Chandra and JVLA observations of HST Frontier Fields cluster MACS J0717.5+3745
R. J. van Weeren, G. A. Ogrean, C. Jones, W. R. Forman, F. Andrade-Santos, Connor J. J. Pearce, A. Bonafede, M. Brüggen, E. Bulbul, T. E. Clarke, E. Churazov, L. David, W. A. Dawson, M. Donahue, A. Goulding, R. P. Kraft, B. Mason, J. Merten, T. Mroczkowski, P. E. J. Nulsen, P. Rosati, E. Roediger, S. W. Randall, J. Sayers, K. Umetsu, A. Vikhlinin, A. Zitrin
DDraft version April 24, 2018
Preprint typeset using L A TEX style AASTeX6 v. 1.0
CHANDRA AND JVLA OBSERVATIONS OF HST FRONTIER FIELDS CLUSTER MACS J0717.5+3745
R. J. van Weeren (cid:63) , G. A. Ogrean † , C. Jones , W. R. Forman , F. Andrade-Santos , Connor J. J. Pearce ,A. Bonafede , M. Br¨uggen , E. Bulbul , T. E. Clarke , E. Churazov , L. David , W. A. Dawson ,M. Donahue , A. Goulding , R. P. Kraft , B. Mason , J. Merten , T. Mroczkowski , P. E. J. Nulsen ,P. Rosati , E. Roediger ‡ , S. W. Randall , J. Sayers , K. Umetsu , A. Vikhlinin , A. Zitrin † Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA Department of Physics, Stanford University, 382 Via Pueblo Mall, Stanford, CA 94305-4060, USA Department of Physics and Astronomy, University of Southampton, Southampton SO17 1BJ, UK Hamburger Sternwarte, Universit¨at Hamburg, Gojenbergsweg 112, D-21029 Hamburg, Germany Kavli Institute for Astrophysics and Space Research, Massachusetts Institute of Technology,77 Massachusetts Avenue, Cambridge, MA 02139 U.S. Naval Research Laboratory, 4555 Overlook Ave SW, Washington, D.C. 20375, USA Max Planck Institute for Astrophysics, Karl-Schwarzschild-Str. 1, 85741, Garching, Germany Space Research Institute, Profsoyuznaya 84/32, Moscow, 117997, Russia Lawrence Livermore National Lab, 7000 East Avenue, Livermore, CA 94550, USA Department of Physics and Astronomy, Michigan State University, East Lansing, MI 48824, USA Department of Astrophysical Sciences, Princeton University, Princeton, NJ 08544, USA National Radio Astronomy Observatory, 520 Edgemont Road, Charlottesville, VA 22903, USA Department of Physics, University of Oxford, Keble Road, Oxford OX1 3RH, UK ESO - European Organization for Astronomical Research in the Southern hemisphere,Karl-Schwarzschild-Str. 2, D-85748 Garching b. M¨unchen, Germany ICRAR, University of Western Australia, 35 Stirling Hwy, Crawley WA 6009, Australia Dipartimento di Fisica e Scienze della Terra, Universit`a di Ferrara, Via Saragat 1, I-44122 Ferrara, Italy E.A. Milne Centre for Astrophysics, Department of Physics & Mathematics,University of Hull, Cottinton Road, Hull, HU6 7RX, UK Cahill Center for Astronomy and Astrophysics, California Institute of Technology, MC 249-17, Pasadena, CA 91125, USA Institute of Astronomy and Astrophysics, Academia Sinica, PO Box 23-141, Taipei 10617, Taiwan (cid:63)
Clay Fellow; † Hubble Fellow; ‡ Visiting Scientist
ABSTRACTTo investigate the relationship between thermal and non-thermal components in merger galaxyclusters, we present deep
JVLA and
Chandra observations of the HST Frontier Fields clusterMACS J0717.5+3745. The
Chandra image shows a complex merger event, with at least four compo-nents belonging to different merging subclusters. NW of the cluster, ∼ . (cid:38)
20 keV ICM gas detected by
Chandra near the relic location provides additional support for this re-acceleration scenario.
Keywords:
Galaxies: clusters: individual (MACS J0717.5+3745) — Galaxies: clusters: intracluster a r X i v : . [ a s t r o - ph . C O ] J a n van Weeren et al. medium — Radiation mechanisms: non-thermal — X-rays: galaxies: clusters INTRODUCTIONMerging galaxy clusters are excellent laboratoriesto investigate cluster formation and to explore howthe particles that produce cluster-scale diffuse radioemission are accelerated. A textbook example ofan extreme merging cluster is MACS J0717.5+3745.MACS J0717.5+3745 was discovered by Edge et al.(2003) as part of the MAssive Cluster Survey (MACS;Ebeling et al. 2001) and is located at a redshift of z = 0 . . ± . (cid:38)
20 keV (Ma et al. 2008, 2009; Limousin et al.2012, 2016). A study of the Sunyaev-Zel’dovich (SZ) ef-fect provided further evidence for the presence of shock-heated gas (Mroczkowski et al. 2012). Mroczkowskiet al. (2012) and Sayers et al. (2013) found evidencefor a kinetic SZ signal for one of the subclusters, con-firming the large velocity offset ( ≈ − ) foundearlier from spectroscopy data (Ma et al. 2009). A re-cent study reported the detection of a second kineticSZ component belonging to another subcluster (Adamet al. 2016). Connected to the cluster in the southeastis a ∼ M vir = (3 . ± . × M (cid:12) ; Umetsu et al. 2014), complex mass distributionand relatively shallow mass profile (Zitrin et al. 2009),the cluster provides a large area of sky with high lensingmagnification, and is thus selected as part of the ClusterLensing And Supernova survey with Hubble (CLASH,Postman et al. 2012; Medezinski et al. 2013) and theHST Frontier Fields program (Lotz et al. 2014, 2016)to find high-z lensed objects.Ma et al. (2009) reported decrements in theICM temperature near two of the subclusters ofMACS J0717.5+3745, which they interpret as remnantsof cool cores. For one of these subclusters, Ma et al.(2009) measured a temperature 5.7 keV temperature,suggesting this component is still at the early stage ofmerging. Diego et al. (2015) found that one of the darkmatter components (the one furthest to the NW) has asignificant offset from the closest X-ray peak. Significantoffsets between the lensing and X-ray peaks are expected [email protected] in the case of a high-speed collision in the plane of thesky.Previous radio studies of the cluster have focused ondiffuse radio emission that is present in the cluster (vanWeeren et al. 2009b; Bonafede et al. 2009; Pandey-Pommier et al. 2013). The cluster hosts a giant radiohalo extending over an area of about 1.6 Mpc. Po-larized emission from the radio halo was detected byBonafede et al. (2009). The radio luminosity (1.4 GHzradio power) is the largest known for any cluster, inagreement with the cluster’s large mass and high globaltemperature (e.g., Cassano et al. 2013).The cluster also hosts a large 0.7–0.8 Mpc radio relic.It has been suggested that the radio relic in the clustertraces a large-scale shock wave which originated fromthe ongoing merger events (van Weeren et al. 2009b),or alternatively, from an accretion shock related to thelarge-scale filament at the southeast (Bonafede et al.2009).In the standard scenario (Enßlin et al. 1998) for radiorelics, particles are accelerated at the shock via the Dif-fusive Shock Acceleration (DSA) mechanism in a firstorder Fermi process (e.g. Drury 1983). A problem withthis interpretation is that shock Mach numbers in clus-ters are typically low ( M (cid:46) JVLA ) and
Chandra obser-vations of lensed radio and X-ray sources located behindMACS J0717.5+3745. In this work, we present the re-sults of the
Chandra and
JVLA observations of the clus-ter itself. A
Chandra analysis of the large-scale filamentto the southeast is described in a separate letter (Ogreanet al. 2017). The data reduction and observations aredescribed in Section 2. The radio and X-ray images, andthe spectral index and ICM temperature maps are pre-sented Sections 3 and 4. This is followed by a discussionand conclusions in Sections 5 and 6. In this paper weadopt a ΛCDM cosmology with H = 70 km s − Mpc − ,Ω m = 0 .
3, and Ω Λ = 0 .
7. With the adopted cosmol-ogy, 1 (cid:48)(cid:48) corresponds to a physical scale of 6.387 kpc at z = 0 . handra JVLA observations of MACS J0717.5+3745 OBSERVATIONS & DATA REDUCTION2.1.
