Spectral decomposition of starbursts and AGNs in 5-8 micron Spitzer IRS spectra of local ULIRGs

0.25 MB·English

by  E. Nardini G. Risaliti

Mon.Not.R.Astron.Soc.000,1–5(2002) Printed3February2008 (MNLATEXstylefilev2.2)
Spectral decomposition of starbursts and AGNs in 5–8 µm
Spitzer IRS spectra of local ULIRGs
E. Nardini,1 G. Risaliti,2,3 M. Salvati,2 E. Sani,1
8
M. Imanishi,4 A. Marconi1 and R. Maiolino5
0
0 1 Dipartimento di Astronomia, Universit`adi Firenze, L.goE. Fermi 2, 50125 Firenze, Italy. E-mail: [email protected]
2 2 INAF - Osservatorio Astrofisico di Arcetri, L.goE. Fermi 5, 50125 Firenze, Italy
3 Harvard-Smithsonian Centerfor Astrophysics, 60 Garden St. Cambridge, MA 02138 USA
n
4 National Astronomical Observatory, 2-21-1, Osawa, Mitaka, Tokyo181-8588, Japan
a
5 INAF - Osservatorio Astronomico di Roma, viadi Frascati 33, 00040 Monte Porzio Catone (RM), Italy
J

8
2
ReleasedXxxxXxxxxXX

]
h
p ABSTRACT
- We present an analysis of the 5–8 µm Spitzer-IRS spectra of a sample of 68 local
o
Ultraluminous Infrared Galaxies (ULIRGs). Our diagnostic technique allows a clear
r
t separation of the active galactic nucleus (AGN) and starburst (SB) components in
s
the observed mid-IR emission, and a simple analytic model provides a quantitative
a
[ estimate of the AGN/starburst contribution to the bolometric luminosity. We show
that AGNs are ∼ 30 times brighter at 6 µm than starbursts with the same bolomet-
1
ric luminosity, so that even faint AGNs can be detected. Star formation events are
v confirmedas the dominant power sourcefor extreme infraredactivity, since ∼85%of
4
ULIRG luminosity arises from the SB component. Nonetheless an AGN is present in
0
3 the majority (46/68) of our sources.
4
Key words: galaxies: active; galaxies: starburst; infrared: galaxies.
.
1
0
8
0
: 1 INTRODUCTION ence between the 3-µm to bolometric ratios in AGNs and
v starbursts(∼twoordersofmagnitudelarger intheformer).
Xi Ultraluminous Infrared Galaxies (ULIRGs, LIR > 1012L⊙) This makes the detection of the AGN component possible
are thelocal counterpartsof thehigh-redshift objects dom- even when the AGN is heavily obscured and/or bolometri-
r
a inatingthecosmicbackgroundinthefar-infrared andmilli- cally weak compared to thestarburst. Howevertheoriginal
metricbands.Unveilingthenatureoftheirenergy sourceis prescription islimited bythelowqualityoftheavailable L-
fundamental in order to understand the star formation his- bandspectra of ULIRGs(R06, Imanishiet al. 2006), which
toryandtheobscuredAGNactivityinthedistantUniverse. makes the results on individual sources highly uncertain,
Since their discovery, several infrared indicators have except for the ∼10–15 brightest objects.
been proposed to determine whether the central engine in
ULIRGs is an AGN or a starburst (SB). The presence of Atpresentwehaveextendedtheanalysistothe5–8µm
high-ionizationlinesinthemid-IRspectraofULIRGspoints spectralband,usingtheobservationsoftheIRSinstrument
to AGN activity, while intense PAH emission features are (Houck et al. 2004) onboard Spitzer. We disentangled the
typical of starburst environments (Genzel et al. 1998; Lau- AGNandSBcontributionstotheobserved5–8µmemission
rent et al. 2000). Recently, the absorption feature of amor- of ULIRGs by combining average spectral templates repre-
phous silicate grains centered at 9.7 µm has also been used sentingthedifferentpropertiesofthetwophysicalprocesses
together with the PAH emission to assess the nature of the atwork.ThehighqualityofSpitzer-IRSdata,inadditionto
obscuredpowersource(Spoonetal.2007).Analternateway therelativelylowdispersionoftheintrinsiccontinuumprop-
todisentangleAGNsandSBsinULIRGshasbeenproposed erties of both AGNs and starbursts in this spectral range,
by Risaliti et al. (2006, hereafter R06), based on the sepa- allowsamuchmoreaccuratedeterminationoftheAGN/SB
rationofthetwocontinuumcomponentsin3–4µmspectra. componentsthanpossibleatotherwavelengths(e.g.X-rays)
This method has been successfully applied to a sample of or with other diagnostic methods based on emission lines.
