The Limited Impact of Outflows: Integral-field Spectroscopy of 20 Local AGNs

We used the results of the ionized gas kinematics of ∼39,000 type 2 AGNs at z < 0.3 from Woo et al. (2016). The study has several advantages as an outflow census of type 2 AGNs. First, Woo et al. (2016) uniformly selected a large number of type 2 AGNs from the SDSS DR7 (Abazajian et al. 2009) using the emission-line diagnostics. Second, they measured the systemic velocity based on the stellar absorption lines, since the systemic velocity provided by the SDSS pipeline is uncertain because redshift measurements are partly based on emission-line features. Third, they provided the ionized gas kinematics by measuring the first moment (velocity) and the second moment (velocity dispersion) of [O iii].

We selected AGNs with strong outflow signatures from the parent sample of Woo et al. (2016) (Figure 1). First, we limited the redshift range to z < 0.1 in order to have enough spatial resolution and extent of outflows for the given IFU field of view (FOV). Then, we selected 3396 AGNs (∼9%) with a relatively high dust-corrected [O iii] luminosity (log ${L}_{[{\rm{O}}\mathrm{III}],\mathrm{cor}}\gt 41.5$ erg s⁻¹) since the outflow fraction increases with the [O iii] luminosity (e.g., Bae & Woo 2014; Woo et al. 2016). We further selected 491 AGNs (∼14% of the luminosity-limited sample) with strong outflow signatures in [O iii] kinematics, that is, with [O iii] velocity offset ${v}_{[{\rm{O}}{\rm{III}}]}$ > 100 km s⁻¹ or [O iii] velocity dispersion ${\sigma }_{[{\rm{O}}{\rm{III}}]}$ > 300 km s⁻¹. The selected AGNs are only ∼1% of the parent sample, consisting of the best candidates for outflow studies with local AGNs. Among the selected AGNs, we observed 18 AGNs with IFUs (see Table 1). Note that we observed two additional targets (J1100+1124 and J1520+0757), which satisfy the selection criteria for gas kinematics but with a slightly lower [O iii] luminosity (${L}_{[{\rm{O}}\mathrm{III}],\mathrm{cor}}\sim {10}^{41}$ erg s⁻¹). We also note that the AGNs presented in the pilot study by Karouzos et al. (2016a, 2016b) are a subsample with stronger outflow signatures than ours (e.g., ${L}_{[{\rm{O}}\mathrm{III}],\mathrm{cor}}\gt {10}^{42.0}$ erg s⁻¹).

Zoom In Zoom Out Reset image size

Table 1.  Targets and Observations

SDSS name	${\alpha }_{2000}$	${\delta }_{2000}$	Redshift	log ${L}_{[{\rm{O}}\mathrm{III}],\mathrm{cor}}$	Class	Obs. date	Telescope/Instrument	Seeing	texp
	hh mm ss.s	dd mm ss
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
J0130+1312	01 30 37.8	+13 12 52	0.07272	42.01	S	2014 Dec 26	Magellan/IMACS-IFU	0.9	2400
J0341−0719	03 41 34.9	−07 19 25	0.06603	41.90	S	2014 Dec 26	Magellan/IMACS-IFU	0.7	7200
J0806+1906	08 06 01.5	+19 06 15	0.09799	42.34	S	2014 Dec 26	Magellan/IMACS-IFU	0.8	3600
J0855+0047	08 55 47.7	+00 47 39	0.04185	41.63	S	2013 Apr	VLT/VIMOS-IFU	1.0	10050
J0857+0633	08 57 59.0	+06 33 08	0.07638	41.74	S	2014 Apr 06	Magellan/IMACS-IFU	1.0	7200
J0911+1333	09 11 24.2	+13 33 20	0.07966	42.17	S	2014 Dec 27	Magellan/IMACS-IFU	0.8	5400
J0952+1937	09 52 59.0	+19 37 55	0.02445	41.52	S	2014 Apr 05	Magellan/IMACS-IFU	1.0	3600
J1001+0954	10 01 40.5	+09 54 32	0.05638	42.11	S	2014 Apr 06	Magellan/IMACS-IFU	1.0	5400
J1013−0054	10 13 46.8	−00 54 51	0.04258	42.18	S	2014 Apr 04	Magellan/IMACS-IFU	1.0	3600
J1054+1646	10 54 23.8	+16 46 53	0.09498	42.39	L	2014 Apr 27	Magellan/IMACS-IFU	0.7	7200
J1100+1321	11 00 03.9	+13 21 50	0.06420	42.10	S	2014 Dec 26	Magellan/IMACS-IFU	0.8	7200
J1100+1124	11 00 37.2	+11 24 55	0.02700	40.88	S	2013 Apr	VLT/VIMOS-IFU	0.8	8040
J1106+0633	11 06 30.7	+06 33 34	0.04044	41.90	S	2013 Apr	VLT/VIMOS-IFU	0.8	8040
J1147+0752	11 47 20.0	+07 52 43	0.08271	42.33	L	2014 Dec 28	Magellan/IMACS-IFU	0.9	7200
J1214−0329	12 14 51.2	−03 29 22	0.03382	42.41	S	2014 Apr 04	Magellan/IMACS-IFU	0.7	3600
J1310+0837	13 10 57.3	+08 37 39	0.05211	41.76	S	2014 Apr 04	Magellan/IMACS-IFU	1.0	3600
J1440+0556	14 40 18.0	+05 56 34	0.06105	42.45	S	2014 Apr 06	Magellan/IMACS-IFU	1.1	5400
J1448+1055	14 48 38.5	+10 55 36	0.08930	42.83	S	2014 Apr 06	Magellan/IMACS-IFU	1.3	5400
J1520+0757	15 20 33.7	+07 57 12	0.04343	41.11	S	2014 Apr 06	Magellan/IMACS-IFU	1.3	5400
J2039+0033	20 39 07.0	+00 33 16	0.04835	42.15	S	2014 May 21	Magellan/IMACS-IFU	1.0	10800

Notes. (1) SDSS name of AGN; (2) R.A. (J2000); (3) Decl. (J2000); (4) redshift derived from the brightest spaxel of observed spectra; (5) extinction-corrected [O iii] luminosity in logarithm (erg s⁻¹); (6) class from the emission-line diagnostics, i.e., Seyfert (S), and LINER (L); (7) date of observation (local); (8) telescope and instrument; (9) seeing size (''); (11) exposure time (seconds).

Download table as:  ASCIITypeset image

We calculate the AGN bolometric luminosity ${L}_{\mathrm{bol}}$ as $\mathrm{log}{L}_{\mathrm{bol}}=3.8+0.25\mathrm{log}{L}_{[{\rm{O}}\mathrm{III}],\mathrm{cor}}+0.75\mathrm{log}{L}_{[{\rm{O}}{\rm{I}}],\mathrm{cor}}$, where ${L}_{[{\rm{O}}{\rm{I}}],\mathrm{cor}}$ is the extinction-corrected [O i]λ6300 luminosity (Netzer 2009). We infer the black hole mass (${M}_{\mathrm{BH}}$) by adopting the ${M}_{\mathrm{BH}}-{\sigma }_{\star }$ relation (Park et al. 2012), where ${\sigma }_{\star }$ is the stellar velocity dispersion obtained from the MPA-JHU catalog of SDSS DR7.⁷ We find that AGNs with strong outflows show no significant difference in the distributions of redshift, Eddington ratio, and emission-line ratios (e.g., Baldwin et al. 1981), compared to the AGNs in the luminosity-limited sample (Figure 2). Thus, we assume that the selected AGNs are a random subsample of the luminosity-limited sample.

Zoom In Zoom Out Reset image size

We observed 17 out of 20 type 2 AGNs with the IFU of the Inamori Magellan Areal Camera (IMACS-IFU) on the Magellan telescope at Las Campanas Observatory. We used the $f/4$ (long mode) camera with a recently upgraded Mosaic 3 CCD (8k × 8k). We used the 300 lines/mm grating with a tilt angle of 60, providing a large wavelength range of 3420–9550 Å with a spectral resolution R ≈ 1600. The choice of $f/4$ camera provides an FOV of 5'' × 45 with 600 fibers, and each spaxel (spatial pixel) has a size of 02 in diameter on the sky. Given the range of redshift of our targets, we explore the central region within several kiloparsecs from the nucleus, which provides detailed information on the gas outflows in the central region of the host galaxies.

The observations were performed during seven nights in 2014 April, May, and December with a seeing of 07–13 in April and May and 07–09 in the December run. In each afternoon, we obtained bias images, HeNeAr arcs, dome flats, and sky flats if possible. During the night, we initially took an acquisition image with the IFU to make sure the target was located at the center of the FOV. During our first run (April 6), however, the acquisition was poorly performed because of an unexpected spatial offset of the IFU images on the CCD. Hence the targets of the night failed to be located at the center of the FOV. For the other nights, after locating the target at the center of the FOV, we took science exposures and an arc image at the same position of the telescope. For each observing run, we observed spectrophotometric standard stars for flux calibration.

For the IMACS-IFU data reduction, we utilized the IDL-based software P3D,⁸ which is a general-purpose data-reduction tool for fiber-fed IFUs (Sandin et al. 2010). Since P3D does not officially support the IMACS-IFU data reduction yet, we modified the parameter files in the code for the IMACS-IFU instrument setup. First, we constructed a master bias frame, then subtracted the bias frame from all exposures. We also constructed a combined dome-flat image with the IFU as a reference frame to trace the dispersed light. Using the combined dome-flat image, we obtained a tracing solution for each target. Since we did not obtain a dome-flat image after each science exposure, we manually shifted the dome-flat image for each science exposure to properly trace the dispersed light. The shift is, on average, ∼5 pixels on the CCD in the cross-dispersion direction (y axis). After we obtained the individual tracing solutions, we applied the solution to the HeNeAr arc frame obtained after each target in order to get a wavelength solution. We used a fifth-order polynomial function to obtain the wavelength solution, resulting in a residual rms of ∼0.02 Å.

