Galactic-scale Feedback Observed in the 3C 298 Quasar Host Galaxy

Observations were taken using the integral field spectrograph OSIRIS (Larkin et al. 2006) with the upgraded grating (Mieda et al. 2014) behind the LGS-AO system at W.M. Keck Observatory on 2014 May 19 and 20, (UT). The quasar was used for tip/tilt correction, while the laser tuned to 589.2 nm created an artificial star on-axis for higher order corrections. We used the Hn3 (May 19) and Jn1 (May 20) filters with a plate scale of 100 milliarcseconds (mas) per lenslet with a position angle of 103°. In these modes, OSIRIS has a field of view of 32 × 64 in Jn1 and 48 × 64 in Hn3. We took four 600s exposures on-source in each filter, plus an additional 600s pure-sky frame. Each night immediately after the quasar observations, we observed the standard star HD136754 for telluric and flux calibrations. Both nights were photometric with near-infrared seeing of 04–05.

Early ALMA science (cycles 2 and 3) band 4 and band 6 observations (2013.1.01359.S and 2015.1.01090.S, PI: Vayner) were aimed at observing the rotational molecular transition of CO J = 3−2 and J = 5−4 in emission to map the distribution and kinematics of the molecular gas in 3C 298. One 1.8745 GHz spectral window was centered on CO (J = 3−2) (141.87 GHz) and CO (J = 5−4) (236.43 GHz), while three additional spectral windows were set up to map the continuum in each band. The effective velocity bandwidth per spectral window was approximately 4000 $\mathrm{km}\,{{\rm{s}}}^{-1}$ for band 4 and 2,400 $\mathrm{km}\,{{\rm{s}}}^{-1}$ for band 6. Observations were taken in an extended configuration with an approximate angular resolution of ∼04 and ∼03 for bands 4 and 6, respectively. In Table 1 we summarize the observational setup for each band.

Table 1.  ALMA Cycle 2 and 3 Observations Summary

Date	Band	Central Frequencies	Integration	PWV	Antennae	Beam	Linea	Continuum
		(GHz)	Time (minutes)	(mm)	$\#$	Size	σ/beam (mJy)	σ/beam (μJy)
2015 Aug 6	4	141.86	27.84	4.5	39	044 × 041	0.5	48
2016 Sep 9	4	141.85, 140.00, 129.97, 128.01	24.19	2.5	37	039 × 030	0.22	46
2016 Sep 17	6	236.42, 234.5, 220.99, 219.11	19.65	0.7	38	028 × 018	0.37	44

Note.

^aComputed per 34 $\mathrm{km}\,{{\rm{s}}}^{-1}$ channels.

Download table as:  ASCIITypeset image

The data were reduced with the OSIRIS data reduction pipeline version 3.2, which performs standard near-infrared IFS reduction procedures: it performs dark subtraction, adjusts channel levels, removes crosstalk, identifies glitches, cleans for cosmic rays, extracts spectra, assembles a data cube, and corrects for dispersion. We used the scaled sky-subtraction routine for the Jn1 data cubes, which uses families of atmospheric OH-emission lines between multiple frames for a cleaner subtraction. For Hn3 data we used our own custom sky-subtraction routine that scales only the nearest three OH lines in proximity to the quasar Hα emission line. This produced better residuals than the pipeline's scaled sky routine in Hn3. Inspection of individual spaxels in the scaled sky-subtracted data cubes still revealed strong OH-sky line residuals, typically over 2–3 spectral pixels in the wings of the Hα emission line. These spectral pixels were linearly interpolated using the slope from the neighboring two pixels around the strong residuals. The telluric spectrum of the calibration star was extracted over the seeing halo, the hydrogen absorption lines removed and its continuum was divided by a 9400 K blackbody function. The 1D telluric spectrum was normalized and divided into the quasar data cubes. Individual data cubes were then shifted to a common position and combined using a $3\sigma$ clipping algorithm, which is part of the OSIRIS data reduction pipeline. Finally we applied flux calibration to all quasar data cubes using the telluric corrected standard star spectrum and scaling the DN/s/channel to match the expected Jn1 and Hn3 band flux (erg s⁻¹ cm⁻²).

Both the quasar broad-line emission (Hβ and Hα) and quasar continuum originate from gas on parsec scales from the SMBH, and are therefore spatially unresolved in our OSIRIS observations. Wavelength channels of the broad-line emission and/or continuum can be used to construct a pristine quasar image that can then be used for PSF subtraction in the reduced data cube. In Vayner et al. (2016) we discuss in greater detail our PSF subtraction routine and its performance on a set of luminous type-1 radio-quiet quasars at z ∼ 2. In brief, we used broad emission line/continuum channels that should not overlap with the host galaxy emission with spectral channels offset by 5000–10,000 $\mathrm{km}\,{{\rm{s}}}^{-1}$ from the quasar redshift. We carefully select spectral channels that do not coincide with OH-emission lines or regions of low transparency in the near-infrared. PSFs are generated by combining individual data channels and are scaled to the peak pixel values. This empirical PSF is then subtracted from the entire data cube while rescaling to the peak pixel value of the quasar emission per wavelength channel.

After data reduction, the 100 mas mode in OSIRIS suffers from flux misassignment between adjacent spaxels given its enlarged pupil size in the instrument. The reductions still conserve the integrated flux of the source, but neighboring pixels can receive a ∼10% misassignment of flux. This effect is seen in bright stars (H < 16 mag) with excellent AO correction, where flux is misassigned from the bright central spaxel to the row above and below the centroid position of the point source. This effect can be easily identified by evidence of spaxels with an inaccurate spectral shape. Our PSF subtraction routine removes a significant portion of flux from the PSF in these spaxels, but they generally have stronger post-PSF subtraction residuals. We masked the affected spaxles with background flux values calculated by taking a standard deviation in a 1'' × 1'' sky region. We fit a 2D Gaussian to the PSF image and measure a FWHM of 0127 and 0113 in Hn3 and Jn1, respectively. After PSF subtraction of the quasar data cubes, we smooth the Hn3 and Jn1 data sets to a common resolution with a beam FWHM of 02 to improve the signal-to-noise ratio (S/N) in the diffuse parts of the host galaxy.

In this section we investigate the kinematics properties of nebular emission lines in the host galaxy of 3C 298. We inspect all individual spectra that overlap between the two observing modes (Jn1 and Hn3), which amounts to approximately 2,640 spectra. An emission line is identified to be real if the peak intensity is at least three times greater than the noise per wavelength channel, and the emission-line dispersion is larger than the instrumental resolution (0.206 nm for Hn3 and 0.174 nm for Jn1).

