Deep Chandra, HST-COS, and Megacam Observations of the Phoenix Cluster: Extreme Star Formation and AGN Feedback on Hundred Kiloparsec Scales
M. McDonald, B. R. McNamara, R. J. van Weeren, D. E. Applegate, M. Bayliss, M. W. Bautz, B. A. Benson, J. E. Carlstrom, L. E. Bleem, M. Chatzikos, A. C. Edge, A. C. Fabian, G. P. Garmire, J. Hlavacek-Larrondo, C. Jones-Forman, A. B. Mantz, E. D. Miller, B. Stalder, S. Veilleux, J. A. Zuhone
aa r X i v : . [ a s t r o - ph . GA ] A ug Draft version November 13, 2017
Preprint typeset using L A TEX style emulateapj v. 08/22/09
DEEP CHANDRA, HST-COS, AND MEGACAM OBSERVATIONS OF THE PHOENIX CLUSTER:EXTREME STAR FORMATION AND AGN FEEDBACK ON HUNDRED KILOPARSEC SCALES
Michael McDonald , Brian R. McNamara , , Reinout J. van Weeren , Douglas E. Applegate ,Matthew Bayliss , , Marshall W. Bautz , Bradford A. Benson , , , John E. Carlstrom , , ,Lindsey E. Bleem , , Marios Chatzikos , Alastair C. Edge , Andrew C. Fabian ,Gordon P. Garmire , Julie Hlavacek-Larrondo , Christine Jones-Forman , Adam B. Mantz , ,Eric D. Miller , Brian Stalder , Sylvain Veilleux , , John A. ZuHone Draft version November 13, 2017
ABSTRACTWe present new ultraviolet, optical, and X-ray data on the Phoenix galaxy cluster (SPT-CLJ2344-4243). Deep optical imaging reveals previously-undetected filaments of star formation, extending toradii of ∼ × M ⊙ ), young ( ∼ ±
50 M ⊙ yr − . We report a strong detection of O vi λλ > ⊙ yr − ) from the cooling intracluster medium. We confirm the presence of deep X-ray cavities inthe inner ∼
10 kpc, which are amongst the most extreme examples of radio-mode feedback detected todate, implying jet powers of 2 − × erg s − . We provide evidence that the AGN inflating thesecavities may have only recently transitioned from “quasar-mode” to “radio-mode”, and may currentlybe insufficient to completely offset cooling. A model-subtracted residual X-ray image reveals evidencefor prior episodes of strong radio-mode feedback at radii of ∼
100 kpc, with extended “ghost” cavitiesindicating a prior epoch of feedback roughly 100 Myr ago. This residual image also exhibits significantasymmetry in the inner ∼
200 kpc (0.15R ), reminiscent of infalling cool clouds, either due to minormergers or fragmentation of the cooling ICM. Taken together, these data reveal a rapidly evolvingcool core which is rich with structure (both spatially and in temperature), is subject to a variety ofhighly energetic processes, and yet is cooling rapidly and forming stars along thin, narrow filaments.
Subject headings: galaxies: active, galaxies: starburst, X-rays: galaxies: clusters, ultraviolet: galaxies
Electronic address: [email protected] Kavli Institute for Astrophysics and Space Research, MIT,Cambridge, MA 02139, USA Department of Physics and Astronomy, University of Water-loo, Waterloo, ON N2L 3G1, Canada Perimeter Institute for Theoretical Physics, Waterloo, Canada Harvard-Smithsonian Center for Astrophysics, 60 GardenStreet, Cambridge, MA 02138, USA Argelander-Institut f¨ur Astronomie, Auf dem H¨ugel 71, D-53121 Bonn, Germany Department of Physics, Harvard University, 17 Oxford Street,Cambridge, MA 02138 Fermi National Accelerator Laboratory, Batavia, IL 60510-0500, USA Department of Astronomy and Astrophysics, University ofChicago, 5640 South Ellis Avenue, Chicago, IL 60637, USA Kavli Institute for Cosmological Physics, University ofChicago, 5640 South Ellis Avenue, Chicago, IL 60637, USA Argonne National Laboratory, High-Energy Physics Division,9700 South Cass Avenue, Argonne, IL 60439, USA Department of Physics & Astronomy, University of Kentucky,Lexington, KY 40506, USA Department of Physics, Durham University, Durham DH13LE, UK Institute of Astronomy, Madingley Road, Cambridge CB30HA, UK Huntingdon Institute for X-ray Astronomy, LLC D´epartement de Physique, Universit´e de Montr´eal, C.P. 6 128,Succ. Centre-Ville, Montreal, Quebec H3C 3J7, Canada Department of Astronomy and Astrophysics, University ofChicago, 5640 South Ellis Avenue, Chicago, IL 60637, USA Institute for Astronomy, University of Hawaii, 2680 Wood-lawn Drive, Honolulu, HI 96822, USA Department of Astronomy, University of Maryland, CollegePark, MD 20742, USA Joint Space-Science Institute, University of Maryland, College INTRODUCTIONThe hot intracluster medium (ICM) is the most mas-sive baryonic component in galaxy clusters, comprising ∼
12% of the total cluster mass, or ∼ –10 M ⊙ inrich clusters. The bulk of the ICM is very low den-sity ( ≪ − cm − ), and will require ∼
10 Gyr to coolvia thermal Bremsstrahlung radiation. In the centers ofclusters, however, this situation is reversed. The ICMin the inner ∼
100 kpc can reach high enough densitythat the cooling time is short relative to the age of thecluster ( t cool . ⊙ yr − starburst in the central cluster galaxy (for a review, seeFabian 1994). However, despite the fact that roughlya third of all galaxy clusters have short central cool-ing times (e.g., Bauer et al. 2005; Vikhlinin et al. 2007;Hudson et al. 2010; McDonald 2011; McDonald et al.2013b), such massive starbursts are extremely rare (e.g.,McNamara et al. 2006; McDonald et al. 2012b). Thevast majority of clusters which ought to host runawaycooling flows show only mild amounts of star formation(Johnstone et al. 1987; McNamara & O’Connell 1989;Allen 1995; Hicks & Mushotzky 2005; O’Dea et al. 2008;McDonald et al. 2011; Hoffer et al. 2012; Donahue et al.2015; Mittal et al. 2015), suggesting that much of thepredicted cooling at high temperature is not, in fact, oc-curring. Park, MD 20742, USA
By comparing the cooling rate of intermediate-temperature ( ∼ − K) gas (e.g., Bregman et al. 2001;Oegerle et al. 2001; Peterson et al. 2003; Bregman et al.2006; Peterson & Fabian 2006; Sanders et al. 2010, 2011;McDonald et al. 2014a) to the ongoing star formationrate in central cluster galaxies, one can quantify whatfraction of the cooling ICM is able to condense and formstars. These studies typically find that only ∼
1% of thepredicted cooling flow is ultimately converted into stars.Recently, McDonald et al. (2014a) showed that this inef-ficient cooling can be divided into two contributing parts:cooling from high ( ∼ K) to low ( ∼
10 K) temperatureat ∼
10% efficiency, and an additionally ∼
10% efficiencyof converting cold gas into stars. The former term isgenerally referred to as the “cooling flow problem”, andsuggests that some form of feedback is preventing gasfrom cooling out of the hot phase.The most popular solution to the cooling flow problemis that “radio-mode” feedback from the central super-massive blackhole is offsetting radiative losses in theICM (see reviews by McNamara & Nulsen 2012; Fabian2012). Radio jets from the active galactic nucleus(AGN) in the central cluster galaxy can inflate bubblesin the dense ICM, imparting mechanical energy tothe hot gas (e.g., Bˆırzan et al. 2004, 2008; Dunn et al.2005; Dunn & Fabian 2006; Rafferty et al. 2006;Nulsen et al. 2007; Cavagnolo et al. 2010; Dong et al.2010; O’Sullivan et al. 2011; Hlavacek-Larrondo et al.2012, 2014). The ubiquity of radio jets in so-called“cool core clusters” (e.g., Sun 2009) suggests that thetwo are intimately linked, while the correlation of thefeedback strength (e.g., radio power, mechanical energyin bubbles) with the X-ray cooling luminosity providesevidence that this mode of feedback is, on average, suf-ficient to fully offset cooling (e.g., Rafferty et al. 2006;Hlavacek-Larrondo et al. 2012). Further, recent studiesof newly-discovered high- z clusters in the South PoleTelescope 2500 deg survey (Bleem et al. 2015) suggestthat this energy balance between ICM cooling and AGNfeedback has been in place since z ∼ M ∼ . × M ⊙ ) sys-tem is the most X-ray luminous cluster yet discovered(L −
10 keV , = 8 . × erg s − ), with a predictedcooling rate of ∼ ⊙ yr − . However, contrary to thenorm, the central galaxy in the Phoenix cluster harborsa massive starburst ( ∼
800 M ⊙ yr − ; McDonald et al.2013a), a massive reservoir of molecular gas (M H ∼ × M ⊙ ; McDonald et al. 2014c), and a dusty type-2quasar (QSO; McDonald et al. 2012b; Ueda et al. 2013).In McDonald et al. (2012b), we argued that the only fea-sible way to bring such a vast supply of cold gas into thecenter of a rich cluster was via a runaway cooling flow.However, Hlavacek-Larrondo et al. (2014) showed thatradio-mode feedback is operating at a level that ought tooffset radiative cooling in this system, although whetherthis energy has had time to couple to the ICM is uncer-tain. The fact that the central AGN is simultaneously providing strong radiative (quasar-mode) and mechani-cal (radio-mode) feedback suggests that it may be in theprocess of transitioning from a QSO to a radio galaxy,and that the starburst is being fueled by gas that cooledbefore the first radio outburst. However, this interpreta-tion hinges on a single, shallow X-ray observation withthe Chandra X-ray Observatory (10 ks), in which boththe X-ray nucleus and cavities are detected at low signif-icance.In an effort to provide a more complete picture of thissystem, we have obtained deep far-UV spectroscopy andX-ray imaging spectroscopy using the
Hubble Space Tele-scope
Cosmic Origins Spectrograph (HST-COS) and theAdvanced CCD Imaging Spectrometer (ACIS) on the
Chandra X-ray Observatory , respectively. These dataprovide a detailed picture of the young stellar popula-tions, the intermediate-temperature gas (O vi ; 10 . K),and the hot intracluster medium. These new X-raydata represent a factor of >
10 increase in exposure timeover previously-published observations. We have also ob-tained deep optical and radio data via Megacam on theMagellan Clay Telescope and the Giant Meterwave RadioTelescope (GMRT), respectively. Combined, these datawill provide a much more detailed view of the complexinterplay between cooling and feedback in this extremesystem. We describe the reduction and analysis of thesenew data in §
2, along with supporting data at IR andoptical wavelengths. In § § § = 70 km s − Mpc − , Ω M = 0 .
27, and Ω Λ = 0 .
