PROBING THE EXTREME REALM OF ACTIVE GALACTIC NUCLEUS FEEDBACK IN THE MASSIVE GALAXY CLUSTER, RX J1532.9+3021

RX J1532.9+3021 was originally observed in 2001 with the Chandra CCD Imaging Spectrometer (ACIS) in VFAINT mode and centered on the ACIS-S3 back-illuminated chip (10 ks; ObsID 1649) and the ACIS-I3 front-illuminated chip (10 ks; ObsID 1665). It was subsequently observed on 2011 November 16 for 90 ks (ObsID 14009; PI: Hlavacek-Larrondo) in VFAINT mode, significantly improving the image quality and centered on the ACIS-S3 back-illuminated chip with the ACIS-S1, ACIS-S2, ACIS-S4, and ACIS-I3 also switched on.

All three ObsIDs were processed, cleaned and calibrated using the latest version of the ciao software (ciaov4.5, caldb 4.5.6), and starting from the level 1 event file. We applied both charge transfer inefficiency and time-dependent gain corrections, as well as removed flares using the lc_clean script with a 3σ threshold. The net exposure times are shown in Table 1. We then exposure-corrected the image, using an exposure map generated with a monoenergetic distribution of source photons at 1.5 keV. The combined exposure-corrected 0.5–7.0 keV image is shown in Figure 1.

Zoom In Zoom Out Reset image size

The X-ray data were spectroscopically analyzed with xspec (version 12.8.0a). We use the abundance ratios of Anders & Grevesse (1989) throughout this study, and exclude points sources from all regions by identifying them by eye. The latter are only present beyond a radius of 250 kpc. We also mostly use C-statistics throughout the study. The background for each ObsID is selected as a region located on the same chip, devoid of point sources, and far from any cluster emission. It is also chosen to occupy the same location on the sky in all three ObsIDs. Since RX J1532.9+3021 has a high redshift, the cluster occupies only a small fraction of the chips, about 50% within a radius of 1 Mpc, and a background can easily be extracted from the surrounding region. Note that our choice of background does not significantly affect our results. If we use blank-sky observations or select a similar region, located on the second back-illuminated chip (ACIS-S1) for ObsID1649 and ObsID14009, as well as another chip for ObsID1665 (ACIS-I0), we find results consistent with each other.

We use the Kalberla et al. (2005) value for the Galactic absorption throughout this study. If we select a region located between 50 kpc and 200 kpc, and fit an absorbed mekal plasma model to the 0.5–7 keV spectrum while letting the Galactic absorption, temperature, abundance and normalization parameters free to vary, we find that the derived absorption value is consistent with the Kalberla et al. (2005) value to within 1σ, justifying our choice for the Galactic absorption.

There are two datasets available in the XMM-Newton archive for RX J1532.9+3021, see Table 2. We only consider the high spectral resolution XMM-Newton reflection grating spectrometer (RGS) data since these provide better constraints on the amount of cool gas (see Peterson & Fabian 2006 and references therein). Data reduction was accomplished following the standard sas procedures. A point-spread function (PSF) extraction region of 90% and a pulse-height distribution region of 95% were used. The data were also grouped to have at least 25 counts spectral bin⁻¹ and background model spectra were created with rgsbkgmodel.

There are several observations of RX J1532.9+3021 in the Very Large Array (VLA) archive. We only consider the data in A and B configurations, since the spatial resolution is well matched to that of the Chandra data. The data are summarized in Table 3. We reduced them using the standard astronomical image processing system (Greisen 2003).

The combined A and B configuration 1.4 GHz image, with a 2'' spatial resolution is shown in the left panel of Figure 2. The best VLA 325 MHz image at 5'' resolution is shown in the right panel of Figure 2. The 1.4 GHz map reveals a strong core and twin elongated lobes to the east and west that coincide with the X-ray cavities. These lobes extend to the edge of the cavities along the jet direction, but are thinner along the perpendicular direction. The 325 MHz map reveals a jet-like feature along the north to south direction. We further discuss this feature in Section 7.2 in the context of the presence of older outflows, known as ghost cavities.

Zoom In Zoom Out Reset image size

There is also a faint underlying component seen in the maps that extends out to 70 kpc in radius (μJy level). We interpret this component as a mini-halo, similar to what is seen in other cool core clusters (e.g., Mazzotta & Giacintucci 2008). This mini-halo has also been recently reported by Giacintucci et al. (2013), and we show in the middle panel of Figure 2 their Giant Metrewave Radio Telescope (GMRT) 610 MHz radio contours (which will also appear in Kale et al. 2013).

