Inverse-Compton emission from the lobes of 3C 353
J.L. Goodger, M.J. Hardcastle, J.H. Croston, N.E. Kassim, R.A. Perley
aa r X i v : . [ a s t r o - ph ] J a n Mon. Not. R. Astron. Soc. , 1–12 (2007) Printed 6 December 2018 (MN L A TEX style file v2.2)
Inverse-Compton emission from the lobes of 3C 353
J.L. Goodger, ⋆ M.J. Hardcastle, J.H. Croston, N.E. Kassim and R.A. Perley University of Hertfordshire, College Lane, Hatfield, Hertfordshire AL10 9AB, UK US Naval Research Lab 4555 Overlook Ave., SW Washington, DC 20375, USA National Radio Astronomy Observatory, P.O. Box O, Socorro, NM 87801, USA
ABSTRACT
X-ray emission due to inverse-Compton scattering of microwave background photons byelectrons in the lobes of powerful radio galaxies has now been seen in a large number ofobjects. Combining an inverse-Compton model for the lobe X-ray emission with informationobtained from radio synchrotron emission provides a method of constraining the electron pop-ulation and magnetic field energy density, which cannot be accomplished using the radio dataalone. Using six frequencies of new and archival radio data and new
XMM-Newton observa-tions of the Fanaroff & Riley class II radio galaxy 3C 353, we show that inverse-Comptonemission is detected in the radio lobes of this source at a level consistent with what is seenin other objects. We argue that variations in the X-ray/radio ratio in the brighter eastern loberequire positionally varying magnetic field strength. We also examine the X-ray nucleus andthe cluster, Zw 1819.1-0108, spatially and spectrally.
Key words: galaxies: active – X-rays: galaxies – galaxies: individual: 3C 353 – galaxies: jets – radiationmechanisms: non-thermal: X-ray: galaxies: clusters
Extended X-ray emission from the lobes of powerful, FR II(Fanaroff & Riley 1974) radio galaxies and quasars is thought tobe produced by inverse-Compton scattering of the cosmic mi-crowave background (CMB) (Feigelson et al. 1995; Tashiro et. al.1998; Tashiro et al. 2001; Isobe et al. 2002; Hardcastle et al. 2002)and of infrared photons from the core (Brunetti, Setti & Comastri1997). Although the synchrotron self-Compton emission from thelobes can also be modelled, it is negligible compared to the inverse-Compton emission from other populations except in the very small-est lobes (Hardcastle et al. 2002). Recent studies of the integratedX-ray properties of radio lobes have been carried out for large sam-ples (Croston et al. 2005; Kataoka & Stawarz 2005), showing thatthe CMB is the dominant photon population in most cases. Com-bining an inverse-Compton/CMB model for the lobe X-ray emis-sion with information obtained from radio synchrotron emissionprovides a method of constraining the electron population and mag-netic field energy density, which cannot be accomplished using ra-dio data alone. Unlike the jets, the lobe material is not movingrelativistically so there are no beaming effects to consider duringthis analysis (e.g. Mackay 1973). Understanding the contributionsof the electron densities of the synchrotron and inverse-Comptonelectron populations and the magnetic field allows us to estimatedirectly the total energy in the radio source, and thus the amount of ⋆ E-mail: [email protected] energy that can be transferred to the environment, without assum-ing equipartition, and also allows us to investigate the distributionof internal energy within the source.Hardcastle & Croston (2005) recently carried out a spatiallyresolved X-ray inverse-Compton study of the bright X-ray and ra-dio source Pictor A. They used the variation of the X-ray/radio ra-tio across the lobes to investigate the variation in electron densityand magnetic field strength throughout the source. They found thatvariations in either component alone could not explain the observedX-ray and radio properties of the lobe. Similar spatially resolved X-ray studies of other bright FR II radio galaxies are needed to followup these results. Here we report on observations of the radio galaxy3C 353.3C 353 is a FR II radio galaxy associated with the clusterZw 1718.1-0108. Although it is one of the brightest extragalacticsources in the sky at low frequencies, it is relatively poorly studied,presumably due to its low declination and low Galactic latitude.Swain (1996) observed 3C 353 at four radio frequencies with theNRAO Very Large Array (VLA) and published a study of the po-larisation variation across the jets, which favoured a spine-sheathmodel (Swain, Bridle & Baum 1998). Their VLA observations re-vealed filamentary structure within the lobes making 3C 353 an ex-cellent target for a spatially resolved study of electron distributionand magnetic fields. X-ray emission from the cluster associatedwith 3C 353 was first detected by Iwasawa et al. (2000) in
ASCA and
ROSAT images. The radio source resides in a giant ellipticalgalaxy on the edge of the cluster, which Iwasawa et al. found to bebright in the X-ray with a luminous point source coinciding with c (cid:13) J.L. Goodger et al. the radio galaxy’s core. Iwasawa et al. determined global clustertemperatures and used optical observations to identify the cluster’smembers. They confirmed a redshift of z = 0 . for 3C 353, aredshift of z = 0 . for three bright member galaxies, and usedthe velocity distribution to confirm 3C 353 as a member of the samesystem.In this paper we use new XMM-Newton and radio observationsto investigate the nature of the electron distribution and the mag-netic fields in the lobes and hotspots of 3C 353. We also determinethe cluster temperature, density and pressure profiles and exam-ine the cluster’s interaction with the lobes of 3C 353. Throughoutthe paper we use a cosmology in which H = 70 km s − Mpc − , Ω m = 0 . , and Ω Λ = 0 . . The angular scale is 1 arcsec = 0.61 kpc.We define the spectral index in the sense that S ν ∝ ν − α . The pho-ton index Γ = 1 + α . We used a combination of new and archival radio data and new
XMM-Newton
X-ray data to examine the broad-spectrum electronenergy distribution of 3C 353. Our radio data extends to lower fre-quencies than have been studied previously, whilst our X-ray datahas greatly improved sensitivity and resolution compared to the
ASCA data of Iwasawa et al. (2000).
