Extended hard X-ray emission in highly obscured AGN

We investigated the X-ray morphological properties of the obscured AGN using the CIAO image analysis tools installed in SAOImage DS9⁴⁴4ds9; http://ds9.si.edu. Images were created in the 0.3-7.0 keV band (full band), the 0.3-3.0 keV band (soft band), the 3.0-7.0 keV band (hard band), and the 6.0-7.0 keV band (where the Fe K$\alpha$ could dominate) in the following analysis. We employed the sub-pixel binning technique to push for higher spatial resolution, which has been tested and frequently applied to imaging studies of extended emission and X-ray jets (e.g., Wang et al. 2011a, b, c; Paggi et al. 2012). We used a fine pixel size of 0.062$\arcsec$ (1/8 of the ACIS native pixel size) when producing the images.

To quantitatively measure the extent and amount of extended emission, we generated radial surface brightness profiles in different energy bands and different azimuthal sectors (when possible), following the procedure described in our previous work (e.g., Fabbiano et al. 2017, 2018a; Jones et al. 2020; Ma et al. 2020). We extracted radial profiles using concentric annuli out to a radius of 8$\arcsec$ for three sources and a radius of 17$\arcsec$ for two sources, reaching the background level. Off-nuclear point sources within the outer circle were all removed before generating the radial profiles. We started with an annular bin size of 0.5$\arcsec$ and increased the bin size at larger radii to maintain a minimum of 10 counts in each bin. To gauge the magnitude and significance of the extended emission, we compared the radial profiles to the Chandra PSFs for the corresponding energy bands. We modeled the PSF for each given centroid position and energy band using ChaRT⁵⁵5https://cxc.harvard.edu/ciao/PSFs/chart2/ and MARX 5.5.0⁶⁶6https://space.mit.edu/cxc/marx following the CIAO PSF simulation thread⁷⁷7https://cxc.cfa.harvard.edu/ciao/threads/psf.html. The default AspectBlur parameter in simulate_psf⁸⁸8https://cxc.cfa.harvard.edu/ciao/ahelp/simulate_psf.html (blur = 0.07$\arcsec$) was adopted. The radial profiles show the background-subtracted⁹⁹9We used annular background regions centered around the targets at $\sim$25$\arcsec$(inner) to $\sim$40$\arcsec$(outer) in radius, avoiding regions with point sources. surface brightness distribution in units of counts per arcsecond² in each energy band. The PSF radial profiles were generated in the same energy bands and were normalized to the counts within the central 0.5$\arcsec$ radius bin.

Following Ma et al. (2020), we take several measures to quantify the amount and extent of the extended emission in each energy band. We first measured the excess counts over the Chandra PSF outside the 1.5$\arcsec$ radius circle to avoid potential contamination from a nuclear component. We also measured the total excess counts above the PSF, which includes all the extended emission that does not belong to the central point source. Along with the excess counts, we also measured the extended, excess fractions in each energy band. The extended fraction $>$ 1.5$\arcsec$ is defined as the ratio of the excess counts above the Chandra PSF in the 1.5$\arcsec$-8$\arcsec$ or 1.5$\arcsec$-17$\arcsec$ annular region to the background-subtracted, total counts within the 8$\arcsec$ or 17$\arcsec$ radius circle at the given energy band. We define the total excess fraction to be the ratio of the total excess counts above the PSF (including the 0.5$\arcsec$-1.5$\arcsec$ region) to the total net counts within the 8$\arcsec$ or 17$\arcsec$ radius circle at the given energy band. Table 2 lists the excess counts over the Chandra PSF with associated Poisson statistical errors (including the background error), the fraction of the extended emission in the 1.5$\arcsec$-8$\arcsec$ or 1.5$\arcsec$-17$\arcsec$ annular region, and the total excess fraction in each energy band. In cases where we were able to identify azimuthal sectors, e.g., ionization cones and cross-cones, we also measured excess counts in each sector (Table 3).

In addition to the excess counts and excess fractions, we also estimated the fluxes and luminosities of the extended emission (Table 4). We converted the 0.3-3.0 keV and 3.0-7.0 keV excess counts to fluxes and luminosities assuming an absorbed (Galactic absorption) power-law model with a photon index $\Gamma$ of 1.9 for all the sources. Variations on the assumed $\Gamma$ from 1.4 to 2.4 change the estimated fluxes/luminosities by less than 20%.

