Stellar Populations of Galaxies at 6.3 < z < 8.6Mauro GiavaliscoPSU June 6-10, 2010Collaborators:Steve FinkelsteinEros VanzellaAdriano FontanaCasey PapovichNaveen ReddyHarry FergusonAnton KoekemoerMark Dickinson & GOODS teamUMass Amherst

WFC3 Sample SelectionUsing the WFC3 HUDF data, detected 3000 objects in a J+H-band detection image.Computed photometric redshifts of all objects detected at 3.5 ( 2500) in both J125 and H160 using EAZY (Brammer+08).Sample consists of 31 galaxies with 6.3 < zphot < 8.6.23 at 6.3 < zphot 7.5, and 8 and 7.5 < zphot < 8.6.

VLT/Hawk-I Sample SelectionDeep UBVRIZYJK imaging with VTL/FORS and VLT/Hawk-I in 3 fields (including GOODS-S).Total of 15 z-band dropouts at 6.5

Comparison to LBG CriteriaCircles: Gray = z < 6.3, Cyan = 6.3 < z < 7.5, Red = 7.5 < z < 8.6.Brown symbols represent synthesized colors of brown dwarfs, from empirical spectra.Using stars in the WFC3 image, we measured FWHMPSF = 0.18All of our 31 objects are resolved.

Goals

Study their colors, compare UV spectra to those at z~3-5

Characterize the evolution

Can we rule out the null hypothesis that these objects are no different than galaxies at z 3-6?

If so, can we say anything about more exotic stellar populations?

UV (SFR), Optical (stellar mass) LF

Examine the remaining stellar population properties?

What can we confidently say, and what is guesswork?

Stellar PopulationsWe investigated the stellar populations in these galaxies by comparing our observations to CB07 models.We fit data from ACS, WFC3 and Spitzer, assuming that z = zphot.Included Ly emission in the models (see Finkelstein+07,08,09), as it can significantly affect the Y105/J125 fluxes at these redshifts, as the line EW increases as (1+z).Results imply that current NIR NB surveys are not deep enough to see majority of objects.We computed 68% confidence ranges on each fitted parameter simulations.We included the uncertainty on the photometric redshift in these simulations - increases final uncertainties significantly!Age and dust are not well constrained with WFC3 fluxes alone - average of 200 Myr and AV = 0.3 mag - though majority are consistent with very young ages and low/zero extinction.While metallicity is traditionally poorly constrained, the blue colors of these objects rule out Z > Z at 95% confidence, and Z > 0.05 Z at 68% confidence.

UV Colors: z~7 +0.20SED fit to the HawK-I stack: E(B-V) = 0.05 -0.15

Rest-UV ColorsHawk-I: 16 galaxies

UV colors at z~8Galaxies seem to have redder UV color at z~8These probably reflects true scatter of UV SEDLya in H, which would make the colors bluer.But as we will see later, these galaxies do not seem to have much LyaFinkelstein et al. 2010

Constraints on Metallcityfrom SED fittingMetallicity is probably lower than that of z~3 LBG

No evidence for zero metallicity

LBG at z~4, 5 same UV spectral types as at z~3Vanzella, MG, et al. 2009

Absorbers more dust-obscured than emitters; effect of the outflows (Shapley et al. 2003)Vanzella et al. 2009

At z~6, only one type identified, but huge selection effects against other typesAbsorption lines weaker in composite spectrum, most likely diluted because of wavelength shifts due to winds (spectra registered to Lya)We do not know how to stack absorbers[see Vanzella, MG, et al. 2009]

Lya up to z~6.5 (GOODS-S)Vanzella, MG et al. 2009EmittersAbsorbers

Lya observed up to z~6.5in i-dropoutsThe highest-z GOODS i-band dropout (GOODS-N);Also observed by PEARS

Lya stronger at fainter UV luminosity and higher redshiftAmong LBG with Lya emission, there is a trend of increasing Ew with both increasing redshift and decreasing LUV(Vanzella, MG, 2009; see also Stark+10)

