Galaxy Formation and Evolution in Clusters

1
Galaxy Formation and Evolution (in Clusters) 2

• Alice Shapley (Princeton)
• June 14, 15, 16th, 2006

2
Overview and Motivation

• 0. Introductory remarks about galaxy
  formation evolution and why clusters are useful
  for this
• Galaxy evolution in clusters from z0-1
  (emphasis on early-type)
• Galaxy evolution in general from z0-1
• Direct observations of cluster galaxy
  progenitors forming at high redshift (z 2)
• Protoclusters at high redshift (z 2)

3
Recap..
4
Different Types of Galaxies

• Galaxies observed in different forms
• Divide by morphology, color, spectra
• E.g., morphological type E/S0 vs. spiral
• 80 of galaxies in cores of nearby clusters are
  E/S0 (Dressler 1980)
• Galaxies of different types have different
  formation mechanisms

Strateva et al. (2001)
5
History of Galaxy Evolution

• Traditionally, galaxies used to constrain
  cosmological model (Sandage 1961)
• Cosmological tests compare measure of distance
  and redshift e.g., apparent magnitude, m, of
  standard candle (with known M) vs. redshift

Initially used cluster elliptical galaxies as
standard candles
6
History of Galaxy Evolution

• Not only were galaxies brighter in the past
  (i.e., at higher z, M was brighter), but, Tinsley
  (1976) pointed out uncertainties in dM/dt
  translate into unacceptable uncertainties in q0
  (form of IMF, metallicity, star-formation
  history)
• In order to constrain q0, need to know
  evolutionary correction to high precision

Dont use elliptical galaxies to measure cosmic
deceleration!!!!
7
Models of Galaxy Evolution
Of course, there is mutual uncertainty
uncertainty in evolution of galaxies hinders
interpretation of cosmological tests BUT
uncertainty in cosmological model hinders
interpretation of galaxy evolution data
Population synthesis models tell how SED evolves
with TIME -- but we observe galaxy mags, colors,
spectra at different REDSHIFTS Growth of
structure (e.g., the halo mass function and its
evolution) depends on cosmological parameters
8
Models of Galaxy Evolution
Of course, there is mutual uncertainty
uncertainty in evolution of galaxies hinders
interpretation of cosmological tests BUT
uncertainty in cosmological model hinders
interpretation of galaxy evolution data
Population synthesis models tell how SED evolves
with TIME -- but we observe galaxy mags, colors,
spectra at different REDSHIFTS Growth of
structure (e.g., the halo mass function and its
evolution) depends on cosmological parameters
9
Models of Galaxy Evolution

• Late 1970s, motivation for studying distant
  galaxies became not only for cosmological probes,
  but rather for understanding their history and
  formation
• Two basic paradigms for understanding galaxy
  formation
• ? Monolithic collapse
• ? Hierarchical structure formation

10
Monolithic Collapse

• Eggen, Lynden-Bell Sandage (1962) observed
  that metal-poor halo stars in the Milky Way have
  highly elliptical orbits characteristic of system
  in free-fall
• Metal-rich stores have more disk-like
  distribution and kinematics

Color (bluer, metal poorer)
11
Monolithic Collapse

• Interpretation 1010 years ago protogalaxy
  collapsed from intergalactic material, collapse
  was rapid (108 years for equilibrium to be
  reached), big burst of star-formation, formed
  stars with eccentric orbits during collapse, disk
  stars formed later

Color (bluer, metal poorer)
Monolithic collapse classical formation
mechanism for ellipticals and bulges, which are
collections of old stars
12
Hierarchical Stucture Formation

• Whereas monolithic collapse works backwards from
  present using understanding of stellar evolution
  and stellar dynamics (whats the cosmological
  model?), hierarchical structure formation works
  from within the ?CDM cosmological framework,
  provides ab initio model for galaxy formation,
  motivated by CMB and large-scale structure
• Galaxy formation and evolution is a natural
  consequence of the growth of the power spectrum
  of fluctuations by gravitational instability, in
  a universe dominated by dark matter
• Model predicts evolution of dark matter halo
  mass function through merging and accretion

(Springel et al. 2005)
13
Hierarchical Stucture Formation

• While the evolution of the dark matter is now
  fairly well understood (gravity, cosmological
  model), tracing the evolution of the baryons is
  complicated!
• Gas cooling and other hydrodynamical effects,
  star formation and IMF, feedback (from AGN and
  supernovae)
• Unfortunately, it is all these messy baryonic
  processes that translate the population of dark
  matter halos into the galaxies that we observe
  over a range of cosmic epochs.
• Big question whats the best way to constrain
  the baryonic physics of galaxy formation, now
  that there appears to be agreement on underlying
  cosmological model?

