The XXL Survey. XX. The 365 cluster catalogue

30 Pages·2017·3.56 MB·English

A&A620,A5(2018) Astronomy
https://doi.org/10.1051/0004-6361/201731606 &
©ESO2018 Astrophysics
The XXL Survey: second series Special issue
The XXL Survey
(cid:63),(cid:63)(cid:63)
XX. The 365 cluster catalogue
C.Adami1,P.Giles7,E.Koulouridis3,F.Pacaud4,C.A.Caretta1,6,M.Pierre3,D.Eckert9,M.E.Ramos-Ceja4,
F.Gastaldello8,S.Fotopoulou46,V.Guglielmo1,2,31,C.Lidman11,T.Sadibekova3,A.Iovino29,B.Maughan7,
L.Chiappetti8,S.Alis41,B.Altieri32,I.Baldry20,D.Bottini8,M.Birkinshaw7,M.Bremer7,M.Brown21,
O.Cucciati27,34,S.Driver18,19,E.Elmer15,S.Ettori27,42,A.E.Evrard25,26,L.Faccioli3,B.Granett29,33,M.Grootes24,
L.Guzzo29,33,A.Hopkins11,C.Horellou28,J.P.Lefèvre3,J.Liske22,K.Malek36,F.Marulli27,34,35,
S.Maurogordato12,M.Owers11,17,S.Paltani9,B.Poggianti2,M.Polletta8,37,40,M.Plionis10,45,A.Pollo36,38,
E.Pompei5,T.Ponman16,D.Rapetti43,44,M.Ricci12,A.Robotham18,19,R.Tuffs23,L.Tasca1,I.Valtchanov32,
D.Vergani39,G.Wagner13,14,J.Willis30,andtheXXLconsortium
(Affiliationscanbefoundafterthereferences)
Received20July2017/Accepted27November2017
ABSTRACT
Context. Inthecurrentlydebatedcontextofusingclustersofgalaxiesascosmologicalprobes,theneedforwell-definedclustersamples
iscritical.
Aims. The XXL Survey has been specifically designed to provide a well characterised sample of some 500 X-ray detected clusters
suitableforcosmologicalstudies.Themaingoalofpresentarticleistomakepublicanddescribethepropertiesoftheclustercatalogue
initspresentstate,aswellasofassociatedcataloguesofmorespecificobjectssuchassuper-clustersandfossilgroups.
Methods. FollowingfromthepublicationofthehundredbrightestXXLclusters,wenowreleaseasamplecontaining365clustersin
total,downtoafluxofafew10−15ergs−1cm−2inthe[0.5–2]keVbandandina1(cid:48)aperture.Thisreleasecontainsthecompletesubset
ofclustersforwhichtheselectionfunctioniswelldeterminedplusallX-rayclusterswhichare,todate,spectroscopicallyconfirmed.
Inthispaper,wegivethedetailsofthefollow-upobservationsandexplaintheprocedureadoptedtovalidatetheclusterspectroscopic
redshifts.ConsideringthewholeXXLclustersample,wehaveprovidedtwotypesofselection,bothcompleteinaparticularsense:one
basedonflux-morphologycriteria,andanalternativebasedonthe[0.5–2]keVfluxwithin1arcminoftheclustercentre.Wehavealso
providedX-raytemperaturemeasurementsfor80%oftheclustershavingafluxlargerthan9×10−15ergs−1cm−2.
Results. Ourclustersampleextendsfromz∼0toz∼1.2,withoneclusteratz∼2.Clusterswereidentifiedthroughameannumber
of six spectroscopically confirmed cluster members. The largest number of confirmed spectroscopic members in a cluster is 41. Our
updated luminosity function and luminosity–temperature relation are compatible with our previous determinations based on the 100
brightestclusters,butshowsmalleruncertainties.Wealsopresentanenlargedlistofsuper-clustersandasampleof18possiblefossil
groups.
Conclusions. ThisintermediatepublicationisthelastbeforethefinalreleaseofthecompleteXXLclustercataloguewhentheongoing
C2clusterspectroscopicfollow-upiscomplete.Itprovidesauniqueinventoryofmedium-massclustersovera50deg2areaouttoz∼1.
Keywords. galaxies:clusters:general–large-scalestructureofUniverse–galaxies:groups:general–
galaxies:clusters:intraclustermedium
1.Introduction
(cid:63)Based on observations obtained with XMM-Newton, an ESA sci-
Mostgalaxycluster-relatedcosmologicalprobesrelyoncluster
ence mission with instruments and contributions directly funded by
ESAMemberStatesandNASA.BasedonobservationsmadewithESO numbercountsandlarge-scalestructureinformation.X-raysur-
TelescopesattheLaSillaandParanalObservatoriesunderprogrammes veys have had a key role in this framework since the historical
ID 191.A-0268 and 60.A-9302. Based on observations obtained with Einstein observatory Medium Sensitivity Survey (Gioia et al.
MegaPrime/MegaCam, a joint project of CFHT and CEA/IRFU, at 1990). Many other surveys were conducted with the ROSAT
theCanada-France-HawaiiTelescope(CFHT)whichisoperatedbythe observatory, and more recently, XMM-Newton and Chandra
NationalResearchCouncil(NRC)ofCanada,theInstitutNationaldes
Sciences de l’Univers of the Centre National de la Recherche Scien- CNRS. This research has made use of the VizieR catalogue access
tifique (CNRS) of France, and the University of Hawaii. Based on tool,CDS,Strasbourg,France.Thisresearchhasalsomadeuseofthe
observations collected at the German-Spanish Astronomical Centre, NASA/IPAC Extragalactic Database (NED) which is operated by the
CalarAlto,jointlyoperatedbytheMax-Planck-InstitutfürAstronomie Jet Propulsion Laboratory, California Institute of Technology, under
HeidelbergandtheInstitutodeAstrofísicadeAndalucía(CSIC).This contractwiththeNationalAeronauticsandSpaceAdministration.
work is based in part on data products produced at Terapix available (cid:63)(cid:63)Full Table 5 is only available at the CDS via anony-
attheCanadianAstronomyDataCentreaspartoftheCanada-France- mous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or via
Hawaii Telescope Legacy Survey, a collaborative project of NRC and http://cdsarc.u-strasbg.fr/viz-bin/qcat?J/A+A/620/A5
ArticlepublishedbyEDPSciences A5,page1of28
A&A620,A5(2018)
Table1.StatisticsoftheXXL-365-GC,XXL-C1-GC,andXXL-100-GCsamples.
