The cosmic evolution of star formation and metallicity

The cosmic evolution of star formation and metallicity over the last 13 billion years

(an observational perspective)

Andrea Cimatti (INAF - Osservatorio Astrofisico di Arcetri)

OUTLINE

SFR indicatorsHigh-z star-forming galaxies“Fossil” galaxies

Cosmic evolution of SF densityCosmic evolution of stellar mass“Downsizing”

Metallicity indicatorsMetallicity of high-z galaxiesCosmic evolution of the mass-metallicity relation

The 1996+ revolution

HST

ESO VLT JCMT

Keck

The “historical” Lilly – Madau plot

Lilly et al. 1996 Madau et al. 1996

Star formation rate (SFR) main indicators

L(recombination lines) (e.g. Hα) (primary)L(forbidden lines) (e.g. [OII]3727) (empirical, not universal) L(Lya)L(UV continuum) from OB stars (1500-2800 Å)

L(FIR) (and L(MIR) ?) (10-1000μm)L(radio) (1.4 GHz) L(X) (2-10 keV)

Caveat: AGN “contamination”, dust extinction, IMF assumption

Specific star formation rate = SSFR = SFR/(stellar mass) [yr-1]

Small SSFR most mass was already built-up in the pastLarge SSFR significant mass is still building

Star formation

The star formation that we see:

high-z galaxies which are forming stars

Optical selection based on broad-band colors

Steidel et al. 2005

Magenta (“BM” selection): 1.5 < z < 2Cyan (“BX” selection): 2 < z < 2.5Yellow+Green (LBG selection): z ~ 3

BM BX LBG

Optically-selected star-forming galaxies at 1<z<4

(Shapley et al. 2003)

< log M(stars)/Msun > = 10.3 ± 0.5

< SFR > = 30 ± 20 Msun/yr

0 < E(B-V) < 0.3

N ~ 3x10-3 Mpc-3

1/3 < Z/Zsun < 1

(Steidel et al. 2004, Reddy et al. 2005, Shapley et al. 2003, 2005, Erb et al. 2006)

Selected in the optical with the so called BM/BX/LBG color criteria (Steidel et al. 1996, Adelberger et al. 2004)

Photometric candidates at 7 < z < 10

HUDF data. Bouwens et al. 2004, 2005

No secure genuine “primordial” (Pop III)objects identified to date

…

K- to mm-selected dusty starbursts (1<z<5)

E(B-V)>>0.3< log M(stars,,gas)/(Msun) > ~ 11SFR ~ 100-(1000) Msun/yr, Z ~ ZsunN ~ 10-4 Mpc-3 (10-5 for submm galaxies)

Problem for galaxy formation models

(dEROs, SMGs, DRGs, sfBZKs, HyEROs, IEROs…;Totani et al. 2001, Cimatti et al. 2002, Daddi et al. 2004, Chapman et al. 2005; Franx et al. 2003; Chen et al. 04)

SpitzerIRAC-EROs

Submm/mm galaxiesK-selected starbursts

Dusty EROs

High-z dusty AGN

Dust thermal emission from a quasar at z=6.42 CO(3-2) emission from the same quasar(Bertoldi et al. 2003) (Walter et al. 2004)

Many high-z quasars have high FIR luminosity (up to 1e13 Lsun), dust continuum emission consistentwith mass of ~ several x 1e8 Msun and molecular gas with mass of the order of 1e10 MsunSFR > 1000 Msun/yr !?

