A comparative analysis of virial black-hole mass estimates of moderate-luminosity active galactic nuclei using Subaru/FMOS

0.7 MB·English

by  Kenta Matsuoka

DRAFTVERSIONJANUARY14,2013
PreprinttypesetusingLATEXstyleemulateapjv.02/07/07
ACOMPARATIVEANALYSISOFVIRIALBLACK-HOLEMASSESTIMATESOFMODERATE-LUMINOSITYACTIVE
GALACTICNUCLEIUSINGSUBARU/FMOS
K.MATSUOKA1,2,J.D.SILVERMAN3,M.SCHRAMM3,C.L.STEINHARDT3,T.NAGAO4,5,J.KARTALTEPE6,D.B.SANDERS7,
E.TREISTER8,G.HASINGER7,M.AKIYAMA9,K.OHTA5,Y.UEDA5,A.BONGIORNO10,11,W.N.BRANDT12,M.BRUSA13,14,15,
P.CAPAK16,F.CIVANO17,A.COMASTRI15,M.ELVIS17,L.C.HO18,S.J.LILLY19,V.MAINIERI20,D.MASTERS18,M.MIGNOLI15,
M.SALVATO13,J.R.TRUMP21,Y.TANIGUCHI2,G.ZAMORANI15,D.M.ALEXANDER22,ANDY.T.LIN23
DraftversionJanuary14,2013
3
1 ABSTRACT
0 We present an analysis of broad emission lines observed in moderate-luminosity active galactic nuclei
2
(AGNs), typical of those found in X-ray surveys of deep fields, with the aim to test the validity of single-
n epochvirialblackholemassestimates. Wehaveacquirednear-infrared(NIR;1.0- 1.8µm)spectraofAGNs
a uptoz∼1.8intheCOSMOSandExtendedChandraDeepField-SouthSurvey,withtheFiberMulti-Object
J Spectrograph(FMOS)mountedontheSubaruTelescope.Theselow-resolution(R∼600)NIRspectraprovide

0 asignificantdetectionofthebroadHαemissionlinethathasbeenshowntobeareliableprobeofblackhole
1 mass at low redshift. Our sample has existing optical spectroscopy (through programs such as zCOSMOS)
which providesa detection of MgII, a broad emission line typically used for black hole mass estimation at

] z&1. We carryoutaspectral-linefittingprocedureusingbothHαandMgIItodeterminethevirialvelocity
O
of gas in the broad line region, the monochromatic continuum luminosity at 3000 Å, and the total Hα line
C luminosity.Withasampleof43AGNsspanningarangeoftwodecadesinluminosity(i.e.,λLλ∼1044- 46ergs
. s- 1), wefind atightcorrelationbetweentherest-frameultravioletandemission-lineluminositywith a distri-
h
p butioncharacterizedby hlog(λL3000/LHα)i=1.52andadispersionσ=0.16. Thereisalsoacloseone-to-one
- relationshipbetweentheFWHMofHαandofMgIIupto10000kms- 1 withadispersionof0.14inthedis-
o tributionofthelogarithmoftheirratios. Bothofthesethenleadtotherebeingverygoodagreementbetween
tr Hα-andMgII-basedmassesoverawiderangeinblackholemass(i.e.,MBH∼107- 9M⊙). Wedofindasmall
s offsetinMgII-basedmasses,relativetothosebasedonHα,of+0.17dexandadispersionσ=0.32.Ingeneral,
a
theseresultsdemonstratethatlocalscalingrelations,usingMgIIorHα, areapplicableforAGNatmoderate
[
luminositiesanduptoz∼2.

1
Subjectheadings:blackholephysics—galaxies:active—quasars:emissionlines
v
2
3 1. INTRODUCTION
3 Electronicaddress:[email protected]
2 1DepartmentofPhysicsandAstronomy,SeoulNationalUniversity,599 While the study of galaxy evolution has made important
. Gwanak-ro,Gwanak-gu,Seoul151-742,RepublicofKorea stridesinrecentyearsbybeingabletoweighindividualgalax-
1 2ResearchCenterforSpaceandCosmicEvolution,EhimeUniversity,2-5 ies (i.e., determine a mass), the field of quasar research is
0 Bunkyo-cho,Matsuyama790-8577,Japan grappling with the issue of how to measure accurately the
3 3KavliInstituteforthePhysicsandMathematicsoftheUniverse(Kavli
masses of a supermassive black holes (SMBHs) for the dis-
1 IPMU),TheUniversity ofTokyo,5-1-5Kashiwanoha, Kashiwa277-8583,
tant quasar population. Here the challenge is greater due to
: Japan
v 4TheHakubiCenterforAdvancedResearch,KyotoUniversity,Yoshida- the fact thatthe sphere of influenceof a SMBH can onlybe
i Ushinomiya-cho,Sakyo-ku,Kyoto606-8302,Japan resolvedforalimitedsampleofnearbygalaxieswhereasthe
X
5Department of Astronomy, Kyoto University, Kitashirakawa-Oiwake- dynamicalmassofagalaxycaneasilybemeasuredduetoits
r cho,Sakyo-ku,Kyoto606-8502,Japan
large spatial extent. A significant leap forward in our abil-
a 6NationalOpticalAstronomyObservatory,950NorthCherryAveue,Tuc-
ity to both accurately and efficiently measure the masses of
son,AZ85719,USA
7Institute forAstronomy, University ofHawaii, 2680WoodlawnDrive, SMBHs,MBH,atallredshiftswilllikelyleadtonewinsights
Honolulu,HI96822,USA onquestionssuchashowareblackholesfueled,whatitisthe
8DepartamentodeAstronomía,UniversidaddeConcepción,Casilla160-
C,Concepción,Chile
9Astronomical Institute, Tohoku University, 6-3 Aramaki, Aoba-ku, 1200EastCaliforniaBoulevard,Pasadena,CA91125,USA
Sendai980-8578,Japan 17Harvard-SmithsonianCenterforAstrophysics,60GardenStreet,Cam-
10Max-Planck-InstitutfürextraterrestrischePhysik,Giessenbachstrasse1, bridge,MA02138,USA
85748GarchingbeiMünchen,Germany 18TheObservatories ofthe Carnegie Institution forScience, 813Santa
11INAF–OsservatorioAstronomicodiRoma,ViadiFrascati33,00040 BarbaraStreet,Pasadena,CA91101,USA
MontePorzioCatone,Italy 19InstituteforAstronomy,ETHZürich,Wolfgang-Pauli-Strasse27,8093
12Department of Astronomy and Astrophysics, ThePennsylvania State Zürich,Switzerland
20EuropeanSouthernObservatory,Karl-Schwarzschild-Strasse 2,85748
University,UniversityPark,PA16802,USA
GarchingbeiMünchen,Germany
13Max-Planck-InstitutfürextraterrestrischePhysik,Postfach1312,84571
21University of California Observatories/Lick Observatory and Depart-
GarchingbeiMünchen,Germany
mentofAstronomyandAstrophysics,UniversityofCalifornia,SantaCruz,
14DipartimentodiFisicaeAstronomia,UniversitádegliStudidiBologna,
CA95064,USA
vialeBertiPichat6/2,40127Bologna,Italy 22DepartmentofPhysics,DurhamUniversity,SouthRoad,DurhamDH1
