An AGN with an Ionized Gas Outflow in a Massive Quiescent Galaxy in a Protocluster at z = 3.09

Table 1 summarizes the data and their reference of the target. Our target is one of the galaxies confirmed by our NIR spectroscopic observations of massive galaxies in a protocluster at z = 3.09 in the SSA22 field (Steidel et al. 1998) using Multi-Object Infra-Red Camera and Spectrograph (MOIRCS; Ichikawa et al. 2006; Suzuki et al. 2008) on the Subaru telescope (Kubo et al. 2015 hereafter K15; referred J221737.3+001823.2). Hereafter we call the target as J221737.29+001823.4 adopting the source position measured on the new K_s -band image taken with updated MOIRCS (nuMOIRCS; Fabricius et al. 2016; Walawender et al. 2016) (Kubo et al. 2021, hereafter K21) using SExtractor (Bertin & Arnouts 1996). The source is an AGN matching with the X-ray source at a 018 angular separation in the catalog obtained with Chandra in Lehmer et al. (2009a,2009b). Its redshift is confirmed by detecting the [O iii]λ λ 4959,5007 emission, which shows two peaks corresponding to z = 3.0851 ± 0.0001 and 3.0926 ± 0.0003 indicating the complex kinematics of the ionized gas.

Table 1. Summary of the Data

Items	References
Photometry
CFHT MegaCam u⋆	not published (PI: Cowie)
Subaru S-Cam ${BVRi}^{\prime} z^{\prime}$	Matsuda et al. (2004)
	Hayashino et al. (2004)
Subaru MOIRCS JHKs	Kubo et al. (2013)
Subaru nuMOIRCS Ks	K21
HST ACS F814W	in archive (PID9760)
Spitzer IRAC & MIPS 24μm	Webb et al. (2009)
ALMA 1.2 mm	Umehata et al. (2018)
ALMA 3.0 mm	Umehata et al. (2019)
VLA 6 GHz	H. Umehata et al. 2022, in preparation
VLA 3 GHz	Ao et al. (2017)
VLA 1.4 GHz	Chapman et al. (2004)
Chandra 0.5–8 keV	Lehmer et al. (2009b)
Spectroscopy
Subaru MOIRCS VPH-K	K15
Keck MOSFIRE H and K	Umehata et al. (2019)
VLT MUSE	Umehata et al. (2019)

Download table as:  ASCIITypeset image

As we noted in K15 and revisit in Section 3.1, the SED of J221737.29+001823.4 is well-fitted by a quiescent galaxy. Figure 1 shows the rest-frame UVJ color diagram to select quiescent galaxies (e.g., Whitaker et al. 2013). The UVJ colors are interpolated from the best-fit SED described in Section 3.1. J221737.29+001823.4 is classified as a quiescent galaxy similar to another massive quiescent galaxy with ${z}_{\mathrm{spec}}={3.0922}_{-0.004}^{+0.008}$ at R.A., decl. = 22:17:37.25, +00:18:16.0, only 75 ( ≈ 60 in physical kpc) away from the target confirmed in K21. We note that J221737.25+001816.0 is not detected in [O iii] emission and the Chandra and radio data described below but shows weak [O ii] emission that can originate in an AGN. Including our studies, several studies have shown the prevalence of massive quiescent galaxies in protoclusters at up to z = 3.37 (e.g., Kubo et al. 2013; K21; Shi et al. 2021; McConachie et al. 2021). Such massive quiescent galaxies in protoclusters are the most plausible progenitors of typical giant ellipticals today.

Zoom In Zoom Out Reset image size

In the SSA22 protocluster, AGNs have been surveyed using Chandra and Spitzer data (Webb et al. 2009), and those hosted by galaxies ranging from Lyα emitters to submillimeter galaxies have been studied (Geach et al. 2009; Lehmer et al. 2009a, 2009b; Webb et al. 2009; Tamura et al. 2010; Kubo et al. 2013; K15; Umehata et al. 2015, 2019; Monson et al. 2021). Among them, J221737.29+001823.4 is not particularly luminous in the X-rays but is the most [O iii] luminous source observed at this point (K15).

