Optical, Near-IR, and Sub-mm IFU Observations of the Nearby Dual Active Galactic Nuclei MRK 463

MUSE is a state-of-the-art, second-generation VLT IFU instrument (Bacon et al. 2010). Its relatively large field of view, ∼1 arcmin², and broad wavelength coverage, ∼4800–9300 Å, are particularly well-suited to observing the entire physical extents of low-redshift systems (e.g., physical sizes of tens of kpc in a single pointing) across a large spectral range (simultaneous coverage of the H_β, H_α and [O iii] lines, among others). The resolving power of MUSE ranges from 1770 at 4800 Å to 3590 at 9300 Å, which corresponds to velocity resolutions of ∼80 to ∼170 km s⁻¹.

Mrk 463 was observed by the VLT/MUSE as part of program 095.B-0482 (PI: E. Treister). Four other nearby dual AGN were also observed as part of this program and will be presented on subsequent papers. Mrk 463 was observed on 2015 July 18 in two 58 min observation blocks (OBs), IDs 1182834 and 1182837, each comprised of three on-target integrations of 976 s plus overheads. Sky conditions were clear, with <50% lunar illumination, and a reported ambient seeing <08. The final combined, processed, and calibrated VLT/MUSE data cube for Mrk 463 has an effective exposure time of 5856 seconds (∼1.6 hours), an average airmass of ∼1.5, and an average FWHM of ∼07 based on measurements of stars in the field on the reconstructed white filter image.

Data reduction was carried out using the ESO MUSE pipeline version 1.6.2²⁰ (Weilbacher et al. 2014) in the ESO Reflex graphical environment (Freudling et al. 2013). The data reduction stages are standard and include bias and dark current subtraction, flat fielding, wavelength and flux calibration, and astrometric correction. Sky background was subtracted using emission-free regions of the resulting data cube. However, the process is complicated by the large data volume. The MUSE 1' × 1' field of view has 300 × 300 spatial bins (spaxels) and 4000 spectral bins, such that an individual MUSE frame is a few GB in size and data reduction requires several GB for intermediate products and multiple CPU cores to run. First-look visualizations of the resulting combined data cube were done using the QFitsview²¹ software, but not used in our subsequent work. Further analysis, including emission line fitting, velocity maps, etc. of well-isolated narrow emission lines was carried out using IDL Fluxer²² tool version 2.7. More complex line fits, including several overlapping components were carried out using the Pyspeckit Python package (Ginsburg & Mirocha 2011), as Fluxer only allowed a maximum of two simultaneous Gaussian fits.

We obtained near-IR IFU observations of Mrk 463 using the Spectrograph for INtegral Field Observations in the Near Infrared (SINFONI; Eisenhauer et al. 2003) camera mounted at the VLT, as part of program 093.B-0513 (PI: S. Cales). This program aimed to observe four confirmed z < 0.1 dual AGN that are visible from the southern hemisphere, one of which was Mrk 463. SINFONI has a much smaller field of view, 8'' × 8'', and hence these observations only cover the central region around the dual nuclei.

The observations were carried out using the 125 × 250 mas spatial scale in seeing-limited mode (i.e., no adaptive optics), together with the K-band grating, which provides coverage from 1.95 to 2.45 μm and a spectral resolution of ∼4000. The observations were carried out in service mode and spread over four OBs: IDs 1053697 (2014 April 17), 1053700 (2014 May 17), 1053702 (2014 June 18), and 1053704 (2014 July 21). The requested weather conditions were: ambient seeing better than 08, clear skies and airmass <1.5. However, only OBs 1053697 and 1053700 met such conditions; the remaining ones were taken either in a worse seeing and/or in the presence of thin clouds. Therefore, for this analysis, we mostly focus on the first two OBs in order to achieve the highest possible image quality.

The reduction of the SINFONI data was carried out using the dedicated ESO pipeline version 2.9.0 (Modigliani et al. 2007) in Reflex. The following steps were performed: (i) creation of a map of non-linear pixels from flat-field frames obtained with increasing exposure times; (ii) creation of a combined dark frame and hot pixel map; (iii) creation of a master flat field from individual flat-field images; (iv) determination of the optical distortion coefficients and computation of the slitlets' distance table; (v) creation of a wavelength map from a set of arc lamp observations in order to carry out the wavelength calibration; (vi) stacking of individual science and telluric star frames. The flux calibration of the science frames was performed using the Fitting Utility for SINFONI (FUS) package developed by Dr. Krispian Lowe as part of his PhD thesis,²³ based on observations of standard stars taken less than two hours away and with a difference in airmass <0.2 from the science data. As in the case of the VLT/MUSE data, the final combined data cubes were analyzed using the QFitsview tool for a first-look visualization and Fluxer for the Gaussian fitting of each isolated emission lines.

