WISE J233237.05−505643.5: A DOUBLE-PEAKED, BROAD-LINED ACTIVE GALACTIC NUCLEUS WITH A SPIRAL-SHAPED RADIO MORPHOLOGY

The radio continuum observations were carried out at the ATCA using the Compact Array Broadband Backend with a 2 × 2 GHz bandwidth. W2332−5056 was mapped at 1.5 GHz in the 6C configuration on UT 2010 July 16 and 17. The 5.6 GHz and 9.0 GHz data were obtained simultaneously on UT 2010 July 18 and 19 in the 6C configuration (maximum baseline ∼6.0 km) and on UT 2010 August 10 in the hybrid H168 configuration (with five antennas closely positioned within 200 m of each other and the sixth antenna at a distance of 4.5 km). Observations were taken in snap-shot mode over 12 hr (6C) and 6 hr (H168) periods. A total of 25 minutes of integration was obtained at 1.5 GHz and 166 minutes at 9.0 and 5.6 GHz. PKS 1934-63 was used as the flux calibrator while PKS 2311-452 (for 5.6 and 9.0 GHz) and PKS 2333-528 (all three bands) were used for phase calibration. The largest angular scales (θ_LAS) visible under these configurations are 55 at 1.5 GHz, 39 at 5.6 GHz, and 23 at 9.0 GHz.

The data reduction was done using miriad. The final robust weighted images are shown in Figures 1(a)–(c). The primary beam at the highest frequency, 9.0 GHz, is >6', significantly larger than the 1' field of interest for W2332−5056, and hence we did not apply a primary beam correction. The rms of the 1.5 GHz map is 0.62 mJy beam⁻¹ with a beam size of 1312 × 480 and a position angle (P.A.) = 6°. The (u,v) data at 5.6 GHz and 9.0 GHz from the two configurations were combined, resulting in an rms noise of 29 μJy beam⁻¹ and 37 μJy beam⁻¹ in the final synthesized beams of 285 × 151, P.A. = 13° and 180 × 087, P.A. = 8°, respectively. The results of the integrated flux measurements using aips are listed in Table 1. Because the θ_LAS for each band is larger than the angular extent of W2332−5056, we anticipate negligible or minimal missing flux in the ATCA interferometric measurements.

Zoom In Zoom Out Reset image size

Table 1 presents mid-IR photometry of W2332−5056 from the WISE All-Sky Release Atlas Catalog (Cutri et al. 2012). The observations of this field were carried out on UT 2010 May 19 and 20. A total of 12 single frames in each WISE band are coadded. The spatial resolution of WISE is ∼6'' in W1, W2, and W3 and ∼12'' in W4. W2332−5056 is well detected in W1−W3, with a signal-to-noise ratio >15, and marginally detected (4σ) in W4. The mid-IR fluxes are listed in Table 1.

In addition, W2332−5056 was observed in W1 and W2 during the WISE post-cryogenic survey on UT 2010 November 13, 16, 17, and 18. A total of 19 frames were acquired, spread across 3 days due to the satellite maneuvering to avoid the Moon during its Sun-synchronous polar orbit. These second-epoch measurements allow us to test the variability of the AGN in W2332−5056 at mid-IR wavelengths, as discussed in Section 3.4.

Finally, due its projected proximity to SPT-CL J2331−5051, W2332−5056 was also observed in [3.6] during Spitzer Cycle-9 (ID: 60099, PI: Mark Brodwin) on UT 2009 November 26. The pipeline-processed data (PBCD; shown in panel (d) of Figure 1), retrieved from the Spitzer Archive, provide another mid-IR epoch for comparison.

2.3.1. Optical and Near-IR Imaging

2.3.2. Optical Spectroscopy

The optical spectroscopic observations of W2332−5056 were carried out with the Gemini Multi-Object Spectrograph (GMOS; Hook et al. 2004) on the Gemini-S telescope on UT 2011 May 11 and UT 2011 November 28. We obtained two 10 minute integrations observations with the 15 slit and the B600/520.0 disperser in 2011 May, covering 3800–6700 Å with a resolving power R ≡ λ/Δλ = 1700. At z = 0.3447, the Hβ line is near the red end of the wavelength coverage. In 2011 November, in order to cover Hβ and Hα together and confirm the BL Hβ feature observed earlier, we revisited W2332−5056 with Gemini-S/GMOS, with the 15 slit and the R400/705.0 disperser. The wavelength coverage ranged from 6200 Å to 9100 Å with R = 1900. The spectra were processed using IRAF. The white dwarf GJ 318 was used for flux calibration in 2011 May and the DA white dwarf LTT3218 was used in 2011 November. The final combined spectrum is shown in Figure 2.

