Spectral Energy Distributions of Weak Active Galactic Nuclei Associated With Low-Ionization Nuclear Emission Regions

0.44 MB·English

by  Michael Eracleous

ToAppearintheAstrophysicalJournalSupplements
PreprinttypesetusingLATEXstyleemulateapjv.08/22/09
SPECTRALENERGYDISTRIBUTIONSOFWEAKACTIVEGALACTICNUCLEIASSOCIATEDWITH
LOW-IONIZATIONNUCLEAREMISSIONREGIONS
MICHAELERACLEOUS1,2,3,JASONA.HWANG1,3,&HE´LE`NEM.L.G.FLOHIC1,4
ToAppearintheAstrophysicalJournalSupplements
ABSTRACT
Wepresentacompilationofspectralenergydistributions(SEDs)of35weakactivegalacticnuclei(AGNs)
0
inlow-ionizationnuclearemissionregion(LINERs)usingrecentdatafromthepublishedliterature. Wemake
1
useofpreviouslypublishedcompilationsofdata, aftercomplementingandextendingthemwithmorerecent
0
data. The main improvementin the recent data is afforded by high-spatial resolution observations with the
2
ChandraX-RayObservatoryandhigh-spatialresolutionradioobservationsutilizinganumberoffacilities. In

n addition,aconsiderablenumberofobjectshavebeenobservedwiththeHubbleSpaceTelescopeinthenear-IR
a throughnear-UVbandssincetheearliercompilationswerepublished. Thedataincludeupperlimitsresulting
J
fromeithernon-detectionsorobservationsatlowspatialresolutionthatdonotisolatetheAGN.Forthesake

7 ofcompleteness,wealsocomputeandpresentanumberofquantitiesfromthedata,suchasoptical-to-X-ray
1 spectralindices(a ),bolometriccorrections,bolometricluminosities,Eddingtonratios,andtheaverageSED.
ox

Weanticipatethatthesedatawillbeusefulforanumberofapplications.Inacompanionpaper,weuseasubset
]
A ofthesedataourselvestoassesstheenergybudgetsofLINERs.
Subjectheadings:galaxies:nuclei—galaxies:active—X-rays:galaxies
G
.
h
1. INTRODUCTION ing questions related to LINERs and their central engines,
p
suchaswhethertheweakAGNsareresponsibleforpowering
- Low-ionization nuclear emission regions (LINERs) were
o theemission-lineregionsbyphotoionization,whetherornot
identified as a class by Heckman (1980) based on the rela-
r modelsforlow-radiativeefficiencyaccretionflowsprovidea
t tive intensities of their oxygen emission lines. Their defin-
s ing properties are: [O II] l 3727 / [O III] l 5007 > 1 and gooddescriptionof the propertiesof these weak AGNs, and
[a [O I] l 6300/ [O III] l 5007> 1/3. Optical spectroscopic wrehgaiotnisotfhietsrhoolestogfafleaexdyb(ascekeothfethdeiswcuesaskioAnGofNthoensethaendceontthrearl
surveys(e.g., Ho,Filippenko,&Sargent1997a) have shown
related issues in Ho 2008). A good example of such an ap-
1 them to be very common, occurring in approximately 50%
v of the nuclei of nearby galaxies. Suggestions for the ul- plicationistheworkofMaoz(2007),whoindependentlycol-
4 timate power source of the emission lines include (a) a lectedmeasurementsatveryspecificfrequenciesforasubset
2 weak active galactic nucleus (an AGN harboring an accret- oftheobjectsinoursample.
9 ing, supermassive black hole; e.g. Halpern&Steiner 1983; Motivatedbytheabovequestionswesetouttocollectpub-
2 lisheddatafromtherecentliteraturetodefinethespectralen-
Ferland&Netzer 1983), (b) hot stars (either young or old,
. ergydistributions(SEDs) ofweakAGNsin LINERsatpho-
1 e.g.Terlevich&Melnick1985;Filippenko&Terlevich1992;
ton energiesabove1 Ry. Our primaryinterestwas to assess
0 Shields1992;Barth&Shields2000;Binetteetal.1994),and
the energy budgets of LINERs and investigate whether the
0 (c)shocks(e.g.Heckman1980;Dopitaetal.1996,andrefer-
AGN is powerful enough to power the the observed emis-
1 encestherein). Recentradio, UV, andX-ray surveysathigh
sionlinesby photoionization. In the processwe constructed
: spatialresolution,mid-IRspectroscopy,and variabilitystud-
v thefullSEDs(fromradioto X-rayenergies),thusextending
ies have uncovered weak AGNs in the majority of LINERs
i and updating the pioneeringwork of Ho (1999). In this pa-
X studiedsofar,suggestingthattheymakeupa large(perhaps
per we present the data and a number of quantities derived
the largest) subset of all AGNs (e.g. Filhoetal. etal. 2004;
r
fromthemwithoutanyscientificinterpretation.Weanticipate
a Nagar,Falcke,&Wilson2005;Barthetal.1998;Maozetal.
that this compilation of data will be useful for a number of
2005;Hoetal.2001;Terashima&Wilson2003;Dudiketal.
additionalapplications,includingthosementionedinthepre-
2005; Flohicetal. etal. 2006; Gonza´lez-Mart´ınetal. 2006;
viousparagraph. In a followuppaperwe use a subsetofthe
Dudik,Satyapal,&Marcu2009)
datapresentedheretoaddressthespecificquestionregarding
The ubiquity of LINERs suggests that they are an impor-
theenergybudgetofLINERs(Eracleousetal.2009,hereafter
tantcomponentofthenucleiofgalaxiesinthelocaluniverse.
paperII).
