The Role of Galaxy Mergers Among High

Redshift ULIRGsJeyhan S. Kartaltepe

Hubble Fellow - NOAO

Hubble Fellows Symposium2012 March 7

The Role of Galaxy Mergers Among High

Redshift ULIRGsJeyhan S. Kartaltepe

Hubble Fellow - NOAO

Hubble Fellows Symposium2012 March 7

z < 0.3 z~2z~2

What are LIRGs/ULIRGs?LIR(8-1000 mm) > 1011 L LIRGs

LIR > 1012 L ULIRGs, LIR > 1013 L HyLIRGs!!Rare locally …

PAH Complex

DustStellar ContinuumArp 220

Kennicutt et al. 2003

High Redshift (U)LIRGs

ULIRGsLIRGs

Comoving Infrared Energy Density

Low Lum

Le Floc’h et al. 2005

LIRGs dominate IR energy density (and cosmic star formation rate) at z > 0.7 Le Floc’h et al. 2005

ULIRGs as important or dominate by z~2 Caputi et al. 2007; Magnelli

et al. 2009, 2011; Bethermin et al. 2011; Murphy et al. 2011

Important role at the peak of galaxy assembly

MotivationWhat drives star formation and AGN activity

and how does this change as a function of redshift?What is the role of galaxy mergers?Are mergers necessary for the extreme

luminosities of (U)LIRGs?What was the role of ‘cold flow’ gas

accretion in the early universe?How does galaxy morphology correlate with

luminosity and therefore star formation rate?

Properties of Local (U)LIRGs

ULIRGs: IRAS 1 Jy Sample, med(z) = 0.145 (Veilleux, Kim, & Sanders 2002) > 99% are major mergers of gas rich spirals

LIRGs: RBGS, med(z) =0.0082 (Sanders et al 2003, Ishida 2004)

log(LIR) > 11.5 strongly interacting major mergers (65%) doubles (18%) minor interactions (18%)

log(LIR) < 11.5 strongly interacting major mergers (36%) doubles (23%), minor interactions (15%) high luminosity end of normal starforming disks (26%)

Fraction of mergers increases systematically with LIR!

Properties of Local (U)LIRGsFraction of (U)LIRGs with an AGN increases with

LIR Veilleux et al. 1995, 1999; Tran et al. 2001; Yuan

et al. 2010< 20% for LIR < 1011 L

> 50% for LIR > 1012.3 L

Large fraction of composites Mix of SF, AGN, shocks Difficult to disentangle

Yuan et al. 2010

The Merger Scenario Mergers among

ULIRGs are ubiquitous

High AGN fraction among ULIRGs

Leads to the merger scenario (Sanders et al. 1988) Evolutionary

connection: Gas-rich mergers LIRG ULIRG QSO

Eventually form “red and dead” elliptical

Hopkins et al. 2008

The Role of Cold FlowsTheoretical simulations suggest that the high SFRs

of (U)LIRGs can be sustained at high redshift (z~2) by ‘cold flows’ (e.g., Dekel et al. 2009, Davé et al. 2009) Accretion of cold gas along filaments Minor mergers

Observationally supported by apparent lack of mergers among high redshift (U)LIRGs (e.g., Genzel et al. 2006; Forster-Schreiber et al. 2009) Mix of results at high redshift so far High redshift mergers are hard to see Cold flows have yet to be observed!

High Redshift (U)LIRGsAre mergers necessary to produce these extreme

luminosities?Studies at high redshift have been limited so far

Small samples Uncertain luminosities Lack of rest-frame optical imaging

Most previous studies have been based on Spitzer 24 mm selection Pre-Herschel, longer wavelength data has only been

available for small numbers of objects (e.g., Spitzer 70/160 mm, Submillimeter)

Previous High Redshift (z~2) Studies

24 mm / color selected samples e.g., Dasyra et al. 2008; Melbourne et al. 2009; Bussmann et

al. 2009, 2011; Zamojski et al. 2010 Wide range of results

Submillimeter Galaxies (SMGs) Morphology (e.g., Conselice et al. 2003; Swinbank et al. 2010) Kinematics (e.g., Tacconi et al. 2008) Tend to find high fractions of mergers

