Origin of the 12um Emission Across Galaxy Populations from WISE and SDSS Surveys

7.4 MB·English

by  E. Donoso

Draftversion January17,2012
PreprinttypesetusingLATEXstyleemulateapjv.5/2/11
ORIGIN OF 12µm EMISSION ACROSS GALAXY POPULATIONS FROM WISE AND SDSS SURVEYS
E. Donoso1, Lin Yan1, C. Tsai1, P. Eisenhardt2, D. Stern2, R. J. Assef2,6, D. Leisawitz3, T. H. Jarrett5, S. A.
Stanford4
1SpitzerScienceCenter,CaliforniaInstitute ofTechnology, 1200E.CaliforniaBlvd.,PasadenaCA91125,USA
2JetPropulsionLaboratory,CaliforniaInstituteofTechnology, Pasadena,CA91109,USA
3GoddardSpaceFlightCenter,Greenbelt,MD20771, USA
4DepartmentofPhysics,UniversityofCalifornia,Davis,CA95616, USA
5InfraredProcessingandAnalysisCenter,CaliforniaInstitute ofTechnology, Pasadena, CA91125, USAand
2 6NASAPostdoctoral ProgramFellow
1 Draft version January 17, 2012
0
2 ABSTRACT

n We cross-matched Wide-field Infrared Survey Explorer (WISE) sources brighter than 1 mJy at
a 12µmwith the SloanDigitalSky Survey (SDSS) galaxyspectroscopiccatalogto produce a sample of
J ∼105 galaxies at hzi=0.08, the largest of its kind. This sample is dominated (70%) by star-forming

3 (SF) galaxies from the blue sequence, with total IR luminosities in the range ∼ 108−1012L⊙. We
1 identify which stellar populations are responsible for most of the 12µm emission. We find that most
(∼80%)ofthe12µmemissioninSFgalaxiesisproducedbystellarpopulationsyoungerthan0.6Gyr.
]
O Incontrast,the 12µm emissionin weakAGN (L[OIII] <107L⊙)is producedby older stars,with ages
of∼1−3Gyr. WefindthatL linearlycorrelateswithstellarmassforSFgalaxies. Atfixed12µm
C 12µm
luminosity, weak AGN deviate toward higher masses since they tend to be hosted by massive, early-
. type galaxies with older stellar populations. Star-forming galaxies and weak AGN follow different
h
p L12µm–SFR (star formation rate) relations, with weak AGN showing excess 12µm emission at low
- SFR (0.02−1M⊙yr−1). This is likely due to dust grains heated by older stars. While the specific
o star formation rate (SSFR) of SF galaxies is nearly constant, the SSFR of weak AGN decreases by
r
∼3 orders of magnitude, reflecting the very different star formation efficiencies between SF galaxies
t
as and massive, early-type galaxies. Stronger type II AGN in our sample (L[OIII] > 107L⊙), act as
an extension of massive SF galaxies, connecting the SF and weak AGN sequences. This suggests a
[
picture where galaxies form stars normally until an AGN (possibly after a starburst episode) starts

1 to graduallyquenchthe SFactivity. We alsofind that4.6–12µm coloris auseful first-orderindicator
v of SF activity in a galaxy when no other data are available.
3 Subject headings: infrared: galaxies — galaxies: active — surveys
4
9
2 1. INTRODUCTION The rate at which galaxies transform gas into stars is
1. A detailed picture of the present day galaxy popula- one of the most fundamental diagnostics that describes
0 tions, their evolution and emission properties across dif- the evolutionofgalaxies. Ofmajorimportanceis tofind
2 ferentwavelengthsisstillfarfromcomplete. Surveyslike what physical parameter(s) drive changes in the SFR.
1 the Sloan Digital Sky Survey (SDSS; York et al. 2000) As dust-reprocessedlight from young stars is re-emitted
: mainly in the far-infrared (FIR) regime, the FIR lumi-
v have collected large amounts of information in the opti-
nosityisoneofthebesttracersofstarformationactivity
i cal regime,while 2MASS (Skrutskie et al. 2006)and the
X (Kennicutt 1998). It is well known that commonly used
IRAS mission (Neugebauer et al. 1984) have provided
SFR indicators, such as the UV continuum and nebular
r valuable,albeitrelativelyshallow,datasetsfromJ band
a emission line fluxes, require sometimes substantial cor-
upto100µm. Morerecently,theWide-fieldInfraredSur-
rections for dust extinction. Furthermore, these correc-
vey Explorer (WISE; Wright et al. 2010) has completed
tionsarehighlyuncertainanddifficulttoderive. Forthis
mapping the whole sky in the mid and far infrared, at
reason,integratedestimators basedonthe totalinfrared
sensitivities much deeper than any other large-scale in-
(IR) luminosity, either alone or in combination with the
frared survey. For example, WISE is about 100 times
ultraviolet luminosity (e.g. Heckman et al. 1998), and
more sensitive at 12µm than IRAS.
monochromaticestimatorsbasedmainlyon24µmfluxes,
While our understanding of the optical and far-IR
alone or in combination with Hα luminosity (e.g. Wu
properties of galaxies (long-ward of 24µm) has grown
et al. 2005; Alonso-Herrero et al. 2006; Calzetti et al.
steadily,thanksmainlytoSpitzerandHerschel,thespec-
2007;Zhu et al. 2008;Rieke et al. 2009;Kennicutt et al.
tralregionbetween10-15µmremainscomparativelyun-
2009), are increasingly being considered as reliable star
explored. Innormalgalaxies,thelightattheredderopti-
formation indicators for normal and dusty star-forming
calbandsandintheJ,H andK near-IRbandsisclosely
galaxies. The use of any IR flux as a SFR indicator re-
tiedtothetotalmassofthegalaxy,asitisdominatedby
liesontheassumptionthattheIRcontinuumemissionis
the red population of older stars. At wavelengths longer
duetowarmdustgrainsheatedbyyoungstars. However
than∼8µm,emissionfromdustheatedbyyoungerstars
there is also a contribution to dust reprocessed emission
becomesincreasinglyrelevantandbeginstotracethestar
by old stellar populations, associated more with the in-
formation rate (SFR).
2 Donoso et al.
terstellar medium around evolved stars than to recently verse galaxy populations observed by WISE and SDSS
bornstars. Inaddition,somefractionofthe IRluminos- (includingdeeperphotometricSDSSobjects),whilehere
ity may be attributed to active galactic nuclei (AGN), if we focus on 12µm-selected sources with available spec-
present(inSection3.8wefindAGNemissionisofminor tra.
importance in most of our sample). The exact contribu- This paper is organizedas follows. InSection2 we de-
tion of each component is difficult to estimate without scribethesurveysusedinthisworkaswellasexplainthe
detailed spectral analysis. constructionofourjointWISE-SDSSsample. InSection
Chary & Elbaz (2001) and Rieke et al. (2009) found 3wecharacterizethegalaxypopulationsandpresentthe
correlations between the 12µm luminosity and the total results on the analysis of the mid-IR emission and SFR.
