abel et al 2002; Bromm et al 2002):
Pop III IMF may be very
top-heavy, possibly with most of the stars with mass >~ 100 Msun
top row: 1/1000 of a box volume from z=100, to z=18.2
middle row: z=18.2 only and continuously zoom in
bottom row: z=18.2 only
Abel et al (2002)

11. top row: 1/1000 of a box volume from z=100, to z=18.2
middle row: z=18.2 only and continuously zoom in
bottom row: z=18.2 only
Bromm et al (2002)

12. The mass of Pop III stars is likely due to stellar feedback processes
Tan & McKee (2002, 2004):
The mass of Pop III stars is likely
to fall in the range of
M = Msun
due to stellar feedback processes
T=300 K due to H2 cooling
And n=1.e-4 /cm^3, the critical density
Of H2 rotational-vibrational line cooling,
Yields total core mass of 10^3 Msun.
These cores collapse inside-out to form
A protostar, which then grows rapidly
In mass, through accretion in a disk.
Below 30Msun feedback is not important
But above 100Msun feedback becomes
The bottleneck due to (1) HII region
Breakout to large distances where
Escape velocity equals the the
Sound speed of 10km/s
And (2) Lyman-alpha and FUV pressure
In the HII region.

13. Observ. Case: Massive Pop III Stars
M>140 Msun: PISN after core He depletion
(Hedge and Woosley 2002), explosive O and Si
Burning . This instability quickly disrupts. The star, ejects metals and leaves no remnant. Because it is triggered at an unusually Early stage in the star’s evolution. This PISN produces an unusual nucleosynthetic signature, which could appear in the second generation EMP (extremely metal poor) stars in the Galactic halo. Qian Wasserburg (2002) and Oh et al. (2001) used this idea to argue that these stars justify the VMS hypothesis.
When M>260Msun, as the temperatures
Resulting from PI collapse are great enough
To photodisintegrate nuclei, which counters explosive oxygen and silicon burning (Fryer, Woosley & Heger 2001; Heger & Woosley 2002). Little mass ejection and metal enrichment from such supernovae.
M=40-140: form BHs with relatively inefficient metal ejection.
M<40: form neutron stars with more normal enrichment rates (however, see Umeda & Nomoto 2003 with mixing and fallback.
There are 3 clear features in Galactic halo EMP
Abundances: specific Fe-peak element ratios (especially Zn), the widespread presence of r-process elements, and elevated [C,N,O/Fe] ratios.
Iron-peak elements (Cr-Zn): produced in explosive events
r-Process elements (A>100): these elements are thought to be produced by rapid neutron capture in hot, dense, neutron-rich environments during explosive events. Associated with stars M=8-40Msun. The absolute abundances and relative ratios of r-process elements are thus sensitive indicators of core-collapse supernova activity.
The mean [r/Fe] is similar to the solar value at all [Fe/H], but with up to 2 dex scatter at [Fe/H]~-3. The relative abundances (i.e., [Eu/Ba] are also similar to the solar values.
Primary elements (C,N,O): of these direct products of main-sequence stellar nucleosynthesis, C is easily found in EMPs, but N and O are difficult to measure.
Many EMP stars are C-rich relative to Fe.
Tumlinson et al. (2004 argue that VMS have no significant post-He nuclear burning and therefore produce no r elements (HW02). If VMS produce all the Fe up to [Fe/H]~-3, the r elements should be absent instead of appearing at [r/Fe]~ -0.5 as observed.
Wasserburg & Qian (2000) argue that the qualitative change in [r/Fe] at [Fe/H]~-3 presented an ``iron conundrum”; namely, that the wide dispersion in Eu and Ba abundances at [Fe/H}~-3 suggested unrelated sources of
Fe and r elements. They proposed a prompt (P) inventory of Fe production by an initial propulation with large Fe yields but little or no r production. This initial inventory ceased at [Fe/H]~-3, where the onset of high-frequency (H) events (SN II, tau~10^7 yrs) and low-frequency (L) events (SN Ia, tau~10^8yrs) initiated a correlation between r and Fe.
Oh et al. (2001) pointed out that the trends in [Fe/H] and Ba also appear in Si and Ca, elements produced by PISN, but not in C, which is not abundantly produced by PISN, apparently strengthening the positive evidence that VMS dominated metal enrichment from the first stellar generation.
Based on abundance patterns of extremely
metal-poor Galactic stars:
Oh et al. (2001), Qian & Wasserburg (2002):
M>140 Msun
PISN with no r-process elements
Umeda & Nomoto (2004),
Tumlinson, Venkatesan, & Shull (2004):
M=10-140Msun
Type II supernovae/hypernovae

14. Ionizing photon emission efficiency
Pop III M*=10-300Msun: eUV=40, ,000 photons/baryon
Salpeter IMF Z=0.01Zsun : eUV=3500 photons/baryon
eUV(Pop III)
/eUV(Pop II)
=10-30
Bromm, Kudritzkl
& Loeb (2001)

15. Existence of double peaks in h
h = photon production rate/photon destruction rate
= c* fesc (df*/dt) eUV/ C(1+z)3
 Double peaks: the
c*: star formation efficiency (unknown)
fesc: ionizing photon escape fraction (unknown)
eUV: ionizing photon production efficiency
df*/dt: halo formation rate (computable)
C: gas clumping factor (constrained)

