Download - ppt
FIRST LIGHT IN THE UNIVERSEFIRST LIGHT IN THE UNIVERSE
Richard Ellis, Caltech
1. Role of Observations in Cosmology & Galaxy Formation
2. Galaxies & the Hubble Sequence
3. Cosmic Star Formation Histories
4. Stellar Mass Assembly
5. Witnessing the End of Cosmic Reionization
6. Into the Dark Ages: Lyman Drop Outs
7. Gravitational Lensing & Lyman Alpha Emitters
8. Cosmic Infrared Background
9. Future Observational Prospects
Saas-Fee, April 2006
Witnessing the End of Cosmic ReionizationWitnessing the End of Cosmic Reionization
Four important constraints put bounds on the epoch of cosmic reionization and the sources responsible:
• Evolution in the optical depth of Ly absorption in high redshift QSOs (Fan et al Ap J in press, astro-ph/0512082)
• Ubiquity of metals in the intergalactic medium (Songaila 2004)
• Large angular scale power in the temperature-polarization cross-correlation function in the CMB (Kogut et al 2003, Spergel et al 2006)
• Assembled mass density at z~5-6 from HST/Spitzer probes of faint galaxies (Eyles et al 2005, Yan et al 2005, Stark & Ellis 2006)
These motivate us to search the era z > 5 for star forming sources
Gunn-Peterson TestGunn-Peterson Test
GP = -ln T
Do complete troughs mean reionization just ended at z=6.2?
Or is it just natural thickening of the forest as we move to higher z?
Clue: as we approach end of reionization we expect abrupt change in optical depth GP with z
11 SDSS QSOs
Fan et al (2003)
SDSS QSOs (2003)SDSS QSOs (2003)Cosmic variance along available sightlines
T = f()/fcont
SDSS QSOs (2006)SDSS QSOs (2006)
Fan et al (astro-ph/0512082)
SDSS i-z colors in 6600 deg2 used to locate 19 QSOs with 5.74<z<6.42
Combined analysis of Ly,, opacity used to verify GP(z):
Wavelength ranges:
Ly: 1040 < (Å) < (1216) Ly: 970 < (Å) < (1040) Ly: 949 < (Å) < (970)
( ) - region affected by QSO ionization (proximity effect) excluded
E.g. using Ly alpha & Ly beta together
Redshift
Ly
Ly
Ly
Ly
zQSO
zQSO
cosmic variance
How GP effect works
• So even tiny neutral fraction xHI~10-4 gives a complete GP trough
• For reference xHI~10-5 at z~0
• But since GP f , for same nH, (Ly), (Ly) are <6.2,<17.9 smaller
• In practice, conversion from nHI depends on IGM clumpiness
• So absolute comparison of higher order Lyman lines more complicated than identifying relative trends in each
neutral fraction xHI
Comparing Comparing (Ly(Ly) and ) and (Ly(Ly))
Ly Ly
Empirical argument; discontinuity seen in both samples:
For z<5.5 () = 0.85[(1+z)/5]4.3 ; () = 0.38[(1+z)/5]4.3
Data for z > 5.5 is inconsistent: (1+z)10.9+
Dispersion increases likewise: () from 0.30.6
This is the fundamental justification for the end of reionization
Combined Trend in Combined Trend in GPGP(z)(z)
Combining the higher order Lyman lines on an absolute scale is hard
Is there really a break at z ~5.5?
Raw transmission
Summary of the G-P Issues
Optical depth is not very sensitive to ionized fraction
Key issue is rate of change in (z): this distinguishes reionization (overlapping HII regions) from thickening in forest: do we see a break in trends as we approach z>6
Geometry important: knowing UV background can we directly infer sizes of Stromgren spheres at various z?
Line of sight fluctuations and statistics of `dark gaps’ contains information on topology of late stages
Are some lines of sight at z>6 still consistent with large neutral fraction?
Scale of Ionized RegionsScale of Ionized Regions
Mean free path
At z > 5.7 mean free path of ionizing photons approximates clustering scale of SF galaxies, suggesting they are source of UV photons
Declining extent of QSO’s proximity effect: R [(1+z) xHI]-1/3 suggesting higher z QSOs lie in a IGM whose xHI is 14 higher
Extent of proximity effect
Distribution of Dark GapsDistribution of Dark Gaps
Gap or `void’ statistics contain useful information on topology
Define `gap’ as contiguous region where > MIN (e.g. > 3.5)
Regions of high transmission are effectively associated with large HII regions thus their distribution acts like an indirect Ly LF:
probe of neutrality independent of Gunn Peterson
Ly Ly
Metallicity of the Intergalactic MediumMetallicity of the Intergalactic Medium
Keck II + ESI 12 hrs
CIV forest is already present at z=4.5
Songaila & Cowie (2002, 2003)
Ubiquity of CIV in many sight-lines to z~5 indicates earlier SF
Carbon also present in LyCarbon also present in Ly forest forest
CIV was detected in the Ly forest in 1995 with N(CIV)/N(HI)~10-2 -10 -3
CIV is now seen in even the weakest Ly systems (Ellison et al 2000)
How did it get there? These are low column density HI systems
Suggestive of ubiquitous enrichment from early times
Ly forest
log f(N) z-1
log N(CIV)
Only Modest Evolution in (CIV)
CIV
SiIV
Songaila astro-ph/0507649
First Year of WMAP DataFirst Year of WMAP Data
• CMB linear polarization powerful probe of e- scattering at recombination (1 deg) and reionization (>5 deg)
• TE cross power correlation function implies an electron scattering optical depth e = 0.17 0.04 (Kogut et al 2003)
If reionization was instantaneous: z = 17 5 (Bennett et al 2003)
Putting the TE signal in context…Putting the TE signal in context…
More realistic reionization history (an example)More realistic reionization history (an example)
• Adopt WMAP cosmology and P-S formalism
• Assume fraction f of baryons collapse into bound objects and all energy IGM
• Solve for cooling, enrichment and ionization
• Complications (feedback, inhomogeneities) serve only to delay reionization
• Find z~10-12 since stars before reionization must contribute to
Fukugita & Kawasaki (2003)
WMAP 3 year dataWMAP 3 year data
Assume power law CDM
= 0.17±0.08 = 0.09±0.03
Spergel et al (2006): instantaneous reionization at z=113 (2)
Recall: Balancing the Stellar Budget at z~0Recall: Balancing the Stellar Budget at z~0
GALEX, SDSS UV
Spitzer FIR
ACS dropouts
Cole et al 2dF
Integrating the SF history reproduces the stellar mass density at z~0
Star formation history Mass assembly history
Integrated stellar mass density at z~5-6:
Now turn the argument around:
Instead of checking that mass is consistent with past SF we can ask:
What does accumulated stellar mass imply for earlier SF & reionization?
