Transcript ACS & SIRTF GOODS

Massive Galaxies at
high redshift
GOODS
http://www.stsci.edu/science/goods
Lewnidas (Lexi) Moustakas
Space Telescope Science Institute
M Dickinson, H Ferguson, M Giavalisco
R Somerville, T Dahlen, B Mobasher, H Yan
1
The SDSS z~0 age/stellar
mass relation
~3x1010 Mo
age
From Kauffmann et al.
2
Kauffmann et al. 2002
outline
 The public GOODS -- new: Spitzer! - identifying massive field galaxies at redshifts z>1
 color-selected samples & fitting SEDs
EROs, IRAC-EROs, J-K-selection ...
 Conclusions:
1. Galaxies with M>few x 1010Msun are abundant even at z~1, 2, 3
2. Spitzer's rest-frame IR observations are key
 In progress:
 Towards a complete census of masses and SFRs at all z's
 Properties as function of local environment (always in the field)
3
galaxy formation:
an observational goal
 A major goal is to measure the
distribution function of stellar mass and
star formation rates over time and
environment
f(M, M/t, t, d)
This encapsulates the assembly history via
all modes -- quiescent star formation,
starbursts, &c.
 Enter GOODS
4
What is GOODS?
 The -Great Observatories Origins Deep Survey An orchestration of deep observations of the
HDFN and the CDFS regions (~ 300 square arcmin
in total) with the most powerful telescopes over
the widest wavelength range
 30 times larger solid angle than HDFN + HDFS
 Based on large programs with Spitzer, HST,
Chandra, Newton, VLT, and more.
 All datasets and derived products are open to the
public domain
5
A Synopsis of GOODS
 GOODS Space
 HST Treasury
(PI: M. Giavalisco)
 B, V, i, z (27.2, 27.5, 26.8, 26.7)
 Δθ = 0.05 arcsec, or ~0.3
kpc at 0.5<z<5
 SIRTF/Spitzer Legacy
(PI: M. Dickinson)
 3.6, 4.5, 5.8, 8, 24 μm
 Chandra (archival):
 0.5 to 8 KeV
 Δθ < 1 arcsec on axis
 XMM-Newton (archival)
•hold
•hold

GOODS Ground
 ESO, (PI C. Cesarsky), CDFS
 Full spectroscopic coverage in CDFS
 Ancillary optical and near-IR
imaging
 Keck, access through GOODS’ CoIs
 Deep spectroscopic coverage
 Subaru, access through GOODS’
CoI
 Large-area BVRI imaging
 NOAO support to Legacy &
Treasury
 Very deep U-band imaging
 Gemini
 Optical spectroscopy, HDFN
 Near-IR spectroscopy, HDFS
 VLA, ultra deep HDFN (+Merlin,
WSRT)
 JCMT + SCUBA sub-mm maps of
6
HDFN
GOODS-S imaging coverage
VLT/ISAAC J & K coverage
shown (ESO v1.0 public
release, May 2004). ISAAC
H-band covers roughly half that
area.
Chandra coverage shown is only
over the best PSF region (6arcmin).
Complete image covers the whole
GOODS-S field.
7
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
10’
4.5, 8.0 mm
16’.5
Second epoch in August 2004 will
complete CDF-S IRAC observations
5s point source sensitivity (shot noise
only):
3.6, 5.8 mm
Channel
mJy
AB mag
3.6mm
0.11
26.27
4.5mm
0.21
25.57
5.8mm
1.35
23.58
8.0mm
1.66
23.35
8
HUDF
The Hubble
Ultra Deep Field
in GOODS-South
BViz + JH
z850~28
09 march '04
9
Beckwith et al. in prep
1.6 to 8 mm view of the
Hubble Ultra Deep Field
What 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
10
Redward-marching CMDs
Overall color distribution gets
bluer at longer wavelengths.
