Discovery of a Giant Radio Fossil in the Ophiuchus Galaxy Cluster

We obtained from the GMRT archive the observations of the Ophiuchus cluster at 153 MHz and 240 MHz used by M10 and P09 and reduced them using the National Radio Astronomy Observatory Astronomical Image Processing System (AIPS). Details on these observations are summarized in Table 1.

The data were collected in spectral-line observing mode, using the GMRT hardware backend, with a total observing bandwidth of 8 MHz. We carefully inspected the data and found that all observations were partially impacted by radio frequency interference. We used RFLAG to excise visibilities affected by radio frequency interference (RFI), followed by manual flagging to remove residual bad data. Gain and bandpass calibrations were applied using the standard primary calibrators 3C286 at 153 MHz and 3C286 and 3C48 at 240 MHz. The Scaife & Heald (2012) scale was adopted in SETJY to set their flux densities at these frequencies. Phase calibrators, observed several times during the observation, were used to calibrate the data in phase. To reduce the size of the data sets, at the same time minimizing the effects of bandwidth smearing, the calibrated visibilities were averaged appropriately in frequency, with 14 channels at 240 MHz and 21 channels at 153 MHz, each 0.375 MHz and 0.25 MHz wide, respectively. A number of phase self-calibration iterations, followed by a final self-calibration step in amplitude and phase, were applied to the target visibilities. During the self-calibration process, images were made using widefield imaging, decomposing the primary-beam area into a large number of smaller facets. Additional facets were placed on outlier bright sources out to a distance of 10° from the phase center. The final images were produced using widefield and multiresolution imaging in AIPS with the Briggs robust weighting (Briggs 1995) set to 0. At 240 MHz, we combined the two final self-calibrated data sets in the uv plane to produce the final images. The root mean square (rms) noise levels reached in our full-resolution images are reported in Table 1. Our values are significantly lower (∼40%–50%) than the noise levels reached by M10 and P09. Finally, correction for the GMRT primary-beam response was applied using PBCOR in AIPS.⁹

Residual amplitude errors, σ_amp, are estimated to be within 15% at 153 MHz and 10% at 240 MHz (Chandra et al. 2004). Errors on measured flux densities were calculated as

$$\begin{eqnarray}&&{\sigma }_{{S}_{\nu }}=\sqrt{{\left({\sigma }_{\mathrm{amp}}\times {S}_{\nu }\right)}^{2}+{\left({\sigma }_{\mathrm{rms}}\times \sqrt{{N}_{\mathrm{beam}}}\right)}^{2}},\end{eqnarray} \tag{ 1 }$$

where S_ν is the source flux density at the frequency ν, σ_rms is the noise level in the image, and N_beam is the number of beams crossing the source.

Figure 2 presents our GMRT images at 240 MHz at a resolution of 45'' (a) and 120'' (b). The radio emission is dominated by three bright extended radio galaxies. A detailed morphological and spectral study of these sources has been presented by M10. The central diffuse minihalo (G09, M10) is well detected in our images, where it occupies an area ∼200 kpc in radius. The location of the central cD galaxy (yellow cross) coincides with the faint radio point source embedded in the minihalo emission (M10, W16).

Zoom In Zoom Out Reset image size

A distinct region of extended emission (hereafter referred to as relic lobe) is visible southeast of the cluster center. Its brightest part (ridge) coincides with source E in M10, but we see a considerably larger source than in M10, thanks to the higher sensitivity of our GMRT images. The relic lobe extends out to a distance d ∼ 17' (∼570 kpc) from the cluster center and has a radius of r ∼ 68 (∼230 kpc). A GMRT image at 153 MHz at the resolution of 120'' is shown in Figure 3. Due to the lower sensitivity of this image, only the innermost part of the minihalo (r ∼ 130 kpc) and the brightest portion of the relic lobe (d ∼ 400 kpc) are well detected.

