A DEEP CHANDRA ACIS STUDY OF NGC 4151. III. THE LINE EMISSION AND SPECTRAL ANALYSIS OF THE IONIZATION CONE

Figure 1 (top) shows the 0.3–2 keV ACIS image (∼05 pixel⁻¹). Bright structured soft X-ray emission along the northeast–southwest (NE–SW) direction is apparent, together with fainter, less extended emission along the northwest–southeast (NW–SE) direction. The inner r < 2'' nuclear region, where we identified interactions between the radio jet and the optical clouds, has been discussed in Paper II. Here we focus on the kiloparsec-scale extended emission. We first extracted counts from sectors every 10° to create azimuthal surface brightness profiles for the extended X-ray emission (2'' ⩽ r ⩽ 30''), which provides a clear view of its angular distribution, and to identify sectors that contain the bright outflows (Figure 1, bottom). This plot was used to guide further spectral extraction (see Section 2.2.2). The extended emission along the NW–SE direction is interesting as this is the direction of the putative obscuring torus. Figure 2 demonstrates the clear elongation perpendicular to the bicone (in particular the NW sector), using the 0.3–2 keV ACIS image smoothed with the CIAO tool csmooth. We extracted the 0.3–2 keV radial profiles of the NW sector, the bright SW cone, and the faint emission in between (the "control region"), shown in Figure 3. The NW emission is on average five times fainter than the bright SW cone, but brighter than the control region at r < 7'' (4σ significance). Beyond r = 7'', it becomes indistinguishable with the faint emission. The implication on the nuclear obscuring structure will be discussed later in Section 3.2.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

A quick examination of the spectrum of the extended emission extracted from a circular region between a radius of 2'' and 30'' shows the clear presence of strong blended emission lines (Figure 4), corresponding to the emission lines (<2 keV) seen in High Energy Transmission Grating (HETG; Canizares et al. 2005) observations (Ogle et al. 2000). These lines cannot be uniquely resolved with the spectral resolution of the ACIS imaging observation. Most notably the blended lines appear as three strong emission features approximately centered at 0.57 keV (O vii), 0.68 keV (O viii), 0.91 keV (Ne ix) and a few weak lines between 1–2 keV (Mg xi, Si Kα, Si xiii; Figure 4). The blended emission features seen in the ACIS spectrum and the corresponding HETG lines are summarized in Table 1 (see Section 2.2.1).

Zoom In Zoom Out Reset image size

Following Paper II, we created adaptively smoothed images of the NGC 4151 circumnuclear region using the subpixel resolution data. We extracted images from the merged data in three spectral bands dominated by emission lines below 2 keV: 0.3–0.7 keV ("soft band" containing the O vii and O viii emission), 0.7–1.0 keV ("medium band" containing the Ne ix emission), and 1–2 keV ("hard band"). Figure 5(a) presents a false color composite image of the central ∼45'' × 45'' (∼3 kpc on a side), the circumnuclear region of NGC 4151, where the soft, medium, and hard band smoothed images are shown in red, green, and blue, respectively. Figure 5(b) zooms in to the central 1 kpc radius region, emphasizing the bicone emission. For visualization of the faint features, the nuclear region (see Paper II) is saturated and excluded from this figure. The X-ray biconical morphology has been described in Paper I, and here we focus on the spectral differences. The extended bicone emission (>2'') appears rich in 0.3–1 keV emission (red and green). Similar color emission (mixed red/green) is also seen perpendicular to the bicone direction with a hint of harder (blue) emission. Clumps of 0.3–0.7 keV (red; O vii and O viii) emission are also seen; some particularly prominent features are seen in the SW cone ∼6'' and ∼10'' from the nucleus.

Zoom In Zoom Out Reset image size

To further highlight the regions where the strong emission lines arise, we also extracted the X-ray emission in narrow energy intervals (0.53–0.63 keV; 0.63–0.73 keV; 0.85–0.95 keV) to create line-strength images. This is a reasonable approach since the line emission dominates over the weak underlying continuum (Schurch et al. 2004) in these narrow bands. The resulting images are shown in Figure 6. The position of the nucleus is indicated with a black cross. O vii and O viii line emission is prevalent in the extended regions. Ne ix emission, while extended along the general position angle of the large-scale extended emission, is bright in the central r = 2'' but becomes fainter quickly at r ≳ 3'' compared to the O vii emission. We examined similar emission line maps for the weak lines in the 1–2 keV range (Ne x, Ne ix RRC, Mg xi) and found that they are highly concentrated within the inner r = 2'' region. This is better demonstrated in Figure 6(d), where the radial profiles of the line emissions are shown.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Spectra and instrument responses were generated using CIAO version 4.2 and analyzed with XSPEC version 12.6 (Arnaud 1996). Background spectra were taken from a source-free region from the same chip. Spectra were grouped to have at least 20 counts per energy bin to allow χ² fitting. Unless otherwise noted, we restricted our modeling to photon energies above 0.3 keV, where the ACIS calibration is good,⁶ and below 2 keV, above which the nuclear continuum dominates (Wang et al. 2010b). The 90% confidence interval for a single interesting parameter is reported for all fitting results.