16. A closer look: a pre-WMAP model
Evolution of neutral hydrogen fraction
Recent additional constraints:
Wyithe & Loeb (2004):
x=a few x
based on QSO Stromgren
sphere size
Mesinger & Haiman (2004,ApJ):
Haiman & Cen (2005,ApJ):
based on LAE LF
Malhotra & Rhoads (2004,ApJ):
White’s talk (2006):
x~0.03 based on Stromgren
sphere sizes.
Totani’s talk (2006):
X < 0.6 based on GRB spectra
nHI/nHtot & nHII/nHtot
XHI=0.1 – 0.3
@z=6-12 t=
Cen (2003a)
Redshift

17. Evolution of the mean IGM temperature
Mean IGM temperature (K)
Redshift

18. Post-WMAP: implications on Pop III star formation processes
Without Pop III massive stars: te < 0.09
With Pop III massive stars and reasonable star formation efficiency
and ionizing photon escape fraction: te =
With an inefficient metal enrichment process and
Pop III massive stars: te = 0.15 possible
To reach te = 0.17 requires either (1) ns >=1.03
or (2) c*(H2, III) > 0.01, or (3) photon escape fraction very high for Pop III
Cen (2003b)

19. A more detailed calculation (Wyithe & Cen 2006, astro-ph/0602503)
Separate treatments of halo gas and
IGM in metal enrichment
Follow Pop III/II with a gradual
transition determined by metals
Include photoionization feedback
and minihalo screening effects
fcrit/fJeans
Redshift

20. New results from this more detailed calculation (Wyithe & Cen 2006)
Without Pop III massive stars: te < , with a rapidly increasing xHI
to >0.5 by z=8
With Pop III massive stars and reasonable star formation efficiency
and ionizing photon escape fraction: te = , with an extended
plateau of xHI = at z=7-12
With perhaps too generous assumptions about Pop III star formation
processes (very high escape fraction and/or very high star formation
efficiency), te = 0.21 max is possible.
Which one would I bet on? Physical sanity would eliminate the last choice.
Physical reasonableness for Pop III IMF would then argue for the second
choice. So te = seems most likely, same as I got 4 years ago.
Judgement day: Thursday, March 16, 2006 (3rd Yr WMAP results)

21. A 24-billion-particle radiation transfer simulation of detailed cosmological reionization process (Trac & Cen 2006)
Particle mass=2x106, Box size=100Mpc/h, timestep determined by c
Ncell=120003, Spatial resolution=8kpc comoving

22. Ok, bets placed, that is all fine! But,
How much do you REALLY know
about first galaxies?

24. A 21-cm probe of individual first galaxies
using CMB as the background radio source
with an antenna temperature of TCMB
dT = (Ts-TCMB)(1-e-t)
TCMB = 85K,

25. The structure of a first galaxy

26. Threshold by X-ray Background Heating
Number per cubic Mpc
Halo Mass

27. Brightness temperature decrement profile

28. Fundamental Applications with First Galaxies
Probe IMF, ns, mCDM , …
at n(gal)=1.e-6/Mpc3  Dns=0.01 (3s)
Determine Pk:
DV=100 Gpc3 within z=28-32
such as baryonic oscillations, etc., without messy
astrophysical biases
Alcock-Paczynski (AP) test:
assuming each measurement 20% error,
with 10,000 galaxies  Dw=0.012 (3s),
if WM=0.3 (no error) and k=0

29. Abundance of 21-cm absorption halos
Mean IGM temperature (K)
Square arcseconds

30. Typical low high-z galaxies
Theory: MHzG~ Msun
Observed LAEs at z>6: SFR>40Msun/yr (Hu et al 2003;
Kodaira et al 2003),
assuming c*=0.10, tsb=5x107yrs
---> MLAE (total) = 1x1011Msun
Thus, the current observations of z>6 LAEs do not probe the bulk
of first galaxies;
typical observed LAEs at z<6 have SFR~a few Msun/yr (Rhoads et al 2003; Taniguchi et al 2003)
Cen (2003c)

31. Quasar Stromgren spheres
Cen (2003c)
Rs= 4.3x-1/3(N/1.3x1057s-1)1/3(tQ/2x107yr)1/3[(1+zQ)/7.28]-1 Mpc
t(r)= 1.2x(WM/0.27)-1(Wb/0.047)[(Rs2-r2sin2q)1/2-r cosq]-1 ,
where Rs and r are in proper Mpc
Cen & Haiman (2000)

32. (1): probing ionization state of IGM and sizes of Stromgren spheres
Application of high-z galaxies inside quasar Stromgren spheres
(1): probing ionization state of IGM and sizes of Stromgren spheres
Evidently,
(i) x=0.1 and
x=0.01 differentiated
at >6s level
(ii) Rs
determined to
high accuracy; consequently,
tQ determined
accurately
Rs=3Mpc
Cen (2003)
x=0.01
0.1
1.0
Rs=5Mpc

33. Application of first galaxies inside quasar Stromgren spheres, cont.
(2) galaxy luminosity function and spatial
distributions at z>6
(3) probing environment around quasars
(4) probing anisotropy of quasar
emission …

34. Conclusions The universe has a long reionization
process (Wyithe & Cen 2006).
The star formation processes at high-z
may be quite different from those for
low-z and local star formation
3rd+…… year WMAP data should
give us a lot firmer information

35. A profitable way to detect high-z galaxies may be to target high-z observed luminous quasars, which provide a set of interesting applications
Radio observations of 21-cm line may provide a unique way to detect the very first galaxies at z=30-40, which could potentially provide a set of fundamental applications.