Feeding and Feedback in the Powerful Radio Galaxy 3C 120

We started the spectral analysis including a continuum power law with ${\rm{\Gamma }}\simeq 1.7(C/\nu =2407/2220)$. A Galactic absorption of N_H = 1.1 × 10²¹ cm⁻² modeled with wabs in XSPEC is included in all the fits (Kalberla et al. 2005). Equivalent results are obtained using the TBabs model in XSPEC. An inspection of the data in Figure 1 show the possible presence of absorption residuals in the soft X-ray, and possible emission lines. Indeed, an additional neutral absorber intrinsic to the source, with a column density of ${N}_{{\rm{H}}}=(4.3\pm 0.5)\times {10}^{20}$ cm⁻², is highly required (${\rm{\Delta }}C/{\rm{\Delta }}\nu =81/1$). Equivalent results are obtained using either the zwabs or zTBabs models in XSPEC.

Zoom In Zoom Out Reset image size

2.1.1. Emission Lines in the Soft X-Ray Band

In Figure 2 we can see a series of possible emission features in the energy range between E = 0.65–0.85 keV. When modeled with Gaussian emission lines, the rest-frame energies of the two most significant features are at E = 719 ± 3 eV and $E={777}_{-7}^{+2}$ eV, respectively. The parameters of the lines are reported in Table 2. We consider only the lines with a fit improvement of ΔC ≥ 4 for two additional degrees of freedom, corresponding to a detection significance higher than 90%.

Zoom In Zoom Out Reset image size

Table 2.  Best-fit Parameters of the Emission Lines in 3C 120

E	σ	I	EW	ID	ΔC
Soft X-Ray Emission Lines
719 ± 3	8 ± 2	23.0 ± 8.0	8 ± 2	Fe xvii	12
${777}_{-7}^{+2}$	<50a	5.8 ± 3.4	2 ± 1	Fe xviii/O viii	5
Fe K emission lines
${6232}_{-23}^{+8}$	<330a	${0.5}_{-0.2}^{+0.3}$	8 ± 4	Fe Kαc	6
6394 ± 5	<20a	1.9 ± 0.3	32 ± 6	Fe Kα	40
${6703}_{-20}^{+5}$	<40a	${0.6}_{-0.2}^{+0.4}$	11 ± 5	Fe xxv Heα	6
${7055}_{-15}^{+21}$	<80a	1.0 ± 0.4	${20}_{-8}^{+6}$	Fe Kβ	9

Notes. Columns: rest-frame energy in eV; line width in eV; intensity in units of 10⁻⁵ ph s⁻¹ cm⁻²; EW in eV; line identification; C-statistics improvement after including the Gaussian line. The line identifications are derived from the National Institute of Standards and Technology (NIST) database at www.nist.gov Verner et al. (1996) and Kaye & Laby at www.kayelaby.npl.co.uk.

^a90% upper limit.

Download table as:  ASCIITypeset image

The E ≃ 719 eV feature is relatively broad, with a width of σ = 8 ± 2 eV. The closest lines⁹ to the first emission feature are the ionized Fe L-shell doublet of Fe xvii, with transitions respectively at energies of E = 727 eV and E = 739 eV. Another spectral feature close to the observed energy is the O viiradiative recombination continuum (RRC) at E = 739 eV (e.g., Liedahl & Paerels 1996). The only other features around this energy are the Lα and Lβ fluorescence lines from neutral or lowly ionized Fe at E = 705 eV and E = 719 eV, respectively. The possible alternative interpretations of this feature are discussed in the following section using more physical models.

The second emission line at the energy of E = 777 eV may be associated with the Fe xviii $2{\rm{p}}\to 3{\rm{s}}$ doublet at E = 777 eV and E = 779 eV, or with O viii Lyβ at E = 775 eV. The latter possibility is supported by the detection of O viii Lyα at E ≃ 650 eV in the XMM-Newton reflection grating spectrometer (RGS) spectrum (Ogle et al. 2005; Torresi et al. 2012). The XMM-Newton RGS provided better signal-to-noise data at energies below E ≃ 700 eV, while Chandra HETG is superior at higher energies.

