An AzTEC 1.1-mm Survey for ULIRGs in the field of the Galaxy Cluster MS 0451.6-0305
J. L. Wardlow, Ian Smail, G. W. Wilson, M. S. Yun, K. E. K. Coppin, R. Cybulski, J. E. Geach, R. J. Ivison, I. Aretxaga, J. E. Austermann, A. C. Edge, G. G. Fazio, J. Huang, D. H. Hughes, T. Kodama, Y. Kang, S. Kim, P. D. Mauskopf, T. A. Perera, K. S. Scott
aa r X i v : . [ a s t r o - ph . C O ] O c t Mon. Not. R. Astron. Soc. , 1–20 (2009) Printed 29 October 2018 (MN L A TEX style file v2.2)
An AzTEC 1.1-mm Survey for ULIRGs in the field of the GalaxyCluster MS 0451.6 − J. L. Wardlow ⋆ , Ian Smail , G. W. Wilson , M. S. Yun , K. E. K. Coppin , R. Cybulski ,J. E. Geach , R. J. Ivison , , I. Aretxaga , J. E. Austermann , A. C. Edge , G. G. Fazio ,J. Huang , D. H. Hughes , T. Kodama , Y. Kang , S. Kim , P. D. Mauskopf , T. A.Perera , K. S. Scott , Department of Physics, Durham University, South Road, Durham, DH1 3LE, UK Institute for Computational Cosmology, Durham University, South Road, Durham, DH1 3LE, UK Department of Astronomy, University of Massachusetts, Amherst, MA 01003, USA Scottish Universities Physics Alliance, Institute for Astronomy, University of Edinburgh, Blackford Hill, Edinburgh, EH9 3HJ, UK UK Astronomy Technology Centre, Royal Observatory, Blackford Hill, Edinburgh EH9 3HJ, UK Instituto Nacional de Astrof´ısica, ´Optica y Electr´onica, Tonantzintla, Puebla, M´exico Center for Astrophysics and Space Astronomy, University of Colorado, Boulder, CO 80309, USA Harvard-Smithsonian Center for Astrophysics, 60 Garden St. MS-65, Cambridge, MA 02138-1516, USA National Astronomical Observatory of Japan, Mitaka, Tokyo 181-8588, Japan Astronomy & Space Science Department, Sejong University, Seoul, South Korea School of Physics & Astronomy, Cardiff University, Queens Buildings, The Parade, Cardiff, CF24 3AA, UK Illinois Wesleyan University, P.O. Box 2900, Bloomington, IL 61702-2900, USA Department of Physics and Astronomy, University of Pennsylvania, Philadelphia, PA 19104, USA
Accepted year month day. Received year month day; in original form year month day
ABSTRACT
We have undertaken a deep ( σ ∼ . mJy) 1.1-mm survey of the z = 0 . clusterMS 0451.6 − > . in the central 0.10 deg and present the AzTEC map, catalogueand number counts. We identify counterparts to 18 sources (50%) using radio, mid-infrared, Spitzer
IRAC and Submillimeter Array data. Optical, near- and mid-infrared spectral energydistributions are compiled for the 14 of these galaxies with detectable counterparts, which areexpected to contain all likely cluster members. We then use photometric redshifts and colourselection to separate background galaxies from potential cluster members and test the reli-ability of this technique using archival observations of submillimetre galaxies. We find twopotential MS 0451 −
03 members, which, if they are both cluster galaxies have a total star-formation rate (SFR) of ∼
100 M ⊙ yr − – a significant fraction of the combined SFR of allthe other galaxies in MS 0451 −
03. We also examine the stacked rest-frame mid-infrared, mil-limetre and radio emission of cluster members below our AzTEC detection limit and find thatthe SFRs of mid-IR selected galaxies in the cluster and redshift-matched field populations arecomparable. In contrast, the average SFR of the morphologically classified late-type clusterpopulation is ∼ times less than the corresponding redshift-matched field galaxies. This sug-gests that these galaxies may be in the process of being transformed on the red-sequence by thecluster environment. Our survey demonstrates that although the environment of MS 0451 − Key words: submillimetre – galaxies: clusters: individual (MS 0451 − ⋆ E-mail: [email protected]
Galaxy clusters are highly biased environments in which galax-ies potentially evolve more rapidly than in the field. The galaxypopulations of local massive clusters contain mainly early- c (cid:13) J. L. Wardlow et al. type galaxies which define a colour-magnitude relation (CMR)(Visvanathan & Sandage 1977; Bower et al. 1992). However, stud-ies of clusters out to z ∼ suggest that they contain increasingactivity at higher redshifts due to a growing fraction of blue, star-forming galaxies (Butcher & Oemler 1984). Over the same redshiftrange there appears to be a growing deficit in the CMR popula-tion at faint magnitudes, as well as a claimed increasing decline inthe numbers of S0 galaxies, suggesting that the blue, star-forminggalaxies may be transforming into these passive populations withtime (Dressler et al. 1997; Smail et al. 1998; De Lucia et al. 2007;Stott et al. 2007; Holden et al. 2009).The blue, star-forming populations within the clusters are ac-creted from the surrounding field as the clusters grow via gravi-tational collapse. The evolution in the star-forming fraction in theclusters may thus simply track the increasing activity in the fieldpopulation at higher redshifts. The increasing activity in the field isalso reflected in an increasing number of the most luminous (anddusty) star-burst galaxies with redshift (e.g. Le Floc’h et al. 2005):the Luminous Infrared Galaxies (with L FIR > L ⊙ ) and theirUltraluminous cousins (ULIRGs, L FIR > L ⊙ ). These sys-tems will also be accreted into the cluster environment along withtheir less-obscured and less-active population as the clusters grow.Indeed, mid-infrared (mid-IR) surveys of clusters have identifieda population of dusty star-bursts whose activity increases with red-shift (e.g. Geach et al. 2006, 2008). However, these mid-IR surveyscan miss the most obscured (and potentially most active) galax-ies which are optically thick in the rest-frame mid-IR. If they arepresent in clusters – even in relatively low numbers – such ac-tive galaxies will contribute significantly to the star formation rate(SFR) in these environments and the metal enrichment of the in-tracluster medium. Hence to obtain a complete census of the starformation within clusters we need to survey these systems at evenlonger wavelengths, in the rest-frame far-infrared (far-IR), corre-sponding to the observed sub-millimetre and millimetre waveband.Over the past decade or more there have been a numberof studies of rich clusters of galaxies in the sub-millimetre andmillimetre wavebands (e.g. Smail et al. 1997; Zemcov et al. 2007;Knudsen et al. 2008). Many of these studies were seeking to ex-ploit the clusters as gravitational telescopes to aid in the studyof the distant sub-millimetre galaxy (SMG) population behind theclusters, while others focused on the detection of the Sunyaev-Zel’dovich (SZ) emission. Due to the limitations of current tech-nologies direct detection of millimetre sources is restricted to thosewith fluxes S µ m & mJy, or equivalently galaxies with SFRs & M ⊙ yr − – much higher than expected for the generalcluster populations based on optical surveys. Nevertheless thesestudies have serendipitously detected a number of cluster galax-ies, although these are either atypical (e.g. central cluster galax-ies, Edge et al. (1999)) or are not confirmed members (e.g. Best2002; Webb et al. 2005). More critically, with few exceptions thesestudies have all focused on the central 2–3 arcmin of the clusters,where the SZ emission and lensing amplification both peak, andhave not surveyed the wider environment of the cluster outskirtswhere much of the obscured star formation is likely to be occurring(e.g. Geach et al. 2006). The two exceptions are the wide-field sur-vey of the z ∼ . cluster A 2125 by Wagg et al. (2009) and thesurvey of an overdense region of the COSMOS field by Scott et al.(2008) and Austermann et al. (2009a). Wagg et al. (2009) detected29 millimetre galaxies across a ∼ -arcmin diameter region cen-tered on A 2125 of which none are claimed to be members. How-ever, the only redshift estimates available are based on crude radio-to-millimetre spectral indices, which are sensitive to both dust tem- perature and redshift (Blain et al. 2003). The AzTEC/COSMOSsurvey (Scott et al. 2008; Austermann et al. 2009a) covered a num-ber of structures, including an X-ray detected z = 0 . cluster andconcluded that the statistical overdensity of sources was most likelydue to the gravitational lensing of background SMGs by these fore-ground structures.To help to conclusively answer the issue of the obscured star-forming populations in distant clusters we have therefore under-taken a wide-field 1.1-mm survey of the z = 0 . rich clusterMS 0451.6 − −
03) with the AzTEC cam-era (Wilson et al. 2008) on the James Clerk Maxwell Telescope(JCMT). This panoramic millimetre survey can also take advan-tage of the significant archival data available for this well-studiedregion. In particular the panoramic imaging of MS 0451 −
03 from
Spitzer and uniquely, the
Hubble Space Telescope ( HST ), as wellas extensive archival multiwavelength imaging and spectroscopy,which we employ for determining cluster membership of AzTEC-detected galaxies. Our survey probes the rest-frame 700 µ m emis-sion of cluster galaxies – in search of examples of strongly star-forming but obscured galaxies – as well as identifying more lumi-nous and more distant examples of the SMG population. We cancompare our millimetre search for cluster members to the previousmid-IR survey of this cluster by Geach et al. (2006) who uncovereda population of dusty star-forming galaxies which dominate the in-tegrated SFR of the cluster of ± M ⊙ yr − . Our AzTECmap covers ∼ times the area of the 850 µ m SCUBA observa-tions of the central region of MS 0451 −
03 (Borys et al. 2004b),while the depth of σ ∼ . mJy is sufficient to identify ULIRGsindividually and obtain stacked constraints on far-IR fainter clustermembers.We describe our observations and the data reduction in § § § Λ CDM cosmology with Ω M = 0 . , Ω Λ = 0 . and H = 70 kms − Mpc − and all photometry is onthe AB system unless otherwise stated. The AzTEC millimetre camera (Wilson et al. 2008) was installedon the JCMT during November 2005 and operated nearly contin-uously until February 2006. On the JCMT AzTEC has a field-of-view of ∼ -arcmin and a beam of 18-arcsec FWHM. For thisproject we observed a 0.23-deg region centred on MS 0451 −
03 at h m . s , ◦ ′ . ′′ , over a total of 24.5 hours between 2005December 14 and 2006 January 6. Observations were carried outin raster-scan mode with 1200 ′′ long scans in elevation separatedby 9 ′′ steps in azimuth (see Wilson et al. (2008) for details of ourstandard raster scan strategy). Scans were made at a fixed speed of120 ′′ s − . Data from the turnarounds were excised for this analysis.Each observation took 35 minutes to complete and a total of 42 ob-servations were made of the field. Over the set of observations thezenith opacity at 225 GHz, τ , was monitored with the CaltechSubmillimeter Observatory’s opacity meter and ranged from 0.03to 0.17. The average opacity for the set of observations was 0.09,equivalent to an atmospheric column of precipitable water vapourof ∼ mm. c (cid:13) , 1–20 Figure 1.