Then we removed cosmic rays on each science frame by using the PyCosmic routine, developed for a robust cosmic-ray removal of fiber-fed IFU data (Husemann et al. 2012). After the cosmic-ray removal, we extracted the object and sky spectra by using the tracing solutions. Choosing among several methods for spectrum extraction, we used the modified optimal extraction with a predefined Gaussian function from the tracing solution, which gives less noise in spectra and more robust results than the top-hat extraction (Horne 1986; Sandin et al. 2010). IMACS has an internal atmospheric dispersion corrector, so we did not correct the atmospheric dispersion during the reduction because we do not find any wavelength-dependent spatial offset in the final constructed image. After the spectra extraction, we obtained a mean spectrum of sky background by using dedicated sky fibers and subtracted the mean sky spectrum from each science spectrum. Finally, we combined all science exposures and calibrated the flux with a sensitivity function obtained from standard star observations. In addition, to increase the signal-to-noise ratio (S/N) in the outskirts of the FOV, we binned seven adjacent spaxels into a single spectrum. Since the combined spaxels have a size of 06 in diameter on the sky, we still have spaxels smaller than the seeing (07–13) during the observing runs.

For three out of 20 AGNs, we obtained the data with the multipurpose optical instrument VIMOS in IFU mode on the VLT-UT3 (Program ID: 091.B-0343(A), PI: Flohic). We used the MR grating with the GG475 filter, providing a spectral range of 4900–10150 Å with a spectral resolution of ∼720. The VIMOS-IFU has 6400 fibers (80 × 80) without dedicated sky fibers. With the adjustable scales of 067 per fiber, we have a large FOV of 27'' × 27'', which is a factor of ∼30 larger than our IMACS-IFU FOV. Thus, the VIMOS-IFU data allowed us to investigate the impact of gas outflows on galactic scales of ∼10 kpc for the redshift range of our targets. The observing run was executed as a service-mode observation in 2013 April. The exposure time was ∼2.8 hr per target, with 15 × 670 s exposures. We applied a spatial offset of ±2'' for science exposures to compensate for dead or bad fibers in the array.

The reduction for the VLT/VIMOS-IFU data was performed with P3D in the same manner as for the IMACS-IFU data. After that, we combined the multiple exposures by considering the spatial offset of ±2''. For both J1100+1124 and J1106+0633, we discarded three exposures that suffered from a wrong spatial offset during the observations.

For each spaxel, we measured the flux and velocities of emission lines after subtracting the stellar continuum, as similarly performed by Woo et al. (2016). For the stellar continuum, we utilized the pPXF code (Cappellari & Emsellem 2004) to construct the best-fit model using the MILES simple stellar population models with solar metallicity (Sánchez-Blázquez et al. 2006). In this procedure we measured the velocity of the stellar component in each spaxel.

Then, we fit the narrow emission lines, for example, the Balmer lines, [O iii]λ5007, [N ii] doublet, and [S ii] doublet, by utilizing the MPFIT code (Markwardt 2009). Since AGNs with gas outflows generally show a broad wing component, especially in the [O iii] line profile (Greene & Ho 2005; Mullaney et al. 2013; Shen & Ho 2014; Woo et al. 2016), we used two Gaussian functions to represent the broad and narrow components. For the Hα+[N ii] region, we assumed that each narrow and broad wing component of Hα and [N ii] has the same kinematics, that is velocity and velocity dispersion. We also fixed the flux ratio of [N ii]λ6549 and [N ii]λ6583 as one-third for both broad and narrow components (Figure 3).

Zoom In Zoom Out Reset image size

To ensure that the broad component is not fitting the noise in the spectrum, we adopted two conditions for accepting the broad wing component. First, the peak amplitude of the broad component should be a factor of 3 larger than the standard deviation of the residual spectra at 5050–5150 Å. For the Hα+[N ii] region, we choose a larger amplitude of either the Hα or [N ii] broad component as the peak amplitude. Second, the sum of the width (σ) of the two Gaussian components should be smaller than the distance between the peaks of the two Gaussian components. If the broad component is not detected in the fitting, we alternatively fit the emission lines with a single Gaussian function.

Based on the model fit of the emission lines, we calculated the first moment ${\lambda }_{0}$ and the second moment ${\sigma }_{\mathrm{line}}$ of each emission line, which respectively represent the luminosity-weighted velocity and velocity dispersion of the emission line, as

$$\begin{eqnarray}&&{\lambda }_{0}=\displaystyle \frac{\int \lambda f(\lambda )d\lambda }{\int f(\lambda )d\lambda },\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&{\sigma }_{\mathrm{line}}^{2}=\displaystyle \frac{\int {\lambda }^{2}f(\lambda )d\lambda }{\int f(\lambda )d\lambda }-{\lambda }_{0}^{2},\end{eqnarray} \tag{ 2 }$$

where $f(\lambda )$ is the flux at each wavelength λ. We calculated the velocity shift of emission lines by comparing the first moment and the systemic velocity measured from the center of host galaxies. The second moment was corrected for the instrumental resolution measured from the sky emission lines. To estimate the uncertainties for the measurement, we adopted a Monte Carlo realization generating 100 mock spectra by randomizing the flux with noise, and we obtained the measurements from each spectrum. Then we adopted a 1σ dispersion of the distribution of each measurement as the uncertainty.

4.1.1. Morphology

We perform a visual inspection of the spatial distribution of the [O iii] flux. The morphologies are, in general, in good agreement with the flux distribution of the stellar component, showing no noticeable biconical outflow features (Figure 4). For 12 out of 20 AGNs, however, the [O iii] flux distribution shows a lopsidedness from the stellar center of the host galaxy (Figure 5), although it is relatively uncertain due to the poor sampling (06 spaxel in diameter; see Section 2) and the seeing size (07–13). In general, the spaxel with the maximum [O iii] flux has an offset of ∼1 spaxel from the center of the galaxy. Such a lopsidedness in the [O iii] flux distribution is possibly due to the dust obscuration of either the approaching or receding component of biconical outflows (e.g., Crenshaw et al. 2010; Bae & Woo 2014; Fischer et al. 2014; Bae & Woo 2016).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

By performing the kinematic decomposition of [O iii] as described in Section 3.1, we find that 15 our of 20 AGNs have a detectable broad component in the line profile. After separating the narrow and broad components from the line profile, we examine the morphologies of the [O iii] flux distribution of the two components separately (middle and bottom panels in Figure 4, respectively). First, the spaxels with a broad component have a smaller extent than those with a narrow component. This is mainly due to difficulties in detecting a broad component from low S/N spectra at the outskirts of the FOV. Second, the spaxels containing the broad component show diverse morphologies. Among the 15 broad-component-detected AGNs, eight AGNs show a compact, round-shaped morphology, while the other seven AGNs show an irregular (e.g., patchy or elongated) shape.

4.1.2. Kinematics

We present the kinematic properties of [O iii] based on its total, narrow, and broad components. First, we examine the 2D maps of the line-of-sight velocity structure of the [O iii]-emitting region (Figure 6). When we examine the total profile of [O iii], we find that 18 AGNs show blueshifted (14) or redshifted (four) velocity shifts within the central kiloparsec or a more extended region, compared to the systemic velocity of the host galaxy. One AGN (J1013–0054) shows a rotational kinematics, and another AGN (J1440+0556) shows no or little velocity offset beyond the uncertainty in velocity offset measurement within the central kiloparsec region.

Zoom In Zoom Out Reset image size

When we focus on the broad and narrow components separately, the velocity structure looks different from the total component. The majority (10 out of 15) of broad-component-detected AGNs show negative velocity offset, while four AGNs show positive velocity offset and one AGN shows ambiguous velocity structure. The result is consistent with the model predictions of biconical outflows and dust obscuration, which expect a larger number of AGNs with negative velocity offset than of those with positive velocity offset (Bae & Woo 2016). For the 15 broad-component-detected AGNs, the spaxels with a narrow component show mostly either negative velocity offset (five), positive velocity offset (two), rotational features (seven), or no or little velocity offset (one) after removing the broad component (see Table 2).

Table 2.  Characteristics of the Spatially Resolved Velocity Fields of the NLR

	[O iii]			Hα
SDSS name	V[O iii],T	V[O iii],N	V[O iii],B	${V}_{{\rm{H}}\alpha ,T}$	${V}_{{\rm{H}}\alpha ,N}$	${V}_{{\rm{H}}\alpha ,B}$	Disk type
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
J0130+1312	blue	blue	blue	rotation	rotation	rotation	SF
J0341−0719	red	rotation	red	rotation	rotation	rotation	SF
J0806+1906	blue	blue	red	rotation	rotation	rotation	AGN
J0855+0047	blue	blue	blue	blue	rotation	blue	SF
J0857+0633	red	red	⋯	rotation	rotation	⋯	AGN
J0911+1333	blue	blue	blue	blue	rotation	blue	SF
J0952+1937	blue	blue	⋯	rotation	rotation	blue	SF
J1001+0954	blue	rotation	blue	red	rotation	red	SF
J1013−0054	rotation	rotation	ambiguous	rotation	rotation	rotation	SF
J1054+1646	blue	blue	blue	blue	rotation	blue	AGN
J1100+1321	blue	rotation	blue	blue	rotation	blue	SF
J1100+1124	red	red	⋯	red	red	⋯	⋯
J1106+0633	blue	blue	⋯	blue	blue	blue	⋯
J1147+0752	blue	rotation	blue	blue	rotation	blue	SF
J1214−0329	blue	rotation	blue	blue	rotation	blue	SF
J1310+0837	red	red	red	rotation	rotation	⋯	AGN
J1440+0556	systemic	systemic	red	red	red	red	⋯
J1448+1055	blue	red	blue	blue	rotation	blue	AGN
J1520+0757	blue	blue	⋯	blue	blue	blue	⋯
J2039+0033	blue	rotation	blue	rotation	rotation	rotation	SF

Notes. (1) The name of AGN; (2) the characteristics of the velocity field based on the total profile of [O iii]; (3) the same as (2) but based on the narrow component of [O iii]; (4) the same as (2) but based on the broad component of [O iii]; (5) the same as (2) but for the total profile of Hα; (6) the same as (2) but based on the narrow component of Hα; (7) the same as (2) but based on the broad component of Hα; (8) type of disk seen in the narrow component of Hα.

