1

VVDS: Towards a complete census of star formation at 1.4<z<~6

Olivier Le FèvreLaboratoire d’Astrophysique de MarseilleWith the VIMOS VLT Deep Survey team

• Magnitude selected spectroscopic redshift survey– Ultra-deep VVDS sample

• Ly flux limited spectroscopic redshift survey

Improved galaxy census

2

Understanding galaxy evolution at z>1.4 requires a complete census

• Star formation, mass assembly history, are derived from galaxy counts

• Are we making a reasonably complete census of star-forming galaxies ?– Color pre-selection vs. flux

selection• Hard job: only 3-4% of

AB~25 galaxies are at z>1.4

Ast

rono

my

Ast

ro-P

hysi

cs

3

Making a complete census

Hypotheses

f(m,z)

Simulationsg(m,size,SB, color)

Seeing

LBG, BzKDRG, SMG,EROs, NB,…

I, K, 3.6µm,…Em. line

g(m,size,SB,color): uncertain forfaint high-z objects

4

VVDS: magnitude limited samples

Sample Selection redshift # galaxiesSpectro-z

# galaxies z>1.4

Wide IAB≤22.5 0<z<1.5

35000 -

Deep IAB≤24 0<z<5 12000 970

Ultra-Deep

IAB≤24.75

0<z<5 863 409

Ly Flux 2<z<6.5

204 204

A total of ~1580 galaxies with 1.4<z<6.6 and spectroscopic redshifts

Check http://cencosnew.oamp.fr/ for public data

VLT-VIMOS

5

New Ultra-Deep sample

• Magnitude selected IAB≤24.75

• 863 galaxies targeted with VLT-VIMOS

• 262 re-observed from failed IAB≤24

• 18h integrations in blue + 18h integrations in red, 3600-9350A coverage

• 86% complete

6

High S/N Combined

spectra

Lym

an-b

reak

7

« Bean counting »

• Producing accurate redshift distributions N(z,m)

• Completness of redshift measurements– Minimized with re-observed samples

• Two independant observations• Reobserved objects with longer integration

times

– Tested with photometric redshifts

8

VVDS IAB≤24200580% complete

Redshift distribution IAB ≤24

Uncorrected for incompletness

9

Redshift distribution IAB ≤24

VVDS IAB≤24, corrected 200996% complete

Corrected for incompletness Corrected for incompletness using high success rate in re-observedsample with 3700-9300A wavelengthcoverage and 4.5x exposure time

Failed on first attempt IAB≤24 sample, nowre-observed with secure z

10

Redshift distribution IAB ≤24.75

VVDS-Udeep IAB≤24.75863 objects

New ultra-deep sample

z_median=1.486% complete

Well into redshift desert because of the 3600-9350A coverage

11

Aposteriori comparison spectroscopic vs. Photometric

redshifts

LBG

Spectacular agreement:

High quality photometrytrained onHigh quality spectroscopic redshifts

Ilbert et al., in prep.

12

Galaxy number density at z~3

• Use N(z)

• Corrected IAB≤24

• Uncorrected IAB≤24.75

~2x more galaxies in magnitude limited survey than in LBG, for all mag IAB

22

Confirms Le Fèvre et al. 2005

Work in progress

CENSUS IS DIFFICULT AT z~3 WORRISOME FOR z>3

13

Use VVDS to check color-color selection: ugr (LBG z~3)

Confirmed spectro-z2.7≤z≤3.5

Confirmed spectro-zz<2.5 and z>3.5

60% contamination 15% not selected

LBGExquisite photometryFrom CFHTLS

Main source of uncertainty from color selection: the depth of photometry

Black: z<2.7Green: stars

VVDS spectro:

Not-LBG

14

Census of LAE:The VVDS Ly emitters

sample• Search for serendipitous emission in the

slits of the VVDS Deep and Ultra-Deep observations

• Use the large wavelength coverage to isolate Ly systems

• Use the exact knowledge of the sky background to estimate the flux limit F()

A complete flux limited LAE survey

15

What are we looking for?

Slits: 1”x5-20”VVDS main target

Photometrically invisible Lya emitter

Target spectrum

Serendipitous Lya @ 1216Å

• 1200 slits, covering 3.3 arcmin2, 3500-9500Å, exp. times 65000s• 8000 slits, covering 22.2 arcmin2, 5500-9500Å, exp. times 16000s

16

The LAE sample

• Flux limit: ~2x10-18 erg.cm-2.s-1

• 86 Lya from UDeep blue

• 25 Lya from UDeep red

• 27 Lya from Deep• 66 spectroscopy

main targets

204 emitters

The largest LAE sample with spectroscopy confirmed z

17

Combined spectra Ly emitters

Combined spectra Ly line profileAsymetric, as expected

18

6 Ly emitters at 6<z<6.6

19

Redshift and Luminosity distribution

• UDeep blue• UDeep red• Deep

First exploration at faint luminosities 1041 erg/s

20

LAE Luminosity function

No apparent evolution from z=2 to z=6.6

But evolution when including dust and IGM corrections

we reach log(lum)=41erg/s:

we can constrain

• Black: literature z=3-6

=1.75

21

Star formation density from LAE

• No correction for dust• No correction for IGM

absorption

• Constant SFD from z=2 to z=6

• Ly contributes to 20% of the SFD at z=2.5, 30% at z=5, >50% at z=6

Cassata et al., submitted

If Lya emission is coming from Star Formation…

22

Summary

• Galaxy studies at z>1 require complete magnitude-selected deep spectroscopic surveys

Star-forming galaxy census at z>1 is not complete yet

• UV-continuum selection spectroscopic survey counts ~2x more galaxies at z~3 as compared to color-selection (LBG)

Color-color selection has build-in inaccuracy

• Faint Ly emitters with log(Ly )<42 are numerous, LF slope =1.7 at z~2-6

Ly SFRD is 30% of the total SFRD at z~3, becoming dominant at z~6

New deep redshifts surveys are required to make progress