The Main Mode of Galaxy/Star Formation?

Avishai Dekel, HU JerusalemLeiden, September 2008

HU Flow Team Birnboim, Freundlich, Goerdt, Neistein, Zinger

Simulations Teyssier, Pichon, Kravtsov

Massive Disk Buildup & Breakup by Cold Streams at

z=2-3

Outline

• Massive galaxies at high z: ‘’narrow cold streams through hot halos

• Inflow rate into the halo and into the disk

• Smooth flows vs Mergers

• High-SFR galaxies at z=2-3

• Disk breakup and bulge formation

Shock-Heating Scale

Mvir [Mʘ]

Birnboim & Dekel 03 Dekel & Birnboim 06

stable shock

unstable shock

typical halos

6x1011

Mʘ

Keres et al 05

Gas through shock: heats to virial temperaturecompression on a dynamical timescale versus radiative cooling timescale

t cool−1 t compress

−1

Shock-stability analysis (Birnboim & Dekel 03): post-shock pressure vs. gravitational collapse

t compress≡215ρρ≈4

3

RsV

Libeskind, Birnboim, Dekel 08

d(Entropy)/dt

A virial shock in a 3D cosmological simulation: at Mcrit – rapid expansion from the

inner halo to Rvir

At High z, in Massive Halos: Cold Streams in

Hot Halos

Totally hot at z<1

in M>Mshock

Cold streams at z>2

shock

no shock

coolingDekel &

Birnboim 2006

density

Temperature

adiabatic infall

shock-heate

d

cold flows

disk

Analysis of Eulerian hydro simulations by Birnboim, Zinger, Dekel, Kravtsov

Mass Distribution of Halo Gas

M*

Mvir [Mʘ]

all hot

1014

1013

1012

1011

1010

109

0 1 2 3 4 5 redshift z

all cold

cold filamentsin hot medium

MshockMshock>>M*

Mshock~M

*

Cold Streams in Big Galaxies at High z

Dekel & Birnboim 06 Fig. 7

the millenium cosmological simulation

high-sigma halos: fed by relatively thin, dense filaments → cold narrow streams

typical halos: reside in relatively thick filaments, fed ~spherically → no cold streams

Ms~M*

Ms>>M*

Large-scale filaments grow self-similarly with M*(t) and always have typical width ~R* ∝M*

1/3

At high z, Mshock halos are high-σ peaks: they are fed by a few thinner filaments of higher density

Origin of dense filaments in hot halos (M≥Mshock)at high z

At low z, Mshock halos are typical: they reside in thicker filaments of comparable density

Gas Density in Massive Halos 2x1012Mʘ

Ocvirk, Pichon, Teyssier 08

high z low z

M=1012Mʘ

M=1012Mʘ

Critical Mass in Cosmological Simulations

Ocvirk, Pichon, Teyssier 08

Mstream

Mshock

DB06cold filamentsin hot medium

Massive high-z disks by cold narrow streams

Dekel et al. 2008, Nature

MareNostrum AMR simulation 50 Mpc cosmological box 1 kpc resolution Ocvirk, Pichon, Teyssier 2008

Mvir=1012Mʘ at z=2.5

Gas density following dark-matter filaments

Entropy: virial shock & low-entropy streams

log T / ρ2/ 3

Inward gas flux: all in the streams

m= ρ v r r2 [ M⊗ yr−1 rad−2 ].

Another example

Always 3 streams?

Flux per solid angle

Flux per solid angle

Average Assembly Rate into Rvir by

EPSNeistein, van den Bosch, Dekel 06; Birnboim, Dekel, Neistein 07; Neistein & Dekel 07, 08

⟨ M b⟩ vir≈6 . 6 M Θ yr−1 M 12

1 . 15 1z 2. 25 f 0. 165

d ln Mdω

≈ − 2/ π 1/ 2 σ 2 M / q −σ2 M −1/ 2

ω≡δ c

D t q≈2. 2

Growth rate of main progenitor (time invariant):