Zoom In Zoom Out Reset image size

W2332−5056 was serendipitously observed twice with the ACIS-I camera in programs to map the galaxy cluster SPT-CL J2331−5051. The total exposure time was 35.0 ks, split between a guaranteed time observation of 29.0 ks on UT 2009 August 12 (ObsID 9333; PI: Garmire) and a guest observation of 6.0 ks on UT 2009 August 30 (ObsID 11738; PI: Mohr). The X-ray data for the cluster were reported in Andersson et al. (2011). The data were analyzed following standard procedures using the Chandra Interactive Analysis of Observations (CIAO; V4.2) software. We identified "good time intervals" for the observations, yielding a total effective exposure time of 30.0 ks for W2332−5056.

The ATCA maps at 1.5 GHz (20 cm), 5.6 GHz (5.4 cm), and 9.0 GHz (3.3 cm) are shown in panels (a)–(c) of Figure 1. These maps clearly show a mixed morphology between FR-I and FR-II structures that comprise the W2332−5056 system. In the lowest resolution map, at 1.5 GHz (3''× 48), three distinct peaks are observed, including the core and two unevenly bright spots at both ends of the center. The central peak is offset by 30 from the location of the WISE-detected host galaxy (referred to as the core), which indicates significant unresolved 20 cm emission from the extended structure north of the core. At shorter wavelengths, the jet is well resolved and exhibits a complex morphology, described next.

In the 9.0 GHz map (180 × 087, P.A. = 8°), we identify six distinguishable components, marked as A–F in Figure 3: (A) a core (coincident with the host galaxy), a pair of "primary" jets, including (B) a SE jet with a P.A. of 120° and (C) the counter distorted northwest (NW) jet, (D) a linear structure at a nearly orthogonal P.A. of 20°, (E) a luminous hot spot, and (F) an arc close to the core. The fluxes of these radio structures are listed in Table 2.

Zoom In Zoom Out Reset image size

Table 2. Radio Continuum Flux Densities of Resolved Radio Components

Component	ID	F1.4 GHz	F5.6 GHz	F9.0 GHz	$\alpha ^{5.6}_{9.0}$
		(mJy)	(mJy)	(mJy)
Core	(A)	14 ± 1	5.0 ± 0.1	4.6 ± 0.1	−0.18 ± 0.07
Primary jet (SE)	(B)	34 ± 1	11.6 ± 0.2	9.5 ± 0.3	−0.41 ± 0.08
Primary jet (NW)	(C)	57 ± 1	27.8 ± 0.2	18.1 ± 0.3	−0.90 ± 0.03
Linear structure	(D)	...	1.5 ± 0.1	1.1 ± 0.1	−0.70 ± 0.15
Hot spot	(E)	...	21.2 ± 0.1	14.0 ± 0.2	−0.87 ± 0.03
Arc-like structure	(F)	...	...	1.83 ± 0.08a	...

Notes. Component IDs (second column) are marked in Figure 3. F_5.6 GHz and F_9.0 GHz are measured in the beam-matched maps (29 × 16, P.A. = 15°). α, as defined in F_ν∝ν^α, is the spectral index between 9.0 GHz and 5.6 GHz. ^aMeasured in the map of the beam = 180 × 087, P.A. = 8°.

Download table as:  ASCIITypeset image

At the position of the mid-IR and optical peak, a radio core is apparent. The core has a relatively flat spectrum: a spectral index of α = −0.18 ± 0.07 (F_ν∝ν^α) from 5.6 GHz to 9.0 GHz in the maps of matched beam of 29 × 16 at P.A. = 15°. The deconvolved size of the core is <03 at 9.0 GHz, based on single Gaussian fitting of the uniformly weighted map.