“ERO-like” objects get fainter
and fewer, but are still seen out
to H - 5.8 mm color,
corresponding to zERO > 3
Some bright galaxies pop up
strongly at 8mm; presumably
PAH emission from low-z,
brighter galaxies, or “unveiled”
AGN.
11
the red sequence to z>1.4
Rest-frame colormagnitude diagrams,
z~0.2 to z~1.8
These data are from
GOODS & GEMS,
for different sample
selections. The pink
are K-selected. Red
circles are EROs.
See how these glxs
dominate the red
sequence at z~1 etc!
"extremely red galaxies"
12
Somerville & Moustakas et al - in prep
most KAB<22 extremely red
objects are old-star dominated
"EROs"
late
irregular
other
See also:
Yan & Thompson 2004;
Smith et al 2004
Bell et al 2004
early
Space density of early-type EROs is n~2x10-4 Mpc-3
13
Moustakas et al 2004, ApJL
typical (old) ERO SED
Data
An old-elliptical SED
 The spectral energy
distributions of the early-type
EROs basically demand large
ages, T>2Gyr
 This is true even if there is
some 'frosting' of star
formation in place at z~1 (c.f.
the DEEP2 findings)
 This example has a
GOODS:FORS2 redshift,
z=1.19
 The GOODS:FORS2
spectroscopy of ~80 EROs is
being used for line-index
diagnostics - Kuntschner et al,
in prep
14
Moustakas et al in prep
the (dark) art of SED fitting
Population synthesis fitting to
observed SEDs of Lyman Break
Galaxies at z~3 (inclusion of
Spitzer data is forthcoming!).
A large wavelength range is
needed, especially to the rest-IR.
Significant mass
from older
stellar population
can be hidden by
ongoing star
formation,
-> 'maximum M/L
models'
15
Papovich 2002; Dickinson et al 2003
IRAC-Extremely Red Objects
 IRAC-selected with
fn(3.6mm)/fn(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 in B435), but faint.
z - m(3.4mm) vs redshift
16
Haojing Yan et al 2004, ApJ submitted
An “IERO” in the HUDF
ACS
NICMOS
ISAAC
IRAC
17
SED fitting for IEROs
Most IEROs are best-fit with
unreddened 2-component stellar
populations:
• ~2.5 Gyr old stars
• + secondary ~0.1 Gyr burst
• zphot ~ 1.6 to 2.9
-Key result:* In most IEROs, at
z~2ish, OLD STARS
are required.
* Dust does not seem
to be enough.
18
SEDs of the HUDF IEROs
A few objects are poorly fit by
old stellar models (e.g., with
sharply rising flux to 8mm)
Rest-frame K-band luminosities
~0.35 to 5 times present-day
L*K for early-type galaxies,
implying substantial stellar
masses (~1010 - 1011 Msun)
Number density is comparable
to or greater than that of
present-day galaxies with
similar luminosities
19
Completing the census




K-band EROs at z~1-1.5
3.6mm IEROs at z~1.6-2.9
K-band J-K selection -> z~2.5
UV selected LBGs z~2.5-6 (and >6?)
 In progress... collating all the galaxy
populations found to z~2.5 (ish)
 High-redshift teaser: stellar
populations of galaxies at z~5.8
20
z = 5.828 i-dropout in CDF-S
3.6mm
4.5mm Excellent PSF greatly
improves sensitivity at 3.6
and 4.5 mm relative to
proposal expectations.
Many of the brighter z~5-6
galaxies are well-detected
in channels 1+2.
5.8mm
IRAC Ultradeep HDF-N
observations (up to 100h
exposure time) may yield
detections in channels
8.0mm 3+4
21
Stellar population fitting for
z=5.828 galaxy
Typical LBG colors.
Clear evidence for a Balmer break
between K and 3.6mm.
Otherwise blue SED (above & below
break) suggests low reddening, but
this is not well constrained.
