Herschel observations: contraints on dust attenuation and star formation histories at
high redshift
Véronique BuatLaboratoire d’Astrophysique de Marseille
Elbaz’s lectures (david.elbaz3.free.fr/coursJ1.html), adapted from Devriendt+99
Visible
Infrared mm
UV
wavelength
inte
nsit
y
dust
Both UV and IR are related to recent star formationThey are anti-correlated because of dust attenuation
IR selected objects are usually obscured with a low residual emission in UV
Conversely: we expect a UV selection to be dominated by IR faint sources
A few words about the physical link between UV and IR emissions
SFR density traced by combining UV and IR surveys
Takeuchi, Buat & Burgarella, 2005
Two parts in this talk….
• Analysis of the full UV-to-IR Spectral Energy Distributions of galaxies, dust attenuation & star formation at z> 1 Véronique
• UV and IR luminosity functions, evolution of global dust attenuation and star formation rate density with redshift up to z~4 Denis
Outline of the first part1. Fitting the full SED in a physical way2. The GOODS-Herschel project: observed SEDs of distant galaxies3. Dust attenuation curve4. Star formation rates and stellar mass determinations5. The Herschel insight on IR spectral energy distributions of distant galaxies
With Sébastien Heinis, Médéric Boquien, Denis Burgarella & the GOODS-Herschel team
1. Modeling the UV-to-IR SEDs: with or without energy budgets
Muzzin+10Stellar and dust emission fitted separately
Stellaremission
Dust emission
Da Cunha+08, Magphys code
Radiation transfer & multi-components: valid for local resolved galaxies
Popescu+12
Simple recipes: dust attenuation curves, IR librariesValid for large samples of unresolved galaxies
CIGALE code, Noll+09
Focusing on SED fitting with conservation of the energy bewteen UV-optical and IR
Metallicity fixed or variable
Stellar spectra and isochrones depend on Z
Usually Starburst law but not always
A set of templates
Various SFH, one or several populationsMinimum age free or fixed
CIGALE : Code Investigating GALaxy Emission
P.I. D. Burgarella (Noll+09) http://cigale.oamp.fr/
OUTPUT PARAMETERS :All based on a Bayesian analysis• input parameters• Stellar Mass•Dust luminosity•Amount of obscuration•D4000 break, slope of the UV continuum….
2. The GOODS-Herschel, PI D.Elbaz
GOODS-N: 10’x15’ PACS+SPIRE, 100, 160, 250, 250, 500 µm 1 mJy @ 100 µmGOODS-S/CDFS: 10’x10’ PACS 100,160 µm down to the confusion limit at 100 µm ~0.7 mJy
GOO
DS-N
orth
fiel
d
SPITZER data
HERSCHEL data
PACS SPIRE
A public website for Herschel extragalactic and cosmological surveys: GOODS-Herschel, HerMES, HRS,
VNGShttp://hedam.lam.fr/
Building the multi-wavelength sample for objects in the CDFS
Z> 1 : to redshift the UV range in the visible optical (UV rest-frame) + IR thermal data
1<Z<3Combining GOODS-Herschel/PACS data and
Subaru/MUSYC broad and intermediate band filters (Cardamone+10)+IRAC & MIPS data (Dickinson+03)
28 photometric bands: SNR > 5 in optical and NIR + SNR> 3 at 24 µm 236 galaxies 80 sources detected with PACS, upper limits for the other ones
GOODS-H-CDFS field at 24-100-160 µm Subaru Intermediate band filters
Some examples of SEDs, with their best fit obtained with CIGALE Buat+11An absorption feature at 2175 A rest frame clearly seen and modeled.The full UV-to-IR SED fitted.Physical parameters derived from the analysis of their PDF (not from the best SED)
Calzetti et al. 94, 00: from spectroscopic datano bump in attenuation curve of local starburt galaxies, moderate rise in UV
3. Dust attenuation curve in external galaxies at 1<z<2
Buat et al. 2011, 2012Calzetti 01__ MW
---- LMC___SMC
starburst
Extinction law ≠Attenuation law
Ilbert et al. 2009- Cosmos field-30 photometric bands from UV to NIR:« A broad absorption excess at 2175 A seems necessary to explain the UV flux of some starburst galaxies »
Composite spectrum 1.5<z<2.5: evidence for a bump for 30% of the
sourcesNoll+09
But no evidence for a bump in Lyman Break Galaxies at z≈2 (Vijh+03, Reddy+08)
Some previous evidence for the presence of bumps at high z
CIGALE: optimized to study dust attenuation curves
different amount of dust attenuation (and same attenuation curve) for the young and the old stellar population
5500 Å slope
Calzetti et al. (2000) UV bump power law
AmplitudeCentral wavelength Width
Eb= 1.6 +/- 0.4 δ = -0.27 +/- 0.17
20% of the galaxies (47% of those detected in IR) have a confirmed bump (2 σlevel)20% of the galaxies (43% of those detected in IR) have a dust attenuation steeper than the Calzetti law (2 σ level)
General shape of the average attenuation curve consistent with an increase in UV similar to that of the LMC supershell extinction curve (steeper than the Calzetti curve on top of which there is a bump at 2175 A) whose amplitude is similar to that found in the LMC supershell, (45% that of the MW one) .
