Spatially Resolved BPT Mapping of Nearby Seyfert 2 Galaxies

The [O iii] λ5007 and Hα λ6563 lines were observed with the Wide Field and Planetary Camera 2 (WFPC2) on board the HST by various programs (PIDs: 5411, 6332, 6419, 8259). WFPC2 consists of four chips, PC1, WF2, WF3, and WF4. PC1 has a pixel scale of 00455 and the WF chips have a pixel scale of 00996 (Gonzaga et al. 2010). Depending on the redshifts, the [O iii] and Hα line imaging were taken with the narrowband filters if the lines could be covered. Otherwise, linear ramp filters (LRFs) were used, where the central wavelength was tuned to the wavelength of the redshifted lines. Corresponding continuum images were taken with medium- or wide-band filters or LRFs. For the LRFs, the transmissions and passbands are position-dependent, resulting in an upper limit of 14'' on the field of view or the size of an emission line region. Exposure times varied from 400 s to 1600 s for the emission lines and 50 s to 280 s for the corresponding continua. All observations were split into two integrations to allow cosmic ray rejection except for NGC 5643's [O iii] and Hα continuum observations, where only a single exposure was executed. The details of the observations, e.g., instrument/chip, filter, exposure time, observation date, and program ID, are given in Table 1.

Data reduction started with reprocessing all the images with astrodrizzle in the software package Drizzlepac⁵ (Gonzaga et al. 2012). We set the final image pixel scale to 004 to match WFC3 or 005 to match the Advanced Camera for Surveys (ACS) (see Section 2.2), as BPT mapping requires all four emission line images having exactly the same pixel scales, and also because the native WFPC2 images are undersampled. Individual frames were combined to remove cosmic rays and corrected for geometric distortions. In cases where a few cosmic rays or hot pixels remained after running astrodrizzle or in the case of a single exposure, we manually removed them with the Pyraf tool imedit. We also rotated all the images such that north is up and east is left during the astrodrizzle run.

After the initial processing with astrodrizzle, we subtracted the background, which was determined by the median value of line-free regions, for both emission line and continuum images. We then multiplied these images by the flux calibration keyword PHOTFLAM and the FWHM of the filter bandwidth 2.355 × PHOTBW to convert the image units to flux ergcm⁻² s⁻¹ pixel⁻¹. The emission line and corresponding continuum images were aligned according to the positions of the nucleus and stars in the common field. Continuum-free emission line images were generated by subtracting a scaled continuum image from the on-band image. The scaling of the continuum images was estimated from the FWHM of the on-band filter bandpass, and was adjusted when necessary to avoid over-subtraction or under-subtraction.

Due to the relatively large filter bandwidth and close proximity in wavelengths, the contribution of [N ii] λλ6548, 6584 to the Hα filter is significant (typically 30%–50%) and must be subtracted. However, the local [N ii]/Hα ratio is subject to spatial variation in the intrinsic [N ii]/Hα ratio over the entire emission line region (Tsvetanov & Walsh 1992; Ferruit et al. 1999), the wavelength shifts due to velocity gradients (Whittle et al. 1988), and the position on the chips (e.g., position-dependent transmission of the LRFs). To account for the uncertainties in [N ii] subtraction, we first used a typical [N ii] fractional contribution of 40% as a constant rescaling factor and applied it to the [N ii]+Hα images to obtain Hα-only images. Then we varied the [N ii] contribution fraction and tested the effects on the final BPT maps (Section 3).

The final continuum-subtracted emission line maps were used to measure the integrated line fluxes. We measured the integrated fluxes of both the entire emission line region as well as the nuclear region. The line flux measurements and corresponding circular aperture sizes are listed in Table 2. The apertures are centered on the nucleus, the location of which is determined as the peak position of [O iii].

