galaxy formation from infrared to submm

Aug 8th, 2007 MAGPOP Summer School

Galaxy formation from infrared to submm

Michael Rowan-RobinsonImperial College London

1. Extragalactic infrared and submillimetre surveys2. Radiative transfer models for extragalactic infrared sources3. Models for source-counts and background radiation, from submm to ultraviolet


Extragalactic infrared and submillimetre surveys

Michael Rowan-RobinsonImperial College London

Dole et al 2006

most of the starlight ever generated in the universe is emitted at infrared wavelengths, ~ 50% is absorbed by dust and reemitted at far infrared and submillimetre wavelengths


1 micron - 1 mm

- a few terrestial windows

the infrared and submillimetre bands


Pre-IRAS, IRAS

Pre-IRAS: 1969: Caltech 2 Micron Survey (Neugebauer and Leighton) – circumstellar dust shells, BN object1976: The AFGL Survey at 4.2, 11, 19.8 and 27.4 m

(Price and Walker) – cds, HII regionsIRAS:1984: IRAS all sky survey at 12, 20, 60, 100 m

- 30,000 infrared galaxies (measured redshifts of 12000 with S(60)>0.6 Jy - PSCz)

- ir cirrus- ULIRGS, HLIRGS- AGN dust tori- ir dipole, large scale structure


IRAS

Horsehead Nebula


IRAS all-sky survey


Infrared galaxy populations

with IRAS we were able to identify the main infrared galaxy populations

- quiescent galaxies (ir cirrus)- starburst galaxies (prototype M82)- extreme starbursts (prototype A220)- AGN dust tori

but the IRAS survey was not deep enough (z ~ 0.3) to study the cosmological evolution of these populations, though 60 m source-counts showed that evolution is present, at a comparable rate to that seen in radio-galaxies and quasars

an important insight was that as infrared luminosity increased, the proportion of interactions and mergers increased


ISOISO, launched 1996, reached z~1, + spectroscopy


ISO surveysCAM Deep Surveys

Fadda et al, 2001, AA, astro-ph/011412 Franceschini et al, 2002, AA,

astro-ph/0108292Elbaz et al, 2002, AA 384, 848

ELAIS Survey at 6.7, 15, 90, 175 mOliver et al, 2000, MN 316, 749Serjeant et al, 2000, MN 316, 768Efstathiou et al, 2000, MN 319, 1169Serjeant et al, 2001, MN 322, 262Lari et al, 2001, MN 325, 1173Gruppioni et al, 2002, MN 341, L1Rowan-Robinson et al, 2004, MN 351,

1290ISO HDF-N and HDF-S surveys

Oliver et al, 2002, MN 332, 546Mann et al, 2002, MN 332, 549

FIRBACK 175 m surveyDole et al, 2001, AA 372, 264


ISO surveys * main result was very strong increase in star-formation rate in galaxies between z = 0 and 1 (factor ~10) (Rowan-Robinson et al 1997, Flores et al 1999), confirming the result from optical surveys (Lilley et al 1996, Madau et al 1996) and that the rate estimated from optical data without correction for extinction is severely underestimated.

• problems of screen model for extinction correction

• issue of consistency between estimates of star-formation rate from uv, H, radio, far infrared

sfr = 2.2 x 10-10 L60 = 2.5x10-8 LH = 4.5x10-10 L2800A


Dust extinction in local galaxies

<AV> ~ 0.3 in local galaxies(Rowan-Robinson, 2003, MN 344, 13)


comparison of star-formation rate estimates

Daddi et al 2007


HDFN at 850 mHughes et al 1998

impact of JCMT

SHADES

blank field surveys at 850 m showed that we were able to survey the whole universe to z = 5 with ultraluminous ir galaxies


Submillimetre surveysHubble Deep Field North

Hughes et al 1998

Hawaii surveysBarger et al 1998, 1999, Cowie et al 2002, Wang et al 2004

CUDSS surveyEales et al 1999, Webb et al 2003,Clements et al 2004, Ashby et al 2006

UK 8 mJy survey (200 sq arcmin)Scott et al 2001, Fox et al 2001,Ivison et al 2002, Almaini et al 2003

SHADES (0.5 sq deg)Mortier et al 2005, Coppin et al 2006, Ivison et al 2007, Aretxaga et al 2007


submillimetre associations and galaxy redshift distribution

Chapman et al, 2005, ApJ 622, 772 poor spatial resolution of JCMT means that reliable optical or infrared associations can only be made if have millimetreinterferometry or radio associations

why don’t we see submm galaxies at z > 4 ?

is this a selection effect ?