During ALMA Cycle 2, we were granted 5.1 hours of priority B time to observe a sample of four nearby dual AGN, including Mrk 463, in band 6 as part of program 2013.1.00525.S (PI: E. Treister). Four separate spectral windows were defined: one covering the ¹²CO(2–1) molecular transition, providing a 10.7 km s⁻¹ spectral resolution and a 2560 km s⁻¹ bandwidth, and the remaining three covering the mm continuum at ∼220–235 GHz. The observations of Mrk 463 were taken on 2015 August 16 between 22:27 and 23:27 UT (uid://A002/Xa830fc/X4abc), and on 2015 September 16 between 21:01 and 22:01 UT (uid://A002/Xaa305c/X129). For the August 16th observations, 34 antennae were used, with baselines ranging from 43 m to 1.57 km, while on September 16th the same number of antennae were used, distributed between 41 m and 4.5 km baselines. The total on-target time in each observation was 25.7 minutes. Data reduction and analysis were carried out using the Common Astronomy Software Applications (CASA) version 4.5.0.

The requested 3σ sensitivity of the observations was 0.58 mJy beam⁻¹ per ∼10.7 km s⁻¹ channel. This requirement was not quite met, as the early processing carried out by the North America ALMA Science Center yielded a sensitivity of 0.78 mJy beam⁻¹ on the ¹²CO(2–1) spectral window, 34% higher than the goal value. This value was obtained using a Briggs Robust image weighting (Briggs 1995) with a robust parameter of 0.5, which provides a balance between a smaller beam and lower sensitivity. Instead, we decided to re-process the ALMA data using a natural weighting scheme that delivers a higher sensitivity at the expense of a slightly worse spatial resolution. Using this method, we achieved a sensitivity of ∼0.6 mJy beam⁻¹ at 3-σ in the ¹²CO(2–1) line, thus coming much closer to our goal flux limit. The resulting beam size is 03 × 017.

In order to be able to compare emission at different wavelengths from the instruments described above, it is critical to ensure that all the data sets share the same astrometric reference point. This is potentially difficult, given the relatively small overlap in structures at different wavelengths. The ALMA absolute astrometric calibration is better than ∼01, as reported by González-López et al. (2017).

Using the Chandra X-ray data for this system, Bianchi et al. (2008) located the position of the two AGN to within 05. In particular, we derived the exact position using the hard X-ray band, 2–8 keV, because most of the emission can be attributed to the AGN in this case.

In order to obtain line flux maps from the MUSE data cubes, we perform Gaussian fits to the He ii 4685 Å, Hβ 4861 Å, [O iii] 5007 Å, Hα 6563 Å, [N ii] 6548,6583 Å, and [S ii] 6717,6731 Å features. The line fitting procedure is as follows. First, the continuum emission in each pixel is subtracted by performing a fit using a second-order polynomial function between 5000 Å and 8000 Å. Using this broad wavelength range and a relatively small polynomial order allows to obtain a good fit for the spectral continuum in the region of interest while remaining unaffected by the presence of even prominent emission or absorption features. We then performed a Voronoi tesselation of the cube by demanding a minimal signal-to-noise ratio of 10 in the Hα/[N ii] complex in each bin. While this choice has no impact on the central regions of the system, where this condition is met basically on every pixel, it is important in the outskirts. We then assume Gaussian profiles for all emission lines. Given that it is a strong and well-isolated feature, we use the [O iii] 5007 Å line as a template in order to define the widths and relative wavelength offsets for all the subsequent narrow components. We further allow for a secondary line component at a different velocity, also using the [O iii] 5007 Å line as template and adding extra Gaussians with both the line centers and widths as free parameters to account for the possible presence of broad Hα and Hβ emission. In each case, a component is only considered if the peak is detected at a signal-to-noise greater than three.

The resulting images are presented in Figures 2 and 3. In all of these cases, the total flux for all the line components is shown. Some of the same key features seen in Figure 1 and discussed in the previous section are clearly visible in these images as well. However, other remarkable components are now detected as well. Particularly important are the detections of extended Hα gas to the northwest of the system and the "stream" of Hα blobs to the southwest, seen in Figure 3. Additionally, we can identify an absorption region to the west of the Mrk 463W nucleus in the Hβ 4861 Å map.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The flux ratios of these emission lines can be used in order to determine the nature of the ionization source (e.g., Baldwin et al. 1981; Veilleux & Osterbrock 1987; Kewley et al. 2006). Figures 4 and 5 show the maps for the [O iii]/Hβ, [S ii]/Hα, and [N ii]/Hα ratios, computed using the sum of the flux in the narrow components for each emission line. According to Kewley et al. (2006), any source/region with [O iii]/Hβ > 10 is nearly guaranteed to be dominated by AGN ionization, independent of the value of [N ii]/Hα or [S ii]/Hα. While shocks appear to be widespread in major galaxy mergers, they typically have lower [O iii]/Hβ ratios, ∼1–5, compared to active nuclei (Rich et al. 2011, 2015). Hence, just from examination of Figure 4, we can see several potential AGN-dominated regions. In order to investigate the nature of the ionization energy source(s), we selected five outstanding zones in the system, highlighted in Figures 4 and 5: the two nuclear regions, the southern emission line region, the northern clump, and a representative Hα emitting region in the northwest. The locations of these five regions on the [O iii]/Hβ versus [N ii]/Hα and [S ii]/Hα diagnostic diagrams are shown in Figure 6. As can be clearly seen, both nuclei, the southern emission line region, and the northern clump are consistent with being dominated by the AGN energy output. In contrast, region #5 appears to fall on the star-forming locus. These diagrams indicate that the influence of the AGN on the surrounding material is widespread and can be detected at large distances, even ∼10 kpc away from the nuclei.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Further hints about the nature of the ionization sources in this system can be obtained by exploring emission lines with higher ionization energies. To this end, Figure 7 shows the emission maps for the He ii 4685 Å and [S ii] 6717/31 Å lines, which have ionization energies of 54.4 eV and 23.3 eV, respectively. As can be seen in the figure, highly ionized lines, particularly He ii, are much more concentrated, mostly on the Mrk 463E nucleus and the emission line regions, while Mrk 463W is barely detected. As presented by Shirazi & Brinchmann (2012) and later by Bär et al. (2017), the He ii 4685 Å emission and He ii/Hβ ratio can be used to identify AGN even in sources missed by classification methods based on emission lines at lower ionization energies (Sartori et al. 2015). According to Shirazi & Brinchmann (2012), sources or regions with log(He ii/Hβ) > −1 can be classified as AGN dominated. This is the case for the Mrk 463E nucleus and its surrounding ionization cone, for the southwest emission line region but not for the Mrk 463W nucleus. This suggests that the AGN in Mrk 463E is energetically more important for the system than the one in Mrk 463W.