Figure 3 shows the spectral index map, obtained by calculating the index between the VLA 1.4 GHz and 325 MHz images. This map was produced by first convolving the two images to a similar resolution of ≈5'', and then computing the spectral index for regions where the radio emission was above 3σ_rms. The spectral index (α) is defined such that the flux density scales as S_ν∝ν^−α. Interestingly, Figure 3 shows that the core has a flat, and even slightly inverted radio spectrum. The extended emission on the other hand is characterized by a steep spectrum with α ≈ −2, which is typically seen in mini-halos and indicates aging of the relativistic particles producing the radio emission (e.g., Ferrari et al. 2008).

Zoom In Zoom Out Reset image size

RX J1532.9+3021 was observed with the Hubble Space Telescope (HST) as part of the Cluster Lensing And Supernova survey with Hubble (Postman et al. 2012). The right panel of Figure 1 shows the F814W image taken with the Advanced Camera for Surveys (ACS) Wide Field Channel (WFC). It contains the redshifted Hα emission line (Hαλ6562) and highlights well the complex filamentary structure of the BCG. We also show the F475W and F625W images taken with the same camera later in this work to further highlight the filaments. These contain respectively the [O ii] λ3732 and [O iii] λ5007 lines.

We use the F225W image taken with the Wide Field Camera 3 UVIS imager to compute a UV star formation rate (SFR) associated with the galaxy. The extended filaments are not seen in this filter, which corresponds to a rest-wavelength of ≈1737 Å. We derive the total SFR associated with the BCG within a circular region centered on the galaxy with a radius of 35 (≈18 kpc). Beyond this region, no significant UV emission is detected. We use the Kalberla et al. (2005) value for the Galactic reddening assuming N_H = (2.21 ± 0.09) × 10²¹ A_V (Güver & Özel 2009) and R = 3.1, to correct the HST F225W, F275W and F336W images for Galactic reddening. This results in an E(B − V) of 0.0335 ± 0.0015 and we use the reddening curve of Cardelli et al. (1989).

To estimate the intrinsic reddening in the galaxy, we then make use of the F275W and F336W images taken with the same camera, while assuming that the monochromatic UV luminosity is flat in the 1500–2800 Å wavelength range (Kennicutt 1998). This allows us to derive an intrinsic flux density in the F225W band, and therefore an SFR for the galaxy in this band. We correct the flux density on a pixel-to-pixel basis. The average intrinsic reddening correction is E(B − V) = 0.102, peaking at E(B − V) ∼0.2 in the central 0.5 arcsec. We note here that there are significant uncertainties arising both from the reddening corrections and the background determination of the UV filters. These dominate the uncertainties from counting statistics. We propagate them in our determination of the error for the SFR. We also assume that the UV emission is dominated by star formation, and not line emission.

Taking these factors into account, the flux density corrected for Galactic reddening is f_F225W = (6.0 ± 2.9) × 10⁻²⁸ erg cm⁻² s⁻¹ Hz⁻¹. After correction for intrinsic reddening, the flux density is f_F225W = (1.2 ± 0.6) × 10⁻²⁷ erg cm⁻² s⁻¹ Hz⁻¹ leading to an SFR of 76 ± 38 M_☉ yr⁻¹ (Kennicutt 1998).

To study the temperature and metallicity distribution across the cluster, we bin different regions together using a contour binning algorithm which follows the surface brightness variations (Sanders 2006).

We begin by analyzing the large-scale structures within the central 500 × 500 kpc. We bin the regions so that they each contain at least 3600 counts, initially smooth the image with a signal-to-noise ratio of 15 and restrict the lengths to be at most two times that of a circle with the same area. We extract a 0.5–7 keV spectrum for each region and fit an absorbed single mekal model to the data while using C-statistics. We left the temperature, abundance, and normalization parameters free to vary. The resulting temperature and metallicity maps are shown in Figure 10. For the temperature, the errors on each value vary from 4% in the inner regions to 8% in the outer regions, and for the abundance they vary from 16% (inner) to 36% (outer). The average temperature in the eastern direction (over four bins) is 6.0 ± 0.2 keV, whereas the temperature to the west, located within bins of similar radius, is 7.4 ± 0.2 keV. The east direction is therefore statistically cooler than the west, and may be the result of a past merger between the cluster and a cooler subcluster along the east-west direction. There is also a south–west region that appears more metal rich (0.7 Z_☉) than the average value seen throughout the cluster (0.4 Z_☉), similar to what is seen in the Ophiuchus cluster (Million et al. 2010), but this increase lies within 2σ of the neighboring bins. Also seen, is a contrast in the temperature structure to the west, where the temperature jumps from 5 keV to 7 keV at a radius of around 65 kpc. This could indicate the presence of a cold front. We further investigate this in the following section. Note that the evidence of multi-phase gas found in Section 3 mostly contributes to the inner 50 kpc, and does not affect the large-scale temperature and abundance map significantly beyond the inner few bins.

