The XMM-LSS survey: the Class 1 cluster sample over the extended 11 deg 2 and its spatial distribution
N. Clerc, C. Adami, M. Lieu, B. Maughan, F. Pacaud, M. Pierre, T. Sadibekova, G. P. Smith, P. Valageas, B. Altieri, C. Benoist, S. Maurogordato, J. P. Willis
MMon. Not. R. Astron. Soc. , 1–21 (2002) Printed 17 September 2018 (MN L A TEX style file v2.2)
The XMM-LSS survey: the Class 1 cluster sample over theextended 11 deg and its spatial distribution N. Clerc (cid:63) , C. Adami , M. Lieu , B. Maughan , F. Pacaud , M. Pierre ,T. Sadibekova , G. P. Smith , P. Valageas , , B. Altieri , C. Benoist ,S. Maurogordato , J. P. Willis Max Planck Institut f¨ur Extraterrestrische Physik, Postfach 1312, 85741 Garching bei M¨unchen, Germany. Aix-Marseille Universit´e, CNRS, LAM (Laboratoire d’Astrophysique de Marseille) UMR 7326, 13388, Marseille, France. School of Physics and Astronomy, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK H. H. Wills Physics Laboratory, University of Bristol, Tyndall Ave, Bristol BS8 1TL, UK Argelander-Institut f¨ur Astronomie, University of Bonn, Auf dem H¨ugel 71, 53121 Bonn, Germany. Laboratoire AIM, CEA/DSM/IRFU/SAp, CEA Saclay, 91191 Gif-sur-Yvette, France. Institut de Physique Th´eorique, CEA Saclay, 91191 Gif-sur-Yvette, cedex, France. CNRS, URA 2306, 91191 Gif-sur-Yvette, cedex, France. ESAC, Villafranca del Castillo, 28692 Madrid, Spain. Laboratoire J. L. Lagrange OCA-CNRS-UNSA, BP4429, Nice Cedex 04, France. Department of Physics and Astronomy, University of Victoria, 3800 Finnerty Road, Victoria, BC, Canada
Accepted 2014 August 8. Received 2014 July 8; in original form 2014 March 24.
ABSTRACT
This paper presents 52 X-ray bright galaxy clusters selected within the 11 deg XMM-LSS survey. 51 of them have spectroscopic redshifts (0 . < z < . z phot = 1 .
9, and all together make the high-purity ”Class 1” (C1) clustersample of the XMM-LSS, the highest density sample of X-ray selected clusters with amonitored selection function. Their X-ray fluxes, averaged gas temperatures (median T X = 2 keV), luminosities (median L X, = 5 × ergs/s) and total mass esti-mates (median 5 × h − M (cid:12) ) are measured, adapting to the specific signal-to-noiseregime of XMM-LSS observations. Particular care is taken in deriving the sample se-lection function by means of realistic simulations reproducing the main characteristicsof XMM observations. The redshift distribution of clusters shows a deficit of sourceswhen compared to the cosmological expectations, regardless of whether WMAP-9 orPlanck-2013 CMB parameters are assumed. This lack of sources is particularly notice-able at 0 . (cid:46) z (cid:46) .
9. However, after quantifying uncertainties due to small numberstatistics and sample variance we are not able to put firm (i.e. > σ ) constraints on thepresence of a large void in the cluster distribution. We work out alternative hypothe-ses and demonstrate that a negative redshift evolution in the normalization of the L X − T X relation (with respect to a self-similar evolution) is a plausible explanationfor the observed deficit. We confirm this evolutionary trend by directly studying howC1 clusters populate the L X − T X − z space, properly accounting for selection biases.We also point out that a systematically evolving, unresolved, central component inclusters and groups (AGN contamination or cool core) can impact the classification asextended sources and be partly responsible for the observed redshift distribution. Weprovide in a table the catalogue of 52 clusters together with their measured properties. Key words: cosmology: observations – catalogues – galaxies: clusters: general –X-rays: galaxies: clusters. (cid:63)
E-mail: [email protected]
Studying the spatial distribution of galaxy clusters in a vol-ume of cosmological size enables the cartography of large- c (cid:13) a r X i v : . [ a s t r o - ph . C O ] A ug N. Clerc et al. scale structure in the Universe through its most massivebuilding blocks. The number counts distribution of clustersis therefore an excellent test for cosmological models and thegrowth of structure (e.g. Borgani et al. 2001; Henry et al.2009; Vikhlinin et al. 2009; Mantz et al. 2010; Rozo et al.2010; Pierre et al. 2011; Benson et al. 2013).From an observational point of view, galaxy clustersare advantageously high signal astrophysical sources: as ex-pected from simple scaling arguments (e.g. Kaiser 1986), themost massive virialized objects are also the largest in sizeand their observable properties scale up with mass. This isespecially true in X-ray wavelengths because of the largeamount of X-ray photons emitted by the hot ( T ∼ K)intra-cluster baryonic gas trapped in their deep potentialwells. With typical X-ray luminosities of 10 − ergs/s,clusters can be detected up to large cosmological distances ina systematic and controlled way. On the other hand, galaxyclusters are rare objects and robust analyses of their spa-tial distribution (e.g. redshift distribution, 2-point correla-tion function...) and their ensemble properties (e.g. scalingrelations, mass distribution...) require medium to large areasurveys in order to accumulate statistical power. These char-acteristics motivated the assembly of large samples of galaxyclusters detected in X-rays. In particular, studies based onthe ROSAT all-sky survey (Truemper 1993) such as RE-FLEX (B¨ohringer et al. 2001), NORAS (B¨ohringer et al.2000), the ROSAT North Ecliptic Pole Survey (Henry et al.2001), the 400d (Burenin et al. 2007) delivered solid cosmo-logical results based on galaxy clusters (e.g. Schuecker et al.2003; Vikhlinin et al. 2009; Mantz et al. 2010). During thelast decade, the very sensitive XMM-Newton (Fassbender etal. 2011; Willis et al. 2013) and Chandra (e.g. Tozzi et al.2013) observatories revealed the presence of their character-istic emission beyond redshifts of 1, and even up to z ∼ with XMM pointed observations, reaching a sensitivityof 10 − ergs/s/cm for extended sources in the [0.5-2] keV band. Thanks to the wide and complementary multi-wavelength coverage (from radio to γ -rays) and dedicatedfollow-up effort, it constitutes a relevant field for studies ofgalaxy clusters and groups. The complete X-ray source cat-alogue is published in Chiappetti et al. (2013) along withthe optical associations. The XMM-XXL survey is currentlyextending its area to 50 deg following a similar strategyas for the detection and characterization of galaxy clusters,with international support and expertise (Pierre et al. 2014,in prep.) The sample used in this work is unique in terms ofX-ray and spectroscopic redshifts completenesses: (i) spec-troscopic redshifts (from observations of cluster galaxy mem-bers) enable to position clusters in 3D space and accuratederivation of their physical properties: gas temperature, lu-minosity, physical size; (ii) a trade-off between completeness,purity and assessment of selection effects has been carefullydesigned: Pacaud et al. (2006) indeed demonstrated the ex-istence of an uncontaminated sample of extended sourcesdetected on XMM-LSS images called ”C1”, 29 of them weredetected in the first 5 deg of the XMM-LSS survey (Pacaudet al. 2007).In this study, we focus on the redshift distribution of the complete set of 52 XMM-LSS C1 clusters and its cosmo-logical modelling. Results from the catalogue of sources de-tected in the Planck survey by the Sunyaev-Zeldovich effectindicated a deficit of clusters at all redshifts when comparedto expectations from the Planck CMB cosmological model(Planck Collaboration et al. 2013b). Tension could be alle-viated by modifying the mass-observable relation. Studiesbased on galaxy clusters indeed appeal for a simultaneousmodelling of the cosmological halo mass function, the mass-observable relations (scaling laws and their evolutions withredshift) and modelling of selection effects (e.g. Pacaud etal. 2007; Pratt et al. 2009; Vikhlinin et al. 2009; Mantz etal. 2010; Allen, Evrard, & Mantz 2011; Clerc et al. 2012b;Planck Collaboration et al. 2013b). Therefore, part of ourresults concerns the L X − T relation of C1 clusters and itsevolution. This latter point is particularly debated in cur-rent studies of clusters detected in XMM data. Pacaud etal. (2007) pointed out the importance of selection biases insuch studies. Reichert et al. (2011) found a negative trend(relative to self-similar expectations) by analysing an hetero-geneous sample of objects. Both the XMM Cluster Survey(XCS, Hilton et al. 2012) and the XMM CLuster Archive Su-per Survey (X-CLASS, Clerc et al. 2012b) indicated a nega-tive evolution in the normalization of the relation. However,differences in the selection and analyses methods make suchcomparisons difficult.This paper is organized as follows. In Section 2 we de-scribe the dataset and our choice of sample for this study.An in-depth characterization of the survey selection func-tion is presented in Section 3. Derivation of cluster proper-ties is detailed in Section 4. The spatial distribution and theluminosity-temperature relation of clusters in the sample areshown in Section 5, and we discuss further the modelling ofthe redshift distribution in Section 6. Section 7 summarizesour findings.Throughout this paper, we assume Ω m = 0 .
3, Ω Λ = 0 . h = 0 . H = 100 h km.s − , except otherwisestated. In particular, our discussion of the redshift distribu-tion (Sect. 5 and 6) alternates between WMAP-9 (Hinshawet al. 2013) and Planck CMB (Planck Collaboration et al.2013a) cosmologies. In all that follows, M δ (= M δc ) is themass within a sphere of radius R δ (= R δc ), inside which themass density is δ times the critical density of the Universeat the considered redshift. Transformations between differ-ent values of δ will assume a NFW profile (Navarro, Frenk,& White 1997). The complete XMM-LSS 11 deg source catalogue is pre-sented in full extent by Chiappetti et al. (2013). In this sec-tion, we briefly recall the main characteristics of the 11 deg XMM-LSS survey and our procedure for detecting sources,with particular emphasis on the confirmation of C1 galaxyclusters.
The XMM-LSS survey is located at R . A . = 02 h m and δ = − ◦ (cid:48) and consists of 98 XMM pointings on a ∼ . ◦ × . ◦ contiguous footprint (Fig. 1). It represents c (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure 1.