Moreover, they trace the population of AGNs at the lowest
In §2 we present the sample of galaxies and their basic
luminosities. The data available from the most modern ob-
properties. In §3 we present the SEDs, discuss extinction
serving facilities allow us to address a number of outstand-
corrections, and comment on a number of objects that war-
1Department of Astronomy and Astrophysics, The Pennsylvania State rantspecialattention. Finally, in §4 we presenta numberof
University,525DaveyLab,UniversityPark,PA16802 quantitiesderivedfromthedata,namelytheoptical-to-X-ray
2CenterforGravitationalWavePhysics,ThePennsylvaniaStateUniver- spectralindices(a ox),thebolometricluminosities,Eddington
sity,104DaveyLab,UniversityPark,PA16802 ratios,andtheaverageSED.Weprovidedetailsofhowwese-
3Department of Physics & Astronomy, Northwestern University, 2131 lectedthedataandhowweconvertedthemeasuredquantities
TechDrive,Evanston,IL60208
to“monochromatic”luminosities.Weemphasizecaveatsand
4Department ofPhysicsandAstronomy, University ofCalifornia, 4129
FrederickReinesHall,Irvine,CA92697 sources of uncertainty in our methods. We also catalog the
2 Eracleous,Hwang,&Flohic
originaldata(withappropriatereferencestothesourceofthe
information) so that one can reproduce our procedureswith TABLE1
modifications,ifdesired. SAMPLEOFGALAXIESANDTHEIRBASICPROPERTIES
Distance(Mpc)
2. TARGETSANDTHEIRPROPERTIES
Hubble
WeconstructedasampleofobjectsclassifiedasLINERsor Galaxy Typea Tullyb SBFc PNLFd Othere
transitionobjectsby Hoetal. (1997b)5 basedon the follow- (1) (2) (3) (4) (5) (6)
ing considerations. All objects were required to have mea- NGC0266 SB(rs)ab 62.4
sured optical emission line luminosities and X-ray observa- NGC0404 SA(s)0 2.4 3.0 3.1
NGC1097 SB(rl)b 14.5
tions with Chandrathat yielded a measurementof the AGN
NGC1553 SA(rl)0 13.4 17.2
X-rayspectrumoranupperlimittoitsX-rayflux. Thehigh NGC2681 SBA(rs)0/a 13.3 16.0
spatial resolution of Chandra allows us to spatially separate NGC3031(M81) SA(s)ab 3.6 3.6 3.6
NGC3169 SA(s)a 19.7
the AGN emission from most of the diffuse, thermal X-ray
NGC3226 E2 23.4 21.9
emission.WeshowedpreferencetoobjectsobservedintheX-
NGC3379(M105) E1 8.1 9.8 10.4
ray bandwith longexposuretimes i.e., objectswhoseX-ray NGC3507 SB(s)b 19.8
spectrahavehighsignal-to-noiseratio(S/N).Insuchcases,a NGC3607 Sa(s)0 19.9 21.2
NGC3608 E2 23.4 21.3
modelfit to the AGN spectrumcan givea directestimate of
NGC3628 SAb 7.7
thetotalextinctiontowardtheAGNcontinuumsource.More- NGC3998 SA(r)0 14.0 13.1
over,anysoft,unresolved,circumnuclearX-rayemissioncan NGC4111 SA(r)0 17.0 13.9
be separated spectrally. A significant subset of the sample NGC4125 E6 24.2 22.2
NGC4143 SAB(s)0 17.0 14.8
objectswereimagedintheUVwiththeHSTandanoverlap-
NGC4261(3C270) E2-3 30.0 31.6 36.1
pingsubsethadwellsampledSEDs inthe entirerangefrom NGC4278 E1-2 9.7 14.9
the radio to X-ray bands. We took advantage of these over- NGC4314 SB(rs)a 9.7
lappingsubsetsinorderto determineupperlimitstotheUV NGC4374(M84,3C272.1) E1 16.8 17.1 18.0
NGC4438 SA(s)0/a 16.8 11.3
fluxeswhentheywerenotmeasureddirectlyandtodetermine
NGC4457 SAB(s)0/a 17.4 10.7
bolometricluminositiesfromthedata,aswedetailin§3. NGC4486(M87,3C274) E0-1 16.8 14.9 15.8
Wereliedheavilyonpreviouslypublishedcompilationsof NGC4494 E1-2 9.7 15.8 16.8
data(e.g.,Ho1999;Ptaketal.2004)andonpaperspresenting NGC4548(M91) SBb(rs) 16.8 17.9 15.0
NGC4552(M89) E 16.8 14.3 19.6
largecollectionsofmeasurementsintheradio,UV,orX-ray
NGC4579(M58) SAB(rs)b 16.8 21.0
bands (e.g., Nagaretal, 2005; Doietal. 2005b; Maozetal. NGC4594(M104) SA(s)a 9.2 9.1 9.6 9.3
2005; Flohicetal. etal. 2006). We note that the work of NGC4636 E/S0 17.0 13.6 18.1
NGC4736(M94) (R)SA(r)ab 4.3 4.8 4.4 4.7
Maozetal. (2005) is particularly important because it uses
NGC5055(M63) SA(rs)bc 7.2
variability in the UV continuum to check whether the nu- NGC5866 S0 15.3 14.3 15.1
clear source in a LINER is an AGN. The previously pub- NGC6500 SAab 39.7
lishedcompilationswereupdatedandsupplementedwithdata NGC7331 SA(s)b 14.3 12.2 14.5
that became available later. Because of possible contami-
nation of the AGN emission by circumnuclearemission, we aHostgalaxyHubbletypesweretakenfromthecatalogofTully(1988).
adoptedmeasurementsfrom high-spatialresolutionobserva- bDistancesfromthecatalogofTully(1988).
cDistancesobtainedusingthesurfacebrightnessfluctuationmethod.SeeTonryetal.
tionswhichisolatetheAGN.Inthenear-IR,optical,andnear-
(2001). FollowingJensenetal.(2003),wehavecorrectedthedistancemodulusre-
UV bands, when given a choice we adopted measurements portedbyTonryetal.(2001)bysubtracting0.16mag.
fromimagesratherthanfromspectra. Thehighspatialreso- d Distancesobtainedusingtheplanetarynebulaluminosity functionmethod. See
lutionofChandraandtheHSTwereoptimal(seeexampleim- Herrmannetal.(2008)forNGC4736andcompilationofCiardulloetal.(2002)and
agesinEracleousetal.2002;Flohicetal. etal.2006). Inthe referencesthereinforothergalaxies.
e Distancesobtained usinga variety of different methods, as follows: NGC 404,
radioband,possible confusionfromcircumnuclearemission
NGC3031, andNGC4736usingthetipoftheredgiantbranch(TRGB)method
(orinsomecasesjets)isalsopossible,thereforewereliedon Karachetsevetal.(2002)(thedistancetoNGC3031obtainedbytheCepheidmethod
high-resolution VLA or VLBA/VLBI observations. For the by Freedmanetal. 1994, agrees exactly with the TRGB distance); NGC 4261,
NGC4552,andNGC4636usingthefundamentalplanerelation(seeGavazzietal.
sakeofcompleteness,wealsoincludemeasurementsmadeat
1999); NGC 4438, NGC 4457, and NGC 4579 using Tully-Fisher method (see
lowspatialresolution,butwetaketheseasupperlimitstothe
Gavazzietal.1999);NGC3031,NGC4548andNGC7331usingtheCepheidvariable
fluxoftheAGN.Includedinthesampleare13LINERswith method(Freedmanetal.2001,1994).
actual X-ray and UV (2500 A˚) measurements, 11 LINERs
withactualX-raymeasurementsbutnoUVmeasurements,6 types,theLINERtypesaccordingtoHoetal.(1997b)ofour
LINERs with actual UV (2500 A˚) measurements and X-ray samplegalaxies,andtheblackholemassesderivedfromthe
upperlimits,and5LINERswithX-rayupperlimitsonly. stellar velocitydispersion. We describeand discussthe data
The galaxies of the resulting sample are listed in Table 1 below.
alongwith their morphologicaltypes and distances. Table 2
listsGalacticreddening,centralstellarvelocitydispersionand 2.1. Distances
inferred black hole mass, and LINER type for the nuclei of
We have compileddistance measurementsfromthe litera-
thesamplegalaxies. TheGalacticreddeningwastakenfrom
ture that are based on a variety of techniques, which we in-
Schlegeletal. (1998). InFigure1 we showthe distributions
cludeinTable1. All35galaxieshavedistancescatalogedin
ofthedistancesreportedbyTully(1988), themorphological
Tully(1988),whichweredeterminedusingamodelforpecu-
liar velocitiesand assuming H =75 kms−1Mpc−1. These
5Hereafter, werefertoallobjectscollectively as“LINERs.” In§2.2we 0
arelistedincolumn(3)ofTable1. Forasignificantfraction
re-examinetheirclassificationsbasedonmorerecentlydevelopedcriteriaand
concludethattheycanallberegardedasLINERs. ofthegalaxiesinourcollection,distancemeasurementsbased
SEDsofWeakAGNsinLINERs 3
onmoredirecttechniquesareavailable,whichwealsolistin
Table 1. More specifically, 24 galaxies have had their dis-
tances determined via surface brightness fluctuations (SBF)
by Tonryetal. (2001), which we list in column (4) of Ta-
ble1aftermakingasystematiccorrectionof−0.16magtothe
distance modulus, followingJensenetal. (2003). In column
(5)ofTable1,welistdistancesto8galaxies,determinedvia
theplanetarynebulaluminosityfunction(PNLF)methodand
drawnmostlyfromthecompilationofCiardulloetal.(2002),
and references therein. Distances determined by any other
methodareincludedincolumn(6)ofTable1. Inthispaper,
weadoptthedistancesfromTully(1988)sothattheluminosi-
tieswederivecanbecompareddirectlytootherquantitiesre-
portedbyHoetal.(1997b),whousedthesamedistances(e.g.
