AGN−Host Interaction in IC 5063. I. Large-scale X-Ray Morphology and Spectral Analysis

The Chandra observations are merged using the CIAO tool reproject_obs, ¹⁰ adopting the deepest observation ObsID 21466 as a reference frame. To produce the most accurate merged data set to enable subpixel analysis with image pixel size 1/8 ACIS pixel (0492), we explored different ways of aligning the data from each observation.

2.1.1. Aligning Off-nuclear Point Sources

2.1.2. Aligning the Nuclear Peak Positions

2.1.3. Merging Subarray and Full-array Mode Observations

To obtain the final merged image, we combined the nuclear-aligned merged image and the full-array observation (7878) by aligning their off-nuclear point-like sources (Section 2.1.1). From this image, we estimated an $\mathrm{FWHM}={1.371}_{-0.024}^{+0.022}$ native pixels to the Gaussian model, lower than the one derived from the nuclear-aligned merged image, but consistent at ∼1.3σ.

Figure 1 shows both the reprojected images of each observation and the merged image obtained through the best alignment method. These are 1/8 subpixel images, adaptively smoothed with the dmimgadapt ²⁰ tool, adopting minimum and maximum smoothing logarithmic scales of 2 and 15 native pixels, respectively, a minimum number of 4 counts under the kernel, and 15 scales to use, at the energy band 0.3–7.0 keV. The black crosses mark the peak positions (at R.A. = 20:52:02.311, decl. = −57:04:07.623) used as a reference for the alignment of the subarray mode observations.

Zoom In Zoom Out Reset image size

The nuclear spectrum exhibits both strong hard (>3 keV) X-ray emission and a soft excess at lower energies. These characteristics are suggestive of direct nuclear coronal emission in a leaky absorber model (Reichert et al. 1985) with a high covering fraction of 99.2% ± 0.2%, plus a reflection component. Alternatively, the spectrum can also be fitted (albeit formally less well; Table 4) by the model used in previous work on IC 5063 by Vignali et al. (1997) and Tazaki et al. (2011). This model fits the hard part of the X-ray spectrum with a power-law and reflection PEXRAV component associated with an absorption model, and it fits the soft excess with a simple power law. Although both models require a reflection and two power-law components, in the leaky absorber model the two power laws represent a single partially absorbed continuum, while in the approach used by Vignali et al. and Tazaki et al. the two components are decoupled, with quite different photon indices and intrinsic absorptions. The covering factor and the high column density we found from the fit of the nuclear spectrum are consistent with the more recent work by Balokivić et al. (2018), where the torus geometry of the IC 5063 NuStar data was studied in detail.

At energies >3 keV we detect a 6.4 keV neutral Fe Kα line (14σ), with an $\mathrm{EW}\sim {178}_{-41}^{+55}\,\mathrm{eV}$ consistent with both Vignali et al. (1997) and Tazaki et al. (2011) for IC 5063. This low EW is typical for Seyfert 1 galaxies (<0.5 keV), while Seyfert 2 galaxies usually exhibit Fe Kα EW ∼ 0.5–2 keV (Singh et al. 2011; Shu et al. 2011). However, some Seyfert 2 galaxies have similarly low EW to IC 5063, e.g., Mrk 348 (EW ∼ 40 eV; Singh et al. 2011) and Mrk 477 (EW ∼ 100–300 eV; Hernández-García et al. 2015). These low EWs may be indicative of an "unobscured" Seyfert 2 galaxy (see, e.g., Brightman & Nandra 2008; Bianchi et al. 2012), in which, although lacking the broad-line region, the hard X-ray emission appears unabsorbed. In our case Inglis et al. (1993) detected broad emission lines in polarized flux, thus suggesting a more complex structure of the nucleus in which the broad-line region is obscured, while the primary X-ray emission can escape unattenuated. If, instead of calculating the EW from the total continuum, we measure it with respect to the PEXMON reflected continuum component (see Table 4), we obtain $\mathrm{EW}\sim {1.96}_{-0.62}^{+1.23}\,\mathrm{keV}$, which is consistent with the standard scenario of an emission line reflected from circumnuclear material (Smith et al. 1993; Singh et al. 2011).

The most prominent feature below 3 keV is at ∼1.8 keV. We associate this feature with a blend of the Mg xii (1.745 keV; ∼5σ) and Si xiii (1.865 keV; ∼2σ) transitions, which are an indication of photoionized emission (Koss et al. 2015). Liu et al. (2019) suggested that the Si xiii line can also be related to outflowing hot gas.

Multicomponent fits to photoionization and thermal emission (Table 4) show that the nucleus is dominated by a low-photoionization ($\mathrm{log}U\sim -1.6$), high column density ($\mathrm{log}[{N}_{{\rm{H}}}/{\mathrm{cm}}^{2}]\sim 23.5$) gas component, which allows us to reproduce prominent emission lines at ∼1.7–1.9 keV and ∼2.3 keV. A second, high-photoionization component ($\mathrm{log}U\sim 1.5$), with similarly high column density absorption, can model the soft excess continuum. In Figure 5 we report the largest and lowest photoionization parameters used in the same model to fit the spectra of different regions of other Seyfert galaxies. The ionization parameters we estimate for the nuclear region of IC 5063 are compatible with those reported in literature. Alternatively, a collisional component with temperature kT ∼ 1.30 keV and a shock model with kT ∼ 2.87 keV are equally acceptable (see Table 4).

Zoom In Zoom Out Reset image size

The nuclear region used for the extraction of the spectrum (Section 4) encloses the unresolved nucleus (which dominates the hard emission), the inner regions of the [O iii]/Hα ionization bicone (Colina et al. 1991), and most of the region of interaction with the radio jets (Morganti et al. 1998). This could explain the complexity of the spectral parameters. In particular, the two temperatures may be consistent with shock velocities of ∼800 and ∼1200 km s⁻¹, respectively (see Table 2), assuming ${v}_{\mathrm{shock}}=\sqrt{16{kT}/3\mu }$ (Wang et al. 2014). Both the collisionally and shock-ionized models predict ISM densities n_e ∼ 0.1 cm⁻³ (assuming a filling factor η = 1), which are an order of magnitude less than the ISM density estimated in the nucleus of ESO 428-G014 (n_e ∼ 4 cm⁻³; Fabbiano et al. 2018a, Table 9) and similar to those in the outer optical arcs in Mrk 573 (n_e ∼ 0.14 cm⁻³; Paggi et al. 2012). This result is puzzling, as these low densities are inconsistent with those expected in gas-rich disk galaxies. They may be explained as an overestimate of the filling factor due to clumpiness in the ionized gas, or to our seeing the emission coming from a thin skin of hot gas above the galaxy plane, possibly due to an AGN wind host–disk interaction (Maksym et al. 2019). A third possibility is that the low densities could be real and due to strong AGN winds evacuating the ISM from the galaxy in these locations. Table 2 lists how the physical parameters depend on the filling factor η. In the table we give values for η = 1. Of note are the cooling times of ∼10⁷ yr, which could indicate a transient phenomenon as also indicated in Mukherjee et al. (2018) simulations, which found a cooling time lower than the dynamical time in the core ($\mathrm{log}\,{t}_{\mathrm{cool}}/{t}_{\mathrm{dyn}}\leqslant -2$).