Table 2 summarizes the measured magnitudes and fluxes of the target. The u^⋆ to 8.0 μm photometry was taken in the same way as K21. The u^⋆ to K_s , and F814W-band images are convolved to match the PSF to a FWHM of ≈ 1.0 arcsec and measured fluxes with a 20 diameter aperture. In the IRAC 3.6–8.0 μm photometry we apply aperture correction computed by using the K_s -band image to match with u^⋆ to the K_s band (see detail in Kubo et al. 2013). Then we corrected the PSF matched photometries multiplying the total (Kron flux measured on the nuMOIRCS image) to the aperture photometry ratio in the K_s band. Major emission lines shifting to the bandpasses were checked spectroscopically and subtracted from these broadband measurements. [O ii], Hβ, and [O iii] were subtracted from the H- and K-band fluxes, respectively, while Hα and [N ii] are shifted out of the bandpasses. Lyα and [C iv] were ignorable as they were observed with the Multi-Unit Spectroscopic Explorer (MUSE) on the Very Large Telescope (VLT; Umehata et al. 2019; 2 σ = 0.3 × 10⁻¹⁸ erg s⁻¹ cm⁻² arcsec⁻²) but are not detected. Type-2 AGNs show strong Lyα in general (e.g., Norman et al. 2002; Mignoli et al. 2019). Indeed, Lyα fluxes of a few 10⁻¹⁶ erg s cm⁻² s⁻¹ were detected in the narrow-line AGNs at z ∼ 3 with [O iii] fluxes similar to J221737.29+001823.4 (Law et al. 2018). Although J221737.29+001823.4 itself showed no strong Lyα, it is likely associated with the extended Lyα nebulae, which indicate the presence of abundant intergalactic gas (Umehata et al. 2019). Thus it is natural to consider that Lyα photons of J221737.29+001823.4 are absorbed and/or scattered by circum/intergalactic media to form the extended Lyα nebulae (e.g., Hennawi et al. 2009), although the detail of Lyα damping mechanism is beyond the scope of this study.

Table 2. Photometry

Items	Values
u⋆ (mag)	>27.2 (2σ)
B (mag)	27.28 ± 0.26
V (mag)	26.77 ± 0.20
R (mag)	26.57 ± 0.18
$i^{\prime}$ (mag)	26.12 ± 0.18
$z^{\prime}$ (mag)	25.79 ± 0.21
F814W (mag)	26.66 ± 0.32
J (mag)	>24.58 (2σ)
H (mag)	>24.66 (2σ)
Ks (mag)	22.58 ± 0.08
3.6 μm (mag)	22.15 ± 0.06
4.5 μm (mag)	21.92 ± 0.05
5.8 μm (mag)	21.86 ± 0.16
8.0 μm (mag)	21.67 ± 0.18
24 μm (μJy)	<100 (3σ)
1.2 mm (μJy)	<75 (3σ)
3.0 mm (μJy)	<43 (3σ)
6 GHz (μJy)	9.56 ± 0.85
3 GHz (μJy)	13.07 ± 1.64
1.4 GHz (μJy)	<48 (4σ)
0.5-2 keV (erg cm−2 s−1)	${1.0}_{-0.4}^{+0.6}\times {10}^{-16}$
2-8 keV (erg cm−2 s−1)	${11.2}_{-3.3}^{+4.3}\times {10}^{-16}$

Download table as:  ASCIITypeset image

J221737.29+001823.4 was observed by the Atacama Large Millimeter/submillimeter Array (ALMA) in Band-6 and Band-3, which give upper limits at 1.2 mm and 3 mm (Umehata et al. 2015, 2017, 2018, 2019). J221737.29+001823.4 was also observed by the Karl G. Jansky Very Large Array (VLA) C band (6 GHz), S band (3 GHz; Ao et al. 2017), and L band (1.4 GHz; Chapman et al. 2004), and is detected in 3 and 6 GHz. Here we briefly explain the recent C-band observations (details will be presented in H. Umehata et al. 2022, in preparation). The field of view covers an entire field of ALMA deep survey field in the SSA22 (Umehata et al. 2018), which includes J221737.29+001823.4. Observations were carried out in the A and B configurations covering 4.2–8.2 GHz with a total on-source time of 89 hours. After data reduction and imaging with VLA Common Astronomy Software Applications (CASA), the resultant map has a synthesized beam size of 089 × 079 (PA = 2471) and an rms level 0.35 μJy beam⁻¹ at the phase center. We measure the 3 GHz and 6 GHz fluxes using CASA/imfit. We also put a point-source upper limit at 1.4 GHz. We put an upper limit of Spitzer Multi-Band Imaging Photometer (MIPS) 24 μm (Webb et al. 2009; see also Kubo et al. 2013). We use the Chandra X-ray flux values listed in Lehmer et al. (2009b). Its radio flux is around the detection limit of the VLA COSMOS 3 GHz survey (Smolčić et al. 2017), and the X-ray flux is lower than the detection limit of the Chandra COSMOS Legacy survey (Marchesi et al. 2016). To summarize, J221737.29+001823.4 is an object hardly detected by deep and wide X-ray and radio surveys to date.

