The properties of radio galaxies and the effect of environment in large-scale structures at z&nbsp

25 Pages·2017·3.66 MB·English

MNRAS472,998–1022(2017) doi:10.1093/mnras/stx1984
AdvanceAccesspublication2017August3
The properties of radio galaxies and the effect of environment in
∼
large-scale structures at z 1
Lu Shen,1‹ Neal A. Miller,2 Brian C. Lemaux,1 Adam R. Tomczak,1 Lori M. Lubin,1
Nicholas Rumbaugh,3 Christopher D. Fassnacht,1 Robert H. Becker,1 Roy R. Gal,4
Po-Feng. Wu5 and Gordon Squires6
1PhysicsDepartment,UniversityofCalifornia,Davis,OneShieldsAvenue,Davis,CA95616,USA
2DepartmentofMathematicsandPhysics,StevensonUniversity,1525GreenspringValleyRoad,Stevenson,MD21153,USA
3NationalCentreforSupercomputingApplications,UniversityofIllinois,1205WestClarkSt.,Urbana,IL61801,USA
4InstituteforAstronomy,UniversityofHawai’i,2680WoodlawnDrive,Honolulu,HI96822,USA D
o
5Max-PlanckInstitutfu¨rAstronomie,Ko¨nigstuhl17,D-69117,Heidelberg,Germany w
n
6SpitzerScienceCentre,CaliforniaInstituteofTechnology,M/S220-6,1200E.CaliforniaBlvd.,Pasadena,CA91125,USA lo
a
d
e
d
Accepted2017August1.Received2017July31;inoriginalform2017February14 from
h
ttp
s
ABSTRACT ://a
c
a
In this study, we investigate 89 radio galaxies that are spectroscopically confirmed to be d
e
membersoffivelarge-scalestructures(LSSs)intheredshiftrangeof0.65≤z≤0.96.Based m
ic
onatwo-stageclassificationscheme,theradiogalaxiesareclassifiedintothreesub-classes: .o
u
p
activegalacticnucleus(AGN),Hybrid,andstar-forminggalaxy(SFG).Westudytheproperties .c
o
of the three radio sub-classes and their global and local environmental preferences. We find m
/m
AGN hosts are the most massive population and exhibit quiescence in their star formation n
ra
activity. The SFG population has a comparable stellar mass to those hosting a radio AGN s
/a
butareunequivocallypoweredbystarformation.Hybrids,thoughselectedasanintermediate rtic
population in our classification scheme, were found in almost all analyses to be a unique le
/4
type of radio galaxies rather than a mixture of AGN and SFGs. They are dominated by a 72
/1
high-excitation radio galaxy population. We discuss environmental effects and scenarios for /9
9
each sub-class. AGN tend to be preferentially located in locally dense environments and in 8
/4
thecoresofclusters/groups,withthesepreferencespersistingwhencomparingtogalaxiesof 06
2
similarcolourandstellarmass,suggestingthattheiractivitymaybeignitedinthecluster/group 2
0
6
virializedcoreregions.Conversely,SFGsexhibitastrongpreferenceforintermediate-density b
y
globalenvironments,suggestingthatdustystarburstingactivityinLSSs islargelydrivenby g
u
e
galaxy–galaxyinteractionsandmerging. s
t o
n
Key words: galaxies: active – galaxies: clusters: general–galaxies: evolution–galaxies: 1
1
groups:general–galaxies:starformation–radiocontinuum:galaxies. M
a
rc
h
2
0
Bestetal.2005,2007;Mauch&Sadler2007;Kauffmann,Heck- 2
1 INTRODUCTION 3
man&Best2008)andgenerallyretainthesepropertiesouttoz∼2
Two main galaxy populations are detected at radio wavelengths: (Malavasietal.2015)atleastforradiogalaxyatlog(M∗/M(cid:5))>10.
star-forming galaxies (SFGs) and galaxies with an active galac- Sinceradiowavesarenotaffectedbydustextinction,asevidenced
tic nucleus (AGN, e.g. Miley 1980; Condon 1992; Mauch & bythetightfar-infraredradiocorrelation(FRC)observedfordusty
Sadler2007;Smolcˇic´ etal.2008;Padovanietal.2009,2011).In SFGs (Condon 1992; Condon, Cotton & Broderick 2002; Bell
bothcases,thedominantsourceoftheradioemissionissynchrotron et al. 2003; Sargent et al. 2010; Appleton et al. 2014; Magnelli
radiation from relativistic electrons accelerated by supernova or et al. 2015), radio luminosity is widely used as a star formation
powered by AGN, with a sub-dominant component of free–free rate (SFR) indicator (e.g. Yun, Reddy & Condon 2001; Hopkins
radiationfromHIIregions(e.g.Condon1992). etal.2003;Belletal.2003,2005).Thescatteranddifferencesof
Radio AGNs are typically found in massive quiescent galaxies certainpopulationsintheFRCalsoallowfortheseparationofSFGs
with older stellar populations (Miller & Owen 2002; Best 2004; fromAGN(e.g.Padovanietal.2011;Bonzinietal.2013).
RadioAGNsareoftensegregatedintotwosub-populations:high-
excitation radio galaxy (HERG) and low-excitation radio galaxy
(cid:2)E-mail:[email protected] (LERG). These sub-populations are primarily classified based on
(cid:6)C 2017TheAuthors
PublishedbyOxfordUniversityPressonbehalfoftheRoyalAstronomicalSociety
RadiogalaxiesinLSSsatz∼1 999
thepresenceorabsenceofhigh-excitation(HE)emissionlinesin luminosity, and study their environmental preferences. We also
the spectra of their host galaxies (Hine & Longair 1979; Laing compare such environmental preferences of the three radio sub-
et al. 1994). Various methods have been applied to separate ra- classeswithacontrolsampleofspectroscopicallyconfirmedmem-
dio AGN sub-classes by means of radio morphology, the radio bers. For the purpose of studying environmental effects on radio
luminosity,andopticalexcitationdiagrams(e.g.Evansetal.2006; galaxies, rather than environmental properties driven solely by
Smolcˇic´ 2009; Bonzini et al. 2013; Padovani et al. 2015). It is colourorstellarmass,wecarefullyselectanon-radiocontrolsam-
foundthatthesetworadioAGNpopulationsaredifferentinternally; plematchedinthestellarmassandrest-framecolourforeachradio
HERGs are powered by radiative-mode (or quasar-mode, cold- sub-classandperformacomparisonontheglobalandlocalenvi-
mode)accretion,whileLERGsarepoweredbyjet-mode(orradio- ronments.
