THE ORIGIN OF DOUBLE-PEAKED NARROW LINES IN ACTIVE GALACTIC NUCLEI. I. VERY LARGE ARRAY DETECTIONS OF DUAL AGNs AND AGN OUTFLOWS^*

Comerford et al. (2012) present long-slit spectroscopy obtained at the Lick, Palomar, and MMT observatories of a subsample of 81 SDSS galaxies with double-peaked narrow AGN emission lines at 0.03 < z < 0.36 and 14.5 < r < 18.5. These galaxies were selected from a parent sample of 340 double-peaked AGNs derived from the 3 catalogs that have identified active galaxies in SDSS with double-peaked [O iii] emission lines: Wang et al. (2009), Liu et al. (2010a), and Smith et al. (2010).

From the 81 galaxies presented in Comerford et al. (2012), we have chosen as targets the full sample of 18 SDSS galaxies with double-peaked narrow emission lines that have (1) VLA FIRST detections, indicating they should be detectable with the VLA; and (2) optical long-slit spectra that exhibit two relatively compact AGN emission components with angular separations of at least $0\buildrel{\prime\prime}\over{.} 2,$ so that the VLA A-configuration at frequencies $\geqslant 8\;{\rm{GHz}}$ can resolve double radio cores spatially without difficulty. All galaxies are Type 2 AGNs with redshifts between 0.04 < z < 0.22 (Table 1). The measured angular separations of the two optical emission components on the sky for the sample are $0\buildrel{\prime\prime}\over{.} 21\lt {D}_{[{\rm{O}}\;{\rm{III}}]}\lt 0\buildrel{\prime\prime}\over{.} 87,$ which correspond to expected physical separations of 0.35–2.3 kpc. Details on the long-slit observations and data reduction can be found in Comerford et al. (2012).

Table 1.  List of Double-peaked AGNs Observed with the VLA

Source	R.A. (J2000)	decl. (J2000)	z	(kpc/'')	Flux Cal.	Phase Cal.
SDSS J000249+004504	00 02 49.07	$+00\;45\;04.8$	0.0868	1.62	3C48	J0016–0015
SDSS J000911–003654	00 09 11.58	$-00\;36\;54.7$	0.0733	1.39	3C48	J0016–0015
SDSS J073117+452803	07 31 17.55	$+45\;28\;03.2$	0.0835	1.57	3C138	J0735+4750
SDSS J073656+475946	07 36 56.47	$+47\;59\;46.8$	0.0962	1.78	3C138	J0735+4750
SDSS J080218+304622	08 02 18.65	$+30\;46\;22.7$	0.0766	1.45	3C138	J0751+3313
SDSS J084624+425842	08 46 24.51	$+42\;58\;42.8$	0.2192	3.54	3C138	J0836+4125
SDSS J085841+104122	08 58 41.76	$+10\;41\;22.1$	0.1480	2.58	3C138	J0839+0319
SDSS J091646+283526	09 16 46.03	$+28\;35\;26.7$	0.1423	2.50	3C138	J0915+2933
SDSS J093024+343057	09 30 24.84	$+34\;30\;57.3$	0.0610	1.18	3C286	J0927+3902
SDSS J102325+324348	10 23 25.57	$+32\;43\;48.4$	0.1271	2.27	3C286	J1022+3041
SDSS J102727+305902	10 27 27.90	$+30\;59\;02.4$	0.1245	2.23	3C286	J1022+3041
SDSS J111201+275053	11 12 01.78	$+27\;50\;53.8$	0.0472	0.93	3C286	J1150+2417
SDSS J115249+190300	11 52 49.33	$+19\;03\;00.3$	0.0967	1.79	3C286	J1150+2417
SDSS J115822+323102	11 58 22.58	$+32\;31\;02.2$	0.1658	2.84	3C286	J1150+2417
SDSS J155619+094855	15 56 19.30	$+09\;48\;55.6$	0.0678	1.30	3C286	J1608+1029
SDSS J162345+080851	16 23 45.20	$+08\;08\;51.1$	0.1992	3.30	3C286	J1608+1029
SDSS J171544+600835	17 15 44.05	$+60\;08\;35.7$	0.1569	2.71	3C286	J1657+5705
SDSS J225420–005134	22 54 20.99	$-00\;51\;34.1$	0.0795	1.50	3C48	J2257+0243

Download table as:  ASCIITypeset image

We obtained NRAO VLA observations of 18 galaxies with double-peaked, spatially offset narrow AGN emission lines at X-band (8–12 GHz). The observations were carried out in the A-configuration for 12 hr with two 1024 MHz wide basebands centered at 8.5 and 11.5 GHz. The correlator was configured with 8 spectral windows, each of which contained 64 channels with a frequency resolution of 2 MHz. The observations consisted of 7–8 minute scans of the target sources, interspersed with 1–1.5 minute scans of nearby phase reference sources. Switching angles between target and reference pairs ranged between 0.5° and 4°. Therefore, over the 12 hr period, between 28 and 32 minutes of integration time was obtained for each of the 18 targets. The 18 galaxies, together with the flux density and phase calibrators used for each galaxy, are listed in Table 1. The uncertainty of the flux density calibration is typically 3% at these frequencies (Perley & Butler 2013).

We flagged, calibrated, and imaged the X-band data with the Common Astronomy Software Applications (CASA) package. Phase-referenced (Stokes I) images of each galaxy were produced by applying the CLEAN task with a Briggs robust parameter of −0.5 to achieve good spatial resolution with only a modest loss of sensitivity. For the analysis, we smoothed the upper 11.5 GHz baseband image to match the resolution of the lower 8.5 GHz baseband using the CASA task IMSMOOTH. These two images were then combined to generate an image of the spectral index distribution across the source. This observational approach avoids spectral index uncertainty that results from intrinsic source variability. We also generated a total intensity image by integrating all the spectral windows in the two basebands. The radio properties of each galaxy at 8.5 and 11.5 GHz (both in native resolution and smoothed to match the resolution of the 8.5 GHz image) are given in Tables 1–5, and radio images with contours are shown in Figures 1–18. The images show the central $2^{\prime\prime} \times 2^{\prime\prime}$ of each galaxy to display its detailed morphology. None of the galaxies show structures above the 3σ rms level outside this region. In some galaxies, two compact sources are detected in the total intensity image. We label these sources "C1" and "C2," with C1 being the brightest of the two.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In Comerford et al. (2012), three morphological parameters from the long-slit spectra were used in an attempt to constrain what mechanisms produce double-peaked narrow AGN emission lines: the projected spatial separation between the two [O iii] emission features (${D}_{[{\rm{O}}\;{\rm{III}}]}$), the true position angle of the two emission components on the sky (${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$), and the spatial extent of the [O iii] emitting region. While a quantitatively accurate analysis of ${D}_{[{\rm{O}}\;{\rm{III}}]}$ and ${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$ was performed, the spatial extent of the emission was only measured by eye. The visual method used in Comerford et al. (2012) gives two qualitative classifications: compact and extended. While this approach is useful to identify very extended emission line regions (likely produced by AGN outflows), it is less helpful in moderately extended/compact regions (like the 18 galaxies discussed in this paper, as indicated by Comerford et al. 2012). First, it does not provide a quantitative measurement of the size of the emitting line region, so sources classified as compact might be in reality extended (especially if the emission is faint), and second, the detection of "compact" [O iii] emitting regions should not be taken as evidence for or against AGN outflows or dual AGNs (especially when the real size in parsecs is not known), because extended emission line regions (>10 kpc) have been observed in confirmed dual AGNs (such as NGC 6240 and Mrk 266) and compact emission line regions (<1 kpc) have been observed in single AGNs with outflows (Müller-Sánchez et al. 2011).