AzTEC S/N contours for the MS 0451 −
03 field shown over the corresponding Subaru R -band image; smoothed (dot-dashed) X-ray contourshighlight the cluster centre. We also plot a 5-arcmin radius circle (dashed) illustrating the sub-region analysed in § > of the maximum value, where sources are extracted. In this region AzTEC contours shown are at σ intervals, startingat . σ . 15 arcsec radius circles mark the 1.1-mm source detections labelled in order of decreasing S/N (Table 2) and squares highlight sources which arepotential cluster members ( § The 42 individual AzTEC observations were reduced andcombined using the publicly available AzTEC Data ReductionPipeline V1.0 – a set of IDL code that is optimised for detect-ing point sources in blank-field AzTEC surveys. This is the samepipeline that has been used in previous blank-field AzTEC/JCMTanalyses including Scott et al. (2008), Perera et al. (2008) andAustermann et al. (2009b).Since the reduction of these observations follows the same procedure and uses the same code as previous AzTEC studies, werefer the reader to Scott et al. (2008) for a thorough review of thereduction steps. Here we focus on the particulars of this field andthe measured characteristics of the MS 0451 −
03 map.Since we are interested in counterparts to our detected SMGsit is critical that we understand the pointing accuracy of themaps. In the reduction pipeline, fine corrections to the JCMT’spointing model are applied to each observation based on reg- c (cid:13) , 1–20 J. L. Wardlow et al. ular observations of a nearby QSO ( α = 04 h m . s , δ = − ◦ ′ . ′′ ) in the same manner as described in Scott et al.(2008) and Wilson et al. (2008). The residual astrometric error forthe field is measured by stacking the AzTEC map at the positionsof radio sources in the field (see § ∼ σ and demonstrate an overall systematic astrometric shift inthe AzTEC map of ∆ α = − . ′′ ± . ′′ , ∆ δ = 1 . ′′ ± . ′′ withrespect to the radio reference frame. This correction is negligiblein Dec, but the significant RA offset is applied to the AzTEC dataprior to further analysis and source extraction.The AzTEC data are flux calibrated as described inWilson et al. (2008) from nightly observations of Uranus over theJCMT run. The final calibration accuracy is ∼ .The primary products which come out of the AzTEC pipelineare a map representing the average filtered point source response,a map of the of the weight of each pixel in the sky flux map (rep-resenting the uncertainty in flux of each pixel), and a map of thesignal to noise estimate in each pixel. All maps are made on thesame ′′ × ′′ grid and all maps have been Wiener filtered to op-timize sensitivity to point sources. The central ∼ . deg of theAzTEC map is relatively uniform, with all pixels having > ofthe peak weight. Sources are extracted in the central region of themaps ( § ∼ . mJy. Contours of ‘signal-to-noise’ (S/N), overlayed on the Subaru R-band image of the fieldare shown in Fig. 1. Significant multiwavelength archival data exist for theMS 0451 −
03 field, which are summarised in Table 1 anddescribed below.
Geach et al. (2006) obtained
Spitzer
MIPS 24 µ m observations of a0.23-deg area of MS 0451 −
03, excluding the central 25-arcmin .The full details of the reduction and source extraction process aredescribed in Geach et al. (2006). The central region of the clusterwas part of Guaranteed Time Observations ( Spitzer program 83) sothese data were obtained from the archive and incorporated into themosaic. The 5- σ catalogue detection limit corresponds to 200 µ Jy.
Archival observations of the MS 0451 −
03 field at 1.4 GHzwere obtained using the National Radio Astronomy Observa-tory’s (NRAO ) Very Large Array (VLA), combining 9 hours ofdata obtained in 2002 June in the VLA’s BnA configuration with16 hours of A-configuration data taken in 2006 February (ProjectIDs AN109 and AB1199, respectively). The nearby calibrator,0503+020, was used to track amplitude and phase, with absoluteflux and bandpass calibration set via 0137+331. The now-standard1.4-GHz wide-field imaging approach was adopted (Owen et al.2005; Biggs & Ivison 2006), using spectral-line mode ‘4’ to ac-quire data with an integration time of 3 s (10 s in BnA) and using the IMAGR task in the Astronomical Image Processing System (
AIPS )to map out the primary beam using a mosaic of 37 images, with 17 NRAO is operated by Associated Universities Inc., under a cooperativeagreement with the National Science Foundation. more distant radio sources covered by additional facets. Several it-erations of self-calibration and imaging resulted in a noise level of11 µ Jy beam − , with a . ′′ × . ′′ synthesised beam at a positionangle 0.0 ◦ .Sources and corresponding fluxes are obtained from SE X - TRACTOR (Bertin & Arnouts 1996), using the S/N map for detec-tion and the flux map as the analysis image; sources are extractedwhere a minimum of 10 contiguous pixels (each is 0.16 arcsec )have S/N > . The resulting catalogue has a 3- σ flux limit of 51 µ Jy and is corrected for bandwidth smearing. We verify the statis-tical properties of our catalogue by comparing source counts withBiggs & Ivison (2006). Notably, the source density is not signifi-cantly enhanced towards the cluster centre, allowing us to use thecounts across the whole field in our statistical calculation of AzTECcounterparts ( § Observations of MS 0451 −
03 in the U -band were taken withthe MegaPrime camera on the Canada France Hawaii Telescope(CFHT) and reach a 3- σ detection limit of 25.1 mag in a 0.25-deg field. Standard reduction techniques were employed, customisedfor MegaPrime data (C.-J. Ma & H. Ebeling, private communica-tion), details of which can be found in Donovan (2007).In addition we observed a . -deg area centered on the clus-ter core on 2007 October 09 through the z ′ filter with the PrimeFocus Imaging Platform mounted on the 4.2-m William HerschelTelescope (WHT). A total integration time of one hour was ob-tained and reduced using standard techniques and calibrated usinga Sloan Digital Sky Survey standard field, yielding a 3- σ limitingmagnitude of 24.2 mag.Finally, the BVRI imaging employed here comes from thePISCES survey (Panoramic Imaging and Spectroscopy of ClusterEvolution with Subaru; Kodama et al. 2005). Detection limits arelisted in Table 1. In addition K -band imaging to a 3- σ depth of20.1 mag over 0.12-deg centred on MS 0451 −
03 was obtainedfrom the Wide-Field Infrared Camera (WIRC; Wilson et al. 2003)on the Palomar Hale telescope. Standard reduction methods wereemployed and are detailed in G. Smith et al. (in prep.). Thesedata were used in the previous analyses of Geach et al. (2006) andMoran et al. (2007a,b).
The
Spitzer
InfraRed Array Camera (IRAC; Fazio et al. 2004) ob-servations of the ′ × ′ field centered on the MS0451 field wereobtained as a part of the Cycle 5 General Observer (program 50610,PI: M. Yun) on 2009 March 18. Each tile was observed for a total of1500 seconds in each of the four IRAC bands (3.6, 4.5, 5.8, and 8.0 µ m) in full array mode with a 15 position dither pattern with 100second exposures at each position. The new data are combined withthe archival IRAC data (program 83) to produce the final mosaicimages using C LUSTER G RINDER (Gutermuth et al. 2009), whichis an
IDL software package that utilizes standard Basic CalibratedData (BCD) products from the Spitzer Science Centre’s standarddata pipeline. The angular resolution of the final mosaic ranges be-tween 2.0 to 2.5 arcsec depending on the observing band; the lim-iting depths are given in Table 1. c (cid:13) , 1–20 Table 1.
Summary of the observationsFilter Instrument Detection ReferenceLimit a U CFHT – MegaPrime 25.1 mag Donovan (2007) B Subaru – Suprime-Cam 26.6 mag Kodama et al. (2005) V Subaru – Suprime-Cam 25.8 mag Kodama et al. (2005) R Subaru – Suprime-Cam 25.1 mag Kodama et al. (2005) I Subaru – Suprime-Cam 24.2 mag Kodama et al. (2005) z ′ WHT – PFIP 24.2 mag This work K Palomar Hale – WIRC 20.1 mag G. Smith et al., in prep.3.6 µ m Spitzer – IRAC 23.0 mag This work4.5 µ m Spitzer – IRAC 23.2 mag This work5.8 µ m Spitzer – IRAC 21.4 mag This work8.0 µ m Spitzer – IRAC 22.1 mag This work24 µ m Spitzer – MIPS 120 µ Jy Geach et al. (2006)1.1 mm JCMT – AzTEC 3.6 mJy This work1.4 GHz VLA 51 µ Jy This work a . σ minimum catalogue limit in the roughly constant noise regionof the map; in the other bands we quote σ limits. To estimate redshifts photometrically we require that the photome-try in all filters samples the same emission from the source so an ac-curate spectral energy distribution (SED) can be built. To meet thisrequirement we degrade all the optical images to match the worstseeing – the ( V -band) in which the FWHM is 1.54 arcsec. Each see-ing convolved image is astrometrically matched to the USNO cat-alogue with a set of unsaturated and unblended stars spread acrossthe frames. We use SE XTRACTOR (Bertin & Arnouts 1996) on the R -band image to detect objects with a minimum of ten adjacent0.2-arcsec pixels at least 1.5 σ above the background to providea source list and then use the APPHOT routine in
IRAF to extract3-arcsec diameter aperture photometry at these positions in eachconvolved image. These measurements are then aperture correctedassuming a point source, to yield total magnitudes. We report 3- σ detection limits in Table 1.The resolution of the IRAC images is significantly lower thanthe optical so for source extraction we consider these data sepa-rately, although an equivalent procedure is followed. Sources aredetected on the 8 µ m image with SE XTRACTOR (Bertin & Arnouts1996) and are required to have a minimum of 4 adjacent 0.9-arcsecpixels at least 2 σ above the background. The 8 µ m band is cho-sen so that we can use the IRAC colours to constrain counterpartsfor otherwise unidentified SMGs ( § APPHOT routine in IRAF toextract 3.8-arcsec diameter fluxes. We also ensure that all SMGswhich are identified in Submillimeter Array (SMA), radio or 24 µ m data and have IRAC counterparts in any bands are extracted, re-moving the requirement for an 8 µ m detection. The extracted fluxesare corrected for the aperture losses, employing the factors calcu-lated by the SWIRE team (Surace et al. 2007), resulting in totalmagnitudes, which can be directly compared to our optical cata-logue. Figure 2.