4000 A break
Stellar mass estimate ~1.5x1010 Msun
which is slightly larger than typical
for L* LBGs at z~3
observed wavelength
22
A STScI mini-Workshop
on massive galaxies
27-29 September 2004
23
Conclusions
The rest-infrared data are important!
 Multi-l SED-fitting good for subtleties
 In the field, we find many massive
galaxies (M*>few x1010Msun) out to high-z
The space densities are significant,
n~10-4-10-3 Mpc-3, so important as model
constraints (see RSS talk)
In progress: clustering/environment
24
The near future
3.6mm
4.5mm
5.8mm
8.0mm
Stand by for the
GOODS *Ultradeep*
IRAC observations
-andthe 24mm MIPS data
in both fields
Eucaristw!
25
26
J-K color for z~2-3 selection
threshold color
J-K>1.37 (AB)
J-K>2.3 (Vega)
Recent application of this criterion & of photometric redshift:
van Dokkum et al 2003; Franx et al 2003; Daddi et al 2004
27
Moustakas et al in prep
jk - selection
3s
10s
limit
limit
oo == X-ray
X-ray sources
sources
 The sample I use here
is Ks-selected,
restricted to SNRK>10
 There are formal Jband dropouts that
are included
 Total sample size: 131
galaxies, ~1 arcmin-2
 X-ray sources are
tracked, so two
samples explored
 'wx' - X-ray sources
 'nx' - remaining obj's
28
Moustakas et al in prep
jk - LBG comparison
C.f. Steidel et al 2004 for z~2ish work
 Perhaps half of the jk sample
would be too faint for
ground-based R (rest-UV)
selection to work...
 NOT too faint for z850
selection, though (eg from
GOODS). z850<26 for all jk
galaxies!
 The surface densities are
comparable, ~1 arcmin-2
 The UV colors are only
somewhat red -- V-z~1mag
29
Distribution of jk sources
134 arcmin2
30
HUDF
The Hubble
Ultra Deep Field
in GOODS-South
BViz + JH
z850~28
09 march '04
31
Beckwith et al. in prep
jk - HUDF morphologies
~10'' x 8'', ACS z-band, 0.03''/pix
32
Moustakas et al in prep
Spatial associations
There is early evidence of
strong spatial
correlations (Daddi et al
2004)
Xray
Xray
Our own w(q) & x(q)
measurement is in prog.
The visual associations are
dramatic, and there is
clearly strong
correspondence with
distinct X-ray sources
~1 arcmin across
33
jk - stellar masses
 very early results show rest-frame colors suggest stellar masses quite
comparable to EROs, ~1010Mo and higher
 space densities may be comparable to EROs, as well
 ages are less constrained, still -- stay tuned.
-Possible implications EROs' progenitors were already fully in place upon formation?
 Star formation rates must have been high and sustained earlier
-Questions How do AGN (and environment) figure in this picture?
 What are their star formation rates?
34
jk - X-ray sources
Constraining the photon index
 There are 19/131 X-ray sources
= 15% of the sample.
 Considering the X-ray sources, and
a typical redshift of z~2.2, we
constrain the photon index G and
the in situ obscuring HI column, NH:
 G~1.2 & NH~1.2x1022 cm-2
 Luminosity/object LX>1043 erg s-1
 Largely => OBSCURED AGN
35
Moustakas, Bauer, Immler et al in prep
jk - X-ray stack results
Counts distributions
80 'clean' objects used
for this stack
E(keV)
Ns
full
soft
hard
0.5-8
0.5-2
2-8
6.1
5.5
4.3
fX(cgs)
log(LX)
7x10-18
42.1
41.6
2.5x10-17
42.2
36
Moustakas, Bauer, Immler et al in prep
AGN vs Star Formation
 The observed soft and hard fluxes imply a photon index of
around G ~ 1.8.