We obtained VLA observations at 1.6 GHz and 4.8 GHz from theVLA archive as well as carrying out new observations at 327 MHzat the VLA and at 620 MHz at the NCRA Giant Meter-wave RadioTelescope (GMRT) in 2006. We also obtained images of 3C 353at 74 MHz and 8.4 GHz (see below). All the radio data were re-duced in
AIPS . The radio observations are summarized in Table 1and Fig. 1 shows an example of the image quality achieved.
The 74-MHz observations used are those discussed byKassim et al. (2007) and the data reduction is described inthat paper.
Our 327 MHz observations of 3C 353 were taken at the VLA onthe 27th May 2006 using the nearby Pie Town antenna of the VeryLarge Baseline Array (VLBA) to increase the long baseline of ourA configuration data set by a factor of two. We carefully monitoredthe Pie Town antenna during the calibration steps to ensure that thelong baselines survived.Initial bandpass calibration using 3C 286 resulted in smoothphase and amplitude variation across all channels. As a flux cali-brator, 3C286 is slightly resolved at this frequency. If the fractionalbandwidth is great enough that the intrinsic visibility of 3C 286varies across the pass band, a model is required (Lazio et al. 2005).At 327 MHz with the Pie Town antenna, we found that this effectis negligible so 3C286 was used as the flux calibrator without amodel, effectively treated as a point source. 3C 286 was used tocalibrate the phase across all baselines. The resultant calibrateddata set was flagged to remove noise then spectrally averaged using
SPLAT , averaging every 4 channels to reduce the number of chan-nels to 6, whilst producing a single source data set. The calibrated data included baselines up to 80 k λ , twice what would have beenachieved with only the A configuration.The B-configuration data set was bandpass calibrated using3C 286 to calibrate the channel gains, then flux using 3C 286 andphase calibrated with . After flagging and phase cross-calibrating with the A configuration data set, the data sets weremerged and a multi-facet deep cleaned map of 106 fields was pro-duced. The dynamic range of the final map is 580:1 with a resolu-tion of . × . arcsec. Our 620 MHz observations were taken at the GMRT on 19th July2006. The two observing frequencies (upper and lower sidebands)were calibrated separately. Preliminary bandpass calibration wasperformed using 3C 286 followed by flagging and one iteration of flgit to remove RFI from all sources. This flagged data set was thenbandpass- and flux-calibrated before being manually flagged andanother iteration of flgit applied. The data set was spectrally aver-aged into a multi-source data set, cutting the number of channelsfrom 128 to 6, whilst applying the bandpass calibration. This dataset was finally flux- and phase-calibrated before 3C 353 was split into a single-source file. The flux and phase calibrator used was1743-038. One iteration of phase self-calibration was performedbefore the data from the two observing frequencies were split intoindividual channels and recombined using
DBCON to a single dataset containing all of the averaged channels. This combined data setwas phase self-calibrated twice more and the final image has reso-lution . × . and dynamic range
590 : 1 . Observations from A-, B- and C-configurations were flux- andphase-calibrated separately using NRAO530 as the phase calibra-tor. Due to the allocation of the observing frequencies in the C-configuration data set, the data at the two frequencies were splitand recombined using
DBCON to match the frequency allocationsof the A- and B-configuration data sets. Once combined, the datawere phase self-calibrated and deep cleaned, with 100,000 itera-tions of
CLEAN applied. The final image has a resolution of 1.7 ×
500 : 1 . For maximum uv coverage, we use B-, C- and D-configuration datasets, each of which were phase- and flux-calibrated and flagged. Wecombine two epochs of D-configuration observations to increasethe signal to noise.The final combined data set was phase and amplitude self-calibrated, normalizing the gain. There was no significant reduc-tion in the maximum flux when compared to phase-only self-calibration, but the image quality was improved. The final imagehas a resolution . × . arcsec and a dynamic range of 1180:1. The X band images of 3C 353 were kindly provided by Alan Bridle.The deep cleaned images have a dynamic range of 2500:1 and aresolution of . × . arcsec. The details of the reduction aredescribed by Swain (1996). c (cid:13) , 1–12 nverse-Compton emission from the lobes of 3C 353 D E C L I NA T I ON ( J ) RIGHT ASCENSION (J2000)17 20 40 35 30 25 20-00 57 003058 003059 0030-01 00 00
Figure 1.
Table 1.
Radio Observation DetailsBand Frequency Telescope:Config. Date Duration Bandwidth Phase Calibrator Dynamic Range Res.(MHz) (s) (MHz) (arcsec)4 73.8 VLA 07/03/1998 7740 1.3 3C405 2090:1 . × . P 327.3 VLA:A+PT 27/05/2006 25660 3.125 3C286 580:1 . × . VLA:B 04/10/1998 3010 0.098 1416 + . × . L 1665/1385 VLA:A 19/05/1968 20710 12.5 NRAO530 500:1 . × . VLA:B 14/08/1968 6640 25.0 NRAO530VLA:C 16/09/1985 1441 50.0 1741-038C 4848/4898 VLA:B 14/08/1986 6799 50.0 NRAO530 1180:1 . × . VLA:C 22/08/1985 1679 50.0 1741-038VLA:D 15/11/1993 532 50.0 1730-130VLA:D 28/09/1984 3360 50.0 1725+044X 8440/8452 VLA:BCnD 13/03/1994 12.5 2500:1 . × . We observed 3C 353 on the 25th August 2006 and 17th February2007 with
XMM-Newton
EPIC MOS1, MOS2 and pn cameras. Theinitial observation (Set 1) from 25th August 2006 yielded 39 522 sand 39 525 s for MOS1 and MOS2 respectively and 34 042 s for thepn camera, whereas the second observation on the 17th February2007 (Set 2) was for 10 473 s and 10 487 s for MOS1 and MOS2and 5 653 s for pn. The pn camera was in Extended Full Framemode for both observations and the MOS cameras were in FullFrame mode.The data sets were initially processed using
SAS version 7.0.0.Standard EPIC MOS and PN pipelines and the standard filters
XMMEA EA and
XMMEA EP were applied. The MOS data setswere filtered to include single, double, triple and quadruple events(PATTERN
12) whereas the PN data set was filtered to includeonly single and double events (PATTERN c (cid:13) , 1–12 J.L. Goodger et al. D E C L I NA T I ON ( J ) RIGHT ASCENSION (J2000)17 20 40 35 30 25 20-00 56 3057 003058 003059 0030-01 00 0030 D E C L I NA T I ON ( J ) RIGHT ASCENSION (J2000)17 20 40 35 30 25 20-00 56 3057 003058 003059 0030-01 00 0030 D E C L I NA T I ON ( J ) RIGHT ASCENSION (J2000)17 20 40 35 30 25 20-00 56 3057 003058 003059 0030-01 00 0030 D E C L I NA T I ON ( J ) RIGHT ASCENSION (J2000)17 20 40 35 30 25 20-00 56 3057 003058 003059 0030-01 00 0030
Figure 2.
