Galaxies as Sources of Reionization

Galaxies as Sources of Reionization

Haojing Yan (Carnegie Observatories)

Reionization Workshop at KIAAJuly 10, 2008

Haojing Yan (Carnegie Observatories)

Reionization Workshop at KIAAJuly 10, 2008

Outline• Luminosity Function of Galaxies at z 6

— UV LF has a very steep faint-end slope

• Stellar Masses of Galaxies at z 6

— some high-mass, “old” galaxies already in place; but they are not likely the dominant reionzation sources.

• Implications for (HI) Reionization

— dwarf galaxies did it!

• An Unanswered Question at z 6 — evolution of LF at the bright-end?

Part I

LF of Galaxies at z 6 (5.5 z 6.5)

Source(s) of ReionizationYan & Windhorst 2004, ApJ, 600, L1

Critical value fromMadau, Haardt & Rees 1999

Contribution from reionizing sources

• Galaxies can account for the necessary reionizing photons, if the LF Galaxies can account for the necessary reionizing photons, if the LF has a steep faint-end slope; dwarf galaxies are important contributors.has a steep faint-end slope; dwarf galaxies are important contributors.

To z<30 mag, 108 i-dropouts found in the HUDF

(Yan & Windhorst 2004, ApJ, 612, L93; YW04)

Note: ~ 1.5 mag deeper than Bunker et al. (2004; MNRAS, 355, 374)

• By pushing to the very limit of the HUDF, we start to be able to address the LF faint-end slope at z~6.

i’ z’

z’=29.23

z’=29.97

Detection Reliability at z>28.5 mag Level

z=5.83; Dickinson et al. (2004)

z=5.9; Malhotra et al. (2005)

ACS Grism Observations of HUDF (GRAPES; Malhotra et al. 2005)

z=6.0

z=6.1

z=6.4

• GRAPES: i-dropouts success rate of ~ 90% in the HUDF to z~27.5 mag

•Our HUDF z 6 candidate sample supports a very steep UV LF faint-end slope:

α = -1.8 to -1.9

• Dwarf galaxies can provide sufficient (re)ionizing photons at z 6

YW04 Constrain to the UV LF at z 6

Recent Result Confirms the Steep Faint-end Slope (Bouwens et al.

2006)

506 i-drops: UDF, UDF-Pars, GOODS But compare to YW04: M* = -21.03, * = 4.6x10-

4/Mpc3

4.6x10-3

Msun/yr/Mpc3

1.1x10-2

Msun/yr/Mpc3

SFR is still uncertain by 2x

“Lilly-Madau Diagram”

Luminosity Function of z 6 LAE• LAE : ~ 1/4 of the entire galaxy population (based on

results at z~3), but still very important — easier to identify; current redshift record holder is the LAE at z=6.96 (Iye et al. 2006)

• LAE as probe of the reionization epoch : neutral IGM — Lya line suppressed — LAE number drop (e.g., Marilada-Escude 1998; Malhotra & Rhoads 2001)

• LAE at z 6 are usually selected at two narrow windows at z=5.7 & 6.5 in order to avoid strong night-sky lines

Evolution of LAE LF from z=5.7 t0 6.5

• Malhotra & Rhoads (2004): no evolution seen; IGM ionized up to z=6.5

• Haiman & Cen (2005): not necessarily; local HII bubble permits escape of Lya photons and the suppression is not as large; <XHI> up to 25%

Better Statistics from Subaru Deep Field

Shimasaku et al. (2006) Kashikawa et al. (2006)

• Kashikawa et al. (2006): strong evolution from z=5.7 to z=6.5 !

• Significant fraction of HI at z=6.5 ?? WMAP zreion ~ 11.4?

Part II

Stellar Masses of Galaxies at z 6

Stellar Mass Assembly History in Early Universe

• Stellar mass density & SFR density: = ∫SFR dt

• Need measurements at rest-frame optical (and beyond) to reduce biases caused by dust extinction and short-lived stars when converting light to mass

• Study at high-z made possible by Spitzer IRAC

• GOODS Spitzer Legacy Program has played a critical role

3.6μm 4.5μm

5.6μm 8.0μm

z =5.83 galaxy

IRAC Sees z ~ 6 Galaxies in HUDF

z=5.83

z=5.9

zp~5.9

Three i-drops in HUDF securely detected by IRAC

Yan et al. 2005, ApJ, 634, 109

• Some high-mass (a few x 1010Msun) galaxies were already in place by z6 (age of Universe < 1.0 Gyr)

• A few hundred Myr old (formed at z>>6)

• Number density consistent with CDM simulation from Nagamine et al. (2004)

Some Major Conclusions from SED Fitting

See also Eyles et al. (2005)

CDFS, 3.6μm HDFN, 3.6μm

Extending to Entire GOODS(Yan et al. 2006, ApJ, 651, 24)