JVLA observationsJVLA observations of MACS J07175+3745 were ob-tained in the L-, S-, and C-bands, covering the frequencyrange from 1 to 6.5 GHz. An overview of the frequencybands and observations is given in Table 1. The to-tal recorded bandwidth was 1 GHz for the L-band, and2 GHz for the S- and C-bands. For the primary cali-brators we used 3C138 and 3C147. J0713+4349 was in-cluded as a secondary calibrator. All four polarizationproducts were recorded.The data were reduced with
CASA (McMullin et al.2007) version 4.5 and data from the different observingruns were all processed in the same way. The data reduc-tion procedure is described in more detail in van Weerenet al. (2016b). To summarize, the data were calibratedfor the antenna position offsets, elevation dependentgains, global delay, cross-hand delay, bandpass, polar-ization leakage and angles, and temporal gain variationsusing the primary and secondary calibrator sources. RFIwas identified and flagged with the
AOFlagger (Offringaet al. 2010). The calibration solutions from the primaryand secondary calibrator sources were applied to thetarget field and several rounds of self-calibration werecarried out to refine the gain solutions.After the individual datasets were calibrated, the ob-servations from the different configurations (for the samefrequency band) were combined and imaged together.One extra round of self-calibration was carried out, us-ing the combined images, to align the datasets from thedifferent configurations.Deep continuum images were produced with
WSClean (Offringa et al. 2014) in the three different frequencybands. We employed the wide-band clean and multi-scale algorithms. Clean masks were employed at allstages and made with the
PyBDSM source detection pack-age (Mohan & Rafferty 2015). The final images werecorrected for the primary beam attenuation using thebeam models provided by
CASA .Images were made with different weighting schemesto emphasize different aspects of the radio emission. Anoverview of the image properties is given in Table 2. Wealso produced deeper images by stacking the L-, S-, andC-band images (equal weights) after convolving them toa common resolution .2.1.1. Spectral index maps The large change in the primary beam size prevents a sim-ple joint deconvolution and would have required the computa-tionally expensive wide-band A-Projection algorithm (Bhatnagaret al. 2013).
For making radio spectral index maps, we imaged thedata with
CASA , employing w-projection (Cornwell et al.2005, 2008) and multiscale clean (Cornwell 2008) with nterms=3 (Rau & Cornwell 2011). Three separate con-tinuum images (corresponding to the L-, S-, and C-bands), were created at reference frequencies of 1.5, 3.0,and 5.5 GHz, respectively. Inner uv-range cuts were em-ployed based on minimum uv-distance provided by theC-band data. In addition, we used uniform weighting tocorrect for differences in the uv-plane sampling. Differ-ent uv-tapers were used to produce images at resolutionsof 1.5 (cid:48)(cid:48) , 2.5 (cid:48)(cid:48) , 5 (cid:48)(cid:48) and 10 (cid:48)(cid:48) . The remaining minor differ-ences in the beam sizes (after using the uv-tapers) weretaken out by convolving the images to the same resolu-tion. The images were corrected for the primary beamattenuation.We created the spectral index maps by fitting a firstorder polynomial through the three flux measurementsat 1.5, 3.0 and 5.5 GHz in log ( S ) − log ( ν ) space. Thespectral index thus represents the average spectral in-dex in the 1.0–6.5 GHz band, neglecting any spectralcurvature. Pixels with values below 2 . σ rms in any ofthe three maps were blanked.2.2. Chandra Observations
MACS J0717.5+3745 was observed with
Chandra fora total of 243 ks between 2001 and 2013. A summary ofthe observations is presented in Table 3. The datasetswere reduced with CIAO v4.7 and CALDB v4.6.5, fol-lowing the same methodology that was described byOgrean et al. (2015). ObsID 1655 was contaminatedby flares even after the standard cleaning was applied.Given that the exposure time of ObsID 1655 is < −
12 keV count rate as thecorresponding ObsID.Point sources were detected with the CIAO script wavdetect using wavelet scales of 1, 2, 4, 8, 16, and32 pixels and ellipses with radii 5 σ around the centersof the detected sources. These point sources were ex-cluded from the analysis.2.3. Chandra Background Modeling
Background spectra were extracted from a region out-side 2 . van Weeren et al. Table 1 . JVLA ObservationsObservation date Frequency coverage Channel width Integration time On source time a LAS b [GHz] [MHz] [s] [hr] [ (cid:48)(cid:48) ]L-band A-array 28 Mar, 2013 1–2 1 1 5 . . .
25 970L-band D-array 9 Aug, 2014 1–2 1 5 2 .
25 970S-band A-array 22 Feb, 2013 2–4 2 1 5 . . .
25 490S-band D-array 3 Aug, 2014 2–4 2 5 3 .
25 490C-band B-array Sep 30, 2013 4.5–6.5 2 3 5 . .
25 240C-band D-array Aug 2, 2014 4.5–6.5 2 5 3 .
25 240 a Quarter hour rounding b Largest angular scale that can be recovered by these observations
Table 2 . JVLA image propertiesfrequency resolution weighting a uv-taper r.m.s. noise[GHz] [ (cid:48)(cid:48) ] [ (cid:48)(cid:48) ] [ µ Jy]1–2 1 . (cid:48)(cid:48) × . (cid:48)(cid:48) Briggs – 5.11–2 2 . (cid:48)(cid:48) × . (cid:48)(cid:48) Briggs 2 4.91–2 5 . (cid:48)(cid:48) × . (cid:48)(cid:48) uniform 5 7.91–2 10 . (cid:48)(cid:48) × . (cid:48)(cid:48) uniform 10 152–4 0 . (cid:48)(cid:48) × . (cid:48)(cid:48) Briggs – 1.92–4 2 . (cid:48)(cid:48) × . (cid:48)(cid:48) Briggs 2 2.02–4 5 . (cid:48)(cid:48) × . . (cid:48)(cid:48) × . (cid:48)(cid:48) uniform 10 6.24.5–6.5 1 . (cid:48)(cid:48) × . (cid:48)(cid:48) Briggs – 2.24.5–6.5 3 . (cid:48)(cid:48) × . (cid:48)(cid:48) Briggs 2 2.04.5–6.5 5 . (cid:48)(cid:48) × . (cid:48)(cid:48) uniform 5 2.44.5–6.5 10 . (cid:48)(cid:48) × . (cid:48)(cid:48) uniform 10 3.9 a For all images made with Briggs (1995) weighting we used robust=0 . ing emission from the Local Hot Bubble (LHB), anabsorbed thermal component describing emission fromthe Galactic Halo (GH), and an absorbed power-lawcomponent describing emission from unresolved pointsources. We used photoelectric absorption cross-sectionsfrom Verner et al. (1996), and the elemental abundancesfrom Feldman (1992). The hydrogen column densityin the direction of MACS J0717.5+3745 was fixed to8 . × cm − , which is the sum of the weighted av-erage atomic hydrogen column density from the Leiden-Argentine-Bonn (LAB; Kalberla et al. 2005) Survey andthe molecular hydrogen column density determined by Willingale et al. (2013) from Swift data. The tempera-tures and the normalizations of the thermal componentswere free in the fit, but linked between different datasets.The temperature and normalization of the LHB compo-nent are difficult to constrain from the
Chandra data(its temperature is ∼ . ROSAT spectrum extracted from an annuluswith radii 0 .
15 and 1 degrees around the cluster cen-ter (which is beyond R ). The index of the power-lawcomponent was fixed to 1.41 (De Luca & Molendi 2004).The normalizations of the power-law components of the Chandra spectra were free in the fit, but the power-lawnormalizations of ObsIDs 16235 and 16305 were linkedsince the observations were taken close in time. Thepower-law normalization of the
ROSAT spectrum wasfixed to 8 . × − photons keV − cm − s − arcmin − (Moretti et al. 2003). The instrumental background-subtracted spectra were modeled with xspec v12.8.2(Arnaud 1996). The Chandra spectra were binned toa minimum of 1 count/bin, and modeled using the ex-tended C-statistic (Cash 1979; Wachter et al. 1979). Thespectra were fitted in the energy band 0 . − Thebest-fitting sky background parameters are summarizedin Table 4. Throughout the paper, the uncertainties forthe X-ray derived quantities are quoted at the 1 σ level,unless explicitly stated. RADIO RESULTS3.1.
