Quantifying the suppression of the (un)-obscured star formation in galaxy cluster cores at 0.2 ≲ z ≲ 0.9
L. Rodríguez-Muñoz, G. Rodighiero, C. Mancini, P. G. Pérez-González, T. D. Rawle, E. Egami, A. Mercurio, P. Rosati, A. Puglisi, A. Franceschini, I. Balestra, I. Baronchelli, A. Biviano, H. Ebeling, A. C. Edge, A. F. M. Enia, C. Grillo, C. P. Haines, E. Iani, T. Jones, M. Nonino, I. Valtchanov, B. Vulcani, M. Zemcov
MMNRAS , 1–32 (2018) Preprint 24 December 2018 Compiled using MNRAS L A TEX style file v3.0
Quantifying the suppression of the (un)-obscured starformation in galaxy cluster cores at 0.2 (cid:46) z (cid:46) L. Rodr´ıguez-Mu˜noz (cid:63) , G. Rodighiero , C. Mancini , P. G. P´erez-Gonz´alez , ,T. D. Rawle , E. Egami , A. Mercurio , P. Rosati , A. Puglisi , ,A. Franceschini , I. Balestra , I. Baronchelli , , A. Biviano , H. Ebeling ,A. C. Edge , A. F. M. Enia , C. Grillo , , C. P. Haines , E. Iani ,T. Jones , , M. Nonino , I. Valtchanov , B. Vulcani , M. Zemcov Dipartimento di Fisica e Astronomia “G. Galilei”, Universit`a degli Studi di Padova, Vicolo dell’Osservatorio 3, I-35122, Italy Departamento de Astronom´ıa y Astrof´ısica, Universidad Complutense de Madrid, Av. Complutense s/n, C.P. 28040, Madrid, Spain Centro de Astrobiolog´ıa, Instituto Nacional de T´ecnica Aeroespacial, Carretera de Ajalvir km 4, Torrej´on de Ardoz, Madrid, E-28850, Spain ESA/Space Telescope Science Institute (STScI), 3700 San Martin Drive, Baltimore, MD 21218, USA Steward Observatory, University of Arizona, 933 N. Cherry Ave, Tucson, AZ 85721, USA INAF-Osservatorio Astronomico di Capodimonte, via Moiariello 16, 80131 Napoli, Italy Dipartimento de Fisica e Scienze della Terra, Universit`a degli Studi di Ferrara, via Saragat 1, 44122 Ferrara, Italy Laboratoire AIM-Paris-Saclay, CEA/DSM-CNRS-Universit´e Paris Diderot, IRFU/Service d’Astrophysique, CEA Saclay,Orme des Merisiers, F-91191 Gif-sur-Yvette, France University Observatory Munich, Scheinerstrasse 1, D-81679 Munich, Germany IPAC, Mail Code 314-6, Caltech, 1200 E. California Blvd., Pasadena, CA 91125, USA INAF-Osservatorio Astronomico di Trieste, via G. B. Tiepolo 11, I-34131, Trieste, Italy Institute for Astronomy, University of Hawaii, Honolulu, HI 96822, USA Centre for Extragalactic Astronomy, Department of Physics, Durham University, South Road, Durham DH1 3LE, UK Dark Cosmology Centre, Niels Bohr Institute, University of Copenhagen, Juliane Maries Vej 30, DK-2100 Copenhagen, Denmark Dipartimento di Fisica, Universita degli Studi di Milano, via Celoria 16, I-20133 Milano, Italy INAF - Osservatorio Astronomico di Brera, via Brera 28, I-20121 Milano, Italy Department of Physics and Astronomy, PAB, 430 Portola Plaza, Box 951547, Los Angeles, CA 90095-1547, USA Department of Physics, University of California Davis, 1 Shields Avenue, Davis, CA 95616, USA Herschel Science Centre, European Space Astronomy Centre, ESA, E-28691 Villanueva de la Ca˜nada, Spain Center for Detectors, School of Physics and Astronomy, Rochester Institute of Technology, Rochester NY 14623, USA
Accepted XXX. Received YYY; in original form ZZZ
ABSTRACT
We quantify the star formation (SF) in the inner cores ( R / R ≤ (cid:46) z (cid:46) Herschel
Lensing Survey and theCluster Lensing and Supernova survey with
Hubble . These programmes, covering therest-frame ultraviolet to far-infrared regimes, allow us to accurately characterize stellarmass-limited ( M ∗ > M (cid:12) ) samples of star-forming cluster members (not)-detectedin the mid- and/or far-infrared. We release the catalogues with the photometry, pho-tometric redshifts, and physical properties of these samples. We also quantify the SFdisplayed by comparable field samples from the Cosmic Assembly Near-infrared DeepExtragalactic Legacy Survey. We find that in intermediate- z cluster cores, the SF activ-ity is suppressed with respect the field in terms of both the fraction ( F ) of star-forminggalaxies (SFG) and the rate at which they form stars ( SFR and s SFR = SFR / M ∗ ).On average, the F of SFGs is a factor ∼ SFR and s SFR typically ∼ z ∼ Key words: galaxies: clusters: general – galaxies: evolution – galaxies: star formation– catalogues (cid:63)
E-mail: [email protected] (cid:13) a r X i v : . [ a s t r o - ph . GA ] D ec L. Rodr´ıguez-Mu˜noz et al.
Galaxies appear to be distributed into two fairly distinctgeneral groups (e.g., Kauffmann et al. 2003, Bell et al. 2004,Baldry et al. 2004, Haines et al. 2017): a population of rel-atively red, quiescent galaxies (i.e., where the star forma-tion activity has already been quenched), which are char-acterized by spheroid-dominated morphologies; and a pop-ulation of rather blue, star-forming galaxies (SFGs), withdisk-dominated morphologies. Understanding the nature ofthe processes that make a galaxy a member of either cate-gory at any cosmological epoch is one of the longest standingunsolved problems in astrophysics.The fraction of red/quiescent/early-type galaxiesamong the whole population scales with the stellar mass( M ∗ ) of the galaxies up to z ∼ z ∼ dichotomy be-tween (still) star-forming and quenched galaxies, should bedriven (independently; Peng et al. 2010) by the impact onthe evolution of galaxies of two kind of processes: thosesomehow related to the stellar mass of the galaxies theyquench, and therefore, responsible for the so-called massquenching ; and those linked to physical processes takingplace in high density environments, responsible for the so-called environmental quenching . The physical nature of thesequenching processes and its evolution with redshift remainscontroversial.A plethora of works have studied the star formation(SF) activity within galaxy clusters at different redshifts asto quantify the environmental influence on galaxy evolution(e.g., Dressler et al. 1997, Poggianti et al. 1999; Poggianti2003, De Lucia et al. 2007, Saintonge et al. 2008, Finn et al.2010, Vulcani et al. 2011). This large body of work gives ev-idence for a significant transformation of galaxy populationsin clusters since z ∼
1. Already three decades ago, Butcher &Oemler (1984, see also Butcher & Oemler 1978) found thatthe fraction of blue cluster members increases from zero inthe local universe to ∼
20% by z ∼ z ∼ post-starburst (PSB;e.g., Poggianti et al. 2009, Muzzin et al. 2014, Paccagnellaet al. 2017), and jellyfish galaxies (e.g., Smith et al. 2010,Poggianti et al. 2017). Also, first CO observations in z ∼ s SFR ; defined as the ratio between the
SFR and the M ∗ of a galaxy) in the inner regions of clus- ters (i.e., within the virial radius, R virial ) with respect tothe field, with values typically ∼ SFR and M ∗ found for the star-forming fieldgalaxies up to z ∼ main sequence (MS)of SFGs. The existence of the MS is interpreted as the prooffor a typical mode in which the galaxies form stars (e.g., Ren-zini & Peng 2015). The tightness of the correlation (0.3 dexscatter; e.g., Whitaker et al. 2012b) is interpreted as a pos-sible consequence of the short time-scale of the dominantquenching process (Peng et al. 2010) moving the field SFGsout of the MS. As a consequence, the displacement of thecluster members MS towards lower SFR values could implythat the dominant quenching mechanisms in rich environ-ments are different (e.g., slow quenching mechanisms couldpopulate the region below the MS with transition galaxies ontheir way to be turned off; Haines et al. 2015, Haines et al.2013, Paccagnella et al. 2016). However, other works suchas Peng et al. (2010), Finn et al. (2010), Wijesinghe et al.(2012), or Tyler et al. (2013) find the same
SFR distribu-tion in clusters as in the field at intermediate redshifts. Thesediscrepancies appear to be due to a combination of differentfactors such as observational biases (e.g.,
SFR detectionlimit), different sample selection functions, and cluster-to-cluster differences (e.g., Geach et al. 2006, Alberts et al.2016).A variety of mechanisms have been proposed as the re-sponsible for environmental quenching (see reviews by, e.g.,Boselli & Gavazzi 2006 and Haines et al. 2007): gravitationalinteractions with the potential well of nearby galaxies or thecluster itself, also known as harassment (Moore et al. 1996);removal and thermal heating of the interstellar medium ofthe galaxies by the interaction with the intra-cluster medium(ICM), the so-called ram-pressure stripping (RPS; Gunn &Gott 1972, Poggianti et al. 2017); the removal of the hotgas reservoirs of the halo of galaxies, or strangulation , andsubsequent halt of the supply of material needed to sustainthe SF, leading up to the eventual starvation (Larson et al.1980). These mechanisms shape the evolution of galaxies indifferent time-scales, probably with different efficiency de-pending on the properties of both galaxies and clusters, andthe particular circumstances under which the infall takesplace (see, e.g., Boselli & Gavazzi 2006, Berrier et al. 2009).Furthermore, it has also been proposed that the environmen-tal impact on these SFGs starts in early stages of the infallif the accreted galaxies are bound up in small groups ( pre-processing ; e.g., Haines et al. 2015). Distinguishing amongthese mechanisms remains challenging, and relies on the de-tailed study and accurate quantification of the changes suf-fered by the SF processes and structural properties of thegalaxies in rich environments.Recently, a number of state-of-the-art surveys have tar-geted massive galaxy clusters at intermediate redshift withthe main goal of exploring low-luminosity galaxies at highredshift taking advantage of the gravitational lensing phe-nomenon (e.g.,
Hubble
Frontier Fields, Lotz et al. 2017). Inthis work, we aim at shedding light on the impact of envi-ronment on the star-forming activity in galaxies populatingclusters by using these surveys to study the cluster inhabi-tants themselves.
MNRAS , 1–32 (2018)
Un)-obscured star formation in cluster cores We focus our analysis on 24 X-ray selected (i.e., withtotal masses ∼ ∼ × M (cid:12) ) clusters targeted by the Herschel
Lensing Survey (HLS; Egami et al. 2010), a far-infrared (FIR) and sub-millimetre survey using the ESA
Herschel
Space Observatory, and the Cluster Lensing andSupernova survey with Hubble (CLASH; Postman et al.2012), a deep optical and near-infrared (NIR)
Hubble
SpaceTelescope program, as well as by other NIR and mid-infrared (MIR)
Spitzer programs. The sample extends be-tween 0.187 ≤ z ≤ M ∗ -selected samples of cluster SFGs. The use of Herschel observations complementing optical and NIR data guaran-tees a proper quantification of the SF shrouded by dust.Indeed, SFGs detected in the MIR and/or FIR (M-FIR)often have optical colours consistent with those of passivelyevolving galaxies and therefore, they are easily missed bystudies limited to the optical or NIR regimes. Not quantify-ing the contribution of these obscured processes can lead toan under estimation of the true level of SF by a factor ∼ the same analysis to the optical-to-FIR publicly availablephotometry on three of the fields targeted by the CosmicAssembly Near-infrared Deep Extragalactic Legacy Survey(CANDELS; Grogin et al. 2011, Koekemoer et al. 2011).This article is organized as follows: Section 2 describesthe cluster sample and corresponding data. Section 3 de-scribes our approach to combining the different photometricdata and building the multi-wavelength catalogue we use toderive photometric redshifts (Section 4) and physical proper-ties of galaxies through a SED-fitting approach (Section 5).In Section 6, we detail our procedure to select cluster mem-bers using spectroscopic and photometric redshifts estima-tions. The final cluster members samples of SFGs are pre-sented in Section 7 and further characterized in Section 8.The quantification of the SF activity in the core of theseclusters is discussed in Section 9. Finally, an interpretationof our results is given in Section 10, and a summary and themain conclusions of this work are given in Section 11.Throughout this work we assume a flat ΛCDM cosmol-ogy with H =70 kms − Mpc − , Ω m =0.3, and Ω Λ =0.7. Star-formation rates and stellar masses are based on a Salpeter(1955) initial mass function (IMF). The catalogues of star-forming cluster members associ-ated to this paper, including multi-wavelength photometry,photometric redshifts, and physical properties, can be down-loaded from the public flavour of the Rainbow
CosmologicalDatabase (P´erez-Gonz´alez et al. 2008, Barro et al. 2011a,b). The
Herschel Lensing Survey (HLS; Egami et al. 2010) isa large imaging survey of galaxy clusters in the far-infrared(FIR) and sub-millimetre using the ESA
Herschel Space Ob-servatory (Pilbratt et al. 2010). HLS provides deep PACS(Poglitsch et al. 2010) and SPIRE (Griffin et al. 2010) imag-ing (see Section 2.3) for a sample of 65 X-ray-luminous (i.e.,massive) clusters of galaxies in the redshift range between0.2 (cid:46) z (cid:46) Cluster Lensing and Supernova survey withHubble (CLASH; Postman et al. 2012). CLASH is a Multi-Cycle Treasury Program with the aim of providing ultra-deep photometry of 25 X-ray selected, massive ( ∼ ∼ × M (cid:12) ) galaxy clusters in a total of 16 passbandsusing HST ACS/WFC, WFC3/UVIS, and WFC3/IR (seeSection 2.1 for details). CLASH clusters are drawn heavilyfrom the Abell and MACS cluster catalogues (Abell 1958,Abell et al. 1989, Ebeling et al. 2001, Ebeling et al. 2007,Ebeling et al. 2010, Mann & Ebeling 2012).The wealth of photometric and spectroscopic dataavailable for this galaxy clusters sample, that we callCLASH+HLS, enables the accurate identification and char-acterization of their galaxy population (e.g., Annunziatellaet al. 2016, Maier et al. 2016, Balestra et al. 2016). Indeed,CLASH+HLS clusters have been extensively studied in pre-vious works. CLASH photometry together with spectroscopyfrom different surveys (see Section 2.4) have provided strongconstraints on the cluster inner mass distributions and pro-files (e.g., Zitrin et al. 2015, Biviano et al. 2013, Annunzi-atella et al. 2014). Also, their dynamical state and substruc-tures have been analyzed through different techniques, suchas the Sunyaev-Zel’dovich effect (SZ; Sunyaev & Zel’dovich1972, Rumsey et al. 2016) and X-ray surface brightness anal-ysis (see Rumsey et al. 2016 and references therein), as wellas lensing (e.g., Zitrin et al. 2013, Grillo et al. 2015) and http://rainbowx.fis.ucm.esMNRAS000
Herschel Space Ob-servatory (Pilbratt et al. 2010). HLS provides deep PACS(Poglitsch et al. 2010) and SPIRE (Griffin et al. 2010) imag-ing (see Section 2.3) for a sample of 65 X-ray-luminous (i.e.,massive) clusters of galaxies in the redshift range between0.2 (cid:46) z (cid:46) Cluster Lensing and Supernova survey withHubble (CLASH; Postman et al. 2012). CLASH is a Multi-Cycle Treasury Program with the aim of providing ultra-deep photometry of 25 X-ray selected, massive ( ∼ ∼ × M (cid:12) ) galaxy clusters in a total of 16 passbandsusing HST ACS/WFC, WFC3/UVIS, and WFC3/IR (seeSection 2.1 for details). CLASH clusters are drawn heavilyfrom the Abell and MACS cluster catalogues (Abell 1958,Abell et al. 1989, Ebeling et al. 2001, Ebeling et al. 2007,Ebeling et al. 2010, Mann & Ebeling 2012).The wealth of photometric and spectroscopic dataavailable for this galaxy clusters sample, that we callCLASH+HLS, enables the accurate identification and char-acterization of their galaxy population (e.g., Annunziatellaet al. 2016, Maier et al. 2016, Balestra et al. 2016). Indeed,CLASH+HLS clusters have been extensively studied in pre-vious works. CLASH photometry together with spectroscopyfrom different surveys (see Section 2.4) have provided strongconstraints on the cluster inner mass distributions and pro-files (e.g., Zitrin et al. 2015, Biviano et al. 2013, Annunzi-atella et al. 2014). Also, their dynamical state and substruc-tures have been analyzed through different techniques, suchas the Sunyaev-Zel’dovich effect (SZ; Sunyaev & Zel’dovich1972, Rumsey et al. 2016) and X-ray surface brightness anal-ysis (see Rumsey et al. 2016 and references therein), as wellas lensing (e.g., Zitrin et al. 2013, Grillo et al. 2015) and http://rainbowx.fis.ucm.esMNRAS000 , 1–32 (2018) L. Rodr´ıguez-Mu˜noz et al.
Table 1.
Description of the galaxy cluster sample. We display the following information: (1) Cluster ID; (2-3) Coordinates of the clustercentre as in Postman et al. (2012); (4) redshift Postman et al. 2012; (5) Velocity dispersion (we use the value σ cl =1600 km s − whenno observational estimation was found in the literature); (6) Radius within which the mean density is 200 times the critical density atthe redshift where the cluster is located ( ∼ R virial according to the simulations of Evrard et al. 1996; we use R =2000 kpc, see forinstance Umetsu et al. (2014), for those cases for which no precise value was found in the literature); (7) The SF activity of the BCG asquantified through the emission of the UV, corrected for extinction ( SFR
BCG , UV , corr . ; Donahue et al. 2015), and the emission in the FIR( SFR
BCG , TIR ; Rawle et al. 2012a); (8) cool-core tracer C parameter as published by Donahue et al. (2016); (9) number of spectroscopicredshifts within the area covered by the CLASH catalogue ( ∼ ). Note: + a Gelleret al. (2014); b Mercurio et al. (2003); c G´omez et al. (2012); d Balestra et al. (2016); e Biviano et al. (2013); f Ebeling et al. (2007); g Annunziatella et al. (2016); h Newman et al. (2013); i Rosati et al. (2014); j Coe et al. (2012); k Karman et al. (2015); l Huchra et al.(2012); m Ebeling et al. (2014); n Treu et al. (2015) and Schmidt et al. (2014); o Ravindranath & Ho (2002); p Cohen & Kneib (2002); q Shectman et al. (1996); r Abazajian et al. (2009); s σ cl and R derived using the value of the mass within R ( M ) from Umetsuet al. (2014).ID RA Dec z σ cl R SFR
BCG , UVcorr . / TIR C z spec [J2000] [J2000] [km s − ] [kpc] [ M (cid:12) yr − ](1) (2) (3) (4) (5) (6) (7) (8) (9)A0383 02:48:03.40 -03:31:44.9 0.187 931 +5959 a +10 − a ± ± a,h A0209 01:31:52.54 -13:36:40.4 0.206 + +88 − b +50 − g ± b,i,g A2261 17:22:27.18 32:07:57.3 0.224 1524 s s ± j RBS1748 21:29:39.94 00:05:18.8 0.234 1600 2000 2.9 ± s +97 − h ± h MS2137 21:40:15.18 -23:39:40.7 0.313 1257 s +140 − h ± +230 − c s ± i,k MACS1931 19:31:49.66 -26:34:34.0 0.352 1339 s s ± s s ± s s ± l MACS1720 17:20:16.95 35:36:23.6 0.387 1296 s s ± +12 − d +110 − d ± d,n,m MACS0429 04:29:36.05 -02:53:06.1 0.399 1140 s s ± +53 − e +100 − e ± e MACS0329 03:29:41.56 -02:11:46.1 0.450 1165 s s ± s s ± o,p,q MACS1311 13:11:01.67 -03:10:39.5 0.494 1600 2000 5.8 ± +120 − f s ± m MACS0717 07:17:32.63 37:44:59.7 0.545 1660 +120 − f s ± l,m MACS1423 14:23:47.76 24:04:40.5 0.545 1300 +120 − f ± ± m MACS2129 21:29:26.06 -07:41:28.8 0.570 1400 +120 − f ± m MACS0647 06:47:50.27 70:14:55.0 0.584 900 +170 − f s ± +130 − f s ± ± l,r kinematics of galaxy populations (e.g., Girardi et al. 2015).Despite the X-ray selection, that generally favours highly re-laxed clusters, the sample is found to be not homogeneouslydynamically relaxed (Postman et al. 2012, Rumsey et al.2016). Finally, a number of works have studied in detailthe brightest cluster galaxies (BCG) of the CLASH+HLSsystems. For instance, Donahue et al. (2015) and Donahueet al. (2016) carried out a study on the morphology and SFactivity of these peculiar galaxies, using the rest-frame UVimaging provided by CLASH. Furthermore, they also char-acterized the intra cluster gas in the vicinity of the BCGsand beyond, by analysing the X-ray emission of the innercluster cores. Complementary, Rawle et al. (2012a) studiedthe obscured SF activity undergone by the BCGs of the mas-sive clusters observed by HLS, and its dependence with theX-ray gas cooling times for cool-core (CC) clusters .In the following subsections, we describe the photomet-ric and spectroscopic datasets available on the cluster fields Cool-core clusters are defined as those systems with X-ray cool-ing times < (see Table 2 & 3 for a summary of their main characteris-tics), as well as other ancillary data found in the literature. Hubble optical and near-infrared photometry
In this work, we use the CLASH photometric dataset pub-lished by Postman et al. (2012). This data release containsthe photometry performed on the HST
ACS/WFC (F435W,F475W, F606W, F625W, F775W, F814W, and F850LP),WFC3/UVIS (F225W, F275W, F336W, and F390W), andWFC3/IR (F105W, F110W, F125W, F140W, and F160W)deep imaging of 25 massive intermediate redshift clusters.Object detection and photometry is accomplished usingSExtractor (Bertin & Arnouts 1996) in dual image modeusing a weighted sum of the ACS/WFC and WFC3/IRimages (see Postman et al. 2012 for details on the
HST data reduction, catalogue build up, and main characteris-tics). These catalogues cover an area of ∼ , limitedby the WFC3/IR images ( ∼ × ), and therefore, https://archive.stsci.edu/prepds/clash/MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores Table 2.