Download table as:  ASCIITypeset image

Second, we examine the 2D maps of velocity dispersion of [O iii] (Figure 7). The maps of total [O iii] profile provide a hint for the mixture of narrow and broad components in the FOV. The range of velocity dispersion in the central region is from ∼200 to ∼600 km s⁻¹, which is much larger than the stellar velocity dispersions of the host galaxy. As we separate the broad and narrow components, the velocity dispersion of the broad component becomes larger up to ∼800 km s⁻¹, while that of the narrow component is, in general, broadly consistent with the stellar velocity dispersion. In the maps of the narrow component, we also find spaxels with a relatively large velocity dispersion at the boundary of spaxels with a broad component, which is possibly due to an unremoved broad component in the line profile.

Zoom In Zoom Out Reset image size

Third, we investigate the radial distribution of the [O iii] velocity and velocity dispersion for each component (Figure 8). In most cases, the absolute velocities of the broad components tend to become smaller as a function of distance from the center, while those of the narrow components show a flat distribution as a function of distance. The velocity dispersions of the broad component are at least a few times larger than the stellar velocity dispersion and are also decreasing as a function of distance. In J0911+1333, for example, the velocity dispersions of the broad component are a factor of 4−6 larger than the stellar velocity dispersion. Such large velocity dispersions indicate a nongravitational origin, that is, AGN outflows. In comparison, the velocity dispersions of the narrow component are larger than the stellar velocity dispersion by a factor of ∼2 at the center and become comparable to the stellar velocity dispersion at larger distance. We find that the trends in velocity dispersion of narrow and broad components are qualitatively similar for all 15 broad-component-detected AGNs.

Zoom In Zoom Out Reset image size

Last, we compare the [O iii] velocity and velocity dispersion for each component (Figure 9). We clearly see that the narrow and broad components are located in a different locus on the velocity–velocity dispersion (VVD) diagram. For example, J0911+1333 shows narrow components (blue dots) with larger velocity offsets compared to the broad components (red dot), while the velocities obtained from the total profile show a velocity in the middle of the velocities for the narrow and broad components. The narrow components are located within 300 km s⁻¹ of the velocity offset, while the broad components show a blueshifted velocity offset from −500 to −100 km s⁻¹. The mean velocity dispersion of the narrow component is ∼200 km s⁻¹, while that of the broad component is ∼550 km s⁻¹. The rotational features and smaller range of velocity dispersion indicate that the narrow-component kinematics are related to the gravitational potential of the host galaxy, while the broad-component kinematics are due to nongravitational phenomena.

Zoom In Zoom Out Reset image size

4.2.1. Morphology

We perform a visual inspection of the morphology of the Hα flux distribution (Figure 10). Similarly to [O iii], the morphology of the Hα flux distribution follows the flux distribution of the stellar component in the host galaxy. We also find a lopsidedness in Hα flux in six out of 20 AGNs (Figure 11). Interestingly, such asymmetry is more frequently found in [O iii] (12 out of 20) than in Hα. All objects with Hα flux lopsidedness also show [O iii] flux lopsidedness, implying that what causes the lopsidedness in the flux distribution is commonly affecting both Hα- and [O iii]-emitting regions, but is less significant in Hα.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

By performing a multi-Gaussian decomposition for the Hα+[N ii] lines, we find that 19 out of 20 AGNs have a detectable broad component in the Hα line profile (bottom panel of Figure 10). Similarly to [O iii], the spatial distribution of the broad Hα component is smaller than that of the narrow component. Also, the morphologies of the spaxels with a Hα broad component are mostly irregular, but are somewhat comparable to those of the [O iii] broad component. We find that all AGNs with an [O iii] broad component (15) also have a broad component in Hα, implying a common physical origin for both broad components in [O iii] and Hα. Also, there are four AGNs having a broad component in Hα but not in [O iii]. Among them, three AGNs have a lower S/N in [O iii] than in Hα, while the remaining AGN (J1106+0633) has comparable S/N in [O iii] and Hα. For this object, it is possible that the [O iii] broad component was not revealed by our VLT/VIMOS observations, since the configuration has a low spectral resolution (R ∼ 720).

4.2.2. Kinematics

We examine the velocity structure of the Hα-emitting region (Figure 12). When we focus on the total component of Hα, we find that 13 AGNs show either a negative (nine) or positive (four) velocity offset, while the other seven AGNs show velocity structures similar to those of the stellar component, that is, rotational or systemic velocity. If we examine the narrow component of Hα, we find that the velocity of the narrow Hα is consistent with systemic or rotational velocity in most objects (16/20). Among the remaining four AGNs, two AGNs show negative velocity offset, and the other two AGNs show positive velocity offset. After removing the broad component in the total profile, the velocity map of the narrow component of Hα has the signature of Keplerian disk rotation (see Table 2). Rotational disk features are found in the narrow component of Hα of 16 AGNs in our sample. Seven AGNs have a rotational disk feature in the narrow component of both Hα and [O iii], which are consistent with one another.

Zoom In Zoom Out Reset image size

Similarly to [O iii], the maps of the velocity dispersion of the total Hα show clear signs of the mixture of narrow and broad components in the central region (Figure 13). The large velocity dispersion and spatial concentration of the broad component in Hα also support its nongravitational origin. After removing the broad component, the narrow components are, in general, consistent with the stellar velocity dispersion of the host galaxy.

Zoom In Zoom Out Reset image size

Also, the distributions of Hα velocity and velocity dispersion as a function of distance are qualitatively similar to those of [O iii] (Figure 14), while the scales of the Hα velocity and velocity dispersion are generally smaller than those in [O iii]. Similarly, the narrow and broad components are clearly separated in the VVD diagram, as we found in the [O iii] (Figure 15).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The size of the NLR $({R}_{\mathrm{NLR}})$ is typically measured based on the flux distribution of the emission lines, without taking into account the outflow properties (e.g., Bennert et al. 2002; Schmitt et al. 2003; Liu et al. 2013a). To quantify the size of the outflows, we defined outflow size ${R}_{\mathrm{out}}$, which is the radius where the [O iii] velocity dispersion is equal to the stellar velocity dispersion of its host galaxy (Karouzos et al. 2016a). Here we use the two sizes, $({R}_{\mathrm{NLR}})$ and $({R}_{\mathrm{out}})$, to investigate the properties of the NLR and outflow kinematics, respectively.

First, we measure the flux-weighted mean size of the NLR (Husemann et al. 2014) for ${R}_{\mathrm{NLR}}$ as

$$\begin{eqnarray}&&{R}_{\mathrm{NLR}}=\displaystyle \frac{\int {Rf}(R){dR}}{\int f(R){dR}},\end{eqnarray} \tag{ 3 }$$

where R is the distance from the center and f(R) is the flux at a given distance. We only use the spaxels classified as composite or AGN from the emission-line diagnostics (e.g., Baldwin et al. 1981; Kauffmann & Heckman 2009). Then, we correct for seeing effects by subtracting the seeing size in quadrature, resulting in a ∼14% decrease in size. Note that ${R}_{\mathrm{NLR}}$ for all targets is resolved compared to the seeing size. The mean ${R}_{\mathrm{NLR}}$ is ∼880 ± 360 pc for our sample with the mean ${L}_{[{\rm{O}}{\rm{III}}]}$ of ${10}^{40.7}$ erg s⁻¹, where ${L}_{[{\rm{O}}{\rm{III}}]}$ is extinction-uncorrected [O iii] luminosity (see Table 3).

Table 3.  ${L}_{[{\rm{O}}{\rm{III}}]}$ and the Sizes of the NLR and Outflows of the Type 2 AGNs

SDSS name	log ${L}_{[{\rm{O}}{\rm{III}}]}$	log ${R}_{\mathrm{NLR}}$	log ${R}_{\mathrm{out}}$
(1)	(2)	(3)	(4)
J0130+1312	40.63	2.98 ± 0.20	2.37a
J0341−0719	40.66	2.90 ± 0.21	3.14 ± 0.08
J0806+1906	40.91	3.22 ± 0.15	2.60a
J0855+0047	40.41	3.16 ± 0.16	3.75 ± 0.08
J0857+0633	40.35	3.14 ± 0.14	2.69 ± 0.08
J0911+1333	41.01	2.99 ± 0.21	3.48 ± 0.08
J0952+1937	40.05	2.67 ± 0.15	3.01 ± 0.08
J1001+0954	40.79	2.91 ± 0.18	3.33 ± 0.08
J1013−0054	40.54	2.79 ± 0.18	2.82 ± 0.08
J1054+1646	41.06	2.93 ± 0.28	3.44 ± 0.08
J1100+1321	40.91	2.88 ± 0.22	3.28 ± 0.08
J1100+1124	39.55	3.11a	3.37 ± 0.08
J1106+0633	40.77	2.96 ± 0.17	3.73a
J1147+0752	41.28	2.77 ± 0.35	3.33 ± 0.08
J1214−0329	40.76	2.55 ± 0.25	2.80 ± 0.08
J1310+0837	40.89	2.86 ± 0.19	2.64 ± 0.08
J1440+0556	40.59	2.71 ± 0.31	3.02 ± 0.08
J1448+1055	41.31	2.97 ± 0.24	3.46 ± 0.08
J1520+0757	39.75	2.66 ± 0.25	3.04 ± 0.08
J2039+0033	40.90	2.86 ± 0.18	2.87 ± 0.08

Notes. (1) Name of the AGN; (2) extinction-uncorrected [O iii] luminosity in logarithm with uncertainty (erg s⁻¹); (3) the flux-weighted size of the NLR in logarithm with uncertainty (pc); (4) the kinematic size of outflows in logarithm with uncertainty (pc). Note that we adopt a different cosmology for the sizes and luminosity calculation as H_o = 70 km s⁻¹ Mpc⁻¹, ${{\rm{\Omega }}}_{{\rm{\Lambda }}}$ = 0.70, and ${{\rm{\Omega }}}_{m}$ = 0.30, in order to be consistent with the results in the literature.