2.2.1. The 30'' Radius Extended Emission as a Testbed

We first characterize the X-ray spectrum of bright extended emission extracted from the 2'' ⩽ r ⩽ 30'' circular region, exploring a phenomenological spectral fitting approach that was broadly adopted in previous studies at CCD resolution (Yang et al. 2001; see also Ogle et al. 2000; Smith & Wilson 2001). The "continuum" in the ACIS spectrum (a combination of true underlying continuum and highly blended weak emission lines) was described by a smooth bremsstrahlung emission. We emphasize that the bremsstrahlung emission here should not be assigned any physical origin, whereas the derived line emission should be considered as a lower limit in such an approach.

Repeating the technique used to fit the spectrum of the nucleus (Wang et al. 2010b), narrow emission lines were then added to the model. The Chandra HETG and XMM-Newton Reflection Grating Spectrometer (RGS) spectra (Ogle et al. 2000; Schurch et al. 2004; Armentrout et al. 2007) were used as a guide for the identification of the soft X-ray lines. The X-ray spectrum of the nucleus of NGC 4151 is notoriously complex and variable. The hard X-ray continuum is dominated by a power-law component suppressed by a partially covering absorber and a Compton reflection component (Weaver et al. 1994; Schurch et al. 2004; De Rosa et al. 2007; Wang et al. 2010b; Lubinski et al. 2010), and the absorption consists of both a mildly ionized "warm absorber" and a cold absorber (Kraemer et al. 2005; González-Martín 2008). The PSF scattered contribution from the nuclear emission is taken into account in the spectral fitting, using the nuclear component fixed at the best-fit nuclear spectral model and normalization (Wang et al. 2010b) scaled to the expected fraction from PSF simulations. The extraction region contains 4.7% of the 0.3–2 keV nuclear emission due to PSF scattering, corresponding to ∼25% of the extracted counts. This can be taken as the upper limit of contamination to the extended line emission, as the line emission dominates the nuclear spectrum.

The identified emission line blends are summarized in Table 1 and the spectrum is shown in Figure 7 (see also Figure 4). The absorption column required by the fit is consistent with the line-of-sight Galactic column toward NGC 4151 (N_H = 2.1 × 10²⁰ cm⁻²; Murphy et al. 1996). The X-ray luminosity of the extended emission (between 130 pc and 2 kpc from the nucleus) corrected for the Galactic absorption is L_{0.3–2 keV} = (1.1 ± 0.2) × 10⁴⁰ erg s⁻¹, compared with L_2–10 keV = 2 × 10⁴² erg s⁻¹ for the unabsorbed nuclear source (Wang et al. 2010b).

Zoom In Zoom Out Reset image size

However, such a simplified model does not provide constraints on the emission mechanism, and more physically meaningful spectral models are needed. Fitting the data with combinations of absorbed optically thin thermal emission with solar abundances (APEC model; Smith et al. 2001) was attempted but gave poor results (reduced χ²_ν > 3). The abundances were then allowed to vary from solar abundance in the APEC model using the variable abundance APEC model (VAPEC), in which Z_O, Z_Ne, Z_Si, and Z_Fe were left free while other elements were fixed at solar abundance. When Z is allowed to vary, the fit generally improves but still is far from satisfactory (χ²_ν ∼ 2). Moreover, the required abundances become unphysically low (Z < 0.01 Z_☉) as the fit attempts to reproduce a smooth continuum.

To have a self-consistent model, we have made use of the Cloudy photoionization modeling code (Ferland et al. 1998). Although the resolution of the ACIS CCD spectrum prevents us from performing line-based diagnostics, e.g., deriving line ratios of the Helium-like triplets, spectral fitting with photoionization models still provides useful constraints about the photoionization status (as in Bianchi et al. 2006, 2010; Gonzalez-Martin et al. 2010). Using Cloudy version C08.00, which enables a Cloudy/XSPEC interface (Porter et al. 2006), we modeled the <2 keV X-ray spectrum assuming an open plane-parallel geometry ("slab"). The dimensionless ionization parameter (Osterbrock & Ferland 2006) is defined as U = Q/(4πr²cn_H), where n_H is the hydrogen number density, r is the distance to the inner face of a model slab, c is the speed of light, and Q = ∫^∞_13.6 eVL_ν/hν is the emitting rate of hydrogen ionizing photons (s⁻¹) by the ionizing source. We assumed the broken power law used by Kraemer et al. (2005) for the AGN continuum (see also Armentrout et al. 2007). We varied U and N_H, the column density of the slab, to create spectral models covering a grid of parameters (−3 ⩽ log U ⩽ 3, 19.0 cm⁻² ⩽ log N_H ⩽ 23.5 cm⁻²), which were fed to XSPEC.