73. We assume z =0 .
596 for the Phoenix cluster, which is based on opticalspectroscopy of the member galaxies (Ruel et al. 2014;Bleem et al. 2015). DATABelow, we summarize the acquisition and reduction ofnew data used in this study, along with a brief summaryof supporting data published in previous works.2.1.
Chandra: X-ray Imaging Spectroscopy
The Phoenix cluster (SPT-CLJ2344-4243) was ob-served by
Chandra in 2011 ( obsid : 13401; PI: Garmire)for a total of 11.9 ks as part of a large X-ray surveyof SPT-selected clusters (PI: Benson). To understandin detail this extreme system, we have obtained an ad-ditional 117.4 ks ( obsid s: 16135, 16545; PIs: McDon-ald, Garmire), resulting in a combined exposure timeof 129.3 ks (see Figure 1) and a total of 88,042 countsin the central 1 Mpc and in the energy range 0.5–8.0keV. All observations were performed with ACIS-I. Thedata from each observation were individually processedin the standard manner, including cleaning and filteringfor background flares, using CIAO 4.6 and CALDB 4.6.3.The X-ray background was modeled using three distinctcomponents. First, the re-scaled ACIS blank-sky back-ground observations provide an adequate representationof both the particle-induced and unresolved cosmic X-ray background. A second component, modeled as a soft(0.18 keV apec ) excess, accounts for Galactic interstel-lar medium emission (Markevitch et al. 2003). Finally,a third component, modeled as a hard (absorbed 40 keV bremss ) excess, accounts for the fact that a larger-than-normal fraction of the CXB may be unresolved in shorterexposures. The latter two components were simultane-ously fit to the on-source spectrum and an off-sourcespectrum extracted from a blank region of the targetfield, at a physical separation of >
100 kpc
Fig. 1.—
Chandra
ACIS-I 0.5–8.0 keV image of the Phoenix clus-ter, representing a total exposure of 129.3 ˙ks. This deep exposurereveals a relatively relaxed morphology with the majority of thecounts being concentrated in the central ∼
100 kpc.
In order to search for structure in the X-ray surfacebrightness, images were made in several bandpasses in-cluding a soft (0.5–2.0 keV) and hard (4.0–8.0 keV)band, which trace the hot ICM and central AGN, re-spectively. To search for faint structure in the soft im-age, we subtract a two-dimensional model comprised ofthree beta functions with a shared center, position an-gle, and ellipticity. The residual image after this modelwas subtracted was smoothed both adaptively, using csmooth , and with fixed-width Gaussians. The for-mer technique provides the highest-quality image overlarge scales, while the latter has a lower (but non-zero)likelihood of artificially creating features out of noise.Thermodynamic profiles were measured by extract-ing on-source spectra in concentric annuli, where westrive for ∼ dsdeproj (Sanders & Fabian 2007; Russell et al. 2008). Depro-jected spectra were background subtracted using off-source regions (the cluster only occupies one of the fourACIS-I chips) and fit over the range 0.5–10.0 keV using xspec (Arnaud 1996). We model the X-ray spectrumwith a combination of Galactic absorption ( phabs ) andan optically-thin plasma ( mekal ), freezing the absorbing http://cxc.harvard.edu/ciao/ahelp/csmooth.html column ( n H ) to the Galactic value (Kalberla et al. 2005)and the redshift to z = 0 . HST-COS: Far UV Spectroscopy
The Cosmic Origins Spectrograph (COS) is a medium-to-high resolution UV spectrograph on HST with a 2.5 ′′ diameter aperture. The broad UV coverage and sensi-tivity of this instrument make it ideal for simultaneouslydetecting coronal line emission (e.g., O vi ) and the stel-lar continuum in young stellar populations. We designedour observational setup (ID 13456, PI: McDonald) to beable to detect O vi λλ , we model the throughput across the aper-ture, yielding a two-dimensional estimate of the through-put over the extent of the central cluster galaxy. Oursetup yields >
75% throughput over the full extent of thestarburst, with >
50% throughput out to the edge of thecentral galaxy. Figure 2 shows our estimate of the two-dimensional throughput map compared to optical andnear-UV images from HST WFC3-UVIS and COS of thecentral cluster galaxy.We use the G160M grating with a central wavelengthof 1577˚A, providing coverage from 1386–1751˚A. This cor-responds to rest-frame 870–1100˚A at z = 0 . β , O vi λλ ii λ ∼
20 ks on-source for each spectrum. This exposuretime was chosen to allow simultaneous modeling of theO vi lines and the UV continuum, to sufficient depth torule out a 1% efficient cooling flow in the absence of anO vi detection.The observed data were binned to ∼ . The re-sulting spectra were modeled using the latest versionof Starburst99 (v7.0.0; Leitherer et al. 1999) . Thesemodels have sufficient spectral resolution ( < < λ < Z ⊙ (Galactic)and Z ⊙ / β , O vi λλ ii λ χ . Thebest-fitting models are shown in Figures 3 and 4, andwill be discussed in § § Fig. 2.—
Left panel: HST WFC3-UVIS image of the central galaxy in the Phoenix cluster, from McDonald et al. (2013a). Middle panel:HST-COS near-UV image on the same scale, showing that the UV continuum comes from scales larger than the central AGN. Right panel:Combined spectroscopic throughput from two pointings of HST-COS. White contours show the HST WFC3-UVIS i -band imaging, whichis well covered by the COS apertures. MegaCam Optical Imaging
The Phoenix cluster (SPT-CLJ2344-4243) was ob-served on 30 August 2013 with the Megacam instrumenton the Clay-Magellan telescope at Las Campanas Ob-servatory, Chile (McLeod et al. 1998). Exposures weretaken in g (4 ×
200 sec), r (4 ×
600 sec + 2 ×
120 sec), and i (4 ×
400 sec) passbands in standard operating mode andcover the inner 12 ′ (5 Mpc) of the cluster. For this workwe consider only the inner ∼ ′ – the remaining datawill be published in an upcoming weak lensing study(Applegate et al. in prep). Images were processed andstacked using the standard Megacam reduction pipelineat the Smithsonian Astrophysical Observatory (SAO)Telescope Data Center. Color photometry was calibratedusing Stellar Locus Regression (SLR; High et al. 2009).For a more detailed description of the data reduction andcalibration, the reader is directed to High et al. (2012).2.4. Giant Meterwave Radio Telescope: 610 MHz
Giant Metrewave Radio Telescope (GMRT) 610 MHzobservations of SPT-CL J2344–4243 were taken on June14 and 15, 2013. The total on source time resulting fromthese two observing runs was about 10 hrs, with a usablebandwidth of 29 MHz. The data were reduced with theAstronomical Image Processing System (AIPS), Parsel-Tongue (Kettenis et al. 2006) and Obit (Cotton 2008),following the scheme detailed in Intema et al. (2009).Reduction steps include flagging of radio frequency in-terference (RFI), bandpass and gain calibration. Fol-lowing that, several cycles of self-calibration were car-ried out to refine the calibration solutions. Direction-dependent gain solutions were then obtained towardsseveral bright sources within the field of view. The datawas imaged by dividing the field of view into smallerfacets (Perley 1989; Cornwell & Perley 1992). This cor-rects for the non-coplanar nature of the array and thedirection-dependent calibration. For more details aboutthe data reduction the reader is referred to Intema et al.(2009) and van Weeren et al. (2014).2.5. Supporting Data: UV, Optical, IR We also include in this paper optical multi-band imag-ing from HST WFC3-UVIS (McDonald et al. 2013a) andoptical imaging spectroscopy (McDonald et al. 2014c).For a detailed description of a specific dataset, we di-rect the reader to the aforementioned papers. Relevantfeatures of the various datasets are summarized below.Optical imaging was acquired on the cluster corefor McDonald et al. (2013a) in five WFC3-UVIS bands:F225W, F336W, F475W, F625W, and F814W. Whilerelatively shallow, these data provided the first resolvedview of the starburst in the central cluster galaxy. Ad-ditionally, large-area ground-based data at z -band wasobtained with MOSAIC-II on the Blanco 4-m telescope(for details, see Song et al. 2012; Bleem et al. 2015).Optical imaging spectroscopy of the central galaxy wasobtained from Gemini-S GMOS in IFU mode, and pre-sented in McDonald et al. (2014c). These data spanrest-frame 3000–6000˚A, and reveal complex emission-linenebulae in and around the central galaxy. The morphol-ogy of this gas is, for the most part, consistent with thenear-UV morphology. To allow a direct comparison ofHST-COS and GMOS spectroscopy, we extract aperturespectra from the GMOS IFU data, matching the COSthroughput as a function of radius within the aperture.This allows us to simultaneously model the UV and op-tical continuum, as well as compare measured emissionline fluxes in the UV and optical. RESULTSBelow, we summarize the main results that emergefrom the analysis of these deep UV, X-ray, optical, andradio data. We defer a discussion of these results in thegreater context of the cluster’s evolution to § Dust and Young Stellar Populations in the BCG
Using aperture-matched spectra from both HST-COSand Gemini-S GMOS (McDonald et al. 2014c), com-bined with broadband photometry spanning the spectralgap at 1200–3500˚A (McDonald et al. 2013a), we can con-strain the relative contribution of young and old stellarpopulations to the UV+optical continuum, along withthe effects of reddening on the spectrum over nearly anorder of magnitude in wavelength. In Figure 3, we show
Fig. 3.—
UV-through-optical spectrum of the central galaxy in the Phoenix cluster. The optical IFU spectrum, from McDonald et al.(2014c), and broadband photometry, from McDonald et al. (2013a), have been aperture matched to the far-UV HST-COS spectrum basedon Figure 2. Thin colored curves show the best-fit stellar population model reddened by a variety of Galactic extinction models fromCardelli et al. (1989), while thick orange and purple lines show the best-fit models using extinction models from Calzetti et al. (1994).The featureless, grey extinction model from Calzetti et al. (1994) provides a better match to the spectrum, which appears to lack thecharacteristic 2175˚A absorption feature. The best-fit stellar population model consists of a highly-reddened, young (4.5 Myr), metal-poorstarburst. A continuous starburst model provides a qualitatively similar fit, although the strength of the Balmer jump (inset) appears tobe more consistent with a recently-quenched starburst. the full UV+optical spectrum for the combined area ofthe two HST-COS apertures (Figure 2). We overplota series of best-fitting model spectra, comparing differ-ent reddening models (Cardelli et al. 1989; Calzetti et al.1994) and star formation histories. All models include avariable-mass old (6 Gyr) stellar population. We notethat the extinction curve from Cardelli et al. (1989) wascalibrated using observations of stars in our galaxy andthe Magellanic clouds, while Calzetti et al. (1994) wascalibrated on nearby starburst and blue compact galax-ies. We find, as did Calzetti et al. (1994), that a rela-tively gray, “feature-less” extinction curve provides thebest fit to the data, suggesting that the size distributionof the dust is skewed towards larger grains. The dataalso suggest a lack of the characteristic 2175˚A bump,which is the strongest absorption feature in the interstel-lar medium but tends to be missing in starburst galaxies(Calzetti et al. 1994). Fischera & Dopita (2011) suggestthat high levels of turbulence can both flatten the cur-vature of the extinction law and wipe out the 2175˚A ab-sorption feature. This is consistent with McDonald et al.(2014c), where we show that the ionized gas has signifi-cant velocity structure ( h σ v i ∼
300 km s − ) and is con-sistent with being ionized primarily by shocks.Assuming the flatter extinction curve of Calzetti et al.(1994), we find an overall good fit ( χ dof = 1 .