Table 2. Physical Parameters of the Collisionally and Shock-ionized Gas in All the Regions

Region	Model [tot] a	V η	L0.3–10 keV	ne η−1/2	Eth η1/2	Pth η−1/2	tcool η1/2	M η1/2	vshock
		(1065 cm3)	(1039 erg s−1)	(cm3)	(1055 erg)	(10−12 dyne cm−2)	(107 yr)	(106 M⊙)	(km s−1)
Nucleus	APEC [AC]	0.14	3.2	${0.11}_{-0.06}^{+0.07}$	${0.5}_{-0.2}^{+0.4}$	248 ± 114	${0.5}_{-0.2}^{+0.4}$	${1.2}_{-0.7}^{+0.8}$	${779}_{-301}^{+255}$
	PSHOCK [PC]	⋯	6.2	${0.12}_{-0.05}^{+0.06}$	${1.4}_{-0.3}^{+1.7}$	671 ± 162	${0.7}_{-0.2}^{+0.9}$	${1.4}_{-0.5}^{+0.7}$	${1200}_{-940}^{+1029}$
NW cone	APEC [AC]	4.10	1.4	${0.012}_{-0.006}^{+0.006}$	${1.4}_{-0.5}^{+1.1}$	22.9 ± 8.4	${3.1}_{-1.1}^{+2.4}$	${4.0}_{-2.0}^{+2.0}$	${712}_{-255}^{+368}$
SE cone	APEC [AC]	4.10	1.0	${0.011}_{-0.007}^{+0.009}$	${1.9}_{-0.9}^{+1.9}$	30.4 ± 14.9	${6.1}_{-3.0}^{+6.2}$	${4.0}_{-2.5}^{+3.1}$	${826}_{-309}^{+401}$
Cross-cone	PSHOCK [PP]	27.9	3.9	${0.018}_{-0.010}^{+0.008}$	${9.7}_{-3.0}^{+6.3}$	23.3 ± 7.1	${7.8}_{-2.4}^{+5.1}$	${41}_{-24}^{+19}$	${584}_{-309}^{+245}$
	PSHOCK [PP]	⋯	4.2	${0.006}_{-0.002}^{+0.002}$	${5.1}_{-1.5}^{+2.4}$	12.2 ± 3.6	${3.8}_{-1.1}^{+1.8}$	${14}_{-6}^{+4}$	${719}_{-235}^{+283}$
	PSHOCK [PC]	⋯	3.7	${0.005}_{-0.003}^{+0.003}$	${7.6}_{-2.7}^{+6.4}$	18.1 ± 6.3	${6.5}_{-2.2}^{+5.5}$	${13}_{-7}^{+7}$	${926}_{-413}^{+516}$
	PSHOCK [PA]	⋯	5.7	${0.006}_{-0.002}^{+0.002}$	${10}_{-2}^{+8.1}$	24.7 ± 4.7	${5.7}_{-1.0}^{+4.5}$	${14}_{-5}^{+5}$	${1044}_{-465}^{+665}$
	APEC [PA]	⋯	1.7	${0.005}_{-0.003}^{+0.004}$	${2.0}_{-0.5}^{+2.2}$	4.8 ± 1.3	${3.7}_{-1.0}^{+4.0}$	${11}_{-7}^{+9}$	${521}_{-317}^{+274}$
	APEC [AC]	⋯	2.2	${0.006}_{-0.003}^{+0.003}$	${6.2}_{-2.6}^{+3.8}$	14.8 ± 6.2	${8.8}_{-3.7}^{+5.4}$	${14}_{-8}^{+7}$	${786}_{-274}^{+255}$
	APEC [AA]	⋯	3.6	${0.008}_{-0.003}^{+0.003}$	${7.7}_{-2.3}^{+3.4}$	18.5 ± 5.6	${6.7}_{-2.0}^{+3.0}$	${18}_{-7}^{+7}$	${779}_{-213}^{+200}$
	APEC [AA]	⋯	3.6	${0.008}_{-0.005}^{+0.005}$	${2.0}_{-1.0}^{+1.8}$	4.7 ± 2.3	${1.7}_{-0.8}^{+1.6}$	${18}_{-12}^{+12}$	${388}_{-158}^{+200}$

Note.

^a "Model" represents the template from which we derived the values, and the complete configuration of the model to which it belongs is "tot," where A = APEC, C = CLOUDY, P = PSHOCK. We assumed a filling factor η = 1. We therefore show all the parameters as a function of the filling factor. Errors are quoted at 1σ significance.

Download table as:  ASCIITypeset image

Figures 2 and 3 clearly show the presence of X-ray emission stretching E–W along the bicone direction out to ∼2 kpc from the nucleus. This emission is particularly prominent in the soft band (<3 keV). Extended soft (<3 keV) X-ray emission has been observed in Seyfert 1.5–2 galaxies, spatially correlated with the [O iii] emission (Bianchi et al. 2006, 2010; Levenson et al. 2006), suggesting a common origin largely due to photoionization, and partially to collisional ionization, of circumnuclear clouds (Wang et al. 2011b; Matt et al. 2013). Prominent emission lines are found in the spectrum below 3.0 keV, typically observed in CT Seyfert (Wang et al. 2011c; Fabbiano et al. 2018a; Maksym et al. 2019; Jones et al. 2020). These lines include Ne x, Mg xi, Si xiii, Si xiii, S xv, and the Fe l complex. Given the ACIS spectral resolutions, these lines are typically blended in the observed spectra (Table 5 in Appendix B). The best-fit models of the bicone spectra need to include, at least, one photoionization component. Focusing on the best-fit models consisting of two photoionized phases, Figure 5 shows that the NW cone photoionization parameters put it at the periphery of the Seyfert 2 distribution, toward lower U_low. In this case, the presence of two quite different photoionization phases in the bicone direction could be justified by the X-shape morphology of the ionization cones (Colina et al. 1991), implying the coexistence of two net separated regions differently illuminated by the central AGN. In both cone regions the higher-photoionization component can be replaced by a collisionally ionized component with temperature kT ≈ 1.2 keV.

The resulting average ISM density is ∼0.01 cm⁻³ for both sides, where we are considering average density of thermal gas in a spherical shell with angle 60° and from ∼0.5 to ∼3.4 kpc. As in the nucleus, these densities are lower than those reported for other Seyfert 2 galaxies (Paggi et al. 2012; Fabbiano et al. 2018a; Maksym et al. 2019). This suggests more clumpiness of the hot gas in IC 5063. Since these densities are ∼1/10 of those observed in the nuclear region, the cooling times are correspondingly longer, ∼3.0 × 10⁸ yr (Table 2).

