ALMA pin-points a strong over-density of U/LIRGs in the massive cluster XCS J2215 at z=1.46
Stuart M. Stach, A. M. Swinbank, Ian Smail, Matt Hilton, J. M. Simpson, E. A. Cooke
DDraft version October 11, 2018
Preprint typeset using L A TEX style emulateapj v. 12/16/11
ALMA PIN-POINTS A STRONG OVER-DENSITY OF U/LIRGS IN THE MASSIVE CLUSTER XCS J2215 AT Z = 1.46 Stuart M. Stach, A. M. Swinbank, Ian Smail, Matt Hilton, , J. M. Simpson E. A. Cooke Draft version October 11, 2018
ABSTRACTWe have surveyed the core regions of the z = 1.46 cluster XCS J2215.9 − ∼
500 kpc ( ∼ (cid:48) ). For six of these galaxies we also obtain CO(2–1) and CO(5–4) line detections, confirming them as cluster members, and a further fiveof our millimetre galaxies have archival CO(2-1) detections which also place them in the cluster.An additional two millimetre galaxies have photometric redshifts consistent with cluster membership,although neither show strong line emission in the MUSE spectra. This suggests that the bulk ( ≥ ∼ Keywords:
Galaxies: clusters: individual: (XMMXCS J2215.9 − INTRODUCTION
Galaxy clusters present a convenient laboratory for thestudy of environmental influences on galaxy formationand evolution due to the large variety in environmentswithin a relatively small observable area, from the high-density cores to the low-density outskirts. Observationalstudies of clusters in the local universe show that theircores are dominated by metal rich, gas-poor early-type(lenticulars, or S0s, and ellipticals) galaxies with littleor no current star-formation activity. In contrast, late-type, actively star-forming disk galaxies are found pref-erentially in the outskirts of clusters and in the surround-ing lower-density field, yielding a so-called “morphology–density” relation (Dressler 1980; Bower et al. 1992; Whit-more et al. 1993; Bamford et al. 2009).This correlation of galaxy star-formation activity andmorphology with environment in the local universe (e.g.Lewis et al. 2002; Gomez et al. 2003; Balogh et al. 2004;Kodama et al. 2004) is suggestive of environmental pro-cesses being at least partly responsible for the quench-ing of star formation in the early-type galaxies in high-density regions. Potential environmental processes whichcould drive this include galaxies interacting with the intr-acluster medium (ICM) causing ram pressure stripping oftheir interstellar gas (Gunn & Gott III 1972), or “stran-gulation,” where the continued accretion of gas from Centre for Extragalactic Astronomy, Department of Physics,Durham University, South Road, Durham, DH1 3LE, UK; email:[email protected] Astrophysics and Cosmology Research Unit, School ofMathematics, Statistics and Computer Science, University ofKwaZulu-Natal, Durban 4041, South Africa Academia Sinica Institute of Astronomy and Astrophysics,No. 1, Sec. 4, Roosevelt Rd., Taipei 10617, Taiwan their surroundings is cut off (Larson et al. 1980); galaxymergers leading to dramatic changes in galaxy’s structureand the triggering of a starburst which rapidly consumestheir gas (Merritt 1983); and tidal interactions, whichcan enhance star formation (Aguilar & White 1985). Ul-timately, each of these processes acts to reduce the gassupply and eventually shut off star formation, and all actpreferentially on galaxies in higher density regions.At higher redshift, it has been shown that the fractionof blue star-forming disk galaxies found in clusters in-creases (Butcher & Oemler Jr 1978; Arag´on-Salamancaet al. 1993). A similar behaviour has also been seenin these clusters using star-formation tracers that areless sensitive to dust extinction, such as mid-infraredemission. Indeed, 24 µ m surveys of actively star-forminggalaxies using the MIPS instrument on the Spitzer SpaceTelescope have found increasing numbers of starburstsin clusters out to z ∼ ∝ (1 + z ) γ with γ ∼ γ ∼ z ∼ a r X i v : . [ a s t r o - ph . GA ] N ov An ALMA survey of U/LIRGs in XCS J2215 at z = 1.46 alent to that of the field (Brodwin et al. 2013; Darvishet al. 2016) and in some z (cid:38) z > µ m emission, which is of-ten used as a star-formation tracer, becomes increasinglyproblematic due to the presence of strong, redshiftedemission from polycyclic aromatic hydrocarbon (PAH)and silicate absorption features which fall in the band.As a result, studies of more distant clusters have fo-cused on the far-infrared/submillimetre wavebands andhave uncovered evidence of a continued rise with red-shift in the activity in overdense regions at z >
1, astraced by an increasing population of the most stronglystar-forming, dusty (Ultra-)Luminous InfraRed Galax-ies (U/LIRGs) (e.g. Webb et al. 2005, 2013; Tran et al.2010; Popesso et al. 2012; Smail et al. 2014; Noble et al.2016). These studies have uncovered mixed evidence ofa reversal in the star-formation–density relation in clus-ter cores at high redshift. For example, a “reversal”has been claimed in some massive clusters at z > ∼ − − z = 1.62. However,these highly star-forming galaxies did not reside in thedensest regions of the cluster and instead the core wasalready populated with passive red galaxies, a trend alsoseen by Newman et al. (2014) in a z = 1.8 cluster, whichhas a cluster core dominated by a quiescent galaxy pop-ulation. These results suggest that a massive quiescentpopulation in some z ∼ − µ m maps obtained by Ma et al. (2015).In this paper, we present Atacama Large Millime-ter/submillimeter Array (ALMA) interferometric obser-vations of dust continuum and CO emission of galaxiesin the central region of XCS J2215. Our observations in-clude a 1.2 mm mosaic of a 500 kpc diameter region en-compassing the central four SCUBA-2 850 µ m sources de-tected by Ma et al. (2015) (hereafter, Ma15). Our ALMAdata provide us with the means to study the U/LIRGpopulation in this cluster in the millimetre at resolutionsan order of magnitude higher than that provided fromcurrent single-dish bolometer cameras and with muchgreater sensitivity. We use our ALMA continuum ob-servations to robustly identify the 850 µ m counterparts.We then searched for emission lines arising from molec-ular gas in cluster members. At the cluster redshift, our ALMA observations in Band 3 and Band 6 covertwo transitions commonly seen in star-forming galaxies: CO(2–1) and CO(5–4). We employ these detectionsto confirm the cluster membership of U/LIRGs seen to-ward the cluster core and to estimate their molecular gascontent and physical properties.This paper is structured as follows: § §