∼50 nearby ULIRGs (Risaliti, Imanishi & Sani 2007, sub- In this paper we present our decomposition method, and
mitted) and provided an estimate of the average AGN/SB discuss a simple analytical model providing a quantitative
contribution to ULIRGs. The key reason for using the con- estimate of theAGN/SBcontribution tothebolometric lu-
tinuumemission at λ≃3–4 µm as adiagnostic isthediffer- minosity of each source.
2 E. Nardini et al.
2 OBSERVATIONS AND DATA REDUCTION
Inordertoperformadetailedstudyofarepresentativesam-
pleof ULIRGsin thelocal Universe, weselected 68 sources
with z < 0.15 and a 60-µm flux density f > 1 Jy. Most
60
of the objects are taken from the IRAS ULIRG 1 Jy sam-
ple (Kim & Sanders 1998), but a few more sources in the
southernhemispherehavealsobeenincluded.Thefluxlimit
at 60 µm ensures an unbiased selection with respect to the
relative AGN/SB contributions.
IRS observations were obtained within three differ-
ent programs: PID 105 (PI J.R.Houck), PID 2306 (PI
M.Imanishi), PID 3187 (PI S.Veilleux). The coadded im-
agesprovidedbytheSpitzerScienceCenter (afterthetreat-
ment with pipeline version S13.0) have been background-
subtracted by differencing thetwo observations in the nod-
ding cycle. The spectra havebeen extracted and calibrated
following thestandardprocedureforpoint-likesourceswith
the package SPICE. The flux uncertainties have been esti-
mated from source and background counts (in e−/s). Fi- Figure 1.Comparisonbetween theSBtemplateofB06(dashed
redline)andourtemplate(solidblueline),constructedfromthe
nally,weperformedasmoothconnectionbetweentheShort-
emission of the 5 brightest SB-dominated ULIRGs. The shaded
Low spectral orders, with no necessity of relative scaling.
area showsthe1-σ rmsdispersion inthe5ULIRGspectra.The
Outofthe68spectra,wealreadypublished48inIman-
verticallong-dashed green lines enclosethefittingregion.
ishi et al. 2007. Six more spectra are shown in Armus et
al. 2006, 2007. The remaining 14 spectra are analyzed here
forthefirsttimeandwillbefullypresentedinaforthcoming
AGN. Our recent L-band analysis of bright ULIRGs
paper (Nardiniet al. 2008, in prep.).
shows that the intrinsic AGN emission is due to hot dust
grainsandthefluxdensityiswelldescribedbyafeatureless
power law with a fixed spectral index: f ∝ λ1.5. Here we
ν
3 THE 5–8 µm AGN/SB SEPARATION
adoptthesamespectralshapeupto8µm,inagreementwith
DespitethediversityoftheglobalIRSspectraofpureAGNs new Spitzer observations of a large sample of local type 1
andpureSBs,andthecomplexityofthephysicsinvolved,lit- quasars (Netzeret al. 2007).
tledispersionisseenatwavelengthsshortwardofthe9.7µm An active nucleus is much more compact than a cir-
silicate feature. This makes possible the use of universal cumnuclearstarburstregion.Asaconsequence,thenear-IR
AGN/SB templates to reproduce the spectral properties of radiationduetothindustreprocessingcanbeitselfstrongly
ULIRGsinthe5–8µminterval.Inthefollowingwedescribe reddened because of a compact absorber along the line of
thetemplates adopted in ourmodel. sight. We therefore introduce an exponential attenuation
Starburst. The mid-IR spectral features of local SB factor e−τ(λ), where the optical depth follows the conven-
galaxies show very little variations from one object to an- tional law τ(λ) ∝ λ−1.75 (Draine 1989). A similar correc-
other in the 5–8 µm wavelength range, while larger differ- tion is not needed in the SB template. We stress that this
encesarepresentinthe∼9–30µmband(Brandletal.2006, does not imply that the starburst spectrum is not affected
hereafter B06). In order to check if this is the case in the byinnerobscuration;thepossibleeffectsofthisobscuration
ULIRG luminosity range as well, we analyzed the sources (whichareclearatlongerwavelengths,e.g.inthesilicateab-
in our sample estabilished to be starburst-dominated by sorption features at 9.7 and ∼18 µm) are however already
multi-band studies. We did not find significant variations accounted for in the adopted observational template.