Catalogue detection limits (see Table 1) in each filter for a z = 0 . galaxy overlayed with Arp 220 and M82 SEDs (Silva et al. 1998)scaled to our . σ This study aims to identify 1.1-mm detected ULIRGs in the z =0 . galaxy cluster MS 0451 −
03. A priori, based on the space den-sity of ULIRGs at z ∼ . (Le Floc’h et al. 2005), and the overden-sity of LIRGs in clusters at this redshift (Geach et al. 2006), we ex-pect to find, at most, only a couple of cluster members in our survey.The large background population coupled with this small numberof expected members makes this study challenging. Therefore, wefocus on techniques to identify sources and confirm cluster mem-bership. Similarly to other SMG studies, a fraction of our sourcesare unidentified in the radio, mid-IR or optically. However, basedon an Arp 220 SED, cluster members which are detectable aboveour 1.1-mm flux limit are also expected to be brighter than the cata-logue limits in all other bands (Fig. 2). Therefore, any unidentifiedsources are likely to be part of the background population. c (cid:13) , 1–20 J. L. Wardlow et al.
Table 2.
AzTEC galaxies in the MS 0451 −
03 field in order of decreasingS/N. Source names correspond to AzTEC positions in the J2000 epoch.SMGs with robust multiwavelength counterparts ( § a . + . − . a . + . − . a . + . − . b . + . − . . + . − . . + . − . b . + . − . . + . − . . + . − .
10 MMJ 045407.14-030033.9 4.3 4.3 1.0 . + . − .
11 MMJ 045358.12-025233.3 4.3 4.3 1.0 . +1 . − .
12 MMJ 045426.76-025806.5 4.2 4.2 1.0 . +1 . − .
13 MMJ 045356.09-025031.4 4.2 5.4 1.3 . +1 . − .
14 MMJ 045424.53-030331.7 4.1 4.2 1.0 . +1 . − .
15 MMJ 045328.86-030243.3 4.1 5.0 1.2 . + . − .
16 MMJ 045354.64-030004.0 4.0 4.0 1.0 . +1 . − .
17 MMJ 045431.35-025645.8 4.0 4.0 1.0 . + . − .
18 MMJ 045411.57-030307.5 4.0 4.1 1.0 . + . − .
19 MMJ 045415.53-025125.0 3.9 4.1 1.0 . +1 . − .
20 MMJ 045345.88-030440.0 3.9 4.0 1.0 . +1 . − .
21 MMJ 045358.49-025601.4 3.9 3.8 1.0 . +1 . − .
22 MMJ 045403.10-025006.8 3.9 4.7 1.2 . +1 . − .
23 MMJ 045357.46-030237.1 3.9 4.0 1.0 . +1 . − .
24 MMJ 045420.10-030658.2 3.8 4.0 1.0 . +1 . − .
25 MMJ 045422.17-030445.8 3.8 3.9 1.0 . +1 . − .
26 MMJ 045349.69-025807.1 3.8 3.8 1.0 . + . − .
27 MMJ 045421.17-030740.2 3.8 3.9 1.0 . + . − .
28 MMJ 045345.06-025722.6 3.7 3.7 1.0 . +1 . − .
29 MMJ 045442.54-025455.0 3.7 4.5 1.2 . +1 . − .
30 MMJ 045444.78-030030.9 3.7 3.8 1.0 . +1 . − .
31 MMJ 045411.73-025712.7 3.7 3.6 1.0 . +1 . − .
32 MMJ 045411.17-031019.2 3.6 4.1 1.1 . +1 . − .
33 MMJ 045352.72-030130.9 3.6 3.7 1.0 . +1 . − .
34 MMJ 045454.20-025919.3 3.6 4.6 1.3 . +1 . − .
35 MMJ 045359.27-030327.4 3.5 3.6 1.0 . +1 . − .
36 MMJ 045340.09-030334.2 3.5 3.6 1.0 . +1 . − . a These SMGs were observed with the SMA and detected ( § b These SMGs were observed with the SMA but not formally detected( § Millimetre sources are identified as local maxima with S/N > . and are selected in the roughly uniform noise ( σ ∼ . mJy, ∼ ) region of the AzTEC signal-to-noise map (with weightsof at least half of the maximum). In Table 2 we present these 36detections in order of decreasing S/N. Measured source fluxes andtheir errors are given by the peak pixel value and correspondingnoise value respectively; deboosted fluxes are calculated follow-ing the method of Perera et al. (2008). We determine the positionsof sources on the sub-pixel level by calculating the centroid af-ter weighting pixels close to the brightest pixel according to thesquare of their fluxes. Fig. 1 shows the Subaru R -band image withAzTEC contours of constant S/N and the detected 1.1-mm sources Figure 3.
Differential number counts using deboosted fluxes fromthe AzTEC survey of MS 0451 −
03 compared to those from theAzTEC/SHADES survey (Austermann et al. 2009b). We see that across thewhole area of our survey there is no strong overdensity of 1.1 mm sources inMS 0451 −
03 compared to the AzTEC/SHADES blank-field survey; henceit is apparent that the majority, but not necessarily the entirety, of the mil-limetre sources in the cluster field are background galaxies. However, in thecentral 5 arcmin radius of the map ( ∼ . Mpc at z = 0 . ) there appearsto be a slight overdensity of sources. This could be due to a combination ofthe Sunyaev-Zel’dovich effect, gravitational lensing of background sourcesby MS 0451 −
03, and potentially a contribution from 1.1 mm cluster mem-bers. The solid line is the Schechter function fit to the number counts in theentire AzTEC map, and the dotted lines represent the survey limits - thelower and upper lines represent the whole AzTEC MS 0451 −
03 field andthe central 5 arcmin radius, respectively. For clarity number counts in thecentral 5 arcmin radius of MS 0451 −
03 and the AzTEC/SHADES surveyare offset slightly in flux. marked. Based on the analyses of AzTEC data in Scott et al. (2008)and Perera et al. (2008) we expect that 3–4 (8–10%) of the sourcesin our . σ catalogue are false detections with only ∼ . and 0amongst those with S/N > . and respectively.To examine whether there is an overdensity of AzTEC galax-ies in MS 0451 −
03 compared to similar field surveys – poten-tially indicating a significant number of obscured ULIRG clus-ter members – we calculate the number counts in MS 0451 − −
03 map, by trimming the map to this area and re-peating the calculations with the sky model, deboosting, and com-pleteness estimates based on the full maps, since these are not ex-pected to change across the field. The number counts for the wholeMS 0451 −
03 survey area, and the central 5 arcmin (which repre-sents a physical scale of ∼ . Mpc at z = 0 . ), compared to theAzTEC/SHADES survey are presented in Fig. 3. c (cid:13)000
03 survey area, and the central 5 arcmin (which repre-sents a physical scale of ∼ . Mpc at z = 0 . ), compared to theAzTEC/SHADES survey are presented in Fig. 3. c (cid:13)000 , 1–20 Across the whole field MS 0451 −
03 does not exhibit a sourceexcess at 1.1 mm compared to a blank field. This suggests that, asexpected, there is not a dominant population of luminous obscuredstar-forming galaxies in MS 0451 −
03. Nevertheless, it is possiblethat a small number of the sources are still cluster members. In-deed, our number counts analysis (Fig. 3) suggests there may be asmall overdensity of sources in the central 5-arcmin radius regionof the AzTEC map. Based on integral number counts and their er-rors at 1.1 mm down to 1 mJy, derived from bootstrap samplingthe Posterior Flux Densities of sources (as described in detail inAustermann et al. (2009b)) a tentative overdensity is apparent atthe ∼ . σ level. If real this overdensity could arise from a com-bination of the Sunyaev-Zel’dovich effect, the gravitational lensingof background sources by the cluster, and potentially 1.1 mm clus-ter members. To discover whether any of this excess is caused by1.1-mm bright cluster galaxies we must first accurately locate thecounterparts to the 1.1-mm emission. The large AzTEC beam (18-arcsec FWHM on the JCMT), coupledwith the high spatial density of optical sources makes identificationdifficult unless precise positions are determined for the SMGs. Theobserved mm/sub-mm emission from SMGs represents rest-framefar-IR emission from dust-reprocessed starlight or AGN activity.Therefore, by exploiting the far-IR–radio correlation (e.g. Condon1992; Garrett 2002) and the low spatial density of radio sources, itis possible to employ deep ( σ ∼ µ Jy ), high-resolution, interfer-ometric radio observations to determine accurate source positionsfor ∼ µ m observations can be used foridentification, a method which has proved useful in confirming ten-tative radio counterparts, or providing positions for a small numberof the radio undetected SMGs (e.g. Ivison et al. 2004; Pope et al.2006; Ivison et al. 2007; Chapin et al. 2009). In addition, mid-IRcolours have been used to isolate potential cluster SMGs – al-though this emission does not track far-IR output and so it is anindirect indicator of the luminous far-IR source (Ashby et al. 2006;Pope et al. 2006; Yun et al. 2008).An alternative approach is submillimetre interferometry (e.g.Iono et al. 2006; Younger et al. 2007; Cowie et al. 2009), whichis preferable to these traditional radio and mid-IR identificationmethods because it samples the same part of the Spectral EnergyDistribution (SED) as our AzTEC imaging. However, sourcesgenerally require individual observations due to the small field-of-views of such instruments, commanding prohibitive exposuretimes to accurately locate all sources in a catalogue.We employ all three methods in this work, using millimetreinterferometry where available, and radio and mid-IR imaging asan alternative. The details of each of these identification methods isdiscussed in the following sections. In total we identify 18 AzTECcounterparts (50% of the sample; Table 3), of which 14 (39% ofthe total) also have optical counterparts (Table 4). In Fig. 2 wedemonstrate that at z ∼ . our multiwavelength observations aredeep enough to detect galaxies above our AzTEC flux limit andhence provide counterparts, assuming SEDs typical of Arp 220 orM82. Therefore, our unidentified sample is not expected to containcluster members. Of the optically-unidentified SMGs, one lies outof the field of the Subaru, CFHT and IRAC observations, two arecontaminated by nearby saturated stars and one is fainter than ourdetection limits at all wavelengths. We discuss the AzTEC galax- ies and corresponding counterparts on a source by source basis inAppendix A. The five brightest SMGs presented here were observed at µ mwith the SMA in October and November 2007. On-source inte-gration times of 5–6 hours were employed, yielding maps with σ µ m ∼ . mJy. The SMA configuration resulted in a beamof ∼ arcsec. These data are discussed in detail in Wardlow et al.(in prep.).Of the five sources observed with the SMA MMJ 045438.96,MMJ 045447.55 and MMJ 045433.57 were detected with fluxesof 8.4, 5.5 and 8.3 mJy respectively. The remaining two targets,MMJ 045431.56 and MMJ 045413.35, were undetected, but haveAzTEC 1.1 mm detections with SNR = 5 . and . respec-tively. Unlike this MS 0451 −
03 study, previous SMA observa-tions of AzTEC galaxies in the COSMOS field detected all seventargets (Younger et al. 2007). However, the faintest of the COS-MOS SMGs has deboosted 1.1 mm flux of 5.2 mJy, compared to3.7 and 3.6 mJy for MMJ 045431.56 and MMJ 045413.35 respec-tively. Therefore, based on the deboosted fluxes it is possible thatMMJ 045431.56 and MMJ 045413.35 are too faint to be detectedin our observations, although there are examples of bright SMGsthat are undetected with the SMA (e.g. Matsuda et al. 2007). Al-ternatively, the lack of SMA counterparts could suggest that ei-ther the µ m to 1.1 mm flux ratios of these sources are lowerthan expected (i.e. the dust is colder, or they are higher redshiftthan spectroscopically identified SMGs), that these galaxies ex-hibit extended far-IR emission on scales ≫ arcsec ( ≫ kpc at z = 2 ; for example, from a merger-induced starburst),or that the AzTEC beam contains a blend of multiple SMGs.Our radio and SMA observations have similar resolutions, there-fore, extended or multiple radio counterparts can indicate the ex-tended or multiple nature of (sub)millimetre emission. Based onthe radio emission it is likely that MMJ 045431.56 is composedof multiple (sub)millimetre sources, as discussed in Appendix A.However, MMJ 045413.35 is more likely to be a single resolved(sub)millimetre source, with a lower 890-to-1100 µ m ratio thanexpected. These issues are discussed further in Wardlow et al. (inprep.). We use the positions of the three SMA detected SMGs inthe following analysis.We note that the proposed radio identifications ofMMJ 045438.96 and MMJ 045433.57 are coincident with theirSMA positions. The other SMA detected galaxy, MMJ 045447.55,lies ∼ arcsec from an elliptical galaxy with bright radio andmid-IR emission, which, due to the high fluxes is formally a‘robust’ identification, but is unlikely to be the true source of themillimetre emission. MMJ 045431.56 and MMJ 045413.35 whichwere targeted but not detected with the SMA are identified throughradio and mid-IR counterparts respectively. c (cid:13) , 1–20 J . L . W a r d l o w e t a l . Table 3.