3. We then discuss these in § §
5. We assumea ΛCDM cosmology with Ω M = 0.3, Ω Λ = 0.7 and H = 70 km s − Mpc − , which gives an angular scale of8.5 kpc arcsec − at z = 1.46. We adopt a Chabrier initialmass function (IMF) (Chabrier 2003) and any magni-tudes are quoted in the AB system. OBSERVATIONS AND DATA REDUCTION
XCS J2215.9-1738
XCS J2215 provides an excellent opportunity to studythe nature of star-formation activity in the central re-gions of a high-redshift cluster. At z = 1.46 it is oneof the most distant clusters discovered in X-rays (Stan-ford et al. 2006), with extensive multiwavelength follow-up (Hilton et al. 2007, 2009, 2010; Hayashi et al. 2010,2014). Of particular relevance here is the SCUBA-2 sur-vey of the clusters by Ma15 which discovered an over-density of submillimetre galaxies (SMGs) in its core.Unlike other (proto-)clusters studied at high redshifts(e.g. CLG J0218 Rudnick et al. 2012; Lotz et al. 2013;Hatch et al. 2016), XCS J2215 appears structurally well-developed. By combining XMM-Newton and
Chandra observations, Hilton et al. (2010) (hereafter H10) de-rived an X-ray luminosity of L X = 2.9 +0 . − . × erg s − and an ICM temperature T = 4.1 +0 . − . keV. Employingthe R –velocity dispersion relation of Carlberg et al.(1997); where R is the radius from the cluster centrewithin which the mean density is 200 times the criticaldensity at the redshift of the cluster, H10 used an itera-tive method to estimate a line-of-sight velocity dispersionof σ v = 720 ±
110 km s − from the 31 galaxies with spec-troscopic redshifts within R = 0.8 ± (cid:48)(cid:48) .The velocity distribution of the galaxies, however, didshow signs of bimodality, suggesting that the cluster maynot be a completely relaxed and virialized system.Within the central 0.25 Mpc of the cluster Hayashiet al. (2010, 2014) found 20 [O ii ] emitters with dust-free star-formation rates (SFR) > M (cid:12) yr − . Using Spitzer /MIPS, H10 found a further three bright 24- µ msources with estimated SFRs of ∼ M (cid:12) yr − withinthe central 0.25 Mpc. However, as noted earlier, at z = 1.46 the broad PAH feature at 8.6 µ m and potentialsilicate absorption features are redshifted into the 24 µ mMIPS band, complicating the measurements of SFRsfrom this mid-infrared band. To provide a more robustcensus of luminous dusty galaxies Ma15 obtained sen-sitive, longer wavelength observations with SCUBA-2 at850/450 µ m of XCS J2215. These observations were com-bined with JVLA observations at 1.4 GHz and archivalimages and photometry from Hubble Space Telescope ( HST ), Subaru, and
Spitzer (respectively: Dawson et al.2009; Hilton et al. 2009, 2010) to study the U/LIRGs inthe cluster. Ma15 detected seven submillimetre sources tach et al. Figure 1.
Upper panel : the ALMA 1.25 mm (Band 6) mosaic of XCS J2215 taken from six overlapping pointings covering a 500 kpcdiameter region in the cluster core. We detect 14 > σ continuum detections, demonstrating a very significant overdensity of millimetresources in this region marked by circles and numbered. We list their properties in Table 1. We also overlay the SCUBA-2 850 µ m S/Ncontours from Ma15 starting at 2 σ and increasing in steps of 1 σ (dashed lines showing the equivalent negative contours). Lower panel :a slightly zoomed three-colour
HST image (F125W, F140W, and F160W), with our ALMA detections labelled, showing the rest-frame V -band morphologies of the millimetre sources. We highlight source CO (labelled “CO”) detections from this study or archival CO(2-1) which confirm clustermembership (eleven sources) or photometric redshifts suggesting possible membership (IDs 2, 9)
An ALMA survey of U/LIRGs in XCS J2215 at z = 1.46 above a 4- σ significance cut within R (100 (cid:48)(cid:48) radius),an order of magnitude above the expected blank-fieldcounts. A further four fainter ( > σ ) 850 µ m sources weredetected which were confirmed through counterparts in Herschel /PACS 70 µ m, 160 µ m and MIPS 24 µ m. Theprobabilistic identification of counterparts to these sub-millimetre sources in the mid-IR and radio associated 9of the 11 with galaxies that had spectroscopic or pho-tometric redshifts that suggested that they are clustermembers. The total SFR from these potential U/LIRGcluster members yields an integrated SFR within R of > M (cid:12) yr − , this suggests that XCS J2215 is oneof the highest SFR clusters known at high redshifts (Maet al. 2015).We note that after submission of this paper Hayashiet al. (2017) published an ALMA Band 3 CO(2-1)study of XCS J2215 which overlaps with our observa-tions. Their concentration on just CO(2-1) enabledthem to take deeper integrations over a slightly widerfield and hence allowed them to detect fainter sources;however, where our observations overlap we obtain sim-ilar results.