among the spectra, concluding that a fixed template can Summarizing, the different contributions to the ob-
be used to represent the 5–8 µm SB component in local served energy output of a ULIRG can be parametrized as
ULIRGs.Webuiltsuchtemplateusingthefivebrighestob- follows:
jectsamongthepureSBsinoursample(IRAS10190+1322, fobs(λ)=fint (1−α )usb+α uagne−τ(λ) (1)
IRAS 12112+0305, IRAS 17208−0014, IRAS 20414−1651 ν 6 6 ν 6 ν
(cid:2) (cid:3)
and IRAS 22491−1808), whose underlying continuum has where α is the AGN contribution to the 5–8 µm intrinsic
6
been reproduced by a power law and normalized at 6 µm flux density fint, while usb and uagn are the SB and AGN
6 ν ν
before adding.OurSBtemplateis shown in Fig.1,together templatesnormalized at6µm.Apartfromthefluxnormal-
with its dispersion in the whole IRS spectral band and the ization,ourmodelcontainsonlytwofreeparameters,i.e.α
6
B06 template. The little spectral dispersion below 8 µm and the optical depth to the AGN τ(6 µm). They are both
among SBs of different luminosity (to be compared with shown in Tab.1.
their large differences at longer wavelengths) is in itself an Additional high-ionization emission lines and molecu-
interestingresult,whichshouldbefullyinvestigatedthrough lar absorption features (due to ices and aliphatic hydrocar-
detailedemissionandradiativetransfermodels.Concerning bons), whenever present, were fitted by means of gaussian
this we only notice that such a remarkable similarity can profilesexceptforthewatericeabsorptionat∼6µm,repro-
result from the spatial integration over a large number of ducedwith thelaboratory profile from theLeiden database
individualstar-forming regions. corresponding to pureH O ice at 30 K.
2
AGN and starburst in ULIRGs 3
ofAGNactivity,andcanbeused(a) totesttheconsistency
ofourdecompositionmethodand(b)toestimatetherelative
AGN and SB contributions to the bolometric luminosity of
oursample. Wedefinethe 6-µm tobolometric ratio as:
ν fint
R= 6 6 . (2)
(cid:18) FIR (cid:19)
whereFIR isthetotalinfraredflux,estimatedasinSanders
&Mirabel(1996).Sincetheintegratedinfraredluminosityof
ULIRGs coincides almost exactly with their bolometric lu-
minosity,Risafairapproximationtothefractionoftheto-
talenergyoutputthatisintrinsicallyemittedinthe5–8µm
range. Reminding that the intrinsic AGN/SBcontributions
are α fint and (1−α )fint respectively, and decomposing
6 6 6 6
FIR as FIaRgn +FIsRb, a simple connection between R and α6
is brought out:
RagnRsb
R= , (3)
α Rsb+(1−α )Ragn
6 6
provided that Ragn and Rsb, the equivalents of R for pure
(unobscured) AGNs and pure SBs, are defined as in Eq.2.
We fitted the theoretical R(α ) relation (Eq.3) to our data
6
considering Ragn and Rsb as free parameters, and found:
logRagn =−0.49+0.12 and logRsb =−1.93+0.03. (4)
−0.13 −0.03
WenotethatRagn turnsouttobesomewhathigherthantra-
ditional estimates based on AGN spectral energy distribu-
Figure 2. Three representative examples of the typical 5–8 µm tions:forexample,wederivelogRagn ∼−0.6from theSED
spectral shapes of ULIRGs. The differences among the spectra
of Elvis et al. (1994). This suggests that the local quasars
are entirely due to the different AGN contribution and its ob-
(mostly PG quasars) used to build the mentioned SED can
scuration. Whenever the AGN is the dominant power source, as
becontaminatedtosomeextentbyastarburstcontribution,
inMrk231, astrongcontinuum almostobliterates thePAHfea-
tures. On the contrary, in Mrk 273 the AGN is fainter and the in agreement with recent studies (Netzeret al. 2007).