IFU Kinematics of massive star forming galaxies (SMGs/BzKs, etc.) e.g., Genzel et al. 2008; Forster-Schreiber et al. 2009 Mixed results – 1/3 mergers, 1/3 rotation dominated, 1/3

dispersion dominated

GOODS-HerschelHerschel coverage of both GOODS fields (PI: D. Elbaz)

Deepest Herschel data taken! Full GOODS-N field: 1.6 mJy (@ 100 mm) Central part of GOODS-S: 0.6 mJy (@ 100 mm)

Small area coverage but very deepWell suited for identifying high

redshift (z > 1) sources100-500 mm coverage is closer

to the peak of emission Better for constraining LIR

and Tdust

GOODS-N GOODS-S

Identifying Merging Galaxies

Easy in the local universe!

Nearby Examples: The

Mice & NGC 520

Identifying Merging Galaxies

Easy in the local universe!

At high redshift surface brightness dimming becomes a problem…

Hibbard & Vacca 1997

Identifying Merging Galaxies

And band shifting becomes important!Rest-frame UV versus Rest-frame optical

ACS I band

WFC3 H band

CANDELSCosmic Assembly Near-infrared Deep Extragalactic

Legacy Survey (PIs: S. Faber & H. Ferguson)Multi-cycle HST Treasury ProgramWFC3 NIR imaging of portions of 5 deep fields

GOODS-N, GOODS-S, UDS, COSMOS, EGS

UDS Mosaic

GOODS-Herschel & CANDELS

A match made in heaven!

Rest-frame optical imaging for z~2 galaxies

Ideal for probing structure of z~2 ULIRGs

GOODS-Herschel ULIRG Sample

• 52 ULIRGs with CANDELS imaging

• First complete, FIR selected sample of ULIRGs at z~2!

• Additional 70 LIRGs over this z range

<z> ~ 2.2

<log(LIR)> ~ 12.26

• Focus on all ULIRGs with z = 1.5 - 3.0 in GOODS-S

Comparison SampleSelected 260 comparison galaxies (5 for each ULIRG)

Not Herschel detected less luminous z~2 populationMatched to redshift and H magnitudeRandomized and classified

ULIRGs + comparison Visual classification scheme Classified by me + 3-5 other

classifiers Analyzed agreement Eventually compare to

quantitative merger selections

Visual Classification Scheme

Main Morphologi

cal Class

Interaction Class

Results

Merger+Int+IrrKartaltepe et al. 2012

Comparison with z ~ 1 COSMOS 70 mm

277 sources 0.8 < z < 1.2 Relatively shallow, but large

area coverage log(LIR) = 11.3 - 12.9

GOODS-Herschel North+South 100/160 mm Smaller area, significantly

deeper 293 sources w/ 0.8 < z < 1.2 Log(LIR) = 10.6 – 12.4

Classified using ACS imaging and same scheme

”Main Sequence” and Starburst Galaxies

Correlation between star formation rate (SFR) and stellar mass True locally (e.g., Brinchmann et al. 2004) and at high

redshift (e.g., Noeske et al. 2007; Daddi et al. 2007) Evolves with redshift (increase in gas fraction?)

Galaxies elevated above this relation Starbursts

Noeske et al. 2007

”Main Sequence” and Starburst Galaxies

Elbaz et al. 2011

Rodighiero et al. 2011

What is the role of mergers among starbursts and “main sequence” galaxies at z~2?

Are z~2 (U)LIRGs Starbursts?Main Sequence (MS)

from Elbaz et al. 2011

MS x3

Kartaltepe et al. 2012

Are z~2 (U)LIRGs Starbursts?

Starbursts

Starbursts

Kartaltepe et al. 2012

Starbursts

Starbursts

Kartaltepe et al. 2012

Are z~2 (U)LIRGs Starbursts?Disks:

med(sSFR/sSFRMS) = 2.4

Starbursts

Starbursts

Kartaltepe et al. 2012

Are z~2 (U)LIRGs Starbursts?Interactions &

Mergers:med(sSFR/sSFRMS) =

3.8

Starbursts

Kartaltepe et al. 2012

What are z~2 “Main Sequence” Galaxies?