Armentrout et al. (2007) successfully modeled the XMM-Newton/RGS soft X-ray spectrum of NGC 4151 with three photoionization components: a high-ionization component (log U = 1.3, log N_H = 23.0), a medium-ionization component (log U = 0, log N_H = 23.0), and a low-ionization component (log U = −0.5, log N_H = 20.5). In our initial attempt, we included all of these components, with the ionization parameters and the column densities fixed at these values and the normalizations set free. In addition, the photoionized emission was absorbed with a line-of-sight column density, N_{H, los}. The fit to the ACIS spectrum was poor (χ²_ν ⩾ 6). This is not a total surprise, given that these best-fit parameters are optimized for the brightest photoionized material close to the nucleus (10⁻³ to 1 pc; Armentrout et al. 2007) that dominates the soft X-ray spectrum of NGC 4151, whereas the emission of interest here arises on the r ∼130 pc–1.5 kpc scale.

When the (U, N_H) parameters for the photoionized components were set free to vary, we were able to obtain a statistically satisfactory fit (χ²_ν = 1.1), reproducing the observed ACIS spectrum of the large-scale extended X-ray emission well (Figure 8). The line-of-sight column density required by the fit is N_{H, los} = 2 × 10²⁰ cm⁻², consistent with the Galactic column toward NGC 4151 (Murphy et al. 1996). Two components, a high-ionization phase (log U = 0.8, log N_H = 20.0) and a low-ionization phase (log U = −0.25, log N_H = 19.4), are required to produce both the hydrogen-like and helium-like neon and oxygen line emissions. The Cloudy model predicts gas pressures of 6 × 10⁻⁷ dyne cm⁻² and 1.3 × 10⁻⁷ dyne cm⁻² for the high-U component and low-U component, respectively. If they are collocated, additional pressure—perhaps from a hot intercloud medium in a multiphase ISM (Elvis et al. 1983; Ogle et al. 2000)—is needed to reach pressure equilibrium. Note that the log U = −0.25 component closely resembles the "Low" model component in Armentrout et al. (2007) and the "D+Ea" X-ray absorption component in Kraemer et al. (2005).

Zoom In Zoom Out Reset image size

These model components still left small residuals at ∼0.5 keV (the N vii Lyα) and 1.02 keV (the Ne x Lyα). Adding in a third photoionized component can further improve the fit (best-fit parameters log U = 1.23, log N_H = 19.2), but the improvement appears insignificant when an F-test is performed (at the 70% significance level). Thus, the two-component model is used as a template for the following analysis of subregions.

Given the satisfactory spectral fits with photoionized components and lack of clear signatures of collisional ionization in the ENLR, how much thermal emission from hot gas may be present in the circumnuclear region of NGC 4151? We obtain a constraint on the possible presence of the hot ISM by invoking a collisionally ionized component in the best-fit photoionization model. After adding a thermal emission component (APEC) to the photoionized emission, the fit is improved with a kT = 0.81 ± 0.05 keV (at the >99% significant level by F-test) and the residual at 0.9–1.1 keV is mostly accounted for. This thermal component may contribute ≲ 12% of the total soft X-ray emission, L_{th, 0.5–2 keV} = 6 × 10³⁸ erg s⁻¹. As a cautionary note, although the APEC component improves the fit statistically, it remains possible that the residuals are due to photoexcitation of Fe L shell and the enhanced resonance lines (e.g., Ogle et al. 2003; Kinkhabwala et al. 2003) that are inadequately accounted for with the Cloudy photoionization model components. This will be explored with the high-resolution HETG spectrum in future work. We will discuss the possible presence of the thermal component further.

We emphasize that, because of likely complex geometrical dependencies and absorption between these components in the kiloparsec-scale emission, this modeling may still be oversimplistic to simultaneously reproduce the observed features in the broad band.

2.2.2. Spectral Fitting of Selected Subregions

To perform spectral analysis of the extended X-ray emission, the large-scale region was divided into four sectors with P.A.s as outlined in Figure 1. Extraction regions are termed the "SW," "NW," "SE," and "NE" regions, respectively (Table 2; see Figure 1). The bisectors of NE and SW correspond to the approximate P.A. of the ENLRs (e.g., Robinson et al. 1994). They are further divided into subregions along the radial direction (e.g., SW 1, 2, 3, and 4; NE 1, 2, and 3), with ∼1000 counts (0.3–2 keV) in the fainter outer regions (SW 4 and NE 3). The NW and SE sectors cover regions (2'' ⩽ r ⩽ 15'') perpendicular to the bicone, which is the direction of the large-scale weak bar (Asif et al. 1998).