Large dispersions among galaxies
Derivation of a mean dust attenuation curve with a UV bump in the CDFS galaxies with 1<z<2
C00
LMC
SMC
SMC
MW
LMCC00
C00: Starburst law, Calzetti.00
3. Stellar mass and Star Formation Rate derivation for 1<z<3 galaxies
Buat+13, (A&A, under revision)
Impact of Different Star Formation Histories: • exp(-t/τ), age free, decl. τ model 1 stellar pop.•exp(t/τ), age fixed (zf ≈ 8), rising. τ model 1 stellar pop.•An old (age free or fixed) decl. τ pop. + a young, age free pop with a constant SFR 2 st. pop.
When IR data are included SFRs are well determined and are non dependent on the SFHsUncertainties remain larger for Mstar estimates
0.12 dex
0.17 dex
Base line model: 2 pop, both age free
Fits without IR data (MIPS and PACS data omitted): the SFR are poorly measured, stellar masses remain unchanged due to uncertainties on dust attenuation
Stellar mass and Star formation rate derivation (cont’d)
Buat+13 Impact of including or not IR data
w/o IR
with IR
W/o
IR-w
ith IR
Consequences on the measure of the Specific Star Formation Rate:SSFR = SFR/Mstar
Adopted Limit for Starburst galaxiesElbaz+11
Dispersion of the Main Sequence of Galaxies defined with IR data or accurate measures of dust attenuation(Noeske+07, Rodighiero+11, Salmi+12)
log(
SSFR
-with
out I
R)-lo
g(SS
FR-w
ith IR
))
Sargent+12
Dale & Helou (2002)64 semi-empirical templates
Chary & Elbaz (2001) 105 empirical templates
-Draine & Li (2007) models (Draine+07): very useful with Herschel data
4. IR SEDs: new insights from Herschel
Dale+12
Templates for the distant universe dependent on the SFR activity , Elbaz+11
« normal galaxies » Starbursting galaxies
•Main Sequence: on average SFR-M* relation, even if very luminous, PAH are prominent•Starburst galaxies: SFR/M* higher than for Main Sequence, less PAH, warmer dust
DIFFERENT FROM LOCAL TEMPLATES OF SIMILAR LUMINOSITY:The contribution of PAHs for high-z luminous galaxies is larger than in nearby galaxies of same IR luminosity
When using local templates of CE01, Lir calculated with 24 µm is overestimated at z=2, (Nordon+12)
@ z=2 8 μm rest frame, PAH dominated
« ex
act »
log(
LIR)
Original CE01 templates With the New Main Sequence templateL IR
from
mon
ochr
omati
c flu
x/L IR
(tot
al-a
ll M
IPS-
Hers
chel
dat
a)
With Herschel, the situation remains complex….
Ciesla+13, in prep.
To be optimistic: Lir estimates do not strongly depend on the adopted template when several bands are available
Casey 12Ciesla, 12, PhD thesisHerschel Reference Sample, SPITZER+ SPIRE data
Conclusions•Fitting the full SED from the UV to the IR with a conservation of the enery balance is very powerful to constrain dust attenuation and consequently SFRs•The attenuation curve of galaxies with 1<z<2 is steeper than the law for local starbursts and is found to exhibit a bump at 2175 A with an amplitude similar to that of the LMC2 extinction curve, •Without IR data, SFRs can be over or underestimated by factors larger than 2 inducing a similar uncertainty for the SSFRs. •Stellar masses derivations are more sensitive to the adopted star formation history than SFRs. •IR SEDs at all z are constrained with Herschel data, modifying the local templates, but more work is needed to build reliable, detailed SEDs
All these analyses are based on data from the CDFS obtained in the framework of the GOODS, GOODS-Herschel and MUSYC projectsThank You!