In order to provide the base for a quantitative analysis, we have measured the size of the [O iii] emitting region using the Akaike information criterion (AIC), defined by Akaike (1974). Briefly, the method works as follows (see Figure 19): we first select a two-dimensional (2D) window with a 100 pixel width in the spectral direction centered on [O iii] so that it contains the entire emission line plus continuum on both sides of the line, and a 20''–30'' width on the spatial direction. Then, for each row in the spatial direction, we fit both a two-parameter and five-parameter model. These models correspond to a line with a given slope (two parameters) and a single Gaussian combined with a line with a given slope (five parameters). Then, we used the AIC to determine which model is a better fit. The AIC is defined as: AIC = ${\chi }^{2}+2k,$ where k is the number of parameters. It penalizes for the use of more parameters when a fit with more variables is unnecessary. We used the corrected AIC_c for finite sample size: AIC_c = AIC $+2k(k+1)/(n-k-1)$, where n is the sample size. A smaller AIC_c indicates a better relative fit between the two models. We applied this criterion for each row in the spatial direction and determined the last row where the more complex Gaussian + linear fit has a lower AIC_c statistic (the Akaike width). This corresponds to the row of last significant emission, and therefore the size of the [O iii] emitting region. The uncertainties were estimated using Monte Carlo techniques. The method involves adding noise to the galaxy spectra, and refitting the result to yield a new set of Akaike widths. After repeating this 500 times, the standard deviation of the distribution of the Akaike width was used as the uncertainty for the size of the [O iii] emitting region.

Zoom In Zoom Out Reset image size

With the Gaussian fits in hand, we created surface brightness radial profiles of the [O iii] flux distribution for each position angle of the slit, by measuring the area under the Gaussian of each spectrum along the spatial direction. We compared these profiles with the radial profiles of the continuum emission and the point-spread function (PSF; see Figure 19). The continuum measurements were taken at both sides of the [O iii] emission line in windows of 20 pixels along the wavelength axis. We fit a Gaussian in the spatial direction for each continuum window, and then took the average of the two Gaussians. The PSF profiles (Gaussian) were constructed in a similar way, but using standard stars observed closest in time to the science frames and a single spectral window of 40 pixels. For each galaxy, we measured the spatial extent of the continuum and [O iii] emission at 5% and 50% of the peak flux.

The results are shown in Table 6. An example of the radial profiles of the three components ([O iii] flux, galaxy continuum, PSF) is shown in Figure 19 for the case of J1023+3243. For completeness, we present the measurements obtained at each position angle of the slit in Table 6, but the discussion presented here is based only on the PA_slit exhibiting the most extended emission (column 4 in Table 6). It is also important to point out that these measurements are position-angle-dependent, and therefore the results presented here represent a lower limit on the size of the emitting region, except in those cases where one of the PA_slit is consistent (within the errors of ±15°) with ${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$ (the position angle of the most extended emission, Comerford et al. 2012).

The continuum emission is resolved (${\mathrm{FWHM}}_{\mathrm{cont}}\;\gt$ FWHM_PSF) in all cases, but the [O iii] emission is resolved in only 10 galaxies (${\mathrm{FWHM}}_{[{\rm{O}}\;{\rm{III}}]}\;\gt$ FWHM_PSF). In fact, most of the [O iii] emission consists of a marginally resolved/unresolved core and extended emission out to several kiloparsecs. Also, in all cases the FWHM of the continuum is larger than the FWHM of the [O iii] emission. However, in three galaxies, the spatial extent of the [O III] emission at 5% of the peak flux is larger than the size of the continuum at the same percentage of the peak value. One of them (J1715+6008) has in fact the most extended [O iii] emitting region of the whole sample (∼30 kpc). In order to compare the extent of the [O iii] emission in all galaxies, we have plotted in Figure 20 the surface brightness radial profiles of the [O iii] flux distribution. The results will be discussed in Section 4.1.

Zoom In Zoom Out Reset image size

Four measurements are used to characterize the radio emission of the sample galaxies: radio morphology, size of the central bright sources, spectral index value of the compact sources, and spatial distribution of the spectral index (spectral index image). The radio morphologies of the double-peaked AGNs in our sample comprise three categories: (1) galaxies presenting a single radio source, (2) galaxies with extended radio features above the 3σ level, and (3) galaxies showing two independent radio sources (two bright peaks of emission). We used the task IMFIT in CASA to fit a single elliptical Gaussian to each of the compact sources detected in each galaxy (except for J1158+3231 in which we fitted simultaneously two elliptical Gaussian components, see Figure 14). The parameters of the Gaussians (flux density, size, and position angle) are summarized in Tables 2–5. A source is considered compact if it is unresolved or its size is comparable to the size of the beam (within the errors).

From the flux densities at 8.5 and 11.5 GHz, we obtained an X-spectral index α (where $S\propto {\nu }^{\alpha }$) for each of the Gaussian components. In general, compact flat and slightly steep-spectrum radio sources ($\alpha \gtrsim -0.8$) are interpreted as AGNs.⁸ Spectral index images of AGNs typically show a flat spectrum in the region of the core, though the spectrum can be inverted if significant synchrotron self-absorption is occurring (Edwards et al. 2000). These sources are described as young AGNs in the early stages of development, before expansion to a larger size (the spectral peak of an expanding radio source moves toward lower frequencies; Vollmer et al. 2008), and are common among Type 2 quasars (Lal & Ho 2010). Jet components usually have steeper spectra ($\alpha \lesssim -0.8$) due to aging of the electron population, as higher energy electrons lose energy via synchrotron radiation faster than those at lower energies (e.g., Readhead et al. 1979; Fomalont 1981; Marscher 1988; Pushkarev et al. 2012). Thus, neglecting edge effects (the spectral index images become more uncertain with decreasing continuum brightness) in AGNs where the core dominates (no evidence for radio jets down to the limits of sensitivity and spatial resolution), only a constant flat spectral index is observed in the region of the radio core, with typical values in the range $-0.8\lesssim \alpha \lesssim 1.$ On the other hand, spatial variations of the spectral index (an overall steepening of the spectra with distance, sometimes even within the size of the VLA beam) is a common characteristic of radio jets (Hovatta et al. 2014).

We also measured the position angle of the extended radio emission from the lowest $3\sigma$ contours in the total intensity images (PA_radio; Table 7). For the galaxies showing two compact radio cores, this corresponds to the position angle of the vector between the two sources on the sky. This angle was not measured in galaxies presenting only a single radio core, since this corresponds to the position angle of the elliptical Gaussian (Table 2).

Table 2.  Summary of Radio Properties of the Single AGNs

				Beam Parameters		Source Parameters
SDSS Name	Source	Frequency	rms	${\theta }_{M}\times {\theta }_{m}$	PA	${\theta }_{M}\times {\theta }_{m}$	PA	Flux Density	α
		(GHz)	(μJy beam−1)	($^{\prime\prime}$)	($^\circ {\rm{E}}$ of N)	('')	($^\circ {\rm{E}}$ of N)	(mJy)
J0736+4759	1	8.5	10.2	0.19 × 0.16	−19.9	0.18 × 0.17	3.8	1.0	−0.61 ± 0.14
	1	11.5	12.5	0.17 × 0.12	−178.0	0.18 × 0.12	1.8	0.79
	1	11.5	11.3	0.19 × 0.16	−19.9	0.18 × 0.17	1.7	0.83
J0802+3046	1	8.5	7.5	0.18 × 0.16	−51.2	0.23 × 0.18	−8.6	0.53	−0.19 ± 0.16
	1	11.5	9.3	0.16 × 0.13	1.7	0.21 × 0.13	−8.1	0.45
	1	11.5	9.5	0.18 × 0.16	−51.2	0.23 × 0.14	−7.7	0.50
J0846+4258	1	8.5	7.8	0.18 × 0.15	−39.7	0.21 × 0.18	−5.0	0.17	−0.41 ± 0.18
	1	11.5	10.1	0.16 × 0.12	−9.1	0.16 × 0.12	−6.5	0.12
	1	11.5	10.5	0.18 × 0.15	−39.7	0.18 × 0.15	−4.5	0.15
J0916+2835	1	8.5	8.9	0.19 × 0.17	−77.4	0.2 × 0.18	107.1	1.20	−0.66 ± 0.13
	1	11.5	9.9	0.16 × 0.14	−5.5	0.17 × 0.14	5.1	0.90
	1	11.5	9.5	0.19 × 0.17	−77.4	0.18 × 0.15	5.7	0.98
J1112+2750	1	8.5	7.7	0.19 × 0.18	84.0	0.19 × 0.19	0.8	0.72	−0.34 ± 0.14
	1	11.5	8.4	0.17 × 0.15	14.0	0.17 × 0.15	178.4	0.52
	1	11.5	8.8	0.19 × 0.18	84.0	0.19 × 0.16	−1.5	0.65
J1556+0948	1	8.5	8.8	0.25 × 0.19	−62.5	0.26 × 0.19	−61.0	0.80	0.73 ± 0.14
	1	11.5	17.9	0.20 × 0.16	82.0	0.22 × 0.16	79.5	0.96
	1	11.5	17.5	0.25 × 0.19	−62.5	0.25 × 0.19	77.7	1.0

Notes. The uncertainties in the size and PA of the sources are 003–005 and 4°–7°, respectively, from IMFIT in CASA. Errors on the measured flux densities are approximately 3.5%, dominated by the uncertainty in the flux density scale. It corresponds to the sum of the error reported by IMFIT in CASA and the 3% calibration error, added in quadrature.