Table 2. Spectral Fits to the Extended X-Ray Emission

Region	Source	log NH, losa	log U1	log NH, 1	log U2	log NH, 2	Nuclear PSFb	χ2/dof	F0.5 − 2 keV
	Counts (0.3–2 keV)	(× 1022 cm−2)		(cm−2)		(cm−2)	Counts (0.3–2 keV)		(erg s−1 cm−2 arcsec−2)
NE 1	2432	0.04 ± 0.02	0.86	19.7	−0.22	19.7	380	63/65	6.9 × 10−15
NE 2	1814	0.07 ± 0.03	1.05	19.8	0.06	20.1	200	60/53	1.3 × 10−15
NE 3	913	0.03 ± 0.01	0.49	19.9	−1.87	23.5	140	26/29	1.7 × 10−16
SW 1	2096	0.02	0.83	20.0	−0.38	19.4	270	60/57	1.0 × 10−14
SW 2	1667	0.02	1.06	20.0	−0.42	19.8	130	58/45	2.9 × 10−15
SW 3	1560	0.02	1.00	19.6	−0.50	19.6	100	54/46	7.2 × 10−16
SW 4	966	0.02	0.34	19.0	−1.28	23.5	80	34/30	4.5 × 10−17
NW	1050	0.22 ± 0.02	1.03	19.2	−0.98	19.5	400	35/35	3.1 × 10−16
SE	1081	0.16 ± 0.04	1.26	20.3	−1.16	19.9	530	28/33	2.2 × 10−16
O vii-rich	2626	0.02 ± 0.02	0.14	20.3	−1.22	20.3	200	60/56	4.3 × 10−15
Ne ix-rich	2141	0.07 ± 0.02	0.6	19.6	−0.21	20.4	210	50/56	4.0 × 10−15

Notes. ^aThe line-of-sight absorbing column density toward the photoionized emission. For the SW regions, the fitted N_{H, los} prefers values lower than the Galactic column toward NGC 4151, and was fixed at N_H = 2 × 10²⁰ (Murphy et al. 1996). ^bExpected counts from the simulation of PSF scattered nuclear emission.

Download table as:  ASCIITypeset image

Corresponding to the variations in color, spectral differences are clearly visible when the ACIS spectra across sectors are compared (Figure 9). The fitting results are summarized in Table 2, and Figures 10, 11, and 12 show the corresponding spectra and the fits. We find the following.

• 1.  
  Except for the outermost regions (NE 3 and SW 4), a high-ionization component is present in all regions (log U ∼ 0.83–1.26). Lower ionization components are present along the bicone direction (NE–SW), with log U values ranging between −0.5 and 0. Recall that the fit to the integrated spectrum, which includes emission from all the subregions, finds the log U ∼ 0.8 and log U ∼ −0.25 components.
• 2.  
  In the NE 3 and SW 4 regions, a very low ionization component (log U > −1) and a high column density (log N_H = 23.5) are required by the fit to improve the residuals in the Mg i Kα and Si i Kα. An F-test indicates that this fluorescent component significantly improves the fit at 98% confidence.
• 3.  
  The second photoionized component in the NW–SE regions, perpendicular to the ionization cone, shows a lower ionization (log U ∼ −1).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We also extracted spectra of the X-ray emission in the bicone regions that are prominent in 0.3–0.7 keV emission (or "O vii" rich, visually red in Figure 5) and in 0.7–1 keV emission (or "Ne ix" rich, visually green in Figure 5). The spectral fits are shown in Figures 13(a) and (b), respectively. Whereas the fitted ionization parameter for the "Ne ix" rich emission is close to the average value, the "O vii" rich emission is generally of lower ionization, likely due to higher number density. This is consistent with the appearance of clumpy material in the optical.

Zoom In Zoom Out Reset image size

Kiloparsec-scale extended X-ray emission is ubiquitously observed in nearby obscured Seyfert galaxies with ENLRs (Bianchi et al. 2006, 2010; Guainazzi et al. 2009; Gonzalez-Martin et al. 2010), but the ionization mechanism for this emission has been a subject of active debate. For NGC 4151, Yang et al. (2001) proposed a two-phase model for the extended X-ray emission, including a photoionized warm gas and a hotter collisionally ionized gas (see also Komossa 2001). The relatively low electron temperatures (T ∼ 10⁴–10⁵ K) inferred from the RRC (see Liedahl 1999 for review) and the line ratios from the He-like triplets strongly support the presence of a photoionized component (Ogle et al. 2000; Schurch et al. 2004; Kraemer et al. 2005). However, does the observed soft spectrum entirely consist of photoionized emission, or does collisional ionization become more important than the photoionization mechanism beyond the nuclear region (Armentrout et al. 2007)?

3.1.1. Evidence for Photoionization as the Dominant Ionization Mechanism

Several of our results point to photoionization as the dominant mechanism.