mode,hot-mode)accretion(e.g.Ciotti,Ostriker&Proga2010;Best The paper is outlined as follows. Section 2 introduces the five
&Heckman2012).Intheformercase,theaccretionisefficientnear LSSs.InSection3,wediscusstheobservationaldata.Wepropose
theEddingtonrate,whileinthelattercase,theaccretionisconsid- a classification scheme in Section 4 to divide radio galaxies into
erablylessefficient.Additionally,theblackholesthattheycontain AGN,Hybrids,andSFGs.InSection5,weanalysetheproperties
typically have very different masses, with the former population ofthehostgalaxiesofthethreeradiogalaxysub-classesandtheir
D
hostingblackholesthatare,onaverage,approximatelyanorderof environmentalpreferences.Wethen,inSection6,discusstheresults o
w
magnitudelessmassive(Hickoxetal.2009;Best&Heckman2012). ofthethreeradiosub-classesseparatelyandleadtoascenariofor n
lo
Environments,anothersubjectanalysedinthepaper,playanim- eachsub-class.Finally,inSection7,wesummarizeallourresults. a
d
portantroleontriggeringandquenchingprocesses.ForSFGs,this Throughoutthispaperallmagnitudes,includingthoseintheinfrared ed
role is evidenced by the relationships of galaxy colour, morphol- (IR),arepresentedintheABsystem(Oke&Gunn1983;Fukugita fro
m
ogy,stellarmass,andSFRwithvariousmeasuresofenvironment etal.1996).Allequivalentwidthmeasurementsarepresentedinthe h
(e.g.Dressler1980;Pengetal.2010;Gru¨tzbauchetal.2011;Peng restframeandweadopttheconventionofnegativeequivalentwidths ttp
s
ewthailc.h2p0r1o2c;eWsseatczceoluenttasl.fo2r0t1h3e)s.eHreolwateivoenrs,hiitpiss.nItohtafsulbleyeunnpdreorpsotosoedd cqourorteesdpoinndpirnogpetrouanfitesa.tWureeaodboseprtvaedcoinnceomrdiasnsicoen(cid:3).Acolllddidsatarkncmesataterer ://ac
a
that ram-pressure stripping (RPS), harassment, and tidal forces cosmologywithH0=70kms−1Mpc−1,(cid:4)(cid:3)=0.73,and(cid:4)M=0.27, dem
wouldactwithdifferentefficienciesandoverdifferenttime-scales, andaChabrierinitialmassfunction(IMF;Chabrier2003). ic
dependingonthepropertiesofthegalaxies andontheirenviron- .o
u
ments(e.g.Byrd&Valtonen1990;Abadi,Moore&Bower1999; 2 THE ORELSE LARGE-SCALE STRUCTURE p.co
Bahe´ etal.2013;Merluzzietal.2016).Meanwhile,AGNmaybe m
largelytriggeredthroughgalaxymergingorstrongtidalinteractions SAMPLE /m
n
(e.g. Kauffmann & Haehnelt 2000; Hopkins et al. 2006, 2008), In this section, we give a picture of the five LSSs in our sample. ra
s
though other environmental or secular processes may also be re- ThesefivefieldsaretheoriginalfieldsfromtheORELSEsurvey /a
sponsible (Kocevski et al. 2012, and reference therein). Resul- thatwere observedwithcoordinated VLAandChandraobserva- rtic
le
tant AGN activity can subsequently either trigger (positive feed- tions(seeRumbaughetal.2012fordetailsonthelatter).Theyspan /4
7
back)orsuppress(negativefeedback)starformationactivity(e.g. 7–15Mpc in the plane of the skyand 50–250Mpcalong the line 2
/1
Mullaney et al. 2012; Zubovas et al. 2013; Lemaux et al. 2014b; ofsight(LOS).ThefiveLSSsaredefinedbasedonboththeLOS /9
Hickoxetal.2014;King&Pounds2015). direction(asdeterminedfromthevelocityoffsets)andtheprojected 98
RadioAGNsarefoundpreferentiallynearorinclusters/groups distancetothenearestcluster/group.We,here,relyonRumbaugh /40
6
at low redshift (e.g. Miller & Owen 2002; Best 2004; Argudo- etal.(2012,2016)fortheLSSredshiftboundaries,determinedby 2
2
Ferna´ndezetal.2016)withsuchapreferenceappearingtopersist visuallyexaminingtheirredshifthistogram,anddesignedtoinclude 06
uptoz∼2(e.g.Magliocchettietal.2004;Lindsayetal.2014;Hatch allgalaxiesineachoverallLSS.Meanwhile,thespectroscopiccov- by
etal.2014;Malavasietal.2015).Assuch,radioAGNsarewidely eredtypicallyextendsto5Mpcfromanygivencluster/groupwithin gu
used to detect protoclusters at z > 1 (e.g. Best 2000; Venemans theredshiftrangeofeachLSS.Forthesefivefields,wehavefullyre- es
etal.2007;Wylezaleketal.2013).HERGsandLERGsalsoshow ducedradiocatalogues[ofthe16fieldsinORELSEwithVLA/Karl t on
different environmental preferences, with the former restricted to G.JanskyVeryLargeArray(JVLA)imaging]accompanyingfully 11
low-density environments and the latter occupying a wider range reducedphotometricandspectroscopiccatalogues.Moreover,they M
a
of densities (e.g. Donoso et al. 2010; Gendre et al. 2013). Radio spanthefullrangeofORELSEstructures,intermsofhalomassand rc
h
SFGs, conversely, tend to be located in lower global and local dynamics,albeitwithoutthehigherredshiftstructureswhichwill 2
0
density environments at low redshift (e.g. Miller & Owen 2002; beincludedinfutureworks.Wesummarizethecentralposition,the 23
Best2004). numberofclustersorgroups,andtherangeofvelocitydispersions
Inthisstudy,weinvestigate89radiogalaxies thatarespectro- ineachfield,seeTable1,while,thespectroscopypropertieswillbe
scopicallyconfirmedmembersoffivelarge-scalestructures(LSSs) describedinSection3.3.
at z ∼ 1 drawn from the Observations of Redshift Evolution in
LargeScaleEnvironments(ORELSE;Lubinetal.2009)survey.The
2.1 SC1604
ORELSEsurveyisasystematicsearchforLSSsaroundanoriginal
sampleof20galaxyclustersinaredshiftrangeof0.6<z<1.3 TheSC1604superclusterwasoriginallydiscoveredinGunn,Hoes-
withthefieldofviewofallfields∼0.25–0.5deg2.Thegoalisto sel&Oke(1986)asaclustercandidate,andthenconfirmedbyOke,
studygalaxypropertiesoverawiderangeoflocalandglobalenvi- Postman&Lubin(1998).TheLSSofSC1604spansaredshiftrange
ronments.Inthispaper,weuseVeryLargeArray(VLA)1.4GHz of0.84<z<0.96,consistingoffiveclustersandthreegroups,with
imagingtolocateradiogalaxiesbymatchingthemtoalargedata aclusterbeingσν >550kms−1.Manydetailedstudieshavebeen
setofspectroscopicallyconfirmedmembersandseparatetheminto performedonthisstructure(Lubinetal.1998a,2000;Postman,Lu-
threesub-classes:AGN,Hybrid,andSFG.Wecomparetheprop- bin&Oke2001;Best2002;Gal&Lubin2004;Riekeetal.2004;
erties of their hosts’ galaxies in colour, stellar mass, and radio Postmanetal.2005;Homeieretal.2006;Galetal.2008;Kocevski
MNRAS472,998–1022(2017)
1000 LShenetal.
Table1. PropertiesofORELSELSS.
Field RAa Dec.a <zspec>b Confirmed σ zspec Comfirmed
(J2000) (J2000) clusters/groupsc ranged LSSrange memberse
SC1604 16:04:15 +43:21:37 0.90 8 300–800 0.84–0.96 520
SG0023 00:23:52 +04:22:51 0.84 5 200–500 0.82–0.87 246
SC1324 13:24:35 +30:18:57 0.76 4 200–900 0.65–0.79 421
RXJ1757 17:57:19.4 +66:31:29 0.69 1 862.3±107.9 0.68–0.71 75
RXJ1821 18:21:32.4 +68:27:56 0.82 1 1119.6±99.6 0.80–0.84 129
aCoordinatesforSC1604,SG0023,andSC1324arethemedianofcentralpositionsofclusters/groups,whileRXJ1757andRXJ1821aregivenasthecentroid
ofthepeakofdiffuseX-rayemissionassociatedwiththerespectivecluster.
bAveragezspecofmembergalaxiesineachLSS.
cClusters/groupsarespectroscopicallyconfirmedusingthemethodpresentedinGaletal.(2008).
dGalaxy LOS velocity dispersion in units of km s−1, measured within 1Mpc. For SC1604, SG0023, and SC1324, the range of velocity dispersions of
D
clusters/groupsisgiven.ForRXJ1757andRXJ1821,thevelocitydispersionofthesingleclustersisgiven. o
eNumberofspectroscopicobjectsconfirmedintheLSSredshiftrange,withqualityflagQ=3or4,18.5≤i(cid:7)≤24.5(formoredetailsseeSection3.3). wn
lo
a
d
etal.2009;Lemauxetal.2010,2011,2012;Kocevskietal.2011; 3 OBSERVATIONS AND REDUCTIONS e
d
Rumbaughetal.2012,2013;Ascasoetal.2014;Wuetal.2014; In this section, we present the observations and the reduction of fro
Rumbaughetal.2016). m
radio, photometric, and spectroscopic data. We also describe the h
optical matching technique used to confirm radio galaxies with ttp
s
2.2 SG0023 spectroscopiccounterparts.Bytheendofthissection,weconfirm ://a
c
89radiogalaxiesinthefiveLSSs. a
The SG0023 structure was originally discovered in Gunn et al. d
e
(1986),andconfirmedincludingtwoadditionalgroupsbyOkeetal. m
ic
(1998). The LSS spans the redshift range of 0.82 < z < 0.87, .o
consistingoffivemerginggalaxygroups(Lubinetal.2009,1998a; 3.1 Radioobservations up
.c
Lemauxetal.2016).Manyadditionalstudieshavebeenperformed Each of the five LSSs were observed in 2006–2009 in the same om
on this structure (Postman, Lubin & Oke 1998; Lubin, Postman manner using the VLA at 1.4GHz in its B configuration, where /m
n
& Oke 1998b; Best 2002; Rumbaugh et al. 2012, 2013; Lemaux the resulting full width at half-maximum (FWHM) resolution of ra
etal.2016;Rumbaughetal.2016). the synthesized beam is about 5arcsec. Given the approximate s/a
31arcmin diameter field of view (i.e. the FWHM of the primary rtic
le
2.3 SC1324 beam),weoptedtoobserveeachtargetusingasinglepointingof /4
7
the VLA with the exception of SC1324 for which two pointings 2
TheSC1324superclusterwasoriginallydiscoveredinGunnetal. wereusedtocoverthefullextentofthestructure.Astheobserva- /1/9
(1986), with two clusters confirmed. Subsequent studies have in- tionswerecompletedpriortotheupgradeoftheVLA,weusedthe 98
creasedthestructuretothreeclustersandfourgroup,spanningthe ‘fourmode’correlatorsettingwhichconsistsofseven3.125MHz /40
redshiftrangeof0.65<z<0.79(Postmanetal.2001;Lubin,Oke channelsateachoftwointermediatefrequencies.Thismodewas 62
2
&Postman2002;Lubin,Mulchaey&Postman2004;Rumbaugh thebestcompromiseforsensitivity(fromawidertotalbandwidth) 06
etal.2012,2013,2016). withlesserdistortionofthebeam(thesmallerindividualchannels by
reduces bandwidth smearing). Each LSS was observed over the gu
e
courseofseveraldaysusingindividualtracksofseveralhoursper s
2.4 RXJ1757 day to insure good (u, v)-coverage with a nearly circular synthe- t on
The RXJ1757.3+6631 cluster (hereafter RXJ1757, described in sizedbeam.Netintegrationtimeswerechosentoresultinfinal1σ 11
Rumbaugh et al. 2012), was discovered as part of the ROSAT sensitivitiesofabout10µJyperbeam.Detailsoftheobservations M
a
North Ecliptic Pole (NEP) survey, and identified as NEP200 maybefoundinTable2. rc
h
(Gioiaetal.2003),andconsistingofonlyoneclusteratz=0.69. In general, the procedures for calibration and imaging of the 2
0
Becausewedetectlikelysub-structuresassociatedwithRXJ1757, data paralleled those presented in Miller et al. (2013) except that 23
weconsiderthissystemtobeaLSS.ThisLSSspanstheredshift theLSSdatacorrespondtoasingleor,atmost,twopointingsper
rangeof0.68<z<0.71,andhasonlylimitedstudies(Rumbaugh target field and required substantially fewer facets to handle the
etal.2012,2013). effectsofskycurvature.Theobservationsoneachdateincludeda
gain calibrator (3C48 for SC0023 and 3C286 for all other LSSs)
along with a nearby (ranging from 3.4◦ to 7.6◦) phase calibrator.