Continuum images AtomDB version is 2.0.2 The same energy band was used for all the spectral fits pre-sented in this paper. handra JVLA observations of MACS J0717.5+3745 Table 3 . Summary of the
Chandra observations.ObsID Instrument Mode Start date Exposure time (ks) Filtered exposure time (ks)1655 a ACIS-I FAINT 2001-01-29 19.9 15.84200 ACIS-I VFAINT 2004-01-10 59.0 52.616235 ACIS-I FAINT 2013-12-16 70.2 63.416305 ACIS-I VFAINT 2013-12-11 94.3 82.6 a ObsID 1655 was excluded from the analysis, see Section 2.2
Table 4 . Best-fitting X-ray sky background parameters.Model component Parameter Chandra ROSATLHB kT (keV) 0 .
135 (fixed) 0 . +0 . − . norm (cm − arcmin − ) 7 . × − (fixed) 7 . +0 . − . × − GH kT (keV) 0 . +0 . − . . +0 . − . norm (cm − arcmin − ) 2 . +0 . − . × − . +2 . − . × − Power-law Γ 1 .
41 (fixed) 1 .
41 (fixed)norm at 1 keV (photons keV − cm − s − arcmin − ) ObsID 4200 7 . +0 . − . × − . × − (fixed)ObsIDs 16235 4 . +0 . − . × − ObsIDs 16305 van Weeren et al.
The high-resolution (0 . (cid:48)(cid:48) × . (cid:48)(cid:48) ) 2–4 GHz S-bandimage is shown in Figure 1. Combined wide-band L-, S-,and C-band images at 1.6 (cid:48)(cid:48) and 5 (cid:48)(cid:48) resolution are shownin Figure 2. The most prominent source in the imagesis the large filamentary radio relic, with an embeddedNarrow Angle Tail (NAT) galaxy ( z = 0 . z = 0 . JVLA images at resolu-tions of 1.6 (cid:48)(cid:48) , 2.7 (cid:48)(cid:48) , and 5 (cid:48)(cid:48) – to highlight details aroundthe radio halo area – are shown in Figure 3.For the relic, we measure a largest linear size (LLS) of ≈
800 kpc, similar to previous studies. Our new imagesare significantly deeper and have better resolution thanprevious studies of this source. They reveal many newdetails in the relic and show that the relic has a signifi-cant amount of filamentary structure on scales down to ∼
30 kpc. Small scale filamentary structures have alsobeen seen for other relics, such as the Toothbrush clus-ter (van Weeren et al. 2012, 2016a), A3667 (R¨ottgeringet al. 1997), A3376 (Bagchi et al. 2006) and in particularfor Abell 2256 (Clarke & Enßlin 2006; van Weeren et al.2009a; Owen et al. 2014).We also note several narrow filaments of emission orig-inating from the relic. These are marked with red arrowsin Figure 3. These filaments have lengths of 50–150 kpcand widths as small as 10 kpc. In addition, there are twolarger regions of extended emission that are connectedto the radio relic. These extended regions are markedwith blue arrows.The radio halo component extends to the south andnorth of R3 and west of the R2 (see Figure 1 for thelabeling). Our images also reveal a significant amountof structure around the radio halo, including several fil-amentary features. They are marked with black arrowsin Figure 3. The brightest of these is connected with theR3 component of the main radio relic and has a LLS ofabout 200 kpc.Another prominent ∼
200 kpc NS elongated radio fil-ament is located at the northern outskirts of the cluster.This filament has a well defined boundary on its easternside, while it fades gradually towards the west. A fainter filament with a similar size and NS orientation is locatedNE of it. Evidence of two other filaments, with a NWorientation, are seen at the northern boundary of theradio halo. Another ∼
100 kpc long structure is locatedat the southern end of the radio halo. Three additionalembedded filamentary structures in the radio halo arefound west of R2. These are marked with dashed-lineblack arrows ( Figure 3). We also find two enhance-ments in the halo emission which we marked with whitedashed-line circles.Bonafede et al. (2009) suggested the presence of faintradio emission to the SE, along the large-scale galaxyfilament (not to be confused with the smaller radio fil-aments in the cluster) in the 325 MHz image from theWENSS survey (Rengelink et al. 1997). We do not findevidence for this in our deeper observations. We specu-late that the emission seen at 325 MHz could have beenthe result of blending of several compact sources dueto the low-resolution of the 325 MHz image. We alsonote that the emission is not found in the low-frequencyGMRT 610 and 235 MHz observations published byPandey-Pommier et al. (2013).3.1.1.
Spectral index maps
Spectral index maps at resolutions of 10 (cid:48)(cid:48) , 5 (cid:48)(cid:48) , 2.5 (cid:48)(cid:48) ,and 1.2 (cid:48)(cid:48) are shown in Figure 4. As explained inSection 2.1.1, these were made by fitting straightpower-laws through flux measurements at 1.5, 3.0, and5.5 GHz. The spectral index uncertainty maps areshown in Appendix A.The central NAT source shows clear evidence of spec-tral steepening in the higher resolution spectral indexmaps, from about − . − . (cid:48)(cid:48) resolution, as a function of distance fromthe radio core are displayed in Figure 5. In the lower res-olution spectral index maps this steepening is reduced,which is expected, since the reduced resolution causesmixing of emission from nearby regions (i.e., the relic)with flatter spectra. Evidence for spectral steepeningalong the tails is also found at the far SE tailed source( z = 0 . − . − . z = 0 . − . − . ∼ . (cid:48) ( ∼
80 kpc) from thecore), while the core has an inverted spectrum with α ≈ +0 .
5. Little steepening is seen along the lobesof the source, from about − . − .
8. The LAS of3.5 (cid:48) corresponds to a physical size of 560 kpc at thesource redshift.Radio relics often display spectral index gradients,with the spectral index steepening in the direction of handra JVLA observations of MACS J0717.5+3745 Figure 1 . S-band 2–4 GHz combined
JVLA
A-, B-, C-, and D-array image made with Briggs (1995) weighting ( robust=0 ). Theimage has a resolution of 0 . (cid:48)(cid:48) × . (cid:48)(cid:48) and a r.m.s. noise level of 1.9 µ Jy beam − . Components of the radio relic are labeledas in van Weeren et al. (2009b). van Weeren et al. the cluster center (e.g., Clarke & Enßlin 2006; Giacin-tucci et al. 2008; van Weeren et al. 2010; Bonafede et al.2012; Kale et al. 2012; Stroe et al. 2013). This spec-tral steepening is explained by synchrotron and InverseCompton (IC) losses in the post-shock region of an out-ward traveling shock front. We also find a spectral in-dex gradient in MACS J0717.5+3745, with the spectralindex decreasing from about − . (cid:46) − . − . − .
6. No clearspectral index trends were reported by Bonafede et al.(2009), but this can be explained by the lack of signal-to-noise compared to our new
JVLA observations.The spectral index distribution for the radio relicaround the central NAT source is more complex.This is not too surprising given that the relic inMACS J0717.5+3745 has an irregular asymmetric mor-phology, likely the result of the complex quadruplemerger event, and is projected relatively close to thecluster center, implying that the structure is not neces-sarily observed close to edge-on (Vazza et al. 2012).We also find evidence for EW spectral steepening,from about − . − .
5, across the brighter northernfilament (blue region, Figure 5). However, the uncer-tainties are significant as is indicated on the plot. Thebright filament just north of R4 has a flatter spectrum( − .
1) than the halo emission in the vicinity. This isalso the case for the filament at the southernmost partof the radio halo. The radio halo spectral index variesbetween − . ≈ − X-RAY RESULTS4.1.
Global X-ray Properties
In Figure 6, we present the
Chandra . − Chandra images with
JVLA radio contours and the surface mass density (derivedfrom a lensing analysis, Ishigaki et al. 2015) overlaid,are shown in Figure 7. An optical Subaru-CHFT imageoverlaid with X-ray contours is shown in Figure 9. Fourdifferent substructures (A–D) are labeled, following Maet al. (2009).The X-ray emission of the cluster is complex, consist- ing of a bar-shaped structure to the southeast with asize of 800 ×
300 kpc. The bar consists of two sepa-rate components (C and D, see Figures 6 and 9). Thesetwo components are likely associated with two separatemerging subclusters and are also detected in the masssurface density map. X-ray surface brightness profilesacross the bar along two rectangular boxes is presentedin Figure 8. These profiles show that the western edgeof the bar is cut off more abruptly than the eastern edge.We did not attempt to fit a density model to the edgebecause of the unknown (and likely complex) geometry.The brightest part of the ICM consists of a V-shapestructure, which is associated with major mass compo-nent B. To the northwest, an elongated, bullet-like X-raysubstructure is seen, with a sharp boundary on its north-ern edge. This structure seems to be associated withmass component A, and is also seen in the mass sur-face density map from Ishigaki et al. (2015). However,Johnson et al. (2014) and Limousin et al. (2016) placethe center of the westernmost mass component about0.5 (cid:48) east of the center of the X-ray component. We dis-cuss this “fly-through” bullet-like core in more detail inSection 4.3. A small “clump” of gas is found just northof the bar (again best seen in Figure 6, located at thecyan circle in Figure 7).From Figure 7 we find that the radio filament north ofR4 is aligned with the SW part of the V-shaped struc-ture. The southernmost radio filament (Figure 3) coin-cides with the southern end of the X-ray bar. The twonorthern filaments (north of R1) are located in the faintX-ray outskirts of the cluster.In the SE, the radio halo emission roughly follows theoutline of the bar. North of R4 the halos follows thebright X-ray region consisting of the V-shaped structureand emission north of it. The western part of the clusteris devoid of diffuse radio emission.To measure the global X-ray properties of the clus-ter, we extracted
Chandra spectra in a circle with aradius of R = 1 .