The above-mentioned measurements are based on the simulated Chandra PSF. Potential caveats due to the uncertainties of the simulated PSF are discussed in the Appendix.

We discuss the morphology, extended component, and radial profiles of individual sources in the following section.

NGC 678 is classified as an edge-on, barred spiral galaxy [SB(s)b] at $z$ = 0.00946 (NED; $D$ $\sim$ 42 Mpc; 1$\arcsec$ $\sim$ 200 pc) with a column density of log $N_{\rm H}$ = 23.40 ${}^{+0.59}_{-0.50}$ cm^-2 (Ricci et al., 2017).

Figure 1 shows the Chandra images of NGC 678 at the 0.3-3.0 keV and 3.0-7.0 keV bands. The soft X-ray emission appears to be elongated more in the NE-SW direction while the hard band image shows the opposite. An off-center X-ray source is detected $\sim$6.5$\arcsec$ ($\sim$1.3 kpc at the galaxy distance) southwest of the NGC 678 nucleus with an estimated 0.3-7 keV flux of 4.9 $\times$ 10^-14 erg s^-1 cm^-2 and a luminosity of 9.7 $\times$ 10³⁹ erg s^-1 if associated with NGC 678, which could be an ultraluminous X-ray source (Swartz et al., 2004). However, we cannot completely rule out the possibility that this source is part of an ionization cone structure. This off-center X-ray source is excluded from our following analysis of the extended emission.

As shown in Figure 2, we generate the radial profiles in the full, soft, and hard bands. The soft X-ray emission shows total excess counts well detected at 6.6$\sigma$ above the PSF out to the outer radius (Table 2). The 3.0-7.0 keV band radial profile shows excess emission out to at least 5.5$\arcsec$ ($\sim$ 1.1 kpc) with a total of 49.5 $\pm$ 11.1 excess counts detected at 4.4$\sigma$, although the excess counts beyond 1.5$\arcsec$ are not significant (2.0$\sigma$). The total excess fraction in the 3.0-7.0 keV hard band is about 20%. We also checked the 6.0-7.0 keV band, and the total excess counts in this band are detected above 3$\sigma$ although most of the excess counts are in the inner 0.5$\arcsec$-1.5$\arcsec$ region (out to $\sim$ 300 pc).

IC 1657 is a nearly edge-on ($i$ = 78^∘) barred spiral [SB(s)bc] galaxy at $z$ = 0.01195 (NED; $D$ $\sim$ 53 Mpc; 1$\arcsec$ $\sim$ 252 pc) hosting a Seyfert 2 nucleus (Véron-Cetty & Véron, 2006) with log$N_{\rm H}$ = 23.40${}^{+0.13}_{-0.09}$ cm^-2 (Ricci et al., 2017). IC 1657 has been studied using optical IFUs (Dopita et al., 2015; López-Cobá et al., 2020). Optical emission line maps reveal a fan-like extended narrow line region emerging mainly from one side (E) perpendicular to the highly inclined N-S galaxy disk. An ionization cone with an opening angle close to 90^∘ is visible as revealed by strong [N ii] emission (Figure 6 in López-Cobá et al. 2020), extending at least $\sim$11$\arcsec$ ($\sim$ 2.8 kpc) from the galaxy disk plane. The galaxy disk itself is dominated by H ii regions as revealed by the H$\alpha$ emission, while ionized clumps present slightly stronger [O iii] emission at the outskirts of the disk (López-Cobá et al., 2020).

Figure 3 shows that IC 1657 has a relatively low surface brightness core in the soft band, and the soft X-ray emission is elongated and distributed nearly vertically along the N-S direction, which aligns with the host galaxy disk. There is barely any emission in the perpendicular direction. The excess emission in the soft band is well detected (Table 2). The hard band emission does not seem to follow the trend in the soft X-rays. IC 1657 has a total excess counts of 57.7 $\pm$ 12.6 (4.6$\sigma$) in the hard band with about 19% of the emission in the extended component, but most of the excess emission is found in the inner region out to 1.5$\arcsec$ ($\sim$ 378 pc) (Figure 4).

Given the strong azimuthal dependence of the soft X-ray emission, we further divided the region into two azimuthal sectors to examine excess emission in these two sectors separately, the soft X-ray elongation sector and the sector perpendicular to it (cross sector). The extended soft X-ray emission is not detected in the cross sector (Table 3). The hard band turns out to have marginally more excess counts in the cross sector with 33.5 $\pm$ 8.8 counts at 3.8$\sigma$ than in the soft X-ray elongation direction with 24.9 $\pm$ 8.7 counts at 2.9$\sigma$, which is contrary to the soft X-ray morphology.