Download table as:  ASCIITypeset image

Finally, we estimated the possible contribution from star formation to the radio flux density of our sources. Following Kauffmann et al. (2003), typical Hα star formation rates for our sample are about $1\;{M}_{\odot }\;{\mathrm{yr}}^{-1}$, leading to an expected contribution of only about 0.02 mJy at 1.4 GHz (Murphy et al. 2011). Assuming a flat spectrum between −0.3 and −0.1, this corresponds to $\lesssim 0.01$ mJy at 11.5 GHz. This is less than $10\%$ of the flux density of each of the sources. The only exception is Source 2 in J1023+3243 in which the 11.5 GHz flux density due to star formation is over a factor of 3 lower than our measured 11.5 GHz flux of 0.03 mJy. Nevertheless, this suggests that an additional emission source, such as a weak AGN, would be needed to provide the missing flux density. Therefore, in our sample galaxies, a contribution from star formation is expected to be negligible. Furthermore, all the galaxies in our sample have optical line ratios that are consistent with AGN-dominated emission rather than star formation (Wang et al. 2009; Liu et al. 2010a).

Based on the VLA data alone, we classify the 18 galaxies into four categories.⁹

• 1.  
  Single AGNs: The VLA images show a single compact radio source. The spectral index is flat or inverted ($\alpha \gtrsim -0.8$) and constant in the spectral index images (neglecting edge effects). These are typical characteristics of an AGN. Six galaxies are included in this category (Table 2).
• 2.  
  Radio Jets: Extended jet-like radio structures are detected above the $3\sigma$ level in seven galaxies (Table 3). The morphologies mostly consist of a small-diameter radio component (in most cases unresolved or marginally resolved) and extended linear features that emanate from one side, or from opposite sides of the radio core. The small central radio components have a steep spectrum $\alpha \lt -0.9$ (Table 3), which probably corresponds to combined emission from the radio core (the AGN) and the radio jet. The only exceptions are J0009–0036 (which shows an inverted spectrum α = 0.36), and J0930+3430 (which has a flat spectrum α = −0.35). Therefore, in these two cases, the radio emission of the core is dominated by the AGN. In all cases, the spectral index images show variations of the spectral index with distance along the linear features. It is obvious from the difference in the core and jet spectral indices that some steepening in the index is occurring along the jet, with values of the core and jet components in the range $-3\lesssim \alpha \lesssim 2,$ typical of jets in AGNs (Hovatta et al. 2014). Note that the spectral index image for J0731+4528 could not be obtained, since the source was not detected at 11.5 GHz. However, this indicates that the source has a very steep spectrum and is consistent with the jet interpretation.
• 3.  
  Dual AGNs: Three galaxies show two spatially separated radio sources in our VLA images (Table 4, Figures 10, 14 and 16). The properties of each of the sources are consistent with those of single AGNs as described above (they are compact and the spectral index is flat and constant within the FWHM of the sources), which confirm the presence of two AGNs in these galaxies. There is no evidence of extended radio emission, or diffuse radio emission between the sources, which suggests that the morphological structures most likely are not associated with radio jets. This fact and the lack of axial symmetry between the two sources suggest that they are not symmetric hotspots of an obscured AGN¹⁰ (An & Baan 2012; Wrobel et al. 2014b). This is supported by the spectral index analysis. For high-power compact symmetric objects (CSOs), spectral indices steepen with source separation (An & Baan 2012). Such trends have not yet been investigated for low-power CSOs. However, if a similar trend exists, the spectral indices shown in Table 4 would then be very atypical for the source separations of 0.6–1.6 kpc. Therefore, based on the radio data alone, we conclude that three galaxies in our sample host dual AGNs.¹¹
• 4.  
  Ambiguous: The total intensity images show two compact radio sources, but the properties of one of them cannot be measured either at 8.5 or 11.5 GHz. The two sources are embedded in general diffuse emission, which suggests the presence of radio jets. Therefore, the two sources in the total intensity images could be interpreted either as knots or hotspots in radio jets, or dual AGNs. Two galaxies exhibit these characteristics (Table 5, Figures 1 and 11), and will be discussed in detail in Section 3.2.5.

Table 3.  Summary of Radio Properties of the Extended Radio Sources

				Beam Parameters		Source Parameters
SDSS Name	Source	Frequency	rms	${\theta }_{M}\times {\theta }_{m}$	PA	${\theta }_{M}\times {\theta }_{m}$	PA	Flux Density	α
		(GHz)	(μJy beam−1)	('')	($^\circ {\rm{E}}$ of N)	('')	($^\circ {\rm{E}}$ of N)	(mJy)
J0009–0036	1	8.5	11.0	0.27 × 0.19	−44.0	0.30 × 0.20	30.0	7.8	0.36 ± 0.13
	1	11.5	16.0	0.19 × 0.17	−56.1	0.20 × 0.18	57.6	8.2
	1	11.5	16.2	0.27 × 0.19	−44.0	0.27 × 0.19	56.7	8.7
J0731+4528	1	8.5	5.2	0.25 × 0.17	−37.1	0.38 × 0.21	138.7	0.17	...
	1	11.5	9.6	0.17 × 0.12	−177.3	...	...	...
	1	11.5	8.5	0.25 × 0.17	−37.1	...	...	...
J0858+1041	1	8.5	7.4	0.2 × 0.18	81.9	0.22 × 0.2	85.6	0.90	−1.23 ± 0.14
	1	11.5	8.4	0.17 × 0.14	25.3	0.19 × 0.17	172.3	0.59
	1	11.5	7.9	0.2 × 0.18	81.9	0.2 × 0.18	172.9	0.62
J0930+3430	1	8.5	6.6	0.18 × 0.16	−41.0	0.18 × 0.17	106.6	0.20	−0.35 ± 0.16
	1	11.5	7.6	0.16 × 0.13	−3.9	0.19 × 0.13	144.7	0.17
	1	11.5	8.0	0.18 × 0.16	−41.0	0.19 × 0.15	147.9	0.18
J1152+1903	1	8.5	7.6	0.19 × 0.18	29.7	0.24 × 0.18	30.9	0.20	−1.18 ± 0.17
	1	11.5	7.8	0.18 × 0.14	19.1	0.24 × 0.15	27.6	0.11
	1	11.5	8.0	0.19 × 0.18	29.7	0.24 × 0.16	26.5	0.14
J1715+6008	1	8.5	9.2	0.25 × 0.15	66.4	0.31 × 0.25	139.3	1.12	−1.1 ± 0.13
	1	11.5	18.8	0.22 × 0.11	67.5	0.26 × 0.21	140.5	0.66
	1	11.5	18.5	0.25 × 0.15	66.4	0.27 × 0.22	138.6	0.80
J2254–0051	1	8.5	7.3	0.25 × 0.19	−49.1	0.29 × 0.19	114.8	0.20	−0.95 ± 0.17
	1	11.5	8.8	0.18 × 0.17	71.1	0.21 × 0.19	70.5	0.15
	1	11.5	8.6	0.25 × 0.19	−49.1	0.25 × 0.19	67.2	0.15

Note. Same as Table 2, but for the extended radio sources.

