Download The VLA COSMOS Survey

The VLA-COSMOS survey:
Tracing star-forming and AGN
galaxies through cosmic time
Vernesa Smolčić (Caltech)
E. Schinnerer (MPIA), C.L. Carilli (NRAO), M. Bondi (INAF),
P. Cilliegi (INAF), G. Zamorani (INAF), K. Jahnke (MPIA),
M. Sargent (MPIA) & the (VLA-)COSMOS collaboration
Radio emission at 1.4 GHz (20cm)

Dominated by synchrotron radiation

Two dominant populations in extragalactic
radio surveys:
1
Star forming (SF) galaxies
Radio emission is not sensitive to dust
2
Active galactic nuclei (AGN)
Radio emission directly traces the population of low radio
power AGN, deemed important for galaxy formation
Condon 1992
Star forming galaxies
Radio – IR correlation
M82
1
.
4
G
H
z
2
0
Condon 1992
c
m
1
.
4
G
H
z
z
~
5
Bell 2003
van der Kruit 1971; Helou et al. 1985; Condon
et al. 1992, Yun et al. 2001; Bell 2003; Obric et
al. 2006; Mauch & Sadler 2007
Cosmic star formation history
Short-wavelength radiation (e.g. UV) sensitive to dust
 radio emission overcomes this bias
Compilation based on
different star formation
estimators (Hα, OII, UV;
Hopkins 2004)
Star formation rate density
[M / yr / Mpc3]
AGN feedback
1. QUASAR MODE
2. RADIO MODE
-
- Once a static hot
gas halo forms
around the galaxy
- Modest BH mass
growth
- Radio outflows heat
surrounding gas 
truncation of further
stellar mass growth
-
Merger driven
Vigorous BH mass growth
Qusar wind gas expells
gas out of the galaxy’s
center 
termination of quasar
& starburst phase
Not necessarily linked
to radio outflows
Faber et al. 2007
Allows good reproduction of
observed galaxy properties
Croton et al. 2006; Bower et al. 2006; Sijacki et
al. 2006, Hopkins et al. 2006…
Different phases of
galaxy merger (gas);
MPA
galaxy cluster MS0735.6+7421
(z=0.2); white = HST, blue =
Chandra, red = VLA; NASA.gov
2. Radio mode
1. Quasar mode
Allows good reproduction of
observed galaxy properties
Luminosity function of galaxies
Croton et al. 2006
Croton et al. 2006: mean BH accretion rate per
unit volume averaged over the entire simulation
This theoretically derived
curve can directly be
inferred from radio
observations
HOWEVER
Deep radio data (rms<15μJy/beam) of
a large sample needed !!!
The faint (<1 mJy) radio population


1.4 GHz (20 cm) differential radio source counts (normalized to Euclidian
space) flatten below 1 mJy
 rise of a new population not contributing significantly at higher flux levels
The composition of this faint radio population is highly controversial (Seymour et
al. 2004, 2008, Simpson et al. 2006, Fomalont et al. 2007, Padovani et al. 2007, Smolcic et al. 2008, Kellermann et al.
2008)
Differential 20 cm source counts (norm. to Euclidian space)
star forming gals +
low-power AGN
 Robust SF/AGN
classfier needed !!!
n S2.5 (sr-1 Jy1.5)
sub-mJy radio population:
Bondi et al. (2008)
FIRST / NVSS
S (mJy)
Cambridge
The COSMOS Survey
The COSMOS survey
COSMOS overview (Scoville et al. 2007)
 2 □O equatorial field
X-ray to radio imaging (~30 bands)
 galaxy photo-z accuracy, 0.7%
(Ilbert et al 2008)
 quasar photo-z accuracy, 1.5%
(Salvato et al. 2008)
 spectroscopy (VLT-VIMOS + Magellan-IMACS)
5σ depth for all
existing data and the
expected 5σ depth
for upcoming or
ongoing guaranteed
time observations
Radio view of COSMOS field:
VLA-COSMOS 20 cm survey
NRAO Very Large Array
VLA-COSMOS core team: Schinnerer, Smolčić, Carilli, Bondi, Ciliegi,
 Scoville,
NRAO Very
Large Array
Bertoldi,
Blain, Impey, Jahnke, Koekemoer,
Le Fevre, Urry, Martínez Sansigre, Wang, Datta, Riechers
VLA-COSMOS team: Schinnerer (PI)

Smolcic,
Bondi,
Ciliegi, Scoville,
Large
projectCarilli,
(275hr)
: Schinnerer
et al. (2004, 2007)

Bertoldi,
Blain,
Impey, Jahnke,
~ 2,400
sources
(catalogKoekemoer,
- public)
O

Le Fevre,
Martinez
Wang, Datta
~ 2 □Urry,
; mean
rms Sansigre,
~ 10 Jy/beam,
1.5” resolution


unique complementary COSMOS data set enabling
Pilot project
(10hr):
A array evolution
(Schinnererthrough
et al. 2004)
studies
of AGN/SF
cosmic times
~ 250 sources (catalog - public)