Zoom In Zoom Out Reset image size

Although the VLT/SINFONI near-IR IFU data only cover a smaller ∼8'' × 8'' region at the center of the Mrk 463 system, as shown in Figure 1, they are an important complement to the MUSE optical maps. Figure 8 shows the near-IR continuum between 2.1 and 2.4 μm and the continuum-subtracted integrated Paα hydrogen emission line at 1.8751 μm in the rest frame. For this and all other near-IR emission lines, the continuum subtraction was carried out by performing a simple first-order polynomial fit to the adjacent spectral regions on each side of the line. The line fitting was carried out assuming a single Gaussian functional form, which was visually deemed appropriate and reasonable, considering that the SINFONI data only cover the central region of the system. No cut in signal to noise ratio of the resulting line was used in constructing the map. Both nuclei are clearly detected on the continuum map, showing a very similar morphology to the optical continuum image presented in Figure 1. The Paα map, in contrast, reveals new components. In particular, we can marginally detect an emission blob between the two nuclei and another emitting region to the west of the Mrk 463W nucleus. While both of these regions are also visible in the Hα map, they are not detected in [O iii], suggesting that they are star-forming regions. However, as will be later shown in Section 5.1, the region between the nuclei is subject to moderate extinction, A_Hα ∼ 2, and therefore it is possible that the lack of detection of [O iii] in this area is due to obscuration.

Zoom In Zoom Out Reset image size

The Paα emission line provides a measurement of the star formation rate. Using the SINFONI cube, together with the conversion from Paα luminosity to star formation rate provided by Rieke et al. (2008), we derive values of 26.5 M_⊙ yr⁻¹ for the East nucleus and 0.75 M_⊙ yr⁻¹ for the West one. These star formation rates derived from the Paα line are fully consistent with those of 30 M_⊙ yr⁻¹ for Mrk 463E and <10 M_⊙ yr⁻¹ for Mrk 463W reported by Evans et al. (2002). However, it is important to point out that, as shown in Figure 6, both nuclei appear to be dominated by the AGN emission, at least for the optical lines, and thus these SFR estimates can be considered as upper limits.

The near-IR data also map other emission lines such as [Si vi] and Hydrogen Brγ at rest-frame wavelengths of 1.962 and 2.1655 μm, respectively; these are both presented in Figure 9. Emission from [Si vi] is detected at high significance from Mrk 463E and weakly from Mrk 463W, thus further confirming the AGN nature of both nuclei. However, in contrast, Brγ is only weakly detected from Mrk 463E.

Zoom In Zoom Out Reset image size

With our ALMA Cycle 2 data, we detect both ¹²CO(2–1) and the mm continuum in the central region of the Mrk 463 system. Figure 10 shows the velocity-integrated continuum-subtracted ¹²CO(2–1) emission map, together with contours for the rest-frame 232 GHz continuum. As can be seen, the strongest CO(2–1) emission is found in a region to the northeast of the Mrk 463E nucleus, although significant emission is also located around each nucleus, as well as between them. Interestingly, only a small fraction of the CO(2–1) emission overlaps with the AGN location.

Zoom In Zoom Out Reset image size

Defining a ∼4'' × 3'' region centered on the Mrk 463E nucleus that encloses the majority of the flux, we obtain a total flux density of 11.4 ± 0.12 Jy km s⁻¹. In this case, the error bars were estimated by measuring the rms in nearby source-free regions of the same area. Previous observations of this source at lower resolution, ∼2'', yielded total unresolved fluxes of 7.2 ± 0.8 Jy km s⁻¹ (Alloin et al. 1992) and 6.8 ± 0.9 Jy km s⁻¹ (Evans et al. 2002) for the ¹²CO(1–0) transition. Following Solomon & Vanden Bout (2005), we assume that the CO emission is thermalized and optically thick, such that the intrinsic line luminosity is independent of J, and hence we expect the ¹²CO(2–1) flux density to be four times higher than ¹²CO(1–0). Therefore, our high-resolution ALMA observations resolve ∼50% of the CO emission for the Mrk 463E nucleus extended over scales of ∼4''. Using a region of the same area centered on the Mrk 463W nucleus, we derive a line flux density of 4.5 ± 0.12 Jy km s⁻¹. This is consistent with the estimates of Evans et al. (2002), who also found a flux ratio of ∼3:1 between the Mrk 463E and Mrk 463W nuclei, thus suggesting that the extended component encompasses both nuclei—as expected, given the observed nuclear separation.