93) to thedata spanning rest-frame 800–6000˚A. The best-fittingmodel (Figure 3; purple curve) is a 4.5 Myr-old starburstwith a total zero-age mass of 2 . × M ⊙ . This cor-responds to <
1% of the total stellar mass of the cen-tral galaxy.
The fit quality is only slightly reduced( χ dof = 1 .
97) if we assume a continuous star formationhistory, with the best-fit model representing a ∼
317 M ⊙ yr − over the past &
15 Myr (Figure 3; orange curve).These two models disagree on the strength of the Balmer jump at ∼ ∼ < λ rest < ∼ § Z = Z ⊙ / ∼
5) strongerabsorption lines which are not present in the observedspectrum. We stress that, despite the extremely high starformation rate inferred by the UV spectrum, the youngpopulation constitutes <
1% of the total stellar mass inthe central galaxy. This is consistent with our picture ofBCG formation, since this mode of star formation must be suppressed to prevent central cluster galaxies fromgrowing too massive (and luminous) by z ∼ ∼ . Coronal Emission: O vi λλ Fig. 4.—
Far-UV spectra for the two HST-COS pointings de-scribed in § β , O vi λλ vi ] λ ii λ < The emissivity curve of O vi λλ ∼ . K, making it an excellent probe of gas at in-termediate temperatures between the “hot” ( > K,probed by X-ray emission lines) and “warm” ( ∼ K,probed by optical emission lines) phases. Emis-sion from O vi has been detected in several nearbygalaxy clusters (Oegerle et al. 2001; Bregman et al. 2006;McDonald et al. 2014a), allowing an independent esti-mate of the ICM cooling rate. Here, we attempt a simi-lar analysis to these earlier works on the central ∼
15 kpcof the Phoenix cluster (see Figure 2).In Figure 4 we show the red-side UV spectra for ourtwo HST-COS pointings. These spectra are modeled us-ing the best-fitting young stellar population described in § § β ,O vi λλ ii λ ∼ Fig. 5.—
High-ionization UV-optical emission line ratios forthe combined HST-COS apertures shown in Figure 2. Overplot-ted are model expectations for a pure cooling flow (Ferland et al.1998; Chatzikos et al. 2015), photoionization from a dust-free AGN(Groves et al. 2004), and radiative shocks (Allen et al. 2008). Inred we show the measured line ratios from the combined HST-COS and Gemini GMOS spectra shown in Figure 3. The blue andgreen points show the residual when photoionization from youngstars and a central AGN have been removed, respectively. Theincreasingly-large errorbars represent our uncertainty in the detailsof these photoionization models. In purple we show the expectedshift if a 5,000 M ⊙ yr − cooling flow were subtracted from thesedata, demonstrating the inability of these data to adequately con-strain the cooling rate of the ICM. that they are empirically calibrated and lacking the fullpopulation of stars included in synthetic optical spectra(see Leitherer et al. 2014). We note that He ii λ ii λ vi λλ f OV I = 2 . ± . × − erg s − cm − and f OV I = 2 . ± . × − erg s − cm − for the north-ern and southern apertures, respectively. The placementof these apertures was chosen so that the spectra couldbe added (i.e., the combined throughput never sums to >
1; see Figure 2), meaning that the total O vi lumi-nosity in the central ∼
15 kpc of the Phoenix cluster isL
OV I = 7 . ± . × erg s − . For comparison,Bregman et al. (2006) found L O vi ∼ × erg s − for the Perseus cluster, using UV spectroscopy from theFUSE satellite.Under the naive assumption that 100% of theO vi λλ ∼ . K, we can estimate the ICM cooling rateat intermediate temperatures. Using the latest cloudy code (Ferland et al. 1998; Chatzikos et al. 2015), and as-suming initial plasma properties matched to the mea-sured values in the inner ∼
150 kpc of the cluster ( Z =0 . Z ⊙ , kT = 10 keV), we find ˙M O vi = (L O vi / . × erg s − ). For the measured luminosity quoted above, thiscorresponds to a cooling rate of 55,000 M ⊙ yr − . How-ever, there are additional sources of ionization that maybe dominating the line flux here, specifically photoioniza-tion from the central AGN and heating from shocks. InMcDonald et al. (2014c) we show that both of these ion-ization sources are contributing to the line flux in high-ionization lines such as [O iii ] λ ii λ ′′ wide aperture centered on the nu-cleus. Assuming a range of AGN photoionization models(Groves et al. 2004) to estimate the O vi emission, basedon the observed He ii λ vi emission coming from thenucleus, without having any actual spatial informationfrom the COS spectroscopy. The inferred contributionto each line from the AGN was subtracted, allowing usto estimate AGN-free line ratios in the optical and UV..The resulting line ratios, which have both stellar andAGN photoionization removed (Figure 5, green points),remain consistent with radiative shocks with v ∼
300 kms − (Allen et al. 2008). The large error bars, resultingfrom our combined uncertainty in the stellar and AGNphotoionization fields, are consistent with the full rangeof magnetic fields tested by Allen et al. (2008). This isconsistent with our earlier work (McDonald et al. 2014c),in which we argued that the warm (10 K) gas is predom-inantly heated by radiative shocks, based on multiple op- tical line ratio diagnostics (e.g., [O iii ]/H β , [O ii ]/[O iii ],as well as the gas kinematics. In this earlier work, weshowed evidence for a high-velocity, highly-ionized plumeof gas extending north from the central AGN, along thesame direction as our two COS pointings. Such highly-ionized, high-velocity signatures are typically not ob-served in low- z clusters, where the warm gas tends tobe only weakly ionized.Figure 5 demonstrates that, even after removing ion-ization contributions from young stars and AGN, the ob-served high-ionization line ratios are inconsistent witha pure cooling flow model from cloudy (Ferland et al.1998; Chatzikos et al. 2015). However, we demonstratethat including a 5000 M ⊙ yr − cooling flow to the model(representing only ∼
10% of the observed O vi flux, purplepoint in Figure 5) does not significantly change the ob-served line ratios, which remain consistent with shocks.Further, we expect additional ionization from both par-ticle heating (e.g., Ferland et al. 2009) and mixing (e.g.,Fabian et al. 2011). The former likely contributes ata low level throughout the cluster core. Given thatthe radio flux of the AGN is fairly typical of BCGswith significantly less star formation, we expect parti-cle heating to be negligible compared to other ionizationsources in the extreme environment of the Phoenix clus-ter. On the other hand, mixing of hot and cold gas islikely contributing significantly given the abundance ofmultiphase gas in this system. Mixing of the hot andcold gas ought to result in an intermediate-temperaturephase, which would likely be near ∼ K (Fabian et al.2011). This would lead to substantial O vi emission forrelatively low rates of net cooling from the hot to coldphase. Thus, we conclude that, as a result of the myriadof additional high-ionization sources, including shocks(e.g., McDonald et al. 2012a), mixing (e.g., Fabian et al.2011), and particle heating (e.g., Ferland et al. 2009), weare unable to constrain the properties of the cooling ICM,which is contributing negligibly to the total O vi flux.Additional high-ionization lines such as O vii or Fe xiv would improve these constraints, perhaps allowing an es-timate of the amount of gas cooling through ∼ K.3.3.
Star-Forming Filaments on 100 kpc Scales
In McDonald et al. (2013a) we presented high-angular-resolution broadband imaging of the cluster core, re-vealing complex filaments of star formation on scales of ∼
40 kpc. While providing unmatched angular resolution,these data were relatively shallow ( ∼ g , r ) and a redder band( z ) from the MOSAIC-II camera on the Blanco 4-mtelescope. In the cluster core, the new data showthat the star-forming filaments initially identified withHST (McDonald et al. 2013a) extend significantly fur-ther than previously thought. To the south, a pair offilaments extend for ∼
40 kpc each, while to the north-west a third filament extends for ∼
65 kpc. The mostextended filament extends due north from the centralgalaxy for ∼
100 kpc (highlighted in Figure 6). This is
Fig. 6.—
Optical g, r, z image of the inner region of the Phoenix cluster (SPT-CLJ2344-4243). The central cluster galaxy is located atthe center of the image. Extending radially from this galaxy are several blue filaments, likely sites of ongoing star formation. We highlightthe longest filament to the north, which is extended for ∼
100 kpc. In the lower right we show a zoom-in of the central galaxy, based onhigher-resolution HST data in the same bandpasses. These data were originally presented in McDonald et al. (2013a) and reveal a complex,filamentary morphology in this starburst galaxy. the most extended star-forming filament yet detected ina cool core cluster, exceeding the well-studied filamentsin the nearby Perseus (60 kpc; Conselice et al. 2001;Canning et al. 2014), Abell 1795 (50 kpc; Cowie et al.1983; McDonald & Veilleux 2009), and RXJ1532.9+3021(50 kpc; Hlavacek-Larrondo et al. 2013) clusters. Thesefour filaments, particularly the northern pair, are excep-tionally straight, similar to the northern filament in thePerseus cluster. This morphology has been used to arguefor relatively low turbulence in the core of the Perseuscluster(e.g., Fabian et al. 2008), although counter argu-ments can be made on the basis of ICM density fluctua-tions (e.g., Zhuravleva et al. 2015).The filamentary emission is observed in both the g and r bands for all four filaments, providing preliminary evi-dence that this is continuum, rather than line, emission.At the rest frame of the Phoenix cluster, the g -bandspans 2400–3300˚A. This wavelength range is relativelyfree of strong emission lines, with Mg ii λ v &
500 km s − ) ra-diative shocks in a dense ( n &
100 cm − ) medium.The total amount of rest-frame UV emission in theouter filaments – those that were not included inMcDonald et al. (2013a) – is relatively small. For ex-ample, the outer 50% of the northern filament (50–100 kpc) contributes ∼ ∼ X-ray Surface Brightness Maps " k p c S m oo t hed ( . ") " k p c Un s ha r p M a sk ed Fig. 7.—
Upper panel: Gaussian smoothed (FWHM = 2 ′′ ) 0.5–2.0 keV image of the central ∼
100 kpc of the Phoenix cluster. Thisimage clearly shows the pair of cavities ∼
10 kpc to the north andsouth of the X-ray peak, despite the lack of any additional process-ing. Lower panel: Unsharp masked image of the same region asabove. This residual image highlights the small-scale structure inthe inner region of the cluster, showing a pair of highly-significantcavities to the north and south and overdense regions to the eastand west.