Originating from the core, the primary jet extends about 14'' to the NW, punctuated by a typical FR-II galaxy Doppler-boosted hot spot that is much brighter than the core. In contrast to the linear jet ("B") to the SE, the NW jet shows an asymmetric zig-zag morphology. The deviation is 10° from the axis of the SE jet, while a deviation of 30° is seen on the opposite side. Moreover, there appear to be distinct differences across the radio bands, notably closer to the core, better revealed in Figure 1(f). The jet is more curved at 5.6 GHz on the SE side and relatively straighter on the NW, spatially offset from the 9.0 GHz emission. The nature of this offset is under investigation with new, deeper radio observations that we are currently pursuing.

The faint linear structure "D," pointing to the center of the core, extends 5''–7'' from the central core, is visible at 5.6 GHz, and notably at 9.0 GHz (appearing as "green" in Figure 1(f)), extending over >2 beams. The emission at 1.5 GHz is also extended in the same direction, indicated by the elongation of the core in the low resolution 1.5 GHz map. The coalignment is shown in both 9.0 GHz and 5.6 GHz of the radio core and the resolved host in optical imaging (white/magenta in the figure). This structure has signal-to-noise >30. It is at a P.A. of 20°, which is distinct from the P.A. of the elongated ATCA beams at 5.6 GHz and 9.0 GHz (13° and 8°, respectively). Moreover, it does not match the uncleaned dirty beam structure of a relatively weak central core (the peak-to-peak ratio between the core and component "D" is ∼8). Structure "F" has an unusual shape resembling a curved, arc-like structure, possibly the result of a three-dimensional radio structure seen in projection.

The primary jets extend to 70 kpc on both sides at a P.A. of 120°. To our detection limit, there is no clear jet-like structure that connects directly to the core on the axis of the primary jet in the central 12 kpc (25). The western jet exhibits a complex behavior that may indicate interactions with the ISM and intergalactic medium. The structure of the eastern counter jet is more linear, with a maximum P.A. change of ≲10°. The pitch angle along the jet reaches 20° in the 5.6 GHz map, whereas it is within 5° at 9.0 GHz. If the driving mechanism that produces this "zig-zag" morphology in the jets is periodic, the estimated period is on the order of 0.25 Myr based on the length between off-axis peaks of the serpentine primary jet.

The asymmetric brightness of the jets suggests that the western side of the primary jet is on the side near to us. This is because jets can be brighter (fainter) by a factor of D³ if they are moving toward (away from) the observer (discussed in Meier et al. 2001), where D ≡ [Γ(1 + (v/c)cos θ)]⁻¹ is the Doppler factor for a relativistic jet at Lorentz factor Γ for given velocity v and the angle between the axis of the jet and the observer's line of sight θ. However, θ for the primary jets is likely to be >80° (e.g., the maximum for jets in the face-on plane) based on the relatively small brightness difference between the primary jets on each side.

The integrated fluxes of identified lines are listed in the Table 3. The presence of the high ionization line [Ne v]3426 Å, as well as the ratio of [O iii]5007 Å to NL Hβ indicate the AGN nature of W2332−5056. Additional evidence for an AGN is provided by broad Hα and Hβ line widths, as shown in Figure 2. Both have asymmetric profiles with respect to the NL. The FWHM of Hα is ∼11,000 km s⁻¹, and the full width at quarter maximum is ∼16,000 km s⁻¹. For Hβ, the peak of the BL component is centered at ∼6453 Å or 3800 km s⁻¹ blueshifted with respect to the systemic host galaxy redshift of 0.3447, consistently measured by both the narrow emission lines as well as the CaHK absorption. Similarly, the peak of the BL Hα is at 8730 Å or 3200 km s⁻¹ blueshifted with respect to the host galaxy systemic redshift. Objects with asymmetric, very broad Balmer lines whose peaks are shifted from the corresponding NLs are usually referred to as DPEs. We note a few narrow emission features (marked as "?" in the inset of Figure 2) straddling Hα, unlikely to be [N ii] or [S ii], which could potentially be associated with Hα components shifted by −1700 and +4000 km s⁻¹ with respect to the galaxy systemic velocity. The different blueshifts for the broad Hα and Hβ are the result of contamination from these unidentified components. From the difference between the scaled Hα and Hβ profiles, we detect no significant broad Fe ii emission near Hβ within ±10,000 km s⁻¹. We do not see clear evidence for very BL regions in other spectral features.