emission line luminosities). None of the conclusionsof this
paper or paper II depend sensitively on the distance. How-
ever, we emphasize that differentapplications(e.g., detailed
studyandmodelingoftheSEDs)mayrequireamoreaccurate
distancethanthatofTully(1988).Insuchacasethedataand
derivedquantitieswepresentherecanbeeasilyconvertedto
adifferentdistance,viaasimplescaling.
2.2. SpectroscopicClassificationandLINERTypes
Incolumn(5)ofTable2welisttheLINERtypesaccording
toHoetal.(1997b),whicharebasedontherelativeintensities
ofthenarrowemissionlines.Theobjectsincludedinoursam-
ple are classified in this scheme either as pure LINERs (de-
notedbyL),orastransitionobjects,withdiagnosticlineratios
intermediatebetweenLINERsandHIIregions. Inabout1/3
oftheobjectsinthissample,therearedetectablebroadwings
ontheHa lineHoetal.(1997c);thesearetermed“type1.9”
LINERsandareidentifiedwitha“1”incolumn(5)ofTable2,
whileallotherobjectsarelabeledwitha“2.”Iftherewasany
ambiguity or uncertainty in the classification, both possible
classesarelisted.ThenucleusofM81hasanuncertainclassi-
fication;itcanbeeitheraLINERoraSeyfert.Sincethework
ofHoetal.(1997b),morerecentclassificationschemesusing
thesamediagnosticlineratioshavebecomeavailable,suchas
those of Kewleyetal. (2001), Kauffmannetal. (2003), and
Kewleyetal. (2006). Thus, we have applied these classifi-
cation schemes to the relative intensities of diagnostic lines
measured by Hoetal. (1997b) and report the outcome of
this exercise in column (6) of Table 2. We list the location
of each nucleus in the [O I]/Ha vs [O III]/H, [N II]/Ha vs
[O III]/H, and [S II]/Ha vs [O III]/Hdiagram¯s, with L de-
noting th¯e LINER region, S denoting¯the Seyfert region, H
denotingtheHIIregion,andCdenotingthe“composite”re-
gionin the [NII]/Ha vs [OIII]/Hdiagram(intermediatebe-
¯
tweenLINERsorSeyfertsandHIIregions).Usingthesecri-
teria, six objects are classified as “S/L/S,” however they fall
very close to the Seyfert-LINER boundary in the [O I]/Ha
vs [O III]/Hand [SII]/Ha vs [OIII]/Hdiagrams. Similarly,
¯ ¯
NGC4314isclassifiedas“L/C/H”butitfallsveryclosetothe
Composie-LINERand HII-LINER boundariesin [NII]/Ha
vs[OIII]/H,and[SII]/Ha vs[OIII]/Hdiagrams,respectively.
The mix¯of LINER types in our sa¯mple is dictated by the
availabilityofdata. Therefore,theremaysomebiasesinher-
itedfromthesurveysfromwhichweadoptedthedata. More
specifically, the surveys we have relied on targeted radio-
FIG.1.—Distribution ofthebasic properties ofthe hostgalaxies ofthe
sampleLINERs. Theblackholemassesinpanel(d)werederivedfromthe bright, UV-bright, and X-ray bright objects. Thus, we find
stellarvelocitydispersions,usingequation(1). NGC266,NGC3507,and that “transition” objects (according to Hoetal. 1997b) are
NGC4438arenotincluded inthelasthistogram because their blackhole under-represented in our sample relative to their number in
massesarenotavailable. NGC404isalsonotplottedinthelasthistogram
thesurveyofHoetal.(1997a),while“type1.9”LINERsare
becauseitisoutofrange,withlog(M/M⊙)=5.3. over-represented. We also note that our search for data, al-
4 Eracleous,Hwang,&Flohic
2.3. StellarVelocityDispersionsandBlackHoleMasses
TABLE2
PROPERTIESOFLINERNUCLEI Thestellarvelocitydispersionsofthesamplegalaxieswere
taken from the Hypercat database6, with the following ex-
Galactic LINERType ceptions: the values for NGC 404, NGC 4278, NGC 4314,
E(B−V)a s ⋆b log NGC 4579, NGC 4736, NGC 5055, NGC 6500 are from
Galaxy (mag) (kms−1) (MBH/M⊙)c Hod Kewleye Barthetal. (2002), while the value for NGC 1097 is from
(1) (2) (3) (4) (5) (6)
Lewis&Eracleous (2006). The black hole masses were es-
NGC0266 0.069 ... ... L1 L/L/L timatedfromthestellarvelocitydispersionsviatheM –s
BH ⋆
NGC0404 0.059 40 5.3 L2 L/C/L
relationship(Ferrarese&Merritt2000;Gebhardtetal.2000;
NGC1097 0.027 196 8.1 L1 L/L/S
NGC1553 0.013 177 7.9 L2/T2 ?/L/? Tremaineetal.2002),namely,
NGC2681 0.023 108 7.1 L1 L/L/H
NGC3031 0.080 162 7.8 S1.5/L1 L/L/S log(M/M )=a +b log(s /s ) , (1)
⊙ ⋆ 0
NGC3169 0.031 163 7.8 L2 L/L/L
NGC3226 0.023 193 8.1 L1 L/L/L wherea =8.13±0.06,b =4.02±0.32,ands =200kms−1
NGC3379 0.024 207 8.2 L2/T2 L/L/L 0
(Tremaineetal. 2002). For the following galaxieswe adopt
NGC3507 0.024 ... ... L2 L/L/H
NGC3607 0.021 224 8.4 L2 L/L/S theblackholemassesmeasuredfromspatiallyresolvedstel-
NGC3608 0.021 192 8.1 L2/S2 S/L/S lar and/or gas kinematics: NGC 3031 (Boweretal. 2000;
NGC3628 0.027 171 7.9 T2 S/L/S
Devereuxetal. 2003), NGC 3998 (deFrancescoetal. 2006)
NGC3998 0.016 305 8.4 L1 L/L/S
NGC4111 0.015 148 7.6 L2 L/L/H NGC 4261 (Ferrareseetal. 1996), NGC 4374 (Boweretal.