2.5 RXJ1821
The(u,v)dataforthesecalibratorswereeditedtoremoveobvious
The RXJ1821.6+6827 cluster (hereafter RXJ1821; described in interferenceandotheraberrationaldata,andtheresultinggainand
Lubin et al. 2009; Rumbaugh et al. 2012), was discovered in the phasecalibrationswereappliedtothetargetfields.Thetargetfields
ROASTNEPsurvey,andidentifiedasNEP5281,consistingofonly wereimagedusinga‘flys-eye’patternofsevenfacets,witheach
oneclusteratz=0.8168.Similarly,welikelyfindsurroundingsub- facethaving10241.5arcsecpixelsperside.Wide-fieldmapsmade
structureassociatedwithRXJ1821.ThisLSS,spanningtheredshift at lower resolution identified bright sources outside this primary
rangeof0.80<z<0.84,hasbeenstudiedbyGioiaetal.(2004), beamcoveringflys-eye,andthesesourcesalongwithbrightsources
Lubin et al. (2009), Lemaux et al. (2010), and Rumbaugh et al. insidetheflys-eyereceivedtheirowndedicatedsmallerfacetsfor
(2012,2013,2016). imaging. For each specific observation date, (u, v) data sets were
MNRAS472,998–1022(2017)
RadiogalaxiesinLSSsatz∼1 1001
Table2. VLAradioobservationcharacteristics.
Field Code Observationdatesa RAb Dec.b rmsc Beam
(J2000) (J2000) (µJy) size
SC1604 S7218 06July07,06July10,06July13,06April09 16:04:13.2 +43:13:51 9.3 5.3arcsec×5.0arcsec
SG0023 S8597 07November06,07November07,02November09,07November10 00:23:50.7 +04:22:45 13.9 4.4arcsec×4.4arcsec
SC1324N S9484 09February18,09February19,09February25,09February26,09March02,09March04 13:24:49.5 +30:51:34 11.4 4.7arcsec×4.5arcsec
SC1324S S9484 09February18,09February19,09February25,09February26,09March02,09March04 13:24:42.5 +30:16:30 10.5 4.7arcsec×4.5arcsec
RXJ1716 SA598 09February23,09February27,09February28,09March01 17:57:19.8 +66:31:39 10.5 4.9arcsec×4.7arcsec
RXJ1821 SA598 09February23,09February27,09February28,09March01 18:21:38.1 +68:27:52 9.3 4.9arcsec×4.7arcsec
aForSC1604,thisisstartingdayoftheobservations,whileforotherfields,thesearethedatewhentheobservationwastaken.
bSpatialpositionforthephasecentreofobservation.
crmssensitivityforfinalimageassociatedwithalldataforthatpointing.
D
generatedandself-calibrated.The(u,v)datawerethenaveragedto bridgeAstronomySurveyUnit.2 Additional,JandKsbandimag- o
w
produceasingle(u,v)datafilecorrespondingtothecompletesetof ingwastakenusingtheCanada–France–HawaiiTelescope(CFHT) nlo
observations(i.e.combiningallobservationdates)foreachtarget Wide-fieldInfraRedCamera(WIRCam;Pugetetal.2004)mounted ad
e
LSSfield.Finalroundsofimagingandself-calibrationwerethen ontheCFHTandwasreducedthroughtheI’iwipre-processingrou- d
performedonthese(u,v)data.Inthesefinalimagingsteps,thedata tinesandTERAPIX.IRimagingat3.6,4.5,5.8,and8.0 µm(5.8and fro
m
for a specific LSS target were imaged separately by intermediate 8.0 µmonlyavailableforSC1604)wastakenusingtheSpitzertele- h
frequency,polarization,andinroughly2-hblocksinLocalSidereal scopeInfraredArrayCamera(IRAC;Fazioetal.2004).Thebasic ttp
s
Time. The resulting images were then combined using variance calibrated data images provided by the Spitzer Heritage Archive ://a
weightingtoarriveatthefinalimagesforeachtargetLSSfield. werereducedusingtheMOSAICKERANDPOINTSOURCEEXTRACTOR ca
Thefinalimageswerethenusedtogeneratesourcecatalogues. (Makovoz&Marleau2005)packageaugmentedbyseveralcustom de
m
TheNationalRadioAstronomyObservatory’s(NRAO)Astronom- InteractiveDataLanguagescriptswrittenbyJ.Surace. ic
icalImageProcessingSystemtaskSADcreatedtheinitialcatalogues Photometry was obtained by running Source Extractor .ou
byexaminingallpossiblesourceshavingpeakfluxdensitygreater (SEXTRACTOR; Bertin & Arnouts 1996) on point spread function p.c
o
thanthreetimesthelocalrmsnoise.Wetheninstructedittoreject (PSF)-matchedimagesconvolvedtotheimagewiththeworstsee- m
allstructuresforwhichtheGaussianfittedresulthasapeakbelow ing. Magnitudes were extracted in fixed apertures to ensure that /m
n
fourtimesthelocalrmsnoise.BecauseGaussianfittingworksbest themeasuredcoloursofgalaxiesareunbiasedbydifferentimage ra
s
fcorerautendrebsyolvSAeDdhaanvdinmgasrugbintraallcyterdesthoelvGedaussosuiarcnesfi,tsrefsriodmuatlhieminagpeust qetuaall.it2y01fr5o)mwaismuasgeedtfooriSmpaitgzee.r/AIRlsAoC, tihmeagpeasc,kdaugeetoT-tPhHeOTlar(gMeePrSliFn /artic
le
images were inspected in order to adjust the catalogue. This step ofthesedatathatcanblendprofilesofnearbysourcestogetherand /4
7
addedthoseextendedsourcespoorlyfittedbyGaussians.Peakflux contaminatesimpleaperturefluxmeasurements.SeeLemauxetal. 2
/1
density, integrated flux density, and their associated flux density (2016)formoredetails. /9
errors(σ)aregeneratedbySAD.Weusethepeakfluxdensityunless Spectralenergydistribution(SED)fittingwasperformedinthree 98
theintegratedfluxislargerbymorethan3σ thanthepeakfluxfor stages.Atthefirststage,aperturemagnitudeswereinputtothecode /40
6
eachindividualsource. EasyandAccuratezphot fromYale(EAZY,Brammer,vanDokkum 22
0
& Coppi 2008). This includes six templates derived from a non- 6
negativematrixfactorizationdecompositionoftheProjectd’E´tude by
3.2 Photometricobservationsandcatalogues des GAlaxies par Synthe`se Ev´olutive (PEGASE; Fioc & Rocca- gue
Comprehensivephotometriccataloguesareconstructedforallfive Volmerange 1997) template library with one additional template st o
fromMaraston(2005)representinganoldstellarpopulation.This n
LSSs.Wesummarizetheavailableopticalandnear-infrared(NIR) 1
templatesetisgenerallymoreeffectiveatidentifyingthelocation 1
observations for each LSS and the reduction of these data in Ta- M
ofstrongspectralbreaksinbroad-bandphotometrythanusingin-
blesC1andC2.Afulldescriptionofthereductionprocesswillbe a
dividual templates. We refer the reader to Brammer et al. (2008) rc
giveninTomczaketal.(2017).Opticalimagingwastakenwiththe h
for a more thorough discussion of the templates used to fit for 2
Large Format Camera (LFC;Simcoe et al. 2000) on the Palomar 0
2
5-mtelescope,usingSloanDigitalSkySurvey(Doietal.2010)-like photometricredshifts. 3
r(cid:7), i(cid:7), and z(cid:7) filters, reduced in the Image Reduction and Analysis Foreachobject,thezphot isestimatedfromaprobabilitydistri-
butionfunction(PDF,hereafterP(z)),generatedbycalculatingthe
Facility (Tody 1993), following the method in Gal et al. (2008) .
