The Search for Forming Galaxies

1
The Search for Forming Galaxies

• Chris ODea
• Space Telescope Science Institute

• Acknowledgements
• Mauro Giavalisco
• Harry Ferguson

2
Outline

• Hierarchical Galaxy Formation
• Star Formation Stellar Evolution
• Searches for Forming Galaxies
• Narrow Band Optical Searches
• GPS Quasars
• High-Z Radio Galaxies
• The Hubble Deep Fields
• Lyman-Break Galaxies
• Sub-mm/IR
• Star Formation History of the Universe

3
Hierarchical Galaxy Formation
(Virgo consortium)
4
Hierarchical Galaxy Formation The Paradigm

• At recombination (z1160), the universe is very
  homogeneous smooth
• There is a spectrum of density perturbations
  gravitational potential fluctuations are
  independent of length scale
• Low mass clumps collapse first and merge to form
  galaxies
• Larger scale structure builds slowly as galaxies
  form - groups, clusters, super clusters.

e.g., Kauffmann etal. 1993, MNRAS, 264, 201
5

• Blow up of dark matter density in the region
  around a rich cluster in a simulation of a ?CDM
  universe at z0.

Jenkins etal 1998, ApJ, 499, 20
6

• Numerical models of structure formation in 4
  cosmologies. (dark matter density is plotted).
• All simulations are normalized to reproduce the
  abundance of rich galaxy clusters today.
• However, the power spectrum of the simulated dark
  matter distribution is not consistent with that
  of observed galaxies.

Jenkins etal 1998, ApJ, 499, 20
7
Star Formation Stellar Evolution
8
Star Formation

• Evolution of the UV-Optical SED of a continuous
  star burst.
• The SED brightens in the UV around 3 Myr and then
  reddens only slightly with time.

1 solar mass/yr with solar metals and Salpeter
IMF 1-100 M? (Starburst99 code).
9
Star Formation

• Evolution of the UV-Optical SED of an
  instantaneous star burst.
• The SED brightens in the UV around 2 Myr and then
  reddens and fades as the stars evolve.

106 M? burst with solar metals and Salpeter IMF
1-100 M? (Starburst99 code).
10
SED of Instantaneous Burst

• Broadband spectrum of instantaneous burst reddens
  and dims are the population evolves (massive hot
  stars die first).

Devriendt etal. 1999, AA, 350, 381
11
Star Formation in a Merger

• N-Body simulation of evolution of galaxies with
  dusty starbursts showing old stellar population.

Mass distribution of old stars projected onto
(x,y) plane at each time T for the merger model.
Each frame is 105 kpc. Merger is
prograde-retrograde. (Bekki Shioya 2001, ApJS,
134, 241).
12
Star Formation in a Merger

• N-Body simulation of evolution of galaxies with
  dusty starbursts showing gas and new stars.

Mass distribution of gas and new stars projected
onto (x,y) plane at each time T for the merger
model. Each frame is 105 kpc. Merger is
prograde-retrograde. (Bekki Shioya 2001, ApJS,
134, 241).
13
Star Formation in a Merger
Star formation rate depends on the accumulation
of dense gas in the central region.

Time evolution of star formation rate in solar
masses/yr in the merger.
Time evolution of gas mass accumulated within the
central regions.
(Bekki Shioya 2001, ApJS, 134, 241).
14
Star Formation in a Merger

• Time dependence of SED depends on time dependence
  of star formation rate.
• IR and sub-mm luminosity increases during peak of
  star formation (when gas is efficiently
  transported to galaxy center).
• In later stages, gas is rapidly consumed, and UV
  and IR luminosity declines.

Spectral energy distribution of a merger as a
function of time. Model includes gas and dust.
Time given in Gyr. (Bekki Shioya 2001, ApJS,
134, 241). 104 Å 1µ.
15
Star Formation in a Merger

• Effect of dust is to remove UV light and
  re-radiate in the IR.

Spectral energy distribution of a merger (top)
with gas and dust, and (bottom) without.
Corresponds to maximum SFR in the merger. Bekki
Shioya 2001, ApJS, 134, 241. 104 Å 1µ.
16
Integrated Spectra of Galaxies

• Spectra reflect the large difference in SFR as a
  function of Hubble type.

Fluxes Normalized at 5500 Å. (Kennicutt 1992,
ApJS, 79, 255)
17
SRF vs Hubble Type

• Line EQW scales with stellar birthrate parameter
  (b) and Hubble type.