spectraloutlineofastarburstisclearlyidentified.Asimilarspec- AccordingtoEq.4,AGNsare∼30timesmoreluminous
trumisexhibitedbyIRAS20551−4250, butthefeaturesareless at 6 µm than starbursts with the same bolometric lumi-
prominentandthecontinuumissteeper:thissourceharboursan nosity. We are now able to quantify the AGN contribution
obscuredAGN.Ineachpanel,inadditiontothedata(greenfilled (αbol =FIaRgn/FIR) to the total infrared luminosity of each
circles)andtheirbestfits(blackthinline),wehaveincludedthe source:
reddened AGN (red dot-dashed line) and starburst (blue dotted
α
line)components. α = 6 , (5)
bol α +(Ragn/Rsb)(1−α )
6 6
where Ragn/Rsb ≃ 28. The values of α are listed in
bol
Fig.2 shows the spectral decomposition of three rep- Tab.1. Our estimates for the ∼15 brightest sources are in
resentative ULIRGs. In spite of the great diversity of the
good agreement with those of R06 and with the Genzel et
observed spectra, our simple model provides a good fit of
al. (1998) and Laurent et al. (2000) mid-IR diagnostic di-
each spectrum in the sample: the residuals from the best
agrams. Considering the whole sample, our results can be
fits are in all sources smaller than 10% at all wavelengths
compared with the optical classification and with L-band
(though the fits are not formally acceptable in a statistical
andhardX-raystudies,whenavailable.Asubstantialagree-
sense, with a reduced χ2 &2, due to the small error bars
ment is obtained in all cases. It is worth noticing that the
and the remaining unfitted minor features). In particular,
optical classification alone gives incompleteinformation: all
both the PAH emission and the continuum are always well
the sources classified as Seyferts show clear traces of AGN
reproduced: this implies that the large variations in the 5–
activity,but7outof8amongtheULIRGswithα >0.25
bol
8µmspectralshapeofULIRGsareentirelyduetotheAGN and τ > 1 are indeed classified as LINERs or H ii regions.
contribution and its obscuration. A detailed analysis of the
LINERsareagainconfirmedtoberatherheterogeneouswith
results for each source and a physical interpretation of pe-
respect to the nature of their energy source. Such ambigui-
culiarcasesarethesubjectofaforthcomingpaper(Nardini
ties can besolved by applyingour diagnostic.
et al. 2008, in prep.).
ByinvertingEq.5therelationbetweenRandα takes
bol
the neat form R=α Ragn +(1−α )Rsb, and is plotted
bol bol
inFig.3a.Asafinalcheckwehavecomputedforeachsource
4 AGN/SB BOLOMETRIC CONTRIBUTIONS thefollowing quantities:
Thelargedifferencebetweenthe5–8µmtobolometricratios Ragn = ν6α6f6int and Rsb = ν6(1−α6)f6int . (6)
inAGNsandSBsimpliesthatthisratioisitselfanindicator (cid:18) αbolFIR (cid:19) (cid:20) (1−αbol)FIR (cid:21)
b b
4 E. Nardini et al.
Table 1. Spectral parameters for the 68 sources in our sample. α6: AGN contribution to the intrinsic continuum emission at 6 µm
(in percent). τ: Optical depth of the AGN component at 6 µm (we assume τ = 0 for the sources with no detected AGN). αbol: AGN
contributiontothebolometricluminosity(inpercent).Theerrorsinαbol areduetothestatisticaluncertaintybothinthefluxamplitude
oftheAGN/SBcomponentsandintheratiosRagn andRsb.Thesystematiceffectsarediscussedinthetext.O/X/L:SB/AGN/LINER
classificationbasedonoptical,X-rayandL-bandspectroscopy.A:AGN,L:LINER,A*:AGN,tentativedetection.References:1:Veilleux
etal.1999,2:Veilleuxetal.1995,3:Ducetal.1997,4:Iwasawaetal.2005,5:Severgninietal.2001,6:Franceschinietal.2003,7:Balestra
etal.2005,8:Vignatietal.1999,9:Imanishietal.2003,10:Imanishietal.2006,11:Risalitietal.2006,12:Sanietal.2007.†:Sources
withsimultaneousαbol >0.2andτ >1.