A single Gaussian profile provides a good fit to the 486.1 nm Hβ emission line that is redshifted into Jn1 at an observed wavelength of 1186 nm. The 495.9, 500.7 nm [O iii] lines are redshifted into the Jn1 band at an observed wavelength of 1210 and 1221 nm. The 495.9, 500.7 nm [O iii] lines are fit simultaneously, each with a single Gaussian profile. The position and width of the 495.9 nm line are held fixed to the redshift and width of the 500.7 nm line at each spaxel. The line ratio between the [O iii] lines are held fixed at 1:2.98 (Storey & Zeippen 2000). The 495.9, 500.7 [O iii] lines in several spaxels in the northwest and southeast regions require two Gaussian profiles for a good fit. The [N ii] 654.9, 658.5 nm, Hα 656.3 nm, and [S ii] 671.7 673.1 nm are redshifted into the Hn3 filter. Spaxels with detected Hα and both [N ii] emission lines are fit together with three Gaussian profiles. The position and width of the [N ii] lines are fixed to the Hα redshift and width, with a flux ratio between [N ii] 654.9, 658.5 nm of 1:2.95. Generally, a single Gaussian profile fit to Hα and each [N ii] line provides a satisfactory fit (${\chi }_{R}^{2}\sim 1\mbox{--}2$), with the exception of several spaxels in the southeast region, where an additional broad Hα component is necessary. The Hα, Hβ, and [O iii] broad-line components have similar velocity dispersions and offsets. The lower S/N of the [S ii] doublet make it challenging to fit a single Gaussian profile to each emission line, therefore at each spaxel where [S ii] is detected, we fit a single Gaussian profile to the combined signal. Regions where no [N ii] or [S ii] lines are detected have only a single Gaussian fit to Hα. Limits on the [N ii] and [S ii] lines are derived by inserting a Gaussian profile with the same width as Hα at the expected location of the emission line based on the Hα redshift, with a peak flux twice greater than the standard deviation of the noise. Similarly to a real detection, we integrate the inserted emission line to obtain a limit.

To construct 2D flux maps, we integrate each emission line from $-3\sigma$ to $+3\sigma$, where σ is derived from the line fit. The error on the line flux is calculated by taking a standard deviation of every three spectral channels and summing in quadrature over the same spectral region where the nebular emission line is integrated. Velocity maps for each emission line are relative to the redshift of the quasar's broad-line region, which is calculated by fitting a single Gaussian profile to the broad Hβ line constructed by spatially integrating the cube over the seeing halo before PSF subtraction. The velocity dispersion map has the instrumental PSF subtracted out in quadrature using the width of the OH-emission sky lines. The errors on the velocity offsets and dispersions are based on 1σ errors associated with the least-squares fit.

In Figure 1 we show the line-integrated emission of Hα, [O iii], and [S ii] in a three-color image composite alongside the [O iii] radial-velocity and dispersion maps of 3C 298. A distinct extended broad velocity emission region is cospatial with radio synchrotron emission emanating from extended jet/lobes. The radio VLA observations are taken from Mantovani et al. (2013; Project code: AJ206). We downloaded fully reduced clean map of 3C 298 at 8485.100 MHz.⁸ The centroid of the point source with the flattest spectral slope is associated with the optical location of the quasar. We extract the image centered on the quasar and rotate to a position angle of 103° to match our OSIRIS observations.

Zoom In Zoom Out Reset image size

We generate integrated spectra for three distinct outflow regions identified based on their radial velocity (±400 km s⁻¹) and dispersion (${V}_{\sigma }\gt 500$ $\mathrm{km}\,{{\rm{s}}}^{-1}$), see Figure 1. For the purposes of the figure, the northwestern outflow is identified as the redshifted outflow or "Outflow-R," and the northeastern outflow is identified as the blueshifted outflow or "Outflow-B," and the third outflow in the southeastern direction is identified as "AGN-Outflow." The AGN outflow probably belongs to a secondary nucleus in the 3C 298 system becuase of its isolated nature and because the ionized emission does not extend from the quasar and does not coincide with the quasar jet/lobes. The AGN outflow is in close proximity to the dynamical center of a rotating disk in the 3C 298 system that belongs to a second merging galaxy (see Section 3.5 for further discussion). Each spectrum is then fit with multiple Gaussian profiles. Outflow-R is best fit with a combination of two relatively broad (${V}_{\sigma }\sim 500$ km s⁻¹) Gaussian profiles in the [O iii] lines, Hα, and [N ii], which we interpret as a signature of outflowing gas along the line of sight. The 500.7 nm [O iii] line requires an additional relatively narrow (${V}_{\sigma }\sim 80$ km s⁻¹) component for a good fit that has no counterpart in Hα, [N ii], or [S ii]. A faint Hβ line is detected, and is fit with a single Gaussian component that potentially matches the broad components of [O iii] and Hα. See the top row of Figure 2 for the integrated spectrum in J and H band along with the fit. Components A (green curve) and B (red curve) are the broad lines, while component C (blue curve) is the narrow line found only in [O iii]. The white contour in the right column shows the region over which the data cube is spatially integrated. The spectrum of the blueshifted outflow (Outflow-B) region is best fit with a single broad Gaussian component in Hβ, [O iii], Hα, and [N ii]; the second row of Figure 2 shows the spectra along with the fit. In the redshifted outflow region the lines are very broad, so the [S ii] doublet is blended, making it very hard to fit the individual lines. The lines are slightly narrower in the blueshifted outflow region, and each emission line in [O iii] and Hα requires only a single Gaussian component. This made it possible to fit the [S ii] doublet using a single Gaussian profile for each emission line.

Zoom In Zoom Out Reset image size

We construct a high S/N spectrum over the entire AGN-Outflow region. Hα and [O iii] are fit with two components: a broad-line blueshifted Gaussian for the outflow, and a narrow component that we interpret as part of the AGN/quasar narrow-line region. Hβ, [N ii], and [S ii] are fit with a single Guassian narrow-line component.

In this section we investigate potential ionizing sources for distinct regions of the host galaxy. Line ratio maps are constructed by taking ratios of integrated flux maps [O iii], Hα, [N ii], and [S ii]. The Hβ line is only detected in a small number of individual spaxels; sensitivity and dust obscuration prohibits us from detecting this line over similar sized regions to other detected nebular emission lines. We construct an Hβ map over the region where Hα is detected by assuming case-B recombination (${F}_{{\rm{H}}\beta }={F}_{{\rm{H}}\alpha }/2.89$).

