Spatially Resolved Spectroscopy of Narrow-line Seyfert 1 Host Galaxies

In order to study the host galaxies underneath the bright point-like AGN emission, we decomposed the data cubes into AGN and host contributions using the software package QDeblend^3D (Husemann et al. 2013, 2014, 2016). A detailed description of the deblending algorithm can be found in the above articles. The basic method relies on the fact that the spatial distribution of the broad-line flux in the data cube provides a measure of the PSF at the central broad-line wavelength, given that the broad Balmer lines of unobscured AGNs originate from an unresolved region of $\lt 1\,\mathrm{pc}$.

Assuming that the central-most spectrum in the data cube is dominated by the AGN, the AGN contribution across the IFU field of view is removed by normalizing the AGN spectrum in absolute flux according to the PSF at a given spaxel. Depending on the AGN–host galaxy contrast, even the central spectrum can have a significant host galaxy contribution. Therefore, the host galaxy contribution is iteratively removed from the AGN spectrum while preserving the host galaxy surface–brightness distribution extrapolated from the circumnuclear spectra (after the first iteration) to the center for each wavelength slice.

The good spectral resolution of the WiFeS data allowed us to separate the broad lines from the narrow lines for a PSF measurement, even though the broad lines of NLS1s are by definition narrower than those of BLS1s. As the PSF changes with wavelength due to atmospheric dispersion, the cubes obtained from the WiFeS blue and red arms were treated separately during the deblending process, using the PSFs derived from the broad lines of Hβ and Hα, respectively.

The morphology and surface brightness profile for most of the targeted AGN host galaxies have not been studied in detail before. Therefore, we obtained a model of the host galaxy surface–brightness distributions by fitting the synthetic g and r broadband images extracted from the WiFeS data cubes. The fits were derived using GALFIT (Peng et al. 2002, 2010) and assume a Sérsic profile for the host galaxies (as well as for a companion galaxy, where applicable). For the convolution kernel, we used the PSFs estimated via QDeblend^3D from the Balmer line closest in wavelength. Bad spaxels and foreground stars were masked out during the fitting process. We explored two simple models with Sérsic indices of n = 1 and n = 4, respectively, and selected the model with the lower chi-square to derive the magnitudes of the AGN and the host galaxy as well as the effective radius. Figure 1 shows an example of a spectrum in the IFU data cube before and after AGN decomposition, which demonstrates that the AGN contribution is efficiently removed by the deblending procedure.

Zoom In Zoom Out Reset image size

The underlying host galaxy spectrum in the AGN-deblended data—consisting of a stellar continuum and possibly emission lines—was fitted using the spectral synthesis model code PyParadise, which is an extended Python version of paradise (Walcher et al. 2015). PyParadise takes a set of template continuum spectra to reconstruct the best-fitting continuum shape and a Gaussian profile for each emission line. The modeling process is briefly outlined in the following. Further details of the software can be found in Walcher et al. (2015).

In a first step, all library spectra and all spectra in the data cube were normalized by a running mean filter, where regions of strong emission-line or sky-line residuals were replaced by a linear interpolation over the masked wavelength ranges. Then the template library was smoothed to the instrumental resolution of the target spectra. A single spectrum of the library was chosen in order to estimate the systematic redshift and velocity dispersion by convolving the spectrum with a corresponding kernel as part of an MCMC algorithm. The entire library was then convolved with the maximum likelihood kinematics, and a nonlinear negative least-squares minimization was used to find the best linear combination of the library spectra to match the data. In three subsequent iterations, the best template combination was again used to refine the kinematics. After the best-fitting stellar continuum model was removed from the real data, the emission lines were fitted in the residual spectra. Here, all lines were modeled with simple Gaussian profiles and coupled in redshift and intrinsic rest-frame velocity dispersion while taking into account variations in spectral resolution.

A Monte Carlo bootstrap was used to estimate errors on the emission lines, including template mismatches of the stellar population spectra library. For each spectrum, the fitting process giving the best-fitting stellar kinematics was repeated 200 times after modulating the spectra according to the flux noise per spectral pixel and considering just 80% of the library spectra randomly chosen for each realization. Likewise, the emission-line fits were repeated in order to measure errors for the line redshift, the velocity dispersion, and the line flux based on the rms of the results. The entire process—except for the initial smoothing of the template library—was performed on a spaxel-by-spaxel basis, providing maps for the stellar kinematics, the line fluxes, and the gas kinematics and velocity dispersion across the field of view. In order to improve the signal-to-noise ratio, the data cubes for MS 22549-3712, WPVS 007, and TON S180 were binned to 3'' per pixel.

The AGN emission lines of the five sources were measured in the nuclear spectra obtained by integrating the host-deblended AGN data cubes over the full PSF. The emission lines were fitted simultaneously in two wavelength regions around Hβ and Hα using a superposition of Gaussian profiles together with Fe ii templates from Kovačević et al. (2010). The underlying AGN continuum was sampled in the line-free regions of the spectra and subtracted using a first-order polynomial fit for each of the two wavelength regions.