15INAF–OsservatorioAstronomicodiBologna, ViaRanzani1,40127 3LE,UK
Bologna,Italy 23Institute of Astronomy and Astrophysics, Academia Sinica, Taipei
16NASA/JPLSpitzerSciencecenter, California InstituteofTechnology, 10617,Taiwan
2 MATSUOKAETAL.
connectionwith its host galaxy, and how do SMBHs evolve the broad-line region is not so simple (Wangetal. 2009;
withinacosmologicalframework. Trakhtenbrot&Netzer 2012), especially for the most lumi-
Spectroscopyenablesustoprobethekinematicsofionized nous quasars (Marzianietal. 2009; Steinhardt&Silverman
gas within the vicinity of a SMBH in distant active galac- 2011;Marzianietal.2013)thatcanhavesignificantoutflows
tic nuclei (AGNs) and luminous quasars to infer their black possiblyinresponsetoamoreintenseradiationfield.Further-
holemasses. Traditionally,emissionlines(e.g.,CIV, MgII, more,itisnon-trivialtodisentanglethebroadMgIIlinefrom
Hβ, and Hα) detected in the optical and velocity-broadened Fe emission that sits at its base, especially for lower mass
between 2000- 20000 km s- 1 are used to probe the gravi- blackholesandmaybeeventhoseatthehighmassend.
tational potential well of a SMBH. This lower limit on the We present first results of a near-infrared spectroscopic
velocity width has been set somewhat arbitrarily since there survey of broad-line AGNs (BLAGNs) primarily in the
existsa well-knownpopulationof bothtype1 AGNshaving COSMOS field using the Fiber Multi-Object Spectrograph
narrowerlinewidths(i.e.,NLS1;Osterbrock&Pogge1987) (FMOS) mounted on the Subaru Telescope. With FMOS,
and those with intermediate-massblackholes(Greene&Ho we now have the capability to simultaneously acquire near-
2004,2007). infraredspectraof ∼200targetsoverafieldofviewof0.19
There are methods to determine the luminosity-weighted square degrees. In this study, we report on the comparison
radialdistancebetweenthebroad-lineregion(BLR)andcen- between the Hα emission line profile, detected in the near-
tral source, RBLR, for AGN (z.0.4) through reverberation- infrared, with that of Mg II present in previously available
mappingcampaigns(e.g.,Bahcalletal.1972;Peterson1993; optical spectra. We aim to establish how effective recipes
Bentzetal.2009)basedonBalmerlines. Evenwiththecom- (established locally) are to measure single-epoch black hole
plex nature of the BLR, this characteristic radius is tightly masses outto z∼1.8 and forBLAGNs of lower luminosity
correlatedwith its luminosity(Kaspietal. 2000, 2005), thus (i.e., lower black hole mass), as comparedto those foundin
providing a means to infer such a distance to the BLR in theSDSS.OursampleissupplementedwithAGNsfromthe
large quasarsamples based solely on luminosity. Then cou- ExtendedChandra Deep Field-South (ECDF-S) Survey that
pled with velocity information provides a viral mass esti- reach fainterdepths, in X-rays, than the Chandra/COSMOS
mate based on a single-epoch spectrum. Such techniques surveyandimproveourstatisticsatlowerblackholemasses.
have been applied to large quasar samples most notably In Section 2, we fully describe the FMOS observations in-
the Sloan Digital Sky Survey (SDSS). A number of studies cludingtargetselection,datareduction,andsuccessratewith
(e.g., Vestergaard&Osmer 2009; Steinhardt&Elvis 2010; respect to detecting the Hα emission line. We describe our
Shenetal.2011)clearlydemonstratethatsuchsampleseffec- methodforfittingbroademissionlinesinSection3. Ourre-
tivelyprobeSMBHsaboveMBH&109M⊙atz∼2(anepoch sults are described in Section 4. Throughout this work, we
ofmaximalblackholeactivity)duetothewideareacoverage assumeH0=70kms- 1 Mpc- 1, ΩΛ=0.7,ΩM =0.3,andAB
andshallowdepth. Itisimportanttokeepinmindthatthese magnitudes.
black hole masses are based on calibrations using lower lu-
2. NEAR-INFRAREDSPECTROSCOPYWITH
minosityAGNsatlowredshift;theirapplicationtoluminous
SUBARU/FMOS
quasarsathighredshiftis notwellsolidifiedwith reverbera-
tionmapping(Kaspietal.2007). The capability of FMOS (Kimuraetal. 2010) to simulta-
Deep surveys, such as COSMOS, GOODS and AEGIS, neously acquire near-infrared spectra for a large number of
are effectiveprobesof blackholeaccretionat lowermasses. objectsovera wide field offers greatpotentialfor studies of
Giventhattheblackholemassfunctionissteeplydecliningat galaxies(Yabeetal.2012;Roseboometal.2012)andAGNs
logM &9(Shen&Kelly2012),studiesoftheglobalpop- (Nobutaetal.2012)athighredshift.Overacircularregionof
BH
ulationwithSDSSaresusceptibletolargeuncertaintieswhen 30′ in diameter, it is possible to place up to 400fibers, each
extrapolatedtolowermasses(Kelly&Shen2012).Whileno- witha1.′′2aperture,acrossthefield. Todetectemissionlines
ble attempts have been made to characterize the low-mass inAGNsoverawiderangeinredshift,weelecttousethelow-
end (logM . 9; Tamuraetal. 2006; Hopkinsetal. 2007; resolutionmodethateffectivelycoverstwowavelengthinter-
BH
Merloni&Heinz2008),thesestudieshavebeenbasedonlu- valsof1.05- 1.34µm(Jband)and1.43- 1.77µm(Hband)
minosity and do not consider virial velocities. These deep simultaneously. Thespectralresolutionisλ/∆λ≈600,thus
survey fields have considerable X-ray coverage that can be a velocity resolution FWHM ∼500 km s- 1 at λ=1.5 µm,
utilizedto selectAGNsthatmitigatebiasesincurredby host suitable for the study of broad emission lines of AGN. Un-
galaxydilutionandobscuration. Suchselectionthenhasthe fortunately,this mode requiresan additionalopticalelement
potentialtoeffectivelyprobethelowerluminosityAGNpopu- (i.e.,VPHgrating)thatreducesthetotalthroughputto ∼4%
lationthatmaybepoweredbylowermassblackholesorthose at 1.3 µm and impactsthe limiting depth reachablein a few
accreting at sub-Eddington rates. Followup optical spectro- hoursofintegrationtime.