The NIR spectroscopic observations were performed using MOIRCS (K15) and Multi-Object Spectrometer For Infra-Red Exploration (MOSFIRE; McLean et al. 2012) on the Keck-I telescope (see Umehata et al. 2019 and H. Umehata et al. 2022, in preparation for details). We show the slit positions in Figure 2. The observations were conducted using three slit position angles, 5 (slit A), 105 (slit B), and 58 (slit C) degrees. For slits A and B, we conducted a K-band spectroscopy for 3.2 and 3.3 h net exposure with MOSFIRE on 2020 September 11 and 12 (Umehata et al. 2019 and H. Umehata et al. 2022, in preparation). For slit A, an H-band spectrum was obtained with 2.5 h exposure with MOSFIRE on 2020 September 11 (see detail in K21). The slit widths were 07 for all the MOSFIRE observations. For slit C, we conducted a K-band spectroscopy with 3.8 h exposure with MOIRCS on the Subaru telescope on 2012 October 29 (K15). The location of slit C was offset from the center of the target in order to observe the other two targets simultaneously. The slit width for the MOIRCS observation was 08. All observations were conducted in good seeing conditions with FWHM PSF sizes of 05–08. Both MOIRCS and MOSFIRE observations were performed by two-position mask-nod sequence dither along slits to perform sky subtraction accurately. The calibration of the fluxes was performed using one telluric standard star at similar air masses taken before or after the observations each night. The spectral resolutions for MOSFIRE K (H) and MOIRCS K band at the given configurations were 3620 (3660) and 1700, respectively.

Zoom In Zoom Out Reset image size

Figure 2 pipeline, shows the spectra obtained using the MOIRCS and MOSFIRE. The spectra obtained by both instruments are corrected of slit losses. [O ii]λ λ3727, 3729, Hβ, and [O iii] are detected significantly. Although the observing methods and conditions are not uniform, the Hβ and [O iii] fluxes measured at each slit location do not differ greatly. In the case of J221737.25+001816.0 in K21, which was taken with the same masks as both slit A and slit B, the differences between the slit-loss-corrected fluxes from MOSFIRE spectroscopy and the Kron fluxes based on the MOIRCS imaging in H and K_s bands were ∼20%. Thus the flux calibration in this study is almost correctly performed.

We used X-CIGALE (Yang et al. 2020), which models the SED of a host galaxy and X-ray-to-radio emission from an AGN simultaneously. The model parameters used in X-CIGALE are summarized in the Appendix. Briefly, we included a stellar population synthesis model with dust attenuation, dust emission, an optical to infrared emission model for an AGN, X-ray power-law emission, and radio synchrotron emission. For stellar population models, we adopted a Chabrier (2003) IMF and a Bruzual & Charlot (2003) model with solar metallicity, and dust attenuation law from Calzetti et al. (2000). We assumed a delayed exponentially declining SFH model described as,

$$\begin{eqnarray*}&&\mathrm{SFR}(t)\propto \displaystyle \frac{({t}_{o}-t)}{{\tau }^{2}}\times \exp (-({t}_{o}-t)/\tau )\,\mathrm{for}\,0\,\leqslant t\,\leqslant {t}_{o},\end{eqnarray*}$$

where t is the look-back time, t_o is the look-back time onset of star formation, and τ is the e-folding time at which the SFR peaks. We use the dust emission templates from Dale et al. (2014), which are parameterized by a power-law slope α of dM_d (U) ∝ U^−α dU, where M_d is the dust mass, and U is the radiation field intensity and AGN fraction. The AGN fraction of this component is set to zero as we included another AGN model. X-CIGALE adopts an energy balance principle for the emission from stars, i.e., the energy emitted by dust in the IR corresponds to the energy absorbed by dust in the UV to optical.

We adopted models of rest-frame UV to IR emission from an AGN according to Fritz et al. (2006). Here we fixed the parameters at default values of X-CIGALE except for position angle (ψ) and AGN fraction. The parameters fixed here can affect the mid-IR (MIR) SED shape, but this is out of the scope of this study as the MIR flux of our target is not well constrained. The X-ray power law is parameterized by a photon index Γ. The Fritz et al. (2006) and X-ray models are constrained not to have an optical to X-ray luminosity ratio significantly different from the empirical values (Just et al. 2007). The radio synchrotron emission in X-CIGALE is parameterized with a power-law spectral slope α where radio flux F_ν ∝ ν^−α , with radio–IR correlation q_IR (Helou et al.1985).

We did not include the nebular emission in X-CIGALE that is useful for modeling young star-forming galaxies but not important for a massive quiescent galaxy like our target. Our target has strong nebular emission lines from an AGN, which is not included in the AGN model templates, and their contributions to the SED were subtracted in advance. Low-mass X-ray binaries (LMXB), high-mass X-ray binaries (HMXB), and hot gas from a host galaxy were also considered for the X-ray model, but according to the recipe adopted in X-CIGALE, their contribution to X-ray luminosities are expected to be 2 or more orders of magnitude lower than that of the AGN X-ray luminosity of our target. As shown in Figure 2, as it is hard to determine the systemic redshift of the host galaxy, hereafter we adopt z = 3.085 to calculate the physical properties. We also ran X-CIGALE with a fixed position angle ψ = 0°, where the contribution of the AGN is negligible at rest-frame UV to optical wavelength.