We create three line ratio maps log([O iii]/Hβ), log([N ii]/Hα), and log([S ii]/Hα). In Figure 3 we plot the line ratios on a standard BPT diagram, log([O iii]/Hβ) versus log([N ii]/Hα). Empirical (Kauffmann et al. 2003) and theoretical (Kewley et al. 2001) curves separating photoionization by O-type stars versus quasar ionization are shown. Generally, values that lie above these curves represent photoionization by an AGN, while points below represent photoionization by newly formed O-type stars, tracing regions of active/recent star formation. Points on the BPT diagram are color coded to match the line ratio map of log([O iii]/Hβ) (middle panel in Figure 3). The 1'' central region (red and orange in Figure 3) is mainly photoionized by the quasar in an extended narrow-line region (ENLR), as shown by very high log([O iii]/Hβ) ≳ 0.8 values. 658.5[N ii] is detected over the red region in individual spaxels, while over the orange region, we place a 2σ flux limit. Even within the limits, the orange region falls in the AGN photoionzation portion of the BPT diagram due to very high log([O iii]/Hβ) values. The star formation region (blue) has low log([O iii]/Hβ) < 0.8 and log([N ii]/Hα) < 0.5 values, corresponding to ionization from newly formed stars. We generate a spectrum over the Star Formation region and easily identify and fit a single Gaussian to the [O iii], Hα, [N ii], and a Hβ line with an S/N of 2 identified (bottom panel of Figure 2).

Zoom In Zoom Out Reset image size

Star-forming regions are spatially offset from the large outflows, and their kinematics and morphologies agree with a potential tidal feature induced by the merging galaxies. Outflow regions (yellow and green) correspond to the highest log([O iii]/Hβ) line ratios. These regions are inconsistent with ionization by young stars, and are photoionized by quasars and shocks. Gas in outflow regions is moving at sufficiently high velocity, so the Mach number of the wind is high, and strong radiative shocks are expected. Using shock models from Allen et al. (2008), we find that a large portion of observed line ratios agree with shock models for gas with an electron density between ${10}^{2}\mbox{--}{10}^{3}$ cm⁻³ and shock velocities of 1000 km s⁻¹. In Figure 4 we plot the log([O iii]/Hβ) versus log([S ii]/Hα)values over regions where [S ii] was detected in individual spaxels. We overlay shock models on this BPT diagram. A large fraction of the nebular line ratios in the redshifted outflow region (green) and blueshifted outflow region (yellow) are consistent with these models. Dark red/blue lines in the model represent higher velocity gas and stronger magnetic parameter (see Allen et al. 2008 for further details). Furthermore, the region that is photoionized by AGN shows a spatial profile that drops off as $1/{R}^{2}$ in nebular emission, while in the region where we see powerful outflows, the profile is more complex, suggesting multiple ionizing sources. Both regions show similar extinction values, as we discuss below.

Zoom In Zoom Out Reset image size

In order to investigate how dust can alter our photoioniziation measurements, we construct a high S/N spectrum over the ENLR (red in Figure 3) to estimate the amount of dust obscuration from the Hα/Hβ ratio. Using the Calzetti et al. (2000) law, we find an average ${A}_{V}=0.07$ over the ENLR (the red region in Figure 3), based on an Hα/Hβ ratio of 2.9. This would on average increase the ratio of log([O iii]/Hβ) only by a factor of 1.05 for the ENLR; this is not large enough to alter our results. We find that in portions of the ENLR and redshifted outflow region, the ${{\rm{A}}}_{{\rm{v}}}$ value is consistent with zero extinction in the ionized gas. Similar results are found over the outflow regions, suggesting negligible extinction in the ionized gas. Over the star-forming region (blue region in Figure 3), we measure an ${A}_{V}$ value of 1.2 based on an Hα/Hβ ratio of 4.0. The observed non-detection of Hβ in individual spaxels is consistent with the derived integrated extinction value and the expected Hβ S/N relative to [O iii] and Hα emission lines.

To the southeast of the quasar, there is a kinematic feature resembling a rotating galactic disk that is offset by ∼200 $\mathrm{km}\,{{\rm{s}}}^{-1}$ from the quasar BLR, with a projected rotational velocity of ±170 $\mathrm{km}\,{{\rm{s}}}^{-1}$. In this section we outline the modeling done on the radial-velocity map of [O iii] emission to confirm the rotating-disk nature of this feature. We isolate this feature for the disk fitting, using a box size of 07 × 17 centered on the gradient feature in the [O iii] radial-velocity map. The boxed region is selected to exclude any strong broad emission from the conical outflow. The [O iii] velocity field is modeled by fitting a 2D arctangent disk model given by

$$\begin{eqnarray}&&V(r)=\displaystyle \frac{2}{\pi }{V}_{\max }\arctan \left(\displaystyle \frac{r}{{r}_{\mathrm{dyn}}}\right),\end{eqnarray} \tag{ 1 }$$

where V(r) is rotation velocity at radius r from the dynamical center, V_max is plateau velocity, and r_dyn is the radius at which the arctangent function has a turn over to a decreasing slope. The measured line-of-sight velocity from our observations relates to V(r) as

$$\begin{eqnarray}&&V={V}_{0}+\sin i\cos \theta V(r),\end{eqnarray} \tag{ 2 }$$

where

$$\begin{eqnarray}&&\cos \theta =\displaystyle \frac{(\sin \phi ({x}_{0}-x))+(\cos \phi ({y}_{0}-y))}{r}.\end{eqnarray} \tag{ 3 }$$

The radial distance from the dynamical center to each spaxel is given by

$$\begin{eqnarray}&&r=\sqrt{{(x-{x}_{0})}^{2}+{\left(\displaystyle \frac{y-{y}_{0}}{\cos i}\right)}^{2}},\end{eqnarray} \tag{ 4 }$$

where ${x}_{0},{y}_{0}$ is spaxel location of the dynamical center, V₀ is velocity offset at the dynamical center relative to the redshift of the quasar broad-line region, ϕ is the position angle in spaxel space, and i is the inclination of the disk.

We fit the seven-parameter velocity field model to a selected region of our observed [O iii] velocity map using a nonlinear least-squares routine. The selected box region tries to exclude broad emission from the extended quasar outflow and a tidal feature that seems to show its own distinct kinematic structure. Our best fit has a ${\chi }_{R}^{2}=0.7$ with the following values: ${x}_{0},{y}_{0}=-0\buildrel{\prime\prime}\over{.} 49$, +094 relative to the centroid of the quasar, V_max = 209 ± 33 $\mathrm{km}\,{{\rm{s}}}^{-1}$, ${r}_{\mathrm{dyn}}=2.2\pm 0.98$ kpc, $i=72^\circ \pm 4^\circ$, $\phi =-109^\circ \pm 6^\circ$, and ${V}_{0}=170\pm 9\,\mathrm{km}\,{{\rm{s}}}^{-1}$ relative to the redshift of the quasar broad-line region. Figure 5 shows the region selected for disk fitting with the modeled 2D velocity profile and residuals between the model and observed velocities. The disk is offset from the quasar by 106 or ∼9 kpc, which is evidence of a merging disk system in 3C 298. We obtain a dynamical mass of $1.3\pm 0.8\times {10}^{10}\,{M}_{\odot }$ within a radius of ${r}_{\mathrm{dyn}}=2.2\,\mathrm{kpc}$. The dynamical center of this disk is in the vicinity of the isolated AGN outflow identified in Figure 2. With currently available archival Chandra ACIS observations, we are unable to explore the X-ray properties of this secondary AGN candidate. Unfortunately, there is strong asymmetry in the Chandra PSF along this position angle from the quasar. Future Chandra observations at a specific spacecraft roll angle will allow us to search for X-ray emission from this secondary AGN and explore the potential dual AGN nature of this system.