14
Hierarchical Stucture Formation A comment on the
meaning of formation

• Possible difference between redshift at which
  XX of stars formed vs. redshift at which XX of
  stars were assembled into one unit

From de Lucia et al. (2005), on formation of
elliptical galaxies Semi-analytic model (w/AGN
feedback) grafted onto Millennium DM Simultion
Star-formation
Mass Assembly
15
Why Clusters are Useful

• Clusters useful for galaxy evolution studies
  because (based on various identification
  techniques X-ray, optical/red-sequence, lensing,
  SZ) a cluster provides a large samples of
  galaxies at the same redshift and relatively
  compact field, now to z1.45 (Stanford et al.
  2006)
• Also, close proximity of galaxies with each
  other and ICM allows for study of environmental
  effects in high density environments
  (gravitational and hydrodynamical)
• Complexities to join in timeline, need to
  understand how clusters at high redshift relate
  to clusters at lower redshift (e.g., in terms of
  mass) -- also need to understand variation in
  cluster populations at each redshift before
  connecting cluster galaxies at different redshifts

16
I. Galaxy Evolution in Clusters from z0-1
17
Evolution of E/S0 galaxies

• Collectively refer to E/S0 as early-type
  galaxies (may be some ambiguity in classifying
  each type), the type that make up 80 of galaxy
  population in cores of nearby clusters. HST
  important for morph. Classification at higher
  redshift.
• Evidence that stars in these galaxies formed at
  zgt2 (evidence for passive evolution? monolithic
  collapse?)
• ? Evolution of colors
• ? Evolution of Color-Magnitude (CM) relation
• ? ?M/LB from evolution in Fundamental Plane

18
Evolution of E/S0 galaxies

• Early-type galaxies in local clusters form a
  homogeneous class

• Color-magnitude diagram in Virgo/Coma scatter
  is 0.05 mag, of which 0.03 mag is observational
  error
• Physical sequence is increasing metallicity at
  increasing mass
• Small scatter around relation implies that stars
  (galaxies) formed at zgt2
• (Bower et al. 1992)

19
Evolution of E/S0 galaxies

• Early-type galaxies in local clusters form a
  homogeneous class

• Color-magnitude diagram in Virgo/Coma scatter
  is 0.05 mag, of which 0.03 mag is observational
  error
• Physical sequence is increasing metallicity at
  increasing mass
• Small scatter around relation implies that stars
  (galaxies) formed at zgt2
• (Bower et al. 1998)

20
Evolution of E/S0 galaxies

• Early-type galaxies in local clusters form a
  homogeneous class

• z0 Fundamental Plane
• Relationship among velocity dispersion, surface
  brightness, and effective radius
• Implies M/L?M0.24 with small scatter (20),
  which also implies small age scatter at fixed mass

(Jorgensen et al. 1996)
21
Evolution of E/S0 galaxies

• Evolution in E/S0 colors vs. z can give us some
  clues about early-type galaxy formation

• Ellis et al. (1997) look at CM relation in
  clusters at z0.5 (use HST for morphological
  separation)
• CM-relation has same slope at z0.5 as z0, small
  scatter, which does not increase at fainter
  magnitudes
• Tight scatter at z0.5 can be understood if bulk
  of sf occurred 5-6 Gyr ago zgt2

Central 1 Mpc
z0.56
22
Evolution of E/S0 galaxies

• Evolution in E/S0 colors vs. z can give us some
  clues about early-type galaxy formation

• Ellis et al. (1997) look at CM relation in
  clusters at z0.5 (use HST for morphological
  separation)
• CM-relation has same slope at z0.5 as z0, small
  scatter, which does not increase at fainter
  magnitudes
• Tight scatter at z0.5 can be understood if bulk
  of sf occurred 5-6 Gyr ago zgt2

