1. Cosmological Reionization -- Allahabad, India
High-Redshift Galaxies and the Reionization of the Universe: Insight from Recent
WFC3/IR Observations
Rychard Bouwens
(UC Santa Cruz / Leiden)
Cosmological Reionization -- Allahabad, India
February 17, 2010

2. Shuttle Servicing Mission SM4
WFC3
UCSC 02/01/10 RJB

3. Special Thanks to My Collaborators
With a Special Thanks to:
Garth Illingworth, Marijn Franx, John Blakeslee, Holland Ford, Rodger Thompson, Louis E. Bergeron, Massimo Stiavelli, Dan Magee, Ivo Labbe, Pieter van Dokkum, Dan Coe, Larry Bradley, Valentino Gonzalez
ACS GTO team: Holland Ford, Garth Illingworth, Mark Clampin, George Hartig, Txitxo Benitez, John Blakeslee, Rychard Bouwens, Marijn Franx, Gerhardt Meurer, Marc Postman, Piero Rosati, Rick White, Brad Holden, Dan Magee + many other team members
UDF-IR team: Rodger Thompson, Garth Illingworth, Rychard Bouwens, Mark Dickinson, Pieter van Dokkum, Dan Eisenstein, Xiaohui Fan, Marijn Franx, Marcia Rieke, Adam Riess
HUDF09 WFC3 IR team: Garth Illingworth, Rychard Bouwens, Marijn Franx, Pieter van Dokkum, Massimo Stiavelli, Ivo Labbe, Michele Trenti, Marcella Carollo, Pascal Oesch, Dan Magee

4. Galaxy Formation & Evolution at z>6
Key Science Interests
1) Galaxies as possible reionization sources
-- This follows from evidence from z~6 SDSS quasars and 7-year WMAP optical depth measurements that the universe was likely reionized between z~6 and 11…
UCSC 02/01/10 RJB

5. Galaxy Formation & Evolution at z>6
Key Science Interests
1) Galaxies as possible reionization sources
-- This follows from evidence from z~6 SDSS quasars and 7-year WMAP optical depth measurements that the universe was likely reionized between z~6 and 11…
2) The luminosity and masses of galaxies at these epoches are likely to build up very rapidly to z~3.
3) Galaxies at these epochs are likely to show unique and very interesting stellar populations (new IMFs, zero metallicities,
and no dust)
UCSC 02/01/10 RJB

6. What are the implications for
What does the new WFC3/IR instrument tell us about galaxy evolution at z>6?
What are the implications for
reionization?

7. But first let me remind you about results from LF studies at z~4-6?

8. Faint-end Slope of the UV Luminosity Function
High Luminosity
Galaxies
Low Luminosity
Shallow
Bouwens et al. 2007
Reddy et al.
2008
Need the deepest data to do this well, so use fields like HUDF!
Faint-end
Slope
Steep
For such steep faint end slopes, the volume density of
lower luminosity galaxies is substantial:
50% of the UV luminosity density is below 0.06 L*
50%
Bouwens et al. 2007
(see also Beckwith et al. 2006
and Oesch et al. 2007)

9. Evolution of the UV Luminosity Function
Bright
Hierarchical Buildup
AGN Feedback?
Downsizing
M*UV
Faint
Redshift
Bouwens, Illingworth, Franx, & Ford 2007
UCSC 02/01/10 RJB

10. Let us look at quick demonstration of how well WFC3/IR works!
What does the new WFC3/IR instrument tell us about galaxy evolution at z>6?
Let us look at quick demonstration of how well WFC3/IR works!

11. Demonstrating Performance of WFC3/IR
NICMOS
72 orbits
WFC3/IR
16 orbits
Region of the HUDF

12. Let us talk about some of the early science results...

13. Exciting Ultradeep WFC3/IR Program
Deepest
optical data
HUDF09 WFC3/IR
program
CDF-South GOODS
HUDF09 WFC3 IR team: Garth Illingworth, Rychard Bouwens, Marijn Franx, Pieter van Dokkum, Massimo Stiavelli, Ivo Labbe, Michele Trenti, Marcella Carollo, Pascal Oesch, Dan Magee
Should find z>=7 galaxies
UCSC 02/01/10 RJB

14. August 26 - Sept 6, 2009

15. 16 z~7 z-dropouts from our 4.7 arcmin2 HUDF09 observations over the HUDF
(Oesch et al. 2009)
Excellent S/N
As an example, here are the z-dropout
candidates identified previously
(Bouwens et al. 2004, 2008; Oesch et al. 2009)
Oesch et al. (2010)
but see also McLure et al.; Bunker et al.; Yan et al; Finkelstein et al.