Table 2.  Observed Emission Line Fluxes

Sourcename	${F}_{[{\rm{O}}\ \mathrm{III}]}^{\mathrm{int}}$	${F}_{[{\rm{O}}\ \mathrm{III}]}^{\mathrm{nuc}}$	${F}_{{\rm{H}}\alpha }^{\mathrm{int}}$	${F}_{{\rm{H}}\alpha }^{\mathrm{nuc}}$	Aint	Anuc
	(erg cm−2 s−1)	(erg cm−2 s−1)	(erg cm−2 s−1)	(erg cm−2 s−1)	(arcsec)	(arcsec)
NGC 1386	(1.27 ± 0.30) × 10−12	(4.72 ± 0.10) × 10−13	(4.98 ± 0.61) × 10−13	(1.38 ± 0.17) × 10−13	4	0.4
NGC 5643	(2.14 ± 0.22) × 10−12	(8.62 ± 0.27) × 10−13	(5.63 ± 0.79) × 10−13	(2.25 ± 0.24) × 10−13	4.8	0.35
NGC 2273	(3.42 ± 0.45) × 10−13	(1.24 ± 0.35) × 10−13	(2.22 ± 0.25) × 10−13	(5.72 ± 0.35) × 10−14	2.5	0.4
NGC 3081	(6.49 ± 0.35) × 10−13	(1.88 ± 0.09) × 10−13	(2.28 ± 0.38) × 10−13	(5.38 ± 0.37) × 10−14	2	0.32
Mrk 573	(1.41 ± 0.08) × 10−12	(3.42 ± 0.11) × 10−13	(4.53 ± 0.41) × 10−13	(1.01 ± 0.09) × 10−13	5	0.5
NGC 7212	(8.22 ± 0.42) × 10−13	(3.03 ± 0.14) × 10−13	(2.90 ± 0.28) × 10−13	(9.09 ± 0.72) × 10−14	3	0.3
NGC 7674	(4.74 ± 0.27) × 10−13	(3.59 ± 0.16) × 10−13	(2.67 ± 0.26) × 10−13	(1.76 ± 0.08) × 10−13	1.3	0.35
	${F}_{[{\rm{S}}\ \mathrm{II}]}^{\mathrm{int}}$	${F}_{[{\rm{S}}\ \mathrm{II}]}^{\mathrm{nuc}}$	${F}_{{\rm{H}}\beta }^{\mathrm{int}}$	${F}_{{\rm{H}}\beta }^{\mathrm{nuc}}$	Aint	Anuc
NGC 1386	(7.08 ± 0.14) × 10−13	(6.35 ± 0.12) × 10−14	(4.53 ± 0.17) × 10−13	(3.80 ± 0.21) × 10−14	4	0.4
NGC 5643	(4.39 ± 0.09) × 10−13	(7.32 ± 0.13) × 10−14	(3.07 ± 0.15) × 10−13	(4.61 ± 0.18) × 10−14	4.8	0.35
NGC 2273	(3.19 ± 0.09) × 10−13	(4.69 ± 0.12) × 10−14	(8.64 ± 0.35) × 10−14	(1.01 ± 0.22) × 10−14	2.5	0.4
NGC 3081	(1.95 ± 0.05) × 10−13	(2.76 ± 0.10) × 10−14	(1.94 ± 0.11) × 10−13	(3.12 ± 0.20) × 10−14	2	0.32
Mrk 573	(4.10 ± 0.10) × 10−13	(5.62 ± 0.17) × 10−14	(3.55 ± 0.25) × 10−13	(6.15 ± 0.18) × 10−14	5	0.5
NGC 7212	(1.99 ± 0.07) × 10−13	(4.37 ± 0.12) × 10−14	(1.76 ± 0.10) × 10−13	(3.38 ± 0.22) × 10−14	3	0.3
NGC 7674	(1.67 ± 0.05) × 10−13	(8.15 ± 0.15) × 10−14	(1.08 ± 0.11) × 10−13	(5.18 ± 0.15) × 10−14	1.3	0.35

Note. F^int is the integrated line flux of the NRL/ENLR measured in a circular aperture with a radius of A_int. F^nuc is the integrated line flux of the nuclear region measured in a circular aperture with a radius of A_nuc.