In Figure 7 we show a smoothed 0.5–2.0 keV imageof the central ∼
100 kpc in the Phoenix cluster. Withoutany additional processing, the pair of cavities in the inner ∼
10 kpc, reported initially by Hlavacek-Larrondo et al.(2014), are evident. An unsharp masked image of thecore (lower panel of Figure 7) reveals significant structurein the inner ∼
30 kpc, with overdense regions surroundingthe pair of cavities. The cavities, located ∼
10 kpc to thenorth and south of the central AGN, are detected at asignificance of ∼ σ in this residual image. The unsharpmask technique is only sensitive to substructure on asingle angular scale, so it is unsurprising that we detectno large-scale asymmetries beyond the core region in thisresidual image.In order to look for ICM substructure on multiple an- gular scales, we model the 0.5–2.0 keV surface bright-ness map with a sum of two beta models and a con-stant background, using sherpa . The two beta mod-els, meant to represent the core and outer ICM, sharea common center, ellipticity, and position angle. Wenote that, since the X-ray spectrum of the central AGNis highly obscured (McDonald et al. 2012b; Ueda et al.2013), there is no central point source detected in thissoft energy band. During the fit, the cavities (Figure7) and point sources were masked to prevent any spuri-ous positive or negative residuals. The best-fitting modelwas subtracted, with the residual images shown in Fig-ure 8. This residual image shows a significant amountof structure in the central ∼
150 kpc. On scales largerthan ∼
200 kpc (0.15R ) there is little structure in theresidual image (see rightmost panel in Figure 8). If thecluster had recently experienced a recent minor merger,one would expect to find large-scale spiral-shaped surfacebrightness excesses that would be visible on >
100 kpcscales in the residual image (e.g., Roediger et al. 2011;Paterno-Mahler et al. 2013). For comparison, the large-scale spiral feature in Abell 2029 (Paterno-Mahler et al.2013) represents an excess surface brightness of 22% at aradius of ∼
150 kpc. At similar radii in the Phoenix clus-ter, the surface brightness uncertainty in the 0.5–2.0 keVimage is ∼ minor merger.In general, the cluster appears relaxed – an observationcorroborated by Mantz et al. (2015), who find a markedabsence of any large-scale “sloshing” features in this sys-tem based on an independent analysis, labeling it as oneof the most relaxed clusters in the known Universe.Directly to the northwest and southeast of the X-raypeak there are a pair of cavities, detected with S/N = 25,and consistent in location and size to those reported inHlavacek-Larrondo et al. (2014). The northern cavityappears to be “leaking” to the west, while the southerncavity is slightly extended to the east. If these cavitiesare indeed not confined to a simple bubble morphology,then the estimates of the AGN power derived in previousworks (Hlavacek-Larrondo et al. 2014) are significantlyunderestimating the full power output of the AGN. In-terestingly, the four most extended, linear filaments ofyoung stars (Figure 6) appear to trace the AGN outflow,with the northern and southern cavities occupying thespace between the two filaments in the respective direc-tion. This may be due to the expansion of the cavitycompressing the surrounding gas, leading to more rapidcooling along the edges of these bubbles.We note that these two X-ray cavities are at a com-mon distance of ∼
18 kpc from the X-ray peak, which isrelatively small given the massive size of the cool core( ∼ ∼ http://cxc.harvard.edu/sherpa4.4/0
18 kpc from the X-ray peak, which isrelatively small given the massive size of the cool core( ∼ ∼ http://cxc.harvard.edu/sherpa4.4/0 Fig. 8.—
Residual X-ray surface brightness images (0.5–2.0 keV) after a double beta model (shared center, ellipticity, position angle)has been subtracted. Left: Residual image smoothed with a fine kernel (Gaussian FWHM = 2.2 ′′ ). This image highlights the negativeresiduals (red) near the central AGN (white cross) which are assumed to be X-ray cavities. The dashed radial lines depict the locationsof the most extended star-forming filaments (Figure 6). This panel demonstrates that the most extended star-forming filaments appear toavoid the central X-ray cavities. There are also three regions of enhanced (blue) surface brightness, surrounding the X-ray cavities, whichmay be sites of enhanced cooling. Middle: Residual image smoothed with a medium kernel (Gaussian FWHM = 3.5 ′′ ). This image iszoomed out by a factor of 3.2 to highlight large-scale features, such as the spiral-shaped overdensities (blue) and the asymmetric centralcavities (red). Right: Residual image smoothed with a coarse kernel (Gaussian FWHM = 7 ′′ ). This image is zoomed out by an additionalfactor of 2, to show the relative lack of structure on scales larger than ∼
200 kpc (0.15R ). We highlight a potential set of “ghost cavities”(e.g., McNamara et al. 2001) to the north and south, ∼
100 kpc from the cluster center. These may be the remnants of a prior epoch ofstrong radio-mode feedback, although they are only marginally detected (see Table 2). to the surrounding ICM. We will return to this line ofreasoning in § >
100 kpc (see central panel of Figure 8). Themorphology of this overdense region is reminiscent of aninfalling cloud of cool gas. Alternatively, the fact thatboth the northern bubble and the northern overdensityshare the same curvature suggests that this may sim-ply be the outer rim of a bubble expanding in a dense,cool core. To classify these with more certainty, we re-quire deep enough X-ray data to produce temperatureand metallicity maps.On larger scales, there may be a pair of “ghost” cavitiesto the north and south of the cluster center. These arelocated at distances of ∼
100 kpc from the X-ray peak andare detected most readily in a heavily-smoothed residualimage (right panel of Figure 8). We will return to adiscussion of the significance of these cavities in § Radio Jets, Ionized Outflows, and X-ray Cavities
In McDonald et al. (2014c), we present evidencefrom spatially-resolved optical spectroscopy of a highly-ionized outflow to the north of the central cluster galaxy.This outflow was identified via the presence of a peak in
Fig. 9.—
This figure shows the smoothed residual maps from Fig-ure 8 with 600 MHz radio contours overlaid in white. The beamshape of the 600 MHz data is shown in the upper right corner.These contours highlight radio emission centered on the X-ray nu-cleus and elongated in the north-south direction, in the same direc-tion as the most significant set of cavities. At large radii, the radioemission appears to bend in the direction of the “leaky” cavities.The extended radio emission to the southeast appears to coincidewith the most extended cavity, at a distance of ∼
100 kpc. Wealso highlight the direction of the ionized wind (inset), discussedin McDonald et al. (2013a), with a yellow arrow. high-ionization optical emission lines (e.g., [O iii ], He ii )roughly 20 kpc north of the galaxy nucleus. In Figure 91 TABLE 1Deprojected Thermodynamic Profiles from X-ray Spectra
Radii [kpc] kT [keV] P [keV cm − ] K [keV cm ] Z [Z ⊙ ]7– 27 6.1 +1 . − . +0 . − . +3 . − . +0 . − .
27– 54 8.0 +0 . − . +0 . − . +3 . − . +0 . − .
54– 108 9.9 +1 . − . +0 . − . +9 . − . +0 . − . +1 . − . +0 . − . +26 . − . +0 . − . +4 . − . +0 . − . +294 − +0 . − . +7 . − . +0 . − . +1350 − +0 . − . Note . — The central 1 ′′ has been clipped due to the presence ofan X-ray point source (central AGN). (inset), we show that this highly-ionized gas lies in thesame direction as the northwestern cavity in the X-rayemission. This may indicate that strong AGN feedback,which is excavating the central cavities, is also heatingthe multiphase gas, presumably via shocks. This is sim-ilar to what is observed in Abell 2052 (Blanton et al.2011).Figure 9 also shows the extended morphology of the600 MHz radio data from GMRT. These data show emis-sion centered on the X-ray peak and central galaxy, ex-tended in the same north-south direction as the cavi-ties. Interestingly, the extended emission spreads to thesoutheast and northwest, similar to the “leaking” cavi-ties shown in Figure 8. The most extended emission inthe southeast appears to trace the morphology of the po-tential “ghost” cavities, identified in the previous section(see right panel of Figure 8). This low-significance cav-ity may, instead, be an extension of the primary southerncavity, with the radio emission leaking out to the south-east in a similar manner to the northern cavity. Thisseems to be the case based on the central panel of Figure8. Alternatively, the presence of isolated cavities at largeradii may signal two distinct bursts of AGN feedback,separated by a relatively small amount of time. We willdiscuss these scenarios, among others, in § Thermodynamics of the ICM
In Figure 10 and Table 1 we show the results of ourX-ray deprojection analysis (see § ∼ r ∼
500 kpc, 0.4R ) Fig. 10.—
Thermodynamic profiles from X-ray spectroscopy.Spectra are deprojected using dsdeproj (Sanders & Fabian 2007;Russell et al. 2008) and fit with a combination of absorption fromneutral gas ( phabs ) and plasma emission ( mekal ). Uncertaintyregions are shown with diamonds. For comparison, we show ex-pectations (where available) based on averages of low- and high- z clusters in similar mass ranges. The central temperature, pressure,and entropy in Phoenix are at the extreme end for low- z clusters,and are even more extreme compared to other clusters at z > . z (Vikhlinin et al. 2006; Pratt et al. 2007; Baldi et al.2007; Leccardi & Molendi 2008b) and high- z clusters(Baldi et al. 2012; McDonald et al. 2014b), which typi-cally peak at 0.3–0.4R . The factor-of-three drop inthe temperature in the inner region is on the high endof what is observed in massive clusters (Vikhlinin et al.2006; McDonald et al. 2014b).The central pressure (1.1 keV cm − ) is exceptionallyhigh – the highest measured in any cluster to date –consistent with the expectation for a strong cool coreembedded in one of the most massive clusters known.The pressure drops off rapidly with radius, consistentwith the expectation from the universal pressure pro-file (e.g., Arnaud et al. 2010; Planck Collaboration et al.2013; McDonald et al. 2014b). The pressure profile inthe Phoenix cluster appears more consistent with z ∼ z ∼ . r .