Figures 3 and 4 show that the bicone emission is present also at energies >3 keV. Harder extended components (in both the continuum emission above ∼3 keV and the neutral 6.4 keV Fe Kα line) have been detected with Chandra in AGNs (see Bauer et al. 2015; Fabbiano et al. 2017; Jones et al. 2020, 2021; Ma et al. 2020). The bicone spectra clearly show a roughly flat hard X-ray continuum in IC 5063, which needs to be fitted with a reflection component, whose normalization is weaker (by a factor 2) and photon index is steeper (Γ = 2.7) compared to that of the nuclear spectrum (Γ = 1.45). In agreement with Fabbiano et al. (2017), we suggest that the steeper hard X-ray continuum seen in extended regions implies that this is due to the scattered and/or fluorescent intrinsic emission escaping unattenuated in the bicone direction.

Table 3 gives both the percentages of the counts with respect to the nuclear region and the excess counts over the PSF, in the merged image and in the different sectors (as defined in Section 4) and energy bands. At energies <3 keV the percentage of counts in the bicone regions is significantly larger than the percentage predicted for the nuclear spillover (Appendix B.1). Of these excess counts, 66% are from within the bicone sectors, which represent 1/3 of the total external area.

A neutral iron emission line is observed in both the bicone spectra, and most of this feature is modeled by a reflection PEXMON component (e.g., George & Fabian 1991; Matt et al. 1991) with solar abundance, while a small (∼16%) contribution is reproduced with the photoionization CLOUDY model. However, in the energy band 6.1–6.6 keV, which is dominated by this Fe Kα emission, 60% (44/73) of the counts in the 2''–15'' annulus are in the NW cone. This excess of Fe Kα emission corresponds to the protrusion at a projected distance of 5'' (1.2 kpc) in Figure 4 at these energies. This protrusion of the neutral iron emission in the NW cone is well visible in the azimuthal surface brightness profile in Figure 6, showing a 2σ–3σ significance. The Fe Kα emission spatially correlated with the most intense radio hot spot suggests the presence of high-density clouds in this region. These clouds may fluoresce because of the interaction with AGN photons escaping in the jet direction. Alternatively, X-rays may also be produced locally by jet-induced shocks and interact with the clouds (see Section 5.5).

Zoom In Zoom Out Reset image size

Figure 3 and Table 3 show significant extended emission in the cross-cone region. This emission is more prominent at energies <1.5 keV and extends out to a radius of ∼3 kpc from the nucleus. Figure 7 shows the 0.3–1.5 keV image, adaptively smoothed with Gaussian kernels ranging from 0.5 to 30 image 1/8 of instrumental pixel in 30 iterations and 10 counts under the kernel. The color scale has been chosen to minimize the visual impact of the nuclear and bicone emission.

Zoom In Zoom Out Reset image size

Extended soft X-ray emission, perpendicular to the bicone axis, has been observed in other AGNs (Wang et al. 2011c; Paggi et al. 2012; Fabbiano et al. 2018a; Maksym et al. 2019; Jones et al. 2020, 2021). This emission is not expected from the classical AGN unification paradigm (Antonucci 1993). A possible explanation could be that the nuclear torus may be porous and allows part of the nuclear photoionizing continuum to escape (Nenkova et al. 2008). The volume of the cross-cone is 4 times the volume of the bicone, and photons with energies 0.3–1.5 keV in the cross-cone are 73% of those in the bicone. Therefore, following Fabbiano et al. (2018a), the transmission of the torus in the cross-cone direction is 20%, twice that found for ESO 428-G014. Alternatively, this emission may be due to hot outflows from the galactic disk of IC 5063, caused by the interaction of the jet with the ISM, as predicted by the simulations of Mukherjee et al. (2018). This scenario would be in agreement with the conclusions in Maksym et al. (2020b), Maksym et al. (2020a), and Venturi et al. (2021), which found suggestions for lateral outflows as a consequence of the radio jet−ISM interaction, detecting "dark rays" in Hubble Space Telescope (HST) near-infrared data perpendicular to the galaxy disk as a suggestion for the presence of large-scale dust, a low-ionization [S ii] emission loop, and high velocity dispersion of Hα and [O iii] emission lines in the cross-cone area, respectively. The best phenomenological model of the cross-cone spectrum (see Appendix B.3.3) consists of a steep Γ = 3 power law plus a flat reflection component (Γ < 1.2). The spectrum shows an excess at 0.9–1 keV, which could be due to blended Fe l emission lines (e.g., Fe xxi [1.009 keV]) and may also include Ne ix [0.915 keV] and/or Ne x [1.022 keV] lines. It does not exhibit the strong emission lines at ∼1.8 keV or ∼2.3 keV observed in the nuclear and bicone spectra, which can only be reproduced by photoionization models. In the cross-cone region, we find no evidence for a neutral iron line at 6.4 keV.

The cross-cone spectrum can be fitted with a mix of any two components among photoionization, collisional ionization, and shock ionization models. The principal photoionization component has $\mathrm{log}U\sim 0.8$ and very low column densities $\mathrm{log}({N}_{{\rm{H}}}/{\mathrm{cm}}^{2})\lt 19.7$. For the fit with two components of photoionized gas, the second phase has a higher ionization parameter $\mathrm{log}U\sim 1.7$ and $\mathrm{log}({N}_{{\rm{H}}}/{\mathrm{cm}}^{2})\lt 20.6$. Figure 5 shows that both the ionization parameters for the cross-cone spectrum are larger than those found by Paggi et al. (2012) for the cross-cone of Mrk 573. These high values may suggest that the emission is not due to photoionization from AGN photons escaping from a leaky torus but instead is prevalently thermal.

The temperatures we find for shock-ionized gas range from 0.5 to 3 keV, while the collisionally ionized gas temperatures are between 0.3 and 1.4 keV. The normalization values of the thermal models imply nominal densities of the emitting gas n_e ∼ 0.006 cm⁻³ (see Table 2), a factor ∼2 lower than those in the bicone region. Given the large volume, the range of masses (∼(1–4) × 10⁷ M_⊙) in the cross-cone is larger than what we observed in the bicone regions. Collisionally and/or shock-ionized gas would be consistent with the predictions of the Mukherjee et al. (2018) simulations, in which, in the later stage of the jet−ISM interaction, gas at T ∼ 10⁷ K would be swept away from the disk in the form of large-scale filamentary winds with velocities >500 km s⁻¹. The observed gas densities, however, are a factor of 10 lower than expected for perpendicular filaments in Mukherjee et al. (2018) simulations. Alternatively, if we assume that this thermal gas has a density n_e ∼ 0.1–0.3 cm⁻³, as predicted by Mukherjee et al. (2018) (see their Figures 2 and 4), we derive a filling factor η < 0.01. This is exactly consistent with the filling factor lower than 1% estimated by Oosterloo et al. (2017) in ALMA observations of molecular gas (CO transitions). Hence, we suggest that the filling factor is far from unity and that the hot and cold phases of the gas, in the inner jet-affected regions, exhibit a similar clumpiness. The cohabitation of the hot X-ray-emitting and molecular CO-emitting gas in the jet-disturbed circumnuclear regions is observed in other objects (e.g., Grossová et al. 2019; Feruglio et al. 2020). That these two different gas phases share the same clumpiness suggests that the radio jet impact has a similar fragmentation effect on these two phases. According to a typical scenario, the high pressure due to the jet compresses and accelerates fast, shock-driven, lateral outflows, increasing the density and temperature of the molecular gas (Oosterloo et al. 2017) and producing both soft X-ray photons by shock and hard X-ray photons by scattering by dense clouds (Fabbiano et al. 2017). However, the question remains whether this cold phase was already present in the circumnuclear medium, mixed with the hot-phase gas, before the impact with the radio jet, or whether it represents gas that, for some reason, has cooled down or has not warmed up. A careful spatial analysis comparing cold molecular gas and its hot phase in the jet-affected region is planned to get a more complete picture. In conclusion, emission from shocked and/or collisionally ionized gas is preferred over photoionized gas for the cross-cone region in IC 5063.