Radio and 24 µ m counterparts of AzTEC galaxies in MS 0451 −
03. All matches within 10-arcsec are listed, those secure identifications at 1.4 GHz or 24 µ m with P . are shown in bold and tentativeassociations ( . < P . ) are also presented. Images and a discussion of each source are presented in Appendix A.Source 1.4GHz position a S . b P . µ m position c S µ m d µ m P µ m µ m–1.4GHz e RA Dec Separation RA Dec Separation Separation(J2000) ( µ Jy) (arcsec) (J2000) ( µ Jy) (arcsec) (arcsec) f h m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′ g h m . s − ◦ ′ . ′′ < - - - - < - - -04 h m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′ f h m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′ h h m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′ fh m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′ - - < - - - h m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′
10 MMJ 045407.14 04 h m . s − ◦ ′ . ′′ h m . s − ◦ ′ . ′′
12 MMJ 045426.76 i h m . s − ◦ ′ . ′′ < - - - - < - - -
15 MMJ 045328.86 j - - ∼ - - h m . s − ◦ ′ . ′′ -
17 MMJ 045431.35 04 h m . s − ◦ ′ . ′′ - - < - - -
18 MMJ 045411.57 j - - ∼ - - h m . s − ◦ ′ . ′′ -
26 MMJ 045349.69 - - < - - h m . s − ◦ ′ . ′′ -
27 MMJ 045421.17 j - - ∼ - - h m . s − ◦ ′ . ′′ -28 MMJ 045345.06 j - - ∼ - - 04 h m . s − ◦ ′ . ′′ i h m . s − ◦ ′ . ′′ < - - - - < - - - a ∼ . -arcsec. b µ Jy. c µ m positions have typical uncertainties of ∼ . -arcsec. d µ m fluxes have typical errors of 40 µ Jy. e µ m–1.4GHz separations consistent with the same counterpart are italicised. f The radio and mid-IR positions agree with the SMA position ( § g MMJ 045447.55 is identified from SMA observations (Wardlow et al., 2009b), but a bright elliptical galaxy ∼ . arcsec from the AzTEC position is also a formal radio and mid-IR identification. We considerthe radio and mid-IR galaxy to be a chance association – at P = 0 . we expect 1–2 chance associations in our catalogue. h There is an additonal 176- µ Jy 24- µ m source, with P = 0 . , coincident with the radio counterpart of MMJ 045431.56. It is considered related to the AzTEC detection due to the radio identification. i MMJ 045426.76 and MMJ 045442.54 have no radio or mid-IR counterparts but are identified on the basis of their IRAC colours (Yun et al. 2008). j There is a ∼ σ radio peak coincident with the 24- µ m counterpart. c (cid:13) R A S , M N R A S , In this work we use both 1.4-GHz VLA and 24- µ m Spitzer
MIPSobservations to locate the AzTEC galaxies in our sample. We iden-tify SMG counterparts independently at radio and mid-IR wave-lengths before comparing these for each AzTEC source. In ourmaps the surface density of radio galaxies is lower than the mid-IR,the positional accuracy is greater ( ∼ . arcsec compared to ∼ . arcsec), and the link from the radio emission to the far-IR is tighter.Therefore, if the 24 µ m and 1.4 GHz identifications disagree weconsider the radio position as the more reliable counterpart.To ensure that no genuine associations are missed we searchup to 10 arcsec from each AzTEC position. This correspondsto a ∼ σ search radius for the lowest signal-to-noise ratio(SNR) sources, where σ is the error on AzTEC position, given by . − , and the FWHM is of the instrument beam(Ivison et al. 2007). We reject counterparts with more than 10 percent probability of being chance associations using the P-statistic ofDownes et al. (1986) (see also Pope et al. 2006; Ivison et al. 2007).Counterparts found with P < . are considered robust and thosewith . < P < . tentative. We present all identifications inTable 3. Yun et al. (2008) studied SMGs securely identified with SMA, ra-dio, or 24 µ m data and found that they typically have redder IRACcolours than the submillimetre-faint foreground galaxy population.They proposed a selection criteria for SMG counterparts, based onIRAC colours and found that the SMA detected galaxies in theirsample, whether radio and mid-IR identified or not, were all re-covered by this method. We verify that this selection criteria forour SMA, radio and 24 µ m-identified SMG counterparts and useit to search for counterparts to our otherwise unidentified AzTECsources and identify two further galaxies. Unlike the Yun et al.(2008) SMA-detected SMGs, MMJ 045447.55, our optically, radioand 24 µ m faint, but SMA-detected SMG, is not detected in any ofthe IRAC wavebands to the depth of the survey listed in Table 1. −
03 AzTEC galaxies
Our study seeks to identify potential millimetre sources in the clus-ter population of MS 0451 −
03. There are spectroscopic observa-tions of 1639 galaxies within our field, but nevertheless none ofthe AzTEC galaxies have been spectroscopically observed. There-fore, we use photometric methods to separate any potential AzTECdetected cluster members from ‘typical’ z ∼ SMGs. We firstapply two simple colour tests – the advantage of these is that itis easy to understand the biases in the sample – before apply-ing a more sophisticated photometric redshift analysis. We usethe S µ m /S . and S . /S . flux ratios to estimatethe redshifts of identified AzTEC galaxies and then also considerthe BzK selection criteria of Daddi et al. (2004) to separate thehigher redshift ( z > . ) SMGs from potential cluster members.Finally, we test whether these conclusions are reliable by usingthe spectroscopic redshifts and multi-wavelength photometry ofthe SMGs in Borys et al. (2005), and apply these findings to ourAzTEC MS 0451 −
03 sources to obtain photometric redshifts forour galaxies.
In Fig. 4 we plot S µ m /S . versus S . /S . forAzTEC MS 0451 −
03 SMGs. For comparison we plot 850 µ msources from Ivison et al. (2007), extrapolated from the observed850 µ m to the equivalent . mm flux assuming a power law spec-trum of the form S ν ∼ ν . . As expected both sets of SMGs liebroadly within the same region of colour-colour space, adding con-fidence to our detections and identifications. The redshift tracksof the local star-forming galaxy Arp 220, and the higher redshiftHR10 ( z = 1 . ) (based on the SEDs of Silva et al. (1998)) sug-gest that the AzTEC galaxies have redshifts of . z . . ,in concordance with the SMG population (Chapman et al. 2005).However, potential degeneracy between redshift and dust tempera-ture in the templates makes this method unreliable for identifyingcluster galaxies.We next determine whether any of the SMGs are likely to lieat low redshift and hence potentially be members of MS 0451 − BzK selection of Daddi et al. (2004) to separate z > . galaxies from those at z < . Of the ∼ SMGs with spectro-scopic redshifts presented in Chapman et al. (2005) ∼ have z > . , suggesting that the BzK selection should enable usto remove the majority of background sources from our sample.Fig. 5 shows the optically identified SMGs in MS 0451 −
03 whichare covered by our B , z and K imaging, in addition to Daddi et al.(2004) selection criteria for high redshift ( z > . ) galaxies; SMGswith z phot < . are highlighted ( § BzK selection alone, at least two SMGs are low redshiftand therefore potential cluster members; several additional galax-ies lie close to the border or have photometric limits which couldplace them in the low redshift region. Therefore, we next carry outa full photometric redshift analysis of the whole sample using H Y - PERZ (Bolzonella et al. 2000) to calculate redshifts for the SMGs c (cid:13) , 1–20 J . L . W a r d l o w e t a l . Table 4.
Optical and near-IR photometry for the detected SMG counterparts with derived photometric redshifts. Potential cluster members are shown in bold ( § U B V R I z K µ m 4.5 µ m 5.8 µ m 8 µ m z phot > . ± ± ± ± ± ± ± ± ± ± . +0 . − . ± ± ± ± ± ± ± ± ± ± . +0 . − . ± ± ± ± ± ± ± ± ± ± ± . +0 . − . a MMJ 045421.55 24.92 ± ± ± ± ± ± ± ± ± ± ± . + . − . > . > . > . > . > . > . > . ± ± ± ± . +0 . − .