“Main Sequence”Non-interacting disks: 57%Interactions & Mergers: 24%

Starbursts

Kartaltepe et al. 2012

Are z~2 Starbursts Mergers?

StarburstsNon-interacting disks: 42%Interactions & Mergers: 50%

ConclusionsMorphologies of z~2 ULIRGs span a wide

range‘Disks’ make up a significant fraction (many

irregular)40% non-interacting, 60% total

~45% are mergers or interactionsAdditional ~25% are irregular (minor mergers?)Comparable to fractions at z~1, slightly lowerMore likely to be interacting pairs than advanced

mergers z~2 ULIRGs are more like z~0 LIRGs

ConclusionsMost (U)LIRGs on the main sequence are

non-interacting disks (57%)BUT significant (24%) number are

mergers/interactionsSignificant enhancement of SFR for a short

time during the mergerHalf of starbursts are clear

mergers/interactions Large fraction of “Irregular Disks”

Enhanced role of minor mergers?

Future Directions(aka: Next Year’s Talk)

The results presented were small numbers! Need rest of CANDELS fields + expanded

Herschel coverage for firm conclusions/statistics

Herschel+CANDELS – PI: M. Dickinson (OT2)Coming soon!

NIR spectroscopic follow-upStarburst-AGN emission line diagnosticsWhat is the role/contribution of the AGN?Large international Subaru FMOS program of

COSMOS

Starburst Montage

Fiber Multi-Object Spectrograph (FMOS)Low-res (R~600) Hi-res (R~2200) In low-res mode,

can cover 0.9-1.8 mm at once

400 fibers30’ diameter FOV

Ideal for COSMOS!

AGN Fraction

5% at low LIR

>70% for ULIRGs

100% of HyLIRGs

log (LIR)

AGN

Frac

tion

AGN Fraction increases systematically with LIR (as it does locally)!

Kartaltepe et al. 2010a

Spectroscopic AGN Selection

Spectroscopic selection (e.g., Baldwin et al. 1981, Veilleux & Osterbrock 1987, Kauffmann et al. 2003, Kewley et al. 2006, Yuan et al. 2010)

Classes: SF, AGN dominated and compositesDAGN measure separates relative contribution

Kewley et al. 2006 Yuan et al. 2010

Spectroscopic AGN Selection

At low redshift, we can classify 283 galaxies from COSMOS 70 mm sample using standard line diagnostics 0 < z < 0.5 (where Ha still observable, 38% of sample at these

redshifts) log(LIR) = 8.8 – 12.1, <log(LIR)> = 10.7

AGN

HII

Comp.

Beyond z>0.5, lose Ha and [NII] lines, but have [OIII] and Hb out to z~1

Color as a proxy for [NII]/Ha? (e.g., Yan et al. 2010)

Appears to work well for normal galaxies locally…

At higher redshifts….

Yan et al. 2010

Applied to 70 mm Sample

500 galaxies with [OIII] and Hb

0 < z < 1 (~40% of sample in this range)

log(LIR) = 8.5 – 12.5, <log(LIR)> = 11.1

30 ULIRGs, 256 LIRGs70 mm selected

galaxies just aren’t red!

AGN

Mass-Excitation

Method of Juneau et al. 2011 uses mass instead

Applied to 70 mm Sample

Same 500 galaxies with O[III] and Hb

Appears to work better for infrared selected galaxies

X-ray mostly in AGN region

Trend with LIR

AGN

What about higher redshifts?

Current and future surveys with NIR spectrographs Until now, only small samples possible with longslit

spectographs Several multi-object spectrographs now online or will

be soon MOIRCS, FMOS, LUCIFER, FLAMINGOS2, MOSFIRE, etc.

Will become possible to use BPT type analysis at z = 1 - 3 for large samples

COSMOS FMOS survey will cover 1000’s of IR selected galaxies and AGN over the next few years

Sample Spectra

Preliminary FMOS Results

~60 (U)LIRGs at 1.0 < z < 1.6 with coverage of all four lines

Preliminary FMOS Results

Same sources on Mass-Excitation Diagram