Radio contour images of 3C 353 at 327 MHz (top left), 620 MHz (top right), 1.67 GHz (bottom left) and 4.8 GHz (bottom right) all convolved to7.5 × σ × (1,2,4,...) mJy/beam where σ = The MOS and PN data sets were energy filtered to includeenergies between 0.5 and 5 keV before using the
SAS command evselect to generate images from the filtered events files. The task eexpmap was then used to generate exposure maps without vi-gnetting. We chose not to include vignetting correction in theseimages as it incorrectly weights up the particle background at theedge of the field, creating artefacts in the image. The images fromthe separate detectors were normalized to the Set 1 PN count rate(in order to remove the PN chip gaps) and then combined. The ex-posure maps were combined then divided by the total filtered live-time. The combined normalised image was finally divided by thescaled combined exposure map. The combined image was Gaus-sian smoothed with a kernel of 10.4 arcsec to highlight the clustersand radio lobes using the
CIAO command aconvolve .X-ray spectra were extracted using especget , a script that com-bines the evselect , arfgen and rmfgen commands in SAS , from re-gions defined to examine the emission from the northern and south-ern sub-cluster regions, and the east and west lobes. As the core andlobes are small regions near the pointing centre, so that vignettingis not important, we used local background subtraction. However,for the cluster, vignetting needs to be considered. The filtered eventfiles were weighted using evigweight before spectra were extractedusing a double subtraction method, (e.g. Arnaud et al. 2002). Back-ground template files for the field of view were made using filescreated by Read & Ponman (2003) and scaled to the same particlebackground level as our observations. A local background regionaway from the galaxy and cluster was defined using DS
9. A tem-plate background spectrum was subtracted from the source spec- trum events file to account for instrument and particle noise beforea local background region was used to subtract residual backgroundemission due to the differences in the Galactic/extragalactic back-ground levels of the source and scaled background datasets. Dueto the low signal to noise of Set 2, only Set 1 could be used in thedouble subtraction method applied to the cluster. The spectra werethen binned to 20 counts per channel after background subtraction,ignoring the first 20 channels for the MOS cameras and the first 50channels for the pn camera.
In this section we discuss the results of the X-ray spectral fittingand radio flux density measurements. X-ray fitting was carried outusing
XSPEC v 11.3 in the energy range 0.3 – 7.0 keV. Where ther-mal models were fitted we used a redshift of 0.03.
The column density towards 3C 353 is uncertain. Iwasawa et al.(2000) adopt a value of . × cm − based on the H I mea-surements of Dickey & Lockman (1990). However, they note thatthe visual extinction in the direction of 3C 353 would correspondto a column density of a factor ∼ higher. From the Galactic dustmeasurements of Schlegel et al. (1998) we estimate an A V of 1.4mag, which would correspond to a column density of ∼ . × c (cid:13) , 1–12 nverse-Compton emission from the lobes of 3C 353 D E C L I NA T I ON ( J ) RIGHT ASCENSION (J2000)17 21 00 20 45 30 15-00 52545658-01 000204060810 D E C L I NA T I ON ( J ) RIGHT ASCENSION (J2000)17 20 40 35 30 25 20-00 57 003058 003059 0030-01 00 0030
East Hotspot West Hotspot
Figure 3.
Radio contours of 3C 353 at 1.67 GHz with a 0.3-7.0 keV
XMM-Newton image (MOS 1, MOS 2 + pn) of the X-ray emission from the radio lobesand the cluster Zw 1718.1-0108 (left) Gaussian smoothed with a kernal of 10.4 arcsec; the same image zoomed in to show the inverse-Compton emission fromthe lobes (right), with the East radio/X-ray hotspot and West radio hotspot labelled. for standard Galactic gas/dust ratios. The true value is likely to liesomewhere between these extremes.We therefore used our new data to estimate a column den-sity directly. We extracted spectra for bright point sources around3C 353 using local background regions. In Set 1 there were 5 pointsources bright enough for spectroscopy. We fitted power-law mod-els with free absorption to these sources and found a weighted mean N H = (1 . ± . × cm − . As this is consistent with therange estimated above, we adopt the mean value in the analysis thatfollows.The X-ray point sources are not shown in Fig. 3 and were ex-cluded from all spectra extracted where the cluster was examined.The emission from the radio galaxy was also excluded. In the X-ray, the nuclear region was defined with a radius of 40 arc-sec to include as many nuclear photons as possible while excludingthe X-ray emission from the radio lobes. The background was es-timated from a local background region positioned above the radiogalaxy so as not to include any emission related to the radio galaxy.In the radio, the nuclear region was much smaller, adjusted in eachdata set to be a close fit to the nuclear emission. The peak intensitywas determined using jmfit for each frequency, the results of whichare shown in Table 2.We initially attempted to fit single power-law and thermalmodels to this spectrum; however, the fits were poor ( χ = 466 . for 259 d.o.f. and χ = 2242 . for 259 d.o.f. respectively). Wetherefore fitted a model which has been shown (Hardcastle et al.2006) to provide a good fit to the nuclear X-ray emission from Table 2.