• 1.  
  Emission Line Gas Morphologies. First of all, the morphological agreement between soft X-ray and optical [O iii] emission is very strong. In particular, there is a good spatial correlation between [O iii] line emission and the three strong X-ray emission lines (Figures 6(a)–(c)). There is clearly an overall morphological coincidence with the ionized gas bicone traced by [O iii] emission, as noted in previous X-ray studies (e.g., Ogle et al. 2000; Yang et al. 2001). A good agreement between the peaks of optical and X-ray line emissions in the inner r = 3'' is found.
• 2.  
  [O iii]-to-soft X-Ray Ratio of the HST-resolved Clouds. Bianchi et al. (2006) studied a sample of eight Seyfert 2 galaxies, and suggested that the same kiloparsec-scale gas photoionized by the AGN continuum can simultaneously produce the observed X-ray and [O iii] emission. Our HRC and ACIS studies of the inner 2'' (<130 pc) emission in NGC 4151 (Wang et al. 2009 and Paper II) have shown that, for most of the clouds close to the nucleus, the ratios of [O iii] to soft X-ray emission are consistent with the typical ratios in Bianchi et al. (2006), the exception being clouds collisionally ionized by the radio jet.Our subpixel-processed ACIS images of NGC 4151 enable us to directly compare the extended X-ray emission and those of the resolved [O iii] clouds (Figure 6). The main cloud features are labeled in Figure 14. Using the calibrated continuum-subtracted HST/WFPC2 F502N image (Kaiser et al. 2000), we measured the [O iii] fluxes for the clouds and listed them in Table 3. ACIS spectra were extracted from the same regions and fitted with the continuum-plus-emission line-model to obtain their 0.5–2 keV X-ray fluxes. Figure 15 shows the [O iii] to soft X-ray ratio for the distinct cloud features at various radii to the nucleus (∼150–1500 pc). The ratios listed in Table 3 are consistent with a constant [O iii]/X-ray flux ratio of ∼15 for the 1.5 kpc range spanned by these measurements. The radially constant [O iii]/X-ray ratios indicate that the relative contributions of the low- and high-ionization gas phases at different radii are similar, as pointed out by our referee. If the [O iii] and X-ray emission arise from a single photoionized medium, as discussed in Bianchi et al. (2006), this further implies that the ionization parameter does not vary significantly to a large radial distance and the density is approximately decreasing as r⁻², as expected for a freely expanding nuclear wind based on mass conservation (Crenshaw et al. 2000; Bianchi et al. 2006).
• 3.  
  Spectral Fitting. The spectral analysis shows that a model with just two photoionized components of high- and low-ionization parameters can describe well the observed X-ray spectra in various regions across the circumnuclear area, whereas in the spectral fits that invoke only collisionally ionized components, strong emission line residuals persist. This supports the dominant role of nuclear photoionization.
• 4.  
  Kinematics of the Ionized Gas. The full velocity field of the NLR for NGC 4151 was mapped by Kaiser et al. (2000), which supports a hollow biconical outflow model for these clouds. At distances exceeding 4'' (∼260 pc) from the nucleus, Kaiser et al. (2000) find that the gas cloud kinematics are consistent with circular rotation in a gravitational field, identical to the host ISM. Therefore, except for a few clouds in the nuclear region that appear disturbed by the impact of the radio jet (Kaiser et al. 2000; Mundell et al. 2003; see also Paper II), there is no evidence for strong kinematic shocks or any strong shocks in the outflow, which is consistent with the gas being photoionized by the nuclear radiation.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 3. Measured X-Ray and [O iii] Fluxes for the Clouds

Cloud	Distance	Distance	[O iii] Flux	0.5–2 keV Flux	$F_{[\rm O\,\mathsc{iii}]}/F_{[0.5\hbox{--}2\, {\rm keV}]}$
Number	to Nuc.	to Nuc.
	('')	(pc)	(10−14 erg s−1 cm−2)	(10−15 erg s−1 cm−2)
1	2.6	169	29.8 ± 1.9	25.7 ± 5.2	11.6
2	2.7	175	21.1 ± 2.0	15.2+1.0− 5.4	13.8
3	3.2	208	2.6 ± 0.4	2.4+0.3− 1.3	10.7
4	4.8	312	7.9 ± 0.9	5.4+5.0− 0.7	14.5
5	3.5	228	1.2 ± 0.4	1.6+0.2− 1.0	7.5
6	4.7	306	1.9 ± 0.2	4.7+0.2− 3.5	4.1
7	6.1	397	10.2 ± 6.0	6.6+0.5− 3.1	15.2
8	10.7	696	3.9 ± 0.7	4.2+0.2− 2.5	9.6
9	17.9	1164	0.9 ± 0.1	0.2+0.4− 0.2	43.4
10	17.7	1151	1.8 ± 0.1	0.9+0.1− 0.8	20.3

Note. The [O iii] flux is measured from F502N image.

Download table as:  ASCIITypeset image

Based on all the above considerations, we conclude that photoionization by the nucleus is most likely the dominant ionization mechanism in the ENLR of NGC 4151, confirming the suggestions of previous studies (Ogle et al. 2000; Schurch et al. 2004; Kraemer et al. 2005; Armentrout et al. 2007; Wang et al. 2009 b). It is reassuring to establish a role for nuclear photoionization in the extended NLRs that is consistent with the grating studies. However, there is also evidence that points to a more complex emission mechanism than a single photoionized medium (e.g., Bianchi et al. 2006). We discuss these results and their implications below in Sections 3.2 through 3.4.