2.1.2. Emission Lines in the Fe K Band

In Figure 3 we show the ratio between the data and the absorbed power-law continuum in the Fe K region between E = 5.5–7.5 keV. We note the presence of several emission features.

Zoom In Zoom Out Reset image size

The emission line at energy E = 6394 ± 5 eV is detected with very high significance (${\rm{\Delta }}C/{\rm{\Delta }}\nu =40/2$), and it is clearly identified with the Fe Kα fluorescence emission doublet from neutral or lowly ionized material (Kα₁ at E = 6391 eV and K${\alpha }_{2}$ at E = 6404 eV). The emission line at $E={6232}_{-23}^{+8}$ eV, though less significant, is at the exact energy expected for the Compton shoulder of the main Fe Kα line of E ≃ 6240 eV (e.g., Matt 2002; Watanabe et al. 2003; Yaqoob & Murphy 2011).

The emission line at $E={7055}_{-15}^{+21}$ eV is instead at an energy consistent with the accompanying Fe Kβ fluorescence line at E = 7058 eV. If this interpretation is correct, then the equivalent width of this feature is estimated to be ≃60% of the Kα. This is much larger than the typical value of ≃11% expected for the Kβ from atomic physics (e.g., George & Fabian 1991). One possible explanation for this discrepancy could be that the Kβ line is blended with other features. The closest emission feature would be Fe xxvi Lyα at  E = 6970 eV. However, this possibility seems unlikely because the Kβ line is unresolved.

The Fe Kβ/Fe Kα ratio of emission line photons that escape from the obscuring torus may be higher than the tabulated value when the medium is optically thick to absorption of either of the two emission lines. This is due to the differential absorption opacities for the Fe Kα and Fe Kβ lines. Moreover, possible clumpiness in the torus may affect this ratio. In the next section we explore more physically motivated models for these features and discuss a possible interpretation for the discrepancy in the equivalent width ratio.

The emission feature at $E={6703}_{-20}^{+5}$ eV is consistent with the energy of the Fe xxv Heα resonance emission line at E = 6700 eV. A similar feature was also observed in lower-energy resolution spectra from Suzaku and XMM-Newton (e.g., Tombesi et al. 2010b, 2014; Gofford et al. 2013; Lohfink et al. 2013). The limited signal-to-noise and bandwidth in the Fe K region of the present Chandra HETG spectrum does not allow us to detect the broad Fe K emission feature previously reported in a Suzaku observation (Kataoka et al. 2007). The parameters of the lines are reported in Table 2.

After the initial phenomenological modeling, we performed a more physically motivated fit of the spectrum. The starting model was an absorbed power law.

2.2.1. Emission Lines in the Fe K Band

2.2.2. Emission Lines in the Soft X-Rays

2.2.3. Absorption in the Soft X-Ray and Fe K Bands

We performed a search for possible absorption features in both the soft X-rays and in the Fe K band. In the soft X-ray band we do not find absorption lines with a statistical significance higher than 99%, indicating that, if present, the WA is either highly ionized or it has a low column density. This is consistent with previous studies using XMM-Newton RGS (Ogle et al. 2005; Torresi et al. 2012). Using an xstar photoionized absorption table and assuming a typical ionization of logξ ≃ 2.5 erg s⁻¹ cm, as detected in the WA of other BLRGs (Reeves et al. 2009; Torresi et al. 2010, 2012), we estimate an upper limit of the column density of just N_H < 5 × 10¹⁹ cm⁻².

Absorption lines indicative of UFOs in the Fe K band have been reported in two Suzaku observations of 3C 120 performed in 2006 and 2012 (Tombesi et al. 2010b, 2014). The outflow in the 2006 Suzaku observation was detected in the energy range E ≃ 7–7.3 keV, with velocity v_out ≃ 0.08c, ionization parameter logξ ≃ 3.8 erg s⁻¹ cm, and column density N_H ≃ 1 × 10²² cm⁻² (Tombesi et al. 2010b). The 2012 observation showed an outflow at E ≃ 7.7 keV with higher velocity and ionization, i.e., v_out ≃ 0.16c, logξ ≃ 4.9 erg s⁻¹ cm, and N_H > 2 × 10²² cm⁻² (Tombesi et al. 2014).

