Title: The Evolution of Galaxies over the Last Half of Cosmic Time by S' M' Faber and the DEEP
1The Evolution of Galaxies over the Last Half of
Cosmic TimebyS. M. Faber and the DEEP DEIMOS
Teams
- Supported by CARA, UCO/Lick Observatory, and the
National Science Foundation
2Hubble Types two kinds of galaxies
Frei et al., AJ, 111,174, 1996
3HST galaxy populations in HDF-N
Age
Driver et al. 1998
z
There is a major transition at z 1.4. Red
galaxies appear to end there, and a population
of blue irregulars and compacts appears.
4Primordial density ripples are the seeds of
galaxies and large-scale structure today
Simulation courtesy of Springel, White, and
Hernquist
5Theory LCDM cosmology predicts that all galaxies
formed at least partly by mergers
Dark halo merger tree
Scale Factor Halos
Within the currently favored cosmology (Lambda
Cold Dark Matter, LCDM) structure forms
hierarchically, from the bottom-up. Dark matter
halos (and possibly the galaxies they host) are
built by a series of discrete merging events.
- Z3
Major progenitor 3.9 x 1011 M?
12 distinct halos (gt 2.2 x
1010 M?) - Z1
Major progenitor 1.5 x 1012
6 distinct halos (gt 2.2 x 1010
M?) - Z0
One aalaxy-sized halo roughly the
size of the Milky Way, Mass2.9 x 1012 M?
Wechsler et al. 2002
Figure courtesy of Risa Wechsler
6Local scaling laws relate L, R, v, and IoA
clue to how baryons populate dark halos
7Tully Fisher Relation for Spirals
The TF relation is the correlation between
rotation speed and absolute magnitude for disk
galaxies. W is total linewidth, which is close to
but not exactly 2 x vrot.
Kannappen et al., ApJ, 123, 2358, 2002
8Typical disk surface brightness profiles
Pure exponentials would be straight lines. The
exponential scale length ? is a measure of the
size of the baryonic disk.
Courteau, ApJS, 103, 363, 1996
9Spheroids follow the r1/4 law
Again, there is a radial scale length, here
usually called Reff, which is the radius that
encloses half the light.
10Kormendy laws for ellipticals and other
dynamically hot systems
Ellipticals
Globular clusters
Dw spheroidals
Kormendy, ApJ, 295, 73, 1985
11Fundamental Plane for Spheroids
The Fundamental Plane correlates Re, surface
brightness, and ? for elliptical galaxies.
The Fundamental Plane for Coma and other nearby
cluster ellipticals
Fundamental Plane edge on
Fundamental Plane face on
Jorgensen et al., MN, 280, 167, 1996
12A toy evolution model
Fall and Efstathiou 1970, Faber 1982,
Blumenthal et al. 1984, Mo, Mao, and Whilte 1998
- Assumptions
- Each galaxy relates homologously to its DM halo
- Radius, circular velocity, and mass all scale
in proportion to the analogous quantities of the
halo at exactly the same epoch (Mo, Mao, White).
- M/L does not evolve, so that L can be equated
with mass. Orcorrections are made based on
colors and stellar population theory to derive
baryonic mass from light.
13Some predictions of the model
Then the zeropoints of scaling laws evolve as
follows
Changes we are looking for
Faber et al., Rome Conference on Disk Galaxies,
2000
14(No Transcript)
15The DEEP2 Survey
-
- U.C. Berkeley M. Davis (PI), A. Coil, M.
Cooper, B. Gerke, R. Yan, C. Conroy - U.C. Santa Cruz S. Faber (Co-PI), D. Koo, P.
Guhathakurta, D. Phillips, C. Willmer, B. Weiner,
R. Schiavon, K. Noeske, A. Metevier, L. Lin, N.
Konidaris, G. Graves - U. Hawaii N. Kaiser, G. Luppino
- LBNL J. Newman, D. Madgwick
- U. Pitt. A. Connolly JPL P.
Eisenhardt - Princeton D. Finkbeiner Keck G.
Wirth - K survey (Caltech) K. Bundy, C. Conselice, R.