Approximate for LCDM

M=2x1012Mʘ z=2.2 dM/dt ~ ~ 200 Mʘyr-

1

May explain high-SFR galaxies if - a similar flux penetrates to the disk - it is gas rich - SFR follows rapidly

Inflow Rate into the Disk

At z~~2-3, M~~1012Mʘ, the input rate into the disk is comparable to the infall rate into the virial shock, most of it along narrow streams

Conditional Distribution of Gas Inflow Rate

mergers >1:10

⟨ M b⟩ ≈6 . 6 M Θ yr−1 M 12

1. 15 1z 2 . 25

P M ∣ M

smooth flows

n M =∫0

∞

P M ∣ M n M dM

Assume scaling of P(Mdot|M)

M b≈6. 6 M Θ yr−1 M 12

1. 15 1z 2 . 25

P(M) by Sheth-Tormen

Comoving Number Density of Galaxies as a function of gas inflow

rate

Gas inflow rate > SFR but by a small margin SFR very efficient!

Star-Forming Gal’s

Sub-Millimeter Gal’s

SFR=

prediction, e.g., n=2x10-4 SFR<200

Streams in 3D: partly clumpy

Half the stream mass is in clump >1:10

Birnboim, Zinger, Dekel, Kravtsov

Inflow Rate into the Disk

50% of the flux is in mergers > 1:10

but the duty cycle is < 10%

n M =∫0

∞

P M ∣ M n M dM

Fraction of Mergers

BzK/BX/BM are mostly mini-minor mergers <1:10, i.e. smooth flowsBright SMG are half-and-half mergers >1:10 and smooth flows

SFG: Stream-Fed Galaxies

At a given Mdot, 75% of the galaxies are fed by smooth flows

Streams – Clumpy Disk - Bulge

One stream with impact parameter ~~ Rdisk

determines the disk spin, while other streams generate turbulence

The streams provide continuous rapid gas supply into a disk Jeans instabilityAt z>2, the streams maintain high dispersion:

Giant clumps. They interact, lose AM, and coalesce into a compact spheroid - “classical bulge” (Noguchi 99; Elmegreen, Bournaud, Elmegreen 08)

M V 2 t circ≈Mσ 2 σ2

V 2≈0 . 1

RJeans≈ 3σ 2

V 2 Rdisk

merger

hhalos

accretion

disk

Old Paradigm

radiative cooling

cold

hot

cold gas young stars

spheroid

old stars

hot halo

spheroidclumpy

disk

New Paradigm

z>2

clumpydisk

spheroid

cold streams

hot halo

thick disk

Mv>1012

z<1

new disk

Detectable by absorption:

External source: c.d.>20 cm-2 at 30% sky coverage

Internal source: c.d.>21 cm-2 at 5% sky coverage

Column density of cold, in-streaming gas

Here is what it should look like in Hα

M*

Mvir [Mʘ]

all hot

1014

1013

1012

1011

1010

109

0 1 2 3 4 5 redshift z

all cold

cold filamentsin hot medium

MshockMshock>>M*

Mshock~M

*

When and where did most stars form?

Dekel & Birnboim 06

at z=2-3 in galaxies of ~~1011 Mʘ by cold streams

Conclusions: SFG = Stream-Fed Galaxies

At z=2-3,=2-3, “disks” of ~~1011Mʘ grow rapidly via narrow cold gas streams through shock-heated halos

Penetration: disk input rate ~~ halo entry rate ~ 200 M~ 200 Mʘ ʘ yryr-1-1

Streams are half Streams are half clumps >1:10 and half smooth, merger duty cycle <0.1

Abundance: SFR>150 at n~~3x10-4 , SFR>500 at n~~6x10-5

Most sBzK/BX/BM are fed by smooth streams in halos 1012-13Mʘ

Half the SMG are mergers >1:10

At z>2, streams generate rotation & maintain gas-rich & turbulent disk giant clumps (1) SFR & (2) coalescence into a spheroid

The cold streams should be detectable at z>2 in absorption and emission (DLAS? Lyman-limit? Lyman-alpha emitters?)

Observed SFGs SFR must closely follow gas input rate