Emission line galaxies

Line emitting galaxies are generallyfound with narrow-band imaging or “slitless” spectroscopy (1 < z < 6.6)

McCarthy et al. 1999, Hu et al. 2002, Glazebrook et al. 2004, Kurk et al. 2004,Malhotra et al., Rhoads et al. , Taniguchiet al. 2005, Bunker et al., Doherty et al. 2006

Lya at z=6.56 (Hu et al. 2002)

Kurk et al. 2004

Lya at z=6.54

OPTICAL SELECTION

NEAR-IR SELECTION

The star formation that we do not see:

“fossil” galaxies which had star formation

Old passive spheroids at z>1

E/S0 galaxiesPassively evolving1 < z < 21 – 4 Gyr oldM(stellar) > 1011 Msun

Problem for galaxy formationmodels

z(SF onset) > 2 – 3Short-lived, powerful starbursts

It is possible to derive SF history from spectra

Cimatti et al. 2002, 2004, McCarthy et al. 2004, Daddi et al. 2005, Saracco et al. 2005

A massive galaxy candidate at z~6.5

Photometric candidate (no spectroscopic redshift)Consistent with a galaxy at z=6.5 with a large stellar

Stellar mass of 6e11 Msun (!) z(form) > 9

Alternative: very dusty starburst at z=2.5

(Mobasher et al. 2005)See also Eyles et al. 2005, Yan et al. 2006

Other massive galaxy candidates at 5 < z < 8

z J H K 3.6 4.5 5.8 8.0 24 micron

HST ACS: B+V+I+zK ≥ 25 (AB)

STACKING (3”x3”)

Rodighiero et al. 2006

The cosmic evolution

The Lilly-Madau plot 10 years ago

The Lilly – Madau plot now

Hopkins et al. 2005, 2006

Hatched and green: 24μmRed star: radioBlue: optical/UV

DLAs

Cosmic evolution from “archeology” of z~0 galaxies

Heavens et al. 2004

SDSS data + MOPED

Dependence on luminosity and sample selection

K-selected samples miss a significant fraction of SF galaxies with L<L*At all z, L>L* galaxies contribute only 1/3 to the SFR densityL<L* galaxies are the dominant sites of star formationSFR density ~constant at 1<z<4, drops by 2x at z~4.5(Gabasch et al. 2005)

“Downsizing” (Cowie et al. 1996)

Thomas et al. 2004

Constraints from σ, Hβ, Mgb, <Fe>, stellar populations

More massive spheroids form earlier and fasterFormation time scales independent of environment~1-2 Gyr younger in low density environmentsMass assembly almost completed around z~1

(see also Cimatti, Daddi & Renzini 2006)

Dependence on mass and environment

Mass-dependent SFHs for z~0 galaxies(Heavens et al. 2004)

Latest results confirm that massivegalaxies which dominated cosmic SFat z~3 are in clusters today, whereasgalaxies dominating SF at z~0 inhabit low density regions (Poggianti et al. 2006,Sheth et al. 2006)

Early-type galaxies

The evolution of the stellar mass function

Stellar mass function evolution pergalaxy type (Bundy et al. 2006)Shaded areas = 1 σ confidence regions

Increase of N(red) mirrored by decrease of N(blue)

Fractional contributions of red and bluegalaxy populations to the stellar mass function

Largest sample analyzed to date:DEEP2 spectroscopy + optical-NIR SEDs>8000 galaxies over 1.5 square degrees (4 fields)

M(tr)

Specific star formation

Feulner et al. 2005

SSFR = SFR/M(stars)

Higher in lower massgalaxies at all redshifts

Oldest stars in largestmass galaxies

Massive galaxiesare in a quiescent stateat z<2 (no significantchange in stellar mass) Strong increase of<SSFR> at z>2-3 for most massive galaxies

Downsizing of SF

See also Juneau et al. 2005,Caputi et al. 2006

Metallicity

Metallicity indicators

IONIZED GAS

R23 = ([OII]3727+[OIII]4959,5007) / Hβ (Pagel et al. 1979)N2 = [NII]6584 / Hα (Denicolò et al. 2002)O3N2 = ([OIII]5007/Hβ)/([NII]6584/Hα) (Pettini & Pagel 2004)R23 + [OIII]5007/[OII]3727 (Nagao et al. 2006)[NeIII]3869/[OII]3727 (Nagao et al. 2006)