Source α6 τ αbol O/X/L Source α6 τ αbol O/X/L
ARP220 75±1 1.40±0.01 9.8+−32..77 L1/SB4/SB10 IRAS14197+0813 75±2 2.10±0.14 10+−54 –/–/–
IRAS00091−0738† 90±1 1.81±0.04 25+8 SB1/–/– IRAS14252−1550 <20 <0.04 <0.9 L1/–/SB10
−7
IRAS00188−0856 96±1 0.60±0.02 45±9 L1/–/A*10 IRAS14348−1447 51+−35 <0.09 3.7+−21..15 L1/–/SB11
IRAS00456−2904 <1.0 0 <0.04 SB1/–/– IRAS15130−1958 91+1 <0.01 28+9 A1/–/A10
−3 −11
IRAS00482−2721 <54 <0.04 <4.1 L1/–/– IRAS15206+3342 52±1 0.30±0.02 3.9+−11..72 SB1/–/SB10
IRAS01003−2238† 96±1 1.58±0.02 49±9 SB1/–/– IRAS15225+2350 89±1 0.77±0.02 22+7 SB1/–/SB10
−6
IRAS01166−0844 87±1 1.19±0.06 20+7 SB1/–/– IRAS15250+3609 94±1 0.91±0.01 35+9 L2/SB6/–
−6 −8
IRAS01298−0744† 98±1 1.79±0.02 74+8 SB1/–/– IRAS15462−0450 90+1 <0.01 25+10 A1/–/–
−9 −16 −18
IRAS01569−2939 85±1 1.13±0.04 17+6 SB1/–/– IRAS16090−0139 89±1 0.69±0.01 23+7 L1/–/A*10
−5 −6
IRAS02411+0353 <17 <0.01 <0.8 SB1/–/– IRAS16156+0146 90±1 0.37±0.01 25+8 A1/–/–
−6
IRAS03250+1606 <3.4 <0.11 <0.2 L1/–/A*10 IRAS16468+5200 85±1 0.77±0.02 18+6 L1/–/SB10
−5
IRAS04103−2838 56±1 0.08±0.02 4.5+−21..04 L1/–/– IRAS16474+3430 <4.9 <0.01 <0.2 SB1/–/A*10
IRAS05189−2524 91+−14 <0.01 28+−812 A1/A5/A12 IRAS16487+5447 21±1 <0.04 1.0+−00..54 L1/–/A*10
IRAS08572+3915 99±1 0.44±0.01 85+5 L1/–/A10 IRAS17028+5817 <1.2 0 <0.05 L1/–/A*10
−6
IRAS09039+0503 61±1 0.71±0.04 5.5+−21..57 L1/–/A*10 IRAS17044+6720 91±1 0.32±0.01 26+−86 L1/–/A10
IRAS09116+0334 <1.8 <0.01 <0.07 L1/–/A*10 IRAS17179+5444 84±1 0.31±0.02 16+6 A1/–/A10
−4
IRAS09539+0857† 91±1 1.85±0.03 27+8 L1/–/SB10 IRAS17208−0014 <7.9 <0.01 <0.4 L3/SB6/SB11
−7
IRAS10190+1322 <0.3 0 <0.02 SB1/–/SB10 IRAS19254−7245 89±1 0.21±0.08 23+9 A3/A6/A11
−7
IRAS10378+1109 73+−19 <0.01 9.0+−34..76 L1/–/A*10 IRAS20100−4156 86±1 0.47±0.02 19+−65 SB3/A6/SB11
IRAS10485−1447 60±1 0.16±0.03 5.3+−21..47 L1/–/A*10 IRAS20414−1651 <2.2 <0.07 <0.09 SB1/–/SB10
IRAS10494+4424 <0.4 0 <0.02 L1/–/A*10 IRAS20551−4250† 90±1 1.19±0.01 25+8 L3/A6/A12
−6
IRAS11095−0238† 94±1 1.22±0.01 35+9 L1/–/SB10 IRAS21208−0519 <0.9 0 <0.04 SB1/–/SB10
−8
IRAS11130−2659 84±1 1.15±0.02 16+6 L1/–/– IRAS21219−1757 99+1 <0.01 83+13 A1/–/A10
−5 −4 −48
IRAS11387+4116 <0.8 0 <0.03 SB1/–/SB10 IRAS21329−2346 43±2 <0.07 2.7+−10..39 L1/–/A*10
IRAS11506+1331 54+−121 <0.01 4.2+−12..99 SB1/–/A*10 IRAS22206−2715 <9.3 <0.17 <0.4 SB1/–/–
IRAS12072−0444† 94±1 1.06±0.01 37+9 A1/–/A10 IRAS22491−1808 <1.4 0 <0.06 SB1/SB6/–
−8
IRAS12112+0305 <22 <0.02 <1.0 L1/SB6/SB11 IRAS23128−5919 48±1 0.36±0.03 3.4+−11..50 L3/A6/A11
IRAS12127−1412 98±1 0.24±0.08 60+−1154 L1/–/A10 IRAS23234+0946 30+−23 <0.05 1.6+−00..86 L1/–/SB10
IRAS12359−0725 54+−243 <0.01 4.2+−23..29 L1/–/A*10 IRAS23327+2913 73±1 0.86±0.03 9.1+−32..87 L1/–/SB10
IRAS13335−2612 <0.6 0 <0.02 L1/–/– MRK231 93±1 <0.12 32+8 A1/A6/A10
−7
IRAS13454−2956 59+−128 <0.01 5.0+−23..28 A1/–/– MRK273 67+−14 <0.01 7.0+−22..87 A1/A7/A10