ALMA Band 6 Observations
We obtained 1.25 mm continuum and simultaneous CO( J = 5–4) observations of the core of the XCS J2215cluster using ALMA covering the core of the cluster in-cluding four of the SCUBA-2 sources identified by Ma15.These Band 6 observations were carried out on 2016 June19 (project ID: 2015.1.00575.S). To cover the CO(5–4)emission lines we set two spectral windows (SPWs) tocover the observed frequencies from 232.7 to 236.4 GHzor ∆V ∼ − which comfortably covers the ex-pected 720 ±
110 km s − velocity dispersion of the clus-ter. A further two SPWs were centred at 248.9 and251.4 GHz, where no visible emission lines are expected,for continuum imaging. Each SPW had a bandwidthof 1.875 GHz with a spectral resolution of 3.904 MHzfor the emission-line SPWs (corresponding to a veloc-ity resolution of 4.97–5.01 km s − ) and a spectral res-olution of 31.250 MHz for the continuum SPWs (37.3–37.6 km s − ). At these frequencies the full-width-half-maximum (FWHM) of the primary beam is ∼ (cid:48)(cid:48) ; there-fore, a mosaic of six pointings was required to map thecentral 500 kpc diameter covering the cluster core (Fig.1). The observations were conducted with forty-two 12 mantennae where the bandpass calibration was obtainedfrom J2258 − − Common Astronomy Software Application ( casa v4.6.0 (McMullin et al. 2007)). The observation used aconfiguration which yielded a synthesised beam in Band6 for the six pointings of ∼ . (cid:48)(cid:48) × . (cid:48)(cid:48)
47 (PA ∼
78 deg.).The resulting continuum maps were created with the clean algorithm using multi-frequency synthesis modewith a natural weighting to maximise sensitivity. Weinitially created a dirty image from the combined SPWsfor each field and calculated the rms noise values. Thefields are then initially cleaned to 3 σ and then mask-ing boxes are placed on sources > σ and the sourcescleaned to 1.5 σ . The six fields were then combined to create a final image for source detection with an rms σ . = 48 µ Jy beam − at its deepest point, shown inFig. 1. ALMA Band 3 Observations
As well as the Band 6 mosaic, we also obtaineda single pointing in Band 3 centered on the clus-ter to cover CO(2–1) emission from gas-rich clus-ter members. These observations were carried out on2015 August 7 using thirty-nine 12 m antennae (projectID:2013.1.01213.S), using J2258 − − − ). At this observing frequency,the FWHM of the primary beam is ∼ (cid:48)(cid:48) and thereforethe central ∼
500 kpc of the cluster (Fig. 1) could be cov-ered in a single pointing. The same reduction approachwas taken for the Band 3 observations as used for theBand 6 data, to create channel maps with a velocity res-olution of 50 km s − and a noise level in each channel of0.3–0.8 mJy beam − . Source Detection
To search for sources in the 1.25 mm continuum mapwe used
Aegean (Hancock et al. 2012) to identify > σ detections. As part of this source extraction, we con-structed a noise map for the mosaic by deriving standarddeviations of the flux density in a box around each pixelwith a size comparable to the synthesised beam. Brightpixels are rejected in each box using a 3 σ clipping toavoid real sources contaminating the noise map. The out-side edge of the ALMA mosaic was then trimmed to thehalf-width half-maximum radius of the primary beamsand source extraction was performed within this regionwhich had a maximum noise of σ rms = 0.09 mJy beam − .Based on this noise map we detect 14 S/N > σ candi-date sources from the Band 6 continuum map shown inFig. 1 and listed in Table 1. All continuum sources havecorresponding K s , i , r and H band counterpartswithin 0 . (cid:48)(cid:48)
5, (Fig. 1 & 2) and to estimate the reliability ofthese detections we perform the same detection routinein the negative source map which yields zero detectionsat > σ .We compare this number of detections with the blank-field 1.2 mm number counts of from the Hubble UltraDeep Field (Aravena et al. 2016; Dunlop et al. 2016) (seealso Oteo et al. (2016)). For sources brighter than a fluxlimit of ∼ ∼ ± ∼ × overdensity of millimetre sources inthe central projected 500 kpc of XCS J2215 seen in Fig. 1.To search for CO emission lines we first adopted atargeted search by extracting spectra from the Band 3and 6 datacubes at the positions of the fourteen 1.25 mmcontinuum detections. In addition, we extract spectraat the positions of the 25 spectroscopic members fromH10, the 46 sources from H09 with photometric redshiftsindicating possible cluster membership and the 20 [O ii ]emitters from Hayashi et al. (2014) that are within thefootprint of the ALMA observations. tach et al. Figure 2.
Thumbnails showing the ALMA Band 6 continuum (top row of each panel), K s (middle row of each panel) and three-colour HST
WFC3 images (1.25, 1.40, and 1.60 µ m, lower row of each panel), of the S/N > K s and hence likely to below mass), suggesting that dynamical interactions may be triggering the strong star formation in these galaxies. Each of these thumbnailsis centred on the positions given in Table 1. Six of the brightest sources in the ALMA continuum map: IDs 3, 6, 7, 8, 11, 12 additionallyyield CO(2–1) and CO(5–4) emission-line detections, while IDs 1, 5, 10, 13, 15 have archival CO(2–1) detections from Hayashi et al.(2017) confirming that these are members of the cluster. Each thumbnail is 6 (cid:48)(cid:48) × (cid:48)(cid:48) with major ticks every 1 . (cid:48)(cid:48) σ contour for the integrated CO(5–4) map across the FWHM of the detected lines, showing thehigh- J CO emission is aligned with the dust continuum.
We detect six significant emission lines in the Band 3data, all corresponding to bright dust continuum sources(IDs: 3, 6, 7, 8, 11, 12 in Fig. 1). One of these sources,ID 6, also has a redshift from H10: z = 1.454, consis-tent with our CO-derived measurement, Hayashi et al.(2017) also report CO(2–1) detections for all six ofthese sources with redshifts consistent with our own. Weidentify all of these lines as CO(2–1) from cluster mem-bers and plot the spectra for these in Fig. 3. Applying thesame procedure on the Band 6 cube yielded significantdetections of CO(5–4) from just the same six sourcesand these are also shown in Fig. 3 and the line emis-sion is contoured over the continuum in Fig. 2 showingthe high- J gas is colocated with the rest-frame 500 µ mdust emission. For the CO(5–4) emission lines, we sub-tracted the continuum emission in the uv data usingthe uvconstsub task in casa and, by averaging data across channels, created continuum-subtracted channelmaps at a velocity resolution of 50 km s − with an rmsof 0.3–0.4 mJy beam − . We find no individually detected CO(2–1) or CO(5–4) emission from any of the othersources in the spectroscopic or photometric samples.For the galaxies where CO emission lines were de-tected, we calculated the intensity-weighted redshift.This was calculated for both the CO(5–4) and CO(2–1) lines, and for both lines for all six sources the derivedredshifts were in excellent agreement, as can be seen inFig. 3. The redshift values reported in Table 2 are themeans of the redshifts derived from the two transitions.Line widths are derived from fitting Gaussian profilesto the binned spectra using scipy.curve fit in
Python weighted by the uncertainty in the spectra.We attempted a blind search for CO emission lines inthe Band 3 and 6 cubes by collapsing them in 300 km s − An ALMA survey of U/LIRGs in XCS J2215 at z = 1.46 . . . . . . Velocity wrt CO line (km s − ) . . . . . . F l u x D e n s i t y ( m J y B e a m − ) − −
902 98 − . . . . . . − − − −
738 262 − − − − − −
81 919 1919 −
204 796 1796
299 1299 2299 . . . . . . Velocity wrt Cluster Redshift (km s − ) Figure 3.
Spectra of the six CO detected cluster members with the top row showing the CO(2–1) emission in Band 3 and the bottomrow the CO(5–4) lines which fall in Band 6. In each case the best-fitting single Gaussian profiles are overlaid. We see detections of linesin both transitions in all six sources, confirming these millimetre sources as gas-rich cluster members. The lower velocity axis is centeredon the peak velocity of the Gaussian fit to the CO(2–1) line whilst the upper velocity axis shows a velocity scale relative to the nominalcluster redshift of z = 1.460. The spectra for IDs 3, 7, 8, 11, 12 are binned to 100 km s − resolution however for ID 6 the data was binnedat 50 km s − due to their narrow line width. The lack of spectral coverage at velocities > − for CO(5–4) IDs 8, 11 and 12 isdue to the edge of the spectral window. The dashed red vertical lines show the velocity at which the CO(2–1) Gaussian’s peak for eachof our six detections. wide bins (similar to the observed FWHM of the targeteddetections) and stepping the bins in 100 km s − incre-ments across a velocity range − σ v < v < +2 σ v where σ was the given velocity dispersion of the cluster from H10. Aegean was used to detect peaks in these collapsedchannel maps looking for > σ detections. This proce-dure recovered the six previously identified line emittersbut did not uncover any additional blind detections. Wedo see hints ( < . σ ) of further CO(5–4) detectionsat the locations and redshifts of two of the CO(2–1)detections from Hayashi et al. (2017) (ALMA.B3.04 andALMA.B3.08) however these are all faint and none ofthem make our selection cut and so they are not consid-ered for the rest of the paper.
