Transcript LIRGs and cosmic star formation

Conference summary
Catherine Cesarsky
ESO
Moriond, March 2005
When UV meets infrared
• (and everything from gamma rays
to radio)
• Do we see the same sources in
UV and IR?
GALEX
IRAC GOODS
24 micron MIPS
Summary
1. By selection, UV galaxies and IR galaxies have very different
characteristic IR/UV ratios (the means differ by a factor of 10).
2. The morphological and stellar mass distributions of the two
populations have good overlaps (> 70%). IR galaxies tend
to be more massive and earlier types, with an excess of interacting
galaxies, and UV galaxies to be less massive and later types.
3. UV galaxies are less clustered than IR galaxies.
4. Galaxies with the highest SFR (>100 M ๏ /yr, Ltot > 1012 L ๏),
are missed in the UV samples.
5. A population of low metallicity (< 1/10 solar), low mass (<10^9 M )
dwarf UV galaxies (prototype I Zw 18) are `IR quiet’, with the
๏
IR/UV ratio ~ 0.3 or less. They occupy only a few percent of a
UV selected sample.
UV/mid-IR comparison of two LIRGs
Images: HST/STIS UV - Contours: ISOCAM 7μm
7μm/UV ~ 800:10:35
7μm/UV ~ 330:160:190
At z~2: UV --> R-/I- band & ISO/CAM 7μm -> Spitzer/MIPS24.
The poor spatial at z~2 will result in blending of the emission from the unresolved interacting components. An
increased scatter will thus be introduced in the observed optical to mid-IR colors of these galaxies, leading to a
systematic underestimation of their dust content.
Charmandaris, Le Floc’h, Mirabel, ApJ, 2004, 600, L15
VC
Moriond 2005
• Do we need UV to
understand star
formation?
• YES, at least in some cases
(low obscuration)
Rest UV Traces Star Formation
Over Large Range of
Specific Star Formation & SFR/Area
b
M*
M*

M*
 M * / tage 
Low Surface
Brightness
Galaxies
Luminous UV
Galaxies
What gives???
Sgas ranges 20:1
Early Type Gals
Milky Way

Sgas 1.4 ranges 70:1
Rest UV
Traces Star
Formation
Shortcut
to SFH
Over Large Range of Specific Star Formation
•
b-parameter vs. NUV-r color
– Obtain b from color alone
– Works when no spectra are
available
– Valuable for high z
– Spread in x-direction due to
internal extinction
M*
M*
b