NGC4125 0.019 227 8.4 T2 L/L/L 1998), NGC 4486 (Machetto 1997), and NGC 4594
NGC4143 0.013 214 8.3 L1 L/L/L (Kormendyetal. 1996). In the case of NGC 3031 separate
NGC4261 0.018 309 8.7 L2 L/L/L
measurements from stellar kinematics, gas kinematics, and
NGC4278 0.029 261 8.6 L1 L/L/L
NGC4314 0.025 117 7.2 L2 L/C/H the central stellar velocity dispersion give nearly identical
NGC4374 0.040 308 9.2 L2 L/L/L values (within 5%). In the case of NGC 4261, the black
NGC4438 0.028 ... ... L1 L/L/L hole mass from gas kinematics differs from that obtained
NGC4457 0.022 96 6.9 L2 L/L/H
fromthecentralstellarvelocitydispersionbyafactorofonly
NGC4486 0.022 333 9.5 L2 L/L/L
NGC4494 0.021 145 7.6 L2 L/L/L 1.6. In seven cases where the stellar velocity dispersion is
NGC4548 0.038 144 7.6 L2 L/L/L reported both in the Hypercat database and by Barthetal.
NGC4552 0.041 203 8.2 T2 L/L/L (2002), the black hole masses are within a factor 1.6 or less
NGC4579 0.041 165 7.8 L1 L/L/L
from each other. There are four large discrepancies, as fol-
NGC4594 0.051 241 9.0 L2 L/L/L
NGC4636 0.028 203 8.2 L1 L/L/L lows: in NGC 404 where the stellar velocity dispersions re-
NGC4736 0.018 112 7.1 L2 L/L/L ported in the Hypercat database and by Barthetal. (2002)
NGC5055 0.018 108 7.1 T2 L/L/H leadtoblackholemassesthatdifferbyafactorof3.6,while
NGC5866 0.013 159 7.7 T2 S/L/S
in NGC 3998, NGC 4486, and NGC 4594 mass determina-
NGC6500 0.090 214 8.3 L2 L/C/L
NGC7331 0.091 138 7.5 T2 S/L/S tionsfromthecentralstellarvelocitydispersionandspatially
resolved stellar or gas kinematics differ by factors between
aReddeningcausedbytheISMoftheMilkyWay;fromSchlegeletal.(1998). 2.7 and 3.5. Considering all cases with multiple determi-
bThestellarvelocitydispersion,takenfromtheHypercatdatabase,withthefollowing nations of the black hole mass, the values of D log(M/M⊙)
exceptions: NGC404,NGC4278,NGC4314,NGC4374,NGC4579,NGC4736, are evenly distributed about zero with a standard deviation
NGC5055,NGC6500fromBarthetal.(2002);NGC1097fromLewis&Eracleous of 0.34 (amounting to a factor of approximately 2). There
(2006).
cThelogoftheblackholemassinM⊙.Derivedfromstellarvelocitydispersionsusing arethreegalaxiesforwhichwe couldnotestimate the black
equation(1),withthefollowingexceptions:forNGC3031itwasderivedfromstellar hole masses because we could not find the necessary data:
andgaskinematics(Boweretal.2000;Devereuxetal.2003), whileforNGC4261 NGC 266, NGC 3507, and NGC 4438. The distribution of
andNGC4374itwasderivedfromgaskinematics(Ferrareseetal.1996;Boweretal.
black hole masses is shown in Figure 1d. Their values span
1998).
d SpectroscopicclassificationaccordingtoHoetal.(1997b), withtheexceptionof therange5.3<log(M/M⊙)<9.5.
NGC1097andNGC1553.TheLINERtypesforthesetwogalaxiesweretakenfrom
Phillipsetal.(1984)andPhillipsetal.(1986),respectively. L1=LINERwithabroad
Ha line,L2=LINERwithoutabroadHa line,T2=intermediateemissionlineratios 3. SPECTRALENERGYDISTRIBUTIONS
betweenLINERandHIIregion,S=Seyfert;combinationsindicateintermediateline 3.1. DataCompilation
ratiosbetweentwoclasses.
e Spectroscopic classification based on the criteria of Kewleyetal. (2001), InTable 3 we presentthe data makingup the SEDs of in-
Kauffmannetal.(2003),andKewleyetal.(2006).Thethreedesignationsrefertothe dividual galaxies in the form of monochromatic luminosity
lvosc[aOtioIInI]o/Hfthdeiaogbrajemcst,inretshpee[cOtivIe]/lHy.aL=vsL[IONEIIRI],/HS=¯,S[NeyIfIe]/rHt,aH=vHs[IOIrIeIIg]i/oHn¯,,aCn=d“[cSoImI]/pHoas- versusfrequency(i.e.,n Ln vsn ). Thesedataweretakenfrom
ite,”i.e.,int¯ermediatebetweenLINERsorSeyfertsandHIIregionsinthe[NII]/Ha vs sourcesintheliterature,aslistedinTable3.Weincludeinthis
[OIII]/Hdiagram. table monochromatic luminosities with and without correc-
¯
tions for extinction (see discussion of extinction corrections
below). As we noted in §2, we include primarily measure-
thoughextensive,wasnotexhaustive,withtheresultthatwe ments made at high spatial resolution (<1′′); in some cases
mayhaveoverlookedafewobjects.Itisnotclearwhetherthe observationsat lower spatial resolution are used, but the re-
relative number of LINER types in our sample should have sultingmeasurementsaretreatedasupperlimits. Upperlim-
an effect on our conclusions. If the properties of the AGN its fromnon-detectionsare also listed in this table. In cases
are related to the LINER class, then the composition of the where UV (2500 A˚) measurements or limits are not avail-
samplemayinfluenceanyaveragepropertiesderivedfromit. ablebutX-raymeasurementsorupperlimitsare,weinferred
Weattempttoassesssucheffectsattheendof§4.1usingour
estimatedbolometricluminosities. 6http://www-obs.univ-lyon1.fr/hypercat
SEDsofWeakAGNsinLINERs 5
UVupperlimitsfromthe2keVmonochromaticluminosities data) are listed in Table 4. Using these parameterswe have
andtheassumptionthattheoptical-to-X-rayspectralindexis calculatedthevaluesofn Ln at0.5,1,and10keV,whichwe
a <1.5.Wejustifythisassumptionin§4.1,below.Column listinTable3(thethreehighest-frequencyvaluesforeachob-
ox
(3),labeled“Observedn Ln ,”givesthemonochromaticlumi- ject).WenotethattheAGNsinLINERsaretypicallydetected
nosity obtained from the observed flux density without any at0.5and1keV,andoftenalsoat10keV,soitisfairtouse
corrections. Columns (4) and (5), labeled “Corrected n Ln ,” themodelstoevaluaten Ln attheseenergies. InFigure2we
givethemonochromaticluminosityafterminimum(following plottheX-raycomponentoftheSEDasathicksolidlinebe-
Calzettietal. 1994) and maximum (following Seaton 1979) tween 0.5 and 10 keV (already correctedfor extinction)and
extinctioncorrections. We discussthesecorrectionsindetail weextrapolateitto100keV,althoughonlyforreference.