MUSE AO Observations
Observations of the central 1 × R = 16.1 mag star located95 (cid:48)(cid:48) from the field center was used to correct for the re-maining atmospheric tip-tilt. In total, we observed thecluster core for 3.5 ks, which was split into four 860 s ex-posures, each of which was spatially dithered by ∼ (cid:48)(cid:48) .Between exposures, the IFU was also rotated through90 degrees to improve the flat-fielding along slices in thefinal datacube. We used the standard spectral range,which covers 4770–9300 ˚A and has a spectral resolutionof R = λ / ∆ λ = 4000 at λ = 9200˚A (the wavelength of the[O ii ] at the redshift of the cluster sample) – sufficient toresolve the [O ii ] λλ esorex pipeline which extracts, wavelength-calibrates, flat-fieldsthe spectra, and forms each datacube. Each observation was interspersed with a flat-field to improve the slice-by-slice flat field (illumination) effects. Sky subtractionwas performed on each sub-exposure by identifying andsubtracting the sky emission using blank areas of sky ateach wavelength slice, and the final mosaics were thenconstructed using an average with a 3 σ clip to rejectcosmic rays, using point sources in each (wavelength col-lapsed) image to register the cubes.For each source in our 1.2 mm catalog, we extract aspectrum by integrating the datacube within a 0 . (cid:48)(cid:48) ii ] emission at the cluster red-shift. For all the galaxies with CO detections, we detectthe [O ii ] emission line with signal-to-noise ranging from5 to 100. We fit the [O ii ] emission doublet and deriveredshifts that agree with the CO (within their 1 σ error)in all cases. We report the [O ii ] emission-line redshiftsin Table 1. ANALYSIS AND RESULTS
Our ALMA survey of the central regions of XCS J2215has resolved the overdensity of four submillimetre sourcesin the core of the cluster into 14 separate 1.25 mm contin-uum sources (Fig. 1). The four brightest 1.25 mm contin-uum components each correspond to one of the SCUBA-2 sources discovered by Ma15: our ID 11 corresponds toSCUBA-2 source CO(2–1)and CO(5–4) emission with redshifts that place themwithin the cluster. In addition, we match the remainingfainter continuum sources to the Hayashi et al. (2017) CO(2–1) catalog, finding five further matches: IDs1, 5, 10, 13, 14. Matching to the H10 redshift cata- tach et al. z = 1.301 making it an in-terloper in the foreground of the cluster and we do notconsider it a cluster member in the following analysis.In the remaining two continuum-selected ALMA sources(ID 2 and 9) we do not detect any emission lines in theMUSE spectra, which covers 4770–9300 ˚A (which corre-sponds to z = 0.28–1.50 for [O ii ] emission). However, wenote that both of these two galaxies in the H09 pho-tometric redshift catalogue have redshifts that are con-sistent with being possible cluster members, althoughthe absence of lines in the MUSE spectra either sug-gests they are highly obscured, or they lie at higher red-shift than z = 1.50. For the 11 spectroscopically con-firmed millimetre-selected cluster members we derive arest-frame velocity dispersion of σ = 1040 ±
100 km s − .This is marginally higher than the σ = 720 ±
110 km s − determined by H10 for the cluster members within thecluster core. This difference is not statistically significant( ∼ σ ), but the sense of the difference is consistent withthe expectation that the millimetre-selected sources arelikely to be relatively recently accreted galaxies whichhave yet to fully virialise.Our ALMA observations provide precise positions forthe sub/millimetre emission and so unambiguously iden-tify the counterparts in the optical and near-infraredwavebands, as shown in Fig. 1 and Fig. 2. Over halfof these sources have companions on scales of ∼ (cid:48)(cid:48) –3 (cid:48)(cid:48) ,although more than half of these are faint or undetectedin the K s band, suggesting they have relatively modeststellar masses. Nevertheless, this is some indication thatclose tidal interactions or minor mergers may be the trig-ger for the starburst activity seen in this population. SED Fitting
We estimate the far-infrared luminosities for each ofour continuum sources by fitting their far-infrared andsubmillimetre photometry using a library of galaxy tem-plate SEDs from Chary & Elbaz (2001); Draine et al.(2007); Rieke et al. (2009). We use our 1.25 mm con-tinuum fluxes along with fluxes from the lower resolu-tion single-dish observations from SCUBA-2 at 450 and850 µ m (Ma et al. 2015) and archival Herschel
PACS dataat 100 and 160 µ m (see Santos et al. 2013). Due to thelow resolution for the single-dish observations, the fluxesfor the individual sources were estimated by deblendingthese maps using the method detailed in Swinbank et al.(2013) using the ALMA detections as positional priors asdescribed in Ma15. We calculate the infrared luminosityfrom integrating the best-fitting SEDs for each galaxy inthe wavelength range 8–1000 µ m and from this derivedthe far-infrared luminosity assuming the sources lie at thecluster redshift (Table 1). The far-infrared luminositiesshow a ∼
50% dispersion at a fixed 1.25 mm flux, but theformal error bars are consistent with a single ratio, andhence there is no strong evidence for a variation in SEDshape within our small sample. In particular, we notethat we obtained CO detections for six of the brightest1.25 mm continuum sources, only three of which fall inthe top five brightest sources based on the far-infraredluminosities. This may indicate that the far-infrared lu-minosities may be less reliable than adopting a singlerepresentative SED model and fitting this just to the 1.25 mm continuum flux. We also caution that if thereare systematic differences in the dust SEDs of galaxies inhigh-density environments, e.g., due to stripping of dif-fuse cold gas and dust components (Rawle et al. 2012),then this will not be captured by the templates in ourlibrary.We next estimate the star-formation rate from the far-infrared luminosities using the Kennicutt (1998) rela-tion and assuming a Chabrier IMF. For the 14 ALMAcontinuum sources we derive L IR in the range of (1.7–9.1) × L (cid:12) and a median (3.6 +2 . − . ) × L (cid:12) whichcorresponds to SFRs of ∼ M (cid:12) yr − (Table 1).The derived luminosities of L IR = 10 –10 L (cid:12) classifythese cluster galaxies as LIRGs with the brightest on theULIRG boundary.Integrating the ongoing star formation in the spectro-scopically confirmed millimetre-selected cluster memberswe derive a total SFR in the central ∼
500 kpc of thecluster of > ∼ M (cid:12) yr − . Including the photometricallyidentified members (but excluding ID 4), this increases to > ∼ M (cid:12) yr − . This is comparable to the total SFR es-timated by Ma15 within R = 0 . ∼ (cid:48)(cid:48) ), eventhough that region is much larger than the extent ofour current ALMA survey of the central R ≤ .