M*
 M * / tage 
NUV-r  b
H and UV radial profiles
Thilker, Meuer, et al
• Radial profile differences
seen in other galaxies
• Not all galaxies show H
deficit
UV
Ha
•Star clusters as
indicators/
demonstrators of star
formation
• Do we need IR to
understand star
formation?
• YES, especially for the
brightest galaxies
• Can the different star
formation indicators be
reconciled?
• Sometimes…
Ha/UV in SDSS
Treyer, Johnson, et al.
• Ha/UV shows
systematic trend
Higher LUV, Blue NUV-r
L(Ha)/L(UV)~Kennicutt
Low LUV, Red NUV-r
L(Ha)/L(UV) > Kennicutt
SFRs as estimated by UV, [OII] & IR
(Hammer et al, Venice 2003, proceedings, astro-ph/0401246)
OII line & UV luminosities underestimate SFR values by
factors 5 to 100 for starbursts & LIRGs !
SFRNUV vs. SFRdust
log SFRdust (Msun yr-1)
Quite good agreement on average but...
log SFRNUV (Msun yr-1)
Two different trends are observed:
At low values of ANUV,
the
dust
emission
underestimates
the
total SFR because of
the non negligible NUV
emission.
At high values of ANUV,
the
NUV
emission
underestimates
the
total SFR.
Problem with ANUV?
●
log SFRNUV/SFRdust
●
log SFRNUV (Msun yr-1)
Estimating extinctions and SFRs at z ~1
(Flores et al, 2004, A&A 415, 885)
FORS2/ISAAC: 16 ISO galaxies, 0.4< z <1
- extinction corrected H SFRs are close
to mid-IR estimates (Elbaz et al, 2002) for
SFR < 150 MO/yr (i.e. below ULIRGs)
 more robust SFR estimates
-luminous IR galaxies (not ULIRGs)
dominates the cosmic star formation
density at z~1
(confirmed by Spitzer, Le Floch et al, 2004)
 less than 20% of the star formation
density is coming from extremely dust
enshrouded regions
•Deep IR surveys: do
we understand what
we see?
•Probably, but…
EBL: optical vs IR
CIRB~ 1.5 OPT IGL
In local universe, about 30%
bolometric light in IR; LIRGs, ULIRGs
produce 2% of bolometric luminosity
However,distant universe is IR.
Due to LIRGs? How distant?
LW3
z=0
Typical galaxy
spectra
0.5
1
1.5
2
LW3 15
LW2 6.7 
K-corrections
140 m
CIRB peak:
Individual galaxies peak: 60 to 100 m
Peak shifted to 140 m
if z=0.4 to 1.3 (<z>~0.85)
15 m
8 m
z=0.85 
140 m
z=0
80 m
ISOCAM deep surveys in LW3 (12-18 m):
Ideal to detect redshifted PAH for z~0.85 (or in
general at z<1.5)
Number Counts
•
Roughly in agreement with
ISOCAM results
•
•
•
•
Some confused ISOCAM sources
are resolved by Spitzer
The HDF-N pilot study is not an
unbiased survey
Marleau et al. (2004) find 24 m
number counts peak at fainter
flux than 15 m counts
difference b/w 15 and 24 m
counts is not the result of
confusion of ISOCAM sources or
systematic differences between
the observatories
From the MIR ?
M82
(Laurent et al. 2000)
(disque)
Local universe : correlation MIR – LIR (Elbaz et al, 2002)
correlation radio-MIR (Codon 1992, Yun et al, 2001)
or radio is a tracer of LIR
MIR
+ local
templates or correlations
=> FIR=>
LIR => SFR
IR vs ISOCAM
15 m
IR vs IRAS
12 m
15 m vs IRChary
& Elbaz 2001
Dale & Helou, 2002
Lagache et et al, 2004
……..
Kennicutt 1998
The PAH bump exists at z=0.7
SED of a LIRG at z=0.69 (LIR~1011.1 L,SFR~22 Myr-1)
15m ISOCAM
24m Spitzer-MIPS
LIRGs and cosmic star formation
50 % stars born z<1.5 (70 % universe age)
36 %
@ z<1 (57 %)
67 %
@ z<2 (76 %)
W*
Proportion of present-day stars born in LIRGs > 50 %
==> Common phase experienced by all/most galaxies...
General 24m differential counts
(this work, Chary et al. 2004, Papovich et al. 2004)
Model predictions S24/S15 as a function of z, S24
S > 2-3 mJy dominated by
objects with S24/S152-2.5
S  0.3 mJy dominated by
objects with S24/S15 1.5
S < 0.2-0.3 mJy dominated by
objects with S24/S15 > 2-3
-> NEW POPULATION !
R-band mag versus Flux@24μm
80% completeness
limit at 24μm
 VERY hard to be complete in the redshift identification
at any 24μm flux, using VVDS/GOODS/COMBO-17
Rencontres de Moriond, March 6-12th 2005
IR luminosities in the CDFS
2635 sources
with redshifts
* Modest IR emitters
at 0<z<0.5
* ULIRGs : quite
rare at 0<z<1
* LIRGS: significant
contribution at
z>0.5
80% completeness
limit
Rencontres de Moriond, March 6-12th 2005
* More « normal »
starbursts are not
negligible neither
Star formation history at z<1
et al.2004
2004 _ _ _ _ _
Compilation byLagache
Hopkins
Blain et al. 2002 . . . . . .
total
Chary & Elbaz 2001
11
LIR <10 L .
11
LIR >10 L .
ULIRGs
 LIRGs/ULIRGs dominate beyond z~0.7
Rencontres de Moriond, March 6-12th 2005
Star formation history at z<1
AGN contribution ??
* ISO/XMM : <20% (Fadda et al. 2002)
* X-ray +IR bkg synthetic models :
<5% (e.g., Silva et al. 2004)
First Spitzer results : <15% of
sources flagged as AGNs by VVDS
& COMBO-17 (see also SWIRE,
Franceschini et al. 2005)
 LIRGs/ULIRGs dominate beyond z~0.7
Rencontres de Moriond, March 6-12th 2005
Summary
* 55~65 % of 24μm sources at z<1
for flux>80μJy
3.5
* At 0<z<1, L* evolves at least by (1+z)
( exclude a pure density evolution)
* IR luminous galaxies start to dominate
the SFRH at z>0.6
* LIRGs+ULIRGs = 70% of SFR at z=1
* Need a better understanding of IR
SEDs : IRS GTO, MIPS SED mode...
Cornell University - Ithaca, December 1st 2004
• Is galaxy formation (the
building up of galaxies) regular
or episodic?
• Mostly episodic, even if we
don’t know for sure why.
LIRGs: potentially double their masses in ~0.8 Gyr
SFR: IR & H
Red dots: LIRGs (20-200 MO/yr)
Full squares: starbursts
(<20M/yr)
SFR: [OII]3727
Open symbols
From BE00:
Brinchman & Ellis 2000
How to account for the high LIRG fraction
(15% of intermediate mass galaxies) ?
A specific population ?
LIRGs are continuously forming stars during 3.3 Gyrs (z=1  z=0.4)
 they would multiply their masses by 2 x (3.3/0.8)=8.2 !!
BUT no trace of recent formation of massive galaxies,
dominated by E/S0, with 3 1011<Mstar<31012MO
• Do we understand ultra
luminous star forming
galaxies?
• Yes, although debate on
role of AGN not completely
closed
The first 18
low-resolution
IRS spectra
of ULIRGs
Diversity!
is the name of the game…
VC
Moriond 2005
Results of submm surveys
• Highly luminous (ULIRG) systems
• SFR ~ 1000 M yr-1
• Massive systems
• Evidence for outflowing winds
Progenitors
of massive
elliptical galaxies?
• Do we understand
Luminous star forming
galaxies?
• Errrrr, well…
Stellar properties of distant LIRGs
• b parameter: SFR/<SFR> = 5 +/-3
• Burst duration
~ 108 years
• Burst stellar mass fraction
~ 5-10 %
• M/Lz ~ 0.3 (SDSS 1.6)
• Stellar masses:
<M*>
~ 5 x1010 M
Large UVLGs = LIRGs ?
•
•
•
•
UV Luminosity Density from UVLG x30 from z=0 to z=1
25% of FUV luminosity density at z=1 from UVLG
SFR from LIRGs x20 from z=0 to z=1
> 70% of dust-enshrouded SFR density at z=1 from LIRGs
Goldader et al. (2002)
Burgarella et al.
(2005)
Conclusions