Fabbiano et al. (2018a) noticed that the extent of the large-scale kiloparsec-size emission of the obscured AGN ES0 428-G014 decreases with increasing photon energy, and they suggested that this effect may be related to a more central concentration of the denser molecular clouds responsible for the reflection and scattering of the higher-energy nuclear photons in the galaxy disk. More recently, Jones et al. (2021) have confirmed this effect for a sample of five CT AGNs studied with Chandra. We find a similar behavior in IC 5063, as shown by both Table 3 and Figure 8, which compares the large-scale extent of the emission of IC 5063 in different energy bands with that of ESO 428-G014, following Fabbiano et al. (2018a). Figure 8 shows that at energies <4 keV in both AGNs the extent is larger in the cone direction. The decrease in the extent of the X-ray emission, at energies <4 keV, along the cross-cone direction in IC 5063 is similar to that found for ESO 428-G014. The extent of the bicone regions in IC 5063 is a steeper function of energy than in ESO 428-G014.

Zoom In Zoom Out Reset image size

Jones et al. (2021) associate the energy dependence of the extent of the bicone with its inclination relative to the galactic disk. However, this angle is small (ΔPA ∼ few degrees) both for ESO 428-G014 (Riffel et al. 2006) and for IC 5063 (Colina et al. 1991), indicating that something else may contribute to this different energy dependence. Future investigations of a larger sample of CT AGNs/Seyfert 2 galaxies are needed to further probe this point.

At higher energies (>4 keV) the extents along the bicone and cross-cone direction in IC 5063 are similar, indicating symmetrical extent of the emission. Similar cases have been found in Jones et al. (2021). However, Table 3 shows that the emission along the cross-cone is less significant than that in the bicone.

The hard emission is likely due to reflection off molecular clouds, while the soft X-ray emission is likely from both thermal optically thin and AGN-photoionized low-density ISM. The possible differences in the radial energy dependence between IC 5063 and ESO 428-G014 probably then reflect differences in the molecular cloud distributions in the two galaxies.

Figure 3 shows that the radial surface brightness profile of the bicone region at energies <1.5 keV shows a noticeable "flattening" in the region within ∼3'' (i.e., ∼700 pc), a radius roughly consistent with the terminal hot spots of the radio jet in IC 5063 (Morganti et al. 2007). We speculate that this enhancement of X-ray surface brightness may be connected with the interaction of the radio jet with the ISM in the same region (e.g., Sutherland et al. 1993; Falcke et al. 1998; Gallimore et al. 2006; Wang et al. 2009; Mukherjee et al. 2018). The NW cone spectrum shows a ≈3σ broadband feature consistent with Fe xxv at ∼6.7 keV. Figure 9 shows a 1/8 subpixel image in the 6.5–6.8 keV energy band, smoothed with a kernel radius of 10 subpixels, as a proxy of the distribution of the ionized Fe xxv emission. The radio ATCA contours at 24 GHz (Morganti et al. 1998) are superimposed in green. This image shows a protrusion in the NW sector at a distance of 800 pc from the nucleus, which is spatially correlated with the 24 GHz radio hot spot. To estimate the significance of this extended ionized iron feature, we plotted the surface brightness azimuthal profile inside the 2''–35 annulus (purple in Figure 9) in the 6.5–6.8 keV image, with a bin size of 60°. The azimuthal profile is shown in Figure 10, in which the dashed red lines separate the different sectors and the dashed green line represents the average value estimated in all the sectors, excluding the NW cone. We find 2.4σ and 3.7σ significance of the surface brightness excess in the NW sector relative to the average value and to the background, respectively. This excess is consistent with the image in Figure 9. We do not detect a similar significant feature in the SW cross-cone sector. The Fe–Kα 6.7 keV line is typically associated with emission from collisionally (or shock) ionized gas (e.g., NGC 6240; Netzer et al. 2005; Feruglio et al. 2013; Wang et al. 2014; Fabbiano et al. 2020). The Fe xxv line has been assumed to have a nuclear origin in most works; however, it is observed to be extended out to 40 pc in NGC 4945 (Marinucci et al. 2017) and to 2.1 kpc in NGC 6240 (Netzer et al. 2005), in which it appears like a bridge connecting two merged nuclei (Fabbiano et al. 2020). In IC 5063 this feature is seen only in the NW cone region, where outflowing multiphase gas has been observed (Morganti et al. 2007; Tadhunter et al. 2014; Dasyra et al. 2015, 2016; Morganti et al. 2015; Oosterloo et al. 2017). We speculate that its origin, as well as that of the perturbed gas, may be related to the interaction of the radio jet with the ISM.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We first fitted the nuclear spectrum with a "leaky absorber" model applying a partial covering absorption model ²⁴ to a power law (XSPOWERLAW ²⁵ ): partcov(xsphabs) xspowerlaw.

Following previous works on obscured AGNs (Levenson et al. 2006; Fabbiano et al. 2018a), we added components to the model, guided by the ≳2σ contiguous residuals and F-test ²⁶ results. We first added a simple reflection (PEXRAV ²⁷ ) component with solar abundance, forcing it to have the same photon index of the power law. Then, we included the most prominent emission lines one at a time and left their energy and normalization free to vary, while fixing the width of the lines to the Chandra spectral resolution (∼100 eV). We then introduced additional emission lines to remove remnant contiguous significant residuals at ≳2σ.

The data, best-fit model, and residuals are shown in Figure 11. The model yields a power-law component and a reflection component with photon index Γ = 1.45 ± 0.10 and N_H ≃ (2.9 ± 0.1) × 10²³ cm⁻² with a covering fraction of 99.2% ± 0.2%, which is clearly required to get a good fit of the soft excess. These results are consistent with the values observed in Seyfert 2 galaxies (e.g., Cappi et al. 2006; Vasylenko et al. 2015). In addition, there are a total of five emission lines, including Fe–Kα emission with an equivalent width $(\mathrm{EW})\simeq {178}_{-41}^{+55}$ eV. Table 4 gives the best-fit parameters, as well as the energies and normalizations of the emission lines.