From a large sample of nearby spiral galaxies
(Kennicutt 1998, ARAA,36, 189).
18
Narrow Band Searches

• A proto galaxy forming stars at a rate of 100
  M?/yr should produce a Lya luminosity 1043
  ergs/s (e.g., Thompson etal, 1995, AJ, 110, 963).
• Yet, with some exceptions (see next viewgraph)
  Lya from possible proto galaxies is rarely
  detected in deep narrow band searches (Thompson
  etal 1995 Stern Spinrad, 1999, PASP, 111,
  1475)
• This implies that the galaxies are obscured by
  dust.

19
Extended Lya Emission

• Two large, bright, diffuse Lya blobs in a
  protocluster region at z3.09
• The blobs are similar to those seen around
  powerful radio galaxies, but these are
  radio-weak.
• They could be excited by obscured AGN or they
  could be large cooling-flows.

(Steidel etal, 2000, ApJ, 532, 170)
20
High z GPS Quasars

• A significant fraction of radio-loud quasars at
  high z (gt2) tend to be GPS.
• GPS quasars tend to be at high z (gt2)
• Possibly, the high z quasars are GPS because the
  radio sources are confined to small scales (lt100
  pc) due to dense gas in the host circumnuclear
  region.
• The presence of the dense gas necessary to
  confine a powerful quasar (gt 1010 M?), suggests
  that the host is a proto galaxy.

(ODea 1998, PASP,110, 493)
21
Radio Galaxies

• (Carilli 2000)

22
Radio Galaxies at High z

• Powerful radio galaxies are detectable out to
  high z.
• They are generally bright L Ellipticals with
  old stellar populations rather than proto
  galaxies.

Van Breugel etal. 1999, ApJ, 518, L61
23
The Hubble Deep Fields
24
(No Transcript)
25
HDF Census

• 3000 Galaxies at U,B,V,I
• 1700 Galaxies at J, H
• 300 Galaxies at K
• 9 Galaxies at 3.2mm
• 50 Galaxies at 6.7 or 15mm
• 5 Sources at 850mm
• 0 Sources at 450mm or 2800mm
• 16 Sources at 8.5 GHz
• 150 Measured redshifts
• 30 Galaxies with spectroscopic z gt 2
• lt20 Main-sequence stars to I 26.3
• 2 Supernovae
• 0-2 Strong gravitational lenses
• 6 X-ray sources

Ferguson, Dickinson Williams 2000, ARAA, 38, 667
26
Advantages and disadvantages of a pencil-beam
survey

• Normalized by galaxy luminosity function. Shows
  the number of L volumes.
• Volume is smallest at low z where most of cosmic
  time passes.
• (Ferguson etal. 2000, ARAA, 38, 667)

27
Galaxy Counts

• Galaxy number counts favor ?CDM cosmologies.
• Galaxies are more numerous than simple
  no-evolution models (esp at U)

Ferguson etal 2000, ARAA, 38,667
28
WFPC2 NICMOS Imaging

• Selected galaxies from the HDF-N at a range of z.
  Left B, V, I Right I, J, H.
• Morphologies are similar in both optical and
  near-IR.

Ferguson etal. 2000, ARAA, 38, 667
29
Galaxy Morphologies

• Higher fraction of irregular peculiar galaxies
  than seen locally.
• Qualitatively supports hierarchical galaxy
  formation.
• LSB galaxies and bursting dwarf galaxies dont
  dominate the counts.

Abraham et al. 1996, Baugh et al. 1996, Ferguson
Babul 1998
30
Galaxy Sizes at z3

• The galaxies at z3 are small but luminous, with
  half-light radii 1.8 ltr1/2lt 6.5 h kpc and
  absolute magnitudes -21.5 gt M(B) gt -23.

Blue magnitude vs half-light radius for High-Z
HDF galaxies and a representative sample of local
galaxies. (Lowenthal etal 1997, ApJ, 481, 673)
31
F814W
32
F606W
33
F450W
34
F300W
35
STIS 2300?
36
STIS 1600Å
37
Lyman Break Galaxies
38
Lyman-Break Galaxies

• Color selection of star-forming galaxies from the
  
• 912 Å continuum discontinuity
• Effects of cosmic opacity
• Photoelectric absorption
• Line blanketing
• and moderate dust obscuration
• Makes identification of distant galaxies easy
  with optical/near-IR multi-band imaging
• Very efficient 90 at z3, 50 at z4
• Current best way to test ideas on galaxy
  formation

39
Spectral Features due to Hydrogen
(Valenti 2001)
40
Lyman-Break selection
(Giavalisco 2001)
41
Lyman-Break selection
(Giavalisco 2001)
42

• Expected colors of high z Lyman break galaxies
  are well defined, and not sensitive to reddening.

