star formation at high redshift: observational progress and theoretical perspectives romeel davé...

Star Formation at High Redshift:Star Formation at High Redshift:Observational Progress and Observational Progress and

Theoretical PerspectivesTheoretical Perspectives

Romeel DavéRomeel Davé

movie by B. Oppenheimer

Cosmic SF HistoryCosmic SF History

Madau et al 1996

Hopkins & Beacom 2006

x10 from z=1-0

~flatz=4-1

risingto z~4

- Systematics >~x2- Sample overlap?

Breaking down cosmic SFBreaking down cosmic SF Bursty or quiescent?Bursty or quiescent? How important are mergers?How important are mergers? How much is obscured?How much is obscured? Stellar mass growth vs. SFR?Stellar mass growth vs. SFR? Is hi-z SF systematically different?Is hi-z SF systematically different?

Main Sequence of Galaxy Main Sequence of Galaxy

FormationFormation

SFR ~ MSFR ~ M**0.7-0.90.7-0.9, scatter 0.3 dex., scatter 0.3 dex.

Evolves down independently of MEvolves down independently of M** SFR quiescent, not burst, dominatedSFR quiescent, not burst, dominated

Noeske etal 07z~0.2-1.1

Daddi etal 07z~1.4-2.5

Elbaz etal 07z~0.8-1.2

Labbe etal 10z~7

SFR from mergers?SFR from mergers?

Close pairs: <10% global SF in >3:1 mergersClose pairs: <10% global SF in >3:1 mergers Morphological: <30% global SF in interactingMorphological: <30% global SF in interacting BUT– Asymmetry: ~50% z~1 SF in “mergers”BUT– Asymmetry: ~50% z~1 SF in “mergers” Non-interacting galaxies dominate SFRDNon-interacting galaxies dominate SFRD

Robaina et al 2009

Jogee et al 2009

Shi+09

Bouwens+09

Obscured star formationObscured star formation

Most SF obscured; peaks at z~2.Most SF obscured; peaks at z~2. LIRGs/ULIRGs dominate to z~1,2…but don’t LIRGs/ULIRGs dominate to z~1,2…but don’t

interpret as local ULIRG analogs!interpret as local ULIRG analogs!

Dust productionSF to disks

Seymour+09

SFR vs. dMSFR vs. dM**/dt/dt

dM*/dt << SFR (+IMF) at z~2-3dM*/dt << SFR (+IMF) at z~2-3 Completeness? (Reddy+09)Completeness? (Reddy+09) Is SFR at z~2-3 fundamentally different?Is SFR at z~2-3 fundamentally different?

RD 08

Wilkins+ 09

How Do We Understand All How Do We Understand All

this?this? ComputersComputers!!

A Not-totally-wrong Story A Not-totally-wrong Story for Star-forming Galaxiesfor Star-forming Galaxies

Cold flowsCold flows Smooth accretionSmooth accretion Regulation by galactic outflowsRegulation by galactic outflows The Galaxy-IGM connectionThe Galaxy-IGM connection Truncation by mergers/AGNTruncation by mergers/AGN

Cold mode Cold mode accretionaccretion

Cold flows dominate gas infall for ALL star-forming galaxies.

Hot halos for >1012 M

InfallfromIGM

disk

Rvir

Rcool

Dekel+ 09 z=2, AMR

At z>~1-2, flows penetrate hot halo.

At low-z, flows break up into clouds (HVCs?)

z=2, 1012M halo z=1, 1013M halo

Keres+ 09

Cold Mode -- in every hydro sim ever doneCold Mode -- in every hydro sim ever done

SFR tracks accretionSFR tracks accretion

SFGs at all masses, epochs SFGs at all masses, epochs process accreted gas into process accreted gas into stars on t<<tstars on t<<tHH

Keres+09

Accretion is SmoothAccretion is Smooth

Accretion is ~smooth Accretion is ~smooth (& filamentary).(& filamentary).

Mergers subdominant.Mergers subdominant.