Radio Core FluxesBand Frequency Flux Luminosity(MHz) (Jy) (W Hz − sr − )P 327.3/329.6 0.097 1.5 × - 620.5/633.3 0.089 1.4 × L 1665/1678 0.116 1.8 × C 4848/4898 0.140 2.2 × X 8440/8452 0.149 2.3 × narrow-line radio galaxies, consisting of a power-law component atGalactic absorption (see Section 3.1), a second power-law compo-nent with redshifted intrinsic absorption and a kT = 1 keV thermalcomponent. The best fitting model had χ = 268 for 237 d.o.f. Thecore spectrum is shown in Fig. 4. The photon indices of the unab-sorbed and absorbed power-law components were Γ = 1 . ± . and . ± . respectively, and the intrinsic column density was . ± . × cm − . The unabsorbed X-ray component has a1 keV luminosity density of . × W Hz − sr − , whereas theabsorbed X-ray component with the absorption removed and ex-trapolated to include energies from 2–10 keV has a luminosity of . × erg s − . We discuss the interpretation of the nuclearemission in Section 4.1. By considering the X-ray emission from regions encompassing theeast and west lobes but excluding the hotspots and the core, we ex-tracted spectra from the combined X-ray data set using local back- c (cid:13) , 1–12 J.L. Goodger et al.
Figure 4.
X-ray spectrum of the nucleus in the energy range 0.3–7.0 keV.The model plotted is the dual power-law + thermal model as described inSection 3.2. For clarity only the PN data are plotted. ground regions which we fitted with both thermal emission modelsand power laws. Assuming a circularly symmetric model for thecore, we determined that ∼ per cent of the counts in the Eastlobe spectra are from the core whereas the core counts account fora quarter of those in the West lobe spectra. Using the XMM-Newton calibration files, we determined the fraction of the core spectrun ex-pected to be scattered into the West lobe region to be 4 per cent ofthe total core emission, and so we included a fixed component inthe West lobe spectral fit consisting of the best-fitting core modelwith a normalisation fixed at 0.04 times that of the core. The detailsof the core spectrum are given in Section 3.2. Due to the low signalto noise of Set 2, the West lobe was undetected in this data set andso our spectra were extracted from Set 1 only.The background emission from the cluster environment is theprimary contributor to the uncertainty in the lobe flux densities.To best take account of this variation we used a truncated annularbackground region for the East lobe where the lobe appears to inter-act with the cluster. We investigated the effect of using local back-ground regions above and below the galaxy; the best fitting power-law models gave consistent photon indices but led to an increaseof a factor 2 in the flux density of the East lobe. These backgroundregions only take account of the emission from one of the two sub-clusters leading us to believe that a truncated annular backgroundregion is a more accurate measure of the cluster’s contribution. Forthe West lobe, a local background region located to the south of thegalaxy, at a similar distance from the brightest cluster emission tookaccount of the emission from the southern sub-cluster sufficientlyto allow a good fit to the spectra.The power-law model gave better fits than the thermal modelfor both lobes with χ = 39 . for 34 d.o.f for the East lobe and χ = 23 . for 24 d.o.f for the west lobe. The photon indices were Γ = 1 . ± . and Γ = 1 . ± . respectively, corespondingto α = 0 . ± . and α = 0 . ± . . Figure 5 shows the best-fitting power-law model for the East and West lobe X-ray data. Thelow frequency X-ray spectral index, measured between 327 MHzand 1.67 GHz, is ∼ . for both lobes, so that the X-ray spectralindices of both lobes are consistent with a CMB inverse-Comptonscattering model. We discuss the nature of the lobes in Section 4.2. Both the East and West hotspots are clearly detected in all radiofrequencies. The West hotspot is undetected in the X-ray; how-ever, the East hotspot has an X-ray counterpart . ± . arcsec( . ± . kpc) from the centre of the radio hotspot. This has toolow signal to noise for spectral analysis to be performed, with only60 counts between 0.3 and 7.0 keV. In recent Chandra observations(Kataoka et al., private communication), the same feature is de-tected but the offset was measured as . ± . arcsec. A jet knotwas also detected in the vicinity of the East X-ray hotspot suggest-ing that the difference in our offset measurements could be due tocontamination from this knot.The inverse-Compton analysis performed on the lobes can-not be applied here as the electron populations responsible for anyinverse-Compton X-ray and radio synchrotron emission reside indifferent regions. X-ray hotspots in radio galaxies of similar radioluminosity to 3C 353 are generally thought to be produced by syn-chrotron emission, and so this is likely to be the dominant emissionprocess for this X-ray hotspot as well. The offsets measured be-tween the X-ray and radio hotspots are consistent with what is seenin some other sources (Hardcastle et al. 2002, 2007; Erlund et al.2007). Unfortunately, the southernmost section of the southern sub-clusterextends beyond the field of view of the
XMM-Newton cameras;however, a comparison to the
ASCA image of Iwasawa et al. (2000)shows that the peak of the X-ray surface brightness is included inour data.We extracted spectra from the northern and southern sub-clusters using a double subtraction method with annular local back-ground regions. The northern and southern sub-cluster regions areshown in Fig. 7. Thermal models fitted using mekal in X
SPEC give acceptable fits ( χ = 258 for 175 d.o.f and χ = 544 for 490 d.o.f for the northern and southern sub-clusters respec-tively) with kT = 3 . ± . keV for the northern sub-cluster and kT = 4 . ± . keV for the southern sub-cluster both fitted with anabsorption of N H = 1 . × cm − . The best-fitting models areshown in Fig. 6. Iwasawa et al. (2000) determined the temperatureof the combined sub-clusters to be kT = 4 . ± . keV using aGalactic absorption of N H = 1 . × cm − whereas when theupper limit ( . × cm − ) was used, the temperature dropped to . ± . keV which is consistent with the mean temperature ofthe cluster. The difference in temperature of ∼ keV supports theidea that the two sub-clusters are originally separate componentsundergoing a merger. Taking a north-south slice through the clusterusing rectangular regions of × arcsec from 17:20:50.074,-00:53:13.77 (Fig. 8), there is no evidence for a shock and no in-crease in surface brightness, indicating a non-violent interaction be-tween these components.A thermal model was fitted to annular regions in the northernand southern sub-clusters. The results are shown in Table 3. Theresidual local background was accounted for using regions in a rel-atively source free area of the field of view. Using an annular back-ground region gave consistant temperatures for both sub-clusters.The northern sub-cluster shows a linear increase in temperaturewith radius with a best fit of T ( r ) = 0 . r + 1 . keV whilethe southern sub-cluster is isothermal within the errors. 3C 353’sEast lobe lies at a radius of 250 – 350 arcsec from the centre ofthe northern sub-cluster but it may also be affected by the south- c (cid:13) , 1–12 nverse-Compton emission from the lobes of 3C 353 Figure 5.