Following the prescription described in Sections 2.1 and 2.2 of Solomon & Vanden Bout (2005), particularly their Equation (3):

$$\begin{eqnarray*}&&{L}_{\mathrm{CO}}^{{\prime} }=3.25\times {10}^{7}{S}_{\mathrm{CO}}{\rm{\Delta }}v{\nu }_{\mathrm{obs}}^{-2}{D}_{L}^{2}{(1+z)}^{3},\end{eqnarray*}$$

we calculate a CO(2–1) luminosity for the emission surrounding the Mrk 463E nucleus of ${L}_{\mathrm{CO}}^{{\prime} }$(2–1) = 3.173 × 10⁸ K km s⁻¹ pc². Next, using the relation M_gas = M(H₂) = α ${L}_{\mathrm{CO}}^{{\prime} }$, and α = 4.3 M_☉ (K km s⁻¹ pc²)⁻¹ (Bolatto et al. 2013), we obtain a total gas mass in this region of 1.36 × 10⁹ M_☉. Similarly, for the material surrounding the Mrk 463W nucleus, we obtain ${L}_{\mathrm{CO}}^{{\prime} }$(2–1) = 1.253 × 10⁸ K km s ⁻¹ pc² and M_gas = 5.39 × 10⁸ M_☉. These differ from the gas masses derived by Evans et al. (2002), primarily due to the different values of α assumed. Indeed, Evans et al. (2002) used a value of α = 1.5 M_☉ (K km s⁻¹ pc²)⁻¹ for this source and reported a total molecular gas mass of 10⁹ M_☉, whereas we obtain ∼7 × 10⁸ M_☉ while assuming their α value.

Using the VLT/MUSE data, we analyze the kinematics from two strong and well-isolated emission lines: Hβ 4861 Å and [O iii] 5007 Å. Figure 11 shows the position–velocity (p–v) diagram centered on the Mrk463-E nucleus with position angles (PAs) of 0° (i.e., north–south) and 75° using the [O iii] 5007 Å line. In this case, we assume a reference velocity of 15226 km s⁻¹, which is 130 km s⁻¹ higher than the systemic heliocentric velocity of 15096 km s⁻¹ that corresponds to the redshift of 0.050355 reported by Falco et al. (1999), which we used for the rest of the analysis. This was done to match the center of the potential additional stellar disk to the west of Mrk 463W discussed in Section 4.2. A clear velocity gradient can be seen on the north–south slit, as discussed below. This structure is very spatially localized, as it is not present on the PA = 75° diagram. The southern emission line is visible on the PA = 0° slit region at ∼10''. As can be seen, this region is roughly at the systemic velocity. Similarly, part of the emission associated with the Mrk463W nucleus can be seen on the PA = 75° diagram, at ∼4'' away from the eastern one.

Zoom In Zoom Out Reset image size

Figure 12 shows the velocity and velocity dispersion maps for the [O iii] 5007 Å emission line. The velocity gradient associated with the eastern nucleus that was visible on the p–v diagram shown in Figure 11 is present here as well. As first suggested by Hutchings & Neff (1989) and later discussed by Chatzichristou & Vanderriest (1995), this structure is consistent with the presence of a two-sided ionization cone, which also aligns with the extended radio emission detected from this nucleus. While a detailed modeling of this biconical outflow is beyond the scope of this paper, we can use the observed emission line features in this region to derive physical properties of the ionized gas outflow. As presented by Müller-Sánchez et al. (2011), the mass outflow rate is given by

$$\begin{eqnarray*}&&{\dot{M}}_{{\rm{out}}}=2{m}_{p}{n}_{e}{v}_{\max }{Af},\end{eqnarray*}$$

where m_p is the proton mass, n_e is the density of the ionized gas, A is the lateral surface area of one cone of the outflowing region, and f is the filling factor of the ionized gas. The factor of two accounts for the two sides of the ionized cone. The value of n_e can be estimated from the observed ratio of the [S ii] 6716/6731 Å emission lines, as described by Osterbrock & Ferland (2006). We find that, on average, the [S ii] ratio in the region of the biconical outflow has a value ∼0.7. According to the prescription of Osterbrock & Ferland (2006) and assuming a temperature of ∼10,000 K, we estimate that n_e should be in the range from 1000 to 5000 cm⁻³ and adopt a value of 3000 cm⁻³. For the filling factor, f, which cannot be derived directly from observations, a range of 0.01 < f < 0.1 is commonly assumed (e.g., Storchi-Bergmann et al. 2010). However, our assumed n_e value is much higher than in those cases, and considering the n_e ∝ f^−0.5 relation proposed by Oliva (1997), we assume a value of f = 0.001, which is 10 times larger than the value used by Nevin et al. (2018). We consider a maximal velocity of v_max = 350 km s⁻¹, as measured on the [O iii] velocity map. The value of A is obtained considering that the physical distance from the nucleus to the location of the maximal velocity is ≃3.3 kpc and that the cone radius at that position is ≃0.9 kpc, thus yielding A = 9.62 × 10⁶ pc². Next, using the expression described above, we obtain that ${\dot{M}}_{\mathrm{out}}=512(f/0.001)$ M_⊙ yr⁻¹. This value is on the high side of the range of mass outflow rates found previously in systems hosting biconical outflows (Müller-Sánchez et al. 2011, 2016; Nevin et al. 2018) in moderate-luminosity AGN and similar to those found in high redshift quasars (Brusa et al. 2015) and ULIRGs (e.g., Veilleux et al. 2005).