^aUpper limit of the sizes.

Download table as:  ASCIITypeset image

To estimate the uncertainty in ${R}_{\mathrm{NLR}}$, we construct 100 mock spatial distributions of the [O iii] region by randomizing the flux for each spaxel including noise, and we measure ${R}_{\mathrm{NLR}}$ in the same manner. Then we adopt 1σ of the distribution as the uncertainty of ${R}_{\mathrm{NLR}}$. In addition to this uncertainty, we add 10% of ${R}_{\mathrm{NLR}}$ to account for the uncertainty in the seeing size, and we also add a half spaxel size (∼03) to take into account the uncertainties from the spatial sampling. Note that ${R}_{\mathrm{NLR}}$ of J1100+1124 is regarded as the upper limit of ${R}_{\mathrm{NLR}}$ since the target is observed using VLT/VIMOS, which has a large FOV, and the flux-weighted size includes some spaxels of spiral arms classified as the composite region. These spaxels might be contaminated with shock-induced line emissions in the emission-line diagnostics.

Second, we measure the outflow size ${R}_{\mathrm{out}}$ based on the 1D distributions of the [O iii] velocity dispersion as a function of distance from the center (Figure 16). We obtain the mean value of the [O iii] velocity dispersion from the spaxels as a function of distance, then determine the radius where the [O iii] velocity dispersion is equal to the stellar velocity dispersion of the host galaxy, after linearly interpolating the mean values of the [O iii] velocity dispersion (orange lines). We also correct the measured size by subtracting the seeing size in quadrature. We assume 20% uncertainty in ${R}_{\mathrm{out}}$. We obtain that the mean ${R}_{\mathrm{out}}$ is ∼1800 pc, which is about a factor of 2 larger than the mean ${R}_{\mathrm{NLR}}\sim 880$ pc (see Table 3). The result is in good agreement with the result from Karouzos et al. (2016a), who reported that ${R}_{\mathrm{out}}$ is a few times larger than ${R}_{\mathrm{NLR}}$. Four AGNs (J0130+1312, J0806+1906, J0857+0633, and J1310+0837), however, show smaller ${R}_{\mathrm{out}}$ than ${R}_{\mathrm{NLR}}$ due to their low [O iii] velocity dispersion compared to the stellar velocity dispersion of the host galaxy, indicating that stellar velocity dispersion may not be the best parameter to separate the nongravitational outflow signatures from the gravitation kinematics.

Zoom In Zoom Out Reset image size

Based on our measurements of both ${R}_{\mathrm{NLR}}$ and [O iii] luminosity, here we present the size–luminosity relationship (Figure 17). Since the relationships for type 1 and type 2 AGNs are in good agreement with one another (Schmitt et al. 2003), we also include 29 high-luminosity type 1 AGNs from the literature with mean ${L}_{[{\rm{O}}{\rm{III}}]}={10}^{42.4}$ erg s⁻¹(Husemann et al. 2013, 2014) for which the R_NLR was measured in a way that is consistent with our study. We also include six type 2 AGNs obtained from the Gemini/GMOS-IFU (Karouzos et al. 2016a, 2016b). For this comparison, we use the extinction-uncorrected [O iii] luminosity and apply the cosmological parameters used in the work of Husemann et al. (2014): ${{\rm{\Omega }}}_{{\rm{\Lambda }}}$ = 0.70 and ${{\rm{\Omega }}}_{m}$ = 0.30. We assume 10% uncertainty in ${L}_{[{\rm{O}}{\rm{III}}]}$. To fit the size–luminosity relation, we apply a forward regression method using the FITEXY code in the IDL library (e.g., Park et al. 2012), obtaining

$$\begin{eqnarray}&&\mathrm{log}{R}_{\mathrm{NLR}}=(0.41\pm 0.02)\times \mathrm{log}{L}_{[{\rm{O}}{\rm{III}}]}-(14.00\pm 0.77).\end{eqnarray} \tag{ 4 }$$

The slope ∼0.41 ± 0.02 is consistent with the slope reported by Husemann et al. (2014; 0.44 ± 0.06).

Zoom In Zoom Out Reset image size

The size–luminosity relation has been reported with different slopes based on different samples and methods, resulting in various physical interpretations for the relation. For example, Schmitt et al. (2003) found a slope of 0.33 ± 0.04 for local type 1 and 2 Seyfert galaxies by measuring the size and estimating the [O iii] luminosity from the HST narrowband imaging data. In addition, Liu et al. (2013a) found a slope of 0.25 ± 0.02 for type 2 quasars and Seyfert galaxies based on heterogeneous data from IFU and long-slit observations. To explain the physical conditions in the NLR of the sources, they argued that the pressure inside NLR clouds (P) and the gas density (n) drop as radius r increases, that is, $P(r)\propto {r}^{-2}$ and $n(r)\propto {r}^{-2}$, resulting in an ionization parameter (U) independent of radius. In contrast, other studies reported a slope of ∼0.5 (Bennert et al. 2002; Husemann et al. 2014; Hainline et al. 2013). For example, Bennert et al. (2002) found a slope of 0.52 ± 0.06 for luminous Seyferts and quasars based on HST narrowband imaging data. Similarly, Hainline et al. (2013) reported a slope of 0.4–0.5 for type 2 quasars and Seyfert galaxies, but they used the luminosity at 8 μm as a proxy of AGN luminosity rather than [O iii] luminosity, arguing that the AGN luminosity is more directly traced by the luminosity of 8 μm than of [O iii]. To explain the slope of ∼0.5, these studies adopted a simple model that assumes a constant ionization parameter and the density for the clouds, which is not the case for our sample (see Section 5.1).

Similar to the photoionization size (${R}_{\mathrm{NLR}}$)–luminosity relation, we examine the relationship between ${R}_{\mathrm{out}}$ and [O iii] luminosity (Figure 18). Although the dynamic range of [O iii] luminosity is rather small in our sample, we find no clear relationship between ${R}_{\mathrm{out}}$ and [O iii] luminosity, presumably due to the dynamical timescale of the outflows. If we consider the relatively low outflow velocity ∼1000 km s⁻¹ and ${R}_{\mathrm{out}}\,\sim$ 1–2 kpc, it will take (1–2) × ${10}^{6}$ yr to reach ${R}_{\mathrm{out}}$ for the outflow, which is much larger than the photoionization timescale for the NLR. As a result, ${R}_{\mathrm{out}}$ may not show a clear relationship with ${L}_{[{\rm{O}}{\rm{III}}]}$, while ${R}_{\mathrm{NLR}}$ does. To further constrain whether there is a positive relationship between ${R}_{\mathrm{out}}$ and [O iii] luminosity, we may need to obtain ${R}_{\mathrm{out}}$ from higher luminosity type 2 AGNs.

Zoom In Zoom Out Reset image size

In previous sections, we find both gravitational and nongravitational kinematics in the NLR, and the nongravitational kinematics are closely related to rotation, as we noticed from the velocity maps of the narrow component of Hα. The properties of rotational disks in the sample are worth investigating, since they might provide useful hints on AGN feedback and coevolution.

By using the integrated spectra within ${R}_{\mathrm{NLR}}$, we classify the disk represented by the narrow Hα into two groups: (1) SF-type (log [N ii]/Hα ≥ 0) and (2) AGN-type (log [N ii]/Hα < 0). Eleven AGNs are classified as SF-type, five AGNs are classified as AGN-type, while four AGNs have no or ambiguous rotation in Hα (summarized in Table 2). We use the line ratio of the narrow components of Hα and [N ii] for disk classification, since we clearly see the rotational feature in the narrow component in Hα but not in [O iii], which strongly represents nongravitational kinematics.

For the two groups, we compare the ${D}_{n}(4000)$ and ${\rm{H}}{\delta }_{A}$ indices of the central region of AGNs (3'' in diameter). We obtained the indices from the MPA-JHU catalog of SDSS DR7 galaxies. The two groups show clearly different ranges of indices. The SF-type group has ${D}_{n}(4000)=1.37\pm 0.13$ and ${\rm{H}}{\delta }_{A}=2.83$ ± 1.32, while the AGN-type group has ${D}_{n}(4000)=1.71$ ± 0.15 and ${\rm{H}}{\delta }_{A}=-0.08$ ± 0.92, showing that the SF-type group has smaller ${D}_{n}(4000)$ and stronger ${\rm{H}}{\delta }_{A}$ than that of the AGN-type group, as expected.

We also compare the distribution of specific star formation rate (SSFR) as a function of the stellar mass (M_*) (Figure 19). We also adopted the SSFR and M_* from the MPA-JHU catalog of SDSS DR7 galaxies. The SSFRs for the whole galaxy were estimated by combining the synthetic models on the integrated spectra and photometric information (Brinchmann et al. 2004), which provides SFRs sensitive over the past 10⁸–10⁹ yr (Kennicutt 1998). The AGNs with SF-type disks (blue dots) have higher SSFRs (log SSFR = −10.0 ± 0.4 yr⁻¹) and smaller stellar mass (log M_* = 10.6 ± 0.3 M_⊙), while the AGNs with AGN-type disks (red dots) have smaller SSFRs (log SSFR = −11.3 ± 0.7 yr⁻¹) and larger stellar mass (log M_* = 11.2 ± 0.3 M_⊙). The results consistently indicate that the AGNs with SF-type disks have ongoing star formation at a level similar to that of star-forming galaxies of similar stellar mass (see Woo et al. 2017). We will compare the energetics of the AGNs with SF- and AGN-type disks and discuss the feedback scenario for the AGNs in Section 6.4.