Ellis, P. Eisenhardt - Groth Strip Spitzer MIPS IRAC, GALEX
16DEEP2 basics
- 4 Fields 14 17 52 30 (Extended Groth
Strip) 16 52 34 55 (zone of
very low extinction) 23 30 00
00 (on deep SDSS strip) 02 30
00 00 (on deep SDSS strip) - Field dimensions 30 by 120 for fields 2,3,4
15 ? 120 for
Groth Strip - Primary Redshift Range z0.75-1.4,
pre-selected using BRI photometry to eliminate
objects with zlt0.75 - Magnitude limit R lt 24.1
- Grating and Spectra 1200 l/mm 6500-9100 Å
OII 3727Å doublet visible for
0.7ltzlt1.4 - Resolution 1.0 slit FWHM1.7Å ? 68 km/s
17DEEP2 vs local redshift surveys
SDSS
2dF
LCRS
DEEP2
z0
CFASSRS
PSCZ
z1
18Masks tiled across a 42x28 CFHT pointing
Standard field is 120x28 1 sq deg
19 DEIMOS during assembly
20DEIMOS on Keck Nasmyth
DEIMOS
21DEIMOS Masks and Detector
- Slit masks are curved to match the focal plane
and imaged onto an array of 2k ? 4k CCDs - Readout time for full array (150 MB!) is 56
seconds (8 amplifier mode)
- The detector is a mosaic of 8 2K x 4K CCDs from
MIT/Lincoln Laboratories. The CCDs are
high-resistivity, red-sensitive devices that are
45 ? thick, with a peak QE of 85 and enhanced QE
of 23 at 10,000 A.
22Pre-selected photo-zs gt 0.7
23Typical redshift distribution
Photo-z color cut is working very well
We are currently measuring redshifts for 70 of
the targets. Nearly all failures are at higher z
(Steidel 2003).
24The Groth Strip has no color cut
The Extended Groth Strip in DEEP2 extends down to
z 0 4,000 galaxies now have redshifts in this
field Large scale structure walls are
visible Color bimodality red/blue
25Progress on the 1Hour Survey
- 90 planned Keck nights in total
- Started in July 2002
- 36 nights have been used (40)
- 54 nights remain (60)
- 5,500 redshifts were measured from 7,000 spectra
in the 2002 season (1,600 beyond z 1) - 2,700 redshifts have been processed so far from
the 2003 season (all in the Groth Strip) - 28,000 spectra have been obtained to date 40
- Planned ending date Fall 2005
26Data beyond z 1 are increasing strongly
Galaxies exist in large numbers beyond z
1. There is no redshift desert beyond z1.
27Luminosity FunctionsandColor Bimodality
28An important discovery the red sequence
The color-magnitude diagram from SDSS
The color-magnitude sequence of early-type
galaxies.
Hogg et al.,
29COMBO-17 Color bi-modality to z1.1
25,000 galaxies
17-color photo zs
R-band selected to R 24
Bell et al. 2004
30DEEP2 sees the same color bi-modality to z1.4
Correcting for M/L, red galaxies are more massive.
Sloan finds the same red/blue division locally
and puts a fuzzy dividing line at 3 x 1010
solar masses.
Willmer et al. 2004
31R-band selection limits the depth of the
luminosity function
Note how the magnitude limit changes slope in the
CM diagram with redshift. It is vertical when
the color used for restframe absolute magnitude
(here the B band) is redshifted into the filter
bandpass used for the sample selection (here the
R band.
Willmer et al. 2004
32R-band selection limits the depth of the
luminosity function
The limiting absolute magnitude for the ALL
function at z 1 is -21.2 B mag.
Willmer et al. 2004
33Luminosity functions COMBO-17 and DEEP1 agree
well to z 0.8
Restframe B band
- Red dots are COMBO-17, redshifts are photo zs.
- Dark squares are DEEP1, redshifts are
spectroscopic. - Both surveys go to R24
- Curves are COMBO functions at z 0.0-0.3.
- Blue galaxies function shifts to brighter mags.
- Red galaxies function shifts down and to left.
All
All
Red
Red
Blue
Blue
Willmer et al. 2004
34DEEP2 and COMBO-17 agree out to z 1.2
Restframe B band
Restframe B band
All
- Blue galaxies L brightens by about 0.8 mag at
z 1, but number density is constant. - Red galaxies L brightens by about 1.6 mag at
z 1, but number density is lower.