in§3.2below. TheindividualSEDsareshowngraphicallyin In a number of cases, the X-ray emission from the AGN
Figure2, in a separate panelforeachgalaxy; the“Observed is not detected, but an upper limit is available from the ob-
n Ln ”isrepresentedbyfilledpoints,witharrowsdenotingup- servations. These limits are expressed in Table 4 as upper
per limits, while open points show n Ln after the maximum limitsonthenormalizationoftheX-rayspectrum,foranas-
extinctioncorrection. sumedvalueofaphotonindexofG =1.8.7 Inthecaseofone
With the exceptionof the X-ray data, measurementswere galaxy,NGC3379, the normalizationofthe X-rayspectrum
made in relatively narrow bands (D n /n <0.2), centered at a isnotanupperlimitbutitwasderivedundertheassumption
∼
specific frequency. Whenever multiple measurements were ofG =1.8,becauseofthelowS/NoftheX-rayspectrum.The
available at approximately the same frequency, these were monochromaticluminositiesoftheseobjectsgiveninTable3
averaged together and their fractional standard deviations, arealsoidentifiedasupperlimits.
s /(n Ln ), are given in a separate column of Table 3. These
3.2. ExtinctionCorrections
variationscanbecausedbyintrinsicvariabilityofthesource
(specifically in the near UV; cf, Maozetal. 2005) or differ- To facilitate extinction corrections, we have listed the
ences in spatial resolution. In the former case the temporal values of the color excess (or reddening), E(B −V),
fluctuationsareoftenlessthan10%ontimescalesofabouta associated with the Galactic ISM (taken from
year. However,Maozetal.(2005)havealsofoundsomeex- Schlegel,Finkbeiner,&Davis, 1998) in Table 2. The
amples of larger variations (up to 50%) on these short time model fits to the X-ray spectra yield a value for the total
scales. Moreimportantly,theyhavefoundthatontimescales equivalenthydrogencolumn density (N , listed in Table 4).
H
ofseveralyearstoadecadetheamplitudeofthefluctuations These models typically assume the photoelectric absorption
canreacha factorofa few. In the case ofnear-IRandradio cross-sections of Morrison&McCammon (1983), who
data,thefluctuationscanbeaslargeas∼60%. Wehavealso adopted the elemental abundances of Anders&Ebihara
encounteredan exampleof evenmore extremevariabilityin (1982). In Table 4, we also list the corresponding value of
theradioband,whichwediscussbrieflyin§3.3. the total color excess, derived from the following relation
TheX-raydataareintheformofbroad-bandspectra,typi- betweenthevisualextinction,A , and thehydrogencolumn
V
callyspanningtheenergyrange0.5–10keV.ManyoftheX- density(fromPredhl&Schmitt1995,applicabletotheMilky
rayspectracanbefittedwithsimplepower-lawmodels,mod- Way)
ified by interstellar photoelectric absorption in nearly neu- N /A =1.79×1021cm−21mag−1, (3)
H V
tral gas. In some cases, more complex models are needed,
consisting of a power law plus optically thin emission from andassumingthatRV≡AV/E(B−V)=3.2.Asexpected,in
a thermal plasma (see, for example, Eracleousetal. 2002; mostcases the total reddening,E(B−V)X correspondingto
Flohicetal. etal. 2006, and references therein). In such the hydrogen column determined from the X-ray spectra, is
cases, the power-law component is attributed to the AGN largerthanthereddeningcausedbytheGalacticISM.Intwo
and the thermal plasma component is ascribed to spatially- cases,NGC404andNGC3031,E(B−V)exceedsE(B−V)X
unresolved, circumnuclear emission. Thus, we have taken by 0.019 and 0.008 magnitudes respectively, which is well
thepower-lawcomponentfromsuchmodelstorepresentthe withintheuncertaintiesofthecolumndensitiesderivedfrom
emission from the AGN. In all cases, we adopt the power- X-ray observations. In Figure 3a, we show the distribution
lawmodelasaconvenientparameterizationofthedata. This ofvaluesofE(B−V)X oftheobjectsinoursampletoillus-
modeldescribesthe intrinsic photonenergyspectrum(num- tratethatE(B−V)X<0.6inabout31/33oftheobjectsand
ber of emitted photons per unit energy interval) as N(E)= E(B−V)X<0.2inabout22/33oftheobjects(weexclude2
N (E/E )−G , where E is a fiducial energy (typically E = objectswithextremelyhighextinction,inwhichnomeaning-
10kev),G0isknownast0he“photonindex,”andN =N(E0)is fulcorrectioncanbemade).ThevaluesofE(B−V)Xshould
0 0 beregardedwithcautionbecauseoftheassumptionsinvolved
the“normalizationconstant”(withunitsofcm−2s−1 keV−1).
inderivingthem.MostuncertainistheassumptionofaGalac-
With this convention, and keeping in mind that fn (E0) = ticgas-to-dustratio,whichisinherentintherelationbetween
hN E (wherehisPlanck’sconstant),wecanwritetheflux
0 0 thehydrogencolumndensityandthevisualextinctiongiven
densityspectrumas
inequation(3). Furthermore,thereisalsothepossibilitythat
E 1−G the lines of sight to the UV and X-ray sources do not pass
fn (E)= fn (E0) throughexactlythesamecolumnofabsorbingmaterial.
E
(cid:18) 0(cid:19) We applied extinction corrections to the observed
N E 1−G monochromatic luminosities as follows: Using the val-
=0.663 1cm−2s−01keV−1 1keV mJy. (2) ues of the total reddening, E(B−V)X, for each galaxy
(cid:18) (cid:19)(cid:18) (cid:19)
Theparametersdescribingthepower-lawX-rayspectrum(G 7 This value of G is the median value found in the large sample of
Flohicetal.etal. (2006). The derived limits are not very sensitive to the
and N0, for E0 =1 kev), as well as the equivalenthydrogen valueofG (seethediscussioninFlohicetal.etal.2006,andfootnote(a)in
column density (NH; inferred from fitting the model to the Table4).Forexample,changingG by±0.2changesthefluxbyonly∼10%.
6 Eracleous,Hwang,&Flohic
FIG.2.—Thespectralenergydistributionsofindividualgalaxies
inoursample. Thedatapointsrepresentmeasurementsinamod-
eratelynarrowband(D n /n <∼0.2),centeredataspecificfrequency.
Whenevermultiplemeasurementswereavailable atapproximately
thesamefrequency,thesewereaveragedtogetherandtheirstandard
deviation is represented as a vertical error bar (see discussion
in §3.1 of the text). The filled circles represent measurements
withoutextinctioncorrections, whiletheopencirclesarethesame
measurementsafterextinctioncorrections(seedetailsin§3.2ofthe
text). Thethicksolidlinesrepresentthepower-law componentof
the best-fitting model to the0.5–10 keV X-rayspectrum (already
correctedforextinctionandextrapolatedto100keV,forreference).
Inthe case ofNGC4261, wedo notplot the optical points after
extinctioncorrections;seethediscussionin§3.3ofthetext.
THE COMPLETE FIGURE WILL BE AVAILABLE IN
THEELECTRONICVERSIONOFTHEJOURNAL.