WealsouseR-,I-,andZ-bandopticalimagingfromSuprime-Cam χ2 ofthephotometrieswithrespecttoPEGASEmodels.Aseparate
setoffittingwasperformedutilizingalibraryofstellartemplates
(Miyazakietal.2002)ontheSubaru8-mtelescope,reducedwith
(Pickles1998)todeterminestarsfromgalaxies.Inthesecondstage
theSDFRED2pipeline(Ouchietal.2004)supplementedbyseveral
routinesTraitementE´le´mentaireRe´ductionetAnalysedesPIXels of the fitting process, the code EAZY was run again using zphot or
(TERAPIX).1 Some J- and K-band data were taken with the United high-qualityspectroscopicredshiftswhenavailabletoderiverest-
framemagnitudesforallphotometricobjects.Magnitudeswerecor-
Kingdom Infrared Telescope (UKIRT) Wide-Field Camera (WF-
rectedtoinfinitybyapplyingthedifferencebetweenMAG_APER
CAM;Hewettetal.2006)mountedontheUKIRTandwasreduced
and MAG_AUTO in the detection band to all magnitudes when
usingthestandardUKIRTprocessingpipelinecourtesyoftheCam-
drivingphysicalparametersfromSEDfitting.Asforthefinalstage,
1http://terapix.iap.fr 2http://http://casu.ast.cam.ac.uk/surveys-projects/wfcam/technical
MNRAS472,998–1022(2017)
1002 LShenetal.
SEDfittingonaperture-correctedmagnitudeswasperformedwith bolometricIRluminosity,L .Aluminosity-independentconver-
TIR
the Fitting and Assessment of Synthetic Templates (FAST; Kriek sion of flux to LTIR was first suggested at by the stacking results
etal.2009)code,withthesameredshiftusedinthesecondstage. ofPapovichetal.(2007).Sincethenthistechniquehasbeensup-
BecausethetemplatesusedwithEAZYcannotbeusedtoestimate portedbyseveralstudiesacrossabroadrangeofredshiftsandstellar
physicalpropertiessuchasstellarmassweemploythestellarpop- masses(seee.g.Muzzinetal.2010;Wuytsetal.2011;Tomczak
ulation synthesis (SPS) library of Bruzual & Charlot (2003) as- etal.2016).
sumingaChabrier(2003)stellarIMFandsolarmetallicityforthis Weapplyaqualityflagforeachphotometricdetectiontoinclude
partoftheanalysis.Fordustextinction,weadopttheCalzettietal. detectionsabove3σ inthedetectionimagewithcoverageinatleast
(2000)attenuationcurve.Weadoptdelayedexponentiallydeclin- five images, and exclude stars, saturations,3 and bad SED fitting
ingstarformationhistories(SFRs∝t×e−t/τ),allowinglog(τ/yr) (∼1percent worst χ2 values). Photometric objects are also re-
torangebetween8.5and10instepsof0.5,log(age/yr)torange stricted to be in the same spatial area as the spectroscopic ob-
between8and10instepsof0.2,andA torangebetween0and servation in each field and observed i(cid:7) band 18.5 ≤ i(cid:7) ≤ 24.5 to
V
3instepsof0.1.OnlythestellarmassandV-bandattenuationin maximize the completeness of spectroscopy, while still including
magnitudes (A ) derived from this fitting are used in this paper. thevastmajorityofourspectroscopicallyconfirmedmembers.
V D
It is important to note that the parameter space used in our SPS Moreover, we use a wider LSS photometric redshift range for o
w
modellingissmallerthaninothercontemporarystudies.Neverthe- photometricobjectswithoutconfirmedz ,duetotheuncertainty n
spec lo
less,wehavetestedouranalysisusingstellarmassesderivedfrom of z , which varies from field to field. We use z selection a
phot phot d
amoreextendedparameterspaceandfindthatourresultsremain described in Appendix B to select the photometric members (see ed
unchanged. TableB1fornumberofphotometryintheLSSs).Thezphotredshift from
The precision and accuracy of the photometric redshifts rangedefinedusingequation(B1). h
were estimated from fitting a Gaussian to the distribution of ttp
a(znsdpewc−aszfophuont)d/(t1ob+ezσsp(cid:9)ezc/)(1m+eza)s∼ur0e.m03enwtsithinatchaetarsatnrogpeh0i.c5o≤utlzie≤rr1a.t2e 3.3 Spectroscopicobservations s://aca
((cid:9)z/(1+z)>0.15)of5–10percentforallfivefieldstoalimitofi(cid:7) Spectroscopictargetswereselectedbasedontheopticalimagingin de
≤24.5.Aslightsystematicoffsetfromzero((cid:9)z/(1+zphot)∼0.01) the r(cid:7), i(cid:7), and z(cid:7) from LFC,following the methods in Lubin et al. mic
wasnoticedforallfivefields.Thevalueofthisoffset,multiplied (2009).Wetargetedobjectswithprioritygiventoreddergalaxies .ou
by (1 + zphot), was applied to all raw zphot values (see Lemaux atthepresumedredshiftsoftheLSSs,whichwasintendedtopref- p.c
o
etal.2016;Tomczaketal.2017formoredetails).Forgalaxieswith erentiallyconfirmredsequencemembersoverbluercounterparts, m
knownorsuspectedAGNinoursamplebasedonourradioclas- generallyincludinggalaxieswithani(cid:7)-bandmagnitudebrighterthan /m
n
sification,wechecktheirstellartemplatefitting.Theyallprovide 24.5. In addition, for certain masks, we prioritized X-ray and ra- ra
s
dgiofofedrefintstitfoAthGeNo/bHseyrbvreiddptehmotpolmateestrwy,etrheuisnMste∗aadndusAeVd.wouldnotbe dwiiothdtehteecDteedepobIjmecatgsi.nTghaenodpMticualltis-pOebcjteroctscSoppeyctwroagsrapprihm(aDriElyIMtaOkeSn; /artic
le
We make use of Spitzer Multiband Imaging Photometer for Faberetal.2003)ontheKeckII10mtelescope.DEIMOSisanef- /4
Spitzer (MIPS) 24µm observations for estimating rest-frame to- ficientinstrumentforaspectroscopicsurveyofhigh-redshiftLSSs, 72
talIR(TIR)luminosities(8-1000µm,LTIR),whichareavailablefor because of its large field of view (16.7arcmin × 5.0arcmin) and /1/9
SC1604, SG0023, and RXJ1821. Due to the large PSF of these itscapabilityofpositioningupto120+galaxiesperslitmask.We 98
data (PSF (cid:2) 4arcsec), profiles of nearby sources would blend used the 1200 line mm−1 grading with 1arcsec-wide slit at vari- /40
6
together and contaminate simple aperture flux measurements. To ouscentralwavelengths(seeTable3),withwavelengthcoverageof 22
help mitigate this effect, we employ T-PHOT (Merlin et al. 2015). approximately2600Å.Totalexposuretimesare1–4.5hpermask, 06
In brief, the methodology of T-PHOT is as follows. First, the posi- whichvariedbasedonobservationalconditionsandthemagnitude by
tions and morphologies of objects are obtained for use as priors distributions of targets. A few additional redshifts from the Low gu
e
Abarsneodutosn1a99se6g)m) oefntaatihoinghmeraprepsrooludtuicoendibmyagSeE.XCTRuAt-CoTuOtRs(aBreerttainke&n RadedseodluftoiornSCIm16a0g4in,gSGS0p0e2c3tr,oamndetRerX(JL1R82IS1;(sOekeeOekteaelt.a1l.919959)8w;Gerael st on
from this higher resolution image which are then used to create &Lubin2004;Gioiaetal.2004). 11
low-resolution24µmmodelsofeachobjectbysmoothingwitha DuringthereductionofDEIMOSdata,serendipitousdetections Ma
providedconvolutionkernel.Thesemodelsarethensimultaneously (hereafterserendips)wereidentifiedbyeyeinthespatialprofileof rch
fittothe24µmimageuntiloptimalscalefactorsforeachobjectare slits.Ifsuchdetectionsexisted,aserendipitousspectrumwasgener- 20
obtained as assessed through a global χ2 minimization. We run atedusingthemethoddescribedinLemauxetal.(2009).Redshifts 23
T-PHOT in ‘cells-on-objects’ mode where when fitting a model for andqualityflag(Q)weredeterminedforalltargetsserendipitous
a given object only neighbours within a 1arcmin ×1arcmin box detections.SeeGaletal.(2008)andNewmanetal.(2013)formore
centredontheobjectareconsideredinthefitting.Afterthisinitial detailsonthequalityflags.Weusespectroscopicobjectsonlywith
sequence, T-PHOT is then rerun in a second pass in which regis- Q=3and4inthispaper.Q=3meansonesecureandonemarginal
teredkernelsgeneratedduringthefirstpassareutilizedtoaccount featurewereusedtoderivetheredshift.TargetsofQ=4areatleast
formildastrometricdifferencesbetweentheinputimages.Output twosecurefeatureswereused.Q=−1flagmeansthetargetwas
fluxesarethetotalmodelfluxfromthebestfit. definitelydeterminedtobestar.Q=1or2indicatedthatwecould
To estimate LTIR, we make use of the IR spectral template in- notdetermineasecureredshift.Moreover,thesamei(cid:7) bandmag-
troduced by Wuyts et al. (2008). This template is constructed by nitude use flag (18.5 ≤ i(cid:7) ≤ 24.5), as applied to the photometric
averagingthelogarithmofalltemplatesintheDale&Helou(2002) catalogues,wasappliedtothespectroscopicsamples,tokeepthese
IRspectrallibrary.Foreachgalaxy,wescaletheIRspectraltem-
plate,shiftedtothegalaxy’sredshift(z whenavailable)toobtain
spec
itstotal24µmfluxestimate.Wethenshiftthetemplatebacktothe 3Definedascaseswheremorethan20percentofthepixelsinthephoto-
restframeandintegratebetween8and1000µmtreatingthisasthe metricaperturearesaturatedinthedetectionimage.