Zoom In Zoom Out Reset image size

Table 4. Best-fit Parameters and ${\chi }_{{\rm{R}}}^{2}$/dof of Empirical (top) and Physical (bottom) Best-fit Models Used to Fit IC 5063 Nuclear Spectrum (29242 Counts at 0.3–7.0 keV) in the 0.3–8.5 keV Energy Band

Best-fit Empirical Models
Model		χ2/dof	Γ a	normpl (photons cm−2 s−1)	${N}_{{\rm{H}}}^{\mathrm{pl}}\,(\times {10}^{22}{\mathrm{cm}}^{-2})$	normrefl (photons cm−2 s−1)	${N}_{{\rm{H}}}^{\mathrm{refl}}\,(\times {10}^{22}{\mathrm{cm}}^{-2})$
(A) Leaky absorber		0.960/421	1.45 ± 0.10	${2.54}_{-0.44}^{+0.45}\times {10}^{-3}$	b ${29.11}_{-1.16}^{+1.19}$	${1.58}_{-0.55}^{+0.61}\times {10}^{-3}$	⋯
(B) 2 pl + refl		0.932/414	1.7 c	${3.76}_{-0.22}^{+0.18}\times {10}^{-3}$	${34.98}_{-0.86}^{+0.92}$	${7.62}_{-1.41}^{+1.79}\times {10}^{-3}$	${3.11}_{-0.96}^{+1.11}$
			2.2 c	2.27 ± 0.16 × 10−5	⋯	⋯	⋯
Emission Lines
	Energy (keV)	Flux (10−6 photons cm−2 s−1)		Identified Emission Lines d
(A)	1.32 ± 0.02	0.87 ± 0.27		Mg xi [1.331 keV]
	1.79 ± 0.01	1.13 ± 0.23		blend Mg xii [1.745 keV] + Si xiii [1.865 keV]/Fe xxiv [1.778 keV]
	2.33 ± 0.03	0.65 ± 0.30		Si xiii [2.346 keV]
	2.48 ± 0.02	0.93 ± 0.29		S xv [2.461 keV]
	6.39 ± 0.01	21.1 ± 1.5		Fe–Kα [6.4 keV]
(B)	1.01 ± 0.02	0.93 ± 0.50		Fe xxi [1.009 keV]/Ne x [1.022 keV]
	1.33 ± 0.01	1.28 ± 0.28		Mg xi [1.331 keV]
	1.47 ± 0.02	0.47 ± 0.21		Mg xii [1.473 keV]
	1.76 ± 0.02	${1.23}_{-0.40}^{+0.28}$		Mg xii [1.745 keV]/Fe xxiv [1.778 keV]
	1.85 ± 0.03	${0.75}_{-0.31}^{+0.34}$		Si xiii [1.865 keV]
	2.33 ± 0.03	0.67 ± 0.29		Si xiii [2.346 keV]
	2.47 ± 0.02	0.80 ± 0.28		S xv [2.461 keV]
	6.39 ± 0.01	20.2 ± 1.9		Fe–Kα [6.4 keV]
Best-fit Photoionization Models
${\chi }_{{\rm{R}}}^{2}$/dof	Γ a	${N}_{{\rm{H}}}^{\mathrm{pl}}$ ( × 1022 cm−2)	CvrFract (%)	$\mathrm{log}\,U$	$\mathrm{log}\,({N}_{{\rm{H}}}/{\mathrm{cm}}^{-2})$	kT (keV)	EM (10−6 cm−5)
0.966/427	${1.41}_{-0.15}^{+0.12}$	${29.14}_{-1.68}^{+1.49}$	${99.7}_{-0.3}^{+0.2}$	$-{1.60}_{-0.08}^{+0.09}$	>23.5
				${1.50}_{-0.11}^{+0.09}$	${22.95}_{-0.15}^{+0.12}$
0.970/427	1.31 ± 0.11	${28.26}_{-1.26}^{+1.33}$	99.4 ± 0.2	$-{1.50}_{-0.09}^{+0.05}$	>23.5	${}^{\mathrm{th}}{1.21}_{-0.18}^{+0.13}$	${4.98}_{-1.84}^{+2.13}$
0.978/427	1.31 ± 0.11	${28.17}_{-0.50}^{+0.41}$	99.4 ± 0.4	$-{1.50}_{-0.05}^{+0.05}$	>23.3	${}^{\mathrm{sh}}{2.87}_{-1.76}^{+2.11}$	${6.49}_{-0.90}^{+1.73}$

Notes. For the phenomenological models we report energy, normalization, and identification of the emission lines we detect.

^a Same photon index for the power-law and reflection components. ^b Column density associated with a partial covering model with covering fraction 99.2% ± 0.2%. ^c Photon indices are fixed according to Vignali et al. (1997) and Tazaki et al. (2011). ^d Identification emission lines from atomdb.org database.

EM is the normalization of the collisional (APEC) and shock (PSHOCK) ionization model equivalent to $\tfrac{{10}^{-14}}{4\pi {[{D}_{A}(1+z)]}^{2}}\int {n}_{e}{n}_{{\rm{H}}}{dV}$, with D_A the angular distance and n_e and n_H the electron and hydrogen density, respectively; ^(sh/th) indicates temperature and normalization of the shock/thermal model.

Download table as:  ASCIITypeset image

A.1.1. Consistency with Previous Works on IC 5063

We estimated a total observed L_{2–10 keV} ∼ 4 × 10⁴² erg s⁻¹ similar to that of Tazaki et al. (2011) (L_{2–10 keV} ∼ 5.6 ×10⁴² erg s⁻¹), both in a ≈25-radius region, but one order of magnitude lower than that of Vignali et al. (1997) (2 ×10⁴³ erg s⁻¹). We fitted the soft excess power-law model used by both Vignali et al. (1997) and Tazaki et al. (2011) to the nuclear spectrum of IC 5063. This model consists of a power law plus a 6.4 keV Gaussian emission line, i.e., the neutral iron Kα emission, and a reflection plus a fixed power-law component with a photon index Γ = 1.7, as well as an associated intrinsic absorption (see Tazaki et al. 2011). For the Compton reflection component PEXRAV we fix solar abundance and the cosine of the inclination angle of the scattering disk at $\cos (\mathrm{Incl})=0.45$. The soft (<2 keV) X-ray excess is modeled with an unabsorbed Γ = 2.2 power law, as often found in the spectra of Seyfert 2 galaxies (e.g., Guainazzi & Bianchi 2007; Bianchi et al. 2009).

Figure 12 shows the data, best-fit model, and residuals, and Table 4 lists the best-fit parameters. By fixing the photon indices as used in Vignali et al. (1997) and Tazaki et al. (2011), we find a similar column density for the power law ((3.50 ± 0.09) ×10²³ cm⁻²) and reflection (${3.11}_{-0.96}^{+1.11}\times {10}^{22}\,{\mathrm{cm}}^{-2}$) components. This suggests that the shape of the nuclear spectrum of IC 5063 has remained unchanged from 1994 (Vignali et al. 1997) to 2009 (Tazaki et al. 2011) to the present.

