The highest-frequency detection of a radio relic: 16-GHz AMI observations of the `Sausage' cluster
Andra Stroe, Clare Rumsey, Jeremy J. Harwood, Reinout van Weeren, Huub J. A. Röttgering, Richard D. E. Saunders, David Sobral, Yvette C. Perrott, Michel P. Schammel
MMon. Not. R. Astron. Soc. , 000–000 (0000) Printed 31 August 2018 (MN L A TEX style file v2.2)
The highest-frequency detection of a radio relic: 16-GHz AMIobservations of the ‘Sausage’ cluster
Andra Stroe (cid:63) , Clare Rumsey , Jeremy J. Harwood , Reinout van Weeren ,Huub J. A. R ¨ottgering , Richard D. E. Saunders , , David Sobral , Yvette C. Perrott ,Michel P. Schammel , Leiden Observatory, Leiden University, P.O. Box 9513, NL-2300 RA Leiden, The Netherlands Astrophysics Group, Cavendish Laboratory, JJ Thomson Avenue, Cambridge, CB3 0HE School of Physics, Astronomy and Mathematics, University of Hertfordshire, College Lane, Hatfield, Hertfordshire AL10 9AB, UK Harvard Smithsonian Center for Astrophysics (CfA - SAO), 60 Garden Street Cambridge, MA 02138, US Kavli Institute for Cosmology Cambridge, Madingley Road, Cambridge CB3 0HA, UK I.N.A.F. - Osservatorio Astronomico di Roma, via Frascati 33, 00040 - Monte Porzio Catone (Roma), Italy
31 August 2018
ABSTRACT
We observed the cluster CIZA J2242.8+5301 with the Arcminute Microkelvin Imager at 16GHz and present the first high radio-frequency detection of diffuse, non-thermal cluster emis-sion. This cluster hosts a variety of bright, extended, steep-spectrum synchrotron-emittingradio sources, associated with the intra-cluster medium, called radio relics. Most notably, thenorthern, Mpc-wide, narrow relic provides strong evidence for diffusive shock accelerationin clusters. We detect a puzzling, flat-spectrum, diffuse extension of the southern relic, whichis not visible in the lower radio-frequency maps. The northern radio relic is unequivocallydetected and measures an integrated flux of 1 . ± . < Key words: acceleration of particles, radiation mechanisms: non-thermal, shock waves,galaxies: clusters: individual: CIZA J2242.8+5301, radio continuum: general
Radio relics are diffuse, strongly-polarised, Mpc-wide synchrotronobjects found at the periphery of disturbed galaxy clusters (e.g.Feretti et al. 2001). Relics are thought to trace large-scale, fast,outward-travelling shock fronts (Mach numbers up to 4) inducedby major mergers between massive clusters (Ensslin et al. 1998;Sarazin 2002; Feretti et al. 2012). These objects usually extend per-pendicularly to the merger axis of their host cluster and display nar-row transverse sizes, resulting from a spherical-cap-shaped regionsof diffuse emission seen side-on in projection (Feretti et al. 2012).Integrated radio spectral indices of elongated relics below < . − . < α < − . F ν ∝ ν α ) and the spectradisplay no curvature up to ∼ µ G level magnetic fields at theoutskirts of clusters (e.g. Bonafede et al. 2009, 2010). Due to low (cid:63)
E-mail: [email protected] acceleration efficiencies, mildly-relativistic (rather than thermal)electrons likely cross the shock surface multiple times by diffusingback through the shock after each passage. These re-acceleratedelectrons then exhibit synchrotron radio emission.CIZA J2242.8+5301 (‘Sausage’ cluster; Kocevski et al. 2007;van Weeren et al. 2010) hosts a remarkable example of double,Mpc-wide, narrow radio relics. Twin relics are thought to form aftera head-on collision of two roughly equal-mass clusters (Roettigeret al. 1999). The northern relic (RN) is bright (0 .
15 Jy at 1 . . α int = . ± .
04 (Stroe et al. 2013). RN displays spectral in-dex steepening and increasing curvature from the outer edge of therelic towards the inner edge, thought to be due to synchrotron andinverse Compton losses in the downstream area of a shock with aninjection spectral index of ∼ − .