 The estimated rest-frame LX(2-8keV) ~ 1042 erg s-1
AGN
If the obscuration is high, the hard-X-ray
flux is absorbed, so the photon index G
will be larger.
The X-ray luminosity and G are consistent
with Seyfert-level AGN activity. Optical
spectroscopy (van Dokkum et al 2003,
Daddi et al 2004) do reveal some AGN
features in the z~2 galaxies.
Large population of obscured AGN?
Star Formation
For a ~Salpeter IMF, and star formation
rates somewhat above a few Msun/yr,
there is a tight relation between SFR
and LX, which arises from high-mass Xray binaries and supernovae.
SFRX ~ 100 Msun/yr [Grimm et al 2003]
SFRUV ~ few Msun/yr [Kennicutt 1998]
"Ultraluminous infrared galaxies"?
37
Comparison with UVselected galaxies at z~2
 Adelberger et al 2004; Steidel et al 2004; Reddy & Steidel 2004





The redshift ranges can be comparable
The rest-UV colors are similar
~50% of jk galaxies would be missed by R-limit, but not by z-limit
The implied X-ray and (uncorrected) UV SFRr are comparable
The pure AGN fraction is similar; it may be higher for jk galaxies
All of these points suggest that results from UV-selected surveys
are somewhat incomplete; and that AGN may in fact be more
adundant than indicated so far.
38
Questions & implications
We are missing at least some of the
mass and star formation at z~2-3
What is the relation of jk's with submm bright z>2 ULIRGs?
There may be a significant amount of
hidden AGN activity at earlier times.
39
J-K colors of SCUBA glxs
our reddest jk
our color cut
our 10s limit
 Many (most??) SCUBA
sources are at <z>=2.4
(Chapman et al 2003)
 The majority have IR
counterparts & many
have similar J-K colors
(Frayer et al 2004)
 The surface densities
are comparable; but
the Frayer sources are
magnified by
foreground cluster.
40
Frayer et al 2004
A picture
 It seems that even at <z>~2.2, the progenitors of
massive galaxies are already in place. Are these
galaxies freshly 'assembled'? Or did that happen
much earlier, still? Why and how would 'monolithic'
collapse happen? This is a major challenge...
 Even so, a lot is happening at that time. There is a
lot of obscured AGN activity, that may be tracing
something else. Morphologies are quite varied.
 I suspect we're missing even more from the picture
at z~3-4, where we might see the 'pieces' of these
most massive galalxies, fall into place.
41
Clustering evolution - theory
 Press-Schechter theory gives the
abundance and clustering strength of
dark matter halos
 Similar global galaxy properties may
be (should be) connected to the dark
matter somehow
 This connection can be made neatly
with the 'occupation function'
42
Moustakas & Somerville 2002
dark matter halo masses
"bias" comes from the clustering,
There can be many galaxies in each
which fixes the 'minimum' DM halo mass
dark matter "halo", or none. The
average behavior can be parametrized
with the Halo Occupation Function,
or Distribution (cf Wechsler's talk)
N(M>Mmin) = (M/M1)a
Mmin - threshold halo mass
M1 - 'typical' mass
a - mass function slope
43
Moustakas & Somerville 2002
galaxies' dark matter halos
 The occupation function
parameters can be
constrained through the
measured clustering
strength and the space
density
 Here we plot the results
for z~0 ellipticals, z~1.2
EROs, and z~3 LBGs
44
linear bias
correlation scale
clustering evolution
 The simplest model hasa
galaxies following the dark
matter they're associated with
-- 'galaxy conserving model'
(Fry 1996)
 See the behavior of
populations with properties
established at different
redshifts. Do they 'connect'?
45
glazebrook et al
The "Gemini Deep Deep
Survey", GDDS, stellar
space density meas'mt.
Comparison is to
'GALFORM' models,
Granato, Baugh.
Are hierarchical models
then, dead??
Glazebrook et al. 2004 : comparison with low baryon-density models...
46