All
Red
Red
Blue
Blue
Willmer et al. 2004
35DEEP2 and COMBO-17 agree out to z 1.2
Restframe B band
All
- Blue galaxies L brightens by about 0.8 mag at
z 1, but number density is constant. - Red galaxies L brightens by about 1.6 mag at
z 1, but number density is lower.
All
Red
Red
Blue
Blue
Willmer et al. 2004
36CFRS luminosity function evolution
- Claimed results
- Red-galaxy function does not evolve to z 1.
DOES NOT AGREE. - The blue-galaxy function rises and steepens at
the faint end. DATA AGREE,
DIFFERENT INTERPRETATION. - Mag limit was only R 22.5. New data are 1.5
mag deeper.
Steepening
Lilly et al. 1995
37Luminosity Functions Since z 1
Blue galaxies
Dimming of 0.5-1.0 mag constant number density
Red galaxies
Dimming of 1.7 mag rising number density with
time
38Color evolution implies only modest changes in M/L
DEEP1
RC3
39Color evolution implies only modest changes in M/L
0.4 mag
0.7 mag
DEEP1
RC3
40 Blue Galaxy Scaling LawsCompare to Toy Model
41Toy evolution model (Mo, Mao, and White 98)
Consider a model in which each galaxy relates at
all epochs homologously to its DM halo, so that
radius, circular velocity, and mass density scale
directly with analogous halo quantities. Assume
further that M/L does not evolve, so that L can
be equated with mass. Then the zeropoints of
scaling laws evolve as follows
Changes we are looking for
Faber et al., Rome Conference on Disk Galaxies,
2000
42Stellar Mass vs Size
COMBO-17 Disk galaxies Mass-radius relation
Completeness Map
43COMBO-17 Disk galaxies Mass-radius relation No
shift in zeropoint vs. time
Toy model prediction
44The toy model says
The zeropoints of scaling laws evolve as follows
Changes we are looking for
Faber et al., Rome Conference on Disk Galaxies,
2000
45Now look at the TF relation
The zeropoints of scaling laws evolve as follows
Changes we are looking for
Faber et al., Rome Conference on Disk Galaxies,
2000
46Roughly half of one DEEP2 mask
OH sky lines
- Each slitmask has 140 objects over an 8k x8k
array. The average slit length is 5 with a gap
of 0.5 between slits. We tilt slits up to 30
degrees to trace the long axis of a galaxy.
47 Flexure Compensation CCDs
48A fully automated reduction pipeline removes the
sky lines
Thank you, UC Berkeley!
O II 3727
SDSS spectral pipeline code by Schlegel et al.
allowed us to rapidly develop a full 2d and 1d
spectral reduction pipeline that is completely
automated
A few percent of one DEEP2 mask, rectified,
flat-fielded, CR cleaned, wavelength-rectified,
and sky subtracted. Note the resolved OII
doublets. Shown is a small group of galaxies
with velocity dispersion ? ? 250 km/s at z?1.
Note the clean residuals of sky lines. High
dispersion improves sky subtraction.
49 Poisson-limited sky subtraction
Plot shows residual of flux from b-spline sky
model in region of sky emission lines, in units
of local RMS. Smooth curve is gaussian, width 1.
50High spectral resolution also enables kinematics
studies
- Improved internal velocity measurements with
high-resolution DEIMOS data
Resolved OII doublet with 220 km/s separation
8 arcsec
8 arcsec
51Many rotation curves are marginally resolved
spatially
Cooper et al. 2004
Four 2-d spectra showing resolved, tilted OII
emission, and derived circular velocity Vc(r).
All curves tend to look linear at low spatial
resolution.
52For most of the others, we are able to measure
integrated linewidths
Fit lines to 1- D extracted spectrum. Kinematic
measurement even when spatially unresolved
Altogether we measure linewidths for 80 of all
blue galaxies
Weiner et al. 2004
53GOODS-N Keck Treasury TF relation
- Keck Treasury Redshift Survey
- Team Keck and DEEP DEIMOS spectra, 1420
redshifts in GOODS-N - Spectra are now public
- Zeropoint offset of 1.5 magnitude at z 1
Z 0
Weiner et al. 2004
54DEEP2 TF relation Season 1
3200 galaxies beyond z 0.65 Similar results
offset is 1.5-2 mag at z 1
Weiner et al. 2004
55Summary Blue Galaxy Luminosity Evolution
- Luminosity function shift ? M 0.5-1 mag -
Color change ? M 0.5-1
mag - Number density Roughly
constant - TF zeropoint shift ? M
1.5-2 mag bigger! Implies that v is SMALLER
at fixed mass.