In most heavily obscured AGN, an ionization cone is present in soft X-rays whenever an optical ionization cone is visible and they normally align with each other, but this is not the case in IC 1657. The optical ionization cone appears perpendicular to the soft X-ray extent, which aligns with the H ii dominated galaxy disk, and [O iii] is found mostly at the periphery of the star formation region. Instead, as shown earlier, it is the hard X-ray emission that is found perpendicular to, rather than along, the host galaxy plane. We will discuss possible origin of the cone structure and the extended hard X-ray emission in Section 5.

NGC 5899 is a barred spiral galaxy [SAB(rs)c] (Ann et al., 2015) at $z$ = 0.00864 (NED; $\sim$ 38 Mpc; 1$\arcsec$ $\sim$ 183 pc) and hosts a Seyfert 2 nucleus Véron-Cetty & Véron (2006) with log $N_{\rm H}$ = 23.03${}^{+0.04}_{-0.03}$ cm^-2 (Ricci et al. 2017; see also Baloković 2017).

As shown in the Chandra images (Figure 5) and radial profiles (Figure 6), NGC 5899 is the only AGN among this sample that does not show extended soft X-ray emission. In the 3.0-7.0 keV band, there are some excess counts detected at 3.9$\sigma$ above the PSF with a total excess fraction of $\sim$ 8% (Table 2), but beyond 1.5$\arcsec$ ($\sim$ 275 pc) the radial profile is basically consistent with the PSF (i.e., no excess emission).

NGC 454E is an early-type galaxy at $z$ = 0.01213 (NED; $D$ $\sim$ 54 Mpc; 1$\arcsec$ $\sim$ 255 pc) in a pair of interacting galaxies (Johansson, 1988; Stiavelli et al., 1998) with log$N_{\rm H}$ = 23.30${}^{+0.04}_{-0.03}$ cm^-2 (Ricci et al., 2017). NGC 454E is also identified as a new member of the class of changing-look AGN’ ($N_{\rm H}$ varying from $\sim$ 1 $\times$ 10²⁴ cm^-2 to $\sim$ 1 $\times$ 10²³ cm^-2; Marchese et al. 2012; Baloković 2017), i.e., AGN that show significant variation of the absorbing column density along the line of sight (Matt et al., 2003).

The soft X-ray emission in NGC 454E shows some excess counts above the PSF mostly in the inner 0.5-1.0$\arcsec$ region (Figures 7 and 8). In the hard band, there are also some excess counts in the inner 0.5-1.0$\arcsec$ region but they are not statistically significant, and extended emission beyond 1.5$\arcsec$ is not detected (Table 2).

ESO 234-G050 appears to be a low surface brightness spiral galaxy¹⁰¹⁰10ESO 234-G050 was classified as a blue compact dwarf (BCD) elliptical galaxy by Aguero (1993). ($z$ = 0.00877; $D$ $\sim$ 39 Mpc; 1$\arcsec$ $\sim$ 185 pc; NED) in the DECam Legacy Survey imaging (Dey et al., 2019). It hosts a Seyfert 2 nucleus (Aguero, 1993; Onori et al., 2017) with log $N_{\rm H}$ = 23.08${}^{+0.12}_{-0.18}$ cm^-2 (Ricci et al., 2017).

ESO 234-G050 exhibits prominent extended soft X-ray emission preferentially distributed along the NW-SE direction as shown in Figure 9, which resembles an ionization bicone structure. The hard X-ray band has a concentrated surface brightness distribution in the center (Figures 9 and 10), but no extended emission is detected above 3$\sigma$. Since the soft X-ray emission has a strong azimuthal dependence, we split the data into two azimuthal sectors, one in the NW-SE direction along the soft X-ray elongation and one in the NE-SW direction, to examine excess emission in each sector separately. ESO 234-G050 does not have enough counts for us to produce radial profiles for each azimuthal sector though. The extended soft X-ray emission in the soft X-ray elongation sector is well detected as expected. There is also some excess emission detected in the cross sector, mostly in the inner 0.5$\arcsec$-1.5$\arcsec$ region. For the hard band, there are slightly more excess counts in the soft X-ray elongation sector than in the cross sector. However, none of them is detected above 3$\sigma$.