3.1.2. Constraints on the Presence of Collisionally Ionized Gas

Figure 6(a) clearly shows that one distinct O vii emission blob "A" (at 6'', ∼390 pc SW from the nucleus) does not align with a bright [O iii]-emitting cloud, designated as "K1" in Robinson et al. (1994) and offset by 1'' to the SE of "A" (see Yang et al. 2001). We find this is also the case for a more distant X-ray blob "B" (10'' SW) where the closest [O iii] emission cloud is 2'' to the SW. Similar displacements between the X-ray enhancement and the [O iii] blob are also clearly found in the O viii line emission (Figure 6(b)). Note that the Chandra astrometry is accurate to 03 (Paper I); therefore the misalignment between features at this spatial scale is significant and indicates that the origin of the optical/X-ray emission needs further investigation.

The radial location of X-ray blob "A" lies at the boundary between NLR and ENLR identified in the optical (Robinson et al. 1994), where the density drops rapidly from n ∼ 1600 cm⁻³ to n ∼ 200 cm⁻³ (Penston et al. 1990; Robinson et al. 1994). We can estimate the ionization parameter of K1. From the nuclear spectrum (Wang et al. 2010b), we obtain an ionizing luminosity L_{13.6 eV–100 keV} ∼ 10⁴⁴ erg s⁻¹. Assuming a number density of 200 cm⁻³ (Penston et al. 1990; Robinson et al. 1994) and a radial distance of 1.2 × 10²¹ cm to the nucleus, we obtain an ionization parameter log U = −2. This is lower than, but in agreement with, the value of log U ∼ −1.7 derived from the [O iii]/[O ii] ratio in Robinson et al. (1994), supporting the photoionization assumption.

Since the X-ray blob and the [O iii] cloud are located at the same radial distance, their spatial offset suggests that the X-ray blob is in a higher ionization phase than the [O iii]-bright cloud. A likely explanation is that the X-ray blob has a lower density in the clump. If their densities are similar, the difference might arise from a higher ionizing flux received at the surface of the X-ray blob, perhaps due to less screening than in the direction toward the [O iii] cloud.

Our spectral fitting results show that a range of ionization states are needed to obtain good fits, in agreement with the multiple ionization parameters required in the grating spectra (e.g., Armentrout et al. 2007).

A high-ionization (log U ∼ 1) component is found in the spectral fitting for all regions surrounding the nucleus, including the NW and SE sectors perpendicular to the ionization cone (Table 2). A low-ionization (log U ∼ −1) component is also present in the NW–SE direction that has lower U than found in the cone direction (NE–SW). This strongly indicates that there is no continuous absorbing "torus" blocking all the ionizing photons. This hypothesis is supported by the agreement between the observed H₂ line flux in the "torus" direction and the value predicted from X-ray irradiation by the active nucleus (Paper I), and by the measurement of H i absorption in the nuclear region, where high levels of clumping on the smallest scales is found (Mundell et al. 2003). We note that a "patchy torus" or clumpy wind arising from the accretion disk in which the clouds are optically thick (e.g., Elitzur & Shlosman 2006) may allow leakage of nuclear continuum for such a high-U component. However, it cannot reproduce the log U ∼ −1 component in the directions where the continuum is blocked.

Extended emission along the "torus" direction is clearly present in the large-scale X-ray images shown in Figures 2 (indicated by the arrows) and 5(a). The base of this emission is at the location of the nucleus, and thus may indicate ionization by the leaked photons in this direction. Moreover, we found an intriguing coincidence between radial regions of enhanced X-ray surface brightness and the P.A. of outflowing optical clouds identified by the HST study (Das et al. 2005). They pointed out that these faint clouds flowing close to the plane of the torus would need to be ionized by leaked nuclear radiation, which is attested to by our X-ray data. The low U suggests a lower ionizing flux received by some clouds, likely due to nuclear continuum filtered by "warm absorbers" (e.g., Krongold et al. 2007) or the cone walls (e.g., Kraemer et al. 2008). Evaluating the difference between the ionization parameters in the direction of the torus and in the bicone, we estimate that approximately 1% of the total ionizing flux is seen by the NW–SE clouds.

Previous work on the NLR kinematics of NGC 4151 have established that the bicone geometry in NGC 4151 results in one side of the conical outflow exiting the galactic plane at a small angle, partly intersecting the host disk (e.g., Robinson et al. 1994; Crenshaw et al. 2000; Das et al. 2005; Storchi-Bergmann et al. 2009). Thus, part of the host ISM is likely illuminated by the AGN radiation. Our X-ray work finds that multiple phases of photoionized gas are present in the ENLR of NGC 4151. The fitting results for the "O vii" rich (red in Figure 5) and "Ne ix" rich (green in Figure 5) regions indicate that there is a range of ionizations, with a lower ionization phase associated with the clumpy clouds and the higher ionization with the diffuse outflow. There is a lowest ionization (log U ∼ −2) component at larger radii in the SW and NE sectors. This component may come from high-density material in the host disk plane, perhaps the nuclear spirals, illuminated by the nuclear radiation.

We further note that interactions between the biconical outflow and the host galaxy ISM, similar to the radio jet–cloud interaction in the nuclear region, may contribute to the heating of any hot gas present in the host disk. Currently there is no evidence in the gas kinematics that supports any strong interaction (Kaiser et al. 2000). However, as we discuss below, there is some morphological evidence that, at the largest radii, the outflow runs into a clump in the inhomogeneous ISM and perhaps results in local shock heating of the gas.