30 kpc) entropy is 19.2keV cm , consistent with what is found for nearby coolcore clusters (e.g., Cavagnolo et al. 2009; Hudson et al.2010). The entropy profile shows no evidence of ex-cess entropy (e.g., Cavagnolo et al. 2009) in the inner re-gion, following the “baseline” profile of Voit et al. (2005)from the central bin to the outermost bin at ∼ z and high- z systems (e.g., Bautz et al. 2009; Walker et al. 2013;Reiprich et al. 2013; Urban et al. 2014; McDonald et al.2014b).The measured metallicity profile is consistentwith the declining profile found in low- z clus-ters (e.g., De Grandi et al. 2004; Baldi et al. 2007;Leccardi & Molendi 2008a). We note that, giventhe large uncertainties, the metallicity profile is alsoconsistent with a constant value as a function of radius.The ratio of the cooling time to the freefall time( t cool /t ff ) in the cores of nearby clusters correlates withthe presence of multiphase gas, presumably signaling alink between the warm and hot phases (McCourt et al.2012). Following Gaspari et al. (2012), we estimate theratio of the cooling time to the freefall time in the coreof Phoenix using the following equations: t cool = 32 ( n e + n i ) n e n i Λ( T, Z ) (1) t ff = (cid:18) rg ( r ) (cid:19) / (2)We estimate g(r) assuming that the core is in hydrostaticequilibrium, and the cooling function (Λ( T, Z )) follow-ing Sutherland & Dopita (1993), assuming solar metal-licity and n i = 0 . n e . The resulting t cool /t ff profile isshown in Figure 11. The minimum value of t cool /t ff ∼ ∼
20 kpc from the central AGN, which corre-sponds to the radius within which the bulk of the starformation is contained (see Figure 6 and McDonald et al.2013a). This minimum value is on par with the minimumvalue of t cool /t ff in some nearby cool core clusters, whichcan vary from ∼ ∼
20 (Voit & Donahue 2014). How-
Fig. 11.—
Ratio of the cooling time ( t cool ) to the free-fall time( t ff ) as a function of radius for the Phoenix cluster. The shadedgrey band represents our uncertainty, which is primarily driven byuncertainty in the temperature profile. We highlight t cool /t ff = 10with a horizontal dashed line, which appears to be the approximatethreshold for condensation in nearby clusters (McCourt et al. 2012;Sharma et al. 2012; Gaspari et al. 2012). We also highlight themaximum extent of four distinct star-forming filaments (Figure6) with thin vertical lines, and the average extent with a thickvertical line. The range of filament radii is shown as a shadedblue region. This figure demonstrates that the condensation of thecooling ICM may be fueling star formation out to radii of ∼
100 kpcin this extreme system. ever, such low values of t cool /t ff are typically reached at r <
10 kpc (Gaspari et al. 2012).Interestingly, t cool /t ff remains below 10 (the approxi-mate threshold for ICM condensation in nearby clusters(McCourt et al. 2012; Sharma et al. 2012; Gaspari et al.2012)) out to a radius of ∼
60 kpc, which is nearly themaximum extent of the star-forming filaments shown inFigure 6. This, combined with the presence of large-scale star-forming filaments, suggests that condensationof the cooling ICM is important in this system out toexceptionally large radii.We will return to this discussion of the cluster’s ther-modynamic properties later, to quantitatively comparecooling and feedback processes in the cluster core. DISCUSSIONWe have presented new X-ray, ultraviolet, optical, andradio data on the core of the Phoenix cluster (SPT-CLJ2344-4243), providing the most complete picture ofthis extreme system to date. Below, we discuss theimplications of these new data in the context of pre-vious observations and interpretations (McDonald et al.2012b, 2013a; Ueda et al. 2013; McDonald et al. 2014c;Hlavacek-Larrondo et al. 2014).4.1.
A Revised Estimate of the Central Galaxy StarFormation Rate ⊙ yr − . In this earlier work, weassume that, in the absence of dust, the slope of theUV spectral energy distribution (SED) should be flat(Kennicutt 1998). This allowed us to compute an ex-tinction correction based on the UV color. However, theamount of UV extinction and the age of the stellar popu-lation are degenerate with regards to the slope of the UVSED. Ideally, we would like an alternate, age-free, esti-mate of the reddening, which we obtained via Balmer lineratios in McDonald et al. (2014c). This reddening mapshowed a peak value of E ( B − V ) = 0 . E ( B − V ) ∼ . ∼ R V = 3 . ⊙ yr − .This estimate of the star formation rate may be arti-ficially high. Calzetti et al. (1994) show that starburstgalaxies tend to lack the rapid rise in the UV extinc-tion curve that is characteristic of the Milky Way andLMC/SMC extinction curves. This is confirmed in Fig-ure 3, where we show that a shallow, featureless extinc-tion curve (e.g., Calzetti et al. 1994) provides a muchbetter match to the UV-through-optical spectrum of thecentral galaxy in the Phoenix cluster. Using the best-fitinstantaneous-burst model from Figure 3, and averagingover the lifetime of the burst (4.5 Myr), we infer a re-duced star formation rate of 490 M ⊙ yr − .As we show in Figure 2, the throughput of the com-bined HST-COS apertures is only ∼ ∼ ⊙ yr − . More likely, the extinction in the ex-tended filaments is lower than in the inner few kpc.Assuming a much more conservative extinction in theaperture correction ( E ( B − V ) = 0 .
15, representing theminimum value measured in McDonald et al. (2014c))yields SFR = 570 M ⊙ yr − . Assuming that the cor-rect answer lies somewhere between these extrema, weinfer that the aperture- and extinction-corrected, time-averaged star formation rate for the central galaxy inthe Phoenix cluster is 610 ±
50 M ⊙ yr − , where theuncertainty here represents only the uncertainty in theextinction and aperture corrections. We note that thisUV-based estimate is consistent with the recent IR-basedestimate from Tozzi et al. (2015) of 530 ±
50 M ⊙ yr − ,suggesting that these different methods may be converg-ing on the right answer.The dominant systematic uncertainty in this estimateis how, exactly, the star formation has proceeded. Forexample, an instantaneous burst of average age 2 Myr,which is preferred by the UV-only spectral fit (Figure 5),would have a time-averaged star formation rate of 1200 M ⊙ yr − before aperture correction . On the other hand,if the star formation has been proceeding at a constantlevel for the past 15 Myr, the time-averaged rate can beas low as ∼
300 M ⊙ yr − (Figure 3). This uncertaintycould be reduced by obtaining deep, rest-frame near-UV( ∼ ∼
30 kpc. Instead, thesedata provide improved constraints on the extent andlarge-scale morphology of the star-forming filaments.4.2.
The Origin of the Star-Forming Filaments
In the central ∼
30 kpc, the UV emission is highlyasymmetric (Figure 6), indicative of a vigorous, turbu-lent starburst. However, at large radii the young starsare oriented along thin, linear filaments, akin to sys-tems like Perseus (Conselice et al. 2001) and Abell 1795(McDonald & Veilleux 2009). In McDonald et al.(2012b) we demonstrate that the fuel for the observedstar formation must originate within the cluster core.A scenario in which cold gas is brought into the corevia infalling gas-rich galaxies and/or groups is unfeasiblefor this system, given the extreme ICM density (efficientram-pressure stripping) and amount of gas needed to fuelsuch a starburst (i.e., multiple gas-rich compact groups).A popular scenario for the origin of star-forming fila-ments in nearby cool core clusters involves cool gas beingdrawn from the cluster core in the wake of buoyant radio-blown bubbles (e.g., Fabian et al. 2003; Churazov et al.2013). Figures 6 and 8 demonstrate that the most ex-tended filaments surround, rather than trail, the mostsignificant set of cavities, inconsistent with the uplift sce-nario. Indeed, the fact that the star-forming filamentsextend beyond these cavities in radius suggests that thecool gas was present before the current epoch of feedback.However, this does not rule out a scenario in which thesefilaments were uplifted by a previous episode of feedback,with the bubbles responsible for this action having risento such radii that they are undetectable.It may be that no additional mechanism is neces-sary to produce the extended, star-forming filamentsobserved in the core of the Phoenix cluster. Recentwork by McCourt et al. (2012), Sharma et al. (2012),and Gaspari et al. (2012) has shown that local thermody-namic instabilities can develop when the ratio of the cool-ing time to the free-fall time ( t cool /t ff ) is less than 10. InFigure 11, we show this ratio as a function of radius forthe Phoenix cluster, assuming hydrostatic equilibrium inthe computation of t ff . The cooling time is shorter than10 times the free-fall time over an unprecedented ∼
60 kpcin radius, suggesting that the star-forming filaments maybe fueled by local thermodynamic instabilities in the hotICM, which rapidly condense and then “rain” down ontothe central cluster galaxy (Voit et al. 2015). Further, weexpect a map of t cool /t ff to be asymmetric, since thegravitational potential is roughly spherically symmetric(in the core), while the gas density is not (see Figure 8).For example, the spiral-shaped overdensity at ∼
50 kpc4north of the cluster center will likely have t cool /t ff < t cool /t ff ∼
10 and t cool ∝ /n e . Thus, despite the factthat t cool /t ff >
10 at r >
60 kpc, there are likely over-dense regions in the ICM with t cool /t ff <
10 out to thefull extent of the star-forming filaments at r ∼
100 kpc.If the star-forming filaments did indeed condense outof the hot ICM, there ought to be evidence of cooling inthe X-ray spectrum. Using the latest
Chandra data, weestimate the classical cooling rate following White et al.(1997), using the equation:˙ M ( i ) = L X ( i ) h ( i ) + ∆ φ ( i ) − [∆ φ ( i ) + ∆ h ( i )] P i ′ = i − i ′ =1 ˙ M ( i ′ ) h ( i ) + ∆ φ ( i ) (3)where the i and i ′ indices refer to given annuli, L X ( i )is the bolometric X-ray luminosity in a given annulus, h ( i ) ≡ kT ( i ) /µm p is the energy per particle-mass ofthe hot gas for a given annulus, and ∆ φ ( i ) is the changein gravitational potential across shell i . The first termon the right corresponds to the cooling rate in the ab-sence of any additional heating, while the second term onthe right is the correction factor to account for the factthat infalling gas will simultaneously be heated gravi-tationally. Carrying out this sum in the inner 100 kpc,we estimate a classical (luminosity-based) cooling rateof 3300 ±
200 M ⊙ yr − , consistent with the estimateof 2700 ±
700 M ⊙ yr − from McDonald et al. (2013a).Assuming the star formation rate of 613 M ⊙ yr − from § ∼
20% of the pre-dicted cooling flow is converted into stars. This numberis very close to the expected star formation efficiencyfrom cool gas (10–15%; McDonald et al. 2011, 2014a),leaving open the possibility that cooling at the hot phasemay be ∼ χ = 128.75 for 128 degrees of free-dom). Likewise, recent XMM-Newton observations pub-lished by Tozzi et al. (2015) find relatively weak coolingsignatures in the spectrum at temperatures of 0.3–3.0keV, with measurements by the MOS (620 +350 − M ⊙ yr − )and PN (210 +145 − M ⊙ yr − instruments yielding valuessignificantly lower than the luminosity-based estimate.This may be because AGN feedback has recently haltedcooling at high temperatures (see § vii , O viii , Fe xvii – xx ; Peterson & Fabian2006) are redshifted to . . Chandra effective area correction arehigh due to contamination (O’Dell et al. 2013). Further,the disagreement between the spectroscopic cooling ratesderived by the pn and MOS detectors on
XMM-Newton (Tozzi et al. 2015) suggest that systematic uncertain-ties in our understanding of the soft X-ray response ofthese detectors (on both
Chandra and
XMM-Newton ) will dominate any estimate of the spectroscopic coolingrate. Thus, while there is no direct evidence for coolingin the low-resolution X-ray spectrum, we can not rule outthe hypothesis that the star-forming filaments are beingfueled by local thermodynamic instabilities in the ICM.The
Chandra
X-ray data presented here offer furthersupport for a local fuel supply. The short central cool-ing time ( t cool < yr) falls below the threshold forthe onset of cooling instabilities in the hot atmosphere(e.g., Rafferty et al. 2008). It also meetings the t cool /t ff criterion for cooling instabilities (e.g., McCourt et al.2012; Sharma et al. 2012; Gaspari et al. 2012; Voit et al.2015). However, the level of star formation lies belowthat expected from unimpeded cooling. Phoenix har-bors a powerful central radio-AGN that is apparentlyreducing the cooling rate by ∼ K (i.e., Fe xvii , O vii ) as inother clusters (Peterson & Fabian 2006), is inconsistentwith simple, isobaric cooling models, indicating that thegas may be cooling through other channels (e.g., mix-ing) or has recently been quenched (e.g., Li & Bryan2014). Future observations with high-resolution X-raygrating/microcalorimeter spectrometers (e.g., Astro-H)will help to break this degeneracy by providing firmconstraints on the rate of radiative cooling from high( ∼ K) through low ( ∼ K) temperatures for a sam-ple of nearby, cool core clusters.4.3.