10 MMJ 045407.14 22.83 ± ± ± ± ± ± ± ± ± ± ± . +0 . − .
12 MMJ 045426.76 > . ± ± ± ± ± ± ± ± ± ± . +0 . − .
15 MMJ 045328.86 > . > . > . > . > . > . > . ± ± ± ± . +0 . − .
17 MMJ 045431.35 24.45 ± ± ± ± ± ± ± ± ± ± ± . + . − .
18 MMJ 045411.57 24.83 ± ± ± ± ± ± ± ± ± ± ± . +0 . − .
26 MMJ 045349.69 26.41 ± ± ± ± ± ± > . - - - - . +0 . − .
27 MMJ 045421.17 25.42 ± ± ± ± ± ± ± ± ± ± ± . +0 . − .
28 MMJ 045354.06 25.61 ± ± ± ± ± ± ± . +0 . − .
29 MMJ 045442.54 25.00 ± ± ± ± ± > . ± ± ± ± ± . +0 . − . a This is the combined photometry for C1 and C2 as discussed in § c (cid:13) R A S , M N R A S , Figure 4. S µ m /S . versus S . /S . for SMGs in theMS 0451 −
03 field. We differentiate between sources with robust and ten-tative detections ( . < P . ), and highlight SMGs which satisfythe BzK selection criteria of Daddi et al. (2004) (Fig. 5). Sources withoutboth mid-IR and radio detections are plotted at the 3 σ detection limit ofthe respective catalogues (Table 1). For comparison we also plot SHADESidentifications (Ivison et al. 2007), converted to expected observed 1.1 mmfluxes as discussed in the text; error bars of this sample are omitted forclarity. Both sets of SMGs lie broadly within the same region of colour-colour space, providing confidence in our identifications. Redshift tracksof the z = 1 . SMG HR 10 and local starburst Arp 220 (based on SEDsfrom Silva et al. (1998)) suggest that the AzTEC sources generally have . z . . , in concordance with Chapman et al. (2005). Having located the MS 0451 −
03 SMGs we now need to testwhether any are potential cluster members. Photometric redshiftshave been calculated for several sets of SMGs using various codesand spectral templates, generally designed for use on ‘normal’ lowredshift galaxies (e.g. Clements et al. 2008). However, SMGs arelocated at high redshifts and powered by dusty starbursts whichcan be contaminated by AGN meaning that the stellar templatesderived from low redshift galaxies may not be appropriate.Typically studies of SMGs do not have both spectroscopicand photometric information so testing of photometric redshiftresults is difficult. Often results from codes such as H
YPERZ (Bolzonella et al. 2000) or I
MPZ (Babbedge et al. 2004) are com-pared to those from other, apparently cruder, methods of redshiftestimation – such as the radio to far-IR -(sub)-mm spectral indices(e.g. Carilli & Yun 1999; Yun & Carilli 2002; Aretxaga et al. 2003;Clements et al. 2008). There are two main problems with this ap-proach – firstly, the comparison is made with a set of results whichthemselves are not known to be correct, and secondly, the errorson the comparison redshifts are typically large, such that a gen-eral agreement can be confirmed but nothing more. In cases wherespectroscopic information has been obtained for a sub-sample ofgalaxies typically they are small (e.g. Pope et al. 2005).In this work we expand on previous analyses to examine howwell redshifts can be constrained for a spectroscopically confirmedsample of typical SMGs (median z ∼ . ). We employ the sam- Figure 5. B − z vs. z − K colour-colour plot of the optically identifiedAzTEC galaxies in MS 0451 −
03 which lie within the field-of-view of our B , z and K observations. The lines of separation between passive and star-forming z > . BzK galaxies (pBzK and sBzK respectively), z < . galaxies, and stars (Daddi et al. 2004) are shown. Two SMGs occupy the z < . region, suggesting they are potential cluster members. We dis-tinguish between galaxies with z phot > . and z phot < . based onphotometric redshifts calculated in § BzK selection and full pho-tometric analysis broadly agree as to the high and low redshift samples giv-ing more confidence in our ability to isolate millimetre sources in the clus-ter from the dominant background population. As expected, by this criteriamost SMGs are deemed to be actively star-forming. ple of 12 SMGs from Borys et al. (2005) which have spectroscopicredshifts from Chapman et al. (2003, 2005) and Pope et al. (2008).Photometry in 12 bands from U to 8 µ m is presented in Borys et al.(2005). We note that SMM J123622.65 has a rest-frame UV/opticalspectrum with z = 2 . (Chapman et al. 2005; Swinbank et al.2004) but mid-IR spectroscopic analysis suggests z = 1 . ± . (Pope et al. 2008), therefore, we exclude it from the statistical anal-yses. We also exclude SMM J123712.05 from the sample as it is un-detected at wavelengths shorter than 2 µ m and the power-law shapeat longer wavelengths means that the redshift cannot be derived.We use the H YPERZ package (Bolzonella et al. 2000) for ourphotometric redshift estimates. The program calculates expectedmagnitudes in the observed filters for given SEDs of a library ofmodel star-formation histories, with different ages, reddening andredshifts; the observed and expected magnitudes are compared ineach filter. We use the Bruzual & Charlot (1993) spectral templatesprovided with the H YPERZ package which represent star-formationhistories resulting in SEDs which match local ellipticals (E), Sb, asingle burst (Burst) and a constant Star-formation rate (Im). Red-shifts between 0 and 7 are considered. SMGs are known to be dustysystems, therefore, we initially allow reddening of A V ,in steps of 0.2 using the Calzetti et al. (2000) reddening law. Agesof the galaxies are required to be less than the age of the Universeat the appropriate redshift.In Fig. 6, we compare the spectroscopic with our photomet- We use H
YPERZ (cid:13) , 1–20 J. L. Wardlow et al.
Figure 6.
Spectroscopic versus photometric redshifts for the SMGs inBorys et al. (2005). Galaxy SMMJ 123622.65 has rest-frame UV/opticalspectra with z = 2 . (Chapman et al. 2005; Swinbank et al. 2004) butmid-IR analysis places it at z = 1 . ± . (Pope et al. 2008), therefore,SMMJ 123622.65 is plotted at each redshift. Utilising 12-band photometryresults in constraints that are sufficient for separating high-redshift SMGsfrom those at z ∼ . . ric redshifts for the Borys et al. (2005) galaxies; error bars shownare the H YPERZ quoted 99% confidence intervals which representthe σ limits more realistically than the H YPERZ quoted 68% in-tervals for this sample. The average scatter in ∆ z = z phot − z spec is . ± . , although this is smaller for the z . SMGs( ∆ z = 0 . ± . ) than for the more distant sample ( ∆ z =0 . ± . ), demonstrating that when testing photometric red-shifts of these galaxies it is important to consider a comparisonsample with the same redshift distribution as the population understudy. Our results are unchanged if we restrict A V to less than 2.5although a tighter limit than this does affect the results – increasing ∆ z . AGN contamination in the 8 µ m IRAC filter is possible forSMGs (Hainline, L. J. et al. in prep.), therefore we repeat the pho-tometric redshift calculations as above but excluding the 8 µ m datafrom the analysis. For all the SMGs ∆ z no8 µ m ∆ z allbands yield-ing an average scatter ∆ z = 0 . ± . for the high redshift( z > . ) galaxies – consistent with the view that the 8 µ m fluxesare contaminated.We also find no change if we allow a full range of templates:E, S0, Sa, Sb, Sc, Sd, Burst, and Im spectral types, and minimalchange if only we restrict the templates to only Burst and Im models– despite the fact that when the full range of templates is allowedmost of the galaxies are best-fit by the burst and elliptical models.We therefore find it unlikely that the addition of any further pure-stellar templates will improve the redshift accuracy.In this work we are interested in separating low redshift ( z ∼ . ; potential cluster member) SMGs from the typical high redshift( z ∼ ) population, and it is clear from Fig. 6 that this is possiblebased on the photometric redshift resolution. We use H
YPERZ to calculate the redshifts of the SMGs in the fieldof MS 0451 −
03 from our optical, near-IR and mid-IR photome-try. The Bruzual & Charlot (1993) spectral templates included withH
YPERZ are used, and, as discussed in § A V , and redshifts to z . Bright, re- solved galaxies with uncharacteristically high primary photometricredshifts are considered to lie at the calculated secondary solutions.The redshift estimate of each counterpart is presented in Ta-ble 4, and in Fig. 7 we show the SMG photometric redshifts ver-sus AzTEC 1.1-mm fluxes. Although, most of the AzTEC sourcesare high redshift background galaxies, two – MMJ 045421.55 andMMJ 045431.35 – are possible cluster members, and are discussedfurther below. We note that as reported by other authors (e.g.Pope et al. 2005) there appears to be a weak trend of redshift withmillimetre flux – the brightest millimetre galaxies lie at higherredshifts. Our sample has a median photometric redshift of 1.2and an interquartile range of z = 0 . – . – lower than spectro-scopic studies (e.g. Chapman et al. 2005, h z i = 2 . ) but simi-lar to other photometric studies (e.g. h z i = 1 . ; Clements et al.2008). We note that if we exclude the potential cluster membersMMJ 045421.55 and MMJ 045431.35 from this analysis the me-dian redshift of the field SMGs in this study is 1.8 and the interquar-tile range is z = 0 . – . .Although in § µ m in-formation can sometimes improve the accuracy of photometric red-shift estimates of SMGs, the redshifts reported in Table 4 includethe 8 µ m photometry. This is because for the galaxies with onlyIRAC detections the resulting lack of information about the loca-tion of the . µ m stellar peak means that only weak redshift con-straints are possible if we remove the 8 µ m photometry. Critically,the inclusion or exclusion of the 8 µ m information does not affectMMJ 045421.55 and MMJ 045431.35 which still both have photo-metric redshifts consistent with the cluster.In Fig. 5 we showed the B − z versus z − K colour-colour plotfor AzTEC galaxies in MS 0451 −
03 and used the
BzK selectioncriteria (Daddi et al. 2004) to separate galaxies above and below z = 1 . . We similarly group the SMGs based on the full photo-metric analysis from H YPERZ and find that both methods broadlyagree. In Fig. 7 we show the redshifts and K -band magnitudes ofMS 0451 −
03 SMGs in comparison to the spectroscopic sample ofSMGs examined by Smail et al. (2004), and local ULIRGs fromKim et al. (2002) and Stanford et al. (2000). The apparent K -bandmagnitudes of our AzTEC sample are consistent with their esti-mated redshifts, when compared with spectroscopic SMG surveysand local ULIRGs, suggesting that our photometric redshifts arereasonable. The general agreement between our photometric red-shift analysis, BzK and K -band magnitudes supports the derivedredshifts of the AzTEC sources.The potential cluster members, MMJ 045421.55 andMMJ 045431.35, are discussed in detail here. The other identifiedSMGs are examined in Appendix A. MMJ 045421.55
MMJ 045421.55 is identified through radioemission 2.2 arcsec from the AzTEC centroid, which lies betweentwo optical galaxies (2.1 and 1.6 arcsec from the northern (C1)and southern (C2) galaxies respectively). There is also an IRACsource at the location of the radio emission, which appears slightlyextended towards C2 (Fig. A1). The three possible explanationsfor this system are: C1 and C2 are interacting and both millimetrebright; C2 is the millimetre counterpart; or neither C1 nor C2 areresponsible for the millimetre emission. Each of these possibilitiesis discussed below.Individual photometric analysis of C1 and C2 (excluding theIRAC information) suggests that they are both potential clustermembers, and with a separation of 2.6 arcsec - correspondingto ∼ kpc at z = 0 . – they could be in the early stagesof a merger. Although high-resolution (0.1 arcsec) archival HST c (cid:13) , 1–20 Figure 7.