Zoom In Zoom Out Reset image size

In conclusion, we use the leaky absorber as the best-fit model, as it is the simpler one and has only marginally worse χ² (Table 4).

We investigated the physical mechanisms responsible for the X-ray emission in the nucleus, by fitting the emission lines in the spectrum with a combination of photoionization (CLOUDY; Ferland et al. 1998), optically thin thermal (APEC; Foster et al. 2012), ²⁸ and shock (PSHOCK; Borkowski et al. 2001) ²⁹ models. We followed a similar procedure to that used for the spectral analysis of ESO 428-G014 (Fabbiano et al. 2018a). In particular, the photoionization model consists of a grid of values produced with the CLOUDY c08.01 package. The variables in CLOUDY are the ionization parameter ³⁰ (log U = [−3.00:2.00] in steps of 0.25) and hydrogen column density (log N_H = [19.5:23.5] in steps of 0.1) through the irradiated slab of gas. The APEC model generates a spectral emission from a collisionally ionized, and optically thin, diffuse hot (10⁴ < T_e < 10⁹ K) plasma assuming a thermal collisional ionization equilibrium and that the collisional excitation dominates. The collisional excited plasma may be powered by a shock-confined outflow (see Maksym et al. 2019). We also considered the PSHOCK model, as it allows for modeling a total or partial collisionless heating of electrons in the shock front (Borkowski et al. 2001). In particular, Borkowski et al. (2001) assume a plane-parallel shocked plasma model with constant post-shock electron and ion temperature, element abundances, and ionization timescale, providing a useful approximation for supernova remnants, but more generally for all cases in which X-ray emission is produced in a shock front.

We added the physical models to those of the leaky absorber plus reflection models (Appendix A.1). The PEXRAV model was replaced by the reflection PEXMON ³¹ model (Nandra et al. 2007), which self-consistently generates iron and nickel emission lines.

We initially set the normalizations of the partially absorbed power-law and reflection components to zero to allow the inclusion of new models in the soft (<3 keV) band. After fitting the new models, all the parameters of the leaky absorber model were allowed to vary. The choice of the best fit is based on the statistical F-test, the shape of the residuals as the ratio between data and model, and the physical plausibility of the fit parameters, as discussed below.

We started considering the physical models individually and then increased the complexity of the model by adding additional components up to a maximum of three physical models. We estimated the significance of the improvement due to an additional component with the F-test.

The strong emission feature at ∼1.7–1.9 keV, likely arising from a blend of Mg xii and Si xiii emission lines, is a clear signature of the presence of photoionized gas: the collisional (APEC) and shock (PSHOCK) models fail to reproduce this emission feature. Therefore, we at least require one photoionization CLOUDY component in each model below.

A single CLOUDY model gives a fit with good overall statistics, i.e., reduced χ² ≈ 1, but still leaves significant (∼4σ) residuals at low energies <0.7 keV, at ∼1.7–1.9 keV and ∼2.3 keV (see Figure 13).

Zoom In Zoom Out Reset image size

To minimize these residuals, we fitted the spectrum with the following two-component combinations: two photoionization, photoionization + collisional ionization, and photoionization + shocked ionization models (Figure 14). Based on the F-test, we obtain a significant improvement in all cases, leaving only ∼2.5σ significance contiguous residuals at <0.7 keV and at ∼1.7–1.9 keV and ∼2.3 keV, where the strongest emission lines are located. The best-fits are obtained by considering two CLOUDY components (${\chi }_{{\rm{R}}}^{2}/\mathrm{dof}=0.966/427$). Adding an APEC or PSHOCK component gives similar reduced ${\chi }_{{\rm{R}}}^{2}/\mathrm{dof}$ (0.970/427 and 0.978/427, respectively). We find no improvement with three-component models. The best-fit parameters are reported in Table 4. The best-fit model indicates the presence of both low ($\mathrm{log}\,U=-1.6\pm 0.1$) and high ($\mathrm{log}\,U=1.5\pm 0.1$) photoionization gas with high column density ($\mathrm{log}\,[{N}_{{\rm{H}}}/{\mathrm{cm}}^{2}]\gt 22.9$) consistent with the high column density ($\mathrm{log}\,[{N}_{{\rm{H}}}/{\mathrm{cm}}^{2}]={23.46}_{-0.02}^{+0.03}$) of the direct power law in the same fit.

Zoom In Zoom Out Reset image size

In the other two cases the fits suggest a low-photoionization, high-density gas and either collisional emission with temperature kT = 1.3 ± 0.2 keV or shocked gas with $kT={2.87}_{-1.76}^{+2.11}\,\mathrm{keV}$.

To model the spectra of the extended regions, we need first to evaluate and remove the contamination of the strong nuclear spectrum, spilling outside the central 2'' region in each region, due to the PSF wings. ³² The contribution of the nuclear emission in the external regions is estimated from the PSF model of the merged image. This PSF model was obtained by combining 500 PSF realizations produced for each observation with the ChaRT and MARX 5.5.0 tools, to give the same signal-to-noise ratio as the observation. As Chandra PSFs change with energy, we produced multiple PSF models at different energy bands. To evaluate the residual emission due to the nucleus in each sector, we computed, in each PSF image, the ratio between the net counts in a given external region and in the central circle of 2'' radius. These ratios represent the percentage of nuclear emission contributing to the spectral emission in each external region and are used to rescale the best fit of the nuclear spectrum, thus producing the final nuclear spillover models.

Figure 15 shows the percentage of nuclear emission contribution in each region versus energy. Notice that, for the single cross-cone and bicone sectors, the PSF spillover is approximately equal to and below 1%. For the entire annular extended region the fraction of nuclear spillover reaches a peak of 5% at 5 keV. However, the PSF models are subject to statistical uncertainties estimated to be less than 10% within 2'' of the centroid for on-axis PSFs, with an additional uncertainty of ∼5% for off-axis PSFs. ³³ We therefore assume that errors of our nuclear spillover are ∼10%–20%.

Zoom In Zoom Out Reset image size

Another contamination to the X-ray spectra of the extended regions, although less significant than nuclear spillover, derives from the emission of the stellar population. However, while this X-ray stellar contribution is negligible with respect to the nuclear spectrum, it may be of greater relevance in regions farther away from the AGN. In particular, the XRBs are the main contributors in the 2–10 keV band (Persic & Rephaeli 2002).