16. Are these sources really redshift z~7 galaxies?
The properties of these sources are:
1) Very strong z-Y break
2) Very blue Y-J colors redward of the break
3) Non-detection at optical wavelengths
Very unusual properties for sources found in the
real universe
Typical contaminants tend to be low-mass stars, but in a multi-color space (z-Y, Y-J, J-H), there is a nice separation between z~7 star-forming galaxies and low-mass stars.
Contamination from Photometric Scatter: Simulations give contamination rates <= 1%

17. UV Luminosity Functions
Now, use the z~7 search results to derive constraints on z~7 UV LF
UV Luminosity Functions
z~7
Old NICMOS z~7 LF (Bouwens et al. 2008;
Oesch et al. 2009)
z~7
z~4
Log # mag-1 Mpc-3
z~6
Bright
Faint
Oesch et al. 2010
UCSC 02/01/10 RJB

18. 5 z~8 Y-dropouts from our 4.7 arcmin2 HUDF09 observations over the HUDF
(Bouwens et al. 2009)
Are these sources really redshift z~8 galaxies?
Extended, so not low-mass stars...
Contamination from photometric scatter < 1%

19. UV Luminosity Functions
Now, use the z~8 search results to derive constraints on z~8 UV LF
UV Luminosity Functions
z~8
z~4
Log # mag-1 Mpc-3
z~7
Bright
Faint
Oesch et al. 2010; Bouwens et al. 2010
UCSC 02/01/10 RJB

20. Total WFC3/IR Data Set Deepest optical data WFC3/IR ERS HUDF09 WFC3/IR
program
New WFC3/IR ERS samples
~17 z~7 z-dropouts
~6 z~8 Y-dropouts
Second HUDF09 sample
CDF-South GOODS
~17 z~7 z-dropouts
~7 z~8 Y-dropouts
Total z>=7 WFC3/IR sample
~50 z~7 z-dropouts
~25 z~8 Y-dropouts
UCSC 02/01/10 RJB

21. UV LFs at z~7-8 from Wide area + Ultra-deep Observations with WFC3/IR
UV Luminosity Functions
Preliminary
Log # mag-1 Mpc-3
~50 z~7 galaxies
~25 z~8 galaxies
50 z~7 galaxies,
25 z~8 galaxies
Bright
Faint
Bouwens et al. 2010
UCSC 02/01/10 RJB

22. UV Luminosity Functions
Wide area + Ultra-deep Observations can be used to more accurately constrain the UV LF at z~7-8
UV Luminosity Functions
Preliminary
68% and 95% confidence intervals ϕ*
Faint
Bright
Bouwens et al. 2010
UCSC 02/01/10 RJB

23. Integrate the UV LFs at z~7 and z~8, one derives the
SFR density
Bouwens et al. (2010)
Star Formation
Rate density

24. ⦗ ???? What does this mean for reionization? Change in ionization
state
Recombination Rate
Ionization Rate
dQHII
dnion/dt
QHII = − ⦗ nH dt
trec
QSO + Galaxy + Annihilating DM(?)
~3%
unknown
????
trec = (0.6 Gyr) C3 ((1+z)/7)3

25. What fraction of universe do galaxies reionize?
Fraction of HI ionized
~100%
~54%
~28%
~14%
z~6
z~7
z~8
z~9
Clumping Factor ~3
(Bolton et al. 2005; Pawlik et al. 2009)
Escape Fraction ~20%
see also Oesch et al. 2009

26. What does such a LF evolution
mean for reionization?
Fraction of HI ionized
QHII
~100%
~54%
~28%
~14%
z~6
z~7
z~8
z~9
Thompson optical depth
τ ~ 0.05
vs. WMAP-7
τ ~ /
see also Oesch et al. 2009

27. Purported Upturn in SFR density
One other solution to reionization of the universe was proposed by Yan et al
based upon 15 purported z~8 galaxies and 20 z~9-10 candidates....
Yan et al. (2010)
Purported Upturn in SFR density

28. We decided to test this hypothesis:
However, most of the Yan et al z~9-10 J-dropout candidates are very close to extended foreground galaxies
We decided to test this hypothesis:
Plot distance of dropouts to
closest extended galaxy
Our z~7 sample
Distribution of distances to extended sources for blank area on image
e.g.,
Yan et al. (2009) z~9-10 Candidates
Clear association of Yan et al z~9-10 sample with foreground galaxies at
% confidence
Suggests their sample is not reliable and likely full of contaminants
Yan z~9-10 sample
Yan et al z~8 sample also shows this association with foreground galaxies at
99.2% confidence
Suggests their sample is not reliable and likely full of contaminants
But expect no association between z~9-10 J-dropouts and extended foreground galaxies