Figure 16 gives a large-scale (1' × 1') view of the key ISM components in the circumnuclear region of NGC 4151, illustrating the directions of the kinematic major axis of the host galaxy (green solid line, P.A. ∼ 22°; Pedlar et al. 1992; Mundell et al. 1999), the large-scale "weak fat bar" (green dotted line, P.A. ∼ 130°; Mundell & Shone 1999), the ENLR bicone (cyan line, P.A. ∼ 65°; Evans et al. 1993), and the distribution of H i (Mundell et al. 1999) and CO (Dumas et al. 2010) gas, respectively.

Zoom In Zoom Out Reset image size

In Paper I we noted that the NE soft X-ray emission reaches as far as 12'' (780 pc) from the nucleus, but that the surface brightness then decreases significantly by a factor of five. Comparing the continuum-subtracted Hα emission (Knapen et al. 2004) with the extended soft (0.3–1 keV) X-ray emission (Figure 17(a)), the ionized gas traced by bright Hα emission is mainly located in the r ∼ 10'' biconical region along the NE–SW direction centered on the nucleus, which also closely follows the soft X-ray emission. An arc-like feature in the Hα image (∼15'' × 2'') is clearly present at the r = 10'' location (Figure 16) at the edge of the X-ray emission, and the northern CO gas lane is also found here (Figure 17(b), and Dumas et al. 2010).

Zoom In Zoom Out Reset image size

Figure 17(b) provides a multiwavelength, enlarged view of this region, showing the spatial relations between the soft X-ray emission, the Hα, the CO gas lanes, and the H i distribution (Mundell & Shone 1999). The termination of the NE X-ray emission appears to be closely associated with the presence of the dense cold neutral material (traced by the CO and H i emission) in the galactic disk. No similar feature is seen to the SW of the nucleus.

The origin of the Hα arc (Figure 17(a)) at the terminus of the NE X-ray emission is intriguing. It is spatially very close to the northern CO gas lane and is one of the dust arcs identified in the optical V−I color map (Asif et al. 1998), which are gaseous compressions driven by the bar potential (Mundell et al. 1999; Dumas et al. 2010). In this context, the Hα arc may be a more clumpy continuation of the dense CO arc and photoionized by the AGN radiation. The rate of ionizing photons from the AGN is approximately 10⁵³ s⁻¹ (Kraemer et al. 2005; cf. L_{13.6 eV–100 keV} ∼ 10⁴⁴ erg s⁻¹, Wang et al. 2010b). Assuming that the arc has a line-of-sight depth that is comparable to its projected length (∼1 kpc), the arc's intercepted ionizing photon rate is 8 × 10⁵¹ s⁻¹ at its distance to the nucleus (970 pc), a covering factor of almost 10%. This photon budget should be treated as an upper limit as some fraction of the ionization photons will be absorbed or scattered on the way to the arc. If we assume that the gas is in photoionization equilibrium, for the case B effective recombination rate of Hα (Osterbrock 1989), a rate of 8 × 10⁴⁹ s⁻¹ ionizing photons s⁻¹ is needed to produce the observed Hα luminosity of the arc (L_Hα ∼ 8 × 10³⁷ erg s⁻¹), which is only 1% of the maximal rate of available ionizing radiation in the direction of the arc (for a covering factor f_c = 10%). The scenario where part of the CO gas lane is photoionized by the AGN and produces the Hα arc seems plausible with a modest covering factor of f_c ∼ 0.1%, implying that the arc has a filamentary structure with a thickness of a few parsecs.

On the other hand, we consider the possibility that the arc could be a bow shock feature from interaction between the biconical outflow and dense gas in the host disk piled in the CO gas lanes by the large-scale stellar bar. Assuming that the X-ray emission is due to shock heating as the outflow encounters the dense CO lane, and that the shock is strong and adiabatic, we estimate the shock velocity v_s, adopting a postshock temperature T_ps = (3/16)(μv²_s/k_B) (Lehnert et al. 1999), where μ is the mass per particle (CO and HI) and k_B the Boltzmann constant. For the kT ∼ 0.3 keV X-ray temperature measured at the arc, the required v_s is ∼150 km s⁻¹, which is much higher than the local sound speed c_s = 10 km s⁻¹ in the ∼10⁴ K Hα-emitting gas. Although there is no direct measurement at the exact position of the Hα arc, the measured outflow velocity in [O iii] at r = 8'' NE of the nucleus is fully consistent with this value of v_s, e.g., cloud 28, which has v = 190 ± 32 km s⁻¹ (Kaiser et al. 2000).