Zoom In Zoom Out Reset image size

Data reduction was performed using CASA (Common Astronomy Software Applications; McMullin et al. 2007) versions 4.4 and 4.7. At the observed frequencies, synchrotron emission from the quasar dominates molecular CO or dust emission, and has sufficient S/N to perform self-calibrations directly on the science source. We used the CASA CLEAN function to establish a model for the synchrotron continuum through several interactive runs with clean masks centered on high S/N features. Cleaning is performed with Briggs weighting using a robust value of 0.5 with a pixel scale of 005. We used the gaincal function to perform phase corrections. The self-calibrated data were then cleaned again with further phase corrections until we no longer saw a significant improvement in the S/N on the continuum. The final root mean square (rms) improved by a factor of 6–8 in the continuum images. We perform continuum subtraction in UV space by fitting a first-order polynomial to channels free of line emission from the host galaxy, and subtract this fit from the rest of the channels.

Data cubes were imaged using clean with a plate scale of 005 per pixel, a spectral channel size of 34 $\mathrm{km}\,{{\rm{s}}}^{-1}$, and natural weighting to improve the S/N in the diffuse structure of the host galaxy at the cost of angular resolution. For cycle 3 data, clean masks were placed on CO emission with S/N > 5 with the same mask applied to all channels. The resulting beam sizes and rms values are reported in Table 1.

We find that the cycle 2 band 4 data cube is a factor of 2 noisier due to poorer weather conditions at the time of observations. Faint emission is detected with a peak S/N of 3.6σ for CO (J = 3−2) emission over a number of spaxels in the east/northeast direction from the quasar. Applying UV tapering to a resolution of 098 × 08 with Briggs weighting (robust = 0.5) improves the S/N to 5, but the line is not spatially resolved. For the analysis of CO (J = 3−2), we therefore defer to cycle 3 data taken under superior weather conditions.

We align the ALMA and OSIRIS data cubes by matching the position of the optical quasar emission to the unresolved quasar synchrotron emission centroid in the continuum map. The centroid of the quasar matches the location of the unresolved point source with the flattest spectra seen using high-resolution centimeter observations of 3C 298 with MERLIN (component B1/B2 in Fanti et al. 2002). This location is typically associated with the quasar core. Finally, we rotate the ALMA data cube to a position angle of 103° to match the OSIRIS observations.

In this section we describe the analysis of resolved CO (3−2) and (5−4) 3D spectroscopy. Initial inspection revealed a high S/N $\gtrsim \,5$ detection of CO (J = 3−2) and (J = 5−4) emission. We extracted spectra over the beam size centered on the detected regions. We then collapsed the data cube over spectral channels where the CO features are detected, and constructed S/N maps by dividing the integrated emission by the standard deviation computed in an empty-sky region. Spaxels that showed an emission line with an S/N ≥ 3 were fit with a Gaussian profile. We constructed a flux map by integrating over the line ($-3\sigma$ to $+3\sigma$). We created a velocity map by computing the Doppler shift of the emission-line centroid relative to the redshift (z = 1.439) of the broad-line region (calculated from Hβ), and generated a velocity dispersion map from the Gaussian fit. Figure 6 shows the CO emission, radial velocity, and dispersion maps.

Zoom In Zoom Out Reset image size

3.7.1. Molecular Gas Velocity Field Modeling

In this section we outline the modeling performed on the CO (3−2) radial-velocity map to derive a dynamical mass for the inner ∼2 kpc of the host galaxy. We find that the majority of the CO emission is concentrated at the spatial location of the quasar and has a secondary peak offset by 16 kpc that is spatially coincident with the recent star formation from the OSIRIS BPT analysis (see Figure 3 and Section 3.4). CO (3−2) emission concentrated on the quasar shows a distinct velocity gradient resembling a rotating disk, with an extent of about 08 with a velocity difference of ±150 $\mathrm{km}\,{{\rm{s}}}^{-1}$. Similar to Section 3.5, we model the molecular disk with a hyperbolic arctangent function and derive the following properties about the disk: ${\chi }_{R}^{2}\,=0.6:{x}_{0},{y}_{0}=0\buildrel{\prime\prime}\over{.} 05,0\buildrel{\prime\prime}\over{.} 0$ relative to the centroid of the quasar, V_max = 392 ± 65 $\mathrm{km}\,{{\rm{s}}}^{-1}$, ${r}_{\mathrm{dyn}}=2.1\pm 0.9$ kpc, $i=54\buildrel{\circ}\over{.} 37\,\pm 6\buildrel{\circ}\over{.} 4$, $\phi =5\buildrel{\circ}\over{.} 3\pm 1\buildrel{\circ}\over{.} 28$, and ${V}_{0}=-13.0\pm 3.15\,\mathrm{km}\,{{\rm{s}}}^{-1}$. Fits of the disk along with the residuals are shown in Figure 7. We derive a dynamical mass $1.7\pm 0.9\times {10}^{10}\,{M}_{\odot }$ within ${r}_{\mathrm{dyn}}=2.1\,\mathrm{kpc}$ using the disk inclination derived from the velocity modeling.

Zoom In Zoom Out Reset image size

Alternatively, we collapsed the data cube along the spectral direction from −150 $\mathrm{km}\,{{\rm{s}}}^{-1}$ to +150 $\mathrm{km}\,{{\rm{s}}}^{-1}$ to produce a CO (3−2) image of the molecular disk. Using a 2D Gaussian, we obtained a major axis value of 0374 ± 005 and a minor axis of 0155 ± 0098 corrected for the beam size. Assuming the molecular disk can be approximated by an oblate spheroid, we derived the inclination angle using the following formula (Holmberg 1946):

$$\begin{eqnarray}&&{\rm{i}}={\cos }^{-1}\sqrt{\displaystyle \frac{{({\rm{b}}/{\rm{a}})}^{2}-{{\rm{q}}}_{0}^{2}}{1-{{\rm{q}}}_{0}^{2}}},\end{eqnarray} \tag{ 5 }$$

where a and b are the major and minor axes, and q₀ is the axial ratio for an edge-on disk taken to be 0.13. We derived a dynamical mass of $1.0\pm 0.5\times {10}^{10}\,{M}_{\odot }$ using a major axis radius of 1.6 ± 0.215 kpc and an inclination angle of $66\buildrel{\circ}\over{.} 6\pm 17\buildrel{\circ}\over{.} 8$. The dynamical masses and disk inclination angles obtained with these two methods agree within the errors. With two distinct galactic disks (see Section 3.5) in the 3C 298 system, we find evidence for an intermediate- to late-stage merger.