In this table we show an overview of the photomet-ric bands used in this work: (1) name of the observing bandand instrument; (2) effective wavelength of the filter; (3) medianFWHM of the PSF in arcseconds; (4) name of the project to whichthe data belongs. ( ∗ ) Spitzer Programs + ) Spitzer Programs λ eff FWHM Project(1) (2) (3) (4)WFC3-F225W 237.84 nm 0 (cid:48)(cid:48) .08 CLASHWFC3-F275W 271.47 nm 0 (cid:48)(cid:48) .08 CLASHWFC3-F336W 335.86 nm 0 (cid:48)(cid:48) .07 CLASHWFC3-F390W 393.22 nm 0 (cid:48)(cid:48) .07 CLASHACS-F435W 436.33 nm 0 (cid:48)(cid:48) .08 CLASHACS-F475W 475.05 nm 0 (cid:48)(cid:48) .08 CLASHACS-F606W 596.11 nm 0 (cid:48)(cid:48) .08 CLASHACS-F625W 630.97 nm 0 (cid:48)(cid:48) .08 CLASHACS-F775W 770.59 nm 0 (cid:48)(cid:48) .08 CLASHACS-F814W 807.31 nm 0 (cid:48)(cid:48) .09 CLASHACS-F850LP 905.26 nm 0 (cid:48)(cid:48) .09 CLASHWFC3-F105W 1.06 µ m 0 (cid:48)(cid:48) .13 CLASHWFC3-F110W 1.15 µ m 0 (cid:48)(cid:48) .13 CLASHWFC3-F125W 1.25 µ m 0 (cid:48)(cid:48) .14 CLASHWFC3-F140W 1.40 µ m 0 (cid:48)(cid:48) .14 CLASHWFC3-F160W 1.54 µ m 0 (cid:48)(cid:48) .15 CLASHIRAC-3.6 µ m 3.56 µ m 2 (cid:48)(cid:48) .1 ∗ IRAC-4.5 µ m 4.50 µ m 2 (cid:48)(cid:48) .1 ∗ IRAC-5.8 µ m 5.74 µ m 2 (cid:48)(cid:48) .2 ∗ IRAC-8.0 µ m 7.93 µ m 2 (cid:48)(cid:48) .2 ∗ MIPS-24 µ m 23.84 µ m 5 (cid:48)(cid:48) + PACS-100 µ m 102.25 µ m 8 (cid:48)(cid:48) HLSPACS-160 µ m 165.59 µ m 12 (cid:48)(cid:48) HLSSPIRE-250 µ m 253.13 µ m 18 (cid:48)(cid:48) HLSSPIRE-350 µ m 355.87 µ m 25 (cid:48)(cid:48) HLSSPIRE-500 µ m 511.19 µ m 36 (cid:48)(cid:48) HLS they mainly sample the very inner cluster cores. An angulardistance of 2.0 arcmin corresponds to 375 kpc and 932 kpcfor the lowest and largest redshifts in the sample, respec-tively. The total area covered, including the 24 clusters, is ∼
135 arcmin . The exposure times of the frames vary be-tween 2000 and 5000 s, reaching average (5 σ ) limiting ABmagnitudes of ∼
26. A summary of the properties of thedataset is shown in Table 2.
Spitzer near and mid-infrared photometry
A series of programs with
Spitzer have covered all CLASHclusters with IRAC 3.6 and 4.5 µ m bands. Furthermore,40% of them have also been observed with IRAC 5.8 and8.0 µ m channels, and 50% has been covered by MIPS 24 µ mband. These data were extracted from the Spitzer
Heritagearchive . Spitzer images reduction, source detection, andphotometry were carried out as described in P´erez-Gonz´alezet al. (2005) and P´erez-Gonz´alez et al. (2008), for MIPSand IRAC, respectively. Briefly, the data reduction was car-ried out with MOPEX (Mosaicking and Point-source Ex-traction), the package provided by the
Spitzer
Science Cen-ter for reducing and analysing imaging data. In the case http://irsa.ipac.caltech.edu/applications/Spitzer/SHA of IRAC, the source detection and photometry were car-ried out with SExtractor (Bertin & Arnouts 1996), usingthe same procedure as Huang et al. (2004). Photometry wasperformed using a small circular aperture, and an aperturecorrection was applied to get the total flux. IRAC beamsizes are 2.1, 2.1, 2.2, and 2.2 (cid:48)(cid:48) respectively for increasingwavelengths. The average sensitivities reached at 5 σ are 1.4,1.5, 4.5, 4.2 µ Jy. In the case of MIPS images, characterizedby a larger point-spread function, the photometry was ex-tracted by PSF fitting. Several detection passes are used inorder to make catalogues as complete as possible, in spiteof the significant source confusion. The MIPS 24 µ m beamsize is 5 (cid:48)(cid:48) . The average MIPS 24 µ m limiting flux at 5 σ is234 µ Jy. In Table 2 and 3 we summarize the properties ofthese photometric catalogues. We report the heterogeneoussensitivities reached by IRAC and MIPS imaging on the dif-ferent CLASH clusters. In particular, MIPS 24 µ m limitingfluxes vary between 77 and 852 µ Jy.
Herschel far-infrared photometry
This study employs the PACS 100, 160 µ m, and SPIRE 250,350, 500 µ m imaging provided by HLS for all the clusters.We use the catalogues created by the HLS team followingthe methodology presented by P´erez-Gonz´alez et al. (2010)and Rawle et al. (2010, 2016). Source catalogues and pho-tometry in all bands were obtained with standard PSF fit-ting methodology, relying on a set of fixed IRAC and MIPSprior position catalogues. PACS imaging at 100 and 160 µ mhas mean 5 σ flux limits of 4.7 and 8.7 mJy, while in thethree SPIRE bands, the typical 5 σ limits are 19.4, 15.3, and13.7 mJy, respectively for the 250, 350, and 500 µ m bands.The beam sizes for the five Herschel bands (sorted by in-creasing effective wavelength) are 8, 12, 18, 25, and 36 (cid:48)(cid:48) ,respectively.
One of the programs with a greater contribution to our spec-troscopic redshift sample is the spectroscopic survey car-ried out on the 13 southern CLASH clusters with the Vis-ible Multi-Object Spectrograph (VIMOS; Le F`evre et al.2003) mounted on the Very Large Telescope (VLT), the so-called CLASH-VLT survey (CLASH-VLT Large Programme186.A0.798; P.I.: P. Rosati; Rosati et al. 2014). We refer thereader to Biviano et al. (2013) and Balestra et al. (2016) fordetails on spectroscopic data, target selection, and perfor-mance statistics of the mentioned project. We also make useof spectroscopic redshift measurements from the Grism LensAmplified Survey from Space (GLASS; Schmidt et al. 2014;Treu et al. 2015), a large
Hubble Space Telescope programaimed at obtaining grism spectroscopy of the HFF. Besidesthese, we also gather spectroscopic redshifts from other sur-veys (see Table 1 for a complete list of the works included).Finally, we also retrieve redshifts through NASA/IPAD Ex-tragalactic Database (NED), mainly from the 2MASS Red-shift Survey (Huchra et al. 2012), and the Seventh DataRelease of the Sloan Digital Sky Survey (Abazajian et al.2009). In Section 4 we describe the properties of the finalspectroscopic sample.
MNRAS000
MNRAS000 , 1–32 (2018)
L. Rodr´ıguez-Mu˜noz et al.
Table 3.
Limiting fluxes (5 σ ) of the Spitzer and
Herschel photometric catalogues used in this work. F lim [ µ Jy] F lim [mJy] Spitzer /IRAC
Spitzer /MIPS
Herschel /PACS
Herschel /SPIRECluster 3.6 µ m 4.5 µ m 5.8 µ m 8.0 µ m 24 µ m 100 µ m 160 µ m 250 µ m 350 µ m 500 µ m(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)A0209 2.0 1.7 5.2 5.4 268.7 4.6 9.1 14.6 14.0 10.7A0383 2.7 2.2 6.7 6.3 317.6 4.8 9.4 14.8 13.7 10.8MACS0329 1.3 1.3 – – – 4.5 8.5 19.9 15.8 14.7MACS0416 1.2 1.2 – – – 4.7 8.5 19.2 14.9 13.9MACS0429 1.3 1.3 – – – 4.5 8.3 21.0 16.7 14.4MACS0647 1.1 1.3 – – – 4.8 10.8 23.1 20.3 14.7MACS0717 1.7 1.9 – – 133.3 4.7 9.2 17.8 15.9 12.0MACS0744 2.2 3.0 1.2 1.8 – 4.4 8.4 14.5 14.1 11.3A0611 1.0 1.0 – – 380.6 4.8 8.4 15.0 13.9 11.1MACS1115 1.3 1.4 – – – 4.7 8.7 20.4 16.1 14.5MACS1149 0.9 0.9 – – – 4.7 8.6 15.1 15.3 14.9MACS1206 1.1 1.1 – – 305.7 4.5 10.3 25.8 21.9 18.3CLJ1226 3.6 3.6 2.0 1.8 131.7 6.5 11.3 22.2 18.3 18.6MACS1311 1.2 1.4 – – – 4.7 8.4 20.1 15.6 14.2RXJ1347 1.7 1.5 4.5 2.7 143.7 4.3 7.8 21.1 18.6 18.5MACS1423 1.4 1.8 – – 95.5 5.2 9.5 14.2 12.6 10.3RXJ1532 1.2 1.2 – – 180.3 4.8 8.4 18.3 14.5 13.5MACS1720 0.9 0.8 – – – 4.7 8.7 19.6 14.9 13.0A2261 1.9 1.9 5.8 4.6 108.5 4.6 8.9 20.0 16.1 14.0MACS1931 3.6 2.7 – – 851.9 4.5 8.7 19.5 15.2 13.4MACS2129 1.9 1.7 1.1 1.6 112.6 5.2 13.7 33.5 28.3 29.2RBS1748 1.2 1.2 – – 311.8 4.5 9.4 15.3 14.5 11.4MS2137 1.8 1.6 7.1 7.7 97.7 5.1 9.4 14.5 13.3 11.1AS1063 2.2 1.7 6.6 6.0 76.9 4.8 7.7 14.7 14.6 10.9 We merge the photometric datasets described in the previ-ous section to obtain UV-to-FIR SEDs for all the sourcesin the catalogues released by CLASH. To this end, we usethe
Rainbow
Cosmological Database (P´erez-Gonz´alez et al.2008, Barro et al. 2011a,b) and associated software package.We use CLASH catalogues as parent catalogues to take ad-vantage of the high resolution of HST imaging. However,this requires taking special care of the inevitable blending ofsources in bands with poorer resolution, as well as possiblecounterpart misidentification.In the following subsections, we describe the strategythat we use for the build-up of our multi-wavelength photo-metric catalogue.
Initially,
Rainbow searches for counterparts of our parentcatalogue in the rest of the bands. In practice, each catalogueis cross-matched to the CLASH positions.
Rainbow takesinto account possible astrometry offsets between the bandsby re-aligning each pair of them using the positions of sev-eral sources in small 1 (cid:48) × (cid:48) boxes around a given source. Thesearch radii we use to find counterparts candidates are 1 (cid:48)(cid:48) .5,2 (cid:48)(cid:48) .5, 2 (cid:48)(cid:48) .5, 4 (cid:48)(cid:48) .0, 9 (cid:48)(cid:48) .0, 9 (cid:48)(cid:48) .0, and 12 (cid:48)(cid:48) .0 for IRAC, MIPS 24 µ m,PACS 100 and 160 µ m, and SPIRE 250, 350, and 500 µ m cat-alogues. These values are chosen in order to cope with thetypical WCS offsets between different images, as well as un-certainties in the determination of the center for faint MIPSand Herschel sources. We note, however, that a compari-son of the CLASH vs MIPS/
Herschel coordinates for secure(i.e., bright) mid- and far-IR sources points out that the typ- ical WCS uncertainty is ∼ (cid:48)(cid:48) .2 for IRAC, ∼ (cid:48)(cid:48) .4 for MIPS, ∼ (cid:48)(cid:48) .4 for PACS, and ∼ (cid:48)(cid:48) .3 for SPIRE. In Section 3.3 wetake into account both the search radius and the WCS accu-racy measurements to discuss how many HST counterpartswe find for each M- and FIR source, and how we select themost likely among the former.
The IRAC photometry is recomputed on CLASH positionsfollowing a deconvolution method detailed in Barro et al.(2011a). The procedure is similar to that used in, e.g.,Grazian et al. (2006), Wuyts et al. (2008), Williams et al.(2009), or Wang et al. (2010), and briefly consists on theconvolution of the PSF of the higher resolution image tothe IRAC PSF and a subsequent scaling of the flux of eachsource in a way that the total flux equals the emission of theblended source in the lower resolution image.
Given the larger beam sizes of the M/FIR bands, a sim-ple cross-correlation of the optical/NIR and M/FIR cata-logues frequently assigns the same M/FIR source to differ-ent optical/NIR counterparts (especially when using
HST images). On average, the relaxed search radii we use tocross-match catalogues lead to the assignation of each MIPS24 µ m, PACS, and SPIRE source to 2, 5, and 32 opti-cal/NIR sources, respectively. However, within the WCS ac-curacy measurements there are, on average, 1 optical/NIRsource for each detection in MIPS 24 µ m, PACS, andSPIRE 250 µ m and 250 µ m, and 2 optical/NIR sources for MNRAS , 1–32 (2018)
Un)-obscured star formation in cluster cores each SPIRE 500 µ m source. These latter values are more in-formative of the level of uncertainty in our cross-matchingprocedure and reliability of the counterparts identification,as well as possible blending affecting the low resolutionbands.Due to the large difference between the resolution ofCLASH and M/FIR bands, it is not advisable to apply adeblending procedure such as it was done on IRAC photom-etry. Instead, we limit our approach to the identification ofthe most likely counterpart, or dominant contributor to theM/FIR fluxes, among the multiple short wavelength coun-terparts assigned to the same M/FIR sources. The fact thatthe FIR catalogues are built using IRAC and MIPS 24 µ mpriors guarantees a consistent framework to link the pho-tometry across the whole wavelength range. Different stud-ies have addressed the task of identifying counterparts ofFIR/Sub-millimetre galaxies in shorter wavelengths (e.g.,Alberts et al. 2013), avoiding using simply the shortest dis-tance match with the aim of achieving a more physicallydriven identification. Our approach steps through the N-to-FIR wavelength range and evaluates which of the IR SEDsof the multiple candidates is most likely to be associatedwith the M/FIR detection.We first set local and average SNR limits in the FIRbands. These limits are 2 σ and 3 σ for MIPS and Herschel bands (see Table 3, where we show the flux values corre-sponding to the 5 σ detection in each band and cluster).The 2 σ is used to maximize the information available toidentify the FIR counterparts, however, we clarify that wedo not consider MIPS 24 µ m fluxes below 3 σ detections inthe rest of the work. Then, we select as the optical/NIRcounterpart of each MIPS 24 µ m source the brightest candi-date in the reddest IRAC band available. Then, we shift thismethodology to larger wavelength bands. We select as theoptical/NIR counterpart of each PACS source the brightestcandidate in MIPS 24 µ m. When MIPS is not available, weuse the reddest IRAC band in which the source is detected.Finally, we select as the optical/NIR counterpart of eachSPIRE source, the brightest candidate in the reddest PACSband available, if any. Otherwise, MIPS 24 µ m and IRACbands are used. If different optical/NIR candidates presentvery similar fluxes (within 1 σ ) in the band that is used toidentify the counterpart, we impose a criterion of minimumdistance, and therefore, we select as the optical/NIR coun-terpart the galaxy with the closest position to the M/FIRsource. In all cases described, the MIPS, PACS, and SPIREfluxes of the CLASH sources that are not identified as realcounterparts are flagged and they are not used subsequently.Therefore, each M/FIR source is assigned to a single opti-cal/NIR source. We note that using IRAC as a tracer ofPACS or SPIRE emitters can lead to spurious associations.This is because NIR and FIR trace different components andprocesses in the galaxies. In the clusters with MIPS cover-age, the average fraction of Herschel sources’ optical coun-terparts identified by their IRAC fluxes is 20% and 32%for PACS and SPIRE, respectively. These values increase,however, in those fields without MIPS photometry, reaching91% and 49%, respectively. These cases are flagged for fur-ther check. After a thorough visual inspection of the outputof our procedure, we detect only obvious mismatch casesin galaxies located in the border of the
HST /WFC3 im-ages. We have identified a number of galaxies suffering from over-deblending in the CLASH catalogues, which means thatthe photometry of these galaxies are divided into differentsources. In these cases, the flux of the MIR and FIR cata-logues are generally assigned to source corresponding to thecentral region of the galaxy.
Photometric redshifts ( z phot ) are computed using the EAZY code (Brammer et al. 2008), specifically conceived for thistask.