As shown in Figure 2, all [O iii] line profiles have significantly redshifted tails. Hβ shows no significant emission from a broad-line region (FWHM > 2000 km s⁻¹). We measured the fluxes and line profiles of the emission lines by fitting them with combinations of Gaussian profiles. The composites of two or three Gaussian profiles are often used to fit the [O iii] line profiles of AGNs (e.g., Harrison et al. 2014; Zakamska & Greene 2014). Zakamska & Greene (2014) found that three-component Gaussian models are better than two- or one-component Gaussian models for 400 out of 568 obscured AGNs found in the Sloan Digital Sky Survey (SDSS). According to Villar-Martín et al. (2011), such a three-component Gaussian model consists of two narrow components and one very broad (FWHM ≳ 1000 km s⁻¹) component. Such a broad component can appear as a distinct tail in a spectrum, but it was not identified well in our target maybe due to the low sensitivity. As we describe later, the line profiles of our target are approximated well by combinations of the two narrow Gaussian profiles without another very broad component. Thus to compare the line properties properly with literatures, we fitted each [O iii] with a combination of two Gaussian components.

The line widths, central wavelengths, and intensity of the two Gaussians are free parameters. [O iii]λ4959 was fitted simultaneously with [O iii]λ5007, adopting the same model profile and a fixed line ratio [O iii]λ4959/[O iii]λ5007 = 0.33. First, we subtracted the residual sky from each spectrum. The spectra significantly affected by OH airglow, which were automatically defined as regions where σ > 1 − 5 × 10⁻¹⁷ erg cm⁻² s⁻¹ Å⁻¹, were masked in advance (gray shaded in Figure 2) and not used in the fits. We then fitted the above models with Markov Chain Monte Carlo using emcee (Foreman-Mackey et al. 2013) in Python. The 5th, 50th, and 95th percentiles of the samples in the marginalized distributions are quoted as means and uncertainties. The line center is defined as velocity v₅₀ at which 50% of the line flux accumulates. The line width is defined as W₈₀, the velocity width of the line containing 80% of the total flux (W₈₀ = v₉₀ − v₁₀; see details in Harrison et al. 2014). We also measured a dimensionless relative asymmetry = ((v₉₅ − v₅₀) −(v₅₀ − v₀₅))/(v₉₅ − v₀₅) (Zakamska & Greene 2014).

The fitting procedure for Hβ and [O ii] was similar to that of [O iii], but we adopted different models. Hβ was fitted with the best-fit model profiles for [O iii] for each slit multiplied with a free parameter. The [O ii] doublet was fitted using a combination of two Gaussian profiles with the same line widths and fixed separation of the central wavelengths but with the line ratio as a free parameter. Finally, we corrected their instrumental broadening for W₈₀ assuming ${W}_{80,\ \mathrm{corr}}\,={[{{W}_{80}}^{2}\mbox{--}{{W}_{80,\ \mathrm{inst}}}^{2}]}^{1/2}$, where the W₈₀ is the original value obtained in the spectral fitting and, W_{80, inst} is W₈₀ for a Gaussian with the width owed to the instrumental broadening ≈ 60 km s⁻¹. The corrections are less than 2% of the original values. We evaluated the spatial extent of the [O iii] along the slits by taking the average spatial profiles between 20420 and 20490 Å.

Figure 2 shows the best-fit spectra and spatial extent of the [O iii] emission line at each position. The red curves show the best-fit profiles. The best-fit model profiles for [O iii] give reasonable fits for Hβ while the observed [O ii] may trace only a blue part of the [O iii] as the green vertical lines in Figure 2 show the z₅₀ of [O ii] corresponding to the [O iii].

Figure 3 and Table 3 show the best-fit SED and parameters of J221737.29+001823.4, respectively. The AGN dominates the rest-frame UV emission. Similar to our previous SED modeling without an AGN component (K15), the best-fit SED is that of a massive quiescent galaxy but due to the contribution of the AGN to the rest-frame UV, old stars are more dominant where the best-fit age and stellar mass become larger. The reduced χ² value for the ψ = 0 model is larger than that for the ψ free model, but still the ψ = 0 model can fit the observed fluxes well. Therefore, we cannot reject the possibility that J221737.29+001823.4 has in fact a smaller ψ, and the AGN contribution on the SED at rest-frame UV is overestimated.

Zoom In Zoom Out Reset image size

Table 3. SED Parameters

Model	Parameter	Value (free ψ) a	Value (ψ = 0) b
(1)	(2)	(3)	(4)
	Reduced χ2	0.91	1.60
Stellar	agemain/Gyr	1.68 ± 0.29	1.40 ± 0.04
	τmain/Myr	45 ± 54	200 ± 0
	E(B − V)/mag	0.12 ± 0.04	0.09 ± 0.02
	age⋆ c /Gyr	1.59 ± 0.29	1.00 ± 0.36
	M⋆/1010 M⊙	9.91 ± 1.67	6.22 ± 0.33
	SFR/M⊙ yr−1	0.05 ± 0.02	3.41 ± 0.84
Dust	α	0.71 ± 0.70	0.14 ± 0.11
Fritz06	ψ/deg	51 ± 3	0
	fAGN	0.16 ± 0.16	0.14 ± 0.14
X-ray	Γ	0.67 ± 0.20	0.65 ± 0.21
	$\mathrm{log}({L}_{2-10\ \mathrm{keV}}$/erg s−1)	43.2 ± 0.1	43.2 ± 0.1
Radio	qIR	0.61 ± 0.28	0.61 ± 0.27
	α	0.61 ± 0.18	0.73 ± 0.22
	$\mathrm{log}({L}_{1.4\ \mathrm{GHz}}$/erg s−1 Hz −1)	31.05 ± 0.06	31.08 ± 0.09

Notes.