The left-hand panel shows photometric redshifts of SMGs against the 1.1 mm AzTEC fluxes. Two SMGs - MMJ 045421.55 and MMJ 045431.35- have photometric redshifts consistent with being cluster members and are highlighted; we discuss these two sources in detail in § K -band magnitudes of MS 0451 −
03 SMGs compared to the spectroscopic sample presented in Smail et al. (2004), and localULIRGs from Kim et al. (2002) and Stanford et al. (2000). Both datasets occupy the same parameter space suggesting it is unlikely that any of our photometricredshifts are extreme outliers.
F814W imaging shows no evidence of disturbance (Fig. A1), theexpected tidal tails can be low surface brightness features, mak-ing them difficult to detect. Such interactions are widely known totrigger dusty starbursts in which the radio emission appears to belocated between the optical nuclei (e.g. systems similar to VV 114;Frayer et al. 1999; Le Floc’h et al. 2002; Iono et al. 2004).With the aim of measuring redshifts we targeted C1 and C2with the ISIS long-slit spectrograph on the WHT during servicetime in 2009 February. The total integration time was one hour in σ = 1 . -arcsec seeing and standard reduction techniques wereemployed. A faint continuum was observed (the two targets areblended), but no features suitable for redshift measurement weredetectable. A possible faint ( ∼ σ ) emission feature at 7852 ˚Ais visible, which, if real, is most likely to be [O III ] λ ˚A at z = 0 . – placing the galaxy in a small group just behind thecluster. However, the feature is tentative and inconclusive.On the assumption that C1 and C2 lie at the same redshift,and that the millimetre, radio and IRAC fluxes are emitted from amerging system as described above, we combine the optical fluxesfrom C1 and C2. The optical to near-IR photometric redshift of thewhole system is z = 0 . +0 . − . which, as shown in Fig. 7, agreeswith the z versus K ULIRG trend. Therefore, if C1 and C2 are aninteracting system, MMJ 045421.55 a possible cluster member.The radio counterpart lies only . σ from C2 (compared to . σ from C1), and the IRAC emission of MMJ 045421.55 appearsextended towards C2. Therefore, in the situation where C1 and C2are unassociated we find it most likely that the C2 is the counter-part to the IRAC, radio and millimetre flux. In this case we ob-tain a photometric redshift of z = 0 . +0 . − . – once again placingMMJ 045421.55 in the region of MS 0451 − MMJ 045431.35
The counterpart to MMJ 045431.35 is securelyidentified 5.4 arcsec from the AzTEC centroid through its radioemission. The source is detected in all our
Spitzer and ground-based imaging and resolved in the
HST
F814W image into merginggalaxies with centroids separated by ∼ . arcsec ( ∼ kpc at z = 0 . ) and tidal tails between them (Fig. A1). We suggestthat a dusty starburst triggered by the interaction between the twogalaxies is causing the millimetre emission from this system. Thephotometric redshift is calculated as z = 0 . ± . , makingMMJ 045431.35 a possible cluster member.If MMJ 045421.55 and MMJ 045431.35 are cluster membersat z = 0 . SED-fitting suggests they each have L FIR ∼ × L ⊙ and thus SFR ∼
50 M ⊙ yr − . We also find that theyare colder than typical z ∼ SMGs, with dust temperatures of T d = 15 ± K (for β = 1 . ), or T d = 30 ± K (for β = 1 . ), com-pared to T d ∼ K and β = 1 . for archetypal SMGs. Such prop-erties are not unprecedented – the spectroscopic survey of SMGs byChapman et al. (2005) contained examples of sources with equiva-lent millimetre-to-radio flux ratios at z ∼ . , suggestive of galax-ies containing cold dust. Similarly, in the SCUBA Local UniverseGalaxy Survey (SLUGS) Dunne et al. (2000) surveyed local IRAS -bright galaxies with SCUBA at 850 µ m and utilised the combined IRAS and SCUBA photometry for SED fitting. This sample of lo-cal galaxies has T d = 35 . ± . K, and β = 1 . ± . . In-deed, cold low-redshift galaxies are easier to detect at 850 µ m or1.1-mm than their hotter counterparts. This is because colder dustproduces emission which peaks at longer wavelengths than hotdust. Therefore, we do not find it unreasonable that one or both c (cid:13) , 1–20 J. L. Wardlow et al. of MMJ 045421.55 and MMJ 045431.35 are cluster members at z = 0 . with T d ∼ K and β = 1 . .If both MMJ 045421.55 and MMJ 045431.35 are members ofMS 0451 −
03, their combined SFR is ∼
100 M ⊙ yr − – a sig-nificant fraction of the SFR of all the cluster galaxies within 2Mpc ( ± ⊙ yr − ; Geach et al. 2006) MMJ 045431.35 lies ∼ . Mpc from the cluster centre (about half of the turnaroundradius), but MMJ 045421.55 is much closer to the centre: ∼ Mpcin projection. Notably, both of these systems are most likely merg-ers. Although not conclusive, if they are both cluster members, thissuggests that existing starbursts are not instantaneously suppressedas they are accreted into the cluster environment, and can surviveuntil at least 1 Mpc from the cluster centre. Alternatively the galaxypairs could have been accreted into the cluster – suggesting that ac-creted galaxies can retain their gas reserves during infall into clus-ters (Geach et al. 2009).
We can also investigate obscured star-formation of the generalgalaxy population in MS 0451 −
03 below the flux limit of ourAzTEC map by stacking fluxes at the positions of known clustergalaxies. Optical galaxies in MS 0451 −
03 were morphologicallyclassified by Moran et al. (2007b) using the scheme defined byAbraham et al. (1996), which we group into early-types and late-types for this analysis. We also define a mid-IR sample of galaxieswhich are bright at 24 µ m, based on the catalogue of Geach et al.(2006) with S µm > µ Jy. To reduce contamination we con-sider only those galaxies with spectroscopic redshifts and to searchfor environmental dependencies we also examine spectroscopicallyidentified field galaxies. The cluster members are required to have . < z < . and the field population is outside this win-dow. The field samples have median redshifts of 0.58, 0.46, and0.28 with interquartile ranges of z = µ m,AzTEC 1.1 mm and VLA 1.4GHz fluxes. We note that the bothgalaxy samples are optical magnitude limited due to the require-ment for a spectroscopic redshift. Therefore, the stacked SEDs maynot be representative of the entire population, in particular the mostobscured galaxies are likely to fall below the optical magnitudelimit. However, we expect such selection effects to equally affectthe cluster and field samples allowing us to compare populationsbetween the two density regimes.Any galaxies within 9 arcsec (the radius of the AzTEC beam)of an AzTEC map pixel with S/N > ± . are removed prior tostacking the AzTEC map. The contribution from each AzTEC-faintgalaxy is weighted by the inverse of the squared noise at that pixelto calculate the weighted mean 1.1 mm flux of each population.Similarly we calculate the clipped weighted mean radio flux of eachsample using our 1.4 GHz VLA map, correcting for bandwidthsmearing, and stack 24 µ m image in the same way. The stackingparameters are given in Table 5 and the resulting SEDs are pre-sented in Fig. 8.We fit the mid-IR and late-type galaxies with templateSEDs from Dale & Helou (2002) and calculate the correspond-ing SFRs based on the far-IR luminosity (Kennicutt 1998). Mid-IR cluster galaxies without bright AzTEC counterparts have SFR = 16 . ± . ⊙ yr − which is consistent the SFR estimateof . ± . ⊙ yr − for the mid-IR field population. In contrast,the late-type galaxies in the field have SFR = 3 . ± . ⊙ yr − ,compared to SFR = 2 . ± . ⊙ yr − in the cluster. Both clus- ter and field early-type populations are undetected at 1.1 mm and24 µ m and SED fitting to the 24 µ m limits suggest, on average,SFR < .
06 M ⊙ yr − .Since our samples are flux limited, the different redshift dis-tributions will affect the derived SFR. The extent of this effect onour results is tested by calculating an observed SFR for each sam-ple by fitting templates with S µ m = 200 µ Jy at the redshift ofthe galaxies included in the stacks. Due to the generally higher red-shift of the field sample, we find that identical mid-IR populationswould be observed with SFR . ± . times lower in the clus-ter than the field. In fact the stacked cluster population is observedwith SFR . ± . times lower than the field population, there-fore, to within ∼ σ the mid-IR samples are consistent. However,the equivalent test for late-type galaxies, where the field sample is,on average, lower redshift, suggests that, due to the different lumi-nosity distance, identical populations in our analysis would appear2.6 times more active in the cluster than in the field. In fact, thecluster late-type galaxies are less active than the field. Therefore,the cluster environment is suppressing star-formation activity in thelate-type population.Geach et al. (2006) estimated the total SFR within 2 Mpc ofthe core of MS 0451 −
03 as ±
100 M ⊙ yr − , based on con-verting 24 µ m fluxes of colour-selected 24- µ m detected galaxiesto total IR luminosities with the average SED from Dale & Helou(2002). Our analysis presented in this paper includes radio and 1.1mm data enabling us to better characterise the cluster populations,and by stacking we can probe a fainter population. Within 2 Mpcof cluster core we calculate SFR > ±
50 M ⊙ yr − from thespectroscopically confirmed mid-IR and late-type galaxies. To thiswe can then add
50 M ⊙ yr − for MMJ 045421.55, the potentialULIRG within 2 Mpc of the cluster core. In this study we have investigated the dust-obscured star-formingpopulation of the galaxy cluster MS 0451 −
03; we utilise 1.1-mmobservations to study obscured star-formation in MS 0451 −
03. Wepresent a σ ∼ . mJy AzTEC map of the central 0.1 deg ofMS 0451 −
03, within which 36 sources are detected at S/N > . .We use radio, 24 µ m, IRAC and SMA observations to preciselylocate 18 of these SMGs.We calculate the reliability of photometric redshifts for SMGsand find that they are able to remove the bulk of background con-tamination, allowing us to isolate potential cluster members us-ing our optical, near- and mid-IR photometry. Based on these red-shifts we find two SMGs which are possible cluster members:MMJ 045421.55 and MMJ 045431.35. These systems are both re-solved into close pairs of galaxies by our ground-based and HST imaging, suggesting that interactions have triggered their starbursts.If they are cluster members both of these SMGs contain cold dustwith T d ∼ K, for β = 1 . (similar to SLUGS galaxies;Dunne et al. 2000) and have SFRs of ∼
50 M ⊙ yr − each.Geach et al. (2006) compared obscured activity, based on µ m emission, in MS 0451 −
03 and Cl 0024+16 at z = 0 . , andfound that Cl 0024+16 has SFR ∼ times that of MS 0451 − −
03 is underactive and Cl 0024+16 overactiveat µ m. Therefore, it is likely the other galaxy clusters, includingCl 0024+16, could contain ULIRGs in significantly larger numbersthan we find in MS 0451 − c (cid:13) , 1–20 Figure 8.
