high redshift galaxies in the great observatories origins deep survey (goods) mark dickinson...

High Redshift Galaxies in the Great Observatories Origins Deep Survey

(GOODS)

Mark Dickinson(National Optical Astronomy Observatory)

for the GOODS team

29 September 2004 Massive Galaxies through Cosmic Time

GOODS: Great Observatories Origins Deep Survey

The Great Observatories Origins Deep Survey

(GOODS)GOODS unites the deepest surveys from NASA’s Great Observatories (Spitzer, Hubble,

Chandra), ESA’s XMM-Newton, and the best ground-based observatories, to establish ultradeep reference fields with public data sets to study the evolution of galaxies and active galactic nuclei over the broadest accessible range of redshift and cosmic time.

GOODS provides:• Two fields, two hemispheres, 60x larger than historical Hubble Deep Field• “Near-HDF” depth 4-band HST/ACS imaging (HST Treasury Program: M. Giavalisco PI)

– All observing completed & data released

• Deepest Spitzer coverage from 3.6 – 24 m (Spitzer Legacy Program: M. Dickinson, PI)– Data-taking 3/4 complete; first public release in October 2004

• Deepest X-ray coverage from Chandra & XMM-Newton • Extensive supporting observations (NOAO, ESO, Gemini, Keck, ATCA, VLA, etc.)• Public release of data and data products

Primary science goals:• The mass assembly history of galaxies• Census of energetic output from star formation and supermassive black holes• Emergence of the Hubble Sequence• Supernovae at high redshift and the cosmic expansion• Dark matter distribution through weak gravitational lensing• Measurements or limits on the discrete source component of the EBL



Hubble ACS image of theGOODS CDF-S (detail)

Complete v1.0 data release now availablehttp://www.stsci.edu/science/goods



1st epoch Spitzer GOODS CDF-S IRAC images

First epoch CDF-S IRAC data observed in February 2004:• 23.2 hours/position x 4 pointings• ~60% of field covered in each IRAC channel• ~20% of field has 4-channel overlap, including the HUDF

Second epoch in August completed CDF-S IRAC observations

5 point source sensitivity (shot noise only):

4.5, 8.0 m

3.6, 5.8 m

10’

16’.5

Channel

Jy AB mag

3.6m 0.11 26.27

4.5m 0.21 25.57

5.8m 1.35 23.58

8.0m 1.66 23.35



GOODS Spitzer/IRAC view of the Hubble Ultradeep

FieldIRAC extends our view of the distant universe to wavelengths 5x redder than Hubble/NICMOS.

What Spitzer/IRAC sees:

• Light from longer-lived, red stars that dominate the mass of galaxies, redshifted to IRAC wavelengths

• Starlight and active galactic nuclei obscured by dust

• Potentially capable of seeing extremely distant objects, z > 7, which are invisible to optical telescopes



IRAC Extremely Red Objects

(H. Yan et al. 2004)• IRAC-selected with f(3.6m)/f(z850)

> 20 (AB color > 3.25

• Like (R-K)Vega > 5 “ERO” criterion, but shifted to redder bandpasses.

• We may expect that this will select ERO-like galaxies at z > 1.5 to 2

• 17 objects in HUDF area after excluding ambiguous cases due to blending

• 2 are undetected in ACS HUDF; others are detected (even down to B435) but faint.

• All have J-K > 2.3; 5/17 have J-K > 3. Four overlap Chen & Marzke sample of 9 red HUDF objects.



IERO example

HST/ACS

HST/NICMOS

Spitzer/IRAC

VLT/ISAAC



17 IEROs



SED fitting for HUDF IEROs

Most IEROs can be fit with unreddened 2-component stellar populations:

• ~2.5 Gyr old stars• + secondary ~0.1 Gyr burst• zphot ~ 1.6 to 2.9



“Weighing” old galaxies at high redshift

Red colors & spectral shape require presence of old stars at 1.6 < z < 3.5, with <zmed> = 2.4

For Chabrier IMF: stellar masses 0.1-1.6 x 1011 Msun (median 5e10). Compare to M*(z=0) = 8e10, or typical z=3 LBG M=6e9.

21.9 < KAB < ?? (faintest have KAB > 25), with median <K> = 23.8 (Kvega ~ 22)

2 objects have X-ray detections (AGN).



Space density of IEROs

Highly uncertain!• Based on HUDF alone so far - 13 objects in “refined sample”,

field-to-field variance?• No secure estimate yet of selection & completeness (but

presumably the number is a lower limit)• No empirical measure of volume selection function; we

assume a volume-limited sample with 1.6 < z < 2.9 and M(KAB) < -21.3

Nevertheless…barreling ahead:• MK < -21.3 : n = 3.2e-4 Mpc-3

• MK < -23.0 : n = 1.3e-4 Mpc-3

• 14% and 20% of present-day early-type galaxy space density, assuming 1.0 mag of passive fading to z=0

• M < -23 sample all at 2.4 < z < 2.9: ~50% of z=0 (Egals)



HDFN-JD1



HDFN-JD1 from Spitzer

Spitzer IRAC data clearly favors old, z~3 stellar population for HDFN-JD1.