Zoom In Zoom Out Reset image size

In addition, we also detect a rapidly outflowing region to the northwest of the Mrk 463E nucleus, reaching speeds of up to ∼−600 km s⁻¹. This region was already identified as #4 in Figure 4. As shown in Figure 6, this region has the highest value of the [O iii]/Hβ ratio and is clearly located in the AGN region, which confirms that this outflowing material is being ionized by the AGN, most likely in Mrk 463E, given its spatial location and the luminosity of its nucleus. Furthermore, this outflowing region appears to be spatially connected to the biconical outflow. The velocity dispersion map for the [O iii] 5007 Å line shows a broad emission line component in the central region of Mrk 463E. Both the north and south edges of the biconical outflows also show relatively high-velocity dispersions, reaching up to ∼600 km s⁻¹, suggesting the presence of shocks in the boundary of the outflowing regions. The remaining regions are characterized by relatively low-velocity dispersions, up to ∼200 km s⁻¹, close to the spectral resolution of VLT/MUSE.

The velocity and velocity dispersion maps, as measured from the Hβ 4861 Å emission line, are shown in Figure 13. As for the [O iii] 5007 Å line, a very clear gradient, ranging from −350 km s⁻¹ in the north to ∼150 km s⁻¹ in the south, centered on the Mrk 463E nucleus, is clearly visible. This gradient is also associated with the biconical outflow presented above. This map does not include the absorption line region located to the west of the Mrk463W nucleus, which will be studied in detail in the following subsection. Both nuclei appear to be consistent with the systemic velocity, or with very small offsets. Similarly, the southern emission line region is at the systemic velocity and has no discernible structure. However, there is a clear tail of atomic gas emission traced by the Hβ 4861 Å transition, which includes the southern emission line region visible in the [O iii] map, presenting a velocity gradient ranging from ∼100 km s⁻¹ to >200 km s⁻¹. This appears to reflect the velocity gradient along a tidal tail from the ongoing dynamical interaction and is consistent with the velocity gradient seen in HI observations of other systems (e.g., Hibbard & van Gorkom 1996).

Zoom In Zoom Out Reset image size

The velocity dispersion profile for the Hβ 4861 Å line reveals the presence of a relatively broad component with an FWHM of up to ∼700  km s⁻¹ in the nuclear region Mrk 463E, consistent with the location of the hard X-ray emission reported by Bianchi et al. (2008). The arc-like structure to the northeast of the Mrk 463E, which also shows high-velocity dispersion values, is explained by the presence of two superimposing velocity components, one associated with the edge of the biconical outflow reaching velocities of ∼400 km s⁻¹ and the underlying galaxy at the systemic velocity. The rest of the system presents only narrow components with a width of ∼200 km s⁻¹ or less, consistent with being unresolved at the MUSE spectral resolution.

We can further use the VLT/MUSE data cube to measure the kinematic properties of the stars in the Mrk 463 system. In order to achieve this, we performed a new Voronoi tessellation (Cappellari & Copin 2003), this time aimed to reach a minimum signal-to-noise ratio of 30 per bin in the optical continuum near the Mg b feature at ∼5200 Å in the rest frame. The resulting bin map is shown in Figure 14. Next, for each bin, we simultaneously masked the most prominent emission lines and fitted the optical continuum and absorption lines using the Penalized Pixel-Fitting (pPXF) method presented by Cappellari & Emsellem (2004). The optical spectrum at each Voronoi bin was fitted using stellar templates from the MILES library (Falcón-Barroso et al. 2011), which were then used to derive properties such as the stellar velocity and velocity dispersion, h₃ and h₄ Hermite polynomial coefficients, etc. A more detailed description of this procedure is given by Cappellari (2017). In Figure 14, we show the reduced χ² for each Voronoi bin. Fits were not obtained for the central regions of each galaxy, as the optical spectra in these are dominated by AGN light.

Zoom In Zoom Out Reset image size

In Figure 15, we present the stellar velocity and velocity dispersion maps created by performing this spectral fitting. While the stellar velocity map appears rather flat, some structures are visible. In particular, there is a clear region to the northwest of the Mrk 463W nucleus, presenting a relative line-of-sight velocity of ∼200  km s⁻¹. In addition, a negative velocity region can be observed to the south of the Mrk 463W nucleus. These features could potentially be related to the ongoing major merger, such as tidal plumes from a second (or later) close pass of the nuclei. Interestingly, this structure overlaps spatially with a region dominated by Hβ in absorption, as can be seen in Figure 2. This, together with the fact that this region does not have a counterpart on the emission line velocity maps presented in Figures 13 and 12, strongly suggests that this region is mostly devoid of ionized gas. While very faint, this structure can be seen as an extra component in the optical continuum, as shown in Figure 1.