Could the extended hard X-ray emission be explained by the X-ray binary (XRB) population, which has a fairly hard spectrum? The expected X-ray luminosities from XRBs were estimated by using scaling laws depending upon the dominant XRB populations. In most cases, the X-ray emission is located within the central bulge of the galaxy, where the low-mass XRBs (LMXBs) dominate. Here we used the $L_{\rm X}$-$L_{\rm K}$ scaling relation for LMXBs from Boroson et al. (2011). Since the X-ray emission only occupies the central region rather than the entire host galaxy, we also corrected the K-band luminosities for the region where X-ray extent is seen (8$\arcsec$ or 17$\arcsec$ circle defined in Section 3), using images from the 2MASS survey (Jarrett et al., 2003). For three sources (NGC 424, NGC 1125, and NGC 3281) where the X-ray emitting region encloses both the bulge and spiral arms where star formation occurs, we also estimated the expected X-ray emission from high-mass XRBs (HMXBs). We used the scaling relation between $L_{\rm X}$ and SFR for HMXBs in Mineo et al. (2012, 2014). The expected HMXB $L_{\rm X}$ was not corrected for the X-ray emitting region due to complex distribution of star forming regions. These $L_{\rm X}^{\rm HMXB}$ estimates can then be considered as upper limits. Whenever both the SFR and stellar mass are available, we also derived the expected $L_{\rm X}$ from the relation with both SFR and stellar mass (Lehmer et al., 2019).

To facilitate comparison with the measured X-ray luminosities in Table 4, we list the expected X-ray luminosities in the 0.3-3.0 keV and 3.0-7.0 keV bands separately in Table 5, using a typical soft-to-hard ratio of $\sim$1.3 according to the spectral modeling of XRBs (e.g., Boroson et al. 2011). Figure 12 shows the comparison between the measured extended X-ray luminosities (Table 4) and the expected luminosities from XRBs (Table 5) for both the soft and hard bands.

In all of the sources, the measured extended hard X-ray luminosities are higher than the expected X-ray luminosities due to XRBs by factors of $\sim$ 2 - 25. The X-ray emission from XRBs could well explain some of the hard extents if not all. There is still excess hard X-ray emission that can not be explained by XRBs, suggesting a connection with the AGN. For the extended soft X-ray emission, the measured luminosities in two sources, ESO 005-G004 and 2MASXJ0025+6821, are consistent with the expected X-ray luminosities from XRBs. For the remaining 10 cases, the excess over the XRB prediction ranges from a factor of $\sim$ 1.5 to 25.

In well-studied CT AGN, which show spectacular ionization bicones and remarkable correspondence between high surface brightness X-ray features and the radio jet and optical line emission, it has been suggested that the extended hard emission is due to scattering in the ISM of photons escaping the nuclear region in the direction of the ionization cone (e.g., ESO 428-G014 (Fabbiano et al., 2017, 2018a, 2018b); IC 5063 (Travascio et al., 2021); see also Jones et al. 2020). The ALMA and SINFONI observations of CO(2-1) and molecular hydrogen ($H_{2}$) of ESO428-G014 have confirmed that the 3-6 keV continuum and Fe K$\alpha$ emission are due to scattering from dense ISM clouds (Feruglio et al., 2020).

The hard emission seen in the cross-cone may indicate a porous torus or the predicted hot cocoon around the nucleus due to jet-ISM interaction. XRBs could potentially provide an alternative explanation for the extended hard X-ray emission in the cross-cone direction. Previous well-studied CT AGN with deep Chandra observations show that the cross-cones account for about 1/3 of the total excess emission in the hard band (Fabbiano et al., 2017; Jones et al., 2021). If we assume the same fraction for this sample (sources in Table 4 that show clear cone-like structures) and also assume half of the expected XRB emission is located in the cross-cone region, then the estimated fractions of the 3-7 keV cross-cone excess emission that could be explained by the expected XRBs are below 30% for three of them and up to 70% for ESO 137-G034.

Since most of the extended hard emission cannot be explained by XRBs, we further explore the origin associated with the AGN and compare with other extended hard X-ray detected AGN in the following sections.