Layout of the 98 XMM observations constituting theXMM-LSS 11 deg survey. Positions of C1 clusters presented inthis work are overlaid as black squares. The on-axis, clean, expo-sure time of each pointing is shown by the colour scale. The sizeof each XMM observation is on scale and corresponds to the 13 (cid:48) radius circular area analysed around each pointing centre. a sub-area of the larger XMM-XXL survey (Pierre et al.2011) for which a full description of the observation strat-egy and quality is given in Pierre et al. (2014, in prep.) Eachof those pointings corresponds to a single observation withthe EPIC detectors (MOS1, MOS2 and PN) in full frameimaging mode, spanning a field of view of roughly 30 (cid:48) diam-eter each.The centre of each observation is defined by theexposure-weighted location of the optical axes of all threeXMM telescopes. The work presented in this study relies onX-ray data collected by each pointing up to an off-axis radiusof 13 (cid:48) . The total geometrical area of the survey amounts to11.1 deg , 3.0 deg (27%) of that area consists in overlapsbetween adjacent pointings (Fig. 1). The remaining area(8.1 deg ) is covered by one unique pointing. For each in-dividual observation, an event list is created and filteredfrom solar proton flares by automatic inspection of the high-energy light-curves. High count-rate periods associated toparticle flares are flagged and removed for each of the threedetectors separately. Cleaned exposure times after flares re-moval amount to ∼
10 ks for the majority of the pointings(Fig. 1), although spatial fluctuations in the survey depth doexist and will be modeled in the selection function (Sect. 3).
Our source detection procedure builds upon the algorithmdescribed in Pacaud et al. (2006, 2007) with several revi-sions as detailed in Chiappetti et al. (2013). These revisionsmainly consist in a transcription of the algorithm code fromIDL to Python and a correction in the relative astrometry.Individual detector images are created in the [0.5-2] keVband from cleaned event lists and binned in pixels of 2 . (cid:48)(cid:48) width. Sources are detected on each pointing image sepa-rately: in a first step, all three EPIC images are co-added,then filtered using the mr filter multi-resolution algorithm(Starck, Murtagh, & Bijaoui 1998). Such a filtering ade- quately accounts for the Poisson nature of the noise in thebackground and source areas. A SExtractor (Bertin &Arnouts 1996) pass over this image provides an initial de-tection list and a first guess for their positions. In a secondstep, each source from this list is characterized thanks to amaximum likelihood fitting algorithm (
XAmin , Pacaud etal. 2006) specifically developed to assess the extension ofsources in XMM-LSS data. Among the output parametersof interest are the detection likelihood quantifying the signif-icance of the detection relative to a case without source andthe extension likelihood , comparing the case of an extendedsource and a point-like source. These likelihood values aredefined (and corrected) so as to match a 2-parameter χ statistics, hence convertible into equivalent probabilities fol-lowing P = exp( − L/ Following the methodology presented in Pacaud et al.(2006), we define a C1 (Class 1) sample of sources by fil-tering on
XAmin output parameters, namely the extension ( > (cid:48)(cid:48) ), the extension likelihood ( >
33) and the detectionlikelihood ( > and 3 were deblended astwo distinct, non-C1 sources (XLSSC 12, 28 and 74 ; seeimages in App. C of this paper and Fig. C1 of Pacaud et al.2007). In all three cases, such deblending arose because ofpoint sources present close to the cluster centre. It is a directconsequence of the multi-scale approach implemented in ourdetection algorithm. On the other hand, 5 sources within the6 deg area explored by Adami et al. (2011) were promoted Lower-purity sample defined by a decrease in the extension like-lihood from 33 to 15.c (cid:13)000
33) and the detectionlikelihood ( > and 3 were deblended astwo distinct, non-C1 sources (XLSSC 12, 28 and 74 ; seeimages in App. C of this paper and Fig. C1 of Pacaud et al.2007). In all three cases, such deblending arose because ofpoint sources present close to the cluster centre. It is a directconsequence of the multi-scale approach implemented in ourdetection algorithm. On the other hand, 5 sources within the6 deg area explored by Adami et al. (2011) were promoted Lower-purity sample defined by a decrease in the extension like-lihood from 33 to 15.c (cid:13)000 , 1–21
N. Clerc et al. as C1 sources with this new processing, 4 of them previouslyclassified as C2. We attribute these differences to changes inthe
XAmin pipeline version, the event list processing and/orthe XMM-SAS version and calibration data. Such changescan be viewed as additional ”noise” in the images on topof the usual Poisson and background noise. However theyare not handled by the noise model of our pipeline and thusinduce variations in the final source list near the selectionthresholds. The maximal list of C1-classified sources (eitherby one or the other processing runs) contains 54 entries. Wemanually discarded one detection whose characterization asan extended source is doubtful, since it suffers from severeprojection effects due to a bright X-ray emitting star and alarge X-ray cluster (XLSSC 061 in Table 1 and Fig. C1) inthe foreground.
Our procedure for confirming and validating C1 sources asgalaxy clusters involves optical spectroscopic observationsand is fully described in Adami et al. (2011). Spectroscopicfollow-up campaigns were dedicated to the confirmation ofC1 clusters, making use of different observing facilities in or-der to cover the range of redshifts encountered in the sample.Selection of spectroscopic targets was based on ugriz opti-cal imaging from the CFHT-LS survey, choosing in prioritybright red-sequence galaxies in the vicinity of the clusterX-ray emission. Each spectrum was reduced and its corre-sponding redshift measured by several independent persons.A final redshift value and associated quality flag was as-signed by a moderator. We refer to Adami et al. (2011) forthe meaning of these flags in terms of redshift reliability.Additional galaxy redshifts from the VVDS deep and ultra-deep surveys (Le F`evre et al. 2005) and the Subaru DeepSurvey (Ueda et al. 2008) as well as redshifts collected fromthe NED were added to the sample. Spectroscopic data isstored in Cesam and will be publicly released in the end ofthe XMM-XXL survey.Cluster redshift validation was first based on identifi-cation of groups of galaxies sharing similar radial velocitiesalong a line of sight by using the gapper method (e.g. Bi-viano et al. 1997). We assigned membership in each putativegroup using a physical radius of 500 kpc around the X-rayposition, computed using a cosmological angular distanceat the mean group redshift. We used catalogues of galax-ies with photometric redshifts derived from the CFHT-LSWide imaging survey and inspected galaxy density maps inphotometric redshifts slices centered around each group. Aclear overlap between the X-ray isophotes and the densitymap at the (true) redshift of the source is expected. In thiscase, the nature of the X-ray source is confirmed as a galaxycluster and its redshift validated (”C1 confirmed” clusters).One source out of the 53 selected C1 candidates could nothave any related spectroscopic observation and falls outsideof the CFHT-LS footprint, preventing the derivation of aphotometric redshift. Noting that it was classified as a C2 http://ned.ipac.caltech.edu/, NASA/IPAC ExtragalacticDatabase. source by our former pipeline, it is discarded from this anal-ysis.Table 1 lists the C1 cluster catalogue including all 52sources validated as bona-fide C1 clusters for this work. Itrepresents a superset of Pacaud et al. (2007) and Adamiet al. (2011) C1 samples and a subset of the full XMM-LSS source catalogue (Chiappetti et al. 2013). All but one(98%) have spectroscopically validated redshifts. We notethat XLSSC 035 and XLSSC 048 are not ”C1 confirmed clus-ters” according to the definition above, because of their cur-rent low number of securely identified spectroscopic mem-bers (see Adami et al. 2011, for a discussion). The remainingcluster (XLSSU J021744.1-034536) is located at z phot = 1 . In order to obtain statistically relevant results from thepresent cluster sample, we derive a survey selection func-tion. It is straightforwardly associated to the detection pro-cess described in previous section and our procedure followsclosely Pacaud et al. (2006); Elyiv et al. (2012); Clerc et al.(2012b).
We first compute a selection function for each individualpointing. To this purpose, we designed simulations of realis-tic XMM observations containing point-sources representa-tive of the observed AGN logN-logS (Moretti et al. 2003) andextended sources. The latter are represented as β -modelswith β = 2 /
3, with varying core radii and fluxes, placed atrandom positions on the detectors. Our simulations span avariety of detector exposure times ( T exp = 3, 7, 10, 20, 40 ks)and particle background levels ( b = 0.25, 0.5, 1, 2, 4 timesthe nominal value of Read & Ponman 2003). For each valueof the exposure time ( T exp ), of the background value ( b ),cluster extent (input core-radius ext ) and input count-rate( cr ), 2500 to 3500 fake XMM pointings were produced, eachof them containing 4 to 8 extended sources. All simulatedobservations went through the same detection and charac-terization pipeline as used for real data. As discussed earlier,our simulations show that contamination of a C1 selectedsample by point-like and spurious sources is very low, rang-ing from 1% to 3% depending on exposure time and back-ground level. This contamination rate results from simula-tions where the point-source population is spatially uncor-related from the extended sources. We will discuss in Sect. 6an extreme case, in which point-sources systematically pop-ulate the centre of clusters in simulations. The observationalselection function is the probability of detecting and charac-terizing an extended source as a C1. Output detection lists c (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Table 1.
The XMM-LSS C1 sample of galaxy clusters. ( a : XLSSU J021744.1-034536). Quoted uncertainties reflect 68% confidenceintervals limits. (1) Cluster redshift, in brackets is the number of cluster members with spectroscopic redshifts and ”L” stands for”literature” (Abell 329). ( b : photometric redshift from Willis et al. 2013, c : see note in Table 2 of Adami et al. 2011). (2) Absorbed fluxin units 10 − ergs/s/cm measured in the [0.5-2] keV band in a circular aperture of 0.5 Mpc at the cluster redshift. (3) As computedfrom a M − T relation (see text). (4) Bolometric luminosity within R , units 10 ergs/s. (5) Rest-frame [0.5-2] keV luminosity,units 10 ergs/s. (6)-(7) Mass estimates, units 10 h − M (cid:12) from two different methods (Sect. 4.3.1 and 4.3.2). (8)
1: Pacaud et al.(2007), 2: Berg´e et al. (2008), 3: Adami et al. (2011), 4: Willis et al. (2013), 5: Abell, Corwin, & Olowin (1989). xlssc
R.A. Dec z [ N z ] R spec F [0 . − T X R L bol L [0 . − M M Ref.J2000 J2000 ( (cid:48)(cid:48) ) 0.5 Mpc (keV) (Mpc) (ergs/s) (ergs/s) M1 M2 (1) (2) (3) (4) (5) (6) (7) (8)
060 33.668 -4.552 0.14 [L] 360 116 . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . +1 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . +1 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . +1 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± . a b
45 1 . ± . . +5 . − . . ± . . ± . . ± . . ± . . ± .
06 0 . ± .
03 7.1 0.9 -077 34.522 -3.656 0.20 [4] 80 4 . ± . . +1 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . +3 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . +2 . − . . ± . . ± . c ] 27 1 . ± . . +1 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± .
05 2.5 2.3 -035 35.950 -2.858 0.07 [1 c ] 60 6 . ± . . ± . . ± .
03 0 . ± .