69 Mpc (Mantz et al. 2010) aroundRA = 07 h m . s ◦ (cid:48) (cid:48)(cid:48) . The spectrawere instrumental background-subtracted, and modeledas the sum of absorbed thermal ICM emission and skybackground emission. The sky background model wasfixed to the model summarized in Table 4. The tem-perature, metallicity, and normalization of the thermalcomponent describing ICM emission were left free in thefit.We measured T = 12 . +0 . − . keV, Z = 0 . ± . Z (cid:12) , and a 0 . − . . ± . × erg s − . 4.2. Temperature Map
To map the ICM temperature, we used
CONTBIN (Sanders 2006) to bin the surface brightness map handra JVLA observations of MACS J0717.5+3745 Figure 2 . Left: Deep wide-band combined L-, S-, and C-band image with a resolution of 1.6 (cid:48)(cid:48) . Contour levels are drawn at[1 , , , . . . ] × σ rms . These individual L-, S-, and C-band images were made with Briggs (1995) weighting ( robust=0 ). Right:Deep wide-band combined L-, S-, and C-band image with a resolution of 5 (cid:48)(cid:48) . Contours are plotted in the same way as in the leftpanel, with the exception of the lowest contour level. The lowest contour level comes from the 10 (cid:48)(cid:48) resolution image and is drawnat 5 σ rms . The 5 (cid:48)(cid:48) and 10 (cid:48)(cid:48) resolution images were made with uniform weighting and tapered to these respective resolutions.
383 kpc1 arcmin
Figure 3 . Wide-band 1.0–6.5 GHz images of the cluster at resolutions of 1.6, 2.7 and 5 (cid:48)(cid:48) . The weighting schemes to make the 1.6and 5 (cid:48)(cid:48) resolution images is given in the caption of Figure 2. The 2.7 (cid:48)(cid:48) resolution image was made with Briggs (1995) weighting( robust=0 ) and a uv-taper. These wide-band images reveal a significant amount of fine-scale structure in the extended radioemission. Narrow filaments extending from the relic are indicated with red arrows, diffuse components extending from the relicwith blue arrows, and (small) filaments in the general region of the halo with black (dashed-line) arrows. Two enhancements inthe radio halo emission are indicated with white dashed-line circles. van Weeren et al.van Weeren et al.
Figure 3 . Wide-band 1.0–6.5 GHz images of the cluster at resolutions of 1.6, 2.7 and 5 (cid:48)(cid:48) . The weighting schemes to make the 1.6and 5 (cid:48)(cid:48) resolution images is given in the caption of Figure 2. The 2.7 (cid:48)(cid:48) resolution image was made with Briggs (1995) weighting( robust=0 ) and a uv-taper. These wide-band images reveal a significant amount of fine-scale structure in the extended radioemission. Narrow filaments extending from the relic are indicated with red arrows, diffuse components extending from the relicwith blue arrows, and (small) filaments in the general region of the halo with black (dashed-line) arrows. Two enhancements inthe radio halo emission are indicated with white dashed-line circles. van Weeren et al.van Weeren et al. smoothed to a “signal”-to-noise of 10 in individual re-gions with a uniform “signal”-to-noise ratio of 55. Here,by “signal” we refer not only to the ICM signal, butrather to ICM and sky background signal combined; thenoise is the instrumental background emission. We ex-tracted total spectra and instrumental background spec-tra from each of the individual regions, and modeledthem as the sum of absorbed thermal emission from theICM and sky background emission. The parameters ofthe sky background model were fixed to the values inTable 4. The ICM metallicity was fixed to 0 . Z (cid:12) . Fig-ure 10 shows the resulting temperature map. An inter-active version of the map, which includes uncertaintieson the best-fitting spectral parameters at the 90% con-fidence level, is available at https://goo.gl/KtE33D .In Figure 10, we show the temperature map with over-laid X-ray and radio contours. Ma et al. (2009) arguedthat the V-shaped region (subcluster B) contains a coolcore remnant with a temperature of ∼ ∼
12 keV in the V-shapedregion. The results reported by Ma et al. (2009) werebased only on ObsID 4200. Neither using only ObsID4200, nor changing the region used to measure the tem-perature allowed us to obtain a temperature lower than8 keV (with the 90% confidence level uncertainties con-sidered). We also did a separate analysis that followedthat of Ma et al. (2009) more closely: we used blank-skyevent files, fitted the ICM with a MEKAL model, fixedthe abundance to 0 . . × cm − . Again, the temperature we obtainedwas above 9 keV at the 90% confidence level.Our temperature map reveals an extremely hot regionin the SSE part of the cluster center, with a tempera-ture (cid:38)
20 keV. This hot region is associated with thebar-shaped region of enhanced surface brightness seenin Figure 6. Ma et al. (2009) reported another possiblecool core remnant in the W part of this region, wherethey measured a temperature of 8 . ± . σ un-certainties, region A22 in their publication). This tem-perature was significantly lower than the temperaturesreported in adjacent regions, which all had >
15 keVgas. Choosing a region that approximates that of Maet al. (2009), we measure 13 . +4 . − . keV (1 σ uncertain-ties). While this temperature is consistent with thatmeasured by Ma et al. (2009), it is also consistent withthe temperatures of the adjacent regions.In conclusion, we find temperatures above ∼
10 keVthroughout the ICM, with a temperature peak of >
20 keV in the X-ray bright, bar-shaped region SSE ofthe radio relic. Similarly, Mroczkowski et al. (2012)did also not report temperatures below ∼
10 keV usingXMM-Newton and
Chandra observations of the cluster.Therefore, we do not confirm the temperatures of the cool regions reported by Ma et al. (2009). The V-shapedregion does seem to be cooler than its immediate sur-roundings, but not at the level as reported by Ma et al.(2009). 4.3.
Fly-Through Core
Approximately 0.7 Mpc NW from the cluster center,there is a X-ray core (Figure 11) with a tail extending ∼
200 kpc towards the SE, roughly in the direction of thelarge-scale galaxy filament in the SE. This morphologysuggests that this core, seen “flying” through the ICM ofMACS J0717.5+3745 and ram-pressured stripped by thecluster’s dense ICM, traveled NW along the SE filamentand is seen after it traversed the brightest ICM regions.In essence, the core is analogous to a later stage of thegroup currently seen within the filament.The core is embedded (at least in projection) in theICM of MACS J0717.5+3745. To determine the core’sphysical properties, we modeled the contamination fromthe ICM by extracting spectra N and S of the core.These spectra were modeled with a thermal componentwith a metallicity of 0.2 solar. We assumed the spectralproperties were the same in the N and S regions. Thespectra of the core were modeled as the sum of emis-sion from the contaminating ICM and from the coreitself. The spectra of the core and of the regions Nand S of it were modeled in parallel. The best-fittingresults are summarized in Table 6 and the regions areindicated on Figure 11. The temperature of the core,6 . +1 . − . keV, is consistent with the temperatures N andS of the core, in regions that are approximately at thesame distance from the cluster center as the core. Wealso compared the core temperature with the temper-atures ahead of (NW) and behind (SE) the core. Thetemperature decreases from 10 . +2 . − . keV behind thecore, to 5 . +1 . − . keV ahead of the core. From thesetemperature measurements, we therefore find no evi-dence of a core colder than its surroundings, nor ofa temperature discontinuity (either a shock or a coldfront) ahead of the core.A cold front and a shock front would be expectedahead of the core, similarly to the features seen in theBullet Cluster (Markevitch et al. 2002) and in front ofthe group NGC 4839 infalling into the Coma Cluster(Neumann et al. 2001), see also the review by Marke-vitch & Vikhlinin (2007). We searched for possible ev-idence of a cold/shock front by modeling the surfacebrightness profile of the group. The sector from whichthe surface brightness profiles was extracted is shown inFigure 12 (left panel). We chose an elliptical sector withan opening angle and ellipticity aligned with a possibleedge observed by eye in the surface brightness map. Themodel fitted to the surface brightness profile is shown inthe right panel of Figure 12. The surface brightness pro- handra JVLA observations of MACS J0717.5+3745 Table 5 . Best-fitting parameters of the broken power-law model fitted to the surface brightness of the core. Uncertainties arequoted at the 1 σ level. The region from which the surface brightness profile was extracted, as well as the profile and best-fittingmodel, are shown in Figure 12. α α Normalization r break n /n Sky background[photons cm − s − arcmin − ] [arcmin] [photons cm − s − arcmin − ] − . +1 . − . . +0 . − . . +1 . − . × − . +0 . − . . +0 . − . (1 . ± . × − Power-law index at r < r break . b Power-law index at r > r break . c Density jump across the discontinuity.