Shen et al. (2011) determined a single-epoch black hole mass for 3C 298 using three calibration methods: ${10}^{9.57\pm 0.03}\,{M}_{\odot }$ using the MgII-black hole mass relationship calibrated from their study, ${10}^{9.37\pm 0.03}\,{M}_{\odot }$ using the Vestergaard & Osmer (2009) calibration, and ${10}^{9.56\pm 0.01}\,{M}_{\odot }$ using the McLure & Dunlop (2004) relation. The systematic differences between black hole mass measurements are typically due to differing quasar samples and the properties of the emission lines (e.g., Hβ) that were used to calculate the SMBH mass.

With both the derived dynamical and black hole mass, we can compare the nuclear region of 3C 298 to the local black hole ${M}_{\mathrm{bulge}}\mbox{--}{M}_{\mathrm{BH}}$ relationship. In Figure 8 we plot the 3C 298 black hole and bulge dynamical mass relative to the local scaling relation from Häring & Rix (2004), Sani et al. (2011), and McConnell & Ma (2013). Compared to all three relationships, the 3C 298 bulge dynamical mass is either ∼2–2.5 orders of magnitude lower than what is expected, or the black hole is ∼2 orders of magnitude higher than what is expected for its bulge mass. If we assume that the 3C 298 bulge is more extended than what is measured by ALMA, then we can use the tidal feature at 21.5 kpc as an estimate for the total enclosed mass. To do this, we integrate all spaxels with a detected CO (3−2) line and remove the outflow contribution by simultaneously fitting a broad and a narrow Gaussian component to the outflow region. We measure a maximum line width of 369.3 $\mathrm{km}\,{{\rm{s}}}^{-1}$ at 90% intensity, and find a maximum circular velocity of $184.6/\sin (i)\,\mathrm{km}\,{{\rm{s}}}^{-1}$ at 21.5 kpc from the quasar. Using this maximum circular velocity, we obtain a total enclosed mass at this radius of $\sim 2.7\times {10}^{11}\,{M}_{\odot }$, assuming the smallest measured inclination angle from the molecular disk. This value should be considered as a maximum limit for the total enclosed mass of the 3C 298 system. This enclosed dynamical mass at 21.5 kpc from the quasar still places 3C 298 an order of magnitude above the local scaling relation. The radius used for this calculation is also significantly larger than the typical bulge radius at this redshift (Bruce et al. 2014; Sachdeva et al. 2017), and there is no guarantee that the entire enclosed stellar mass will end up in the newly created bulge when the merger is completed. This implies a delay in growth of the galactic bulge for the 3C 298 host galaxy with respect to its SMBH. Averaging SMBH masses and dynamical masses determined for 3C 298, we obtain the following values: ${M}_{\mathrm{dyn},\mathrm{bulge}}=1.35\pm 0.5\times {10}^{10}$, and ${M}_{\mathrm{SMBH}}=3.23\pm 1.1\times {10}^{9}$. Errors are added in quadrature with the SMBH mass standard deviation included to address the systematic uncertainty associated with using various calibration methods.

Zoom In Zoom Out Reset image size

3.7.2. Molecular Gas Mass

In this section we derive the molecular gas mass from the spatially integrated CO (3−2) and (5−4) emission presented in Table 2.

Table 2.  Measured Molecular-line Intensities

Region	${I}_{\mathrm{CO}(3-2)}$	${I}_{\mathrm{CO}(5-4)}$
Molecular Disk	0.63 ± 0.035 Jy km s−1	1.26 ± 0.063 Jy km s−1
Tidal Featurea	0.13 ± 0.013 Jy km s−1	0.1 ± 0.025 Jy km s−1

Note.

^aMolecular gas associated with the star-forming region shown in blue in Figure 3.

Download table as:  ASCIITypeset image

In order to measure the molecular gas in the 3C 298 system, we convert the CO (3−2) flux into ${L}_{\mathrm{CO}(3-2)}^{{\prime} }$ using the following equation from Carilli & Walter (2013),

$$\begin{eqnarray}&&{L}_{\mathrm{CO}(3-2)}^{{\prime} }=3.25\times {10}^{7}{{\rm{S}}}_{\mathrm{CO}(3-2)}{\rm{\Delta }}{\rm{v}}\\ &&\quad \times \,\displaystyle \frac{{D}_{{\rm{L}}}^{2}}{{(1+z)}^{3}{\nu }_{\mathrm{obs}}^{2}}{\rm{K}}\,\mathrm{km}\,{{\rm{s}}}^{-1}\,{\mathrm{pc}}^{2}.\end{eqnarray} \tag{ 6 }$$

We convert into CO (1–0) luminosity (${L}_{\mathrm{CO}(1-0)}^{{\prime} }$) using an ${L}_{\mathrm{CO}(3-2)}^{{\prime} }/{L}_{\mathrm{CO}(1-0)}^{{\prime} }$ ratio of 0.97 (Carilli & Walter 2013). Using the typical ${\alpha }_{\mathrm{CO}}$ value from nuclear star bursts and quasars ($0.8\,{\rm{K}}\,\mathrm{km}\,{{\rm{s}}}^{-1}\,{\mathrm{pc}}^{2}$)⁻¹), we derive a total molecular gas mass equal to $8.75\pm 0.4\times {10}^{9}\,{M}_{\odot }$. The total molecular gas includes all regions with a fitted CO (3−2) line in Figure 6. The molecular gas disk contains $6.6\pm 0.36\times {10}^{9}\,{M}_{\odot }$, while the active star formation region that lies 16 kpc away contains $1.44\pm 0.14\times {10}^{9}\,{M}_{\odot }$. We measure a 2σ molecular gas limit of $1\times {10}^{9}\,{M}_{\odot }\left(\tfrac{{\alpha }_{\mathrm{CO}}}{0.8}\right)\left(\tfrac{{V}_{\mathrm{FWHM}}}{200\,\mathrm{km}\,{{\rm{s}}}^{-1}}\right)$ per beam. This corresponds to a molecular gas surface density of $104\,{M}_{\odot }\,{\mathrm{pc}}^{-2}\,\left(\tfrac{{\alpha }_{\mathrm{CO}}}{0.8}\right)\left(\tfrac{{V}_{\mathrm{FWHM}}}{200\,\mathrm{km}\,{{\rm{s}}}^{-1}}\right)$. These estimates are valid for the inner few arcseconds around the phase center (quasar).