The signal-to-noise and limited high-energy bandwidth of the Chandra HETG are not adequate to clearly detect such features. Nonetheless, in Figure 3 there is a hint of absorption features in the energy range between E ≃ 7.2–7.4 keV. We checked for the presence of an outflow using an xstar photoionization table with an input power-law continuum slope of Γ = 1.7 and a turbulence velocity of 1000 km s⁻¹. We find that an outflow may be present at the 99% level using the F-test (${\rm{\Delta }}C/{\rm{\Delta }}\nu =11/3$). However, we note that the significance may be lowered down to about 90%–95% considering a blind line search between E = 7.0–7.5 keV, which corresponds to about 10 intervals at a HEG resolution of 50 eV. The best-fit indicates an outflow velocity of v_out = 0.068 ± 0.002c, an ionization parameter log$\xi ={3.48}_{-0.10}^{+0.17}$ erg s⁻¹ cm, and a column density ${N}_{{\rm{H}}}=({3.3}_{-1.6}^{+2.4})\times {10}^{21}$ cm⁻². These parameters suggest the presence of a UFO with characteristics similar to the one detected in the 2006 Suzaku observation (Tombesi et al. 2010b, 2014). The final best-fit model is shown in Figure 4 and the parameters are listed in Table 3.

Zoom In Zoom Out Reset image size

Table 3.  Best-fit Model of the Chandra HETG Spectrum of 3C 120

Power-law Continuum
Γ	1.771 ± 0.008
Neutral abs
logNH (cm−2)	20.67 ± 0.05
Gsmooth × nei emission
FWHM (103 km s−1)	${2.4}_{-1.8}^{+1.4}$
kT (keV)	${0.7}_{-0.1}^{+0.9}$
τ (1011 s cm−3)	${1.3}_{-1.0}^{+1.2}$
EM (1064 cm−3)	${2.4}_{-1.6}^{+1.2}$
MYTorus
θ (deg)	20
logNH (cm−2)	>24.8b
Xstar emission
logξ (erg s−1 cm)	${3.75}_{-0.38}^{+0.27}$
NH (1021 cm−2)	>10b
Xstar absorption
logξ (erg s−1 cm)	${3.48}_{-0.10}^{+0.17}$
NH (1021 cm−2)	${3.3}_{-1.6}^{+2.4}$
vout(c)	0.068 ± 0.002
C-statistic
C/ν	2256/2209
Fluxa
0.5–2 keV	3.4
2–7 keV	4.1

Notes.

^aIntrinsic power-law continuum flux in units of 10⁻¹¹ erg s⁻¹ cm⁻². ^b90% lower or upper limit.

Download table as:  ASCIITypeset image

The soft X-ray band of 3C 120 below E ≃ 1 keV is rather complex and the current Chandra data show a series of emission lines in the energy range between E ≃ 0.6–0.9 keV.

The first report of possible thermal emission lines in the soft X-ray spectrum of 3C 120 was published by Petre et al. (1984) using Einstein Observatory data. The excess emission between E = 0.5–2 keV was modeled with a thermal bremsstrahlung component with a temperature near 1 keV. The signature of this component was the presence of weak emission lines at energies consistent with Fe L, Mg K, and Si K. Similar soft X-ray structures were also reported in the XMM-Newton and Suzaku spectra of 3C 120 (Torresi et al. 2012; Tombesi et al. 2014). The most intense of these features are likely associated with ionized Fe L transitions.

We tested several different alternatives for the origin of the emission in the soft X-rays and found that a non-equilibrium ionization collisional plasma model with an electron temperature of ${kT}=({0.7}_{-0.1}^{+0.9})$ keV provides both the best-fit and a physically plausible explanation. This model allows us to approximate the condition in shocks or hot plasmas.