X-ray spectra for the East (left) and West (right) lobes in the energy range 0.3–7.0 keV. The East lobe shows the MOS 1, MOS 2 and PN cameraswith the power-law model while the West lobe shows only the PN data for clarity with the dual power-law + thermal contribution from the core as well as thebest-fitting power-law model from the lobe emission as described in Section 3.3.
Figure 6.
X-ray spectra for the northern (left) and southern (right) sub-clusters in the energy range 0.3–7.0 keV. The MOS1, MOS2 and PN data sets are shownin each case with the best-fitting thermal model as described in Section 3.5.
Table 3.
Temperature profiles for the northern and southern sub-clustersNorthern SouthernRadius kT χ /d.o.f. Radius kT χ /d.o.f.(arcsec) (keV) (arcsec) (keV)0 – 50 . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . ern sub-cluster. The annular temperature at this radius ( kT =3 . ± . keV) is consistant with the global northern sub-clustertemperature ( kT = 3 . ± . keV) as well as the temperature mea-sured by the slice at this radius ( kT = 3 . ± . keV). We thereforeadopt a temperature of . ± . keV for the environment of 3C 353.As 3C 353 interacts mainly with the northern sub-cluster andas the southern sub-cluster is not fully covered within the XMM-Newton field of view, surface brightness and pressure profiles wereconstructed for the northern sub-cluster only. To determine the cen- tre of the northern sub-cluster we used a centroiding routine foreach of the
XMM-Newton cameras. Annular regions were used toextract a surface brightness profile, excluding the chip gaps, X-raypoint sources, emission from 3C 353 and from the southern sub-cluster and the position of the missing chip in the MOS 1 cam-era. An annular region was used to account for the residual localbackground emission. The surface brightness profiles for each cam-era were fitted with a model consisting of a β model convolvedwith the XMM-Newton
PSF based on the on-axis parametrisationdescribed in the
XMM-Newton
CCF files XRT1 XPSF 0006.CCF,XRT2 XPSF 0007.CCF and XRT3 XPSF 0007.CCF before beingfitted with a β model (Fig. 9).We examined the annuli in quadrants to highlight any den-sity variation in a north-south or east-west direction. The north-ern quadrants and the south-east quadrant were consistent. We ex-cluded the south-west quadrant from the final profile to ensure thatX-ray emission from the radio galaxy, and any local disturbances inthe cluster gas, were removed. The resulting surface brightness pro-files were consistent for all cameras. The joint best-fitting model,for all cameras, had a radius of 194 arcsec and β = 0 . . Consid-ering a σ confidence interval for 2 interesting parameters, β was c (cid:13) , 1–12 J.L. Goodger et al.
Figure 7.
Northern and southern sub-cluster regions from which spec-tra were extracted to determine the sub-cluster global temperatures. Thecrossed regions are those which were excluded from the analysis and in-clude the X-ray point sources and 3C 353. The crosses indicate the limits ofthe slice used. The inner edge of a large annular region masking the edge ofthe camera can be seen at the bottom of the image. unconstrained (though we consider values below 0.35 and above0.9 unrealistic), whereas the core radius of the northern sub-clusterwas limited to between 184 and 412 arcsec. We discuss the resultsof this section in Section 4.3.
The double-peaked nuclear X-ray spectrum, consisting of a heav-ily absorbed component together with one with only Galacticabsorption, is typical of what is observed in narrow-line radiogalaxies (NLRG) (e.g. Sambruna, Eracleous & Mushotzky 1999;Grandi, Malaguti & Fiocchi 2006; Belsole, Worrall & Hardcastle2006; Hardcastle, Evans & Croston 2006). However, when we plotthe unabsorbed luminosity of the heavily absorbed nuclear com-ponent in 3C 353 against its 178-MHz total radio luminosity, asdone by Hardcastle et al. (2006) (their figure 3) we find that 3C 353lies almost 1 order of magnitude below the correlation they de-termined for NLRGs: that is, 3C 353 is overluminous in the radiofor its X-ray luminosity or underluminous in the X-ray for its ra-dio luminosity, and lies closer to the region of the plot occupiedby low-excitation radio galaxies (LERGs). By contrast, the unab- T e m p e r a t u r e ( k e V ) Offset (arcsec) 0 1 2 3 4 5 6 0 200 400 600 800 1000 1200 T e m p e r a t u r e ( k e V ) Offset (arcsec)
Figure 8.
Temperature variation of north-south slice through clusterwith rectangular regions of × arcsec from a zero of offset at17:20:50.074, -00:53:13.77. Figure 9.