scopic observations are enabling single-epoch virial black Accurate removal of the bright sky backgroundwhen ob-
hole mass estimates (down to 107M⊙) based on the proper- servingfromthegroundwithnear-infrareddetectorsremains
ties of their broad emission lines and continuum luminosity the primary challenge. With FMOS, an OH-airglow sup-
(Trumpetal.2009;Merlonietal.2010). pression filter (Maiharaetal. 1993; Iwamuroetal. 2001) is
MassestimatesforthesehigherredshiftAGNs,whichactu- built into the system that significantly reduces the intensity
allyconstitutethemajorityofthepopulationindeepsurveys of strong atmospheric emission lines that usually plague the
such as COSMOS, rely on MgII or CIV since the Hβ line JbandandHband. Equallyimportant,thereisthecapability
(usedtocalibraterecipesbasedonlocalsamples)isnolonger (cross-beam-switching; CBS hereafter) to dither targets be-
availableintheopticalwindowatz&0.8. Whiletheassump- tween fiber pairs effectively measuring the sky background
tion that the MgII line is produced from the same physical spatially close to individual objects and through the same
region as Hβ may hold (McLure&Jarvis 2002; Shenetal. fibersasthesciencetargets. Inthismode,twosequentialob-
2008), there are studies that indicate that the physics of servationsaretakenbyoffsettingthetelescopeby60′′ while
BHMASSESTIMATESOFMODERATE-LUMINOSITYAGNS 3
keepingthetargetwithinoneofthetwofibers. Thetradeoff
is that only 200 fibers are available in this CBS mode. We
referthe readerto Kimuraetal.(2010) forfulldetailsofthe
instrumentanditsperformance.
Below, we briefly describe our observing program using
FMOS including the selection of type 1 AGNs, data reduc-
tion, and success rate with respect to the detection of broad
emissionlines. Completedetailsofourprogramwillbepre-
sentedinSilvermanetal.(inprep.)alongwiththefullcatalog
of emission-line propertiesof broad-lineAGNs in the COS-
MOSandECDF-S.
2.1. Targetselection
Our primary selection of AGNs is based on their hav-
ing X-ray emission detected by Chandra (Elvisetal. 2009;
Civanoetal.2012)withinthecentralsquaredegreeofCOS-
MOS (hereafter C-COSMOS). The high surface density of
AGNs(∼2000perdeg2)atthelimitingdepths(f0.5- 2.0keV>
2×10- 16ergscm- 2s- 1)ofC-COSMOSfieldensuresthatwe
makeadequateuseofthemultiplexcapabilitiesofFMOS.In
addition,twoFMOSpointingswereobservedfurtheroutfrom
thecenteroftheCOSMOSfieldthuswerelieduponthecata-
logofopticalandnear-infraredcounterpartsto XMM-Newton
sources(Brusaetal.2010)forAGNselection.
We further require that optical spectroscopy (Trumpetal.
2007;Lillyetal.2009)isavailableforeachsourcethatyields
a reliable redshift and detection of at least a single broad
(FWHM>2000kms- 1)emissionline,namelyMgIIinmany
cases. Wethenspecificallytargetedthosewithspectroscopic
redshiftsthatallowsustodetecteitherHβ(1.2<z<1.7and
1.9<z<2.6),Hα(0.6<z<1.0and1.2<z<1.7),orMgII
(2.8<z<3.8and4.1<z<5.3)intheobservedFMOSspec-
tralwindowsinlow-resolutionmode,i.e.,1.05- 1.34µmand
1.43- 1.77µm. Fibers are assigned to BLAGNs (forwhich FIG.1.—SelectionandcompletenessoftheCOSMOSAGNswithFMOS
wecandetectemissionlinesofinterest)withalimitingmag- spectra. Each panel shows the distribution of the observed sample (open
histogram)andthosethatyieldblackholemassestimatesbasedontheHα
nitudeofJAB=23. ThoseatJAB<21.5aregivenhigherpri- emissionline(filledblackhistogram)for(a)X-ray0.5- 10.0keVfluxand
oritytoensurethatthissample(oflowerdensityonthesky) (b)J-bandmagnitude.
is well represented in the final catalog; this also effectively
improvesuponoursuccess rate ofdetectionbothcontinuum of Hawaii in S10B- S11A. Weather conditions were accept-
and line emission. Due to the sensitivity of FMOS and the able although clouds, mainly cirrus, reduced our observing
low numberdensity of AGNs at z>3, our sample has very efficiency. The typical seeing was ∼ 1′′ with considerable
fewdetectionsofMgIIinthenearinfrared. variationacrossthenights.
In addition, we have acquired FMOS observations of X- WeelectedtousetheCBSmodewhiletakingtwosequen-
ray selected AGNs in the ECDF-S (Lehmeretal. 2005) sur- tial exposureseach of 15 minutes for each position (namely
vey with the inclusion of those that are only detected in the A and B positions hereafter). These pairs of exposure were
deeper2- 4Msdata (Luoetal. 2010; Xueetal. 2011) in the repeatedmultipletimesinordertoreachaneffectivetotalin-
central region that covers GOODS. This deeper X-ray field tegrationtimeof2.0- 3.5hourson-source. Sometimeislost
offersthepotentialtoextendthedynamicrangeofourstudy torefocusingandrepositioningfibersatregularintervalsdur-
intermsofblackholemassandEddingtonratio. We specif- ing the fullobservation. In the early data, only onespectro-
ically select AGNs, as mentioned above, based on their op- graph(IRS1) was available thus∼100fibers were available
tical propertiesdeterminedthroughdeepspectroscopiccam- forsciencetargets.