~ 1 sqrdeg;
rms ~et
30al. Jy/beam
Deep
project (62hr):
Schinnerer
(to be submitted)

~ 1 □O; rms ~ 7 Jy/beam
Large project (275hr): A+C array (Schinnerer et al. 2007)

~ 3,642
sources
(catalog
- public)
327
MHz project
(24hr):
Smolčić
et al.
(in prep)
O

sqrdeg;
rms ~ 15(10) Jy/beam;
~ 2 □~ 2(1)
; rms
~ 0.5mean
mJy/beam





Deep project (62hr): A array
~ 1 sqrdeg; rms ~ 7-8
Jy/beam (central 30’)
What have we (so far) learned
from VLA-COSMOS?
The composition of the sub-mJy radio population
@ z≤1.3: ~ 350 SF &
~ 600 AGN gals.
Baldwin-Phillips-Terlevich (1981) diagram
Differential 20 cm source counts
n S2.5 (sr-1 Jy1.5)
New rest-frame color-based method for
separating SF from low-luminosity AGN
galaxies (i.e. Seyfert, LINERs; Smolčić et al. 2008a) 
applied to VLA-COSMOS data
FIRST / NVSS
Cambridge
S (mJy)
Bondi et al. (2008)
Sub-mJy radio population:
1) not dominated by star forming galaxies
2) fair mix of SF and (low-L) AGN galaxies
Smolčić et al. (2008; ApJS; 177, 14)
Kauffmann et al. (2003), Kewley et al. (2001,2006),
Obrić et al. (2006), Smolčić et al. (2006, 2008a)
The radio - (F)IR correlation

-

Current focus on (Sargent et al., in
prep.):
quantification of selection effects in
view of future deep EVLA & Herschel
data
statistically sound treatment of flux
limits using survival analysis
 evolution of radio-IR relation for star
forming systems out to z~1
Future work:
Little or no evolution
of the IR/radio ratios
at least out to z~1
Smolčić et al. (2008); Sargent et al. (in prep)
- Effects of environment (E. Murphy et al., in prep)
- separation of star forming systems into
different classes of objects (e.g. optical
morphology, mass)
- stacking of radio population at faint IR
fluxes
The dust-unbiased cosmic star
formation history @ z≤1.3 from
the VLA-COSMOS survey
Good agreement between
VLA-COSMOS and

20cm lumiosity functions
for VLA-COSMOS star
forming galaxies (blue)
Cosmic star
formation history

previous radio results
(1 order of magnitude
smaller sample; Haarsma
et al. 2000)
other SFRD estimates
from Hα, OII, UV, IR
with dust correction
applied where needed
Dust attenuation at
intermediate redshifts is
well understood
Smolčić et al. (2009, ApJ, 690, 610)
Probing SFRs at high z via stacking
COSMOS Lyman break galaxy sample of Lee, Capak et al.
Stacking detection:
U band drop-outs (2.5 < z < 3.5)
Median flux: 0.90 ± 0.21 μJy
<SFRradio> = 31 ± 7 MSUN/yr
<SFRUV> ~ 17 MSUN/yr
 dust attenuation factor ~1.8 <<
standard attenuation factor of 5
Star formation history derived
from UDS/UKIDSS BzK
selected galaxies stacked in
radio (Dunne et al. 2008)
(Steidel et al. 1999, Adelberger & Steidel 2000,
Reddy & Steidel 2004)
Carilli et al. (2008; ApJ, 689, 883)
Dust attenuation at high
redshifts may be smaller than
at lower redshifts
c
The evolution of VLA-COSMOS
(weak) radio AGN
Ledlow & Owen
(1996) FRI / FRII
diagnostic plot for
VLA-COSMOS AGN
20cm lumiosity
functions for VLACOSMOS AGN (red)
Qualitative agreement
between cosmological
model and observations is
very encouraging for the
idea of ‘radio mode’
feedback
Smolčić et al. (ApJ, sub.)
Volume averaged
mechanical
heating rate
Comoving BH
accretion
rate density
Summary & EVLA outlook
VLA-COSMOS:
 Composition of sub-mJy radio population: fair
mix of SF and low-power AGN galaxies
 z ≤1.3:
 Cosmic evolution of VLA-COSMOS SF
and AGN galaxies
 First observational insight into ‘radio
mode’ feedback beyond the local universe
 z ~ 3: stacking down to 1μJy levels that EVLA
will be able to observe
VLA-COSMOS Large Project limits
EVLA-COSMOS:
 Deeper 20 cm imaging:
 probing radio LIRGs (>10 MSUN/yr) through cosmic time
 complete sample of ULIRGs (>100 MSUN/yr) out to high z
 probing weak radio AGN out to high z  testing cosmological models
 6 cm imaging:
 high resolution: radio morphology, composite objects
 spectral indices
 probing thermal (free-free) radio emission for z>3.5