01 1.3 1.1 1,3028 35.985 -3.100 0.30 [8] 45 2 . ± . . +0 . − . . ± . . ± . . ± . . +1 . − . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . +1 . − . . ± . . ± . . ± . . ± . . ± .
04 0 . ± .
02 1.0 1.1 -053 36.114 -4.836 0.50 [7] 60 2 . ± . . +8 . − . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . +0 . − . . ± .
02 0 . ± .
01 1.3 1.0 1,3001 36.238 -3.816 0.61 [17] 90 7 . ± . . +0 . − . . ± . . ± . . ± . . ± . . ± .
04 0 . ± .
02 3.6 2.4 -008 36.337 -3.801 0.30 [11] 63 1 . ± . . +0 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . ± .
05 0.399 0 . ± .
01 0 . ± .
01 1.0 1.1 1,3052 36.568 -2.665 0.06 [5] 72 12 . ± . . +0 . − . . ± .
02 0 . ± .
01 1.4 1.1 1,3009 36.686 -3.684 0.33 [8] 54 2 . ± . . +0 . − . . ± . . ± . . ± . . +1 . − . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . +1 . − . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . +0 . − . . ± . . ± . (cid:13)000
01 1.4 1.1 1,3009 36.686 -3.684 0.33 [8] 54 2 . ± . . +0 . − . . ± . . ± . . ± . . +1 . − . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . +1 . − . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . +0 . − . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . +0 . − . . ± . . ± . (cid:13)000 , 1–21 N. Clerc et al. were correlated in position with input source lists, and theselection function P C , [ θ ,θ ] ( T exp , b, cr, ext ) was derived byselecting only sources satisfying the C1 criteria (see Sect. 2)and detected within a given a range of off-axis values [ θ , θ ]. The next step in characterizing the survey selection functionconsists of linking simulations to real observations. Each ac-tual XMM pointing is assigned its proper selection functionby interpolating along the ( T exp , b ) grid used for simulations.We estimated background levels ( b ) in real observations bymatching the average local background level around each de-tected point-like source to the value seen in simulated point-ings, similarly to Elyiv et al. (2012); Clerc et al. (2012b).The 8.1 deg consisting of non-overlapping area (i.e.unique to each pointing) was treated by averaging all 98individual pointing selection functions calculated up to 10 (cid:48) off-axis radius (i.e. [ θ , θ ] = [0 , (cid:48) ]). The remaining 3 deg mainly (97%) consist of area shared by exactly two adja-cent pointings, whose centers are spaced by approximately20-25 arcmin. Since cluster detection is performed indepen-dently on each pointing, we estimated the selection functionof each overlap between pointing A and pointing B by: P ( AB, overlap) = P ( A ) + P ( B ) − P ( A ) P ( B ) (1)where P ( A ) and P ( B ) are the respective selection functionsof each pointing, considering a [ θ , θ ] = [10 (cid:48) , (cid:48) ] range.Fig. 2 shows the averaged selection function of the 11 deg area. It takes into account each unique (non-overlapping)patch of the survey as well as overlaps, weighted by theirrespective geometrical area. The noticeable sharp decreasein probability for high-countrate, small-size, sources corre-sponds to a morphological misclassification into point-likesources by the automated algorithm.Such a formulation of the survey selection function onlydepends on observational quantities (apparent size and flux).Assuming a cosmological model and a set of mass-observablerelations, it can be rewritten in terms of a limiting mass as afunction of redshift. We provide an example in Fig. 3, assum-ing Planck 2013 cosmology and illustrate the changes due tothe choice of model, specifically the luminosity-temperaturerelation. An X-ray spectrum was extracted around each cluster po-sition in a circular aperture. Similarly to Pacaud et al.(2007), a background annulus is chosen so that it doesnot contain emission from the cluster. The spectral extrac-tion radius ( R spec ) is optimized on the basis of the signal-to-noise estimated from the cluster surface-brightness pro-file. Background-subtracted spectra are fitted with XSpec v.12.8.0 (Arnaud 1996) using a single-temperature APECplasma model (v.2.0.1) and assuming a galactic hydrogendensity column given by Kalberla et al. (2005). Metallicityabundances were fixed at 0.3 times the solar value, except
Figure 2.
The average XMM-LSS 11 deg C1 selection functionin X-ray observables domain, as derived from simulations of real-istic XMM-LSS observations. Contours represent the probabilityof detecting and classifying as C1 an extended source with a sur-face brightness profile following a β -model ( β = 2 /
3) of given coreradius and given total flux (or count-rate). Pointing to pointingdifferences in sensitivity and pointings overlaps are taken intoaccount and weighted according to their area on sky.
Figure 3.
The XMM-LSS 11 deg C1 selection function in mass-redshift plane, as derived from the observational function shownon Fig. 2. Planck CMB cosmology is assumed, conversion frommass to temperature follows Arnaud, Pointecouteau, & Pratt(2005) and two different luminosity-temperature relations aretested: M12 (Maughan et al. 2012, ’ALL’ sample) and P09 (Prattet al. 2009, ’ALL’ sample), both following self-similar evolution. for XLSSC 60 . The median temperature measured in thesample is 2.1 keV, with a typical uncertainty of ∼ Its spectrum contains enough photons to enable a simultaneousfit of temperature and metallicity and we find an abundance valueof 0 . ± . Z (cid:12) within R spec .c (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg A comparison with previously measured values for the29 clusters in common with Pacaud et al. (2007) is presentedin Appendix A, with an attempt to disentangle between thedifferent causes of discrepancies. We reach the conclusionthat a change in APEC models slightly impacts the tem-peratures for the coolest systems, while other results agreewell within the error bars: changes in the X-ray process-ing,
XSpec version and plasma models only create scatteraround the one-to-one relation.Assuming the M − T X relation of Sun et al. (2009)(converted into a R c − T X relation, see their Table 6 forthe ”Tier 1-2+clusters” sample), we assign a value of R c to our clusters: R (Mpc) = 0 . h − (cid:16) T (cid:17) . E ( z ) − (2)where E ( z ) = H ( z ) /H is the normalized Hubble constant. X-ray cluster fluxes were measured in the [0.5-2] keV banddirectly on images created from cleaned event lists. Twomethods were tested and compared: (i) modelling the sur-face brightness radial profile, and (ii) integrating the sourceflux in growing circular apertures (”growth curve analysis”B¨ohringer et al. 2004; ˇSuhada et al. 2012), as applied inAdami et al. (2011); Clerc et al. (2012b). We detail here ourprocedures.(i) The first method follows Pacaud et al. (2007) byassuming a one-dimensional β -model (Cavaliere & Fusco-Femiano 1976) and three free parameters: angular core-radius, β and normalization. A local background level is es-timated by means of a double-component model (vignettedand unvignetted) adjusted in a source-free area over each ofthe three XMM EPIC detectors. A local, one-dimensional,analytic PSF model (as in Arnaud et al. 2002) is convolvedto the model β -profile and accounts for the telescopes spa-tial resolution. Model and data profiles are binned to ensurea minimal 3- σ signal-to-noise ratio in each bin. χ statisticsprovide a best-fit value and confidence levels on a β -coreradius grid. The normalization is derived from the numberof counts collected in the fit area. Given the generally lowsignal-to-noise ratios of C1 clusters and the high backgroundlevels in XMM images, χ contours are degenerate in thetwo-dimensional β -core radius parameter space (e.g. Alshinoet al. 2010). We ranked χ contours and surface brightnessprofiles according to their level of degeneracy and found thatobjects with more than 300 net counts in the [0.5-2] keVband provide well-behaved χ surfaces. For those 30 clus-ters, the 3-parameter model is then considered as a gooddescription of its surface brightness profile. For the remain-ing 22 objects, we instead forced β = 2 / R spec in Fig. 4).(ii) The second method does not assume any model asfor the cluster surface brightness profile. Cluster emissivityis integrated in circular annuli of growing sizes around the Figure 4.
Comparison of [0.5-2] keV absorbed fluxes derived withtwo different methods for all clusters in the sample. Measurementsreported on the x-axis were obtained by fitting a β -model to theX-ray surface brightness profile. The y-axis corresponds to anaperture photometry measurement (growth curve analysis). Thiscomparison is performed within a radius R spec , defined as an op-timal signal-to-noise extraction region (see text). The inset zoomsover a crowded area in the plot. cluster centroid. Background subtraction is controlled viaan annulus whose size is adjusted by hand, and a double-component model (vignetted+unvignetted) is fitted over thepixels inside the annulus. This background model is thentransported at the source location. Surrounding sources de-tected by the detection pipeline are masked out and anyremoved area is accounted for by assuming circular symme-try of the count-rate profile. Since this method involves man-ual intervention, we checked for its robustness by comparingthe results of two independent measurers (T. Sadibekova,N. Clerc). In all 22 cases that underwent this comparisonwe found results in agreement within 1- σ error bars.Finally, conversion factors from count-rates to physi-cal, galactic-absorbed, fluxes were computed using XSpec and the best-fit APEC model derived from X-ray spectralanalysis (Sect. 4.1). Fig. 4 displays a comparison of fluxesmeasured by methods (i) and (ii). It shows that both agreewithin their 1- σ error bars. This comparison is shown withinthe radius R spec , specifically chosen for maximizing thesignal-to-noise ratio of each cluster emissivity (see Sect. 4.1and numerical values in Table 1). Hence, discrepancies dueto β -model extrapolation or background removal uncertain-ties are kept at their lowest level. Finally, we show the goodagreement between these calculations and previously pub-lished values (Pacaud et al. 2007) in Appendix A. We derived X-ray bolometric and [0.5-2] keV (rest-frame)luminosities by combining physical fluxes measured in the[0.5-2] keV (observer frame) band with the best-fit spectralmodel found in Sect. 4.1. We quote in Table 1 luminosities c (cid:13)000
Comparison of [0.5-2] keV absorbed fluxes derived withtwo different methods for all clusters in the sample. Measurementsreported on the x-axis were obtained by fitting a β -model to theX-ray surface brightness profile. The y-axis corresponds to anaperture photometry measurement (growth curve analysis). Thiscomparison is performed within a radius R spec , defined as an op-timal signal-to-noise extraction region (see text). The inset zoomsover a crowded area in the plot. cluster centroid. Background subtraction is controlled viaan annulus whose size is adjusted by hand, and a double-component model (vignetted+unvignetted) is fitted over thepixels inside the annulus. This background model is thentransported at the source location. Surrounding sources de-tected by the detection pipeline are masked out and anyremoved area is accounted for by assuming circular symme-try of the count-rate profile. Since this method involves man-ual intervention, we checked for its robustness by comparingthe results of two independent measurers (T. Sadibekova,N. Clerc). In all 22 cases that underwent this comparisonwe found results in agreement within 1- σ error bars.Finally, conversion factors from count-rates to physi-cal, galactic-absorbed, fluxes were computed using XSpec and the best-fit APEC model derived from X-ray spectralanalysis (Sect. 4.1). Fig. 4 displays a comparison of fluxesmeasured by methods (i) and (ii). It shows that both agreewithin their 1- σ error bars. This comparison is shown withinthe radius R spec , specifically chosen for maximizing thesignal-to-noise ratio of each cluster emissivity (see Sect. 4.1and numerical values in Table 1). Hence, discrepancies dueto β -model extrapolation or background removal uncertain-ties are kept at their lowest level. Finally, we show the goodagreement between these calculations and previously pub-lished values (Pacaud et al. 2007) in Appendix A. We derived X-ray bolometric and [0.5-2] keV (rest-frame)luminosities by combining physical fluxes measured in the[0.5-2] keV (observer frame) band with the best-fit spectralmodel found in Sect. 4.1. We quote in Table 1 luminosities c (cid:13)000 , 1–21 N. Clerc et al. measured within R as estimated for each cluster (Eq. 2).The median bolometric (resp. soft-band) luminosity of oursample is 4 . × ergs/s (resp. 2 . × ergs/s) within R . Individual uncertainties are dominated by count-ratemeasurement uncertainties (including background removal)and their median level is 11%. The ultimate quantity of interest describing galaxy clustersis their individual, total, mass as derived from X-ray data.Given the low signal-to-noise ratios associated to each clus-ter, we choose to only provide rough estimates and no errorbars. However, we illustrate results obtained by two differentmethods, each one depending on different observables.