EAZY is a template-fitting code based on χ mini-mization between observed photometry and a set of 6 SEDtemplates. Among them, 5 templates are generated followingthe Blanton & Roweis (2007) non-negative matrix factoriza-tion algorithm with PEGASE stellar population synthesismodels (Fioc & Rocca-Volmerange 1997) and a calibrationset of synthetic photometry derived from semi-analytic mod-els. The last one is a dusty starburst model, and it is addedto the set in order to compensate for the lack of dusty galax-ies in the calibration photometric sample.The achievable quality of photometric redshifts dependsstrongly on the quality of the photometric dataset itself,and the wavelength domain it covers (e.g., Pacifici et al.2012). In particular, it benefits from high-quality photom-etry sampling strong continuum features (e.g., Lyman orBalmer breaks). In this sense, the 16 CLASH broadbandphotometric points enable high levels of accuracy in the pho-tometric redshift estimation (Jouvel et al. 2014, Molino et al.2017, Connor et al. 2017). In order to make use of the wholepotential of our dataset, we fit not only the whole wave-length range covered by CLASH, but also the IRAC photo-metric points. Furthermore, for those clusters with availablespectroscopic samples we perform a zero-point fine-tuning(following the methodology by Barro et al. 2011a,b) to ac-count for mismatches between the CLASH colours and theSED-fitting template library colours, or other hypotheticalsystematic problems. The median absolute zero-points usedare 3% and 5% for CLASH and IRAC bands, respectively. We assess the quality of the z phot obtained for each clus-ter by comparing them against the available and reliable z spec . We cross-correlate CLASH dataset with the spectro-scopic catalogues using a radius of 0 (cid:48)(cid:48) .5. The total referencespectroscopic sample is composed of 1034 spectroscopicallyconfirmed galaxies within the area of the WFC3 imaging (i.e.the area covered by the photometric catalogues) over the 24CLASH+HLS clusters we analyse. This sample is by defini-tion inhomogeneous, as can be expected of the combinationof studies designed with different scientific objectives and se-lection criteria. It extends between 0.1 9, with the 90%of the galaxies at z< 2. Figure 1 displays the distribution of z spec (empty histogram), and the distribution of magnitudesin the ACS/F814W band (empty histogram; nested panel). The reliability of the z spec is given by the spectroscopic surveysin the form of a quality flag normally linked to the number andSNR of the spectral features identified on the spectrum, that areused to calculate the redshift.MNRAS000 2. Figure 1 displays the distribution of z spec (empty histogram), and the distribution of magnitudesin the ACS/F814W band (empty histogram; nested panel). The reliability of the z spec is given by the spectroscopic surveysin the form of a quality flag normally linked to the number andSNR of the spectral features identified on the spectrum, that areused to calculate the redshift.MNRAS000 , 1–32 (2018) L. Rodr´ıguez-Mu˜noz et al. z spec m AB, F814W 16 19 22 25 28 31 Figure 1. Distribution of z spec for our spectroscopic sample(1034 galaxies; empty histogram). The distribution of the red-shifts of the 378 spectroscopically confirmed cluster membersis given in red. In this figure, we show the distribution up to z =2, which contains 90% of the sample. The nested panel showsthe corresponding distribution of magnitudes in the ACS/F814Wband. A number of quantities have been used in the literatureto quantify the behaviour of the data points in this diagram(see, e.g., Pell´o et al. 2009), either in terms of scatter, aswell as the presence of outliers and systematic offsets. Inthe last decade, the normalized median absolute deviation( σ NMAD ; Hoaglin et al. 1983) of the difference between the z phot and the z spec (∆ z = z phot − z spec ) has been frequentlyused to characterize the scatter of the distribution of z phot (e.g., Ilbert et al. 2009). A typical photometric redshift errordistribution has tails that clearly depart from a pure Gaus-sian distribution, in addition to a relatively large fraction ofoutliers. The σ NMAD estimator manages to achieve a stableestimate of the spread of the core of the z phot distributionwithout being affected by the mentioned tails. It is definedas σ NMAD = 1 . × median (cid:18) | ∆ z − median (∆ z ) | z spec (cid:19) . (1)Following the notation by Barro et al. (2011b), we considerthe fraction of catastrophic outliers, η , defined as those casesfor which | ∆ z | / (1 + z spec ) > . . (2)Finally, in order to characterize the systematic offsets of thephotometric redshifts obtained, δ , we use the expression δ = ∆ z/ (1 + z spec ) . (3)When compared with the spectroscopic sample, our photo-metric redshift estimations present σ NMAD =0.04, and 8% ofcatastrophic outliers (see Figure 2). The outliers are typ-ically either faint sources with noisy photometry in HST and/or IRAC bands (e.g., high redshift galaxies, objects lo-cated in the border of the CLASH catalogues) or galaxies forwhich the IRAC photometry seems to be contaminated bybright nearby objects. We do not identify systematic effects, z ph o t . . A0209A0383MACS0329MACS0429MACS0717MACS1206MACS1311MACS1720A2261 Outliers: 8 % σ NMAD : 0.04 (1+ z )∆ z /(1+ z spec ) = 0.20∆ z /(1+ z spec ) = 0.04 z spec ∆ z / ( + z s p ec ) - . . Figure 2. Evaluation of the z phot quality. The black and reddashed lines show, respectively, the accuracy reached by our re-sults considering the whole spectroscopic sample and the defini-tion of outlier. The vertical lines mark the redshift of each cluster(Table 2). with an average δ = − . 01. These values are comparable withthose published by Jouvel et al. (2014) for CLASH clusters.As we are using the z phot to select cluster members,we also assess their quality using only a subsample of spec-troscopic members. We follow the selection criteria used byMolino et al. (2017, see Section 4.2) in order to be able tocompare our results with theirs. The cluster members refer-ence spectroscopic sample is formed by galaxies for which thedifference between its z spec and the cluster redshift (∆ z cl )fulfills | ∆ z cl | ≤ . 01. Also, in order to guarantee an opti-mal sampling of the optical and NIR SED, only galaxiesdetected at least on 14 CLASH bands are considered. Us-ing these criteria we select 378 galaxies (see red histogramin Figure 1). In this case, our photometric redshift esti-mations present σ NMAD =0.03, and 2% of catastrophic out-liers. These values are comparable with to those obtainedby Molino et al. (2017): σ NMAD =0.02, and η< δ =0.01. MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores In order to derive the physical properties of the galaxiesfound on CLASH+HLS fields, we apply a SED-fitting anal-ysis to the entire dataset gathered and described in the pre-vious sections. We use the Rainbow Cosmological Databasesoftware package (P´erez-Gonz´alez et al. 2008; Barro et al.2011a,b) to fit, on the one hand, the optical/NIR photome-try (CLASH & IRAC), and on the other hand, the M/FIRphotometry (MIPS & Herschel ). In both cases, we fix theredshifts derived with EAZY or, when available, the z spec .In particular, the optical/NIR fitting code performs a χ minimization between the observed data and a set ofsemi-empirical template SEDs computed from spectroscopi-cally confirmed galaxies modeled with PEGASE stellar pop-ulation synthesis models (Fioc & Rocca-Volmerange 1997).In particular, we use the templates generated by P´erez-Gonz´alez et al. 2008 (see their Appendix B) assuming a sin-gle stellar population with a exponentially declining star for-mation history (SFH; SFR ( t ) ∝ e − t/τ ) with a time-scale ( τ )varying between 1 Myr (instantaneous burst) and 100 Gyr(constant SFH) and an age that can take values between1 Myr and 13.5 Gyr. We also assume a Salpeter (1955) IMFspanning stellar masses from 0.1 to 100 M (cid:12) , metallicity ( Z )values 0.005, 0.0.02, 0.2, 0.4, 1.0, 2.5, and 5.0 Z (cid:12) , extinctionbetween 0 and 5 mag, and a Calzetti et al. (2000) attenu-ation law. We complement the set of templates with QSOand AGN empirical templates drawn from Polletta et al.(2007) that account for the galaxies whose UV-to-NIR emis-sion is domitated by an AGN. In the case of the M/FIRSED-fitting, the χ minimization is performed between theobserved photometry and the typical dust emission modelsby Chary & Elbaz (2001), Dale & Helou (2002), Rieke et al.(2009), and Draine & Li (2007). The M ∗ of each galaxy is estimated by Rainbow fromthe average scale factor required to match the templatemonochromatic luminosities to the observed fluxes, weightedwith the photometric errors. The random uncertainty of the M ∗ is derived from the dispersion in the mass-luminosityrations in the different bands. The average expected uncer-tainty in the estimations of M ∗ taking into account varia-tions in Z , SFH, or IMF are within 0.3 dex (P´erez-Gonz´alezet al. 2008). We take advantage of our rich dataset to analyse the SFactivity undergone by the galaxies in these fields in termsof total SFR ( SFR TOT ). Similarly to previous works (seeKennicutt & Evans 2012 and references therein), we considerthat the total SF activity of a galaxy can be derived fromthe combination of (1) the UV luminosity emitted by youngstars that is able to escape from the inter-stellar medium(ISM), and (2) the UV luminosity that is absorbed by theISM and re-emitted in the M/FIR regime. We use the recipeof Bell et al. (2005), which is based on the calibration of Kennicutt (1998): SFR TOT = SFR TIR + SFR UV (4) SFR TIR /M (cid:12) yr − = 1 . × − L TIR /L (cid:12) (5) SFR UV /M (cid:12) yr − = 5 . × − L /L (cid:12) (6)where L TIR is the integrated total IR luminosity and L is the rest-frame monochromatic luminosity at 2800 ˚A (un-corrected for extinction).We compute L TIR by integrating the best-fit Draine &Li (2007) dust emission templates between 8 to 1000 µ m. Aswe mentioned previously, we use four different libraries ofdust emission models in our analysis. The main differencesbetween these models are the prominence of the PAHs andtheir dependence with the total IR luminosity, as well asthe ratio between the mass of hot and cold dust. A discus-sion on these properties is beyond the scope of this paper,nevertheless, we use all these template sets to include thedifferences between the assumptions made by them in theuncertainty of the total IR luminosity. Therefore, the L TIR values given in this work are derived from the Draine & Li(2007) libraries, whereas the uncertainties are the RMS ofthe L TIR estimations using the 4 template libraries. We havechecked that the differences between the luminosities givenby the best fitting templates of each library are of the orderof (cid:46) L interpolating the best fitted op-tical/NIR empirical template at 2800 ˚A(rest-frame). Thiswavelength is covered by observational data over the wholeredshift range of interest.Obviously, this formalism can only be used in the caseof galaxies detected in the M/FIR. For those galaxies notdetected by MIPS or Herschel , we compute SFR TOT bycorrecting the UV luminosities (i.e., SFR UV ) for dust at-tenuation ( A UV ) following the expression SFR TOT = SFR UV , corr . = SFR UV × . A UV (7)where the SFR UV is obtained using Equation 6.Meurer et al. (1999) demonstrate that local starburstgalaxies exhibit a relatively tight, monotonic relation be-tween the ratio between the UV and the TIR luminosity( IRX ) and the UV slope ( β ). Through this relationship,they derive a relation between the extinction of the UV (inparticular, the attenuation at 1600˚A) and the β itself, pro-viding a simple relation that can be applied to correct UVluminosities. However, this and other typical attenuationrecipes based on the UV slope (e.g., Calzetti et al. 1994)are derived for extreme starburst galaxies, while the sourcesfor which we need the correction (i.e. those not-detected inthe M/FIR) are less extreme SFGs. Thus, using those ex-pressions can lead to an overestimation of the extinctionand an overcorrection of the UV luminosity. Therefore, wederive an extinction correction optimized for our work (seeAppendix B).In what follows, the values of the SFR TOT refer to the SFR UV , corr . (Equation 7, in which we use our own A UV ),except in those cases when the M/FIR is available, where The UV continuum slope is defined by assuming that the UVregime of the SED of a galaxy can be described by a power law( ∝ λ β ; Calzetti et al. 1994, Meurer et al. 1999).MNRAS000 SFR UV , corr . (Equation 7, in which we use our own A UV ),except in those cases when the M/FIR is available, where The UV continuum slope is defined by assuming that the UVregime of the SED of a galaxy can be described by a power law( ∝ λ β ; Calzetti et al. 1994, Meurer et al. 1999).MNRAS000 , 1–32 (2018) L. Rodr´ıguez-Mu˜noz et al. l og M ∗ [ M (cid:12) ] z l og L T I R [ L (cid:12) ] - l og S F R T I R [ M (cid:12) y r − ] ClustersField Figure 3. Top panel : Distribution of M ∗ of the cluster membersand field galaxies samples (IRAC 4.5 µ m 3 σ detection) with red-shift. The mass limits obtained for each cluster as it is describedin the text are marked with black diamonds. Bottom panel : Varia-tion with redshift of the total infrared luminosities (left axis) and SFR TIR (right axis) of the parent samples of M-FIR detectedsources. In the case of the field, we show the sample prior to the SFR TIR cuts described in the text. Black symbols indicate the L TIR and SFR TIR corresponding to the limiting flux of MIPS24 µ m (squares) or PACS 100 µ m (diamonds). For each clusteronly the deepest limit is represented. In both panels, red symbolsindicate cluster galaxies, grey symbols indicate field galaxies. Thenumber of field galaxies in this plot has been down-sampled to30% of the original sample size, for visualization purposes. we consider the addition of the SFR TIR and the SFR UV (Equation 4). The most unambiguous way to identify cluster members re-lies on accurate spectroscopic redshifts. However, the ac-quisition of complete z spec samples remains infeasible ex-cept for a relatively small and bright fraction of the galaxypopulation. Indeed, using photometric redshifts to estimatethe distances to galaxies has become a fundamental aim ofgalaxy surveys conducted during recent years (e.g., Ilbertet al. 2009, Barro et al. 2011b). Although less accurate thanspectroscopic ones, photometric redshifts provide a way toestimate distances for galaxies too faint for spectroscopy orsamples too large to be practical for complete spectroscopiccoverage. Given the incomplete and inhomogeneous spec-troscopic coverage of our sample of clusters we are forced to Table 4. Summary of some of the quantities used for the iden-tification of cluster members and an evaluation of the technique:(1) Cluster ID; (2) number of spectroscopic members as definedby Equation 8; (3) σ NMAD derived for the individual clusters; (4)number of σ NMAD to be used in the integration of the P ( z ); (5)membership probability threshold; (6) completeness level (%); (7)fraction of interlopers (%).ID z σ NMAD n P thr K I (1) (2) (3) (4) (5) (6) (7)A0383 33 0.02 3 0.30 91 8A0209 50 0.04 3 0.75 92 7A0611 21 0.03 3 0.55 95 5AS1063 71 0.06 1 0.15 87 10MACS0416 84 0.09 2 0.75 86 13MACS1206 51 0.06 3 0.85 88 11RXJ1347 13 0.07 1 0.25 85 13MACS1149 160 0.12 2 0.85 91 9MACS0717 83 0.05 3 0.75 89 10MACS2129 11 0.09 1 0.70 64 27 use criteria to select cluster members based either on z spec or z phot .The spectroscopic cluster members are identified asthose galaxies with z spec within the redshift range definedby the redshift of the cluster, z cl , and its velocity disper-sion, σ cl . In Table 1 we show the values we use and thecorresponding references. In practice, we use the followingcriteria (see Cava et al. 2009): | z cl − z spec | < × σ cl × (1 + z cl ) (8)For those cases in which a z spec is not available, ourmember selection relies on the redshift probability distribu-tion, P ( z ), given by EAZY instead on the individual z phot associated to each galaxy. This approach captures all thephotometric redshift information, which can significantly re-duce the impact of the catastrophic errors in the z phot - z spec plane (e.g., Fern´andez-Soto et al. 2002). This is of key im-portance to our work, as it translates into a smaller con-tamination with foreground and background sources in ourcluster members selection. In particular, we use the methoddeveloped by Pell´o et al. (2009) based exclusively on photo-metric redshift estimates. This approach modifies the tech-nique presented by Brunner & Lubin (2000) in order to takeadvantage of the P ( z ). It calculates a probability of beinga cluster member ( P member ) integrating P ( z ) within a red-shift range centred in the redshift of the cluster z cl and witha width (∆ z ) related to the accuracy of the photometricredshifts (see Section 4.1). P member = (cid:90) z cl +∆ zz cl − ∆ z P ( z ) dz (9)In our case, we use ∆ z = n × σ NMAD × (1+ z cl ). Applyingthis technique to those galaxies for which we have a reliablespectroscopic redshift we can calibrate the cluster memberselection, which means to find a probability threshold ( P thr )over which a galaxy is considered to be a cluster member,given a certain n . Table 4 shows the values of n and P thr we find to maximize the completeness level ( K ) and mini-mize the percentage of interlopers ( I ) for those clusters withspectroscopic members. Table 4 also gives the values of K and I for each case. We reach K > 80% and I < 20% (limitingvalues used also by Pell´o et al. 2009) for 9 out of the 10 MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores clusters with more than 10 spectroscopic cluster membersavailable. In the case of MACS2129, the cluster with fewerspectroscopic members available (11), we retrieve K =64%and I =27%. Still, the members sample we derive for it in-cludes 73% of correct cluster members. For those clustersfor which less than 10 spectroscopic redshifts were available,we use the average value of n , and the probability thresh-old derived for the other individual clusters: n =2, P thr =0.5.The reader can find examples of the application of a similarselection procedure in the works by (e.g.) Eisenhardt et al.