Table 6 . Parameters of the regions used for the spectralanalysis of the fly-through core. The regions are shown inFigure 11. Uncertainties are quoted at 1 σ level.Model Component Temperature a Normalization b Core 6 . +1 . − . . +0 . − . × − N+S of Core 7 . +1 . − . . +0 . − . × − Ahead of Core 5 . +1 . − . . +0 . − . × − Behind Core 10 . +2 . − . . +0 . − . × − Units of keV. b Units of cm − arcmin − for the thermal components, andphotons keV − cm − s − arcmin − at 1 keV for the power-law components. file extracted from the circular sector is well-fitted by abroken power-law density model. In this profile, there isan edge near ∼ . (cid:48) – 0 . (cid:48) . The best-fitting model has adensity jump of 3 . ± . ≈ . (cid:48) from the center ofthe sector. The best-fitting parameters for the brokenpower-law model are summarized in Table 5.The density discontinuity is at the very edge of thecore. Therefore, we speculate that the discontinuity isassociated with a cold front rather than with a shockfront. The failure to find a temperature discontinuityassociated with the density jump is likely due to poorcount statistics and emission from hot gas projected ontothe core. The latter also dilutes the observed densityjump, in which case our measurement of the jump am-plitude is only a lower limit . DISCUSSION5.1.
Origin of the radio relic
Radio relics are thought to trace relativistic elec-trons that are accelerated or re-accelerated at shocks.The presence of a powerful radio relic in the clusterMACS J0717.5+3745 is therefore consistent with thecluster undergoing a violent merger event. In fact,the
Chandra temperature map indicates that the relictraces a hot shock-heated region with temperatures of This applies to the situation were the emission from the hotgas exceeds the emission from outside the jump in the brokenpower-law model ∼
20 keV and higher. If we interpret the observed spec-tral index trends across the relic, Figure 5, as due to elec-trons cooling in the post-shock region, then the shockshould be located at the eastern boundary of the relicand the post-shock region is located to the west of that.We extracted temperatures on the eastern side of therelic ( T , the putative pre-shock region; regions 1 and 3)and around the putative shock downstream region ( T ;regions 2 and 4). The regions are indicated in Fig-ure 13. For the northern part of the relic, we find T = 20 . +5 . − . keV and T = 20 . +12 . − . keV (regions 1, 2).For the southern part we measure T = 27 . +8 . − . keV and T = 16 . +3 . − . − keV (regions 3, 4). So it is hard to sayfrom the temperatures where the pre- and post-shockregions are. Since the relic is at least partly located inthe cluster outskirts (the R1 and R2 part) and the X-rayemissivity is roughly proportional to the density squaredthe Chandra temperatures do not necessarily probe theactual pre- and post-shock gas but rather hot regions ofhigher density, with the relic projected close to it. Thisis particularly relevant for the southern part of the radiorelic. We also do not detect any X-ray surface bright-ness edges associated with the relic. This might implythat the shock surface is not seen very close to edge-on and/or projection effects are important, or the Machnumber is rather low.Therefore we conclude that given the complexity ofthe merger event and unknown projection effects, theprecise relation between the relic and location of thehot gas remains uncertain.5.1.1.
Acceleration mechanisms
For relics, an important question is by which mecha-nism the synchrotron emitting electrons are accelerated.The standard scenario proposed by Enßlin et al. (1998)is that particles are accelerated at shocks via the DSAmechanism. A problem with this scenario is that shocksin clusters generally have Mach number of M (cid:46)
3, andthe acceleration of electrons from the thermal pool isthought to be very inefficient for these low Mach num-bers, in apparent conflict with the presence of brightradio relics. In this case an unrealistic fraction of theenergy flux through the shock surface (Macario et al.2011; Eckert et al. 2016; van Weeren et al. 2016a) needs2 van Weeren et al.van Weeren et al.
3, andthe acceleration of electrons from the thermal pool isthought to be very inefficient for these low Mach num-bers, in apparent conflict with the presence of brightradio relics. In this case an unrealistic fraction of theenergy flux through the shock surface (Macario et al.2011; Eckert et al. 2016; van Weeren et al. 2016a) needs2 van Weeren et al.van Weeren et al.
Figure 4 . Spectral index maps at 10 (cid:48)(cid:48) , 5 (cid:48)(cid:48) , 2.5 (cid:48)(cid:48) , and 1.2 (cid:48)(cid:48) resolution (top left to bottom right). Black contours are drawn atlevels of [1 , , , . . . ] × σ rms and are from the S-band image. Pixels with values below 2 . σ rms in the individual maps wereblanked. The corresponding spectral index uncertainty maps are displayed in Figure A1. handra JVLA observations of MACS J0717.5+3745 γ -rays. Vazza & Br¨uggen(2014); Vazza et al. (2015, 2016) show that the ob-served γ -rays upper limits are in tension with the rela-tive acceleration efficiency of electrons and protons thatis expected from DSA. Tension with DSA has also beenfound from the discrepancy between the measured Machnumbers from X-ray observations and the radio spectralindex (see Equation 3) for some relics (e.g., Itahana et al.2015; Akamatsu et al. 2015; van Weeren et al. 2016a).PIC simulations show that electrons can be acceler-ated from the thermal pool via the SDA mechanism,which would solve some of the problems with DSA (Guoet al. 2014a,b; Caprioli & Spitkovsky 2014). Anothermodel to solve the low acceleration efficiency of stan-dard DSA is that of re-acceleration of fossil electrons(e.g., Markevitch et al. 2005; Giacintucci et al. 2008;Kang & Ryu 2011; Kang et al. 2012; Pinzke et al. 2013).These fossil electrons could, for example, originate fromthe (old) lobes of radio galaxies. Indeed observationsprovide some support for this scenario because of thecomplex morphologies of some relics, suggesting a linkwith a nearby radio galaxy in a few select cases (Giovan-nini et al. 1991; van Weeren et al. 2013; Bonafede et al.2014; Shimwell et al. 2015). The most compelling casefor re-acceleration has been found in the merging clusterAbell 3411-3412 (van Weeren et al. 2017). Here a tailedradio galaxy is seen connected to a relic. In additionspectral flattening is observed at the location where thefossil plasma meets the relic and at the same locationan X-ray surface brightness edge is observed.5.1.2. Evidence for re-acceleration in MACS J0717.5+3745
We argue that the NAT galaxy inMACS J0717.5+3745 provides another compellingcase for particle re-acceleration because (1) the NATgalaxy is a spectroscopically confirmed cluster mem-ber,(2) we observe a morphological connection betweenthe relic and NAT source, (3) there is evidence for hotshock-heated gas at the location of the radio relic (withthe caveat of unknown projection effects), and (4) wecan trace the spectral index across the tails of thisgalaxy until they fade into the relic. After fading intothe relic the spectral index flattens again (Figure 5,right panel magenta points), as is expected in the caseof re-acceleration.For a NAT source, we expect to start with a power-law radio spectrum, the radio spectrum then steepensprogressively along the tails of the NAT source due tosynchrotron and IC losses. Apart from spectral steepen-ing, the spectral curvature should also increase along thetails due to these energy losses. When the fossil electrons pass through the shock, they are re-accelerated and thespectral index flattens again. In MACS J0717.5+3745we observe this expected trend.In the case of re-acceleration, the radio injection spec-tral index is set by the Mach number of the shock, unlessthe index of the fossil distribution is flatter than whatwould be produced by the re-acceleration process. Fol-lowing Markevitch et al. (2005), we start with a power-law momentum fossil electron distribution f fossil ( p ) ∝ p − s fossil , (1)the distribution after re-acceleration (not consideringenergy losses) can be given by f inj , re ( p ) ∝ p − s inj , re . (2)For DSA the injection index is given by s inj , dsa = 2 M + 1 M − s fossil < s inj , dsa then s inj , re = s fossil .Thus for weak shocks, or a flat distribution of fossilplasma, the shape of the radio spectrum will be pre-served under re-acceleration. If s fossil > s inj , dsa we have s inj , re = s inj , dsa , so spectral shape is what we wouldnormally expect from DSA. Note that the radio spec-tral index is related to electron momentum distribution(with index s ) as α = − ( s − /
2. In summary, for re-acceleration the index of the momentum distribution isgiven by s inj , re = s fossil for s fossil < M +1 M − s inj , dsa for s fossil > M +1 M − . (4)At the location where the NAT source inMACS J0717.5+3745 fades into the relic, the spectralindex is steep with α (cid:46) − s (cid:38)
5) and that wouldsuggest that we are in the regime s fossil > s inj , dsa andthe spectral index of the relic follows what would beexpected in the case of DSA. If the spectral index is setby the Mach number, we would need at least a shockwith M = 2 . α inj = − . M (cid:38) . M = 2 . ∼ Shape of the fossil electron distribution beforere-acceleration van Weeren et al.van Weeren et al.