The emission-line fits for the continuum-subtracted AGN spectra are shown in Figure 2. The broad components of Hβ and Hα were modeled with a superposition of two Gaussian profiles (labeled "broad" (b) and "very broad" (vb) components, respectively). For consistency, the very broad Gaussian component was included for all sources in order to improve the fit to the broad-line wings, which were otherwise underrepresented. Our modeling approach for the broad Hβ and Hα components is in line with a larger number of previous studies that conclude that the broad Hβ and Hα components in many NLS1s are best approximated via two Gaussian components or a Lorentzian profile (e.g., Dietrich et al. 2005; Xu et al. 2007; Mullaney & Ward 2008; Cracco et al. 2016; Schmidt et al. 2016).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

During the fit, the corresponding components of Hβ and Hα were kinematically tied to each other in terms of central velocity and line width. The kinematics of He ii and He i was tied to the broad Gaussian component. The line profiles of all other lines—including the narrow components of Hβ and Hα—were based on a single narrow Gaussian profile. In order to improve the convergence of the fit and the decomposition of the complex Hβ and Hα profiles, all narrow components were forced to have the same relative velocity and line width. An additional blueshifted component of the [O iii] $\lambda \lambda 4959$,5007 lines was included in the models when the [O iii] lines showed obvious asymmetries. As a further constraint for all fits, we imposed constant flux ratios of 3.05 and 2.948 for the doublet lines [N ii] $\lambda \lambda 6581,6548$ and [O iii] $\lambda \lambda 5007,4959$ (main as well as blueshifted components), respectively (e.g., Schirmer et al. 2013).

The kinematics of the Fe ii templates was tied to the broad Gaussian component of the Hβ and Hα model. We used the narrowest Fe ii templates provided by Kovačević et al. (2010) and convolved the templates to the corresponding Gaussian width using a Gaussian kernel. The instrumental spectral resolution of the WiFeS observations was taken into account during the convolution. The intrinsic width of the available Fe ii templates of $\sim 500\ \mathrm{km}\,{{\rm{s}}}^{-1}$ (Gaussian sigma) imposed a lower limit on the possible Fe ii widths probed by the fitting procedure. For WPVS 007 and TON S180, the width of the broad Gaussian reference component was below this limit ($\sim 412\ \mathrm{km}\,{{\rm{s}}}^{-1}$ and $\sim 306\ \mathrm{km}\,{{\rm{s}}}^{-1}$, respectively), so that the Fe ii width was effectively decoupled from the reference component.

The line fluxes and kinematics derived from the fit to the host-deblended AGN spectra of the five sources are listed in Table 3. The narrow lines in all five sources are characterized by velocity dispersions between 100 and 240 $\mathrm{km}\,{{\rm{s}}}^{-1}$. The velocity dispersions of the broad and very broad Gaussian profiles used to model the broad Hα and Hβ components are in the range of 306–652 km s⁻¹and 1060–2100 $\mathrm{km}\,{{\rm{s}}}^{-1}$, respectively. The narrowest line profiles in the two broad-line components are shown by TON S180. Since the narrow-line contribution to the Balmer lines in TON S180 and MCG-05-01-013 is not well defined, these two sources were fitted without any narrow Hα, Hβ, or [N ii] components. Including a narrow-line contribution in the fit can result in broader broad-line components, as these would be more confined to the broad base of the line. Regarding the values discussed in this section, broader broad-line components would be reflected in a larger $\mathrm{FWHM}\,({\rm{H}}\beta )$, a larger BH mass, and smaller Eddington ratios.

Table 3.  Line Fluxes and Kinematics Determined from the Host-deblended AGN Spectra

Line	Wavelength	ESO 399-IG20	MCG-05-01-013	MS 22549-3712	TON S180	WPVS 007
	(Å)
		Flux (${10}^{-16}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$)
Fe ii	4434 − 4684	1060 ± 20	180 ± 10	410 ± 10	1670 ± 70	820 ± 10
He ii	4685.7	143 ± 9	13 ± 5	89 ± 7	120 ± 30	17 ± 6
H${\beta }_{{\rm{n}}}$	4861.3	49 ± 6	⋯	49 ± 7	⋯	20 ± 4
H${\beta }_{{\rm{b}}}$	4861.3	810 ± 20	190 ± 9	510 ± 20	530 ± 50	750 ± 10
H${\beta }_{\mathrm{vb}}$	4861.3	1090 ± 30	200 ± 10	160 ± 20	1340 ± 80	600 ± 20
[O iii]bl	4958.8	⋯	⋯	50 ± 20	⋯	50 ± 10
[O iii]	4958.8	178 ± 6	15 ± 3	70 ± 20	120 ± 30	30 ± 10
[O iii]bl	5006.8	⋯	⋯	150 ± 20	⋯	140 ± 10
[O iii]	5006.8	531 ± 6	46 ± 3	200 ± 20	360 ± 30	90 ± 10
[O i]	6300.2	27 ± 5	⋯	10 ± 4	30 ± 20	9 ± 3
[O i]	6363.7	30 ± 5	⋯	26 ± 4	30 ± 20	15 ± 3
[Fe x]	6374.5	37 ± 5	⋯	22 ± 4	⋯	2 ± 3
[N ii]	6548.0	32 ± 7	⋯	67 ± 10	⋯	31 ± 4
H${\alpha }_{{\rm{n}}}$	6562.8	82 ± 7	⋯	270 ± 20	⋯	301 ± 8
H${\alpha }_{{\rm{b}}}$	6562.8	3840 ± 60	1100 ± 30	1770 ± 50	1950 ± 70	2650 ± 20
H${\alpha }_{\mathrm{vb}}$	6562.8	2710 ± 50	740 ± 30	610 ± 30	3420 ± 70	1640 ± 30
[N ii]	6583.3	96 ± 7	⋯	200 ± 10	⋯	94 ± 4
He i	6678.2	120 ± 10	⋯	41 ± 7	40 ± 20	48 ± 6
[S ii]	6716.3	80 ± 5	⋯	33 ± 3	⋯	47 ± 3
[S ii]	6730.7	59 ± 5	⋯	20 ± 3	⋯	39 ± 3
[O iii]λ5007/Hβtotal		0.272 ± 0.006	0.118 ± 0.009	0.49 ± 0.04	0. 19 ± 0. 02	0.17 ± 0.01
${R}_{4570}\,=$ Fe ii${}_{4434-4684\mathring{\rm{A}} }/{\rm{H}}{\beta }_{{\rm{b}}+\mathrm{vb}}$		0.56 ± 0.01	0.46 ± 0.03	0.61 ± 0.03	0. 89 ± 0.06	0.61 ± 0.01
[S ii]λ6716/[S ii]λ6731		1.3 ± 0.1	⋯	1.7 ± 0.3	⋯	1.2 ± 0.1
		Systemic velocity ($\mathrm{km}\,{{\rm{s}}}^{-1}$)
Narrow componenta		7545 ± 2	9161 ± 9	11664 ± 4	18430 ± 20	8585 ± 1
		Relative velocity ($\mathrm{km}\,{{\rm{s}}}^{-1}$)
Broad componentb		+208 ± 4	+150 ± 10	+250 ± 7	+200 ± 20	+83 ± 2
Very broad componentc		+200 ± 20	+360 ± 50	+10 ± 70	+240 ± 30	−140 ± 20
Blueshifted [O iii]d		⋯	⋯	−510 ± 50	⋯	−240 ± 20
[S ii]e		⋯	⋯	+170 ± 10	⋯	−
		Velocity dispersion ($\mathrm{km}\,{{\rm{s}}}^{-1}$)
Narrow componenta		148 ± 2	130 ± 10	231 ± 6	220 ± 20	98 ± 2
Broad componentb		636 ± 6	652 ± 9	645 ± 8	306 ± 7	412 ± 3
Very broad componentc		1830 ± 30	2070 ± 70	2100 ± 100	1060 ± 20	1390 ± 20
Blueshifted [O iii]d		⋯	⋯	350 ± 40	⋯	220 ± 10
[S ii]e		⋯	⋯	99 ± 9	⋯	−