Surface brightness profile for the northern sub-cluster showingthe PN data set with the best-fitting β model which has β = 0 . and coreradius 194 arcsec. sorbed component is consistent with the correlation found by Hard-castle et al. between unabsorbed X-ray luminosity and nuclear 5-GHz radio emission (their figure 3).The optical emission-line classification of 3C 353 is uncer-tain. Laing et al. (1994) use the ratio of the flux in the [O III ]line to that in the H α to distinguish between the NLRGs andLERGs. The best published optical spectrum of 3C 353 is that ofSimpson et al. (1996), and their quoted line fluxes clearly place3C 353 below the [O III ]/H α cutoff of 0.2 proposed by Laing etal: thus by this definition 3C 353 would be a LERG. However, ifthe line fluxes are corrected for reddening using values determinedfrom the Schlegel, Finkbeiner & Davis (1998) dust maps (see Sec-tion 3.1), then the emission-line ratio rises to . ± . , wherethe errors are derived from the normal 10 per cent errors of Simp- c (cid:13) , 1–12 nverse-Compton emission from the lobes of 3C 353 son et al. and are almost certainly too low: within the errors wecannot say whether 3C 353 should be classed as a LERG or NLRG.The nature of 3C 353’s nucleus both in the optical and in the X-rayremains ambiguous. Using the method of Croston et al. (2005), the X-ray observationswere compared to the predictions of an inverse-Compton model,based on synchrotron modelling of the radio data sets using
SYNCH (Hardcastle et al. 1998). The broad-band spectra for the lobes areshown in Fig. 10. The measured 1-keV flux densities of the Eastand West lobes are . ± . and . ± . nJy. The equipartitionmagnetic field strengths for the East and West lobes are 0.89 nTand 0.84 nT respectively, while if we assume that all the X-rayemission is inverse-Compton in origin then the measured flux den-sities correspond to magnetic field strengths of . ± . nTand . ± . nT. Therefore, while the weakly detected Westlobe has a measured magnetic field strength within errors of thepredicted equipartition value, the East lobe’s measured magneticfield strength exceeds the predicted equipartition value by a factor ∼ so that B obs /B eq ≃ . . The West lobe’s measured mag-netic field strength is also consistent with a substantial departurefrom equipartition. This implies that the lobes of 3C 353 are elec-tron dominated, which is not unusual considering the range of fieldstrengths in the sample of Croston et al. (2005). This factor wouldbe reduced for 3C 353 if the lobes were not in the plane of the sky.The measured magnetic field strengths in the lobe correspondto internal pressures of . ± . × − Pa and . ± . × − Pa for the East and West lobes respectively. (If the alternativelocal background regions discussed in Section 3.3 were used, themagnetic field strengths would be reduced by ∼ . nT and theinternal pressure of the East lobe would double, but this does notaffect our conclusions.)Using the radio data sets, we constructed a spectral index mapbetween the 1.67-GHz and 327-MHz images, Fig 11. It revealed avariation of ∆ α ∼ . across both lobes. Within the radio lumi-nous region of the East lobe, excluding the hotspots, the spectralindex is roughly constant, α = 0 . ± . despite the filamentarystructure seen in Fig. 1, whereas in the West lobe, α = 0 . ± . excluding the hotspot. We note that the radio lobes do not appearto be entirely separate, and thus consider the region in between theradio luminous lobes, north and south of the core to be an inter-lobe region, which cannot be unambiguously associated with ei-ther lobe. This inter-lobe region has a relatively steep spectrum of α = 0 . ± . whereas the hotspot in both lobes exhibits a flatterspectrum of α = 0 . ± . . Fig. 11 shows the spectral index mapbetween 1.67 GHz and 327 MHz with contours from the 327-MHzmap.The X-ray/radio ratio was determined for the hotspot, lobeand inter-lobe regions, using the high signal to noise of the X-raydata set. The 327 MHz radio map of resolution . × . arcsecwas used with the convolved, exposure corrected X-ray map ofenergies 0.3 to 7.0 keV. The X-ray/radio ratio was found to be afactor of 4 greater in the steep inter-lobe region than in the flat,hotspot regions. If the magnetic field strength and the number den-sities of both the inverse-Compton and synchrotron emitting elec-tron are constant across the lobe, the X-ray/radio ratio would alsobe constant, but this is not what we see in 3C 353. Similar resultshave been seen in other radio galaxies, notably 3C 452 (Isobe et al.2002) and Pictor A (Hardcastle & Croston 2005). Hardcastle & Croston consider the following three models to explain the vary-ing X-ray/radio ratio:(i) Some other emission process could boost the X-ray emissionin the inner regions;(ii) The central regions contain more low-energy electrons rela-tive to the outer regions;(iii) The magnetic field strength varies as a function of position.A contribution from the core was included in the X-ray spec-tral analysis which also takes account of the contribution of the un-absorbed X-ray emission associated with the jet and the absorbedX-ray emission from the accretion disk. Emission from the corecannot therefore boost the X-ray emission in the inner regions.Our spectral fits show no evidence for a contamination of the lobespectra by galaxy-scale thermal emission. Neither could a boost bedue to the galaxy’s proximity to the cluster as local backgroundregions were used to account for any thermal emission from thesub-clusters. For inverse-Compton scattering of nuclear photons tobe the dominant process, the core bolometric luminosity needs tobe at least W. This is not unrealistic, however modelling thelobe surface brightness for this scenario using the results of Brunetti(2000) as described by Hardcastle et al. (2002) and with 3C 353 inthe plane of the sky, reveals a gradient across the lobe which is notobserved in either the X-ray image nor the X-ray/radio ratio. Usingthe spectral energy distributions of Haas et al. (2004), we deduceda typical IR spectrum for 3C 353 by scaling a dual power-law fitto the spectral energy distribution of 3C 33. 3C 33 was chosen as ithas a similar luminosity and is classified as a NLRG/FR II. The IRspectrum was normalised by the ratio of low-frequency radio lumi-nosities of the two sources. The predicted flux density for the nu-clear inverse-Compton emission is . × − Jy at 1 keV, which isa factor of ∼ fainter than the predicted CMB inverse-Comptonemission and the observed flux density ( . and . nJy respec-tively). Thus we can rule out model (i).If we consider the magnetic field to be constant at the mea-sured lobe averaged magnetic field strength of 0.39 nT and applyan inverse-Compton model to the lobes, we find the emission at327 MHz traces electrons with γ ≃ whilst the measuredinverse-Compton emission traces electrons at γ ≃ . As thecritical frequency for synchrotron emission goes as γ , we find thata variation in the X-ray/radio ratio of a factor of 4 requires a varia-tion in the spectral index between 10 MHz and 327 MHz of ∼ . .Even if the equipartition magnetic field strength of 0.89 nT is as-sumed so that the 327 MHz emission traces electrons of γ ≃ ,the observed spectral index variation requires a factor ∼ . vari-ation in the X-ray/radio ratio which is still much lower than ob-served. This is also the case with Pictor A (Hardcastle & Croston2005); a variation in the low-energy electron densities alone (modelii) cannot explain the variation of X-ray/radio ratio across the lobe.Alternatively, if we assume constant electron densities for boththe synchrotron and inverse-Compton emission electrons and alsothat the magnetic field strength does not vary along the line ofsight, we find that the observed variation in the X-ray/radio ratiothen requires a variation in the magnetic field of at most a factor of . . For a given frequency, this means the spectral index observedat low frequencies, between 1.67 GHz and 327 MHz, for regionsof high X-ray/radio ratio should correspond to the spectral indexbetween 3.5 GHz and 825 MHz for low X-ray/radio ratio regions.We consider the spectral indices between 4.8 GHz and 1.67 GHz tolimit the spectral indices of the low X-ray/radio ratio regions andfind that they exceed the upper limit predicted by the observed X-ray/radio ratio. From this comparison, we cannot rule out the pos- c (cid:13) , 1–12 J.L. Goodger et al.