29. Where does this leave us with the SFR density?
Yan et al. (2010)
Bouwens et al. (2010)

30. IRAC What can we learn about reionization from Stellar Mass Estimates?
IRAC images
IRAC
rest-frame
optical at
z~7
Stellar
Mass
Valentino Gonzalez

31. Stellar Masses of z~7 Galaxies
z~7 SED Fit
z~2 SED Fit
(bad fit)
Flux
1030 ergs s-1
cm-1 Hz-1
Mass = 4.2 x 109 Msol
Age = 398 Myr
Gonzalez et al. 2009
UCSC RJB

32. Total WFC3/IR z~7-8 samples of use for stellar mass density estimates
WFC3/IR ERS
HUDF09 WFC3/IR
program
HUDF09 sample
~17 z~7 z-dropouts
~7 z~8 Y-dropouts
New WFC3/IR ERS samples
~17 z~7 z-dropouts
~6 z~8 Y-dropouts
CDF-South GOODS
UCSC 02/01/10 RJB

33. Stellar Mass Density Divide by cosmic time...
Implies higher value of SFR density...
and hence higher thompson optical depth, but
still does not match τ ~ / WMAP-7 value...
New WFC3/IR Results
Labbe et al. 09b
Gonzalez et al. 2009
UCSC RJB

34. What can learn from the UV colors of z>~7 galaxies?

35. fλ ~ λβ Galaxies at z>6: What type of colors do they
have in the UV-continuum?
What is the UV slope β?
UV-continuum slope β depends upon the age, metallicity, and dust content of a star-forming population
UV-continuum slope β most sensitive to changes in dust content
The power law slope of UV continuum:
fλ ~ λβ
UCSC RJB

36. not fit with standard stellar population models
z~7 galaxies from ultra-deep WFC3/IR observations of the HUDF:
What about their UV colors?
red
“more dusty”
Bouwens et al. 2009
UV slope
dust free
not fit with standard stellar population models
blue
“more dust-free”
bright
Luminosity
faint
Bouwens et al. 2010
UCSC RJB

37. not fit with standard stellar population models
What about their UV colors?
versus redshift
red
“more dusty”
Lower UV Luminosities (~0.1 L*)
UV slope
blue
dust free
not fit with standard stellar population models
“more dust-free”
Redshift
Bouwens et al. 2010
UCSC RJB

38. What might a UV slope β ~ −3 imply?
--> (Lyman-Continuum Escape Fraction May be > 30%...)
Nebular Continuum Emission Must Not Be That Important
To produce very blue β’s (i.e., −3) we require hot, young stars...
Very Hot Stars β ~ −3 (blue)
Emitted Light
1. Lyman-Continuum Escape Fraction Large (>~ 30%)
β ~ −3
Ionized Gas (from hot stars) β ~ −1.7 (red)
Total (Hot Stars + Ionized Gas) β ~ −2.3
→ Nebular Continuum Emission Keeps Us from producing very blue β’s
Possible Solutions?
2. Models overpredict nebular continuum emission (?)
Bouwens et al. 2010
Aspen 2010 RJB

39. What can we learn about galaxy formation and evolution from observations of very high redshift galaxies
WFC3/IR allows us to very efficiently identify galaxies at high redshift. We identify 50 z~7 z-dropouts and ~25 z~8 Y-dropout galaxies in the new observations
The UV LF at z~7-8 is quite consistent with the extrapolation from lower redshift
SFR density continues to decrease towards higher redshift to our search limits
Implies steady decrease in ionized fraction QHII towards high redshift, i.e.,
~56% at z~7, ~28% at z~8, 14% at z~9, ... with tau ~ 0.05
The UV-continuum slopes beta we measure at z~7 are very blue, particularly towards very low luminosities and may suggest modest Lyman-Continuum Escape Fractions, i.e., > 30%.
We can use the deep IRAC data -- in combination with the near-IR data -- to do stellar population modelling for z~7 and z~8 sources and estimate stellar masses and hence SFR densities.
These estimated SFR densities also allow us to estimate the number of reionizing photons at high-redshift... Again there is a tension with the WMAP-7 measurement.
UCSC RJB