paigns(Szokolyetal. 2004; Silvermanetal. 2010). As with
2.3. DataReduction
theCOSMOSsample,weplacehigherpriorityonthebrighter
AGNs(R <22)whiletargetingthefaintercaseswithlower WeusethepubliclyavailablesoftwareFIBRE-pac(FMOS
AB
priority. Image-BasedREductionpackage;Iwamuroetal.2012). The
reductionroutinesare basedon IRAFtasks althoughseveral
2.2. Observations
stepsareprocessedbyadditionaltoolswrittenbytheFMOS
Wehaveacquirednear-infraredspectrawithSubaru/FMOS instrument team. Since our observation are carried out us-
of broad-line AGNs from the COSMOS and ECDF-S sur- inganABABnoddingpatterninCBSmodeofthetelescope,
veys. Themajorityofthedatawasobtainedduringopenuse aneffectiveskysubtraction(A- B)canbeperformedusingthe
timethroughNAOJoverthreenightsinDecember2010(ID; twodifferentskyimages:An- Bn- 1andAn- Bn+1takenbefore
S10B-108)andtwonightsinDecember2011(ID;S11B-098). andafterthen-thexposure. Aftertheinitialbackgroundsub-
Additional targets were observed during other programsbe- traction,a crosstalksignalisremovedbysubtracting0.15%
ing carriedoutin the COSMOS field throughthe University for IRS1 and 1% for IRS2 from each quadrant. The differ-
4 MATSUOKAETAL.
ence in the bias between the quadrants is corrected to make
theaverageovereachquadrantequal. Wefurtherapplyaflat
fieldcorrectionusingadomelampexposure. Badpixelsare
maskedthroughoutthereductionprocedure.Additionalsteps
include the distortion correctionand the removalof residual
airglowlines. Thisprocedureiscarriedoutforbothpositions
A and B. Individual frames are combined into an averaged
imageandanassociatednoiseimage.Finally,thewavelength
calibrationiscarriedoutbasedonareferenceimageofaTh-
Ar emission spectrum. Individual one-dimensional science
anderrorspectraare extractedbothtobe usedforthe fitting
ofemissionlineprofiles.
Weperformafirstattemptatfluxcalibrationbyusingspec-
tra of bright stars, usually 1- 2 per spectrograph. A single
stellar spectrum for each spectrograph is chosen to apply a
correctionbasedonthespectroscopicmagnitudeandphotom-
FIG. 2.—Bolometric luminosity as a function ofredshift forAGNs in
etryfrom2MASS.Animprovementoftheabsolutefluxlevel
COSMOS(redfilledcircles), ECDF-S(redopencircles), andSDSS(grey
isrequiredtoaccountforapertureeffects.Therefore,wescale andblacksymbols).
thefluxofeachFMOSspectraofourAGNstomatchthedeep
infraredphotometryusingtotalmagnitudesavailablefromthe resolutionmode(seeFigure19ofKimuraetal.2010).
UltraVISTA survey (McCrackenetal. 2012) available over
the COSMOS field. While scale factors can reach as high 3. SPECTRALLINEFITTING
as∼4,themedianvalueis1.64.
Wemeasurethepropertiesofbroademissionlinespresent
2.4. Samplecharacteristicsandcompleteness in optical and near-infrared spectra to derive single-epoch
black hole mass estimates. For the emission lines of inter-
We haveobservedover100type1AGNsinthecombined
est here (i.e., Mg II and Hα), we specifically measure the
COSMOSand ECDF-Sfieldsto date. InFigure1, we show
fullwidthathalfmaximum(FHWM),totalluminosityofthe
the X-ray flux and NIR magnitude distribution of the 108
emissionlineinthecaseofHα,andcontinuumluminosityat
AGNs(originallyidentifiedasChandraX-raysources)inthe
3000Å. Due to the moderateluminositiesof the AGN sam-
COSMOS field that have 0.6< z< 1.8, a redshift interval
ple,therecanbeanon-negligiblehostgalaxycontributionthat
where we are capable of detecting Hα in the FMOS spec-
impactsthe estimate oftheAGN continuumat redderwave-
troscopic window. The distributions are shown for both the
lengths;therefore,wechosetousetheHαlinefluxratherthan
observedobjectsandthe56havingasignificantdetectionof
thecontinuumluminosity. Fortunately,themulti-wavelength
a broad Hα emission line (at the expected wavelength) that
photometryoftheCOSMOS,includingtheHSTimaging,en-
subsequentlyyieldedablackholemassestimate.
ablesustodeterminethelevelofsuchcontaminationthatwill
Priortoourobservations,wehadnoempiricalassessment
befullyassessedinafuturestudy.Emissionlinesarefitusing
oftheperformanceofFMOSthusAGNweretargetedtofaint
aprocedureasoutlinedbelowthatenablesustocharacterize
infraredmagnitudes,nowunderstoodnottobefeasibleusing
thelineshapeforeventhosethathaveaconsiderablelevelof
the low-resolution mode for the faintest objects (J ∼23).
AB
noise.
Asshown in Figure1(b), we havea reasonablelevelof suc-
Our fitting procedure of the continuum and line emission
cess (71%) with the detection of broad emission lines at
utilizes MPFITFUN, a Levenberg-Marquardt least squares
brightermagnitudes(J <21.5). Unfortunately,thesuccess
AB
minimizationalgorithmas availablewithintheIDL environ-
rateissignificantlyloweratfaintermagnitudesthusdropping
ment. Even though this routine has well-known computa-
to50%fortheentiresamplewithJ <23. Itisworthhigh-
AB
tionalissues, thisalgorithmiswidelyused duetoits ease of
lightingthatasurveydepthofJ <21.5isabouttwomagni-
AB
use and fast execution time. The routine returns best-fit pa-
tudesfainterthancurrentnear-infraredspectroscopicobserva-
rametersandtheirerrorsaswellasmeasureofthegoodness
tionsofSDSSquasars(Shen&Liu2012)atsimilarredshifts.
ofthe overallfit. We furtherdescribe the individualcompo-
InFigure2,wedemonstratethisbyplottingthebolometriclu-
nents required to successfully extract a parameterization of
minosity(basedonL andabolometriccorrectionof5.15;
3000
thebroadcomponentusedindeterminingvirialmasses. Itis
Shenetal.2011)ofourAGNcomparedtothosefromSDSS
worthrecognizingthat each line has its own advantagesand
surveys.
disadvantagesthatneedtobeconsideredcarefullyespecially
Greene&Ho(2005)describeindetailthebenefitsofusing
when fitting data of moderate signal-to-noise ratio (S/N). A
Hαforblackholemassmeasurements. Inparticular,theHα
finalinspectionofeachfitbyeyeisperformedtoremoveob-
emissionlineisstronger(∼3×)thusmoreeasilydetectedas
viouscaseswhereabroadcomponentisnotadequatelydeter-
compared to Hβ. This is clearly evident from our observa-
minedalmostexclusivelyduetospectrahavinglowS/N.