3.2.1. Parameterization with Gaussian Profiles

We first attempted to parameterize the BL components with two Gaussian profiles without excluding narrow emission lines in order to measure the FHWM and fluxes of all lines. The two BL Gaussian profiles and one NL component are used for both Hα and Hβ in the model, allowing us to fit independently Hβ and Hα. The narrow absorption features (between rest-frame 4866.4–4908.2 Å and 6529.3–6552.4 Å) are excluded from the fitting process. For the [O iii]4959 Å and 5007 Å lines, one BL Gaussian component and one NL component are assumed, while the [N ii]6583 Å is fit with a single NL Gaussian component. The redshift and linewidth of the [O iii]4959 Å and [O iii]5007 Å lines are assumed to be equal, with a fixed flux ratio $F_{ [{\rm O}\,\scriptsize{III}]5007}/F_{[{\rm O}\,\scriptsize{III}]4959} = 3$. The results ($\chi ^{2}_{\rm {red}}=1.85$, degrees of freedom (dof) = 3607) are shown in Figure 4 and Table 4.

Zoom In Zoom Out Reset image size

Table 4. Emission Line Parameters

Line	ΔV	FWHM	Flux
	(km s−1)	(km s−1)	(10−16 erg cm2 s−1)
Hα
NL	0 ± 40	390 ± 50	9 ± 1
BL-blue	−4410 ± 50	9180 ± 30	519 ± 2
BL-red	+5330 ± 130	9110 ± 60	229 ± 3
Hβ
NL	−70 ± 30	360 ± 60	1.6 ± 0.2
BL-blue	−4270 ± 90	8790 ± 10	74 ± 1
BL-red	+8860 ± 130	8800 ± 20	48 ± 1
[O iii] λ5007
NL	+70 ± 10	480 ± 10	33 ± 1
BL	+50 ± 50	1300 ± 120	8 ± 1
[N ii] λ6583
NL	−10 ± 10	340 ± 60	6 ± 1

Notes. The results of Gaussian fitting of the emission lines. The second column, ΔV, represents the velocity offset with respect to the rest-frame wavelength of the line at a redshift of 0.3447. For each emission line, one BL component and one NL component is included, except for Hα and Hβ, for which two BL components are employed. The [O iii]4959 Å (not shown in the table) and [O iii]5007 Å lines are forced to have the same ΔV and FWHM, with a 1:3 line flux ratio.

Download table as:  ASCIITypeset image

The blue BL components are centered at ≲ − 4300 km s⁻¹ with respect to the NL, while the red component peaks at > + 5300 km s⁻¹. Both blue and red BL components have FWHMs ≳8800 km s⁻¹. The flux of the blue components is ∼1.5–2.3 times that of the red components. The fitting residuals shown in the lower panels of Figure 4 indicate that the model represents the data near both Hα and Hβ fairly well, except for the blocked region and the very red wing of Hβ (ΔV > 11, 000 km s⁻¹).

3.2.2. Spectroscopic Modeling—The Case of a Thin, Elliptical Accretion Disk

In the literature, the DPE feature, such as we see in W2332−5056, has often been interpreted as a thin accretion disk surrounding a single AGN (Bonning et al. 2007; Shields et al. 2009; Eracleous et al. 2009). We test the disk-emitter scenario by fitting the Hα and Hβ profiles with models based on line emission from an elliptical accretion disk (Eracleous et al. 1995). The asymmetric blueshifted BL features in both Hα and Hβ in this model are due to Doppler-boosted line emission from the disk. All NL features (Hα, Hβ, [O iii]4959 Å, [O iii]5007 Å, [N ii]6584 Å) and an unexpected "absorption-like" region redward of NL Hβ (6535–6600 Å) are excluded from the model fitting. The eccentricity, e, angle of periastron, ϕ, inclination angle, i, and inner/outer radii, r_in/r_out, are free parameters for fitting. Each model fit Hα and Hβ with a radius-dependent power-law emissivity (r) = ₀ × r^−q for the disk, where r is the radius of the emission region from the center and q is the power-law index. Two q indices are used as independent parameters, q_Hα and q_Hβ, for the emissivities of Hα and Hβ, respectively. These emissivity assumptions for the photoionization models (Collin-Souffrin & Dumont 1989; Dumont & Collin-Souffrin 1990a, 1990b) were developed by Chen & Halpern (1989) and are extensively used by others, including Eracleous & Halpern (2003).