65. The cluster contains a faintercounter-relic towards the south, a variety of diffuse patches of emis-sion and a number of radio head-tail galaxies (Stroe et al. 2013).Relics have been primarily studied at low radio frequencies( < . ∼
40 radiorelics with published spectra (Feretti et al. 2012) have measure- c (cid:13) a r X i v : . [ a s t r o - ph . C O ] M a r A. Stroe et al. ments up to 2 . z = . .
191 Mpc. All images are in the J2000 coordinate system.
For our analysis, we combine the existing WSRT and GMRT obser-vations with new AMI observations. We use the WSRT and GMRTdatasets presented in Stroe et al. (2013) and refer the reader to thatpaper for details of the data reduction. In summary, the data wereflagged, bandpass and gain calibrated and bright sources in the fieldwere removed using the ‘peeling’ technique (Noordam 2004). Atotal of three frequencies were observed with the GMRT: 153, 323and 608 MHz and four with the WSRT: 1 .
2, 1 .
4, 1 . . AMI is a dual array of interferometers located near Cambridge,UK. The Small Array (SA) and the Large Array (LA), observe over13 . − . . −
20 m, while the LA has eight 12 . −
110 m, giving the instruments sen-sitivities to complimentary ranges of angular scale. Observationstowards the X-ray cluster centre were taken between July 2012 andFebruary 2013 on both SA and LA, with the field observed with asingle pointing with the SA and with a series of multi-point hexago-nal raster observations on the LA. The northern relic itself was alsoobserved with the LA with four pointings centred along its axis. Forall observations, flux calibration was performed using observationsof 3C 48, 3C 286 and 3C 147, with 3C 286 calibrated against VLAmeasurements (Perley & Butler 2013). Raw data were flagged forhardware errors, shadowing and interference and phase and ampli-tude calibrated using the in-house software package
REDUCE (AMIConsortium: Davies et al. 2009). All of the reduced LA data wereconcatenated into a single uv dataset before mapping.Unlike the WSRT and GMRT arrays, that measure total in-tensity I, both AMI arrays measure the single polarisation StokesI+Q. For the spectral work, it is necessary to correct the AMI val-ues to make them consistent with those from the other telescopes.4 . −
140 rad m − implies a rotationof 24 degrees between 5 and 16 GHz. We assume the same degreeof polarisation at 16 GHz as at 4 . / ( φ ) , where φ is the angle between the electric fieldvector and the orientation of the I+Q AMI feeds, which is verticalon the sky. We take into account the variation of φ along the relic.The unpolarised RN flux is added to the corrected polarised flux: I = ( I + Q )+ ( I + Q ) / ( φ ) . The I integrated flux den-sity is obtained by decreasing 24% from the I+Q value. Figure 1 shows separate CLEANed maps for the LA and SA data,using ‘Briggs’ weighting (Robust set to 2.0 to enhance diffuseemission, Briggs 1995). The SA map resolution is 3 . × . ×
22 arcsec. The RMS noisein the SA map is ∼ . − near the northern radio relic,while in the LA map it is 35 µ Jy beam − . To produce directly-comparable, multi-frequency radio images, anumber of steps were taken before combining the maps for thestudy of the integrated spectrum. Due to the very low SA resolu-tion compared to the WSRT and GMRT maps, we chose to com-bine only the LA map with the other datasets. We imaged the datausing the CLEAN algorithm with the same pixel size (1 arcsec perpixel), image size and uniform weighting. The uv-coverage of theLA samples densely down to a uv-distance of 0 . λ . Therefore,only GMRT and WSRT data beyond a uv distance of 0 . λ wereused, so that our radio maps image approximately the same spatialscales on the sky. We simulated an LA observation of a uniformbrightness distribution with the angular dimensions of the northernrelic as measured by the LA, with the uv coverage and pointingsused for the real LA observation. We found that the LA could beresolving out a negligible part of the largest scale diffuse emission.The uv-cut is necessary for extended sources, as inconsistent inner-uv coverages can lead to non-comparable integrated fluxes. All ofthe maps were primary beam corrected and convolved to the beamof the AMI LA map.We adopt an absolute flux-scale