Near-ir Surveys2MASS all-sky survey at J, H, K (to 15.8, 15.1, 14.3 mag.)

– http://www.ipac.caltech.edu/2mass/UKIDDS survey of 7500 sq deg in JHK (K=18.3)

-http://www.ukidss.org/

FIR SurveysSPITZER surveys (GTO - various, FLS - 4 sq deg, SWIRE - 49 sq deg, GOODS - 0.1 sq deg, AEGIS - 1 sq deg, COSMOS - 1 sq deg) at 3.6, 4.5, 5.6, 8, 24, 70, 160 m

- http://ssc.spitzer.caltech.edu/

ASTRO-F all-sky survey in 6 bands at 9-180 m- http://www.akari.org.uk


2MASS• 2MASS provides a better picture of galaxy distribution at z<0.03


Layered SPITZER Surveys• Wide–shallow FLS GTO-shallow SWIRE

– greatest volume 4 8.5 49 sq deg

– rare luminous objects

– large-scale structure

• Confusion-limited GTO-deep GOODS-IRAC– maximum information 2.5 sq deg 300 sq arcmin

on faintest resolved sources

• Ultra-deep GTO-ultra GOODS-24 m– confusion distribution 150 300 sq arcmin


M51Sombrero

combined 3.6, 8 and 24 m images(SINGS consortium)


M82N2207/IC2163

Stefan’s Quintet


SPITZER SWIRE survey

49 sq deg in 6 areas, at 3.6, 4.5, 5.8, 8, 24, 70, 160 m


ELAIS N1: 9 sq. deg

SWIRE 3.6 m survey in ELAIS-N1


Photometric redshifts

method based on fixed galaxy and AGN templates, two passes through data to help identify QSOs and AGN dust tori, and selected priors Rowan-Robinson et al 2007


Photometric redshiftsSWIRE-VVDS sample (with VVDS team, PI LeFevre)

VIRMOS-VLT Deep Survey spectra>1000 sources~3% rms in (1+z)<1% outliers

~ IRAC 3.6 and 4.5 m big help in reducing outliers

z = 1 2 3 4

red: gals, blue QSOs


Photometric redshiftsAll SWIRE Catalogue

VVDS: 9 optical bandsN1,N2: 5 optical bandsLockman: 3-4 optical bndsCDFS: 3 optical bandsXMM-LSS: 5 optical bands

z = 1 2 3 4

SWIRE Photometric Redshift Catalogue contains over 1 million redshifts, 10% have z >2, 4% have z > 3, 20% detected at 24 m, 1% at 70 or 160 m

red: galaxies, blue QSOs


rms, % outliers, as function of number of bands


Redshift distributions

>=3 bands from U-8m

left: Suburu XDS, R<27.5

below: ELAIS-N1, r<23.5: for optically blank sources use 3.6-8 m for phot-z


Lir/Lopt• for galaxies with an ir excess, we fit ir templates (cirrus, M82, A220, AGN dust torus) and estimate Lir (reasonably accurate if have 70 m detection)• for cirrus galaxies, Lir/Lopt is a measure of optical depth of ism• for star-forming galaxies, Lir/Lopt is the specific star formation rate


Lir/Lopt for starbursts


star-forming ellipticals •* Lir/Lopt (specific star formation rate) versus Lopt (~ M•) for galaxies with elliptical galaxy template fits

* includes objects like Arp 220, whose star formation is heavily obscured

•Black: cirrus•Red: M82 starburst•Green: A220 starburst


ultraluminous cirrus gals(L), star-forming ellipticals(R)


AKARI• Japanese mission, 68 cm cooled telescope, first all-sky far infrared survey since IRAS, 90 and 140 m, sensitivity probably comparable to IRAS FSS, but much better spatial resolution, so in principle may be able to construct deeper all-sky sample than PSCz