MNRAS472,998–1022(2017)
RadiogalaxiesinLSSsatz∼1 1003
Table3. DEIMOSspectroscopicobservationcharacteristics.
LSS Central Approximatespectral Exposuretime Averageseeing Numberof Numberof
λ(Å) coverage(Å) range(s) range(arcsec) mask targetsa
SC1604 7700 6385–9015 3600–14400 0.50–1.30 18 2445
SG0023 7500–7850 6200–9150 5700–9407 0.45–0.81 9 902
SC1324 7200 5900–8500 2700–10800 0.44–1.00 12 1237
RXJ1757 7000–7100 5700–8400 6300–14730 0.47–0.82 6 728
RXJ1821 7500–7800 6200–9100 7200–9000 0.58–0.86 6 593
aIncludingspectroscopictargetwith18.5≤i(cid:7)≤24.5,includingsirendips.
twocataloguesconsistent.Welistthenumberofspectroscopically redshifts (referred to as radio-spec matches) and within the LSS
confirmedmembers,withintheLSSredshiftrangeinTable1. redshiftrange(seeTable1),referredtoasradioconfirmedgalaxies.
D
Weobtain272radio-specmatchesfromtheradio-photmatches,71 o
w
ofwhicharemembersoftheLSSs(seeTable4fornumbersofradio- n
3.4 Opticalmatching photmatches,radio-specmatches,andradio-confirmedgalaxiesin loa
d
Priortobeginningthefulloptical–radiomatchingprocess,therel- eachLSS). ed
ative astrometry of the two sets of imaging was compared in the Radiosourcesmayhaveextendedmorphology,however,inthe fro
m
followingway.Thepositionsofradiopointsourceswithadetec- maximum LR technique, each radio source has been assumed to h
tionsignificancelargerthan10ineachfieldweredrawnfromthe be a point source coincident with the optical centre. To account ttp
catalogues generated in Section 3.1 and were nearest-neighbour forastrophysicalandastrometricoffsets,weuseσj =2arcsecas s://a
matchedtothephotometriccatalogueineachcorrespondingfield. thesearchradiusandrantheopticalmatchingagain.Weexamined ca
The median offsets of the radio sources relative to their nearest- theadditionalmatchesbyoverlayingradiofluxcontoursforeach de
m
neighbouropticalcounterpartsinαandδweremeasuredforeach match on the optical detection image to determine whether they ic
individualfield.Theoffsetswerealwayslessthan0.4arcsecandtyp- are extended radio objects and confirm by eye if they are truly .ou
icallymuchless.Theinverseoftheseoffsetsweresubsequentlyap- radioobjectswithspectroscopiccounterparts.Withthismethod,we p.c
pliedtothefullradiocatalogueforeachfield.Theresultantnormal- added18radiogalaxieswithspectroscopicallyconfirmedredshifts om
izedmedianabsolutedeviation(Hoaglin,Mosteller&Tukey1983) totheradio-confirmedsample.Thenumbersofextraradiogalaxies /m
n
ofthecorrectedradiopositionsofthe>10σsourcesrelativetothose confirmedineachoftheLSSarelistedinTable4. ra
s
oftheiropticalcounterpartswasfoundtobe0.25arcsec,whichis Toaccountforradiodoubles,i.e.thosegalaxiesforwhichtwo /a
considerablysmallerthantheradius(1arcsec)chosenforoptical– distinctradioobjectsarephysicallyassociatedwithanindividual rtic
radiomatching. opticalobject,asinthecaseofFanaroff–RileytypeIandIIsources le/4
ouTrorasdeiaorcshouforcrersadtoiotghaelapxhioetsowmiethtriinctchaetainlodgivuiedsu.aWlLeSuSsse,twheemmaatxcih- d(Fiuasnaarnodffra&ntRhieleoypt1ic9a7l4m),awtcehiunsgedagσajin=us5inagrctsheecsapsecthtreossceoaprcicharllay- 72/1/9
mumlikelihoodratio(LR)techniquedescribedinRumbaughetal. confirmedmembers.Weexaminedtheadditionalmatchesbyover- 98
(2012),whichwasdevelopedbySutherland&Saunders(1992)and layingradiofluxcontoursforeachmatchontheopticaldetection /40
6
also used by Taylor et al. (2005), Gilmour et al. (2007), and Ko- imagetodetermineandconfirmbyeyeiftheyaretrulyradiodou- 22
cevskietal.(2009).Themainstatisticcalculatedineachcaseisthe bles.WefoundoneradiodoubleinSC1324.Theradiofluxdensity 06
LR,whichestimatestheprobabilitythatagivenopticalsourceisthe ofthissourceisthecombinedfluxdensityoftworadiosources. by
genuinematchtoagivenradiosourcerelativetothearrangementof Inordertoconfirmourspectroscopicsampleisnotbiased,we gu
e
thetwosources(cid:2)arisingby(cid:3)chance.TheLRisgivenbytheequation pwehrefonrmtheedpthhoetfoomlleotwriicngcosutenptse.rpWaertsusdeozpnhoottfhoarvreadaios-epchuortemzasptecch,etso, st on
LRi,j= wiexp −σ2ri2,j/2σj2 (1) stheelescetgpaoltaexniteisalaLndSSthmeeramdbioe-rcso.Tnfihremraeddiog-aplahxoitessatmhaptleariencmluadtcehsebdotihn 11 Ma
j thesameradius(σj =1arcsec).Thissampleisdescribedinmore rch
Here,r istheseparationbetweenobjectsiandj,σ isthepositional detailinSection3.2andAppendixB,namedasradio-photsample. 2
i,j j 0
errorofobjectj,whereweuse1arcsecaspositionalerrorofallradio Thenumberofradio-photsampleobjectsineachLSSarelistedin 23
detection(Condon1997),andw2=1/n(<m)istheinverseofthe Table4.
i i
numberdensityofopticalsourceswithmagnitudefainterthanthe When a radio source matches to a spectroscopic target with a
observedi(cid:7)-bandmagnitude.Theinclusionofthelatterquantityis blendedserendip,wewererequiredtodeterminetherealspectro-
designedtoweightagainstmatchingtofainteropticalobjectsthat scopiccounterpart.ThereareeightsuchcasesacrossthefiveLSS:
aremorelikelytohavechanceprojections.Foreachradiosource, sixinSC1604,oneinSG0023,andoneinRXJ1821.Wedetermined
we carried out a Monte Carlo (MC) simulation to estimate the thespectroscopiccounterpartofagivenradiosourcebasedontwo
probabilitythateachopticalcounterpartisthetruematchusingthe methods.First,thespatialseparationbetweentheradiosourceand
LRs.Weadoptthesamethresholdformatchingtosingleordouble spectroscopictargetanditsserendipbasedonacarefulde-blending
objectsasRumbaughetal.(2012). oftheobjectwhenpossible:thefurthertheopticalobjectfromthe
The optical matching is done to the overall photometric cata- radioobject,thelesslikelyitwillbematchedtotheradioobject.
logues, aimed to increase the probability of the true match. We Incaseswherethisde-blendingwasnotconvincing,weexamined
obtainatotal589radio-photmatchesinthemaximumlikelihood thetwo-dimensionspectrumoftheblendedsourcesandchosethe
matching. In this paper, we mainly analyse radio objects which brighter of the traced emission, since the probability of a given
have photometric counterparts with spectroscopically confirmed radio object matched to the brighter optical object is higher. We
MNRAS472,998–1022(2017)
1004 LShenetal.
Table4. Numberofradiosources,andphotometricandspectroscopicMatches.