Similarly to Pacaud et al. (2007) we can estimate the masswithin R by using:(i) a mass-temperature relation (Sun et al. 2009, seeEq. 2) in order to obtain a value of R ,(ii) the best-fit β -model profile injected into the equationof hydrostatic equilibrium under the assumption of isother-mality of the intra-cluster medium (e.g. Ettori 2000), whichreads: M ( M (cid:12) ) = 1 . × β R c T x x , (3)where β and R c (Mpc) are the best-fit β -model parame-ters found in Sect. 4.2, T is expressed in keV and x = R c /R .Mass estimates obtained by this method are listed inTable 1 under column label ”M1”. In principle, we note thatassuming a R − T relation is unnecessary and redundantsince the equation linking R to M is univocal. We dis-cuss our choice further in Appendix B. The second method starts from growth-curve flux measure-ments. They provide luminosities integrated within a cylin-drical aperture of any given size. We make use of a L X − M relation from Sun (2012) and iteratively find the value of R that leads to a converged set of L X, and M val-ues. This approach is similar to the one presented by ˇSuhadaet al. (2012) for the XMM-SPT cluster sample. However, incontrast to the XMM-SPT clusters, temperature measure-ments are available in this work. They are used to model thecluster X-ray spectrum (using APEC v.2.0.2) and convertfrom instrumental count-rates to physical quantities. Wechecked that this iterative analysis returns values of L X, consistent with the computation described in Sect. 4.2.2within their respective error bars, although the underlyingscaling relations used in deriving the radius R differ. Val-ues obtained with this method are quoted in Table 1 undercolumn label ”M2”. Figure 5.
Comparison of mass estimates for the 52 clusters inthis sample. These are obtained with two independent method,both based on X-ray data:
Method 1 assumes hydrostatic equilib-rium and a gas distribution following a β -model. Method 2 relieson a given luminosity-mass relation (Sun 2012). Clusters locatedright of the dashed line have their surface brightness profile well-described by a 2-parameter β -model, while clusters on the leftwere imposed β = 2 / Figure 5 compares mass estimates obtained from these twomethods. Although the scatter is high and differences up toa factor of ∼ k -correction), while it directlyenters Method 1 (see Eq. 3). The correlation between thesevalues thus reflects the intrinsic correlation between mass,luminosity and temperature in the intra-cluster medium.Based on these results, we find that the median mass ( M c )of clusters in the present sample is ∼ × h − M (cid:12) . Fig. 6 shows the redshift distribution of all 52 sources en-tering this analysis (plain blue histogram). Typical uncer-tainties on cluster redshift are much smaller than the binsize of ∆ z = 0 . × − ,Adami et al. 2011). A model distribution is superimposed,whose derivation follows the steps presented in Pacaud etal. (2007); Pierre et al. (2011) (see also Clerc et al. 2012a).In brief, a model halo distribution (Tinker et al. 2008) isprojected onto the X-ray observable space (+redshift) us-ing standard scaling relations (Arnaud, Pointecouteau, &Pratt 2005; Maughan et al. 2012) and XMM detector re-sponses. Evolution of these scaling relations follows a self-similar model. Our observational selection (shown on Fig. 2)filters out undetectable clusters. To this purpose it assumes c (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure 6.
The XMM-LSS 11 deg C1 redshift distribution in∆ z = 0 . ±
12 sources. 1- σ uncertaintiesare computed analytically and account for both shot-noise (redbars) and sample variance (see text and Valageas et al. 2011). a fixed physical core-radius of 180 kpc for all clusters. Theresulting distribution is further integrated over all quan-tities but redshift and provides the expected dn/dz dis-tribution. This quantity is summed in finite redshift binsand uncertainties on the histogram are computed followingthe formalism described in Valageas et al. (2011). They ac-count for both shot-noise (i.e. √ n i ) and sample variancein each bin. These uncertainties are almost uncorrelated(e.g. C ij,i (cid:54) = j (cid:28) (cid:112) C ii C jj ). Given the number of assumptionsinvolved here, Sect. 6 will discuss further the hypothesesentering the derivation of this model histogram. L X − T scaling relation and evolution The luminosity-temperature relation ( L X − T ) is one of themain studied scaling laws of X-ray clusters (see e.g. Ar-naud & Evrard 1999; Pratt et al. 2009; Mittal et al. 2011;Maughan et al. 2012; Takey, Schwope, & Lamer 2013). It re-flects the history of heating and cooling of the intra-clustergas (see e.g. Voit 2005) and relates two major cluster prop-erties ultimately linked to the total cluster mass (Kaiser1986). Taking advantage of the wide redshift range spannedby our sample, we address the evolution in the normalizationof this scaling law assuming various local relations selectedin the literature. We pay particular attention to the role ofselection effects in this analysis. L X − T relation in the sample Figure 7 shows the relationship between the bolometric lu-minosity and the temperature measured for the 52 clustersin this sample. The X-ray luminosity correlates well with gastemperature in all three redshift slices displayed on Fig. 7as expected from basic scaling arguments (Kaiser 1986). Weplot on the same figure a selection of recent scaling relations (Arnaud & Evrard 1999; Pratt et al. 2009; Maughan et al.2012; Sun 2012). Although these scaling relations were de-rived with samples of clusters spanning relatively wide red-shift ranges, the numerical values corresponding to these re-lations were published for the local Universe ( z = 0). Hencein the following we will refer to them as ”local” scaling re-lations. Each of them is evolved self-similarly to the medianredshift in each panel of Fig. 7 ( z = 0 . , . , . L X ( T, z ) = L ( T, × E ( z ). The slope of these extrapolatedrelations appears compatible with our data points, as is thenormalization. However, this visually good agreement hasto be checked against selection effects and we describe ourfindings in the next paragraph. L X − T normalization and impactof selection biases Since many objects in the sample are close to the detectionthreshold, it is necessary to correctly account for selectioneffects before any attempt to interpret the data point distri-bution in Fig. 7 (e.g. Pacaud et al. 2007; Mantz et al. 2010;Allen, Evrard, & Mantz 2011; Reichert et al. 2011). A typi-cal misinterpretation can arise from Malmquist bias: intrin-sically brighter objects are favorably present in the sample,which translates into an average luminosity at a given tem-perature and redshift higher than the true expected lumi-nosity. Such a bias is increasingly important as the intrinsicscatter in the studied scaling law is high. This is the case forthe L X − T relation (of the order of σ ln L | T ∼ .
6, e.g. Prattet al. 2009). We follow the same approach as in Pacaud etal. (2007) for studying the evolution in the normalization ofthe relation from our sample. Namely, we model at each red-shift the L X − T relation assuming a local ( z = 0) relationand a normalization following E ( z ) (1 + z ) α . The exponent α is a free parameter in the analysis. The scatter σ ln L | T iskept at its z = 0 value. The resulting distribution is foldedwith the survey selection function (Fig. 2) after passing itthrough the XMM instrumental response. The likelihood ofeach cluster being drawn from this particular model is com-puted taking into account uncertainties on the temperature and neglecting uncertainties on luminosities.Repeating this procedure for a range of values in α en-ables the derivation of 68% confidence intervals for α , asquoted in Table 2 (under column ”corrected”). The result-ing value depends on the assumed local ( z = 0) scalingrelation because of the differences in their slopes, normal-izations and scatters. Fig. 8 visually illustrates our results.The disagreement between data points and best-fit scalingrelations is only apparent and due to the impact of selectioneffects. This is reflected in the numbers shown in Table 2(column ”uncorrected”) where the same procedure is ap-plied, but the selection function is artificially neglected (i.e.we assume that P C = 1). In the latter case, the normaliza-tion is found to be positive or mildly negative (with respectto self-similar evolution, α = 0), while a proper account forselection effects clearly hints toward a negative evolution.This result is in agreement with the findings of Reichert et In truth, the full C-statistic given by
XSpec from the spectralfitting and converted into a probability distribution as a functionof T X .c (cid:13)000
XSpec from the spectralfitting and converted into a probability distribution as a functionof T X .c (cid:13)000 , 1–21 N. Clerc et al.
Figure 7.
The luminosity-temperature relation of our cluster sample. Luminosities are bolometric and measured within R as inferredfrom a mass-temperature relation (see text). Temperatures are measured within an aperture maximizing the signal-to-noise of the spectralfit. Several local scaling relations are overplotted for comparison, each being evolved following a self-similar evolution L X ( T, z ) ∝ E ( z ):P09 (Pratt et al. 2009), AE99 (Arnaud & Evrard 1999), S12 (Sun 2012) and M12 (Maughan et al. 2012). A typical σ ln L | T = 0 . z = 0 . , . , . Table 2.