Zoom In Zoom Out Reset image size

Next, we analyze the inner 150 × 150 kpc region which shows interesting features especially at soft X-rays (see right panel of Figure 11). To highlight the thermodynamic properties of these features, we bin the regions so that they follow the surface brightness contours of the soft 0.3–1.0 keV image. We initially smooth the image with a Gaussian function (σ = 1.3 pixel), and then bin the regions so that they each contain at least 400 counts. We also restrict the lengths to be at most 1.3 times that of a circle with the same area. However, we extract the spectrum from each region in the full 0.5–7 keV range, which will contain more than 400 counts (around 900 counts). We use C-statistics and show the results in Figure 11. Here, the errors vary from 7% in the inner regions to 11% in the outer regions for the temperature, and from 30% (inner) to 55% (outer) for the abundance. Due to the large error bars on the abundance, we only focus on the temperature structure. The number of counts from the current data are also not sufficient to fit a two-temperature model to each of the regions selected. We would need at least two to three times the number counts per region to provide a good fit (or regions two to three times larger). Therefore, although we found evidence for multi-temperature gas within the inner 50 kpc, we focus here on the small-scale temperature structure, choose small regions containing around 900 counts each, and only fit a single mekal model to each region. Both the small-scale and large-scale temperature maps show that the cluster has a cool core, with the temperature dropping from 8 keV in the outskirts to 4 keV in the central 100 kpc.

Zoom In Zoom Out Reset image size

The plume-like feature mentioned in Section 4 stands out clearly in the right panel of Figure 11, indicating that the plume most likely contains cooler material. However, when fitting an absorbed mekal model to the Chandra spectrum of the plume, we find that the temperature and abundance are consistent to a 1σ level with those of a surrounding region, located within the same radius as the plume but excluding the plume. The data therefore remain too poor to determine clearly how much cooler the plume is, and deeper observations are needed.

We now proceed to do a more in depth analysis of the thermodynamic properties of the cluster, by computing thermodynamic profiles.

We first select annular regions containing roughly a signal-to-noise ratio of 90 (8000 counts), and push the analysis to larger radii. The projected and deprojected temperature, metallicity, electron density, and electron pressure profiles are shown in Figure 12. For each spectrum, we fit an absorbed mekal model while using C-statistics and we deprojected the spectra using the standard projct mixing model in xspec. Note that the multi-temperature gas discussed in Section 3 only contributes to the inner 3 annuli, and the thermodynamic profiles of the hotter component are consistent with those shown in Figure 12.

Zoom In Zoom Out Reset image size

Figure 12 shows that beyond a radius of 100 kpc, the projected metallicity profile decreases with radius from Z ≈ 0.6 solar to Z ≈ 0.3 solar. This figure also shows the characteristic drop in temperature with decreasing radius seen in cool core clusters. However, there is a distinct, interesting feature at a radius of 65 kpc (see dashed line). Here, the deprojected temperature increases from 4.5 keV to 6 keV, while the deprojected electron pressure profile breaks at this radius, such that the pressure remains constant across the front. The latter is a typical characteristic of cold fronts.

To further investigate the potential cold front, we compute thermodynamic profiles along the same four sectors as Figure 7. We select regions so that they each contain a signal-to-noise ratio of 45 (2000 counts), and do not correct for deprojection effects due to the limited number of counts. The number of counts are also not sufficient to fit a two-temperature model to the data. The results are shown in Figure 13, and the strongest temperature jump observed lies along the western direction and a radius of 65 kpc. Interestingly, the front coincides with the edge of the western X-ray cavity as well as the edge of the radio mini-halo. We further discuss this front in Section 7.1.

Zoom In Zoom Out Reset image size

Using the Chandra X-ray images, we searched for evidence of a central point source. The hard 2–8 keV image reveals three bright pixels in the central region that coincide with the nucleus as seen from the HST image (R.A. = 15^h32^m5377, decl. = +30°20'5944; see inset of Figure 8). Here, the nucleus in the HST image corresponds to the brightest point within the central 20 kpc. Taking a circular region with a radius of 1'' centered on these bright pixels, we compare the total counts to the 3σ upper limit of the surrounding background. The background is taken as a surrounding annulus with an interior radius of 1'' and an outer radius of 2'', and separately as a surrounding annulus with an interior radius of 2'' and an outer radius of 4''. In both cases, the counts within the central 1'' lie well within the 3σ upper limit of the background, and there is no clear evidence for an excess associated with a central point source.

To further show this, we calculate the hardness ratio, defined as (H − S)/(H + S). Here, H represents the hard number of counts within a certain region as seen from the 2–8 keV image, and S represents the soft number of counts from the same region, but taken from the 0.5–2 keV image. We compute the hardness ratio for the central 1'' region and find a ratio of −0.50 ± 0.04, where the uncertainty is determined by assuming that the counts are governed by Poisson statistics. The hardness ratio of the surrounding 2''–4'' region is −0.57 ± 0.02. As seen, the ratios lie within 2σ of each other, and there is no clear evidence that the central 1'' has a harder spectrum than the surrounding cluster emission.