Zoom In Zoom Out Reset image size

In Figure 4, we overlay our GMRT 240 MHz contours on the low-frequency continuum images from GLEAM in the 72–103, 103–134, 139–170 and 170–231 MHz subbands. At the lowest GLEAM frequencies (panels (a) and (b)), the relic lobe is fully detected with good spatial coincidence with the GMRT image. The 139–170 MHz and 170–231 MHz subband images, instead, only detect its northern ridge well. The ridge is also visible in the VLSSr image at 74 MHz, shown in Figure 5. Its peak emission is detected at a 4σ level (1σ = 180 mJy beam⁻¹). The fainter emission from the minihalo is not detected at this sensitivity level. Finally, in Figure 6 we compare the 240 MHz emission to a higher-frequency VLA image at 1477 MHz (from G19), showing the central minihalo and only hints of the much lower surface brightness emission from the relic lobe.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We measured the flux density of the relic lobe in the GMRT, GLEAM, VLSSr and VLA images within a circular region of radius r = 68, centered on R.A._J2000 = ${17}^{{\rm{h}}}{12}^{{\rm{m}}}50\buildrel{\rm{s}}\over{.} 2$ and decl._J2000 = −23^d31^m13s. This region corresponds to the X-ray cavity size inferred from the XMM-Newton image (Figure 10) and is entirely filled by the relic lobe at 240 MHz (Figure 2). Table 2 summarizes the relic lobe flux density at all frequencies. Errors include the local image rms and flux calibration uncertainty (Equation (1)). Differences between the flux-density scales adopted for the GMRT and VLSSr (Scaife & Heald 2012), GLEAM (Baars et al. 1977), and VLA data (Perley & Butler 2013) are estimated to be at most ∼5% (Perley & Butler 2017).

For the VLSSr, we corrected the flux measured on the image for the clean bias and adopted a flux uncertainty of 20%, as appropriate for extended sources detected at a very low signal-to-noise ratio (Lane et al. 2014). This high uncertainty also accounts for the source location at the edge of the VLSSr mosaic, as fluxes can be affected by uncertainties in the primary-beam model at large distances from the pointing center.

For GLEAM, a systematic flux uncertainty of 8% is expected for sources in the same decl. range as the Ophiuchus cluster (Hurley-Walker et al. 2017). Due to the very low Galactic latitude of Ophiuchus, the GLEAM fluxes reported in Table 2 have been corrected for contamination due to Galactic diffuse synchrotron emission, to which GLEAM is highly sensitive thanks to the very short uv spacings of the MWA (e.g., Su et al. 2017). The GLEAM images of the region containing the Ophiuchus cluster in fact exhibit large-scale structures of Galactic emission, which affect flux-density measurements. Figure 7 shows a 3° × 3° region, centered on Ophiuchus, of the 170–231 MHz GLEAM image, where large variations of foreground emission are visible across the image. In particular, the Ophiuchus cluster appears to be located at the boundary of a bright region of Galactic emission. To estimate the contribution of this Galactic foreground, we have extracted the flux density within a set of eight circular source-free regions of the same size as the region used to measured the flux density of the relic lobe. As shown in Figure 7, the regions were carefully placed around the cluster to sample a wide range of surface brightness variations. The following average values were obtained for each of the four GLEAM subband images: −2.41 Jy, −2.44 Jy, −0.89 Jy and +1.01 Jy, going from the lowest- to the highest-frequency subband (since the large-scale Galactic structure is not deconvolved in the GLEAM images, the Galactic contribution can be positive or negative; Hurley-Walker et al. 2017). Finally, these values were used to correct the flux densities measured for the relic lobe in the corresponding subband images.

Zoom In Zoom Out Reset image size

Using the flux-density measurements in Table 2, we derived the integrated radio spectrum of the relic lobe between 74 and 1477 MHz, shown in Figure 8. The spectrum is very steep and well described by a power law with a spectral index α_tot = 2.4 ± 0.1. There is no indication of spectral curvature over the interval of frequency covered by our data. If we use only the low-frequency data points (74–240 MHz), we obtain a slope α_low = 2.7 ± 0.2, which is consistent within the errors with α_tot. Similarly, if we compute the spectral index from the GLEAM points alone, the resulting slope is 2.5 ± 0.3. The two GMRT measurements (blue points) give a flatter spectrum (α = 1.7 ± 0.4). However, the 153 MHz flux is likely underestimated due to the missing structure in the relic lobe at 153 MHz compared to the 240 MHz image (Figures 2 and 3). The 153 MHz flux density is also lower than the GLEAM flux density at a similar frequency of 154 MHz, though the two values are still consistent within the error bars (Table 2).