^a The values and errors calculated with X-CIGALE. ^b Calculated with a fixed position angle ψ = 0°. ^c The stellar mass weighted age.

Download table as:  ASCIITypeset image

The best-fit SED from X-CIGALE at IR is lower than the upper limit of the observed fluxes. Adopting the average 1.2 mm flux to IR luminosity L_IR relation for the SED library in Danielson et al. (2017), the conservative upper limit of L_IR is ∼ 0.9 − 2.0 × 10¹¹ L_⊙ taking the 95% confidence interval. It corresponds to SFR < 9 − 21 M_⊙ yr⁻¹ using the L_IR to SFR conversion in Kennicutt & Evans (2012). Hereafter we use L_IR = 2.0 × 10¹¹ L_⊙ and SFR = 21 M_⊙ yr⁻¹ as a a conservative upper limit.

Table 4 lists the strength, line center and line width at each slit position. Using the largest flux value at slit B, J221737.29+001823.4 has a [O iii] luminosity of $\mathrm{log}({L}_{[{\rm{O}}\ \mathrm{III}]}\,/\,\mathrm{erg}\,{{\rm{s}}}^{-1})\,=43.28\pm 0.01$. The observed X-ray luminosity is $\mathrm{log}({L}_{2-10\mathrm{keV}}/\mathrm{erg}\,{{\rm{s}}}^{-1})=43.2\pm 0.1$. Thus L_{[O III]} is more than 10 times larger than that expected from the empirical X-ray to [O iii] luminosity relation for type-1 AGNs in the local universe (e.g., Ueda et al. 2015). Furthermore, our target has a flat X-ray SED (Γ = 0.67 ± 0.20), which suggests significant absorption in the X-rays where Γ ∼ 1.8 is typical (e.g., Yang et al. 2016; Liu et al. 2017). We then estimated the intrinsic X-ray luminosity based on the [O iii] luminosity. Adopting the X-ray to [O iii] luminosity relation for Seyfert-1 AGNs in the local universe (Ueda et al. 2015), J221737.29+001823.4 has an intrinsic X-ray luminosity of $\mathrm{log}({L}_{2-10\mathrm{keV},\mathrm{intrinsic}}/\mathrm{erg}\,{{\rm{s}}}^{-1})\approx 44.4$. Applying the X-ray to bolometric correction as in Marconi et al. (2004), the source has a bolometric luminosity of $\mathrm{log}({L}_{\mathrm{bol}}/\mathrm{erg}\,{{\rm{s}}}^{-1})\approx 46.0$, which would make it a moderately luminous QSO. Assuming a scaling relation between the black hole mass and the host galaxy stellar mass of M_BH ≈ 0.002 M_⋆, following Aird et al. (2012), J221737.29+001823.4 may have M_BH ≈ 2 × 10⁸ M_⊙ and Eddington luminosity L_Edd = 1.3 ×10³⁸ M_BH/M_⊙ ≈ 2.6 × 10⁴⁶ erg s⁻¹. Thus it may have a relatively large Eddington ratio of L_bol/L_Edd ∼ 0.4.

Table 4. [O iii] Line Profile

Slit	z	Voffset a	FWHMcorr	fraction b
		(km s−1)	(km s−1)
(1)	(2)	(3)	(4)	(5)
A	${3.0832}_{-0.0001}^{+0.0002}$	⋯	${242}_{-17}^{+24}$	${0.68}_{-0.06}^{+0.04}$
	${3.0905}_{-0.0036}^{+0.0005}$	${533}_{-265}^{+36}$	${253}_{-24}^{+83}$	${0.32}_{-0.04}^{+0.06}$
B	${3.0832}_{-0.0001}^{+0.0002}$	⋯	${229}_{-17}^{+25}$	${0.67}_{-0.07}^{+0.04}$
	${3.0905}_{-0.0027}^{+0.0005}$	${535}_{-245}^{+32}$	${240}_{-21}^{+79}$	${0.33}_{-0.04}^{+0.07}$
C	${3.0848}_{-0.0005}^{+0.0005}$	⋯	${147}_{-54}^{+69}$	${0.81}_{-0.22}^{+0.08}$
	${3.0914}_{-0.0037}^{+0.0009}$	${484}_{-247}^{+73}$	≲180	${0.19}_{-0.08}^{+0.21}$

Notes.