For two of the brighter, spatially resolved [O iii]-emitting clouds seen in the HST images (clouds 1 and 2 in Figure 14; see also Figure 4 in Kaiser et al. 2000), we were able to constrain the number density for the X-ray-emitting gas with Cloudy modeling of their X-ray spectra when an inner radius r = 100 pc to the nucleus and an ionizing photon rate Q = 10⁵³ are set. The best-fit density is log n_H = 2.9 ± 0.5 cm⁻³. We use this number density to estimate the mass-loss rate and the kinematic power of the hot phase outflow, which are key quantities that gauge the impact of AGN outflow on the host galaxy ISM.

Following Barbosa et al. (2009) and Storchi-Bergmann et al. (2010), we calculate the mass-loss rate using $\dot{M}_{x}$ = n_Hm_Hv_rC_gA, where v_r is the radial velocity (taken as the approximate outflow velocity), m_H the proton mass, C_g the gas filling factor, and A the area in the cross section of the outflow. For a radial distance of r = 4 × 10²⁰ cm (130 pc) and the hollow cone geometry of Das et al. (2005), we obtain A = πr²(sin ²θ_out − sin ²θ_in) = 1.2 × 10⁴⁰ cm². Adopting C_g = 0.11 (Storchi-Bergmann et al. 2010) and v_r ∼ 750 km s⁻¹ (Crenshaw et al. 2000; Storchi-Bergmann et al. 2010), we obtain a mass outflow rate $\dot{M}_{x}=2.1\, M_{\odot }$ yr⁻¹ and a kinetic luminosity for the outflow $L_{{\rm outflow}}=1/2\dot{M}_{x}v^2 \sim 1.7 \times 10^{41}$ erg s⁻¹, or ∼0.3% of the bolometric luminosity of the AGN (L_bol = 7.3 × 10⁴³ erg s⁻¹; Kaspi et al. 2005).

The $\dot{M}_{x}$ derived from the hot phase outflow is comparable to the value derived from near-IR studies of ionized gas in NGC 4151 (Storchi-Bergmann et al. 2010), which estimated a mass outflow rate at ≈1.2 M_☉ yr⁻¹ along each cone. The $\dot{M}_{x}$ value is over 10 times higher than the outflow rate derived from UV and optical spectra (0.16 M_☉ yr⁻¹; Crenshaw & Kraemer 2007), but the latter is measured much closer to the nucleus (at 0.1 pc). The optical-emitting gas likely has a much smaller filling factor than the X-ray-emitting gas, which explains the difference in the measurements. If our study measured an outer part of the same outflow, it could have accelerated to a higher velocity or have loaded ISM between 0.1 pc and 100 pc, and hence a higher $\dot{M}_{x}$.

How does $\dot{M}_{x}$ compare to the accretion rate of the supermassive black hole in NGC 4151? The bolometric luminosity of NGC 4151, L_bol = 7.3 × 10⁴³ erg s⁻¹ (Kaspi et al. 2005), is about 1.2% of its Eddington luminosity, which is L_Edd ∼ 6 × 10⁴⁵ erg s⁻¹ for the black hole mass M_BH = 4.57^+0.57_{− 0.47} × 10⁷ M_☉ of NGC 4151 (Bentz et al. 2006). Adopting $L_{{\rm bol}}=\eta \dot{M}_{{\rm accr}} c^2$ and η = 0.1 for the efficiency of the accretion disk (Shakura & Sunyaev 1973), this implies that the central engine of NGC 4151 is accreting at $\dot{M}_{{\rm accr}}=0.013 \,M_{\odot }$ yr⁻¹. Since the measured outflow rate is 160 times the accretion rate necessary to feed the active nucleus, there must be either significant entrainment of the host galaxy's ISM while the nuclear outflow is expanding, or part of the gas fueling the nucleus is ejected before radiating, or a high fraction (over 99%) of the X-ray gas does not take part in the outflow and is illuminated by the AGN "in situ." However, the latter seems unlikely as the grating spectra strongly indicate that the emission lines have outflow velocities of 200–300 km s⁻¹ (e.g., Armentrout et al. 2007). Significant mass loading of the outflow by the host ISM may be most plausible and common, since previous estimated NLR outflow rates in Seyfert galaxies (∼0.1–10 M_☉ yr⁻¹) generally exceed the accretion rate by tenfold (Veilleux et al. 2005).

Nevertheless, even with such a high $\dot{M}_{x}$ in NGC 4151, the kinematic power L_outflow suggests that only 0.3% of the available accretion power is extracted to drive the outflow, similar to the findings of Storchi-Bergmann et al. (2010) and Holt et al. (2006, 2011). For a sample of Seyfert NLR outflows (NGC 4151 not included) studied in the [S iii] λ9069 emission, Barbosa et al. (2009) reported a lower L_outflow/L_bol = 10⁻⁵ to 10⁻⁴. As pointed out in Mathur et al. (2009), this poses a problem for the majority of quasar feedback models which require a higher fraction of the accretion power of the black hole (∼5% of L_bol to unity) to be thermally coupled to the host ISM. Our L_outflow/L_bol is in rough agreement with a two-stage feedback model recently proposed by Hopkins & Elvis (2010), which requires only 0.5% of the radiated energy to drive the initial hot outflow to efficiently shut down star formation in the host.