Sample Selection N C1+C2+C3 N C1 N C2 N C3
XXL-365-GC AllC1clusters 365(341) 207(183) 119(119) 39(39)
+spectros.C2/C3
XXL-100-GC 100brightestclusters 100(99) 96(95) 4(4) 0
Notes.Numberswithinparenthesesarethenumbersofspectroscopicallyconfirmedclustersfortheconsideredselection.
produced surveys such as the XMM-LSS, XMM-COSMOS, cluster luminosity function and of the luminosity–temperature
XMM-CDFS and Chandra-Ultra-Deep surveys (Pierre et al. relation. The results of spatial analyses performed on the clus-
2004; Hasinger et al. 2007; Comastri et al. 2011; Ranalli et al. ter catalogue (search for super-clusters and fossil groups) are
2013).Followingthispath,itisnowclearthatclustercosmolog- presented in Sect. 6. Notes on the newly detected structures
icalstudiescanonlyberigorouslyperformedbysimultaneously and recent redshift measurements are gathered in Appendices.
fitting a cosmological model, the cluster selection function and Throughout the paper, for consistency with the first series of
the physical modelling of the cluster evolutionary properties in XXLpapers,weadopttheWMAP9cosmology(Hinshawetal.
whichever band the cluster selection has been performed (e.g. 2013, with Ωm = 0.28, ΩΛ = 0.72, and H0=70kms−1Mpc−1),
Allen et al. 2011). X-ray cluster cosmology is especially well except if explicitly stated. From the semantic point of view, we
suited to such an approach, because the properties of the X-ray also mention that the structures called clusters in the present
emitting intra-cluster medium can be ab-initio predicted with paperarenotverymassivestructures,butareintermediate-mass
goodaccuracy,eitherusingananalyticalmodelorbymeansof concentrationsinthemassrangebetweengroupsofgalaxiesand
hydrodynamicalsimulations. verymassiveclustersofgalaxies.
The XMM-XXL project (XXL hereafter) covers two areas
of 25 deg2 each with XMM-Newton observations to a sensitiv-
2.SelectionoftheX-rayclustersample
ityof∼5×10−15ergs−1cm−2 (forpointsources);thetwoareas
are centred at: XXL-N (02h23(cid:48) −04◦30(cid:48)) and XXL-S (23h30(cid:48) TheX-raypipelineandtheclusterselectionprocedurealongwith
−55◦00(cid:48)).Inafirststep,XXLaimsatin-depthclusterevolution- the XXL selection function are extensively described in XXL
arystudiesoverthe0 < z < 1rangebycombininganextensive PaperII.Werecallherethemainsteps.
data set over the entire electromagnetic spectrum. In a second Our detection algorithm (the same version of Xamin used
andultimate stepwe aimata standalonecosmological analysis in XXL Paper II, cf. also Faccioli et al. 2018, hereafter XXL
(Pierreetal.2016,hereafterXXLPaperI)andtheX-raycluster Paper XXIV) enables the creation of an uncontaminated (C1)
catalogueconstitutesthecoreofthewholeproject:itsconstruc- clustersamplebyselectingalldetectedsourcesinthe2D[EXT;
tionalongwiththedeterminationoftheclustermultiwavelength EXT_STAT] output parameter space. The EXT parameter is a
parametersfollowsaniterativeprocessdemandingspecialcare. measure of the cluster apparent size and the EXT_STAT param-
Inthisprocess,thespectroscopicconfirmationoftheX-rayclus- eter quantifies the likelihood of a source of being extended.
tercandidateshasoccupiedacentralplaceintheprojectoverthe The EXT_STAT likelihood parameter is a function of cluster
last5yr.Inafirstpublication(Pacaudetal.2016,hereafterXXL size, shape and flux. This parameter depends on the local
Paper II) we presented the hundred brightest galaxy clusters XMM-Newtonsensitivity.
(XXL-100-GC) along with a set of preliminary scientific anal- Simulations enable the definition of limits for EXT and
yses,includingtheX-rayluminosityfunction,spatialcorrelation EXT_STAT above which contamination from point sources is
studiesandacosmologicalinterpretationofthenumbercounts. negligible, providing the C1 sample. Relaxing slightly these
The present, and second, release is the last before the publica- limits, we define a second, deeper, sample (C2) to allow for
tionofthecompleteclustercatalogue.Thiswilloccurwhenthe 50% contamination by misclassified point sources; these can
ongoing C2 cluster spectroscopic follow-up is completed. The easily be cleaned up a posteriori using optical versus X-ray
main goal of present article is to make public and describe the comparisons. Initially, the total number of such C2 cluster
properties of the second release, as well as of associated cata- candidates was 195 and more than 60% are already spectro-
loguesofmorespecificobjectssuchassuper-clustersandfossil scopicallyconfirmed(seebelow).Wedefinedathirdclass,C3,
groups.Thepresentsamplecontainsthecompletesubsetofclus- corresponding to (optical) clusters associated with some X-ray
tersforwhichtheselectionfunctioniswelldetermined(namely, emission, too weak to be characterised; the selection function
theC1selection)plusallX-rayclusterswhichare,todate,spec- of the C3 sample is therefore undefined. Initially, most of the
troscopically confirmed. The C1 and C2 classes are defined as C3 objects were not detected in the X-ray waveband and are
in XXL Paper II and will be described below. Altogether, this located within the XMM-LSS subregion. We refer the reader
amounts to 365 clusters and is referred to as the XXL-365-GC to Pierre et al. (2004) for a more detailed description of these
sample (cf. Table 1). Along with the cluster list itself, we pro- classes.
videanupdateoftheX-rayclusterpropertiesandoftheirspatial With the present paper, we publish all C1 clusters (XXL-
distributionaspresentedinthe2016XXL-100-GCpublications. C1-GC hereafter, cf. Table 1) supplemented by the C2 and C3
Theclusterparametersderivedinthepresentpublicationsuper- clusterswhicharespectroscopicallyconfirmed.C3clusterswere
sede the XXL-100-GC ones, even thought the consistency (see not specifically targeted, but were sometimes confirmed as by-
below)isverygood. productsofexistinggalaxyspectroscopicsurveys.Table2gives
In the next section, we describe the construction of the statisticsoftheXXL-365-GCsampleintermsofC1,C2,andC3
currentsample.Section3givesadetailedaccountofthespectro- clusters.Thisamountsto207C1(amongthem,183spectroscop-
scopicvalidationprocedure.Wepresenttheclustercataloguein icallyconfirmedtodate,4withsomespectroscopybutneeding
Sect.4.Section5providesupdateddeterminationsoftheX-ray moredata,13withaphotometricredshift,and7withoutredshift
A5,page2of28
C.Adamietal.:TheXXLSurvey.XX.
Table2.StatisticsoftheXXL-365-GCsampleintermsofC1,C2,and Table3.DetailsofthethreeESOPIruns.
C3clusters.