uncertainty of 10 per cent forthe GMRT and WSRT data, following Stroe et al. (2013). This un-certainty results from telescope pointing errors and imperfect cali-bration. AMI flux scale errors are well-described by 5 per cent ofthe flux (AMI Consortium: Davies et al. 2011). The left panel of Figure 1 shows a 1 . . λ , all of the diffuse emission visibleat lower frequencies is recovered in the AMI SA map.The northern relic (RN) displays an arc shape, but the emis-sion is mixed with radio galaxies H and B and diffuse source I (seealso right panel, Figure 1). c (cid:13) , 000–000 he highest-frequency detection of a radio relic Figure 1. ‘Briggs’-weighted AMI 16 GHz images (robust=2). Contours drawn at [ , , , ] × σ RMS . Left : AMI SA. The beam size 3 . × . ∼ . − . The grey intensity shows a low-resolution (3 arcmin) WSRT 1 . Right : AMI LA in intensity and contours, at 44 arcsec ×
22 arcsec resolution, with σ RMS ≈ µ Jy beam − . Source labelling from Stroe et al. (2013) is shown. The complex of diffuse emission towards the south of the clus-ter arises as a blending of sources RS, J, A and tailed-radio sourceF. At lower frequencies, radio phoenix J is much brighter than therelic RS (Stroe et al. 2013). Since J has a much steeper spectrumthan RS, they contribute comparably to the flux at 16 GHz. Puz-zling is the ∼ σ upper limiton the WSRT flux (giving ∼
15 mJy), we would expect the spec-tral index of this extension to be flatter than − .
5. While the peakin the WSRT emission is towards the west, at the location of com-pact radio galaxy A, and it progressively wanes towards the east,the AMI SA emission shows the opposite trend. The peak at theemission is located where no counterpart is seen in the WSRT map.There could be some point-source contamination, but this shouldbe minimal as the LA finds no significant sources in the area.
The right panel of Figure 1 shows the AMI LA map (imaged withrobust=2). The higher resolution enables a better deblending ofsources, but the poorer inner uv-coverage leads to loss of flux onlarge scales. This is evident as most diffuse sources (RS, J) detectedin the SA effectively disappear in the LA map.The northern relic is detected at the 11 σ level at peak fluxand clearly separated from its neighbouring source H towards thewest. Only the central, brightest part of source R1 is visible. Wealso detect sources labelled A, B, C, D and E as point sources withhigh S/N ( > σ ). The nucleus of tailed-radio galaxy F is detectedat 10 σ , but its steep spectrum tail is not recovered, as expected(see Stroe et al. 2013). The ‘extension’ is not detected in the high-resolution 16 GHz, suggesting it may have a diffuse nature. Figure 2 and Table 1 present the spectrum of RN. The flux densi-ties are measured in fixed boxes in uniform-weighted maps. Notethat because of the uniform weighting, RN is detected at 6 σ levelsignificance at peak. We use a least-squares method to fit a singlepower law to the integrated flux-density of the relic from each of the eight radio maps, at common resolution and with the commonuv-cut. This fitting takes into account a total flux error computed asthe quadrature of the flux scale error of 10 per cent for the GMRTand WSRT measurements and 5 per cent in the AMI LA, and theRMS noise in each map multiplied by the square root of the numberof beams contained in the box we measure the flux in.From spatially-resolved, low-frequency observations of RN,we found a ∼ − . . − .
06 (Stroe et al. 2013). Figure 2 shows in the dotted line theinjection spectrum of the freshly-accelerated electrons, while thedashed line presents the integrated spectrum, as derived from thelow-frequency data. A single power law fit ( α int = − . ± . χ of 163 (solid line in Figure 2). The 16 GHz measurement lies 12 σ below the extrapolation of the low-frequency spectrum. Radio relics are thought to form at the wakes of travelling shockfronts produced by the major merger of galaxy clusters (Ferettiet al. 2012). The physical processes underlying their formation,such as the injection and ageing mechanism, can be constrained us-ing high-frequency measurements, which have not been performeduntil now. Here, we present the 16 GHz measurement of a relicthrough AMI observations of the ‘Sausage’ cluster.