CAVEAT: shock-ionized gas, AGN photoionization

ISM and STARS

Optical absorption features in E/S0 galaxies (e.g. Lick indices, Fe4383)Iron absorption lines at 2000-3000 Å (e.g. Savaglio et al. 2004)UV absorption features (e.g. 1370 Å, 1425 Å, 1978 Å, Rix et al. 2004)Metal absorption lines in DLAs (e.g. Pettini et al. )

Nagao et al. 2006 Emission line indicators

Stellar mass – ionized gas metallicity relation

Tremonti et al. 2004 (SDSS)

Stellar vs. ionized gas metallicity

Gallazzi et al. 2005 (SDSS)

Mass – metallicity relation at 0.4 < z < 1.0A M-Z relation exists at <z>~0.7 and evolves with redshiftAt a given mass, a galaxy at z~0.7 has lower metallicity vs. z~0Evolution more rapid at lower masses. Massive galaxies have Z(solar) at z~0.7 (bulk of SF completed)A more rapidly declining SF in more massive galaxies is consistent with the results (downsizing…)

(see also Carollo & Lilly 2001, Lilly et al. 2003, Kobulnicky & Kewley 2004, Maier et al. 2004, 2006)

Savaglio et al. 2005(CFRS + GDDS samples)

Optically-selected star-forming galaxies at z~2

Shapleyet al.2004

Erb etal. 2006

Mass – metallicity relation at z~2

Erb et al. 2006Optically-selected

Metal-rich starbursts at z>2

Submm galaxies (Tecza et al. 2004, Swinbank et al. 2005) Distant Red Galaxies (J-K>2.3) (van Dokkum et al. 2004) K-band bright optically-selected galaxies (BX) (Shapley et al. 2004) BzK-selected starbursts (De Mello et al. 2004)

Very few observations

Emission line ratios and UV absorptions suggest solar to super-solar metallicities

Metallicity at z>3

Metal abundance derived from R23

1/10 < Z/Zsun < 1 (highly uncertain)

For the only certain galaxy: 1/6 < Z/Zsun < 1/2

Pettini et al. 2001

A cautionary tale…

[OII], Hβ, [OIII], Hα, [NII]Line ratios imply AGN and/or shock ionization (winds)

H-band spectrum only low metallicityK-band spectrum only high-metallicity

(van Dokkum et al. 2005)

z ~ 2.5 K-selected

The problem of “missing metals”

For a given IMF and a mean stellar yield (e.g. <y>=2.4%, Madau et al. 1996), the total amount of metals formed by a given time t is:

ρ(Z,t) = <y> ∫ dρ(stars,t)/dt

Only a fraction of the expected metals is actually seen in galaxies !

At z~2 : 5% DLAs 5% Submm galaxies 5% Distant Red Galaxies15% Optically-selected star-forming galaxies~30% (50-60% if corrected for incompleteness) (Bouché et al. 2006a, 2006b)

At z~3 the problem is even more serious :5-10% Lyman-break galaxies

The rest could be in hot phase with T~106 K (e.g. Ferrara, Scannapieco & Bergeron 2005)

Multi-wavelength surveys are needed to unveil diverse populations of high-z star-forming galaxies (but no Pop III objects detected yet)

The cosmic SFR density increases rapidly to z~1-2, but evolution unclear at z > 2

The old, massive, passive E/S0 galaxies already present at 1< z < 2 require star formation onset at z > 2-3 and short-lived powerful starbursts

Dusty, massive, high-metallicity starbursts at z~2-3: E/S0 progenitors ?

Mass is more important than environment in driving galaxy evolution

“Downsizing”: massive galaxies form stars earlier and faster

A mass-metallicity relation exists up to z ~ 2 (“downsizing” evolution)

Only a fraction of the expected metals is seen in galaxies

New generation of hierarchical merging models start to agree better with obs

The global picture