This sample of heavily obscured, though not CT, AGN was selected under the same criteria as the CT AGN sample in Ma et al. (2020) except for the lower absorbing column densities. We also utilize exactly the same metrics to quantify the amount and extent of the extended component. Now we have a larger sample of obscured AGN with systematically measured excess counts and excess fractions to investigate whether the measured quantities correlate with any physical parameters such as AGN bolometric luminosity $L_{\rm bol}$, black hole mass $M_{\rm BH}$, Eddington ratio $\lambda_{Edd}$, $N_{\rm H}$ etc., which may shed light on the origin of the extended hard X-ray emission.

In Ma et al. (2020), we found a moderate correlation between the total excess fraction and log$N_{\rm H}$ among the 9 CT AGN (also including ESO 428-G014 and NGC 7212), i.e., CT AGN with a higher log$N_{\rm H}$ tend to have a higher total excess fraction. Jones et al. (2021) presented extended hard X-ray emission of several CT AGN with accumulatively deep Chandra observations that enable studies of trends along the ionization cone direction versus the cross-cone direction. Jones et al. (2021) found a strong correlation in the ionization cone between the total (hard X-ray) excess fraction and log$N_{\rm H}$, while the trend is not observed in the cross-cone region.

Now that we have a larger sample, we add our new sources to Figure 11 to revisit this potential correlation. The orange squares are the five obscured AGN in this work. The gray circles are the seven CT AGN plus two individually selected CT AGN (ESO 428-G014 and NGC 7212) in Ma et al. (2020). The total excess fractions of these AGN are measured over all azimuthal angles. We also include in Figure 11 the CT AGN from Jones et al. (2021) with the excess fractions measured in the ionization cone direction. Our sample in this work greatly expands the log$N_{\rm H}$ parameter space, however, there is still a gap at log$N_{\rm H}$ $\sim$ 23.5 - 23.8 cm^-2. The previously observed moderate trend in CT AGN by Ma et al. (2020) or the strong correlation in the ionization cone by Jones et al. (2021) is diluted rather than strengthened by the new sources. We still need more heavily obscured AGN that cover a wider range of column densities to better probe this relation. The dilution of the trend could also suggest that CT AGN and non-CT AGN may have different relations, or there could be a hidden parameter that would separate this sample into subsamples in which the correlation might hold.

No correlations have been found between the total excess fraction and other AGN parameters, i.e., $L_{\rm bol}$, $M_{\rm BH}$, or $\lambda_{Edd}$. We also use the estimated fluxes or luminosities of the extended components instead of the total excess fraction, and we do not see trends either.

In addition, we should point out that the Chandra observations of the sources in our survey (the seven CT AGN in Ma et al. 2020 and the five obscured AGN in this work) are relatively shallow ($\sim$ 10 - 55 ks) compared to the previously individually selected CT AGN (ESO 428-G014 and the five CT AGN in Jones et al. 2021), which have accumulatively deep observations ($\sim$ 100 - 540 ks). Therefore, we should use caution when interpreting results from mixed shallow and deep observations.

Since the seven CT AGN in Ma et al. (2020) and the five obscured AGN in this work were systematically selected and observed under the same survey strategies, here we compare these two samples in terms of the detection rate, spatial extent, and total excess fraction of the extended hard X-ray component. In Ma et al. (2020), five out of the seven CT AGN show extended emission in the 3.0-7.0 keV band detected at $>$ 3$\sigma$ above the PSF. In this sample, we detected the extended hard X-ray emission in three out of the five AGN. There is no significant difference between the average total excess fractions or fluxes/luminosities of the two samples. However, the CT AGN sample, on average, has an apparently larger hard X-ray extent than this non-CT sample. As discussed in Ma et al. (2020) for CT AGN, the amount of nuclear obscuration may be connected to the dense molecular clouds, i.e., CT clouds, in the host galaxy, which provide the materials needed to scatter the X-ray photons and produce the extended hard X-ray component. The extent of the hard X-ray emission also requires that some of the CT clouds must be coming from much father out in the galaxy than just in the torus. This also explains the observed larger extent in the CT AGN than in the non-CT AGN. Nevertheless we must bear in mind that both samples are still small, and we should revisit this after obtaining large enough samples.