Field Radio Radio-phot Radio-spec Radio-phot Radio Radio
sourcesa matchesb matchesc sampled confirmede extraf
SC1604 501 193 109 42 32 9
SG0023 313 95 53 12 6 2
SC1324 565 165 36 49 21 4+1
RXJ1757 258 91 45 6 3 0
RXJ1821 201 45 29 13 9 3
aRadiosourceswithasignificance>4σ inatleastoneoftheradiointegratedorpeakflux.
bRadiosourcesmatchedtophotometriccounterpartsinthefield,restrictedwith18.5≤i(cid:7)≤24.5andthesamespatialareaascorresponding
spectroscopicimaging.
cRadiosourcesmatchedtospectroscopiccounterpartsinthefield,restrictedwith18.5≤i(cid:7)≤24.5andqualifyflag.
dRadioobjectsthathavecounterpartsofphotometricmembersintheLSSsinσj=1arcsec.Forphotometricmembers,seeexplanation
D
inSection3.2. o
eRadioobjectsmatchedinσj=1arcsecthathavespectroscopicallyconfirmedcounterpartsintheLSSs. wn
fRadioextraconfirmedmembersinσj=2arcsec+confirmedradiodouble.Thefinalnumberofradioconfirmedsampleisaddingthe loa
lasttwocolumns. de
d
determined that the primary spectroscopic targets are the correct normalgalaxies(Condon1992;Kauffmannetal.2008;DelMoro from
matchestoalleightradioobjects. etal.2013),andalsoconsistentwiththeselectionusedbyotherstud- h
ies (e.g. Hickox et al. 2009). Applying this criteria, we select 20 ttps
AGNsfromtheradio-confirmedsample(seeTable4).Wenotethat ://a
4 TWO-STAGE RADIO CLASSIFICATION theradioluminositycriteriameansthatallSFGsandHybridshave ca
d
Radioemissioncanhaveanoriginfromsynchrotronradiationand luminositiesbelow1023.8;however,becauseofthesecondcolour– em
free–freeradiation,bothofwhichareactiveprocessesforanSFG. SRLcriteria(seeSection4.2),AGNscanhaveradioluminosities ic.o
lessthan1023.8aswell. u
Conversely,theradioemissioninAGNisdominatedbysynchrotron p
.c
andrelatedprocesses.Therefore,weneedamethodtodistinguish o
m
betweenradioemissionduetostarformationandthatduetoAGN /m
activity.Wedefinea‘hybridgalaxy’(hereafterHybrid)asaradio 4.2 Colour–SRLcriteria nra
galaxywithemissionpotentiallyassociatedwithbothSFandAGN s/a
processes. Thesecondstageofourclassificationisthecolour–SRLmethod. rtic
TheFIRandradioluminositiesofSFGsarefoundtobetightly Thedustextinctioncorrectionappliedtotherest-frameMU −MB le/4
related, known as FRC (e.g. de Jong et al. 1985; Helou, Soifer colourfollowsthemethodinCalzettietal.(2000)withAV,derived 72
& Rowan-Robinson 1985; Helou et al. 1988; Condon 1992; Yun from photometric catalogues, as the colour excess of the stellar /1
etal.2001).TheTIR(8−1000µm)/radioluminosityratioqTIR,one continuumEs(B−V)foreachspectroscopicobject.Studieshave /998
indicatoroftheFRC,hasbeenwidelyusedtodisentangleAGNsand shown that AGN and SFGs occupy different areas in the 4000Å /4
0
SFGs(e.g.Helouetal.1985;Belletal.2003;Sargentetal.2010; breakDn(4000)versusradioluminositynormalizedbystellarmass 62
Padovanietal.2009),sincetheFIRemissionat250–1000µmcomes diagramforradiogalaxiesuptoz<1.3(Bestetal.2005;Smolcˇic´ 20
6
fFruormtheorpmtiocarell,ytthheicFkRreCgihoanssobfeeannifnotuenndseltyoSbFeGc(oKnestnannitcuattt1z9<98)2. MetUal−. 2M00B8()d.uTshtecoPreraercstoedn)croersrte-flraatimoneccooleofufirciiesn∼t0o.f65D,nu(4si0n0g0)spaencd- by gu
(Sargent et al. 2010), though, more recently, studies indicated a troscopicallyconfirmedgalaxieswhichhave(i)areliablemeasure- es
mildevolutionontheslopeof(1+z)γ (Maoetal.2011;Bourne mentofDn(4000),(ii)a3σ detectioninMIPS,and(iii)whichwere t on
et al. 2011; Schleicher & Beck 2013), where γ = −0.12 ± 0.04 matchedtoasignificant(>4σ)radiosource.Becauseofthisstrong 1
1
(Magnellietal.2015).Welackmid-IRorFIRobservationinsome correlationandourinabilitytomeasureDn(4000)foreveryradio M
of our cluster fields to derive TIR and thus are unable to adopt galaxy,weusetherest-frameMU−MBcolourinstead. arc
this method for the full sample. As a solution to this problem, Wecalibratethecolour-SRLseparationwithqTIR classification. h 2
weutilizeMIPS24µmobservations fromSC1604,SG0023,and WeutilizeMIPSmid-IRobservationsfromSC1604,whichisthe 02
3
RXJ1821,toderiveq ,andtodevelopatwo-stageclassification onlyfieldwithdeep,wide-fieldMIPSobservationamongstthefive
TIR
thatreliesonL andcolour–stellarmassnormalizedradiolu- fields, and two narrow-field (7arcmin × 7arcmin) observations
1.4GHz
minositydensity(colour–SRL),usingtheM(U)−M(B) as fromSG0023andRXJ1821.Tomaximizeoursamplepopulation,
rest-frame
colour versus L modulated by the galaxy stellar mass. We wetakeallradiosourceswithhigh-qualityspectroscopiccounter-
1.4GHz
obtain107radiogalaxiesfromradio-specmatchesatallredshifts partsintheentirefieldwhichalsohave3σ 24µmdetections.We
whichhave24µmdetectionsat≥3σ (seeSection3.2),notingthat obtain107galaxiesfromwhichtoobtainourclassificationrulesfor
thesegalaxiesarenotnecessarilyconfirmedtotheLSSs. the colour–SRL method via qTIR equation adopted from Lemaux
etal.(2014b):
(cid:4) (cid:5) (cid:4) (cid:5)
4.1 Radioluminositycriteria q =log 3.939×1026LTIR −log L1.4GHz . (2)
TIR 3.75×1012W WHz−1
We set the primary criteria to be that radio galaxies with
log(L ) > 23.8 are classified as AGNs. Studies have found Themaincontributiontotheuncertaintyonq comesfromthe
1.4GHz TIR
thatalmostallgalaxiesatz∼1withradiopowerabovethisthresh- spectraindex(α),whichistypicallyintherangeof0.4≤α≤1.0
old are AGN, and is consistent with the maximal radio output of for80percentofalltypesofgalaxiesintheredshiftrange0<z<4
MNRAS472,998–1022(2017)
RadiogalaxiesinLSSsatz∼1 1005
D
o
w
n
lo
a
Figure1. qTIRhistogram.Including107radioobjectswithspectroscopic de
aconldoumreidd-IsRolicdoulinntee,rpcaerntstr.eFditatitnqgTbIRy=ad1o.5u0ble(σ-G=aus0s.i6a0n)manoddeqlT,IpRlo=tte2d.2i5n d from
(reσgi=on0s.2ar1e),dmesaigrkneadteadstothbeevAerGtiNca,lHdyabsrhiedd,alinndesS.FTGheclaresdsi,figerdeeans,qaTnIRd.blue Ffuingcutrioen2o.fTthheesrteesllta-frrammaessMnoUrm−alMizBed(1co.4rreGcHtezdlfuomridnuosstityexdtienncstiitoynf)oarsalal https
radio-confirmedgalaxies(blackdots)inthefiveLSSs.Theredandblue ://a
(Lemauxetal.2014b),withqTIR tobeaffectedupto20percent. dashedlinesareusedtoseparatevariousradiosub-classesinthecolour– ca
d
However,noneofourresultsaresensitivelychangedatthislevel. SRL scheme. Dots with orange open pentagons are radio galaxies with e
m
inBFyigfi.t1ti,nwgeaadroeubableleGtaoussseipaanramteodrealditootghaelaqxTiIeRshiinsttoogAraGmNshdoowmn- lmoogn(Ld1s.)4GarHez)co>lo2u3re.8d.bDyottshewiritqhToIRpecnlamssairfikceartsio(nsqmuaertehso,dt,riaanndglaerse,aunsedddsiae-t ic.ou
thecolour–SRLcriteria. p
inatedandstar-formation-dominatedregions.Radiogalaxieswith .c
thehigherq Gaussianfitting,centredatq =2.25(σ =0.21), om
representtheTIRSF-dominatedpopulation,whiTlIeRtheGaussianfitting Tsaambplele5i.neNaucmhbLeSrSos.f AGNs, Hybrids, and SFGs in the radio confirmed /mn
with lower qTIR, centred at 1.50 (σ = 0.60) includes the AGN- ras
dwoimthi<naqted p>op=u2la.1ti6o±n.0O.0u8rasttazr∼fo1r,muastiinognaGcaoumspsilaentefisttteilnlagr-amgraesess- Field Total AGNs Hybrids SFGs /artic