Mechanical AGN Feedback
To estimate the enthalpy released by the radio jets, E = 4 P V , we measured the cavity sizes and surroundingpressures (e.g., Churazov et al. 2000; McNamara et al.2000). The X-ray cavities are seen in the raw image withwell-defined, elliptical shapes. We measured their loca-tions and sizes using the residual image after smoothingit with a gaussian kernel (FWHM = 3 pixels). Their ra-dial centroids, projected major and minor axes, and en-ergetics are presented in Table 2 and shown in Figure 12.We assume the cavities are ellipsoidal volumes with axesperpendicular to the plane of the sky and equal to theprojected major axes (upper limit) and minor axes (lowerlimit). The mean jet power assumes the cavities rosebuoyantly to their current locations in the plane of thesky following (Bˆırzan et al. 2004; McNamara & Nulsen2007).For projected distances of 17 kpc, we find a rise timeof ∼ . × yr, assuming the bubbles rise at the soundspeed, to a more likely ∼ − × yr, assuming thebubbles rise buoyantly (Bˆırzan et al. 2004). These fig-ures depend on the local gravitational acceleration mea-sured from the X-ray pressure profile, assuming hydro-static equilibrium. The gas pressure and mean gas den-sity at the locations of the cavities were found to be3 × − erg cm − and 0 .
15 cm − , respectively. For a to-tal cavity enthalpy of 4 pV = 1 . − . × erg, we find amean jet power of 1 . − . × erg s − . This mechan-ical power places Phoenix among of the most energeticAGN outbursts known. The total radio luminosity (10MHz – 10 GHz), assuming a constant spectral index of α = − .
35 (McDonald et al. 2014c), is 3 . × erg s − ,or roughly one percent of the jet power, consistent withother nearby galaxy clusters (Bˆırzan et al. 2008).The total mechanical power output measured in the5 TABLE 2Properties of X-ray Cavities
Cavity r [ ′′ ] r [kpc] a [ ′′ ] a [kpc] b [ ′′ ] b [kpc] pV [10 erg] t buoy [10 yr] P cav [10 erg s − ] S/N Central Cavities
North 2.6 17.3 1.7 11.3 1.3 8.6 1.9 – 2.5 2.1 – 5.1 0.8 – 2.5 24.7South 2.6 17.3 2.2 14.6 1.3 8.6 2.5 – 4.2 1.8 – 5.8 0.9 – 4.7 26.2Total (inner) 4.4 – 6.7 1.7 – 7.2
Potential Ghost Cavities
North 17.0 114.6 6.6 44.8 4.6 31.1 18 – 26 8.7 – 13 1.8 – 3.8 7.4South 14.4 97.2 8.6 58.4 2.9 19.3 11 – 34 6.4 – 14 1.0 – 6.7 8.1Total (outer) 29 – 60 2.8 – 10.5
Note . — : Range reflects uncertainty in the three-dimensional bubble shape. : Range reflects combined uncertainty in three-dimensional bubble shape and bubble rise time (assuming buoyant rise). : Based on residual image (Figure 8). Ratio of cavity depth to noise level at the same radius. core of the Phoenix cluster is ∼ − × erg s − .This is a factor of ∼
15 kpc50 kpc
Fig. 12.—
Upper panel: Smoothed, residual 0.5-2.0 keV image ofthe inner ∼
50 kpc of the Phoenix cluster (see Figure 8). Contourshighlight structure in the positive (dashed) and negative (solid)residuals. The red cross highlights the position of the central AGN,while red lines show the size and shape of the ellipses used todetermine cavity energetics. Lower panel: Similar to upper panel,but zoomed out by a factor of ∼ earlier study, which resulted in a factor of ∼ ∼ ∼ L cool,r< kpc = 9 . ± . × erg s − ). Given the extreme classical coolingrate in the inner 100 kpc (3300 M ⊙ yr − ), even a smallmismatch in energetics over a short time could lead tothe observed star formation rate of ∼
600 M ⊙ yr − in thecentral galaxy.In Figures 8 and 12 we show the locations of two po-tential “ghost” cavities, at radii of ∼
100 kpc. These po-tential cavities are detected with signal-to-noise of ∼ ∼ . − . × erg s − ,or anywhere from a third to 100% of the cooling lumi-nosity. This range overlaps well with the energy in theongoing outburst at small radii, suggesting that the me-chanical power has been roughly constant over the pasttwo epochs of feedback, spanning ∼
100 Myr.In summary, there is strong evidence for ongoingmechanical-mode feedback in the inner ∼
20 kpc of thePhoenix cluster, and weaker, but still convincing, ev-idence for a prior epoch of feedback roughly 100 Myrago. These episodes of feedback have highly uncertainenergetics (factor of a few in P cav ), but are consistentwith the energy required to offset cooling losses. Thisscenario is further supported by the fact that the coolingtime at r > ∼
100 Myr), suggesting that perfect balancein a given feedback epoch is not necessary at larger radii.Future studies with significantly deeper X-ray and low-frequency radio data will provide a more clear picture ofboth the duty cycle and energetics in these radio-modeoutbursts. 4.4.
A Transitioning AGN? . × erg s − . This high X-ray fluxis relatively stable, with no significant fluctuations ob-served between the initial Chandra and Suzaku obser-vations in 2011 and 2012, and the most recent Chan-dra observations in 2014. Radio AGN with compara-ble radiative and mechanical powers are uncommon, asare radiatively-efficient AGN at the centers of clusters(Russell et al. 2010; O’Sullivan et al. 2012; Walker et al.2014). The Eddington luminosity for a 10 M ⊙ blackhole is 1 . × erg s − , which would place Phoenix at afew percent of the Eddington rate, close to the expectedtransition between radio-mechanical AGN and quasars(Churazov et al. 2005; Russell et al. 2013). StandardAGN theory posits that the form of power output de-pends on the specific accretion rate: objects accretingabove a few percent of the Eddington rate are dominatedby radiation and those below are dominated by mechan-ical outflows. That Phoenix exhibits the characteristicsof both a quasar and a radio galaxy, suggesting that itmay be in transition between the two states.Based on their discovery of neutral iron emissionfrom the AGN using the Suzaku observatory, Ueda et al.(2013) have argued that the BCG is a Type 2 quasar witha high absorption column density. Assuming a bolomet-ric correction factor of 130 (Marconi et al. 2004), theyargue that the total unabsorbed AGN power would be6 × erg s − . This figure implies a black hole accre-tion rate exceeding 100 M ⊙ yr − , which would be thehighest rate known in a brightest cluster galaxy. Sucha high bolometric correction implies a high infrared lu-minosity, where most of the radiation from the buriedAGN should be escaping. The total infrared luminosityemerging from the quasar and surrounding star forma-tion is only L IR = 3 . × (McDonald et al. 2012b).This luminosity is comparable to both the mechanicaland X-ray energy fluxes, but lies two orders of magnitudebelow the bolometric luminosity quoted by Ueda et al.(2013). We conclude that the bolometric correction tothe nuclear X-ray luminosity is at most a factor of afew to ten, consistent with the lower values reportedby Vasudevan & Fabian (2007) for high-luminosity AGN.Therefore the mechanical power of Phoenix’s radio jetsis comparable to the radiation emerging from the quasar,suggesting that it may currently be undergoing a transi-tion from “quasar–mode” to “radio–mode”.4.5. The Radio Mini-Halo
The 610 MHz image of the core of the Phoenix clus-ter previously revealed the presence of a central com-pact radio source, associated with the BCG, and dif-fuse 400 −
500 kpc emission surrounding this centralsource (van Weeren et al. 2014). The diffuse emissionis classified as a radio mini-halo and, at a redshift of0 . ± . ± . ± . × W Hz − ,scaling with a spectral index of − . M =1 . × M ⊙ ; McDonald et al. 2012b), and turbulentvelocities scale with halo mass (based on structure for-mation theory), it is feasible that the Phoenix clustermaintains a high enough turbulent velocity in its coreto continuously re-accelerate a population of relativis-tic electrons. Alternative models, which can also ex-plain the presence of radio mini halos, invoke secondaryelectrons that are produced by collisions between CRprotons and thermal protons (e.g., Pfrommer & Enßlin2004; Fujita et al. 2007; Keshet & Loeb 2010). Thesesecondary models are only successful in explaining asmall range of observed properties (e.g., ZuHone et al.2014). 4.6. Is the Phoenix Cluster Unique?