ESOId Instrument Duration SemestersNb
Classes XXL-365-GC Spect. ≥3redshifts
191.A-0268 FORS2 132h 4
C1 207(114/93) 183(105/78) 160(96/64) 191.A-0268 EFOSC2 15n 4
C2 119(59/60) 119(59/60) 70(42/28) 089.A-0666 FORS2 15h 1
C3 39(38/1) 39(38/1) 31(31/0) 60.A-9302 MUSE 3h 1
All 365(211/154) 341(202/139) 261(169/92)
Notes. Col. 1: considered classes. Col. 2: numbers within the total
when possible. The targets were chosen in order to favour the
XXL-365-GCsample.Col.3:numbersofspectroscopicallyconfirmed
cluster confirmation by galaxies within the X-ray contours. We
clusterswithintheXXL-365-GCsample.Col.4:numbersofspectro-
note that the X-ray contours are created from a wavelet filtered
scopicallyconfirmedclusterswithatleastthreespectroscopicredshifts
photonimage.Thecontoursarerunineachframefortherange
within the XXL-365-GC sample. Numbers within parentheses are for
thenorthernandsouthernareas. between0.1cts/pxcorrespondingtothetypicalbackgroundlevel
forexpositiontimeof10ks(∼10−5cts/s/px)andamaximalvalue
intheframespacedby15logarithmiclevels.
estimation),119C2and39C3.TheC1selectionprovidesacom-
(a) We made extensive use of the ESO optical facilities
pletesampleinthetwo-parameterspaceoutlinedabove.Inorder
(NTT/EFOSC2andVLT/FORS2).WeweregrantedthreePI
to allow straightforward comparisons with different X-ray pro-
allocations,includingaLargeProgramme(191-0268)anda
cessingmethods,wegive,forinformationonly,theapproximate
pilotprogramme(089.A-0666).Wegivethedetailsofthese
completeness flux limit of the XXL-365-GC sample computed
newPIESOprogrammesinTable3.
from simulated detections. We performed the measurements
FORS2 and EFOSC2 galaxy targets were first choosen
within a radius of 1arcmin around the cluster centre (defined
accordingtotheirstrategicalplaceinsidetheclusters,taking
from the X-ray data). We assume, as in XXL Paper II, that
intoaccountthealreadyknownredshiftsfromothersurveys,
the XMM-Newton count-rates are computed in the [0.5–2]keV
and their location regarding the X-ray contours. Then, we
bandandconvertedintofluxesassuminganEnergyConversion
putasmanyslitsaspossibleonotherobjects.Wemeasured
Factor(ECF)of9.04×10−13ergs−1cm−2/(cts/s).Thecomplete-
thespectroscopicredshiftsbymeansoftheEZcode(Garilli
nessfluxlimit(the100%completenessfluxlimitaveragedacross
et al. 2008) that was already used for the VIPERS sur-
the entire survey area) is then ∼1.3×10−14ergs−1cm−2. We
vey (Guzzo et al.2014; Scodeggio et al. 2017). We adopted
emphasisethatsinceafluxof10−14correspondsto∼100photons
the same approach: the only operation that required human
on-axisfor10ksexposures(MOS1+MOS2+PN),uncertainties
interventionistheverificationandvalidationoftheEZmea-
are large, which may affect the cluster ranking as a function of
suredredshift.Eachspectrumisindependentlymeasuredby
thefluxby10%ormore.
two team members. At the end of the process, discrepant
redshifts are discussed and homogenised. The quality of
the redshift measurements is defined as in the VVDS and
3.Spectroscopicredshifts
VIPERSsurveys:
3.1.Collectingthespectroscopicinformation – Flag0:noreliablespectroscopicredshiftmeasurement;
– Flag 1: tentative redshift measurement with a ∼50%
The spectroscopic surveys conducted on the XXL fields are
chancethattheredshiftiswrong.Theseredshiftsarenot
listed in XXL Paper I (Table 3). We provide below a short
used;
descriptionofthisratherheterogeneousdataset.Inordertoper-
– Flag2:confidenceestimatedtobe>95%;
formthespectroscopicvalidationandfurtherdynamicalstudies
– Flag 3 and 4: highly secure redshift. The confidence is
of the XXL clusters, all available spectroscopic information on
estimatedtobehigherthan99%;
galaxieslocatedintheXXLfieldshasbeenstoredintheCEntre
– Flag9:redshiftbasedonasingleclearfeature,giventhe
dedonnéeSAstrophysiquesdeMarseille1.Theirastrometrywas
absence of other features. These redshifts are generally
matched with the CFHTLS T0007 catalogue2 for XXL-N and
reliable.
withtheBCScatalogue(Desaietal.2012)forXXL-S.Thepub-
(b) We also made use of the AAOmega instrument on the
lic and private surveys stored in CESAM and relevant to XXL
AAT. A first observing campaign was published in Lidman
aredescribedinthefollowing.Allinall,thetotalnumberofred-
et al. (2016, hereafter XXL Paper XIV), while supplemen-
shiftspresentintheCESAMdatabaseare∼145000and∼8500
taryobservationsdonein2016willbeincludedinChiappetti
fortheXXL-NandXXL-Sfieldsrespectively(asofDecember
etal.(2018,hereafterXXLPaperXXVII).Forthefirstrun,
2016,includingmultiplemeasurements).
cluster galaxies were the prime targets and we used Runz
(Hintonetal.2016)tomeasureredshifts.X-rayAGNinthe
3.1.1.XXLextendedsourcesspectroscopicfollow-up XXL-S field were the prime targets for the second run and
campaigns onlysparefibreswereputonclustergalaxies.WeusedMarz
(Hintonetal.2016)tomeasureredshifts.Foreachspectrum,
We conducted our own spectroscopic follow-up to complement
weassignaqualityflagthatvariesfrom1to6.Theflagsare
the already available public spectroscopic data sets. C1 clus-
identicaltothoseusedintheOzDESredshiftsurvey(Yuan
ters were the primary targets, but we also targeted C2 clusters
et al. 2015). We used AAT quality flags 3 or 4 which are
equivalenttotheESOflags2,3,or4.
(c) We also obtained Magellan spectroscopy at Las Campanas
1 http://www.lam.fr/cesam/ observatory from an associated survey (A. Kremin, priv.
2 http://www.cfht.hawaii.edu/Science/CFHTLS/T0007/ comm.).Weonlyusedthe262mostreliableredshifts.
A5,page3of28
A&A620,A5(2018)
(d) We collected redshifts at the William Herschel Telescope representingagoodcompromisebetweenthespectrographs
(WHThereafter, cf.Koulouridiset al.2016,hereafter XXL resolutionandthepossiblerealdifferencebetweenredshifts,
PaperXII).Redshiftsweremeasuredandqualityflagswere atthe3-σlevel).Thispointstostronglydiscrepantredshifts
assignedinthesamewayasfortheESOdata. for5%ofthesample.Acomparablepercentageisexpected
inGuzzoetal.(2014)fortheVIPERSsurvey.Wetherefore
concludethatoursampleissimilartotheVIPERSsurveyin
3.1.2.RedshiftsfromtheXMM-LSSsurvey
termsofincompatibleredshifts(cf.Scodeggioetal.2017).
WeincludedallredshiftsobtainedfortheXMM-LSSpilotsur- (b) Formeasurementswithin±3×600kms−1,thestatistical1-σ
vey (11 deg2 precursor and subarea of XXL-N, Pierre et al. redshiftscatteris∼0.00049×(1+z).Thisrepresentsalmost
2004).ThesampleisdescribedinAdamietal.(2011). 150kms−1.WenotethatFig.1maygivethefeelingthatthe
dispersion is much larger at low redshifts. However, this is
mainly due to the fact that many objects are concentrated
3.1.3.Literaturedata
along the zero difference level. The statistical 1-σ uncer-
The XXL-N area was defined to overlap with the VIPERS sur- tainty is for example ∼0.00049 at z ≤ 1 and ∼0.00057 at
vey(VIMOSPublicExtragalacticRedshiftsurvey:Guzzoetal. z≤0.5.