Download table as:  ASCIITypeset image

All the [S ii] λλ6716, 6731 and Hβ λ4861 imaging data in this work are from our HST program 15350 (P.I. Maksym). We obtained the [S ii] and Hβ narrow line and the [S ii] continuum images with either WFC3/UVIS or ACS/WFC, depending upon the redshift and available filters. No additional observations were taken for the Hβ continua as the existing [O iii] continuum images can be used for Hβ as well. Exposure times varied from 750 s to 1034 s for the lines and from 190 s to 314 s for the [S ii] continua. WFC3 images have a native pixel scale of 004 (Gennaro et al. 2018) and ACS has a pixel scale of 005 (Lucas et al. 2018). The WFC3 and ACS images display faint charge trails following hot pixels, cosmic rays, and bright stars in the direction of CCD readout, which need charge transfer efficiency (CTE) correction. For all the filters except the LRFs, we used the CTE-corrected images (*flc.fits) by the HST pipeline. The HST pipeline is not capable of correcting CTE for the LRFs though.

We first reprocessed the images with astrodrizzle, setting the final image scales to the native pixel sizes of WFC3 or ACS. Subsequent processing of [S ii] and Hβ is the same as the procedures described in Section 2.1. No contamination due to strong emission lines is expected in these filters. The line flux measurements on the continuum-subtracted line maps are given in Table 2.

Mrk 573 (z = 0.0172, D = 74.6 Mpc, 1'' = 349 pc) is known for its extended emission line region with rich features in the ionization bicone (e.g., Falcke et al. 1998; Ferruit et al. 1999). There are two sets of arcs and multiple emission knots in between. A triple radio source is coincident with the nucleus and the inner emission knots (e.g., Ulvestad & Wilson 1984; Falcke et al. 1998).

Figure 1 shows the continuum-subtracted [O iii] λ5007 and Hα line maps observed with WFPC2, and the [S ii] λλ6716, 6731 and Hβ line maps taken with ACS in the inner 1.8 kpc of Mrk 573. [S ii] and Hβ are fainter than [O iii] and Hα, and the outer arcs are almost invisible. Using these line maps, we created the BPT diagram (left panel) and the spatially resolved BPT image (right panel) with each pixel color-coded according to its BPT type as shown in Figure 2. The boundaries of different excitation mechanisms are defined in Kewley et al. (2006). Red corresponds to Seyfert-like activity, yellow denotes LINER-like activity, and the transition zone is coded as green. Blue represents line ratios typical of H ii regions. Only pixels detected above 3σ in all lines are used. H ii pixels are excluded by this criterion, and most pixels that meet this criterion are located within the inner arcs. The spatially resolved, color-coded map clearly demonstrates that the nucleus and the ionization bicone are dominated by Seyfert-like activity, closely wrapped by a thin layer (∼20 pc) of Seyfert–LINER transition, and surrounded by LINER-type emission (∼250 pc in thickness). This structure is highly consistent with the recently identified LINER "cocoon" feature revealed in our pilot HST BPT mapping of NGC 3393 (Maksym et al. 2016), although the LINER region in Mrk 573 appears thicker than in NGC 3393. Nevertheless, both sources are limited by the depths of the observations and therefore the full extent of the LINER cocoon has yet to be revealed by deep observations. In this work, we have also identified a thin layer of Seyfert–LINER transition and shown the smooth transition from Seyfert to LINER emission for the first time.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Revalski et al. (2018) also constructed spatially resolved BPT diagrams in 1D, although along the NLR of Mrk 573 using emission line ratios measured from the spectra taken with the HST/STIS long-slit and ground-based Dual Imaging Spectrograph at the Apache Point Observatory. For the HST/STIS spectrum, they measured line ratios at each spatial distance (spatial resolution = 01) from the nucleus along the 02 wide slit. The slit position of −712 spatially samples the bright nuclear emission and inner and outer arcs, but does not pass through the bright emission line knots. The ground-based 2'' wide slit covers most of the NLR features, and Revalski et al. also measured line ratios at each distance (spatial resolution = 04) within 2'' of the nucleus. Their N–BPT, S–BPT, and O–BPT diagrams show that all their data points lie in the Seyfert AGN regime. There are a few points near the Seyfert–LINER border and they interpret the emission as gas exposed to an absorbed or filtered ionizing continuum. In addition, they performed multi-component photoionization modeling and the results are consistent with pure AGN ionization from the central source. So they conclude that they do not see any evidence for shock ionization in those regions. However, a single spatial dimension cannot simultaneously capture the radial and azimuthal variations in emission. They do not see LINER-type most likely because the line ratios are measured along 1D, where both Seyfert and LINER regions within the slit width can be mixed and unresolved and dominated by Seyfert-type line ratios. We will further discuss the origin of LINER activity and relation to shock excitation in Section 5.