(2008), Vulcani et al. (2011), and Brodwin et al. (2013).Thorough studies of SED-fitting code performance haveidentified and quantified their tendency to derive overconfi-dent P ( z ). This means that the confidence intervals derivedfor the z phot are too narrow. Given that we base our photo-metric cluster members identification on the P ( z ) providedby EAZY , we perform a simple check to evaluate the im-pact of this effect on our work. In practice, we check that thedistribution of spectroscopic redshifts in the cluster is com-parable with the distribution obtained combining the pho-tometric redshifts P ( z ) (Sheth & Rossi 2010). Additionally,we perform the check described by Wittman et al. (2016)through which we find that the overconfidence of the P ( z )we use can be corrected broadening it by applying a con-volution with a σ =0 . P ( z )leads to a different calibration of the membership determi-nation method with smaller P thr . The main objective of our study is to compare the SF ac-tivity that takes place in the inner region of intermediateredshift clusters with the typical observed in lower den-sity environments (i.e., field). In this section, we describethe different galaxy samples from which we derive the re-sults of this work. In the rest of the article the samplesare frequently subdivided in three increasing redshift bins(0.2 10. Our final cluster members sample con-tain 1518 galaxies.We have performed a comparison between the clus-ter members we select using our approach and the mem-bers catalogues published by Connor et al. (2017) for allCLASH clusters. On average, 90 +3 − % of the galaxies withlog M ∗ /M (cid:12) > 10 in each of our samples have a counterpartin their general catalogues. Among them, 87 +9 − % are alsoconsidered cluster members by Connor et al. (2017). Finally,only a 6 +14 − % of galaxies included in the cluster members cat-alogues of their publication are not included in our clustermembers samples. Therefore, in this range of stellar massesthe differences are within our estimated levels of complete-ness and contamination. In order to build a reference sample to which compare theproperties of the cluster members, we make use of the out-standing datasets available on three of the CANDELS fields(Grogin et al. 2011, Koekemoer et al. 2011). In particular,we focus on both the GOODS fields (Giavalisco et al. 2004;see Sections A1, A2) and COSMOS (Scoville et al. 2007; seeSection A3).Using an analogous approach to that described in Sec-tions 3, 4, and 5, we create multi-wavelength cataloguesand derive the photometric redshifts and physical proper-ties (e.g., M ∗ , SFR ) of the galaxies in CANDELS cata-logues. Then, we apply the same spectroscopic and photo-metric redshift criteria to select a field sample correspondingto each cluster members sample in terms of redshift range.Then, for each field sample, we select only the galaxies witha > σ detection in IRAC 4.5 µ m band and a IRAC 4.5 µ mflux larger than the 3 σ detection limit of each correspond-ing cluster sample. Figure 3 represents the distribution ofthe field samples in the M ∗ - z plane.The final field parent sample contains 7466 systems withlog M ∗ /M (cid:12) > 10. We exclude the 360 galaxies without arobust mass estimation (see previous section). We divide the samples of field and cluster galaxies into star-forming and passive using the rest-frame U − V vs V − J MNRAS , 1–32 (2018) L. Rodr´ıguez-Mu˜noz et al. U − V . . . . . Clusters UVJ-SFClusters UVJ-PClusters MIRClusters FIR z< V − J U − V . . . . . z ≥ Field (68%)Field Figure 4. UV J -diagram for the cluster members (circles) andfield galaxies (grey contours and points) in two redshift bins (toppanel 0 UV J -diagram). Differentworks (e.g., Wuyts et al. 2007, Williams et al. 2009) have evi-denced the power of the UV J -diagram to select pure samplesof either quiescent and SFGs (e.g., Wuyts et al. 2007, Bram-mer et al. 2011, Whitaker et al. 2012a, Whitaker et al. 2015).In particular, we identify passive galaxies (hearafter, UV J -P) following the recipes by Williams et al. (2009) for theredshift bins 0 Figure 5. s SFR TOT vs. M ∗ for the star-forming cluster mem-bers (blue points) and field galaxies (grey contours and points) inthe two redshift bins in Figure 4. The cluster members detectedin the FIR are highlighted with larger blue circles and a red bor-der. The black lines represent the MS by Renzini & Peng (2015)scaled to the median redshift of the corresponding bin consideringan evolution with redshift of the s SFR of the shape (1+ z ) . ± . (Sargent et al. 2012). of galaxies considering the uncertainties in the syntheticphotometry. We retrieve ≤ 1% differences in the numbercounts of either category and sample. We find that in theclusters (field) samples, 25% (5%) of SFGs could be classifiedas passive given their error bars and 28% (22%) of passivegalaxies could be classified as SFGs. We have checked thatexcluding the galaxies in the vicinities of the limits betweenthe UV J -P and the UV J -SF loci do not change the resultsof our work significantly. This is probably due to the factthat these transition galaxies present similar properties oneither side of the border.In Figure 4, we show the UV J -diagram for the clusterand field samples. As we can see, some galaxies detected inthe FIR (i.e., presumably SFGs) are located in the region MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores l og S F R T O T [ M (cid:12) y r − ] MS @ Field (68 %)FieldClustersClusters FIR UVJ-SF < z < - . . . . . l og S F R T I R [ M (cid:12) y r − ] MS @ Field (68 %)FieldClustersClusters FIR M-FIR < z < - . . . . . l og S F R T O T [ M (cid:12) y r − ] < z < UVJ-SF - . . . . . l og S F R T I R [ M (cid:12) y r − ] < z < M-FIR - . . . . . log M ∗ [ M (cid:12) ] l og S F R T O T [ M (cid:12) y r − ] < z < UVJ-SF7.5 8.5 9.5 10.5 11.5 - . . . . . log M ∗ [ M (cid:12) ] l og S F R T I R [ M (cid:12) y r − ] < z < M-FIR7.5 8.5 9.5 10.5 11.5 - . . . . . Figure 6. SFR TOT vs M ∗ relation for the star-forming cluster members in our study split up in three increasing z bins (top, middle,and bottom panels). On the (left-) right-hand panels, we include the ( UV J -SF) M-FIR galaxies across the whole mass range. The SFR TOT refers to the SFR TIR + SFR UV for those galaxies M-FIR detected, and SFR UV,corr otherwise. Blue points always representthe distribution of clusters members in both cases. Those galaxies detected in the FIR (i.e., Herschel ) are shown with larger blue pointshighlighted with red borders. grey contours represent the distribution (68 confidence levels) of field galaxies. We also display the MS byRenzini & Peng (2015, black lines) scaled to the median redshift of the corresponding subsample of cluster members considering a trendof s SFR with redshift ∝ (1 + z ) . ± . (Sargent et al. 2012). The shaded areas represent the selection criteria used to build the finalsamples of UV J -SF and M-FIR galaxies (i.e. they represent the cut in M ∗ , and SFR TIR ).MNRAS000 TIR ).MNRAS000 , 1–32 (2018) L. Rodr´ıguez-Mu˜noz et al. CLASH0001783CLASH0001653CLASH0000032A0209A0209A03830.2000.2100.190z spec z spec z spec RGB F814W F160W IRAC 3.6 IRAC 4.5 MIPS 24 PACS 100 PACS 160 SPIRE 250 SPIRE 350 SPIRE 500 Figure 7. Thumbnails of 3 cluster members from the M-FIR sample ordered by increasing redshift. From left to right we display aRGB image (5” × HST /ACS/F814W, F606W, and F435W, following the methodology by Lupton et al. (2004), 30” × HST /ACS/F814W and HST /WFC3/F160W bands followed by Spitzer /IRAC and MIPS, and Herschel /PACSbands, and 90” × 90” postage stamps in the Herschel /SPIRE bands, all ordered by increasing wavelength. When there is a difference inthe sizes of two adjacent frames, we mark with an orange square the size of the smallest on the largest. On the left side we show theID of the object, the name of the cluster, the redshift and either if it is photometric or spectroscopic. The thumbnails of the rest of thesample can be found as online material. theoretically populated by passive galaxies. This contami-nation has been reported in the past (see, e.g., Dom´ınguezS´anchez et al. 2016) and evidences the necessity of a cor-rection of the aforementioned selection criteria. In the final UV J -SF ( UV J -P) samples, we include (exclude) both thegalaxies located in the SFGs locus of the UV J -diagram andthose detected in the M/FIR (see Section 7.4) independentlyof their position in the UV J -diagram. This correction in-creases (decreases) 1% (1%) and 2% (5%) the number ofstar-forming (passive) galaxies in the cluster and field sam-ples, respectively.The UV J -SF ( UV J -P) samples built in CLASH-HLSclusters and the field include 443 (1075) and 4649 (2817)log M ∗ /M (cid:12) > 10 galaxies, respectively.An alternative methodology to select SFGs uses athreshold of s SFR under which a galaxy is considered tobe passive (e.g., Kimm et al. 2009). In Figure 5, we rep-resent the s SFR TOT - M ∗ diagrams for the UV J -SF sam-ples in the redshift bins of Figure 4. We can see thatour UV J -SF selection criteria corresponds approximately tolog s SFR TOT /yr − (cid:38) − . UV J -SF samples selected in the clustersand the field on the SFR TOT - M ∗ plane. The blue shadedarea illustrates the effective definition of the UV J -SF sam-ples considered in the rest of the work. For comparison, wealso represent the MS defined by Renzini & Peng (2015,black line) scaled to the median redshift of the bin, assum-ing and evolution with redshift of the s SFR of the shape(1 + z ) . ± . (Sargent et al. 2012). We notice a systematicoffset of the distribution of cluster SFGs towards lower SFR at fixed M ∗ (see also Figure 5). The quantification of thisdifference can be found in Section 9.3. In order to build comparable samples of galaxies(log M ∗ /M (cid:12) > 10) detected in the M- and/or FIR (M-FIRsamples), we perform the following steps. First, we selectgalaxies with at least a 3 σ detection in one of the M- and/or FIR bands available (i.e., MIPS 24 µ m, PACS 100 & 160 µ m,and SPIRE 250, 350 & 500 µ m), and flux larger than thelimiting fluxes at 3 σ level in the clusters (see Table 3 forthe limiting fluxes at 5 σ detection level). These galaxies arerepresented in the bottom panel of Figure 3. Then, we selectonly the 50 (1496) clusters (field) galaxies for which the es-timated SFR TIR is larger than the (conservative) SFR TIR limits obtained for each cluster (black symbols in the bot-tom panel of Figure 3). Figure 7 shows the thumbnails ofthe cluster members detected in the M- and/or FIR. Finally,we consider galaxies with SFR TIR > M (cid:12) yr − to obtain acomparable set of samples of galaxies throughout the wholeredshift range. This value is larger than the SFR TIR lim-its of our sample, except for the four furthest clusters. Ourfinal M-FIR samples include 36 cluster members and 974field galaxies. On the right-hand half of Figure 6, we dis-play the distribution of these samples on the SFR TIR - M ∗ plane. The red shaded area marks the M ∗ and SFR TIR cutsperformed to define the samples.It is worth mentioning that we perform a visual inspec-tion of each cluster member selected as a M-FIR emitter.We exclude spurious MIPS 24 µ m sources without a coun-terpart in longer wavelengths (e.g., sources on Airy ring fea-tures), galaxies in the borders of the images that are selectedas counterparts of M/FIR sources with coordinates outsidethe area covered by CLASH catalogues, or galaxies sufferingfrom over-deblending in the CLASH catalogues.Interestingly, we find 8 BCGs detected in the M/FIRout of 24 clusters, which corresponds to 33% of our sample.This percentage is consistent with the results of the studyconducted by Rawle et al. (2012a) using HLS data on a sam-ple of 68 massive galaxy clusters spread out in the redshiftrange between 0.08 Un)-obscured star formation in cluster cores bright cluster members has been observed to increase rapidlyfrom 3% up to 65% for galaxies with increasing L TIR valuesvarying from 10 L (cid:12) to > . L (cid:12) in clusters within theredshift range 0.15 TIR from tens to thou-sands of M (cid:12) yr − . Our M-FIR sample of cluster members in-cludes 25 LIRGs and 1 ULIRGs (within CLJ1226, the high-est redshift cluster) and our M-FIR sample of field galaxiesincludes 639 LIRGs, and 10 ULIRGs. These numbers cor-respond to comparable percentages of LIRGs and ULIRGswithin the M-FIR samples in clusters and field. MNRAS , 1–32 (2018) L . R od r ´ ı g u e z - M u ˜ n o z e t a l . Table 5. Number of galaxies selected with the different criteria used to build the final samples of star-forming cluster members ( R /R < > σ in IRAC 4.5 µ m and M ∗ > M (cid:12) ; we show within parenthesesthe number of cluster members without the M ∗ cut; (3) number of galaxies selected as star-forming using the UVJ diagram and/or detected in the MIR and/or FIR ( M ∗ > M (cid:12) ),what we call the UV J -SF sample ; (4) cluster members with M ∗ > M (cid:12) detected in the MIR and/or FIR with a SFR TIR > (cid:12) yr − , what we call M-FIR sample ; we show thetotal number without SFR TIR cut within parentheses; (5 & 6) fraction of UV J -SF and M-FIR galaxies, respectively, obtained using as reference the number of cluster members with M ∗ > M (cid:12) ; (7 & 8) median and quantiles 16 th and 84 th values of the M ∗ of the UV J -SF and M-FIR samples respectively; (9) median and quantiles 16 th and 84 th values of the SFR TOT for the UV J -SF sample obtained as the addition of the SFR TIR and the SFR UV when the former is available, and the SFR UV , corr in the rest of the cases; (10) medianand quantiles 16 th and 84 th values of the SFR TOT for the M-FIR sample obtained as the addition of the SFR TIR and the SFR UV ; (11 & 12) median and quantiles 16 th and 84 th values of the s SFR TOT for the UV J -SF and the M-FIR sample, respectively. Cluster ID Members UV J -SF M-FIR F UV J − SF F M − FIR M ∗ , UVJ − SF M ∗ , M − FIR SFR TOT ,UV J − SF SFR TOT , M − FIR s SFR TOT ,UV J − SF s SFR TOT , M − FIR (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)a0383 11 (53) 2 0 (0) 0.17 ± +0 . − . – 0.46 +0 . − . – -10.24 +0 . − . –a0209 23 (72) 8 0 (1) 0.35 ± +0 . − . – 0.68 +0 . − . – -10.19 +0 . − . –a2261 30 (192) 7 0 (3) 0.23 ± +0 . − . – 0.55 +0 . − . – -9.73 +0 . − . –rbs1748 14 (48) 5 0 (0) 0.36 ± +0 . − . – 0.06 +0 . − . – -10.25 +0 . − . –a0611 18 (34) 6 0 (0) 0.33 ± +0 . − . – 0.48 +0 . − . – -9.85 +0 . − . –ms2137 6 (11) 2 0 (0) 0.33 ± +0 . − . – -0.18 +0 . − . – -10.46 +0 . − . –as1063 28 (48) 15 1 (1) 0.54 ± ± +0 . − . −−−− +1 . − . −−−− -10.29 +0 . − . -9.15 −−−− macs1931 22 (47) 8 0 (0) 0.36 ± +0 . − . – 0.48 +0 . − . – -9.71 +0 . − . –macs1115 18 (31) 4 0 (0) 0.22 ± +0 . − . – 0.01 +0 . − . – -10.32 +0 . − . –rxj1532 14 (28) 4 0 (0) 0.29 ± +0 . − . – 0.53 +0 . − . – -9.89 +0 . − . –macs1720 15 (37) 4 0 (0) 0.27 ± +0 . − . – 0.62 +0 . − . – -9.83 +0 . − . –macs0416 34 (53) 12 1 (1) 0.35 ± ± +0 . − . −−−− +0 . − . −−−− -9.85 +0 . − . -9.03 −−−− macs0429 8 (21) 4 0 (0) 0.50 ± +0 . − . – 0.86 +0 . − . – -9.98 +0 . − . –macs1206 35 (73) 15 1 (1) 0.43 ± ± +0 . − . −−−− +0 . − . −−−− -9.84 +0 . − . -9.97 −−−− macs0329 13 (34) 8 0 (0) 0.62 ± +0 . − . – 0.54 +0 . − . – -9.88 +0 . − . –rxj1347 28 (44) 12 0 (0) 0.43 ± +0 . − . – 0.32 +0 . − . – -9.93 +0 . − . –macs1311 22 (42) 8 1 (1) 0.36 ± ± +0 . − . −−−− +0 . − . −−−− -9.85 +0 . − . -9.18 −−−− macs1149 42 (82) 20 0 (0) 0.48 ± +0 . − . – 0.47 +0 . − . – -9.92 +0 . − . –macs0717 57 (72) 8 0 (0) 0.14 ± +0 . − . – 0.35 +0 . − . – -9.97 +0 . − . –macs1423 26 (30) 7 1 (1) 0.27 ± ± +0 . − . −−−− +1 . − . −−−− -10.00 +0 . − . -9.12 −−−− macs2129 17 (18) 4 1 (1) 0.24 ± ± +0 . − . −−−− +0 . − . −−−− -9.75 +0 . − . -8.71 −−−− macs0647 17 (24) 7 0 (0) 0.41 ± +0 . − . – 0.63 +0 . − . – -10.08 +0 . − . –macs0744 20 (37) 9 0 (0) 0.45 ± +0 . − . – 0.68 +0 . − . – -9.76 +0 . − . –clj1226 33 (57) 18 0 (0) 0.55 ± +0 . − . – 0.49 +0 . − . – -10.08 +0 . − . –Total 551 (1188) 197 6 (10)Median 0.2 As in Table 5, for the galaxies at 0.1 < R /R < Cluster ID Members UV J -SF M-FIR F UV J − SF F M − FIR M ∗ , UV J − SF M ∗ , M − FIR SFR TOT ,UV J − SF SFR TOT , M − FIR s SFR TOT ,UV J − SF s SFR TOT , M − FIR (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)a0383 10 (55) 1 0 (0) 0.10 ± −−−− – -0.19 −−−− – -10.28 −−−− –a0209 3 (20) 1 0 (1) 0.33 ± −−−− – 1.13 −−−− – -9.45 −−−− –a2261 22 (140) 5 0 (1) 0.23 ± +0 . − . – 0.51 +0 . − . – -10.22 +0 . − . –rbs1748 12 (53) 1 0 (1) 0.08 ± −−−− – 0.95 −−−− – -10.07 −−−− –a0611 30 (64) 7 1 (2) 0.23 ± ± +0 . − . −−−− +0 . − . −−−− -9.81 +0 . − . -8.96 −−−− ms2137 13 (25) 3 1 (1) 0.23 ± ± +0 . − . −−−− +0 . − . −−−− -9.95 +0 . − . -9.25 −−−− as1063 24 (48) 6 0 (0) 0.25 ± +0 . − . – -0.14 +0 . − . – -10.47 +0 . − . –macs1931 27 (56) 3 0 (0) 0.11 ± +0 . − . – 0.24 +0 . − . – -10.32 +0 . − . –macs1115 23 (54) 9 1 (1) 0.39 ± ± +0 . − . −−−− +0 . − . −−−− -9.80 +0 . − . -9.41 −−−− rxj1532 12 (31) 1 0 (1) 0.08 ± −−−− – 1.03 −−−− – -9.19 −−−− –macs1720 26 (65) 8 1 (1) 0.31 ± ± +0 . − . −−−− +0 . − . −−−− -9.83 +0 . − . -9.53 −−−− macs0416 24 (63) 5 1 (1) 0.21 ± ± +0 . − . −−−− +0 . − . −−−− -9.96 +0 . − . -8.91 −−−− macs0429 12 (41) 5 0 (1) 0.42 ± +0 . − . – 0.04 +0 . − . – -10.55 +0 . − . –macs1206 42 (86) 9 0 (0) 0.21 ± +0 . − . – 0.18 +0 . − . – -10.08 +0 . − . –macs0329 27 (59) 6 0 (0) 0.22 ± +0 . − . – 0.37 +0 . − . – -10.03 +0 . − . –rxj1347 29 (67) 2 0 (0) 0.07 ± +0 . − . – 1.19 +0 . − . – -9.48 +0 . − . –macs1311 27 (61) 8 2 (2) 0.30 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.54 +0 . − . -8.84 +0 . − . macs1149 70 (158) 16 3 (3) 0.23 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.67 +0 . − . -8.90 +0 . − . macs0717 80 (107) 15 6 (6) 0.19 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.52 +0 . − . -9.27 +0 . − . macs1423 17 (29) 2 0 (0) 0.12 ± +0 . − . – 0.57 +0 . − . – -9.49 +0 . − . –macs2129 33 (38) 1 0 (0) 0.03 ± −−−− – 0.94 −−−− – -9.27 −−−− –macs0647 27 (54) 14 1 (1) 0.52 ± ± +0 . − . −−−− +0 . − . −−−− -9.50 +0 . − . -8.80 −−−− macs0744 33 (56) 6 0 (0) 0.18 ± +0 . − . – 0.91 +0 . − . – -9.75 +0 . − . –clj1226 38 (60) 15 0 (0) 0.39 ± +0 . − . – 0.46 +0 . − . – -9.80 +0 . − . –Total 661 (1490) 149 17 (23)Median 0.4 As in Table 5, for the galaxies at 0.2 < R /R < Cluster ID Members UV J -SF M-FIR F UV J − SF F M − FIR M ∗ , UV J − SF M ∗ , M − FIR SFR TOT ,UV J − SF SFR TOT , M − FIR s SFR TOT ,UV J − SF s SFR TOT , M − FIR (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)a0383 6 (14) 1 0 (1) 0.17 ± −−−− – 0.26 +0 . − . – -10.53 −−−− –a0209 0 (0) 0 0 (0) – – – – – – – –a2261 0 (0) 0 0 (0) – – – – – – – –rbs1748 0 (0) 0 0 (0) – – – – – – – –a0611 5 (11) 2 0 (1) 0.40 ± +0 . − . – 0.69 +0 . − . – -9.48 +0 . − . –ms2137 10 (24) 1 0 (0) 0.10 ± −−−− – 1.13 −−−− – -8.94 −−−− –as1063 5 (10) 2 0 (0) 0.40 ± +0 . − . – 0.31 +0 . − . – -10.34 +0 . − . –macs1931 18 (33) 4 0 (0) 0.22 ± +0 . − . – 0.19 +0 . − . – -9.99 +0 . − . –macs1115 13 (30) 2 0 (0) 0.15 ± +0 . − . – 0.13 +0 . − . – -10.01 +0 . − . –rxj1532 9 (25) 0 0 (0) – – – – – – – –macs1720 19 (49) 10 4 (4) 0.53 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.72 +0 . − . -9.10 +0 . − . macs0416 9 (25) 3 0 (0) 0.33 ± +0 . − . – 0.55 +0 . − . – -9.80 +0 . − . –macs0429 12 (51) 6 0 (1) 0.50 ± +0 . − . – 0.24 +0 . − . – -9.85 +0 . − . –macs1206 17 (35) 4 0 (1) 0.24 ± +0 . − . – 0.62 +0 . − . – -9.86 +0 . − . –macs0329 27 (85) 6 0 (0) 0.22 ± +0 . − . – 0.34 +0 . − . – -9.81 +0 . − . –rxj1347 12 (29) 4 0 (0) 0.33 ± +0 . − . – 0.37 +0 . − . – -9.98 +0 . − . –macs1311 9 (27) 3 0 (0) 0.33 ± +0 . − . – 0.59 +0 . − . – -9.81 +0 . − . –macs1149 15 (48) 7 1 (1) 0.47 ± ± +0 . − . −−−− +0 . − . −−−− -9.80 +0 . − . -8.98 −−−− macs0717 13 (17) 0 0 (0) – – – – – – – –macs1423 7 (12) 2 1 (1) 0.29 ± ± +0 . − . −−−− +0 . − . −−−− -9.11 +0 . − . -8.86 −−−− macs2129 17 (22) 3 0 (0) 0.18 ± +0 . − . – 0.78 +0 . − . – -9.95 +0 . − . –macs0647 12 (36) 5 0 (0) 0.42 ± +0 . − . – 0.78 +0 . − . – -9.31 +0 . − . –macs0744 32 (57) 17 2 (2) 0.53 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.79 +0 . − . -8.45 +0 . − . clj1226 39 (62) 15 5 (5) 0.38 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -8.95 +0 . − . -8.84 +0 . − . Total 306 (702) 97 13 (17)Median 0.6 As in Table 5, for the field samples. Field ID Galaxies UV J -SF M-FIR F UV J − SF F M − FIR M ∗ , UV J − SF M ∗ , M − FIR SFR TOT ,UV J − SF SFR TOT , M − FIR s SFR TOT ,UV J − SF s SFR TOT , M − FIR (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)A0383 97 (528) 67 16 (43) 0.69 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.74 +0 . − . -9.34 +0 . − . A0209 114 (648) 84 18 (57) 0.74 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.78 +0 . − . -9.37 +0 . − . A2261 124 (635) 83 18 (52) 0.67 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.75 +0 . − . -9.31 +0 . − . RBS1748 137 (657) 95 25 (65) 0.69 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.72 +0 . − . -9.28 +0 . − . A0611 109 (598) 71 15 (43) 0.65 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.60 +0 . − . -9.09 +0 . − . MS2137 139 (792) 74 15 (40) 0.53 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.71 +0 . − . -9.23 +0 . − . AS1063 160 (789) 78 16 (43) 0.49 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.58 +0 . − . -9.15 +0 . − . MACS1931 160 (462) 78 16 (46) 0.49 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.57 +0 . − . -9.15 +0 . − . MACS1115 151 (674) 72 16 (38) 0.48 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.61 +0 . − . -9.15 +0 . − . RXJ1532 181 (699) 95 22 (64) 0.52 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.54 +0 . − . -9.21 +0 . − . MACS1720 224 (919) 107 31 (59) 0.48 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.49 +0 . − . -9.28 +0 . − . MACS0416 220 (851) 106 31 (58) 0.48 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.45 +0 . − . -9.22 +0 . − . MACS0429 220 (795) 106 31 (60) 0.48 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.42 +0 . − . -9.22 +0 . − . MACS1206 250 (936) 155 36 (55) 0.62 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.46 +0 . − . -9.18 +0 . − . MACS0329 270 (972) 173 43 (68) 0.64 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.44 +0 . − . -9.19 +0 . − . RXJ1347 270 (1066) 176 47 (84) 0.65 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.44 +0 . − . -9.22 +0 . − . MACS1311 408 (1218) 256 67 (83) 0.63 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.45 +0 . − . -9.21 +0 . − . MACS1149 470 (1414) 296 72 (72) 0.63 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.60 +0 . − . -9.21 +0 . − . MACS0717 469 (970) 297 80 (80) 0.63 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.52 +0 . − . -9.22 +0 . − . MACS1423 470 (1036) 297 69 (69) 0.63 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.59 +0 . − . -9.21 +0 . − . MACS2129 539 (1462) 333 41 (41) 0.62 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.55 +0 . − . -9.06 +0 . − . MACS0647 484 (1218) 311 54 (54) 0.64 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.55 +0 . − . -9.03 +0 . − . MACS0744 860 (1706) 533 113 (113) 0.62 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.29 +0 . − . -9.03 +0 . − . CLJ1226 940 (1576) 706 82 (82) 0.75 ± ± +0 . − . +0 . − . +0 . − . +0 . − . -9.20 +0 . − . -8.86 +0 . − . Total 7466 (22621) 4649 974 (1469)Median 0.2 As a step prior to the evaluation of the SF within clustercores and how it compares to the SF in the field, we explorethe stellar mass function (SMF) of the samples presented inthe previous section. The SMF is a fundamental observablefor the study of the evolution of galaxy populations. Fur-thermore, overlooking hypothetical differences in the SMFof field and cluster samples can lead to a misinterpretation ofthe physics behind the level of SF quantified in the followingsections.In the top panels of Figure 8, we display the SMF forclusters and field galaxies (log M ∗ /M (cid:12) > 10) divided intobins of redshift. We include only galaxies at R < R , i.e.,the R range homogeneously covered along the whole redshiftrange. We exclude the BCGs in our analysis. We correctfor different cluster richnesses by randomly re-sampling thegalaxy population of each cluster using the average samplesize of each redshift bin. Then, to render the field and clus-ter samples statistically comparable, we re-sample each fielddrawing randomly the number of galaxies in the correspond-ing cluster sample. The uncertainties are estimated from thecombination of 500 bootstraps. Then, we model the data byfitting a Schechter function (Schechter 1976) to the SMF.The form of the function isΦ ( M ∗ ) d M ∗ = Φ ∗ (cid:18) M ∗ M ∗ (cid:19) α e − M∗M∗ d M ∗ M ∗ , (10)with M ∗ being the characteristic mass, α the low-massslope, and Φ ∗ the normalization. The normalization is evalu-ated by requiring that the integral of the Schechter functionover the stellar mass range considered equals the fraction ofgalaxies in the sample fitted with respect the total sample.In Table 9 we report the best-fit parameters. The functionprovides overall reasonable fits, although we report a quitelarge scatter of the data points for some of the samples. Thisis probably due to the limited number counts we work with.In the bottom panels of Figure 8, we display the fractionof UV J -P and UV J -SF galaxies in each stellar mass bin.The plots are not perfectly symmetric because we do notfix the median value of each mass bin. We do not representthe stellar mass distribution of the M-FIR sample becauseits size is not statistically significant for this analysis. Themedian value of stellar mass corresponding to each sampleis marked in the upper panels of the same figure (see alsoTable 9).We compare the best-fitting Schechter parameters withthose published recently by van der Burg et al. (2018) forcluster and field galaxies at 0.5 UV J -P populations, for which we derive larger values:log M ∗ =11.30 +0 . − . , and 11.32 +0 . − . , respectively. Regard-ing α , they retrieve − . +0 . − . , − . +0 . − . , and − . +0 . − . for the whole population, the quiescent, and the star-forming samples of the clusters, and − . +0 . − . , − . +0 . − . , and − . +0 . − . for the field. In this case, our results are com-patible with theirs within the error bars.In the first two redshift bins, there are no large differ-ences between the SMF of the whole population of galaxiesin the field and the clusters, with values of the slope and theknee of the Schechter function within the 1 σ errors (see Ta-ble 9). This result has been found in previous works at inter-mediate and high redshift (e.g., Vulcani et al. 2012, Vulcaniet al. 2013, van der Burg et al. 2013, Nantais et al. 2016).On the contrary, the highest redshift bin displays large dif-ferences between the cluster and the field best-fit Schechterfunctions. We claim these differences are mainly due to apoor sampling of the cluster SMF. In fact, data points inthe stellar mass range including 80% of the stellar mass ofboth cluster and field samples are compatible within the er-ror bars.We report hints of a different behaviour of the SMFs offield and clusters and their evolution with z when we splitthe galaxy populations in UV J -SF and UV J -P. At the low-est redshift, the UV J -P SMF appears to present a steeper α than the field, which is not obvious in the second redshiftbin. This makes the UV J -P SMF present a shape apparentlymore similar to the field UV J -SF stellar mass distributions(excluding normalization differences). Balogh et al. (2001)also find that while in the field environment the SMF ofSFGs has much steeper faint-end slope than that for pas-sive galaxies, in the clusters, the passive galaxies have also asteep faint-end. Annunziatella et al. (2014) find that for the z =0.44 (our second redshift bin) cluster MACS1206 (alsoincluded in our sample), the SMF of SFGs is significantlysteeper than the SMF of passive galaxies at the faint end.This is in agreement with our best-fitting SMFs in the inter-mediate redshift bin. Furthermore, they find a smaller slopesSMF for passive cluster galaxies in the inner core of clusters( R /R (cid:46) α and log M ∗ for the UV J -SF and UV J -P samples in the clusters and in the field areoverall compatible within the error bars. The only signif-icant difference appears in the value of the log M ∗ forthe UV J -SF samples in the lowest redshift bin: 10.55 +0 . − . and 11.11 +0 . − . for the clusters and the field, respectively.Other works have also reported the lack of significant differ-ences between the SMF of star-forming and passive galax-ies in different environments (i.e., Vulcani et al. 2013). The UV J -SF and UV J -P SMF evolution with redshift is alsomild in terms of the best-fitting Schechter parameters α andlog M ∗ , and considering our resolution.In the first two redshift bins, we find that the galaxypopulation in massive clusters is clearly dominated by qui-escent galaxies all the way down to M ∗ =10 M (cid:12) , whichis in agreement with (e.g.) van der Burg et al. (2018).The largest mass bins are dominated by stochasticity giventhe small number of galaxies included. Peng et al. (2010)predicts that the SMFs of passive and SFGs should cross(crossing mass) at log M ∗ /M (cid:12) ≈ M ∗ /M (cid:12) ≈ 10 in the second redshift bin. In the thirdredshift bin, the contribution of UV J -SF and UV J -P sam-ples to the whole population of clusters is ≈ MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores Φ . . . . ClustersAll UV J -SF UV J -PFieldAll UV J -SF UV J -P 0.19 10 10.5 11 11.5 0.60 Top panels : Stellar mass distribution within R /R < M ∗ /M (cid:12) > 10) divided in bins ofredshift. On the upper part of each panel, we mark the median stellar mass of every sample. Bottom panels : Relative fraction of UV J -Pand UV J -SF galaxies as a function of stellar mass. Table 9. We report: Median log M ∗ (and 1 σ intervals), best-fitting Schechter parameters (and 1 σ intervals) and reduced χ for thedifferent samples. Sample < log M ∗ /M (cid:12) > log M ∗ /M (cid:12) α Φ ∗ χ We report: Median log M ∗ (and 1 σ intervals), best-fitting Schechter parameters (and 1 σ intervals) and reduced χ for thedifferent samples. Sample < log M ∗ /M (cid:12) > log M ∗ /M (cid:12) α Φ ∗ χ UV J -P and UV J -SF type fractions derivedby Nantais et al. (2016) for z ∼ M ∗ /M (cid:12) < UV J -P and UV J -SF galaxies tend to converge and even to beinverse towards higher mass bins. Other previous studies(e.g., Quadri et al. 2012, Nantais et al. 2016, Papovich et al.2018) have claimed a rapid increase in the number densityof low- and intermediate-mass (log M ∗ / M (cid:12) < z ≈ M ∗ completeness levels hampers the analysis of a pos-sible evolution of the distribution of stellar mass at such lowvalues.It is worth noting that numerous works (e.g., Annun-ziatella et al. 2016) find that passive cluster galaxies arebetter fitted by a double Schechter function, revealing theexistence of two sub-populations of red cluster membersthought to have followed distinct evolutionary paths. Onthe one hand, a population of high mass galaxies thought tobe quenched by processes scaling with stellar mass, and onthe other hand, a population of low-mass galaxies quenchedby environmental processes (Peng et al. 2010). These com-posite SMF of red passive galaxies have also been observedin the field in works such as, e.g., Drory et al. (2009) andBaldry et al. (2012). However, the evidence for these doubleSchechter functions (i.e., an upturn at low stellar masses) isonly visible at log M ∗ /M (cid:12) (cid:46) 10 (Drory et al. 2009), belowthe mass limit of our work. In this section, we present a quantification of the SF ac-tivity hosted by cluster members and field galaxies withlog M ∗ /M (cid:12) > 10, as traced by the UV and the M- and FIR. Figure 9 (left hand panel) shows the fraction ( F ) of UV J -SFand M-FIR galaxies ( F UV J − SF and F M − FIR , respectively;Section 7.4) in the clusters ( R /R < assuming a standard normal distribution. On theright panel, we show the median F and quantiles 16 th and84 th (in the shape of error bars) in the same redshift binsof Figure 6. We also include with larger symbols the frac-tions obtained at 0.1 < R /R < < R /R < F with redshift, we fit to The confidence interval of a point sample estimate of the pop-ulation proportion at 1 σ can be derived considering a standardnormal distribution with the expression (cid:112) p (1 − p ) /n , where n isthe size of the sample and p is the proportion. Both of them mustsatisfy the condition that n p ≥ n (1 − p ) ≥ the data points (fraction for each individual cluster within R /R < α (1 + z ) β , where α corresponds to the value of F at z =0, and β describes itsevolution with redshift (with larger values of β meaning asteeper trend). This methodology is also applied by (e.g.)Haines et al. (2013) and Alberts et al. (2014). The corre-sponding curves and 1 σ confidence intervals (generated us-ing Monte Carlo simulations) are over-plotted in Figure 9with a coloured line and a shaded area around it, respec-tively. Table 10 shows the α and β values of the best-fit. Inthe case of the M-FIR samples, we fit only the clusters witha SFR TIR limit below 10 M (cid:12) yr − ( z< F within clusters is much smallerthan in the field for both UV J -SF and M-FIR samples. Onaverage, F UV J − SF in clusters seems to be approximately 1 / F M − FIR in clusters drop down tovalues not significantly different to zero. Assuming the samefraction of M-FIR galaxies among the SFGs in clusters andfield, the expected average F M − FIR for the former would be ∼ SFR TIR > (cid:12) yr − ) and/or dustysystems in the inner cores of clusters at intermediate red-shifts.Figure 9 also displays different evolutions of F for clus-ters and field with z . The latter displays mild increas-ing trends for F UV J − SF and F M − FIR , which vary with β =0.2 ± β =0.2 ± F remains ∼ 60% forthe UV J -SF samples between z =0.19-0.89. Flat/mild trendsfor the fraction of the star-forming population of galaxies inthe field at intermediate redshifts ( z< 1) are also found byBrammer et al. 2011 and Darvish et al. 2017. In particular,the latter gives 70% of fraction of SFGs which is compara-ble with our results, although there is a larger offset betweenthese numbers and the 40% given by the former. These dif-ferences are likely due to the sample selection criteria. Thefraction of M-FIR galaxies remain also constant ( ∼ z> SFR TIR detectable for this clusters is larger thanthe value used to select M-FIR galaxies.If we now focus on the clusters, we can see that, de-spite the cluster-to-cluster variations (which reach ∼ UV J -SF and M-FIR samples a trendresembling the Butcher & Oemler (1984) effect, in whichthe fraction of SFGs in clusters is observed to increase withredshift. In this case, the trends are fitted with β =1 . ± . β =7 . ± . UV J -SF and M-FIR samples, re-spectively. The fraction of UV J -SF galaxies within clustersincreases from 28% at z ∼ z ∼ L TIR > × L (cid:12) and R 3% at z =0.02 to ∼ 10% at z =0.3 with β =5.7 +2 . − . .The fraction varies between ∼ 1% at z =0.15 and ∼ 4% at z =0.3 considering only R (cid:46) R . Finally, the contributionof M-FIR galaxies to the whole SFGs population ( UV J -SFsample) remains ∼ 23% in the field, and varies from 0% to19% in the clusters between z ∼ z ∼ MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores z F . . . . . U V J -SF; Clusters ( R / R < < R / R < < R / R < R / R < < R / R < < R / R < U V J -SF; FieldM-FIR; Field z ∝ ( + z ) . ∝ ( + z ) . ∝ ( z ) ∝ ( z ) Figure 9. UV J -SF and M-FIR fractions. In the left panel we consider only cluster members at R /R < R /R < < R /R < < R /R < Best-fitting parameters derived from the fit of the evolution with redshift of the F , and median SFR TOT and s SFR TOT forall the UV J -SF and M-FIR samples in the clusters and in the field. For the clusters we include the results only for R /R < α (1 + z ) β . The units of β are M (cid:12) yr − and yr − in the case of the fit of SFR and s SFR ,respectively. The reduced χ for each case are shown in the last column. The fits of the UV J -SF samples are performed using the datapoints spread out the whole redshift range. In the case of the M-FIR we fit only reported redshift ranges.Quantity Environment Subsample z -range α β χ F Cluster ( R /R < UV J -SF 0.19-0.89 0.25 ± ± ± ± UV J -SF 0.19-0.89 0.56 ± ± ± ± SFR TOT Cluster ( R /R < UV J -SF 0.19-0.89 1.82 ± ± ± ± UV J -SF 0.19-0.57 3.36 ± ± ± ± s SFR TOT Cluster ( R /R < UV J -SF 0.19-0.89 (0.67 ± × − ± ± × − ± UV J -SF 0.19-0.89 (1.24 ± × − ± ± × − ± (2016) reports very little evolution of the ratio of dusty andnon-dusty star-forming galaxies as a function of stellar massthroughout this same redshift range.The average values of F UV J − SF and F M − FIR do notpresent a clear trend with R . In fact, all of them are com-patible with the curve fitted to the fractions at R /R < R /R < The environmental quenching efficiency ( QE env ; van denBosch et al. 2008, Peng et al. 2010, Balogh et al. 2016) is MNRAS000 Cluster ( R /R < UV J -SF 0.19-0.89 (0.67 ± × − ± ± × − ± UV J -SF 0.19-0.89 (1.24 ± × − ± ± × − ± (2016) reports very little evolution of the ratio of dusty andnon-dusty star-forming galaxies as a function of stellar massthroughout this same redshift range.The average values of F UV J − SF and F M − FIR do notpresent a clear trend with R . In fact, all of them are com-patible with the curve fitted to the fractions at R /R < R /R < The environmental quenching efficiency ( QE env ; van denBosch et al. 2008, Peng et al. 2010, Balogh et al. 2016) is MNRAS000 , 1–32 (2018) L. Rodr´ıguez-Mu˜noz et al. defined as QE env = ( F P , cluster − F P , field ) / F SF , field , (11)where F P , cluster and F P , field are the fraction of passive galax-ies in the cluster and field, respectively, and F SF , field is thefraction of SFGs in the field.In Figure 10, we show the QE env in the clus-ter cores ( R /R < 3. Our value of F UV J − SF for clus-ters (field) in the highest redshift bin is 0.50 +0 . − . (0.69 +0 . − . )which leads to smaller values of the passive fraction thantheir 0.88 +0 . − . . Our results at z ∼ z ∼ QE env within cluster-centric dis-tances of 1 Mpc or R , while we focus on the inner clustercore, where the fraction of passive galaxies is expected to belarger.The dependence of the QE env with stellar mass is un-der debate. While some works (e.g., Peng et al. 2010, vander Burg et al. 2018) claim environmental quenching to beindependent of mass quenching, others (e.g., Lin et al. 2014,Kawinwanichakij et al. 2017) have detected an increasingtrend of the QE env with stellar mass. The bottom panel ofFigure 10 shows the values of QE env obtained for galaxies at R < R in two stellar mass bins (10.0 < log M ∗ / M (cid:12) < < log M ∗ / M (cid:12) ). As we can see, only in the first red-shift bin the QE env appears significantly larger for the moremassive galaxies. This QE env appears larger also if we splitthe sample at lower masses, but the significance of the resultdecreases. Darvish et al. (2016) claims that environmentalquenching efficiency is almost independent of stellar mass at z< 1, except for galaxies with log M ∗ / M (cid:12) > SFR and s SFR A complementary quantification of the SF activity in clus-ters tackles the question whether beyond the decrease in F shown in Figure 9, the impact of the cluster environmentmodifies the distribution of the rates at which the remain-ing SFGs form stars. In Figure 11 (top and bottom left-handpanels), we display, as a function of redshift, the median SFR and s SFR of each cluster ( R /R < UV J -SF and M-FIR galaxies. The error bars aredetermined using the bootstrap technique to derive the 1 σ confidence intervals, and thus, they represent the spread inthe SFR and s SFR of each subsample, not the intrinsicerror of the estimation of these parameters ( ∼ < R /R < < R /R < SFR and s SFR with redshift, we again fit the median values (of the individ-ual clusters) using a function of the shape α (1+ z ) β . Regard-ing the M-FIR samples, we only fit those data points corre-sponding to clusters at z< Q E e n v . . . . . Nantais+16Cooke+16Balogh+16 van der Burg+13Quadri+12 R / R < M ∗ /M (cid:12) > z Q E e n v . . . . . R / R < < log M ∗ /M (cid:12) ≤ < log M ∗ /M (cid:12) Figure 10. Top panel : Environmental quenching efficiency forgalaxies with log M ∗ /M (cid:12) > 10 in three z bins. Bottom panel : En-vironmental quenching efficiency calculated for two mass bins(10.0 < log M ∗ /M (cid:12) ≤ M ∗ /M (cid:12) > QE env values given by Quadri et al.