56Scorecard for toy model
- V vs. M change is opposite in sign to that
predicted by toy model. Distant galaxies rotate
slower than locally at fixed M, not faster.
- R vs. M does not change. Smaller disks at fixed
M are expected at high z but are not seen.
Changes we are looking for
Faber et al., Rome Conference on Disk Galaxies,
2000
57Blue-galaxy morphologies
Randomly selected blue galaxies in GOODS-North
ordered by redshift z 0.65 ?1.4 in z band
Redshift
58Red-Sequence Galaxies
59The red sequence contains a wide range of
morphologies
A selection of red-sequence and related galaxies
in GOODS-N. Peculiars, spirals, and normal
E/S0s are evident. Harker et al. 2004
Normal spiral
Merger
Post-merger
Normal E/S0
Blue center
Blue ring
60The red sequence contains a wide range of
morphologies
A selection of red-sequence and related galaxies
in GOODS-N. Peculiars, spirals, and normal
E/S0s are evident. Harker et al. 2004
Color maps
Color maps (blue is dark)
Normal spiral
Merger
Post-merger
Normal E/S0
Blue center
Blue ring
61Three classes dominate
62Same classes seen in GOODS-N
63DRGs populate the valley and increase in numbers
at higher redshift
.they may be related to the dusty EROs seen
beyond z 1.5
Groth Strip
GOODS-N
DRGs are blue squares
Weiner et al. 2004
64Most red-sequence galaxies have large bulges and
high concentrations, like normal E/S0s
Red
Weiner et al. 2004
65But even normal E/S0s are often disturbed
Van Dokkum Ellis 2003
Red-sequence E/S0 galaxies in HDF-N. 40 of all
spheroidal galaxies to 23 R mag are disturbed.
Roughly 1/3 of these show blue centers and are
also candidate AGNs.
66However, to R24 at z 0.7, 85 are normal
E/S0/Sas
Normal E/S0s
Normal spirals, some edge-on
Irregulars and peculiars
Bell et al. 2003
67Fundamental plane of field spheroidal galaxies
to z 1
2 mag
?SB 2 mag _at_ z 1 Open circles are disks with
big bulges
Gebhardt et al. 2003
68FP implies 2 mag fading for field E/S0s at z 1
if galaxies are not merging and changing their
structure
?SB versus redshift from FP
Green are field galaxies from DEEP1
Blue are cluster galaxies from literature
X
Fading implied by red luminosity function
Gebhardt et al. 2003
69Butmean integrated U-B colors are close to those
of local E/S0s
70Color vs magnitude evolution for E/S0s
Gebhardt et al. see little color evolution
despite large surface bness evolution. This
suggests a frosting model in which 94 of the
stars formed at z gt2 while the remaining 6
continued to form with ? 7 Gyr. Trager et al.
(2002) also argue that nearby E/S0s have frosting
populations of younger stars.
U-B is rather flat!
Gebhardt et al. 2003
71Stacking improves S/N dramatically. Stacked
spectra of 50 red field galaxies at z 0.8
ltzgt 0.78
Schiavon et al. 2004
72Single-burst Balmer ages are only 2-3 Gyr at z
0.85
Schiavon et al. 2004
73A single-burst model to fit the Balmer lines
forms as recently as z 1.35
Joins red clump at z 1.1
Forms at z 1.35
74But Balmer lines of nearby E-S0s do not fit a
single-burst modelnearby lines too strong
Continuing high Balmer strength is more
consistent with frosting models
H? too strong here
as are constant colors
Schiavon et al. 2004
75But Balmer lines of nearby E-S0s do not fit a
single-burst modelnearby lines too strong
Continuing high Balmer strength is more
consistent with frosting models
H? too strong here
Frosting models imply ongoing star formation. Is
this detected?
as are constant colors
Schiavon et al. 2004
76O II 3727 suggests ongoing star formation in 30
of distant red-sequence galaxies
O II
Konidaris et al. 2004
77OII is below detection limit
When we see OII, the emission is broad
An example of a blue galaxy w/ rotation
Wavelength
78No clear trend with morphology
Strong O II
Weak O II
No O II
79Balmer lines increase more rapidly in field E/S0s
than in clusters, same as FP offsets
(field)
Cluster E-S0s
80Current semi-analytic models dont match the CM
diagram
SAM models for z 0 galaxies, courtesy of Rachel
Somerville
Standard
Enhanced merging
WE DONT KNOW HOW TO QUENCH RED GALAXIES
81Current semi-analytic models dont match the CM
diagram
Observed CM diagrams at z 0.9 and 1.1.