The CO (3−2) and (5−4) emission in some spaxels shows relatively broad (${V}_{\sigma }\gt 270\,\mathrm{km}\,{{\rm{s}}}^{-1}$) emission, greater than the escape velocity $\sqrt{2}{V}_{\mathrm{rotational}}\sim 270\,\mathrm{km}\,{{\rm{s}}}^{-1}$ at the observed edge of the rotating disk. In the position–velocity diagram (Figure 7), broad blueshifted emission is seen, with velocities ranging from −400 to 200 $\mathrm{km}\,{{\rm{s}}}^{-1}$. This emission spatially lies away from the ordered ±150 $\mathrm{km}\,{{\rm{s}}}^{-1}$ rotation profile. When we integrate over these spaxels, the CO (3−2) and (5−4) emission profiles resemble both an outflow with broad emission and narrow emission that is likely emanating from the molecular disk. We fit a double Gaussian component to the CO (3−2) line and a single Gaussian to the CO (5−4) data. We measure a FWHM of 624 ± 49 $\mathrm{km}\,{{\rm{s}}}^{-1}$ for the broad-line emission in CO (3−2) and 687 ± 18 $\mathrm{km}\,{{\rm{s}}}^{-1}$ in CO (5−4). The broad emission line represents either an extended outflow originating from the nuclear region of the 3C 298 quasar or from the molecular disk itself. Figure 9 shows spectra of the CO (3−2) and (5−4) emission regions with their corresponding Gaussian fits. The broad emission in both lines is predominantly found on the blueshifted side of the rotating disk. The total molecular gas in the outflow is $3.3\pm 0.1\times {10}^{9}\,{M}_{\odot }$. In Section 4 we measure the molecular gas outflow rate, kinetic energy, and momentum to understand its impact on the galaxy.

Zoom In Zoom Out Reset image size

We obtain archival Hubble Space Telescope (HST) observations of 3C 298 to quantify properties of the stellar populations of the host galaxy (GO13023; P.I. Chiaberge). Detailed descriptions of the observations are available in Hilbert et al. (2016). In summary, observations were taken in the F606W (${\lambda }_{p}=588.7$ nm, width = 218.2 nm) and F140W (${\lambda }_{p}=1392.3$ nm, width = 384.0 nm) filters, which cover the rest-frame wavelength range of 196.6 nm–286.1 nm and 492.1–649.6 nm, respectively. These wavelength ranges bracket the 4000 Å break feature in galaxy spectra. Two observations were taken in each filter for a total exposure time of 1100.0 s (F606W) and 498.46 s (F140W).

We use the nearby star SDSS J141908.18 + 062834.7 to construct a point-spread function (PSF) for the F606W filter. The PSF star and quasar have a similar magnitude in the F606W filter with a similar g–r color as measured in SDSS. Both the PSF star and quasar are saturated, so they share similar bleeding and diffraction patterns on the detector. In the F140W filter, the quasar is unsaturated and is about 0.8 magnitudes brighter, so we combine two unsaturated stars in the field to produce a final PSF with a matching S/N in the diffraction spikes structure. We extract a 12'' × 12'' box centered on the quasar and the PSF, and scale the flux of the PSF image to match the peak of the quasar emission and then subtract the two images. We also tried to match only the flux in the diffraction spikes that did not overlap with the structure in the host galaxy, and obtained a similar scaling factor. In both filters the inner 1'' is dominated by noise from the PSF subtraction, the diffraction spikes at position angles of 0°, 180°, 225° and along the bleeding pattern in the F606W filter that extends about 05 along PA 0° from the quasar. However, the majority of the host galaxy lies between position angles of 45° and 170°, where the structure is least affected by residual noise. The quoted residual structure is for an image at a PA of 103°, matching the observations of OSIRIS and ALMA.

We convert electron counts at each pixel into flux density (Jy) using the 'PHOTFNU' header value. We resize the pixels to 100 mas plate scale to match OSIRIS and ALMA data by using the flux-conserving IDL frebin routine. We convert the maps into AB magnitude/arcsec² and construct a color map of the host galaxy. Reliable colors are extracted down to a surface brightness limit of 23.5 AB mag/arcsecond² for the majority of the host galaxy, with the exception of the redshifted outflow region that falls in the area dominated by residual noise from PSF subtraction.

Figure 10 presents the WFC3 F606W (left) and F140W (middle) after PSF subtraction in 01/pixel scale. The right panel shows the resolved host galaxy colors (F606W-F140W). Overlaying Hα contours on the color map shows that the star-forming regions identified in OSIRIS (blue region in Figure 3) are nicely aligned with regions with bluer colors. Additionally, we overlay the ALMA CO (3−2) observation. The molecular clump offset by 16 kpc from the quasar matches bluer regions where clumpy structure is seen in the F606W observations. This is evidence for young stellar populations in these regions, as would be expected from on-going star formation.

Zoom In Zoom Out Reset image size

Two available filters for the 3C 298 system are ideally placed to bracket the break feature at a rest-frame wavelength of 4000 Å found in galaxy spectra. Thus we can easily identify young (<100 Myr) stellar populations without any ambiguity, as we attribute blue colors (F606W-F140W < 1.5) in these filters to stellar populations dominated by young O and B stars. Redder colors in these filters can come from older stellar populations (100 Myr–1 Gyr) or from young dusty star-forming regions, and thus we are unable to say much about older stellar populations without additional filters.

HST colors provide strong constraints on the resolved star formation and stellar population history across the 3C 298 system. We approximate the HST color map with the flexible stellar population synthesis (FSPS) code (Conroy et al. 2009; Conroy & Gunn 2010), using a five-parameter star formation history model. We start the FSPS code with an exponential star formation history that is followed by a burst of star formation 2 Gyr later, using a solar metallicity and Salpeter IMF. Using the FSPS code, we find that varying the star formation history and e-folding the timescale τ from 0.1 to 10² before the starburst does not significantly affect the age of the stellar population after the starburst.

The average F606W-F140W color over the blueshifted outflow region is best explained with a stellar population that has an age of 25 Myr after the burst. According to ionized gas emission, dust does not affect the age estimated in this region since the Hα and Hβ ratio suggests a low E(B–V) ∼ 0 value. However, ALMA observations reveal that part of the molecular disk centered on the quasar has extended emission into the blueshifted outflow region, suggesting that a part of the region can have higher dust extinction. The age of the outflow region and the modeled rest-frame U–V color imply that the stellar population resides in the transition zone between the red sequence and the blue cloud in the U–V versus M_V color diagram (Bell et al. 2004; Sánchez et al. 2004; Hopkins et al. 2008). Furthermore, u–r modeled colors at this age suggest that the region lies where post-starburst galaxies reside in the u–r versus M_stellar diagram (Wong et al. 2012). The extracted 3C 298 model spectra shows strong Balmer absorption lines with a strong 4000 Å break, characteristic of a post-starburst galaxy. Stellar colors over the blueshifted outflow region are consistent with a strong episode of star formation that was abruptly halted.