Keres etal 09

Generic predictionsGeneric predictions Tight SFR-MTight SFR-M** relation, relation,

evolves indep of Mevolves indep of M** Mergers/bursts sub-Mergers/bursts sub-

dominant in SFRD.dominant in SFRD. Disturbed morphology Disturbed morphology

due to rapid accretiondue to rapid accretion Early galaxies grow Early galaxies grow

rapidly.rapidly.

… … too too rapidly!rapidly!

Need feedback.Need feedback. RD 08

Bournaud+ 07: Unstable disk sim

Data: Bouwens+06 z~6RD et al 06

Feedback Regulates SFFeedback Regulates SF

Halo mass function, scaled by b/m.

Baldry+ 08

Cosmic Feedback CycleCosmic Feedback Cycle

Outflows: Observations & Outflows: Observations & ImplicationsImplications

Common at z~1+: Σ

SFR>>0.1 M⊙/kpc2

ΔvISM

~ hundreds km/s Local SBs, z~1 SFG:

vwvcirc Momentum-driven winds?

If so, mass loading factor

η1/vcirc

Note: Similar scalings may arise from other physical mechanisms.

M82: Spitzer 8μ

Martin 2005

100

Correct (early) evolution of cosmic SFR IGM enriched to match CIV, OVI absorbers Mass-metallicity with right slope & scaller High gas fractions in low-M galaxies DLA kinematics w/enough wide systems Non-grav entropy in intracluster gas

… all from including one simple scaling of outflows with galaxy mass.

Such Outflows Yield…Such Outflows Yield…

Mass in outflows >> Mass in stars

Winds recycle a lot; recycled wind accretion dominates at z<~1.

Many baryons blown out of even large halos

Outflows keep galaxies gas rich (not poor).

MZR set by balance of inflows and outflows, not gas processing rate

… … Some Surprising Some Surprising ImplicationsImplications

Mwind ~ 3 M*

Median Nrecyc~3

BDO & RD 08

Z~M1/3

The “Equilibrium” The “Equilibrium” MZR ModelMZR Model

z=2

Finlator&RD 08

Z ~ y M*/Macc ~ y (1+)-1

If vw>vesc, constant constant Z. For vw<vesc, 0, Zy. Produces a feature in MZR-- not seen.

For mom-driven winds:

~M*-1/3~vc

-1 Z(M*)~M*1/3

Zgas set by an equilibrium between recent inflow and outflows.

680 km/s

340 km/s

vw~vcirc

Recycling sets the GSMFHalo mass function, scaled by b/m.

Baldry+ 08

trec short;Lots of recycled wind mode

trec longer;Lessrecycled wind mode

trec > thubble;Norecycled wind mode

trec irrelevant;Need to quench star formation!

trec irrelevant;Need to quench star formation!

Galaxies as “Gas Galaxies as “Gas Processing Engines”Processing Engines”

Obtain gas via cold, smooth accretionObtain gas via cold, smooth accretion Creates tight evolving MCreates tight evolving M**-SFR relation.-SFR relation.

Process into stars (some) or eject into Process into stars (some) or eject into outflows (most) on a short timescale.outflows (most) on a short timescale. SFR, gas content, metallicity set by SFR, gas content, metallicity set by

balance of inflow vs outflow.balance of inflow vs outflow. Ejected material recycles, moreso in Ejected material recycles, moreso in

larger galaxies.larger galaxies. Sets shape and evolution of galaxy stellar Sets shape and evolution of galaxy stellar

mass function.mass function.

SFR-MSFR-M**: Close : Close but no cigarbut no cigar

Models get right:Models get right: SlopeSlope ScatterScatter General evolutionGeneral evolution

But But amplitude amplitude evolution is evolution is wrong!wrong!

Data has higher Data has higher SFR at given M*.SFR at given M*.

But wait… cold But wait… cold accretion sets accretion sets maximum maximum rate!rate!

RD 08

•Green: Millenium SAM

•Red, magenta: SPH sims

•Blue: Data (=0.3)

Birthrate evolution: Huh?Birthrate evolution: Huh?

Discrepant during the peak dusty SF epoch.Discrepant during the peak dusty SF epoch. Something different? Calibrations? IMF?Something different? Calibrations? IMF?