^a The velocity offset of this component from the brighter component. ^b The fraction of this component in the total flux.

Download table as:  ASCIITypeset image

J221737.29+001823.4 has a radio spectral index α = 0.61 ± 0.18, where F_ν ∝ ν^−α , which is a typical value of radio synchrotron emission (e.g., Smolčić et al. 2017). The luminosity at 1.4 GHz is given as ${L}_{1.4\ \mathrm{GHz}}\,={F}_{1.4\mathrm{GHz}}4\pi {{D}_{L}}^{2}{(1+z)}^{(\alpha -1)}$ where D_L is the luminosity distance, and F_1.4GHz is the flux at 1.4 GHz, which is 0.021 ± 0.003 mJy from X-CIGALE. Then we find a radio luminosity of $\mathrm{log}({L}_{1.4\ \mathrm{GHz}}$ /erg s⁻¹ Hz⁻¹) = 31.05 ± 0.06. It would be classified as a radio-quiet AGN adopting the criteria suggested by Xu et al. (1999). The logarithmic ratio of the infrared and radio fluxes q_IR of J221737.29+001823.4 is 0.61 ± 0.28 for the best-fit model of X-CIGALE, where q_IR is defined as ${\mathrm{log}}_{10}({F}_{\mathrm{FIR}}/(3.75\times {10}^{12}{Hz})/{F}_{1.4\ \mathrm{GHz}})$ (Helou et al. 1985). Using a conservative upper limit on L_IR, q_IR is ≲1.4. It is lower than that of pure star-forming galaxies with 2.4 in Ivison et al. (2010) and comparable to radio-loud AGNs at low redshift (e.g., Toba et al. 2019). At high redshift, QSOs selectable by wide-field surveys like the SDSS have q_IR similar to that of star-forming galaxies (Ibar et al. 2008; Murphy 2009), while radio sources with low q_IR like local radio-loud AGNs are found by deep surveys like the VLA COSMOS 3 GHz survey (Delhaize et al. 2017). Thus J221737.29+001823.4 is one of the most local radio-loud AGN-like objects at high redshift selectable by radio surveys to date.

Table 5. Detected Emission Lines

Slit	line	z50	W80,corr	Rel. asym.	Flux	[OII]λ3727/λ3729
			(km s−1)		(10−17 erg cm−2 s−1)
(1)	(2)	(3)	(4)	(5)	(6)	(7)
A	[OIII]λ5007	${3.0850}_{-0.0002}^{+0.0002}$	${1174}_{-50}^{+133}$	${0.14}_{-0.03}^{+0.03}$	${21.0}_{-0.8}^{+0.8}$	⋯
	Hβ				${2.2}_{-0.2}^{+0.2}$	⋯
	[OII]λ λ3727, 3729	${3.0821}_{-0.0010}^{+0.0012}$	${354}_{-128}^{+128}$		${1.4}_{-0.3}^{+0.3}$	${1.0}_{-0.3}^{+1.8}$
B	[OIII]λ5007	${3.0851}_{-0.0002}^{+0.0002}$	${1148}_{-47}^{+140}$	${0.15}_{-0.03}^{+0.04}$	${22.7}_{-0.9}^{+0.9}$	⋯
	Hβ				${2.0}_{-0.2}^{+0.2}$	⋯
C	[OIII]λ5007	${3.0857}_{-0.0005}^{+0.0006}$	${903}_{-91}^{+254}$	${0.07}_{-0.07}^{+0.11}$	${19.9}_{-2.1}^{+2.6}$	⋯
	Hβ				${2.0}_{-0.5}^{+0.5}$	⋯

Download table as:  ASCIITypeset image

Figure 4 shows the 6 GHz image and the flux contours of the radio emission overlaid on the K_s -band image. J221737.29+001823.4 is well resolved at 6 GHz, where the image component sizes after deconvolution of the beam are (0.99 ± 0.05) × (0.52 ± 0.05) arcsec and PA = 1512 ± 46. There is also a diffuse component with an orientation different from that of the bright central source. The source lies at the low-power end of the distribution for compact steep spectrum radio galaxies (e.g., Gelderman & Whittle 1994; O'Dea 1998), which are believed to be young systems evolving into more extended radio galaxies or systems with spatial growth restricted by the dense interstellar medium (ISM). The radio position angle is different from the slit directions for NIR spectroscopy, and then it is not clear whether the [O iii] traces the radio morphology. However, the similar spatial extents of [O iii] and radio emission can indicate their physical connection though both are limited by the observational depths.