Download table as:  ASCIITypeset image

Table 4.  Summary of Radio Properties of the Confirmed Dual AGNs

				Beam Parameters		Source Parameters
SDSS Name	Source	Frequency	rms	${\theta }_{M}\times {\theta }_{m}$	PA	${\theta }_{M}\times {\theta }_{m}$	PA	Flux Density	α	${D}_{\mathrm{radio}}$	${D}_{[{\rm{O}}\;{\rm{III}}]}$
		(GHz)	(μJy beam−1)	($^{\prime\prime}$)	($^\circ {\rm{E}}$ of N)	($^{\prime\prime}$)	($^\circ {\rm{E}}$ of N)	(mJy)		('')	('')
J1023+3243	1	8.5	7.2	0.18 × 0.16	−43.0	0.22 × 0.16	131.0	0.25	0.25 ± 0.16	0.45 ± 0.05	0.58 ± 0.03
	1	11.5	8.3	0.16 × 0.13	−2.2	0.23 × 0.16	148.3	0.27		0.42 ± 0.06
	1	11.5	8.7	0.18 × 0.16	−43.0	0.22 × 0.16	146.6	0.27		0.42 ± 0.06
	2	8.5	7.2	0.18 × 0.16	−43.0	0.26 × 0.25	178.0	0.06	−0.29 ± 0.19	0.45 ± 0.05	0.58 ± 0.03
	2	11.5	8.3	0.16 × 0.13	−2.2	0.15 × 0.13	59.0	0.03		0.42 ± 0.06
	2	11.5	8.7	0.18 × 0.16	−43.0	0.22 × 0.16	121.8	0.055		0.42 ± 0.06
J1158+3231	1	8.5	8.0	0.18 × 0.17	−49.0	0.2 × 0.18	35.4	0.3	−0.47 ± 0.17	0.22 ± 0.04	0.27 ± 0.01
	1	11.5	8.5	0.17 × 0.14	−174	0.19 × 0.14	37.8	0.25		0.22 ± 0.04
	1	11.5	8.8	0.18 × 0.17	−49.0	0.2 × 0.18	30.6	0.26		0.22 ± 0.04
	2	8.5	8.0	0.18 × 0.17	−49.0	0.19 × 0.17	35.4	0.29	−0.49 ± 0.17	0.22 ± 0.04	0.27 ± 0.01
	2	11.5	8.5	0.17 × 0.14	−174.0	0.18 × 0.15	37.8	0.23		0.22 ± 0.04
	2	11.5	8.8	0.18 × 0.17	−49.0	0.19 × 0.17	30.6	0.25		0.22 ± 0.04
J1623+0808	1	8.5	8.7	0.27 × 0.19	−62.1	0.28 × 0.20	−60.3	0.24	0.26 ± 0.16	0.47 ± 0.05	0.37 ± 0.06
	1	11.5	10.3	0.22 × 0.17	−80.7	0.25 × 0.18	87.8	0.25		0.44 ± 0.05
	1	11.5	10.8	0.27 × 0.19	−62.1	0.27 × 0.20	88.6	0.26		0.47 ± 0.05
	2	8.5	8.7	0.27 × 0.19	−62.1	0.27 × 0.20	−67.4	0.10	−0.17 ± 0.18	0.47 ± 0.05	0.37 ± 0.06
	2	11.5	10.3	0.22 × 0.17	−80.7	0.23 × 0.17	14.0	0.09		0.44 ± 0.05
	2	11.5	10.8	0.27 × 0.19	−62.1	0.27 × 0.19	5.9	0.095		0.47 ± 0.05

Note. Same as Table 2, but for the confirmed dual AGNs. The parameter ${D}_{\mathrm{radio}}$ indicates the distance between the two strongest radio sources detected in each image. The uncertainty in ${D}_{\mathrm{radio}}$ was calculated via standard error propagation of the errors in the positions of the two sources. ${D}_{[{\rm{O}}\;{\rm{III}}]}$ indicates the projected spatial separation between the two [O iii] emission features (from Comerford et al. 2012).

Download table as:  ASCIITypeset image

Table 5.  Summary of Radio Properties of the Ambiguous Cases

				Beam Parameters		Source Parameters
SDSS Name	Source	Frequency	rms	${\theta }_{M}\times {\theta }_{m}$	PA	${\theta }_{M}\times {\theta }_{m}$	PA	Flux Density	α	${D}_{\mathrm{radio}}$	${D}_{[{\rm{O}}\;{\rm{III}}]}$
		(GHz)	(μJy beam−1)	($^{\prime\prime}$)	($^\circ {\rm{E}}$ of N)	($^{\prime\prime}$)	($^\circ {\rm{E}}$ of N)	(mJy)		('')	('')
J0002+0045	1	8.5	6.6	0.22 × 0.18	−46.0	0.24 × 0.21	96.1	0.35	−0.62 ± 0.15	0.39 ± 0.05	0.58 ± 0.05
	1	11.5	10.3	0.19 × 0.17	−73.0	0.25 × 0.18	77.3	0.25		...
	1	11.5	10.5	0.22 × 0.18	−46.0	0.25 × 0.18	77.6	0.29		...
	2	8.5	6.6	0.22 × 0.18	−46.0	0.38 × 0.27	72.1	0.22	...	0.39 ± 0.05	0.58 ± 0.05
	2	11.5	10.3	0.19 × 0.17	−73.0	...	...	...		...
	2	11.5	10.5	0.22 × 0.18	−46.0	...	...	...		...
J1027+3059	1	8.5	7.4	0.18 × 0.16	−47.0	0.26 × 0.20	93.8	0.18	−0.83 ± 0.19	...	0.31 ± 0.02
	1	11.5	8.0	0.16 × 0.13	−64.2	0.28 × 0.19	107.2	0.14		0.22 ± 0.04
	1	11.5	8.7	0.18 × 0.16	−47.0	0.26 × 0.19	113.4	0.14		0.22 ± 0.04
	2	8.5	7.4	0.18 × 0.16	−47.0	...	...	...	...	...	0.31 ± 0.02
	2	11.5	8.0	0.16 × 0.13	−64.2	0.17 × 0.16	4.6	0.08		0.22 ± 0.04
	2	11.5	8.7	0.18 × 0.16	−47.0	0.18 × 0.16	−4.9	0.08		0.22 ± 0.04

Note. Same as Table 4, but for the ambiguous cases.

Download table as:  ASCIITypeset image

Table 6.  Summary of Results from the Analysis of the Spatial Extent of [O iii] Emission