In the region where star formation is identified in Figure 3 based on BPT diagnostics and in the molecular gas clump found in CO, using the same initial conditions in the FSPS code with the outflow region, we find a stellar population with an age of 20 Myr after a burst of star formation. Without any dust correction, this age should be taken as an upper limit. Ahead of the jet/blueshifted outflow, we find an even younger stellar population. From the Hα emission, we can place a limit of 0.3 ${M}_{\odot }\,{\mathrm{yr}}^{-1}\,{\mathrm{kpc}}^{-2}$ across this region. The ALMA observation of the CO (3−2) transition yields a limit of 1 × 10⁹ ${M}_{\odot }$ on the molecular gas. This suggests that the current star formation rate is relatively low. However, the blue colors indicate recent star formation activity. The stellar populations in this region would then be associated with a young age of 7 Myr after the burst. It is likely that the star formation event was very recent and short (6–10 Myr), which we deduce from the lack of Hα emission that would otherwise be present if star formation persisted for a time longer than 6 Myr.

Ages quoted in this analysis could be altered by additional quasar emission from light scattered off dust grains or electrons in the ISM, making the colors systemically bluer. However, polarimetric observations of 3C 298 reveal that the source is not heavily polarized, with an optical polarization of only 0.77 ± 0.39% (Stockman et al. 1984). Given that the bluest regions also show star formation from other indicators (e.g., Hα and CO (3−2) emission), this implies that the majority of UV emission comes from young massive stars and not from scattered quasar light. The young stellar population observed >16 kpc beyond the blueshifted outflow is even farther away than the region that is photoionized by the quasar, indicating that ionizing radiation from the quasar does not reach this region.

The redshifted and blueshifted outflow regions resemble a cone-like structure. For a conical outflow with constant density ($\overline{\rho }$), we use the outflow rate equation from Cano-Díaz et al. (2012),

$$\begin{eqnarray}&&\dot{M}={{vR}}^{2}{\rm{\Omega }}\overline{\rho },\end{eqnarray} \tag{ 7 }$$

where v is the velocity of the material in the outflow that is assumed to be moving at a constant rate over the cone, R is the radial extent of the cone, and Ω is the opening angle of the cone. For $\overline{\rho }=M/V=\tfrac{3M}{{R}^{3}{\rm{\Omega }}}$, the outflow rate equation simplifies to

$$\begin{eqnarray}&&\dot{M}=3\displaystyle \frac{{Mv}}{R}.\end{eqnarray} \tag{ 8 }$$

When we assume that individual ionizing clouds in the outflow have the same density, we can derive an ionized gas mass from the line-integrated Hα luminosity and the average electron density over the outflow using (Osterbrock & Ferland 2006)

$$\begin{eqnarray}&&{M}_{\mathrm{gas}\mathrm{ionized}}=0.98\times {10}^{9}\left(\displaystyle \frac{{L}_{{\rm{H}}\alpha }}{{10}^{43}\,\mathrm{erg}\,{{\rm{s}}}^{-1}}\right){\left(\displaystyle \frac{{n}_{{\rm{e}}}}{100{\mathrm{cm}}^{-3}}\right)}^{-1}.\end{eqnarray} \tag{ 9 }$$

The electron density is derived from the line ratio of the [S ii] lines in the blueshifted outflow region using a Gaussian fit to each component of the doublet. We find a ratio of ([S ii] 6717/[S ii]6731 = 1.14), which yields an electron density of 272 cm⁻³ using the getTempDen code part of PyNeb (Luridiana et al. 2015) package. We calculate the ionized gas mass from the [O iii] line using the method presented in Cano-Díaz et al. (2012). This method requires knowledge of the ionized gas-phase metallicity at the site of the outflow region. We assume a metallicity value set to solar for a conservative estimate (i.e., for less than solar, the gas mass increases). For the outflow velocity we use the v₁₀ parameter, which is the velocity value where 10% of the line is integrated. This value is calculated from the model fits to the spatially integrated [O iii] and Hα lines for each outflow region. The wings of the emission lines most likely give the true average velocity of the outflow, as the lower velocities seen in the line profiles are probably due to projection effects of the conical structure (Cano-Díaz et al. 2012; Greene et al. 2012). The measured properties of the the outflow regions are presented in Table 3.

Table 3.  3C 298 Ionized Outflow Properties

Region	${L}_{{\rm{H}}\alpha }$	${L}_{[{\rm{O}}{\rm{III}}]}$	${M}_{{\rm{H}}\alpha }$	${M}_{[{\rm{O}}{\rm{III}}]}$	v	R	${\dot{M}}_{{\rm{H}}\alpha }$	${\dot{M}}_{[{\rm{O}}{\rm{III}}]}$
	×1044 $\mathrm{erg}\,{{\rm{s}}}^{-1}$	×1044 $\mathrm{erg}\,{{\rm{s}}}^{-1}$	×108 ${M}_{\odot }$	×108 ${M}_{\odot }$	$\mathrm{km}\,{{\rm{s}}}^{-1}$	kpc	${M}_{\odot }\,{\mathrm{yr}}^{-1}$	${M}_{\odot }\,{\mathrm{yr}}^{-1}$
Outflow-R	0.29	1.8	10	3.5	1703	4.7	1110	390
Outflow-B	0.09	0.3	3.2	0.6	1403	3.4	405	77
Outflow-AGN	0.03	0.04	0.8	0.064	1400	3	115	9

Download table as:  ASCIITypeset image

For the redshited and blueshifted outflow regions, we obtain a combined outflow rate of 467 ${M}_{\odot }\,{\mathrm{yr}}^{-1}$ and 1515 ${M}_{\odot }\,{\mathrm{yr}}^{-1}$ for [O iii] and Hα, respectively, while for the AGN outflow region, we obtain an outflow rate of 9 ${M}_{\odot }\,{\mathrm{yr}}^{-1}$ and 115 ${M}_{\odot }\,{\mathrm{yr}}^{-1}$ for [O iii] and Hα. The Hα outflow rate is likely a better representation of the ionized outflow rate, since Hα is capable of probing denser regions where [O iii] would be collisionally deexcited. The observed Hα outflow rate is similar to what is observed in other quasar surveys (Carniani et al. 2015), where a higher outflow rate is measured in Hα than in [O iii]. The total kinetic energy and luminosity of the ionized outflow are $3.5\times {10}^{58}$ erg and $1.2\times {10}^{45}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$. The kinetic luminosity is consistent with the value usually invoked in theoretical models (Hopkins & Elvis 2010) as necessary to affect star-forming properties in the host galaxy. Over the outflow regions, we measure a 2σ star formation rate upper limit of 0.3 ${M}_{\odot }\,{\mathrm{yr}}^{-1}\,{\mathrm{kpc}}^{-2}$, by using a Gaussian profile with 2σ peak flux and a FWHM of 200 $\mathrm{km}\,{{\rm{s}}}^{-1}$, with the multi-Gaussian fit of the outflowing gas subtracted. The properties of the multiple outflow components are summarized in Table 3.