selectedT(IMR∗≥1010M(cid:5))sampleofSFGsat0<z<2.3(Magnelli SC1604 41 14 17 10 le/4
SG0023 8 2 3 3 7
et al. 2015). Since there is a wide overlap between the AGN and 2
lSoFwreergGioanuss,swiaen)c,hSoFoGsestAoGhNavetoqhav≥eq2T.I2R5<,an1d.5H0y(bthriedcsetnotpreopoufltahtee SRCX1J3127457 235 112 61 81 /1/998
theintermediateregionof1.50<TIRq <2.25. RXJ1821 12 4 5 3 /40
Plotting these 107 radio galaxieTsIRin the rest-frame MU − MB Total 90 33 32 25 622
wbveeertwssueestenLth1te.4hGesHetzph/arMerae∗ti(roMand(cid:5)iloi)npe(oss,epeumlFaatirigko.end2s)a,cslwadsesaifsfioheuednddbliyaneqcslTeIiRan.rTFseihgpe.arre2af,toifrooenr, tihfewseeicnosntedadstadgroepatshewrealld.ioThluums,inoousritryescurlittseraiarefolarrAgeGlyNucnlaasffseifictcead- 06 by gu
e
sfuerrvtheesrausstehaesgarocluansdsitfircuathtiofonrctrhiteerriaadiinotchleasfsuilfilcraatdiioons,awmepcleo.uIlfdqtTeIsRt tviaolnu.eWofemfoasusn-dwethigahttaedferwadiSoFlGumsiannodsiHtyy,vbarildusesaareslparregseeanstmatanlayrgoef st on
therelevanceofouralgorithmbyadoptingprecisionandrecall.4 AGNs.Theselargevaluesare,however,theresultofthesegalaxies 11
havinglowerstellarmassvaluesthantheirradioAGNcounterparts. M
a
Weshouldnotethatspectroscopicincompletenessmightaffectour rc
h
4.3 ResultofAGN/Hybrid/SFGclassificationandbias radio sample, specifically we are likely underestimating the per- 2
0
Using the above two-stage classification, we classified our radio centageofSFGsintheradiosample.Howeverthiswillnotaffect 23
ourconclusionsaswewilldiscussinSection5.
galaxysampleintothreesub-samples:AGN,Hybrid,andSFG.This
classificationisintendedtoidentifytheprocessdominatingtheirra-
dioemissionanddoesnotnecessarilyexcludetheexistenceofother 4.4 X-raycrossmatch
sub-dominant components. We show the classification result in a
Cross-matchingradiogalaxieswithX-rayconfirmedAGNs,drawn
colour–SRLdiagraminFig.2.Thenumberofeachsub-classare
from Rumbaugh et al. (2016), we found a total of five out of 89
summarizedinTable5.Wenoticethat95percentofAGNselected
inthefirststagewithlog(L )>23.8areclassifiedasAGNin radiogalaxieshaveX-raycounterparts(seeTable6).Threeofthese
1.4GHz
areclassifiedasAGNsinourtwo-stageradioclassificationscheme,
consistentwithAGNemittingatbothwavelengthregimesandin
linewithotherstudies(e.g.Hickoxetal.2009;Lemauxetal.2014b).
4Theprecisionisthefractionofthenumberofsourcesbeingselectedina
Interestingly, one of the X-ray AGN is classified as Hybrid and
givensub-classinbothqTIR andcolour–SRLoverthenumberofsources
selectedusingcolour–SRLforthatsamesub-class.Recallisthefractionof another one as SFG. This may be explained by the host galaxy
thesamenumeratorintheprecisionoverthenumberofsourcesselected having X-ray bright in nucleus activity, while the radio emission
usingqTIRforthatsamesub-class. is from concurrent star formation activity. It is also possible that
MNRAS472,998–1022(2017)
1006 LShenetal.
Table6. RadioandX-raycross-match.
Field Radio RA Dec. Radio X-ray X-rayfull log(η)b Colour Typed
ID power IDa Luminositya offsetc
SC1604 502 241.0647 43.1713 23.28 0 6.269E+43 0.871 0.038 Hybrid
SC1604 568 241.1076 43.2126 23.67 2 2.111E+43 0.584 0.160 AGN
SC1604 681 241.1567 43.1494 24.30 6 6.175E+42 0.871 0.038 AGN
RXJ1821 198 275.2819 68.3941 23.72 2 6.240E+42 −0.418 0.336 AGN
RXJ1821 237 275.3494 68.4424 23.15 0 7.728E+42 0.215 0.375 SFG
aX-rayIDandfull-bandluminosityadoptedfromRumbaughetal.(2016).
blog(η)parameterisdescribedinSection5.5.
cColouroffsetparameterisexplainedinSection5.1.
dThetypeofradiogalaxyisbasedonourtwo-stageclassificationscheme.
D
Table7. Summaryofaveragepropertiesofthethreeradiopopulations. o
w
n
Radio (cid:11)Colour(cid:12)a Colour log((cid:11)M∗/M(cid:5)(cid:12))c log((cid:11)η(cid:12))d log((cid:11)1+δgal(cid:12))e log((cid:11)L1.4GHz(cid:12)) loa
type offsetb de
d
AGN 4.62 0.10 11.16±0.03 −0.74 0.67 24.01±0.05 fro
m
Hybrid 3.38 -0.74 10.74±0.07 −0.55 0.58 23.21±0.02 h
SFG 3.36 -0.71 10.95±0.03 −0.51 0.48 23.21±0.02 ttp
s
aMedianvalueofrest-frameMNUV−Mr. ://a
bMedianvalueofcolouroffsetparameter,derivedinSection5.1. ca
cWithstandarderrorofthemedianlisted. de
dη=Rproj/R200×|(cid:9)ν|/σν,seeexplanationinSection5.5. mic
eSeeexplanationonlog(1+δgal)inSection5.5. .ou
p
.c
Table8. Summaryofp-valuesfromK-Stest. o
m
/m
Styapmeaple–sample Colourb Cofoflsoeutcr log((cid:11)M∗/M(cid:5)(cid:12)) log((cid:11)η(cid:12))d log((cid:11)1+δgal(cid:12))e log((cid:11)L1.4GHz(cid:12)) nras
/a
AGN–Hybrid 10−5 10−6 10−3 0.44 0.10 10−11 rtic
AGN–SFG 10−8 10−6 0.01 0.10 0.24 10−9 le/4
Hybrid–SFG 0.18 0.09 0.05 0.66 0.91 0.93 72
Hybrid–mixed6 10−6 0.02 10−4 0.09 0.08 10−4 /1
Radio–spec 0.004 0.006 10−14 0.49 0.15 – /99
8
aTwocomparedgalaxysamples. /40
bRest-frameMNUV−Mr. 622
cColouroffsetparameter,derivedinSection5.1. 0
6
dη=Rproj/R200×|(cid:9)ν|/σν,seeexplanationinSection5.5. by
eSeeexplanationonlog(1+δgal)inSection5.5. gu
fAsampleofmixtureofAGNandSFGs,drawnwiththesamenumberofgalaxiesintheHybridpopulationfromAGNandSFGpopulationsfor100trials, e
s
allowingtherelativenumberofAGNandSFGtovaryineachdraw. t o
n
1
1
X-ray dominated in one stage of the AGN evolution, while radio thehostgalaxiesaresignificantlydifferentamongthesethreeradio M
dominatedinalaterstage(Hickoxetal.2009). galaxypopulations. a
rc
h
2
0
2
5 RADIO GALAXY AND HOST PROPERTIES 3
5.1 Radiogalaxycolours
Withtheseclassificationsinplace,wearenowabletobeginadeep
analysisofdifferentpopulationsofradiogalaxiestocomparethe We use a two-colour selection technique proposed by Williams
propertiesofAGNs,Hybrids,andSFGs.Wehavecombinedallradio et al. (2009) to divide the galaxies into two categories: qui-
galaxiesinthefiveLSSforacomplete,statisticalanalysisofradio escent and star-forming (SF). We adopt the rest-frame of
galaxies across a variety of environments in the redshift range of M − M versus M − M colour–colour diagram following
NUV r r J
0.65≤z≤0.96.Witharelativelysmallredshiftrange,wechoose separationlinesfromLemauxetal.(2014b),wheregalaxieswith
to ignore redshift-driven evolutionary effects. In this section, we M −M >2.8(M −M)+1.51andM −M >3.75are
NUV r r J NUV r
showthepreferencesofthethreesub-classesofradiohostgalaxies consideredquiescent.Weshowthecolour–colourdiagramforspec-
in terms of optical colour, stellar mass, radio luminosity, spectral troscopicallyconfirmedmembers(blackdots)andradiogalaxies,
properties,andenvironments.Wesummarizemedianvaluesofthese markedwithcolouropensquaresdependingontheirradiotype,in
properties for the three radio populations in Table 7 and present Fig.3,toppanel.Theadvantageofcolour–colourdiagrams,rather
theresultsofKolmogorov–Smirnovstatistic(K-S)testscomparing than colour–magnitude, is that dusty SFGs move to the right top
eachpairofpopulationsforallpropertiesinTable8.Wefindthat along a diagonal axis, instead of mixing with quiescent galaxies.