Clipped weighted-average SEDs of the 24 µ m, late- and early-type populations of MS 0451 −
03 and the field. In the case of non-detections the 3 σ noise limit is marked by arrows. The field populations are calibrated to their mean redshifts of 0.58, 0.46, and 0.39 for the mid-IR galaxies, late- and early-types and respectively. We also show the best-fitting SEDs and corresponding SFRs (Kennicutt 1998) from Dale & Helou (2002) for the mid-IR and late-typegalaxies; since the early-type populations are only detected at 1.4 GHz we cannot select one best-fitting SED and instead display all the Dale & Helou (2002)SEDs on this panel. Although the observed populations have similar activity levels, in the late-type galaxies this is due to the flux limits of our observationsand the different luminosity distances between the samples. If the cluster and field late-type populations were equivalent, the cluster should have SFR ∼ times higher than observed. We deduce that the cluster environment has caused a reduction in the SFR of this population, and that these galaxies are probablyin the process of transformation onto the red sequence. Table 5.
Parameters and results of stacking on known cluster and field galaxies in the 24 µ m, 1.1 mm and 1.4 GHz images. SFRs are based onSED fits, except for the early-type galaxies, which are based purely on 24 µ m detection limits. Non-detections are represented by σ limits.Since the samples are flux limited and have various median redshifts we also present the expected observed SFR of the field mid-IR and late-type samples were they to have z median = 0 . ( § N µ m a S µ m b N . a S . b N . a S . b z median SFR SFR ( z = 0 . )( µ Jy) ( µ Jy) ( µ Jy) ( M ⊙ yr − ) ( M ⊙ yr − )Cluster mid-IR galaxies 14 ± < . ± . . ± . . ± . Field mid-IR galaxies 84 ± ± . ± . . ± . . ± . Cluster late-type 167 . ± . < . ± . . ± . . ± . Field late-type 315 . ± . < . ± . . ± . . ± . Cluster early-type 148 < . < . ± . < . < . Field early-type 35 < . < . ± . < . < . a The number of galaxies used in the stacking – excluding those close to the edge, off-image, or near detected sources b Clipped weighted-average flux
To further investigate the obscured star-forming populationwhich lies below the limit of our AzTEC observations we createcomposite SEDs of spectroscopically confirmed mid-IR, early- andlate-type cluster members and compare them to the correspondingfield populations. As expected we find that both early-type popula-tions are undetected at both 24 µ m and 1.1 mm and so are unlikelyto be actively forming large numbers of stars ( < . ⊙ yr − ).The 24 µ m galaxies are significantly more active than the morpho-logically classified late-types with SFRs ∼
15 M ⊙ yr − versus ∼ ⊙ yr − on average. We find that the star-formation activityin the cluster late-type population, compared to a redshift-matchedfield population is quenched and ∼ times lower than expected.Mid-IR galaxies do not show this trend suggesting the more in-tense activity in these systems is more robust to environmental in-fluences. We find that the total SFR > ±
50 M ⊙ yr − in thecentral 2 Mpc of MS 0451 −
03. However, if MMJ 045421.55 is a cluster member it has
SFR ∼
50 M ⊙ yr − and lies 1 Mpc fromthe cluster centre, taking the total SFR within 2 Mpc to & ⊙ yr − . ACKNOWLEDGEMENTS
We thank an anonymous referee for helpful comments whichgreatly improved the clarity of this paper. We would also like tothank C. J. Ma and Harald Ebeling for providing us with reducedCFHT U -band images, Richard Ellis for supplying reduced HST
WFPC2 images, and Sean Moran for the usage of spectroscopiccatalogues and attributed information.J. L. W., I. R. S and J. E. G. acknowledge support from the Sci-ence and Technology Facilities Council (STFC), and K. E. K. C.acknowledges support from an STFC Fellowship. K. S was sup-ported in part through the NASA GSFC Cooperative Agreement c (cid:13) , 1–20 J. L. Wardlow et al.
NNG04G155A. Support for this work was provided in part by theNSF grant AST 05-40852 and the grant from the Korea Science &Engineering Foundation (KOSEF) under a cooperative agreementwith the Astrophysical Research Center of the Structure and Evo-lution of the Cosmos (ARCSEC). Additional support for this workwas provided by NASA through an award issued by JPL/Caltech.The James Clerk Maxwell Telescope is operated by The JointAstronomy Centre on behalf of the Science and Technology Fa-cilities Council of the United Kingdom, the Netherlands Organi-sation for Scientific Research, and the National Research Councilof Canada. This work is based in part on observations made withthe
Spitzer Space Telescope , which is operated by the Jet Propul-sion Laboratory, California Institute of Technology under a con-tract with NASA. This work also made use of the Spitzer Archive,which is operated by the Spitzer Science Center. Based in part ondata collected at Subaru Telescope, which is operated by the Na-tional Astronomical Observatory of Japan. Based on observationsobtained with MegaPrime/MegaCam, a joint project of CFHT andCEA/DAPNIA, at the Canada-France-Hawaii Telescope (CFHT)which is operated by the National Research Council (NRC) ofCanada, the Institute National des Sciences de l’Univers of the Cen-tre National de la Recherche Scientifique of France, and the Univer-sity of Hawaii. The William Herschel Telescope and its service pro-gramme are operated on the island of La Palma by the Isaac NewtonGroup in the Spanish Observatorio del Roque de los Muchachos ofthe Instituto de Astrof´ısica de Canarias.
REFERENCES
Abraham R. G., van den Bergh S., Glazebrook K., Ellis R. S.,Santiago B. X., Surma P., Griffiths R. E., 1996, ApJS, 107, 1Aretxaga I., Hughes D. H., Chapin E. L., Gazta˜naga E., DunlopJ. S., & Ivison R. J., 2003, MNRAS, 342, 759Ashby M. L. N. et al., 2006, ApJ, 644, 778Austermann J. E. et al., 2009a, MNRAS, 393, 1573Austermann J. E. et al., 2009b, preprint (astro-ph/0907.1093)Babbedge T. S. R. et al., 2004, MNRAS, 353, 654Bertin E., Arnouts S., 1996, A&AS, 117, 393Best P. N., 2002, MNRAS, 336, 1293Biggs A. D., Ivison R. J., 2006, MNRAS, 371, 963Blain A. W., Barnard V. E., Chapman S. C., 2003, MNRAS, 338,733Bolzonella M., Miralles J.-M., Pell´o R., 2000, A&A, 363, 476Borys C., Scott D., Chapman S., Halpern M., Nandra K., Pope A.,2004a, MNRAS, 355, 485Borys C. et al., 2004b, MNRAS, 352, 759Borys C., Smail I., Chapman S. C., Blain A. W., Alexander D. M.,Ivison R. J., 2005, ApJ, 635, 853Bower R. G., Lucey J. R., Ellis R. S., 1992, MNRAS, 254, 601Bruzual A. G., Charlot S., 1993, ApJ, 405, 538Butcher H., Oemler Jr. A., 1984, ApJ, 285, 426Calzetti D., Armus L., Bohlin R. C., Kinney A. L., Koornneef J.,Storchi-Bergmann T., 2000, ApJ, 533, 682Carilli C. L., Yun M. S., 1999, ApJ, 513, L13Chapin E. L. et al., 2009, preprint (astro-ph/0906.4561)Chapman S. C., Blain A. W., Ivison R. J., Smail I. R., 2003, Na-ture, 422, 695Chapman S. C., Blain A. W., Smail I., Ivison R. J., 2005, ApJ,622, 772Chapman S. C., Smail I. R., Ivison R. J., Blain A. W., 2002, MN-RAS, 335, L17 Clements D. L. et al., 2008, MNRAS, 387, 247Condon J. J., 1992, ARA&A, 30, 575Cowie L. L., Barger A. J., Wang W.-H., Williams J. P., 2009, ApJ,697, L122Daddi E., Cimatti A., Renzini A., Fontana A., Mignoli M.,Pozzetti L., Tozzi P., Zamorani G., 2004, ApJ, 617, 746Dale D. A., Helou G., 2002, ApJ, 576, 159De Lucia G. et al., 2007, MNRAS, 374, 809Donovan D. A. K., 2007, PhD thesis, University of Hawai’i atManoaDownes A. J. B., Peacock J. A., Savage A., Carrie D. R., 1986,MNRAS, 218, 31Dressler A. et al., 1997, ApJ, 490, 577Dunne L., Eales S., Edmunds M., Ivison R., Alexander P.,Clements D. L., 2000, MNRASEdge A. C., Ivison R. J., Smail I., Blain A. W., Kneib J.-P., 1999,MNRAS, 306, 599Frayer D. T., Ivison R. J., Smail I., Yun M. S., Armus L., 1999,AJ, 118, 139Garrett M. A., 2002, A&A, 384, L19Geach J. E., Smail I., Best P. N., Kurk J., Casali M., Ivison R. J.,Coppin K., 2008, MNRAS, 388, 1473Geach J. E., Smail I., Moran S. M., Treu T., Ellis R. S., 2009, ApJ,691, 783Geach J. E. et al., 2006, ApJ, 649, 661Gutermuth R. A., Megeath S. T., Myers P. C., Allen L. E., PipherJ. L., Fazio G. G., 2009, ApJS, 184, 18Holden B. P. et al., 2009, ApJ, 693, 617Iono D., Ho P. T. P., Yun M. S., Matsushita S., Peck A. B.,Sakamoto K., 2004, ApJ, 616, L63Iono D. et al., 2006, ApJ, 640, L1Ivison R. J. et al., 2007, MNRAS, 380, 199Ivison R. J. et al., 2004, ApJS, 154, 124Ivison R. J. et al., 2002, MNRAS, 337, 1Ivison R. J. et al., 2005, MNRAS, 364, 1025Kennicutt Jr. R. C., 1998, ApJ, 498, 541Kim D.-C., Veilleux S., Sanders D. B., 2002, ApJS, 143, 277Knudsen K. K., van der Werf P. P., Kneib J.-P., 2008, MNRAS,384, 1611Kodama T. et al., 2005, PASJ, 57, 309Le Floc’h E., Charmandaris V., Laurent O., Mirabel I. F., GallaisP., Sauvage M., Vigroux L., Cesarsky C., 2002, A&A, 391, 417Le Floc’h E. et al., 2005, ApJ, 632, 169Matsuda Y., Iono D., Ohta K., Yamada T., Kawabe R., HayashinoT., Peck A. B., Petitpas G. R., 2007, ApJ, 667, 667Molnar S. M., Hughes J. P., Donahue M., Joy M., 2002, ApJ, 573,L91Moran S. M., Miller N., Treu T., Ellis R. S., Smith G. P., 2007a,ApJ, 659, 1138Moran S. M., Ellis R. S., Treu T., Smith G. P., Rich R. M., SmailI., 2007b, ApJ, 671, 1503Owen F. N., Keel W. C., Ledlow M. J., Morrison G. E., WindhorstR. A., 2005, AJ, 129, 26Perera T. A. et al., 2008, MNRAS, 391, 1227Pope A., Borys C., Scott D., Conselice C., Dickinson M.,Mobasher B., 2005, MNRAS, 358, 149Pope A. et al., 2008, ApJ, 675, 1171Pope A. et al., 2006, MNRAS, 370, 1185Scott K. S. et al., 2008, MNRAS, 385, 2225Silva L., Granato G. L., Bressan A., Danese L., 1998, ApJ, 509,103 c (cid:13) , 1–20 Smail I., Chapman S. C., Blain A. W., Ivison R. J., 2004, ApJ,616, 71Smail I., Edge A. C., Ellis R. S., Blandford R. D., 1998, MNRAS,293, 124Smail I., Ivison R. J., Blain A. W., 1997, ApJ, 490, L5Stanford S. A., Stern D., van Breugel W., De Breuck C., 2000,ApJS, 131, 185Surace J. A. et al., 2005,http://swire.ipac.caltech.edu/swire/astronomers/publications.htmlStott J. P., Smail I., Edge A. C., Ebeling H., Smith G. P., KneibJ.