Steidel etal 1999, ApJ, 519, 1
43
Steidel etal 1999, ApJ, 519, 1
44

• Color color plot of real data.
• 207/29,000 satisfy the color selection criteria.
  
• Blue circles are objects with spectroscopic
  3.7ltzlt4.8. And yellow objects are interlopers.

Steidel etal 1999, ApJ, 519, 1
45
Lyman-Break Technique

• NOT photometric redshift
• Just effective set of selection criteria
• Requires follow-up spectroscopic identification
  to be useful

46

• Keck-LRIS spectra
• Rslt25.5
• Texp2-4 hr
• ??12 Å
• Similar to local SF galaxies
• Richness of features from
• Interstellar gas
• Nebular gas
• Stars
• Presence of OB stars
• Varying Lya

Giavalisco 2001
47

• Keck-LRIS spectra
• Rslt25.5
• Texp2-4 hr
• ??12 Å
• Similar to local SF galaxies
• Richness of features from
• Interstellar gas
• Nebular gas
• Stars
• Presence of OB stars
• Varying Lya

Giavalisco 2001
48
Large survey

• Results of spectroscopic follow up of color
  selected LBGs.
• The two samples are consistent with having
  similar colors.

Steidel etal 1999, ApJ, 519, 1
49
The Nature of LBGs

• What is the link between LBGs and the local
  populations?
• Are LBGs small sub-galactic systems that will
  merge to form more massive galaxies, as predicted
  by hierarchical cosmologies (CDM)?
• What is their mass distribution?
• Regardless, their stars must be old
• Can they be the progenitors of the spheroids?
• What is their metallicity?
• What are their stellar mass and age?

50

• HST morphology
• Observed mostly only faint LBGs (mgtm)
• Small size r1/21-3 kpc
• Dispersion of properties both disk-like and
  spheroid-like observed
• Rest-UV and rest-optical morphologies similar

51
Radial Profile WFPC2 NICMOS
52
The HDF-N HST WFPC2 NICMOS-3
53
The HDF-N HST WFPC2 NICMOS-3
54
Results From Morphology

• Disk-like and spheroid-like structures observed
• Compact and fragmented/irregular/diffuse
  structures observed. Merging?
• Sizes smaller than present-day L galaxies
  similar to big bulges and intermediate-luminosity
  Ellipticals
• No obvious evidence for much older, larger
  structures. UV morph. Opt morph.
• NOTE HST has mostly imaged faint (mgtm) LBGs

55
Observing the Rest-Frame Optical SED

• MOTIVATIONS
• Estimate metallicity (O abundance) from optical
  nebular lines
• Estimate dynamics (hence mass)
• Estimate reddening (hence SFR)
• Estimate age and stellar mass
• Two complementary samples GB HDF
• and two methods Keck near-IR spectroscopy and
  HST multi-band photometry

56
Keck NIRSPEC K-band spectra of LBGs
R7-14 Å Texp5-18 Ksec
Pettini et al. 2001
Wavelength (µm)
57
ISAAC K-band spectra of LBGs
Wavelength (µm)
58
NIRSPEC H-band spectra of LBGs
59
Detecting the continuum in K-band
60
The metallicity of LBGs

• Key measure if progenitors of spheroids, LBGs
  must be metal rich
• Measures from the O23 index
• R23(OIIOIII)/Hß
• Measures are double-valued
• Rest-frame optical spectroscopy to target
• OII, Hbeta, and OIII lines (in the near-IR)
• KeckNIRSPEC and VLTISAAC spectra in H and K
  band
• VERY DIFFICULT observations

61
(No Transcript)
62
The Metallicity of LBGs vs Normal Galaxies
Metallicity-luminosity for local galaxies from
Kobulnicky Koo (2000) adjusted for cosmology.
Purple box shows the location of the LBGs where
are over luminous for their metallicity. (Pettini
etal. 2001, ApJ, 554, 981).
63
The Metallicity of LBGs

• 0.1ltO/H/O/H?lt 1
• In two cases O/H/O/H?0.3 (see
  Kobulniky and Koo 2001)
• LBGs are relatively metal rich systems
• More metal enriched than DLAs
• Less enriched than inner regions of AGNs
• Metallicity comparable to the Solar neighborhood

64
Dynamics from the nebular lines

• Idea is to use velocity width of nebular lines as
  dynamical indicator
• It is found
• 50ltslt115 km/s
• Returns masses in the range
• M a few 1010 M?
• within r1/22-3 kpc