SDSS ID	PAslit	${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$	Total Sizea	FWHMcont	${\mathrm{FWHM}}_{[{\rm{O}}\;{\rm{III}}]}$	Size 5% Cont	Size 5% [O iii]	FWHMPSF
			''(kpc)	('')	('')	('')	('')	('')
J0002+0045	68	63.9	7.80 (12.64)	4.09	2.27	8.68	7.80	1.59
	158		7.80 (12.64)	3.25	2.06	6.99	4.58
J0009–0036	23	54.2	4.03 (5.60)	1.23	0.77	2.65	1.85	0.94
	67		5.18 (7.20)	1.35	0.89	2.83	2.05
J0731+4528	21	6	3.12 (4.90)	4.34	3.13	9.12	3.12	3.67
	111		7.02 (11.02)	4.36	2.36	9.15	5.39
J0736+4759	86	98.7	4.03 (7.17)	2.78	1.85	5.79	4.03	0.91
	176		4.03 (7.17)	1.79	1.20	3.77	3.12
J0802+3046	80	150.0	9.34 (13.54)	4.09	3.13	8.52	7.49	3.71
	170		10.11 (14.66)	4.89	3.13	10.17	8.17
J0846+4258	38	175.7	2.88 (10.20)	1.81	1.50	3.81	2.78	0.99
	128		2.88 (10.20)	2.51	1.33	5.23	2.88
J0858+1041	17	157.3	5.45 (14.06)	3.37	2.27	7.01	5.20	2.09
	107		7.00 (18.06)	3.25	2.11	6.81	5.96
J0916+2835	17	40.8	7.00 (17.50)	2.73	2.14	5.73	5.28	2.09
	107		7.80 (19.50)	2.89	2.12	6.06	5.38
J0930+3430	21	105.9	4.67 (5.51)	2.04	1.68	4.33	3.84	3.04
	111		7.00 (8.26)	3.82	2.33	7.98	5.73
J1023+3243	17	137.4	7.00 (15.90)	5.11	2.52	10.66	5.81	2.09
	107		5.72 (13.00)	3.72	2.46	7.75	5.72
J1027+3059	13	76.3	4.61 (10.28)	1.60	1.04	3.37	2.43	1.13
	103		5.76 (12.84)	1.55	1.19	3.27	3.00
J1112+2750	12	4.9	10.92 (10.16)	2.98	2.86	6.22	6.87	2.41
	102		10.14 (9.43)	4.14	2.51	8.77	10.07
J1152+1903	17	60.6	7.00 (12.53)	3.85	2.00	8.04	5.79	2.09
	107		5.45 (9.76)	3.20	2.18	6.71	5.45
J1158+3231	8	13.0	2.30 (6.53)	1.24	0.74	2.66	1.90	1.02
	98		2.88 (8.18)	1.05	0.86	2.24	1.97
J1556+0948	46	134.7	6.24 (8.11)	2.54	1.74	5.53	6.24	1.98
	174		3.11 (4.04)	2.96	1.82	6.19	3.11
J1623+0808	46	28.9	2.88 (9.50)	1.67	1.58	3.48	2.88	1.02
	136		3.46 (11.41)	1.66	1.34	3.46	3.46
J1715+6008	31	145.6	10.92 (29.60)	2.57	0.88	5.62	6.89	1.63
	121		4.68 (12.68)	2.47	1.02	5.41	4.68
J2254–0051	35	117.4	2.88 (4.32)	1.25	0.76	2.68	2.25	0.94
	55		3.46 (5.20)	1.74	0.49	3.66	2.42

Note.

^aSize of the [O iii] emitting region (based on the Akaike information criterion; see Section 2.3). The errors in ${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$ and total size are shown in Table 7, but omitted in this table for compactness.

Download table as:  ASCIITypeset image

These four categories represent our best effort to interpret these galaxies based on the current VLA data. If additional structures exist in the central $3^{\prime\prime} \times 3^{\prime\prime}$ of these galaxies (the diameter of the SDSS fiber), then they would have to have radio emission below the detection limit in our images (the rms image noises were typically 10 μJy beam⁻¹, giving a 3σ detection threshold of ∼30 μJy beam⁻¹), or projected separations smaller than our angular resolution (∼018). Keeping this in mind, we adopt this classification scheme in our following discussion.

Combining the observed radio properties and the measurements obtained from our optical long-slit spectroscopy (Comerford et al. 2012), we now discuss the scenarios that may give rise to the double-peaked narrow emission lines in our AGN sample.

Our method makes use of morphological parameters derived from optical data.¹² Two parameters are crucial for the identification process: the position angle of the [O iii] emission (${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$) and the photometric position angle of the major axis of the galaxy (PA_gal). In galaxies exhibiting two compact radio sources in the total intensity images (dual AGNs and ambiguous cases, Section 3.1), the projected spatial separation between the two [O iii] emission features (${D}_{[{\rm{O}}\;{\rm{III}}]}$) is also relevant.¹³

${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$ and ${D}_{[{\rm{O}}\;{\rm{III}}]}$ are taken directly from Comerford et al. (2012). Since we observed each galaxy at two position angles, the full spatial separation of the two emission components on the sky as well as their position angle could be determined (see details in Comerford et al. 2012). We used SDSS photometry to derive PA_gal, except in J0002+0045, J0736+4759 and J1715+6008, where we used HST F106W images (Figure 21, P.I. Liu). The main results of the identification process are summarized in Table 7, and each case is discussed individually in the next subsections.

Zoom In Zoom Out Reset image size

Table 7.  Summary of Results

SDSS Name	Radio Classification	PAradioa	${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$ b	PAgalc	Size [O iii]d	Origin Double Peakse
		($^\circ {\rm{E}}$ of N)	($^\circ {\rm{E}}$ of N)	($^\circ {\rm{E}}$ of N)	(kpc)
J0002+0045	ambiguous	64.8 ± 6.4	63.9 ± 5.0	64.8 ± 3.1	12.64 ± 1.29	ambiguous
J0009–0036	two-sided radio jet	40.5 ± 6.8	51.2 ± 3.9	135.3 ± 3.9	7.20 ± 0.43	radio jet-driven outflow
J0731+4528	two-sided radio jet	108.3 ± 5.9	6.2 ± 6.7	154.5 ± 6.7	11.02 ± 1.08	AGN-driven outflow
J0736+4759	single AGN	...	98.7 ± 3.8	94.0 ± 3.0	7.17 ± 0.74	rotating disk
J0802+3046	single AGN	...	150.0 ± 1.9	170.5 ± 1.9	14.66 ± 0.87	AGN-driven outflow
J0846+4258	single AGN	...	175.7 ± 8.6	51.9 ± 8.6	10.20 ± 1.59	AGN-driven outflow
J0858+1041	one-sided radio jet	147.0 ± 7.8	157.3 ± 3.5	85.3 ± 3.5	18.06 ± 0.83	radio jet-driven outflow
J0916+2835	single AGN	...	40.8 ± 1.3	81.4 ± 1.3	19.50 ± 1.22	AGN-driven outflow
J0930+3430	one-sided radio jet	177.0 ± 9.2	105.9 ± 2.2	110.3 ± 2.2	8.26 ± 0.70	AGN-driven outflow
J1023+3243	dual AGNs	141.3 ± 8.3	137.4 ± 8.5	125.3 ± 8.5	15.90 ± 1.73	dual AGNsf
J1027+3059	ambiguous	85.2 ± 8.7	76.3 ± 2.4	167.5 ± 2.4	12.84 ± 0.64	ambiguous
J1112+2750	single AGN	...	4.9 ± 6.5	87.4 ± 6.5	10.16 ± 0.84	AGN-driven outflow
J1152+1903	two-sided radio jet	48.6 ± 7.7	60.2 ± 5.1	37.0 ± 5.1	12.53 ± 0.99	radio jet-driven outflow
J1158+3231	dual AGNs	21.2 ± 7.1	13.0 ± 4.5	8.0 ± 4.5	8.18 ± 0.9	dual AGNs
J1556+0948	single AGN	...	134.7 ± 6.6	159.8 ± 6.6	8.11 ± 1.16	AGN-driven outflow
J1623+0808	dual AGNs	29.5 ± 7.5	27.9 ± 4.8	46.4 ± 4.8	11.41 ± 0.9	dual AGNs
J1715+6008	one-sided radio jet	147.5 ± 7.0	145.6 ± 6.8	33.9 ± 3.5	29.60 ± 1.48	radio jet-driven outflow
J2254–0051	one-sided radio jet	110.2 ± 7.5	117.4 ± 5.3	114.8 ± 5.3	5.20 ± 0.64	radio jet-driven outflow

Notes.

^aPosition angle of the extended radio emission. This was measured from the lowest $3\sigma$ contours in the total intensity images. For the dual AGNs, this corresponds to the position angle of the vector between the two compact radio cores on the sky. The typical error is $\sim 8^\circ .$ This angle was not measured in galaxies presenting only a single radio core, since this corresponds to the PA of the elliptical Gaussian (Table 2). ^bPosition angle of the two components of [O iii] emission, from Comerford et al. (2012). The typical error in ${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$ is $\sim 7^\circ$ (Comerford et al. 2012). ^cPosition angle of the photometric major axis of the galaxy, from SDSS photometry (with the same error as in Column 4, Comerford et al. 2012), except in J0002+0045, J0736+4759, and J1715+6008, where we used HST F106W images (Figure 21). ^dTotal extent of the [O iii] emitting region from Table 6. ^e"AGN-driven outflows" refers to the cases which are not radio-jet driven. Radiation pressure, magnetic fields, and thermal winds from the AGN could all power the outflow (e.g., Müller-Sánchez et al. 2011). See Section 3.2.5 for details on the ambiguous cases. ^fAn outflowing velocity component is present in this galaxy and might have a significant impact on the velocity offsets of the two peaks (Sections 3.2.1 and 4.1).