IRAS13509+0442 <0.6 0 <0.03 SB1/–/SB10 NGC6240 65+−68 0.64±0.24 6.5+−43..81 L2/A8/A12
IRAS13539+2920 <0.4 0 <0.02 SB1/–/SB10 4C+12.50 97±1 0.24±0.02 59+10 A1/–/–
−11
IRAS14060+2919 <0.4 0 <0.02 SB1/–/SB10 UGC5101† 92±1 1.09±0.03 30+9 L2/A9/A10
−8
Theresults areshown inFig.3band provethatourde- tribution.Theseelementsstrongly affecttheoverallmid-IR
composition method is reliable in estimating the AGN/SB emissionlongwardofthesilicateabsorptionfeature,andcan
contributions both to the 6-µm and to the bolometric lu- be investigated only through an analysis of the whole IRS
minosity of local ULIRGs. In fact, the estimated 6-µm to spectrum.
bolometric ratios of the SB component in the composite 2)Whilethe5–8µmtemplatesseemtohavelittledispersion
sources (which are located at the bottom right of the plot (asdiscussedin detailinSection3),thespread inthe6-µm
andareheavilydependentonourmodeling)arefullyconsis- tobolometricratiosRagn andRsb canbemuchhigher,mak-
tent,withintheerrors,withtheratiosofthepurestarbursts ingourestimatesoftheAGN/SBbolometricfractionsmore
(located at thebottom left and directly computedfrom the uncertainthanthoseinthe5–8µmband.Theuncertainties
measured 6-µm and IRAS fluxes). This success is promis- in α reported in Tab.1 are obtained assuming the mean
bol
ing in anticipation of a forthcoming study about the role ratios, with theerrors on themean givenin Eq.4.However,
of black hole accretion and star formation in the intense the dispersion around the best fit in Fig.3a is significantly
infrared activity characterising thedistant galaxies. larger. We therefore consider this dispersion (0.3 dex, con-
stant at all values of α ) as the actual uncertainty in the
bol
However,itisimportanttokeepinmindthemainlim- bolometric ratios of the individual sources. The numerical
itations of our approach: resultsontheindividualesourcesareanywaypreciseenough
1)Thenarrowwavelengthrangeusedinthisworksimplifies to estabilish which is the dominant source of the observed
thedecompositionanalysis,butpreventsusfromacomplete luminosity.
studyof thedustcomposition, density andgeometrical dis-
AGN and starburst in ULIRGs 5
spectraof68localULIRGs,observedwiththeSpitzerSpace
Telescope. We have been able to detect an AGN in more
than 60% of our sources, and estimate its contribution to
the bolometric luminosity. In a statistical sense, we con-
firmthatlocalUltraluminousInfraredGalaxiesarepowered
for ∼85% by intense star formation and for the remaining
∼15%byAGNactivity.Ourmethodprovestobesuccessful
in unveilingan intrinsically faint or obscured AGN inside a
ULIRG. In this context we also put on a sound basis our
initial assumption that the wavelength interval 5–8 µm is
an appropriate spectral range in order to search for AGNs:
an AGN turns out to be approximately 30 times more lu-
minous at 6 µm than a starburst with the same bolometric
luminosity.