Zoom In Zoom Out Reset image size

Although the small notches of the spectra left from the model spectra indicate more complex kinematics, each spectrum is well approximated by a combination of two Gaussian profiles. The obtained spectra have large relative asymmetries 0.07–0.15 and line widths of ≈1000 km s⁻¹. The spatial extent of [O iii] (≈15 kpc in Figure 2) of J221737.29+001823.4 is much larger than the typical size of a massive quiescent galaxy at high redshift (a few physical kiloparsec; e.g., Shibuya et al. 2015), though the size of the host galaxy cannot be evaluated robustly because the K_s -band image can be significantly contaminated by [O iii] emission. These line profiles observed at each slit indicate a significant outflow of ionized gas spread over the host galaxy. Such galactic wide-scale outflows of ionized gas have been observed among AGNs in the local universe and at high redshift (e.g., Harrison et al. 2014; Genzel et al. 2014; Harrison et al. 2016; Law et al. 2018; Kakkad et al. 2020).

Table 5 summarizes the ingredient of each spectrum. At all slit locations, a spectrum consists of two narrow components. The redshifts of the bluer components at slits A and B match with that of [O ii] within errors, while the redshifts of both components at slit C are redder than [O ii]. Therefore, J221737.29+001823.4 consists of one component also detected with [O ii] at z ≈ 3.0832 (i) and at least two redshifted components (a bluer component at slit C (ii) and a redder components at slits A, B, and C (iii), at the resolution of this study). Comparing with local AGNs with [O iii] line width as wide as J221737.29+001823.4 in literature (Villar-Martín et al.2011; Fu et al. 2012; Zakamska & Greene 2014), J221737.29+001823.4 has components with narrower line widths and a larger velocity offset between the first and second brightest components. Then the wide line widths of the whole line profiles of J221737.29+001823.4 are due to the large velocity offsets of the two components, while those are due to broad components for most of the type-2 AGNs. The broader components found in such type-2 AGNs are generally blueshifted from the narrower components and regarded as highly disturbed outflowing gas. In the case of J221737.29+001823.4, component (iii) seen at all slit positions likely comes from the gas near the central SMBH. As the emission lines are spatially extended at all slit directions, components (i) and (iii), which dominate the emission-line fluxes at slits A and B and slit C, respectively, likely come from the outflowing extended emission-line region. The narrower line width indicates that the outflowing gas is not highly perturbed like typical type-2 AGNs with wide line widths. Maybe the outflowing gas of this system is not disturbed because of young age or lack of intergalactic gas to interact, or because the wing of the gas is highly attenuated by dust.

Here we estimate the mass outflow rate of J221737.29+001823.4 following Fiore et al. (2017), using the [O iii] W_80,corr and the spatial extent and Hβ luminosity for slit A, which are similar to those for slits B and C. Based on the Hβ luminosity, the mass of the outflowing ionized gas is estimated as

$$\begin{eqnarray*}&&{M}_{{\rm{H}}\beta }=7.8\times {10}^{8}C\left(\displaystyle \frac{{L}_{H\beta }}{{10}^{44}}\right){\left(\displaystyle \frac{\langle {n}_{e}\rangle }{{10}^{3}}\right)}^{-1},\end{eqnarray*}$$

following Osterbrock & Ferland (2006) and Carniani et al. (2015), where $C=\langle {n}_{e}{\rangle }^{2}/\langle {n}_{e}^{2}\rangle$ (set to unity, which is a conservative lower limit); a gas temperature T = 10⁴ K is assumed. Adopting n_e = 200 cm⁻³ following Fiore et al. (2017), we find ${M}_{{\rm{H}}\beta }={7.2}_{-0.7}^{+0.7}\times {10}^{7}$ M_⊙.

The mass outflow rate of ionized gas is calculated as,

$$\begin{eqnarray*}&&\dot{M}=3\times {v}_{\max }\times {M}_{{\rm{H}}\beta }/R,\end{eqnarray*}$$

where v_max is the wind maximum velocity, and R is the radius at which the mass outflow rate is computed. v_max is defined as W_80,corr /1.3, and R is defined as the maximum radius at which the high-velocity gas is detected. Here we use half of the extent of [O iii] for each slit, ≈7.1 kpc, corrected of the spatial resolution with FWHM PSF of ≈4.5 kpc on average. We find a mass outflow rate 22 ± 3 M_⊙ yr⁻¹. Its mass loading factor $\eta =\dot{M}$/SFR is around unity at lower limit. $\dot{M}$ Note that our estimate does not include outflowing gas at a neutral or molecular state. Including them, the total mass outflow rate and thus η could be a few to 10 times larger than that of the ionized gas (e.g., Rupke & Veilleux 2013; Carniani et al. 2015; Fiore et al. 2017; Fluetsch et al. 2019).

We compared the mass outflow rate, AGN bolometric luminosity, SFR, and specific SFR (sSFR) of J221737.29+001823.4 with results in Fiore et al. (2017) who summarized the outflow properties of QSOs at low and high redshift in the literature, and Leung et al. (2019) who investigated the outflow properties for X-ray AGNs at 1.4 < z < 3.8 in Figure 5. We used the upper limit on SFR from the 1.2 mm flux ( < 21 M_⊙ yr⁻¹) and the stellar mass estimated assuming ψ = 0 to obtain sSFR as a conservative limit. J221737.29+001823.4 is not a special object; its ionized gas mass outflow rate is similar to those of AGNs with similar bolometric luminosities. As previous studies have shown, the SFRs of AGNs with strong outflows range widely from quiescent (sSFR < 10⁻¹⁰ yr⁻¹) to starbursting.

