Interpreting stellar populations in a cosmological context

1
Interpreting stellar populations in a
cosmological context

• rachel somerville
• MPIA

with thanks to the GOODS GEMS teams, S. Faber,
B. Allgood, J. Primack, A. Dekel, R. Wechsler
2
Stellar populations can be used to weigh
galaxies
Bell et al. 2003
Papovich et al. 2002
3
massive galaxies (both old/evolved and dusty/star
forming) are being discovered in significant
numbers at redshifts as high as z2
stellar mass
Fontana et al. 2004 (K20) Glazebrook et al. 2004
(GDDS) Brinchmann Ellis 2000 Cohen et al.
2000 Rudnick et al. 2004 (FIRES) Drory et al.
2004 (MUNICS) van Dokkum, et al. 2004
Dickinson et al. 2003 (HDFN)
4
Do massive galaxies at high redshift pose a
crisis for CDM?
local galaxies mgt2.5E10 Msun mgt1.0E11 Msun
LBGs
K20
EROs
sub-mm
these kinds of observations could refute CDM,
but so far they do not pose a problem. n.b. all
theorists agree on this
SDSS QSOs
5
the overcooling problem
halo mass function coolingSF squelching SN
FB merging suppressed in clusters
need to suppress cooling and/or star formation
in massive halos to fit z0 stellar mass
function and luminosity functions
6
Stellar mass assembly history comparison with
LCDM models
Glazebrook et al. 2004
Fontana et al. 2004
7
stellar mass assembly history
good agreement with observational
estimates Glazebrook et al. (GDDS) Rudnick et
al. (FIRES) Dickinson et al. (HDFN) Fontana et
al. (K20) Borch et al. (COMBO-17) Somerville et
al. (GOODS)
Tecza et al. 2003 (SMGs)
IMFKroupa
8
SAMs vs. N-bodyhydro

• most hydro simulations overpredict r today
  because feedback is relatively ineffective
• when strong outflows included, results agree well
  with SAMs (e.g. Springel Hernquist, Nagamine
  et al.)

semi-analytic models
N-bodyhydro Springel Hernquist Nagamine
et al.
9
We use stellar populations to trace the star
formation history

• models do well at reproducing optically
  identified star forming populations at z3-6
  (LBGs) global SFR
• difference between SAM and SPH (Springel
  Hernquist) at zgt3 is due to small mass galaxies

SPH
SAM
10
why do galaxies come in two basic types?

• spheroidal, dynamically hot
• red colors
• strong absorption lines
• predominantly old stars
• little recent star formation

• thin disk
• dynamically cold
• supported by rotation
• blue colors
• strong emission lines
• broad range of stellar ages,
• ongoing star formation

11
galaxy colors (and many other properties) are
strongly bimodal
red
color
blue
bright
faint
luminosity
SDSS
Baldry et al. 2003
12
red
color
blue
bright
faint
luminosity
SDSS
Baldry et al. 2003
13
The two types are divided by a critical mass
old, no recent star formation, high
concentration/surface brightness
3x1010 Msun
old
young
young, recent star formation, low
concentration/surface brightness
Kauffmann et al. 2003
14
what is the role of environment?
increasing density--gt
the color of the red sequence is
almost independent of environment but the
fraction of galaxies in the red sequence vs.
the blue cloud is a strong function of local
density
decreasing luminosity--gt
u-r
(u-r)
Balogh et al. 2004
15
the red sequence color bimodality seen at z1!
rest U-V color
also seen in the DEEP2 redshift survey (Willmer
et al. in prep)
rest V magnitude (luminosity)
Bell et al. 2003
16
in hierarchical models, merger history
determines galaxy morphology
Milky Way galaxy
cluster of galaxies
17
Color-magnitude distribution
SDSS
SAM
18
predicted color distributions are not bimodal
-21.5
-20.5
-22.5
black SDSS purple SAM
-18.5
-19.5
19
model prediction color-magnitude relation at
high redshift
rest U-V color
colored points meet Rlt24 COMBO-17 selection
criterion
rest V magnitude (luminosity)
20
rest U-V color
rest V magnitude (luminosity)
Bell et al. 2003
21
models produce enough bright/massive/bulge dominat
ed galaxies -- but they are too blue
red B/Tgt0.5 blue B/Tlt0.5 cyan tmrg lt 0.5 Gyr
red E/S0 blue S/Irr cyan merger
GEMS
22
not enough EROs
13.5 5.8 3.2 1.0 0.5 0.1
KABlt22
GOODS
rss et al. 2004 GOODS ApJL
23
Results from state-of-the-art numerical
hydrodynamic simulations are very similar
Dave et al., see also Nagamine et al.
24
Why are red galaxies red?