Although this sample presents a relatively smaller hard X-ray extent than the CT AGN sample, these non-CT but still heavily obscured AGN show hard X-ray emission extending to at least $\sim$250 pc in radius, which is beyond the traditional dusty torus in the AGN unified model (e.g., Urry & Padovani 1995; Ramos Almeida & Ricci 2017). Using a simple formula in Barvainis (1987), we estimated the outer torus radii for the five obscured AGN following Ma et al. (2020), and they are all within 60 pc. Also, the amounts of extended hard X-ray emission relative to the total emission of these heavily obscured AGN are not negligible. Therefore, we should take this component into account to improve torus modeling.

All the AGN in this sample have high quality, broad band NuSTAR spectra covering 3-79 keV, which will be presented in a future publication. In principle, the obscuring torus geometry such as the scale height, opening angle, inclination angle, and torus covering factor in clumpy torus models (Nenkova et al., 2008; Elitzur, 2008) can all be extracted from (e.g., NuSTAR) X-ray spectral modeling (e.g., Murphy & Yaqoob 2009; Baloković et al. 2018; Tanimoto et al. 2019). However, the existence of reprocessed emission on $>$100 pc scales is currently not accounted for by any published models. Ignoring this new component can lead to biased estimates of the torus covering factor and/or its average torus column density. It is possible to constrain the opening angle from Chandra imaging of the extended X-ray component, and we know the fraction of total emission that belongs to the extended component, which reduce the uncertainties in the remaining torus model parameters.

About two dozen CT AGN with prominent biconical narrow line regions have now been imaged with Chandra (Fabbiano & Elvis, 2022). Almost all of these have the bicone intersecting the host galaxy disk and interacting with the host ISM, including shocked radio jets and fluorescing molecular clouds, leading to complex X-ray emission regions (Fabbiano & Elvis, 2022). They are objects in which we directly see AGN Feedback in action.

The conical structure in IC 1657, however, emerges perpendicular to the host plane. What would naturally have been assumed to be an AGN ionization bicone in the soft X-rays turns out to be consistent with the prediction from X-ray binary scaling relations, first found by Fabbiano (1989, 2006). The extended soft X-ray emission has an X-ray luminosity of $L_{\rm X}$ $\sim$ 1.3 $\times$ 10⁴⁰ erg s^-1. The predicted full-band $L_{\rm X}$ is 1.3 $\times$ 10⁴⁰ erg s^-1 using the stellar mass and star formation rate (SFR) from López-Cobá et al. (2020) and the X-ray/SFR ratio in Lehmer et al. (2019) with $\sim$40% of $L_{\rm X}$ in the soft band. In contrast, the hard X-ray emission shows more extended emission in the perpendicular direction, in line with the optical cone. Deeper Chandra data is needed to reveal a potentially more complete hard X-ray morphology.

López-Cobá et al. (2020) conducted BPT mapping of IC 1657 using the MUSE IFU data and found the spaxels associated with the ionized cone fall in the region occupied by shock ionization according to the predicated line ratios from theoretical models. The line ratios at the cone region are more compatible with the SF-driven wind rather than AGN-driven wind according to Sharp & Bland-Hawthorn (2010). Therefore they conclude that shock ionization produced by a SF-driven outflow seems to be the most likely explanation for the ionized cone.

The origin of the extended hard X-ray emission along the perpendicular direction is elusive. One possibility is that the hard X-ray emission is still within the galaxy disk. There is then plenty of ISM gas for AGN photoionization, or a radio jet¹¹¹¹11Unger et al. (1989) observed IC 1657 with the VLA in a hybrid B-C configuration but do not show a map of the radio emission (unresolved). to shock against. There also could be molecular clouds for the nuclear photons to interact with, as is the most likely explanation for the extended hard X-ray emission in ESO 428-G014 (Fabbiano et al., 2017, 2018a). If instead the hard X-ray emission extends beyond the galaxy disk, as hinted by the current Chandra data, there would be little material to interact with. IC 1657 could then be an extragalactic analog of the Milky Way Fermi Bubble where the central super massive black hole probably released vast amounts of energy that powers high-energy jets in the past (Su et al., 2010). Simulations of jet-ISM interactions (e.g., Mukherjee et al. 2018) also describe outflows with hot gas of kT $\gg$ 1 keV that do not cool efficiently and might have a low surface density in the hard band. Again, deep Chandra data is required to confirm the extent and reveal a complete morphology of the hard X-ray emission.

The other sources lack multi-wavelength data that inhibit further interpretation. Nevertheless, this newly discovered, extended hard X-ray component opens up a new window to investigating how the supermassive black hole interacts with and impacts the host galaxy.