Did Galaxies Reionize the Universe? Richard Ellis, Caltech CIFAR February 2010

Cosmic Reionization The third major event in cosmic history! When did it happen? Were star-forming galaxies responsible? What feedback processes were involved & how did these affect subsequent assembly? Major progress in 2009 in using galaxies as tracers of reionization With Dan Stark (IoA), Brant Robertson (CIT), Masami Ouchi (OCIW) Andy Bunker (Oxford), Ross McLure, Jim Dunlop (Edinburgh),

Lyman break galaxies: Rest-frame UV continuum discontinuity High Redshift Star Forming Galaxies Lyman alpha emitters: Located via narrow band imaging

When? WMAP Polarization Data rejects instantaneous reionization at z~6-7 Process is likely extended over 6<z<20 CMB studies do not pinpoint the responsible cosmic sources  = 0.17 ± 0.08 (WMAP1, 2003)  = 0.09 ± 0.03 (WMAP3, 2007)  = 0.087± (WMAP5, 2009)  = 0.088± (WMAP7, 2010) WMAP5 Instantaneous reionization Dunkley et al 2009 Jarosik et al 2010

HST IRAC VLT Balmer break Eyles et al (2005) z spec =5.83 M = M  Age > 100 Myr SF since z~7-10 Stellar Mass Implies Much Activity z > 7 Egami + (2005) z=6.8 M ~ M  Age 100 – 450 Myr SF since z~ Stellar mass density at z~5-6 implies past SF in low luminosity galaxies may be sufficient for reionization, especially if escape fraction is high

Early SF: Rapid drop in  CIV from z=4.5 to z=6  3.5 drop in  CIV (4  ) over 300 Myr (4.7 < z < 5.8) Suggests rapid enrichment since z~9 (c.f. Oppenheimer et al vzw model) Caveats: ionization changes, blending (c.f Becker et al), cosmic variance If absorbers representative: Z IGM (z~6) ~ Z  (depends on ioniz n ) Puzzle: implies too few escaping photons (6<z<9) to keep IGM ionized Ryan-Weber et al MN 395, 1476 (2009)

Probes of Reionization: - Lyman  emitters Strong emitter Faint emitter Ly  damping wing is absorbed by HI and thus valuable tracer of its presence. In weaker systems, it may be a sensitive probe of reionisation Complicated by dust? (Dayal et al ) Santos MN 349, 1137 (2004) McQuinn et al MN 381, 75 (2007)

Ouchi et al (2010) 1 deg 2 SXDS with 608 photometric & 121 spectroscopic LAEs Contrast with LBGs: no evolution 3<z<5.7 Tantalizing fading (0. m 3) seen in the LF of Ly  emitters over a small redshift interval 5.7< z< 6.6 (150 Myr) Does this mark the end of reionization: an increase in x HI (e.g. x HI ~0.6 at z~7)? Without dust constraints hard to be sure (Dayal et al 2009) z=6.5 z=5.7 SXDS ~1.0 deg 2 includes cosmic variance errors ~30% A Rapid Drop in Lyα Emitters from 5.7<z<6.6?

Keck Spectroscopic Survey of 4 < z < 7 LBGs Target B, V, I, z drops in GOODS/UDF from Stark et al (09) ACS/IRAC catalog 8-16 hr exposures with DEIMOS to m AB =26.5 (emission lines to m AB ~27.5) Keck/DEIMOS: 361 B drops, 141 V drops, 45 i drops, 17 z-drops = 564 targets VLT/FORS2 retro-selected + same criteria: 195 targets (Vanzella et al) B V i’ M UV Probing sub-L* B,V drops VLT Keck Stark et al (2010) MN submitted

Ly  fraction vs UV luminosity and extinction Strong correlation between extinction- inferred from UV slope β (flux ~ λ β ) and presence/absence of Lyα Strongly suggests low luminosity galaxies are relatively dust-free So their Lyα fraction might be valuable probe of reionization W λ >50Å X(Lyα) β ΔβΔβ Stark et al (2010) MN submitted LBGs with Lyα more dust

Lyα Emission Statistics versus Redshift Keck LBG Spectroscopic Survey z >7 LBG spectroscopic limits Keck survey avoids confusion with dust as can simultaneously measure β Do not confirm 5.7<z<6.6 drop seen in the Subaru LAE LF But rising fraction with z raises issue: why no z > 7 candidates with Lyα?!! Stark et al (2010) MN submitted redshift X(Lyα)

Hubble WFC3 High z Stampede WFC3/IR: nm 2.1  2.3 arcmin field of view 0.13 arcsec pixel times survey power of NIC3 UDF 4.7 arcmin 2 60 orbits in YJH Reaches m AB ~29 (5  ) The unruly mob! Bouwens et al Oesch et al Bunker et al McLure et al Bouwens et al Yan et al Labbé et al Bunker et al Labbé et al Finkelstein et al

z >7 candidates from WFC3 UDF campaign McLure et al (2009) 15 z > 7 candidates z’ Y J H SED  2 (z) 3 IR filters c.f. 2 leads to more secure photometric redshifts and reliable UV continuum slopes

But beware..uncertain redshifts still an issue.. zY J H

Declining Density of Star Formation 3 < z < 8 Bouwens et al astro-ph/

WFC3 Progress – I: z~7 Luminosity Function Oesch et al, Bunker et al 2009 Ouchi 09 (Subaru) WFC3 UDF α = ± 0.33 (Oesch) α = ± 0.65 (Ouchi) NIC UDF - ~25 z-band dropouts to Y AB ~28.5 corresponding to 6.5<z<7.5 - Steep faint end slope: low star formers ~1 M  yr -1 dominant

WFC3 Progress – II: Strong UV Continua? WFC3 data provides Y+J+H data and first reasonable estimate of the slope  of the stellar continuum where f( )   : remarkably steep values   -3 ! Strong trend of increasingly steep UV continua for high z, low L sources Bouwens et al 2009 Bunker et al 2009

What Might This Mean? (..if correct..) - Can reproduce  > -2.5 with dust-free young stars with Z~0.1Z  - For  ~ -3 need very low metallicities, extremely young bursts or top-heavy IMF with implied high escape fraction f esc > 0.2 Schaerer 2003; Bouwens et al ; Finkelstein et al

Upshot: Did galaxies reionize the Universe? Emission rate of ionizing photons Mpc -3 vs C HII and f esc compared with abundance of HST star-forming galaxies Excellent prospects for improving statistical test using ongoing HST programs Ouchi et al

Did Galaxies Reionize the Universe? Probably

Testing the High z Stellar IMF? Integral of star formation history Observed stellar masses Various workers have proposed top-heavy IMF to explain: - intense SF in high z galaxies (Baugh et al 2005) - mismatch between integral of SF and assembled stellar mass (Wilkins et al 2008) Ivan Baldry

SNLS light curve g max = 25.7 r max = 25.2 i,max = 25.1 Determining Rate of SNIIn M  ) SN LBG UV luminosity density of searched LBGs c.f. ♯ of SNIIn – tests IMF slope Cooke, RSE et al Keck LRIS spectrum -- Lyα at z=2.32 z SN SN SN SN SN SN *