HERSCHEL3.6 m passively cooled telecope operating at 50-500 m

layered survey will be carried out by SPIRE and PACS teams in guaranteed time, widest area 70 sq deg in 9 areas


PLANCKPLANCK will carry out a shallow all-sky extragalactic point-source survey, which will detect many high-z very luminous submm galaxies


Radiative transfer models for extragalactic infrared sources

• radiative transfer models for ir sources

• cirrus models for local quiescent galaxies

• models for starburst, UKIRGs, HLIRGs

• applications to Spitzer galaxies, submm galaxies

• IRS spectra and their interpretation


ingredients for models for seds of infrared sources

• model for interstellar grains [ Mathis et al 1977, Draine and Lee 1984, Rowan-Robinson 1992, Desert et al 1990, Siebenmorgen and Krugel 1992, Dwek 1998]

• assumed density distribution for dust [ ~r-, HII region physics (Yorke 1977, Efstathiou et al 2000)]

• dust geometry [ spherically symmetric, axisymmetric (Efstathiou and RR 1990, 1991, 1995, Pier and Krolik 1992, Granato et al 1994, 1997, Silva et al 1998), clumpy [Rowan-Robinson 1995, Hoenig et al 2006]

• radiative transfer code [Rowan-Robinson 1980, Efstathiou and RR 1990, Pier and Krolik 1992, Krugel and Siebenmorgen 1994, Granato et al 1997, Silva et al 1998, Popescu et al 2000, Hoenig et al 2006]


Radiative transfer models for infrared sources

• spherically symmetric dust clouds

- first accurate code 1980 (R-R, ApJS 234, 111)

- circumstellar dust shells 1981-3

- starbursts and ULIRGs (RRE, 1993, MN 263, 675; ERRS, 2000)

- cirrus galaxies (ERR, 2003)

• axially symmetric dust clouds

- first accurate code 1990 (Efstathiou and R-R, MN 245, 275)

- protostars 1991

- AGN dust tori 1995


the radiative transfer equation

The intensity of radiation I(r,) satisfies the equation

dI/ds = - n(r) C,ext I + n(r) Cabs B [T(r)] + n(r) C,sc (’) I (’) d

where Cabs = a2 Qabs ,

C,sc = a2 Q,sc (’)

C,ext = Cabs + C,sc (’) d


interstellar dust grains

size 50 A - 0.1 m (and larger ?)

composition: amorphous C graphite amorphous silicates crystalline silicates SiCPAHs

Brownlee particle


discovery of PAHs

Leger and Puget, 1984,AA 137, L5


IRAS - cirrus

south celestial pole


Cirrus models for local galaxies• assume optically thin ism, extinction AV (<1, 0.4-0.9)

• BC starburst models, age t*, exponential decay time

• characterise galaxies by single mean intensity, = bolometric intensity/solar neighbourhood intensity (~2-5)

• for local galaxies, t* = 0.25 Gyr, = 5-11 Gyr

(Efstathiou and Rowan-Robinson 2003, MN 343, 322)


IRAS - star forming regions

constellation OrionLMC


IRAS - ultraluminous infrared galaxies

Arp 220

Soifer et al, 1984, ApJ 283, L1: the remarkable infrared galaxy Arp 220


Eftstathiou, R-R, Seibenmorgen, 2000, MN 313, 734• embedded phase, t < 107 yrs• expanding neutral shell, t = 107-108 years•at 108 yrs, indistinguishable from cirrus

Models for starburstgalaxies


galaxy sed model fits from GRASIL (Silva et al 1998)


seds of ultraluminous infrared galaxies

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

L:ISO R:SPITZER


IRAS - AGN dust toriMiley et al, 1984, ApJ 278, L79: A 25 m component in 3C390.3


Infrared templates

(Rowan-Robinson 2001)


Hyperluminous infrared galaxies

Rowan-Robinson, 2000, MN 316, 885

starburst dominated

IRAS F10214, z=2.3 galaxy

Teplitz et al 2006


dust torus dominated


SPITZER-IRS: IRAS F00183-7111, hyperluminous infrared galaxy

•IRS spectrum of the hyperluminous ir galaxy F00183-7111= IRAS P00182-7112(Spoon et al 2004)