Download table as:  ASCIITypeset image

3.2.1. Optical Spectral Signature of Dual AGNs

3.2.2. Radio Jet-driven Outflows

3.2.3. AGN Wind-driven Outflows

The radio images of the galaxies classified as single AGNs do not provide sufficient information to identify the origin of the double-peaked, spatially offset narrow AGN emission lines (Table 2). In the absence of extended radio emission and/or a secondary radio core, the most likely scenarios would be either an AGN outflow that is not radio-jet-driven or disk rotation. Using morphological information from our optical long-slit data, we can distinguish between these scenarios. Disk rotation is ruled out when the emission line gas visible in the long-slit spectra is not located in the plane of the galaxy (${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$ ≁ PA_gal). However, in the cases where ${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$ ∼ PA_gal, the situation is more complicated since the double-peaked narrow emission lines could be produced either by disk rotation or by an AGN outflow that is intersecting the plane of the galaxy (Müller-Sánchez et al. 2011). In these cases a more detailed analysis of the gas kinematics is necessary to distinguish between these two scenarios.

In five galaxies classified as single AGNs (J0802+3046, J0846+4258, J0916+2835, J1112+2750, and J1556+0948), the two components of [O iii] emission are located at relatively large angles from the galactic disk¹⁵ , indicating that the ionized gas is not rotating in the plane of the galaxy (${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}\;\nsim$ PA_gal). Therefore, we conclude that the double-peaked narrow emission lines in these objects are produced by an AGN wind-driven outflow (Table 7).

J0731+4528 (Table 3) shows an extended jet-like feature along a position angle of ∼108°, almost perpendicular to the position angle of the two [O iii] emission components (∼6°, Figure 3). The distinct orientations of the two structures (the radio jet and the [O iii] gas) suggest that the radio jet is not influencing the kinematics of the ionized gas on scales of hundreds of parsecs (${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$ ≁ PA_radio). Therefore, we conclude that the double-peaked narrow emission lines in this galaxy are not produced by radio-jet-driven outflows. Since ${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$ ≁ PA_gal, the gas is not rotating in the plane of the galaxy, and therefore the most plausible explanation is that the double-peaked [O iii] lines are produced by an AGN outflow that is not radio-jet-driven.

As was the case for J1715+6008 in the previous section, we now discuss in more detail two galaxies in this category which show additional weak radio components (J1112+2750), or additional components at other wavelengths (J0916+2835).

In J0916+2835 a secondary stellar component has been detected at near-IR wavelengths 123 north of the nucleus of the galaxy (Fu et al. 2012). This component, however, does not have a counterpart in [O iii] emission (Fu et al. 2012), is very weak at 8.5 GHz (∼2σ level in our 8.5 GHz image), and is not present in our 11.5 GHz image. Furthermore, the size of the [O iii] emitting region is ∼20 kpc, which is one of the most extended regions in our sample (Figure 20 and Table 7). Deeper radio observations may enable a firm detection of this secondary source and confirm the system as dual AGNs. However, based on the data we have, we consider J0916+2835 as a galaxy with a single radio core, and the optical long-slit spectroscopy suggests an AGN-driven outflow.

In J1112+2750, a second source of radio emission is detected at ∼09 NW from the central radio core (Figure 12). This source, however, is very weak (3.4σ at 8.5 GHz and 2.2σ at 11.5 GHz). At this sensitivity, the source is not compact, which argues against the presence of a secondary radio core and instead suggests a hotspot-like radio feature. Since its radio properties could not be well measured, we consider this object as a galaxy with a single radio core. The central source is compact and shows a flat spectrum consistent with an AGN (Table 2).

3.2.3.1. J0930+3430: An AGN Wind-driven Outflow in the Plane of the Galaxy

The 8.5 GHz image of J0930+3430 shows a one-sided radio jet that is almost perpendicular to the plane of the galaxy and the [O iii] emission (Figure 9). In the 11.5 GHz image, the flux density of this structure is comparable to the residuals in the cleaning process, making its association with the radio jet uncertain. If this structure is not real, then J0930+3430 would exhibit the same radio properties of single AGNs (Section 3.1). However, even if the jet is real, it is not influencing the kinematics of the ionized gas, as indicated by the difference in position angles (${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}\;\nsim$ PA_radio).

Since ${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$ ∼ PA_gal (Table 7, Figure 9), the origin of the double-peaked narrow emission lines in this galaxy is in the plane of the galaxy. As mentioned in Section 3.2.3, in order to differentiate between disk rotation and AGN outflows intersecting the plane of the galaxy, we need to study in more detail the kinematics of the gas in our long-slit spectra. For the purposes of this paper, we look for evidence of disturbed kinematics in the spectral profiles of the emission lines.¹⁶ Multiple kinematic components in narrow emission lines are usually associated with outflows (see, e.g., Crenshaw et al. 2010 and references therein). Since all our galaxies show double-peaked spectral profiles, we consider more than two kinematic components to be evidence for an outflow.¹⁷ High radial velocities also are ideal tracers of outflows as they discount the likelihood that the observed kinematics could be due to rotation, where the line of sight velocity is rarely greater than 400 km s⁻¹ (Müller-Sánchez et al. 2011). Asymmetric wings (either blueshifted or redshifted) are also indicative of outflow kinematics (Crenshaw et al. 2010; Harrison et al. 2014). Finally, the presence of broad components (broader than forbidden lines from the NLR, typically ∼500 km s⁻¹ FWHM; Osterbrock & Pogge 1987), also suggest the presence of outflows, particularly when they are located at some distance from the nucleus (Müller-Sánchez et al. 2011).

The spectral profiles of J0930+3430 show three of the characteristics mentioned above¹⁸ : multiple components, asymmetric wings, and broad components. As can be seen in Figure 22, the two-Gaussian fits in the central region (from $-0\buildrel{\prime\prime}\over{.} 4$ to 04) do not provide a good match to the observed profiles, indicating the presence of a third component. Furthermore, the profiles at distances larger than ∼−07 show a broad asymmetric blue wing (∼600 km s⁻¹), which on the other side of the nucleus (at distances larger than ∼07) becomes the dominant component. At these locations, an asymmetric redshifted wing appears. All these characteristics provide enough evidence to conclude that J0930+3430 hosts an AGN outflow in the plane of the galaxy.

Zoom In Zoom Out Reset image size

3.2.4. J0736+4759: A Rotating Disk

As is the case for all of the galaxies in Table 2, in J0736+4759 we detect only a single flat-spectrum radio source. The ionized gas emission is spatially coincident with the galaxy's major axis (Figure 4), suggesting that the source of the double-peaked narrow emission lines is in the plane of the galaxy (${\mathrm{PA}}_{[{\rm{O}}\;{\rm{III}}]}$ ∼ PA_gal). Recent HST observations of this galaxy support this interpretation. As can be seen in Figure 21, the [O iii] emission and the near-IR continuum are spatially coincident. As was the case for J0930+3430, a more detailed analysis of the gas kinematics is needed in order to distinguish disk rotation from AGN outflows in the plane of the galaxy.

The spectral profiles of J0736+4759 do not show evidence for disturbed kinematics as described in Section 3.2.3.1. Two Gaussian components provide a good fit to the data (Figure 23). Furthermore, the two components are narrow (FWHM < 500 km s⁻¹), have low radial velocities (<150 km s⁻¹), and are symmetric. Although there might be a configuration in which an outflow produces narrow symmetric double-peaked profiles (Crenshaw et al. 2010), disk rotation provides a simpler explanation for the observed properties of J0736+4759.¹⁹ This is supported by the work of Smith et al. (2012), who studied a sample of double-peaked galaxies with symmetrical profiles and nearly similar components (like J0736+4759). Based on their statistical sample, they argued that a sample of double-peaked galaxies with nearly symmetrical line components have diagnostics in agreement with rotational disk models. We therefore conclude that the most likely origin of the double-peaked narrow emission lines in J0736+4759 is disk rotation.