The modeling was performed by fitting a grid of free parameters and adopting the set of parameters that yields the lowest reduced χ² value. We tested elliptical disk models for eccentricities of e = 0.0–0.8 (shown in Figure 5 along with the best-fit circular model). The high eccentricity e = 0.6 model provides a statistically significant improvement to explain the BL profiles ($\chi ^{2}_{\rm {red}}=0.88$, dof = 1887), although models with e > 0.2 generally provide a good fit ($\chi ^{2}_{\rm {red}}\le 1.02$). The best-fit parameters of this model are as follows: e = 0.6, q_Hα = 2.4, q_Hβ = 3.2, i = 42°, FWHM local broadening σ = 3300 km s⁻¹, $r_{\rm disk}^{\rm inner}=240 r_{\rm G}$, and $r_{\rm disk}^{\rm outer}=4500 r_{\rm G}$, where r_G is the Schwarzschild radius of the central BH.

Zoom In Zoom Out Reset image size

This model with a thin disk and offset BL regions can reasonably explain the Hα and Hβ line profiles in W2332−5056, other than an apparent "absorption" feature at ΔV ∼ 1000 km s⁻¹ redshifted with respect to Hβ. This absorption-like profile is not observed for Hα, likely because it overlaps with [N ii]6583 Å emission at the systemic redshift of the galaxy. The nature of this feature is yet unknown.

The AGN hosted by W2332−5056 is strongly detected in Chandra ACIS-I data, with a total of 1790 source counts in the energy range 0.3–7 keV. It is the brightest X-ray source in the field. In the deeper observation (ObsID 9333), our target is 93 off-axis, where the Chandra/ACIS 50% encircled energy radius is 4'' at 1.4 keV. We extracted the source distorted by the off-axis point-spread function (PSF) using a relatively large 12'' radius circle. The background was from a 30'' circular, source-free region. Response matrices and effective areas were then determined for the target position and the spectrum was binned to have at least 20 counts per spectral bin. The resulting spectrum was analyzed using XSPEC (V12.6.0).

W2332−5056 appears spatially extended in the Chandra data, but the distorted off-axis PSF complicates the interpretation of this spatial structure. We used the CIAO task arestore to perform PSF deconvolution on the deeper image (ObsID 9333) with a PSF simulated by ChaRT in the energy range of 0.3–7 KeV. Two weak, extended X-ray regions ∼45 from the central X-ray peak at P.A.s of ∼30° and ∼200° are marginally detected; together they contain ∼0.6% of the X-ray photon counts.

Using the standard definition of hardness ratio, HR ≡ (H − S)/(H + S), where H is the hard counts between 2–10 keV and S is the soft counts between 0.5–2 keV, the derived HR = −0.05 is typical of extragalactic X-ray sources (e.g., Stern et al. 2002). Our best absorbed power-law fit ($\chi ^2_{{\rm red}}=1.09/70$ dof) to the spectrum has an observed photon index Γ = 1.51 ± 0.08, a total absorbing column of $N_{\rm H} = 5.31^{+0.92}_{-0.86}\, \times 10^{21}\, {\rm cm}^{-2}$, an observed soft X-ray flux of $f(0.5\hbox{--}2\, {\rm keV}) =2.32^{+0.34}_{-0.32}\, \times 10^{-13}\, {\rm erg}\, {\rm cm}^{-2}\,{\rm s}^{-1}$, and an observed hard X-ray flux of $f(2\hbox{--}10\, {\rm keV})= 1.01^{+0.20}_{-0.17}\, \times 10^{-12}\, {\rm erg}\, {\rm cm}^{-2}\,{\rm s}^{-1}$. The corresponding intrinsic, or unobscured, X-ray luminosities are $L(0.5\hbox{--}2\, {\rm keV}) = 1.72^{+0.06}_{-0.06}\, \times 10^{44}\,{\rm erg}\, {\rm s}^{-1}$ and $L(2\hbox{--}10\, {\rm keV}) = 4.14^{+0.32}_{-0.28}\,\times 10^{44}\, {\rm erg}\, {\rm s}^{-1}$.