SEDsofWeakAGNsinLINERs 7
TABLE3
SPECTRALENERGYDISTRIBUTIONSOFINDIVIDUALLINERS
Observed Correctedn Ln (ergs−1)a
n n Ln Referencesc
Galaxy (Hz) (ergs−1) Calzetti(min) Seaton(max) s /hn Ln ib andNotes
(1) (2) (3) (4) (5) (6) (7)
NGC266 1.700×109 1.43×1037 ... ... 1 d
2.300×109 2.14×1037 ... ... 1 d
5.000×109 6.06×1037 ... ... 1 d
8.400×109 1.41×1038 ... ... 1 d
1.199×1015 <2.05×1040 <1.18×1041 <2.88×1041 e
1.210×1017 1.20×1040 ... ... 2
2.420×1017 1.81×1040 ... ... 2
2.420×1018 7.22×1040 ... ... 2
NGC404 1.500×1010 <1.34×1035 ... ... 3 f
4.286×1010 <2.95×1036 ... ... 4 f
1.149×1013 <1.80×1040 ... ... 5 g
3.686×1014 4.74×1039 4.74×1039 5.26×1039 6
9.085×1014 1.94×1039 2.06×1039 2.58×1039 7 h
1.199×1015 1.27×1039 1.39×1039 1.88×1039 7 h
1.322×1015 2.82×1039 3.12×1039 4.65×1039 8
1.210×1017 7.00×1036 ... ... 9
2.420×1017 4.95×1036 ... ... 9
2.420×1018 1.57×1036 ... ... 9
THECOMPLETETABLEWILLBEAVAILABLEINTHEELECTRONICVERSIONOFTHEJOURNAL
aThemonochromaticluminositiesfrom0.1to1m maftercorrectionforextinction.TheminimumcorrectionemploysthestarburstextinctionlawofCalzettietal.(1994),whilethe
maximumcorrectioncorrespondstotheMilkyWaylawofSeaton(1979). Detailsaregivenin§S:dataofthetext. The“observed”X-rayluminositiesalreadyhavethiscorrection
builtin.ThedistancesusedarethoseofTully(1988);seeTable1andthediscussionin§2.1ofthetext.
bThefractionaldispersioninthemonochromaticluminosityincaseswhereanumberofmeasurementswereaveragedtogether.See§3.1ofthetextfordetails.
cReferences. –(1)Doietal.(2005a);(2)Terashima&Wilson(2003);(3)Nagaretal,(2005);(4)Nagaretal.(2000);(5)Saryapaletal.(2004);(6)Chiabergeetal.(2005);(7)
Maozetal.(2005);(8)Maozetal.(1995);(9)Eracleousetal.(2002).
dSeedetailednotesonthisobjectin§3.3ofthetext.
eUpperlimitton Ln(2500A˚)derivedbyassuminga ox=1.5.See§4.1ofthetextfordetails.
fThislimitisaresultofanon-detection.
gThislimitisaresultofcontaminationofthesourcebyother,neighboringsources;theobservationsweretakenthroughalargeaperture.
hTheobservedUVfluxmostlikelyoriginatesinhotstarsintheimmediatevicinityofthenucleus,notintheAGN.Seethediscussionin§3.3ofthetext.Thevaluesofa oximplied
bytheUVfluxisextremelyhighforsuchalow-luminosityAGNthus,theUVfluxlistedherecanbetakenasagenerousupperlimittotheUVfluxoftheAGN.
8 Eracleous,Hwang,&Flohic
we computed the extinction corrections for points in
the SED between 0.1 and 1 m m using the following
five extinction laws: the Milky Way extinction laws
of Seaton (1979) and Cardelli,Clayton,&Mathis (1989),
the Large and Small Magellanic Cloud extinction laws
of Korneef&Code (1981) and Bouchetetal. (1985), re-
spectively, and the starburst galaxy extinction law of
Calzetti,Kinney,&Storchi-Bergmann(1994).
AlloftheMilkyWayandMagellanicCloudlawsagreewell
with each other in the near-IR, optical, and near-UV bands
butnotin the far UV (see, for example, Figures1 and 21in
Calzettietal.1994). Thedifferencesaremostpronouncedin
the far-UV band, at l <1500 A˚. In the near-UV the Milky
WayandLMCextinctioncurvesarenotmonotonic(because
of the “2200 A˚ bump”) with the result that the curves cross
each other. Thus, the law that gives the highest extinction
correction changes with wavelength and also with the value
ofE(B−V) . Toillustratetheeffectsofanddifferencesbe-
X
tween the above extinction laws, we plot in Figures 3b and
3c, respectively, the transmission factor (i.e., the fraction of
flux that is transmitted) at 2500 A˚ according to each of the
extinctionlawsandtherelationsbetweenthedifferenttrans-
missionfactors. ThestarburstextinctionlawofCalzettietal.
(1994)differsconsiderablyfromtheotherfourlaws. Consid-
ering the Milky Way and Magellanic Cloud extinctionlaws,
we find that for E(B−V) <0.2 (22/33 of our objects) the
X
transmission factors range between 0.29 and 0.32 and for
E(B−V) <0.5(31/33ofourobjects)thethetransmission
X
factorsrangebetween0.02and0.03.
Inthe endwe adoptedthe Seaton(1979) law, whichleads
to the largest corrections. We note, however, that if we had
adopted any other of the Milky Way of Magellanic Cloud
laws,thedifferenceinthecorrectedfluxwouldhavebeenof
order 10% or less. The resulting, corrected monochromatic
luminositiesare also includedin column(5) ofTable 3. For
reference,incolumn(4)ofTable3wealsolistthemonochro-
maticluminosityaftercorrectingwiththestarburstextinction
law of Calzetti,Kinney,&Storchi-Bergmann (1994), which
yieldstheminimumcorrectioncomparedtotheotherlaws.
We draw attention to the uncertaintiesinvolvedin our ex-
tinctioncorrections:themostimportantuncertaintyistheas-
sumptionoftheMilkyWaydust-to-gasratio,whichisincor-
poratedintoequation(3),andouradoptedvalueofR =3.2.
V
Wealsoemphasizethattheextinctioncorrectionsweadopted
aresuitableforourspecificapplication(assessingtheenergy
budgetsofLINERs;seepaperII),butmaynotbeappropriate
forotherapplications. Morespecifically,inpaperIIweesti-
matetheionizingluminositiesoftheweakAGNsinthissam-
ple in order to assess whether they can power the observed
emission lines. Therefore, we take the highest possible ex-
tinctioncorrectionso as to erron theside of caution, i.e., to
overestimatetheUVluminosityratherthanunderestimateit.
3.3. NotesonIndividualObjects FIG.3.—Reddeningvaluesandextinctioncorrectionsfor33ofthe35ob-
jectsinoursample(NGC3169andNGC4261areexcludedbecausethecol-
NGC266.– The radio properties of the AGN in NGC 266 umndensityissolargethattheyshouldbeinvisibleintheUV).(a)Thedistri-
have been studied by Doi,Kameno,&Inoue (2005a) butionofthevaluesofE(B−V)X,thereddeningderivedfromtheX-raycol-
umndensity(listedinTable4).Asthishistogramshows,31/33objectshave
usingobservationsatmultipleepochs. Theyfoundsig-
E(B−V)X<0.6and22/33objectshaveE(B−V)X<0.2.(b)Thefractionof
nificantvariabilityinboththeradioluminosityandthe transmittedfluxat2500A˚ forthefiveextinctionlawsthatwehaveexplored,
shape ofthe radiospectrum. Here we adopttheir data withlinestylesasfollows(toptobottom): dot-dashedfortheStarburstlaw
fromVLBAobservationstakenonasingleepoch,2003 ofCalzettietal.(1994),dashedfortheLMClawofKorneef&Code(1981),
dottedfortheLMClawofBouchetetal.(1985),andtriple-dot-dashed and
March8.
solidfortheMilkyWaylawsofCaredellietal.(1989)andSeaton(1979),
NGC404.– A UV spectrum of the nucleus of this galaxy respectively. (c) Comparison of extinction laws via the ratios of 2500 A˚
transmissionfactors. The“Seaton/Calzetti”ratiorepresentsthefullrangeof
shows prominent absorption features from hot stars
possibleextinctioncorrections,whilethe“Seaton/Korneef”ratiorepresents
therangeofofpossiblenon-starburstcorrections.