Zoom In Zoom Out Reset image size

3.2.5. Ambiguous Cases

In Section 2.3 we developed a quantitative technique to revise the previous visual analysis of the spatial extent of sources by Comerford et al. (2012). Here we discuss the implications for each of the scenarios that give rise to the double-peaked narrow emission lines in Section 3.2.

The FWHM of the [O iii] emission ranges from 1 to 3.5 kpc with a mean FWHM${}_{[{\rm{O}}\;{\rm{III}}]}=2.3\;{\rm{kpc}},$ while the total size of the [O iii] emitting region ranges from 5 to 30 kpc with a mean size of 12.2 kpc (Figure 20 and Table 7). These measurements reflect the size of the compact bright core and the extended component, respectively. Interestingly, galaxies with outflows cover the whole range of sizes (the most compact [O iii] emitting region is observed in a galaxy with outflows, J2254–0051), while the only galaxy classified as rotating disk is compact (FWHM and total size are smaller than the mean values). The three dual AGNs discovered in this work appear in the three regimes (as compared with the mean value of total size of [O iii] emission): compact (J1158+3231), moderately extended (J1623+0808), and extended (J1023+3243). This suggests that searches for dual AGNs should not use the assumption that the [O iii] emitting region in dual AGNs is compact, since the region ionized by the combined effect of the two AGNs rotating in a galactic potential is probably larger than in the case of single AGNs. Furthermore, the gas kinematics in this region should also be complex with inflows and outflows in each of the two AGNs (as in the prototypical dual AGN system NGC 6240; Engel et al. 2010). Finally, we note that some of the normalized [O iii] radial profiles (Figure 20) exhibit a bump on the descending part of the curve after the FWHM level. This is only observed in galaxies with outflows (and one dual AGN system, J1023+3243), and can be attributed to low surface brightness outflowing clumps which appear as diffuse components in the 2D spectra. The fact that J1023+3243 exhibits this component suggests that one or both of the AGNs also has visible emission from gas kinematics or jets. This is supported by the large size of the [O iii] emitting region observed in this galaxy, and the fact that ${D}_{[{\rm{O}}\;{\rm{III}}]}$ is slightly larger than ${D}_{\mathrm{radio}}$ (Section 3.2.1 and Table 4).

4.1.1. Size–Luminosity Relation

An important question in the study of AGNs is how the size of the NLR scales with the luminosity of the central source. So far, most of the work on this subject has been concentrated on Seyferts (Fraquelli et al. 2003; Schmitt et al. 2003; Bennert et al. 2006) and nearby quasars (Bennert et al. 2002; Greene et al. 2011; Hainline et al. 2013; Husemann et al. 2013; Liu et al. 2013), but this is an unexplored territory for double-peaked AGNs.²⁰

The sample and measurements being used in this paper are ideal to determine if there is a correlation between NLR size and luminosity in double-peaked AGNs.²¹ If such a correlation exists for these galaxies, we can use this to improve our knowledge about the ionization structure of the NLR and the formation of double-peaked narrow emission lines in galaxies, as well as derive an extra input in photoionization/shock models.

We plot the derived NLR sizes (Table 7) against ${L}_{[{\rm{O}}\;{\rm{III}}]}$ (the total [O iii] luminosity within the $3^{\prime\prime}$ SDSS fiber; Wang et al. 2009; Liu et al. 2010a; Comerford et al. 2012) in Figure 24. This plot also presents a linear fit to the data (dashed line), which gives ${R}_{\mathrm{NLR}}\propto \;{L}_{[{\rm{O}}\;{\rm{III}}]}^{0.52\pm 0.14},$ with a correlation coefficient of 0.67. This relation goes as approximately ${R}_{\mathrm{NLR}}\propto \;{L}_{[{\rm{O}}\;{\rm{III}}]}^{1/2},$ similar to the result found for type 2 quasars by Bennert et al. (2002), but steeper than the observed relationship between ${R}_{\mathrm{NLR}}$ and ${L}_{[{\rm{O}}\;{\rm{III}}]}$ for Seyfert 2 galaxies (${R}_{\mathrm{NLR}}\propto \;{L}_{[{\rm{O}}\;{\rm{III}}]}^{0.33},$Schmitt et al. 2003). These results contrast with the shallow slope of 0.22–0.25 found for radio-quiet obscured quasars (Greene et al. 2012; Liu et al. 2013).²²

The NLR sizes from the galaxies in our AGN sample are slightly larger than what has been measured in other studies of active galaxies for the same range of [O iii] luminosities (${L}_{[{\rm{O}}\;{\rm{III}}]}$ between ${10}^{40.5}$ and ${10}^{42.5}$ erg s⁻¹; see Hainline et al. 2013 and references therein). As can be seen in Figure 24, our galaxies lie above the fit from Liu et al. (2013; slope = 0.25 ± 0.02; dotted line). We attribute this and the steeper slope of the observed double-peaked AGNs to an additional contribution of shocks to the ionized gas emission. All of the distinct scenarios proposed for the origin of double-peaked narrow lines (Section 3.2) suggest the presence of shocks (including dynamical processes in rotating disks and dual AGNs in galaxy mergers; see, e.g., Medling et al. 2015). Shock-enhanced emission can alter measurements of physical properties in host galaxies, including size and morphology of the NLR and galaxy's emission line ratios (Allen et al. 2008). It is also possible that since several double-peaked AGN are related to interacting systems, presenting tidal tails and bridges (Liu et al. 2013), these interactions can provide debris at large distances from the nucleus, which can be ionized by the central engine, increasing the size of the NLR. This could be also the case for the QSOs studied in Bennert et al. (2002), which also have a slope of ∼0.5 and are usually found in merging systems (Martini 2004).

Zoom In Zoom Out Reset image size

CSOs are small (∼1 kpc) radio sources that have symmetric double lobes or jets. The compact double components show steep radio spectra, similar to the lobes in large radio galaxies. In some CSOs, the core is clearly visible (Peck & Taylor 2000; Gugliucci et al. 2005), whereas in most CSOs the central cores are too weak to be detected. The dominant theory for the small size of these objects is that they are young radio sources that could grow into larger radio galaxies.

The "S-symmetry," sometimes observed in CSOs (Tremblay et al. 2009; Deane et al. 2014), has long been predicted to be associated with the presence of binary black holes (Begelman et al. 1980). Therefore, these systems are of great interest, since they might be tracers of merger-induced SMBH growth and potential gravitational wave emitters. To date, the closest separation SMBH binary system (∼7 pc), 0402+379, has been found in a CSO (Rodriguez et al. 2006). More recently, Deane et al. (2014) re-analyzed archival VLA data of the dual AGN system SDSS J150243.09+111557.3 and found that the point-source-subtracted VLA 5 GHz residual image contained extended "S-shaped" radio emission in the southern nucleus. They interpreted this as evidence for a binary SMBH system and later, through VLBI observations with a spatial resolution of 9 × 6 mas, confirmed the presence of two compact flat-spectrum radio sources, converting this object into a triple SMBH system. This interpretation, however, is being challenged by recent VLBA observations with three times better spatial resolution, which reveal two extended radio sources at the position of the southern nucleus (Wrobel et al. 2014b). This morphology is consistent with a double-hotspot scenario, in which both hotspots are energized by a single obscured SMBH.

J0009–0036, J0731+4528, and J1152+1903 present morphologies reminiscent of CSOs, and might be included in this class: the compact radio core resides between <1 kpc jets that have approximately an "S-symmetry" and a steep spectrum. Further observations with VLBI would be required to address the putative presence of binary SMBHs in these systems.

As can be seen in Table 8, the three dual AGNs identified in this work share several common characteristics: they are located at similar redshifts (z ∼ 0.13–0.2), they exhibit peak velocity splittings in the range 370–470 km s⁻¹ (∼420 km s⁻¹), and the projected spatial separation of the two AGNs is in the range 0.62–1.55 kpc (∼1 kpc). Furthermore, in all three cases the dual AGNs are located in the plane of the galaxy (${\mathrm{PA}}_{\mathrm{radio}}\;\sim$ PA_gal). Finally, while J1023+3243 and J1623+0808 appear to be single non-interacting galaxies in SDSS images, J1158+3231 has a companion at a projected separation of $\sim 10^{\prime\prime}$ (∼28 kpc). However, there is no evidence for tidal tails or arcs in any of the three galaxies. All of these characteristics suggest that the three systems are in the same evolutionary stage of a merger event.

