History and modes of star formation in field galaxies:

1
History and modes of star formation in field
galaxies A new picture from AEGIS Kai Noeske,
Lick Observatory and the AEGIS
Collaboration DEEP2, IRAC MIPS, GALEX, Palomar
K teams B. Weiner, S. Faber, D. Koo, A. Coil, J.
Lotz, M. Cooper, M. Davis, P. Guhathakurta, S.
Kassin, J. Newman, A. Phillips, J. Primack, R.
Somerville, C. Willmer G. Fazio, J. Huang, G.
Willson, P. Barmby G. Rieke, E. Le Floc'h, C.
Papovich K. Bundy, C. Conselice, R. Ellis C.
Martin, T. Small, D. Schiminovich, M.
Rich DEEP06, October 13, 2006
2

• AEGIS
• The All-Wavelength
• Extended Groth Strip
• International
• Survey
• 0.25x2 deg
• DEEP2Keck /DEIMOS spectra 10,000 precision
  redshifts, galaxy kinematics
• HST V,I (700 sq arcmin-2xGOODS)
• Very deep
• - Spitzer (IRAC, MIPS)
• - GALEX (NUV, FUV)
• - Chandra ACIS
• - VLA 20cm
• - Ground-based deep U- to K-imaging
• Data access MAST deep.berkeley.edu or contact
  teams for collaborations
• Release mid 2007, ApJ special issue press
  release Oct/Nov 2006

3
Star formation since z1 (e.g. Cowie et al. 96,
Guzman et al. 97, Brinchmann Ellis 00, Heavens
et al. 04, Juneau et al. 05, Bauer et al. 05,
Bell et al. 05, Perez-Gonzalez et al. 05, Feulner
et al. 05, Papovich et al. 06, Caputi et al. 06)
Perez-Gonzalez et al. 2005
Heavens et al. 2004
Bell et al. 2005
Co-moving star formation rate density declined by
x10 Highest galaxy SF rates increase with
z Mass dependency of galaxy star formation
histories highest SFR per unit stellar mass
observed in least massive galaxies massive
galaxies formed bulk of stellar mass faster and
earlier
4
When and how did galaxies of what mass form their
stars? Data are rapidly accumulating, but we
still have no comprehensive picture more
detailed view from joint interpretation of AEGIS
data
5
AEGIS data

• 3,000 AEGIS field galaxies, spectroscopic DEEP2
  redshifts z 0.2 1.1
• IR, UV, emission line star formation tracers
• MIPS 24?m (cf. Le Floc'h et al. 2005)
• DEEP2 Ha, Hb, OII, raw extiction corrected
  (Weiner et al. 2006)
• GALEX NUV, raw extinction corrected star
  formation rates (incl. DEEP/CFHT photometry D.
  Schiminovich)
• Stellar masses (SED fits DEEP BRI, Palomar K
  K. Bundy)
• HST morphologies (GINI/M20, J. Lotz) rest-frame
  colors (C. Willmer)

6
SFR vs M Spitzer MIPS 24?m data
24?m completeness
Stellar mass completeness
Noeske et al. 2006 (ApJL, submitted)
24 mm SFR recover extincted, and fully obscured
star formation - essential at high masses
- sensitivity limited only recover IR-luminous
sources (upper envelope of star-forming
galaxies, mostly massive)
7
Emission line SFRs increased range of SFR, mass
Noeske et al. 2006 (ApJL, submitted)
Adding emission line star formation rates
increases sensitivity, yields SFR for majority
of galaxies in stellar mass completeness
range results tested for consistency with GALEX
UV SFR
8
A more detailed view of star formation properties

Noeske et al. 2006 (ApJL, submitted)
1) Fiducial star-forming galaxies 24?m sources,
or blue emission line galaxies (2/3 of sample)
9
A more detailed view of star formation properties

Noeske et al. 2006 (ApJL, submitted)
1) Fiducial star-forming galaxies 24?m sources,
or blue emission line galaxies (2/3 of
sample) 2) Galaxies not detected in 24?m or
emission lines early-type HST morphologies, red
sequence likely not significantly star-forming
(1/3 of sample)
10
A more detailed view of star formation properties