WE DONT KNOW HOW TO QUENCH RED GALAXIES!
82The red-galaxy luminosity function is evolving
strongly
Consistent with jlum constant!
All
Yet populations are fading by 1-2 mags
jlum constant
Red
Stellar mass in red galaxies must at least double
since z 1 (COMBO-17)
Blue
COMBO-17 DEEP2
83Red sequence stellar mass evolution?
B-band luminosity density
U-V color
(Bell et al.)
U-V colors of COMBO-17 suggest passive fading by
at least x3. Fundamental plane suggest fading by
x4-6.
Yet total luminosity density in red sequence
galaxies stays CONSTANT.
Total stellar mass in red-sequence galaxies must
be rising by x3-6 from z 1 to now (Bell et al.)
84 Two extreme models for red-galaxy formation
The L model star formation continues during
mass assemblygalaxies grow bright and blue and
THEN fade. Bell et al. point out that there are
no galaxies in the requisite part of the CM
diagram below z 1, so evolution would have to
occur BEFORE z 1. But then red mass would NOT
TRIPLE since z1.
Empty ever since z1!
L model
85Two extreme models for red-galaxy formation
The Z model star formation ceases DURING mass
assembly. The final stage is then mainly STELLAR
mergers. The process must be finely tuned to keep
jlum constant. Are there enough mergers after z
1 to support this? Can we see this happening!
Z model
86A third model for red-galaxy formation
The unveiling model Red galaxies are heavily
obscured before appearing on the red-sequence.
Their precursors are not visible in the CM
diagram at all. To test count fraction of faint
dusty galaxies vs. z in SIRTF surveys.
Z model
87Conclusions
- The CM diagram is bimodal to beyond z 1. We
should start thinking of red and blue galaxies as
two separate classes. - Blue galaxies already well formed by z 1 and
settling thereafter? - -- Constant number density, slowly fading
stellar populations. - -- Small linewidths perhaps because disks not
yet settled? - -- Does mass continue to accrete? Do radii
evolve? - Red galaxies a dynamic population only partly
in place by z 1? - -- Rising numbers with time, increasing total
stellar mass. - -- Evidence of continuing star formation and
dynamical disturbances. - -- Yet red galaxies are basically gas poor.
What quenches red galaxies?
88Update on Marc Davis...
- Marc suffered a stroke in late June his recovery
and rehabilitation is ongoing, and will continue
through much of this year. - He is now visiting campus most afternoons,
attending team meetings, reading email, etc. His
participation increases every week, but the top
priority for now remains rehab.
89(No Transcript)
90But the volume emissivity of red galaxies stays
constant to z 1
The lines show the fading of PURELY passive
stellar populations. If populations fade
passively, then mass density of stars in red
galaxies must rise by x2 since z1.
COMBO-17 Bell et al. 2003
91Passive fading is indicated by high color
evolution (reddening)
COMBO-17 sees ?(U-V) 0.40 mag since z 1
(restframe).
92Bulges of disk galaxies are also very red by z 1
CM diagram for bulge components of disk
galaxies Bulge colors are even redder than the
ridgeline of the CM relation for a distant
cluster at z 0.83.
Koo et al. 2004
93Linewidths as a measure of circular velocity
DEEP1 spectra in HST WFPC Groth Strip Vrot
measured by Nicole Vogt with full modeling Good
correlation between linewidth and Vrot s 0.6
Vrot predicted (Rix et al. 1997)
Weiner et al. 2004
94DEEP1 and DEEP2 color vs. redshift
DEEP1 600 zs
DEEP2 5500 zs --- 10 of total
DEEP2 starts at z 0.75
95Luminosity density evolution in COMBO-17
Restframe B and R increase by only 1.5 280 nm
increases by x5, indicating more star formation
at z 1.