Figure 10.
Broad-band spectrum for the East (left) and West (right) lobes of 3C 353 . A synchrotron emission model (solid line) is fitted to the radio data(asterisks). The measured X-ray flux is represented by the dot with the bowtie indicating the error. The dotted line shows the inverse-Compton X-ray fluxprediction and the dash-dotted line shows the predicted synchrotron self-Compton emission, both with the equipartition magnetic field strength. sibility that a varying magnetic field alone could be responsible forthe observed variations in the X-ray/radio ratio and the spectral in-dices. We therefore considered a colour-colour diagram using themethod of Katz-Stone, Rudnick & Anderson (1993). In their studyof Cygnus A, they argue that a single curve on a colour-colourdiagram is consistent with a homogeneous distribution of rela-tivistic electrons together with varying the magnetic field strengthacross the source. The colour-colour diagram for 3C 353, shown inFig. 12, has a similar single curve and so is consistent with thispicture. In Pictor A, Hardcastle & Croston (2005) argued that thedetailed positional variations of radio spectra and radio/X-ray ratiowere not consistent with such a model, but our data are not goodenough to rule it out here.We conclude that a varying electron spectrum alone cannotaccount for the observed variation in X-ray/radio ratio in the lobesof 3C 353 but that a magnetic field strength that varies by a factor of ∼ . throughout the lobes can explain it. Similar conclusions werereached by Hardcastle & Croston (2005) in their study of Pictor A. The surface brightness profile was converted to a pressure profileusing the method of Birkinshaw & Worrall (1993), so that the radiolobe pressures determined with
SYNCH could be directly comparedto the external pressure from the cluster at the position of the ra-dio galaxy. The centre of the East lobe lies ∼ arcsec from thecentre of the northern sub-cluster. At this radial distance where thetemperature of the environment is taken as . ± . keV, the exter-nal pressure is much greater than the internal pressure of the lobes(See Fig. 13). As FR II radio galaxies are expected to be in pres-sure balance or over pressured, either 3C 353 is not in the planeof the cluster or there is an additional contribution to the pres-sure from non-radiating particles such as hot thermal or relativis-tic protons in the lobes. Previous studies of FR II radio galaxies ingroups and clusters have shown that pressure balance can usuallybe achieved without additional protons (e.g. Croston et al. 2005;Hardcastle et al. 2002; Belsole et al. 2004). We therefore assume D E C L I NA T I ON ( J ) RIGHT ASCENSION (J2000)17 20 40 35 30 25 20-00 57 003058 003059 0030-01 00 0030 600 700 800 (cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1) (cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1) (cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)
Figure 11.
Spectral index map between 1.67 GHz and 327 MHz with radiocontours at 327 MHz for levels of . × (1,2,4,...) mJy/beam. The beamsize is shown by the circle in the bottom left corner and the colour bar at thetop shows the mapping of the grey levels to spectral index. that 3C 353 is not in the plane of the cluster and using the Eastlobe, determine that 3C 353 requires a shift in radial position corre-sponding to ∼ arcsecs to be in pressure balance with the north-ern sub-cluster (here we use the best estimate of the lobe pressurefrom Section 4.2). This places 3C 353 ∼ kpc either in front orbehind the centre of the cluster. At this position, the West lobe’s in-ternal pressure is also consistent with the intracluster pressure. Theposition of 3C 353 with respect to the pressure profile is shownin Fig. 13. At this distance from the cluster, the external pressuresseen by the lobes are different by a factor of ∼ . , which may helpto explain the difference in appearance of the lobes, the East hav-ing a spherical appearance whilst the West lobe is elongated withhotspots at the outer edge. c (cid:13) , 1–12 nverse-Compton emission from the lobes of 3C 353 Figure 12.
Colour-colour diagram for 3C 353 made at 7 arcsec resolution.
Figure 13.
Pressure profile for the northern sub-cluster shown by the greyedin region, with the measured pressure of the lobes of 3C 353 at the projecteddistance (solid line) and at the position with the radial shift applied to wherethe East lobe is in pressure balance with the sub-cluster (dashed line). Themeasured inverse-Compton emission was used to determine the pressurewithin each lobe. The temperature of the cluster at the radial distance of3C 353 was used.