W2332−5056 was observed in Spitzer [3.6] in 2009 and in WISE [3.4] between 2010–2011. The median of the [3.4] single-frame fluxes is 1.30 ± 0.05 mJy while the [3.6] flux from a single measurement is 1.311 ± 0.002 mJy; the anticipated color difference between IRAC [3.6] − WISE [3.4] is less than 0.1 mag.¹⁴ No significant mid-IR variation is seen.

At optical wavelengths, the redshifted Hβ spectra were obtained twice in 2011, separated by six months. There is no obvious variation in Hβ between these two epochs. On longer time spans (>3 yr), the central AGN does show significant fluctuations at optical wavelengths. The residual in the differential r'-band imaging (Figure 6) indicates a ∼20% total flux decrease from 2008 November to 2011 November.

Zoom In Zoom Out Reset image size

W2332−5056 is clearly resolved in the SOAR r'-band image (see Figures 1(e) and 7; 065 seeing). We use GALFIT 3.0 (Peng et al. 2002, 2010) to examine its morphology. Inspired by evidence of at least one BL AGN in the system, we use a combination of Sérsic profiles and PSFs to model the host galaxy and unresolved AGN components, respectively. The results are shown in the right panels of Figure 7. The single Sérsic model, single PSF model, and dual PSF models cannot explain the observed morphology; all of them leave significant residual flux (>3%) within a 48 radius of the core. Modeling the image with a Sérsic profile plus a single free PSF, the PSF preferentially selects an off-center optical bright spot, but leaves significant central residual flux (∼7%). The best-fit model ($\chi _{\rm red}^{2} \sim 1.4$, dof = 40388) requires a single Sérsic profile accompanied with two PSF profiles that match the observed brightness distribution to <0.5%. In this model, the Sérsic component, primary PSF component, and secondary PSF components contribute 33.7%, 61.8%, and 4.5% of the total flux, respectively.

Zoom In Zoom Out Reset image size

The Sérsic profile (of the host galaxy) in the best-fitted model has an axis ratio of 0.64 at a P.A. of 48°. It contributes ∼50 μJy (or AB = 19.64) with an effective radius of 136 (6.7 kpc) and has a Sérsic index of 0.92, close to an exponential disk index n = 1. This component has $M_{\rm r^{\prime }} = -21.6$ (AB mag) before k corrections, slightly brighter than the Milky Way. The luminosity is 2.5 × 10¹⁰ L_☉, similar to our spectral energy distribution (SED) model of the host galaxy (blue line in Figure 8). The primary PSF component is centered 016 offset (∼800 pc in projection) from the Sérsic center. This displacement, 25% of the PSF FWHM, appears significant, but could be the result of an imperfect PSF generated from unresolved sources in the field.

Zoom In Zoom Out Reset image size

The secondary unresolved component is 10 (4.9 kpc in projection) offset from the Sérsic center. This component does not show obvious variability. It is too bright ($M_{\rm r^{\prime }} = -19.4$) to be a star cluster inside the host galaxy or a satellite dwarf elliptical galaxy, although it could be a second merging, compact galaxy. It is not likely to be a stellar object unless it is a foreground star along the line of sight. The position angle to the primary compact object is 49°, or ∼30° with respect to the secondary radio jet, making it unlikely to be a bright spot associated with an optical jet of the central AGN. A better understanding of this component will require sensitive, higher-resolution imaging.

The brightest visual companion at slightly greater distance (r > 4''; ∼20 kpc) has $m_{\rm r^{\prime }}$ = 21.6 (AB mag), which corresponds to ∼6 × 10⁹ L_☉ in the r'-band after k-corrections assuming the redshift of W2332−5056. There are no obvious disturbed morphological features brighter than 1/5 of the luminosity of W2332−5056 to suggest that the host galaxy is an early/intermediate-stage major merger system.