The 16 GHz measurement of the northern relic and its integratedspectrum are given in Fig. 2 and Table 1. We find strong evidencefor high-frequency steepening in the integrated spectrum of RN.There are two reasons why this should be considered a robust mea-surement. Firstly, the integrated spectra of point sources in theGMRT, WSRT and AMI LA maps are well described by singlepower laws, implying a correct overall flux scale also for the 16GHz measurements. Secondly, the dense AMI LA uv-coverage atthe shortest spacings indicates minimal loss of flux at large spa-cial scales. All of the lower-frequency measurements (GMRT and c (cid:13) , 000–000 A. Stroe et al.
Table 1.
Integrated radio spectrum of the RN measured in the uniform-weighted radio maps with common uv-cut and resolution. The uncertaintiesof the measurements are computed as the quadrature of the flux error andthe rms noise in each map, multiplied by the square root of the number ofbeams spanned by the source. Note that RN is detected at a total S/N of 24in the integrated spectrum. Taking the lower bound given by the error in theintegrated flux results in a 18 σ detection.Freq. [GHz] 0 .
15 0 .
32 0 . . . . . . . WSRT, < . α ∼ − .
6; van Weeren et al.2010; Stroe et al. 2013). Energy losses due to synchrotron and in-verse Compton processes lead to spectral index steepening and in-creasing spectral curvature in the downstream area (van Weerenet al. 2010; Stroe et al. 2013).Ensslin et al. (1998) modelled the integrated radio spectrumfor such a relic formation scenario. At the shock front the particlesare accelerated to a power-law radio spectrum, followed by lossesthat steepen the spectra. The integrated spectrum results from thesummation of particle spectra spanning a range of ages from differ-ent regions in the downstream area. This is equivalent to the contin-uous injection model which was proposed to explain the integratedspectra of radio galaxies, where the jet deposits freshly acceleratedelectron in the radio lobes at a constant rate (CI; Pacholczyk 1970).In the CI model, the integrated spectrum has a critical fre-quency ν crit , beyond which the spectrum steepens by 0 . ∼
100 MHz), where we observe theaged spectrum. Therefore, simple plane-shock theory in the con-text of DSA predicts that the relic integrated spectral index of asource should be 0 . − . − .
6, which defines a shockfront Mach number of 4 . ± . ∼ − . . − . ± .
04 is consistent with the prediction from theCI model (Stroe et al. 2013).However, by extrapolating the RN low-frequency spectrum,we find that the 16 GHz measurement is in stringent tension withthe CI prediction, at the 12 σ significance level (see Fig. 2). Theintegrated 153 MHz to 16 GHz index is much steeper ( ∼ .
8) thanthe injection index, while if only the high-frequency data is consid-ered, this difference increases to 1 . • As mentioned previously, the Ensslin et al. (1998) model onlyholds for frequencies above the break frequency, where there is abalance between continuously, freshly injected plasma and ageing.In reality, there is a broad frequency range over which the steep-ening takes place. If the break occurs over a range of frequenciesbelow ∼
100 MHz, then the steepening would gradually increasetowards higher frequencies, giving a curved integrated spectrum, asthe one we observe in the northern relic. In cases where the spectral
Figure 2.