2014;Scodeggioetal.2017)andtoencompasstheVVDSsurvey (c) The previous estimates pertain to the full galaxy sample.
(LeFèvreetal.2013).Wethereforeincludedtheredshiftsfrom Wealsoperformedasimilaranalysisontheclustergalaxies
theseVIMOS-basedredshiftsurveys.Theredshiftsaremeasured alone. These galaxies have different types and luminosities
inourownESOspectroscopicfollow-upexactlyinthesameway and are therefore potentially subject to different selections.
asVIPERSandVVDSdid,withthesamequalityflags.Wealso To select these galaxies, we limited the sample to galax-
notethatallredshiftsfromVIPERS,coveringtheredshiftrange ies within one Virial radius and with a velocity within
0.4 ≤ z ≤ 1.2,weremadeavailableforthisanalysispriortothe ±3 × σ , the equivalent galaxy velocity dispersion
v,200
recentpublicrelease(Scodeggioetal.2017). inferredfromscalinglawswithintheVirialradius,fromthe
GAMA, 2dF, 6dF, SDSS: These four catalogues were clustercentre.Wecouldhavetriedtouseinsteadthegalaxy
ingestedandusedwithoutremeasuringtheredshiftofthegalax- velocitydispersioncomputedwithgalaxyredshifts,butour
ies.Theyproviderobustspectroscopicqualityflags.Weconsid- samplingistoosparsetohavepreciseestimations.Thiswill
eredasreliabletheGAMA,2dF,and6dFredshiftswithquality betreatedinafuturepaper.Virialradiusandσ wereesti-
v,200
flags 3 and 4 (e.g. Liske et al. 2015 and Baldry et al. 2014 for mated from X-ray data given in Table F.1 and described in
GAMA, and Folkes et al. 1999 for 2dF), equivalent to the ESO the following. Applying the same method as with the com-
flags 2, 3 or 4. SDSS spectra with “zWarning” between 0 and plete sample, we find an incompatible redshift percentage
16werealsoused.WenotethattheGAMAspectroscopyinside of ∼4% (cf. Fig. 2), even better than for the total sample.
theXXLareaisissuedfromtheGAMAG02fieldwherefibres The1-σredshiftscatteris∼0.00041×(1+z),or120kms−1
werealsointentionallyputonpreliminaryproposedXXLgalaxy in terms of radial velocity uncertainty, also similar to the
targets.G02willbepublicwithintheGAMADR3datarelease estimateforthetotalsample.Finally,wedonotseeanysig-
(Baldryetal.,inprep.). nificantvariationofthe1-σuncertaintybetweenredshifts0
Inaddition,weconsideredothersmallerpublicredshiftcata- and0.9.
logues:Akiyamaetal.(2015)fromSubaru,Simpsonetal.(2006, Thelastissueistoestimatetherelativeweightofthevar-
2012), Stalin et al. (2010), SNLS survey (e.g. Balland et al. ioustelescopesintheclusterredshiftcompilation.Consider-
2009).Weremeasuredandcheckedtheredshiftvaluesforthese ing the sample of cluster galaxies only, we find that ∼45%
surveys, when spectra were available, using the methods devel- are coming from ESO (VIMOS and FORS2 instruments),
oped for our own spectroscopic follow-up. We finally collected ∼45% from AAT (AAOmega instrument), and ∼7% from
and assumed as correct all other redshifts on the XXL areas, SDSS. The remaining ∼3% have various origins (Subaru,
currentlyavailableintheNEDdatabase. WHT,LasCampanas,etc.).
As a remark, for a given object with multiple redshift
measurements,weusedthemeasurementcomingfromthehigh-
3.2.Redshiftreliabilityandprecision
est quality spectrum. We did not notice systematic redshift
Our spectroscopic redshift catalogues come from various tele- differencesintheconsideredsurveys.
scopes, with different instruments, different setups and were
obtained under different observing conditions. We thus needed
3.3.Clusterspectroscopicconfirmation
toevaluateonanobjectivebasistheoverallreliabilityofthedata
set.Althoughwetriedtolimitmultipleobservations,weended StartingfromthelistofextendedX-raysources(C1orC2),the
up with a non-negligible number of galaxies present in differ- clusterspectroscopicconfirmationisaniterativeprocess.
entsurveys.Weusedtheseredundantmeasurementstoevaluate (1) Wefirstcollectedallavailablespectroscopicredshiftsalong
thestatisticalreliabilityofourredshifts.Thesimplestapproach agivenlineofsighttowardsaclustercandidate.Weselected
consists in plotting the redshift difference versus redshift (cf. the spectroscopic redshifts within the X-ray contours and
Fig. 1) for the ∼12000 objects measured twice in the whole searched for gaps larger than 900kms−1 in the resulting
spectroscopic sample. Out of these, 15% had a spectroscopic redshift histogram. This is intended to separate different
quality flag of 4, 61% a quality flag of 3, 24% a quality flag concentrations in the redshift space. We searched for con-
of2,and<1%aqualityflagof9.Weonlyconsiderflags>2in centrations of three or more redshifts between two gaps
thefollowing. and preliminarily assigned the largest concentration to the
(a) To estimate the fraction of incompatible redshifts, we extended source in question. This allows us to estimate the
selectedinFig.1alldoublemeasurementsdifferingbymore angulardistanceofthesourceinquestion.
than±3×600kms−1(600kms−1isatypicalvaluebasedon (2) We then repeated the process, this time within a 500kpc
theVVDSandVIPERSsurveys:cf.LeFèvreetal.2013and radius.Thishassometimesledustoconsiderlargerregions
A5,page4of28
C.Adamietal.:TheXXLSurvey.XX.
Fig.1.Redshiftdifferenceversusredshiftforthe∼12000objectsmea-
sured twice within the spectroscopic survey. The two red dotted lines
represent the ±3×600kms−1 level (cf. Sect. 3.2). We also give the
histogramoftheredshiftdifferencewithinthe[−0.005,0.005]interval.
Fig. 3. Upper panel: y-axis, number of confirmed clusters; x-axis,
number of galaxy redshifts sampling the confirmed clusters. Differ-
ent colours and line styles are from different spectroscopic surveys.
Bottom panel: percentage of galaxy redshifts inside the confirmed
Fig. 2. Redshift difference versus redshift for the galaxies (at <±3×
clusters coming from a given survey and for a given redshift
σ fromtheclustermeanredshiftandwithinoneVirialradius)mea-
v,200 bin. Because of multiple galaxy spectroscopic measurements, the
sured two times within the spectroscopic survey. The two red dotted
sum of the percentages for a given redshift bin is larger than
linesrepresentthe±3×600kms−1level.Wealsogivethehistogramof
100%.
theredshiftdifferencewithinthe[−0.005,0.005]interval.