IR luminosity for small samples of nearby galaxies. Duc Finally, Section 4 summarizes our results and discusses
et al. (2002) found that sources in Abell 1689 with high the implications of this work.
ISO mid-IR color index [15 µm]/[6.75 µm] are mostly Throughout the paper we assume a flat ΛCDM cos-
blue,activelystarforminggalaxies,whilelowmid-IRflux mology, with Ω = 0.3 and Ω = 0.7. We adopt
m Λ
ratios correspond to passive early-type systems. They H =70 km s−1 Mpc−1.
0
suggest that 15 µm emission is a reliable indicator of
2. DATA
obscured star formation. Similarly, shorter wavelength
mid-IR emission such as WISE 12µm is expected to be 2.1. The Wide-field Infrared Survey Explorer Catalog
a practical tracer of star formation activity. WISE has mapped the full sky in four bands centered
An important caveat is that far-infrared and/or ra- at 3.4, 4.6, 12 and 22 µm, achieving 5σ point source
dio measurements are only available for a small frac- sensitivities better than 0.08, 0.11, 1 and 6 mJy, respec-
tion of known galaxies. Early work by Spinoglio & tively. Everypartofthe skyhas been observedtypically
Malkan (1989) using IRAS all-sky data used 12µm to around10times,exceptneartheeclipticpoleswherethe
select unbiased samples of active galaxies with fluxes coverageis much higher. Astrometric precision is better
above 220 mJy and to study their luminosity func- than0.15′′ forhighS/Nsources(Jarrettetal.2011)and
tion. Deep pencil-beam surveys have complemented the angular resolution is 6.1, 6.4, 6.5 and 12′′ for bands
these samples with high redshift data. Seymour et al. ranging from 3 µm to 22 µm.
(2007) conducted a 12µm survey of the ESO-Sculptor This paper is based on data from the WISE Prelimi-
field (700 arcmin2) with the ISO satellite down to 0.24 nary Release 1 (April 2011), which comprises an image
mJy. Rocca-Volmerange et al. (2007) used it to model atlas and a catalog of over257million sources from 57%
mid-IR galaxy counts, revealing a population of dusty, of the sky. An object is included in this catalog if it:
massive ellipticals in ultra luminous infrared galaxies (1) is detected with SNR≥7 in at least one of the four
(ULIRGS).Teplitzetal.(2011)performedimagingofthe bands,(2)isdetectedonatleastfiveindependentsingle-
GOODS fields (150 arcmin2) at 16 µm with the Spitzer exposure images in at least one band, (3) has SNR≥3
spectrometer, finding that ∼15% of objects are poten- in at least 40% of its single-exposure images in one or
tially AGN at their depth of 40-65 µJy. These surveys more bands, (4) is not flagged as a spurious artifact in
illustrate the necessary tradeoff between depth and area at least one band. We refer the reader to the WISE
covered, potentially limiting the statistical significance PreliminaryReleaseExplanatorySupplementforfurther
of results due to cosmic variance. WISE provides the details1 (Cutri et al. 2011).
data to significantly improve the situation. Our sample
of ∼105 galaxies (see below) is over 200 times more sen- 2.2. The MPA-JHU Sloan Digital Sky Survey Catalog
sitive thanIRAS-basedsurveysandcoversanarea∼370 The Sloan Digital Sky Survey (York et al. 2000;
times larger than the GOODS samples. Stoughtonet al. 2002)is a five-bandphotometric (ugriz
In this work we explore the physical properties of bands)andspectroscopicsurveythathasmappedaquar-
12µm-selected galaxies in the local Universe, using a ter of the sky, providing photometry, spectra and red-
large sample of star forming galaxies and AGN with shifts for about a million galaxies and quasars. The
available redshifts and emission line measurements from MPA-JHU catalog (Brinchmann et al. 2004, hereafter
theSDSS.Thisisbyfarthelargest12µmsampletodate, B04) is a value-added catalog based on data from the
and we use it to trace the origin of IR emission across Seventh Data Release (DR7, Abazajian et al. 2009) of
different galaxy populations and to investigate how IR the SDSS2. It consists of almost 106 galaxies with spec-
emission relates to stellar mass. We also explore using tra reprocessed by the MPA-JHU team, for which phys-
12µm luminosity as a proxy for SFR to distinguish in- ical properties based on detailed emission line analysis
tense starburst activity from quiescent star formation. are readily available. Here we give a brief description of
Since weemploy the SDSS spectroscopiccatalog,ourre- the catalog and the methodology employed to estimate
sults apply to relatively bright galaxies at low redshift. SFRs. We refer the reader to the original papers for an
WISE certainly recovers other populations of galaxies, in-depth discussion.
rangingfromlowmetallicitybluecompactdwarfgalaxies The MPA-JHU catalog classifies galaxies according to
at very low redshift (Griffith et al. 2011) to highly ob- theiremissionlines,giventhepositiontheyoccupyinthe
scured sources at high redshift (Eisenhardt et al. 2011). BPT (Baldwin, Phillips & Terlevich 1981) diagram that
Lake et al. (2011) shows that WISE detects L∗ galax- plotsthe[OIII]λ5007˚A/Hβversus[NII]λ6584˚A/Hαflux
ies out to z ∼ 0.8 in the 3.4µm band, while Stern et al.
(2011) shows that WISE is a very capable AGN finder, 1 WISEdataproductsanddocumentation areavailableat
sensitive to both obscured and unobscured QSOs. A http://irsa.ipac.caltech.edu/Missions/wise.html
companion paper by Yan et al. (2011)analyzes more di- 2 TheMPA-JHUcatalogispubliclyavailableat
http://www.mpa-garching.mpg.de/SDSS/DR7/
Origin of 12µm Emission Across Galaxy Populations from WISE and SDSS Surveys 3
Fig.1.—Distributionsofredshift,stellarmass,restframecolor(u−r)0andD4000 forallWISE-SDSS12µm-selectedgalaxies(black),star
forminggalaxies(blue),AGN(red)andcompositegalaxies(green). WealsoshowthedistributionforallopticalMPA-JHUgalaxies(gray),
as well as for galaxies without BPT classification due to the lack of detected emission lines (dotted). The top row shows the complete
samplewhilethelowerthreerowssplitintolow,intermediateandhighIRluminosity. Trianglemarkersindicatethemeansoftherespective
distributions.