Table 8.  Properties of the Dual AGNs Identified in This Work

Source	z	${\rm{\Delta }}v$ a	${L}_{[{\rm{O}}\;{\rm{III}}],\mathrm{blue}}$ b	${L}_{[{\rm{O}}\;{\rm{III}}],\mathrm{red}}$ c	${D}_{\mathrm{AGN}}$ d
		(km s−1)	($\times {10}^{40}$ erg s−1)	($\times {10}^{40}$ erg s−1)	(kpc)
SDSS J102325+324348	0.1271	371 ± 9	6.85 ± 0.86	8.94 ± 0.86	1.02 ± 0.16
SDSS J115822+323102	0.1658	441 ± 3	59.31 ± 2.22	9.91 ± 1.53	0.62 ± 0.11
SDSS J162345+080851	0.1992	469 ± 5	17.05 ± 1.90	42.56 ± 1.79	1.55 ± 0.17

Notes.

^aVelocity offset between the two peaks of [O iii] emission in the SDSS spectrum. ^bEstimated luminosity of the blueshifted component of the [O iii] line in the SDSS spectrum. ^cEstimated luminosity of the redshifted component of the [O iii] line in the SDSS spectrum. ^dPhysical projected separation between the two compact radio sources detected in the total intensity images. This is consistent in all cases with ${D}_{\mathrm{radio}}$ at 8.5 GHz in Table 4.

Download table as:  ASCIITypeset image

The observational evidence for the evolution of merging SMBHs is still rather sparse. Most reported double SMBHs are quasar pairs with separations 10–100 kpc, which are in the initial stage of their merging interaction (e.g., Lotz et al. 2008, and references therein). Direct imaging of active SMBH pairs in spiral galaxies undergoing a major merger is exemplified by the dual AGN separated by ∼6 kpc in Mrk 266 (Mazzarella et al. 2012), while the double radio nuclei with 7.3 pc separation in the elliptical galaxy 0402+37921 suggest the late evolution of a major merger (Rodriguez et al. 2006).

With a separation of ∼1 kpc between the two active SMBHs, J1023+3243, J1158+3231, and J1623+0808 appear to be in an intermediate phase between Mrk 266 and 0402+37921. The host galaxies of J1023+3243 and J1158+3231 are regular spirals, and J1623+0808 appears to be an elliptical galaxy with a disk inside it, suggesting in the three cases the later stages of a major merger or a minor merger. This is supported by recent hydrodynamic simulations of galaxy mergers (Blecha et al. 2013), which suggest that dual AGNs with separations 0.1–1 kpc exhibit peak velocity splittings of ∼500 km s⁻¹ and are located in the plane of the galaxy. All of these pieces of evidence indicate that the three dual AGNs identified in this work are in a short-lived stage (a few hundreds of megayears) when the nuclei are about to enter the final phase of coalescence.

Along with J1023+3243, J1158+3231, and J1623+0808, the following double-peaked AGNs have also been confirmed to contain dual AGNs based on the detection of two compact sources of hard X-rays, or two compact flat-spectrum radio sources²³ (see also Section 4.4): SDSS J110851+065901, SDSS J112659+294442, SDSS J114642+511029, and SDSS J150243+111557 (Fu et al. 2011b; Liu et al. 2013; Comerford et al. 2015). In all cases the SDSS images show disturbed morphologies and tidal features, suggesting ongoing mergers (the major axis of the galaxy is not easily recognizable, except in J112659+294442). In J114642+511029 it is still possible to distinguish in HST images the morphologies of the two interacting galaxies. While in J114642+511029 and J150243+111557 the projected spatial separation of the two nuclei is ∼7 kpc, in J110851+065901 and J112659+294442 the separation is ∼2.5 kpc. The velocity separation of the two peaks is ∼300 km s⁻¹ in J110851+065901, J112659+294442, and J114642+511029 (Liu et al. 2010a), and ∼720 km s⁻¹ in J150243+111557 (which is rather high for a separation of ∼7 kpc, Smith et al. 2010). All of these characteristics suggest that these galaxies belong to a similar evolutionary stage as Mrk 266, in which the progenitor galaxies are still recognizable (or still interacting before they finally merge), and the separation of the two nuclei is in the range 1–10 kpc. Note that J112659+294442 is a minor merger in progress (Comerford et al. 2015).

Double-peaked AGNs have gained popularity in the past few years because of the possibility of them being a signature for dual AGNs. Several candidates have been identified using optical images and spectroscopy. However, dual AGNs confirmations are scarce. This is mainly due to the fact that at optical/near-IR wavelengths the indicators of AGN-related activity are often ambiguous, or cannot be differentiated from the NLR of a single AGN. Indicators such as optical line ratios, optical colors, or the presence of highly ionized gas in most cases can be explained as part of the NLR of a single AGN. Photoionized gas by an AGN does not imply that there is an AGN at that specific location. Because of the proximity between the nuclei in all of these systems, it remains ambiguous whether there are one or two ionizing sources (i.e., AGNs) in each system (see, e.g., Fu et al. 2012 and references therein). For example, the two nuclei of the prototypical dual AGNs NGC 6240 (Komossa et al. 2003) are both heavily obscured in the optical, with emission line ratios characteristic of LINERs (Lutz et al. 1999); another example is Mrk 266 (Mazzarella et al. 2012), whose northern nucleus is optically classified as a composite yet does contain a strong X-ray AGN.

Dual AGNs can only be confirmed by spatially resolving two AGN sources in X-ray or radio observations. Most of the confirmed cases have been discovered serendipitously with Chandra (∼80% of all of the dual AGNs confirmations). Despite this considerable success, detections of closely separated AGNs on scales of hundreds of parsecs to a few kiloparsecs are severely constrained by the limited angular resolution of X-ray telescopes (Chandra has the highest resolution and can only resolve dual AGNs with separations >05). Furthermore, the majority of SDSS galaxies with double-peaked narrow emission lines have estimated X-ray fluxes that are too faint to be detected with reasonable integration times on Chandra (especially considering the evidence for a systematically smaller X-ray-to-[O iii] luminosity ratio in dual AGNs, Liu et al. 2013; Comerford et al. 2015). Finally, as noted by Liu et al. (2013), X-ray confirmations of dual AGN candidates can be challenging and ambiguous, due to the complex nature of X-ray obscuration in mergers.

The VLA in its A-configuration can resolve double radio cores with separations >02 and does not need a pre-selection based on optical AGN diagnostics to boost the success rate (e.g., high [O iii] luminosities). As an example, each of the three dual AGNs discovered in this work would need more than 200 ks with Chandra to detect >10 counts from each of the two AGNs in the system (Heckman et al. 2005, Table 8), and they would not be resolved, since the projected spatial separations of the two AGNs are less than 05 (Tables 4 and 8). However, in some cases, the VLA resolution is not enough to reveal the true radio nature of the sources, particularly in galaxies showing single radio cores or very small projected separations between compact sources (such as the case of J1715+6008). In these cases, the VLA images alone do not provide sufficient information to identify the origin of the double-peaked narrow emission lines. We used the long-slit spectra to distinguish between the different scenarios (Section 3.2). However, the crucial question that remains here is the following: Is there a radio structure within the VLA beam that could be responsible for the double-peaked, spatially offset narrow AGN emission lines? Follow-up observations with VLBI techniques would be ideal to answer this question.

The second limitation to our approach is linked to weak radio sources, particularly at higher frequencies. For example, in J0002+0045 the ambiguities arise because there appear to be two compact radio cores in the total intensity image, but the properties of one them could not be measured at 11.5 GHz. A similar situation occurs in J1112+2750 and J0916+2835 (both galaxies classified as single AGNs; Table 2 and Section 3.2.3).