1. FIRST LIGHT IN THE UNIVERSE
Saas-Fee, April 2006
Richard Ellis, Caltech
Role of Observations in Cosmology & Galaxy Formation
Galaxies & the Hubble Sequence
Cosmic Star Formation Histories
Stellar Mass Assembly
Witnessing the End of Cosmic Reionization
Into the Dark Ages: Lyman Dropouts
Gravitational Lensing & Lyman Alpha Emitters
Cosmic Infrared Background
Future Observational Prospects

2. z > 6 Surveys Represent the Current Frontier
Motivation:
census of earliest galaxies (z=6, =0.95 Gyr) contribution of SF to cosmic reionization constraints on early mass assembly planning effective use of future facilities (ELTs, JWST)
Developing complementary optical/IR techniques:
- continuum dropouts: reducing contamination - HST ACS grism, SEDs with low background - Ly LF: sensitivity to physical state of IGM - strong lensing: extends capabilities at faint end of LF - Spitzer detections: masses & early SF histories

3. Some Key Issues
How effective are the various high z selection methods? L*(z=6)  i~26 where spectroscopy is hard spectroscopic samples biased to include strong L - great reliance on photometric redshifts
Is there a decline in the UV luminosity density 3<z<6? results are in some disagreement differing trends in continuum drops & L emitters
Is the observed UV at z>6 sufficient for reionization? - contribution from (unobserved) faint end of LF? - unusual popns: intense EW(L), steep UV continua?
Significant stellar masses for post-burst z~6 galaxies - how reliable are the stellar masses? - inconsistent with declining SF observed 6<z<10? does this imply an early intense period of activity?

4. Continuum sources probed via dropout technique
z-dropout
Stanway et al (2003)
Traditional dropout technique poorly-suited for z>6 galaxies:
- significant contamination (cool stars, z~2 passive galaxies)
- spectroscopic verification impractical below ~few L*
i-drop volumes: UDF ( ), GOODS-N/S (5.105), Subaru (106) Mpc3
flux limits: UDF z<28.5, GOODS z<25.6, Subaru z<25.4

5. Reducing Contamination from z~2 Passive Galaxies
Addition of a precise optical-infrared color (z - J) can, in addition to the (i - z) dropout cut, assist in rejecting z~2 passive galaxy contaminants.
(Stanway et al 2004)
(i – z)
5.7 < z < 6.5
z~2 passive galaxies
This contamination is ~10% at z~25.6 but is negligible at UDF limit (z~28.5)
(z – J)

6. Contamination by cool Galactic dwarfs - more worrisome
UDF z<25.6 (Stanway et al 2004)
L dwarfs
E/S0
HST half-light radius Rh more effective than broad-band colors
Contamination at bright end (z<25.6) is significant (30-40%)

7. Keck spectroscopy of i-drops: 10.5 hrs zAB < 25.6
z=5.83 L
L-dwarfs contaminate at bright end

8. Counts for i-band drops (GOODS+UDF)
 6 from z~3
Spec limit
GOODS/UDF data to zAB=28.5 consistent with z=3 LBG LF but  6
Bunker, Stanway, Ellis & McMahon MNRAS 355, 374 (2004)

9. ACS dropouts: Luminosity Dependent Evolution?
z=3
Bouwens et al (2006, astro-ph/ ) propose L-dependent evolution - decline in abundance over 3<z<6 mostly for luminous sources
If correct, this affects z-dependent integrated SF density measures corrected to some fiducial luminosity

10. How much steeper could LF slope be at early times?
i-drop counts
N(28.5) = 54
N(30.0) = 108
Yan & Windhorst (2004) extend UDF i-drop search to z(AB)=30.0 (c.f. Bunker et al z(AB)=28.5) and claim  -1.9 after correcting for incompleteness implying a significant increase in integrated density at z~6

11. Decline in UV over 3<z<6 has been controversial
Giavalisco et al 2004
Bunker et al 2004
Poisson errors fail to account for dispersion in claimed number of z~ i-drops, because of varying ways of accounting for contamination plus cosmic variance (10% in GOODS; 40% in UDF)
Bouwens et al 2005 Ap J 624, L5

12. Results from Subaru
HST offers superior photometry & resolution (important for stellar contamination) but SuPrimeCam has much bigger field (each pointing = 2  GOODS-N+S)
Additional photometric bands developed to sort stellar contamination
Shioya et al (2005): used intermediate band 709nm, 826nm to estimate stellar contamination in z~5 and z~6 samples respectively
Shimasaku et al (2005) split z-band into two intermediate filters zB, zR - to measure UV continuum slope
These studies confirm decline indicated via HST studies

13. z~6 dropouts from Subaru - the upshot
SDF dataset > 2  GOODS N+S; cosmic variance ~ 25%
Confirm 5 abundance drop from z~3 to 6 (c.f. Bunker et al, HST)
Luminosity dependent trends - more evolution in massive galaxies?
Remember: this is observed number not dust-corrected SFR

14. Further confirmation of i-drop technique
star
z~2 passive
Malhotra et al (2005) use 40 orbit ACS UDF grism exposures & consider utility of (i - z) color cut vs. contamination
Traditionally use (i - z) > 1.3 (e.g. UDF)
Of 29 z<27.5 candidates with (i - z) > 0.9, 23 are at z~6

15. Determining Abundance Necessary for Reionization
e.g. Madau, Haardt & Rees (1999):
Input details depend on considerable imponderables:
TIGM  10,000 – 20,000 K ?
Teff & Z of stellar population (IMF) ?
C = <HI2 > / <HI>2 simulations suggest C  30 ?
fesc (=1 implies no HI absorption) ?
Leading to factors of 10 uncertainty!