Notes.

^aThe narrow component is used as reference for the systemic velocity. The narrow component includes ${\rm{H}}{\beta }_{{\rm{n}}}$, [O iii], [N i], [O i], [Fe x], [N ii], H${\alpha }_{{\rm{n}}}$, and [S ii], except in the case of MS 22549-3712, where the [S ii] lines were fitted with independent kinematics. ^bThe broad components include ${\rm{H}}{\beta }_{{\rm{b}}}$, ${\rm{H}}{\alpha }_{{\rm{b}}}$, He ii, and He i. ^cThe very broad components include ${\rm{H}}{\beta }_{\mathrm{vb}}$ and ${\rm{H}}{\alpha }_{\mathrm{vb}}$. ^dThe blueshifted [O iii] component corresponds to [O iii]_bl. ^eThe [S ii] lines were fitted with independent kinematics in the case of MS 22549-3712, while in all other cases they shared the kinematics of the narrow component.

Download table as:  ASCIITypeset image

The [S ii]$\lambda 6716,6731$ lines in MS 22549-3712 were found to be redshifted by about 170 km s⁻¹ with respect to the best-fitting central velocity of the other narrow lines and were modeled with independent kinematics. We investigated whether the apparent [S ii] redshift could rather indicate a [O iii] blueshift and artificially result from our fitting method in which all narrow components (especially those of Hβ and Hα) are tied to [O iii]. However, we were unable to reproduce a similarly good fit by tying the narrow components to [S ii] instead of [O iii] while fitting [O iii] with independent kinematics. Since the narrow Hβ and Hα components as well as [N ii] cannot clearly be decomposed without further constraints from more isolated narrow lines, an immediate measure of the relative velocities between those lines, [O iii], and [S ii] is not available. Redshifted [S ii] emission could potentially be an indicator of shock activity in the unresolved NLR.

For the [O iii]$\lambda \lambda 4959$,5007 lines in MS 22549-3712 and WPVS 007, we included an additional blueshifted [O iii] component in order to take into account the blue [O iii] wings. Compared to the main narrow lines, the blue [O iii] components are broader by about 120 km s⁻¹ and are blueshifted by 510 km s⁻¹ and 240 km s⁻¹ in MS 22549-3712 and WPVS 007, respectively.

Table 4 shows the BH masses, Eddington ratios, and the [O iii]$\lambda 5007$ as well as the bolometric luminosities derived from the fits. The FWHM(Hβ) and the σ(Hβ) are based on the full model of the broad Hβ component, i.e., including both the broad and the very broad Gaussian components reported in Table 3. The line shapes of AGN broad lines have been found to correlate with line width in the sense that the FWHM/σ increases with an increasing FWHM of the line (Kollatschny & Zetzl 2011; Peterson 2011). For the five objects studied here, the FWHM/σ ratios from our broad Hβ models range between 1.1 and 1.4. These ratios are clearly different from a Gaussian profile with FWHM/σ ∼ 2.35, but they agree with the results for other narrow-line AGNs in the $\mathrm{FWHM}\,({\rm{H}}\beta )\lesssim 2000\ \mathrm{km}\,{{\rm{s}}}^{-1}$ regime (e.g., Figure 7 in Kollatschny & Zetzl 2013). The origin of the broad-line profiles and the best line-width indicator for virial BH mass measurements have been a long-standing controversy that is beyond the scope of this paper. As in Rakshit et al. (2017), we computed the BH masses based on FWHM(Hβ) together with the relation

$$\begin{eqnarray}&&\mathrm{log}{M}_{\mathrm{BH}}=\mathrm{log}\\ &&\quad \times \,\left[{\left(\tfrac{\mathrm{FWHM}({\rm{H}}\beta )}{1000\mathrm{km}{{\rm{s}}}^{-1}}\right)}^{2}{\left(\tfrac{\lambda {L}_{\lambda }(5100\mathring{\rm{A}} )}{{10}^{44}\mathrm{erg}{{\rm{s}}}^{-1}}\right)}^{0.533}\right]+6.69,\end{eqnarray} \tag{ 1 }$$