We used the star formation rate (SFR)–L_{2–10 keV} correlation in Lehmer et al. (2010) to estimate the total X-ray emission expected from the XRB population in IC 5063. The SFR was obtained from the L_FIR–SFR relation in Satyapal et al. (2005), with a far-infrared luminosity within a ∼12 kpc radius region of L_FIR = 4.7 × 10¹⁰ L_⊙ according to Wiklind et al. (1995) and Morganti et al. (1998). From these calculations, we predicted a total luminosity from the XRBs of ${L}_{2\mbox{--}10\,\mathrm{keV}}={2.5}_{-1.1}^{+5.9}\times {10}^{40}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$. Conservatively, assuming a uniform distribution of the XRB population, we derived the XRB luminosity at 2–10 keV expected in the extended regions, depending on their area relative to the total area of a circle of 12 kpc radius (where the L_FIR was extracted). Based on Persic & Rephaeli (2002), a power law with a fixed photon index Γ = 1.2 as spectral shape for the XRB emission was assumed. In conclusion, in each region the XRB emission at the 2–10 keV energy band is ≈3.5% of the emission due to the nuclear spillover in the same energy band, i.e., 0.1% of the central 2'' counts. The XRB contribution is never more than 2.2% of the observed counts into NW, SE, and cross-cone regions and so can be safely neglected.

We fitted the spectra extracted from the outer regions (i.e., NW cone, SE cone, and cross-cone) with phenomenological models plus emission lines. For each spectrum we fixed the nuclear spillover component (Appendix B.1).

We first fitted a power-law soft excess model as in Appendix A.1. Fitting with only power law plus prominent lines, we obtain a χ² ≈ 1 and few significant residuals at <0.8 keV in the bicone and >3 keV energies in the cross-cone spectra, respectively. The residuals were removed by adding a reflection PEXRAV component with photon index linked to the power law, significantly improving the statistics (χ² ≤ 0.9) according to the F-test.

In the bicone spectra the reflection component is required to fit the hard (>3 keV) X-ray part of the spectrum with a roughly equal contribution from the nuclear spillover component. Instead, in the cross-cone spectrum the nuclear spillover contributes ∼80% ± 20% (Appendix B.1) of the hard X-ray continuum. It could explain the totality of the emission within uncertainties, so that a reflection component would not be required. However, as the nuclear spillover model could change differently at different energies, we decided to keep its amplitude fixed and to add a reflection PEXRAV component with photon index free to vary, in order to remove contiguous residuals at high energy (>5 keV).

In summary, the best-fit models (top panels; Figure 16) of all spectra consist of a power law, a PEXRAV reflection component, and Gaussian emission lines. The best-fit properties and the identified emission lines are reported in Table 5, and we also show the L_{2–10 keV} estimated in each region.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 5. Best-fit Parameters and Reduced ${\chi }_{{\rm{R}}}^{2}$/dof of Empirical (top) and Physical (bottom) Models of the 0.3–7.0 keV Spectra Extracted from Extended Regions

Regions	χ2/dof	Counts (0.3–7.0 keV)	Γ a	normpl ( × 10−6 photons cm−2 s−1)		normrefl ( × 10−4 photons cm−2 s−1)	L2–10 keV (1040 erg s−1)
(A) NW cone	0.600/44	1332	${2.70}_{-0.28}^{+0.30}$	${7.10}_{-0.92}^{+0.69}$		${6.18}_{-3.14}^{5.25}$	2.5
(B) SE cone	0.700/45	1220	${2.70}_{-0.44}^{+0.40}$	${4.12}_{-1.18}^{+0.79}$		${9.79}_{-5.42}^{+8.37}$	2.5
(C) Cross-cone	0.736/91	2283	3.15 ± 0.20; <1.2 (refl)	6.97 ± 0.69		1.05 ± 0.20	6.3
Emission Lines
	Energy (keV)	Flux (10−6 photons cm−2 s−1)		Identified Emission Lines b
(A)	0.89 ± 0.02	1.18 ± 0.40		Ne ix [0.905 keV]/Fe xvii [0.897 keV]
	1.74 ± 0.03	0.17 ± 0.10		Mg xii [1.745 keV]
	1.90 ± 0.05	0.15 ± 0.10		Si xiii [1.865 keV]
	${5.02}_{-0.05}^{+0.06}$	0.19 ± 0.11		Ti xxii [4.977 keV]
	${6.34}_{-0.04}^{+0.34}$	0.48 ± 0.20		Fe Kα [6.4038 keV]
	${6.57}_{-0.04}^{+0.12}$	0.61 ± 0.23		Fe Be-, Li-like Kα [6.629,6.653 keV]/Fe xxv [6.610 keV]
(B)	0.89 ± 0.02	0.67 ± 0.38		Ne ix [0.905 keV]/Fe xvii [0.897 keV]
	${1.76}_{-0.01}^{+0.02}$	0.25 ± 0.10		Mg xii [1.745 keV]
	${2.37}_{-0.06}^{+0.04}$	0.25 ± 0.11		Si xiv [2.377 keV]
	6.38 ± 0.03	0.59 ± 0.21		Fe Kα [6.4038 keV]
(C)	0.86 ± 0.03	1.87 ± 0.58		Ne ix [0.905 keV]/Fe xvii [0.897 keV]
	1.02 ± 0.02	1.25 ± 0.31		Fe xxi [1.009 keV]/Ne x [1.022 keV]
	5.95 ± 0.04	0.64 ± 0.24		Cr xxiv [5.932 keV]
Best-fit Physical Models
Spectrum c	Fit Models	${\chi }_{{\rm{R}}}^{2}$/dof	Γrefl	$\mathrm{log}\,U$	$\mathrm{log}\,({N}_{{\rm{H}}}/{{cm}}^{-2})$	kT (keV)	EM [10−6 cm−5]
NW cone	2 CLOUDY	0.647/53	${2.14}_{-0.53}^{+0.79}$	${1.88}_{-0.11}^{+0.08}$	<19.94
				−2.77 d	<20.73
	CLOUDY + APEC	0.674/53	${2.10}_{-0.52}^{+0.32}$	−2.72	<20.47	${1.01}_{-0.13}^{+0.27}$	${1.84}_{-0.45}^{+0.46}$
SE cone	CLOUDY + APEC	0.610/50	>2.5	$-{1.83}_{-0.12}^{+0.10}$	${22.15}_{-0.24}^{+0.25}$	${1.36}_{-0.19}^{+0.32}$	${1.79}_{-0.71}^{+1.07}$
	2 CLOUDY	0.613/50	${2.49}_{-0.25}^{+0.30}$	1.75 d	${21.61}_{-0.24}^{+0.17}$
				$-{1.72}_{-0.16}^{+0.12}$	${22.33}_{-0.27}^{+0.29}$
Cross-cone	2 PSHOCK	0.750/35	<1.1			e ${0.68}_{-0.19}^{+0.12}$	${28.5}_{-9.7}^{+6.4}$
						e ${1.03}_{-0.11}^{+0.16}$	${3.41}_{-0.56}^{+0.32}$
	CLOUDY + PSHOCK	0.786/35	<1.1	${0.76}_{-0.16}^{+0.20}$	<19.59	e ${1.71}_{-0.34}^{+0.53}$	${2.73}_{-0.82}^{+0.79}$
	PSHOCK + APEC	0.755/35	<1.1			e ${2.17}_{-0.43}^{+0.88}$	${3.14}_{-0.47}^{+0.45}$
						f ${0.54}_{-0.20}^{+0.15}$	${1.95}_{-0.79}^{+1.23}$
	2 CLOUDY	0.788/35	<1.1	${1.75}_{-0.09}^{+0.07}$	<20.6
				${0.75}_{-0.04}^{+0.02}$	<19.7
	CLOUDY + APEC	0.809/35	<1.1	${0.75}_{-0.08}^{+0.05}$	<19.6	f ${1.23}_{-0.15}^{+0.13}$	${3.51}_{-1.03}^{+0.91}$
	2 APEC	0.848/35	<1.1			f ${1.21}_{-0.09}^{+0.08}$	${5.69}_{-0.80}^{+0.80}$
						f ${0.30}_{-0.05}^{+0.08}$	${6.03}_{-2.58}^{+2.50}$