IRAC-detected i-dropouts

CDFS, 3.6μm HDFN, 3.6μm

IRAC-invisible i-dropouts

Difficulty: no photometric info between z’ and IRAC 3.6μm

Have to take a different, simplified approach

(z’-3.6μm) color age for a given SFH M/L for a given SFH at this age stellar mass; repeat for all SFH in the set, and take min, max, median

Stellar Mass Estimates Summarized

• IRAC-detected Sample

Mrep: 0.09 ~ 7.0x1010Msun (median 9.5x109Msun)

Trep: 50 ~ 400 Myr (median 290 Myr)

• IRAC-invisible Sample, using 3.6m upper limit

Upper-limit of Mmax (median 4.9x109Msun)

IRAC-invisible sample stack Random stack

3.6μm

3.6μm mag = 27.44median z’ mag = 27.00

Mmin = 1.5x108

Mrep = 2.0x108 Msun

Mmax = 5.9x109

Stacking of IRAC-invisible i-dropouts

Models courtesy of K. Nagamine;based on simulationsof Nagamine et al. (2004) and Night et al. (2006)

Implications (I): compare to simulation•ΛCDM models seem to be capable of

producing such high-mass galaxies by z 6

Implications (II): Global Stellar Mass Density•Lower limit at z ~ 6: (1.0, 1.6, 6.5) x 106MsunMpc-3

Implications (III): Source of Reionization• Critical SFRD based on

Madau et al. (1999)

• Progenitors of all IRAC-detected z6 galaxies formed simultaneously with the same e-SFH: SFR e-t/

• The progenitors of high-mass galaxies alone CANNOT provide sufficient ionizing photons to sustain the reionization

• Dwarf (low-mass, low-luminosity) galaxies, which could be more numerous, must have played an important role

Part III

Bright-end of LF at z 6

L* & Bright-end of LBG LF

• Bouwens et al. (2006): L*(z=6) = 0.6L*(z=3)

•Effect of large-scale structure ( “cosmic variance”)??

Need Degree-sized Surveys to Minimize Impact of “Cosmic

Variance” at Bright-end

(Millennium Simulation slice at z=5.7)

D1(2h-4d)(overlap SWIRE)

D2 (10h+2d)(w/COSMOS)

D3 D4

16.5’x10’GOODS-Size Area

Bright i-drops in 4-deg2 CFHTLS

Yan et al. (in prep)

Magellan High-z LAE Survey

Yan, McCarthy & Windhorst

Survey Highlights

•Narrow-band imaging in 917nm & 971nm OH-free windows to search for LAE at z ≈ 6.5 & 7.0

•Four IMACS f/2 fields (~ 0.9 deg2); reducing cosmic variance with limited telescope time

•Survey depth (5-) AB=25.0 mag (2.4510-17 erg/s/cm2 for pure-line sources; 7-810-18 erg/s/cm2 for continuum-detected sources)

•Aiming at bright-end of the luminosity function

6.46 — 6.62

6.91 — 7.07

~ 400 Mpc3/arcmin2

(Before upgrading, SITe CCDs)

o(917nm) p(971nm)

Survey Design: Filters

Survey Design: Fields

•Use fields that have public, deep continuum images in multi-bands (especially in z’-band)

•Accessibility from Las Campanas

•CFHTLS Deep D1, D2 & D4 spreading out in RA

Survey Status

• 1-night in Feb. 2007 + 2-night in Mar. 2008, 1 IMACS pointing in COSMOS field (CFHTLS-D2), 20hr in o(917nm)

• 3-night in Jul. 2007, 1 IMACS pointing in CFHTLS-D4, 20 hr in o(917nm)

• Achieved desired depth

COSMOSCFHTLS-D4

1.48o

1.48o

1o

1o

5- source counts

CFHTLSD4NW, 20hr in o

LAE Candidate Selection•Continuum images

from the T0003 release of CFHTLS-D4

•z’-o>0.44 (flin/fcon>1.5) i’-z’>1.3 if detected in z’ non-detection in u’,g’ and r’

•For now only discussing candidates invisible in z’

3 candidates invisible in continuum

o=23.88

o=24.39

o=25.49?

(Now seeking time do spectroscopic identification)

Kashikawaet al. 2006(in SubaruDeep Field)

Rapid Evolution from z=5.7 to 6.6 or not?

Summary• UV Luminosity Function of Galaxies at z 6

— a very steep faint-end slope (lots of dwarf galaxies …)

• Stellar Masses of Galaxies at z 6 — some high-mass, “old” galaxies in place; but not enough

• Implications for (HI) Reionization

— dwarf galaxies did it!

• Unanswered questions at z 6: Bright-end of LF (LBG/LAE) should tell a lot

— degree-sized surveys needed to reduce “cosmic variance”