where M_BH is the BH mass in ${M}_{\odot },\mathrm{FWHM}\,({\rm{H}}\beta )$ is the FWHM of Hβ in $\mathrm{km}\,{{\rm{s}}}^{-1}$, and $\lambda {L}_{\lambda }(5100\ \mathring{\rm{A}} )$ is the 5100 Å luminosity in $\mathrm{erg}\,{{\rm{s}}}^{-1}$ as derived from the continuum fit to the AGN spectra (see Section 2.3). This relation is based on ${M}_{\mathrm{BH}}={{fR}}_{\mathrm{BLR}}{\rm{\Delta }}{v}^{2}/G$, using $f=3/4$, $\mathrm{FWHM}\,({\rm{H}}\beta )$ for the velocity width ${\rm{\Delta }}v$, the gravitational constant G, and the best-fit radius-luminosity relation from Bentz et al. (2013) in order to calculate the broad-line region (BLR) radius R_BLR from the 5100 Å luminosity. The bolometric luminosities and Eddington ratios are estimated using ${L}_{\mathrm{bol}}=9\lambda {L}_{\lambda }(5100\ \mathring{\rm{A}} )$ (e.g., Kaspi et al. 2000) and ${\lambda }_{\mathrm{Edd}}\,={L}_{\mathrm{bol}}/{L}_{\mathrm{Edd}}$, respectively, where ${L}_{\mathrm{Edd}}=1.25\,\times \,{10}^{38}{M}_{\mathrm{BH}}\ [\mathrm{erg}\,{{\rm{s}}}^{-1}]$.

Table 4.  Parameters Derived from the Host-deblended AGN Spectra

Object	FWHM(Hβ)	$\sigma ({\rm{H}}\beta )$	$\mathrm{log}({M}_{\mathrm{BH}})$	log(L([O iii])	$\mathrm{log}({L}_{\mathrm{bol}})$	${\lambda }_{\mathrm{Edd}}$
	$(\mathrm{km}\,{{\rm{s}}}^{-1})$	$(\mathrm{km}\,{{\rm{s}}}^{-1})$	$({M}_{\odot })$	$(\mathrm{erg}\,{{\rm{s}}}^{-1})$	$(\mathrm{erg}\,{{\rm{s}}}^{-1})$
(1)	(2)	(3)	(4)	(5)	(6)	(7)
ESO 399-IG20	1900 ± 20	1450 ± 20	$6.77\pm 0.02$	40.854 ± 0.005	44.05 ± 0.04	$0.15\pm 0.02$
MCG-05-01-013	1850 ± 40	1560 ± 60	$6.53\pm 0.04$	39.97 ± 0.02	43.65 ± 0.08	$0.11\pm 0.02$
MS 22549-3712	1610 ± 20	1190 ± 60	$6.67\pm 0.02$	41.08 ± 0.04	44.14 ± 0.05	$0.24\pm 0.03$
TON S180	1060 ± 50	920 ± 20	$6.88\pm 0.05$	41.51 ± 0.04	45.21 ± 0.04	$1.7\pm 0.2$
WPVS 007	1120 ± 10	980 ± 10	$6.43\pm 0.01$	40.62 ± 0.03	44.29 ± 0.02	$0.58\pm 0.03$

Note. Column 1: Object name. Columns 2 and 3: FWHM and second moment of the broad Hβ line taking into account the superposition of the broad and very broad Gaussian components listed in Table 3. Column 4: Logarithm of the derived BH mass. Column 5: Logarithm of the total [O iii]$\lambda 5007$ luminosity (including the blueshifted component, where applicable). Column 6: Logarithm of the bolometric luminosity, based on the 5100 Å flux. Column 7: Eddington ratio. A detailed description of how the quantities are derived is given in the text. The errors listed in the table are the formal errors propagated from the line-fitting process. The actual uncertainties in the flux-dependent quantities are likely to be larger, since we conservatively estimate a factor of 2 of uncertainty in the flux calibration (see Section 2). This corresponds to uncertainties of $\mathrm{log}(\sqrt{2})\approx 0.2\,\mathrm{dex}$ for $\mathrm{log}({M}_{\mathrm{BH}})$, $\mathrm{log}(2)\approx 0.3\,\mathrm{dex}$ for log(L([O iii])) and $\mathrm{log}({L}_{\mathrm{bol}})$, and a factor of $\sqrt{2}\approx 1.4$ for ${\lambda }_{\mathrm{Edd}}$.

Download table as:  ASCIITypeset image

The parameters of the AGN emission lines of all five objects agree with the classical criteria used to define NLS1s from optical spectra. All objects have FWHM(Hβ) $\lt 2000\ \mathrm{km}\,{{\rm{s}}}^{-1}$—based on the superposition of the two Gaussians used to model the broad Hβ and Hα profiles—and [O iii]/H${\beta }_{\mathrm{total}}\lt 3$ (see Table 3).