We have tested the effect on the resulting resolved BPT map due to uncertainties in subtracting the [N ii] contribution from the Hα filter. We used the [N ii]/Hα line ratio measured at difference distances from the high quality HST/STIS spectrum in Revalski et al. (2018) as a reference. Only a small number of pixels are affected (e.g., BPT type changed) when varying the fractional [N ii] contribution from the maximum to minimum, and our conclusions are unchanged.

NGC 1386 (z = 0.0029, D = 12.4 Mpc, 1'' = 60 pc) is one of the closest Seyfert galaxies, located in (or in the foreground of) the Fornax Cluster (e.g., Storchi-Bergmann et al. 1996a; Rossa et al. 2000). The extended [O iii] and Hα emission line region (Figure 3) is characterized by an ionization bicone with several blobs distributed along the north–south direction, extending over a region of ∼6'' (360 pc) (e.g., Ferruit et al. 2000; Schmitt et al. 2003). The emission along the east–west direction is much narrower, extended by only ∼15 (90 pc). A ring of small emission blobs is located ∼2'' south of the nucleus. The [S ii] and Hβ emission displays a slightly different morphology in the WFC3 images (Figure 3). The far north and south components are much less prominent than those in [O iii] and Hα. Multiple strong, sharp dust lanes are present in the circumnuclear region. The [S ii] and Hβ image scales are stretched to show the dust features. The morphology of the NLR is strongly shaped by the dust features. Mundell et al. (2009) compared the [O iii] emission with VLA radio data at 8.4 GHz. The radio source generally follows the direction of [O iii], although the radio and optical components do not appear to be directly associated (Figure 5 in Mundell et al. 2009).

Zoom In Zoom Out Reset image size

Using the continuum-subtracted line maps in Figure 3, we calculated the line ratios for each pixel and created the BPT diagram and the spatially resolved map color-coded by BPT type (Figure 4). The bright emission blobs show red Seyfert-type activity and transition to yellow LINER regions. A LINER "cocoon" entirely encloses the central emission line region and partially surrounds the northern and southern components. Using the VLT/MUSE IFU, Mingozzi et al. (2019) conducted BPT mapping of the disk and outflow components in NGC 1386, where they show a more complete picture of the LINER emission, permeating in the disk.

Zoom In Zoom Out Reset image size

NGC 3081 (z = 0.00799, D = 34.4 Mpc, 1'' = 164 pc) is a well-known barred spiral galaxy with multiple rings (e.g., Ferruit et al. 2000; Buta et al. 2004). Extended ionized gas is found within ∼1 kpc of the nucleus (Durret & Bergeron 1986; Ferruit et al. 2000). Figure 5 shows the continuum-subtracted line maps within the central 1 kpc × 1 kpc region. The line-emitting region mainly consists of a central component, a northern component, and a fainter southern component, all distributed along the north–south direction forming an ionization bicone. A loop feature is attached to the northern component in the northwest direction, ∼60° away from the bicone axis. Since the [O iii] and Hα images were taken with different chips on WFPC2, the loop is better resolved into small hot spots in Hα on PC1, which has a higher resolution than [O iii] on WF2. The loop feature also appears in the Hβ image and is less obvious in [S ii]. Such a feature has also been discovered in other Seyfert galaxies, e.g., IC 5063 (W. P. Maksym et al. 2020, in preparation).

Zoom In Zoom Out Reset image size

Based on the continuum-subtracted line maps, we conducted BPT mapping and the results are shown in Figure 6. The nucleus and the ionization cone show red Seyfert activity and are well enclosed by a LINER cocoon through a thin, green layer of Seyfert–LINER transition. A small, blue H ii region, where pixels meet the 3σ criterion, is located 25 from the nucleus in the southeast direction, as can be seen in the Hα and Hβ line maps in Figure 5. It is part of a star-forming region in the circumnuclear spiral arms (Buta et al. 2004).