• z = 0.327 (narrow line object), lg Lsb = 13.25


Ltor v. Lsb


Lsb v. Mgas

broken lines show time-scale to convert gas mass into stars

ULIRGs and HLIRGs have bursts on shorter time-scale, or need truncated IMF


cirrus models for SCUBA galaxies• Efstathiou and R-R (2003) found that cirrus models, with slightly higher AV and ( ~ 2-3 times higher than local quiescent galaxies) can also fit high-z galaxies from SCUBA blank-field surveys

• restrict analysis to SCUBA sources:

(a) which have been confirmed by submm interferometry, or

(b) sources from 8 mJy survey which have radio associations

• 70% of sources (16/23) successfully modeled by cirrus model. Note: models fit radio data also.

• assume t* = 0.25 Gyr, = 6 Gyr


SCUBA galaxies


z<0.12 galaxies with cirrus sedsRowan-Robinson et al,2005, AJ 129, 1183 sources with good ISO-ELAIS and SPITZER-SWIRE data

templates


z=0.1-0.9 galaxies

fitted with cirrus or A220 template

A220 model: AV = 200, t* = 26 Myr (Efstathiou and RR 2001)


fitted with cirrus, A220 starburst and AGN dust torus templates

seds of z=0.1-2.2 galaxies/quasars


SPITZER-IRS spectra of ELAIS sources• IRS spectra for 70 ELAIS-

N1 and -N2 sources with S15> 1mJy validate the template fits

• most are ULIRGs, with z =1-3

• Filled circles: optical, ISO, SWIRE

( and MAMBO) data

• Solid curves: model seds

• Red curve: calibrated IRS data

(Hernan-Caballero et al 2006)


seds of submillimetre galaxiesSHADES SXDS Clements et al 2007


what powers ultraluminous infrared galaxies ?

Genzel et al, 1998, ApJ 498, 579


what powers ultraluminous infrared galaxies ?

Spoon et al, 2007, astro-ph/0611918


Models for source counts and background spectrum,

from submm to ultraviolet

• ingredients for counts model at submm to uv wavelengths

• star-formation history, luminosity functions

• assumed seds, parameter estimation

• predicted counts and background intensity


MODELS FOR COUNTS AND BACKGROUND FROM OPTICAL TO SUBMM

(Rowan-Robinson 2001, ApJ, 549, 745,2007, in prep)

* parameterized approach to star formation history

* fitted to infrared and submm counts and background

* 60 m luminosity function derived from PSCz data

* ir and submm seds based on mixture of four components (cirrus, M82-starbust, AGN dust torus, Arp220), proportions depending on luminosity


OTHER RECENT WORK

Franceschini et al, 2001, AA 378, 1 Gispert et al, 2000, astro-ph/005554Xu et al, 2001, ApJ 562, 179Takeuchi et al, 2001, PASJ, astro-ph/009460Pearson et al , 2001, MN 325, 1511Elbaz et al, 2002, AA 384, 848Lagache et al, 2004, ApJS 154, 112Gruppioni et al, 2005, ApJ 318, 9Lagache, Puget, Dole, 2005, ARAA 43, 727


PARAMETERIZED MODEL FOR STAR FORMATION HISTORY* assumed star formation history:

sfr = *(t)/ *(to) = exp Q1 - t/to . (t/to) P

meaning of parameters: t0/Q =sf (cf Bruzual,Charlot 1993) peak sfr when t/to = P/Q, or t = P sf

(essentially the Bruzual and Charlot models with an additional parameter to tune the epoch of peak star formation rate)

* assume 60 µm luminosity function of the form (L) = C* (L/ L* )1- exp-0.5[log10(1+L/L*)/]2

(Saunders et al 1990)* assume luminosity evolution with L*(t)/L*(to) = *(t)/ *(to)