SEDsofWeakAGNsinLINERs 9
TABLE4
X-RAYSPECTRALPARAMETERSANDDERIVEDPROPERTIES
X-RaySpectralParameters
Galaxy N0 NH E(B−V)Xb L2−10keV Lbolc
(NGC) G a (cm−2s−1) (cm−2) (mag) (ergs−1) (ergs−1) REdd a oxd
(1) (2) (3) (4) (5) (6) (7) (8) (9)
0266 1.40 5.223×10−6 1.5×1021 0.262 7.4×1040 2.2×1042 ... ...
0404 2.50 4.490×10−6 2.3×1020 0.040 3.9×1036 1.2×1038 4×10−6 ...
1097 1.64 3.864×10−4 2.3×1020 0.040 4.3×1040 8.5×1041 5×10−5 1.14
1553 1.20 4.438×10−5 3.2×1021 0.559 8.7×1039 4.4×1041 4×10−5 ...
2681 2.00 4.990×10−6 2.7×1021 0.471 1.8×1038 9.0×1039 6×10−6 ...
3031 1.88 1.775×10−3 4.1×1020 0.072 1.9×1040 2.1×1041 3×10−5 1.04
3169 2.60 5.138×10−3 1.12×1023 19.55 1.1×1041 3.3×1042 4×10−4 ...
3226 2.21 4.440×10−4 9.3×1021 1.624 5.0×1040 1.5×1042 1×10−4 ...
3379 1.80e 6.200×10−7 2.75×1020 0.048 1.7×1037 5.1×1038 3×10−8 ...
3507 1.80f <2.400×10−7 1.63×1020 0.029 <3.9×1037 <1.2×1039 ... ...
3607 1.80f <3.000×10−7 1.48×1020 0.026 <5.0×1037 <1.5×1039 <5×10−8 >0.55
3608 1.80f <2.600×10−6 1.49×1020 0.026 <5.9×1038 <1.8×1040 <1×10−6 ...
3628 1.80f <2.000×10−7 2.23×1020 0.039 <4.9×1036 <1.5×1038 <2×10−8 ...
3998 1.88 3.611×10−3 5.82×1020 0.102 2.6×1041 1.4×1043 4×10−4 1.05
4111 1.80f <3.010×10−5 1.40×1020 0.024 <3.6×1039 <1.1×1041 <2×10−5 >1.00
4125 1.80f <2.200×10−6 1.84×1020 0.032 <5.4×1038 <1.6×1040 <6×10−7 ...
4143 1.66 7.331×10−5 1.5×1020 0.026 1.1×1040 3.2×1041 1×10−5 ...
4261 1.56 1.900×10−4 5.0×1022 8.73 1.0×1041 6.8×1041 1×10−5 ...
4278 1.64 1.800×10−4 3.5×1020 0.061 9.1×1039 2.7×1041 5×10−6 ...
4314 2.10 <1.130×10−4 1.78×1020 0.031 <3.1×1037 <9.2×1038 <5×10−7 ...
4374 2.10 3.762×10−5 1.9×1021 0.332 3.5×1039 5.0×1041 3×10−6 0.99
4438 1.80f <9.999×10−6 1.20×1021 0.210 <1.2×1039 <3.5×1040 ... >0.88
4457 1.70 6.815×10−6 9.8×1020 0.171 1.0×1039 3.0×1040 3×10−5 ...
4486 2.17 2.310×10−4 6.1×1020 0.107 1.6×1040 9.8×1041 2×10−6 1.33
4494 1.80 2.352×10−5 3.0×1020 0.052 9.2×1038 2.8×1040 6×10−6 ...
4548 1.70 4.432×10−5 1.6×1022 2.793 5.4×1039 1.6×1041 3×10−5 ...
4552 2.00 6.700×10−6 6.0×1020 0.105 2.6×1039 7.8×1040 4×10−6 1.03
4579 1.50 2.116×10−5 2.54×1020 0.044 1.8×1041 1.0×1042 1×10−4 0.92
4594 1.89 2.450×10−4 1.9×1021 0.332 7.5×1039 4.8×1041 4×10−6 1.23
4636 1.80f <5.100×10−6 1.81×1020 0.032 <6.1×1038 <1.8×1040 <1×10−6 >1.14
4736 1.60 5.600×10−5 3.3×1020 0.058 5.9×1038 1.8×1040 1×10−5 1.36
5055 1.80 2.843×10−5 5.0×1021 0.873 2.0×1038 5.9×1039 4×10−6 ...
5866 1.80f <3.199×10−6 1.47×1020 0.024 <3.1×1038 <9.4×1039 <1×10−6 >0.81
6500 3.10 5.178×10−5 2.1×1021 0.367 5.3×1039 1.6×1041 7×10−6 ...
7331 1.80f <4.001×10−7 8.61×1020 0.150 <3.4×1037 <1.0×1039 <3×10−7 >1.33
aThetypicaluncertaintyinG is±0.2–0.3. ThefractionaluncertaintyintheX-rayfluxisdominatedbytheuncertaintyinG andisgivenbyd fX/fX=0.69d G . Thus,fractional
uncertaintiesintheX-rayfluxareoforder10–20%.
bThereddeningcorrespondingtotheequivalenthydrogencolumndensityreportedinthistable,obtainedusingequation(3)in§3.1ofthetext.
cThebolometricluminosityofNGC3031,NGC3998,NGC4261,NGC4374,NGC4486,NGC4579,andNGC4594wasdeterminedbyintegratingtheS.E.D.Thebolometric
luminosityofNGC1097wasderivedfromamodelfittotheSED(see§3.3).Forallothergalaxiesthebolometricluminositywasobtainedbyscalingthe2–10keVX-rayluminosity
asdiscussedin§4.1ofthetext.
dTheoptical-to-X-rayspectralindex,definedin§4.1ofthetext.ThetabulatedvalueswerecomputedaftercorrectingtheUVfluxaccordingtotheSeaton(1979)lawandusingthe
veaBlueecsaoufseEo(fBt−heVl)oXwfSro/NmocfoltuhmenX(-5ra)yofspthecistrtuambl,eT(mheaxXi-mraaylcpohroretocntioinnd).exTywpaicsaalsusnucmeertdaitnotibeesGon=a 1ox.8arfeor>∼th0e.0p3uarpnodsceaonfbdeearisvhinigghaavsa0lu.0e9fo(sretehtehneodrimscauliszsaiotinonin.§T4h.e1noofrtmhealtiezxatt)i.on
itselfandresultingX-rayfluxarenotupperlimits,butaresubjecttotheassumedvalueofG .
fTheX-rayphotonindexwasassumedtobeG =1.8forthepurposeofderivinganupperlimittothenormalizationandtheX-rayflux.