Wolf et al. 2003
96O II 3727 sample
SDSS E sample
Konidaris et al. 2004
97CM diagrams from Sloan
Red sequence strongest in g-r
Galaxies bluer than this must be bursting
CM relation for envelope
Blanton et al., ApJ, 594, 186, 2003
98Color-color diagrams from Sloan
g-r shows the dichotomy most strongly. Well over
half of the local stellar mass is in red
galaxies. Why are these relations curved? Not
predicted by stellar population models?
Emission? Dust?
Blanton et al.,
99Spectra of red vs. blue Sloan galaxies
Blue galaxies have young stars (Balmer
absorption) and ongoing SFR (emission). Red
galaxies have only old stars.
Blue
Red
Kauffmann et al., MN, 341, 33, 2003. Similar
results from 2Df (Madgwick et al., MN, 343, 871,
2003).
100Why We Need to Understand MergersThe Merger
Fraction Increases Rapidly to z1
Brinchmann Ellis (2000) studied galaxy
morphologies in the Hubble Deep Field (HDF) and
found a distinct rise in the number density of
peculiar (read interacting?) galaxies as a
function of redshift.
Big rise in Peculiars!
- Consistent with Patton et al. (2002) measurement
of the merger rate, as measured by close pairs of
galaxies, in the CNOC2 survey.
Slides in the section by Joel Primack and TJ Cox
101Theory LCDM Cosmology predicts that all galaxies
formed at least partly by mergers
Scale Factor Halos
Within the currently favored cosmology (Lambda
Cold Dark Matter, LCDM) structure forms
hierarchically, from the bottom-up. Dark matter
halos (and possibly the galaxies they host) are
built by a series of discrete merging events.
- Z3
- Major progenitor 3.9 x 1011 M?
- 12 distinct halos (gt 2.2 x 1010 M?)
- Z1
- Major progenitor 1.5 x 1012 M?
- 6 distinct halos (gt 2.2 x 1010 M?)
- Z0
- 1 Galaxy size halo roughly the size of the
Milky Way, Mass2.9 x 1012 M?
Wechsler et al. 2002
102GALEX M31, Sb
103The importance of GALEX
- Cross-calibrate the UV vs. H? (optical) SFR
indicators
- Provide key input to dust-absorption models
- Provide morphological K-corrections for distant
galaxies
M101, Sc
104New work on environment
SDSS CM diagram
Hogg et al. (2003)
105Stellar Mass vs Size
Completeness Map
106(No Transcript)
107(No Transcript)
108(No Transcript)
109(No Transcript)
110Toy model prediction
111EROs are defined as having R-K gt 5
They display a wide variety of morphological
types. Some fraction are normal E/S0s. Some may
be dusty starbursts. ULIRGs? Elliptical
progenitors?
Selection of ERO images from GOODS-S from
Moustakas et al. 2003
To count the number of pure spheroidal
galaxies, we need to weed out and discard
dusty/peculiar/reddened EROs.
112Bruzual Charlot, 2003
Model Spectra
Young
Flux (Arbitrary Units/Å)
Old
Wavelength (Å)
113OII is below detection limit
When we see OII, the emission is broad
Spatial Direction
An example of a blue galaxy w/ rotation
Wavelength
114Models Harker et al, 2004 (in prep)
Star formation spread out
4 Gyr
6 Gyr
8 Gyr
Most Star Formation Early
Specific Star Formation Rate
10 Gyr
2 Gyr
12 Gyr
SDSS Upper Limit Eisenstein et al, 2003
Restframe Color (U-B)
Red
Blue
115Excluded Region born before BB
Specific Star Formation Rate
Restframe Color (U-B)
Red
Blue
116(No Transcript)
117(No Transcript)
118(No Transcript)
119(No Transcript)
120Virtually every ULIRG is found to be a merger or
a recent merger remnant
95 of all ULIRGS are seen to be double or
interacting. Mean separation of nuclei only 2
Kpc. LATE-stage mergers. Bright ULIRGs make
stars at a rate of gt100 M?/yr. Normal galaxies
make stars at a rate of 1 M?/yr.
Borne et al., 2000
121DEIMOS First Light
122(No Transcript)
123(No Transcript)
124(No Transcript)
125(No Transcript)