tions.WesuccessfullydetectabroadHαlinein∼50%ofthe
cases(asmentionedabove)whilethedetectionofHβ isvery
3.1. Hα
low(17%).Evenso,wedodetectHβinafairnumberofcases
up to z∼2.6 thatwill be presentedin the full emission-line WeperformafittotheHαemissionline(ifdetectedwithin
catalog (Silverman et al. in prep.). Any future FMOS cam- the FMOS spectral window) in order to measure line width
paign designed to detect the Hβ emission line (both broad and integrated emission-line luminosity. Based on the spec-
and narrow) should increase the exposure time significantly troscopic redshift as determined from optical spectroscopy,
and/or use the high-resolution mode that has roughly three we select the spectrum at rest wavelengths centered on the
times higher throughputin the H band as compared to low- emission line and spanning a range that enables an accurate
BHMASSESTIMATESOFMODERATE-LUMINOSITYAGNS 5
FIG.3.—ExamplesofourfittingroutinefortheHαlinedetectedinCOS- FIG. 4.—ExamplesofourfitsfortheMgIIlineforthesameAGNas
MOS AGN. In the upper portion of each panel, the observed spectrum is showninFigure3.ThetoppanelshowsthespectralrangearoundMgII.The
showninblackwiththebestfitinred. Thecontinuum (graydashedline) bestfitmodelisindicatedasaredsolidline. Thedifferentcomponentsare
andindividualemission-linecomponents(greygaussiancurves)arealsoin- asindicated: Fe-emission(blue),pseudocontinuum(purple),andGaussian
dicatedrespectively.Inlowerpanels,theresidualsareshownwithsamescale components(greenandbrown).Theresidualisshowninthelowerpanel.In
asthatofupperpanel. theupperrightcornerforeachpanel,weshowtheblackholemassdistribu-
tioncomputedfromourMonteCarlotestsbasedontheuncertaintiesofthe
linewidthandcontinuumluminositymeasurements.
determinationofthecontinuumcharacterizedbyapowerlaw,
fλ∝λ- α.
We employmultipleGaussian componentsto describe the laboratoryvalueof2.96. Thenarrowwidthofthe[NII]lines
lineprofile. Itis commonpracticeto makesuchanassump- isfixedtomatchthenarrowcomponentofHα. Thewidthof
tion on the intrinsic shape of individual components, even thenarrowcomponentsislimitedto420- 800kms- 1(arange
thoughithasbeendemonstratedthatbroad-emissionlinesin notcorrectedforintrinsicdispersion). Thevelocityprofileof
AGN are not necessarily of such a shape (e.g., Collinetal. the broad componentsis characterized by the FWHM, mea-
2006). TheHαlineisfitwithtwoorthreeGaussians(includ- suredusingeitheronegaussianorthesumoftwogaussians.
inganarrowcomponent)andthe[NII]λ6548,6684lineswith Wethencorrectthevelocitywidthfortheeffectofinstrumen-
apairofGaussians. Theratioofthe[NII]linesisfixedatthe tal dispersion to achieve an intrinsic profile width. The Hα
6 MATSUOKAETAL.
luminosity discussed throughoutthis work is the sum of the
broadcomponents.
There are cases for which the fitting routine returns a so-
lutionwiththewidthofthenarrowcomponentpeggedatthe
upperboundof800kms- 1. Itisworthhighlightingthatthis
minimization routine stops the fitting procedure when a pa-
rameters hit a limit thus the returned values are not the true
best-fit values. For these, we inspect all fits by eye and de-
cidewhethersuchanadditionalbroadcomponentisreal. For
manycases, we can use the [OIII]λ5007line profile, within
the FMOS spectral window, to determine whether such a fit
to the narrow line complex is accurate. In addition, we can
usetheavailableopticalspectraforsuchcomparisons. When
the level of significance of the narrow line is negligible, we
rerun the fitting routine and fix the narrow line width to the
spectralresolutionofFMOS, ∼420kms- 1. InFigure3,we
showthree examplesofourfits to the Hα emission line that
spanarangeoflineproperties.
3.2. MgII
We fit the Mg II emission line, as done in
Schramm&Silverman (2012), observed in optical spec- FIG. 5.— Comparison between the monochromatic luminosity at 3000
tra primarily from zCOSMOS (Lillyetal. 2009), Mag- ÅandtheHαemission-lineluminosity. AGNsinoursampleareshownas
ellan/IMACS (Trumpetal. 2007), and SDSS (Yorketal. redcircles. TheopencirclesmarkpublishedobservationsofSDSSquasars
2000). The emission line is modeled by a combination of (greycircles: 0.35<z<0.40;blackcircles: 1.5<z<2.2). Alinearfitto
bothourdataonly(redline)andthejointsample(blackline)isgiven. The
one or two broad Gaussian functions to best characterize distribution oflog(λL3000/LHα)asahistogram isshowninthesubpanel.
the line shape. We first remove the continuum (before Thecolorsoftheirhistogramsaresameasthesymbols. Errorsrepresenta
attempting to deal with the emission lines) by fitting the 1σlevelofuncertainty.
emission in a window surrounding the line. As with Hα,
a power-law function is chosen to best characterize the two decades in luminosity and exhibits a clear correspon-
featureless, non-stellar light attributed to an accretion disk. dence between continuum and line emission. Based on our
WefurtherincludeabroadFeemissioncomponentbasedon AGN sample, we determine the best-fit linear relation to be
an empirical template (Vestergaard&Wilkes 2001) that is log(λL3000)=(0.82±0.08)×logLHα+(9.31±3.47). Forthe
convolvedby a Gaussian of variablewidth and straddlesthe linearfitting,weadopta FITEXYmethod(Pressetal.1992;
baseoftheMgIIemissionline. Aleastsquareminimization Parketal.2012).Thelevelofdispersionofthedataaboutthis
is implemented to determine the best-fit parameters. When fitis0.20dexthatwillcontributetothedispersioninthefinal
possible, we optimize residuals of the fits on a case-by-case massestimates.