ACKNOWLEDGMENTS
We are grateful to the anonymous referee for his/her help-
ful and constructive comments. We acknowledge financial
support from theprin-miur2006025203 grant and the ASI-
INAFgrant I/023/05/0.
REFERENCES
ArmusL.,etal.,2006,ApJ,640,204
ArmusL.,etal.,2007,ApJ,656,148
BalestraI.,BollerT.,GalloL.,LutzD.,HessS.,2005,A&A,442,
469
BrandlB.R.,etal.,2006,ApJ,653,1129(B06)
DraineB.T.,1989,ESASP,290,93
Figure 3. (a) Ratio R between absorption-corrected 6-µm lu- DucP.-A.,MirabelI.F.,MazaJ.,1997,A&AS,124,533
minosity and bolometric luminosity, versus the AGN bolometric ElvisM.,etal.,1994,ApJS,95,1
contributionαbol.TheerrorbarsofRareduetotheuncertainties FranceschiniA.,etal.,2003, MNRAS,343,1181
inthe total infraredfluxFIR andinthe intrinsicAGN fraction, GenzelR.,etal.,1998,ApJ,498,579
α6. The solid lineis the best fit of the R–α6 relation from Eq.3 HouckJ.R.,etal.,2004,ApJS,154,18
(plotted as a function of αbol using Eq.5). (b) Same as above, HouckJ.R.,etal.,2005,ApJ,622,L105
withtheAGNandSBcomponents plottedseparately. ImanishiM.,TerashimaY.,AnabukiN.,NakagawaT.,2003,ApJ,
596,L167
ImanishiM.,DudleyC.C.,MaloneyP.R.,2006, ApJ,637,114
Overall, if we consider as a confidence limit the value ImanishiM.,DudleyC.C.,MaiolinoR.,MaloneyP.R.,Nakagawa
α = 0.01 (i.e. the dispersion around our starburst tem- T.,RisalitiG.,2007,ApJS,171,72
bol
plate), an AGN is present in 46 of the 68 ULIRGs in our Iwasawa K., Sanders D. B., Evans A. S., Trentham N., Miniutti
sample(includingseveralofthoseopticallyclassifiedasHii G.,SpoonH.W.W.,2005, MNRAS,357,565
regions), but it is significant (α & 0.25) only in ∼30% of KimD.-C.,SandersD.B.,1998,ApJS,119,41
bol LaurentO.,MirabelI.F.,CharmandarisV.,GallaisP.,Madden
the cases. The SB process is responsible for almost 90% of
S.C.,SauvageM.,VigrouxL.,CesarskyC.,2000,A&A,359,
theobservedinfraredluminosityofULIRGs,withnosignif-
887
icant (i.e. > 5%) bias due to the sample selection. A simi- NetzerH.,etal.,2007,ApJ,666,806
lar fraction holds for the subsample of 34 sources optically RisalitiG.,etal.,2006,MNRAS,365,303(R06)
classified as LINERs. Our analysis is also consistent with SandersD.B.,MirabelI.F.,1996, ARA&A,34,749
thefindingsaboutthenatureofhigh-redshiftinfrared-bright SaniE.,etal.,2007,arXiv,709,arXiv:0709.1344
galaxiesdetectedin24µmSpitzer MIPSsurveys.IRSspec- Severgnini P., Risaliti G., Marconi A., Maiolino R., Salvati M.,
troscopy shows that they are mostly z ∼1–3 galaxies, with 2001,A&A,368,44
SpoonH.W.W.,MarshallJ.A.,HouckJ.R.,ElitzurM.,HaoL.,
anapparentbiastowardAGN-dominatedsources(Houcket
al. 2005). This is in agreement with the ∼30 times higher Armus L., Brandl B. R., Charmandaris V., 2007, ApJ, 654,
L49
AGNrelativeemission inthe5–8µmrest-framewavelength
Veilleux S., Kim D.-C., Sanders D. B., Mazzarella J. M., Soifer
range we havepointed out.
B.T.,1995,ApJS,98,171
VeilleuxS.,KimD.-C.,SandersD.B.,1999, ApJ,522,113
VignatiP.,etal.,1999, A&A,349,L57
5 CONCLUSIONS
TheuseofaveragetemplatesforAGNandSBemission has
allowedustodisentanglethetwocomponentsinthe5–8µm