25 Mpc(Fig. 1). Thus, our results reinforces the claims thatXCS J2215 demonstrates a very rapid increase in the SFRdensity in the central regions of clusters out to z ∼ CO Line Properties
To derive CO line properties we fit single Gaussiansto each of the Band 3 and 6 emission spectra, whichappear to provide adequate descriptions of the observedline profiles (Fig. 3). Estimates of the line widths weretaken from the FWHM of the Gaussian fits, and the fluxdensity of the CO lines were determined by integratingthe CO spectrum, I CO = (cid:90) +2 σ − σ I ( v ) dv, (1)where σ was taken from the Gaussian fits. Then,the CO luminosities were calculated using the relationgiven in Solomon & Vanden Bout (2005): L (cid:48) CO = 3 . × S CO ∆ vν − D L (1 + z ) − , (2)where L (cid:48) CO is the line luminosity in K km s − pc , S CO ∆ v is the observed velocity-integrated flux densityin Jy km s − , ν is the observed frequency of the emis-sion line in GHz and D L is the luminosity distance inMpc. The FWHM and CO flux densities for both tran-sitions are given in Table 2. For simplicity in compar-ing to the literature we adopted the same values for theconstants α CO = 1, radius of galaxy R kpc = 7 kpc andthe L CO(2-1)/LCO(2-1) ratio of 0 . ± .
13 from Both-well et al. (2013) when deriving M gas and M dyn . Wethen list the estimated gas masses ( M gas ) for the galax-ies based on their CO(2–1) luminosities and adopting α CO = 1 (following Bothwell et al. 2013). We also listthe dynamical masses ( M dyn ) for a disk-like dynamicalmodel with a 7 kpc radius and the average inclination An ALMA survey of U/LIRGs in XCS J2215 at z = 1.46 Table 1
Properties of the ALMA 1.25 mm continuum detections in XCS J2215ID R.A. Dec. S . L FIR
SFR z p z ∗ s z MUSE (J2000) (mJy) (10 L (cid:12) ) ( M (cid:12) yr − ) (H09)1 22 15 58.75 −
17 37 40.9 0.46 ± +1 . − . +30 − +1 . − . † †† −
17 37 41.9 0.49 ± +1 . − . +20 − +2 . − . ... ...3 22 15 58.54 −
17 37 47.6 0.93 ± +4 . − . +60 − +0 . − . −
17 37 50.5 0.21 ± +2 . − . +30 − +0 . − . −
17 37 50.6 0.37 ± +0 . − . +10 − +0 . − . † −
17 37 53.3 0.68 ± +2 . − . +40 − +0 . − . −
17 37 58.0 0.46 ± +1 . − . +30 − +1 . − . −
17 37 59.0 0.88 ± +0 . − . +10 − +1 . − . −
17 37 59.7 0.28 ± +2 . − . +40 − +0 . − . ... ...10 22 15 57.48 −
17 37 59.9 0.18 ± +0 . − . +10 − +0 . − . † −
17 38 14.5 0.98 ± +1 . − . +20 − +0 . − . −
17 38 16.7 0.60 ± +1 . − . +20 − +0 . − . −
17 38 19.4 0.30 ± +1 . − . +20 − +1 . − . † −
17 38 22.3 0.56 ± +2 . − . +30 − +0 . − . † ∗ Spectroscopic redshifts in bold are from CO emission described in this paper, confirmed non-members are in italics . † CO spectroscopic redshifts from Hayashi et al. (2017). † † There are two galaxies separated by < . (cid:48)(cid:48) ∼ − ,both of which are detetced in CO and [O ii ]. for a population of randomly orientated disks (again fol-lowing Bothwell et al. 2013). We note that the derivedvalues are highly dependant on the value of α CO anddue to the SFRs being lower than the ULIRG sources inBothwell et al. (2013), a more Milky Way-like α CO ∼ . α CO results in gas masses for two out of our six detectionsbeing greater than our calculated dynamical masses. Wenote that our dynamical masses, whilst consistent withindependent stellar mass estimations shown below, arebased on adopting a mean inclination angle for popula-tions of randomly oriented disks and an adopted valuefor the galaxy radius of R kpc = 7 kpc however the twogalaxies with gas fractions > α CO ∼ . >
35 kpc which is again unphysical andan order of magnitude greater than previous size estima-tors of CO-emitting regions (Engel et al. 2010). Thissuggests, that for at least these two galaxies, that a MilkyWay-like α CO ∼ . M gas =1 . ± . × M (cid:12) (or M gas =4.3–10.5 × M (cid:12) for α CO = 4 . M dyn =0 . +0 . − . × R kpc M (cid:12) ( M dyn = 6 +2 − × M (cid:12) for R kpc = 7 kpc) and the median gas fraction is relativelylow at f gas = 0 . ± .