The fact that the data require a broadening of this hot emission component indicates that there is an additional component compared to just the thermal one, which for this T ≃ 10⁷ K plasma would be only ${v}_{\mathrm{th}}=\sqrt{2{kT}/{m}_{\mathrm{Fe}}}\,\simeq$ 50 km s⁻¹. Instead, the width of the emission component is much larger and corresponds to a velocity broadening of $\mathrm{FWHM}\,={2400}_{-1800}^{+1400}$ km s⁻¹. This additional velocity component could be either associated with a spherical outflow or an expanding spherical shock. We note that this velocity is much higher than just the thermal broadening or turbulence from a hot interstellar medium,both being typically less than 100 km s⁻¹ (Werner et al. 2009).

We estimate an emission measure of $\mathrm{EM}=({2.4}_{-1.6}^{+1.2})\times {10}^{64}$ cm⁻³. Values of the emission measure of the order of 10⁶⁴ cm⁻³ have also been reported for the hot interstellar medium observed through Fe xvii emission at temperatures of kT ≤ 0.8 keV in the cores of large elliptical galaxies (Werner et al. 2009). Considering the emission measure and assuming a fully ionized plasma, ${n}_{{\rm{e}}}\simeq 1.2{n}_{{\rm{H}}}$, and considering a roughly spherical volume of radius ∼1.3 kpc corresponding to the extent of the region observed with the Chandra HETG of ≃2 arcsec, we then estimate a number density for the gas of n_H ∼ 0.5 cm⁻³. The temperature, density, and extension of the gas are indeed consistent with the values expected for the hot interstellar medium in giant elliptical galaxies (Xu et al. 2002; Werner et al. 2009; Kim & Pellegrini 2012).

Evidence for an excess of emission in the Chandra image from 1 to 3 arcsec with respect to the point-spread function reported in Ogle et al. (2005) is consistent with extended soft X-ray emission. We note that the resolving power of the grating may decrease if the source is extended. For instance, from Figure 8.23 of the Chandra proposers guide¹¹ we see that the MEG resolution could degrade down to $E/{\rm{\Delta }}E\simeq 200$ for a 5 arcsec extended source at a wavelength of 17 Å. Therefore, at E ≃ 700 eV the resolution may degrade down to a worse value of 3.5 eV. However, we note that the observed emission line at E ≃ 720 eV listed in Table 2 is resolved with a width σ = 8 ± 2 eV, which is at least twice larger than the worse energy resolution. Subtracting the 3.5 eV resolution in quadrature may only slightly reduce the broadening to around 7 eV, which is still consistent within 1σ. Moreover, the same line, with consistent broadening, was also reported in the XMM-Newton RGS spectrum by Torresi et al. (2012). Thus, we are confident that the broadening of the line is not affected by the soft X-ray emission region possibly extending up to a few arcsecs.

The total X-ray luminosity between E = 0.1–10 keV estimated for the hot gas in 3C 120 is L_hot ≃ 1.5 × 10⁴² erg s⁻¹. In order to keep the gas hot and prevent a catastrophic cooling flow, there must be some source of energy. An order of magnitude estimate of the cooling timescale for the hot gas associated with the galaxy can be estimated as ${t}_{\mathrm{cool}}=U/{L}_{\mathrm{hot}}$, where U is the internal energy of the gas defined as $U=(3/2){NkT}$, where N is the number of particles and T is the temperature. For an estimated density of n ≃ 0.5 cm⁻³ and considering a spherical region of radius about 1.3 kpc and temperature T ≃ 10⁷ K, we derive $U\simeq 6\times {10}^{55}$ ergs. We can then estimate a short cooling time of the order of ∼1 Myr or up to ∼10 Myr using the prescription in Peterson & Fabian (2006). Therefore, either most of the gas was heated more recently than this timescale or there is an additional source of heating that keeps it hot.

Shocks or outbursts can disturb the typical collisional plasma found in the galactic interstellar medium. The spectrum emitted by this plasma contains diagnostics that have been used to determine the time since the disturbing event, although this determination becomes uncertain as the elements return to equilibrium. For instance, this method has been used to determine the age of supernova remnants or to determine when individual knots in the ejecta were shocked (e.g., Hwang & Laming 2003; Yatsu et al. 2005).