(images are 10x10 or 60x60 kpc)
23
Evolution of E/S0 galaxies

• Evolution in E/S0 colors vs. z can give us some
  clues about early-type galaxy formation

Coma CMD w/0 color evolution

• Ellis et al. (1997) look at CM relation in
  clusters at z0.5 (use HST for morphological
  separation)
• CM-relation has same slope at z0.5 as z0, small
  scatter, which does not increase at fainter
  magnitudes
• Tight scatter at z0.5 can be understood if bulk
  of sf occurred 5-6 Gyr ago zgt2

z0.56
24
Evolution of E/S0 galaxies

• Evolution in E/S0 colors vs. z can give us some
  clues about early-type galaxy formation

• Stanford et al. (1998) look at clusters at
  0.3ltzlt0.9, (again use HST for morph.
• Colors get bluer consistent w/ expectations from
  passive evolution, roughly independent of cluster
  props., CMD slope does not evolve (CMD is M-Z),
  nor does scatter
• Again, consistent with stars being formed in
  single episode at high redshift, relative age
  spread low

slope
scatter
25
Evolution of E/S0 galaxies

• Evolution in E/S0 M/LBvs. z can give us some
  clues about early-type galaxy formation

From Treu et al. (2005)

• Offset in FP 0-pt indicates difference in M/LB
  (see problem)

26
Evolution of E/S0 galaxies

• Evolution in E/S0 M/LBvs. z can give us some
  clues about early-type galaxy formation

Statistics at z1 not great!

• van Dokkum Stanford (2003) look at cluster at
  z1.27, spectra for 3 galaxies, see how they
  relate to local fundamental plane. Offset in FP
  0-pt indicates difference in M/LB
• Mean star-formation age higher than z2

27
Evolution of E/S0 galaxies

• Some unresolved questions at z1

• De Lucia et al. (2004) construct CM relations for
  4 z0.7-0.8 EDisCS clusters, find deficit of
  faint red galaxies, relative to Coma, important
  implications for formation of fainter red
  galaxies
• BUT Andreon et al. (2005) analyze MS 1054-083 at
  z0.83 and find no deficit
• Interloper corrections!

28
Evolution of E/S0 galaxies

• Some unresolved questions at z1

• Homeier et al. (2006) measure CM-relation in
  clusters at z0.9 (part of supercluster), and
  find evidence for scatter increasing at fainter
  magnitudes, consistent with younger ages

29
Evolution of E/S0 galaxies

• Some unresolved questions at z1

• What about field E/S0 galaxies? Treu et al.
  (2005) find difference in M/LB evolution stronger
  for less massive morphologically-selected E/S0
  galaxies over redshift range z0.3-1.2
• Note evidence at z0.4 that field E/S0 are
  younger by 20 than cluster E/S0, zformgt1.5 (van
  Dokkum et al. 2001)

30
Evolution of Galaxy Mix

• While cluster E/S0 appear homogeneous, with
  stars formed at high redshift and passively
  evolving, there is evidence that cluster galaxy
  population mix is evolving
• Multiple ways to consider this, historically,
  which are all correlated
• Evolution in morphological mix (morph-dens
  relation)
• Evolution in mix of red/blue galaxies
  (Butcher/Oemler)
• Evolution in spectral types of galaxies
  (em/abs/EA)

31
Different Types of Galaxies

• Galaxies observed in different forms
• Divide by morphology, color, spectra
• E.g., morphological type E/S0 vs. spiral
• 80 of galaxies in cores of nearby clusters are
  E/S0 (Dressler 1980)
• Galaxies of different types have different
  formation mechanisms

Strateva et al. (2001)
32
Evolution of Galaxy Mix MD

• Morphology-Density Relation in the local
  universe, the fraction of E/S0 galaxies is higher
  in clusters than in less dense environments
  (Dressler 1980)

X-axis is /Mpc2, field is lt10/Mpc2
33
Evolution of Galaxy Mix MD
(Dressler et al. 1997)

• MD relation is present at z0.5, but is different
  from z0 relation
• z0.5 S0 fraction is lower than in clusters at
  z0, while proportion of spirals is higher
• Suggests S0 galaxies in clusters may have evolved
  from spirals

34
Evolution of Galaxy Mix MD

• Out to z1, Smith et al. (2005) find that the
  E/S0 fraction increases from 0.7 to 0.9 in
  highest-density regions
• At lowest densities (field), E/S0 fraction is
  constant at 0.4
• Model early-type population at z1 made up of
  E, subsequent evolution is from transformation of
  infalling spirals into S0 (what about low-density
  environments?)