The most UV luminous galaxies in the combined
GALEX/SDSS sample comprise two populations:



Compact UVLGs appear similar in many respects to
Lyman break galaxies






Large UVLGs – rare, massive disk systems
Compact UVLGs – small systems undergoing intense star
formation
UV Luminosity, star formation rate (selected)
Size
UV extinction
Stellar mass, velocity dispersion
Metallicity
Compact UVLGs may be useful analogs for LBGs
UV Luminous Galaxies (UVLGs)
Dramatic Evolution to z=3 (DS, Ilbert, Arnouts et al)
Total
(1+z)2.5
Luminosity density of
UV luminous
(LBG-analog)
galaxies shows dramatic
evolution: (1+z)5
LFUV,bol > 1010 Lsol
SFR > 10 Msol/yr
Steeper than QSO
LD evolution (Boyle+
Madau et al)
UVLGs produce
a significant fraction
of LD at z = 1
GALEX AIS + IRAS 
Bivariate SF Luminosity Function
1000 GALEX+IRAS
galaxies
LBG
Do AGNs play a
role in galaxy
evolution?
Yes.
Chandra allows to
separate the X-ray
emission from the
nucleus and the
star-forming ring
Jet-Induced Star
Formation in
Centaurus A
S. G. Neff et al.
•
New GALEX data:
– Deep (~27 mag rms)
– Wide field (1.2o)
•
FUV emission (1500A)
detected:
– along jet(s) for >25 kpc
(shocks)
– where jet hits cold
clouds (young stars)
– where inner jet is
disrupted (???)
– possibly around radio
lobes (young stars?)
5 kpc
~
FUV (1500A)
NUV (2300A)
Minkowski’s Object
(cf. van Breugel)
FUV + HI
Neff, Schiminovich et al.
Results for 65 Sey2:
for central (median) 174 pc (65 Sey 2); 121 pc (14-rest)
Heterogeneous star formation histories.
●
10 SSP BC03 ages, Z=1 and 2.5 solar, plus a power law FC.
Some, dominated by old stars (t>2.5Ga), to 80% of the optical light;
Some show strong component of intermediate age stars (100Ma<t<1.4Ga);
Young clusters are ubiquitous (t<25Ma), in some cases to more than 50% of the
light at 4020A and in several to 20%.
Strong FC component also present. This could be a genuine monster or a dusty
young burst.
At least 3 of the 4 components present with significant strength (more than
10%) in any one galaxy.
A simple Ell galaxy + a power law (used many times before) does not apply to
the bulk of Sey 2s.
Problem can be solved with extreme super-winds
>5x1049 erg per solar mass required
Benson (2003)
Massive X-ray outflow in PDS 456
XMM EPIC pn/MOS
Reeves et al. (2003)
Conclusions
 Overwhelming evidence for CDM
 hierarchical structure formation
 Problems with semi-analytical galaxy formation models
- mechanism required to terminate SF in massive gals
- plus other problems…
 AGN feedback is a likely solution
- may be related to the origin of the M/ relation
- could also explain high-mass cut-off & cluster
heating problem
• Are galaxies sensitive to
their large scale
environment?
• Discussed yesterday.
Other problem:
•How to reconcile
integrated and small
scale properties?
Blue Compact Dwarfs
HII region (opt. +IR em. lines)
HI region (UV abs. lines)






NGC1705
NGC253
IZw18
IZw36
Markarian59
SBS0335-052
[N/H]
[O/H]
[Si/H]
[P/H]
[Ar/H]
[Fe/H]
Refs : Lebouteiller et al. (2003), Lecavelier et al. (2003), Aloisi et al. (2003), Thuan et al.
(2005), Thuan et al. (2002), Lee et al. (2003)
V. Lebouteiller – Moriond 2005
6/18
Distant star formation: what
came first?
Consensus (purely
theoretical):
1000 Mo stars