(Maozetal. 1998), indicating that they make a sig- thosefoundinthespectraofhotstars(Maozetal.2005;
nificant contribution to the light at these wavelengths. Maoz 2007). All of these properties suggest that the
The X-rayspectrumis verysoft(see the discussionin nucleus of NGC 404 harbors a compact star-forming
Eracleousetal. 2002), with G =2.5, which is unchar- regionaswellasalow-luminosityAGN.Therefore,we
acteristic of AGNs (cf, Nandraetal. 1997), although includethisobjectinoursample.
an AGN cannot be ruled out based on this observa-
tion. The nuclear source is resolved at both UV and NGC1097.– The SED of the AGN in NGC 1097 and
X-raywavelengthsandithasa“blow-out”morphology modelfitstoitarediscussedbyNemmenetal.(2006).
(Maozetal.1995;Eracleousetal.2002). Ontheother Here, we adopted a subset of the data included
hand,thecompactnucleusoftheUVsourceappearsto in that paper. The IR measurements presented by
bevariablebyafactorofapproximatelytwoonatime Prietoetal. (2005) were taken through a very small
scaleofapproximatelyadecade.Inaddition,thedepths aperture, which isolates the nucleus. However, the
oftheabsorptionlinesappeartobeshallowerthanthe AGN is embedded in an compact, unresolved star-
10 Eracleous,Hwang,&Flohic
burst(Storchi-Bergmannetal.2005)whichmaydomi- the AGN. Therefore, we do not apply any extinction
nate the emission at these wavelengths, therefore, we correctionstothesemeasurementsinTable3nortothe
have designated these measurements as upper limits pointsplottedinFigure2.
to the flux of the AGN. We excluded measurements
taken through large apertures since these were signif- NGC5055.– ProminentabsorptionlinesintheUVspectrum
icantly contaminated by emission from the circumnu- of the nucleus of NGC 5055 suggest that hot stars
clear starburst ring. A significant fraction of the nu- dominate the light (Maozetal. 1998). This conclu-
clearUVfluxappearstooriginatefromacompactstar- sion is supported by the lack of significant UV vari-
burst, as suggested by absorption lines from hot stars abilityandtheextendedmorphologyoftheUVsource
detectedintheHSTspectrum(Storchi-Bergmannetal. (Maozetal. 2005). Although an X-ray source with
2005; Nemmenetal. 2006). By following the best- an AGN-like spectrum is detected (Flohicetal. etal.
fitting SED modelof (Nemmenetal. 2006), we adopt 2006), the equivalent hydrogen column density mea-
onlythefractionoftheUVfluxthatisattributedtothe sured from the X-ray spectrum implies E(B−V) =
AGN.Thismodelincludescontributionsfromaninner, 0.87, which translates into an attenuation by a factor
radiativelyinefficientaccretionflow,andouter,geomet- of100at2500A˚.OurconclusionisthatanAGNmay
rically thin accretiondisk, an obscuredstarburstanda be present in this galaxy, but is not the source of the
jet. The contribution of the jet is appreciable only at observedUVflux.
the lowest radio frequencies, the starburst contributes
NGC6500.– TheUVspectrumofthenucleusofNGC6500
primarily to the near-UV band, the inner, hot accre-
(Maozetal. 1998) has a relatively low S/N, but still
tionflowdominatesinthefar-IRandX-raybands,and
shows absorption features resembling lines from hot
the thin accretion disk dominatesin the near-IR band.
stars. Moreover, the nuclear UV source is extended,
In the same spirit, we have also adopted a bolometric
with no clear “knot” that could be identified with
luminosity of 8.5×1041 ergs−1 from Nemmenetal.
the nucleus (Maozetal. 1995; Barthetal. 1998) and
(2006). In comparison, if we integrate the tabulated
no variability (Maozetal. 2005). The X-ray spec-
SED,weobtainaluminosityof5.1×1041ergs−1.
trum(Terashima&Wilson2003)isindicativeofanob-
NGC3998.– To construct the SED of NGC 3998 we be- scuredAGNwithE(B−V)=0.37,implyinganatten-
gan from the extensive data tabulation of Ptaketal. uation of its 2500 A˚ flux by a factor of 7. Thus this
(2004). We excluded many of the measurements pre- object is very similar to NGC 5055; an AGN is prob-
sentedthereinbecausetheywereobtainedthroughex- ablypresent, butis notthe sourceoftheobservedUV
tremely large apertures that encompass a substantial flux.
fraction of the host galaxy (e.g., from IRAS observa-
tions). Since the AGN in NGC 3998 is rather bright 4. QUANTITIESDERIVEDFROMTHESPECTRAL
comparedtootherobjectsinourcollection,weadopted ENERGYDISTRIBUTIONS
measurements through apertures as large as 3′′ as fair 4.1. Optical-to-X-RaySpectralIndices,Bolometric
measurementsof the AGN luminosity. Measurements Luminosities,andEddingtonRatios
through apertures between 3′′ and 15′′ were taken as
For comparison with other types of AGNs, we have used
upperlimitstotheAGNluminosity.
the rest-frame flux densities at 2500 A˚ and 2 keV to com-
NGC4261.– TheX-rayspectrumoftheAGNinNGC4261 pute the “optical-to-X-ray spectral index” (Tananbaumetal.
has been measured recently by both Chandra 1979),definedas
(Zezasetal. 2005) and XMM-Newton (Gliozzietal.
2003). Bothobservationsyieldthesamespectralindex a ≡−log Ln (2500A˚)/Ln (2keV)
and flux but significantly different equivalent hydro- ox log n (2500A˚)/n (2keV)
(cid:2) (cid:3)
gen column densities (the Chandra spectrum yields
NH =3.7×1020 cm−2, while the XMM-Newton spec- =1+0.38(cid:2)4log (n Ln )2500A˚ ,(cid:3) (4)
trumyieldsNH=5.0×1022cm−2). AUVobservation " (n Ln )2keV #
with the HST, reported by Zirbel&Baum (1998),
yieldsonlyanupperlimitofn Ln <4.9×1038 ergs−1 under the conventionthat Ln (cid:181) n −a ox. Both the UV and X-
at 2300 A˚, which produces a very large dip in the rayflux densitieswerecorrectedforextinction,as described
SED (see Fig. 3 of Ho 1999). Neither of the column in§3.2, sothata ox describesthe shapeoftheintrinsicAGN
densities measured from the X-ray spectra produces spectral energy distribution. We were able to calculate the
a reasonable extinction correction of the UV limit; valuesofa forthe9LINERswithdataavailableat2500A˚
ox
the lower value produces a negligible correction, and2keV.Another6objectshaveavailabledataat2500A˚ but
while the higher value moves the upper limit many only upperlimits at 2 keV; for these we were able to obtain
orders of magnitude above the monochromatic X-ray lowerlimitstoa . Forthreeobjects,NGC404,NGC6500,
ox
luminosity at 0.5 keV. Therefore, we infer that the andNGC5055,wedonotpresentthevalueofa becausewe
ox
AGN is indeed completely extinguished in the UV feelitis unreliable(see the discussion of these three objects
by the large equivalent hydrogen column measured in §3.3). In the case of NGC 404, although an AGN may
from the XMM-Newton spectrum. Thus we derive contributetoobservedUVandX-rayflux,thereissignificant
an upper limit to the near-UV flux by assuming that contaminationbyhotstars,asevidencedbystellarabsorption
a <1.5, which we include in Table 3, and plot in lines in the UV spectrum. In the other two objects, the UV
ox
Figure2. Similarly,wedoubtthatthemeasurementsof light that we observe appears to be dominated by hot stars,
the nucleus of NGC 4261 in the optical band capture withnodiscerniblecontributionfromanAGN.