65
Are the nebular lines good dynamical indicators?
No correlation with with either LUV or MB raises
serious doubts that N.L.s are reliable dynamical
tracers
66
Spatially resolved velocity profiles - 1
67
HST image, F702W
68
Spatially resolved velocity profiles - 2
69
Keck NIRC K-band image, 0.5
70
Gas outflows
Vout 200 - 400 km/s
71
Results from the near-IR spectroscopy

• Estimate of metallicity 0.1ltO/H lt1 solar
• Insight into the extinction law Calzetti law OK
• Mass unconstrained
• Evidence of high-speed outflows (300 km/s)

72
The rest-frame B-band LF
Dickinson, Papovich Ferguson 2001
73
Fitting age and stellar mass
Papovich, Dickinson Ferguson 2001
74
Fitting SED with Broad-band photometry
Papovich, Dickinson Ferguson 2001
75
Stellar Mass and Burst Age
Papovich, Dickinson Ferguson 2001
76
Stuffing in old stars
Papovich, Dickinson Ferguson 2001
77
Stuffing In Old Stars
78
LBGs at z3 and zgt4
The z3 galaxies do not seem to be the same ones
seen at zgt4
79
LBGs at z3 and zgt4
Aging zgt4 ex- LBG should be visible in the HDF
images as red sources. There are no
such galaxies. But we do see zgt4 LBGs. Where are
they at Z3? Recurrent SF? Just bad luck in The
HDF?
80
Conclusions from SED Fitting

• The forming population (the one observed) is
  younger than 1 Gyr
• Unconstrained for how long SF will go on
• Stellar mass smaller, but not too smaller than m
  today M a few 1010 M? (nebular line mass
  really dubious)
• Maybe recurrent SF activity?

81
High-z Galaxy Clustering

• Clustering links mass distribution and physics of
  star formation. Key observable
• Samples are large enough to attempt the measure
• Possible to estimate spatial clustering
• Angular clustering seems reliable and safe measure

82
The Clustering of LBGs

• LBGs are strongly clustered in space
• Correlation lengths rivals that of local galaxies
• Clustering of mass cannot have grown to such an
  extent at z3 in reasonable cosmologies
• Bias galaxies form in biased regions of the mass
  distribution
• In principle, it can constrain the mass spectrum

83
Clustering in the redshift space
The Westphal Field
84
Star Formation History of the Universe
85
UV luminosity and star-formation rates

• SFR is very important parameter for galaxy
  evolution
• If there is no dust obscuration, UV luminosity is
  good tracer of the star-formation rate
• SFR (M?/yr) 1.4x10-28 x LUV(1500 Å)
• (Kennicutt 1998)

86
UV luminosity and star-formation rates

• Star formation rates estimated using UV and Hß
  luminosities are roughly consistent in LBGs.
• (Pettini etal 2001, ApJ, 554, 981)

87
High-z Galaxy Stellar Populations and Extinction
E(B-V)0.4
0.2
0.0
Ferguson etal 2000, ARAA, 38,667
88
Evidence of dust reddening
89
The star-formation rates
90
Luminosity Function of LBGs

• Data are consistent with similar LF at z3 and
  z4.

Luminosity function of LBGs at z34. (Steidel et
al. 1999, ApJ, 519, 1)
91
Rest-Frame Luminosity Function of LBGs

• GB and HDF give similar results.
• Data are consistent with similar LF at z3 and
  z4.
• Possible drop at faint mags at z4.

Luminosity function of LBGs at z34. (Steidel et
al. 1999, ApJ, 519, 1)
92
Star Formation History of the Universe

• Extinction corrected emissivity of star formation
  is constant for zgt1
• Onset of substantial star formation occurs at zgt
  4.5 ?
• Star formation does not show strong peak at z2
  as for quasar activity ?

UV luminosity density as a function of z.
(Steidel et al. 1999, ApJ, 519, 1)
93
Radio and Sub-mm Searches
94
Radio to IR Spectrum of Luminous IR Galaxies
K-correction increases flux density for high-z
objects.
Carilli Yun 2000, ApJ, 530, 618
95
SED of Instantaneous Burst

• IR sub-mm remains bright as a dusty starburst
  spectrum is redshifted.
• Thus, it is relatively easy to detect these
  objects in the sub-mm.

Devriendt etal. 1999, AA, 350, 381
96
Obscured high-redshift galaxies in the HDF
ISO Rowan-Robinson et al. 1997 Desert et al.
1999, Aussel et al, 1999
SCUBA Hughes et al. 1998, Peacock et al. 2000
97
Conclusions
98
The End