(2012), van der Burg et al. (2013), Balogh et al. (2016), Cookeet al. (2016), Nantais et al. (2016). tected in the M- or FIR. Effectively, the fit is performed onlybetween 0.34 SFR and s SFR on average ∼ MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores l og S F R T O T [ M (cid:12) y r − ] . . . . . U V J -SF; Clusters ( R / R < R / R < U V J -SF; FieldM-FIR; Field SF R TIR limit Only z bins U V J -SF; Clusters (0.1 < R / R < U V J -SF; Clusters (0.2 < R / R < z binsM-FIR; Clusters (0.1 < R / R < < R / R < S F R ∝ ( + z ) . S F R ∝ ( + z ) . SF R ∝ ( z ) S F R ∝ ( + z ) . z l og s S F R T O T [ y r − ] - . - . - . - . - . U V J -SF; Clusters ( R / R < R / R < U V J -SF; FieldM-FIR; Field z s S F R ∝ ( + z ) . s S F R ∝ ( + z ) . s S F R ∝ ( + z ) . s S F R ∝ ( + z ) . Figure 11. Top panel : median SFR for the UV J -SF and M-FIR samples. Bottom panel : median s SFR TOT for the UV J -SF andM-FIR samples. Representation as in Figure 9. tween the mass distribution of field and clusters samples (seeSection 8). This can be seen in Figure 5 and Figure 6, wherethe offsets in SFR and s SFR are visible at fixed M ∗ .Figure 11 also displays a clear increasing trend with z of the SFR for both field and cluster UV J -SF sam-ples ( β =2.6 ± β =1.3 ± SFR and s SFR do not show a strong differential evo-lution relative to the field but a systematic offset. Analo-gous trends are found for the s SFR , with β =2.4 ± β =1.2 ± M ∗ distributions driving the variation in s SFR , at leastat log M ∗ /M (cid:12) > 10. A hypothetical impact of the stellarmass distributions of the cluster and field samples wouldtranslate into a different behaviour of the variation of theaverage values of SFR and s SFR with environment, whichis something we do not observe.The high cut in SFR TIR we use to build the M-FIRgalaxy samples translates into a mild increasing trend with z of the median value of the average SFR ( s SFR ) for theM-FIR galaxies in the field, which varies with β =0 . ± . β =0 . ± . MNRAS000 TIR we use to build the M-FIRgalaxy samples translates into a mild increasing trend with z of the median value of the average SFR ( s SFR ) for theM-FIR galaxies in the field, which varies with β =0 . ± . β =0 . ± . MNRAS000 , 1–32 (2018) L. Rodr´ıguez-Mu˜noz et al. values of SFR and s SFR . Also, due to the mentioned SFR TIR constraint we are not able to explore whether theM-FIR samples behave in the same way as the UV J -SFsamples. The M-FIR galaxies with suppressed SF are sim-ply missed by the selection function.A number of works have also identified an offset be-tween the average SFR ( s SFR ) in the clusters and in thefield (e.g., Patel et al. 2009, Vulcani et al. 2010, Haines et al.2015, Haines et al. 2013, Paccagnella et al. 2016). Amongthem, Alberts et al. (2014) find that blue cluster galax-ies ( M ∗ ≥ × M (cid:12) ) present systematically lower aver-age s SFR TIR up to z ∼ Herschel /SPIRE 250 µ m imaging of270 massive galaxy clusters between z ∼ s SFR they re-trieve for clusters at z ∼ z ∼ ∼ -9.70 and ∼ -9.50,respectively) are comparable with ours, as well as their 0.2-0.3 dex differences with the field. This systematic suppres-sion of the level of star-forming activity within rich environ-ments is created by the existence of a numerous populationof transition galaxies located in the lower part of the well-studied MS of SFGs (e.g., Paccagnella et al. 2016, Coendaet al. 2018). Also, Haines et al. (2013) find a 0.2 dex sup-pression of the s SFR in SFGs with log M ∗ /M (cid:12) > 10 and SFR > M (cid:12) yr − within R at 0.15 UV J -SF galaxies at0.2 < R /R < R , where most of the prototypes of galaxiesviolently interacting with the ICM are found (e.g., jellyfishgalaxies, Poggianti et al. 2016; see Boselli & Gavazzi 2006and references therein). The average values of SFR for thecluster M-FIR remain overall compatible with the field val-ues. Instead, the s SFR depart from the field trend at larger R . However, limited number counts of this sample do notallow to extract robust conclusions about this sample. In the previous subsections, we have analysed the SF proper-ties of M ∗ -limited samples of star-forming cluster membersdetected and undetected in the M- and/or FIR. Even thoughwe are able to identify a trend of the SF indices with red-shift, the scatter in the average properties is large. Thesecluster-to-cluster variations have been observed frequentlyin the past, and some works have attempted to quantifythem (e.g., Alberts et al. 2016). This scatter is likely due toa combination of stochastic processes, such as galaxy merg-ers (probably, the limited area covered by our study worsensthis effect), and differences in the properties of the clusters,such as the dynamical state (e.g., Stroe et al. 2015). In thissection, we aim at exploring this latter.Despite the fact of being selected to be largely relaxed,there is disagreement in the literature on the dynamical stateof CLASH sample members (see Rumsey et al. 2016 and ref-erences therein). Given that we are focusing our study on the inner cores of clusters, we use as a proxy of the dynamicalstate of these systems the presence of a CC and the SF ac-tivity undergone by their BCGs. Rawle et al. (2012a) foundthese observables to be strongly correlated, which suggeststhat the SF activity of the BCGs is influenced by the cluster-scale cooling process. In fact, star-forming BCGs seem to beexclusively found in the centers of CC-clusters. However,the separation between cool- and not-cool-core clusters ischallenging. In this work, we use as an indicator of the pres-ence of this feature the parameter C , as defined by Donahueet al. (2016), which is a measure of the concentration of theX-ray emission. More precisely, it gives the ratio betweenthe light within a circular aperture with a 100 kpc radiusand the total light enclosed within a circular aperture witha 500 kpc radius. For CC-clusters, C values are likely > C > C AS1063 =0.19 ± z> UV J -SF and the M-FIR samples ( R /R < F , SFR , and s SFR ) and both the parameter C and the SFR UV of the BCGs extinction corrected ( SFR UVcorr , BCG )provided by Donahue et al. (2015). In order to remove theglobal trends with redshift of the average F , SFR and s SFR that could have an impact on the results, we removethem by normalizing these quantities to the values predictedby the trends fitted for the clusters in the previous subsec-tion at the corresponding redshifts. In each panel of Fig-ure 12, we show the median in three bins of the correspond-ing x-axis parameter populated by the 33% of the clusterssample. Error bars represent the confidence intervals derivedthrough a bootstrap methodology. In the case of the M-FIRsamples, we show with highlighted triangles (black border)the median values of the clusters which contain at least 1object. We use red triangles for the medians calculated con-sidering upper-limits SFR =10 M (cid:12) yr − (our SFR TIR limitfor the M-FIR samples) and s SFR TOT = 3 × − yr − forthose clusters where no M-FIR galaxy is found.If we focus on the upper panels of Figure 12, we see thatthe bins of larger C are marginally dominated by less SFGs.However, the large error bars corresponding to the averageof the M-FIR samples in the first C bin makes the trend notsignificant for this subsamples. In the middle and bottompanels, we do not find a clear correlation between the average SFR or the s SFR and either C or log SFR UVcorr , BCG . 10 DISCUSSION It has long been claimed that galaxies quench more effi-ciently in clusters than in the field (e.g., Butcher & Oemler MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores F / F fi t . . . . UV J -SFM-FIR l og S F R T O T / S F R T O T , fi t - . - . . . . M-FIR (clusters with at least 1 M-FIR galaxy) C l og s S F R T O T / s S F R T O T , fi t - . - . . . . log SFR BCG , UVcorr . [ M (cid:12) yr − ]0.0 0.5 1.0 1.5 2.0 Figure 12. Median values of the F , SFR TOT , and s SFR TOT (top, middle, and bottom panel, respectively) normalized to the valuespredicted by the cluster trends in Figure 9 and 11 at the corresponding redshifts vs the C coefficient given by Donahue et al. (2015)(indicator of the presence of a CC; left hand panels), and the SFR UV of the BCG (right panels; Donahue et al. (2015)) corrected forextinction. The vertical yellow line on the left-hand panels represent the value of C over which the CC are normally located Donahueet al. (2015). We present the averaged values in three equally populated bins of each x-axis parameter. Values derived for the UV J -SFand M-FIR are shown with blue and red symbols, respectively. We use circles to represent the results including all the clusters. In thecase of the clusters where no galaxy was selected in the M-FIR we use an average SFR TOT =10 M (cid:12) yr − (the SFR TIR limit of ourstudy), and an average log s SFR TOT = − . 5. We use triangles to represent the averages found using only clusters with obscured SFactivity in their core (at least 1 M-FIR detected galaxy). ∼ SFR and s SFR do not ap- pear to be the result of differences in the SMF of the galaxysamples studied. Supporting this, Guglielmo et al. (2015)find that galaxies of a given mass have different star forma-tion histories depending on their environment, and therefore,it is not the distributions of galaxy masses in clusters theorigin of the observed dependence of the SF with the envi-ronment. Given that the population of star-forming galaxieswithin massive clusters at the intermediate redshifts probedis thought to be dominated by infalling field galaxies (Kauff-mann 1995), if the quenching of these galaxies were domi-nated by the same processes that turn galaxies off in the field(leading to the global SF decline in the universe since z ∼ MNRAS000 5. We use triangles to represent the averages found using only clusters with obscured SFactivity in their core (at least 1 M-FIR detected galaxy). ∼ SFR and s SFR do not ap- pear to be the result of differences in the SMF of the galaxysamples studied. Supporting this, Guglielmo et al. (2015)find that galaxies of a given mass have different star forma-tion histories depending on their environment, and therefore,it is not the distributions of galaxy masses in clusters theorigin of the observed dependence of the SF with the envi-ronment. Given that the population of star-forming galaxieswithin massive clusters at the intermediate redshifts probedis thought to be dominated by infalling field galaxies (Kauff-mann 1995), if the quenching of these galaxies were domi-nated by the same processes that turn galaxies off in the field(leading to the global SF decline in the universe since z ∼ MNRAS000 , 1–32 (2018) L. Rodr´ıguez-Mu˜noz et al. 2; Madau & Dickinson 2014) the fraction of SFGs shoulddecrease proportionally in both environments (Haines et al.2009). Given the different evolution with redshift we derivefor F UV J − SF in clusters and field, we can say that we arewitnessing the imprint of the impact of environment on theevolution of cluster galaxies ( M ∗ > M (cid:12) ).Our results appear to support the observed evolution ofthe environmental quenching efficiency (van den Bosch et al.2008, Peng et al. 2010, Balogh et al. 2016), defined as thefraction of passive cluster galaxies which would be still star-forming if they were in the field (Nantais et al. 2017), witha major rise since z ∼ strangulation (Larson et al. 1980), whichconsists on the removal of the loosely bound hot halo gasreservoirs by the ICM on long time-scales ( > (cid:46) harassment (Moore et al. 1996).The SFGs infalling into high density environments at z (cid:46) ∼ ∼ overconsumption (McGee et al. 2014), the exhaustionof a gas reservoir through star formation and expulsion viamodest outflows in the absence of cosmological accretion.Maier et al. (2016) also propose it as the explanation for thehigher metallicities found in the accreted cluster galaxies ofMACS0416. It has also been invoked to explain the increas- ing distribution of SFGs with the projected cluster-centricradius (e.g., Alberts et al. 2016, Haines et al. 2015).However, numerous studies have found observationalevidence of rapid quenching mechanisms, such as RPS, thatcan remove the gas of an infalling galaxy in time-scales of theorder of the cluster crossing time ( (cid:46) SFR s evolve un-affected for 2-4 Gyr after infall, and are eventually quenchedrapidly, with an e-folding time of < SFR comparable to those in the field at the same redshift.In addition, Wetzel et al. (2013) propose the quenchingtime-scales do not depend on the halo mass. Interestingly,they claim that up to half of quenched satellites in mas-sive clusters is the result of quenching in infalling groups,namely, pre-processing . Other authors have highlighted theimportance of this phenomenon to explain the propertiesof galaxy populations of intermediate redshift clusters (e.g.,Haines et al. 2015, Ogrean et al. 2015). The cluster-centricdistances we probe in this work ( R /R < UV J -SF cluster core galaxies withlog M ∗ /M (cid:12) > 10 throughout the last 8 Gyr. This is becausethese samples appear to be heavily populated by transi-tion galaxies observed while they quench (Paccagnella et al.2016). However, we cannot rule out the contribution of fastprocesses such as RPS to the enhanced fraction of quenchedgalaxies observed. We also note that our methodology can-not directly select galaxies quenching on short time-scales,such as PSB (e.g., Poggianti et al. 2004, Tran et al. 2007,Muzzin et al. 2014, Paccagnella et al. 2017), as this wouldrequire spectral information, which we lack for more thanhalf of our clusters sample. MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores 11 SUMMARY & CONCLUSIONS We have presented a detailed analysis of the SF activ-ity within 24 massive clusters cores at 0.2 (cid:46) z (cid:46) HST ACS/WFC (F435W,F475W, F606W, F625W, F775W, F814W, and F850LP),WFC3/UVIS (F225W, F275W, F336W, and F390W), andWFC3/IR (F105W, F110W, F125W, F140W, and F160W)imaging. Then, we have combined these catalogues with oth-ers built on Spitzer IRAC (3.6, 4.5, 5.8, and 8.0 µ m) andMIPS (24 µ m) bands, and Herschel PACS (100, and 160 µ m) and SPIRE (250, 350, and 500 µ m), deblending theformer in the position of the CLASH catalogues and se-lecting the most probable UV/optical counterpart for thesources in the rest MIR and FIR bands. Finally, we have alsogathered the spectroscopic information available on thesefields, mainly released by CLASH-VLT and GLASS sur-veys. Consequently, we have derived high quality photomet-ric redshifts ( σ NMAD =0.04, and 8% of outliers) fitting theUV-to-NIR photometry with the EAZY code. We have se-lected cluster members by applying either a spectroscopicredshift criterion or a probabilistic methodology that takesinto account the whole information included in the PDF ofthe photometric redshift estimation. We have used the z phot derived and the Rainbow Cosmological Database softwarepackage to fit, on the one hand, the optical/NIR photometry(CLASH & IRAC), and on the other hand, the FIR photom-etry (MIPS & Herschel ). In this way, we have estimated thephysical properties of the cluster members such as their M ∗ and the rates at which they form stars (as traced by the UVand FIR emission independently). With the aim of buildingup analogous field samples with which compare the resultson clusters, we have applied the same analysis and selectioncriteria on three CANDELS fields. Finally, we have usedsamples of SFGs ( M ∗ > M (cid:12) ) selected using the UVJ-diagram ( UV J -SF samples) to evaluate and compare the SFprocesses in high density environments and the field. Fur-thermore, we have used samples of galaxies ( M ∗ > M (cid:12) )detected in the MIR and/or FIR with SFR TIR > M (cid:12) yr − (M-FIR samples) to explore the obscured SF activity. Tak-ing advantage of the rich dataset available, we have basedour results on the quantification of the total SF, defined aseither the sum of the SF traced by the rest-frame UV emis-sion and the FIR, or the un-obscured SF (traced only bythe rest-frame UV) corrected for the dust extinction withour own optimized recipe.The main results and conclusions of our work can besummarized in the following points: • The SF activity in the inner regions of intermediate- z clusters appears to be suppressed in terms of both thefraction of SFGs and the rate at which they turn gas intostars. • We derive average fractions of UV J -SF galaxies a fac-tor ∼ R /R < R /R < ∼ • We identify increasing trends of F UV J − SF and F M − FIR with z , which evolve faster within clusters ( β =1.1 ± β =7.3 ± R /R < β =0.2 ± β =0.2 ± • UV J -SF cluster members ( R /R < SFR and s SFR typically ∼ UV J -SF fieldgalaxies. Average SFR and s SFR values evolve simi-larly (within the error bars) in clusters, with β =1.3 ± β =1.2 ± β =2.6 ± β =2.4 ± SFR TIR s completeness value given Spitzer /MIPS 24 µ m and Herschel imaging used in thisstudy, we can not explore whether is there a different trendbetween field and clusters dusty SFGs in the average SFR and s SFR . • We find increasing SF activity with cluster-centric dis-tance out to R /R =0.3 in terms of the average SFR and s SFR of the UV J -SF sample. No clear trend is found, how-ever, for the fraction of SFGs. • We do not find an obvious relationship between SF ac-tivity in clusters and the presence of a CC or a BCG formingstars actively.Our results evidence the impact of the cluster environ-ment on the evolution of its inhabitants and favour a dom-inant role of physical processes quenching galaxies slowly.The mechanism typically invoked in these cases is strangu-lation. This process appears to be responsible for the shiftof the average SFR / s SFR exhibited by SFGs in high den-sity environments since z ∼ Rain-bow Cosmological Database. ACKNOWLEDGEMENTS The authors thank Fran¸coise Combes, Carlos L´opez-Sanjuan, Dieter Lutz, Bianca Poggianti and Alvio Renzinifor their suggestions to improve this work. We acknowl-edge funding from the INAF PRIN-SKA 2017 program1.05.01.88.04. L.R.