ratios. This separates galaxies with different ionizing oxygen abundances, 12+log(O/H),are available for star
sourcesasthey populateseparatesequencesonthe BPT forminggalaxiesas calculatedby Tremontiet al.(2004).
diagram. In most galaxies, normal star formation can Thoughout this paper we adopt the spectral line mea-
account for the flux ratios. However, in some cases an surements as well as estimates of SFR, metallicity and
extra source such as an AGN is required. In this paper dust extinction given by the MPA-JHU catalog.
we follow this BPT classificationto distinguish between: The SDSS pipeline calculates several kinds of mag-
(i) star-forming galaxies (classSF andLOWS/N SF nitudes. In this work we have adopted modified Pet-
fromB04), (ii) active galactic nuclei (class AGN and rosianmagnitudes for flux measurements,whichcapture
Low S/N LINER from B04), and (iii) composite sys- a constant fraction of the total light independent of po-
tems that present signatures of the two previous types sition and distance. When appropriate, we have also
(class C from B04). Note that broad-lined AGN like used model magnitudes (modelMag) as they provide the
quasars and Seyfert 1 galaxies are are not included in mostreliable estimates of galaxy colors. Magnitudes are
thesample,astheyweretargetedbydifferentcriteriaby corrected for galactic reddening using the dust maps of
the SDSS. Schlegel, Finkbeiner & Davis (1998).
Star formation rates are derived by different prescrip-
tions depending on the galaxy type. The methodology 2.3. The Joint WISE-SDSS Galaxy Sample
adopted by B04 is based on modeling emission lines us-
ing Bruzual & Charlot (1993) models along with the WehavecrossmatcheddatafromWISEandtheMPA-
CLOUDY photoionization model (Ferland 1996) and JHU catalog to construct a galaxy sample covering an
spectralevolutionmodelsfromCharlot&Bruzual(2008) effective area of 2344 deg2, or 29% of the DR7 (legacy)
to subtract the stellar continuum. To correct lines for spectroscopic footprint. WISE sources were selected to
dustattenuation,MPA-JHUadoptsthemulticomponent have 12µm fluxes above 1 mJy, also requiring objects
dustmodelofCharlot&Fall(2000),wherediscretedust to have clean photometry at 3.4, 4.6 and 12µm. For
clouds are assumed to have finite lifetime and a realistic MPA-JHUsources,weselectedobjectswithde-reddened
spatial distribution. This approach produces SFRs that Petrosian magnitude rpetro < 17.7 and r-band surface
take full advantage of all modeled emission lines. For brightness µr < 23 mag arcsec−2. This selects a con-
AGN and composite galaxies in the sample, SFRs were servative version of the SDSS main galaxy sample (see
estimatedbytherelationshipbetweentheD spectral Strauss et al. 2002). Using a matching radius of 3′′ we
4000
index and the specific SFR (SFR/M⊙ or SSFR), as cal- find 96,217 WISE objects with single optical matches
ibrated for star forming galaxies (see Fig. 11 in B04). (40% of the SDSS sample in the intersection area), and
These estimates have been corrected in the latest MPA- 73sourceswithtwoormorecounterparts. Thelatterare
JHU DR7 release by using improved dust attenuations mostly large extended sources or close interacting sys-
and improved aperture corrections for non-SF galaxies tems of two members. For the rest of this paper we will
following the work by Salim et al. (2007). Gas-phase focusonthesingleIR-opticalmatchesthatconstitutethe
vastmajority(>99.9%)ofthegalaxypopulation. Byus-
4 Donoso et al.
Fig.2.— Distribution of 12µm luminosity for galaxies separated according to spectral type as: SF galaxies (left), composite systems
(middle),andAGN(right).
ingrandomcatalogsgeneratedovertheeffectivearea,the
expected false detection fraction at 3′′ is 0.05%.
Each region of the sky has been observed at least 10
timesbyWISE,withthenumberofobservationsincreas-
ing substantially toward the ecliptic poles. Within our
effective area, the median coverage depth at 12µm is
about 13, varying tipically between 10 and 20. At these
levels,theaveragecompletenessofthesampleisexpected
to be over 90%, as shown in the WISE Preliminary Re-
lease Explanatory Supplement (Sec. 6.6).
3. ANALYSISANDRESULTS
3.1. Derived Properties
We derived stellar masses for all galaxies using the
kcorrect algorithm(Blanton&Roweis2007),whichfitsa
linearcombinationofspectraltemplatestothefluxmea-
surementsforeachgalaxy. Thesetemplatesarebasedon
a set of Bruzual & Charlot (2003) models that span a
wide range of star formation histories, metallicities and
dustextinction. Thisalgorithmyieldsstellarmassesthat
differ by less than 0.1 dex on averagefrom estimates us-
ing other methods (for example, the method based on
fitting the 4000˚A break strength and Hδ absorption in-
dex proposed by Kauffmann et al. 2003a).
To derive restframe colors and monochromatic lumi- Fig. 3.—Distributionofredshift,stellarmassandrestframecolor
nositiesinthe infraredweusedthefitting codeandtem- (u−r)0 for WISE-SDSS 12µm-selected galaxies classified as star
plates3 of Assef et al. (2010), applied to our combined forming(blue). Wealsoshowthedistributionsforallopticalstar-
forminggalaxies(gray)andforstar-forminggalaxieswithout12µm
ugriz photometry plus WISE 3.4 µm, 4.6 µm and 12µm
flux densities above 1 mJy (black). Trianglemarkers indicate the
fluxes. Assef et al. (2010) present a set of low-resolution meansofthedistributions.
empirical spectral energy distribution (SED) templates
SFRcalibration. NoteallopticalSFRsusedinthispaper
for galaxies and AGN spanning the wavelength range
have been corrected for dust extinction.