As one of the most massive cool core clusters known,the Phoenix cluster can be considered, in many ways, a“scaled-up”, but otherwise normal, system. For example,the ∼
100 kpc extent of the star-forming filaments is ex-treme, being the most extended cool filaments observedin a cool core cluster to date. However, we showed inMcDonald et al. (2011) that multiphase gas is typicallyobserved to a maximum radius of r cool in nearby clus-ters, where r cool is defined at the radius within whichthe cooling time is less than 3 Gyr. For Perseus andAbell 1795, this radius corresponds to ∼
60 kpc, while forthe Phoenix cluster we measure r cool = 112 kpc. Thus,while extreme, the extent of these star-forming filamentsare scaling as expected with the overall mass of the clus-ter, and do not challenge our understanding of cool coreclusters. The same can be said for the extreme coolingluminosity, central gas pressure, and radio luminosity.The extreme star formation rate in the core of thePhoenix cluster ( ∼
600 M ⊙ yr − ) may be pointing tosomething unique about this cluster. This starburstaccounts for ∼
20% of the classically-predicted coolingflow, suggesting that cooling may be proceeding very ef-ficiently in this system. For comparison, O’Dea et al.(2008) found, for a sample of nearby clusters with star-forming BCGs, a typical ratio of the star formation7rate to the cooling rate of 1.8 +1 . − . % – roughly an or-der of magnitude lower than what we observe in thePhoenix cluster. However, when the most extreme endof this population is considered – clusters harboring star-burst BCGs such as Abell 1835 (SFR ∼
200 M ⊙ yr − ;McNamara et al. 2006), RX J1504.1-0248 (SFR ∼ ⊙ yr − ; Ogrean et al. 2010), and MACS 1931.8-2634(SFR ∼
170 M ⊙ yr − ; Ehlert et al. 2011) – this ratiojumps to 17 ± ∼ < z < . t cool /t ff = 10, and it does so by a substan-tial margin. The violation of this boundary is thought toinitiate strong AGN feedback via condensation or pre-cipitation of the cooling ICM. This is also reflected inFigure 10, where we show that the ICM entropy pro-file in the Phoenix cluster is consistent with the baselineentropy profile, which represents the expectation in theabsence of any feedback (Voit et al. 2005), at all radii .When considering all 165 known galaxy groups and clus-ters with existing Chandra data and temperatures in therange 4 keV < T X <
15 keV, Cavagnolo et al. (2009)did not find a single cluster with an entropy profile assteep as Phoenix over the range 10 < r < ∼
100 Myr ago. This AGN appears to be heavily influenc-ing the local environment, highly ionizing the cool gas inthe central ∼
10 kpc, driving an ionized outflow north ofthe central galaxy (McDonald et al. 2014c), and inflat-ing bubbles in the dense ICM which may be re-directingthe extended, star-forming filaments (see Figure 9). Un-derstanding the extreme nature of this AGN, specificallywhy it appears to have had transitioned from mechanical-mode to radiative-mode, and now back to mechanical-mode, will likely provide the key to understanding theother oddities in the core of the Phoenix cluster. SUMMARYIn this paper we present a detailed, multi-wavelengthstudy of the core of the Phoenix galaxy cluster (SPT-CLJ2344-4243). The analysis presented here is basedon a combination of new and archival data at radio(GMRT), optical (HST-WFC3, Magellan Megacam), ul-traviolet (HST-WFC3, HST-COS), and X-ray (
Chandra -ACIS-I) wavelengths. The primary results of this lateststudy are summarized as follows: • Complex, star-forming filaments are observed inthe rest-frame ultraviolet to extend up to ∼
100 kpcfrom the central cluster galaxy in multiple direc-tions. These newly-detected filaments extend afactor of ∼ • Modeling the combined UV-optical (rest-frame900˚A–6000˚A) spectrum of the central clustergalaxy reveals a massive (2 . × M ⊙ ), young( ∼ ± ⊙ yr − . We note that this estimate can vary bya factor of ∼ • The best-fitting dust model is significantly “grayer”than that measured for the Milky Way andSMC/LMC, consistent with observations fornearby starburst galaxies and distant quasars. Thedata show no evidence of the 2175˚A bump, suggest-ing that there may be a process either destroying orpreventing the formation of small grains and thosecontributing to this bump (i.e., PAHs). • We detect significant O vi λλ OV I = 7 . ± . × erg s − ) in the inner ∼
15 kpc of the cluster core. This emission is con-sistent with having origins primarily in the centralAGN and in shock-heated gas along a northern ion-ized outflow. We are unable to put constraints onwhat fraction of this emission may originate fromthe cooling ICM – the data are consistent with botha complete lack of cooling and a massive (i.e., 5000M ⊙ yr − ) cooling flow. • We confirm the presence of strong (S/N ∼
25) X-ray cavities in the inner 20 kpc of the cluster core.The total mechanical energy in these cavities isP cav = 2–7 × erg s − , depending on their in-trinsic shape, making this one of the most powerfuloutbursts of radio-mode feedback known. The in-ferred jet power from these cavities is slightly lessthan the cooling luminosity (L cool ∼ erg s − )in the inner 100 kpc. • We find that the bolometric X-ray luminosity of theAGN ( L X,bol = 5 . × erg s − ) corresponds to afew percent of the Eddington luminosity, consistentwith a recent transition from quasars to radio-modeAGN. The similarity of the bolometric luminosityand the mechanical power further supports this pic-ture of an AGN transitioning from a radiatively-efficient mode to a mechanical mode.8 • We find evidence for an additional set of X-raycavities at larger radii ( ∼
100 kpc), suggesting thatthere may have a been prior episode of radio-modefeedback ∼
100 Myr ago. Assuming that both po-tential “ghost” cavities are real, this prior episodeof feedback may have had jet powers of 2.8 – 10.5 × erg s − , which is similar to what is measuredin the inner set of cavities. This implies a relativelyconstant mechanical output of the central AGN be-tween bursts, with a duty cycle of ∼
100 Myr. • The azimuthally-averaged cooling time of theICM is shorter than the precipitation threshold( t cool /t ff = 10) at radii of .
50 kpc, suggestingthat local thermodynamic instabilities in the hotICM may be fueling both the star formation andAGN feedback. Dense substructures extending be-yond 50 kpc likely have exceeded the precipitationthreshold out to much larger radii ( ∼ • We see significant substructure in the inner ∼
200 kpc (0.15R ) of the cluster. The spiralshape of this structure is reminiscent of infallingcool clouds. This may be cooling gas that has beenredirected by the strong mechanical feedback, in-falling cool group-sized halos, or sloshing of the coolcore. The presence of a radio mini-halo supportsthe X-ray observations that the inner core is highlyturbulent. • Outside of the cool core ( r > . R ), the clusterappears relaxed and has thermodynamic proper-ties typical of other clusters at similar mass andredshift.While illuminating in many ways, these new dataleave several questions unanswered. It remains unclearhow such vigorous star formation is sustained in themidst of massive, radio-mode outbursts from the cen-tral AGN. The absolute estimate of the star formationrate in the central galaxy hinges on understanding the star formation history, which would benefit from broader-band near-UV spectroscopy, while confirmation that starformation is being fueled by cooling of the hot ICMawaits high resolution X-ray spectroscopy. Confirmationof the extended cavities at large radii requires deeperX-ray follow-up, alongside high angular resolution low-frequency radio imaging. As the most extreme cool corecluster known, such follow-up studies of this systems willallow a deeper understanding of the complex and ongoingwar between AGN feedback and cooling in dense clustercores. ACKNOWLEDGEMENTSM. M. acknowledges support by NASA throughcontracts HST-GO-13456.002A (Hubble) and GO4-15122A (Chandra), and Hubble Fellowship grant HST-HF51308.01-A awarded by the Space Telescope ScienceInstitute, which is operated by the Association of Univer-sities for Research in Astronomy, Inc., for NASA, undercontract NAS 5-26555. The Guaranteed Time Observa-tions (GTO) included here were selected by the ACISInstrument Principal Investigator, Gordon P. Garmire,of the Huntingdon Institute for X-ray Astronomy, LLC,which is under contract to the Smithsonian Astrophys-ical Observatory; Contract SV2-82024. J.E.C. acknowl-edges support from National Science Foundation grantsPLR-1248097 and PHY-1125897. B.R.M. acknowledgesgenerous financial support from the Natural Sciencesand Engineering Research Council of Canada. R.J.W.is supported by NASA through the Einstein Postdoc-toral grant number PF2-130104 awarded by the ChandraX-ray Center, which is operated by the Smithsonian As-trophysical Observatory for NASA under contract NAS8-03060. J.HL is supported by NSERC through the discov-ery grant and Canada Research Chair programs, as wellas FRQNT. D.A. acknowledges support from the DLRunder projects 50 OR 1210 & 1407, and from the DFGunder project AP 253/1-1. A.C.E. acknowledges supportfrom STFC grant ST/I001573/1. REFERENCESAllen, M. G., Groves, B. A., Dopita, M. A., Sutherland, R. S., &Kewley, L. J. 2008, ApJS, 178, 20Allen, S. W. 1995, MNRAS, 276, 947Arnaud, K. A. 1996, in Astronomical Society of the PacificConference