basis by trying to minimize the number of components. We can compare our data set to more luminous quasars
Absorption features are either masked out or interpolated from the SDSS. In particular, we identify 327 quasars
across. Thefitreturnstwoparametersrequiredforblackhole from the SDSS sample in Shenetal. (2011) with 0.35 <
mass estimates: FWHM and monochromatic luminosity at z < 0.40, that cover a similar luminosity range as our
3000Å. Examplesof ourfits to the MgIIline are presented high redshift sample. These were selected from 1178
inFigure4. quasars at 0.35 < z < 0.40 based on high-quality data
determined by adopting the following criteria: error in
4. RESULTS log(MBH/M⊙) < 0.5, log(FWHMHα/kms- 1) > 2.9, and
We can determine how closely the parameters (i.e., lumi- log(FWHM /kms- 1)>2.9. Inaddition,we includedata
MgII
nosity and FWHM) required to estimate single-epoch black from a recent study by Shen&Liu (2012) that providesthe
holemassesagreebetweentheHαandMgIIemissionlines. emissionlinepropertiesincludingHαlineofhigh-luminosity
Any systematic offset or inherent scatter may only add ad- quasars from the SDSS with 1.5<z<2.2. These samples
ditional uncertainty to the derived masses. We essentially are added to our data shown Figure 5. We clearly see that
want to establish whether or not the kinematics of the BLR SDSS quasars fall along the λL3000-LHα relation as estab-
is consistent with photoionized gas in virial motion around lished above. Furthermore, the SDSS quasars have similar
theSMBH.Weprovideallmeasurementsandderivedmasses dispersion at both low-z and high-z samples to our sample,
inTable1. σ =0.20 and σ =0.15, respectively. We highlight that our
AGNsamplenicelyextendssuchcomparisonsbetweencon-
4.1. Luminosities
tinuum luminosity and line emission at higher redshifts and
OurfirstconcernistodeterminewhethertheHαemission- to lower luminosities. We are able to effectively establish a
line luminosity scales appropriatelywith the UV continuum wider dynamic range, not present in the high-z SDSS sam-
luminosity. For the following analysis, we do not correct pleduetothelimitedluminosityrangearoundlog(λL )∼
3000
forextinctiondueto dustandanycontaminationbythe host 46.2. By merging all three samples, we find the following
galaxy;theimpactofthese,thoughttobesmall,willbequan- relation based on a linear fit: log(λL )=(0.96±0.01)×
3000
tified in a later study. In Figure 5, the continuum luminos- logLHα+(3.00±0.53).
ity,λLλ at3000Å,isplottedagainsttheemission-linelumi- While there is verygoodagreementbetweenthe UV con-
nosity of Hα. Our data (as shown by the red points) spans tinuum and emission line luminosity, there is a small dif-
BHMASSESTIMATESOFMODERATE-LUMINOSITYAGNS 7
TABLE1
CATALOGOFEMISSION-LINEPROPERTIESANDBLACKHOLEMASSES
logL[ergs- 1] logFWHM[kms- 1] logMBH[M⊙]
CIDa Field RA[J2000] DEC[J2000] z Hα 3000Å Hα MgII Hα MgII
178 COSMOS 149.58521 +2.05114 1.350 43.76±0.03 45.35±0.03 3.685±0.013 3.811±0.079 8.68±0.03 8.78±0.16
5275 COSMOS 149.59021 +2.77450 1.400 44.10±0.13 45.50±0.02 3.835±0.028 3.973±0.072 9.18±0.09 9.18±0.14
322 COSMOS 149.62421 +2.18067 1.190 43.16±0.14 45.02±0.02 3.440±0.044 3.581±0.081 7.84±0.12 8.17±0.16
192 COSMOS 149.66358 +2.08522 1.220 43.24±0.14 45.05±0.02 3.339±0.075 3.490±0.035 7.68±0.17 8.00±0.07
157 COSMOS 149.67512 +1.98275 1.330 42.98±0.13 44.84±0.03 3.701±0.040 3.632±0.017 8.28±0.11 8.19±0.04
aTheChandraobs.ID.
ference that may impact, even slightly, our comparison of
the masses. Based on the FMOS sample, the mean ra-
tio hlog(λL3000/LHα)i is 1.54±0.03, slightly higher than
that found for both the low-z and high-z SDSS quasar sam-
ples mentioned above; hlog(λL3000/LHα)i=1.40±0.01 and
hlog(λL3000/LHα)i= 1.36±0.02, respectively. This can be
seenintheinsethistograminFigure5. Ifthiswasduetothe
effect of dust extinction, one would find the opposite trend
withreducedUVcontinuumrelativetotheHαlineemission.
TheremaybeotherexplanationssuchasanunderlyingSED
thatmay bedifferentforX-rayselected samples(Elvisetal.
2012;Haoetal.2012)andhasanimpactontheresponseseen
in photoionized gas, or the effect of the host galaxy on the
aperturecorrectionsthat differsin each band. While this is-
sue is of importance, we reserve a detailed investigation to
subsequentworksinceitisbeyondthescopeofthispaperto
adequately demonstrate such effects. Here, we are primar-
ily concernedwith the magnitudeof a luminosity offsetand
whetheritcontributestoanoffsetbetweenthemasses. With
blackholemassscalingwiththesquarerootoftheluminos-
ity(seebelow),theoffsetinluminosity,asdeterminedabove,
amountstoaverysmalloffsetinlog(MBH/M⊙)of0.07. FIG. 6.— Comparison between the FWHM of Hα with that of MgII.
Red,gray,andblackcirclesarethesameasthoseinFigure5. Alinearfitto
4.2. Velocitywidths
bothourdataonly(redline)andthejointsample(blackline)isverysimilar
to a one-to-one relation indicated by the dashed line. The distribution of
A second pillar for the use of MgII as a black hole mass
indicator is that the emitting-line gas is located essentially log(FWHMMgII/FWHMHα)isalsoshowninthesubpanel.
within the same clouds that emit Balmer emission. While
there are claims that this is the case by comparing the ve-
locityprofileofMgIIwithHβ (e.g.,McLure&Jarvis2002;
Shenetal. 2011), there are reported differences and trends
theclaimbyWangetal.(2009)forashallowervalue. These
that are not well understood (Wangetal. 2009; Shen&Liu
results are supportive of a scenario where the Mg II and the
2012;Trakhtenbrot&Netzer2012).Forexample,MgIItends
Hαemittingregionsareessentiallyco-spatialwithrespectto
tobenarrowerthanHβ withadifferencesignificantlylarger
thecentralionizingsource.
athighervelocitywidths(seeFigure2ofWangetal.2009).