from 0.03 µm to 30 µm, constructed with data from the
NOAO Deep Wide-Field Survey Boo¨tes field (NDFWS,
3.2. General Properties of 12µm Galaxies
Jannuzietal.2010)andthe AGNandGalaxyEvolution
Survey (AGES, Kochanek et al. 2011). The code fits We begin our analysis by exploring the general prop-
threegalaxySEDtemplates thatrepresentanoldstellar erties of the WISE 12µm-selected galaxy sample. The
population (elliptical), a continuously star-forming pop- sampleiscomposedofamixtureof70%SFgalaxies,15%
ulation (spiral) and a starburst component (irregular), AGN, 12% composite galaxies and 3% galaxies lacking
plusanAGNSEDtemplate withvariablereddeningand BPT classificationdue to the absence of detectable lines
IGMabsorption. Thesetemplateshavebeensuccessfully in the spectra (most lack Hα emission). The composi-
usedtotestthereliabilityofIRACAGNselection,andto tion of the MPA-JHU optical sample is 44% SF, 12%
predictthecolor-colordistributionofWISEsources(As- AGN, 6% composite, 37% unclassifiable, which means
sef et al. (2010)). We also use these templates to assess that the 12µm selection is highly efficient in recovering
the relative contribution of AGN to the energy budget. SF systems and avoidsobjects with weak or no emission
InsteadoftryingtoderiveanewcalibrationoftheSFR lines. In terms of the total optical galaxy populations,
in the IR, we take the approach of comparing IR lumi- 61%(SF),53%(AGN)and76%(composite)oftheSDSS
nosities directly to optical dust-corrected SFRs. This galaxies have 12µm flux densities above 1 mJy. In Fig-
makes our results largely independent of any particular ure 1 (top row) we plot the distribution of redshift, stel-
lar mass, D index and restframe u−r color for SF
4000
3 Templatesandcodeareavailableat galaxies,AGN and composite systems, as well as for the
http://www.astronomy.ohio-state.edu/∼rjassef/lrt/ three classes all together. The majority of WISE-SDSS
Origin of 12µm Emission Across Galaxy Populations from WISE and SDSS Surveys 5
Fig.4.—12µminfraredluminosityasafunctionofstellarmassforSFgalaxies(gray),strongAGNwithL[OIII]>107L⊙ (colorpoints,
topleft),andweakAGNwithL[OIII]<107L⊙(colorpoints,topright). ThesolidlineshowsthemedianL12µmofSFgalaxiesasafunction
ofmass. Bottom panelsshowthedistributionofL12µm andstellarageforthevariouspopulationsasindicated.
12µm sources are SF galaxies at hzi = 0.08 with stel- bluer and dominated by younger stars as IR luminosity
lar masses of ∼ 1010.2M⊙; these are typical values for increases. Figure2showsthe redshiftdistributionofthe
the SF class. They clearly populate the blue peak of the restframe 12µm luminosity for our sample. It can be
galaxy bimodal distribution around (u−r) = 1.6 and seen that while high luminosity sources naturally lie at
0
haveinherentlyyoungstellarpopulations(D ∼1.3,or higher redshifts, the redshift distribution of the differ-
4000
agesof∼0.5Gyr). AGNare,asexpected,comparatively ent classes is very similar, except at the lowest redshifts
more massive (M∗ ∼1010.7M⊙), redder ((u−r)0 ∼ 2.1) (z < 0.02) where very few AGN/composite galaxies are
and older (∼1-6 Gyr), dominating the massive end of observed.
the 12µm galaxy distribution. As a population, AGN The absence of red sequence galaxies in our sample is
do not differ significantly (in terms of these properties) not surprising, but there is also a number of SF galax-
in comparison to the corresponding purely optical sam- ies without IR emission due to our flux limit. Figure 3
ple. Composite galaxies present intermediate properties shows the mass, redshift and colors of SF galaxies with
betweentheSFandAGNsamples. Notethatthebulkof and without 12µm emission, as well as for the entire SF
galaxies lacking BPT classification (due to weak or ab- optical sample. On average, SF galaxies not present in
sent emission lines) is missed in our IR-optical sample. our sample have bluer colors, lie at lower redshifts, and
Theseobjectsprimarilypopulatetheredsequenceofthe havestellarmassesaround109.3M⊙, roughlyanorderof
galaxy distribution (e.g. Baldry et al. 2004) and hence magnitude below the 12µm SF galaxy sample. At this
are not expected to be prominent at 12µm. level, galaxies have very little dust mass, and hence can
We then divide the sample into three subsamples of not re-radiate much in the IR.
monochromatic infrared luminosity: faint (L < The main result here is that WISE 12µm-selected
12µm
109.2L⊙), intermediate (109.2L⊙ < L12µm < 1010L⊙), galaxies are primarly typical blue sequence (SF) galax-
ies. It is safe to assume that the majority of blue se-
andbright(L12µm >1010L⊙)sources. Therearenolarge quence galaxies correspond to late morphological types
biaseswithredshift,i.e. thedifferentclassesaresampled
(e.g. Strateva et al. 2001;Shimasakuet al. 2001;Baldry
roughly without preference at all luminosities. Both SF
et al. 2004). AGN and composite objects are also rep-
galaxies and AGN become more massive for higher IR
resented, belonging either to the red sequence or to a
luminosities and we have checked that this also holds in
transitional regime among the two former classes. It
narrowredshiftslices. As wewillshowbelow,thisisdue
is interesting that the detection efficiency is largest for
tothe couplingbetweenthe IRandthe opticalemission.
composite systems, which was also found by Salim et al.
Interestingly, the 12µm SF galaxies change by a factor
(2009) for 24 µm sources that lie in the so called green
of 0.8 dex in mass toward high 12µm luminosities while
valley (e.g. Martin et al. 2007). This is a region located
keeping the same color and stellar content. The AGN
betweentheredandbluecloudsequences,bestidentified
population,whilegettingslightlymoremassive,becomes
6 Donoso et al.
Heckmanetal.(2004)haveshownthat[OIII]fluxisare-
liable estimator of AGN activity. Following these works,
we split the AGN sample by [OIII] luminosity. We see
thatstrongAGN(L[OIII] >107L⊙)haveIRluminosities
considerably larger than weak AGN (L[OIII] < 107L⊙),
followingapproximatelythesamerelationshipwithmass
as SF systems. Weak AGN, instead, lie well below that
correlation and show a larger scatter. We note that the
contribution by star forming regions to the [OIII] flux is
<7% (Kauffmann et al. 2003b).
In the right bottom panel of Figure 4 we compare the
agesof stars in all subsamples as derivedfrom the D
4000
spectral index. SF galaxies have the youngest stellar
populations, peaking at ∼0.5 Gyr, followed by consid-
Fig. 5.—Cumulativefractionofintegrated12µmluminosityfor erably older composite galaxies (∼1.5-2 Gyr). Strong
galaxiesdominatedbystellarpopulationsofagivenage,separated AGN, which dominate the massive end, have interme-
according to spectral type as: SF galaxies (blue), composite sys- diate stellar populations (∼1.5 Gyr) that are closer to
tems(green), strongAGN(black)weakAGN(red).