Zoom In Zoom Out Reset image size

First, we calculate the energetics in each spaxel and examine the variation as a function of radial distance from the center. The ionized gas mass (${M}_{\mathrm{gas}}$) can be estimated as

$$\begin{eqnarray}&&{M}_{\mathrm{gas}}=(9.73\times {10}^{8}{M}_{\odot })\times {L}_{{\rm{H}}\alpha ,43}\times {n}_{e,100}^{-1},\end{eqnarray} \tag{ 5 }$$

where ${L}_{{\rm{H}}\alpha ,43}$ is the Hα luminosity in units of 10⁴³ erg s⁻¹, and ${n}_{e,100}$ is the electron density in units of 100 cm⁻³ (Nesvadba et al. 2006). We estimate n_e for each spaxel by using the [S ii] line ratio (Osterbrock & Ferland 2006). Assuming a gas temperature of 10⁴ K in the NLR, n_e in the central spaxel ranges from 54 to 854 cm⁻³, with the mean value of ${n}_{e}$ ∼ 360 ± 230 cm⁻³, which is consistent with previous studies (e.g., Nesvadba et al. 2006; Karouzos et al. 2016b). We find that both n_e and ${L}_{{\rm{H}}\alpha }$ radially decrease with radial distance, so ${M}_{\mathrm{gas}}$ also decreases outward.

The kinetic energy of gas outflow (${E}_{\mathrm{out}}$) is the summation of bulk energy (${E}_{\mathrm{bulk}}$) and turbulence energy (${E}_{\mathrm{turb}}$):

$$\begin{eqnarray}&&{E}_{\mathrm{out}}={E}_{\mathrm{bulk}}+{E}_{\mathrm{turb}}=\displaystyle \frac{1}{2}{M}_{\mathrm{gas}}({v}_{\mathrm{gas}}^{2}+{\sigma }_{\mathrm{gas}}^{2}),\end{eqnarray} \tag{ 6 }$$

where ${M}_{\mathrm{gas}}$ is the estimated ionized gas mass, and ${v}_{\mathrm{gas}}$ and ${\sigma }_{\mathrm{gas}}$ are the velocity offset and the velocity dispersion measured from the total [O iii] profile, respectively. The momentum (p) can be estimated as ${p}_{\mathrm{out}}={M}_{\mathrm{gas}}{v}_{\mathrm{gas}}$. For the estimation, we only use the spaxels where the [O iii] and [S ii] doublet fluxes have S/N > 3. Since the [S ii] line flux is in general weaker than the Hα flux, the spatial extent of the estimation is mostly focused on the inner ∼2 kpc of the AGNs, which is close to the size of ${R}_{\mathrm{out}}$ of the sample.

We find a general trend that the energy and momentum decrease ∼1 order of magnitude per ∼1 kpc increase in the radial distance in most of the sample (Figure 20). Thus, we obtain distance–energy and distance–momentum relations by combining the energy and momentum per spaxel for the sample (Figure 21). We obtain the error-weighted values of energy and momentum at each bin of 0.5 kpc distance. By assuming the uncertainty of distance as a one-half of a spaxel (∼03), we also obtain the error-weighted values of distance at each bin. Then, we apply a forward regression method using the FITEXY code for the error-weighted values, obtaining

$$\begin{eqnarray}&&\mathrm{log}{E}_{\mathrm{out}}=(54.44\pm 0.05)-(1.14\pm 0.03){D}_{\mathrm{kpc}},\end{eqnarray} \tag{ 7 }$$

$$\begin{eqnarray}&&\mathrm{log}{p}_{\mathrm{out}}=(46.92\pm 0.06)-(1.10\pm 0.03){D}_{\mathrm{kpc}}\end{eqnarray} \tag{ 8 }$$

where D_kpc is the distance in units of kpc. The relationships show that both energy and momentum steeply decrease as a function of distance.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

As we determined two different sizes based on the flux distribution (${R}_{\mathrm{NLR}}$) and kinematics of the NLR (${R}_{\mathrm{out}}$), here we obtain the total energetics of the outflows (i.e., mass outflow rate, energy injection rate, and momentum flux) for the two different sizes and compare the results. We also compare the energetics of the groups of AGNs with different disk properties (Section 4.4) in order to investigate whether nuclear star formation is affected by AGN outflows.

5.2.1. Energetics for Two Different Sizes

To estimate the total energetics of AGN outflows for two different sizes, we construct an integrated spectrum for both ${R}_{\mathrm{NLR}}$ and ${R}_{\mathrm{out}}$ of each AGN, and then we calculate ${M}_{\mathrm{gas}}$ and n_e in the same manner as for the spatially resolved case. We also calculate the [O iii] velocity offset and velocity dispersion from the integrated spectra. As we estimate the integrated energetics using two different sizes, we can directly compare the values and examine the effect of different size estimates on the energetics calculation.

If we assume biconical outflows with radius r and the ionized gas mass M_gas within r, the mass outflow rate ${\dot{M}}_{\mathrm{out}}$ can be calculated as

$$\begin{eqnarray}&&{\dot{M}}_{\mathrm{out}}=\displaystyle \frac{{M}_{\mathrm{gas}}}{{\tau }_{\mathrm{dyn}}}=\displaystyle \frac{3{M}_{\mathrm{gas}}{v}_{\mathrm{out}}}{r},\end{eqnarray} \tag{ 9 }$$

where ${\tau }_{\mathrm{dyn}}$ is the dynamical time for the ionized gas with ${v}_{\mathrm{out}}$ to reach a distance r, and ${v}_{\mathrm{out}}$ is the flux-weighted intrinsic outflow velocity, or bulk velocity of the outflows (Maiolino et al. 2012). Since we use the Hα luminosity from the total profile of Hα for our ${M}_{\mathrm{gas}}$ estimation (Equation (5)), the results can be regarded as an upper limit of ${M}_{\mathrm{gas}}$.

Then the energy injection rate ${\dot{E}}_{\mathrm{out}}$ and the momentum flux ${\dot{p}}_{\mathrm{out}}$ can be calculated as follows:

$$\begin{eqnarray}&&{\dot{E}}_{\mathrm{out}}=\displaystyle \frac{1}{2}{\dot{M}}_{\mathrm{out}}{v}_{\mathrm{out}}^{2},\end{eqnarray} \tag{ 10 }$$

$$\begin{eqnarray}&&{\dot{p}}_{\mathrm{out}}={\dot{M}}_{\mathrm{gas}}{v}_{\mathrm{out}}.\end{eqnarray} \tag{ 11 }$$

5.2.2. Estimation of the Intrinsic Outflow Velocity

To calculate the energy injection rate and the momentum flux (Equations (9)–(11)), we need to properly estimate the intrinsic outflow velocity ${v}_{\mathrm{out}}$ from the observations. We measured ${\sigma }_{[{\rm{O}}{\rm{III}}]}$ and ${v}_{[{\rm{O}}{\rm{III}}]}$, which are closely related to AGN outflows. However, these values are far from the intrinsic ${v}_{\mathrm{out}}$, because the measured values were affected by dust extinction and projection effects, which can be more severe in type 2s than in type 1s (e.g., Bae & Woo 2016). Hence, we need to assume proper outflow models to estimate ${v}_{\mathrm{out}}$, which is difficult to estimate from observations.

For example, Liu et al. (2013b) assumed spherical symmetric or wide-angle biconical outflow models, resulting in the relationship ${v}_{\mathrm{out}}=\,\sim$ 1.9 $\times \,{\sigma }_{[{\rm{O}}{\rm{III}}]}$. This number may not be directly applicable to this study, however, since our sample consists less luminous than theirs, which may indicate a different outflow geometry (Bae & Woo 2016). Our sample also consists of type 2 AGNs, which can be more affected by projection effects. To obtain the relationship of ${v}_{\mathrm{out}}$ and ${\sigma }_{[{\rm{O}}{\rm{III}}]}$ for our sample, we exploit 3D biconical models suggested by Bae & Woo (2016), who successfully reproduced the distributions of the [O iii] kinematics of type 2 AGNs in SDSS (Woo et al. 2016).

In our models, we assume the exponentially decreasing radial flux profile $f(d)={{Af}}_{n}{e}^{-\tau d}$, where f_n is the flux of the nucleus, and A is the amount of dust extinction (0 < A < 1) for each position in 3D. We adopt the range of τ = 3–7 and the size of the bicone as unity. We also assume different velocity profiles v(d) (e.g., linear decrease or increase). Then, the intrinsic flux-weighted ${v}_{\mathrm{out}}$ can be calculated as

$$\begin{eqnarray}&&{v}_{\mathrm{out}}=\displaystyle \frac{\int f(d)v(d){dp}}{\int f(d){dp}},\end{eqnarray} \tag{ 12 }$$

where p represents each position in 3D. The bicone has outer and inner half-opening angles, and we use three different outer half-opening angles of 20°, 40°, and 60°. The inner half-opening angles are fixed as one-half of the outer half-opening angle (Bae & Woo 2016). In our calculations, we use the value of ${\sigma }_{0}$ = ${({v}_{[{\rm{O}}{\rm{III}}]}^{2}+{\sigma }_{[{\rm{O}}{\rm{III}}]}^{2})}^{0.5}$ rather than ${\sigma }_{[{\rm{O}}{\rm{III}}]}$, because ${\sigma }_{0}$ can be a good proxy for the line width without dust extinction (Bae & Woo 2016).

By integrating the 3D bicone models, we calculate the intrinsic ${v}_{\mathrm{out}}$ and ${\sigma }_{0}$ as a function of bicone inclination (Figure 22). While ${v}_{\mathrm{out}}$ is a fixed value, ${\sigma }_{0}$ is minimized when the bicone axis is parallel to the plane of the sky (${i}_{\mathrm{bicone}}$ = 0°) due to the projection effects. On the other hand, if the bicone is inclined to ±40°, which is about the maximum inclination for type 2 AGNs (Marin 2014), ${\sigma }_{0}$ becomes larger due to lower projection effects. As a result, ${v}_{\mathrm{out}}/{\sigma }_{0}$ has a large range from ∼1.5 to ∼2.5 if the outer half-opening angle of the bicone is 40° (middle panel).