If mass density at z~5-6 is greater than can be accounted for by previous SF history, options include:
• extinction: star formed but dust is present
• intrinsically faint contributors: lensing may find them
• upturn in SFH before z~10
Assembled Stellar Mass at z~5-6Assembled Stellar Mass at z~5-6
Stark & Ellis New Astr. Rev. 50, 46 (2006)
€
M*(z) = ρ*(z)dV (z)z= 5
z=10
∫
Estimates of the Stellar Mass Density at z~5-6
Stark, Bunker, Eyles, Ellis & Lacy 2006
GOODS v-drops and i-drops
Cosmic age: ~ 1.2 Gyr (v) + 1.0 Gyr (i); z* AB ~24.5 (v) c.f. 25.0 (i)
Spectroscopic confirmation + continuing Keck campaign (N>40)
IRAC detection for stellar masses: key issue is the uncertainty in mass
z=5.554, ~1.1 1011 M
z=4.831, ~1.6 1011 M
Parameterized Star Formation History
Based on recent compilations by Bouwens, Bunker et al
>0.1L*z=3
entire LF; =-1.7
Measured z~5-6 Mass Density
Stark, Bunker, Eyles, Ellis & Lacy (2006)
SFH only accounts for mass assembled in SF galaxies at z~5, so still a lower limit Luminous
SF galaxies
all
Summary of Lecture #5Summary of Lecture #5
• We have discussed 4 completely independent and indirect probes of cosmic reionization each of which suggests activity - probably involving stars - in the interval 6<z<12
• The HI troughs in SDSS QSOs - the arguments are compelling but rely on a subtle change in properties below and above z~5.5
• Ubiquity of metals in deep intergalactic space - they can only have got there from an early period of SF z>5
• WMAP polarization from electron scattering - not as precise a pointer to the redshift of activity as reported
• The remarkable amount of assembled mass at z~5 - perhaps the best pointer to much SF z>5
HII Regions and the IGM
Strömgren sphere around HII region has radius R(z, SFR, fHI)
Optical depth in the damping wing distance r from ionizing source
To preserve a dark gap, require d > 3.5. In a neutral medium at z~6 this indicates RS~0.34 Mpc and a source with SFR~70 M yr-1, corresponding to L* galaxy. Such sources are rare & inconsistent with one every 100 Mpc. The solution can only be found if
• more abundant, lower L galaxies contribute whose average separation is smaller
• or fHI < 1
Masses for Spectroscopically-Confirmed z~5 Galaxies
In addition to a lower limit for the mass density for 15/30 spectroscopically confirmed sources with clean IRAC detections, we also can deduce (with less confidence) the likely mass contribution of the 107 sources with IRAC detections. Assuming the `clean fields’ are not biased, this can provide an upper limit to the mass density at z~5
CIV pixel optical depth method (POD)CIV pixel optical depth method (POD)
Songaila (2005) AJ 130, 1996
Real data Fake pairs
Inferred Photoionization Rate
For isothermal IGM, log-normal density distribution find GP -0.4
Since b2, dispersion in likely reflects that in UV background
High z data sample over z=0.15 44 comoving Mpc
4 variation in UV b/g on 44 Mpc scales
Studies of DLAs very important:
- dominate IGM HI
- easiest to get precision metal abundances
- best estimate of metallicity in IGM at high z
Damped Lyman Alpha AbsorbersDamped Lyman Alpha Absorbers
Range of H absorbers
Pettini (2003, astro-ph/0303272)
Evolution of Neutral Hydrogen in IGMEvolution of Neutral Hydrogen in IGM
Dominated by Damped
Ly systems NHI> 2 1020 cm-1
Chemical Evolution in DLAsChemical Evolution in DLAs
Missing Metals?Missing Metals?
DLAs have reasonable metallicity but low (HI)
Ly forest has low metallicity but dominates (HI)
Predicted metallicity from previous SF is ~ 1/30th
So where are remaining metals?