The C3 clusters – X-ray sources too faint to be charac-
than the ones defined by the X-ray contours. We checked terisedasC1orC2–thatwepresentinthispaperareonly
whethertheinferredredshiftwascompatiblewiththeprevi- those resulting from the spectroscopic follow-up of X-ray
ousone.Ifyes,weconsideredtheclustertobeconfirmedat sources in the XMM-LSS pilot survey. We did not perform
the considered redshift. If not, we restarted the full process anysystematicclustersearchorfollow-upforthefulllistof
withanotherredshiftconcentration.Inpractice,thisprocess X-raysources.
wasconvergentatthefirstpassforthelargemajorityofthe In Fig. 3, we give the contribution of the major spectroscopic
cases. surveysusedinthepresentpaper.Thisisshowedbothinterms
We kept open the possibility of manually assigning a of the number of clusters with a given number of galaxy red-
redshift to a cluster when the two previous criteria did shifts coming from a given spectroscopic survey, and in terms
not agree (cf. below the peculiar case of XLSSC 035). of number of galaxy redshifts coming from a given survey for
This mainly occurred when dealing with projection effects agivenredshiftbin.ThisforexampleshowsthattheXXLESO
along the line of sight (cf. the eight cases in Appendix B). andXMM-LSSPIallocationswereefficienttoconfirmclusters
Some of the lines of sight were however poorly sampled, in the z∼[0.2–1] range while other major surveys were more
with typically fewer than three redshifts. In this case, we specialised in terms of redshift coverage: VIPERS at z ≥ 0.45,
attempted to confirm the cluster nature of the X-ray source and AAT PI and GAMA at z ≤ 0.7 and z ≤ 0.4 respectively.
by identifying the cluster dominant galaxy (BCG here- In terms of cluster spectroscopic sampling, XXL ESO PI allo-
after) in the i(cid:48) band and close to the X-ray centroid. If the cations enabled us to measure the largest number of galaxy
choice of such a galaxy was obvious and this galaxy had redshifts per cluster (∼5); other surveys yielded various sam-
a spectroscopic redshift, we confirmed the cluster as well. plings. The largest samplings are achieved by the XMM-LSS
This was the case for 30 clusters (with only the BCG), spectroscopicsurvey(mostofthetimeforwellidentifiedpecu-
and for another 50 clusters (with the BCG plus another liarordistantclusters)andbytheGAMAspectroscopicsurvey
concordantgalaxy). fornearbyclusters.
A5,page5of28
A&A620,A5(2018)
Table 4. Mean number of redshifts per cluster line of sight from the 4.2.X-raydirectmeasurements
differentsurveysconsideredinFig.3forthetotalXXLSurvey,north,
andsouthfields. 4.2.1.Luminosityandtemperature
Full details of the analysis of the cluster X-ray properties will
Survey XXL XXL-N XXL-S be found (Giles et al., in prep.), and we outline the main steps
ofthespectralanalysishere.First,weonlyusedthesinglebest
XMM-LSS 9 15 1
pointings for spectral analyses when sources fell on multiple
AATPI / 0 5
pointings. As a conservative approach, the extent of the cluster
VIPERS / 16 0
emissionwasdefinedastheradiusbeyondwhichnosignificant
GAMA / 10 0
clusteremissionisdetectedusingathresholdof0.5σabovethe
ESOPI 2 2 3
background level. Due to the low number of counts and low
signal-to-noise ratio (S/N) of many of the clusters below the
XXL-100-GC threshold, we performed a detailed modelling of
Major surveys such as VIPERS or GAMA have science thebackground,insteadofasimplebackgroundsubtraction.We
objectives related to field studies, and are therefore under- followed the method outlined in Eckert et al. (2011), who per-
represented in Fig. 3 because only a small fraction of these formedthisdetailedmodellingtostudyasourcewhoseemission
redshiftsfallswithinagivencluster.WethereforegiveinTable4 barely exceeded the background. We modelled the non X-ray
the mean numbers of redshifts per line of sight (over the full background(NXB)usingclosedfilterobservations,followinga
redshiftrangeoftheXXLSurvey,andwithinangularradiicorre- phenomenologicalmodel.Forobservationscontaminatedbysoft
spondingto500kpcattheredshiftsoftheclusters).Thisallows protons(wherethecountrateratiobetweenthein-FOV,beyond
us to appreciate the respective contribution of these surveys to 10arcmin, and out-of-FOV regions of the detector was >1.15),
the characterisation of both clusters and projection effects. In we included an additional broken power-law component, with
suchatable,intensivefieldsurveysasVIPERSorGAMAshow the slopes fixed at 0.4 and 0.8 below and above 5keV respec-
theirgreatimportance. tively. The sky background was modelled using data extracted
from an offset region (outside the cluster emission determined
above), using a three-component model as detailed in Eckert
4.Theclustercatalogue
et al. (2011). Within the XSPEC environment, cluster source
Inthissection,wefirstprovideaglobaldescriptionofthesam- spectrawereextractedforeachoftheXMM-Newtoncamerasand
ple.Wethenpresentthedirect(spectral)measurementswemade fitswereperformedinthe[0.4–11.0]keVbandwithanabsorbed
of luminosity, temperature, gas mass, and flux. These measure- APEC(AstrophysicalPlasmaEmissionCode,Smithetal.2001)
mentsareobviouslymorerobustthanusingscalingrelations,but model(v2.0.2),withafixedmetalabundanceofZ =0.3Z .
(cid:12)
they require higher quality data and therefore cannot be com- Wedenotetheluminositywithinr 4 as LXXL ,within
500,MT 500,MT
puted for the whole sample of clusters. Scaling relations were the[0.5–2.0]keVband(clusterrestframe).Luminositiesquoted
therefore used in order to complete the sample for some of the within r are extrapolated from 300kpc (see below) out to
500,MT
followingstudies. r by integrating under a β-profile assuming a core radius
500,MT
r = 0.15r and an external slope β = 0.667 (cf. XXL
c 500,MT
4.1.Sampledescription Paper II). Values for cluster r500,MT are calculated using the
mass–temperature relation of Lieu et al. (2016, hereafter XXL
The C1+C2 clusters are listed in Table 5 which is sorted PaperIV).
according to increasing RA and only the first twenty entries Giventhatwearedealingwithmuchfaintersourcesthanin
are displayed. Blank places in the table are undetermined val- XXLPaperII,itwasnotpossibletomeasureX-raytemperatures
ues. We note that the XLSSC 634 cluster was confirmed by forallclusters.Inparticular,severalC1clusterswerelocatedin
Ruel et al. (2014) with Gemini/GMOS data. The spectroscopi- pointingsaffectedbyflaring,hadverylowcounts,werecontam-
cally confirmed C3 objects are listed in Table G.1. Both tables inated by point sources, or were at very low redshift so with a
are also available in the XXL Master Catalogue browser3 and badspatialcoverage.