The spectrum of the central 1'' region also shows no clear evidence for the presence of non-thermal emission. By taking this region and a surrounding annulus with inner radius 2'' and an outer radius of 3'' as a background, we find that a mekal and power-law model both fit the data equally well. The lack of any obvious non-thermal contribution to the nucleus, even with the new deep 90 ks observations confirms that the central nucleus in RX J1532.9+3021 is radiatively inefficient (see Hlavacek-Larrondo & Fabian 2011).

We therefore derive an upper limit to the nuclear 2–10 keV flux following the method used in Hlavacek-Larrondo & Fabian (2011). The idea is to convert a nuclear count rate into a flux using the pimms Web interface¹⁰. More precisely, we compute the total number of counts within the central 2'' × 2'' square region using the 0.5–7 keV X-ray image of the new ACIS-S deep observations (ObsID 14009). We then subtract the counts of a surrounding background taken as a square annulus with an inner 3'' × 3'' and an outer 4'' × 4'' square, scaled to the same number of pixels. The resulting 3σ upper limit is then converted to a luminosity with pimms, assuming a power-law model with a photon index of 1.9. We find a 3σ upper limit of 5 × 10⁴² erg s⁻¹ in the 2–10 keV energy range. Note that, changing the location of the nucleus by ±2 pixels does not change the results and that the value is lower than the one determined in Hlavacek-Larrondo & Fabian (2011), which is expected since the new observations allow a more precise estimate of the upper limit.

There are two clear X-ray cavities associated with the central AGN. The energy stored within each of the cavities can then be estimated from the following equation:

$$\begin{equation} E_{\rm bubble}=\frac{\gamma }{\gamma -1}PV, \end{equation} \tag{ 3 }$$

assuming that the cavities are in pressure equilibrium with the surrounding medium (e.g., Bîrzan et al. 2004). Here, P is the thermal pressure at the radius of the bubble, V is the volume of the cavity and for a relativistic fluid γ = 4/3, therefore E_bubble = 4 PV.

We also assume that the cavities are of ellipsoidal shape with $V=4{\pi }R^2_{\rm w}R_{\rm l}/3$, where R_l is the projected semi-major axis along the direction of the radio jet axis, and R_w is the projected semi-major axis perpendicular to the direction of the radio jet axis. These radii were measured based on the raw, exposure-corrected 0.5–7 keV image and we assign a 20% uncertainty to R_l and R_w.

To compute cavity powers, we divide the energy by the bubble age given by the buoyant rise time (Churazov et al. 2001) defined as

$$\begin{equation} t_\mathrm{buoyant}= R\sqrt{\frac{SC_{\rm D}}{2gV}} \,, \end{equation} \tag{ 4 }$$

the refill time (McNamara et al. 2000) defined as

$$\begin{equation} t_\mathrm{refill}= 2\sqrt{\frac{r}{g}} \,, \end{equation} \tag{ 5 }$$

or the sound crossing time defined as

$$\begin{equation} t_{c\mathrm{s}}=\frac{R}{c_\mathrm{s}} \,. \end{equation} \tag{ 6 }$$

Here, R is the projected distance from the optical nucleus to the middle of the cavity. S is the cross-sectional area of the bubble ($S={\pi }R_{\rm w}^2$), C_D = 0.75 is the drag coefficient (Churazov et al. 2001), g is the local gravitational acceleration such that g = GM(< R)/R², and r is the bubble radius and is equal to (R_lR_w)^1/2 for an ellipsoidal bubble. Finally, c_s is the sound crossing time of the gas. Between the three time estimates, it is not clear which is the best to use, although they are not expected to differ significantly. All thermodynamic quantities used for these calculations were taken from the deprojected quantities shown in Figure 12. The cavity locations are shown in Figure 14.

Zoom In Zoom Out Reset image size

Deeper exposures continue to discover new cavities, and in particular older X-ray cavities which have risen buoyantly (e.g., Fabian et al. 2011). Weak shocks and sound waves have been seen to contribute significantly to the power output of the AGN, but these have only been seen through deep observations of nearby systems (e.g., Forman et al. 2005). The cavity powers shown in Table 5 should therefore be treated as lower limits to the total mechanical energy being injected by the central AGN. We further discuss the cavity powers in Section 7.3.

We first discuss the potential cold front. Cold fronts are contact discontinuities, thought to originate in part from subcluster merger events and subsequent sloshing of low-entropy gas in cluster cores (see a review by Markevitch & Vikhlinin 2007). In Section 5, we found evidence for a temperature jump from 4.5 keV to 6.5 keV with increasing radius, strongest to the west at a radius of ≈65 kpc (Figure 13). Figure 12 also showed that there is a pressure break associated with the front, such that the pressure inside the front is low. The right panel of Figure 4 also showed a surface brightness edge at r ≈ 65 kpc, shaped such that it resembles those seen in other cold fronts (see the many examples in Markevitch & Vikhlinin 2007 for comparison). All of these results are consistent with the feature being a cold front.