MNRAS472,998–1022(2017)
RadiogalaxiesinLSSsatz∼1 1007
selection criteria, although the radio luminosity criteria does not
limitradioAGNhoststobeonlyredgalaxies.Inaddition,wenote
thatredderindust-correctedM −M doesnotnecessarilyrequire
U B
galaxiestobeclassifiedasquiescentintheM −M/M −M
NUV r r J
colour–colourdiagram,asevidencedbythelargenumberofHybrid
galaxieswhicharesituatedintheSFlocusofthisdiagram.SFGs
are narrowly displayed in the dusty active region, with none of
theSFGslocatedinthequiescentregionconfirmingthatSFGsin
radiosamplesrepresentadustySFGpopulation.Thecolouroffset
histogramofHybridhostgalaxiesshowsthatmostarelocatedin
theSFregionandextendstolowervaluesthanthatofSFGs.
WeemployedtheK-StestontheAGN,Hybrid,andSFGsamples
to test whether any two are consistent with being draw from the
same distribution. The two tailed p-value result of the K-S test
D
givestheprobabilitythattwodistributionsaredrawnfromthesame o
w
distribution.Inthispaper,weadoptthatifthep-value>0.1,we n
lo
cannot reject the hypothesis that the two distributions are drawn a
d
from the same distribution. Otherwise, we say the probability of ed
drawingfromthesamedistributionisverysmall.Wenotethatnone fro
oftheresultswillchangemeaningfully ifweusep-value=0.05 m h
asthethreshold.Thep-valuebetweenthecolouroffsetdistribution ttp
of AGNs and SFGs (or Hybrids) is ∼10−6, which confirms that s://a
AGNhostgalaxiesaredifferentfromHybridsandSFGs.Thehost c
a
galaxiesofSFGandthoseofHybridpopulationsaresimilar,with de
p-value of ∼0.1. Therefore, here, we are not able to confirm a mic
differencebetweenHybridsandSFGs. .o
u
WealsotestedwhethertheHybridpopulationcouldbeamixture p.c
o
ofAGNandSFGs.Wedrew100sampleshavingthesamenumber m
ofgalaxiesaswereintheHybridpopulationfromAGNandSFG /m
n
population,allowingtherelativenumberofAGNandSFGtovary ra
s
ineachdraw.WerantheK-Stestforeachsampleandreturnedthe /a
FMirg−ureM3J.).DTootps:athreesrpeesct-tfrroasmcoepciocalolluyr–ccoonlfiorumreddiamgreammb(eMrsNiUnVth−efiMvrevLeSrsSuss. mofotdheeopf-vpa-lvuaelsu)e.sT(hseeereAspuplteanndtivxalAuefworasth0e.0d2e,fisntritoinognlyofsuthgegemstoindge rticle/4
Radiogalaxiesaremarkedwithopensquares,colouredbytheirtype.The that the Hybrid population is not a mixture of AGN and SFGs, 72
solidlinesshowtheadopteddivisionsbetweentheSFandquiescentgalaxy theoneexceptionbeingthefewcaseswhereSFGsdominatedthe /1/9
populations.Theleftupperregionisthequiescentregion,andtherightlower comparisonsample. 98
regionistheSFregion.Thearrowrepresentsthemovementinthephase However, if we want to compare the percentage of each radio /40
spacewhenastellarcontinuumextinctionofEs(B−V)=0.3isapplied sub-class of the total radio sample, we may be biased by the in- 62
(cmbCoyenamtlfihzrbeemettnreisude)m,tmaabnlee.dmr20rboa0efd0rirs)oa.(-dBbciloooan-tcctfikoormnlmifin1ered:m,csgeocadlalolaguexadrileoabsfxyf(isethehsate)th.ctihoBsettooadtlgtcnoroaummlmo2ubo:refrhcsopiosleftocosutgprrroeacosmctfrfoo,spessictccoaahpllielsiydc- cMseonNmUtapVtliev−teenwMehsrsecnoofMlooNuuUrr,Vstph−eecMstrpores≤cctor2op.si5cco(spsueircevesmya,omsrpeinledceeb,taeaiclcocomonredtseinsugtninrtegopttrhheee- 206 by gue
tpohgortaommeotfripchmoteommbeetrrisc),maenmdbraedrsio(-bplhaocktslainmep,lseca(lheadtcbhyedthceotlootuarlhniusmtobgerarmof, sthpeectwtrooshciosptoicgrraemprseisnenthtaetibvoetntoemsspinanAelpopfeFndigix.3B,w).eInpltohtethleowcoelrouorf st on
scaledbythenumberofradio-photsample).Positivevaluesrepresentobjects offsethistogramforphotometricmembersandthethreesub-classes 11
inthequiescentregion. fromtheradio-photsample,usingthesameclassificationscheme M
a
(AGN-phot,Hybrid-phot,andSFG-phot).AsshowninFig.3,there rc
h
Thisisatypicallimitationwhenperforminganalysissolelywitha isabimodaldistributionforphotometricmembers,howevertheSF 2
0
colour–magnitudediagram. peakisabsentforthespectroscopicallyconfirmedmembers.This 23
We generate a colour offset histogram in the bottom panel of isconsistentwithourspectroscopicsamplenotbeingrepresenta-
Fig.3,withtheupperofthetwohistogramsfortheradio-confirmed tiveofallgalaxiesatbluercolours.Asforradiogalaxies,although
members and the lower for the radio-phot sample, where ‘colour addingthez selectedradiogalaxiesincreasesthetotalradiosam-
phot
offset’istheperpendicularoffsetfromquiescentandSFseparation ple,thecoverageanddistributionofeachradiosub-populationin
line,withpositiverepresentinggalaxyinthequiescentregion.With thecolouroffsetdoesnotchange.Thisindicatesthatspectroscopic
thehelpofthecolouroffset,weareabletoquantitativelyexamine incompletenessdoesnotbiasourresultsfordifferenttypesofra-
thedistributionofradiosub-classeslocatedinthe2Dcolour–colour diogalaxies.Hence,weconfirmthatAGNsareredandquiescent
diagram.Intheradioconfirmedmemberscolouroffsethistogram, galaxies,SFGsarebluer,activeSFGs,andHybridsaredispersedin
weseeaseparationbetweenthepopulationofAGNhostgalaxies colour,thoughwithaslightpreferencetolieintheSFregion.
andSFGhostgalaxies.72percentoftheAGNsamplehascolour
offsetabovezero(i.e.inthequiescentregion).Asmallfractionof
5.2 Stellarmass
theAGNinhabitingtheSFregion,andmostofthemareclosetothe
dividingline(colouroffset<−0.5).AGNsarefoundtobehosted The phenomenon of ‘mass quenching’ results in galaxies with
largely by red, quiescent galaxies, as we expect from our second higher stellar masses being observed with higher quiescent
MNRAS472,998–1022(2017)