Zoom In Zoom Out Reset image size

M10 derived the integrated radio spectrum of the central minihalo and measured a possible steepening of the spectral index from α = 1.4 ± 0.3 in the 153–240 MHz interval to α = 1.60 ± 0.05 in the 240–1477 MHz range. However, the flux-density values used by M10 include part of the newly discovered relic lobe. To investigate how this may impact the behavior of the minihalo spectrum, we rederived the minihalo flux densities after exclusion of the emission contributed by the relic lobe. The faint emission associated with the minihalo is detected in both our GMRT images at 153 and 240 MHz (Figures 1 and 2). The brightest region of the minihalo is also detected in the GLEAM images. However, due to lower angular resolution, it is not possible to effectively separate the minihalo from the central radio galaxy and bright extended source north of the cluster center. G19 measured a minihalo flux density of 62 ± 9 mJy at 1477 MHz, after subtraction of the contribution from discrete radio sources and relic lobe (see their Appendix A). Following G19, we measure a source- and lobe-subtracted flux density of 1224 ± 203 mJy at 153 MHz and 692 ± 70 mJy at 240 MHz for the minihalo. The resulting total spectral index is ${\alpha }_{\mathrm{tot},\mathrm{mh}}=1.32\pm 0.10$ between 153 and 1477 MHz. Contrary to the M10 finding of a possible steepening, the low- and high-frequency indices are found to be consistent within the errors (${\alpha }_{153-240\mathrm{MHz}}=1.26\pm 0.43$ and ${\alpha }_{240-1477\mathrm{MHz}}=1.33\pm 0.10$), indicating that the minihalo spectrum can be described by a single power law, at least up to 1477 MHz. High-sensitivity observations above this frequency are necessary to better constrain the overall shape of the minihalo spectrum and investigate the presence of a high-frequency steepening, as possibly seen in few other minihalos (e.g., Giacintucci et al. 2014).

The Ophiuchus cluster was observed by Chandra ACIS-I in 2014 for a total exposure of 230 ks (ObsIDs 16142, 16143, 16464, 16626, 16627, 16633, 16634, 16635, 16645). Results from this data set were first presented by W16. We extracted an image in the 0.5–4 keV band from these data using the standard procedure (e.g., Wang et al. 2016). The image is shown in Figures 1 and 9 with different colors to emphasize the curious concave X-ray brightness edge reported by W16; see also Figure 1 in W16. This rather sharp edge looks like a boundary of a cavity in the X-ray gas in projection, although the elongation to the south ("southern extension" in Figure 1) appears to be a separate hydrodynamic disturbance and possibly even a small subcluster, as suggested by W16. We point out in Section 5 that the cavity and the hydrodynamic disturbance are not mutually exclusive.

Zoom In Zoom Out Reset image size

The higher-resolution GMRT 240 MHz contours are overlaid on the Chandra image in Figure 9(b), which shows a striking spatial coincidence between the inner edge of the relic lobe (the "ridge") and the X-ray edge that is the boundary of the putative X-ray cavity.

The Ophiuchus cluster was observed with XMM-Newton for 37 ks in 2017 (ObsID 0505150101; Nevalainen et al. 2009). Several offset observations were performed as well, which we do not use here. We extracted images from this data set using the Extended Source Analysis Software (XMM–ESAS)¹⁰ package part of SAS version 11.0.0. We used the standard data cleaning and image technique (Snowden et al. 2008). The exposure-corrected and background-subtracted 0.4–7.2 keV image combining the European Photon Imaging Camera (EPIC) metal oxide semiconductor (MOS) and pn detectors is shown in Figure 10. It shows the same concave brightness edge (green arc) and allows us to trace it to somewhat larger scales. The green circle in panel (d) is selected to trace this edge; its radius is 230 kpc, which is somewhat bigger than in W16. This difference is not meaningful as it is an extrapolation from a small sector, but our larger radius also happens to encompass the giant low-frequency lobe that is seen in GLEAM (Figure 10(d)) and GMRT lower-resolution (Figure 10(c)) images.