From our best-fit with a non-equilibrium ionization model we estimate an ionization timescale of $\tau =({1.3}_{-1.0}^{+1.2})\times {10}^{11}$ s cm⁻³. Figure 1 in Smith & Hughes (2010) shows the typical timescales it takes for different ions to reach equilibrium for different temperatures. For a temperature of T ≃ 10⁷ K, estimated here from the Fe L emission lines, we see that a characteristic timescale to reach 90% of equilibrium would be much longer, τ ≃ 9 × 10¹¹ s cm⁻³. Therefore, the observed gas is about a factor of seven in time away from reaching equilibrium.

Considering the indicative number density of n ≃ 0.5 cm⁻³, we obtain a timescale of ∼10¹¹ s, or about 10,000 years since the main disturbance of the gas occurred. It is tempting to relate such a timescale to the onset or to intermittent outburst activity of the powerful jet or winds in this AGN, perhaps in line with evolutionary scenarios (Reynolds & Begelman 1997; Czerny et al. 2009; Baldi et al. 2015). This scenario is also consistent with the possible thermal emission from isothermal shocks suggested by Halpern (1985) as an explanation of the continuum emission and variability of 3C 120 using data from the Einstein X-ray observatory. The shocks in Halpern (1985) have higher density, velocity (${v}_{s}\sim 20,\,000$ km s⁻¹), and temperature (kT ∼ 7 × 10⁷ K), and are located closer to the black hole (at ∼parsec scales) compared to the ones observed through Fe L line emission here and in Petre et al. (1984). In analogy to supernova remnants, this may indicate a superposition or stratification of AGN shocks with various locations and ages.

Further support for the hot gas being heated by AGN-driven shocks also comes from comparison of the warm mid-IR H₂ emission compared to the thermal emission in radio galaxies. In particular, Ogle et al. (2010) detected the H₂ S(1) line in 3C 120, suggesting that the most likely heating mechanism is jet-driven shocks in a multiphase interstellar medium. Moreover, Lanz et al. (2015) reported a similar luminosity in the mid-IR H₂ lines and in the thermal X-ray emission for radio galaxies whose mid-IR spectra indicate the presence of shocks.

This evidence, together with the large velocity broadening of the hot emission component in 3C 120, about 1000 km s⁻¹, points to the possibility that the velocity width could be due to a spherical outflow or an expanding spherical shocked bubble, perhaps driven by the winds or jet from the central AGN. We note that the Fermi bubbles observed in our Milky Way indeed have comparable temperatures and sizes, and they are estimated to be expanding with a shock speed of v_shock ≃ 1000 km s⁻¹ (Fox et al. 2015; Nicastro et al. 2016).

We note that 3C 120 shows both UFOs and radio jets with powers of P_UFO ∼ 1 × 10⁴⁴ erg s⁻¹ and P_jet ∼ 5 × 10⁴⁴ erg s⁻¹, respectively (Tombesi et al. 2010b, 2014; Torresi et al. 2012). Hot shocked bubbles are indeed expected from theories of AGN feedback driven by winds or jets (e.g., Wagner et al. 2012, 2013; Zubovas & Nayakshin 2014). As the AGN wind sweeps up the ambient medium the forward shock decelerates, producing cooler gas with a temperature of $T\simeq {10}^{7}$ K for a shock velocity of v_shock ≃ 1000 km s⁻¹. Momentum conservation implies that the mean density swept up by the AGN wind is then of the order of n ≃ 0.1 cm⁻³ for an AGN luminosity of $L\sim {10}^{45}$ erg s⁻¹, a shock radius of ${R}_{s}\sim 1\,\mathrm{kpc}$, and a velocity of v_shock ∼ 1000 km s⁻¹. The typical X-ray luminosity for such wind-shocked gas from bremsstrahlung is  ${L}_{{\rm{X}}}\sim {10}^{42}$ erg s⁻¹ (Bourne & Nayakshin 2013; Nims et al. 2015). These parameters are indeed consistent with estimates for the hot X-ray-emitting gas in 3C 120.