-P., Pimbblet K. A., 2007, ApJ, 661, 95Swinbank A. M., Smail I., Chapman S. C., Blain A. W., IvisonR. J., Keel W. C., 2004, ApJ, 617, 64Visvanathan N., Sandage A., 1977, ApJ, 216, 214Wagg J., Owen F., Bertoldi F., Sawitzki M., Carilli C. L., MentenK. M., Voss H., 2009, ApJ, 699, 1843Webb T. M. A., Yee H. K. C., Ivison R. J., Hoekstra H., GladdersM. D., Barrientos L. F., Hsieh B. C., 2005, ApJ, 631, 187Wilson G. W. et al., 2008, MNRAS, 386, 807Wilson J. C. et al., 2003, in Iye M., Moorwood A. F. M., eds,Proc. SPIE. 4841, Instrument Design and Performance for Opti-cal/Infrared Ground-based Telescopes. p. 451Younger J. D. et al., 2007, ApJ, 671, 1531Yun M. S., Carilli C. L., 2002, ApJ, 568, 88Yun M. S. et al., 2008, MNRAS, 389, 333Zemcov M., Borys C., Halpern M., Mauskopf P., Scott D., 2007,MNRAS, 376, 1073
APPENDIX A: NOTES ON INDIVIDUAL SOURCES
In Fig. A1 we present images centered on each AzTEC source, inorder of decreasing S/N. We show Subaru and IRAC colour, and
HST images, with radio and 24 µ m contours. Identified galaxiesare labelled and discussed below.
1. MMJ 045438.96:
SMA observations of this galaxy confirm theidentified radio and mid-IR counterparts. The corresponding redoptical and IRAC source at z phot = 3 . +0 . − . , has a disturbedmorphology in the HST image.
2. MMJ 045447.55:
This galaxy has been located with the SMAbut does not have any radio or 24 µ m counterparts, or optical orIRAC galaxies at the SMA position. Therefore, MMJ 0455447.57could be one of a population of very distant SMGs, or be muchcooler and more obscured than typical SMGs.
3. MMJ 045433.57:
SMA observations confirm the identifiedradio and mid-IR counterparts, corresponding to an optically faintbut near-IR bright galaxy with z phot = 2 . +0 . − . . The HST imageshows three components to this galaxy such that a merger is thelikely cause of the ULIRG.
4. MMJ 045431.56:
We identify a robust extended radio counter-part 9.5 arcsec from the 1.1 mm centroid, which also correspondsto faint mid-IR emission; we consider the corresponding red galaxyat z phot = 0 . +0 . − . to be the likely optical counterpart. The HST image of the region shows a galaxy with a bright core and a lowsurface-brightness tail or edge on spiral arm. However, the pictureof this SMG is complicated by the presence of a statisticallytentative mid-IR counterpart 9.1 arcsec from the 1.1 mm centroidwhich is unassociated with the aforementioned radio emission.The 24 µ m position is coincident with an extended red IRACgalaxy but no optical sources. Further complexity arises from theSMA observation; MMJ 045431.56 was targeted but not formallydetected. However, the SMA data does show two potential sources, close to an extended ∼ σ peak in the radio map, ∼ . arcsecfrom the AzTEC centroid and unassociated with both the radioand 24 µ m counterparts previously discussed. There is also a redIRAC source at this position, and the HST image shows two opticalgalaxies. However, in the IRAC and ground-based optical data thecounterpart is blended with a nearby elliptical galaxy.
5. MMJ 045421.55:
We find reliable coincident radio and 24 µ mcounterparts, corresponding to two optical galaxies separatedby 2.6 arcsec, which are possible interacting cluster members( z phot = 0 . +0 . − . ). The system is discussed in detail in §
6. MMJ 045417.49:
Both mid-IR and strong radio counterpartsare detected although the presence of a nearby, bright star preventsmulti-wavelength optical and near-IR study. Although the stellarhalo is less extended in the
HST image no counterpart is visible.
7. MMJ 045413.35:
The radio and 24 µ m identification agreeswith a 2–3 σ peak in the 890 µ m ‘dirty’ SMA map. Unfortunatelythis source lies outside of the IRAC and much of the ground-basedoptical coverage, although a galaxy is detected in the U -band and HST imaging.
8. MMJ 045412.72:
As potentially part of the gravitationallylensed arc from Borys et al. (2004b) this source is discussed fullyin Wardlow et al. (2009b). Our optical imaging is too shallow todetect the ERO discussed in Borys et al. (2004b), but radio andextended IRAC emission betrays the likely counterpart.
9. MMJ 045345.31:
There are two secure radio and one 24 µ mcounterparts. One of the radio galaxies corresponds to the24 µ m position. The IRAC source has a red component at themid-IR/radio position, but it blended with the emission from anoptically saturated star making further conclusions difficult. It isunclear whether the 24 µ m and radio flux is from the star or abackground SMG counterpart which may be the cause of the redcomponent of the IRAC emission.
10. MMJ 045407.14:
We find coincident reliable radio and 24 µ mcounterparts which are located in a red region at the edge of abright, resolved galaxy which appears disturbed in the HST imageand has z phot = 0 . +0 . − . . The morphology of the galaxy sug-gests that a merger has triggered a dusty starburst region, causingthe millimetre emission. The large size and optical brightnesssupport the low redshift of this galaxy, although if the photometricredshift errors are underestimated it could potentially be a clustermember. Similarly, it is also possible that a background galaxyis the source of the millimetre emission. The detection of COemission lines is required to confirm either scenario.
12. MMJ 045426.76:
There are no radio or mid-IR counterpartsidentified. However, an IRAC galaxy has colours suggestive ofSMG emission (Yun et al. 2008), yielding a counterpart with z phot = 0 . ± . .
15. MMJ 045328.86:
A 24 µ m counterpart is detected 3.8 arcsecfrom the AzTEC position, corresponding to an IRAC source,but no detectable optical flux. The IRAC photometry yields z phot = 1 . +0 . − . . There is coincident faint radio emission( ∼ σ ) at the position of the 24 µ m counterpart, increasing thelikelihood that this is the correct identification.
17. MMJ 045431.35:
A radio counterpart 4.8 arcsec from theAzTEC position coincides with an pair of merging galaxies whichare resolved in the
HST image and have z phot = 0 . ± . ; thispossible cluster ULIRG is discussed further in §
18. MMJ 045411.57: µ m emission betrays the source of themillimetre emission as a galaxy with z phot = 0 . +0 . − . , whichis resolved in the HST image as an interacting pair, separated by < arcsec. There is faint radio emission ( ∼ σ ) ∼ . arcsecto the northeast of the 24 µ m identification, coincident with a red c (cid:13) , 1–20 J. L. Wardlow et al.
Figure A1. × arcsec images centred on each AzTEC galaxy are shown in the left-hand and middle panels; north is up and east is to the left. Inthe left-hand panel we show Subaru data: BVR colour images with solid radio contours at -3 (dotted), 3, 4, 5, 6, 7, 8, 9, 10 × σ are overlayed. The middlepanel contains true-colour images containing the IRAC . . (blue), 5.8 (green) and 8.0 µ m (red) data; dashed contours present 24 µ m flux at 5, 10,20, 30, 40, 50, 100, 200, ...1000 × µ Jy per pixel and the right-hand panel contains × arcsec HST
F814W cutouts centered at the AzTEC position. Inall images circles are centred on the AzTEC positions and have 20 arcsec diameter, corresponding to our radio and mid-IR search radius. 24 µ m and radiocounterparts are highlighted with diamonds and squares respectively; dotted symbols represent tentative counterparts (with . P . ) and the sizesof the symbols are representative of the typical astrometric errors of these data. We mark the positions of SMA detected sources with stars and X-ray sources(Molnar et al. 2002) with crosses. c (cid:13) , 1–20 Figure A1 – continued IRAC source. We conclude that the interaction between the opticalgalaxy pair is producing all of the emission. However, it is possiblethat the radio and IRAC source is unrelated to the optical emission,in this case we can only constrain the redshift to z phot > . .
26. MMJ 045349.69:
A faint optical galaxy with z phot =1 . +0 . − . corresponds to the robust mid-IR counterpart. The galaxy is undetected by HST and the region is not covered by ourIRAC mosaic.
27. MMJ 045421.17:
The mid-IR counterpart is coincidentwith both faint radio emission and a faint red galaxy at z phot = 1 . +0 . − . , which the HST imaging resolves into abright nuclear region and some extended emission. c (cid:13) , 1–20 J. L. Wardlow et al.
Figure A1 – continued
28. MMJ 045345.06:
The galaxy is tentatively identified throughits mid-IR emission, which is coincident with faint radio flux;our IRAC mosaic does not cover the counterpart, but opticalphotometry yields z phot = 1 . +0 . − . .
29. MMJ 045442.54:
Although there are no radio or mid-IRcounterparts MMJ 045442.54 is identified from its IRAC coloursand the photometry of the corresponding faint galaxy yields z phot = 1 . ± . . c (cid:13)000