SF/Composite systems than to weak AGN (see Fig. 1).
in the NUV-r color-magnitudediagram,where SF activ-
In contrast, weak AGN tend to be hosted by early-type
ity is being actively quenched and galaxies are thought
galaxieswith significantly older stars (∼3 Gyr) as found
to be in transitional stage in their migration from the
also by Kauffmann et al. (2003b) after comparing AGN
blue to the red sequence.
hostsizes,surfacedensitiesandconcentrationratioswith
Most galaxies in our sample are either normal lumi-
those of normal early-type galaxies. This highlights the
nosity IR galaxies (L12µm ∼ 109.2−10L⊙; 60%), low lu- importance that young/old stars have in powering the
minosity IR galaxies (L12µm < 109.2L⊙; 31%) or lumi- 12µmemission. ForAGNofroughlysimilarstellarmass,
nous infrared galaxies (LIRGs; L12µm ∼ 1010−10.8L⊙; only when younger stars begin to dominate (and the ac-
9%). However, ULIRGs are also present. There are 114 tive nuclei becomes more powerful) is the IR emission
objects with L12µm > 1010.8L⊙ (roughly equivalent to comparable to actively star-forming systems. Qualita-
LTIR > 1012L⊙ using the conversion of Chary & Elbaz tively, the same result holds if we use the dust-corrected
2001),correspondingtoasurfacedensityof0.049deg−2. r-band absolute magnitude instead of stellar mass.
This is comparable to the 0.041 deg−2 surface density Since galaxies of different ages have very different
IR output, an interesting question to address is how
found by Hou, Wu & Han (2009). These ULIRGs natu-
the 12µm luminosity budget depends upon the age of
rally lie at higher redshift (hzi = 0.2) and populate the
the stellar populations. Figure 5 shows the cumulative
massive end of the SF sequence above ∼ 1011M⊙. We fraction of the integrated 12µm luminosity produced in
reiteratethattheseresultscomefrommatchingWISEto
galaxies of a given age. In SF galaxies, ∼80% of the to-
the relatively bright SDSS spectroscopic sample. WISE
tal IR luminosity is produced by galaxies younger than
galaxypopulationsdowntor ∼22.6areanalyzedinYan
0.6 Gyr. Composite galaxies and strong AGN reach the
et al. (2011).
samefractionatagesof 1.5Gyr and2 Gyr, respectively.
3.3. 12µm Luminosity and Stellar Mass In weak AGN, instead, most of the mid-IR emission is
produced within a range of ∼1-3 Gyr. This inventory
We now have a large sample of 12µm-selected galax-
of 12µm luminosity in the local universe shows where
ies that range from low-IR to ULIRG luminosities, for
the bulk of the IR emission resides, shifting from stellar
whichhighqualityopticalspectraanddust-correctedop-
populations of a few hundred Myr in actively SF galax-
tical SFRs are available. First we examine the relation
ies to a few Gyr in galaxies hosting weak AGN. Thus, it
between L and stellar mass. B04 have shown that,
12µm underlinesagaintheimportantrolethatyoung/oldstars
at least for star forming systems, SFR and stellar mass
have in powering 12µm emission. As we will see later
are strongly correlated in the local universe. There is
inSection3.6,this alsosupportsthe idea thattransition
also evidence that this relationship evolves with redshift
galaxies(i.e. composite/strongAGN) forma smooth se-
(Noeske et al. 2007, Daddi et al. 2007). Although it is
quence that joins highly active galaxies with quiescent
expected that more massive galaxies are naturally more
galaxies.
luminous, it is unclear whether more massive systems
would have more dust emission in the mid-IR. Figure
3.4. 12µm Luminosity and Star Formation Rate
4 shows the correlation between monochromatic 12µm
luminosity and stellar mass for our sample. The corre- We now explore the relationship between infrared lu-
lation is tight for SF systems over nearly three orders of minosityandoptical,dust-correctedSFR.Figure6shows
magnitudeinstellarmass(graypoints,toppanels). Sev- L versus SFR , color-codedby the D spectral
12µm opt 4000
eral studies have found that the distributions of [OIII] index. As discussed by Kauffmann et al. (2003a), the
emission line flux to the AGN continuum flux at X-ray, D index is a good indicator of the mean age of the
4000
mid/far-IR and radio wavelengths (i.e. where stellar stellar population in a galaxy. The dashed line indicates
emission and absorption by the torus are least signifi- the reference relation of Kennicutt (1998), as given by
cant) are very similar for both type I and type II AGN Chary & Elbaz (2001) in terms of 12µm luminosity, to
(Mulchaey et al. 1994; Keel et al. 1994; Alonso-Herrero convert IR luminosity into “instantaneous” SFR. It was
etal.1997). Basedonthis,Kauffmannetal.(2003b)and derivedfromsimple theoreticalmodelsofstellarpopula-
Origin of 12µm Emission Across Galaxy Populations from WISE and SDSS Surveys 7
Fig.6.—Infraredluminosityat12µmasafunctionofdust-correctedstarformationratederivedfromopticalemissionlines,forgalaxies
classified as SF (left), Composite (middle) and strong/weak AGN (right). Linear fits are shown for the cases of: SF galaxies (solid line,
all panels), composite systems (squares, middle panel) and AGN (squares, right panel). The dashed line is the Chary & Elbaz (2001)
conversion between SFR and 12µm luminosity. For sources brighter than 5 mJy at 22µm, we also plot the corresponding L22µm–SFR
relationships(triangles). Inallpanels,themeanstellarageofthedominantstellarpopulationisindicatedbytheD4000 colorscaleonthe
right. Contoursenclose95%,68%and33%ofthedensitydistribution(68% forthecaseofstrongAGN).
tions with ages 10-100 Myr without considering factors driven by the current SF for most AGN. Weak AGN
like metallicity or more complex star formation histo- are predominantely associated with massive, early-type
ries. While this makes it strictly valid only for young galaxiesincreasinglydominatedbyoldstarsatlowSFRs
starbursts, the Kennicutt relation is quite often applied (∼0.1M⊙yr−1). Given their red optical colors, it is un-
to the more general population of star forming galaxies. likely that recent SF could be responsible for their IR
Figure 6 shows that the IR emission from SF galaxies emission. More likely, dust grains heated due to older
(left panel) correlates fairly well with SFR . The cor- stars or an AGN are driving this emission. Note, how-
opt
relation is tighter for high SFRs becoming broader and ever, that in Section 3.8 we will show that the contribu-
slightly asymmetric for low SFRs. A least-squares fit to tion of AGN at 12µm is of minor importance for most
the SF sample is given by AGN, except perhaps for most powerful ones. Only
strong AGN, which are dominated by intermediate-to-
logLSF =(0.987±0.002)logSFR +(8.962±0.003).