The front also coincides with the edge of the radio mini-halo. ZuHone et al. (2013) showed that relativistic electrons can be reaccelerated by the turbulence generated from the sloshing of the core. These reaccelerated electrons then produce the mini-halo emission that is coincident with the regions bounded by the sloshing cold fronts (see also Mazzotta & Giacintucci 2008). This could explain the common location of the cold front and edge of the mini-halo in RX J1532.9+3021, but we note that the cold front is strongest to the west, and at this radius, it also coincides with the edge of the western cavity, similar to what is observed in 4C+55.16 (Hlavacek-Larrondo et al. 2011). If the cold front is only located to the west, it could be simply due to cool gas being dragged out by the central AGN through the outburst (e.g., Churazov et al. 2001). Alternatively, based on Figure 12, if one compares the value of the deprojected pressure at a radius of 30 kpc with the one at a radius of 80 kpc, then it seems that the pressure in this region is higher, consistent with what one would expect from a shock front. The break in the pressure profile seen in Figure 12 could therefore also be interpreted as a potential shock front driven by the expansion of the western cavity similar to the high pressure rims seen around the cavities in the Perseus Cluster (Fabian et al. 2003).

The large-scale temperature map in Section 5.1 revealed that the east direction is cooler (6.0 ± 0.2 keV) than the west (7.4 ± 0.2 keV). Such an asymmetry in the temperature is most likely caused by a past merger of a cooler sub-cluster with the main cluster. If the core is sloshing and generating the turbulence needed to reaccelerate the electrons in the mini-halo, then a past merger could have caused this.

The literature shows several examples of edges which have been interpreted as cold fronts and that have a clear metallicity jump associated with them, but no clear discontinuity in the temperature profile. These include the cold fronts in A2052 (de Plaa et al. 2010), A2199 (Sanders & Fabian 2006), and 2A 0335+096 (e.g., Mazzotta et al. 2003). Others do not show any evidence of a metallicity jump, but clearly show a discontinuity in the temperature profile, such as those in A496 (Dupke & White 2003) and A2204 (Sanders et al. 2005). In the case of 4C+55.16 (Hlavacek-Larrondo et al. 2011) and M87 (Simionescu et al. 2008), there is both a temperature and metallicity jump.

Figure 13 also suggests that there is a metallicity jump to the west at 65 kpc, such that the gas before the front is more metal rich (≈0.75 Z_☉) than gas beyond of the front (≈0.40 Z_☉). The errors bars remain however quite large, and it is not clear from the current data whether the jump is real. Interestingly, there is a second metallicity jump in the western direction at a radius of 40 kpc. This radius coincides with the center of the western cavity, consistent with the idea that metal rich gas is being dragged out by the cavity, similar to what is observed for other clusters (e.g., Kirkpatrick et al. 2011). Here, the metallicity before the cavity is Z ≈ 0.25 Z_☉, whereas at 40 kpc, the metallicity is Z ≈ 0.75 Z_☉. Although we do not show the small-scale abundance map in Figure 11 due to the large errors bars, there is an increase in metallicity associated with the western cavity of about ≈0.5 Z_☉. If we assume that the cavity has a volume of $V=4{\pi }R^2_{\rm w}R_{\rm l}/3$, and that the electron density at the location of the cavity is approximately 0.05 cm⁻³ (from the deprojected quantities), then the excess of iron mass compared to he iron content before the cavity is around 5 ×10⁷ M_☉. Note that, even more significant enhancements of metal-rich gas (10⁸–10⁹ M_☉) along buoyantly rising X-ray cavities have been seen in nearby clusters, such as Sérsic 159-03 (Werner et al. 2011) and Hydra A (Simionescu et al. 2009).

In Hlavacek-Larrondo et al. (2012), we found evidence for only one clear X-ray cavity associated with the central galaxy using the old ACIS-S observations (ObsID 1649). We also found that this cluster had the third strongest cooling luminosity in the 20 MACS clusters with cavities. The only other two stronger cool core clusters were MACS J0947.2+7623 and MACS J1447.4+0827, with cooling luminosities 20%–40% larger respectively. Based only on the western cavity, we concluded that the power stored within the cavity fell short by a factor of ≈2–3 of the cooling luminosity.

With the new 90ks observations, we find evidence for at least two X-ray cavities. We confirm the presence of the western cavity, and report on the newly discovered eastern cavity located at a similar radius and of similar size (see Figure 14). Both cavities also coincident with radio jets as seen from the 1.4 GHz VLA image (see left panel of Figure 2). The direction of the radio jets and the central AGN are also well aligned with each other.