The evolution of the L bolX, − T relation measuredfrom the sample of clusters presented in this work (see Fig. 8).Different local ( z = 0) scaling relations are assumed: P09 (Prattet al. 2009), AE99 (Arnaud & Evrard 1999), S12 (Sun 2012), M12(Maughan et al. 2012). Their normalization is evolved following E ( z ) . (1 + z ) α . ”Corrected” (”Uncorrected”) refers to the best-fitvalue of α found with (without) accounting for selection biases.Best-fit α Best-fit α Local scaling relation Corrected UncorrectedP09 ”ALL, L − T , BCES Ortho.” − . ± . − . ± . σ ln L | T = 0 . − . ± . . ± . L − T S+R” − . ± . − . ± . − R ” (M12) − . ± . . ± . al. (2011); Hilton et al. (2012); Clerc et al. (2012b), althoughthese studies differ in their treatment of selection effects andmodelling.We note here that self-similar evolution does not nec-essarily imply a ∝ E ( z ) scaling of the L X − T normaliza-tion (Maughan 2014). This actually depends on the assumedslope of the M gas − M scaling relation and Maughan (2014)find a E ( z ) . behavior instead. The present significance ofthe negative evolution is thus reduced when expressed rela-tive to this particular scaling. The sample presented in this work is unique in several re-spects: it is drawn from an homogeneous, contiguous, X-raysurvey, it is 98% complete in terms of spectroscopic red-shift availability, spans a wide range of redshifts and massesand the sample selection function is well understood. This issummarized on Fig 9, where each C1 cluster from Table 1 isdrawn at its location in a comoving coordinates frame (the observer is located at the origin and the line of sight is the z -axis). We note that the number of objects (52) presentedin this study is too low to enable robust, quantitative, inter-pretations of the 3-dimensional distribution of clusters in thevolume (e.g. correlation function analysis). However, this isan open window for the on-going XMM-XXL survey (Pierreet al. 2011, Pierre et al. in prep.), which multiplies the sur-veyed area by a factor 5 (in two separate areas).Figure 10 shows the location of galaxies detected in twolarge spectroscopic samples: BOSS-DR10 (Ahn et al. 2013)and VIPERS-PDR1 (Garilli et al. 2013; Guzzo et al. 2013).The declination range was shrunk to only display objects inthe common sky area. Moreover, radial selection effects differfrom one survey to the other, in particular, VIPERS galaxiesare preferentially selected in the 0 . (cid:46) z (cid:46) We focus now on modelling the redshift distribution of clus-ters as presented in Sect. 5.1 and Fig. 6. The first redshiftbin (0 < z < .
1) contains 6 groups (XLSSC 035, 062, 021,054, 011, 052) of low mass and small size. It is thus likelythat our assumption of a fixed core-radius value (180 kpc)fails in this redshift range. For example, assuming a core-radius size of half this value roughly doubles the numberof predicted clusters in this bin. However, given the smallvolume of Universe involved, we refrain from deriving quan-titative results from this bin. Hence, the following resultsrely on the 46 clusters with z > . c (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure 8.
Same data points as Fig. 7. The evolution of local scaling relations now takes into account selection effects – in particularMalmquist bias – and corresponds to our best-fit, 1-parameter, model L X ( T, z ) ∝ E ( z ) . (1 + z ) α . The corresponding best-fit values of α are listed in Table 2 (column ”corrected”). Figure 9.
The tridimensional distribution of XMM-LSS C1 clusters viewed from two perpendicular directions, both orthogonal to theline-of-sight ( z -axis). The observer is located at ( x, y, z ) = (0 , ,
0) in this comoving coordinate system. Each cluster is represented by asymbol whose size is proportional to R as inferred from X-ray temperature measurements (see Eq. 2). The approximate XMM-LSSsurvey boundaries are materialized by two solid lines.c (cid:13)000
0) in this comoving coordinate system. Each cluster is represented by asymbol whose size is proportional to R as inferred from X-ray temperature measurements (see Eq. 2). The approximate XMM-LSSsurvey boundaries are materialized by two solid lines.c (cid:13)000 , 1–21 N. Clerc et al.
Figure 10.
The tridimensional distribution of XMM-LSS C1 clusters (blue circles, sizes proportional to R ) in the context of galaxyspectroscopic surveys. Green triangles stand for galaxies in the SDSS DR10 release (Ahn et al. 2013). Red points stand for VIPERSobjects with flags between 2.X and 9.X (Garilli et al. 2013). Only objects with 33 < R . A <
38 deg and − . < δ < − . (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure 11.
Same figure as Fig. 6, with WMAP-9yr cosmologyreplaced by Planck 2013 CMB cosmology (Planck Collaborationet al. 2013a) in the model derivation (plain line and errors). Thismodel predicts 108 ±
14 C1 clusters in the 0 (cid:54) z (cid:54) The redshift distribution model of Fig. 6 presents a roughagreement with the observed redshift histogram of the sam-ple. Rebinning the histograms such that each model bin con-tains at least 5 objects, we compute a χ value of 18.0 using10 bins. Given that P ( χ (cid:54) .
0) = 96%, we conclude on amarginal agreement between data and this model.Updating the cosmological model to the best-fit modelderived from Planck CMB results (Planck Collaboration etal. 2013a), we obtain a new curve, as shown in Fig. 11. Be-cause of higher values of σ (0.834 instead of 0.821) andΩ m (0.316 instead of 0.279), the total number of predictedclusters is higher in each ∆ z bin. Taking into account uncer-tainties in each bin, we find a χ of 28.0 (in 12 bins). Since P ( χ (cid:54) .
0) = 99 . ∼ .
3% (equivalent to a3- σ rejection).Fig. 11 suggests that the strongest disagreement be-tween data and model occurs in the 0 . (cid:46) z (cid:46) . z ∼
1, ruling out a severe sensitivity effect. Limitingto this ”ad hoc” redshift range provides a χ of 15.4 in 5bins, also leading to model rejection with false-alarm errorprobability of 0.4%. Taking these results at face value letus postulate the existence of a substantial lack of massivestructures around z ∼ .
7. This deficit would be rare enoughto be marginally accounted for in our sample variance calcu-lations. However, in what follows we discuss several physicaleffects that can be held responsible for this observation andargue for a combination of observational and physical effects. L X − T evolution and redshift histogram Model redshift distributions shown in Figs. 6 and 11 dependon the exact scaling relations used for converting clustermasses into observed properties. This is particularly truefor the luminosity-temperature relation, which is involved in
Figure 12.
Same figure as Fig. 11, but assuming an evolution ofthe luminosity-temperature following our best-fit result, α = − . α = 0). Themodel predicts 56 ±
10 C1 clusters. the computation of a cluster mass from its apparent flux. Asshown in Sect. 5.2, our dataset prefers a negative evolutionof the L X − T , namely a normalization evolving as E ( z ) . (1+ z ) − . ± . instead of E ( z ). Fig 12 demonstrates the impactof such an evolution on the modeled redshift distribution,still under the assumption of a Planck 2013 CMB cosmology(i.e. equivalent to Fig. 11). With this evolution, clusters ofa given mass become fainter with increasing redshift, thuswe expect less detections at higher redshifts, simply due toa dimming of these objects. In this case, the χ value is 6.1(in 8 bins), sufficiently low to prevent the model rejection.Hence, the additional assumption of a non self-similar L X − T reconciles the observed redshift histogram with our modeland reduces the significance of the central gap.We note that a complete analysis should simultaneouslyfit a cosmological model and a scaling relation model in orderto properly account for Eddington bias (e.g. Mantz et al.2010). This kind of self-consistent analysis in the signal-to-noise regime of ”XMMLSS-like” clusters is particularly wellhandled by the z -CR-HR method presented in Clerc et al.(2012a) and will be applied to the much larger sample ofXMM-XXL clusters.Finally, we stress that previous results were derived as-suming a local L X − T from Maughan et al. (2012). Theother three local scaling laws considered earlier (Table 2)lead to even higher cluster densities, hence in stronger dis-agreement with this dataset (see also Clerc et al. 2012b fora discussion). Simplifying hypotheses entering the derivation of our selec-tion function do not account for the true physical nature ofgalaxy clusters and possibly impact the model redshift dis-tribution shown on Fig. 6 and 11. We explore in this sectionthe impact of an evolving AGN contamination fraction inthe center of galaxy clusters by means of simple simulationsand a demonstrative toy-model. c (cid:13)000
10 C1 clusters. the computation of a cluster mass from its apparent flux. Asshown in Sect. 5.2, our dataset prefers a negative evolutionof the L X − T , namely a normalization evolving as E ( z ) . (1+ z ) − . ± . instead of E ( z ). Fig 12 demonstrates the impactof such an evolution on the modeled redshift distribution,still under the assumption of a Planck 2013 CMB cosmology(i.e. equivalent to Fig. 11). With this evolution, clusters ofa given mass become fainter with increasing redshift, thuswe expect less detections at higher redshifts, simply due toa dimming of these objects. In this case, the χ value is 6.1(in 8 bins), sufficiently low to prevent the model rejection.Hence, the additional assumption of a non self-similar L X − T reconciles the observed redshift histogram with our modeland reduces the significance of the central gap.We note that a complete analysis should simultaneouslyfit a cosmological model and a scaling relation model in orderto properly account for Eddington bias (e.g. Mantz et al.2010). This kind of self-consistent analysis in the signal-to-noise regime of ”XMMLSS-like” clusters is particularly wellhandled by the z -CR-HR method presented in Clerc et al.(2012a) and will be applied to the much larger sample ofXMM-XXL clusters.Finally, we stress that previous results were derived as-suming a local L X − T from Maughan et al. (2012). Theother three local scaling laws considered earlier (Table 2)lead to even higher cluster densities, hence in stronger dis-agreement with this dataset (see also Clerc et al. 2012b fora discussion). Simplifying hypotheses entering the derivation of our selec-tion function do not account for the true physical nature ofgalaxy clusters and possibly impact the model redshift dis-tribution shown on Fig. 6 and 11. We explore in this sectionthe impact of an evolving AGN contamination fraction inthe center of galaxy clusters by means of simple simulationsand a demonstrative toy-model. c (cid:13)000 , 1–21 N. Clerc et al.
The presence of a central AGN in galaxy cluster is a commonsource of concern for extended source detection algorithms.A central, blended, point-like source increases the total fluxof a cluster in X-rays, thus increasing its detection proba-bility. However, beyond a certain flux the detected sourcecan no longer be classified as a ”secure” extended sourceon morphological grounds only, given the sharply peakedprofile of the blend. We modified our set of XMM clustersimulations by adding a point-source with varying flux atthe center of each β -model. We processed these simulationswith the exact same methodology as for real data and de-rived the C1 detection probability of a cluster as a functionof its count-rate, extent and central contamination.Our raw results are displayed on Fig. 13. As expected,a very bright central AGN (e.g. 80% contamination) causesa misclassification of the detected source and a decreasein the C1 detection probability. On the other hand, a 10-20% contamination in relatively bright and extended clus-ters (e.g. 0 . − .
05 cts/s and 40 − (cid:48)(cid:48) core-radius) slightlyenhances the detection probability (”flux boosting”), whilebright, compact clusters are less affected at this level of con-tamination. The increase in detection with AGN contamina-tion is conspicuous for faint (e.g. 0 . − .