Zoom In Zoom Out Reset image size

It is interesting to check the larger-scale structure of the cluster, in order to look for any ghost cavities on the opposite side. The XMM offsets do not cover the outskirts of the Ophiuchus cluster contiguously, so we have checked the archives of other X-ray observatories looking for a better large-scale image. ROSAT Position Sensitive Proportional Counter (PSPC) and the Advanced Satellite for Cosmology & Astrophysics (ASCA) have an interesting wide coverage; we extracted a ROSAT image because it has a better angular resolution. We used an archival 0.5–2 keV image from a 4 ks ROSAT PSPC pointing toward the cluster center. It is shown in Figure 11 with contours from the lowest-band GLEAM image that shows the fossil lobe. The X-ray statistics are clearly not sufficient to attempt a search for a countercavity. However, the X-ray image shows that the overall cluster gas distribution is elongated in the direction of the lobe, suggesting that the radio and X-ray emission are physically related.

Zoom In Zoom Out Reset image size

The comparison of the radio images and the high-resolution Chandra and XMM X-ray images (Figures 9 and 10) shows a large low-frequency radio lobe that fits hand in glove into the X-ray edge. The spatial coincidence is especially striking between the X-ray edge and the higher-resolution GMRT image (Figure 9b). Neither Chandra nor XMM existing images have the spatial coverage or sensitivity to trace the outer boundaries of the ICM cavity (which would look like a very low-contrast X-ray brightness edge), but the curvature of the visible edge implies that the GLEAM and GMRT radio lobe would fill this putative cavity (Figure 10) as seen in many clusters with AGN lobes, such as Perseus.

What makes this cavity unique is its size and the energy required to create it. The latter has been estimated by W16 to be pV ∼ 5 × 10⁶¹ erg, where p comes from the pressure of the ICM around it and V from the Chandra X-ray edge curvature. This value is of course only an order-of-magnitude estimate, as it involves several assumptions, such as where the cavity is located on the line of sight (which should not be too far from the cluster sky plane because it is still visible in the image) and the ICM density and temperature profiles at the relevant 3D radii. With that caveat in mind, it is interesting to compare the cavity pV and size with the AGN-blown cavities in other clusters. We show such a comparison in Figure 12, using data for individual cavities (as opposed to their pairs, since we see only one in Ophiuchus) from Rafferty et al. (2006), adding the outer cavities in Hydra A (Wise et al. 2007), the recently discovered cavities in SPT-CLJ0528-5300 (Calzadilla et al. 2019), and a pair of giant cavities in A2204 (Sanders et al. 2009; these are seen in the X-ray but not in the radio, so their nature is still ambiguous). The cavity in Ophiuchus would require an AGN outburst that is an extreme outlier in terms of its total energy—orders of magnitude more energetic than those typically seen in clusters, and five times more energetic than the previous record holder, MS 0735+74.

Zoom In Zoom Out Reset image size

If the lobe has started its life as a buoyant bubble injected near the cluster center by the Ophiuchus central galaxy, it would take it at least 240 Myr to rise to the current radius moving with the speed of sound, which is an upper limit on the velocity, so the actual age would be greater than that.

Another method often used to estimate the age of radio galaxy lobes is to look for a break in their radio spectrum and model it. The highest-energy relativistic electrons cool first and the electron power spectrum (and the spectrum of their synchrotron emission) develops an exponential cutoff, which moves to lower energies (and lower radio frequencies) as time passes (e.g., Komissarov & Gubanov 1994). While the radio spectrum of the Ophiuchus cavity is very steep, it is described well by an unbroken power law across our range of frequencies from 74–1477 MHz, and we do not see any high-frequency cutoff (Section 3). We cannot simply assume that the cutoff is outside our frequency band, because for an aged source of such an enormous linear size, the total synchrotron spectrum is likely to be the sum of a patchwork of local spectra of different shapes, which could add up to a power law. A large and old radio lobe is likely to contain very nonuniform and filamentary magnetic fields, which would produce different synchrotron spectra and different local synchrotron cooling times even for a single underlying electron power spectrum (e.g., Tregillis et al. 2004). The resulting total radio synchrotron spectrum thus cannot be usefully related to the spectrum of the emitting electrons (e.g., Eilek & Arendt 1996; Katz-Stone & Rudnick 1997).