We obtained very deep Lya spectroscopy for 7 of our z~7 galaxies (z-band dropouts), with no detectionsAll we found is one possible detection at z=6.972All the other galaxies have no detection down to ~1x1018 erg/s/cm2 (3s)Interesting! Confirmed i-band dropouts have strong Lya emissionFontana et al. 2010

Fontana, Vanzella+10Spectroscopy of z>6.5 candidates20hr of FORS2 on Hawk-I + WFC3 candidates on GOODS-S: only one likely detection (3s)systematic follow-up of z>7 candidates with 8m telescope expensive --> LF untested until JWST/TMT/ELT

Monte Carlo simulations of expected detection rate suggest that the likelihood of such negative result is low ( 16 F = 3.0 +/- 0.3 10-18 erg s-1cm-2 Stark+10

Expected detection rateProbability to detect Lya emission (10s) in N galaxies in our observations

Fontana et al. 2010Lya distrib. consistent with Stanway+07, Vanzella+09, Stark+10

Why no Ly emission?

Expected line flux > 1 x 10-18 erg s-1 cm-2.This is the 3s flux limit of our VLT observationsWhy all of a sudden we dont see Lya any more?Unphysical that LBG selection only works up to z~6.5 What is the reason for the non-detections:Galaxies are more dust-free?The ISM? (e.g. gas accretion)Cosmic opacity?Our flux limit

Evolution of the UV Luminosity FunctionCastellano et al. 2010

Mass Evolution of L* Galaxies

Stellar Masses at z > 7Stellar masses 108 - 109 M (109 at L*), compared to M 1010 at z 3Solid line is the joint probability distribution of mass from the bootstrap simulations.Overall uncertainty on mass is a factor of 10, BUT is a factor of only +2 and -5.The upper limit on mass appears to be well constrained.Test this by fitting a two-population model, where 90% of the mass is forced to form at z = 20.The young age of the Universe at these redshifts limits the amount of mass in old stars - more so than Spitzer.At tUni 500-800 Myr, M/L ratio dominated by stars with M > 3M, or O, B and early A stars.

Mass Evolution of L* GalaxiesTypical (L*) galaxies are lower mass at higher redshift.We confirm a drop in typical galaxy stellar mass, first hinted at, at z 5-6.Gray dots are Ly emitters (LAEs)z 7-8 galaxies look similar LAEs at all redshifts than LBGs at any redshift (cf. S. Malhotras talk).

Cosmic ReionizationBy adding up the rest-UV fluxes of our objects, we can examine how they would impact reionization.We computed the specific luminosity density for the z 7 and 8 samples.With no correction for dust or incompleteness, our objects come within a factor of a few of sustaining reionization for high escape fractions ( 50%).Assumes low clumping factors (Pawlik+09, Finlator+09). Accounting for unseen faint galaxies brings us a factor of 2-3 closer.Escape fractions of ~50% might be reasonable (e.g., Siana+10; Vanzella+10).Check out poster (#5) by Vanzella: detected ion. rad. from LBG at z~3.8 + new f_esc estimates

How else can we observe early star formation?By looking at things that absorb (not emit) light

i.e. by finding the gas released by early starsThis can be done with high spatial sampling by replacing QSO with galaxies to study absorption systemsHere is an example of LBG MgII absorption systemsIntervening gas in an overdensity at z~1.61There are no galaxies with d

Detected PoP III gas?Fe II trough not detected in the stacked spectrum of the four MgII absorbersIt likely implies that the intra-overdensity gas is chemically more pristine than that in the IGM (outflows) of star-forming galaxies at high-zWith more observing power (TMT, EELT), galaxy absorption systems will be a very powerful tool to explore the early universe MG, EV+ 10

SummaryStudy the evolution of star-forming galaxies at z>6.5 within reach (HST+WFC3, 8-m telescopes).Expect significant progress from WFC3 GOODS+AEGIS MCTP.These early galaxies appear bluer than at 26.5 Lya seems to suddenly disappear from these galaxies. Why?UV luminosity function (bright end) evolves rapidly, faster than previous measures.Stellar mass relatively small, growing rapidly.If escape fractions are high, galaxies at these redshifts may be able to sustain reionization (you MUST read E. Valnzella poster here: detected ion. Rad at z~3.8).