35
Evolution of Galaxy Mix BO
Coma z0.02
Cl 00241624 z0.4
Solid E Hatch S0 Clear Sp

• Butcher Oemler (1978) looked at two rich
  clusters at z0.4.
• Color distributions are strikingly different
  from that in Coma, in that 1/3-1/2 of galaxies in
  these clusters have colors of spiral galaxies

36
Evolution of Galaxy Mix BO

• What is nature of blue-type galaxies? Spirals
  and irregulars and post-starburst galaxies
  (strong Balmer absorption lines)
• Infall of blue, late-type galaxies from the
  field, which subsequently lose fuel in denser
  environment

Compilation by van Dokkum (2001), trend with lots
of scatter
37
Evolution of Galaxy Mix Spectra

• Composite spectra of five clusters at z0.5 and
  z0.0, see OII in z0.5 spectra. For example in
  MORPHS sample of 10 z0.4-0.5 clusters, 30 of
  bright cluster galaxies have emission lines
  (Dressler et al. 1999), BUT varies from cluster
  to cluster (IMPORTANT ISSUE!)

38
Evolution of Galaxy Mix Spectra

• The spectrum on top is a post-starburst or
  EA, or KA spectrum.
• The most striking feature is strong Balmer
  absorption lines. What does that mean?
• This spectrum is viewed as a sign that
  star-formation recently stopped

(Poggianti 2004)
39
Evolution of Galaxy Mix Spectra

• Difficult to quantify fraction of emission-line
  galaxies vs. redshift and cluster properties,
  need larger sample of clusters
• SFR above estimated from H? NB imaging (Finn et
  al. 2004)

40
Selection Effects/Biases

• How were clusters selected? (optical, X-ray, SZ)
  Is the measured property correlated at all with
  the cluster selection (B-O in optically-selected
  clusters)
• Morphological classification? How robust is
  this? Especially at high-z. HST is required.
• If galaxy properties depend on cluster
  properties, must understand scatter from cluster
  to cluster at any given redshift
• How do you deal with interlopers?
• Dust???

41
Transformation mechanisms

• Consider both gravitational and hydrodynamical
  effects
• Mergers and strong galaxy/galaxy interactions.
  Most efficient when relative galaxy velocities
  are lower than seen in big clusters in groups
• Harassment Tidal forces from high-speed close
  (50kpc) encounters, effects especially important
  on smaller galaxies (in encounters with larger
  galaxies) and in clusters (1/Gyr), lead to
  disturbed spirals and starbursts, prolate
  morphology with no further star formation (Moore
  et al. 1996)

42
Transformation mechanisms

• Consider both gravitational and hydrodynamical
  effects
• Ram-pressure stripping. Interaction between
  galaxy and ICM. ISM of galaxy can be stripped,
  depends on ICM density and speed of galaxy
  (P?ICMv2), so only important for galaxies
  passing through cluster core (short timescale 107
  yr)
• Strangulation Removal of reservoir of gas
  that can cool and become available for star
  formation, once the galaxy enters more massive DM
  halo, (timescale longer timescale 109 yr),
  incorporated in SAM
• How important is each of these? (e.g. RPS)

43
Transformation mechanisms

• Ram-pressure stripping. Interaction between
  galaxy and ICM. ISM of galaxy can be stripped,
  depends on ICM density and speed of galaxy
  (P?ICMv2), so only important for galaxies
  passing through cluster core, disk gas gets
  stripped
• (simulation by Quilis et al. 2000)