Notes. For the phenomenological models we report energy, normalization, and identification of the emission lines we detect. We report the ${\chi }_{{\rm{R}}}^{2}$/dof of the physical models to the cross-cone spectrum estimated in the 0.3–3.0 keV energy band. The rest of the statistic is evaluated in the 0.3–7.0 keV energy band.

^a Same photon index for the power-law and reflection components, but for the cross-cone spectral model. ^b Identification emission lines from atomdb.org database. ^c For each spectrum we find more than one best-fit physical model. ^d Parameter fixed to the best-fit value or not constrained. ^e Temperature and normalization of the shock/thermal model. ^f Temperature and normalization of the shock/thermal.

Download table as:  ASCIITypeset image

The bottom panels of Figure 16 show the 5.8–6.8 keV energy band, which includes the neutral and ionized Fe K emission lines. In the bicone the 6.4 keV neutral iron transition is double the expected contribution of the nuclear spillover. In the NW and SE cone spectra we find Fe Kα lines with $\mathrm{EW}\,\simeq \,{898}_{-564}^{+1350}$ eV and ≃${796}_{-533}^{+1612}$ eV, respectively. Instead, the Fe Kα line in the cross-cone spectrum is consistent with the nuclear spillover alone.

The NW cone spectrum has a weak broad feature, which can be fitted with the Fe Kα line plus an emission line at ${6.57}_{-0.04}^{+0.12}$ keV, with $\mathrm{EW}\,\simeq \,{1216}_{-612}^{+1249}$eV at 2.7σ significance. This energy is consistent with the Fe xxv 6.7 keV line at 1.1σ and has roughly the same intensity as the neutral iron line.

In this section, we examine the mechanisms responsible for the X-ray emission in extended regions. As in Appendix A.2 for the nuclear spectrum, we fit a combination of CLOUDY, PSHOCK, and APEC (with solar abundances) models. In this case we also include the nuclear spillover contribution (Appendix B.1). We initially fit the spectra with a single physical component and then added additional components as required. We added up to four components to the models as in Appendix A.2.

We show the data/model residuals, ${\chi }_{{\rm{R}}}^{2}$, and degree of freedom of the intermediate spectral fits in Figure 18, while images of the selected best-fit models, with residuals, are reported in the following sections.

Zoom In Zoom Out Reset image size

B.3.1. NW Cone

B.3.2. SE Cone

The SE cone exhibits photoionization features similar to the NW cone spectrum, except for a less intense emission peak at ∼1 keV with respect to the 1.7–1.9 keV emission lines. Fitting these lines requires at least one CLOUDY component of photoionized gas emission in composite models. All the two-component model fits yield ${\chi }_{{\rm{R}}}^{2}\leqslant 1$, but most exhibit significant (≥2σ) contiguous residuals at >3 keV (Figure 18). The model including shock-ionized emission is discarded owing to implausible high temperatures predicted for the shocked gas (>10 keV).

As in the NW cone, a PEXMON reflection is included to fit the hard (>2 keV) X-ray continuum and the 6.4 keV neutral iron emission line. Using a photoionization + PEXMON reflection model reduces the residuals at >3 keV, achieving a reduced ${\chi }_{{\rm{R}}}^{2}=0.983$. However, residuals remain at ≈1.75 and 2.2 keV, suggesting additional spectral components. No significant residuals are found by using three-component models: a reflection and a photoionized phase plus either another photoionization or a collisionally ionized component. Including a shock model creates more residuals (Figure 18). These two best-fit models, as well as residuals, are shown in Figure 19, and the best-fit parameters are reported in Table 5. In both cases we have a low-photoionization gas ($\mathrm{log}\,U\lt -1.7$) with $\mathrm{log}[{N}_{{\rm{H}}}/{\mathrm{cm}}^{2}]\approx 22$ and a reflected PEXMON component with a photon index Γ ≥ 2.5. The remaining emission is ambiguously fitted by either a photoionization component with a large photoionization parameter ($\mathrm{log}\,U=1.75$) and column density $\mathrm{log}[{N}_{{\rm{H}}}/{\mathrm{cm}}^{2}]\approx 21.6$ or a collisional APEC component with temperature kT ≃ 1.36 keV.

Zoom In Zoom Out Reset image size

B.3.3. Cross-cone

The cross-cone spectrum (see Appendix B.2) is different from that of the bicone. It shows an intense hard (>3 keV) X-ray continuum with a similar shape to the nuclear spectrum and also a prominent soft excess around ∼1 keV, decreasing toward 2 keV. The hard spectral emission is dominated by spillover of the nucleus emission. We fit the soft excess with photoionization, collisional, and shock models. We include a PEXRAV reflection component to fit the remaining ∼20% of the >3 keV emission, in excess of the nominal spillover predictions, as with the phenomenological model (Appendix B.2). Most or all of the Fe Kα line in the cross-cone spectrum (≳60%) is due to the nuclear spillover. Because the physical models do not contribute to energies ≳3 keV, we evaluate the fit-statistic only for the soft (0.3–3.0 keV) X-ray part of the spectrum.

Adding model components to the baseline PEXRAV, we find good statistics (${\chi }_{{\rm{R}}}^{2}\gtrsim 1$), but with wide contiguous residuals. Using PEXRAV plus two-component models, we obtain both a reduced ${\chi }_{{\rm{R}}}^{2}\approx 0.8$ and few residuals at ∼1.6 keV, at <2σ significance. Adding additional components does not improve the fit.

We show all the best-fit models in Figure 20 and the respective parameters in Table 5. The temperature of the gas in the collisional and shock phase is always less than or similar to ∼1.2 and 2.1 keV, respectively. The CLOUDY component implies the presence of photoionized gas with $\mathrm{log}U\simeq 0.75$ and column density $\mathrm{log}[{N}_{{\rm{H}}}/{\mathrm{cm}}^{2}]\lt 19.7$. The two CLOUDY component fit also requires gas with high photoionization parameter ($\mathrm{log}U\simeq 1.75$).

Zoom In Zoom Out Reset image size