Zoom In Zoom Out Reset image size

The stellar population fit simultaneously constrains the line-of-sight stellar velocity dispersion, which is presented on the right panel of Figure 15. As for the velocity map, the velocity dispersion in the Mrk 463 system appears to be rather flat and featureless. Again, the region to the northwest of the Mrk 463W nucleus presents an elevated value for the stellar velocity dispersion of ∼400 km s⁻¹, while the average for the system is ∼100 km s⁻¹. This broadening in the line-of-sight velocity dispersion may be evidence of multiple velocity components at this location, perhaps due to the superposition of a secondary tidal tail against a potential stellar disk.

In order to explore the existence of potential disk(s) revealed by kinematical structures, we attempt to fit the stellar kinematics with disk models. Specifically, we use the Keplerian disk models presented by Bertola et al. (1998). We use two disks, one centered on the Mrk 463E nucleus and the other at the center of the velocity gradient found to the west of Mrk 463W. Spatial profiles of the two models are presented in Figure 16. The models fit the data reasonably well, leaving only small velocity residuals. Hence, it is possible to explain the Mrk 463 system as a combination of a bulge and a faint extended disk associated with Mrk 463E, together with a bulge centered on Mrk 463W and a gas-stripped stellar disk to the west of Mrk 463W. There does not seem to be any connection between the ionized gas outflows discussed in Section 4.1 and the potential stellar disks presented here. As discussed in Section 4.1, the outflows in this system appear to be mostly located in the north–south direction and originate from the Mrk 463E nucleus. In contrast, the Mrk 463E stellar disk is oriented with a PA of 75°, i.e., mostly east–west. Similarly, the presumed stellar disk to the west of Mrk 463W does not appear to be connected in any obvious way to any ionized gas outflow.

Zoom In Zoom Out Reset image size

As described above, we would like to further emphasize that it is also possible to understand the observed structures in the stellar kinematics maps via the dynamical effects of the ongoing major galaxy mergers, which is also more natural because it does not require us to assume an ad hoc stellar disk offset by ∼3 kpc from the nucleus of the galaxy. Unfortunately, separating these two scenarios requires observations at higher spectral resolutions than those that can be provided by VLT/MUSE over relatively large angular scales necessary to cover the whole system.

Thanks to the ALMA data, we can analyze the molecular gas kinematics in the nuclear regions of the Mrk 463 system. Figure 17 shows the ¹²CO(2–1) velocity map. Interestingly, the molecular gas regions around each nucleus show very clear velocity gradients, with a range of ∼800 km s⁻¹ in the case of the east region and ∼400 km s⁻¹ next to the west region. Smaller gradients can be seen in a few other locations, including the one between the two nuclei.

Zoom In Zoom Out Reset image size

The eastern emission region presents an incomplete ring-like or infalling spiral structure, possibly centered on the AGN, based on the X-ray position. While there is relatively little CO(2–1) emission detected directly on top of the nucleus, we can use the gas mass and velocity in the immediate vicinity of Mrk 463E to obtain an estimate of the gas infall rate on scales of ∼300 pc around the SMBH. Considering only the material in the central ALMA beam, corresponding to ∼03 or 300 pc or a 150 pc radius, we obtain a total CO mass of ∼3.8 × 10⁷ M_☉. Now, assuming that this gas is falling at ∼200 km s⁻¹ (based on the CO(2–1) velocity measured on the nuclear ALMA beam) and that it is at an average distance of 75 pc, this yields an influx rate of ∼104 M_☉ yr⁻¹. While merely an approximation, this value is over three orders of magnitudes larger than the accretion rate of ∼0.017 M_☉ yr⁻¹ inferred from the X-ray luminosity of the Mrk 463E nucleus reported by Bianchi et al. (2008). This situation is very similar to what was found in the local AGN NGC1068 by Müller Sánchez et al. (2009) and later confirmed by García-Burillo et al. (2014) using high-resolution ALMA data, thus strongly suggesting that SMBH accretion is a highly inefficient process that requires the infalling material to lose most of its angular momentum.

Major galaxy mergers are known for potentially being subject to significant extinction, particularly at optical and UV wavelengths (e.g., Trentham et al. 1999; Bekki & Shioya 2000). Spectroscopically, the traditional way to estimate the amount of optical extinction is based on the observed ratio between the Hα and Hβ lines (e.g., Caplan & Deharveng 1986; Maíz-Apellániz et al. 2004). Taking advantage of the VLT/MUSE observations of the Mrk 463 system, we can then produce a A_Hα map. Following the prescription of Lee et al. (2009), we assume case B recombination with an intrinsic Hα/Hβ ratio of 2.86, a Milky Way extinction curve, and R_V = 3.1 such that A_Hα is given by:

$$\begin{eqnarray*}&&{A}_{{\rm{H}}\alpha }=5.91\mathrm{log}\left(\displaystyle \frac{{f}_{{\rm{H}}\alpha }}{{f}_{{\rm{H}}\beta }}\right)-2.70,\end{eqnarray*}$$