A possible variability of the emitter was reported by Petre et al. (1984). Torresi et al. (2012) reported the following parameters for the emission line in the XMM-Newton RGS spectrum of 3C 120, obtained in 2003: line energy E ∼ 730 eV, flux  $I\simeq 7\times {10}^{-5}$ ph s⁻¹ cm⁻², and line width $\mathrm{FWHM}={4800}_{-2700}^{+2300}$ km s⁻¹. Compared to the parameters of the emission line in our Chandra HETG spectrum reported in Table 2, the main difference in the 11 years spanning between the two observations is a factor of 3 increase in line flux. In the framework of a hot shocked bubble, we would expect an increase in X-ray bremsstrahlung luminosity as the wind shock expands to larger radii (e.g., Nims et al. 2015). Future high-energy resolution observations of the source may confirm this scenario.

On the other hand, we note that the X-ray emission expected from star formation can be estimated as ${L}_{x(0.5-8\mathrm{keV})}\,\simeq 4\times {10}^{39}({\dot{M}}_{* }/1\,{M}_{\odot }\,{\mathrm{yr}}^{-1})$ erg s⁻¹, with a scatter of ≃ 0.4 dex, where ${\dot{M}}_{* }$ is the star formation rate (Mineo et al. 2014). Given the star formation rate in 3C 120 of ${\dot{M}}_{* }\,={2.8}_{-1.0}^{+0.9}$ M_⊙ yr⁻¹ estimated by Westhues et al. (2016) using multi-wavelength data, the resultant X-ray luminosity would be orders of magnitude lower than that observed.

The radio galaxy 3C 120 is an isolated, peculiar S0 galaxy about 25 kpc in diameter. The optical appearance is complex, with multiple dust lanes and two suggestive spiral arms that become radial features at a projected distance of about 6.3 kpc from the core. It is not clear whether the peculiar optical morphology, perturbed rotation curve and wide-spread enhanced star formation are due to the effects of tidal perturbation or interaction of the radio jet with the galaxy (Harris et al. 2004). It has been suggested that the optical data indicate a merger that is near completion (Heckman et al. 1986; Moles et al. 1988).

Indeed, integral field spectroscopy of the central regions of 3C 120 shows evidence of shells in the central kpc region that may be remnants of a past merging event in this galaxy and/or interaction of the radio jet with the intergalactic medium (García-Lorenzo et al. 2005). 3C 120 is a bulge-dominated galaxy probably experienced a merging event with a less massive galaxy. That galaxy was completely disrupted in the merging process, falling in parts that probably produced many of the observed optical structures. There is also evidence that a substantial fraction of its gas has been channeled toward the inner regions, perhaps activating a new feeding phase for the central SMBH (García-Lorenzo et al. 2005).

Indeed, CO emission data add substantial support to the hypothesis that 3C 120 involves a merger of gas-rich galaxies. The CO line profile is broad with a FWHM = 500 km s⁻¹ and a full width at zero intensity (FWZI) of 710 km s⁻¹ (Mazzarella et al. 1993; Evans et al. 2005). The velocity broadening is probably dominated by rotation, but there is also a component coming from disturbances due to the radio jet or AGN wind. The estimated H₂ mass is large, logM(H₂) = 9.79 M_⊙. This corresponds to about twice the molecular gas content of the Milky Way. These CO data corroborate evidence from optical imaging that this and similar radio galaxies originate in collisions and mergers between disk galaxies, as also recently suggested by other authors (e.g., Chiaberge et al. 2015).

Combined radio, optical, and Chandra X-ray images show complex interactions and shocks between the radio jet and colder material. Harris et al. (2004) showed that the radio jet may be impacting a cold, spiral-like structure observable at 4 arcsec (about 2.5 kpc) from the nucleus. Bright radio, X-ray, and optical jet knot emission is observed at this location (Harris et al. 2004). Evidence for complex mixing between hot shocked gas and cold gas due to the interaction between AGN winds or jets has been recently reported in an increasing number of studies (e.g., Morganti et al. 2013; Tadhunter et al. 2014; Feruglio et al. 2015; Tombesi et al. 2015; Dasyra et al. 2016). All these characteristics may place 3C 120 in a comparable parameter space as AGN-dominated ULIRGs (Veilleux et al. 2013; Cicone et al. 2014).