12µm opt youngstellarpopulations,tendtooccupyaregionsimilar
(1) toSFgalaxiesinFigure6. Theyshowaclear“excess”in
This expressionis close to the Chary & Elbaz (2001)re- L12µm atSFRopt ∼0.5M⊙yr−1 thatgraduallydecreases
lation at high SFRs, which is not surprising given that whenstarsgetyoungertowardhigher SFRs. This shows
relation was calibrated using the IRAS Bright Galaxy that the age of stars in an AGN is an important fac-
Sample (Soifer et al. 1987), i.e. luminous galaxies with tor in determining the origin of the 12µm emission. An
L12µm > 109L⊙. The small differences are likely at- expression fitting AGN (weak and strong) is given by
tributabletoluminosity/redshiftselectioneffectsandthe
slight differences between ISO and WISE 12µm filters. logLAGN =(0.582±0.004)logSFR +(9.477±0.002).
12µm opt
More importantly, the agreement is quite good consid- (2)
ering the different origin (FIR vs optical emission lines) Recent work by Salim et al. (2009) compared
of the SFRs. Relative to the Chary & Elbaz (2001)con- NUV/optical SFRs with L calibrated from 24 µm
TIR
version, L12µm is comparatively suppressed by a factor fluxesforredandgreensequenceobjectsatz ∼0.7(cor-
> 1.6 for SFRopt below ∼0.1M⊙yr−1. This is likely respondingtorestframe14µm,closetothe WISE12µm
because low SFR systems have very low stellar masses band). They find large excess IR emission for a given
(<109M⊙),andthereforebecomemoretransparentdue SFR, attributed mainly to older stellarpopulations, and
to the increasing fraction of stellar light that escapes toalesserextenttoanAGN.Wefindbroadlyconsistent
unabsorbed by dust. We note, however, that the spa- results, but for 12µm-selected AGN sources with opti-
tial distrbution of dust in HII regions and/or molecular cal SFRs. Kelson & Holden (2010) have suggested that
cloudscouldalsohaveinfluence(e.g. Leisawitz&Hauser thermally pulsating AGB carbon stars (TP-AGB) with
1988). In addition, we have repeated the test for the agesof0.2-1.5Gyr(correspondingtoD ∼1.2-1.5)can
4000
bluest galaxies in the SF galaxy class, obtaining no sig- also contribute significantly to the mid-IR flux. As seen
nificant differences. This suggests that other effects like in Section 3.3, this does not seem to be important for
metallicity could also be relevant. ourmucholder,weakAGN,andperhapsmarginallyrele-
For AGN, the coupling between optical SFR and IR vantinstrongAGNthathavetypicalstellaragesslightly
luminosity follows a different relationship (right panel abovetheupper1.5Gyrlimit. ThecaseofSFgalaxiesis
of Figure 6). Most AGN lie in a broader distribution interesting because most of the 12 µm luminosity seems
above the instantaneous conversion of Kennicutt, par- to originate from galaxies in the ∼0.3-0.6Gyr age range
ticularly those with SFRopt below ∼1M⊙yr−1. For a andthisluminositycorrelatesrelativelywellwiththeop-
fixed SFR, the IR emission is higher by a factor of sev- tical SFR. While the age ranges seems to overlap, it is
eralrelativeto SFgalaxies,suggestingthat L is not difficult to prove whether TP-AGB dominate the emis-
12µm
8 Donoso et al.
sionornot. FurtherSEDdecompositionandmodelingof
stellarpopulationsisrequiredtofindthefractionofmid-
IR luminosity powered by TP-AGB stars, a task that is
potentially complicated by metallicity effects and uncer-
taintiesintheensemblecolorsofsuchstars. However,the
generalpicture is consistentwithpreviousresults(Salim
et al. 2009; Kelson & Holden 2010) that find evidence
forthe mid-IRbeing sensitiveto starformationoverrel-
atively long (>1.5 Gyr) timescales.
Finally, we consider composite systems, which present
considerableSFactivityalongwithspectralsignaturesof
anAGN (middle panelof Figure 6). By definition, these
objects have up to 40% (see B04) of their Hα emission
coming from a non-stellar origin, though the fraction is Fig.7.— Specific star formation rate (SSFR) as a function of
<15% for most galaxies. With masses, ages and optical 12µmluminosityforSFgalaxies (blue), AGN(red), andcompos-
ite galaxies (grayscale). The metallicity-weighted SSFR for star
colorsintermediatebetweenthe SFandAGNsequences,
forminggalaxiesisindicatedbythedashedgreencontoursandits
composite galaxies closely follow the Kennicutt relation, correspondingy-axisisshownontheright,whereZ=12+log(OH).
except at the low SFR end where older stars once again Toguidethecomparisonofdifferentdistributionsweshowsimple
begin to dominate. A least-squares fit for composite linear fits of the form log(SFR)=a×log(L12µm)+b (dotted lines).
Contoursenclose98%,95%,68%and33%ofthedensitydistribu-
galaxies has a slope intermediate between AGN and SF
tion(68%forthecaseofstrongAGN).
galaxies, and is given by
logLcomp =(0.671±0.003)logSFR +(9.249±0.002). SFR/M⊙,thatmeasuresthecurrentrelativetopastSFR
12µm opt needed to build up the stellar mass of the galaxy. The
(3) SSFR traces the star formationefficiency and its inverse
We note that the optical SFRs utilized here, while not defines the timescale for galaxy formation or the time
ideal, are probably the best estimates publicly available the galaxy requiredto build up its current mass. Higher
for such a large and diverse population of galaxies in valuesofSSFRareindicativeofalargerfractionofstars
the local universe. Other methodologies that use more being formed recently. While ideally we would use gas
sophisticated dust corrections and employ H2, FIR or mass or total mass instead of stellar mass for the nor-
radio data could provide more accurrate values, but are malization, such masses are not easily measured.
difficult to apply across the entire sample and data are Figure 7 shows the SSFR as a function of 12µm lu-
notalwaysavailable (see Saintonge etal. 2011for a cali- minosity for the different galaxy classes. The nearly
brationofSFRbasedonH2 masses). Thisisparticularly flat correlation for SF galaxies means that no matter
relevantfor SFRs in AGN, which canpresent large>0.5 the IR output, the amount of star formation per unit
dex formal errors (see Figure 14 of Brinchmann et al. mass remains relatively constant. SF galaxies display
2004). WehaveverifiedthattheSDSSSFRsinoursam- a weak dependence with L that gets narrower to-
12µm
pleareinbroadagreementwithUV-basedSFRs derived wards higher luminosities. As noted before, a possible
bySalimetal.(2007)(S.Salim,privatecommunication), origin for such residual SSFR could be due to a metal-
with an average offset/scatter of 0.055/0.387 for the to- licity gradient. Calzetti et al. (2007) studied individual
tal sample (0.013/0.334for strongAGN, 0.242/0.567for star forming regions of fixed aperture in nearby galax-
weak AGN). ies with known Paα surface density, and found that low
metallicity galaxies have a small deficit in 24 µm emis-
3.5. 22 µm Luminosity and Star Formation Rate
sioncomparedto highmetallicity galaxies. Relan˜o etal.
