Transcript Stellar Mass Assembly History

GALAXY MASSES…AND BACK
TO DOWNSIZING!
Motivation
Great progress in tracking star formation history but:
- SF density averages over different physical situations (e.g.
bursts, quiescent phases)
- hard to link to theory (witness how CDM is always able to
match the data!)
Stellar mass assembly is a fundamental measurement
- here is where CDM is in deep trouble
- measurements of galaxy masses better link progenitors and
descendants
- most galaxy properties depend on galaxy mass…
Importance of developing mass diagnostics stressed in several early papers:
Broadhurst et al, Kauffmann et al, Cohen et al, Brinchmann & Ellis
1990s N-body, CDM “con gas!”
“No model has succeeded so
well -- structure formation using
dark matter halos, and failed so
miserably -- making galaxies, as
CDM with baryons.” -- Me.
“elliptical””
Cold
Dark
Matter!
“spiral”
Stellar Mass Assembly History (CDM)
Evolution of stellar mass function
Merger trees
z=5,3,2,1,0.5,0
time
z=5,3,2,1,0.5,0
time
z=1
CDM predicts recent growth in assembly of spheroids, slower growth in disks
Dry Mergers?
Caveat: Most observations measure the
ages of the stars in galaxies of different
masses. Young ages are seen for stars in
low mass galaxies and old ages for stars
in massive galaxies..seemingly in
contrast to hierarchical predictions.
van Dokkum (2005) argues high
preponderance of red tidal features & red
mergers in local samples, coupled with a
postulated increase in merger rate (1+z)m
implies significant mass evolution is still
possible in large galaxies:
i.e. stars could be old but assembled mass
could be younger via self-similar merging
of red sub-units (so-called
`dry
mergers’)
Dry Mergers at High Redshift
Clusters: Tran et al
(astro-ph/0505355)
Field: Bell et
al (astroph/0506425)
No good statistics yet on how prevalent this process is
“Dry mergers” -- the latest thing!
Bell et al. 2006 ApJ, 640, 241.
“Dry Mergers in GEMS: The Dynamical Evolution of
Massive Early-Type Galaxies”
By analyzing the images in the Combo-17 survey, Bell et
al. conclude that the typical massive galaxy could have
undergone ~1 dry merger since z ~ 1, more consistent
with the hierarchical picture (at least from the halo
standpoint)).
“Spectroscopic Confirmation of Multiple Red GalaxyGalaxy Mergers at z < 1” -- Tran et al. 2005 ApJ627,L25
The apparent mergers are real…R<14 kpc, ΔV < 165
km s-1. “…these bound pairs must evolve into E/S0
members by z = 0.7…most if not all, of its earlytype members evolved from (passive) galaxy-galaxy
mergers at z < 1”
Identified as a special
epoch in the cluster’s
life, or significant
subcluster merging
(relative velocities
still low), before
virialization
Dry mergers?
Jeltema, Muchaey and collaborators -- study of x-ray groups
redshift
environment
WHY ?
the best predictor of galaxy type is mass! Look at
these luminosity functions for elliptical, S0, and
spiral galaxies:
Mass-to-light ratio
of (star forming)
spirals are ~3-10 x
less than those of
E and S0 galaxies
Luminosity!
And now we also know that these other properties (all really
a function of mass?) go along the Hubble Sequence of
“bulge-to-disk:
•
Environment: “early” types = dense environments
•
Color & star formation: late-types are blue, star forming
•
Mass-to-light ratios: star forming systems produce a lot
more light per unit mass
•
Stellar Age: early galaxies are early, at least their stars
are. Late types are slow developers with younger stars
•
Mass of central black hole: scales with the spheroid
•
Mass of the dark matter halo: scales with (drives?) all of
the above.
Sloan Digital Sky Survey
Blue
Red
Moving through time
3 x 1010Msun
Massive-Passive =
E & S0 galaxies
Blue, star-forming =
spirals, irregulars
Kauffmann et al 2003
Color distribution versus galaxy
magnitude in the Sloan
Baldry et al. 2004
Current star formation rate per unit galaxy stellar mass (M*) vs M*
But
depending on
environment
…
Kauffmann et al. 2004 - Sloan
Measuring galaxy masses: what are the options?
Dynamics: rotation & dispersions
(only for restricted populations)
Gravitational lensing
(limited z ranges)
IR-based stellar masses
(universally effective 0<z<6)
K
Dynamical methods
Rotation curves for disk systems
(e.g. Vogt et al. 1996,1997)
Stellar velocity dispersion for
pressure-supported
spheroidals (e.g. van Dokkum &
Ellis 2003, Treu et al. 2005,
Rettura et al. 2006)
Issue of preferential selection
of systems with “regular”
appearance
The Fundamental Plane:
Empirical correlation between size, μ and *
Considerably superior as a
tracer of evolving mass/light
ratio and assembly history:
Dynamical mass:
- no IMF dependence
- Closer proxy for halo mass
Tough to measure:
-  demands high s/n spectra
- large samples difficult
M = K σ2R/G
Dressler et al. 1987; Djorgovski & Davis 1987; (e.g. Bertin et al. 2002)
Bender Burstein & Faber 1992; Jorgensen et al. 1996
Stellar Masses from Multicolor Photometry
(especially near-infrared)
spectral energy distribution
Mass likelihood function
log mass
K-band luminosity less affected by recent star formation
than optical
Spectral energy distribution  (M/L)K Redshift  LK
hence stellar mass M
log mass
e.g. Kodama & Bower 2003,
or Bundy et al. 2005,2006
What if you don’t know the redshift?
logM
Expected scatter based on
photo-z error distribution
zspec
Catastrophic errors securing photo-z & masses from same photometry
Bundy et al 2006
What if you only have optical photometry?