01 cts/s) clusters,provided their morphology is not too compact and the AGNflux remains reasonable (below ∼ dn/dz We fold the modified selection functions of Fig. 13 into themodel predicting C1 cluster redshift distribution. Resultsare shown in Fig. 14. Each separate histogram shows theexpected density of clusters assuming that all of them arecontaminated by a central AGN at a given level (from 0% to80%) From this analysis, we conclude that high contamina-tion rates dramatically reduce the number of C1 detectionsbecause of misclassifications of the sources. On the otherhand, a slight increase in number of C1 is barely perceptiblefor light AGN contamination (5-10%), lowering the impactof ”flux boosting” on C1 detection. Relying on this sim-ple model, we conclude that mild ( > . < z <
1) redshift range, but we rule out the possibilitythat central point-source flux excess is the cause for the ap-parent increase at z ∼ z ∼ . . ± .
25 and almost uncontaminated in the remaining Note this figure does not use the full 11 deg selection func-tion but the one computed for a 10 ks pointing with nominalbackground level. Figure 14.
Impact of central AGN contamination on the mod-eled C1 redshift distribution. These curves are obtained by foldingthe selection functions of Fig. 13 into our model predicting thenumber density of C1 detections in a typical XMM pointing. Thepercentages correspond to different levels of central AGN contam-ination, assuming that it arises in all clusters at all redshifts.
Figure 15.
Impact of an evolving fraction of central AGN con-tamination onto the modeled cluster redshift distribution. Bluehistogram corresponds to this sample. Red dashed line is identi-cal to Fig. 11, i.e. corresponds to a model in which all clustersare AGN-free. The black line follows a similar model, with a sup-plementary assumption that clusters contain a central AGN. TheAGN relative flux is evolving with redshift as shown in the inset. redshift intervals. This simple model creates a ”gap” in theredshift distribution, simply because of selection effects.
The toy-model depicted in the previous section is an ”adhoc” illustration of the impact of selection effects on theredshift histogram of objects we classify as galaxy clusters.Recent findings in the field of X-ray surveys show that: i) X-ray AGN preferentially live in 10 . M (cid:12) haloes (see e.g.Cappelluti, Allevato, & Finoguenov 2012, for a review) andii) the density of low-luminosity AGN ( L X (cid:46) ergs/s)peaks at redshifts ∼ . − c (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure 13.
Impact of central AGN contamination on the C1 selection function. These curves show the change in C1 detection probabilityof a cluster as a function of a central AGN flux as compared to the probability at zero contamination level. They are derived from realisticsimulations of β -model sources with core-radius R core and various count-rates on the XMM detectors ( CR clu ). Each curve is normalizedto its maximal value and is computed for a typical 10 ks exposure with nominal detector background. of AGN flux in low-mass clusters at redshifts around ∼ . N H and the source luminosity in the [2-10] keV band. As-suming Γ = 1 . N H (10 − cm − ) and L X (10 − ergs/s) and we obtain the total[0.5-2] keV flux emitted by AGN as a function of redshift.As already shown in Ueda et al. (2014), it peaks at z ∼ . (cid:54) M b / ( h − M (cid:12) ) (cid:54) .We finally derive the typical AGN-to-cluster flux ratio inthe [0.5-2] keV band. We assume a typical cluster radius of 0.5 Mpc (hence a volume ∼ . ) and a linear scaling ofAGN number density with matter density (i.e. the numberof AGN per unit volume is x times higher in clusters thanin the field, x being of the order a few hundreds). Integrat-ing this ratio weighted by the number of clusters at a givenredshift provides the series of curves shown on Fig. 16.This model suggests an enhancement of the flux ratiobetween AGN and cluster around z ∼ .
4, and a slight de-crease at higher redshifts. The order of magnitude of thecontamination level must reach ∼ −
50% in order to ac-count for the gap in the C1 redshift histogram (Fig. 15): thisis in rough agreement with the level calculated in our simplemodel (e.g. for x = 500 and L X > ergs/s). Althoughinstructive, this model presents two main caveats, as it ne-glects interactions between AGNs and clusters. First, phys-ical mechanisms activating cluster central AGN differ fromthose in the field (e.g. Fabian 2012). Moreover, AGN in clus-ters may not be centrally concentrated and more likely tospread within the cluster volume (e.g. Branchesi et al. 2007;Haines et al. 2012). c (cid:13)000
50% in order to ac-count for the gap in the C1 redshift histogram (Fig. 15): thisis in rough agreement with the level calculated in our simplemodel (e.g. for x = 500 and L X > ergs/s). Althoughinstructive, this model presents two main caveats, as it ne-glects interactions between AGNs and clusters. First, phys-ical mechanisms activating cluster central AGN differ fromthose in the field (e.g. Fabian 2012). Moreover, AGN in clus-ters may not be centrally concentrated and more likely tospread within the cluster volume (e.g. Branchesi et al. 2007;Haines et al. 2012). c (cid:13)000 , 1–21 N. Clerc et al.
Figure 16.
Outcome of a model for AGN contamination in low-mass clusters. This set of curves shows the typical ratio of AGNto cluster [0.5-2] keV flux based upon luminosity functions fromUeda et al. (2014) and our model for cluster evolution in the rangeof mass 10 − M (cid:12) . We assume that AGN populate the inner0.5 Mpc of clusters with a volume density x = 200 or 500 timesthe field volume density. Different lines show different thresholdsfor the AGN [2-10] keV luminosity. The predicted redshift distribution of XMM-LSS clustersdiscussed so far relies on the assumption of a particular sur-face brightness profile for the clusters. Indeed, the selectionfunction as shown in Fig. 2 is a function of the cluster an-gular core-radius. This core-radius is relative to a β = 2 / R c = 180 kpc. Similarly toSection 6.3, we investigate the impact of such an hypothesison the predicted redshift histogram.In a first step, we relaxed the assumption on β and cre-ated a new set of simulations with different values: β = 0 . β . Unsurprinsingly,changing the value of β at fixed core-radius and flux changesthe C1 detection probability. Higher values lead to sharplypeaked profiles, hence increased detection rates. Very con-centrated clusters ( R c = 10 (cid:48)(cid:48) ) with high β values ( (cid:38) . β -profile with R c = 180 kpc anddifferent β . As precedently in the case of AGN contamina-tion, these results can be interpreted in two complementaryways. First, a possible explanation for the lack of clustersin the current sample can be attributed to selection biasesdue to a (evolving) variation of surface brightness profiles inthe cluster population. Secondly and conversely, statisticalsurface brightness analyses of a (X-ray selected) cluster sam-ple must correctly take into account selection biases, sincesuch a sample may be biased towards sharply peaked profiles(e.g. high values of β or small core-radius values.)In order to estimate the impact of a centrally luminous, cool-core, cluster on our selection function, we followed Eck-ert, Molendi, & Paltani (2011) and modeled such objectswith a ”double- β ” model. Namely, a broad β = 2 / β = 2 / z (cid:46) . Chandra data they foundno change in the cooling properties of those clusters (centralentropy and cooling time). However, they found an evolutionin the distribution of surface brightness profiles, namely adeficit of cuspy, cool-core clusters at high redshifts ( z > . . (cid:46) z (cid:46) . z > .
8, hence leading toa ”gap” in the observed redshift distribution of C1 clusters.Proper assessment of such important evolution trends willideally be addressed by comparing multi-wavelength clusterselection, by targeted
Chandra follow-up observations of thesample and by investigation of numerical simulations.
This work presents a detailed study of a complete sampleof 52 galaxy clusters. All are detected in a X-ray contiguoussurvey with XMM-Newton and covering 11 deg on sky. Weused an improved version of the well-qualified XAmin al-gorithm (Pacaud et al. 2006) to reduce event lists, produceimages in the [0.5-2] keV band and detect sources in theX-ray data. Relying on a set of extensive, realistic simula-tions, we defined a clean and uncontaminated ”C1” sampleof extended sources. A series of follow-up optical observa-tions confirmed the nature of these galaxy clusters. All butone are spectroscopically confirmed and their redshift is de-termined with high accuracy (∆ z < . R , ra-dius estimated from scaling relations. We finally computed c (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure 17.
Impact of varying surface brightness (SB) profiles onthe modeled C1 redshift distribution. These curves are obtainedby folding a set of modified selection functions accounting fordifferent SB profiles into our cosmological model. Our referencesingle- β = 2 / β models appear with dashed lines. An additionalSB model with two superimposed β = 2 / β models, 170 and40 kpc core-radii for the double-component profile). cluster total mass estimates with two different methods, onerelying on the assumption of intra-cluster medium isother-mality and hydrostatic equilibrium, the other on a scalingrelation between luminosity and mass. We interpret theirrough agreement as a reflection of the underlying X-ray scal-ing relations between mass, temperature and luminosity, andas a consequence of the consistency in our measurements.We modeled the C1 cluster selection function acrossthe wide survey area using a set of synthetic simulationsand found a theoretical limiting mass ( M c ) of 1 − × h − M (cid:12) (80% detection probability), depending on red-shift and detailed assumptions of the mass-to-observableconversion. Folding this selection function into a cosmologi-cal model allowed us to compare the observed redshift distri-bution of clusters in the sample to theoretical expectations.Accounting for uncertainties due to small number statis-tics and cosmological sample variance, we find a marginalagreement between the predicted model and the observeddistribution, and we note that results depend on the choiceof model. In particular, assuming the Planck CMB cosmo-logical model leads to predict too high a density of objectsat 0 . < z < .
9, as compared to the current sample.We compared several bolometric luminosity-temperature relations extrapolated from the recentliterature to our data points. Taking advantage of ourknowledge of sample selection effects, we suggest a simpleparametrization for the evolution of the normalization ofthe L X − T and confirm a negative trend with respectto a pure self-similar evolution. This result is observedin numerical simulations (e.g. Short et al. 2010), whenpre-heating of the intra-cluster medium occurs at the earlystage of cluster formation. Interestingly, different assump-tions for local ( z = 0) scaling relations lead to differentresults for their evolution (see numbers quoted in Table 2). This is due to their different slopes and normalizationsand the fact that our sample spans different loci of the L X − T plane as redshift increases. Indeed, selection effectsmove the high-redshift sample to higher temperaturesand luminosities. Importantly, we confirm that mistakenlyneglecting selection effects substantially changes theseconclusions and leads to a quasi self-similar evolution.We concluded this study by investigating the reality ofthe apparent ”gap” in the redshift distribution of clustersbetween 0 . (cid:46) z (cid:46) . • Considering the negative evolution of the L X − T re-lation found earlier, we find a milder disagreement betweenthe Planck 2013 cosmology model and our dataset. We notehowever that a complete analysis should self-consistently ad-dress the cosmological model and the evolution of scalingrelations. • We explored to what extent the observed redshift dis-tribution can be explained in terms of a selection bias dueto central point-source contamination of clusters. Based onour simulations, we cannot attribute the increase of clusterdensity at z = 1 to a detection enhancement (”flux boost-ing”). We propose instead an ad-hoc scenario in which theAGN contamination evolves with redshift, both in its oc-currence and its strength and peaks at 0 . (cid:46) z (cid:46)
1. In-terestingly, combining our cluster population model to theluminosity function of AGN in the Universe points towards asimilar trend, although a more thorough modeling is needed(in paticular using numerical simulations). • We finally described the impact of various surfacebrightness profiles on the C1 selection function. Our resultsshow that the enhanced presence of cuspy, cool-core clustersat low redshifts (e.g. at z < .