We do not detect any significant ionized absorption feature in the soft X-ray. This is consistent with previous studies using XMM-Newton RGS (Ogle et al. 2005; Torresi et al. 2012) and a shorter Chandra HETG (McKernan et al. 2007). Using an xstar photoionized absorption table and assuming the typical ionization of the WA detected in other BLRGs of logξ ≃ 2.5 erg s⁻¹ cm and a velocity consistent with zero at the source rest-frame (Reeves et al. 2009; Torresi et al. 2010, 2012), we estimate a very low upper limit of the column density of a possible WA of just N_H < 5 × 10¹⁹ cm⁻².

From the previous discussion of hot emission in the soft X-ray, we suggest that the lack of a WA may be due to the fact that the temperature of the hot interstellar medium of T ∼ 10⁷ K is much higher than the typical temperature for WAs, T ∼ 10⁵ K (Chakravorty et al. 2009). Therefore, it is plausible that the warm absorbing gas often observed in Seyfert galaxies, if present, has been heated up too much and thus is not observable.

The Fe K band spectrum of 3C 120 in Figure 2 is very rich, showing a series of at least four narrow emission and absorption lines due to iron at different ionization states: an unresolved Fe Kα emission line at E ≃ 6.39 keV, an associated Compton shoulder at E ≃ 6.23 keV, and Fe Kβ at E ≃ 7.05 keV from cold or neutral gas.

The Fe Kα emission line is unresolved, with a FWHM of less than 2300 km s⁻¹, consistent with previous estimates (Yaqoob & Padmanabhan 2004; Shu et al. 2010). This value is comparable to the width of the core of the broad-line region (BLR) observed in optical Hα and Hβ lines (Kollatschny et al. 2014), possibly suggesting a spatial overlap between the cold X-ray material and optical line emitting gas. The narrowness of the line may also be consistent with the low inclination, 20°, observed from the radio jet (Jorstad et al. 2005).

Note that although the FWHM of the Hα/Hβ lines is about 2000 km s⁻¹, the FWZI calculated considering the width of the line at which it reaches the continuum level is much larger, reaching velocities of up to 15,000 km s⁻¹ (Phillips & Osterbrock 1975; Osterbrock 1977). The large difference between the FWHM and the FWZI indicates that the lines should have broad and complex wings. The FWZI is indeed more representative of the full range of velocities in the BLR. Such broad wings could possibly be associated with the part of the BLR that is closer to the black hole, if they are due to rotation and/or a possible fast bipolar outflow component. If the Fe Kα at E ≃ 6.4 keV is related to the BLR, then it may show similar broad wings. However, the signal-to-noise of the current Chandra HETG observation is not high enough to test this possibility.

A detailed characterization using models of X-ray reflection from cold material gives a low reflection fraction of R = 0.22 ± 0.04, consistent with the relatively low EW of the Fe Kα emission line of EW = 32 ± 6 eV. The column density of the torus is estimated to be Compton-thick, with N_H > 6 × 10²⁴ cm⁻² in the equatorial plane if the MYTorus geometry is assumed. The detection of a Compton shoulder to the main Fe Kα line is consistent with these estimates (Awaki et al. 2008). The overall low reflection fraction may indicate a clumpy torus outside the line of sight or a high opening angle of the reflecting material (e.g., Tazaki et al. 2013).

Comparing the extrapolated E = 2–10 keV X-ray luminosity of ${L}_{2-10}\simeq 1.4\times {10}^{44}$ erg s⁻¹ and EW = 32 ± 6 eV derived here for 3C 120 with the Figure 7 in Ricci et al. (2013) we see that the values for this radio galaxy are consistent with those expected in the range between Seyferts and quasars. This is consistent with the X-ray Baldwin effect, which would relate the decrease of the covering factor of the torus with the increase in luminosity (e.g., Iwasawa & Taniguchi 1993).