5) and that wouldsuggest that we are in the regime s fossil > s inj , dsa andthe spectral index of the relic follows what would beexpected in the case of DSA. If the spectral index is setby the Mach number, we would need at least a shockwith M = 2 . α inj = − . M (cid:38) . M = 2 . ∼ Shape of the fossil electron distribution beforere-acceleration van Weeren et al.van Weeren et al. Figure 5 . Left: Regions where spectral indices (shown in the right panel) were extracted. The regions have a width of 2.5 (cid:48)(cid:48) .The region’s colors are matched to the colored data points in the right panel. Right: Computed spectral indices between 5.5and 1.5 GHz in the various regions indicated in the left panel (note that the 3 GHz flux densities were not used to computethe spectral indices in this figure so that we simply have a single spectral index value between the two most extreme frequencypoints). We used the maps at 2.5 (cid:48)(cid:48) resolution (which were also used to compute the spectral index maps). The distance isincreasing from east to west (left to right). The errors shown on this plot only include the statistical uncertainties due to theimage noise (a systematic uncertainty would affect all plotted points in the same way). The x-axis values are offset to aid thevisibility.
Figure 6 . Chandra . − (cid:48)(cid:48) . The dashed-line circle shows R for the cluster. handra JVLA observations of MACS J0717.5+3745 ν break . We fix the injec-tion spectral index to α inj = − .
5. The observed spectrado not show evidence for a strong spectral break. In Fig-ure 14 we plot a model with ν break = 2 GHz. This lackof a spectral cutoff indicates that the spectral ageing ismore complex (e.g., spatially varying magnetic fields)or that there is mixing of radio emission with differentspectra within our measurement regions. This mixingreduces the curvature and moves the spectra closer topower-law shapes (e.g., van Weeren et al. 2012). There-fore the curvature of the fossil particle spectrum remainsunclear, but we can at least conclude that the spectrumis steep. The shape of the fossil distribution will be im-portant input for future modeling and simulations (e.g.,Kang & Ryu 2015; Hong et al. 2015).5.2. Origin of the radio halo and filamentary structures
An interesting question concerns the origin and natureof the radio filaments in the general radio halo area.Are these embedded in the radio halo emission, tracingregions with increased turbulence, or are they similar tothe large relics that trace (re)-accelerated particles atshocks? In the second scenario an additional question iswhether they trace shocks in the denser regions of theICM, or shocks in the cluster outskirts (and in whichcase they can be projected onto the cluster center andradio halo region). The filaments could also only beregions of enhanced magnetic fields, i.e. flux tubes orlarge-scale strands of field.It seems that a shock-origin is preferred, at least forsome of the filaments. There are severe reasons for this(1) the northern filament (above R1), which is locatedin the cluster outskirts, shows an EW spectral indexgradient, and has has a well defined eastern boundary;and (2) the filament above R4 is connected with themain radio relic. So at least two of these filaments areprobably not directly associated with the radio halo. Inaddition, it is possible that the regions indicated withthe white dashed-line circles (Figure 3) are additionalfilaments but projected closer to face-on. However, theycould also just be regions of enhanced magnetic fields.Polarization measurements would provide additionalinformation on the filaments. A high ( (cid:38)
Merger scenario
MACS J0717.5+3745 consists of at least four merg-ing subclusters, as indicated in Figure 9. Subcluster B,corresponding to the V-shaped structure in the
Chan-dra images, has a large line of sight velocity of about3,200 km s − away from us. We speculate that the V-shape could be related to a bullet-like structure seenunder a large projection angle. This would also explainthe lack of an offset between the X-ray gas, dark mat-ter, and galaxies, and is consistent with the large radialvelocity component and the detected kinetic SZ signal(Mroczkowski et al. 2012; Sayers et al. 2013). Interest-ingly, the radio filament above R4 is aligned with theV-shape and is located immediately to the south of it.This could just be a chance alignment. Another pos-sibility is that this filament traces the shock ahead ofsubcluster B.For subcluster D, the galaxy and dark matter peaksare located about 0.4 (cid:48) NW from the X-ray peak of thesubcluster. An offset in this direction would be expecteddue to the effect of ram pressure on the gas (as also sug-gested by Ma et al. 2009), if subcluster D fell in fromthe large-scale galaxy filament to the SE. No clear off-set, between the X-ray peak and dark matter peak, isseen for subcluster C. Adam et al. (2016) reported thedetection of a kinetic SZ signal from subcluster C, withan opposite line of sight velocity with respect to sub-cluster B.Ma et al. (2009) suggested that subcluster A (the fly-through core) fell in from the NW. However, the de-tection of an X-ray edge to the NNE, likely a mergerrelated cold front, suggests that the cluster fell in fromthe SE and the X-ray gas is moving to the N-NW. Thisdirection would be consistent with infall from the large-scale filament to the southeast. Its elongated shape indi-cates it is currently in the process of being ram pressurestripped, see Section 4.3. The associated BCG is locatedslightly (0.2 (cid:48) –0.3 (cid:48) ) to the SE of the X-ray peak. This isdifferent from the situation in the bullet Cluster (Cloweet al. 2006), where the galaxies lead the bullet. Thiscould imply that the dark matter and galaxies are al-ready past pericenter and in the “return phase” of themerger (Ng et al. 2015). Interestingly, the dark matterpeak is located even further to the east (as reported by6 van Weeren et al.van Weeren et al.
Chan-dra images, has a large line of sight velocity of about3,200 km s − away from us. We speculate that the V-shape could be related to a bullet-like structure seenunder a large projection angle. This would also explainthe lack of an offset between the X-ray gas, dark mat-ter, and galaxies, and is consistent with the large radialvelocity component and the detected kinetic SZ signal(Mroczkowski et al. 2012; Sayers et al. 2013). Interest-ingly, the radio filament above R4 is aligned with theV-shape and is located immediately to the south of it.This could just be a chance alignment. Another pos-sibility is that this filament traces the shock ahead ofsubcluster B.For subcluster D, the galaxy and dark matter peaksare located about 0.4 (cid:48) NW from the X-ray peak of thesubcluster. An offset in this direction would be expecteddue to the effect of ram pressure on the gas (as also sug-gested by Ma et al. 2009), if subcluster D fell in fromthe large-scale galaxy filament to the SE. No clear off-set, between the X-ray peak and dark matter peak, isseen for subcluster C. Adam et al. (2016) reported thedetection of a kinetic SZ signal from subcluster C, withan opposite line of sight velocity with respect to sub-cluster B.Ma et al. (2009) suggested that subcluster A (the fly-through core) fell in from the NW. However, the de-tection of an X-ray edge to the NNE, likely a mergerrelated cold front, suggests that the cluster fell in fromthe SE and the X-ray gas is moving to the N-NW. Thisdirection would be consistent with infall from the large-scale filament to the southeast. Its elongated shape indi-cates it is currently in the process of being ram pressurestripped, see Section 4.3. The associated BCG is locatedslightly (0.2 (cid:48) –0.3 (cid:48) ) to the SE of the X-ray peak. This isdifferent from the situation in the bullet Cluster (Cloweet al. 2006), where the galaxies lead the bullet. Thiscould imply that the dark matter and galaxies are al-ready past pericenter and in the “return phase” of themerger (Ng et al. 2015). Interestingly, the dark matterpeak is located even further to the east (as reported by6 van Weeren et al.van Weeren et al.