-M. acknowledges funding support fromthe Universit`a degli studi di Padova - Dipartimento di Fisicae Astronomia “G. Galilei”. GR and CM acknowledge sup-port from an INAF PRIN-SKA 2017 grant. P.G.P.-G. ac-knowledges funding support from the Spanish GovernmentMINECO under grants AYA2015-70815-ERC and AYA2015-63650-P. A.C.E. acknowledges support from STFC grantST/P00541/1. A.M. acknowledges funding from the INAFPRIN-SKA 2017 program 1.05.01.88.04. Finally, we thankthe anonymous referee for the valuable comments and sug-gestions, which led to a substantial improvement of thispaper. Analyses were performed in R 3.4.0 (R Core Team2017). MNRAS000 MNRAS000 , 1–32 (2018) L. Rodr´ıguez-Mu˜noz et al. REFERENCES Abadi M. G., Moore B., Bower R. G., 1999, MNRAS, 308, 947Abazajian K. N., et al., 2009, ApJS, 182, 543Abell G. O., 1958, ApJS, 3, 211Abell G. O., Corwin Jr. H. G., Olowin R. P., 1989, ApJS, 70, 1Alberts S., et al., 2013, MNRAS, 431, 194Alberts S., et al., 2014, MNRAS, 437, 437Alberts S., et al., 2016, ApJ, 825, 72Annunziatella M., et al., 2014, A&A, 571, A80Annunziatella M., et al., 2016, A&A, 585, A160Ashby M. L. N., et al., 2013, ApJS, 209, 22Ashby M. L. N., et al., 2015, ApJS, 218, 33Bai L., Rieke G. H., Rieke M. J., Christlein D., Zabludoff A. I.,2009, ApJ, 693, 1840Baldry I. K., Glazebrook K., Brinkmann J., Ivezi´c ˇZ., LuptonR. H., Nichol R. C., Szalay A. S., 2004, ApJ, 600, 681Baldry I. K., Balogh M. L., Bower R. G., Glazebrook K., NicholR. C., Bamford S. P., Budavari T., 2006, MNRAS, 373, 469Baldry I. K., et al., 2012, MNRAS, 421, 621Balestra I., et al., 2016, ApJS, 224, 33Balogh M. L., Christlein D., Zabludoff A. I., Zaritsky D., 2001,ApJ, 557, 117Balogh M. L., et al., 2016, MNRAS, 456, 4364Barger A. J., Cowie L. L., Wang W.-H., 2008, ApJ, 689, 687Barro G., et al., 2011a, ApJS, 193, 13Barro G., et al., 2011b, ApJS, 193, 30Bell E. F., et al., 2004, ApJ, 608, 752Bell E. F., et al., 2005, ApJ, 625, 23Berrier J. C., Stewart K. R., Bullock J. S., Purcell C. W., BartonE. J., Wechsler R. H., 2009, ApJ, 690, 1292Bertin E., Arnouts S., 1996, A&AS, 117, 393Biviano A., Fadda D., Durret F., Edwards L. O. V., Marleau F.,2011, A&A, 532, A77Biviano A., et al., 2013, A&A, 558, A1Blanton M. R., Roweis S., 2007, AJ, 133, 734Boselli A., Gavazzi G., 2006, PASP, 118, 517Boselli A., et al., 2016, A&A, 596, A11Brammer G. B., van Dokkum P. G., Coppi P., 2008, ApJ, 686,1503Brammer G. B., et al., 2011, ApJ, 739, 24Brodwin M., et al., 2011, ApJ, 732, 33Brodwin M., et al., 2013, ApJ, 779, 138Brunner R. J., Lubin L. M., 2000, AJ, 120, 2851Butcher H., Oemler Jr. A., 1978, ApJ, 219, 18Butcher H., Oemler Jr. A., 1984, ApJ, 285, 426Calzetti D., 1997, AJ, 113, 162Calzetti D., Kinney A. L., Storchi-Bergmann T., 1994, ApJ, 429,582Calzetti D., Armus L., Bohlin R. C., Kinney A. L., Koornneef J.,Storchi-Bergmann T., 2000, ApJ, 533, 682Capak P., et al., 2004, AJ, 127, 180Cava A., et al., 2009, A&A, 495, 707Chabrier G., 2003, PASP, 115, 763Chary R., Elbaz D., 2001, ApJ, 556, 562Cimatti A., et al., 2008, A&A, 482, 21Coe D., et al., 2012, ApJ, 757, 22Coenda V., Mart´ınez H. J., Muriel H., 2018, MNRAS, 473, 5617Cohen J. G., Kneib J.-P., 2002, ApJ, 573, 524Connor T., et al., 2017, ApJ, 848, 37Conroy C., Gunn J. E., White M., 2009, ApJ, 699, 486Cooke E. A., et al., 2016, ApJ, 816, 83Cooper M. C., et al., 2011, ApJS, 193, 14Coppin K. E. K., et al., 2011, MNRAS, 416, 680Cowie L. L., Barger A. J., Hu E. M., Capak P., Songaila A., 2004,AJ, 127, 3137Dale D. A., Helou G., 2002, ApJ, 576, 159 Darvish B., Mobasher B., Sobral D., Rettura A., Scoville N.,Faisst A., Capak P., 2016, ApJ, 825, 113Darvish B., Mobasher B., Martin D. C., Sobral D., Scoville N.,Stroe A., Hemmati S., Kartaltepe J., 2017, ApJ, 837, 16De Lucia G., et al., 2007, MNRAS, 374, 809Dickinson M., et al., 2003, Great Observatories Origins Deep Sur-vey (GOODS) Validation Observations, Spitzer ProposalDom´ınguez S´anchez H., et al., 2016, MNRAS, 457, 3743Donahue M., et al., 2015, ApJ, 805, 177Donahue M., et al., 2016, ApJ, 819, 36Draine B. T., Li A., 2007, ApJ, 657, 810Dressler A., 1980, ApJ, 236, 351Dressler A., et al., 1997, ApJ, 490, 577Dressler A., Rigby J., Oemler Jr. A., Fritz J., Poggianti B. M.,Rieke G., Bai L., 2009, ApJ, 693, 140Drory N., et al., 2009, ApJ, 707, 1595Duc P.-A., et al., 2002, A&A, 382, 60Ebeling H., Edge A. C., Henry J. P., 2001, ApJ, 553, 668Ebeling H., Barrett E., Donovan D., Ma C.-J., Edge A. C., vanSpeybroeck L., 2007, ApJ, 661, L33Ebeling H., Edge A. C., Mantz A., Barrett E., Henry J. P., MaC. J., van Speybroeck L., 2010, MNRAS, 407, 83Ebeling H., Ma C.-J., Barrett E., 2014, ApJS, 211, 21Egami E., et al., 2010, A&A, 518, L12Eisenhardt P. R. M., et al., 2008, ApJ, 684, 905Elbaz D., et al., 2011, A&A, 533, A119Evrard A. E., Metzler C. A., Navarro J. F., 1996, ApJ, 469, 494Fabian A. C., 1994, ARA&A, 32, 277Fadda D., Elbaz D., Duc P.-A., Flores H., Franceschini A., Ce-sarsky C. J., Moorwood A. F. M., 2000, A&A, 361, 827Fern´andez-Soto A., Lanzetta K. M., Chen H.-W., Levine B., Ya-hata N., 2002, MNRAS, 330, 889Finn R. A., et al., 2010, ApJ, 720, 87Fioc M., Rocca-Volmerange B., 1997, A&A, 326, 950Geach J. E., et al., 2006, ApJ, 649, 661Geller M. J., Hwang H. S., Diaferio A., Kurtz M. J., Coe D., RinesK. J., 2014, ApJ, 783, 52Gerke B. F., et al., 2007, MNRAS, 376, 1425Giavalisco M., et al., 2004, ApJ, 600, L93Girardi M., et al., 2015, A&A, 579, A4G´omez P. L., et al., 2012, AJ, 144, 79Grazian A., et al., 2006, A&A, 449, 951Griffin M. J., et al., 2010, A&A, 518, L3Grillo C., et al., 2015, ApJ, 800, 38Grogin N. A., et al., 2011, ApJS, 197, 35Guglielmo V., Poggianti B. M., Moretti A., Fritz J., Calvi R.,Vulcani B., Fasano G., Paccagnella A., 2015, MNRAS, 450,2749Gunn J. E., Gott III J. R., 1972, ApJ, 176, 1Guo Y., et al., 2013, ApJS, 207, 24Haines C. P., Gargiulo A., La Barbera F., Mercurio A., MerluzziP., Busarello G., 2007, MNRAS, 381, 7Haines C. P., et al., 2009, ApJ, 704, 126Haines C. P., et al., 2013, ApJ, 775, 126Haines C. P., et al., 2015, ApJ, 806, 101Haines C. P., et al., 2017, A&A, 605, A4Hoaglin D. C., Mosteller F., Tukey J. W., 1983, Understandingrobust and exploratory data anlysisHuang J.-S., et al., 2004, ApJS, 154, 44Huchra J. P., et al., 2012, ApJS, 199, 26Ilbert O., et al., 2009, ApJ, 690, 1236Jablonka P., Combes F., Rines K., Finn R., Welch T., 2013, A&A,557, A103Jouvel S., et al., 2014, A&A, 562, A86Kajisawa M., et al., 2009, ApJ, 702, 1393Karman W., et al., 2015, A&A, 574, A11Kauffmann G., 1995, MNRAS, 274, 153Kauffmann G., et al., 2003, MNRAS, 341, 33MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores Kawinwanichakij L., et al., 2017, ApJ, 847, 134Kennicutt Jr. R. C., 1998, ARA&A, 36, 189Kennicutt R. C., Evans N. J., 2012, ARA&A, 50, 531Kimm T., et al., 2009, MNRAS, 394, 1131Kocevski D. D., et al., 2011, ApJ, 736, 38Koekemoer A. M., et al., 2011, ApJS, 197, 36Larson R. B., Tinsley B. M., Caldwell C. N., 1980, ApJ, 237, 692Le F`evre O., et al., 2003, in Iye M., Moorwood A. F. M., eds, Soci-ety of Photo-Optical Instrumentation Engineers (SPIE) Con-ference Series Vol. 4841, Instrument Design and Performancefor Optical/Infrared Ground-based Telescopes. pp 1670–1681,doi:10.1117/12.460959Le F`evre O., et al., 2004, A&A, 428, 1043Le F`evre O., et al., 2015, A&A, 576, A79Lewis I., et al., 2002, MNRAS, 334, 673Lin L., et al., 2014, ApJ, 782, 33Lotz J. M., et al., 2013, ApJ, 773, 154Lotz J. M., et al., 2017, ApJ, 837, 97Lupton R., Blanton M. R., Fekete G., Hogg D. W., O’MullaneW., Szalay A., Wherry N., 2004, PASP, 116, 133Lutz D., et al., 2011, A&A, 532, A90Madau P., Dickinson M., 2014, ARA&A, 52, 415Magnelli B., et al., 2013, A&A, 553, A132Maier C., et al., 2016, A&A, 590, A108Mann A. W., Ebeling H., 2012, MNRAS, 420, 2120Marcillac D., Rigby J. R., Rieke G. H., Kelly D. M., 2007, ApJ,654, 825Martis N. S., et al., 2016, ApJ, 827, L25McGee S. L., Bower R. G., Balogh M. L., 2014, MNRAS, 442,L105Mercurio A., Girardi M., Boschin W., Merluzzi P., Busarello G.,2003, A&A, 397, 431Meurer G. R., Heckman T. M., Calzetti D., 1999, ApJ, 521, 64Mignoli M., et al., 2005, A&A, 437, 883Molino A., et al., 2017, MNRAS, 470, 95Moore B., Katz N., Lake G., Dressler A., Oemler A., 1996, Nature,379, 613Mortlock A., et al., 2015, MNRAS, 447, 2Moutard T., Sawicki M., Arnouts S., Golob A., Malavasi N.,Adami C., Coupon J., Ilbert O., 2018, MNRAS, 479, 2147Muzzin A., et al., 2014, ApJ, 796, 65Nantais J. B., et al., 2016, A&A, 592, A161Nantais J. B., et al., 2017, MNRAS, 465, L104Nayyeri H., et al., 2017, ApJS, 228, 7Newman A. B., Treu T., Ellis R. S., Sand D. J., Nipoti C., RichardJ., Jullo E., 2013, ApJ, 765, 24Noeske K. G., et al., 2007, ApJ, 660, L43Ogrean G. A., et al., 2015, ApJ, 812, 153Oliver S. J., et al., 2012, MNRAS, 424, 1614Paccagnella A., et al., 2016, ApJ, 816, L25Paccagnella A., et al., 2017, ApJ, 838, 148Pacifici C., Charlot S., Blaizot J., Brinchmann J., 2012, MNRAS,421, 2002Papovich C., et al., 2018, ApJ, 854, 30Patel S. G., Holden B. P., Kelson D. D., Illingworth G. D., FranxM., 2009, ApJ, 705, L67Pell´o R., et al., 2009, A&A, 508, 1173Peng Y.-j., et al., 2010, ApJ, 721, 193P´erez-Gonz´alez P. G., et al., 2005, ApJ, 630, 82P´erez-Gonz´alez P. G., et al., 2008, ApJ, 675, 234P´erez-Gonz´alez P. G., et al., 2010, A&A, 518, L15P´erez-Gonz´alez P. G., et al., 2013, ApJ, 762, 46Pilbratt G. L., et al., 2010, A&A, 518, L1Poggianti B. M., 2003, Ap&SS, 285, 121Poggianti B. M., Smail I., Dressler A., Couch W. J., Barger A. J.,Butcher H., Ellis R. S., Oemler Jr. A., 1999, ApJ, 518, 576Poggianti B. M., Bridges T. J., Komiyama Y., Yagi M., CarterD., Mobasher B., Okamura S., Kashikawa N., 2004, ApJ, 601, 197Poggianti B. M., et al., 2009, ApJ, 693, 112Poggianti B. M., et al., 2016, AJ, 151, 78Poggianti B. M., et al., 2017, ApJ, 844, 48Poglitsch A., et al., 2010, A&A, 518, L2Polletta M., et al., 2007, ApJ, 663, 81Popesso P., et al., 2011, A&A, 532, A145Popping G., Puglisi A., Norman C. A., 2017, MNRAS, 472, 2315Postman M., et al., 2012, ApJS, 199, 25Quadri R. F., Williams R. J., Franx M., Hildebrandt H., 2012,ApJ, 744, 88R Core Team 2017, R: A Language and Environment for Sta-tistical Computing. R Foundation for Statistical Computing,Vienna, Austria, Ravindranath S., Ho L. C., 2002, ApJ, 577, 133Rawle T. D., et al., 2010, A&A, 518, L14Rawle T. D., et al., 2012a, ApJ, 747, 29Rawle T. D., et al., 2012b, ApJ, 756, 106Rawle T. D., et al., 2014, MNRAS, 442, 196Rawle T. D., et al., 2016, MNRAS, 459, 1626Renzini A., Peng Y.-j., 2015, ApJ, 801, L29Rex M., et al., 2010, A&A, 518, L13Rieke G. H., Alonso-Herrero A., Weiner B. J., P´erez-Gonz´alezP. G., Blaylock M., Donley J. L., Marcillac D., 2009, ApJ,692, 556Rodighiero G., et al., 2011, ApJ, 739, L40Rosati P., et al., 2014, The Messenger, 158, 48Rumsey C., et al., 2016, MNRAS, 460, 569Saintonge A., Tran K.-V. H., Holden B. P., 2008, ApJ, 685, L113Salpeter E. E., 1955, ApJ, 121, 161Sanders D. B., et al., 2007, ApJS, 172, 86Santos J. S., Rosati P., Tozzi P., B¨ohringer H., Ettori S., Big-namini A., 2008, A&A, 483, 35Santos J. S., et al., 2015, MNRAS, 447, L65Sargent M. T., B´ethermin M., Daddi E., Elbaz D., 2012, ApJ,747, L31Schechter P., 1976, ApJ, 203, 297Schmidt K. B., et al., 2014, ApJ, 782, L36Schreiber C., Pannella M., Leiton R., Elbaz D., Wang T., Oku-mura K., Labb´e I., 2017, A&A, 599, A134Scoville N., et al., 2007, ApJS, 172, 1Shectman S. A., Landy S. D., Oemler A., Tucker D. L., Lin H.,Kirshner R. P., Schechter P. L., 1996, ApJ, 470, 172Sheth R. K., Rossi G., 2010, MNRAS, 403, 2137Smith R. J., et al., 2010, MNRAS, 408, 1417Springel V., et al., 2005, Nature, 435, 629Steinhauser D., Schindler S., Springel V., 2016, A&A, 591, A51Stroe A., et al., 2015, MNRAS, 450, 646Sunyaev R. A., Zel’dovich Y. B., 1972, Comments on Astrophysicsand Space Physics, 4, 173Szokoly G. P., et al., 2004, ApJS, 155, 271Tran K.-V. H., Franx M., Illingworth G. D., van Dokkum P.,Kelson D. D., Blakeslee J. P., Postman M., 2007, ApJ, 661,750Treu T., et al., 2015, ApJ, 812, 114Tyler K. D., Rieke G. H., Bai L., 2013, ApJ, 773, 86Umetsu K., et al., 2014, ApJ, 795, 163Vanzella E., et al., 2008, A&A, 478, 83Vulcani B., Poggianti B. M., Finn R. A., Rudnick G., Desai V.,Bamford S., 2010, ApJ, 710, L1Vulcani B., et al., 2011, MNRAS, 412, 246Vulcani B., et al., 2012, MNRAS, 420, 1481Vulcani B., et al., 2013, A&A, 550, A58Vulcani B., et al., 2016, ApJ, 833, 178Vulcani B., et al., 2017, ApJ, 837, 126Wang W.-H., Cowie L. L., Barger A. J., Keenan R. C., Ting H.-C.,2010, ApJS, 187, 251MNRAS , 1–32 (2018) L. Rodr´ıguez-Mu˜noz et al. Wetzel A. R., Tinker J. L., Conroy C., van den Bosch F. C., 2013,MNRAS, 432, 336Whitaker K. E., Kriek M., van Dokkum P. G., Bezanson R.,Brammer G., Franx M., Labb´e I., 2012a, ApJ, 745, 179Whitaker K. E., van Dokkum P. G., Brammer G., Franx M.,2012b, ApJ, 754, L29Whitaker K. E., et al., 2015, ApJ, 811, L12Wijesinghe D. B., et al., 2012, MNRAS, 423, 3679Williams R. J., Quadri R. F., Franx M., van Dokkum P., Labb´eI., 2009, ApJ, 691, 1879Wirth G. D., et al., 2004, AJ, 127, 3121Wittman D., Bhaskar R., Tobin R., 2016, MNRAS, 457, 4005Wuyts S., et al., 2007, ApJ, 655, 51Wuyts S., Labb´e I., F¨orster Schreiber N. M., Franx M., RudnickG., Brammer G. B., van Dokkum P. G., 2008, ApJ, 682, 985Zitrin A., et al., 2013, ApJ, 762, L30Zitrin A., et al., 2015, ApJ, 801, 44van den Bosch F. C., Aquino D., Yang X., Mo H. J., Pasquali A.,McIntosh D. H., Weinmann S. M., Kang X., 2008, MNRAS,387, 79van der Burg R. F. J., et al., 2013, A&A, 557, A15van der Burg R. F. J., McGee S., Aussel H., Dahle H., Arnaud M.,Pratt G. W., Muzzin A., 2018, preprint, ( arXiv:1807.00820 ) APPENDIX A: DATA AVAILABLE ON THECANDELS FIELDS In the following subsections we briefly enumerate the photo-metric and spectroscopic data on the CANDELS fields whichis used in our analysis. A1 GOODS-S We use the multi-wavelength catalogue on theCANDELS/GOODS-S field published by Guo et al.(2013), which combines the CANDELS HST /WFC3F105W, F125W, and F160W bands with data from UV( U band from both CTIO/MOSAIC and VLT/VIMOS),optical ( HST /ACS F435W, F606W, F775W, F814W, andF850LP), and infrared ( HST /WFC3 F098M, VLT/ISAAC Ks , VLT/HAWK-I Ks , and Spitzer /IRAC 3.6, 4.5, 5.8,8.0 µ m) observations. The catalogue is based on sourcedetection in the WFC3 F160W band. Applying themethodology described in Section 3 we complement thecatalogue with MIR photometry in Spitzer /MIPS 24 µ mand 70 µ m from (P´erez-Gonz´alez et al. 2008) and FIRphotometry from the GOODS-Herschel (Elbaz et al. 2011)and PEP (Magnelli et al. 2013) surveys, including PACS100 and 160 µ m, and SPIRE 250, 350, and 500 µ m. Thespectroscopic data are gathered from the VIMOS VLT deepsurvey (Le F`evre et al. 2004), Szokoly et al. (2004), theK20 survey (Mignoli et al. 2005), and other surveys such asthose carried out by (e.g.) Cimatti et al. (2008), Vanzellaet al. (2008). See Guo et al. (2013) for the details. A2 GOODS-N The multi-wavelength catalogue used onCANDELS/GOODS-N is built and described by Barro etal. (in prep.) and includes UV to far IR and radio data.In particular, UV data from GALEX (PI C. Martin),ground-based optical data from U to z bands taken by the Kitt Peak telescope and from the Subaru /Suprime-Cam as part of the Hawaii Hubble Deep Field North project(Capak et al. 2004), 25 medium-bands from the GTCSHARDS (P´erez-Gonz´alez et al. 2013) survey, J , H , and Ks imaging from the Subaru MOIRCS deep survey (Ka-jisawa et al. 2009) and CFHT/WIRCam Ks photometry(Lin in prep.); IRAC 3.6, 4.5, 5.8, 8.0 µ m maps, maps fromSpitzer-GOODS (Dickinson et al. 2003), SEDS (Ashbyet al. 2013) and SCANDELS (Ashby et al. 2015); MIPSdata from FIDEL (PI: M. Dickinson); Herschel from theGOODS-Herschel (Elbaz et al. 2011) and PEP (Magnelliet al. 2013) surveys, including PACS 100 and 160 µ m, andSPIRE 250, 350, and 500 µ m. The spectroscopic redshiftsused are a compilation based primarily on ACS-GOODSredshift survey (Cowie et al. 2004; Barger et al. 2008), theTeam Keck Redshift Survey (Wirth et al. 2004), and theDEEP3 galaxy redshift survey (Cooper et al. 2011). A3 COSMOS We use the multi-wavelength catalogue on the CAN-DELS/COSMOS field published by Nayyeri et al. (2017),which combines the CANDELS HST /WFC3 F105W,F125W, and F160W bands with data from HST /ACSF606W and F814W, CFHT/MegaPrime in the u ∗ , g ∗ , r ∗ , i ∗ , and z ∗ bands, from the Subaru /Suprime-Cam in the B , g + , V , r + , i + and z + , along with twelve intermediate andtwo narrow bands ( ∼ Y , J , H and Ks bands, Mayall /NEWFIRM J , J , J , H , H , K , and Spitzer /IRAC 3.6, 4.5, 5.8, 8.0 µ m bands.Again, we combine this catalogue with MIR photometry in Spitzer /MIPS 24 µ m and 70 µ m from Sanders et al. (2007)and FIR photometry including PACS 100 and 160 µ m fromPEP program (Lutz et al. 2011), and SPIRE 250, 350, and500 µ m from HerMES (Oliver et al. 2012). Among the spec-troscopic surveys gathered we highlight the VIMOS UltraDeep Survey (Le F`evre et al. 2015), zCOSMOS (PI: S. Lilly). APPENDIX B: UV CORRECTION The ratio of the L TIR to L UV , usually referred as IRX , istightly related to the dust attenuation in a galaxy. This is be-cause dust absorbs and scatters mainly UV photons obscur-ing and reddening the galaxy SED at wavelengths (cid:46) µ m.Then, it re-emits the absorbed energy in the IR, at wave-lengths ∼ µ m. Since the work of Meurer et al. (1999)on local starburst galaxies (i.e., extreme SFGs), the rela-tion between the IRX and the slope of the UV ( β ) hasbeen frequently used to estimate the UV dust attenuationof galaxies. In practice, this relation is calibrated for localblue galaxies for which