where f_Hα and f_Hβ are the total emission line fluxes on the Hα and Hβ lines derived following the procedure described in Section 3.1. The resulting map is shown in Figure 18. As can be seen, most of the system is only affected by relatively low extinction levels, A_Hα < 1. However, there are specific areas that appear subject to higher optical obscuration. These include some of the regions where we previously identified high levels of star formation to the east of the Mrk 463E nucleus, and the area between the two nuclei, where we can see values of A_Hα ∼ 1.5–2. While locally important, these relatively modest obscuration levels do not significantly affect our conclusions about the morphologies and kinematics derived from the optical VLT/MUSE IFU maps presented here, and they are even less important for the near-IR analysis shown in Section 3.2.

Zoom In Zoom Out Reset image size

As described in Section 3.1, we identify an emission line region to the south of the system, consistent with AGN powering. This region closely resembles the "Hanny's Voorwerp" structure associated with IC 2497 (Lintott et al. 2009), which was found serendipitously by visual inspection of the galaxy optical images carried out as part of the Galaxy Zoo project (Lintott et al. 2008). Hanny's Voorwerp appears to be ionized by the nearby AGN in IC 2497, which must have been at least 100 times more luminous ∼100,000 years earlier (Schawinski et al. 2010; Keel et al. 2012b; Sartori et al. 2016). For Mrk 463, the southern region is ∼11 kpc away from the Mrk 463E nucleus, which corresponds to ∼40,000 light years. Based on the evident alignment between this emission line region and the ionization cone arising from the Mrk 463E core, we can safely neglect the contribution from the Mrk 463W AGN. In turn, we can estimate the luminosity necessary to supply the ionizing flux required to explain the observed spectrum of the southern region.

In order to carry out this computation, we selected a region in the center of the emission region enclosing most of its flux. For constraints, we used the observed luminosities and their errors in the Hα, Hβ, [He ii] 4686 Å, [O iii] 5007 Å, [N ii] 6583 Å, [S ii] 6717 Å and [S ii] 6731 Å lines, all measured from the VLT/MUSE optical IFU observations. We then performed radiative transfer simulations using the plasma ionization code CLOUDY version 13.04 (Ferland et al. 2013). We considered two different cases: (1) a matter-bounded scenario in which the main parameters are U, the dimensionless ionization parameter, n_H, the hydrogen volume density, and N_H, the hydrogen column density; and (2) a radiation-bounded case, whose only parameters are U and n_H. In both cases, we assumed that the input AGN spectrum was given by a broken power law of the form ${L}_{\nu }\propto {\nu }^{\alpha }$, with α = −0.5 for E < 13.6 eV, α = −1.5 in the range 13.6 < E < 0.5 keV, and α = −0.8 at E > 0.5 keV, consistent with the observed spectrum of local AGN (e.g., Elvis et al. 1994; Kraemer et al. 2009).

For each resolution element in the southern emission region, we used the grid of simulations carried out using CLOUDY in order to find the best fit to the intrinsic emission line ratios, all computed with respect to the Hβ line. From this solution, we can obtain the corresponding value of Q, the number of ionizing photons emitted per unit of time, which is defined by

$$\begin{eqnarray}&&U=\displaystyle \frac{Q(H)}{4\pi {r}_{0}^{2}n(H)c},\end{eqnarray} \tag{ 1 }$$

where r₀ is the distance between the AGN and the cloud. We then use this value to obtain the corresponding intrinsic bolometric luminosity by assuming an intrinsic AGN spectral shape, such as the one given by, e.g., Elvis et al. (1994). The results of the simulations are presented in Figure 19. The best-fitting value for logU can be found between −2.9 and −2.6, with a mean value of −2.7.

Zoom In Zoom Out Reset image size

As can be seen in Figure 19, the resulting fits are not very sensitive to the value of n_H. Indeed, models with the same U and N_H but different values of n_H yield very similar reduced χ². This is critical in our calculation, as the required intrinsic AGN bolometric luminosity is proportional to U × n_H, as can be seen in Equation (1). For the n_H calculation, we assume a fully-ionized gas, as was done for the Voorwerpjes (Keel et al. 2012a). Hence, the atomic hydrogen column density is the same as the electron density, n_H = n_e, which can be estimated from the ratio of the two [S ii] lines. Furthermore, we can neglect the contribution from molecular hydrogen, given that no molecular gas was detected at the position of the southern ionized cloud. From the observed values of the [S ii] 6716/6731 ratio, we estimate an allowed average n_H range in the southern emission region of 10–100 atoms cm⁻³. In turn, this implies bolometric AGN luminosities in the range of 2.7 × 10⁴⁵ erg s⁻¹ to 1.8 × 10⁴⁶ erg s⁻¹. We compare this value with the current AGN luminosity, as observed in X-rays. From Chandra observations, Bianchi et al. (2008) reported a 2–10 keV luminosity of 1.5 × 10⁴³ erg s⁻¹. Using their reported N_H value of 7.1 × 10²³ atoms cm⁻², we can estimate an intrinsic 2–10 keV luminosity of 9.9 × 10⁴³ erg s⁻¹. Assuming a bolometric correction from 2–10 keV of 10 (e.g., Marconi et al. 2004), this implies that the intrinsic AGN luminosity was a factor of ∼3–20 times higher ∼40,000 years ago, when the light that photo-ionized the southern region was emitted. This is similar to the sample of Hanny's Voorwerp-like objects studied by Keel et al. (2017).