3. We estimate stellar masses for our1.2 mm continuum-selected galaxies using their
Spitzer imaging. In particular, we exploit the archival IRACimaging of XCS J2215 to measure IRAC 3.6 µ m magni-tudes for our sources, deriving a median magnitude of21.1 +0 . − . . At z = 1.45, this corresponds to rest-frame H - band, which is sensitive to the underlying stellar massof a galaxy. We exploit the magphys -derived stellarmasses of comparably luminous submillimeter-selectedgalaxies in the Extended Chandra Deep Field South fromda Cunha et al. (2015) and apply their median rest-frame H -band mass-to-light ratio to our sample. Wederive a median stellar mass of M ∗ = 4 +2 − × M (cid:12) which suggests a median gas fraction for our six sourceswith CO detections of f gas = 0 . ± . α CO = 1 . R kpc = 7 kpc is appropriate.Combining these gas masses with the SFRs, weestimate a median gas consumption timescale of200 ±
100 Myrs which is comparable to the crossing-timeof the cluster core. However, as noted above, this ishighly dependent on the choice of α CO , scaling linearly,therefore if a more Milky Way-like α CO is appropriate,then the consumption timescale increases to ∼
800 Myrs,which is comparable to timescales expected for similarmain-sequence galaxies at this redshift ( ∼ ± CO detec-tions for the remaining millimetre sources with archival CO detections indicate gas masses of < ∼ × M (cid:12) andthis may reduce the median gas consumption timescalefor the whole population (although these fainter mem-bers also tend to have lower SFRs, Table 1).For the two continuum sources without spectroscopicredshifts, from their calculated L IR and based on thescatter in the L (cid:48) CO -L IR relation shown in Fig. 5 it is plau-sible that we might not detect CO(5-4) for ID 9. ForID 2, on the other hand, the combination of its faintcontinuum detection and location close to the edge ofthe ALMA primary beam (see Fig. 1) points to a non-detection possibly being a result of insufficient sensitivity.As we have observations of two CO transitions forour ALMA-identified cluster U/LIRGs, we can deter-mine the ratio of the line brightness temperatures be-tween the CO(5–4) and CO(2–1) transitions. Weshow in Fig. 4 the spectral line distributions (SLEDs) tach et al. Table 2
Emission-line properties for CO(2–1) and CO(5–4) detections in XCS J2215 member galaxiesID I CO(2 − FWHM
CO(2 − I CO(5 − FWHM
CO(5 − M gas M dyn (J km s − ) (km s − ) (J km s − ) (km s − ) (10 M (cid:12) ) (10 M (cid:12) )3 0.5 ± ±
90 1.2 ± ±
40 2.4 ± ±
46 0.25 ± ±
30 0.80 ± ±
20 1.0 ± ± ± ±
120 0.6 ± ±
60 1.3 ± ±
58 0.6 ± ±
90 1.0 ± ±
50 2.2 ± ±
411 0.4 ± ±
90 0.8 ± ±
50 1.6 ± ±
312 0.3 ± ±
90 0.7 ± ±
100 1.5 ± ± for our sources compared to other populations and mod-els from the literature. This shows that the cool inter-stellar medium within our cluster LIRGs is less excitedthan comparably luminous local galaxies, although ithas very similar properties to that seen in high-redshift,submillimetre-selected ULIRGs and BzK galaxies. Toquantify this further, we determine a median value ofthe CO(5–4) and CO(2–1) line brightness ratio forour six sources of r / = 0 . ± .
06. We can comparethis to the value derived for statistical samples of high-redshift, submillimetre-selected ULIRGs from (Both-well et al. 2013) and
BzK s from Daddi et al. (2015): r / = 0 . ± . r / = 0 . ± .
13, which yield r / = 0 . ± .
08 for SMGs and r / = 0 . ± . r / = 0 . ± .
09, which yield r / = 0 . ± .
06 for
BzK s. As expected from Fig. 4, these are in agreementto the values we derive and suggests comparable gas ex-citation in our sample of z = 1.46 cluster LIRGs to themore luminous and typically higher-redshift field SMGsstudied by Bothwell et al. (2013), as well as the less lumi-nous BzK s. This in turn suggests that the r / valuesfor the Bothwell et al. (2013) sample should be broadlyapplicable to our sources. DISCUSSION
Our high-resolution continuum observations withALMA have confirmed and significantly expanded theoverdensity of luminous, dusty star-forming galaxiesknown in XCS J2215. Our data also enable us to sur-vey the cluster for massive gas reservoirs, and we find sixgas-rich systems, associated with the typically brighterdust continuum sources. These CO detections, alongwith five sources that have archival CO detections,unambiguously demonstrate that the majority of thesegalaxies are members of the cluster, while photometricredshifts suggest that two of the remaining continuumsources are also possible members.We can use the CO line properties for our sources tocompare to similar observations of other galaxy popula-tions at high and low redshift to understand their physi-cal properties. Hence, while L (cid:48) CO provides a tracer for themolecular gas content in these galaxies, the FWHM ofthe emission lines provides us with a degenerate tracer ofboth the dynamical mass of the galaxy (narrower FWHMsuggests lower mass) and inclination of the galaxy (nar-rower FWHM suggests a more “face-on” galaxy). InFig. 5(a) we compare the L (cid:48) CO(1 − (converted fromour L (cid:48) CO(2 − detections, adopting r / = 0.84) versusFWHM for our six CO-detected galaxies against a sam-ple of local and high-redshift U/LIRGs. As in Bothwellet al. (2013) we overlay the functional form given in Eq. 3
Rotational Quantum Number J upper . . . . . . . I C O ( J k m s − ) XCS J2215Milky Way inner discBzKs avgSMGs avgLocal U/LIRGsConst. BrightnessNarayanan+14 simBournaud+15 simPapadopoulos+12 model
Figure 4. CO SLEDs for the six XCS J2215 LIRGs compared toSLEDs for other populations from the compilation of Daddi et al.(2015). We see that our cluster LIRGs have SLEDs that peakat higher- J than the Milky Way (Fixsen et al. 1999), indicatingthat the interstellar medium in these galaxies is more excited, al-though less excited than local U/LIRGs (Papadopoulos et al. 2012).Our sources appear to be similar to the submillimetre-selected fieldULIRGs studied by Bothwell et al. (2013) and the less luminousBzK from Daddi et al. (2015). We also show model SLEDs fromthe simulations of Narayanan et al. (2012) and Bournaud et al.(2015), and the toy model of Papadopoulos et al. (2012). Thelatter implies that the interstellar medium is a two-phase mix ofstar-forming and non-star-forming gas, with 10% of its gas in thestar-forming phase. All the SLEDs are normalised to the aver-age J = 1 transition for Daddi et al. (2015) BzK average exceptfor the Milky Way and XCS J2215 SLEDs which are normalisedto the