WISE is less sensitive at 22 µm than at 12µm, but a (2007) confirmed that while 24 µm luminosity is a good
significant fraction of our sample (∼30%) has measured metallicity-independent tracer for the SFR of individ-
22µmfluxesabove5mJy(notewefindno22µmgalaxies ual HII regions, the metallicity effect should be taken
without 12µm detection). Similar to the 12µm sample, into account when analyzing SFRs integrated over the
this 22 µm subsample is a mixture comprisedof 65%SF whole galaxy. We test qualitatively for a metallicity ef-
galaxies, 14% AGN and 19% composite systems. How- fectbycalculatingtheSFRperunitmassperunitmetal-
ever, the fraction of strong AGN is 38%, compared to licity,wherethe metallicityisgivenbythe12+log(O/H)
16% for 12µm sources. As 22 µm is closer to the dust gas-phase oxygen abundance derived from optical neb-
emission peak and is not affected by PAH emission fea- ular emission lines. The SF population displays an al-
tures,itisinterestingtocomparewiththe12µmgalaxies most perfectly flat relationship over almost 4 orders of
of our previous analysis. Figure 6 shows the linear fits magnitude in L such that independent of IR out-
12µm
for the 22 µm subsample (triangle markers). These rela- put, a galaxy of given mass and metal content converts
tions are very similar to the 12µm fits (solid line for SF gas into stars at a nearly constant rate. We note that
galaxies, square markers for AGN and composite galax- thesemetallicitiesrepresentthecurrentmetalabundance
ies), supporting the independence of our results on the ratherthantheluminosity-weightedaverageofpaststel-
particularities of a single mid-IR band. lar populations. They also do not suffer from complica-
tionsduetoα-elementenhancementorageuncertainties,
3.6. Specific Star Formation Rate
characteristic of methods relying on absorption-line in-
Given the strong correlations between optical or IR dices. Bond et al. (2011) arrive at a similar conclusion
light and stellar mass, a more interesting metric for regardingaconstantSSFRinnearbygalaxiesusingHer-
comparison is the specific star formation rate, SSFR or schel 250µm and WISE 3.4µm data.
Origin of 12µm Emission Across Galaxy Populations from WISE and SDSS Surveys 9
Compared to SF galaxies, AGN have SSFRs lower by
a factor of ∼10, mainly because of their higher stellar
masses. However, strong AGN lie much closer to the
SF sequence than weak AGN. The former are hosted by
high stellar mass galaxies, but also have young stellar
populationsthat driveupthe SFR athighL . Weak
12µm
AGN do not have this boost in SFR and hence have
lower SSFRs. Once again, composite systems populate
a region intermediate between SF galaxies and strong
AGN. Previous studies (e.g. Brinchmann et al. 2004;
Salim et al. 2007) have shown the relationship between
the SFR and M∗, identifying two different sequences:
galaxies on a star-forming sequence and galaxies with
little or no detectable SF. While the general result is
that the SSFR of massive, red galaxies is lower at 0 < Fig.8.—Specific star formationrate as afunction of restframe
z < 3, the exact dependence of SSFR on mass is still a 4.6-12µm galaxy color (W2-W3)0 for SF galaxies (blue), strong
matter of debate, particularly in view of recent results AGN (dash-dot black) and weak AGN (red). The gray line in-
dicates the median relation of the complete sample. The dashed
that trace the evolution at higher redshift (e.g. Noeske
and dotted lines correspond to linear fits to all sources and to all
et al. 2007; Dunne et al. 2009; but see the discussion by AGN, respectively. Contours enclose 95%, 85%, 68% and 33% of
Rodighiero et al. 2010). In our sample of SF galaxies thedensitydistribution(85%,68%and33%forthecaseofstrong
we find that the dependence of L on M∗ is such AGN).
12µm
thattheefficiency bywhichgasis transformedintostars
larpopulations thatregulate the IR emissionbudget. In
is nearly independent of the IR emission reprocessed by
any case, this suggests that the higher the SSFR, the
the dust. In strong AGN the star formation activity is
more prominent the 12µm IR emission becomes relative
“suppressed” moderately, but considerably more in the
to 4.6 µm, where the latter is expected to strongly cor-
case of weak AGN. This suggests a sequence where a
relate with stellar mass. This shows that the 4.6–12µm
strong-AGN phase is a continuation of the SF sequence
color serves well as a rough first-order indicator of star
at high stellar mass, that gradually turns AGN into a
formationactivityoverthreeordersofmagnitudeinSFR.
population with weakened SF activity and lower L ,
12µm A simple expression fitting all galaxies is given by
dominated by older and redder stars.
Basedonopticaldata,Kauffmannetal.(2003b)found logSSFR=(0.775±0.002)(W2−W3) −(12.561±0.006).
0
thatstrongAGNhostsareindeedpopulatedbyrelatively (4)
young stars, suggesting many of them could be post- IfthegalaxyisknowntohostanAGN,themoreaccurate
starburst systems with extended star formation. With expression is
UVdata,Salimetal.(2007)showedthereis aclosecon-
logSSFR=(0.840±0.008)(W2−W3) −(12.991±0.020).
nection between massive SF galaxies and strong AGN. 0
(5)
Using IR data, we find that the smooth sequence of
These results show that there is a tight link between
galaxies from Figures 6 and 7 supports the hypothesis
stellar mass, current star formation rate and IR color in
of strong AGN being the continuation at the massive-
SF galaxies, which emphasizes the role of the dominant
end of the normal SF sequence. An interesting question
stellar population in regulating star formation.