Short-term (up to decades) AGN variability by factors of ∼5 has been observed in the hard X-ray luminosity before (e.g., Ulrich et al. 1997, and references therein). Hence, our results imply that the AGN activity in Mrk 463E could have changed by up to an order of magnitude for the last ∼40,000. This is consistent with so-called "standard" variability, and thus consistent with the AGN lifetime of ∼100,000 years suggested by Schawinski et al. (2015).

The Chandra X-ray data show that the SMBH in the Mrk 463E nucleus is accreting at a ∼5× higher rate, and is more obscured than the one in Mrk 463W. In turn, most of the larger-scale structures detected by the VLT/MUSE observations, in both continuum and emission lines, appear to be associated preferentially with the Mrk 463E nucleus. Similarly, both the southern emission line region and the outflowing cloud in the north appear to be well-aligned with the apparent biconical gradient, which in turn is also linked to the Mrk 463E AGN. This could be interpreted as the AGN in Mrk 463E being older and more evolved than the one in Mrk 463W, which is then most likely just starting its activity. Additional evidence pointing in that direction was presented in the previous section, where a lifetime greater than 40,000 years was estimated for the Mrk 463E AGN.

Based on the ALMA ¹²CO(2–1) observations presented in Section 3.3, we see that both nuclei contain significant molecular gas reservoirs, ∼10⁹ M_☉ around Mrk 463E and ∼5 × 10⁸ M_☉ surrounding Mrk 463W. We further find ∼3 × 10⁷ M_☉ in molecular gas in clouds in between the two nuclei. This amount of material, albeit small, could be readily available to feed both SMBHs in the near future. Indeed, as presented in Section 4.3, we estimate that the gas infalling rate onto the Mrk 463E nucleus is over three orders of magnitude larger than the SMBH accretion rate inferred from the X-ray luminosity. Hence, we can conclude that both nuclei have the potential fuel to feed quasar-like luminosity systems and increase their SMBH masses significantly, by adding >10⁸ M_☉. However, as presented in Section 4.1, the outflow rate of the ionized gas is similar to the infall rate of the molecular gas onto the central regions of the system. Therefore, while the accretion rate onto the SMBHs is relatively small, there appears to be a rough balance between the material infalling (as molecular gas) and outflowing (as ionized gas). This outflowing material does not, however, appear to affect the star formation processes, as it is highly anisotropic and not aligned with the stellar disks.

Currently, the two nuclei in Mrk 463 are both active, perhaps at a relatively low level, heavily obscured at N_H ∼ 10²³ atoms cm⁻² and separated by ∼4 kpc. As the merger continues, it is expected that the two nuclei will get closer over timescales of 10⁸ years. The molecular gas that is now surrounding each nucleus will likely tend to concentrate such that a fraction of it will likely be available for accretion onto the SMBHs, while at the same time increasing the nuclear obscuration. Depending on the exact (and as-yet unknown) dynamics of the nuclei, the system may pass through a stage analogous to that which the nearby dual AGN NGC6240 is experiencing now, in which both nuclei are separated by ∼1 kpc and are heavily obscured (Komossa et al. 2003), while most of the gas is concentrated between the two nuclei (Tacconi et al. 1999; Iono et al. 2007; G. Privon et al. 2018, in preparation). Alternatively, the gas could remain concentrated on each nucleus as they approach each other, thus creating a configuration similar to what it is observed now in the central region of Arp 220 (e.g., Manohar & Scoville 2017, and references therein). While the amount of ongoing SMBH accretion in Arp 220 is still uncertain (e.g., Barcos-Muñoz et al. 2015), Chandra (Iwasawa et al. 2005) and NuSTAR observations (Teng et al. 2015) suggest that there is at least one heavily obscured, Compton-thick AGN associated with the system, on the western nucleus. In any case, following the scenario suggested by Treister et al. (2010), we can expect that, after ∼1–2 × 10⁸ years, the collapsed nuclear region will start to become unobscured due to the effects of radiation pressure, thus revealing an unobscured quasar, perhaps similar to the one observed in Mrk 231, with strong ionized winds and ultrafast outflows (Feruglio et al. 2015), like those already observed in this system, associated with the Mrk 463E nucleus.

We note that the case of Mrk 463 highlights the difficulty of using the instantaneous AGN luminosity to place objects within evolutionary sequences. At least one nucleus shows evidence for changing luminosities on timescales of 10^4–5 years, which is significantly shorter than those expected for dynamical evolution of the system (e.g., LaMassa et al. 2015; Parker et al. 2016; Gezari et al. 2017). This suggests it is likely problematic to use AGN luminosity as even a broad indicator of evolutionary stage, because one or both nuclei can rapidly rise or drop in luminosity at any point. This further highlights the need for multi-wavelength spatially-resolved spectroscopy over large spatial scales, as presented here, in order to fully understand the evolutionary stage of these complex merging galaxies.