Integrated radio spectrum of the northern relic from 153 MHz upto 16 GHz (see also Table 1). The red cross marks the 16 GHz measurement.The uncertainties include a 10 per cent flux scale error added in quadratureto the σ RMS . A power-law is fitted to the eight frequencies. The dotted lineshows the injection spectrum and the dashed line the integrated spectrum,as derived from high-resolution GMRT and WSRT data (Stroe et al. 2013).The injection and integrated spectra below 2 . σ below the CI prediction. break occurs across the observed frequency range, the results willbe biased to flatter integrated spectra and hence stronger derivedMach number. • The injection spectrum is not a power law. With a pool of ther-mal or pre-accelerated electrons, the injection spectrum is still ex-pected to be a power law (Brunetti & Jones 2014). The accelerationefficiency for electrons beyond γ ≈ × Lorentz factors (equiva-lent to a few GHz-peak emission frequency for µ G magnetic fields)might be smaller than for the lower energy electrons. During theirmultiple crossings of the shock front, the electrons lose energy andradiatively cool during the acceleration, leading to a curved injec-tion spectrum, assuming that the electron mean free path is largerthan its gyro-radius (Keshet et al. 2003). • A gradient of density and/or temperature across the sourcewould result in different spectral components, resulting in a 16-GHz spectrum completely dominated by losses/aged electrons. As-suming an isothermal sphere ICM gas distribution (Sarazin 2002),we calculate a drop of 15 −
25 per cent in electron gas densityacross the 50 − • The magnetic field at the shock location might be strongerthan in the downstream area, as a result of shock compression. Ac-celeration in the presence of ordered, strong magnetic fields at theshock front, combined with turbulent, lower magnetic fields in thedownstream area, could lead to a curved integrated spectrum. Sim-ulations of supernova remnant synchrotron emission under turbu-lent magnetic field conditions suggest that electrons in the cut-offregime can radiate efficiently (Bykov et al. 2008).Nevertheless, higher-resolution data is required for distinguishingbetween these scenarios. At the moment, no relic formation mech-anism can readily explain the high-frequency steepening, thus newtheoretical models have to be developed (Brunetti & Jones 2014). c (cid:13) , 000–000 he highest-frequency detection of a radio relic Towards the south of the cluster, we discover an extension at the8 σ significance level towards the east of RS in the low-resolutionSA AMI map. This source does not have a counterpart in the lowerfrequency data, or in the high-resolution AMI LA map (Figure 1),excluding the possibility of a point source. The extension appearselongated ( ∼ ∼ − .
5. Itsarc-like shape and proximity to RS make the extension an idealcandidate for a relic, but its flat integrated spectral index meansthat the source cannot result from a shock front in the context ofDSA. Striking also is the difference between the spectral index ofthe extension and RS, which points to very different shock prop-erties towards the south and towards the south-west. This could beexplained by different ICM temperature/densities in the two direc-tions. Ogrean et al. (2013) measured a sharp increase in ICM tem-perature in the direction of this extension, followed by a putativeshock with a Mach number of 1 .
2, coincident with the location ofthe radio extension. Such an increase in temperature in the down-stream area of travelling shock fronts has been also found in sim-ulations (Roettiger et al. 1997). The source seems to trace an arc-like shock front, which suggests a shock seen in projection ontothe plane of the sky, which means the radio emission detected is amixture of different age-populations of electrons.
High radio-frequency observations of steep-spectrum, diffuse, clus-ter emission have not previously been made owing to a lack ofsuitable instrumentation. We have observed the ‘Sausage’ mergingcluster at 16 GHz at low (3 arcmin) and high (40 arcsec) resolutionwith the AMI array and we successfully detect diffuse radio relicemission for the first time at frequencies beyond 5 GHz. Our mainresults are: • The northern relic measures an integrated flux density of1 . ± . σ peak detection in a uniformly-weighted map).We investigate in detail its integrated spectrum and conclude thereare clear signs of spectral steepening at high frequencies. If ther-mal electrons are accelerated, the steepening can be caused bya lower acceleration efficiency for the high-energy ( γ > × )electrons, a negative ICM density/temperature gradient across thesource or turbulent downstream magnetic fields amplifying theemission of electrons in the cut-off regime. However, these scenar-ios are unlikely because of low-acceleration efficiencies at weak-Mach-number shocks. Further theoretical modelling is required. • We also detect a peculiar, flat-spectrum ( α int ≈ − .