Figure 7 . Left:
Chandra (cid:48)(cid:48) resolution image and drawnat [1 , , , . . . ] × σ rms . Right: Same image as in the left panel but with the convergence map κ = ΣΣ cr (with Σ (cr) the (critical)mass surface density density) overlaid from Ishigaki et al. (2015). Contour levels are drawn at κ = [1 , . , , × .
8. Thepositions of several mass components from Johnson et al. (2014) and Limousin et al. (2016) are indicated with black and bluecircles, respectively. The cyan circle corresponds to an individual (massive) cluster galaxy (Johnson et al. 2014). D e c ( J ) REG1REG2
Figure 8 . X-ray surface brightness profiles across the bar (SE to NW) in two regions as indicated in the right panel. The barshows a hint of an edge on its western side, located at a distance of about 2.4 (cid:48) . The instrumental background is shown in green,with the uncertainty ranges on the background shown in dashed green lines. handra JVLA observations of MACS J0717.5+3745 (cid:48) ( ≈
150 kpc)and a temperature of 6.8 keV, we compute a sound cross-ing time of 1 × yr for the core. If the X-ray emission isindeed displaced from the dark matter the X-ray clumpwill disperse over the next ∼ yr (assuming there isno dark matter to hold it together). CONCLUSIONSWe presented deep
JVLA and
Chandra observationsof the HST Frontier Fields cluster MACS J0717.5+3745.The radio and X-ray observations show a complexmerger event, involving multiple subclusters. Below wesummarize our findings: • The X-ray temperature map shows that the east-ern part of the cluster is significantly hotter thanthe western part. In the central southeastern partof the cluster the temperatures exceed ∼
20 keV.The hot eastern part of the cluster coincides withthe location of the radio halo and relic. • We find no evidence for the ICM temperatures sig-nificantly less than 10 keV that were reported byMa et al. (2009). • The NW subcluster displays a ram pressure-stripped core, with a surface brightness edge tothe NNE. We speculate that this edge is likely amerger related cold front. • We find evidence that the radio relic inMACS J0717.5+3745 is powered by shock re-acceleration of fossil electrons from a nearby NATsource. • We find an overall EW spectral index gradientacross the radio relic, with the spectral indexsteepening towards the west. • We do not detect density or temperatures jumpsassociated with the radio relic, which could be theresult of the complex merger geometry. Alterna-tively, for re-acceleration the shock Mach numbercould be lower than the M = 2 . • We find several radio filaments in the cluster withsizes of about 100–300 kpc. At least a few of theseare located in the cluster outskirts. That wouldsuggest the filaments are tracing shock waves (andcan thus be classified as a small radio relics). Po-larization observations should provide more infor-mation about the origin and location of these fila-ments within the ICM.
Acknowledgments:
We thank the anonymous refereefor useful comments. The National Radio AstronomyObservatory is a facility of the National Science Foun-dation operated under cooperative agreement by Asso-ciated Universities, Inc. Support for this work was pro-vided by the National Aeronautics and Space Adminis-tration through Chandra Award Number GO4-15129Xissued by the Chandra X-ray Observatory Center, whichis operated by the Smithsonian Astrophysical Observa-tory for and on behalf of the National Aeronautics SpaceAdministration under contract NAS8-03060.R.J.W. is supported by a Clay Fellowship awarded bythe Harvard-Smithsonian Center for Astrophysics. M.Backnowledge support by the research group FOR 1254funded by the Deutsche Forschungsgemeinschaft: “Mag-netisation of interstellar and intergalactic media: theprospects of low-frequency radio observations”. W.R.F.,C.J., and F.A-S. acknowledge support from the Smith-sonian Institution. E.R. acknowledges a Visiting Sci-entist Fellowship of the Smithsonian Astrophysical Ob-servatory, and the hospitality of the Center for Astro-physics in Cambridge. G.A.O. acknowledges support byNASA through a Hubble Fellowship grant HST-HF2-51345.001-A awarded by the Space Telescope ScienceInstitute, which is operated by the Association of Uni-versities for Research in Astronomy, Incorporated, un-der NASA contract NAS5-26555. F.A-S. acknowledgessupport from Chandra grant GO3-14131X. A.Z. is sup-ported by NASA through Hubble Fellowship grant HST-HF2-51334.001-A awarded by STScI. This research wasperformed while T.M. held a National Research CouncilResearch Associateship Award at the Naval ResearchLaboratory (NRL). Basic research in radio astronomyat NRL by T.M. and T.E.C. is supported by 6.1 Basefunding. M.D. acknowledges the support of STScI grant12065.007-A. P.E.J.N. was partially supported by NASAcontract NAS8-03060. Part of this work performed un-der the auspices of the U.S. DOE by LLNL under Con-tract DE-AC52-07NA27344.Part of the reported results are based on observationsmade with the NASA/ESA Hubble Space Telescope, ob-tained from the Data Archive at the Space TelescopeScience Institute. STScI is operated by the Associationof Universities for Research in Astronomy, Inc. underNASA contract NAS 5-26555. This work utilizes gravi-tational lensing models produced by PIs Bradaˇc, Ebel-ing, Merten & Zitrin, Sharon, and Williams funded aspart of the HST Frontier Fields program conducted bySTScI. The lens models were obtained from the MikulskiArchive for Space Telescopes (MAST).
Facilities:
Facility:
VLA,8 van Weeren et al.van Weeren et al.
VLA,8 van Weeren et al.van Weeren et al.
Figure 9 . Subaru B, I, and CFHT Ks band color image of MACS J0717.5+3745 (Medezinski et al. 2013; Umetsu et al. 2014).
Chandra ∝ ( n ) / , with n = [1 , , , . . . ]. The subclusters A–D are labeled as in Ma et al. (2009). Figure 10 . Temperature map of MACS J0717.5+3745. Overlaid are
Chandra . − JVLA radio contours (right; from Figure 3 middle panel). The X-ray contours are drawn at[0 . , . , . , . , . , . , . , . × − photons cm − s − . The radio contours are drawn at [1 , , . . . ] × σ rms . handra JVLA observations of MACS J0717.5+3745 D e c ( J ) Figure 11 . Regions used in the spectral analysis. The re-gions of main interest are drawn in solid lines, while theregions used to characterize the contaminating/surroundingemission are drawn in dashed lines. The best-fitting parame-ters obtained for the gas in these regions are listed in Table 6.
Facility:
CXOAPPENDIX A. SPECTRAL INDEX UNCERTAINTY MAPSSpectral index uncertainty maps are shown in Figure A1 corresponding to a power-law fits through flux measurementsat 1.5, 3.0 and 5.5 GHz. The errors are based on the individual rms noise values in the maps and an absolute fluxcalibration uncertainty of 2.5% at each of the three frequencies.REFERENCES
Adam, R., Bartalucci, I., Pratt, G. W., et al. 2016, ArXiv:1606.07721Akamatsu, H., van Weeren, R. J., Ogrean, G. A., et al. 2015,A&A, 582, A87Arnaud, K. A. 1996, in Astronomical Society of the PacificConference Series, Vol. 101, Astronomical Data AnalysisSoftware and Systems V, ed. G. H. Jacoby & J. Barnes, 17Bagchi, J., Durret, F., Neto, G. B. L., & Paul, S. 2006, Science,314, 791Bhatnagar, S., Rau, U., & Golap, K. 2013, ApJ, 770, 91Bonafede, A., Intema, H. T., Br¨uggen, M., et al. 2014, ApJ, 785,1Bonafede, A., Feretti, L., Giovannini, G., et al. 2009, A&A, 503,707Bonafede, A., Br¨uggen, M., van Weeren, R., et al. 2012,MNRAS, 426, 40Briggs, D. S. 1995, PhD thesis, New Mexico Institute of MiningTechnology, Socorro, New Mexico, USACaprioli, D., & Spitkovsky, A. 2014, ApJ, 783, 91Cash, W. 1979, ApJ, 228, 939Cassano, R., Ettori, S., Brunetti, G., et al. 2013, ApJ, 777, 141Clarke, T. E., & Enßlin, T. A. 2006, AJ, 131, 2900Clowe, D., Bradaˇc, M., Gonzalez, A. H., et al. 2006, ApJL, 648,L109 Cornwell, T. J. 2008, IEEE Journal of Selected Topics in SignalProcessing, vol. 2, issue 5, pp. 793-801, 2, 793Cornwell, T. J., Golap, K., & Bhatnagar, S. 2005, inAstronomical Society of the Pacific Conference Series, Vol.347, Astronomical Data Analysis Software and Systems XIV,ed. P. Shopbell, M. Britton, & R. Ebert, 86–+Cornwell, T. J., Golap, K., & Bhatnagar, S. 2008, IEEE Journalof Selected Topics in Signal Processing, vol. 2, issue 5,pp. 647-657, 2, 647De Luca, A., & Molendi, S. 2004, A&A, 419, 837Diego, J. M., Broadhurst, T., Zitrin, A., et al. 2015, MNRAS,451, 3920Drury, L. O. 1983, Reports on Progress in Physics, 46, 973Ebeling, H., Barrett, E., & Donovan, D. 2004, ApJL, 609, L49Ebeling, H., Barrett, E., Donovan, D., et al. 2007, ApJL, 661,L33Ebeling, H., Edge, A. C., & Henry, J. P. 2001, ApJ, 553, 668Ebeling, H., Ma, C.-J., & Barrett, E. 2014, ApJS, 211, 21Ebeling, H., White, D. A., & Rangarajan, F. V. N. 2006,MNRAS, 368, 65Eckert, D., Jauzac, M., Vazza, F., et al. 2016, MNRAS, 461, 1302Edge, A. C., Ebeling, H., Bremer, M., et al. 2003, MNRAS, 339,913Enßlin, T. A., Biermann, P. L., Klein, U., & Kohle, S. 1998,A&A, 332, 395 van Weeren et al.van Weeren et al.