75, as observed by McDon-ald et al. 2013) could also lead to an apparent gap in theobserved C1 redshift distribution, since cool-core clusters in0 . (cid:46) z < .
75 would be considered as less likely extendedsources by detection algorithms.Comparison of cluster catalogues selected in X-ray andother wavelengths (e.g. optical and S-Z) will confirm thesepossible scenarios. So will do detailed studies with the
Chan-dra observatory by assessing the point-source content andsurface brightness shape evolution of those galaxy clusters.Future results will therefore rely on the larger XMM-XXLsurvey (50 deg at a similar depth), separated in two inde-pendent fields in order to beat sample variance. Echoing ourdiscussion on selection effects and their impact on cosmolog-ical observables (here dn/dz ), a large effort has been under-taken within the XMM-XXL team in order to develop andcompare multi-wavelength cluster selections, to perform de-tailed follow-up of selected samples and analyses of clustersfrom numerical simulations. The full-sky survey of eRosita(Predehl et al. 2010; Merloni et al. 2012) starting early 2016will bring the statistical power ( ∼ X-ray galaxy clus-ters up to z (cid:38)
1) needed to break degeneracies between themultiple assumptions entering the cosmological analysis ofX-ray cluster surveys and will offer an unprecedented tridi-mensional view on the large-scale structure as traced by itsmost massive constituents. c (cid:13)000
1) needed to break degeneracies between themultiple assumptions entering the cosmological analysis ofX-ray cluster surveys and will offer an unprecedented tridi-mensional view on the large-scale structure as traced by itsmost massive constituents. c (cid:13)000 , 1–21 N. Clerc et al.
ACKNOWLEDGMENTS
REFERENCES
Abell G. O., Corwin H. G., Jr., Olowin R. P., 1989, ApJS,70, 1Adami C., et al., 2011, A&A, 526, A18Ahn C. P., et al., 2013, arXiv, arXiv:1307.7735Allen S. W., Evrard A. E., Mantz A. B., 2011, ARA&A,49, 409Alshino A., Ponman T., Pacaud F., Pierre M., 2010, MN-RAS, 407, 2543Arnaud K. A., 1996, ASPC, 101, 17Arnaud M., Evrard A. E., 1999, MNRAS, 305, 631Arnaud M., et al., 2002, A&A, 390, 27Arnaud M., Pointecouteau E., Pratt G. W., 2005, A&A,441, 893Berg´e J., et al., 2008, MNRAS, 385, 695Benson B. A., et al., 2013, ApJ, 763, 147Bertin E., Arnouts S., 1996, A&AS, 117, 393Burenin R. A., Vikhlinin A., Hornstrup A., Ebeling H.,Quintana H., Mescheryakov A., 2007, ApJS, 172, 561Chiappetti L., et al., 2013, MNRAS, 429, 1652Biviano A., Katgert P., Mazure A., Moles M., den HartogR., Perea J., Focardi P., 1997, A&A, 321, 84B¨ohringer H., et al., 2000, ApJS, 129, 435B¨ohringer H., et al., 2001, A&A, 369, 826B¨ohringer H., et al., 2004, A&A, 425, 367Borgani S., et al., 2001, ApJ, 561, 13Branchesi M., Gioia I. M., Fanti C., Fanti R., CappellutiN., 2007, A&A, 462, 449Cappelluti N., Allevato V., Finoguenov A., 2012, AdAst,2012, 1Cavaliere A., Fusco-Femiano R., 1976, A&A, 49, 137Clerc N., Pierre M., Pacaud F., Sadibekova T., 2012, MN-RAS, 423, 3545Clerc N., Sadibekova T., Pierre M., Pacaud F., Le F`evreJ.-P., Adami C., Altieri B., Valtchanov I., 2012, MNRAS,423, 3561Cruddace, R., Voges, W., B¨ohringer, H., et al. 2002, ApJS,140, 239Eckert D., Molendi S., Paltani S., 2011, A&A, 526, A79Elyiv A., et al., 2012, A&A, 537, A131Ettori S., 2000, MNRAS, 311, 313Fabian A. C., 2012, ARA&A, 50, 455Fassbender R., et al., 2011, NJPh, 13, 125014Garilli B., et al., 2013, arXiv, arXiv:1310.1008Gilli R., Comastri A., Hasinger G., 2007, A&A, 463, 79Gobat R., et al., 2011, A&A, 526, A133Guzzo L., et al., 2013, arXiv, arXiv:1303.2623Haines C. P., et al., 2012, ApJ, 754, 97Hasinger G., Miyaji T., Schmidt M., 2005, A&A, 441, 417Henry J. P., Gioia I. M., Mullis C. R., Voges W., BrielU. G., B¨ohringer H., Huchra J. P., 2001, ApJ, 553, L109Henry J. P., Evrard A. E., Hoekstra H., Babul A., MahdaviA., 2009, ApJ, 691, 1307Hilton M., et al., 2012, MNRAS, 424, 2086Hinshaw G., et al., 2013, ApJS, 208, 19Kaiser N., 1986, MNRAS, 222, 323Kalberla P. M. W., Burton W. B., Hartmann D., Arnal c (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg E. M., Bajaja E., Morras R., P¨oppel W. G. L., 2005, A&A,440, 775Le F`evre O., et al., 2005, A&A, 439, 845Mantz A., Allen S. W., Ebeling H., Rapetti D., Drlica-Wagner A., 2010, MNRAS, 406, 1773Mantz A. B., et al., 2014, arXiv, arXiv:1401.2087Maughan B. J., Giles P. A., Randall S. W., Jones C., For-man W. R., 2012, MNRAS, 421, 1583Maughan B. J., 2014, MNRAS, 437, 1171McDonald M., et al., 2013, ApJ, 774, 23Merloni A., et al., 2012, arXiv, arXiv:1209.3114Mittal R., Hicks A., Reiprich T. H., Jaritz V., 2011, A&A,532, A133Moretti A., Campana S., Lazzati D., Tagliaferri G., 2003,ApJ, 588, 696Navarro J. F., Frenk C. S., White S. D. M., 1997, ApJ, 490,493Pacaud F., et al., 2006, MNRAS, 372, 578Pacaud F., et al., 2007, MNRAS, 382, 1289Pierre M., et al., 2004, JCAP, 9, 11Pierre M., Pacaud F., Juin J. B., Melin J. B., Valageas P.,Clerc N., Corasaniti P. S., 2011, MNRAS, 414, 1732Planck Collaboration, et al., 2013, arXiv, arXiv:1303.5076Planck Collaboration, et al., 2013, arXiv, arXiv:1303.5080Pratt G. W., Croston J. H., Arnaud M., B¨ohringer H.,2009, A&A, 498, 361Predehl P., et al., 2010, SPIE, 7732,Read A. M., Ponman T. J., 2003, A&A, 409, 395Reichert A., B¨ohringer H., Fassbender R., M¨uhlegger M.,2011, A&A, 535, A4Rozo E., et al., 2010, ApJ, 708, 645Schuecker P., B¨ohringer H., Collins C. A., Guzzo L., 2003,A&A, 398, 867Short C. J., Thomas P. A., Young O. E., Pearce F. R.,Jenkins A., Muanwong O., 2010, MNRAS, 408, 2213Starck J.-L., Murtagh F. D., Bijaoui A., 1998, Image Pro-cessing and Data Analysis: The Multiscale Approach,Cambridge University Press, Cambridge (GB).ˇSuhada R., et al., 2012, A&A, 537, A39Sun M., Voit G. M., Donahue M., Jones C., Forman W.,Vikhlinin A., 2009, ApJ, 693, 1142Sun M., 2012, NJPh, 14, 045004Takey A., Schwope A., Lamer G., 2013, A&A, 558, A75Tinker J., Kravtsov A. V., Klypin A., Abazajian K., War-ren M., Yepes G., Gottl¨ober S., Holz D. E., 2008, ApJ,688, 709Tozzi P., Santos J. S., Nonino M., Rosati P., Borgani S.,Sartoris B., Altieri B., Sanchez-Portal M., 2013, A&A,551, A45Truemper J., 1993, Sci, 260, 1769Ueda Y., et al., 2008, ApJS, 179, 124Ueda Y., Akiyama M., Hasinger G., Miyaji T., WatsonM. G., 2014, ApJ, 786, 104Valageas P., Clerc N., Pacaud F., Pierre M., 2011, A&A,536, A95Vikhlinin A., et al., 2009, ApJ, 692, 1060Voit G. M., 2005, RvMP, 77, 207Willis J. P., et al., 2013, MNRAS, 430, 134
Figure A1.
Comparison of temperature measurements the 29clusters in common between Pacaud et al. (2007) and this work(Table 1). Error bars delimitate 68% confidence intervals. Thedashed line shows the equality relation.
APPENDIX A: COMPARISON WITH PACAUDET AL. (2007) MEASUREMENTS
As described in Sect. 2, 29 of the 52 clusters presented in Ta-ble 1 pertain to the sample published by Pacaud et al. (2007,hereafter P07). They are issued from a similar datasets andwere analysed through similar methodologies. However, sev-eral changes have occurred between these two analyses: • the XAmin software used to extract sources, leading toslight changes in the masking of sources and definition ofthe optimal extraction radius R spec . • the event list processing and spectral extraction algo-rithms, in particular the XMM-SAS version. • the version of XSpec (from v.11.3.2 to v.12.8.0). • the APEC model, specifically the ATOMDB databasemodels (from v.1.3.1 to v.2.0.1). A1 Updated cluster redshift
XLSSC 35 has been updated to a new redshift value of z = 0 .
07 instead of z = 0 .
17 as quoted in P07. The presenceof a giant elliptical at z = 0 .
069 coincident with the X-raypeak argues in favor of this cluster redshift, although a su-perposition of two layers cannot be ruled out (see discussionin Adami et al. 2011).