Thereareafewnoticeableoutlierswelloutsidethedisper-
OuraimhereistocomparetheFWHMoftheMgIIandHα
sionofthesample.Theseobjectsthenappearasoutlierswhen
emission lines using a sample not yet explored, namely the
comparing their masses based on different lines in the next
moderate-luminosityAGNsathigh-redshiftsin surveyfields
section. Uponinspection,wefindthatthesearetheresultof
suchasCOSMOS.InFigure6,weplottheemission-lineve-
theFMOSspectrahavinglowS/N.Insomecases,theremay
locitywidthbetweenHαandMgIIemissionlines. Basedon
beafittotheHαemissionline,basedondifferentparameter
the FMOS sample only, a positive linear correlation is seen
constraints,thatisequallyacceptabletotheoriginalfitasas-
between the velocity width of the two emission lines with
sessedbyachi-squaregoodnessoffitandhasavelocitywidth
the mean ratio of hlog(FWHMMgII/FWHMHα)i = 0.089± in closer agreement with Mg II. Although, we refrain from
0.022(σ=0.143). Ourdata is in very goodagreementwith
such selective fitting in order to present results that may be
those from the SDSS, i.e., hlog(FWHMMgII/FWHMHα)i = obtainedfrom using similar fitting algorithmson largerdata
0.033 (σ = 0.176) for the low-z sample from Shenetal.
setswheresuchcloseinspectionisnotfeasible.
(2011),hlog(FWHMMgII/FWHMHα)i=0.00(σ=0.076)for
thehigh-zsamplefromShen&Liu(2012),andrecentFMOS
4.3. Virialsingle-epochmasses
resultsfromSXDS(see Nobutaetal.2012). Based ona lin-
earfittothedatashowninFigure6,wemeasureaslopethat We now calculate black hole masses (M ) based on our
BH
is consistent with unity; 0.898±0.132 for FMOS only and single-epochspectrausingboth(i)L andFWHM ,and
3000 MgII
1.001±0.001forFMOS+SDSS,andthatcannotsubstantiate (ii)LHα andFWHMHα. Thiscalculationcanbeexpressedas
8 MATSUOKAETAL.
5. CONCLUSIONS
WehaveinvestigatedtheemissionlinepropertiesofAGNs
in COSMOS and ECDF-S to establish whether Mg II and
Hα provide comparable estimates of their black hole mass.
ThisstudyisthefirstattempttodosoforAGNsofmoderate
luminosity, hence lower black hole mass (107M⊙ <MBH <
109M ), at high redshift that complements studies of more
⊙
luminousquasars. Ourresults clearlyshow thatthe velocity
profilesofMgIIandHαareverysimilarwhencharacterized
by FWHM and the relation between continuum luminosity
andlineluminosityistight.Wethenfindthatvirialblackhole
massesbasedonMgIIandHαhaveverysimilarvaluesanda
levelofdispersion(σ∼0.3)comparabletoluminousquasars
from SDSS. It is important to keep in mind that these re-
sultspertaintospecificcalibrations(McLure&Dunlop2004;
Greene&Ho2005)forestimatingblackholemass. Theuse
ofotherrecipes,suchasprovidedbyMcGilletal.(2008),will
show a discrepancy larger than seen here. To conclude, the
locally-calibratedrecipesforblackholesmassesusingMgII
and Hα are applicableforfainter AGN samples at high red-
shift. These results further support a lack of evolution in
FIG. 7.—ThecomparisonoftheblackholemassestimatedbyHαline the physical properties of the broad line region in terms of
withthatbyMgIIline.Red,gray,andblackcirclesarethesameasthosein
quantities such as α (e.g., Steffenetal. 2006; Greenetal.
Figure5. Thefittobothourdata(redline)andthejointsample(blackline) ox
isnearlyequivalenttoaone-to-onerelation(dashedline).Thedistributionof 2009), emission-line strengths (e.g., Fan 2006), and the in-
log(MMgII/MHα)isalsoshowninthesubpanel. ferred metallicities (Maiolinoetal. 2001; Nagaoetal. 2006;
Matsuokaetal.2011).
follows: As a final word of caution, such estimates of black hole
massarelikelytohaveinherentdispersionasdiscussedabove
log MBH =a+blog λLλ orLline +clog FWHM . andsystematicuncertaintiesthatarenotyetwellunderstood.
(cid:18)M⊙ (cid:19) (cid:18)1044ergss- 1(cid:19) (cid:18) kms- 1 (cid:19) For instance, recipes for estimating black hole mass depend
(1)
ontheassumptionthatthegasispurelyinvirialmotion.This
We explicitly use the recipes provided by Greene&Ho
isunlikelytobetrueforallcasessincebothoutflowsandin-
(2005)andMcLure&Dunlop(2004)forthecasesofHαand
flowsare commonin AGNs. Evenso, thereis evidencethat
MgIIlines,respectively:
thevirialproductofmassandluminosity(asa proxyforthe
(a,b,c)=(1.221,0.550,2.060) forHα, radius to the BLR) is a useful probe of the central gravita-
tionalpotential. Inthe veryleast, itis importanttoestablish
(a,b,c)=(0.505,0.620,2.000) forMgII.
thelevelofdispersioninsuchrelationssinceobservedtrends
WenotethatthecalibrationoftherelationforMgIIhasbeen usually rely on offsets comparableto the dispersion such as
carried out by many studies (e.g., Vestergaard&Peterson theredshiftevolutionoftherelationbetweenblackholesand
2006;McGilletal.2008)andthereareknowndifferencesbe- theirhostgalaxies(e.g.,Pengetal.2006;Merlonietal.2010;
tweenthem. Schramm&Silverman2012).
In Figure 7, we show M for our sample derived
BH
from Hα and Mg II. Our sample spans a range of 7.2 .
log(MBH/M⊙) . 9.5 consistent with that reported by pre-
vious studies of type 1 AGNs in COSMOS (Merlonietal.
2010;Trumpetal.2011)andiscomplementarytothehigher-
L quasarsample at similarredshiftswith log(MBH/M⊙)&9
(Shen&Liu 2012). We find the average (dispersion) in the
blackholemassratioofhlog(MMgII/MHα)iis0.17(σ=0.32)
forourFMOSsample. Theseresultsaresimilartothatdeter- We thank Kentaro Aokiand NaoyukiTamura for their in-
minedfromtheSDSSsample;theaverage(dispersion)inthe valuable assistance during our Subaru/FMOS observations.
blackholemassratio is - 0.05(σ=0.39)atlowz and- 0.03 K.M.acknowledgesfinancialsupportfromtheJapanSociety
(σ=0.20)athighz. Whileanoffsetof0.17dexisseeninthe for the Promotion of Science (JSPS). Data analysis were in
FMOS sample, we conclude that the recipes established us- partcarried outon common-usedata analysiscomputersys-
inglocalrelationsgiveconsistentresultsbetweenMgII-and temattheAstronomyDataCenter,ADC,oftheNationalAs-
Hα-basedestimates. tronomicalObservatoryofJapan(NAOJ).