Series, Vol. 101, Astronomical Data AnalysisSoftware and Systems V, ed. G. H. Jacoby & J. Barnes, 17–+Arnaud, M., Pratt, G. W., Piffaretti, R., et al. 2010, A&A, 517,A92Baldi, A., Ettori, S., Mazzotta, P., Tozzi, P., & Borgani, S. 2007,ApJ, 666, 835Baldi, A., Ettori, S., Molendi, S., & Gastaldello, F. 2012, A&A,545, A41Bauer, F. E., Fabian, A. C., Sanders, J. S., Allen, S. W., &Johnstone, R. M. 2005, MNRAS, 359, 1481Bautz, M. W., Miller, E. D., Sanders, J. S., et al. 2009, PASJ, 61,1117Bˆırzan, L., McNamara, B. R., Nulsen, P. E. J., Carilli, C. L., &Wise, M. W. 2008, ApJ, 686, 859Bˆırzan, L., Rafferty, D. A., McNamara, B. R., Wise, M. W., &Nulsen, P. E. J. 2004, ApJ, 607, 800Blanton, E. L., Randall, S. W., Clarke, T. E., et al. 2011, ApJ,737, 99Bleem, L. E., Stalder, B., de Haan, T., et al. 2015, ApJS, 216, 27Bregman, J. N., Fabian, A. C., Miller, E. D., & Irwin, J. A. 2006,ApJ, 642, 746Bregman, J. N., Miller, E. D., & Irwin, J. A. 2001, ApJ, 553, L125Calzetti, D., Kinney, A. L., & Storchi-Bergmann, T. 1994, ApJ,429, 582 Canning, R. E. A., Ryon, J. E., Gallagher, J. S., et al. 2014,MNRAS, 444, 336Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345,245Cavagnolo, K. W., Donahue, M., Voit, G. M., & Sun, M. 2009,ApJS, 182, 12Cavagnolo, K. W., McNamara, B. R., Nulsen, P. E. J., et al.2010, ApJ, 720, 1066Chatzikos, M., Williams, R. J. R., Ferland, G. J., et al. 2015,MNRAS, 446, 1234Churazov, E., Forman, W., Jones, C., & B¨ohringer, H. 2000,A&A, 356, 788Churazov, E., Ruszkowski, M., & Schekochihin, A. 2013,MNRAS, 436, 526Churazov, E., Sazonov, S., Sunyaev, R., et al. 2005, MNRAS, 363,L91Conselice, C. J., Gallagher, III, J. S., & Wyse, R. F. G. 2001, AJ,122, 2281Cornwell, T. J., & Perley, R. A. 1992, A&A, 261, 353Cotton, W. D. 2008, PASP, 120, 439Cowie, L. L., Hu, E. M., Jenkins, E. B., & York, D. G. 1983, ApJ,272, 29De Grandi, S., Ettori, S., Longhetti, M., & Molendi, S. 2004,A&A, 419, 7Donahue, M., Connor, T., Fogarty, K., et al. 2015, ArXive-prints, arXiv:1504.00598Dong, R., Rasmussen, J., & Mulchaey, J. S. 2010, ApJ, 712, 883Dunn, R. J. H., & Fabian, A. C. 2006, MNRAS, 373, 959 Dunn, R. J. H., Fabian, A. C., & Taylor, G. B. 2005, MNRAS,364, 1343Ehlert, S., Allen, S. W., von der Linden, A., et al. 2011, MNRAS,411, 1641Ekstr¨om, S., Georgy, C., Eggenberger, P., et al. 2012, A&A, 537,A146Fabian, A. C. 1994, ARA&A, 32, 277—. 2012, ARA&A, 50, 455Fabian, A. C., Johnstone, R. M., Sanders, J. S., et al. 2008,Nature, 454, 968Fabian, A. C., Sanders, J. S., Crawford, C. S., et al. 2003,MNRAS, 344, L48Fabian, A. C., Sanders, J. S., Williams, R. J. R., et al. 2011,MNRAS, 417, 172Ferland, G. J., Fabian, A. C., Hatch, N. A., et al. 2009, MNRAS,392, 1475Ferland, G. J., Korista, K. T., Verner, D. A., et al. 1998, PASP,110, 761Fischera, J., & Dopita, M. 2011, A&A, 533, A117Fujita, Y., Kohri, K., Yamazaki, R., & Kino, M. 2007, ApJ, 663,L61Gaspari, M., Ruszkowski, M., & Sharma, P. 2012, ApJ, 746, 94Georgy, C., Ekstr¨om, S., Eggenberger, P., et al. 2013, A&A, 558,A103Gitti, M., Brunetti, G., Feretti, L., & Setti, G. 2004, A&A, 417, 1Gitti, M., Brunetti, G., & Setti, G. 2002, A&A, 386, 456Groves, B. A., Dopita, M. A., & Sutherland, R. S. 2004, ApJS,153, 75Hicks, A. K., & Mushotzky, R. 2005, ApJ, 635, L9High, F. W., Stubbs, C. W., Rest, A., Stalder, B., & Challis, P.2009, AJ, 138, 110High, F. W., Hoekstra, H., Leethochawalit, N., et al. 2012, ApJ,758, 68Hlavacek-Larrondo, J., Fabian, A. C., Edge, A. C., et al. 2012,MNRAS, 421, 1360Hlavacek-Larrondo, J., Allen, S. W., Taylor, G. B., et al. 2013,ApJ, 777, 163Hlavacek-Larrondo, J., McDonald, M., Benson, B. A., et al. 2014,ArXiv e-prints, arXiv:1410.0025Hoffer, A. S., Donahue, M., Hicks, A., & Barthelemy, R. S. 2012,ApJS, 199, 23Hudson, D. S., Mittal, R., Reiprich, T. H., et al. 2010, A&A, 513,A37Intema, H. T., van der Tol, S., Cotton, W. D., et al. 2009, A&A,501, 1185Johnstone, R. M., Fabian, A. C., & Nulsen, P. E. J. 1987,MNRAS, 224, 75Kalberla, P. M. W., Burton, W. B., Hartmann, D., et al. 2005,A&A, 440, 775Kennicutt, Jr., R. C. 1998, ARA&A, 36, 189Keshet, U., & Loeb, A. 2010, ApJ, 722, 737Kettenis, M., van Langevelde, H. J., Reynolds, C., & Cotton, B.2006, in Astronomical Society of the Pacific Conference Series,Vol. 351, Astronomical Data Analysis Software and SystemsXV, ed. C. Gabriel, C. Arviset, D. Ponz, & S. Enrique, 497Kewley, L. J., Dopita, M. A., Sutherland, R. S., Heisler, C. A., &Trevena, J. 2001, ApJ, 556, 121Kirk, B., Hilton, M., Cress, C., et al. 2015, MNRAS, 449, 4010Leccardi, A., & Molendi, S. 2008a, A&A, 487, 461—. 2008b, A&A, 486, 359Leitherer, C., Ekstr¨om, S., Meynet, G., et al. 2014, ApJS, 212, 14Leitherer, C., Schaerer, D., Goldader, J. D., et al. 1999, ApJS,123, 3Li, Y., & Bryan, G. L. 2014, ApJ, 789, 54Mantz, A. B., Allen, S. W., Morris, R. G., et al. 2015, ArXive-prints, arXiv:1502.06020Marconi, A., Risaliti, G., Gilli, R., et al. 2004, MNRAS, 351, 169Markevitch, M., Bautz, M. W., Biller, B., et al. 2003, ApJ, 583, 70Mazzotta, P., & Giacintucci, S. 2008, ApJ, 675, L9McCourt, M., Sharma, P., Quataert, E., & Parrish, I. J. 2012,MNRAS, 419, 3319McDonald, M. 2011, ApJ, 742, L35McDonald, M., Benson, B., Veilleux, S., Bautz, M. W., &Reichardt, C. L. 2013a, ApJ, 765, L37McDonald, M., Roediger, J., Veilleux, S., & Ehlert, S. 2014a,ApJ, 791, L30McDonald, M., & Veilleux, S. 2009, ApJ, 703, L172McDonald, M., Veilleux, S., & Rupke, D. S. N. 2012a, ApJ, 746,153McDonald, M., Veilleux, S., Rupke, D. S. N., Mushotzky, R., &Reynolds, C. 2011, ApJ, 734, 95McDonald, M., Bayliss, M., Benson, B. A., et al. 2012b, Nature,488, 349McDonald, M., Benson, B. A., Vikhlinin, A., et al. 2013b, ApJ,774, 23—. 2014b, ArXiv e-prints, arXiv:1404.6250 McDonald, M., Swinbank, M., Edge, A. C., et al. 2014c, ApJ,784, 18McLeod, B. A., Gauron, T. M., Geary, J. C., Ordway, M. P., &Roll, J. B. 1998, in Society of Photo-Optical InstrumentationEngineers (SPIE) Conference Series, Vol. 3355, OpticalAstronomical Instrumentation, ed. S. D’Odorico, 477–486McNamara, B. R., & Nulsen, P. E. J. 2007, ARA&A, 45, 117—. 2012, New Journal of Physics, 14, 055023McNamara, B. R., & O’Connell, R. W. 1989, AJ, 98, 2018McNamara, B. R., Wise, M., Nulsen, P. E. J., et al. 2000, ApJ,534, L135McNamara, B. R., Wise, M. W., Nulsen, P. E. J., et al. 2001,ApJ, 562, L149McNamara, B. R., Rafferty, D. A., Bˆırzan, L., et al. 2006, ApJ,648, 164Mendygral, P. J., Jones, T. W., & Dolag, K. 2012, ApJ, 750, 166Mittal, R., Whelan, J. T., & Combes, F. 2015, ArXiv e-prints,arXiv:1504.00921Nulsen, P. E. J., Jones, C., Forman, W. R., et al. 2007, in Heatingversus Cooling in Galaxies and Clusters of Galaxies, ed.H. B¨ohringer, G. W. Pratt, A. Finoguenov, & P. Schuecker ,210O’Dea, C. P., Baum, S. A., Privon, G., et al. 2008, ApJ, 681, 1035O’Dell, S. L., Swartz, D. A., Tice, N. L., et al. 2013, in UV,X-Ray, and Gamma-Ray Space Instrumentation for AstronomyXVIII, UV, X-Ray, and Gamma-Ray Space Instrumentation forAstronomy XVIIIOegerle, W. R., Cowie, L., Davidsen, A., et al. 2001, ApJ, 560,187Ogrean, G. A., Hatch, N. A., Simionescu, A., et al. 2010,MNRAS, 406, 354O’Sullivan, E., Giacintucci, S., David, L. P., et al. 2011, ApJ, 735,11O’Sullivan, E., Giacintucci, S., Babul, A., et al. 2012, MNRAS,424, 2971Paterno-Mahler, R., Blanton, E. L., Randall, S. W., & Clarke,T. E. 2013, ApJ, 773, 114Perley, R. A. 1989, in Astronomical Society of the PacificConference Series, Vol. 6, Synthesis Imaging in RadioAstronomy, ed. R. A. Perley, F. R. Schwab, & A. H. Bridle, 259Peterson, J. R., & Fabian, A. C. 2006, Phys. Rep., 427, 1Peterson, J. R., Kahn, S. M., Paerels, F. B. S., et al. 2003, ApJ,590, 207Pfrommer, C., & Enßlin, T. A. 2004, A&A, 413, 17Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. 2013,A&A, 550, A131Pratt, G. W., B¨ohringer, H., Croston, J. H., et al. 2007, A&A,461, 71Rafferty, D. A., McNamara, B. R., & Nulsen, P. E. J. 2008, ApJ,687, 899Rafferty, D. A., McNamara, B. R., Nulsen, P. E. J., & Wise,M. W. 2006, ApJ, 652, 216Reiprich, T. H., Basu, K., Ettori, S., et al. 2013, Space Sci. Rev.,177, 195Reynolds, C. S., Lohfink, A. M., Babul, A., et al. 2014, ApJ, 792,L41Roediger, E., Br¨uggen, M., Simionescu, A., et al. 2011, MNRAS,413, 2057Ruel, J., Bazin, G., Bayliss, M., et al. 2014, ApJ, 792, 45Russell, H. R., Fabian, A. C., Sanders, J. S., et al. 2010, MNRAS,402, 1561Russell, H. R., McNamara, B. R., Edge, A. C., et al. 2013,MNRAS, 432, 530Russell, H. R., Sanders, J. S., & Fabian, A. C. 2008, MNRAS,390, 1207Sanders, J. S., & Fabian, A. C. 2007, MNRAS, 381, 1381Sanders, J. S., Fabian, A. C., & Smith, R. K. 2011, MNRAS, 410,1797Sanders, J. S., Fabian, A. C., Smith, R. K., & Peterson, J. R.2010, MNRAS, 402, L11Sharma, P., McCourt, M., Quataert, E., & Parrish, I. J. 2012,MNRAS, 420, 3174Song, J., Zenteno, A., Stalder, B., et al. 2012, ApJ, 761, 22Sun, M. 2009, ApJ, 704, 1586Sutherland, R. S., & Dopita, M. A. 1993, ApJS, 88, 253Tozzi, P., Gastaldello, F., Molendi, S., et al. 2015, ArXiv e-prints,arXiv:1505.06913Ueda, S., Hayashida, K., Anabuki, N., et al. 2013, ApJ, 778, 33Urban, O., Simionescu, A., Werner, N., et al. 2014, MNRAS, 437,3939van Weeren, R. J., Intema, H. T., Lal, D. V., et al. 2014, ApJ,786, L17Vasudevan, R. V., & Fabian, A. C. 2007, MNRAS, 381, 1235Vikhlinin, A., Burenin, R., Forman, W. R., et al. 2007, in Heatingversus Cooling in Galaxies and Clusters of Galaxies, ed.H. B¨ohringer, G. W. Pratt, A. Finoguenov, & P. Schuecker, 480