The total enthalpy associated with both X-ray cavities (4 PV) is around 3 × 10⁶⁰ erg and the total power is (22.2 ± 8.6) × 10⁴⁴ erg s⁻¹. This is sufficient to prevent the gas from cooling where L_cool ≈ 25 × 10⁴⁴ erg s⁻¹ in the 0.01–50.0 keV range. We show this more clearly in Figure 15. Beyond this energy range, the plasma model does not contribute significantly, and this value of L_cool represents the total cooling luminosity.

Zoom In Zoom Out Reset image size

In comparison, the X-ray cavities in RX J1532.9+3021 are around an order of magnitude more powerful than the inner X-ray cavities of the Perseus Cluster (z = 0.018; Dunn & Fabian 2004) and M87 (z = 0.0044; Forman et al. 2005), and several times that of Hydra A (z = 0.0538; Nulsen et al. 2005). They are also around two times more powerful than those in Cygnus A (z = 0.056), A1835 (z = 0.253), and PKS 0745-19 (z = 0.01028; Rafferty et al. 2006), as well as MACS J1423.8+2404 (z = 0.5449; Hlavacek-Larrondo et al. 2012). These X-ray cavities belong to the extreme realm of AGN feedback, similar to those in MACS J0947.2+7623 (z = 0.354; Cavagnolo et al. 2011) and MACS J1931.8-2634 (z = 0.352; Ehlert et al. 2011), although not as powerful as those in MS 0735.6+7421 (z = 0.216; McNamara et al. 2005).

The right panel of Figure 6 reveals hints of a large northern depression, adjacent to the faint northern shell of the western cavity. This could be due to the western cavity leaking out into the ICM. The leaked relativistic particles could then contribute to the heating (e.g., Guo & Oh 2008), as well as account for the radio mini-halo. Such a phenomenon has been seen in M84 (Finoguenov et al. 2008), as well as in A2052 for the outermost shells of the southern and northern X-ray cavities. The latter appear to be breaking apart and radio plasma is seen to leak out, but only slightly beyond the cavities (Blanton et al. 2011). It is therefore difficult to explain why the leakage in RX J1532.9+3021 extends to such a large scale. Due to the presence of the radio mini-halo, it is also not clear from the radio observations if leakage is occurring.

As mentioned in Section 4, this potential large northern depression could also represent a single older X-ray cavity, or it could represent the accumulation of past western outflows along the northern direction. The surface brightness decrement associated with the depression is on the order of 10%–15% (see Figure 7), whereas the two clear eastern and western cavities have a decrement of 30%. The exact shape of the depression is therefore less well defined. Taking this into account, and considering that projection effects may be important, we assign a radius of 15 ± 7 kpc to the northern depression, where the large uncertainty attempts to account for the unknown factors. Considering that it has roughly the same size as the western cavity, then it is unlikely that the depression represents the accumulation of past western outflows since one would expect it to be significantly larger. If the northern cavity is largely elongated along our line of sight, then the volume could be significantly larger, but this remains unclear. Considering the more likely scenario that it represents one single older outburst, then the power associated with it is ≈5–20 × 10⁴⁴ erg s⁻¹. However, it is important to note that the projected distance from the central AGN to this northern depression is the same as for the western X-ray cavity. Although strong projection effects can account for this, if it were an older X-ray cavity, one would expect it to be located beyond the western X-ray cavity.

Interestingly, the southern optical emission filaments extend beyond the eastern cavity, out to 50 kpc in radius (Figure 9). Some studies suggest that filaments form from cooling of the hot ICM (e.g., McDonald et al. 2010), while others argue that some filaments originate from cool gas being dragged out in the wake of a cavity (e.g., Churazov et al. 2013). The southern filaments may therefore indicate that the past outbursts were aligned along the north to south direction. Even more, the 325 MHz radio map shown in the right panel of Figure 2 shows a diffuse north to south jet-like feature not clearly seen at 610 MHz or 1.4 GHz, consistent with particle aging.

Hence, based on the existence of these southern optical filaments, as well as the north to south 325 MHz jet-like structure, we conclude that RX J1532.9+3021 most likely harbors older AGN-driven outflows along the north to south direction. This implies that either sloshing effects are important and have caused the outflows to change directions, or that the jet is precessing (e.g., Dunn et al. 2006; Sternberg & Soker 2008; Canning et al. 2013). Furthermore, the potential northern X-ray depression lies along this direction, and may therefore represent the wake of the northern older outflow no longer seen at high-frequency radio wavelengths, known as a ghost cavity. Ghost cavities have been seen in other systems, although at much lower redshifts.

If the jet is precessing, then the difference in angle between the location of the current outburst and that of the older outburst is ≈70° for the north/west and ≈50° for the south/east (Figure 14). Assuming that the cavities are rising in the plane of the sky, which is probably not the case, we estimate from the difference in buoyancy rise time that the jet is precessing at ≈20° per 10⁷ yr. This is significantly larger than what Canning et al. (2013) find for A3581, but less than what Dunn et al. (2006) find for the Perseus Cluster. For the latter, the authors find a best-fitting solution consisting of a precession axis inclined by 120° with respect to the line of sight, with an opening angle of 50° and a precession timescale of ≈3 × 10⁷ yr.