REFERENCES
Bahcall,J.N.,Kozlovsky,B.-Z.,&Salpeter,E.E.1972,ApJ,171,467 Fan,X.2006,NewARev.,50,665
Bentz,M.C.,Walsh,J.L.,Barth,A.J.,etal.2009,ApJ,705,199 Green,P.J.,Aldcroft,T.L.,Richards,G.T.,etal.2009,ApJ,690,644
Brusa,M.,Civano,F.,Comastri,A.,etal.2010,ApJ,716,348 Greene,J.E.,&Ho,L.C.2004,ApJ,610,722
Civano,F.,Elvis,M.,Brusa,M.,etal.2012,ApJS,201,30 Greene,J.E.,&Ho,L.C.2005,ApJ,630,122
Collin,S.,Kawaguchi,T.,Peterson,B.M.,etal.2006,A&A,456,75 Greene,J.E.,&Ho,L.C.2007,ApJ,670,92
Elvis,M.,Civano,F.,Vignali,C.,etal.2009,ApJS,184,158 Hao,H.,Elvis,M.,Kelly,B.C.,etal.2012,arXiv:1210.3044
Elvis,M.,Hao,H.,Civano,F.,etal.2012,ApJ,759,6 Hopkins,P.F.,Richards,G.T.,&Hernquist,L.2007,ApJ,654,731
BHMASSESTIMATESOFMODERATE-LUMINOSITYAGNS 9
Iwamuro,F.,Motohara,K.,Maihara,T.,etal.2001,PASJ,53,355 Press, W.H.,Teukolsky, S.A., Vetterling, W. T.,& Flannery, B. P.1992,
Iwamuro,F.,Moritani,Y.,Yabe,K.,etal.2012,PASJ,64,59 Numerical recipes in C. The art of scientific computing, Cambridge:
Kaspi,S.,Smith,P.S.,Netzer,H.,etal.2000,ApJ,533,631 UniversityPress,2nded.
Kaspi,S.,Maoz,D.,Netzer,H.,etal.2005,ApJ,629,61 Roseboom,I.G.,Bunker,A.,Sumiyoshi,M.,etal.2012,MNRAS,426,1782
Kaspi,S.,Brandt,W.N.,Maoz,D.,etal.2007,ApJ,659,997 Schramm,M.,&Silverman,J.D.2012,arXiv:1212.2999
Kelly,B.C.,&Shen,Y.2012,arXiv:1209.0477 Shen,Y.,Greene,J.E.,Strauss,M.A.,etal.2008,ApJ,680,169
Kimura,M.,Maihara,T.,Iwamuro,F.,etal.2010,PASJ,62,1135 Shen,Y.,Richards,G.T.,Strauss,M.A.,etal.2011,ApJS,194,45
Lehmer,B.D.,Brandt,W.N.,Alexander,D.M.,etal.2005,ApJS,161,21 Shen,Y.,&Kelly,B.C.2012,ApJ,746,169
Lilly,S.J.,LeBrun,V.,Maier,C.,etal.2009,ApJS,184,218 Shen,Y.,&Liu,X.2012,ApJ,753,125
Luo,B.,Brandt,W.N.,Xue,Y.Q.,etal.2010,ApJS,187,560 Silverman,J.D.,Mainieri,V.,Salvato,M.,etal.2010,ApJS,191,124
Maihara,T.,Iwamuro,F.,Yamashita,T.,etal.1993,PASP,105,940 Steffen,A.T.,Strateva,I.,Brandt,W.N.,etal.2006,AJ,131,2826
Maiolino,R.,Mannucci,F.,Baffa,C.,etal.2001,A&A,372,L5 Steinhardt,C.L.,&Elvis,M.2010,MNRAS,402,2637
Marziani,P.,Sulentic,J.W.,Stirpe,G.M.,etal.2009,A&A,495,83 Steinhardt,C.L.,&Silverman,J.D.2011,arXiv:1109.1554
Marziani, P., Sulentic, J. W., Plauchu-Frayn, I., & del Olmo, A. 2013, Szokoly,G.P.,Bergeron,J.,Hasinger,G.,etal.2004,ApJS,155,271
arXiv:1301.0520 Tamura,N.,Ohta,K.,&Ueda,Y.2006,MNRAS,365,134
Matsuoka,K.,Nagao,T.,Marconi,A.,etal.2011,A&A,527,A100 Trakhtenbrot,B.,&Netzer,H.2012,MNRAS,427,3081
McCracken, H.J.,Milvang-Jensen, B.,Dunlop,J.,etal.2012,A&A,544, Trump,J.R.,Impey,C.D.,McCarthy,P.J.,etal.2007,ApJS,172,383
A156 Trump,J.R.,Impey,C.D.,Kelly,B.C.,etal.2009,ApJ,700,49
McGill,K.L.,Woo,J.-H.,Treu,T.,&Malkan,M.A.2008,ApJ,673,703 Trump,J.R.,Impey,C.D.,Kelly,B.C.,etal.2011,ApJ,733,60
McLure,R.J.,&Jarvis,M.J.2002,MNRAS,337,109 Vestergaard,M.,&Wilkes,B.J.2001,ApJS,134,1
McLure,R.J.,&Dunlop,J.S.2004,MNRAS,352,1390 Vestergaard,M.,&Peterson,B.M.2006,ApJ,641,689
Merloni,A.,&Heinz,S.2008,MNRAS,388,1011 Vestergaard,M.,&Osmer,P.S.2009,ApJ,699,800
Merloni,A.,Bongiorno,A.,Bolzonella,M.,etal.2010,ApJ,708,137 Wang,J.-G.,Dong,X.-B.,Wang,T.-G.,etal.2009,ApJ,707,1334
Nagao,T.,Marconi,A.,&Maiolino,R.2006,A&A,447,157 Xue,Y.Q.,Luo,B.,Brandt,W.N.,etal.2011,ApJS,195,10
Nobuta,K.,Akiyama,M.,Ueda,Y.,etal.2012,ApJ,761,143 Yabe,K.,Ohta,K.,Iwamuro,F.,etal.2012,PASJ,64,60
Osterbrock,D.E.,&Pogge,R.W.1987,ApJ,323,108 York,D.G.,Adelman,J.,Anderson,J.E.,Jr.,etal.2000,AJ,120,1579
Park,D.,Kelly,B.C.,Woo,J.-H.,&Treu,T.2012,ApJS,203,6
Peng,C.Y.,Impey,C.D.,Ho,L.C.,etal.2006,ApJ,640,114
Peterson,B.M.1993,PASP,105,247