The shape and strength of the neutral Fe Kα emission line suggest that the material feeding the accretion disk, or the torus, may be in the form of Compton-thick, clumpy clouds in an equatorial distribution. This is consistent with a recent model linking the micro and macro properties of AGN feeding and feedback (Gaspari & Sadowski 2017; Gaspari et al. 2017). In this model the inflow occurs via chaotic cold accretion, in which a rain of cold clouds condensing out of the quenched cooling flow is recurrently funneled via inelastic collisions. At the accretion disk scales, the accretion energy is transformed into UFOs and jets, ejecting most of the inflowing mass. The relatively high Eddington ratio of 3C 120 of $L/{L}_{\mathrm{Edd}}\sim 0.3$ is also consistent with the accretion disk being mostly relatively cold and efficient, with episodic thermal or magnetic instabilities linked to jet ejection events (Chatterjee et al. 2009; Tombesi et al. 2010b; Lohfink et al. 2013).

Besides emission and reflection from cold material, an Fe xxv Heα emission line at E ≃ 6.7 keV is also indicative of the presence of a significant amount of highly ionized gas in the central regions of this source. From the photoionization modeling we derive an ionization parameter of log$\xi ={3.75}_{-0.38}^{+0.27}$ erg s⁻¹ cm and a column density of N_H > 1 × 10²² cm⁻². From the definition of the ionization parameter, and assuming a compact spherical shell distribution (${\rm{\Delta }}R/R\leqslant 1$), we can estimate an upper limit on the distance of the emitting material from the X-ray source as ${R}_{\max }={L}_{\mathrm{ion}}/\xi {N}_{{\rm{H}}}$ (e.g., Tombesi et al. 2013a; Gofford et al. 2015). Substituting the estimated column density, ionization parameter, and ionizing luminosity, we derive an upper limit of ≃2 pc.

Absorption residuals in the energy range E ≃ 7.2–7.4 keV are suggestive of a possible UFO with a mildly relativistic velocity of v_out = 0.068 ± 0.002c, a high ionization parameter log$\xi ={3.48}_{-0.10}^{+0.17}$ erg s⁻¹ cm, and a column density of ${N}_{{\rm{H}}}=({3.3}_{-1.6}^{+2.4})\times {10}^{21}$ cm⁻². These parameters are consistent with the UFO detected in a previous 2006 Suzaku observation (Tombesi et al. 2010b, 2014). Considering the previous formula for R_max used for the ionized emitter, and substituting the parameters estimated from the fit, we can derive an upper limit of the distance of the absorber from the X-ray source of less than ≃10 pc.

Interestingly, the ionization parameter, column density, and distance of the Fe xxv emitter and the UFO are comparable, suggesting the possibility that the ionized emission and absorption may originate from the same or related material. Therefore, we may be observing emission and absorption from the same extended outflow, as also evinced from Suzaku observations of another radio galaxy, namely 3C 111 (Tombesi et al. 2013b).

An intriguing phenomenon detected in a number of radio imaging observations of AGN jets in blazars and some radio galaxies, notably 3C 120, is the presence of transverse gradients in the Faraday rotation across the jet at parsec scales (Gómez et al. 2000, 2008, 2011). The origin and nature of such a Faraday screen is still debated. It may be linked to the jet itself, such as a jet sheath with an helical/toroidal magnetic field (e.g., Gabuzda et al. 2014), or magnetized gaseous clouds completely external to the jet (e.g., Zavala & Taylor 2004). In the case of 3C 120, the Faraday screen has been linked to a clumpy sheath of thermal electrons surrounding the radio jet and interacting with the jet at a deprojected distance of 8 pc, with a hydrogen equivalent column density of N_H ∼ 6 × 10²² cm⁻² (Gómez et al. 2000, 2008). The fact that the observed changes in the polarization properties of the underlying jet emission and in the rotation measure seem uncorrelated implies that the Faraday screen and the jet are likely not closely physically related. At the same time, sign reversals in the rotation measure along the jet observed on the timescale of a few/several years imply that the Faraday screen medium is at least mildly relativistic (Gómez et al. 2011).

Consider that the ionized emission/absorption components observed in the Fe K band may be linked to the Faraday screen seen in the radio. In fact, the column density, distance, and relatively large outflow velocity of v_out ≃ 0.07 c are consistent with this possibility and they could naturally explain the observed rotation measure variability (Zavala & Taylor 2004).