Zoom In Zoom Out Reset image size

NGC 7212 (z = 0.0266, D = 116.2 Mpc, 1'' = 535 pc) is a spiral galaxy belonging to a group of three interacting galaxies (e.g., Wasilewski 1981; Falcke et al. 1998). Figure 7 shows the continuum-subtracted line maps in the central ∼3 kpc × 3 kpc region. The extended, asymmetric, diffuse NLR is elongated in the northwest–southeast direction with a strong central component. The WFPC2 observations of [O iii] and Hα are limited by the small field of view of the LRF. Cracco et al. (2011) confirm the presence of an extended ionization bicone up to 3.6 kpc from the nucleus with ground-based observations with a large field of view. There are multiple dust lanes across the galaxy. The dark region northwest of the nucleus is due to a sharp dust lane. The VLA map of NGC 7212 in Falcke et al. (1998) reveals a compact double radio source separated by 07 in the north–south direction, offset from the optical nuclear emission peak.

Zoom In Zoom Out Reset image size

The ionization cone is again dominated by red Seyfert activity as shown in the resolved BPT map (Figure 8). A LINER cocoon is enveloping the Seyfert ionization cone along the eastern and western sides. A small number of pixels in the H ii region of the BPT diagram (left panel) are located in the southern end of the ionization cone (right panel), where stars can form. These star-forming clumps appear to be adjacent to the red Seyfert regions without a LINER layer in between. Such star-forming clumps have also been found in NGC 5643 (Cresci et al. 2015), which are Hα-bright clumps. Most Seyfert regions are still surrounded by a LINER cocoon, and star-forming regions are prevalent in the outer part (Cresci et al. 2015).

Zoom In Zoom Out Reset image size

Using the Magellan Echellette Spectrograph on the Clay telescope, Congiu et al. (2017) extracted spectra from different regions along the slit (PA = 16°) and constructed BPT diagrams to diagnose ionization mechanisms in each region. Most points are clustered in the Seyfert region of the BPT diagrams. The main ionization mechanism in the NLR/ENLR is photoionization by the AGN. Some points are located close to the LINER region, which they explain as a consequence of shocks combined with velocity information. We cannot directly compare our BPT results region by region to theirs though, because their slit only covers a tiny portion of the ionization structures and their region bin size is much larger (>1''). The overall dominance of Seyfert over LINER excitation is consistent with that derived from our maps.

NGC 7674 (z = 0.02903, D = 127.1 Mpc, 1'' = 582 pc) is the brightest member of a compact galaxy group (Verdes-Montenegro et al. 1997; Schmitt et al. 2003). Figure 9 shows the continuum-subtracted line maps in the central ∼2.9 kpc × 2.9 kpc region. They exhibit a loop feature with a hole in emission northwest of the nucleus (∼55° away from the cone axis), which is most prominent in the [O iii] image. One can also identify "arms" extending out from the nuclear region on both the eastern and western sides. NGC 7674 hosts a ∼0.7 kpc S-shaped radio jet (Momjian et al. 2003). Later Kharb et al. (2017) found evidence for the formation of the S-shaped radio structure as due to a black hole merger.

Zoom In Zoom Out Reset image size

Based on the [O iii], Hα, [S ii], and Hβ images, we created the BPT diagram and the resolved map (Figure 10). The nucleus and the eastern arm are dominated by red Seyfert activity. The loop and western arm show Seyfert or Seyfert-to-LINER type. The yellow LINER type permeates and surrounds the nucleus and those features. The blue H ii pixels are likely due to imperfections in the continuum subtraction rather than real star-forming regions, e.g., over-subtraction of Hβ due to cosmic rays on top of the nuclear region of the continuum image.

Zoom In Zoom Out Reset image size

NGC 2273 (z = 0.00614, D = 26.4 Mpc, 1'' = 127 pc) is a ringed galaxy with an inner ring and two pseudo-rings in the outskirt (Aguerri et al. 1998; Ferruit et al. 2000). Figure 11 shows the continuum-subtracted line maps covering line-emitting regions out to the inner ring. The central region shows a jet-like bar structure. The Hα and Hβ images exhibit bright emission components in the north. Mundell et al. (2009) show that the radio emission from NGC 2273 consists of an east–west double source and fainter, extended emission to the northeast and southwest. The jet-like structure in [O iii] extends along the same east–west direction as the radio emission, suggesting interactions between the radio jet and the NLR gas. The diffuse radio emission could be a poorly collimated outflow wind from the nucleus (Christopoulou et al. 1997; Mundell et al. 2009).