5) patch ofdiffuse emission towards the south-east of the cluster, which cannotbe explained by the CI model.The surprising high-frequency spectral steepening results and flat-spectra presented here suggest that the simple CI model, whichhas been widely used in the literature to explain the formation ofradio relics, needs to be revisited. Furthermore, there is a clearneed for high-quality radio observations of relics at cm and mm-wavelengths that resolve radio relics.
ACKNOWLEDGMENTS
We thank the referee for the comments which greatly improvedthe clarity and interpretation of the results. We also thank Gi-anfranco Brunetti, Tom Jones, Martin Hardcastle, Andrei Bykov, Matthias Hoeft, Wendy Williams and Marja Seidel for useful dis-cussions. We thank the staff of the Mullard Radio Astronomy Ob-servatory for their invaluable assistance in the operation of AMI,which is supported by Cambridge University. This research hasmade use of the NASA/IPAC Extragalactic Database (NED) whichis operated by the Jet Propulsion Laboratory, California Instituteof Technology, under contract with the National Aeronautics andSpace Administration. This research has made use of NASA’s As-trophysics Data System. AS acknowledges financial support fromNWO. CR acknowledges the support of STFC studentships. JJHthanks the University of Hertfordshire and the STFC for their fund-ing. RJvW is supported by NASA through the Einstein Postdoctoralgrant number PF2-130104 awarded by the Chandra X-ray Center,which is operated by the Smithsonian Astrophysical Observatoryfor NASA under contract NAS8-03060. DS is supported by a VENIfellowship. YCP acknowledges the support of a Rutherford Foun-dation/CCT/Cavendish Laboratory studentship.
REFERENCES
Davies, M. L., Franzen, T. M. O., Davies, R. D., et al. 2009, MN-RAS, 400, 984AMI Consortium, Davies, M. L., Franzen, T. M. O., et al. 2011,MNRAS, 415, 2708AMI Consortium, Zwart, J. T. L., Barker, R. W., Biddulph, P., etal. 2008, MNRAS, 391, 1545Bonafede, A., Feretti, L., Giovannini, G., et al. 2009, A&A, 503,707Bonafede, A., Feretti, L., Murgia, M., et al. 2010, A&A, 513, A30Briggs, D. S. 1995, PhD thesis, New Mexico Tech, USABrunetti, G., & Jones, T. W. 2014, arXiv:1401.7519Bykov, A. M., Uvarov, Y. A., & Ellison, D. C. 2008, ApJL, 689,L133Dressler, A. 1980, ApJ, 236, 351Drury, L. O. 1983, Reports on Progress in Physics, 46, 973Ensslin, T. A., Biermann, P. L., Klein, U., & Kohle, S. 1998, A&A,332, 395Feretti, L., Fusco-Femiano, R., Giovannini, G., & Govoni, F.2001, A&A, 373, 106Feretti, L., Giovannini, G., Govoni, F., & Murgia, M. 2012, A&Ar,20, 54Giacintucci, S., Venturi, T., Macario, G., et al. 2008, A&A, 486,347Keshet, U., Waxman, E., Loeb, A., Springel, V., & Hernquist, L.2003, ApJ, 585, 128Kocevski, D. D., Ebeling, H., Mullis, C. R., & Tully, R. B. 2007,ApJ, 662, 224Noordam, J. E. 2004, in SPIE Conference Series, Vol. 5489, ed.J. M. Oschmann, Jr., 817–825sOgrean, G. A., Br¨uggen, M., R¨ottgering, H., et al. 2013, MNRAS,429, 2617Pacholczyk, A. G, 1970, Radio astrophysicsPerley, R. A., & Butler, B. J. 2013, ApJS, 204, 19Roettiger, K., Loken, C., Burns, J. O., 1997, ApJS, 109, 307Roettiger, K., Burns, J. O., & Stone, J. M. 1999, ApJ, 518, 603Sarazin, C. L. 2002, in Astrophysics and Space Science Library,Vol. 272, 1–38Stroe, A., van Weeren, R. J., Intema, H. T., et al. 2013, A&A, 555,A110van Weeren, R. J., R¨ottgering, H. J. A., Br¨uggen, M., & Hoeft, M.2010, Science, 330, 347 c (cid:13)000