Gonzalez+ 10

SimulationsSimulations

RD 08

SummarySummary 3 regimes of cosmic SFH: Early growth (z>4),

rapid phase (1<z<4), late decline (z<1) Dominated by fairly smooth, filamentary cold

accretion. Mergers & bursts subdominant. Dust embedded SF peaks @ z~2. Before

that, less dust; after, SF morphology change? Generally, trends well explained by

Cold mode accretion, smooth & filamentary Outflows that eject more mass from small gals.

Mismatch between SFR and M* evolution, both vs. data and theory?

Diff recycling governs GSMF Without wind mode,

GSMF~halo MF (scaled). Wind recycling boosts

higher M* galaxies, flattens GSMF.

Once trecyc > tHubble, reverts to steep slope, modulated by outflows (vzw shallower since ~1/vcirc).

Halo baryonHalo baryoncontentcontent

Winds drive lots of baryons out of halos.

Big galaxies lose baryons early.

MW-like halo today has lost ~½ baryons

Peak SF “efficiency” @ 1013M.

Qualitatively similar to observations.

Winds recycle!Winds recycle! Mwinds >> Mstars

wind ~ 0.2 b ejected ~ 0.5 b Compare: * ~ 0.08 b

Median # ejections: 3

Turnaround radius ~100 physical kpc high-z: Enrich IGM low-z: Halo fountains

Median trecyc ~ 1 Gyr

Constraints from IGM Constraints from IGM enrichmentenrichment Outflows are the only way to enrich IGM z~2-4 CIV absorbers very constraining;

favors momentum-driven wind scalings: vw~vcirc, ~vcirc

-1.

win

d sp

eed

mass loading

Too fewmetals in IGM

IGM too hot

Diffuse IGM unenriched

Too fewmetalsproduced

Momentum-driven wind scalings!

Met

allic

ity

Overdensity

Oppenheimer & RD 2006Oppenheimer & RD 2008Oppenheimer & RD 2009a,b

Gas content fgas(M*) strongly

dependent on outflows. M-D winds keep low-

mass galaxies gas rich by suppressing SF.

Strong outflows make galaxies gas rich, not gas poor.

Galaxies are the antithesis of closed boxes: They process gas as supplied.

Z~M1/3

Mass-Metallicity Mass-Metallicity RelationRelation

z=2

Finlator&RD 08

Conventional thinking: Zgas reflects stage of processing Galaxies lose metals based on .

NO! Why? Outflows are greedy & destructive: They don’t share their energy --

they blow holes!

Analytic 3D simulation

“Wind mode” accretion

Differential RecyclingDifferential Recycling ttrecyclingrecycling goes inversely with mass goes inversely with mass

680 km/s

340 km/s

vw~vcirc

Luminosity functionsLuminosity functions z~6: Large SF

suppression. Rapid consumption

must eject gas. Macc = M*+Moutflow

SFR (1+)-1

z~2-4: Const ~2+ ~ vc

-1 ~1.7 z=0: Faint end slope

~1.3; less steep! (bright end?…fugly)

Data: Bouwens etel 06Models: =0.3, 8=0.9

RD, Finlator, Oppenheimer 06

DLA Kinematics: Outflows?DLA Kinematics: Outflows? Wide separation (v>vrot) DLAs hard to produce;

protogalactic clump infall fails (eg Pontzen etal). Momentum-drive winds puff out gas, produces wide-

separation systems.Prochaska & Wolfe 01

KS test prob

no winds, 8e-5

constant winds, 0.037

MD winds, 0.51

No winds Mom-driven windsConstant winds

S. Hong, Katz, RD etal, in prep

Feedback Feedback TomographyTomography

Background sources Background sources observe outflows in observe outflows in action.action.

Tough now, but will Tough now, but will become routine at become routine at z~2+ in 30m era.z~2+ in 30m era.

COS enables at z<1COS enables at z<1 Also possible in Also possible in

emission around emission around galaxies.galaxies.

The next frontier of The next frontier of observational observational galaxy formation!galaxy formation!