In addition to the usual radio-lobe structure, the East lobe of 3C 353contains a dark circular region of unknown origin at 17:20:36.088,-00:58:44.15 (Fig. 1). Detectable in all radio frequencies except74 MHz (presumably due to its low resolution), the measured fluxesin this region are a factor of 100 above the rms background but area factor of 2 fainter than the surrounding lobe emission. The X-rayimage shows no sign of any feature in this region and neither doesthe Digital Sky Survey (DSS). As the level of the deficit is inde- pendent of frequency we can rule out an foreground absorber andthe required geometry is unrealistic for an obstruction in the lobe.Without further information we are unable to identify this feature.Iwasawa et al. (2000) included optical observations of thecluster region centred on 3C 353 taken with the University ofHawaii 2.2 m telescope in their analysis of Zw 1718.1-0108. Theyidentified three additional massive galaxies, none of which reside inthe northern sub-cluster. Deeper optical observations would help toestablish whether there are any galaxies associated with the north-ern sub-cluster and whether 3C 353 is the dominant member of thissub-cluster despite its position at the edge.
Our results can be summarized as follows: • By fitting an inverse-Compton model to the lobes, we foundthe East lobe to be electron dominated and the West lobe to beconsistent (within the large errors) with equipartition. • We determined that a variation in the electron spectron cannotaccount for the varying X-ray and radio emission alone, but that achange in the magnetic field strength across the lobes is required. • We have obtained a good X-ray spectrum of the nucleus of3C 353. Both the X-ray and optical properties of this source areambiguous but it appears to lie in a region of parameter space in be-tween those normally occupied by narrow-line and low-excitationradio galaxies. • We have detected an X-ray counterpart for the East hotspotoffset by . ± . kpc. • The northern and southern sub-clusters were found to beisothermal with a temperature difference of ∼ keV supportinga model in which they are two originally separate components un-dergoing a merger with no evidence for a violent interaction. ACKNOWLEDGEMENTS
JLG thanks the STFC for a research studentship. MJH acknowl-edges generous financial support from the Royal Society. We alsothank Jun Kataoka and the anonymous referee for helpful adviceand comments. The National Radio Astronomy Observatory is afacility of the National Science Foundation operated under cooper-ative agreement by Associated Universities, Inc. We thank the staffof the GMRT for their help with the observations with that tele-scope: GMRT is run by the National Centre for Radio Astrophysicsof the Tata Institute of Fundamental Research, India. This work ispartly based on observations obtained with
XMM-Newton , an ESAscience mission with instruments and contributions directly fundedby ESA Member States and NASA. Basic research in radio astron-omy at the Naval Research Laboratory is supported by 6.1 basicresearch.
REFERENCES
Arnaud, M. et al., 2002, A&A 390, 27Belsole, E., Worrall, D.M., Hardcastle M.J., Birkinshaw, M. &Lawrence, C.R., 2004, MNRAS 352, 924Besole, E., Worrall, D.M. & Hardcastle, M.J., 2006, MNRAS 366,339Birkinshaw, M. & Worrall, D.M., 1993, ApJ 412, 568Brunetti, G., Setti, G. & Comastri, A., 1997, A&A 325, 898 c (cid:13) , 1–12 J.L. Goodger et al.
Brunetti, G., 2000, Astropart. Phys., 13, 107Croston, J.H., Hardcastle, M.J., Birkinshaw, M. & Worrall, D.M.,2003, MNRAS 346, 1041Croston, J.H., Hardcastle, M.J., Harris, D.E., Belsole, E., Birkin-shaw, M. & Worrall, D.M., 2005, ApJ 626, 733Dickey, J.M. & Lockman, F.J., 1990, ARAA 28, 215Erlund, M.C., Fabian, A.C., Blundell, K.M., Moss, C. & Ballan-tyne, D.R., 2007, MNRAS 379, 498Fanaroff B.L. & Riley, J.M., 1974, MNRAS 167, 31PFeigelson, E.D., Laurent-Muehleisen, S.A., Kollgaard, R.I. & Fo-malont, E.B., ApJ 449, L149Grandi, P., Malaguti, G. & Fiocchi, M., 2006, ApJ 642, 113Haas, M. et al., 2004, A&A 424, 531Hardcastle, M.J., Birkinshaw, M. & Worrall, D.M., 1998a, MN-RAS 294, 615Hardcastle, M.J., Birkinshaw, M., Cameron, R.A., Harris, D.E.,Looney, L.W. & Worrall, D.M., 2002, ApJ 581, 948Hardcastle, M.J. & Croston, J.H., 2005, MNRAS 363, 649Hardcastle, M.J., Evans, D.A. & Croston, J.H., 2006, MNRAS370, 1893Hardcastle, M.J., Kraft, R.P., Worrall, D.M., Croston, J.H., Evans,D.A., Birkinshaw, M & Murray, S.S., 2007, ApJ 662, 166Isobe, N., Tashiro, M., Makishima, K., Iyomoto, N., Suzuki, M.,Murakami, M.M., Mori, M. & Abe, K., 2002, ApJ 580, L111Iwasawa, K., Ettori, S., Fabian, A.C., Edge, A.C. & Ebeling, H.,2000, MNRAS 313, 515Kassim, N.E. et al., 2007, preprint (astroph/0704.3088)Katz-Stone, D.M., Rudnick, L. & Anderson, M.C., 1993, ApJ407, 549Kataoka, J. & Stawarz, L., 2005, ApJ 622, 797Laing, R.A., Jenkins, C.R., Wall, J.V. & Unger, S.W., 1994, ASPC54, 201Lazio, T.J.W., Kassim, N.E., and Perley, R.A. 2005, Low-Frequency Data Reduction at the VLA: A Tutorial for New UsersMackay, C.D., 1973, MNRAS 162, 1Read, A.M. & Ponman, T.J., 2003, A&A 409, 395Sambruna, R.M., Eracleous, M. & Mushotzky, R.F., 1999, ApJ526, 60Schlegel, D.J., Finkbeiner, D.P. & Davis, M., 1998, ApJ 500, 525Simpson, C., Ward, M., Clements, D.L. & Rawlings, S., 1996,MNRAS 281, 509Swain, M.R. 1996 Ph.D. thesis, Univ. RochesterSwain, M.R., Bridle, A.H. & Baum, S.A., 1998, ApJ 507, L29Tashiro et al. 1998, ApJ 499, 713Tashiro, M., Makishima, K., Iyomoto, N., Isobe, N. & Kaneda,H., 2002, ApJ 526, L19 c (cid:13)000