Chris Hayward, CfA (chayward@cfa.harvard.edu)Massive Galaxies Across Cosmic Time 3

November 8, 2010

The contribution of merger-induced starbursts to the SMG population

Collaborators

• Desika Narayanan (Arizona)

• Patrik Jonsson, Lars Hernquist (CfA)

• Dušan Kereš, Phil Hopkins (Berkeley)

• T.J. Cox (Carnegie Observatories)

What drives SF in high-z ULIRGs?

Hopkins, Younger, CCH+10

Submm galaxy selection yields sample of high-z ULIRGs

↑ gas content and compactness at high z → “normal disks” can be ULIRGs

Submm flux not necessarily simply related to LIR b/c dust T variation

Nature of SMG population unclear

Submm counts from SAMs

2005MNRAS.356.1191B

Baugh+05(ν = 850 μm)

•With standard IMF quiescent SF dominates but counts too low•Flat IMF in burst makes bursts dominant

Gadget-2 simulations

• Large suite of major & minor mergers, isolated disks; non-cosmological

• Gadget-2 N-body/SPH (Springel 05)

• Schmidt-Kennicutt SF recipe

• Two-phase ISM of Springel & Hernquist (03)

• BH growth & feedback (Springel+05)

Sunrise radiative transfer• Use Sunrise MC RT code (Jonsson 06; Jonsson, Groves, & Cox 10) in post-

processing to calculate SEDs of simulated galaxies

• Stellar SEDs from Starburst99 (Leitherer+99); Kroupa IMF

• AGN template of Hopkins+07

• Calculate dust attenuation and re-emission

• Solves for dust T iteratively (Juvela 05) to properly treat dust self-absorption -- key for high optical depths encountered in SMGs

• Yields UV-mm SEDs/images from multiple viewing angles

Isolated disk evolution

S850 ∝ SFR0.4

Isolated disk w/ Mhalo = 9e12, Mb = 4e11; initially 60% gas

Submm flux traces SFR well

Hard to produce classical SMG with even with such a massive, gas-rich disk

Merger of two z ~ 2 disks, each w/ Mhalo = 9e12, Mb = 4e11; initially 60% gas

Merger evolution

Two SF regimes:

1. Quiescent disk (during infall)

2. Merger-driven burst

S850 ∝ SFR0.25

S850 ∝ SFR0.4

Bursts inefficient at boosting submm flux (~40x in SFR but <2 in S850) b/c SED significantly hotter at burst and burst does not dominate Lbol

Our model for number counts

1. Submm duty cycles from simulations

2. Merger rates from “semi-empirical” model of Hopkins+10

3. Combine to get number counts:

dN(> Sλ)

dΩ=

∫dN

dV dtd logMbdµdfg(Mb, µ, fg, z)τ(> Sλ,Mb, µ, fg, z)

dV

dΩdz(z)d logMbdµdfgdz

Submm counts from starbursts

1.1 mm

N(>

S)

Merger-driven starburst counts close to observed (Austermann+10) but still below

However, starbursts are not the only way mergers contribute

SMG bimodality

N(>

S)

Engel+10

• Submm bolometers used to identify SMGs have beams ~15” (~130 kpc at z = 2) ⇒ easy

to fit two disks in beam

• Adding two disks is efficient way to boost submm flux

• Will look like “normal” disks; not yet interacting

• Supported by large fraction of SMGs w/ multiple radio counterparts & resolved close pairs (e.g., Engel+10)

• Overlap w/ BzKs Emanuele discussed

Burst+infall counts

Austermann+101.1 mm

N(>

S)

SHADES (0.7 deg2)

Mergers can match counts with standard IMF when both infall and burst stages included

Summary

• We have used high-resolution SPH sims + 3-D RT w/ full dust T calculation to model SMGs; combined with semi-empirical merger rates to predict SMG number counts

• Mergers create SMGs via 2 effects:

1. Starburst induced at coalescence

2. Pre-coalescence: two progenitor disks in submm beam

• Starbursts relatively inefficient at boosting submm flux; infall phase of two disks in submm beam very important

• Match # counts w/ standard IMF

• Unlike local ULIRGs, SMGs are a mix of quiescent and bursting sources; use caution when interpreting observations of high-z ULIRGs, b/c SF mechanism is not uniform