Adam, R., Bartalucci, I., Pratt, G. W., et al. 2016, ArXiv:1606.07721Akamatsu, H., van Weeren, R. J., Ogrean, G. A., et al. 2015,A&A, 582, A87Arnaud, K. A. 1996, in Astronomical Society of the PacificConference Series, Vol. 101, Astronomical Data AnalysisSoftware and Systems V, ed. G. H. Jacoby & J. Barnes, 17Bagchi, J., Durret, F., Neto, G. B. L., & Paul, S. 2006, Science,314, 791Bhatnagar, S., Rau, U., & Golap, K. 2013, ApJ, 770, 91Bonafede, A., Intema, H. T., Br¨uggen, M., et al. 2014, ApJ, 785,1Bonafede, A., Feretti, L., Giovannini, G., et al. 2009, A&A, 503,707Bonafede, A., Br¨uggen, M., van Weeren, R., et al. 2012,MNRAS, 426, 40Briggs, D. S. 1995, PhD thesis, New Mexico Institute of MiningTechnology, Socorro, New Mexico, USACaprioli, D., & Spitkovsky, A. 2014, ApJ, 783, 91Cash, W. 1979, ApJ, 228, 939Cassano, R., Ettori, S., Brunetti, G., et al. 2013, ApJ, 777, 141Clarke, T. E., & Enßlin, T. A. 2006, AJ, 131, 2900Clowe, D., Bradaˇc, M., Gonzalez, A. H., et al. 2006, ApJL, 648,L109 Cornwell, T. J. 2008, IEEE Journal of Selected Topics in SignalProcessing, vol. 2, issue 5, pp. 793-801, 2, 793Cornwell, T. J., Golap, K., & Bhatnagar, S. 2005, inAstronomical Society of the Pacific Conference Series, Vol.347, Astronomical Data Analysis Software and Systems XIV,ed. P. Shopbell, M. Britton, & R. Ebert, 86–+Cornwell, T. J., Golap, K., & Bhatnagar, S. 2008, IEEE Journalof Selected Topics in Signal Processing, vol. 2, issue 5,pp. 647-657, 2, 647De Luca, A., & Molendi, S. 2004, A&A, 419, 837Diego, J. M., Broadhurst, T., Zitrin, A., et al. 2015, MNRAS,451, 3920Drury, L. O. 1983, Reports on Progress in Physics, 46, 973Ebeling, H., Barrett, E., & Donovan, D. 2004, ApJL, 609, L49Ebeling, H., Barrett, E., Donovan, D., et al. 2007, ApJL, 661,L33Ebeling, H., Edge, A. C., & Henry, J. P. 2001, ApJ, 553, 668Ebeling, H., Ma, C.-J., & Barrett, E. 2014, ApJS, 211, 21Ebeling, H., White, D. A., & Rangarajan, F. V. N. 2006,MNRAS, 368, 65Eckert, D., Jauzac, M., Vazza, F., et al. 2016, MNRAS, 461, 1302Edge, A. C., Ebeling, H., Bremer, M., et al. 2003, MNRAS, 339,913Enßlin, T. A., Biermann, P. L., Klein, U., & Kohle, S. 1998,A&A, 332, 395 van Weeren et al.van Weeren et al. D e c ( J ) Figure 12 . Left:
Sector used to model the surface brightness profile in front of the core [GO: needs to be updated].
Right:
Surface brightness profiles and best-fitting model. The instrumental background is shown in blue, with the uncertainty rangeson the background shown in dashed blue lines. For the elliptical sector, the surface brightness is plotted against the major axisof the ellipse. The best-fitting parameters of the broken power-law model are listed in Table 5. region 4region 2region 3region 1
Figure 13 . Regions where we extracted the temperatures ontop of the wide-band 1.0–6.5 GHz radio image.
Fanaroff, B. L., & Riley, J. M. 1974, MNRAS, 167, 31PFeldman, U. 1992, PhyS, 46, 202Fujita, Y., Takizawa, M., Yamazaki, R., Akamatsu, H., & Ohno,H. 2015, ApJ, 815, 116Giacintucci, S., Venturi, T., Macario, G., et al. 2008, A&A, 486,347Giovannini, G., Feretti, L., & Stanghellini, C. 1991, A&A, 252,528Guo, X., Sironi, L., & Narayan, R. 2014a, ApJ, 794, 153—. 2014b, ApJ, 797, 47Hong, S. E., Kang, H., & Ryu, D. 2015, ApJ, 812, 49Ishigaki, M., Kawamata, R., Ouchi, M., et al. 2015, ApJ, 799, 12Itahana, M., Takizawa, M., Akamatsu, H., et al. 2015, PASJ, 67,113
Figure 14 . Radio spectra across the NAT source at the cen-ter of the relic. The normalizations for all spectra are arbi-trary. The upper five spectra are extracted in the magentaregions indicated in Figure 5, starting at the easternmost re-gion. The dotted lines are to guide the eye and connect theflux density measurements at 1.5, 3.0 and 5.5 GHz. The solidblack line corresponds to a power-law model with an expo-nential cutoff with ν break = 2 GHz. The model spectrum isnormalized at our 3.0 GHz flux density measurement. Thebottom spectrum, represented by the dashed line, was com-puted by combining the fluxes from the easternmost relic re-gions with black and red colors (see Figure 5). These parts ofthe relic contain the flattest spectral indices and thus likelycome closest to resembling the radio relic “injection spec-trum”. Spectral index values, between 1,5 and 5.5 GHz, areshown to the left of the spectra. handra JVLA observations of MACS J0717.5+3745 Figure A1 . Spectral index uncertainty maps, corresponding to Figure 4, at resolutions of 10 (cid:48)(cid:48) , 5 (cid:48)(cid:48) , 2.5 (cid:48)(cid:48) , and 1.2 (cid:48)(cid:48) . Jauzac, M., Jullo, E., Kneib, J.-P., et al. 2012, MNRAS, 426,3369Johnson, T. L., Sharon, K., Bayliss, M. B., et al. 2014, ApJ, 797,48Kalberla, P. M. W., Burton, W. B., Hartmann, D., et al. 2005,A&A, 440, 775Kale, R., Dwarakanath, K. S., Bagchi, J., & Paul, S. 2012,MNRAS, 426, 1204Kang, H., & Ryu, D. 2011, ApJ, 734, 18—. 2015, ApJ, 809, 186—. 2016, ApJ, 823, 13Kang, H., Ryu, D., & Jones, T. W. 2012, ApJ, 756, 97Limousin, M., Ebeling, H., Richard, J., et al. 2012, A&A, 544,A71Limousin, M., Richard, J., Jullo, E., et al. 2016, A&A, 588, A99 Lotz, J., Mountain, M., Grogin, N. A., et al. 2014, in AmericanAstronomical Society Meeting Abstracts, Vol. 223, AmericanAstronomical Society Meeting Abstracts van Weeren et al.van Weeren et al.