Prominent Fe ii emission is typically considered an additional characteristic of NLS1s, and all five objects show evidence of Fe ii emission in their nuclear spectra. For comparison with the large samples of NLS1s from the SDSS analyzed by Zhou et al. (2006) and Rakshit et al. (2017), we define the Fe ii strength likewise as the ratio between the Fe ii flux, integrated over the wavelength range 4434–4684 Å, and the flux of the broad Hβ component. As our decomposition of the nuclear spectra includes a broad and a very broad Gaussian profile, we use the sum of these to compute ${R}_{4570}=\mathrm{Fe}\,{{\rm{II}}}_{4434-4684\mathring{\rm{A}} }/{\rm{H}}{\beta }_{{\rm{b}}+\mathrm{vb}}$ as shown in Table 3. Compared to Zhou et al. (2006) and Rakshit et al. (2017), the resulting Fe ii (4434–4684 Å) strength in the five objects agrees with the typical values found for NLS1s. For their NLS1 sample, Zhou et al. (2006) find an average Fe ii strength of ${R}_{4570}=0.83$, which they report to be about twice as high as the ${R}_{4570}\sim 0.4$ of a normal AGN. By comparing the NLS1 sample with the BLS1s from their SDSS parent sample, Rakshit et al. (2017) find a mean Fe ii strength of 0.64 for the NLS1s compared to a mean Fe ii strength of 0.38 for the BLS1s. As can be seen in Table 3, the five NLS1s studied here show a range of R₄₅₇₀ values between ${R}_{4570}=0.46$ for MCG-05-01-013 and ${R}_{4570}=0.89$ for TON S180.

In addition to prominent Fe ii emission, NLS1s have been suggested to show, on average, smaller BH masses and higher Eddington ratios than their broad-line counterparts (e.g., Komossa 2008; Rakshit et al. 2017, and references therein). The five objects generally probe the NLS1 parameter space in terms of Eddington ratios and BH masses. As listed in Table 4, the BH masses of the objects in our study range from ${M}_{\mathrm{BH}}={10}^{6.43}\ {M}_{\odot }$ for WPVS 007 to ${M}_{\mathrm{BH}}={10}^{6.88}\ {M}_{\odot }$ for TON S180. The NLS1 and BLS1 samples drawn from the SDSS Data Release 12 by Rakshit et al. (2017) result in mean BH masses of ${M}_{\mathrm{BH}}={10}^{6.9}\ {M}_{\odot }$ and ${10}^{8}\ {M}_{\odot }$ for NLS1s and BLS1s, respectively. The corresponding dispersions of the $\mathrm{log}({M}_{\mathrm{BH}})$ distributions are $0.41\ {M}_{\odot }$ for the NLS1s and $0.46\ {M}_{\odot }$ for the BLS1s. Compared to these samples, the BH masses of the five objects studied herein agree with the typical BH mass range displayed by NLS1s and are lower than the BH masses typically found for BLS1s.

The highest Eddington ratios among the five sources are displayed by TON S180 (${\lambda }_{\mathrm{Edd}}=1.7$ or $\mathrm{log}{\lambda }_{\mathrm{Edd}}=0.2$) and WPVS 007 (${\lambda }_{\mathrm{Edd}}=0.58$ or $\mathrm{log}{\lambda }_{\mathrm{Edd}}=-0.2$), while ESO 399-IG20, MCG-05-01-013, and MS 22549-3712 show Eddington ratios that are below 0.24 ($\mathrm{log}{\lambda }_{\mathrm{Edd}}\lt -0.62$). Compared to the NLS1s studied by Rakshit et al. (2017), the Eddington ratios of TON S180 and WPVS 007 overlap with the upper and average portion of the NLS1 Eddington ratio distribution, respectively. The Eddington ratios of the other three sources agree with the lower part of the distribution.

An estimate of the electron density in the NLR based on the density-sensitive [S ii] is available for three sources for which the [S ii] lines could be fitted in the nuclear spectra. Assuming an electron temperature of T_e = 10,000 K, ESO 399-IG20 and WPVS 007 show NLR densities of about ${n}_{e}\approx 120\ {\mathrm{cm}}^{-3}$ and ${n}_{e}\approx 240\ {\mathrm{cm}}^{-3}$, respectively. The [S ii]λ6716/[S ii]λ6731 ratio measured for MS 22549-3712 falls into the low-density limit of $\sim 10\ {\mathrm{cm}}^{-3}$. According to Xu et al. (2007), NLS1s can have very low densities in their NLRs in contrast to BLS1s, which typically show ${n}_{e}\gtrsim 140\ {\mathrm{cm}}^{-3}$. The samples of NLS1s and BLS1s studied by Rakshit et al. (2017) do not confirm such a lower-density limit for BLS1s but suggest a weak trend of lower-density NLRs in NLS1s compared to BLS1s. The electron density for WPVS 007 overlaps with the bulk of NLS1 and BLS1 electron densities from that study, while the NLR densities for ESO 399-IG20 and MS 22549-3712 herein agree with the lower-density wing of the distribution presented by Rakshit et al. (2017).

The emission-line maps derived from the AGN-deblended data cubes for the host galaxies of the five objects are shown in Figure 3. All the resulting gas velocity fields show evidence of large-scale gas rotation. Except for MCG-05-01-013, which is obviously a disk galaxy with a bar and/or spiral arm, the spatial resolution is too low in order to unambiguously determine the galaxy type. However, MS 22549-3712 and TON S180 are known to be disk galaxies from high-resolution Hubble Space Telescope images, which suggest a pseudo-bulge in MS 22549-3712 and a classical bulge in TON S180 (Mathur et al. 2012). WPVS 007 was found to be a largely bulge-dominated galaxy with possible evidence of a substructure and a low Sérsic index (Busch et al. 2014). ESO 399-IG20 was classified as an S0 galaxy by Dietrich et al. (2005) based on spectral fitting. The gas velocity field of ESO 399-IG20 is more perturbed, possibly as a consequence of a tidal interaction with the companion to the southwest. The companion, which is evident from the DSS image, is associated with the Hα peak in the southwestern-most corner of the WiFeS Hα map. The g−r colors of the five host galaxies, presented in Section 3.3, are generally consistent with these galaxy-type classifications (cf. e.g., Fernández Lorenzo et al. 2012). Orban de Xivry et al. (2011) proposed that the nuclear activity in NLS1s might be driven by secular evolution. Based on the gas kinematics and suggested galaxy types, it is possible that secular evolution plays a role in at least some of the five objects studied here. The data for MS 22549-3712, TON S180, and WPVS 007 have been binned spatially by 3 × 3 pixels ($3^{\prime\prime} \times 3^{\prime\prime}$) in order to increase the signal-to-noise ratio on the emission lines. Despite the lower spatial resolution in these data, the gas kinematics is likewise characterized by velocity gradients centered on the galaxy nuclei, presumably indicating rotation.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