Zoom In Zoom Out Reset image size

For the IFU sample, we adopt a bicone half-opening angle of 40°, which is consistent with the mean value of half-opening angle estimated for 17 Seyfert galaxies based on HST/STIS data (Fischer et al. 2014). Hence, we adopt the relationship of ${v}_{\mathrm{out}}$ = (2.0 ± 0.5) × ${\sigma }_{0}$ in the energetics calculation for our sample. Alternatively, if the outer half-opening angle of the bicone is 20° (left panel) or 60° (right panel), ${v}_{\mathrm{out}}/{\sigma }_{0}$ ranges from ∼1.5 to ∼4.5 and from ∼1.5 to ∼1.9, respectively. Also, if we use different velocity profiles, the range of ${v}_{\mathrm{out}}/{\sigma }_{0}$ becomes smaller than in the case of a linear decrease.

5.2.3. Estimated Energetics

First, we compare the ionized gas mass and outflow energetics obtained by using two different sizes (Figure 23). The estimated physical quantities within ${R}_{\mathrm{NLR}}$ and ${R}_{\mathrm{out}}$ are listed in Tables 4 and 5, respectively. Note that we could not measure the electron density for two AGNs (J0341−0719 and J1100+1124) when using ${R}_{\mathrm{NLR}}$, and five AGNs (J0341−0719, J0857+0047, J0952+1937, J1100+1124, and J1520+0757) when using ${R}_{\mathrm{out}}$. This is because the estimated electron densities based on [S ii] line ratios are saturated to the lowest density (1 cm⁻³; J0341−0719, J0857+0047, J0952+1937, and J1520+0757), or the [S ii] doublet is affected by the telluric absorption band (J1100+1124).

Zoom In Zoom Out Reset image size

Table 4.  Integrated Physical Properties of the Type 2 AGNs within ${R}_{\mathrm{NLR}}$

SDSS name	log ${L}_{[{\rm{O}}{\rm{III}}]}$	log ${L}_{\mathrm{bol}}$	${\dot{M}}_{\mathrm{acc}}$	${v}_{[{\rm{O}}{\rm{III}}]}$	${\sigma }_{[{\rm{O}}{\rm{III}}]}$	${\sigma }_{0}$	${v}_{\mathrm{out}}$	log ${L}_{{\rm{H}}\alpha }$	ne	${M}_{\mathrm{gas}}$	log ${E}_{\mathrm{kin}}$	${\dot{M}}_{\mathrm{out}}$	log ${\dot{E}}_{\mathrm{out}}$	log ${\dot{p}}_{\mathrm{out}}$
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)
J0130+1312	40.34	43.88	1.35	−175	193	260	520	40.54	370	9.2	54.40	1.5	41.12	33.70
J0341−0719	40.39	43.93	1.50	+144	246	285	570	40.58	122a	30.3	54.99	6.6	41.83	34.38
J0806+1906	40.69	44.23	3.03	−86	187	205	411	40.72	344	14.8	54.39	1.1	40.78	33.47
J0855+0047	40.26	43.80	1.13	−230	285	366	732	40.41	171	14.6	54.89	2.3	41.59	34.02
J0857+0633	40.11	43.66	0.80	+122	177	215	430	40.24	129	13.1	54.38	1.2	40.86	33.53
J0911+1333	40.77	44.31	3.61	−132	429	449	898	41.19	583	26.2	55.32	7.3	42.27	34.62
J0952+1937	39.47	43.01	0.18	−63	345	351	702	40.49	78	38.5	55.28	17.7	42.44	34.89
J1001+0954	40.33	43.87	1.32	−25	548	548	1096	40.48	399	7.3	54.94	3.0	42.06	34.32
J1013−0054	40.11	43.65	0.80	+12	300	300	600	40.14	236	5.7	54.31	1.7	41.29	33.81
J1054+1646	40.72	44.27	3.27	−282	377	471	942	40.64	156	27.6	55.39	9.5	42.42	34.75
J1100+1321	40.56	44.10	2.24	−137	470	490	980	40.32	237	8.6	54.91	3.4	42.01	34.32
J1100+1124	39.53	43.07	0.21	+158	254	299	598	39.95	341b	2.1	53.88	0.3	40.53	33.06
J1106+0633	40.46	44.00	1.78	−176	234	294	587	40.82	520	12.3	54.63	2.5	41.43	33.96
J1147+0752	40.65	44.19	2.73	−358	437	564	1129	40.64	327	12.8	55.21	7.6	42.48	34.73
J1214−0329	40.30	43.84	1.23	−281	331	434	869	40.70	761	6.4	54.69	4.9	42.06	34.43
J1310+0837	40.42	43.97	1.63	+172	250	304	608	40.09	321	3.7	54.13	1.0	41.05	33.56
J1440+0556	40.04	43.58	0.67	+42	480	482	964	39.96	854	1.0	53.98	0.6	41.24	33.56
J1448+1055	40.94	44.48	5.37	−293	576	646	1292	40.74	620	8.7	55.16	3.7	42.29	34.48
J1520+0757	39.16	42.70	0.09	−486	430	649	1297	39.77	54	10.6	55.25	9.2	42.69	34.88
J2039+0033	40.55	44.10	2.20	−71	468	474	947	40.71	251	19.8	55.25	7.8	42.35	34.67

Notes. (1) SDSS name of AGN; (2) extinction-uncorrected [O iii] luminosity in logarithm (erg s⁻¹); (3) AGN bolometric luminosity in logarithm as ${L}_{\mathrm{bol}}=3500\times {L}_{[{\rm{O}}{\rm{III}}]}$ (erg s⁻¹, Heckman et al. 2004); (4) mass accretion rate as ${\dot{M}}_{\mathrm{acc}}={L}_{\mathrm{bol}}/\mu {c}^{2}$ (${10}^{-2}{M}_{\odot }$ yr⁻¹); (5) [O iii] velocity offset with respect to the systemic velocity of the stellar component (km s⁻¹); (6) [O iii] velocity dispersion (km s⁻¹); (7) dust-corrected [O iii] velocity dispersion (km s⁻¹); (8) bulk velocity of the outflows (km s⁻¹) (9) Hα luminosity in logarithm (erg s⁻¹); (10) electron density estimated from the [S ii] line ratio (cm⁻³); (11) ionized gas mass in units of 10⁵ M_⊙ (see Equation (5)); (12) kinetic energy in logarithm (erg); (13) mass outflow rate (M_⊙ yr⁻¹); (14) energy injection rate (erg s⁻¹); (15) momentum flux (dyne).

^aThe estimated electron density from the [S ii] ratio is saturated to the lowest density allowed in the calculation (1 cm⁻³), so we adopt the electron density from SDSS. ^bThe [S ii] doublet is affected by a telluric absorption band, so we adopt the electron density from SDSS.

Download table as:  ASCIITypeset image

Table 5.  Integrated Physical Properties of the Type 2 AGNs within ${R}_{\mathrm{out}}$

SDSS name	log ${L}_{[{\rm{O}}{\rm{III}}]}$	log ${L}_{\mathrm{bol}}$	${\dot{M}}_{\mathrm{acc}}$	${v}_{[{\rm{O}}{\rm{III}}]}$	${\sigma }_{[{\rm{O}}{\rm{III}}]}$	${\sigma }_{0}$	${v}_{\mathrm{out}}$	log ${L}_{{\rm{H}}\alpha }$	ne	${M}_{\mathrm{gas}}$	log ${E}_{\mathrm{kin}}$	${\dot{M}}_{\mathrm{out}}$	log ${\dot{E}}_{\mathrm{out}}$	log ${\dot{p}}_{\mathrm{out}}$
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)
J0130+1312	38.82	42.36	0.04	−106	200	226	453	38.90	57	1.4	53.45	0.8	40.72	33.36
J0341−0719	40.53	44.07	2.09	+126	239	270	541	40.82	⋯	⋯	⋯	⋯	⋯	⋯
J0806+1906	39.80	43.34	0.39	−38	231	234	469	39.92	416	1.9	53.63	0.7	40.68	33.31
J0855+0047	40.42	43.96	1.63	−132	277	307	615	41.02	93	108.2	55.61	3.6	41.63	34.15
J0857+0633	39.73	43.27	0.33	+134	222	259	518	39.85	⋯	⋯	⋯	⋯	⋯	⋯
J0911+1333	40.99	44.54	6.10	−98	355	369	737	41.42	516	49.3	55.43	3.7	41.80	34.23
J0952+1937	39.64	43.18	0.27	−70	180	193	386	40.82	⋯	⋯	⋯	⋯	⋯	⋯
J1001+0954	40.63	44.17	2.62	−57	489	492	984	40.88	536	13.7	55.12	1.9	41.77	34.08
J1013−0054	40.14	43.69	0.86	−5	293	293	587	40.19	187	8.1	54.44	2.2	41.38	33.92
J1054+1646	41.00	44.54	6.18	−260	357	442	884	40.95	151	57.2	55.65	5.7	42.15	34.50
J1100+1321	40.82	44.37	4.09	−126	432	450	901	40.70	122	39.7	55.51	5.7	42.17	34.51
J1100+1124	39.64	43.19	0.27	+140	223	264	528	40.12	⋯	⋯	⋯	⋯	⋯	⋯
J1106+0633	40.64	44.19	271	−129	237	270	540	41.14	250	53.3	55.19	1.6	41.18	33.74
J1147+0752	41.17	44.71	9.07	−314	373	488	976	41.31	428	46.8	55.65	6.6	42.30	34.61
J1214−0329	40.49	44.03	1.89	−231	314	390	779	40.92	686	11.8	54.85	4.5	41.93	34.34
J1310+0837	40.25	43.79	1.09	+165	236	288	576	39.94	289	2.9	53.99	1.2	41.09	33.63
J1440+0556	40.25	43.79	1.09	+28	451	452	905	40.23	803	2.1	54.23	0.5	41.15	33.49
J1448+1055	41.22	44.76	10.22	−254	561	616	1232	41.10	729	16.7	55.40	2.2	42.02	34.23
J1520+0757	39.49	43.03	0.19	−505	466	687	1374	40.18	⋯	⋯	⋯	⋯	⋯	⋯
J2039+0033	40.53	44.07	2.08	−59	408	413	825	40.70	255	19.3	55.12	6.5	42.15	34.53