Zoom In Zoom Out Reset image size

Figure 12 displays the BPT diagram and the spatially resolved map. The nuclear region is dominated by red Seyfert activity and transitions to LINER type. A blue H ii region appears ∼08 to the north of the nucleus. NGC 2273 has been found to host nuclear starburst activity in the central 2'' × 2'' region based on spectral signatures (Gu et al. 2003).

Zoom In Zoom Out Reset image size

The BPT diagram of NGC 2273 displays a special feature that is not seen in other sources. The pixels that satisfy the strong line criterion are distributed in two groups on the BPT diagram, a horizontal group and a diagonal group, suggesting that they may be associated with distinct structures in the resolved map. We further separate the LINER groups in the BPT diagram using different colors (Figure 12, bottom panels). Magenta points denote the pixels with log([O iii]/Hβ) ≥ 0.38, and yellow points belong to the horizontal group with log([O iii]/Hβ) < 0.38. Indeed, the two groups are associated with different structures in the line-emitting region. Most magenta LINER-type pixels are adjacent to the central nuclear region, while most yellow LINER pixels are distributed farther out to the ring structure. We will discuss potentially different excitation mechanisms associated with these two distinct LINER groups in Section 5.

NGC 5643 (z = 0.00399, D = 17.1 Mpc, 1'' = 82 pc) has long been known to exhibit a one-sided ionization cone with filamentary structures (Simpson et al. 1997). Figure 13 shows the continuum-subtracted line maps in the inner r < 5'' (410 pc) region from the nucleus. Both the [O iii] and Hα maps display well-resolved filamentary structures to the east of the nucleus, while barely any emission is detected west of the nucleus. The filaments are also present in the [S ii] and Hβ maps but are much fainter. The Hβ image scale is stretched such that a dark region (nuclear dust structure; Cresci et al. 2015) west of the nucleus in the shape of a crescent moon can be recognized. Faint emission appears southwest of the dark region. Cresci et al. (2015; see also Mingozzi et al. 2019) reveal a more complete, double-sided ionization structure using deep VLT/MUSE integral field observations, although the spatial resolution is lower (088) than our HST observations.

Zoom In Zoom Out Reset image size

Figure 14 shows our HST BPT diagram and color-coded map. Most pixels that meet the 3σ criterion are located in the nuclear region, and are dominated by Seyfert activity. The relatively bright spots in the filamentary structures east of the nucleus also show red Seyfert-type excitation. The central region tends to transition from Seyfert to LINER, but there are no more pixels to reveal a complete LINER cocoon structure. Cresci et al. (2015) also present BPT maps based on their VLT/MUSE data. They not only observed the four lines required in the S–BPT diagram, but also obtained [N ii] and [O i] to make the N–BPT and O–BPT maps. The MUSE data are deep enough and have a large enough field of view to allow them to map the entire ionization bicone, which shows Seyfert activity. The ionization bicone is surrounded by LINER regions, which then transition farther out into the star-forming regions in the host galaxy.

Zoom In Zoom Out Reset image size

NGC 5643 possesses a diffuse, low-luminosity radio jet on both sides of the nucleus, extending ∼30'' long (Leipski et al. 2006). A central radio component overlaps with the optical nuclear emission shown in Figure 13, and the whole radio structure matches well with the full ionization cones shown in Cresci et al. (2015). They also confirm the presence of outflowing gas from the nuclear region based on the high-velocity ionized gas kinematics (further discussion in Section 5).

5.1.1. BPT Line Ratios Insensitive to Dust Extinction

5.1.2. Dust Distribution Affecting NLR/ENLR Morphology and Detectability

Since the ionization cone interiors of our sources are almost entirely Seyfert-type emission, one can simply draw the conclusion that the origin is predominantly due to AGN photoionization. However, fast shocks (v > 500 km s⁻¹), which have a photoionizing precursor, can also produce the Seyfert AGN photoionization-type line ratios on the BPT diagram (Allen et al. 2008; Kewley et al. 2019). For example, the S-shaped ionization structure in NGC 3393 is thought to be predominantly by AGN photoionization, but the S-shape could also be formed by fast shocks with photoionized precursors (Maksym et al. 2016, 2017).