* for each P,Q, parameters L*(to) and are solved for from PSCz data (15000 IRAS galaxies with known z, S(60)≥ 0.6 Jy)


parameterized star-formation history

sfr = *(t)/ *(to) =

exp Q1 - t/to . (t/to) P

meaning of parameters: t0/Q =sf

peak sfr when t/to = P/Q, or t = P sf


assumed seds, fit to counts, predictions•assumed spectral energy distribution, to convert luminosity function to other wavelengths, is mixture of cirrus, M82-starburst, AGN dust torus, Arp220-starburst components (radiative transfer models from Efstathiou et al 2000, Rowan-Robinson 1995), with proportion varying with 60 µm luminosity, to match 12-25-60-100-850 colour-colour and colour-luminosity diagrams

* fit to observed counts at 60, 850 µm, and background intensity at 140, 350, 750 µm to find best (least chi2) values of P,Q:

Ωo P Q

0.3 0.7 3.0 9.0

• best-fitting models then used to predict counts and background spectrum at 0.1 - 1250 m


Infrared templates


colour-colour, colour-luminosity

(Rowan-Robinson 2001)


Luminosity

function at 850 m


Luminosity

function at 60 m


Luminosity

function at 12 m


Luminosity

function at 0.44 m


Counts at 15 and 850 m


Counts at 90 and 175 m


Counts at 0.44 and 1.6 m

(King and RR 2002: improved fits using some density evolution)


Redshift distributions


Star-formation

history

(Rowan-Robinson 2003b)


differential counts at 24 m• 24 m differential counts

(Shupe et al, 2007, Papovich et al 2004)


new model for 24 m counts• 24 m differential counts (Shupe et

al, 2007, Papovich et al 2004)

• new model for ir counts (developed from RR 2001 models): independent evolution for each component, evolution has to flatten off at z < 0.5

M82cirrus

dust tori


SHADES counts at 850 m

Coppins et al, 2007, astro-ph/0609039


differential counts at 70-850 m

• new SWIRE 70, 160 and SHADES 850 m differential counts (Afonso-Luis et al, 2007, in prep,Coppins et al 2007)


contribution of different fir luminosity ranges to star-formation rate

• blue: lg Lfirr/Lo < 11

• orange: 11 < lg Lfirr/Lo < 12

• red: 12 < lg Lfirr/Lo

• green: total

LeFloch et al, 2005, ApJ

Reddy et al 2007


star-formation history to z = 6

Thompson et al 2006

Reddy et al 2007


Frank Low: 1968: the far infrared background

Contribution of Infrared Galaxies to the CosmicBackground

Frank J. LowDepartment of Space Science, Rice University, Houston, Texas, and Lunarand Planetary Laboratory, University of Arizona, Tucson, Arizona

Wallace H. TuckerDepartment of Space Science, Rice University, Houston, Texas

Received 12 July 1968

The far-infrared background due to a superposition of infrared galaxies ofthe type recently observed is computed. It is shown that these galaxiescontribute an amount of energy to the universe roughly equal to thatradiated by the other galaxies and produce a far-infrared backgroundpeaking beyond 50 .m It is likel y that theyacc ount for mos t of theobser vedextragalacti c radi o backgroun d but not 3the ° K microwavebackgroun .d

Phys. Rev. Lett. 21, 1538-41 (1968).


the far ir and submm background

Hauser and Dwek, 2001, ARAA 39, 249


the interpretation of the extragalactic background

the integrated extragalactic background spectrum is a weighted integral of the star-formation history

I = c |oto (t) LZ (t) dt

weighting is by K-correction LZ /L

so submm weighted to higher redshift than far and mid ir

submm background can give strong limit on high z sfr


Models for infrared

background

Rowan-Robinson 2001


QuickTime™ et undécompresseur codec YUV420

sont requis pour visionner cette image.

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

far ir and submm background from stacking analyses


Models for infrared

backgroundRowan-Robinson 2007

now have good bg data at 24, 70, 160 m

models, modified to fit 24 m counts, now also give better fit to background spectrum


History of the universeHistory of the universe


the futureDarwin

TPF

ALMA

Herschel

galaxy formation from infrared to submm

Documents

serjeant et

mmhughes et

surveydole et

madau et

mmoliver et

flores et

efstathiou et

lari et