44
Progenitor Bias

• Contradictory information? Cluster E/S0
  consistent with passive evolution from z1-0.
  Yet, we know there is morphological/color/spectral
  transformation (MD, B-O effect)
• Progenitor Bias states that progenitors of
  youngest E/S0 at low redshift are not classified
  as E/S0 at higher redshift, e.g. z1, leads to
  artificially slow evolution in colors and low
  color scatter at high redshift, and causes an
  overestimate of redshift of when bulk of stars
  formed
• High redshift sample not fair comparison w/ low
  redshift

45
Progenitor Bias
(van Dokkum Franx 2001)
E/S0 progenitors
All progenitors
(Left) Lines shows evolution of M/LB for
galaxies. Only classified as E/S0 when
solid. (Right) Solid line shows evolution in mean
M/LB of E/S0, comparable to single galaxy formed
at very high redshift (dashed), much faster
evolution for all progenitors of z0 E/S0
(long-dashed)
46
Progenitor Bias
(van Dokkum Franx 2001)

• As M/LB compared for E/S0 at different
  redshifts, different sets of galaxies are
  included, evolution can be misinterpreted (i.e.
  simple passive evolution, too high redshift for
  star-formation stopping)
• Simple, analytic model. Needs to be put in
  cosmological context (i.e. do numbers work out?)

47
Concluding Philosophical Comments

• Given many different ways of selecting clusters,
  must understand how observed evolution depends on
  cluster properties, other selection effects
• Many observational questions not resolved yet
  (i.e. is faint end of CM-relation populated in
  clusters at z0.8? How effective are various
  classification schemes vs. z? How does CM-scatter
  evolve to z1? Need statistical sample of FP
  measurements at z1.)
• Many theoretical questions not resolved yet in
  order to gauge importance of transformation
  processes, simulations must be form a realistic
  spiral disk ab initio, and model star-formation
  and feedback correctly in full cosmological
  context -- a tall order

48
Concluding Philosophical Comments

• But, now that we have cosmological framework, we
  can understand how mass builds up in clusters as
  a function of redshift, and interactions among
  dark matter halos
• Figure out which are robust predictions for
  galaxy evolution from cosmological simulation,
  and which are more uncertain (i.e. baryons,
  star-formation, feedback), and what is the best
  way to test them
• Also place these results in a more general
  observational context

49
II. Galaxy Evolution in General from z0-1
50
Global Galaxy Evolution from z1 (averaged over
all environments)

• How do we understand z0 galaxy population as
  the descendants of objects at z1? (new z1
  surveys, e.g. COMBO-17, DEEP2)
• Bimodality in the z0 population
• Evolution in the luminosity function/density,
  for red/blue, or E/S0
• Perhaps more fundamental evolution in the
  stellar mass function/density, and number density
  of galaxies vs. mass
• Other ways of estimating the importance of
  mergers for the evolution in mass functions pair
  counts-gtred-galaxy mergers

51
Evolution of sfr density

• One thing agreed on the sfr density in the
  universe has significantly declined since z1
  (many references)

(from Bouwens et al. 2005)
52
Galaxy population at z0 Bimodality

• SDSS sample of 183,000 galaxies at z0.0-0.2
• Blanton et al. (2003) show distributions in abs.
  mag, colors, surface brightness, light profile
• Bimodality in G-R color (blue/red galaxies)

53
Galaxy population at z0 Bimodality

• Contours indicate densities of 150,000 SDSS
  galaxies in color-magnitude space from Strateva
  et al. (2001)
• (Left) Spectroscopic classification Triangles
  are early-type galaxies open squares are
  late-type
• (Right) Morphological classification Triangles
  are early-type open squares are late-type

54
Galaxy population at z0 Luminosity and Mass
Functions

• Polo reviewed LF
• Cole et al. (2001) construct stellar mass
  function from K-band luminosity function of
  2dF/2MASS galaxies and population synthesis
  models of opt/IR colors, gt1000 sq degrees
• Find
  ?stars6x108M?/Mpc3
• ?stars0.004

Local benchmark
55
Galaxy population at z1

• Bell et al. (2004) use 25,000 COMBO-17 galaxies
  out to z1, covering 0.8 sq. degrees, and
  photometric redshifts
• high redshift galaxies show CMD
• look at evolution of B-band LF of red-sequence
  galaxies (i.e. early-type, E/S0)

56
Galaxy population at z1

• Find B-band luminosity density of red-sequence
  galaxies is constant out to z0.8
• If z0 red-sequence galaxies had all formed at
  higher redshift and evolved passively to z0,
  expected luminosity would have been factor of 2
  higher
• M/L lower at z1, so same luminosity means less
  mass

57
Galaxy population at z1

• Where does this increase in stellar mass for
  luminous red galaxies increase come from?
• Dissipationless merging, truncation of SF in
  some fraction of blue population
• Note there arent enough luminous blue galaxies
  that can fade to become luminous red galaxies
• Big uncertainty COSMIC VARIANCE!