• CDM models produce enough old, massive galaxies.
  the problem is a continuous trickle of star
  formation
• there must be some process that shuts off star
  formation after galaxies have become massive
• this process must be rapid, and seems to be
  connected with the presence of a spheroid
• must work in all environments, but happen to a
  larger fraction of galaxies in dense places

25
toy models

• remove all remaining gas after major mergers
• shut off cooling/SF when MhgtMcrit
• shut off star formation when MgtMcrit
• shut off star formation when M,bulgegtMcrit

26
toy models

• remove all remaining gas after major mergers
• has almost no effect (fresh gas gets accreted)
• shut off cooling/SF when MhgtMcrit
• kills massive galaxies entirely does not produce
  bimodality
• shut off star formation when MgtMcrit
• kills massive galaxies entirely does not produce
  bimodality
• shut off star formation when M,bulgegtMcrit

27
Color-magnitude distribution
SAM gas ejected after major merger
SDSS
28
Color-magnitude distribution
SDSS
SAM SF shut off when MhgtMcrit
29
SF quenched when MhgtMcrit
Mrlt-22.75
(purpleSAM blackSDSS)
-21.75
-20.75
-19.75
-18.75
30
Color-magnitude distribution
SDSS
SAM SF shut off when MbulgegtMcrit
31
Metallicity normalization increased by a factor
of 2
SDSS
SAM SF shut off when MbulgegtMcrit
32
SF quenched when MbulgegtMcrit
Mrlt-22.75
(purpleSAM blackSDSS)
-21.75
-20.75
-19.75
-18.75
33
when do galaxies become quenched?
SF quenched when MbulgegtMcrit
34
Mbulge quenched model
dry mergers?
GEMS
35
AGN the missing link?

• tight observed relation between Mbulge and MBH
• energy emitted expected to be proportional to MBH

Di Matteo, Springel Hernquist 2005
36
AGN feedback by momentum-driven winds
SDSS transition mass
Murray, Quataert Thompson 2004
bulge
fg0.1 fg0.05
BH
observed MBH-s rln
37
caveats

• radiation pressure only one of many physical
  mechanisms whereby AGN can couple to gas
• spherical symmetry assumed -- unrealistic?
• constant opacity assumed -- in reality, a
  function of metallicity
• constant relationship between Mbulge MBH
  assumed at all redshifts

38
momentum wind model
cold gas ejected (and never re-accreted) if
MbulgegtMcrit(s)
still have a cooling flow problem!
39
momentum wind model
cold gas ejected (and never re-accreted) if
MbulgegtMcrit(s)
still have a cooling flow problem!
40
AGNmomentum wind model
-22.75
red sequence improved, and bimodality appears in
the right place, but too many intermediate
luminosity blues still have a cooling flow
problem
-18.75
41
momentum wind model
cold gas ejected (and never re-accreted) if
MbulgegtMcrit(s) cooling shut off in halos with
Vcgt350 km/s standard merging prescription
still have a cooling flow problem!
42
AGN feedback,cooling cutoff,merging
43
AGN-feedback model
too much scatter in red sequence at
high redshiftformation time too late or
too spread out
44
AGN feedback model
too much scatter in red sequence at
high redshiftformation time too late or
too spread out
45
Effervescent heating by giant radio jets

• recent work suggests even columnated jets can
  heat a large filling factor of ICM
• resulting bubbles look similar to those seen in
  Chandra images of some clusters
• Effective in cluster or perhaps group environments

Bruggen, Ruszkowski Hallen 2005
46
runaway QSO growth

• number density of bright QSOs does not turn over
  at low redshift unless an adhoc scaling in
  accretion efficiency is applied, or MBH-s
  relation enforced by hand (e.g. Kauffmann
  Haehnelt Volonteri et al. Wyithe Loeb)

Bromley, rss Fabian 2004, 2005
47
Stellar Populations as fossil relics of star
formation
10 realizations of a Coma cluster
48
real vs. grid-derived age and metallicity
Z from grids
age from grids
actual light-weighted age
actual metallicity
49
SAM Coma
Trager et al. Coma data
50
Dry mergers simulations
Bell, Naab, McIntosh, rss et al.
51
Dry mergers GEMS
52

• Dry mergers visible for 250 Myr
• every luminous E has had 0.5-1 dry merger since
  z1
• in good agreement with expectations from
  hierarchical models

53
Summary

• CDM-based models of galaxy formation that produce
  reasonable agreement with the z0 stellar mass
  function form enough massive galaxies at high zlt2
• But default models do not produce enough massive
  red galaxies, especially at high redshift,
  because of continuous low level star formation.
  need a new process that quenches star formation
  in massive, bulge-dominated galaxies
• momentum-driven winds powered by AGN a promising
  mechanismanother process needed to solve
  cooling flow problem -- but must make enough
  massive galaxies at high redshift!