Noeske et al. 2006 (ApJL, submitted)
1) Fiducial star-forming galaxies 24?m sources,
or blue emission line galaxies (2/3 of
sample) 2) Galaxies not detected in 24?m or
emission lines early-type HST morphologies, red
sequence likely not significantly star-forming
(1/3 of sample) 3) Galaxies with no detection in
24?m, but weak emission lines 2/3 early-type HST
morphologies, red sequence, i.e. large fraction
LINERs (lt20 of sample)
11
Main Sequence of star-forming galaxies
Noeske et al. 2006 (ApJL, submitted)
The majority of star-forming galaxies form a
sequence with a narrow range of SFR at a given
stellar mass and redshift. (Within data limits
at zlt0.5(1), gt95(80) of galaxies with robust
signs of SF lie on the main sequence) Normal
star-forming galaxies, ?SF governed by same set
of few physical processes ? - prior to quenching
of star formation - ?
12
Width of the Main Sequence
Noeske et al. 2006 (ApJL, submitted)
upper limit to intrinsic width of MS 0.3 dex
in SFR at a given mass no change with redshift
to z1 few outliers above
13
Tight Main sequence with similar range of SFR to
z1
Noeske et al. 2006 (ApJL, submitted)
1) Upper limit on the amplitude of SFR
variations/starbursts At a given M and z,
galaxies are 2/3 of the time within a factor of
2 of their average SFR Limit (lt 100 Myr) to
total duration of stronger starbursts Limit to
effect of mergers on SFR - Constraint to
feedback regulation in merger models
14
Tight Main sequence with similar range of SFR to
z1
Noeske et al. 2006 (ApJL, submitted)
2) Range of star formation about the mean is
constant to z1, overall distribution moves to
higher SFR rules out increased (frequency,
amplitude) of starbursts as dominant mechanism of
SFR evolution dominant mode of SF to z1 is
apparently a gradual increase of SFR, with
moderate episodes superposed (Papovich et al.
2005, Reddy et al. 2006, talk by E. Daddi up to
z3?)
15
Luminous Infrared Galaxies starbursts?
LIRGs
Noeske et al. 2006 (ApJL, submitted)
Increase of LIRG number density with z usual
assumption dusty (merger-driven) starbursts,
like LIRGS _at_ z0 (ISO had yielded no whole
picture of SF, tip-of-the-Iceberg) but most
LIRGs at z1 have regular, non-merger
morphologies (Bell et al. 2005, Melbourne et al.
2005, Lotz et al. 2006) AEGIS picture z1 LIRGs
are mostly normal star-forming galaxies,
IR-luminous due to general increase of SF in
individual galaxies with z
16
SF in field galaxies since z1 largely
quiescent, mass-driven?
Brinchmann et al. 2004 (SDSS)
Main Sequence with narrow range of SF persists to
z0, and to low masses
- Tight relations of SFR, O/H with stellar mass
(integrals SF history) - tight stellar
mass-Tully-Fisher relation
Tremonti et al. 2004(SDSS)
McGaugh (2005)
log O/H
log M
Integral of SF history apparently not dominated
by stochastic events Quiescent SF histories,
mass a dominant parameter SF histories at a
given mass are similar (prior to quenching)
17
What drives the dominant gradual decline of SFR
since z1? Gradual processes like gas
exhaustion? A simple model to parametrize the
mass-dependent evolution of SFR
18
t models to parametrize gradually declining
SFR exponentially declining star formation
rates SFR(t) SFR(t0) exp(-t/t) Mass-dependenc
e both the e-folding time t and formation
redshift zf (beginning of star formation) are
functions of the total baryonic (initial gas)
mass (power laws for flexibility) t(M)atMbt (1z
f)(M) azf Mbzf (variable zf
staging) Tinsley equations, instantaneous
recycling closed-box, but automatically include
cold gas loss (heating, winds)
19
? models with coeval start of star formation, zf2
Dotted t(M)?M -0.6 Dot-dash t(M)?M -2 (longer
SF timescales in less massive galaxies) roughly
reproduces high mass end problematic at low
masses cannot reproduce high specific SF rates
Noeske et al. 2006 (ApJL, submitted)
20
High specific SFR at low M, high z universal
late starburst in old galaxies, or delayed onset
of major star formation?
Noeske et al. 2006 (ApJL, submitted)
Assumption of coeval start of SF at high z
implies starbursts in large fraction of
galaxies High specific SF rates would generate
all stellar mass in 3x108 3x109 yr -typical of
starbursts- but 80 complete 1) gt50 of
galaxies would have to be in a higher-than-past
average SF state. 2) specific SFR even increase
to higher z, and slowly decrease towards lower z
not short bursts high SFR part of gradual
decline on Gyr timescales
stellar