Table 5 is available at the CDS. For each source, we provide
(whenavailable):
4.2.2.Gasmass
– the XLSSC identifier (between 1 and 499, or 500 and 999
forXXL-NorXXL-Srespectively; Weanalyticallycomputedgasmassesforclusterswithredshifts
– RAandDec; following closely the method outlined in Eckert et al. (2014,
– theredshiftandthenumberofgalaxiesusedfortheredshift hereafter XXL Paper XIII). Here we briefly recall the vari-
determination; ous steps of the analysis. First, we extract surface-brightness
– theclass,C1,C2(Table5only)orC3(TableG.1only); profiles in the [0.5–2]keV band starting from the X-ray peak
– basic X-ray and X-ray related quantities for the clusters of using the PROFFIT package (Desai et al. 2012). We compute
thepresentrelease(X-rayfluxes,Mgas,500kpc,r500,MT,T300kpc, thesurface-brightnessprofilesfrommosaicimagesoftheXXL
and LXXL ). We note that we give in the present paper the fieldsinsteadofindividualpointings,whichallowsustoimprove
500,MT
value of M , contrary to what was given in XXL the S/N and measure the local background level more robustly
gas,500kpc
PaperXIII; compared to the analysis presented in XXL Paper XIII. The
– a flag indicating whether there is a note on the cluster in surface-brightnessprofilesarethendeprojectedbydecomposing
Appendix G, whether the cluster was already published in the profile onto a basis of multiscale parametric forms.
XXLPaperIIorinformerXMM-LSSreleases,andwhether
theclusterisamemberofthefluxlimitedsample. 4 r500,MT isdefinedastheradiusofthesphereinsidewhichthemean
densityis500timesthecriticaldensityρ oftheUniverseatthecluster’s
c
3 http://cosmosdb.iasf-milano.inaf.it/XXL/ redshift,M isthenbydefinitionequalto4/3π500ρ r3 .
500,MT c 500,MT
A5,page6of28
C.Adamietal.:TheXXLSurvey.XX.
Table5.ListofspectroscopicallyconfirmedC1andC2clustersofgalaxies.
XLSSC α δ z N Class M r T LXXL F Flag
gal gas,500kpc 500,MT 300kpc 500,MT 60
1011 1042 10−15
deg deg M kpc keV ergs−1 ergs−1cm−2
(cid:12)
199 30.192 −6.708 0.339 2 1 73+4 644 2.1+0.2 32±3 67±5 l
−6 −0.3
200 30.331 −6.830 0.333 2 1 48+3 653 2.1+0.3 16±2 31±3 l
−3 −0.4
114 30.425 −5.031 0.233 6 2 40+3 35±8 l
−3
179 30.482 −6.574 0.608 5 1 43+11 14±4 l
−12
113 30.561 −7.009 0.050 9 1 8+1 115±8 l
−1
174 30.592 −5.899 0.235 8 1 41+3 570 1.5+0.1 8±1 25±4 l
−4 −0.1
094 30.648 −6.732 0.886 3 1 106+12 581 3.0+0.5 224±32 48±5 +l
−12 −0.6
196 30.728 −7.652 0.136 8 1 26+2 563 1.3+0.1 4±1 32±4 l
−3 −0.2
178 30.753 −6.285 0.194 2 2 29+3 655 0.8+0.1 3±1 17±3 l
−5 −0.1
156 30.766 −7.101 0.336 4 2 33+3 28±4 l
−3
157 30.865 −6.929 0.585 5 1 70+7 721 3.2+0.8 42±7 19±3 l
−7 −0.7
197 30.923 −7.785 0.439 2 1 107+5 755 3.0+0.4 76±9 97±7 l
−5 −0.5
096 30.973 −5.027 0.520 6 1 89+5 951 5.0+0.9 63±8 36±4 *+l
−5 −0.5
155 31.134 −6.748 0.433 2 1 36+4 576 1.8+0.3 16±3 23±3 l
−5 −0.3
173 31.251 −5.931 0.413 3 1 47+4 930 4.3+0.3 17±2 24±3 l
−4 −0.3
177 31.290 −4.918 0.211 7 2 37+3 22±4 l
−3
102 31.322 −4.652 0.969 3 1 138+7 638 3.9+0.8 167±25 42±4 +l
−7 −0.9
106 31.351 −5.732 0.300 14 1 83+3 777 2.8+0.2 43±3 91±4 +l
−3 −0.3
107 31.354 −7.594 0.436 3 1 67+4 672 2.4+0.4 49±6 56±5 +l
−5 −0.4
160 31.521 −5.194 0.817 4 2 6±4
Notes.Col.1:officialXLSSCname.Cols.2and3:X-rayclustercoordinates.Col.4:clustermeanredshift.Col.5:numberofmeasuredspectro-
scopicredshifts(X:meansredshiftiscomputedfromX-rayspectroscopydirectly).Col.6:XXLclass.Col.7:gasmassinsideaphysicalradiusof
500kpcalongwithlowerandupperuncertainties.Col.8:r .Col.9:X-raytemperaturewithlowerandupperuncertainties.Col.10:LXXL
500,MT 500,MT
X-rayluminosityanduncertaintyinthe[0.5–2]keVrest-frameenergyrange.Col.11:X-rayfluxanduncertaintyasinXXLPaperIIandinthe
[0.5–2]keVband.Col.12:flags:“+”meanstheclusterwasalreadypublishedintheXMM-LSSreleases,*meansthatwehaveanoteonthis
clusterinAppendixG,lmeansthattheconsideredclusterisbrighterthanthefluxcompletenesslimit(∼1.3×10−14ergs−1cm−2),Fmeansthatthe
structureisacandidatefossilgroup.CompletetableisavailableattheCDS.Blankplacesareundeterminedvalues(toolowsignal-to-noiseratio).
Cash (1979) statistics are used to adjust the model to the the shape estimate. Whenever a cluster was detected in several
data,andtheMarkovchainMonteCarlo(MCMC)tool EMCEE pointingswiththesameclassification,wethereforeretainedthe
(Foreman-Mackeyetal.2013)isusedtosamplethelargeparam- onewheretheclusterwasclosesttotheopticalaxis.Theanaly-
eter space. The deprojected profiles are then converted into gas sisthenreliesonasemi-interactiveprocedureinitiallydeveloped
density profiles using X-ray cooling functions calculated using forClercetal.(2012).Itfirstdefinesapreliminarysourcemask
the APEC plasma emission code (Smith et al. 2001). Finally, based on the output of the XXL detection pipeline and allows
the recovered gas density profiles are integrated over the vol- the user to manually correct the mask. Then the signal in a
ume within a fixed physical scale of 500kpc. The gas masses user-definedbackgroundannulusaroundthesourceismodelled
measured for XXL-100-GC clusters using this procedure are with a linear fit to the local exposure map (thus allowing for
consistent with the values published in XXL Paper XIII, with both a vignetted and an unvignetted background component).
ameanvalueM /M =0.984.Formoredetailsontheanaly- Finally, count-rates in each detector are estimated, propagating
new old
sisprocedurewereferthereadertoXXLPaperXIII.InTable5, the errors in the background determination, and turned into a
wegiveonlythegasmassesforclusterswithanuncertaintyon global flux using average energy conversion factors relevant to
thefluxF (seebelow)lowerthanthethirdofthefluxitself.We each field5. Of course the final estimated flux depends some-
60
alsosimilarlydonotprovidegasmassestimatesforC3clusters. what on the chosen background sample. In our case, the sizes
oftheadoptedbackgroundannulivariedsignificantly,reflecting
the large spread in cluster size and flux in the catalogue. They
4.2.3.X-rayflux
ranged from 90 to 300(cid:48)(cid:48) for the inner radius and 180 to 500(cid:48)(cid:48)
TobeabletodirectlycompareourestimateoftheX-rayluminos- for the outer bound. The shifts in the measured fluxes recorded
ityfunction(seenextsection)withtheresultsofXXLPaperII, whenchangingthebackgroundaperturewerealwayswellwithin
weadoptedfortheX-rayphotometrythesameproceduretoesti- the statistical errors, provided that the background annulus was
mateaperturefluxesinaradiusof60(cid:48)(cid:48) (F ).Weperformedthe freefromapparentclusteremission.