A2 Temperature measurements
We first compare the X-ray spectral measurements, throughthe values of temperatures T X . Fig. A1 compares valueslisted in P07 to those published in this work. Taking intoaccount measurement uncertainties, we conclude to a good XMM Science Analysis Software http://atomdb.org/c (cid:13)000
We first compare the X-ray spectral measurements, throughthe values of temperatures T X . Fig. A1 compares valueslisted in P07 to those published in this work. Taking intoaccount measurement uncertainties, we conclude to a good XMM Science Analysis Software http://atomdb.org/c (cid:13)000 , 1–21 N. Clerc et al.
Table A1.
List of X-ray temperature measurement experimentsdesigned to address the discrepancies between previously pub-lished values (Pacaud et al. 2007, P07) and this work. Each linecorresponds to a series of measurements of the 29 clusters in com-mon between the two samples. Line (1) corresponds to valueslisted in Table 1 while line (2) to those listed in Table 1 of P07,as compared in Fig. A1.Extraction/ Event lists
Xspec
APECmask regions version version version(1) this work New 3.2 12.8.0 2.0.1(2) P07
Old 2.1 11.3.2 1.3.1 (3) Old 3.2 11.3.2 1.3.1(4) Old 2.1 12.8.0 1.3.1(5) Old 2.1 11.3.2 1.3.1(6) Old 3.2 12.8.0 2.0.1(7) Old 3.2 12.8.0 1.3.1 agreement between these two series. In order to address thediscrepancies, we defined a series of additional temperaturemeasurements for all clusters by applying changes as listedin Table A1. We summarize our findings as follows, the num-bers referring to this table:(i) (5) provided identical results as P07 (2), proving thatthere is no other source of bias than the one listed in thetable.(ii) (4)-(5) provided identical results, as did (3)-(7). Sinceonly the
XSpec version was changed in these comparisons,we cannot attribute the discrepancies to the spectral fittingroutines in
XSpec .(iii) (6)-(7) led to the identification of a bias in the tem-peratures of cool systems ( T (cid:46) σ differences above 1 keV to 2-3 σ differences below 1 keV).As a straightforward consequence, this bias is also found inthe (1)-(3) comparison.(iv) (1)-(6) provided almost identical results, except forthe presence of two outliers with large error bars. We at-tribute these differences to the changes in spectral extractionregions and the low signal-to-noise of the spectra involved.(v) (3)-(5) and (3)-(2) provided similar results, althoughwith some scatter around the one-to-one relation. This isattributed to the change in event lists creation and spectralextraction routines.In brief, we attribute the scatter around the equality linein Fig. A1 to changes in the event lists processing and ourdefintion of source extraction regions, while the small bias atlow temperatures is attributed to a recent change in APECmodels.Note that the updated redshift value for XLSSC 35leads to an updated value of the temperature that is consis-tent with the previously published value ( T X = 1 . ± . T X = 1 . ± . A3 Flux measurements
We compare on Fig. A2 our flux measurements with thoseobtained in P07. The methodology adopted in P07 is verysimilar to our ”method (i)” in Sect. 4.2. Namely, it re-lies on fitting a surface brightness profile by means of β - Figure A2.
Comparison of [0.5-2] keV flux measurements forthe 29 clusters in common between Pacaud et al. (2007) and thiswork (Table 1) in 0.5 Mpc apertures at the cluster redshift. Errorbars delimitate 68% confidence intervals. The dashed line showsthe equality relation. The clear outlier corresponds to XLSSC 35whose redshift has been updated to 0.07 (instead of 0.17 in P07). models. For this reason, we do not compare fluxes measuredwith method (ii) with those of P07. As this comparison isshown in a radius of 0.5 Mpc at the cluster redshift, we cor-rected our angular apertures from the change of cosmologicalmodel. Note that XLSSC 35 was quoted with a redshift of0.17 in P07, implying a smaller angular aperture than theone used for this work. Overall, both flux measurements arein excellent agreement.
APPENDIX B: DERIVATION OF MASSES ANDASSUMPTION ON R We described in Sect. 4.3.1 a method to estimate the massof a cluster. The first step consists in estimating R froma M − T X relation. However, this can be seen as an un-necessary step since the relation between R and M isstraightforward: M = π ρ c ( z ) R . Combining thisformula with Eq. 3 provides: R = R c (cid:115) . × βT πρ c ( z ) R c − R values do not depend on a scaling relation and canbe used in Eq. 3 to provide a mass estimate. Fig. B1 com-pares the mass estimates obtained by this method (”method1 bis”) and the method presented earlier (”method 1”).Apart from one outlier (XLSSC 080), the agreement is sat-isfactory, showing that our scaling used for inferring R isa reasonable one. c (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure B1.
Comparison of mass estimates for the 52 clusters inthis sample.
Method 1 assumes hydrostatic equilibrium and a gasdistribution following a β -model. Method 1bis is similar but doesnot assume a scaling law for the value of R . Clusters locatedright of the dashed line have their surface brightness profile well-described by a 2-parameter β -model, while clusters on the leftwere imposed β = 2 / APPENDIX C: NOTES ON INDIVIDUALCLUSTERS
Fig. C1 shows X-ray/optical overlays for 23 C1 clusters inthis sample (the other 29 are shown in Pacaud et al. 2007,their Appendix B). • XLSSC 060:
Abell cluster A0329. Our pipelinewrongly deblended multiple components, due to the excep-tional extent and brightness of this source and the presenceof gaps on XMM detectors. They were manually mergedtogether for the purpose of measuring its X-ray proper-ties. Cruddace et al. (2002) measured an unabsorbed, [0.1-2.4] keV flux of 2 . ± . × − ergs/s/cm from ROSATdata. This translates into a [0.5-2] keV, galactic-absorbed,flux of 1 . ± . × − ergs/s/cm , hence entirely consistentwith our XMM value. • XLSSC 079: this cluster is detected on the deepestpointing of the survey (80 ks on-axis) at an off-axis radiusof 8 (cid:48) . The X-ray analysis is therefore severely limited by con-fusion: its extended emission is contaminated by a number ofpoint-sources. Despite our efforts to correctly mask all pointsources, we recommend to consider its temperature and lu-minosity measurements with caution. This is reinforced bythe fact that this cluster appears as a clear outlier in the L X − T diagram (lower-right point in first panel of Fig. 7). • XLSSC 053: this cluster is probably a group of clustersat z ∼ .
5, as hinted by the projected distribution of galaxiesand the faint, large, X-ray surface brightness. An assessmentof its multiple-component nature will be possible thanks tothe analysis of spectroscopic redshifts in the vicinity of thisobject. • XLSSU J021744.1-034536: this cluster was first pre-sented and discussed in Willis et al. (2013). We note thatrecent data obtained with the CARMA interferometer con-firmed the presence of hot gas at the location of this ob-ject via Sunyaev-Zeldovich effect (Mantz et al. 2014). Todate, its spectroscopic confirmation is awaited. A 3 × (cid:48)(cid:48) southwards. Both their spectra in-dicate these objects are probably stars. The red cD galaxy(visible on the image of Willis et al. 2013) partly falls ontothe slit but is too faint to provide a spectrum. A very faintobject located within the X-ray contours at R . A . = 34 . δ = − .
753 (J2000, 2 (cid:48)(cid:48) positional uncertainty) shows oneemission line in all three exposures at λ = 5437 . α emitted by an active emitter at z = 3 .
47. With this data in hand, we cannot exclude thatthe two stars and the active object also emit X-rays contam-inating the galaxy cluster emission.This paper has been typeset from a TEX/ L A TEX file preparedby the author. c (cid:13)000
47. With this data in hand, we cannot exclude thatthe two stars and the active object also emit X-rays contam-inating the galaxy cluster emission.This paper has been typeset from a TEX/ L A TEX file preparedby the author. c (cid:13)000 , 1–21 N. Clerc et al.
Figure C1.
Images of the C1 clusters not presented in Appendix B of Pacaud et al. (2007) and sorted by ascending XLSSC number. Left:X-ray/I-band overlay (7 arcmin on a side). Squares indicate point sources (likelihood >
15, Chiappetti et al. 2013) and red crosses areother detections (likelihood < Top:
XLSSC 009 ( z = 0 . Bottom:
XLSSC 012( z = 0 . (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure C1 – continued Images of the C1 clusters.
Top:
XLSSC 053 ( z = 0 . Bottom:
XLSSC 054 ( z = 0 . (cid:13)000
XLSSC 054 ( z = 0 . (cid:13)000 , 1–21 N. Clerc et al.
Figure C1 – continued Images of the C1 clusters.
Top:
XLSSC 055 ( z = 0 . Bottom:
XLSSC 056 ( z = 0 . (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure C1 – continued Images of the C1 clusters.
Top:
XLSSC 057 ( z = 0 . Bottom:
XLSSC 058 ( z = 0 . (cid:13)000
XLSSC 058 ( z = 0 . (cid:13)000 , 1–21 N. Clerc et al.
Figure C1 – continued Images of the C1 clusters.
Top:
XLSSC 059 ( z = 0 . Bottom:
XLSSC 060 ( z = 0 . (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure C1 – continued Images of the C1 clusters.
Top:
XLSSC 061 ( z = 0 . Bottom:
XLSSC 062 ( z = 0 .
06, located outside of theCFHT-LS W1 footprint).c (cid:13)000
06, located outside of theCFHT-LS W1 footprint).c (cid:13)000 , 1–21 N. Clerc et al.
Figure C1 – continued Images of the C1 clusters.
Top:
XLSSC 064 ( z = 0 . Bottom:
XLSSC 065 ( z = 0 . (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure C1 – continued Images of the C1 clusters.
Top:
XLSSC 072 ( z = 1 . Bottom:
XLSSC 074 ( z = 0 . (cid:13)000
XLSSC 074 ( z = 0 . (cid:13)000 , 1–21 N. Clerc et al.
Figure C1 – continued Images of the C1 clusters.
Top:
XLSSC 075 ( z = 0 . Bottom:
XLSSC 076 ( z = 0 . (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure C1 – continued Images of the C1 clusters.
Top:
XLSSC 077 ( z = 0 . Bottom:
XLSSC 078 ( z = 0 . (cid:13)000
XLSSC 078 ( z = 0 . (cid:13)000 , 1–21 N. Clerc et al.
Figure C1 – continued Images of the C1 clusters.
Top:
XLSSC 079 ( z = 0 . Bottom:
XLSSC 080 ( z = 0 . (cid:13) , 1–21 he XMM-LSS Class 1 cluster sample over the extended 11 deg Figure C1 – continued Images of the C1 clusters. XLSSU J021744.1-034536 (see Willis et al. 2013 for a near-infrared view of thiscluster).c (cid:13)000