Assuming that the X-ray cavities were powered by the release of gravitational binding energy from accretion onto the central black hole, we can estimate the minimum amount of mass the black hole must have accreted to power the outburst with the following equation:

$$\begin{equation} \Delta M_\mathrm{BH}= \frac{E_\mathrm{cavities}}{\eta c^2}. \end{equation} \tag{ 7 }$$

Here, the accretion efficiency (η) is assumed to be equal to 0.1. We find a total mass gain of 2 × 10⁷ M_☉ if we only consider the two clear X-ray cavities, but this amount goes up to ≈0.5 × 10⁸ M_☉ if we also consider the potential ghost cavities. For the two inner X-ray cavities, this translate to an average accretion rate of 0.4 M_☉ yr⁻¹ over the buoyancy rise time of the cavities.

There are no mass estimates of the black hole available in the literature. The BCG is included only in the Two Micron All Sky Survey point source catalog, but the quoted magnitude significantly underestimates the total flux since the PSF extends only to ≈2'' (≈10 kpc) and the total envelope of the BCG extends to some 50 kpc in radius based on the near-infrared HST observations. There are also no hyperleda stellar velocity dispersion measurements. For the purposes of this study, we therefore assume that the black hole mass lies within 10^9–10 M_☉ based on the known black hole masses of other BCGs from modeling (e.g., McConnell et al. 2011). Assuming that the Eddington luminosity is defined as

$$\begin{equation} \frac{L_\mathrm{Edd}}{\rm {{\rm \; erg\rm \, s^{-1}}}} = 1.26 \times 10^{47} \left(\frac{M_\mathrm{BH}}{10^9\,M_{\odot }}\right) \, \end{equation} \tag{ 8 }$$

for a fully ionized plasma, the mechanical power emerging from RX J1532.9+3021 in the form of the two clear X-ray cavities corresponds to 2% of the Eddington luminosity for a 10⁹ M_☉ black hole and 0.2% of the Eddington luminosity for a 10¹⁰ M_☉ black hole.

Such a high Eddington fraction provides some constraints on the possible fuelling mechanisms of the black hole. Since BCGs are embedded in hot atmospheres, one appealing solution is fuelling from the surrounding hot halo (e.g., Allen et al. 2006). This type of accretion, known as Bondi accretion (Bondi 1952), involves a spherical inflow of the hot X-ray emitting gas. For a given atmosphere temperature of T and density of n_e, assuming an adiabatic index of Γ = 5/3, it is given by the following equation:

$$\begin{equation} \frac{\dot{M}_\mathrm{B}}{M_{\odot } \,{\mathrm{yr}}^{-1}} = 0.012 \left(\frac{k_{B}T}{\mathrm{keV}}\right)^{-3/2} \left(\frac{n_e}{\mathrm{cm^{-3}}}\right) \left(\frac{M_\mathrm{BH}}{10^9\,M_{\odot }}\right)^2. \end{equation} \tag{ 9 }$$

To compute the true Bondi accretion rate, we would need to resolve the Bondi radius which is on the order of hundreds of parsecs in RX J1532.9+3021. This is not possible with current X-ray observatories. We therefore only compute a rough estimate of the Bondi accretion rate by considering the temperature and density at the Bondi radius found by Russell et al. (2013) for the most powerful outflow in their subsample of 13 systems with X-ray cavities where they were able to obtain a resolution close to the Bondi radius, NGC 507 (P_cavities ≈ 2 × 10⁴³ erg s⁻¹). The latter has an approximate temperature at the Bondi radius of ≈0.5 keV and an electron of density 0.8 cm⁻³. This corresponds to a Bondi accretion rate of 0.03 M_☉ yr⁻¹ for a 10⁹ M_☉ black hole or 3.0 M_☉ yr⁻¹ for a 10¹⁰ M_☉. If we consider the temperature and density values at the Bondi radius of Allen et al. (2006) for NGC 507, then the accretion rate increases by a factor of two. It is therefore currently difficult to explain the powerful cavities in RX J1532.9+3021 through Bondi accretion for a 10⁹ M_☉ black hole, although the density within the Bondi radius may be even larger than that in NGC 507 considering the extreme cool core properties of the cluster. Interestingly, the BCG in RX J1532.9+3021 has one of the most massive molecular gas detections known Edge (2001; 2.5 × 10¹¹ M_☉). The extensive molecular reservoir surrounding the BCG is therefore significantly larger than the mass required to power the outbursts, making cold accretion also a viable solution to the fueling mechanism of the AGN.