FIR observations is available (e.g.,Calzetti 1997, Meurer et al. 1999) and then, it is used tocorrect the UV luminosity from extinction up to high red-shifts (Meurer et al. 1999). However, important deviationsfrom these relations have been observed (e.g.) for galaxiesforming stars at a lower rates or at different redshifts. Lately,different studies have explored in detailed the physical ori-gin of variations in the IRX - β relation (e.g., Popping et al.2017). In this context, we aim at deriving an optimized dustattenuation correction (i.e. IRX - β relation) that we can ap-ply to those star-forming cluster members fainter than ourobservational limits in MIPS and/or Herschel, and therefore MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores z l og S F R [ M (cid:12) y r − ] - . - . . . . . . SFR TIR SFR UV β I R X -3 -2 -1 0 1 2 AllCalibration sampleMeurer et al. 1999 & Calzetti et al. 2000This work -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 - . - . . . . . . log SFR TIR + SFR UV [ M (cid:12) yr − ] l og S F R [ M (cid:12) y r − ] Cluster members SFR TIR < M (cid:12) yr − SFR UV , corr SFR UV Figure 1. Left panel : SFR TIR (grey contours) and SFR UV (orange contours) versus redshift for all the 1548 M/FIR-detected galaxiesin CANDELS fields with UVJ colours corresponding to SFGs, log M ∗ /M (cid:12) > 10, and 0.1 < z < Central panel : IRX - β relation for the galaxies in CANDELS fields with UVJ colourscorresponding to SFGs, log M ∗ /M (cid:12) > 10, and 0.1 < z < SFR TIR < M (cid:12) yr − of whichthe calibration sample is made of (blue contours). We represent our calibration with a blue line. The black line is the IRX - β fit fromMeurer et al. (1999) modified with a Calzetti et al. (2000) extinction law to the UV wavelength we consider in our study (2800˚A). Right panel : Comparison between SFR TIR + SFR UV and the SFR UV , corr . corrected for dust extinction using our own calibration(Equation B1, blue contours). For comparison we show the distribution of values of SFR UV previous to the dust extinction correction(orange contours). To evaluate the behaviour of our UV correction in the clusters, we represent the comparison between the SFR TOT and the SFR UV , corr . of the cluster members with SFR TIR < M (cid:12) yr − . presumably less star-forming than the starbursts on whichthe calibrations in the literature are defined.Following a similar approach to Dom´ınguez S´anchezet al. (2016), we basically derive a IRX - β relation for a sam-ple of SFGs which are faint M/FIR emitters. In particular,we take advantage of the deep coverage on CANDELS fields(GOODS and COSMOS) to select a subsample of SFGsfainter than the CLASH+HLS fields observational limitsin MIPS and/or Herschel bands. We only consider galax-ies classified as SFGs using an UVJ-diagram, located in theredshift range between 0.1 and 1.0, and with M ∗ /M (cid:12) > 10. InFigure 1 (left panel) we display the distribution with redshiftof SFR TIR and SFR UV of these galaxies (obtained follow-ing Equation 5 and 6, respectively). The calibration sampleincludes the 1548 galaxies with SFR TIR < M (cid:12) yr − (greenhorizontal line).Once the sample is defined, we compute the UVslope for each galaxy using a linear interpolation between1500 ˚A and 2800 ˚A in the best-fit templates given by Rain-bow (Section 5). The typical uncertainty in the β values is ∼ IRX as the ratio of their SFR TIR and SFR UV . In Figure 1 (central panel) we displaythe IRX - β space for the whole field sample of M/FIR emit-ters ( M ∗ /M (cid:12) > 10 and 0.2 IRX - β plane for our calibra-tion sample with a linear function. We derive the followingbest fit expression:A UV = (1 . ± . 04) + (0 . ± . β (B1)Again, following the approach by Dom´ınguez S´anchez et al.(2016), we apply the Meurer et al. 1999 IRX - β relation ( A =4.43 + 1.99 β ) for β values lower than the point inwhich our fit intercepts the relation by Meurer et al. (1999), β =-1.7, and Equation B1 for higher β values.To assess the efficiency of our calibration, we quantifythe scatter of the difference between the SFR TOT derivedas the addition of SFR TIR and SFR UV , and the SFR TOT computed as the SFR UV corrected for dust extinction forour calibration sample (right panel in Figure 1). The valuesvary between -0.38 and 0.26 dex with a median of -0.02 dex.Using the calibration by Meurer et al. (1999) instead wouldhave lead to a median absolute deviation of 0.53 dex. Giventhat we use the calibration built on field galaxies to correctalso the SFR UV of the cluster members not detected inthe M/FIR, we compare how the calibration behaves forthose faint M/FIR cluster members ( SFR TIR < M (cid:12) yr − ).In the right panel of Figure 1, we see that the dust extinctioncorrection behaves similarly in the field and the clusters. Forthe latter, the median absolute deviation is -0.05 dex, andthe differences vary between -0.54 and 0.23 dex. APPENDIX C: CATALOGUES This appendix details the entries of the catalogues released. This paper has been typeset from a TEX/L A TEX file prepared bythe author.MNRAS000 This paper has been typeset from a TEX/L A TEX file prepared bythe author.MNRAS000 , 1–32 (2018) L. Rodr´ıguez-Mu˜noz et al. Table C1. Multiwavelength photometryEntry name Descriptionobject ID of the source in the parent catalogue. This ID is not the CLASH catalogue ID.flux [ µ Jy]err flux [ µ Jy] Table C2. Flags for the MIPS counterpart identification.Entry name Descriptionobject ID of the source in the parent catalogue.MIPS n counterparts Total number of (selection band) counterparts candidates for the MIPS24 source.MIPS ID order ID of the MIPS24 counterpart flagged with the likelihood.The most probable counterpart is flagged with a ‘ 1’.MIPS order The order of likelihood of being the right counterpart of the MIPS source.MIPS discriminator Quantity used to determine the counterpart likelihood order.MIPS fMIPS24 MIPS24 flux [ µ Jy] used for the MIPS24 counterpart identification.MIPS err fMIPS24 MIPS24 flux error [ µ Jy] used for the MIPS24 counterpart identification.MIPS fIRAC80 IRAC80 flux [ µ Jy] used for the MIPS24 counterpart identification.MIPS err fIRAC80 IRAC80 flux error [ µ Jy] used for the MIPS24 counterpart identification.MIPS fIRAC36 IRAC36 flux [ µ Jy] used for the MIPS24 counterpart identification.MIPS err fIRAC36 IRAC36 flux error [ µ Jy] used for the MIPS24 counterpart identification.MIPS distance Distance between the MIPS24 source and the counterpart candidate.MIPS24 snr cuts Flag regarding the SNR cuts applied in MIPS24:0 no-flux, 1 flux > SNR limit, -1 flux < SNR limit.n MIPS24 psf0.25/0.5/1/2 Number of sources in the parent catalogue.n MIPS24 wcs0.25/0.5/1/2 Number of sources in the parent catalogue.n MIPS MIPS24 psf0.25/0.5/1/2 Number of MIPS sources within the MIPS24 PSF.n MIPS MIPS24 wcs0.25/0.5/1/2 Number of MIPS sources within the MIPS24 WCS accuracy.n IRAC MIPS24 psf0.25/0.5/1/2 Number of IRAC sources within the MIPS24 PSF.n IRAC MIPS24 wcs0.25/0.5/1/2 Number of IRAC sources within the MIPS24 WCS accuracy. MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores Table C3. Flags for the PACS counterpart identification.Entry name Descriptionobject ID of the source in the parent catalogue.PACS ID order ID of the PACS counterpart flagged with the likelihood.The most probable counterpart is flaged with a ‘ 1’.PACS discriminator Quantity used to determine the counterpart likelihood order.PACS fPACS160 PACS160 flux [ µ Jy] used for the PACS counterpart identification.PACS err fPACS160 PACS160 flux error [ µ Jy] used for the PACS counterpart identification.PACS fPACS100 PACS100 flux [ µ Jy] used for the PACS counterpart identification.PACS err fPACS100 PACS100 flux error [ µ Jy] used for the PACS counterpart identification.PACS fMIPS24 MIPS24 flux [ µ Jy] used for the PACS counterpart identification.PACS err fMIPS24 MIPS24 flux error [ µ Jy] used for the PACS counterpart identification.PACS fIRAC80 IRAC80 flux [ µ Jy] used for the PACS counterpart identification.PACS err fIRAC80 IRAC80 flux error [ µ Jy] used for the PACS counterpart identification.PACS fIRAC36 IRAC36 flux [ µ Jy] used for the PACS counterpart identification.PACS err fIRAC36 IRAC36 flux error [ µ Jy] used for the PACS counterpart identification.PACS distance Distance between the PACS and the counterpart candidate.PACS order The order of likelihood of being the right counterpart of the PACS source.PACS n counterparts Total number of counterparts candidates for the PACS source.PACS100 snr cuts Flag regarding the SNR cuts applied in PACS100:0 no-flux, 1 flux > SNR limit, -1 flux < SNR limit.PACS160 snr cuts Flag regarding the SNR cuts applied in PACS160:0 no-flux, 1 flux > SNR limit, -1 flux < SNR limit.n PACS100 psf0.25/0.5/1/2 Number of sources in the parent catalogue within the PACS100 PSF.n PACS160 psf0.25/0.5/1/2 Number of sources in the parent catalogue within the PACS160 PSF.n PACS100 wcs0.25/0.5/1/2 Number of sources in the parent catalogue within the PACS100 WCS accuracy.n PACS160 wcs0.25/0.5/1/2 Number of sources in the parent catalogue within the PACS160 WCS accuracy.n PACS PACS100 psf0.25/0.5/1/2 Number of PACS sources within the PACS100 PSF.n PACS PACS160 psf0.25/0.5/1/2 Number of PACS sources within the PACS160 PSF.n PACS PACS100 wcs0.25/0.5/1/2 Number of PACS sources within the PACS100 WCS accuracy.n PACS PACS160 wcs0.25/0.5/1/2 Number of PACS sources within the PACS160 WCS accuracy.n MIPS PACS100 psf0.25/0.5/1/2 Number of MIPS sources within the PACS100 PSF.n MIPS PACS160 psf0.25/0.5/1/2 Number of MIPS sources within the PACS160 PSF.n MIPS PACS100 wcs0.25/0.5/1/2 Number of MIPS sources within the PACS100 WCS accuracy.n MIPS PACS160 wcs0.25/0.5/1/2 Number of MIPS sources within the PACS160 WCS accuracy.n IRAC PACS100 psf0.25/0.5/1/2 Number of IRAC sources within the PACS100 PSF.n IRAC PACS160 psf0.25/0.5/1/2 Number of IRAC sources within the PACS160 PSF.n IRAC PACS100 wcs0.25/0.5/1/2 Number of IRAC sources within the PACS100 WCS accuracy.n IRAC PACS160 wcs0.25/0.5/1/2 Number of IRAC sources within the PACS160 WCS accuracy.MNRAS000 Flags for the PACS counterpart identification.Entry name Descriptionobject ID of the source in the parent catalogue.PACS ID order ID of the PACS counterpart flagged with the likelihood.The most probable counterpart is flaged with a ‘ 1’.PACS discriminator Quantity used to determine the counterpart likelihood order.PACS fPACS160 PACS160 flux [ µ Jy] used for the PACS counterpart identification.PACS err fPACS160 PACS160 flux error [ µ Jy] used for the PACS counterpart identification.PACS fPACS100 PACS100 flux [ µ Jy] used for the PACS counterpart identification.PACS err fPACS100 PACS100 flux error [ µ Jy] used for the PACS counterpart identification.PACS fMIPS24 MIPS24 flux [ µ Jy] used for the PACS counterpart identification.PACS err fMIPS24 MIPS24 flux error [ µ Jy] used for the PACS counterpart identification.PACS fIRAC80 IRAC80 flux [ µ Jy] used for the PACS counterpart identification.PACS err fIRAC80 IRAC80 flux error [ µ Jy] used for the PACS counterpart identification.PACS fIRAC36 IRAC36 flux [ µ Jy] used for the PACS counterpart identification.PACS err fIRAC36 IRAC36 flux error [ µ Jy] used for the PACS counterpart identification.PACS distance Distance between the PACS and the counterpart candidate.PACS order The order of likelihood of being the right counterpart of the PACS source.PACS n counterparts Total number of counterparts candidates for the PACS source.PACS100 snr cuts Flag regarding the SNR cuts applied in PACS100:0 no-flux, 1 flux > SNR limit, -1 flux < SNR limit.PACS160 snr cuts Flag regarding the SNR cuts applied in PACS160:0 no-flux, 1 flux > SNR limit, -1 flux < SNR limit.n PACS100 psf0.25/0.5/1/2 Number of sources in the parent catalogue within the PACS100 PSF.n PACS160 psf0.25/0.5/1/2 Number of sources in the parent catalogue within the PACS160 PSF.n PACS100 wcs0.25/0.5/1/2 Number of sources in the parent catalogue within the PACS100 WCS accuracy.n PACS160 wcs0.25/0.5/1/2 Number of sources in the parent catalogue within the PACS160 WCS accuracy.n PACS PACS100 psf0.25/0.5/1/2 Number of PACS sources within the PACS100 PSF.n PACS PACS160 psf0.25/0.5/1/2 Number of PACS sources within the PACS160 PSF.n PACS PACS100 wcs0.25/0.5/1/2 Number of PACS sources within the PACS100 WCS accuracy.n PACS PACS160 wcs0.25/0.5/1/2 Number of PACS sources within the PACS160 WCS accuracy.n MIPS PACS100 psf0.25/0.5/1/2 Number of MIPS sources within the PACS100 PSF.n MIPS PACS160 psf0.25/0.5/1/2 Number of MIPS sources within the PACS160 PSF.n MIPS PACS100 wcs0.25/0.5/1/2 Number of MIPS sources within the PACS100 WCS accuracy.n MIPS PACS160 wcs0.25/0.5/1/2 Number of MIPS sources within the PACS160 WCS accuracy.n IRAC PACS100 psf0.25/0.5/1/2 Number of IRAC sources within the PACS100 PSF.n IRAC PACS160 psf0.25/0.5/1/2 Number of IRAC sources within the PACS160 PSF.n IRAC PACS100 wcs0.25/0.5/1/2 Number of IRAC sources within the PACS100 WCS accuracy.n IRAC PACS160 wcs0.25/0.5/1/2 Number of IRAC sources within the PACS160 WCS accuracy.MNRAS000 , 1–32 (2018) L. Rodr´ıguez-Mu˜noz et al. Table C4. Flags for the SPIRE counterpart identification.Entry name Descriptionobject ID of the source in the parent catalogue.SPIRE ID order ID of the SPIRE counterpart flagged with the likelihood.The most probable counterpart is flaged with a ‘ 1’.SPIRE discriminator Quantity used to determine the counterpart likelihood order.SPIRE fSPIRE500 SPIRE500 flux [ µ Jy] used for the SPIRE counterpart identification.SPIRE err fSPIRE500 SPIRE500 flux error [ µ Jy] used for the SPIRE counterpart identification.SPIRE fSPIRE350 SPIRE350 flux [ µ Jy] used for the SPIRE counterpart identification.SPIRE err fSPIRE350 SPIRE350 flux error [ µ Jy] used for the SPIRE counterpart identification.SPIRE fSPIRE250 SPIRE250 flux [ µ Jy] used for the SPIRE counterpart identification.SPIRE err fSPIRE250 SPIRE250 flux error [ µ Jy] used for the SPIRE counterpart identification.SPIRE fPACS160 PACS160 flux [ µ Jy] used for the SPIRE counterpart identification.SPIRE err fPACS160 PACS160 flux error [ µ Jy] used for the SPIRE counterpart identification.SPIRE fPACS100 PACS100 flux [ µ Jy] used for the SPIRE counterpart identification.SPIRE err fPACS100 PACS100 flux error [ µ Jy] used for the SPIRE counterpart identification.SPIRE fMIPS24 MIPS24 flux [ µ Jy] used for the SPIRE counterpart identification.SPIRE err fMIPS24 MIPS24 flux error [ µ Jy] used for the SPIRE counterpart identification.SPIRE fIRAC80 IRAC80 flux [ µ Jy] used for the SPIRE counterpart identification.SPIRE err fIRAC80 IRAC80 flux error [ µ Jy] used for the SPIRE counterpart identification.SPIRE fIRAC36 IRAC36 flux [ µ Jy] used for the SPIRE counterpart identification.SPIRE err fIRAC36 IRAC36 flux error [ µ Jy] used for the SPIRE counterpart identification.SPIRE distance Distance between the SPIRE and the counterpart candidate.SPIRE order The order of likelihood of being the right counterpart of the SPIRE source.SPIRE n counterparts Total number of counterparts candidates for the SPIRE source.SPIRE250 snr cuts Flag regarding the SNR cuts applied in SPIRE250:0 no-flux, 1 flux > SNR limit, -1 flux < SNR limit.SPIRE350 snr cuts Flag regarding the SNR cuts applied in SPIRE350:0 no-flux, 1 flux > SNR limit, -1 flux < SNR limit.SPIRE500 snr cuts Flag regarding the SNR cuts applied in SPIRE500:0 no-flux, 1 flux > SNR limit, -1 flux < SNR limit.n SPIRE250 psf0.25/0.5/1/2 Number of sources in the parent catalogue within the SPIRE250 PSF.n SPIRE350 psf0.25/0.5/1/2 Number of sources in the parent catalogue within the SPIRE350 PSF.n SPIRE500 psf0.25/0.5/1/2 Number of sources in the parent catalogue within the SPIRE500 PSF.n SPIRE250 wcs0.25/0.5/1/2 Number of sources in the parent catalogue within the SPIRE250 WCS accuracy.n SPIRE350 wcs0.25/0.5/1/2 Number of sources in the parent catalogue within the SPIRE350 WCS accuracy.n SPIRE500 wcs0.25/0.5/1/2 Number of sources in the parent catalogue within the SPIRE500 WCS accuracy.n SPIRE SPIRE250 psf0.25/0.5/1/2 Number of SPIRE sources within the SPIRE250 PSF.n SPIRE SPIRE350 psf0.25/0.5/1/2 Number of SPIRE sources within the SPIRE350 PSF.n SPIRE SPIRE500 psf0.25/0.5/1/2 Number of SPIRE sources within the SPIRE500 PSF.n SPIRE SPIRE250 wcs0.25/0.5/1/2 Number of SPIRE sources within the SPIRE250 WCS accuracy.n SPIRE SPIRE350 wcs0.25/0.5/1/2 Number of SPIRE sources within the SPIRE350 WCS accuracy.n SPIRE SPIRE500 wcs0.25/0.5/1/2 Number of SPIRE sources within the SPIRE500 WCS accuracy.n PACS SPIRE250 psf0.25/0.5/1/2 Number of PACS sources within the SPIRE250 PSF.n PACS SPIRE350 psf0.25/0.5/1/2 Number of PACS sources within the SPIRE350 PSF.n PACS SPIRE500 psf0.25/0.5/1/2 Number of PACS sources within the SPIRE500 PSF.n PACS SPIRE250 wcs0.25/0.5/1/2 Number of PACS sources within the SPIRE250 WCS accuracy.n PACS SPIRE350 wcs0.25/0.5/1/2 Number of PACS sources within the SPIRE350 WCS accuracy.n PACS SPIRE500 wcs0.25/0.5/1/2 Number of PACS sources within the SPIRE500 WCS accuracy.n MIPS SPIRE250 psf0.25/0.5/1/2 Number of MIPS sources within the SPIRE250 PSF.n MIPS SPIRE350 psf0.25/0.5/1/2 Number of MIPS sources within the SPIRE350 PSF.n MIPS SPIRE500 psf0.25/0.5/1/2 Number of MIPS sources within the SPIRE500 PSF.n MIPS SPIRE250 wcs0.25/0.5/1/2 Number of MIPS sources within the SPIRE250 WCS accuracy.n MIPS SPIRE350 wcs0.25/0.5/1/2 Number of MIPS sources within the SPIRE350 WCS accuracy.n MIPS SPIRE500 wcs0.25/0.5/1/2 Number of MIPS sources within the SPIRE500 WCS accuracy.n IRAC SPIRE250 psf0.25/0.5/1/2 Number of IRAC sources within the SPIRE250 PSF.n IRAC SPIRE350 psf0.25/0.5/1/2 Number of IRAC sources within the SPIRE350 PSF.n IRAC SPIRE500 psf0.25/0.5/1/2 Number of IRAC sources within the SPIRE500 PSF.n IRAC SPIRE250 wcs0.25/0.5/1/2 Number of IRAC sources within the SPIRE250 WCS accuracy.n IRAC SPIRE350 wcs0.25/0.5/1/2 Number of IRAC sources within the SPIRE350 WCS accuracy.n IRAC SPIRE500 wcs0.25/0.5/1/2 Number of IRAC sources within the SPIRE500 WCS accuracy.MNRAS , 1–32 (2018) Un)-obscured star formation in cluster cores Table C5. Redshift and propertiesEntry name Descriptionobject ID of the galaxy in the parent catalogue.z phot EAZY z phot .z spec Spectroscopic redshift.flag Quality of the z spec . Values > (cid:12) .L TIR Total IR luminosity (8-1000 µ m) in L (cid:12) , from the best-fit template (Draine & Li 2007).SFR UV Star formation rate [M (cid:12) yr − ] from the rest-frame monochromatic luminosity at 2800 ˚A.SFR UV corr Star formation rate [M (cid:12) yr − ] from the rest-frame monochromatic luminosity at 2800 ˚A.corrected by extinction using A UV =(1.76 ± ± β .SFR TIR Star formation rate [M (cid:12) yr − ] from the L TIR. U Rest-frame U absolute magnitude from best-fit template. V Rest-frame V absolute magnitude from best-fit template. J Rest-frame J absolute magnitude from best-fit template.MNRAS000