We model the broadband SED of W2332−5056 from the ultraviolet (UV) to the mid-IR with a combination of host galaxy and AGN components using the empirical SED templates of Assef et al. (2010). One challenge is that the source is clearly variable at optical wavelengths (see Section 3.4 and Figure 6). Indeed, the red optical SED observed by Magellan in 2008 is difficult to reconcile with the very blue z' − J color implied by the 2MASS J-band photometry obtained in 1999 unless we invoke variability. We therefore divide the photometric data into three epochs (1991.7–1992.9, 1999.9, and 2008.9–2010.5, in roughly 3 yr time bins) for SED fitting purposes (Figure 8).

The multi-epoch SED fitting was done requiring a common, non-variable host galaxy using a combination of empirical starburst, disk galaxy, and elliptical galaxy templates from Assef et al. (2010) to represent the different age stellar populations in W2332−5056. We required these SED galaxy components to match the morphological modeling discussed in Section 3.5; namely, we required that the r' fluxes from the three galaxy components sum to one third of the total r' flux, or ∼60 μJy. For the AGN component, we adopted a modified AGN SED template in which the warm dust emission at λ > 1 μm is replaced by power-law emission to continue the accretion disk emission to longer wavelengths. This modification was made in order to satisfy a reasonable fit between the r'-band and W2. We also apply foreground dust extinction toward the AGN to accommodate the red optical color. This modified obscured AGN component is reminiscent of the synchrotron emission that characterizes blazars at these wavelengths (Abdo et al. 2010); the WISE colors of W2332−5056 put it close to the blazar strip of Massaro et al. (2012). Blazars, conventionally considered to be the radio galaxies with relativistic jets close to the line of sight, exhibit large amplitude luminosity variations over hours to days (see the review by Ulrich et al. 1997). However, the interpretation of W2332−5056 as a blazar is inconsistent with the radio morphology, nor do we see strong evidence for short-time scale variability at optical and mid-IR wavelengths. We do not address this blazar interpretation further here.

The results of the model fitting are presented in Figure 8. The GALEX observations, obtained between 2003 and 2005, would appear to present a challenge. On the one hand, they could be associated with a variable AGN component observed during an epoch when we do not have other photometry available to use in the SED modeling. On the other hand, they could be associated with the non-variable host galaxy or a combination of the host and variable AGN. In practice, our results do not depend strongly on the GALEX data. If we completely ignore those UV data, our best-fit model implies a heavily obscured AGN during the latest epoch with E(B − V) = 0.81 ± 0.05. For this model, the host galaxy, dominated by the starburst (e.g., irregular galaxy) component, goes through the GALEX photometric points with the best-fit luminosity to the irregular component being L_im = (2.6 ± 1.5) × 10¹⁰ L_☉. Additionally, the $\chi _{{\rm red}}^{2}$ increases from 1.65 (dof = 6) to 3.26 (dof = 7) if we exclude the irregular component in the SED fitting. These suggest that the GALEX photometry corresponds to the host galaxy with little AGN contribution. If we redo the fitting under that assumption, the resulting AGN obscuration basically remains the same, E(B − V) = 0.82 ± 0.04, and the starburst component luminosity becomes slightly more robust, but is essentially unchanged: L_im = (3.2 ± 1.0) × 10¹⁰ L_☉. It is this latter case where we assume that the decade-old GALEX photometry is due to the host galaxy, which we show in Figure 8.

An obscured AGN dominates the optical and WISE photometry for the most recent epoch, while 10 yr ago, when the 2MASS photometry was obtained, our modeling is based on fewer bands but suggests a slightly less luminous AGN with a similar level of obscuration, E(B − V) = 0.8. The AGN reddening in these epochs is slightly higher than the E(B − V) = 0.4 ± 0.1 implied from the X-ray column, N_H ∼ 2.3 ± 0.5 × 10²¹ cm⁻² (Section 3.3), if the conventional Milky Way value of N_H/E(B − V) ∼ 5.9 × 10²¹ cm⁻² (Bohlin et al. 1978; Rachford et al. 2002) applies. Finally, the same fitting was applied to the Digitized Sky Survey (DSS) photometry for the epoch 1991.7–1992.9. Remarkably, the best-fit model shows that no AGN emission component is necessary and is relatively well modeled by the host galaxy component alone. The DSS B-band, taken in 1975, is not included in the fitting due to its early observational epoch, but appears not to affect the fitting results if included. The best-fit AGN amplitude is nominally zero, suggesting that only ∼20 yr in the past W2332−5056 did not have appreciable nuclear activity. We note that, although the best-fit model has relatively larger uncertainty due to the quality of the DSS photometry, the strong variation in the optical band over two decades is significant. The interpolation of our high-quality optical photometry measurements in the current epoch (2008–2010) suggests a flux of ∼300 μJy in the DSS I-band measurement, >10σ higher than the measurement of 80 ± 20 μJy in the 1991–1992 epoch.