According to the line ratio diagnostics presented in the bottom rows of Figure 3, none of the five objects shows clear evidence of an extended NLR at the 2–3 kpc scales probed by our observations. The line ratios for the five sources largely occupy the region dominated by ionization from Hii regions, i.e., the region close to or below the empirical maximum starburst line (the solid black line) from Kauffmann et al. (2003). The binned data for MS 22549-3712 and WPVS 007 show marginal evidence of off-nuclear spaxels extending into the region around and above the theoretical maximum starburst line from Kewley et al. (2001), which could indicate some AGN or shock contribution to the gas ionization (e.g., Rich et al. 2011; Scharwächter et al. 2011; Davies et al. 2014; Dopita et al. 2014). A possible shock activity in MS 22549-3712 has also been pointed out based on the redshifted [S ii] lines in the nuclear spectrum of this object (Section 3.1). However, this interpretation remains speculative given the low signal-to-noise ratio in the host galaxy data for these objects.

Estimates for the global SFRs, stellar masses, and specific SFRs are listed in Table 5. Since the extended ionized gas in all five objects is predominantly excited by star formation, the global SFRs are derived from the integrated Hα luminosities, ${L}_{{\rm{H}}\alpha }$, using the relation from Kennicutt (1998):

$$\begin{eqnarray}&&\mathrm{SFR}\ ({M}_{\odot }\,{\mathrm{yr}}^{-1})=\displaystyle \frac{{L}_{{\rm{H}}\alpha }}{1.26\times {10}^{41}\ \mathrm{erg}\,{{\rm{s}}}^{-1}}.\end{eqnarray} \tag{ 2 }$$

The Hα luminosities are corrected for extinction based on the Balmer decrement using an attenuation correction of the form

$$\begin{eqnarray}&&\displaystyle \frac{{\tau }_{\lambda }}{{\tau }_{V}}=(1-\mu ){\left(\displaystyle \frac{\lambda }{5500\mathring{\rm{A}} }\right)}^{-1.3}+\mu {\left(\displaystyle \frac{\lambda }{5500\mathring{\rm{A}} }\right)}^{-0.7}\end{eqnarray} \tag{ 3 }$$

(Wild et al. 2007, 2011), with $\mu =0.4$, where λ is the wavelength in Å. Expressed in terms of the extinction A_V, this corresponds to a flux correction of

$$\begin{eqnarray}&&{f}_{{\rm{H}}\alpha ,\mathrm{corr}}={f}_{{\rm{H}}\alpha }{10}^{0.4{A}_{V}{\tau }_{{\rm{H}}\alpha }/{\tau }_{V}},\end{eqnarray} \tag{ 4 }$$

where ${f}_{{\rm{H}}\alpha }$ is the flux at the wavelength of Hα. For the spatially well-resolved sources, ESO 399-IG20 and MCG-05-01-013, the extinction correction is applied on a pixel-by-pixel basis. The corresponding extinction maps of these sources are shown in Figure 4. As in Figure 3, only data with a flux signal-to-noise ratio of $\gt 3$, a velocity error of $\lt 30\ \mathrm{km}\,{{\rm{s}}}^{-1}$, and a velocity dispersion error of $\lt 80\ \mathrm{km}\,{{\rm{s}}}^{-1}$ are used. Furthermore, a few unreliable A_V values, which are $\lt 0$ within the uncertainties, are masked out. Since the calculation of A_V involves the fainter Hβ line, the number of pixels with reliable A_V data is smaller than the number of pixels with reliable Hα fluxes. In order to obtain the integrated Hα luminosity, only the Hα fluxes in spatial pixels with reliable A_V measurements are corrected for extinction and integrated. Within the uncertainties, the same Hα luminosities are obtained when integrating over all available spatial pixels with reliable Hα fluxes while adopting an assumed ${A}_{V}\sim 0-1\ \mathrm{mag}$ for the fraction of pixels lacking reliable A_V measurements. Since the signal-to-noise ratios for MS 22549-3712, TON S180, and WPVS 007 are too low to provide reliable A_V measurements, the Hα fluxes for these sources are corrected assuming an average A_V = 1 mag. This value corresponds to the average A_V found for the two well-resolved sources ESO 399-IG20 and MCG-05-01-013. The SFRs for TON S180 are based on the assumption that the Hα emission is dominated by star formation, although the nature of the dominant ionization mechanism cannot be verified using line ratio diagrams (see Figure 3(d)). Thus, the star formation for this source may be considered as an upper limit.