is whether the mid-IR luminosity in powerful AGN is
driven by “normal” ongoing SF, or by hot dust left over
3.8. Effect of AGN on the Energy Budget
after the last episode of SF. Further investigation is re-
quired to fully understand this matter. In the previous sections we noted that the emission
from the AGN could have a significant effect on the
3.7. SSFR Dependence on Mid-IR Color
IR emission, and potentially bias the luminosities of the
Recent work on resolved nearby galaxies has shown a AGN galaxy class. This could be particularly important
definitecorrelationbetweenIRcolorandluminosity. Shi for low SFR galaxies. We test for this effect by esti-
et al. (2011) found that for a variety of sources ranging mating the contribution of the AGN to the total energy
from ULIRGs to blue compact dwarf galaxies, the flux budget for sources classified from the BPT diagram as
ratio f /f traces the SSFR and also correlates either an AGN or as an SF galaxy. To do so, we analyze
24µm 5.8µm
withthecompactnessofstarformingregions. Whileide- the fraction of the 12µm luminosity contributed by each
ally we would like to know the SFR surface density, we of the four templates used in the SED fitting process to
first explore the relation between SSFR and 4.6–12µm our optical+IR photometry, paying particular attention
galaxy color, as shown in Figure 8. Galaxies from the totheAGNcomponent. Figure9showsthemedianfrac-
mainSFsequence(bluecontours)correlatestronglywith tion of 12µm luminosity contributed by each template,
IR color, with strong AGN (black contours) continuing for objects classifiedas SF galaxies. The majority of the
the trend at bluer colors. Weak AGN extend (red con- poweris splitamongthe irregular(Im), spiral(Sbc) and
tours) that relationship remarkably well toward the low elliptical (E) templates, though the AGN contribution
star formation end, albeit with higher dispersion and a becomessignificantforthemostluminoussources,above
slightly steeper slope. Hence, for the same increase in 1010.8L⊙. ThisimpliesthattheAGNhasanegligiblein-
SSFR, AGN experience a smaller variation in IR color fluence in SF galaxies. Figure 10 shows these fractions
than typical SF objects. This is probably due to the for weak and strong AGN galaxy classes. The ellipti-
combinationofa metallicity effect andthe differentstel- cal component is now more prominent in weak AGN of
10 Donoso et al.
Fig.9.—Medianofthe12µmluminosityfractioncontributedby Fig.10.— Same as Figure 9, but for objects classified as weak
thefourtemplatesfromAssefetal.(2010)thatarefittedtoSDSS AGN(dashed) andstrongAGN(solid).
ugriz photometry and WISE 3.4 µm, 4.6 µm and 12µm fluxes
(wealsouse22µmdatawhenavailable). Weincludeonlyobjects 4. SUMMARY
classifiedasSFfromtheBPTdiagram.
In this work we have taken advantage of recently re-
leaseddatafromWISEandSDSStoconstructthelargest
low luminosity, which is not unexpected. In general, the IR-optical sample of galaxies with 12µm fluxes and op-
AGN component is now more important but is still far tical spectra available at hzi∼0.1. This sample allowed
from contributing significantly below 1010.8L⊙. In most us to investigate with high statistical significance how
weak AGN (∼80%) the AGN contribution to the 12µm physicalparameterssuchascolor,stellarmass,metallic-
luminosity is below 40%. About ∼70% of strong AGN ity,redshift,andSFRof12µm-selectedgalaxiescompare
show similar low AGN contributions at 12µm. We note with purely optically selected samples. We have quan-
thatinFigure10weusedWISEaperturephotometryfor tified how the SFR estimates compare for the different
extended, nearby sources and profile-fitting magnitudes spectral types as a function of stellar mass, galaxy age
for unresolved galaxies with χ2 < 3 (see Section 4.5 of andIRcolorinordertopinpointtheunderlyingsourceof
WISE Preliminary Release Explanatory Supplement for 12µm emission and therefore up to what extent it could
furtherdetails). Althoughthedifferencesaresmall,aper- be interpreted as a useful SFR indicator.
ture photometry improves the quality of the SED fit of The main results of this paper can be summarized as
lowluminositygalaxies,asitcapturesthemoreextended follows:
flux of objects at low redshifts.
Note that in both cases, SF galaxies and AGN, the • In general, the WISE-SDSS 12µm-selected galaxy
emission is dominated by the spiral (Sbc) component, population traces the blue, late-type, low mass se-
whichhas a relativelyhigh mid-IRSED. This is because quence of the bimodal galaxy distribution in the
the template, originallyconstructedfromColemanetal. localUniverse. Italsotracesintermediate-typeob-
(1980) and extended into the UV and IR with Bruzual jects with active nuclei, avoiding the bulk of the
& Charlot (2003) synthesis models, also considers emis- red and “dead” galaxies without emission lines.
sion in the mid-IR from dust and polycyclic aromatic Most sources have normal to LIRG luminosities,
hydrocarbons (PAHs). These are added by ad hoc lin- but (few) ULIRGs are also present.
ear combinations of appropriate parts of NGC 4429 and
M82 SEDs obtained by Devriendt et al. (1999). Fig- • The IR emission of SF galaxies and strong AGN,
ure 11 shows the SED fit for AGN with luminosities of dominated by the blue, young stellar population
L[OIII] ∼ 106L⊙ and L[OIII] ∼ 107L⊙. In each case we component,iswellcorrelatedwiththeopticalSFR.
There is a small tendency of low SFR systems to
plottheobjectwiththemedianχ2,i.e. thetypicalfitfor
have slighly lower IR luminosity when compared
sources of those luminosities. The 9 photometric bands
to the canonical relationof Chary & Elbaz (2001).
arewellfittedbythemodelinmostcases. ForAGNwith
L[OIII] >107.5L⊙ wefindtheSEDfitisreasonablygood Tbehceosmeearme olorwe tSrFaRns,plaorwenmt adsusesytsotetmhse tihnactrelaiksienlyg
(althoughingeneralwithhigherχ2)exceptatthe22µm
fractionoflightthatescapesunabsorbedandhence
band. We believe this is caused by the limitation of the
suppresses L . However other effects like the
algorithm (not the templates themselves) to properly fit 12µm
dust distribution or metallicity could be relevant
highly reddened AGN fainter than their hosts, as it is
as well. The latter is shown by the (weak) SSFR
designedto punish the excessive use of reddening on the
dependenceonL . Ingeneral,themid-IRemis-
AGNwhenfewrelevantdatapointsareused. Modifying 12µm
sion at 22µm follows similar correlations seen for
the algorithm slightly to remove this behavior, we are
the 12µm-selected sample, suggesting that these
abletoobtainfitswithbetterχ2 valuesforthese objects
resultsdonotcriticallydependonasingleIRband.
assigningmuchhigherAGNfractionsandreddeningval-
ues,yetthelackoffartherIRdatatodeterminetheorigin • SF galaxies are forming stars at an approximately
ofthe 22µmexcessmakesthese numbers alsouncertain. constantrateper unit massforanIRoutput rang-
Therefore, while these results do not definitely rule out ing over five orders of magnitude. There is small
that the central AGN could have a considerable effect tendency for more luminous objects to have en-
in some extreme sources (e.g. the very strong AGN), it hancedSSFR, whichcouldbe interpretedasasign
certainlyshowsthattheyarenotrelevantformostofthe of SF histories peaking toward later times. How-
galaxy populations analyzed in this paper. ever, this residual dependence seems to be caused