LINERs are more complicated in terms of excitation mechanism or power source (e.g., Ho 2008; Yan & Blanton 2012). Some LINER emission is due to gas ionized by low-luminosity AGN (e.g., Ho et al. 1996; Constantin & Vogeley 2006) or regions exposed to absorbed or filtered nuclear radiation (e.g., Kraemer et al. 2008; Heckman & Best 2014). Further evidence of AGN in LINERs is supported by radio emission. More than half of nuclear LINERs have parsec-scale radio nuclei and sub-parsec radio jets (Heckman 1980; Nagar et al. 2005). In some other cases, shocks by jets or outflows are required to power the LINER emission. Extended LINER emission (i.e., >1 kpc) can be produced by shocks resulting from galactic-scale outflows (e.g., Dopita & Sutherland 1995; Ho et al. 2014, 2016). Slow shocks, which are unable to drive a precursor, can produce LINER-type line ratios on the BPT diagram (Rich et al. 2011; Sutherland & Dopita 2017; Kewley et al. 2019).

The green Seyfert–LINER transition zones at the edge of the ionization cones shown in our resolved maps in Section 4 are likely due to absorbed or filtered AGN ionization. Almost all of the Seyfert 2 galaxies in our sample host radio sources that overlap with the NLR structures, indicating interactions between the optical and radio components. The optical or radio emission could be alternatively explained as the interaction between an AGN disk wind and the surrounding interstellar medium (ISM) (e.g., Venturi et al. 2017; Panessa et al. 2019). Fischer et al. (2013) studied the NLR kinematics of nearby AGN, including the seven Seyfert 2 galaxies in this work, using HST/STIS optical spectroscopy. They found that Mrk 573 (see also Revalski et al. 2018), NGC 5643 (see also Cresci et al. 2015), and NGC 7674 exhibit clear outflow kinematics, while NGC 1386, NGC 2273, NGC 3081, and NGC 7212 show either complex or ambiguous kinematics. The LINER regions in the sources associated with clear radio jets or outflow signatures are most likely due to jet/outflow-induced shocks. Using long-slit spectroscopy, Storchi-Bergmann et al. (1996b, 1996a) found a correlation between shock-sensitive line ratios [N ii]/Hα, [S ii]/Hα and the FWHM line width of [N ii] in a sample of nearby AGN including NGC 3081, which indicates shocks contributing to the ionization of the gas in the circumnuclear region, covering the LINER cocoon region that we have identified.

In different regions of the same galaxy, the LINER emission can be attributed to totally different mechanisms. For example, NGC 2273 is the only galaxy in this sample that displays two distinct groups on the BPT diagram. The relatively high [O iii]/Hβ ratio LINER group is associated with the central optical emission and radio jet, and therefore the excitation mechanism is likely due to the jet–ISM interaction. The flat, relatively low [O iii]/Hβ ratio LINER group is located farther from the nucleus, associated with the ring structure, and not associated with radio jets or outflows, which could be explained as photoionization by absorbed or filtered AGN radiation or regions with high gas densities relative to the AGN ionizing flux. Since NGC 2273 has been found to host nuclear starburst activity, the starburst could also power the LINER emission.

However, one cannot separate excitation sources just based on optical diagnostics when shock excitation is involved (Kewley et al. 2019). In order to verify the regions we suspected to be shock excited, we would need high-resolution IFU spectroscopy and the following criteria must be satisfied as summarized by Kewley et al. (2019). (1) The velocity dispersion distribution is bimodal with the broad component being greater than 80 km s⁻¹. (2) The velocity dispersions correlate with shock-sensitive emission line ratios such as [N ii]/Hα, [S ii]/Hα, or [O i]/Hα. (3) The line ratios are consistent with the predictions from shock models. A 3D diagnostic diagram that combines radius as the x-axis, velocity dispersion as the y-axis, and line ratio as the z-axis is able to separate all three power sources, i.e., star formation, AGN, and shocks (D'Agostino et al. 2019).