FR-II radio sources are almost invariably found to be associated with large elliptical galaxies (Matthews et al. 1964; Owen & Laing 1989). Such hosts typically have de Vaucouleurs' r^1/4 profiles (Owen & White 1991), corresponding to a Sérsic index of 4. However, a class of radio sources hosted in star-forming galaxies has been identified by Norris et al. (2006) and later named Powerful Radio Objects Nested in Galaxies with Star Formation (PRONGS; Mao et al. 2010; Norris et al. 2012). These objects are radio-loud AGNs (≳ 10²⁴ W Hz⁻¹) with radio lobes <200 kpc that are hosted by spiral or starburst galaxies. Mao et al. (2010) discovered 50 PRONGS in a systematic search of a 7 deg² field in the Australia Telescope Large Area Survey (Norris et al. 2006; Middelberg et al. 2008). Prior to this work, two radio-loud, edge-on disk galaxies (NGC 612 and 0313-192) were known in the local universe (Ekers et al. 1978; Ledlow et al. 2001; Keel et al. 2006; Emonts et al. 2008). NGC 612 has a collimated FR-I-like jet and an FR-II-like lobe with a bright hot spot on the other end, sometimes described as a hybrid-morphology radio source (HYMORS; Gopal-Krishna & Wiita 2000), while 0313-192 has a typical FR-I morphology. There are a few other similar nearby examples where radio-loud objects are hosted in disk-like galaxies such as 3C305 (Heckman et al. 1982), 3C293 (van Breugel et al. 1984), B2 0722+30 (Emonts et al. 2009), PKS 1814-636 (Morganti et al. 2011), and SDSS J1409−0302 (Hota et al. 2011). They could be associated with PRONGS but perhaps at a slightly different evolutionary phase. Previous work has suggested that PRONGS may represent an early evolutionary stage for radio-loud galaxies. An example is given by Norris et al. (2012), who present a very long baseline interferometry (VLBI) image of a highly embedded, compact core–jet AGN with 1.7 kpc long jets in IRAS F00183-7111, a ULIRG with L_bol ∼ 9 × 10¹² L_☉ and substantial star formation activity.

The optical morphology (Section 3.5) suggests that the host galaxy of W2332−5056 has a modest disk (Sérsic index ∼0.9), not much bigger than the Milky Way in size and luminosity. With a 5 GHz to 4400 Å continuum flux density ratio ≫10, the W2332−5056 system is clearly hosting a radio-loud AGN according to the definition of Kellermann et al. (1989). Although it hosts a HYMORS-like source, the galaxy of the W2332−5056 system seems to be a normal disk galaxy, making it a candidate for the rare PRONGS classification.

Based on its rest-frame near-UV luminosity νL_ν ∼ 1.1 × 10¹⁰ L_☉ from the host galaxy SED, we estimate a star formation rate of 5 M_☉ yr⁻¹ in W2332−5056, using Equation (1) of Kennicutt (1998) with no significant dust attenuation assumed. The total luminosity of W2332−5056 up to 22 μm is 3.7 × 10¹¹ L_☉. The non-detection by AKARI, IRAS, and the SPT limit the far-IR (FIR) luminosity (L_FIR) to 3.5 × 10¹² L_☉ if an Arp 220-like FIR SED is assumed. However, this Arp 220-like SED model does not match the observed optical-to-near-IR photometry of the system. If a fraction of the star formation activity is highly obscured at optical wavelengths (A_V > 40) but releasing energy via dust emission, the 22 μm (W4) flux density limits L_FIR to <3.5 × 10¹¹ L_☉ or <30 M_☉ yr⁻¹ (i.e., 10% that of Arp 220; Anantharamaiah et al. 2000). Thus, the host galaxy of W2332−5056 is not likely to be undergoing a major starburst. This star formation rate is close to a typical quiescent disk galaxy, unlike the hosts of other PRONGS.