58
Galaxy population at z1

• DEEP2 redshift survey finds the same thing
  (Faber et al. 2006)
• Cimatti et al. (2006) find different result
  less evolution in mass density of massive red
  galaxies (mass dependent evolution)

59
Galaxy population at z1
DEEP2 SDSS shifted to z1

• Another way of looking at it Blanton (2005)
  takes the SDSS sample at z0.1 and observes it
  at z1, assuming no evolution, just
  k-corrections. Then he compares the colors and
  mags with those of the DEEP2 z1 survey.

60
Galaxy population at z1

• In 28x28 GEMS area (deep HST imaging), Bell et
  al. (2005) find pairs of red galaxies (criteria
  for calling it a red merger), luminosity ratios
  lt41, estimate timescale over which it would be
  identified as such and translate counts-gt merger
  rate, and of major mergers experienced by red
  galaxy since z0.7

61
Galaxy population at z1

• Conclusion luminous (MVlt-20.5) present-day
  early-type galaxy experiences 0.5 to 2.0 such
  major mergers since z0.7, van Dokkum (2005)
  finds similar result
• Is this consistent with other observations?
  Models? Uncertainty in merger timescale
  important!

62
Evolution of stellar masses

• Luminosity density and pair counts are both
  methods of looking at evolution, but there is a
  more direct method
• Distribution of stellar masses (of different
  types of galaxies) as a function of redshift
  total stellar mass density as a function of
  redshift abundance of objects of a given stellar
  mass
• (stellar mass of an object can only increase,
  unlike luminosity in a given band)
• With the advent of wide-field and deep K-band
  imaging, estimates of stellar masses for high
  redshift galaxies becomes possible

63
Evolution of stellar masses

• Drory et al. (2005) find that 50 of z0
  stellar mass density in place at z1, 25 at z2
• They estimate stellar mass density by
  constructing stellar mass function
• Compare stellar mass evolution with integral of
  ?sfr(z)

64
Evolution of stellar masses

• Drory et al. (2005) find that 50 of z0
  stellar mass density in place at z1, 25 at z2
• How does it compare wth integral of sfr-history?
• How do you make that plot? How might points at
  high-z need to be adjusted?

65
Evolution of stellar masses

• Bundy et al. (2006) analyze gt8000 DEEP2 data
  with K-band observations over 1.5 sq. degrees,
  model stellar mass functions of red/blue galaxies
  vs. redshift
• Largest set of z1 galaxies with stellar masses
  and redshifts, so they can divide up sample
  (larger area and fainter limit than Drory et al.)

66
Evolution of stellar masses

• Similar result, but emphasize lack of
  significant evolution of total mass function
• Also, favor truncation of star-formation in blue
  galaxies, rather than dry mergers, for explaining
  evolution of red galaxy mass function
• Furthermore, find more massive red galaxies
  assembled first

67
Hierarchical Stucture Formation

• Prediction by de Lucia model more massive
  elliptical galaxies assembled later than less
  massive ones

From de Lucia et al. (2005), on formation of
elliptical galaxies Semi-analytic model (w/AGN
feedback) grafted onto Millennium DM Simultion
68
Concluding Philosophical Comments

• There is not consensus about the evolution of
  stellar mass density and number of red (E/S0)
  galaxies as a function of mass the importance of
  dissipationless (stellar only, dry) mergers vs.
  truncation of star-formation in blue galaxies --
  even for people analyzing the same datasets
• These are important quantities to pin down
  observationally if we are going to constrain
  theories of galaxy formation
• Note we didnt talk about the evolution of
  blue/disk galaxies, or about metals, or about
  clustering