BzK average J = 2 transition. We note that if environ-mental processing has preferentially removed cool material fromthese galaxies, then their measured SLED will appear to be more“active” than it initially started off with. L (cid:48) CO(1 − = ( V / . R . αG , (3)where V is the FWHM of the line, 1.36 α is the COto gas mass conversion factor, R is the radius of the CO(2–1) emission region and G is the gravitationalconstant. We, again, adopt the values of α = 1 and R = 7 kpc (Bothwell et al. 2013). We see that LIRGsidentified with ALMA in XCS J2215 fall within the scat-ter of the properties of local U/LIRGs, but slightly belowthe BzK s population seen at similar redshifts. They mayalso show a marginally shallower trend than the func-tional form given in Eq. 3, although the latter providesa good fit for the higher redshift and higher luminositySMGs in Bothwell et al. (2013). We stress that the con-version of the line luminosities to CO(1–0) may result0
An ALMA survey of U/LIRGs in XCS J2215 at z = 1.46 FWHM (km s − ) L C O ( − ) ( K k m s − p c ) XCS J2215Hayashi+17 z > < z < log L IR (L (cid:12) ) l og L C O ( − ) ( K k m s − p c ) XCS J2215SMGsLocal U/LIRGsBzKs
Figure 5. (a) Variation of L (cid:48) CO(1 − with FWHM of the line for the cluster LIRGs in this work (converting the CO(2–1) line luminosityand FWHM) compared to: the Hayashi et al. (2017) detections from the same cluster, local U/LIRGs (Downes & Solomon 1998), fieldSMGs from Genzel et al. (2010); Bothwell et al. (2013) and
BzK s from Daddi et al. (2010). The solid line is the relation for L (cid:48) CO(1 − given in Eq. 3. Our cluster LIRGs overlap with the local U/LIRG sample, although they appear to have slightly lower inferred L (cid:48) CO(1 − luminosities, at a fixed line width, compared to submillimetre-selected field SMGs and BzKs at a similar redshift. (b) The observed trendof L (cid:48) CO(5 − with L IR for our six CO-detected cluster LIRGs and comparison samples of local U/LIRGs, SMGs, and
BzK s compiled byDaddi et al. (2015). Our cluster LIRGs show L CO(5 − / L IR ratios consistent within the spread of the local U/LIRG population. in systematic uncertainties between samples and individ-ual sources in Fig. 5.Comparing the line widths for CO(2–1) and CO(5–4) emission lines for individual galaxies in our sample, wederive a median ratio of FWHM /FWHM = 0.7 ± CO(5–4) emission may be more ex-tended than CO(2–1). This is the opposite behaviourto that expected if transitions with lower excitation tem-peratures have larger contributions from cool gas on theoutskirts of galaxies (Papadopoulos et al. 2001; Ivisonet al. 2011; Bolatto et al. 2013). This could reflect en-vironmental influences on the gas disks in these clus-ter LIRGs, with the removal of the more diffuse coolinterstellar medium from their extended disks. Simi-lar environmentally driven stripping of cooler materialwas invoked by Rawle et al. (2012) to explain the ap-parently higher dust temperatures seen in the SEDs ofstar-forming galaxies in z ∼ r / ratio than is cur-rently observed, implying that originally they had a lowerexcitation SLED and a higher cold gas and dust mass andgas fraction.As the far-infrared luminosity traces a galaxy’s SFRand L (cid:48) CO traces its gas content we show the ratio of thesetwo observables for the ALMA-detected population inXCS J2215 in comparison to similar galaxies in the low-and high-redshift field in Fig. 5b. To try to limit theeffect of potential systematic errors we plot the line lu-minosities derived directly from our higher-S/N CO(5–4) detections, L (cid:48) CO(5 − , and compare to similar high- J observations of the other populations (following Daddiet al. (2015)). Again, we see that our sources lie withinthe scatter of the local U/LIRG population, althoughthey lie on the high side of the distribution. In compari-son to the linear fit of the local U/LIRG population, our cluster galaxies show a median increase in L (cid:48) CO(5 − fortheir detected L IR of 48 % ±
12 % (or conversely a deficitin L IR at a fixed L (cid:48) CO(5 − ). One possible explanationfor this trend would be if the far-infrared luminositiesof these sources are underestimated due to a relativepaucity of cold dust (a consequence of the estimates oftheir far-infrared luminosities being driven primarily bythe 1.25 mm flux measurements due to their compara-tively small errors in the SED fitting), due to environ-mental processing, compared to the template populationsused to fit their SEDs (see § Environmental affects on the gas and dust incluster U/LIRGs
Looking at the gas and dust properties of our CO-detected galaxies in XCS J2215 we see several hints whichall may be pointing to a relative paucity of cool gas anddust in these systems: (i) the galaxies typically havelow CO(2–1) luminosities at a fixed FWHM, comparedto field galaxies; (ii) the line width measured from the CO(2–1) is typically smaller than that measured for CO(5–4); (iii) at a fixed CO(5–4) line luminosity,these galaxies have lower inferred far-infrared luminosi-ties (which is driven primarily by 1.25 µ m – rest-frame ∼ µ m – flux) than comparable field galaxies. To iso-late these trends in Fig. 6a we plot L (cid:48) CO /FWHM , aproxy of gas fraction, as a function of redshift. TheXCS 2215 galaxies possess similar L (cid:48) CO /FWHM for boththe CO(2–1) and CO(5–4) transitions in compari-son to similar U/LIRGs in the field taken from literature(Bothwell et al. 2013; Carilli & Walter 2013; Zavala et al.2015; Decarli et al. 2016). While in Fig. 6(b), we con-sider the CO luminosity and FWHM ratio for these twotransitions for the cluster galaxies and a sample of sim-ilar redshift field galaxies. We limit the redshift rangefor the field comparison sample to z = 1–2 to try to re-move evolutionary behaviour such as the increasing size tach et al. (cid:48) CO /FWHM with redshift. Unfortunately, thisleaves us with very few appropriate comparison sourcesas we are constrained by field galaxies in our redshiftrange that have observations in both high- and low- J COtransitions. The CO luminosity and FWHM ratios bothshow a tentative trend that the cluster galaxies are com-paratively poorer in the lower-density, cool CO(2–1)gas and also show a smaller CO(2–1) FWHM, suggest-ing that any deficit may be occurring on the outskirts ofthese galaxies. This would be consistent with the strip-ping of the cool, lower-density gas and dust from thegas disks as a result of an environment process (e.g. rampressure stripping) that leaves the more tightly bound,denser CO(5–4) material relatively untouched (Rawleet al. 2012). However, further observations of low- andhigh- J CO are needed of larger samples of high-redshiftcluster and field galaxies are needed to test this sugges-tion.