16. Cosmic Star Formation History
The combined uncertainties in the data (plus uncertain input physics necessary for the predictions) mean we cannot yet convincingly claim (or reject) that the abundance of z>5 sources is sufficient for reionization
Bouwens et al astro-ph/ )

17. The Spitzer Space Telescope Revolution
A modest 60cm cooled telescope can see the most distant known objects and provide crucial data on their assembled stellar masses!
IRAC camera has 4 channels at 3.6, 4.5, 5.8 and 8 m corresponding to 0.5-1m at z~7!
Egami et al (2005) - characterization of a lensed z~6.8 galaxy
Eyles et al (2005) - old stars at z~6
Yan et al (2005) - masses at z~5 and z~6
Mobasher et al (2005) - a galaxy > 1011 M at z~6?

18. Spitzer detections of i-drops at z=6
Eyles et al (2005) MNRAS 364, 443
4 i-drops in GOODS-S confirmed spectroscopically at Keck
Ly  emission consistent with SFR > 6 M yr-1
IRAC detections from GOODS Super-Deep Legacy Program

19. Spectral Energy Distributions of i-drops
#1 z=5.83
#3 z=5.78
VLT K
VLT K
Spitzer + Ly emission constrains present & past star formation
Ages > 100 Myr, probable Myr (7.5<zF<13.5)
Stellar masses: 2-4 1010 M (>20% L*)

20. Independent z~5-6 UDF Spitzer analysis
Yan et al Ap J 634, 109 (2005)

21. Evidence for Unusually Strong UV Continua
Several UDF objects at z~6 show unusually blue (z-J) colors which lie outside the predictions of standard Bruzual Charlot models; a point first noted by Stanway et al (2004) - unclear what this means!
Yan et al Ap J 634, 109 (2005)

22. Spitzer detection of resolved J-drop in UDF
Criterion: (J – H)AB > 1.3 plus no detection in combined ACS
JD2: strong K/3.6m break  potential high mass z~7 source
Mobasher et al (2005) Ap J 635, 832

23. High mass resolved J-drop in UDF + two breaks Mobasher et al (2005)
STARBURST99: z=6.6; EB-V =0.0; Z=0.02, zF>9
BC03: z=6.5; EB-V =0.0; Z=0.004, zF>9
Stellar Mass: 2-7  1011 M dependent on AGN contamination

24. Uncertainty in Redshift and Stellar Mass
~ 25% chance of being z~2.5

25. `Double-Break’ Objects Seem Fairly Common

26. Abundance of Massive Galaxies at z~6: A Crisis?
Abundance of massive galaxies at z~6 with CDM in terms of their implied halo masses, assuming
Scalo IMF
SF efficiency 20%
Find a 1013 M halo in the tiny UDF is a problem!
z = 5.8
Mobasher et al
z = 15
Yan et al
Eyles et al
Barkana & Loeb (2005)

27. Measured z~5-6 Mass Density
all
SFH only accounts for mass assembled in SF galaxies at z~5, so still a lower limit
Luminous SF galaxies
Stark, Bunker, Eyles, Ellis & Lacy (2006)

28. Summary of Lecture #6
Great progress using v,i,z,J-band drop outs to probe abundance of SF galaxies from 3<z<10: Bouwens et al discuss the properties of 506 I-band dropouts to z~29.5!
In practice, these samples are contaminated by foreground stars, z~2 galaxies etc to an extent which remains controversial. We are unlikely to resolve this definitively with spectroscopy until era of ELTs.
Comoving SF rate declines from z~3 to z~6 (and probably beyond) suggesting insufficient 6<z<10 luminous galaxies to reionize Universe
Contribution of lower luminosity systems less clear (lensing: Lecture 7)
Spitzer’s IRAC can detect large numbers of z~5-6 galaxies and it seems many have high masses (one spectacularly so!) and signatures of mature stellar populations - implies earlier activity
Reconciling mature galaxies at z~6 with little evidence for SF systems with 7<z<10 may turn out to be a very interesting result

30. Combining intermediate & broad-bands (Shioya et al)
Stars
Starburst
candidates
Passive
SDF example (for z~5 R-drops) shows how high z galaxies can be separated from interlopers by demanding strong intermediate band depressions (709-i), in addition to a regular (R-i) color cut
Method reduces number of `traditional’ R-drops from 3519 Similar technique using 826nm filter reduces i-drops from 96

31. Intermediate + broad-bands contd.. (Shimasaku et al)
stars
z~6 gals
z~2 passive
Breaking z-band into two components (zR, zB) can insist on blue continuum slope within z-band for z~6 galaxies.
Reduces conventional i-band drops z < 25.4 from 22  12

32. Low Abundance of z~6 Sources
No extinction
Necessary for reionization 6<z<9 (Stiavelli et al 2003)
GOODS+UDF
Bunker et al (2004)
Can rescue situation appealing to:
cosmic variance (unlikely given independent datasets)
steep faint end LF (Yan & Windhorst 2004)
low Z populations (Stiavelli et al 2004)

33. Ages and Stellar Masses
#1 z=5.83
#3 z=5.78
Explored range of BC models including dust:
Ages > 100 Myr, probable range Myr (7.5<zF<13.5)
Stellar masses: 2-4 1010 M (>20% L*)