Zoom In Zoom Out Reset image size

Table 5.  SFR and Host Galaxy Stellar Mass

Object	$\mathrm{log}{L}_{{\rm{H}}\alpha }$	SFR	g − r	mV	$\mathrm{log}{M}_{\star }$	$\mathrm{logsSFR}$
	$(\mathrm{erg}\,{{\rm{s}}}^{-1})$	$({M}_{\odot }\,{\mathrm{yr}}^{-1})$	(mag)	(mag)	$({M}_{\odot })$	$({\mathrm{yr}}^{-1})$
(1)	(2)	(3)	(4)	(5)	(6)	(7)
ESO 399-IG20	40.84 ± 0.04a	0.55 ± 0.05	0.59	14.1 ± 0.8	10.8 ± 0.3	−11.0 ± 0.3
MCG-05-01-013	41.55 ± 0.01a	2.82 ± 0.08	0.44	14.6 ± 0.8	10.6 ± 0.3	−10.2 ± 0.3
MS 22549-3712	40.8 ± 0.4b	0.5 ± 0.4	0.75	16.2 ± 0.8	10.5 ± 0.3	−10.8 ± 0.4
TON S180	41.0 ± 0.4b	0.8 ± 0.6c	0.40	15.7 ± 0.8	10.7 ± 0.3	−10.8 ± 0.4c
WPVS 007	40.6 ± 0.4b	0.3 ± 0.2	0.40	16.5 ± 0.8	9.6 ± 0.3	−10.1 ± 0.4

Notes. Column 1: Object name. Column 2: Extinction-corrected Hα luminosity. Column 3: Hα-based SFR. Column 4: g−r color of the host galaxy. Column 5: V-band magnitude of the host galaxy. Column 6: Stellar mass of the host galaxy. Column 7: Logarithm of the specific SFR. For $\mathrm{log}{L}_{{\rm{H}}\alpha }$ and the SFR, we report the formal errors propagated from the line fitting. For the stellar mass, we assume an error of 0.3 dex, corresponding to the estimated factor of 2 of uncertainty in the flux calibration (see Section 2). Note that the same uncertainty due to the flux calibration may also apply to the Hα luminosity and the SFR.

^aThe integrated Hα luminosity based on a pixel-by-pixel extinction correction, using all pixels with reliable A_V as shown in Figure 4. ^bThe integrated Hα luminosity based on a global extinction correction assuming an average ${A}_{V}=1\pm 1$. ^cThis value is an upper limit (assuming that the Hα luminosity in the host galaxy of TON S180 is indicative of star formation).

Download table as:  ASCIITypeset image

In addition to the SFRs, Table 5 shows estimates of the stellar masses of the five objects. These were derived from the integrated r band magnitude of the best-fitting surface brightness model described in Section 2.1. We used a mass-to-light ratio for the V-band (${{\rm{\Gamma }}}_{V}$) based on the g − r color, following the prescription of Wilkins et al. (2013),

$$\begin{eqnarray}&&\mathrm{log}{{\rm{\Gamma }}}_{V}=1.4\times (g-r)-0.49.\end{eqnarray} \tag{ 5 }$$

Given the low redshift of the objects, any k correction would be small and was ignored for this work. We also ignored a correction for dust attenuation since the amount of dust attenuation for the stellar continuum cannot be estimated from the data themselves.

In Figure 5, the specific SFRs and stellar masses of the five objects are compared to SDSS galaxies from Brinchmann et al. (2004) and to the flux-limited sample of $z\lt 0.2$ type-1 QSOs discussed by Husemann et al. (2014). Since the comparison of the different sources is subject to a mild redshift evolution, we used the linear fits to the specific SFR-stellar mass relation for the three lowest redshift bins from Lara-López et al. (2013) for normalization. The top panel in Figure 5 shows the original data points, which are colored based to their redshift bin, together with the main-sequence relations from Lara-López et al. (2013, Equation (9) and Table 2). The bottom panel shows the normalized data points. All five low-redshift objects discussed here were included in the lowest redshift bin.

Zoom In Zoom Out Reset image size

The five objects show a variety of star formation properties, extending from the main sequence of star-forming galaxies down to lower specific SFRs intermediate to the main sequence and the location of red quiescent galaxies. The general range of star formation properties covered by the NLS1s studied here is very similar to the range covered by the type-1 QSO sample. However, none of the five NLS1s is located above the main sequence. The highest SFR is found for MCG-05-01-013 ($\mathrm{SFR}=2.82\ {M}_{\odot }\,{\mathrm{yr}}^{-1}$), which also shows one of the highest specific SFRs in the sample. MCG-05-01-013 is likely to be in a recent phase of active star formation and is located on the main sequence of SDSS star-forming galaxies in Figure 5. ESO 399-IG20, MS 22549-3712, TON S180, and WPVS 007 have lower Hα SFRs of 0.3–0.8 M_⊙ yr⁻¹. Despite a moderate SFR, WPVS 007 displays the highest specific SFR of the five objects. The stellar mass of WPVS 007 is almost one to two orders of magnitude lower than the stellar masses found for the other four objects or the type-1 QSO sample. Nevertheless, in terms of a specific SFR, WPVS 007 is still consistent with the bulk of low-mass SDSS star-forming galaxies.

The oxygen abundances and metallicity gradients for the host galaxies of the two objects with unbinned data—ESO 399-IG20 and MCG-05-01-013—are shown in Figures 6 and 7. The QSO-host deblending technique is most reliable in the wavelength region close to the lines used to probe the PSF (i.e., Hα and Hβ). We use the [O iii]$\lambda 5007,{\rm{H}}\beta$, [N ii]$\lambda 6583$, and ${\rm{H}}\alpha$ lines in order to derive oxygen abundances based on the O3N2 method (Pettini & Pagel 2004):

$$\begin{eqnarray}&&12+\mathrm{log}{({\rm{O}}/{\rm{H}})}_{{\rm{O}}3{\rm{N}}2}=8.73-0.32\mathrm{log}\left(\displaystyle \frac{[{\rm{O}}\,{\rm{III}}]\lambda 5007/{\rm{H}}\beta }{[{\rm{N}}\,{\rm{II}}]\lambda 6583/{\rm{H}}\alpha }\right).\end{eqnarray} \tag{ 6 }$$