60
measurements on the pointing within which each cluster was
mostsignificantlydetected–asindicatedbytheC1/C2/C3clas- 5 Those assume an APEC v2.0.2 thermal spectrum with T = 2keV
sification. This approach was preferred compared to the other andZ=0.3Z .Thedifferencebetweenthetwofieldscomesfromtheir
(cid:12)
approach consisting of combining all available pointings for a averageabsorbingcolumndensityofn =2.3×1020cm−2 forXXL-N
H
givenclusterasitallowedustokeepgoodspatialresolutionfor and1.25×1020cm−2forXXL-S.
A5,page7of28
A&A620,A5(2018)
4.3.Clusterparametersfromscalingrelations
In order to allow studies of the global properties of the full
sample,wealsoprovidemeanparameterestimatesderivedfrom
scalingrelations(TableF.1).
To estimate luminosity and temperature from scaling rela-
tions(withoutaspectralfit),wefirstextractedtheXMM-Newton
pninthe[0.5–2]keVbandwithin300kpcfromtheclustercen-
tre.Countrateswerecomputedstartingfromvaluesandbounds
fortheintensityS ofthesourceusingcountsandexposuredata
obtained in source and background apertures. The background-
marginalised posterior probability distribution function (PDF)
ofthesourcewasthencalculated,assumingPoissonlikelihoods
forthedetectednumberofsourcecountsandbackgroundcounts
in the given exposure time. The mode of this PDF was deter-
mined,andthelowerandupperboundsoftheconfidenceregion
Fig.4.Comparisonbetweenthetruetemperaturemeasurements(from
were determined by summing values of the PDF alternately
Table5)andestimatesfromthescalingrelations(fromTableF.1).The
above and below the mode until the desired confidence level
dottedandsolidlinesshowthe1:1relationandtheactualregressionto
was attained. When the mode was at S = 0 or the calculation
thedatarespectively.
for the lower bound reached the value S = 0, only the upper
confidenceboundwasevaluated,andwasconsideredasanupper
limit. Regarding the 207 C1 clusters of XXL-C1-GC, only 191
We converted this count rate to the corresponding X-ray are in pointings not affected by flares. All results involving the
luminosity by adopting an initial gas temperature, a metallic- clusterselectionfunctionarethereforebasedonthissubsample
ity set to 0.3 times the solar value (as tabulated in Anders & of191objects.
Grevesse 1989) and the cluster’s redshift (without propagating Fiveamongthese191clustersdonothavearedshiftdetermi-
theredshiftuncertainties).Thesamevalueofthetemperatureis nationandarethereforemodelledusinganincompletenessfactor
usedtoestimater500,MT,usingthemass–temperaturerelationfor in the selection function. Excluding these five, the remaining
thesampleXXL+COSMOS+CCCPinTable2ofXXLPaperIV. sampleof186clustersisusedtocomputetheclusterluminosity
Theluminosityisthenextrapolatedfrom300kpcouttor500,scal function.
(similarasr500,MTbutcomputedduringtheprocessofthecluster Eightoutofthese186clustershavenotemperaturemeasure-
parameters estimate from scaling relations) by integrating over ment and their X-ray luminosity was estimated through scaling
the cluster’s emissivity represented by a β-model with param- relations. This sample of 176 clusters is used to constrain the
eters (rc,β) = (0.15r500,scal,2/3). Hence, a new temperature is luminosity–temperaturerelation.
evaluatedfromthebest-fitresultsfortheluminosity–temperature
relation quoted in Table 2 of Giles et al. (2016, hereafter XXL
Paper III). The iteration on the gas temperature is stopped 5.1.Redshiftdistributionandspectroscopicredshiftsampling
whentheinputandoutputvaluesagreewithinatolerancevalue
Thegalaxyredshiftsamplingofclustersandtheclusterredshift
of5%.
distributions are displayed in Figs. 5 and 6 for various cluster
Usually,thisprocessconvergesinfewsteps(2–3iterations).
selections. Our total sample is the full list of clusters quoted
WeprovideestimatesoftheX-raytemperature,T ,ofthe
300kpc,scal inthepresentpaper,includingthefewnotyetspectroscopically
bolometric luminosity in the [0.5–2]keV range within r ,
500,scal confirmedclustersinTableG.2.
Lbol , of the mass M within r , and of relative
500,scal 500,scal 500,scal We see that the full list is very similar to the list of spec-
errors propagated from the best-fit results of the X-ray tem- troscopically confirmed clusters, cf. top panel of Fig. 5. A
perature, r500,scal, and the bolometric luminosity. A comparison Kolmogorov–Smirnov test shows no difference (at better than
between the measured cluster temperatures and those obtained the 99.9% level) both for the redshift and the redshift sam-
from the scaling relations is displayed in Fig. 4; the observed plingdistributions.Thisfigurealsoshowsthat,amongthenon-
scatter around the 1:1 line simply reflects the intrinsic scatter spectroscopically confirmed clusters, thirteen do not have any
of the luminosity–temperature relation. In some cases (mainly spectroscopicredshift,threeofthemhaveasinglespectroscopic
for C2 clusters), this procedure converges to an M500,scal value redshift(nottheBCG),andonehastwospectroscopicredshifts
thatfallsbelowthemassrangeoftheXXL-100-GCsample(cf. (theBCGbeingnotavailable,spectroscopicconfirmationisnot
XXL Paper IV), used for derivation of the scaling relations. In validatedeither).
thiscase,novaluesaregiven. The XXL-N and XXL-S cluster samples are also similar in
terms of redshift distribution (99.9% level for a Kolmogorov–
Smirnovtest).Wehoweverhaveonaveragemorespectroscopi-
5.Updatedclusterstatistics
callyconfirmedmembers(typicallymorethansixspectroscopic
With the current sample having twice as many C1 clusters redshifts) in the northern field compared to the southern field
as in XXL-100-GC (and 341 spectroscopically confirmed clus- (see below for a more quantitative analysis of the cluster sam-
ters in total), we are in a position to update a number of pling). The probability of having similar samples is only at the
statistical results presented in the 2016 XXL release (a.k.a. 28%levelwithaKolmogorov–Smirnovtest.