Zoom In Zoom Out Reset image size

We show the emission line diagnostics of J221737.29+001823.4 measured at slit A in Figure 6. The left panel shows O32 versus R23, where O32 and R23 are defined as $\mathrm{log}$ ([O iii]/[O ii]) and $\mathrm{log}$ (([O iii]+[O ii])/Hβ), respectively. The right panel shows the [O iii]λ5007/Hβ versus [O ii]λ λ 3726,3729/Hβ line diagnostics. The magenta filled circle shows the line ratio of J221737.29+001823.4. The blue and red diamonds show the line ratios of the bluer and redder components. The black filled contour and blue contour show the star-forming galaxies and AGNs, respectively, selected from the SDSS using spectroscopic data products from the Max Planck Institute for Astrophysics and Johns Hopkins University DR8 catalog (MPA-JHU; Brinchmann et al. 2004; Kauffmann et al. 2004; Tremonti et al. 2004). The star-forming galaxies and AGNs are classified using Baldwin–Phillips–Terlevich (BPT) diagrams (Baldwin et al. 1981) following Kewley et al. (2006). We applied extinction correction using the Hα/Hβ line ratio adopting the Cardelli et al. (1989) extinction law for the SDSS, while no extinction correction was applied for our target.

Zoom In Zoom Out Reset image size

J221737.29+001823.4 has large O32 and [O iii]λ5007/Hβ indices and is at the edge of the distribution of local AGNs. Such line ratios are seen among young star-forming galaxies with high ionization parameters at z = 2 ∼ 3 (Strom et al. 2018), while the AGN is plausibly the dominant ionizing source for our target given the low SFR and the evolved stellar population. Law et al. (2018) finds similarly high [O iii]λ5007/Hβ for the optically faint (R = 20–25) AGNs at z ∼ 2–3, and they are certainly classified as AGNs using the BPT diagram.

We compare the line ratios with models of photoionization by AGNs (Groves et al. 2004) and shock excitation by outflows (Allen et al. 2008). Groves et al. (2004) provides line intensities predicted for dusty and radiation-pressure-dominated photoionization models using the MAPPING III code (Sutherland & Dopita 1993). It describes the intensities of the emission lines from the narrow-line region (NLR) of AGNs with power-law ionizing SED with hydrogen density n_H = 1000 cm⁻³, α = − 1.2 to −2.0, ionization parameter $\mathrm{log}U=-4$ to 0, and metallicity Z = 0.25–2 Z_⊙. Here we show the dusty model in Groves et al. (2004) with α = − 2, $\mathrm{log}U=-4$ to 0, and Z = 0.25–2 Z_⊙. Allen et al. (2008) also uses MAPPING III but provides line intensities predicted for shock excitation. They provide models with pre-shock density n = 0.01–1000 cm⁻³, velocity v < 1000 km s⁻¹, magnetic field B = 10⁻¹⁰ to 10⁻⁴ G, and magnetic parameter B/n^1/2 = 10⁻⁴ to 100 μG cm^3/2. We show the models with n = 1000 cm⁻³, v = 100 to 1000 km s⁻¹, and B = 0.1 to 1000 μ G. We choose α = − 2 and n = 1000 cm⁻³ models as only these models have line ratios close to our target.

The line ratios of J221737.29+001823.4 are close to those of the photoionization models with high ionization parameter $\mathrm{log}U\gt -1$. This indicates that most of its emission originates in the central AGN rather than the shock induced by, e.g., radio jets. The line ratios of individual velocity components are also consistent with the photoionization models. Thus the outflowing gas of J221737.29+001823.4 is likely ionized by the central AGN rather than shock excitation. This is consistent with the spatially resolved spectroscopic analysis in, e.g., Fischer et al. (2017) and Revalski et al. (2018) for local AGNs. We note that there are several known uncertainties in our results. First, the line ratio based on the slit spectroscopy can be a composite of several independent components. For a more detailed discussion, we need a spatially resolved spectroscopy. Second, the dust attenuation is not corrected for our target but is negligible assuming the absorption on emission lines E(B − V)_line =E(B − V)/0.44 ≈ 0.2 measured with X-CIGALE; R23 and [O iii]/Hβ do not change significantly, while O32 can decrease by ∼0.1 and log [O ii]/Hβ can increase by ∼0.1 adopting extinction law of Cardelli et al. (1989).

We can also measure the electron density (n_e ) using the [O ii]λ3727 to λ3729 line ratio (Osterbrock & Ferland 2006). Following Sanders et al. (2016), the line ratio of our target corresponds to an electron density n_e = 10–1500 cm⁻³, which is within the expected range for NLR, 10–10000 cm⁻³ (e.g., Nesvadba et al. 2006; Liu et al. 2013).