Present descendants of cluster U/LIRGs
The final issue we wish to address is, what are the likelyproperties of the present-day descendants of these galax-ies? They are bound in the cluster potential and so theirstellar remnants will reside in a massive cluster of galax-ies at the present day. As we have noted, while thesegalaxies are rapidly forming stars at z = 1.46 and ∼ z ∼ α CO ) this activity is likely to have declined substantiallyas their gas reservoirs are exhausted (this process willbe even quicker if outflows or the further action of en-vironmental processes suggested above accelerate to theremoval of gas). These star-formation events may forma significant fraction of the stellar mass of these systems,up to ∼ M (cid:12) (Ma15), although these estimates arehighly uncertain. Hence, we can conclude that the galax-ies are likely to be massive at the present day and if theirstar formation terminates at z ∼ ∼ z = 1.46 is 9.3 Gyr, and ourexpectation is that the galaxies will rapidly exhaust theircurrent gas supplies (and are unlikely to accrete substan-tial amounts of cold gas from their surroundings). Hence,the stellar populations in their descendants at the presentday are likely to have inferred ages of at least σ Gauss ) can be converted intoan expected velocity dispersion ( σ ) by comparing the ra-tio of our mass estimator for disks with a simple virialequation estimator for a spherical mass distribution, giv-ing us a conversion factor of σ ∼ . σ Gauss . We com-pare the expected properties of these galaxies to those ofsamples of early-type galaxies in local clusters in Fig. 7.We see that most (five out of six) of the CO-detectedLIRGs in XCS J2215 have characteristics similar to those expected for the progenitors of relatively massive early-type galaxies at the present day. CONCLUSIONS
We have analysed ALMA 1.25 mm and 3 mm andMUSE-GALACSI observations of a ∼
500 kpc diameterregion in the core of the z = 1.46 cluster, XCS J2215.Our ALMA observations detect 14 luminous 1.25 mmdust continuum sources within this region (Fig. 1), rep-resenting a ∼ × over-density of sources compared toa blank field. We detect line emission from six of thebrightest of these sources in the 1.25 mm and 3 mm dat-acubes and associate these lines with redshifted CO(5–4) and CO(2–1) transitions (Fig. 3). These lines un-ambiguously identify the millimetre sources as membersof the clusters, while five other continuum sources havearchival CO(2–1) detections that also place them inthe cluster. A further two sources have photometric red-shifts compatible with them being cluster members, butlack spectroscopic redshifts from either ALMA or MUSE(consistent with the expected field contamination in thismap of ∼ > ∼ M (cid:12) yr − in a ∼
500 kpc region,suggested by Ma15’s earlier SCUBA-2 study. Combiningour precise ALMA positions with high-resolution
HST imaging, we see a high fraction of millimetre continuum-selected galaxies with close companions on < ∼ (cid:48)(cid:48) –3 (cid:48)(cid:48) scales(Fig. 2), suggesting that galaxy–galaxy interactions maybe a trigger for their activity, although most of thesecompanions are faint in the K s band, indicating theseare likely to be minor mergers/interactions.We combine the CO(5–4) and CO(2–1) line fluxesfor the cluster LIRGs to derive a median line brightnessratio, r / = L (cid:48) CO(5 − /L (cid:48) CO(2 − = 0 . ± .
06. Thisis comparable to the median ratio estimated for SMGsand
BzK populations at similar and higher redshifts,indicating broadly similar gas excitation in our sampleof z = 1.46 cluster LIRGs to these high-redshift star-forming populations (Fig. 4). We estimate gas masses(assuming α CO = 1) of ∼ × M (cid:12) and a mediangas consumption time-scale of ∼
200 Myrs. This time-scale is comparable to the time for a galaxy to cross thecluster core and so we anticipate that most of these galax-ies will deplete their reservoirs before they exit the regionthey are currently seen in.We also see a possible trend in terms of the gas anddust properties of the millimetre sources compared to z ∼ CO(2–1) luminosities compared to their FWHM thancomparable field galaxies, (ii) the line width measuredfrom the CO(2–1) is typically smaller than that mea-sured for CO(5–4), (iii) at a fixed CO(5–4) line lumi-nosity, these galaxies have lower far-infrared luminositiesthan comparable field galaxies, and (iv) the ratio of the2
An ALMA survey of U/LIRGs in XCS J2215 at z = 1.46 . . . . . z L C O / F W H M ( K k m − s p c ) XCS J2215 CO(2-1)XCS J2215 CO(5-4)Field CO(2-1)Field CO(5-4)
CO(2 1) /FWHM
CO(5 4) L C O ( ) / L C O ( ) XCS J2215Field 1 < z < Figure 6. (a) Variation of L (cid:48) CO /FWHM as a function of redshift. We take XCS J2215 cluster members and compare them to a fieldsample from the literature with published CO(2–1) and CO(5-4) observations. At the cluster’s redshift, the galaxies with line detectionsshow similar L (cid:48) CO /FWHM ratios, within errors, to the field samples at their comparable redshift ( z ∼ L (cid:48) and FWHM for the two transitions that the differences becomeclear. The dashed line represents our redshift cut for sample field galaxies used in the right-hand plot. (b) Plot of the ratios of the COline luminosities against the corresponding FWHM of the two CO transitions. As in the left panel we plot the cluster members fromXCS J2215 and comparison field sources at a similar redshift from the literature which have both low- J CO detections ((2–1) or (1–0)) andhigh- J CO detections ((7–6), (6–5), (5–4), or (4–3)) converted to CO(2–1) and CO(5–4), respectively, using the brightness ratios fromBothwell et al. (2013). We see that the cluster galaxies inhabit the lower left of the plot compared to the small sample of field sourceswith the relevant observations, suggesting that these galaxies may be comparatively poor in lower-density, cool gas in comparison to thefield sample. One possible explanation for this trend is environmental processes stripping of the cooler, less-bound gas from the outskirtsof the galaxies. We overlay an arrow indicating the difference in the median values of the field sample to the cluster galaxies to highlightthe possible transition of a field galaxy to a cluster galaxy and the resulting effect on the low- J CO line properties. . . . . . . log( σ ) A g e ( G y r) XCS J2215Smith et al. (2009)Nelan et al. (2005)
Figure 7.
A plot of the velocity dispersion of local early-typegalaxies to their luminosity-weighted stellar ages, adapted fromNelan et al. (2005). We show the median trend line and dispersionderived by Nelan et al. (2005) and overplot measurements for indi-vidual galaxies in the Shapley Supercluster from Smith et al. (2009)to illustrate the scatter. We plot the velocity dispersions derivedfrom the Gaussian fits to the CO lines for the six CO-detectedmillimetre members in the core of XCS J2215, where their adoptedage is the lookback time to z = 1.46, 9.3 Gyrs. These points there-fore lie where they would appear today if the bulk of their starswere formed in the starburst event we are currently witnessing. CO(2–1) and CO(5–4) CO line luminosities and theFWHM suggest that the cluster galaxies contain a largerfraction of warmer, denser CO(5–4) gas compared tofield galaxies. These trends could be caused by the pref-erential removal of cooler, lower-density material as a re-sult of an environmental process (possibly ram pressurestripping; Rawle et al. (2012)). Larger samples of clustergalaxies are needed to confirm the reality of this trend.Finally, we have demonstrated that these galaxies havesome of the properties of the expected progenitors of the massive, early-type galaxies which dominate the high-density regions of rich clusters of galaxies at the presentday. Specifically, their dynamical masses and stellar agesroughly match those seen in early-type galaxies in localclusters.S.M.S. acknowledges the support of STFC stu-dentship (ST/N50404X/1). A.M.S. and I.R.S. acknowl-edge financial support from an STFC (ST/L00075X/1).I.R.S. also acknowledges support from the ERC Ad-vanced Investigator program DUSTYGAL 321334, anda Royal Society/Wolfson Merit Award. We thankCheng-Jiun Ma, Chian-Chou Chen, Roberto Decarli,Sune Toft, Tomotsugu Goto and Alasdair Thomsonfor their help with the early stages of this project.The ALMA data used in this paper were obtainedunder program ADS/JAO.ALMA tach et al.
HST archives or in pub-lished papers. REFERENCESarchives or in pub-lished papers. REFERENCES