No MIPS 24 m detection: probably not a HR10-like starburst/ERO.



HDF/NIC #1252

Diffuse; UV-detectedJ-H break; zphot ~ 2.7X-ray + radio detectedProbably 24m detected



GOODS/MIPS 24m HDF-N

10.9h MIPS integration per position

Reaches 20 Jy sensitivity (fainter for isolated sources)



Lurid color composite



600 24 m sources 0 < z < 3+



IRAC LBGs out to z ~ 6

3.6m 4.5m

5.8m 8.0m

Excellent PSF greatly improves sensitivity at 3.6 and 4.5 m relative to proposal expectations.

Many of the brighter z~5-6 galaxies are well-detected in channels 1+2.

IRAC Ultradeep HDF-N observations (up to 100h exposure time) may yield detections in channels 3+4



Stellar population fitting for z=5.828 galaxy

Typical LBG colors.

Clear evidence for a Balmer break between K and 3.6m.

Otherwise blue SED (above & below break) suggests low reddening, but this is not well constrained.

Stellar mass estimate ~1.5e10 Msun

which is slightly larger than typical for L* LBGs at z~3



Redward-marching color-magnitude diagrams

Overall color distribution gets bluer at longer wavelengths.

“ERO-like” objects get fainter and fewer, but are still seen out to H - 5.8 m color, corresponding to zERO > 3

Some bright galaxies pop up strongly at 8m; presumably PAH emission from low-z, brighter galaxies, or “unveiled” AGN.



Optical + X-rayHST/ACS + Chandra:

Spitzer detects optically invisible X-ray sources(Koekemoer et al. 2004)

7 out of 200 CDF-S X-ray sources were optically invisible in GOODS/ACS. What might these be?– Unusually faint host galaxies– Extremely red (old or dusty) hosts– Extremely high redshift AGN (z > 7)

All 7 “extreme X-ray/optical objects” (EXOs) are detected by GOODS/Spitzer IRAC observations at 3.6-8 m

5.8 micronsSpitzer/IRAC:



Colors of EXOs

5 EXOs have extremely red ISAAC J-K colors.

One is still undetected at JHK (another has not been observed)

Two objects with H-band data have their breaks between J and H.



SEDs of EXOs

EXOs appear to be a mixed collection.

• A few appear to be old, red host galaxies (like IEROs) at z~3, possibly requiring dust

• Some are far too red in IRAC colors, and evidently require high redshifts and huge obscuration.

• A few are not inconsistent with z > 7 AGN, but this is still far from certain.

• 2nd epoch IRAC + more VLT H-band will really help here!



Summary

IEROs: Evolved galaxies at high redshift (even in HDF-N!)• Strong overlap with other near-IR-selected samples (e.g. FIRES)• Most have trace ongoing SF, but it can be very modest/faint !

Some truly dead objects (or some dust)• Stellar masses are significantly larger than typical z~3 LBGs• Space density is a significant fraction of present-day early-type

galaxies, but much smaller fraction (<10%) of all comparably massive galaxies today.

• We’ll get a better census at the bright end over 30x more GOODS area (limited by the optical sensitivity)

Red objects are a mixed bag (no surprise any more):• Evolved galaxies• Powerful starbursts• AGN (e.g., the EXOs)



Summary

• GOODS IRAC data detect hundreds of LBGs at 2 < z < 4.5, enabling stellar population modeling

• MIPS 24m is detecting LBGs at 2 < z < 3 down to LFIR ~1011 Lsun Calibration to SFRs requires careful comparison of multiwavelength SF indicators, and will rely to some degree on lower-z data.

• IRAC selection and broad wavelength range should enable unified selection/analysis of galaxies at 2 < z < 4, tying together the color-selected samples.

• We detect a handful of LBGs to z~6, and trace SEDs to rest-frame R-band (hopefully to rest 1m with HDF-N ultradeep IRAC data).

• First z~6 examples have stellar masses comparable to z~3 LBGs, but these are the brightest (optically) at their redshift.



Discuss among yourselves

Where do we go from here?

Observational challenges at high redshift:• Are the most massive galaxies enough to really understand evolution?

Probably not - we need a complete census with some dynamic range in mass at all z.

• Need to go ridiculously faint, even for imaging! IEROs: <Kvega>=22, z850 > 26• Spectroscopy: only scratching the surface. What options do we have? Must

this wait for JWST + >=20m telescopes?• Are our surveys adequate for the most massive galaxies? (We need

equivalent to SDSS LRG volumes at z > 1.)• Missing so far: cold gas at high z (except via DLAs - never mentioned here)Connecting populations across redshifts:• Clustering? Kinematics? Metallicity? Stellar populations?• Are any of these individually convincing, or are all required?Theoretical challenges mostly concern gas supply and regulating star formation• What regulates star formation? What turns it off? How does color

bimodality come about?

high redshift galaxies in the great observatories origins deep survey (goods) mark dickinson...

Documents

massive galaxies

goods team slide

cosmic time irac

old galaxies

mass of galaxies

evolution of galaxies

ieros slide

high redshift galaxies