As there is evidence that the emission-line regions in the two host galaxies are largely dominated by star formation (see Section 3.2), the oxygen abundance is calculated for all unmasked spatial pixels. As in Husemann et al. (2014), the oxygen abundance is converted to the T04 (Tremonti et al. 2004) calibration using

$$\begin{eqnarray}12+\mathrm{log}{({\rm{O}}/{\rm{H}})}_{{\rm{T}}04} & = & -738.1193+258.96730x\\ & & -30.057050{x}^{2}+1.1679370{x}^{3},\end{eqnarray} \tag{ 7 }$$

with $x=12+\mathrm{log}{({\rm{O}}/{\rm{H}})}_{{\rm{O}}3{\rm{N}}2}$ (Kewley & Ellison 2008).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

As can be seen in Figures 6 and 7, both of the two well-resolved sources show indications of radial metallicity gradients. In the outer parts of both host galaxy disks, the oxygen abundance decreases with an increasing galactocentric distance. For reasons of comparison, the radius in Figure 7 is shown in units of effective radius, for which we use the effective radii determined from the surface brightness fits described in Section 2.1.

MCG-05-01-013 has a smooth metallicity gradient with a decreasing abundance toward the outer part of the host galaxy at a large galactocentric distance ($\gtrsim 0.7{R}_{{\rm{e}}}$). Closer to the center, at $R\lesssim 0.7{R}_{{\rm{e}}}$, the gradient is flat or even inverted. When using only data at $R\gtrsim 0.7{R}_{{\rm{e}}}$, the abundance gradient of MCG-05-01-013 is $-0.094\ \mathrm{dex}/{R}_{{\rm{e}}}$, based on a linear regression. The limited data for ESO 399-IG20 only probe abundances between about 0.45 and 1.2 R_e, but they show similar indications of a decreasing abundance toward the outer parts of the host galaxy. The corresponding metallicity gradient is found to be $-0.1\ \mathrm{dex}/{R}_{{\rm{e}}}$. Both gradients are very similar to the characteristic gradient of $-0.1\ \mathrm{dex}/{R}_{{\rm{e}}}$ displayed by noninteracting galaxy disks in the local universe (Sánchez et al. 2014). A flattening of the oxygen abundance gradient in the inner regions has been found in galaxy disks with circumnuclear star-forming rings (Sánchez et al. 2014). Such a star-forming ring may be associated with the flattening of the metallicity gradient in the central part of the host galaxy of MCG-05-01-013.

The oxygen abundances for the two well-resolved sources, measured at R_e and extrapolated to the center, are listed in Table 6. The oxygen abundance at R_e is derived from the mean and the standard deviation of all available abundance data between $0.75{R}_{{\rm{e}}}$ and $1.25{R}_{{\rm{e}}}$. This is the same method as used for the $z\lt 0.2$ type-1 QSOs by Husemann et al. (2014), which we adopt for reasons of comparison. The central oxygen abundance shown in Table 6 is extrapolated from the value at R_e using a metallicity gradient of $-0.1\ \mathrm{dex}/{R}_{{\rm{e}}}$. For MCG-05-01-013, we also show the central oxygen abundance resulting from the linear regression for $R\lt 0.7{R}_{{\rm{e}}}$, i.e., taking into account the central flattening of the metallicity gradient.

Table 6.  Oxygen Abundance

Object	Re	$12+\mathrm{log}{({\rm{O}}/{\rm{H}})}_{{\rm{T}}04}$	$12+\mathrm{log}{({\rm{O}}/{\rm{H}})}_{{\rm{T}}04}$
	(arcsec)	at Re	at R0
(1)	(2)	(3)	(4)
ESO 399-IG20	10.1	8.87 ± 0.05	8.97a ± 0.05
MCG-05-01-013	5.6	9.02 ± 0.06	9.12a ± 0.06/8.98b

Notes. Column 1: Object name. Column 2: Effective radius R_e. Column 3: Average oxygen abundance at R_e. Column 4: Extrapolated central oxygen abundance.

^aExtrapolated from R_e assuming −0.1 dex/R_e. ^bExtrapolated via linear regression using the flat gradient found for $R\lt 0.7{R}_{{\rm{e}}}$.

Download table as:  ASCIITypeset image

In Figure 8 (top panel), the two sources with measurements for the central oxygen abundance are plotted in the mass–metallicity plane and are compared to SDSS galaxies and the type-1 QSOs from Husemann et al. (2014). In addition, the plot shows fits to the mass–metallicity relation for the three lowest redshift bins presented by Lara-López et al. (2013). The data points are colored based on their redshift bin. In the bottom panel, the same data points are compared after normalization based on the fitted mass–metallicity relations from Lara-López et al. (2013). The central oxygen abundances for ESO 399-IG20 and MCG-05-01-013 are about 0.1 dex below the low-redshift mass–metallicity relation from Lara-López et al. (2013). (For MCG-05-01-013, this statement is true when considering the flat/inverted central metallicity gradient, resulting in the central abundance marked by the open diamond.) In their type-1 QSO sample, Husemann et al. (2014) found that the trend of bulge-dominated hosts was to preferentially lie at the lower end or below the mass–metallicity relation, while disk-dominated hosts mostly followed the main relation. In comparison to the type-1 QSO sample, the two objects analyzed here seem to occupy an intermediate region between the bulge- and disk-dominated type-1 QSO host galaxies.

Zoom In Zoom Out Reset image size

3.5.1. ESO 399-IG20

3.5.2. MCG-05-01-013

3.5.3. MS 22549-3712

3.5.4. TON S180

3.5.5. WPVS 007