Far-infrared properties of UV selected galaxies from z=4 to z=1.5: unveiling obscured star

formationVéronique Buat & Sebastien Heinis

With the contribution of Laure Ciesla

Based on HerMES/SPIRE data in the COSMOS fieldHeinis, Buat et al. 2013

From Exoplanets to Distant Galaxies: SPICA’s New Window on the Cool Universe18-21 june 2013-University of Tokyo, Japan

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

Only few words about the physical link between UV and IR emissions

We perform a UV rest-frame selection in the COSMOS field @ z=1.5, 3 & 4

Based on photometric redshifts (Ilbert+13)Down to u, r, i ≈ 26 ABmag

What do we find within HerMES/ SPIRE images?

Almost nothing…….

Z = 1.5

Stacking per bin of LFUV

LIR measured by fitting Dale & Helou (2002) templates on SPIRE data

AFUV = f(LIR/LFUV) (Buat+05) LIRGs and sub-LIRGs

Stacking per bin of M*(again on the UV selection)

Z=1.5Z=3Z=4

Stacking per bin of (LFUV,M*)

Dust attenuation increases with M* for a given LFUV

Dust attenuation decreases with LFUV for a given M*

The dispersion in dust attenuation decreases with LFUV

See also Burgarella+06, Buat+09,12

A recipe to derive LIR/LFUV=IRXHeinis+13, very close to be submitted

IRX=log(LIR/LFUV)=IRX0(LFUV)+0.72*log(M*/1010.35)@z=1.5 & 3

SFR versus M* : a well defined ‘Main Sequence’

for star-forming galaxies

SFR= SFR0 M*0.7

slope< 1

see also Noeske+07, Oliver+10, Whitaker+12

slope~1 found by Elbaz+07, Daddi+07, Wuyts+11

(from Kennicutt, 98)

Z=1.5Z=3Z=4

Specific SFR (sSFR=SFR/M*) : very active galaxies at z =3 & 4, a challenge for the models

sSFR(z,M*) high redshift galaxies (z~2.5-4) stay only around 1 Gyr on the Main Sequence, this time increasing with decreasing redshift

What can we do with SPICA?• Herschel was unable to detect individual

galaxies selected in UV at z >= 1.5 (less than 1% of the galaxies directly detected)

Studies based on a stacking technics: average trends only, no or little dispersion measured

• We must increase the number of individual detections if we want to discuss the variety of physical properties of individual galaxies

Individual detections with SPICA?

Dale & Helou 02 templates

α=2

α=1.5

Assuming 50 µJy @ 70 µm

How many galaxies in the COSMOS field?•z=1.5 9419 galaxies/deg2

•z=3 3162 galaxies/deg2

Assuming a Dale & Helou template with α = 2

As a function of LFUV

How many galaxies could be detected in the COSMOS field?•z=1.5 9731 galaxies/deg2

•z=3 3493 galaxies/deg2

•z=4 952 galaxies/deg2

As a function of M*

Which template to measure LIR? What do we learn from

Herschel? Z=0Z=1.5

Ciesla+13, in prep.

The determination of LIR with a single monochromatic measurement might lead to large uncertainties

Several bands might be very useful to constrain the SED

Still some work to be made to refine SEDs……

Ciesla+13, in preparation

Conclusions• To measure the dust emission of UV bright high

redshift galaxies is challenging• HERSCHEL was not able to detect them

individually at z ≥1.5• Deep photometric observations with SPICA @

λ≈70 μm will allow direct detections of several thousands of galaxies per deg2 also observed in optical (UV rest-frame)

• Coordinated deep surveys with SPICA instruments (SAFARI-MCS-FPF) would provide full SEDs of these galaxies, allowing physical analyses.

Stacking per bin of M*(again on the UV selection)

Stacking per bin of LFUV

LIR measured by fitting Dale & Helou (2002) templates on SPIRE data

AFUV = f(LIR/LFUV) (Buat+05) LIRGs and sub-LIRGs