A key ingredient in the mass determination is infrared photometry
which is sensitive to the older, lower mass stars; important z > 0.7
BRI vs BRIK
log  (Mopt)
log
MoptMIR
zspec
Bundy 2006 Ph.D. thesis
log  (MIR)
Einstein Rings
ring arising from single background source
lensing
galaxy
For a compact strong lens aligned with a background source, a ring of
light is seen at a radius depending on the geometry and the lens mass,
i.e. this allows us to measure the mass of the lens
DOWNSIZING EFFECT
IN STAR FORMATION AND
MASS
4 clusters at z=0.7-0.8
EDisCS collaboration
De Lucia et al. 2004
ApJL, 610, L77
Data from Terlevich et
al. (2001)
Smail et al. 1998; Kajisawa et al. 2000,
Nakata et al. 2001, Kodama et al. 2004
-- The effect is seen also in the single-cluster
distributions, despite of the variety of cluster
properties: such a deficit may be a universal
phenomenon in clusters at these redshifts
A deficiency of red galaxies at faint magnitudes
compared to Coma
-- A synchronous formation of stars in all red
sequence galaxies is ruled out, and the comparison
with Coma quantifies the effect as a function of
galaxy magnitude
De Lucia et al. 2004
Observing
late star-forming faint
galaxies becoming
“dwarf ellipticals”
About 10% of the dwarf
cluster population in
the Coma cluster
(see also Tran et al.
2003, Caldwell et al.’s
works, De Propris et
al.)
Poggianti et al. 2004
Downsizing-effect
Going to lower redshifts, the maximum luminosity/mass of
galaxies with significant SF activity progressively decreases.
Active star formation in low mass galaxies seems to be (on
average) more protracted than in massive galaxies.
IN ALL ENVIRONMENTS.
Mass downsizing: Fundamental Plane (Treu et al 2005)
142 spheroidals: HST-derived scale lengths, Keck dispersions
Increased scatter/deviant trends for lower mass systems?
If
log RE = a log s + b SBE + 
Effective mass
ME   2RE / G
So for fixed slope, change in FP intercept i log (M/L)i
Evolution of the Intercept  of the FP
1-3%
of trend:
mass growth
in massive(>10^11.5)
galaxies sinceassembly
z=1.2 –
Strong
lower mass
systems more scatter/recent
20-40% at lower masses
Stellar Mass Functions by Type in GOODS N/S
• No significant
evolution in massive
galaxies since z~1
• In fact, almost no
change in total mass
function with time
above 5 X10^10,
indicating little mass
growth at the high
mass end
2dF
• Bulk of evolution is
in massive Irrs/spirals
Bundy et al (2005) Ap J
634,977
(h=1)
Cimatti et al. 2006 and Brown et al. 2006 emphasize
that, if only a factor of two in mass is added to the
red sequence since z~1, and it is mainly in lower
luminosity (< 1011Msun) galaxies, then simple “running
down” of star formation in disk galaxies, turning them
red, can account for the growth.
A key point to be resolved, and one that may be
telling as to how much the hierarchical picture is in
trouble.
Redshift >1.5 – How many massive galaxies at z=2?
Pioneering study: N=737, H<26.5, zphoto<3, 5 arcmin2
2dF
H=26.5
 SFR(z)
incomplete
 of H-faint low mass galaxies z>1.5
Significant uncertainty estimating contribution
50% of the assembled mass is only in place at a surprisingly low redshift z~1
Integrated SFH underestimates mass assembly: dust, cosmic variance?
Similar HDF-S analysis by Rudnick et al 2003 Ap J 599, 847
Dickinson et al. 2003
Gemini Deep Deep Survey: Stellar Masses
Color pre-selected spectroscopic
sample K<20.6, I<24.5
N=240 in 430 arcmin2 fields
0.5<z<2
Surprising abundance of massive
galaxies at z>1.5
Many are `red and dead’
Glazebrook et al Nature 430, 181 (2004)
Gemini Deep Deep Survey: Slow Mass Assembly
Growth rate slower than semi-analytic models (without AGN feedback)
Rate ~independent of mass so problem for M > M10.5 particularly acute
Glazebrook et al Nature 430, 181 (2004)
Census of Stellar Mass 2<z<3
LBG
DRG
Most M>1011M galaxies are DRGs(77%) - LBGs constitute only 17%
No single technique complete in estimating assembly history
van Dokkum et al 2006
Bower et al 2006, MNRAS 370, 645, “Breaking the hierarchy
of galaxy formation” + Springel et al. 2005, Croton et al. 2006,
Granato and collaborators
Works out a model of ending star formation early by AGN
heating, claiming to restore CDM hierarchical clustering to
good working order.
Sijacki & Springel, 2006 MNRAS, 366, 397
Summary
• Techniques are now well-established for estimating the stellar masses
of galaxies to high redshift; reliability depends on having spectroscopic
redshifts and long wavelength data
• It is now clear that mass assembly since z~2 does not proceed
hierarchically; growth is suppressed in high mass systems at early
times continuing in low mass systems to z~0 (`downsizing’)
• The mass downsizing parallels the star formation downsizing: SF is
quenched above a certain threshold mass whose value declines with
time
• AGN feedback may be able to reproduce this behavior in CDM
models, but further work is needed to understand environmental
dependence of this process: are downsizing trends occurring at a
different rate in clusters vs `field’?
• Massive galaxies are now being found at z>2 in surprising numbers;
many are already passively evolving. This implies much SF activity at
higher redshift