In Section 6.1, we derived a 3σ upper limit of 5 × 10⁴² erg s⁻¹ for the 2–10 keV nuclear luminosity of the central AGN. Radiatively inefficient AGN, with luminosities of <1% Eddington, often lack the optical/UV big blue bump normally seen in more radiatively efficient AGN. The value for the bolometric conversion factor is also not well known for such low-luminosity AGN. Vasudevan & Fabian (2007) find a typical correction factor of 10 for radiatively inefficient AGN, whereas Merloni & Heinz (2007) and Russell et al. (2013) both adopt a correction factor of five for the AGN in BCGs. Here we adopt a factor of 10, but stress that if we were to use a factor of five then our following conclusions would be stronger. Applying this bolometric correction, we find that the radiative output from the central AGN is less than 0.04% Eddington (M_BH = 10⁹ M_☉) or 0.004% Eddington (M_BH = 10¹⁰ M_☉).

Many authors have argued that black holes typically transit between two major states (e.g., Gallo et al. 2003). The first, known as the low-hard state, is characterized by a radiatively inefficient black hole that is accreting at low rates (<1% ${\rm \dot{M}_{\rm Edd}}$), yet capable of driving powerful jetted outflows. The second, known as the high-soft state, is a thermal state characterized by an efficiently accreting black hole (>1% ${\rm \dot{M}_{\rm Edd}}$) in which the jet is quenched.

These two states are illustrated in Figure 16, adopted from Churazov et al. (2005). We stress that Figure 16 is a sketch, and therefore illustrates only qualitatively the properties of the two states. We highlight the approximate position of RX J1532.9+3021 in the figure, by computing the total accretion rate ($\dot{M}$) as the sum required to power the X-ray cavities (for the two clear X-ray cavities) and the bolometric radiative luminosity. We then divide this quantity by the Eddington luminosity for a 10⁹ M_☉ black hole (blue) and a 10¹⁰ M_☉ black hole (red). Although this figure is approximate, it shows nicely that a 10⁹ M_☉ black hole would need to be accreting at very high rates to explain the powerful jets, so much that, the power emerging from the central AGN in terms of radiation is expected to be on the same level as the mechanical power. This is clearly not seen. If the central black hole has a mass of 10¹⁰ M_☉, then the radiative-inefficiency of the central AGN can be easily explained by considering that the AGN is in the low-hard state.

Zoom In Zoom Out Reset image size

Finally, we mention the possibility that the central AGN may be highly obscured. The power then needs to re-emerge in the infrared. Interestingly, RX J1532.9+3021 has the fourth-most infrared luminous BCG in the 62 BCG sample of Quillen et al. (2008, Ł_IR = 22.6 × 10⁴⁴ erg s⁻¹). Furthermore, Gandhi et al. (2009) argued that the intrinsic 2–10 keV X-ray luminosity of an AGN (corrected for absorption) is correlated with its mid-infrared luminosity at 12 μm. If this correlation still holds for BCGs, then by determining a rough estimate of the 12 μm luminosity from the 8 μm and 24 μm measurements of Quillen et al. (2008, ≈10⁴⁴ erg s⁻¹), the corresponding intrinsic 2–10 keV X-ray luminosity should be ≈10⁴⁴ erg s⁻¹. Since we only measure an upper limit of 5 × 10⁴² erg s⁻¹, this would imply that the source is highly obscured. To obtain such a strong attenuation (factor of >20), the absorbing column density would need to be around 10²⁴ cm⁻². If the true luminosity is even higher (≈5 × 10⁴⁴ erg s⁻¹), then the absorbing column density would need to be several 10²⁴ cm⁻², making the source Compton-thick. However, we note that the [O iii] λ5007/Hβ line ratio, based on the line measurements of Crawford et al. (1999) using an aperture of ≈10 kpc, is very low (0.5–0.6). The infrared emission arising from the BCG is therefore more likely dominated by star formation than an AGN (e.g., Quillen et al. 2008; O'Dea et al. 2008).

O'Dea et al. (2008) estimate that the SFR, based on the Spitzer infrared data is ≈100 M_☉ yr⁻¹. We find a consistent UV SFR of 76 ± 38 M_☉ yr⁻¹. The heating from the outflows most likely proceeds in a volume averaged way and is sufficient to offset the cooling on average. However, locally, some cooling in the form of the filaments may occur due local instabilities. In the case of MACS J1931.8-2634, star formation is also occurring even though the cavity power exceeds the X-ray cooling luminosity. This disagrees with the empirical star formation conditions found in Rafferty et al. (2008), suggesting that star formation only occurs in the cavity power is less than the cooling luminosity.

Note that many of the calculations in this section rely on strong assumptions, such as the accretion efficiency η. The latter is a strong function of the inner boundary conditions of the accretion flow and can vary from 0.057 for a non-spinning Schwarzschild black hole to 0.42 for a maximally spinning Kerr black hole (Novikov & Thorne 1973). A rapidly spinning black hole is therefore also an attractive solution to explain the high jet powers observed in RX J1532.9+3021 and the radiative-inefficiency of the central AGN (e.g., Tchekhovskoy et al. 2011).