Currently, such 3D diagnostics are not available for our sources. Chandra X-ray observations, in addition to radio observations, can provide complementary and critical direct measurements of AGN strength and spectroscopic diagnostics of shocks and photoionization, which can be directly compared to narrow-line images (e.g., Wang et al. 2009; Paggi et al. 2012; Maksym et al. 2017; Jones et al. 2020). For example in Mrk 573, Paggi et al. (2012) also found evidence for shocks at the jet–ISM interface based on Chandra X-ray data. A detailed comparison of optical, radio, and X-ray emission of the resolved structures in individual sources is beyond the scope of this paper, and will be presented in future publications.

With the HST's superior spatial resolution, we have identified a loop feature in at least two sources, NGC 3081 (Figure 5) and NGC 7674 (Figure 9). For NGC 3081, the loop feature is bright in [O iii] but is better resolved into small emission line knots in Hα. It also appears in [S ii] and Hβ although less prominently. The loop feature in NGC 7674 is prominent in [O iii] and also appears in Hα, although the resolution of the Hα image is lower and the loop is not well resolved. Both loops align more along the perpendicular direction of the ionization bicone than the bicone direction. The BPT classifications of these two loops are either Seyfert (NGC 3081) or a mixed of Seyfert, Seyfert–LINER transition, and LINER types (NGC 7674). The simplest explanation for the Seyfert-dominated regions is AGN photoionization. However, as discussed above, the origin of LINER emission is complicated and other mechanisms must play a role in shaping this special loop structure.

A very similar loop feature was identified in Seyfert galaxy IC 5063 with HST in the [S ii] and [N ii] line maps perpendicular to the main emission line structure (W. P. Maksym et al. 2020, in preparation). This loop could result from a plume of hot gas ablated by jet–ISM interactions from the galactic plane before escaping laterally. The loops in NGC 3081 and NGC 7674 share the same trend of arising more perpendicular to than along the ionization bicone axis. The loop in NGC 3081 also happens to coincide with the main bipolar outflow described by Schnorr-Müller et al. (2016) in their GMOS IFU study. Fischer et al. (2013) also show a bipolar outflow in NGC 7674 along the north–south direction, possibly intersecting the loop region. So there is probably some outflow–disk gas interaction that is exciting the loops in NGC 3081 and NGC 7674, although the interaction may not be as direct as for IC 5063. We would need additional diagnostics including kinematics in and around the loop region to investigate whether the loops in NGC 3081 and NGC 7674 are due to outflowing gas from jet–ISM interactions or AGN winds. Although less likely given the scales, we cannot rule out the possibility that these loop features are the consequences of post-merger events or tidal forces (e.g., Ji et al. 2014; Keel et al. 2015).

Traditional optical spectroscopic observations, by which these galaxies were classified, place narrow (∼1'') slits on their nuclei, and their nuclear spectra are classified as Seyfert 2 type. Long-slit spectra aligned with the bicone axes will tend to systematically miss the LINER regions surrounding the bicones. Given the common LINER emission in these traditionally classified Seyfert 2 galaxies, we further investigate whether this would affect BPT classification in galaxy surveys. Since we use the Kewley et al. (2006) BPT classification scheme, especially the Seyfert–LINER separation calibrated and validated on emission line galaxies selected from the SDSS, here we conduct a simple experiment to mimic the SDSS's 3'' diameter fiber and calculate the BPT line ratios ("+") using the integrated line fluxes measured within this 3'' diameter circular aperture centered on the nucleus.

The SDSS BPT classifications based on this experiment would be: NGC 5643, NGC 7212, and NGC 7674 remain Seyfert type; Mrk 573, NGC 1386, and NGC 3081 are in the Seyfert–LINER transition zone; only NGC 2273 falls in the LINER region of the BPT diagram.

This simple exercise demonstrates that when we talk about BPT galaxy classifications, it is important to specify the spatial scales (and resolution) associated with the measured BPT line ratios. Only spatially resolved observations like ours or with high-resolution IFUs can disentangle the emission types in different regions of the same galaxy, which are not necessarily in agreement with survey classifications. Admittedly, this simple exercise may not represent real SDSS spectroscopic observations, which requires a more detailed simulation that is beyond the scope of this paper.