Notes. (1) SDSS name of AGN; (2) extinction-uncorrected [O iii] luminosity in logarithm (erg s⁻¹); (3) AGN bolometric luminosity in logarithm as ${L}_{\mathrm{bol}}=3500\times {L}_{[{\rm{O}}{\rm{III}}]}$ (erg s⁻¹, Heckman et al. 2004); (4) mass accretion rate as ${\dot{M}}_{\mathrm{acc}}={L}_{\mathrm{bol}}/\mu {c}^{2}$ (${10}^{-2}{M}_{\odot }$ yr⁻¹); (5) [O iii] velocity offset with respect to the systemic velocity of the stellar component (km s⁻¹); (6) [O iii] velocity dispersion (km s⁻¹); (7) dust-corrected [O iii] velocity dispersion (km s⁻¹); (8) bulk velocity of the outflows (km s⁻¹) (9) Hα luminosity in logarithm (erg s⁻¹); (10) electron density estimated from the [S ii] line ratio (cm⁻³); (11) ionized gas mass in units of 10⁵ M_⊙ (see Equation (5)); (12) kinetic energy in logarithm (erg); (13) mass outflow rate (M_⊙ yr⁻¹); (14) energy injection rate (erg s⁻¹); (15) momentum flux (dyne).

Download table as:  ASCIITypeset image

We find that the ionized gas mass using ${R}_{\mathrm{out}}$ is ∼2.3 ± 1.9 times larger than the mass using ${R}_{\mathrm{NLR}}$ (top left panel), because ${R}_{\mathrm{out}}$ is a factor of ∼2 larger than ${R}_{\mathrm{NLR}}$ (Section 4.3). However, the estimated energetics, that is, outflow velocity (top middle), total energy (top right), mass outflow rate (bottom left), energy injection rate (bottom middle), and momentum flux (bottom right), are overall consistent with each other within uncertainties. The reason for this consistency is mainly the relatively large uncertainties for the estimated quantities, for example, outflow velocity and electron density. Also, the estimated outflow velocity does not change much when we increase the size since we estimate the velocity using the flux-weighted kinematics of [O iii], which is exponentially decreasing from the nucleus. Hence, we will use R_NLR-based physical properties in the following analysis of the energetics of outflows. Note that we use the electron density from SDSS for two AGNs lacking this information from our data (J0341−0719 and J1100+1124).

The estimated ${M}_{\mathrm{gas}}$ is $(1.0\mbox{--}38.5)\times {10}^{5}\,{M}_{\odot }$, and the mass outflow rate ${\dot{M}}_{\mathrm{out}}$ is 0.3–17.7 ${M}_{\odot }$ yr⁻¹. The mass accretion rate ${\dot{M}}_{\mathrm{acc}}$ can be calculated as ${\dot{M}}_{\mathrm{acc}}$ = ${L}_{\mathrm{bol}}/\eta {c}^{2}$, where the η is the accretion efficiency typically assumed to be 0.1, and ${L}_{\mathrm{bol}}$ is the AGN bolometric luminosity estimated as ${L}_{\mathrm{bol}}$ = 3500 × ${L}_{[{\rm{O}}{\rm{III}}]}$, where ${L}_{[{\rm{O}}{\rm{III}}]}$ is the extinction-uncorrected [O iii] luminosity (Heckman et al. 2004). The mean ${\dot{M}}_{\mathrm{out}}$ ∼4.6 ± 4.3 ${M}_{\odot }$ yr⁻¹ is consistent with the values of nearby AGNs (0.1–10 ${M}_{\odot }$, Veilleux et al. 2005). The estimated mean mass outflow rate is about a factor of ∼260 higher than the estimated mean ${\dot{M}}_{\mathrm{acc}}\sim 0.02\,\pm \,0.01\,{M}_{\odot }$ yr⁻¹, indicating powerful mass loading of the AGN outflows by the ISM (e.g., Veilleux et al. 2005; Barbosa et al. 2009).

We also compare the energy injection rate and the momentum flux as a function of AGN bolometric luminosity (Figure 24). The estimated ${\dot{E}}_{\mathrm{out}}$ is ${10}^{40.5-42.7}$ erg s⁻¹, and the estimated ${\dot{p}}_{\mathrm{out}}$ is ${10}^{33.1-34.9}$ dyne. The majority of AGNs (18 out of 20) have a relatively low energy injection rate, about 0.8 ± 0.6% of ${L}_{\mathrm{bol}}$, and also have a relatively low momentum flux, about (5.4 ± 3.6) × ${L}_{\mathrm{bol}}/c$. Both the energy injection rate and the momentum flux for the AGNs are in general increasing as a function of ${L}_{\mathrm{bol}}$. Interestingly, we find two AGNs (J0952+1937 and J1520+0757) with much higher energetics than the majority of AGNs in our sample. Although these two AGNs have relatively low bolometric luminosities (${L}_{\mathrm{bol}}\sim {10}^{42.7-43.0}$ erg s⁻¹), their energy injection rate is 27%–97% of ${L}_{\mathrm{bol}}$, and the momentum flux is 228–449 ${L}_{\mathrm{bol}}/c$.

Zoom In Zoom Out Reset image size

In Sections 4.1 and 4.2, we investigate various properties, such as flux, velocity, and velocity dispersion, of [O iii] and Hα, respectively, in 1D and 2D, finding that the NLR is a mixture of nongravitational, that is, AGN outflows, and gravitational kinematics, that is, rotation or random motion (see also Karouzos et al. 2016a). The nongravitational kinematics are commonly detected in the broad component of both [O iii] and Hα, which are revealed by their large velocity dispersion compared to the stellar velocity dispersion or their large velocity offset with respect to the systemic velocity of the host galaxy. While the broad components of [O iii] and Hα show qualitatively similar nongravitational kinematics, the absolute values of velocity and velocity dispersion of Hα are relatively smaller than those of [O iii].

We find a clear sign of gravitational kinematics (i.e., Keplerian rotation) in the narrow component of Hα for most objects (16 out of 20 AGNs). The other four AGNs, which do not show any rotation in the narrow component of Hα, may also have gravitational kinematics, such as a random motion. In comparison, only eight out of 20 AGNs show the gravitational kinematics (i.e., rotation or random motion) in the narrow component of [O iii], implying that the gravitational kinematics are more significantly presented in Hα than in [O iii]. Consider a simple picture in which biconical outflows and a star-forming disk coexist in the nucleus. Then, the star-forming disk is responsible for the gravitational kinematics, while the biconical outflows are responsible for the outflow kinematics in the NLR. In general, the star-forming region emits stronger Balmer emission than [O iii] emission, so the rotational feature is more distinguishable in Hα than in [O iii]. For the same reason, the outflowing region, which is ionized by an AGN, emits stronger [O iii] than Balmer lines. Thus, the outflow kinematics are more noticeable in [O iii] than in Hα.

In Section 4.4, we further focus on the origin of rotational features in the relation of the narrow component of Hα with star formation. We find that 11 AGNs with SF-type disks show lower ${D}_{n}(4000)$ and higher ${H}_{\delta }$ indices than the AGNs with AGN-type disks, showing that the AGNs with SF-type disks have ongoing star formation, while five AGNs with AGN-type disks do not. The results further support that, even in the AGNs with strong outflows, the star-forming disk and AGN outflows coexist in the nucleus (the central kiloparsec), as several studies pointed out (e.g., Davies et al. 2016, Woo et al. 2017).

While it is clear that the ionized gas outflows of our sample are mainly due to AGN activity (see Section 5.2), we further investigate the driving mechanism of the outflows. Theoretical studies expect that the outflows are energy conserving if the outflowing gas expands as a hot bubble without efficient cooling, while the outflow is momentum conserving if the outflows rapidly cool down and lose their energy (see King & Pounds 2015, for a review). AGN outflows would generate a different impact on the host galaxy depending on whether the outflows are energy  or momentum conserving.

Since the driving mechanism depends on the properties of the accretion wind and the ISM of the host galaxy, it is still under debate which phase is dominant when and at what physical scales (e.g., Silk & Rees 1998; Faucher-Giguère & Quataert 2012; Zubovas & Nayakshin 2014). For example, King & Pounds (2015) imagine a scenario where the outflows are initially momentum conserving due to rapid cooling of the wind. In this case, the outflows lose their energy and cannot remove the ISM, so the SMBH grows until it reaches the ${M}_{\mathrm{BH}}$–${\sigma }_{\star }$ relation. When the SMBH has grown enough, the energy-conserving phase starts. This phase is more energetic than the momentum-conserving phase. As the energy-conserving outflows boost the momentum at larger scales, the outflows sweep out the ISM and may suppress the growth of the SMBH. Faucher-Giguère & Quataert (2012) argue that the energy-conserving outflows are plausible if the wind is fast (>∼10,000 km s⁻¹), or even if the wind is slow (>∼1000 km s⁻¹) with more restricted conditions.