constant SF rate since zf
zf mass-dependent
21
High specific SFR at low M, high z universal
late starburst in old galaxies, or delayed onset
of major star formation?
Noeske et al. 2006 (ApJL, submitted)
High specific SF rates at z1 more consistent
with the early phase of a main epoch of star
formation, gradually decreasing on Gyr
timescales young, gas-rich galaxy onset of this
main SF episode must be recent, else high SFR
would overproduce stellar mass (lt doubling
timescale, 1 Gyr)
stellar
constant SF rate since zf
zf mass-dependent
z
z
22
Staged galaxy formation
New concept staged galaxy formation - massive
galaxies form at high zf less massive galaxies
to start major SF on average later, with zf
ranging from high to low ( No contradiction to
most low mass galaxies being evolved)
z
23
Staged ? models both ? and zf mass-dependent
1) longer SF timescales in less massive galaxies
(classic picture) 2) on average, later onset of
major SF in less massive galaxies (new!) -
reproduces characteristic features of SFR/M (M,z)
to zgt1 - observational and theoretical support
observed mass-stellar age Heavens04,
Thomas05, Feulner05, Reddy, Erb06) LCDM
(Gao05, Neistein06, Wechsler05,
MouriTaniguchi 06) baryon cooling/UV
background (BabulRees 92, Kepner97)
Noeske et al. 2006 (ApJL, submitted)
24
What do t models tell us? 1) t models
parametrize generalized Schmidt law SFR ? (Mgas)n
observed star formation histories appear
consistent with mass-dependent gas depletion 2)
t models include gradual loss of cold gas
(heating, loss) -gt t models can crudely
parametrize gradual quenching processes lt- 3)
t(M) SF efficiency (M) Model parameters shown
here not unique! Limited M range, systematics
of SFR, M IMF ...
25
Summary (NOTE star-forming field galaxies,
Mgt1010 Msun, zlt1) 1) Combining evidence from
AEGIS SF galaxies form a tight Main Sequence,
with a small range (rms lt x 2) of SFR at a given
M,z. Limits strong starbursts, merger effects on
SFR. 2) Gradual decline of SF is more important
than episodic SF/starbursts since z1 LIRGs at
z1 are mostly normal SF galaxies, not short
starbursts - New picture high SFR often not
brief starbursts, but early, gas-rich phase of a
galaxy - 3) Parametrize gradual decline of SF by
mass-dependent t models consistent with gradual
depletion of cold, star-forming gas due to SF
(and feedback / quenching) measure SF
efficiency (M) 4) New scenario Staged galaxy
formation - less massive galaxies form on longer
timescales (classical view) AND start major star
formation later (new) 2 effects contributing to
downsizing t(M), zf(M)
26
Outlook (1) a new benchmark survey for SF, AGN,
environments to zgt1 New Spitzer MIPS Legacy
survey (texpx12!) (in progress, PI M. Dickinson),
DEEP3 (4 yr, KeckII/DEIMOS), proposed, PI
Faber, Co-I Noeske
27
Outlook detailed modelling of mass, SF rate,
color (Monte Carlo models/simple SAMs) - recover
distributions of t, zf at a given mass for SF
galaxies - different morphologies, environments -
assess uncertainties from tracers of SFR, stellar
mass, IMF (collab. with AEGIS Bayesian code J.
Lotz/NOAO)
chance to quantify normal mode of star
formation and isolate additional processes that
quench star formation mergers, AGN, ...
28
t(M), zf(M), dust
t(M)
t(M), zf(M)
t(M), zf(M), dust, scatter (dust) scatter
(t) scatter (zf)
t(M), zf(M), dust, scatter (dust)
29
Star formation tracers extinction, need for
thermal IR
UV, emission lines trace (ionizing) UV photon
flux UV photons absorbed by dust, re-radiated in
thermal IR some SF is completely obscured
extinction-correction of UV tracers not always
valid best SFR approach add UV IR traced SFRs
4
SF from IRUV
SF from rest-360nm, no ext. corr.
Bell et al. 2004
30
Overview I. The DEEP2 galaxy redshift survey /
AEGIS a new deep-(semi) wide
multi-wavelength survey in the Extended Groth
Strip II. Star formation in AEGIS field
galaxies since z1 an improved picture
from joint interpretation of AEGIS data IIa. A
Main Sequence of star-forming galaxies to z1
- gradual decline of star formation in
galaxies dominant since z1 - limited
role of starbursts IIb. A mass-dependent
parametrization of star formation histories
- model low mass galaxies have longer decline
timescales and later onset of major star
formation (components of downsizing) III.
Summary IV. Outlook EGS Spitzer Legacy and
DEEP3/EGS a new benchmark survey for
star formation, AGN, environments, bimodality
since z1.4