In this regard, it may be useful to look at a spectral index map with interesting spatial resolution, which would at least remove the spectral averaging over the face of the lobe. We have recently obtained new, deeper, and more sensitive observations of this source with the upgraded GMRT in Band 2 (130–260 MHz) and Band 3 (250–500 MHz) and we will present the analysis of these new observations in a forthcoming paper.

The spatial coincidence of the X-ray cavity (or at least its inner edge) and the giant, steep-spectrum radio lobe is striking and strongly suggests that we have uncovered a fossil of an enormous AGN outburst. However, there are several puzzles in this scenario. First, the central AGN, the supposed culprit, is rather faint and does not show any jets (W16), nor are there any other more powerful AGNs in the vicinity of the relic lobe. However, the central AGN could have been much more active in the past. The X-ray image shows that sloshing has currently displaced the peak of the cluster cool gas away from the cD galaxy, something that is rarely seen in clusters and that can temporarily starve the AGN of its accretion fuel (Hamer et al. 2012, W16). The gas sloshing seen in the X-ray image may even have been set off by the powerful outburst that has created the giant lobe.

A more difficult puzzle is that we see only one lobe, whereas AGN jets usually come in symmetric pairs and produce a pair of radio lobes in the ICM on two sides of the AGN. Given the likely advanced age of our lobe, we speculate that the counterpart may have propagated into a less-dense ICM on the other side of the cluster and completely faded away. Furthermore, the lobes move together with the ICM in which they are embedded, which breaks the symmetry of the pair (e.g., the outer lobes in Hydra A; Nulsen et al. 2005). Strong cluster-wide bulk gas motions, such as those generated by a merger, may have moved the counterpart lobe somewhere into the outskirts, where it faded. An exotic alternative is that the radio galaxy that gave rise to this bubble is not the cD galaxy, but some other galaxy perhaps in the cluster outskirts; the counterpart lobe would then fade quickly in the low-density gas far from the cluster center.

Another serious problem was posed by W16—how did the cool core survive such an energetic event? There are clusters whose cores appear to have been torn to pieces by AGN outbursts (e.g., A2390, Vikhlinin et al. 2005). The Ophiuchus cool core has a steeper radial density and temperature gradient than most cool cores (W16), which may make it more stable against a disruption that comes from the side rather than from the nucleus. One can imagine long jets from the central AGN piercing the cool core and depositing energy and blowing bubbles right outside the core. On smaller linear scales, this is observed in some cluster galaxies with dense gaseous coronae (e.g., Sun et al. 2007; Sanders et al. 2014) or in remnant compact cool cores (e.g., Cheung et al. 2019), which are pierced by jets emitted by the AGN without disrupting the corona/compact core, and creating bubbles right outside.

The Ophiuchus core does exhibit stronger sloshing motions than most cool cores. It may have been triggered by an asymmetric pair of expanding bubbles deposited outside the core, which may have given the core a push from the side while avoiding destroying the core. It would be interesting to explore such a possibility using hydrodynamic simulations.

On larger linear scales, the ROSAT image (Section 4.3) exhibits a curious elongation in the exact direction of the giant lobe. There are two interesting possibilities here. One is the analogy with Hydra A, where a pair of giant radio bubbles is ensconced in an X-ray cocoon (see Figure 2 in Nulsen et al. 2005), whose outer edge is a weak shock front (Simionescu et al. 2009). Another possibility is cluster-wide sloshing, such as recently found in Perseus (Simionescu et al. 2012) and A2142 (Rossetti et al. 2013), both of which exhibit cold fronts ∼1 Mpc from the center. If the enormous AGN outburst managed to move the cool core and make it slosh, it might induce cluster-wide sloshing as well—its mechanical energy approaches that of minor mergers.

We note here that the hydrodynamic instability of a sloshing front, proposed by W16 to explain the concave brightness edge, and our radio bubble explanation, are not mutually exclusive. The "subcluster," or the surface brightness excess south of the core at one end of the concave feature (Figure 1(b) in W16 and "southern extension" in our Figure 1), exhibits a morphology that makes it likely to be a splash of cool gas from the core generated by the expansion of the bubble and the resulting major hydrodynamic disturbance of the cluster.