Since the momentum flux compared to ${L}_{\mathrm{Bol}}/c$ of our sample is much larger than unity in most cases, we assume that the wind-driven outflows at the nucleus expand to large-scale (kiloparsec) outflows via the energy-conserving phase as

$$\begin{eqnarray}&&{\dot{M}}_{\mathrm{gas}}{v}_{\mathrm{gas}}^{2}\approx f{\dot{M}}_{\mathrm{in}}{v}_{\mathrm{in}}^{2},\end{eqnarray} \tag{ 13 }$$

where ${\dot{M}}_{\mathrm{in}}$ and ${v}_{\mathrm{in}}$ are the initial mass outflow rate and initial velocity in the nucleus, respectively, and f is an energy transfer efficiency from small-scale wind to large-scale outflow. We adopt an efficiency of 0.2 based on recent observational results (Tombesi et al. 2015).

The momentum boost ${\dot{p}}_{\mathrm{gas}}/{\dot{p}}_{\mathrm{in}}$, where ${\dot{p}}_{\mathrm{gas}}={\dot{M}}_{\mathrm{gas}}{v}_{\mathrm{gas}}$ and ${\dot{p}}_{\mathrm{in}}={\dot{M}}_{\mathrm{in}}{v}_{\mathrm{in}}$, can be calculated by combining with Equation (13) as follows:

$$\begin{eqnarray}&&\displaystyle \frac{{\dot{p}}_{\mathrm{gas}}}{{\dot{p}}_{\mathrm{in}}}\approx 0.2\displaystyle \frac{{v}_{\mathrm{in}}}{{v}_{\mathrm{gas}}}.\end{eqnarray} \tag{ 14 }$$

By using this equation, we estimate the values of ${v}_{\mathrm{gas}}$ and ${\dot{p}}_{\mathrm{gas}}$ with the expectations from the energy-conserving outflow by assuming ${\dot{p}}_{\mathrm{in}}={L}_{\mathrm{bol}}/c$ (Figure 25). We find that most AGNs lie within ${v}_{\mathrm{in}}=(0.01\mbox{--}0.3)c$, which is broadly consistent with ultrafast outflows (UFOs; ${v}_{\mathrm{in}}$ ≈ (0.1–0.4)c) detected in X-ray observations (e.g., Tombesi et al. 2015). We do not find, however, any clear evidence for instantaneous quenching of the star formation due to outflows, since 11 out of 20 AGNs still show sign of ongoing star formation (see Sections 4.4 and 6.1).

Zoom In Zoom Out Reset image size

On the contrary, if it were the momentum-conserving phase, the momentum boost would be expected to be 0.2 regardless of ${v}_{\mathrm{gas}}$, which is not the case for our sample. These results imply that energy-conserving outflows due to accretion-disk wind might be the driving mechanism of the ionized gas outflows observed in the sample.

Note that we have two extreme AGNs with a high momentum boost (J0952+1937 and J1520+0757), which lead to unphysically large ${v}_{\mathrm{in}}\gt c$, although the uncertainties are large. One possibility is that the f factor is larger for these AGNs due to largely different ISM composition. For J1520+0757, however, it is difficult to obtain ${v}_{\mathrm{in}}\lt c$ even with f = 1. Another possibility is that the two AGNs may have much lower ${L}_{\mathrm{bol}}$ at present compared to the time when the outflow was launched, resulting in an overestimated momentum boost compared to the true value. We will discuss this scenario in the following section.

In Sections 5.2 and 6.2, we find two AGNs with extreme energetics, while their bolometric luminosities are relatively low (J0952+1937 and J1520+0757). These two AGNs have the lowest electron density among the sample (n_e = 78 and 54 cm⁻³, respectively), implying that the ionized gas has possibly been swept out, which is consistent with their high mass outflow rate (${\dot{M}}_{\mathrm{out}}$ = 17.7 and 9.2 M_⊙ yr⁻¹, respectively). Then, what can cause the discrepancy between their energetics, such as energy injection rate and momentum flux, and the bolometric luminosity? Since we measured the bolometric luminosity of AGNs based on the [O iii] luminosity in the NLR, the difference between the photoionization timescale (∼10⁴ yr) and the dynamical timescale of the outflows (∼10⁶ yr) to reach the NLR may cause the discrepancy. We note that this is the same explanation as for the weak relationship between ${R}_{\mathrm{out}}$ and the [O iii] luminosity (see Figure 18).

Consider the energy-conserving outflow expanding from high-velocity, small-scale winds close to the nucleus (see Section 6.2). As the winds expand, the hot bubble would sweep out the surrounding ISM, and the outflow velocity would become smaller. If we simply consider the constant outflow velocity of 1000 km s⁻¹, it will take 10⁶ yr, which is the dynamical timescale of outflows, to reach the ISM up to ${R}_{\mathrm{out}}$ of 1 kpc. Also, we find that the mean mass outflow rate is ∼4.6 ${M}_{\odot }$ yr⁻¹ for the sample, so the ionized gas within ${R}_{\mathrm{NLR}}$ will be removed after ∼10⁵ yr. Since the outflow would sweep out the ISM in the vicinity of the SMBH within a much shorter timescale ≪10⁵ yr, it is possible that the AGN becomes inactive while the outflows propagate in the NLR (∼kpc). Hence, the AGN can be inactive or less luminous when the powerful outflow reaches the NLR.

From the integrated spectra within ${R}_{\mathrm{NLR}}$, we estimate the physical properties related to the energetics of gas outflows, such as mass outflow rate, energy injection rate, and momentum flux (Section 5.2). We find the mean outflow velocity ${v}_{\mathrm{out}}=\sim 800$ km s⁻¹ and the mean ${R}_{\mathrm{NLR}}=\sim 900$ pc. The dynamical time t_d is given by ${t}_{d}={R}_{\mathrm{NLR}}/{v}_{\mathrm{out}}$ ∼ 10⁶ yr, which is an order of magnitude smaller than the theoretical expectation of 10⁷ yr for quasar lifetimes (e.g., Hopkins et al. 2006). The AGNs have a mass outflow rate ${\dot{M}}_{\mathrm{out}}$ = 0.3–17.7 ${M}_{\odot }$ yr⁻¹ with a mean value of ∼4.6 ${M}_{\odot }$ yr⁻¹, which is a factor of ∼300 larger than the mean mass accretion rate ${\dot{M}}_{\mathrm{acc}}\sim 0.02$ ${M}_{\odot }$ yr⁻¹. For this mean gas mass for AGNs (∼1.4 × 10⁶ ${M}_{\odot }$) and mean mass outflow rate (∼4.6 ${M}_{\odot }$ yr⁻¹), it would take ∼(4.2 ± 3.1) × 10⁵ yr to remove the ionized gas from the ${R}_{\mathrm{NLR}}\sim 1\,\mathrm{kpc}$, which is less than the dynamical timescale of AGNs of ∼10⁶ yr. Such a gas removal timescale (∼4.2 × 10⁵ yr) is comparable to the AGN flickering timescale of ∼10⁵ yr (Schawinski et al. 2015).

From the quantitative estimations, we show that the AGN outflows can remove the ionized gas within ${R}_{\mathrm{NLR}}$ in a reasonably short timescale, implying that star formation within ∼1 kpc of AGN can be affected by AGN outflows. In Section 4.4, for example, we find that ∼30% of AGNs with a disk are AGN-type disks, which is possibly affected by AGN photoionization. The higher ${D}_{n}(4000)$ and smaller ${\rm{H}}{\delta }_{A}$ of the AGNs with AGN-type disks indicate a relatively low star formation rate within the central kiloparsec of the AGNs. We do not find, however, any differences between the AGNs with AGN-type and SF-type disks in terms of outflow energetics (Figures 23–25).

Nonetheless, the most kinematically energetic AGNs of our sample, J0952+1937 and J1520+0757, may provide an insight into AGN feedback. For example, J1520+0757 has no detectable rotational feature in Hα, residing in the "green valley" with log SSFR = −10.9 yr⁻¹. On the other hand, J0952+1937 has an SF-type disk and harbors ongoing star formation with log SSFR = −9.5 yr⁻¹ (see Figure 19). If the outflows in these AGNs are powerful enough to affect the star formation in the host galaxy, their different SSFRs are possibly due to different geometries of outflows with respect to the star-forming disk. This speculation can be tested using high spatial resolution IFU observations combined with a proper modeling of a mixed kinematics of outflows and disk rotation. Alternatively, the outflows may not be powerful enough to immediately quench the star formation in the host galaxy, although the ionized gas shows evidence of powerful outflows. Recent numerical simulations show the inefficiency of AGN outflows in quenching star formation, because the outflows are mostly affecting ionized gas rather than dense molecular gas (e.g., Gabor & Bournaud 2014). This scenario can be tested whether the object shows outflow features in molecular gas or ionized gas via, for example, ALMA observations.

Do AGN outflows also affect star formation on galactic scales (∼10 kpc)? From our results, it would be difficult for our intermediate-luminosity AGNs, because of the steeply decreasing energy and momentum as a function of distance (see Section 5.1). Also, Karouzos et al. (2016b) showed that ∼90% of outflow energy and momentum are within a ∼1–2 kpc region (∼${R}_{\mathrm{out}}$) in type 2 AGNs, implying that energy injected by AGN outflows would become smaller and eventually negligible at larger distance, for example, ∼10 kpc, for the AGN luminosities of our sample.

In Section 4.4, we find that the AGNs with SF-type disks are located on the main sequence of star-forming galaxies, while the AGNs with AGN-type disks are below the main sequence. The global SSFRs of host galaxies of the AGNs with AGN-type disks are on average about one order of magnitude smaller than that in the AGNs with SF-type disks. Does the low SSFR in the sample of AGN-type disks support the AGN feedback in galactic scales? By considering that the AGNs with AGN-type disks are on average a factor of ∼3.5 more massive than the AGNs with SF-type disks (see Section 4.4), it is difficult to conclude that the low SSFRs support the galactic-scale AGN feedback, although it is a plausible scenario. This is because such low SSFRs in massive galaxies can also be accounted for by alternative quenching mechanisms, such as environmental quenching (e.g., Smith et al. 2012; Peng et al. 2015). More IFU-observed samples within similar mass scales may provide a hint for the primary quenching mechanism.