DR1). Detailed analyses of these quantities in the current C1, C2, and C3 cluster distributions are obviously differ-
XXL-C1-GCsamplewillhoweverbethesubjectofforthcoming ent,asdemonstratedbyaKolmogorov–Smirnovtest.C2andC3
papers.Inthispaper,weconcentrateonafewbasicpropertiesof clustershavelowerspectroscopicsamplingthanC1asthesewere
theXXL-C1-GC. not our primary spectroscopic targets. C3 mainly appears as a
A5,page8of28
C.Adamietal.:TheXXLSurvey.XX.
Fig. 5. Distribution of the number of spectroscopic redshifts inside Fig.6.Distributionofthenumberofspectroscopicredshiftsinsideclus-
clusters with a redshift measurement. The insets show the redshift terswitharedshiftestimate.Theinsetsshowtheredshifthistogramsof
histograms of these samples. Top panel: spectroscopic+photometric these samples. Top panel: C1 (red histograms), C2 (blue histograms),
redshift sample (black histograms), and spectroscopic redshift sample andC3(greenhistograms)clusters.Bottompanel:clusterswithalsoa
(redhistograms)clusters.Bottompanel:XXL-N(redhistograms)and flux estimate fainter (black histograms) and brighter (red histograms)
XXL-S (blue histograms) clusters. Photometric redshifts are used in thanthereferencefluxcompletenesslimitof1.3×10−14ergs−1cm−2.
replacement of spectroscopic redshifts in these two histograms when
spectroscopicredshiftsarenotavailable.
subpopulation of intermediate redshift clusters, with also a few
distant(z≥1)structures.
Finally, clusters brighter and fainter than the reference flux
completeness limit of 1.3×10−14ergs−1cm−2 cover almost the
same redshift range. Their redshift distribution is however dif-
ferent (probability of having similar samples only at the 53%
level) with, not surprisingly, a lot more bright clusters at red-
shiftsbelow0.5.Theyalsoareverydifferent(atthe98%level)
in termsof spectroscopicsampling, thebrightest clusters being
betterspectroscopicallysampled.
5.2.X-rayluminositiesandfluxes
Fig.7.X-rayluminosity(LXXL inlogunitofergs−1inthe[0.5–2]keV
We display in Fig. 7 the distribution of cluster luminosities 500,MT
band)distributionofclustershavingaspectroscopicredshiftandalumi-
LXXL (only when available through spectral fit, so C3 clus- nosity determination. Red histogram: the C1 sample; blue histogram:
500,MT
ters are excluded) for the C1 and the C2 samples. In addition, theC2sample.
Fig.8showstheclustermassM (derivedfromscalingrela-
500,scal
tions) distribution for the same subsamples. We note that the lensingmasseswereoverestimated.DeepSubaru-HSCobserva-
cluster masses do not pertain here to direct spectral measure- tionswillprovidehighersignaltonoiseinformationandhelpus
ments (since temperatures are not available for the full sample) understandthecontributionofnon-thermalpressureinthetotal
but were derived using scaling relations; we show these graphs massbudget(Umetsuetal.,inprep.).
toallowglobalcomparisonswithotherclustersamples.InXXL Finally,inordertocomparetheC1andC2subsampleswith
PaperXIII,wementionedthepossibilitythatourtotalCFHTLS theC3subsample,weshowinFig.9the F (fluxwithina60(cid:48)(cid:48)
60
A5,page9of28
A&A620,A5(2018)
Fig. 8. Mass (in log units of M ) distribution of the clusters with a
(cid:12)
spectroscopicredshiftestimate.Redhistogram:theC1sample;bluehis-
togram:theC2sample.Themassdatapointshavebeenderivedfrom
scaling relations based on the cluster luminosities (cf. Sect. 4.3 and
AppendixF).
Fig.9. X-rayflux(F inlogunitofergs−1cm−2,withina60(cid:48)(cid:48) radius Fig. 10. Upper panel: luminosity–temperature relation with the best-
60 fitting models. The light blue circles show the XXL-C1-GC clusters;
inthe[0.5–2]keVband)distributionfortheclustershavingaspectro-
thebest-fittingmodel(includingselectioneffects)isshownbythesolid
scopicredshift.Redhistogram:theC1sample;bluehistogram:theC2
black line, the 1σ uncertainty represented by the grey shaded region.
sample;greenhistogram:theC3sample.Theblackverticallineisthe
estimatedreferencefluxcompletenesslimitof1.3×10−14ergs−1cm−2. Thebest-fittingmodelfittedtothedatausingtheBCESregressionis
shown as the dashed line. Bottom panel: evolution of the luminosity–
temperature relation for XXL-C1-GC. The XXL-C1-GC clusters are
radiusinthe[0.5–2]keVband)distributionofthethreesubsam- representedbythelightbluecirclesandthebest-fittingmodelisgiven
ples.Asexpected,C1clustersarebrighterthantheC2clusters. bytheblacksolidline;thegreyshadedregionhighlightsthe1σuncer-
C3clusterspertaintotwodistinctpopulationsasalreadystated tainty.The“strong”and“weak”self-similarexpectationsaregivenby
intheprevioussectionandshowedinAdamietal.(2011).Alarge thereddashedandbluedashedlines,respectively.
part of them are structures slightly fainter than the C2 clusters,
andafewarebrightanddistantstructures.
totheXXL-100-GCfit,wefindthattheslopeandnormalisation
areconsistent.
5.3.Luminosity–temperaturerelationoftheC1sample We next fit the XXL-C1-GC scaling relation using the
procedure outlined in XXL Paper III, taking fully into account
Figure 10 shows the XXL luminosity–temperature relation for
the selection effects (we refer to Sects. 4.3 and 5.1 in XXL
theXXL-C1-GCsample(bothparametersderivedfromspectral
Paper III for specific details). However, the selection func-
measurements).Afittothedatausingapowerlawoftheform
tion was updated to match the current sample, instead of the
XXL-100-GC selection function previously used. Figure 10
(cid:32) L (cid:33)= E(z)γLTA (cid:32)T (cid:33)BLT (1) (upper panel) shows the XXL luminosity–temperature relation,
L LT T with the best-fitting (bias-corrected) model given by the black
0 0
solid line and the corresponding 1σ uncertainty shown by
wasperformed,where A , B ,andγ representthenormali- the grey shaded region. The best-fitting parameter values and
LT LT LT
sation,slope,andpoweroftheevolutioncorrectionrespectively. their uncertainties are summarised by the mean and standard
Thepowerlawwasfittothedata,firstusingtheBCESorthogo- deviation of the posterior chains for each parameter from a
nalregressioninbasetenlogspace(Akritas&Bershady1996) Markov Chain Monte Carlo output. We used four parallel
assuming self-similar evolution (γ =1). The best fit parame- chains of 50000 iterations each. To test for convergence,
LT
tersaregiveninTable6.ComparingtheXXL-C1-GCBCESfit the stationary parts of the chains were compared using the
A5,page10of28