Similarly, using Equation (8), we compute the molecular outflow rate. As found in Section 3.7, the extent of the broad molecular emission from CO (3−2) and (5−4) lines is approximately 1.6 kpc, and the total molecular gas mass in the outflow region is $3.3\times {10}^{9}\,{M}_{\odot }$ with an outflow velocity of 400 $\mathrm{km}\,{{\rm{s}}}^{-1}$ as measured by V₁₀ parameter. This yields a total molecular outflow rate of 2300 ${M}_{\odot }\,{\mathrm{yr}}^{-1}$. The kinetic energy and luminosity of the molecular outflow are $5.1\times {10}^{57}$ erg and 1 × 10⁴⁴ erg s⁻¹.

Using the virial parameter, the ratio of the free-fall timescale to the dynamical timescale, we can investigate whether the gas is gravitationally bound or unbound in the outflow regions and molecular disk. A value of unity or lower would suggest that the gas is bound and is able to collapse to form stars. A value much greater than one would suggest that the gas is unbound and therefore at the present time should not be collapsing to form stars. Over the outflow region, we measure a virial parameter

$$\begin{eqnarray}&&{\alpha }_{\mathrm{vir}}=5\displaystyle \frac{{\sigma }^{2}R}{{GM}}\approx 164{\left(\displaystyle \frac{\sigma }{500\mathrm{km}{{\rm{s}}}^{-1}}\right)}^{2},\end{eqnarray} \tag{ 10 }$$

using an average velocity dispersion of $\sigma =500\,\mathrm{km}\,{{\rm{s}}}^{-1}$ seen in the ionized outflow, a radius of R = 3 kpc, and the ionized gas mass ($1.32\times {10}^{9}\,{M}_{\odot }$). Using the molecular gas mass of $6.6\times {10}^{9}\,{M}_{\odot }$, an average velocity dispersion of 270 $\mathrm{km}\,{{\rm{s}}}^{-1}$ seen in the molecular outflow and a radius of 1.6 kpc yields ${\alpha }_{\mathrm{vir}}$ of 20.5. Both of these virial parameter estimates suggest that at the present time, the physical conditions in the ionized and molecular outflow regions and in the molecular disk are stable against gravitational collapse, hindering star formation. In contrast, the molecular gas clump in the star-forming region observed in CO (3−2) yields a virial parameter value of ${\alpha }_{\mathrm{vir}}\approx 0.7$. A value close to unity suggests that star formation should be able to proceed in this region, where there are no powerful outflows.

If the quasar-driven wind does not efficiently cool and adiabatically expands, then the initial driving mechanism is capable of transferring its kinetic energy into mechanical energy in the galactic-scale outflow. The momentum flux of the outflow should therefore be greater than ${L}_{\mathrm{bol}}/{\rm{c}}$ (Zubovas & King 2012). We measure a total momentum flux (${\dot{P}}_{\mathrm{outflow}}=\dot{M}\times v$) of $2.1\times {10}^{37}$ dynes over the entire outflow region. We use the 3000 Å quasar luminosity to compute a bolometric luminosity value of $1.04\times {10}^{47}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ (Runnoe et al. 2012). Comparing the outflow momentum flux to the quasar radiation momentum flux (${L}_{\mathrm{bol}}/c$) of $3.5\times {10}^{36}$ dynes yields a ratio ${\dot{P}}_{\mathrm{outflow}}/{\dot{P}}_{\mathrm{quasar}}$ (loading factor) of 6. In principle, this ratio is a lower limit, since there can still be diffuse molecular gas in the ionized outflow regions whose CO emission is below the sensitivity of ALMA. Based on theoretical work by Faucher-Giguère et al. (2012) and Zubovas & King (2012), the measured loading factor of >6 suggests that the wind is energy conserving. The total kinetic luminosity of the outflow is also about 2% of ${{\rm{L}}}_{\mathrm{bol}}$, consistent with the above theoretical prediction. The observed extended outflow properties of 3C 298 strongly suggest that radiation pressure by itself is unable to drive the outflow. Star formation is also incapable of driving the observed outflow because the expected terminal velocity for a supernova-driven galactic-scale wind is only $\sigma \sim 200\,\mathrm{km}\,{{\rm{s}}}^{-1}$ (Murray et al. 2005), far lower than the observed velocities ($\sigma \sim 500\,\mathrm{km}\,{{\rm{s}}}^{-1}$) seen in the biconical wind.

The outflow could be induced by either the jet and/or BAL winds from the quasar. We infer a jet pressure

$$\begin{eqnarray}&&{P}_{\mathrm{jet}}\approx \displaystyle \frac{{L}_{\mathrm{jet}}\times t}{3{\rm{V}}},\end{eqnarray} \tag{ 11 }$$

of $3\times {10}^{-8}$ dynes cm⁻² using the jet kinetic luminosity of $1.4\times {10}^{47}$ erg s⁻¹, derived using the 1.4 GHz (570 MHz observed) radio flux-jet power relation (Bîrzan et al. 2008; Cavagnolo et al. 2010), assuming spherical volume at a radius of 9 kpc and a timescale of 3 Myr. This is an order-of-magnitude calculation given the approximations in the jet luminosity, the lifetime (t) of the jet, and the volume. Observations of other BAL quasars indicate that the kinetic luminosity of outflows can become as high as 10⁴⁶ $\mathrm{erg}\,{{\rm{s}}}^{-1}$ (Chamberlain et al. 2015), which is close to the kinetic luminosity for the 3C 298 jet and would exert a pressure of 10⁻⁸ dynes cm⁻². This means that a BAL-type wind could potentially drive the extended 3C 298 outflow. However, there is no observed BAL in the optical and infrared spectra of 3C 298. The lack of an observed BAL in 3C 298 could be due to an orientation effect where our line of sight does not overlap with the absorbing outflowing gas. If this is not the case and there is no BAL, the jet of 3C 298 still has the necessary ram pressure capable of inducing a large-scale outflow.

The gas pressure (P = nkT) in the ionized ([S ii]) gas is ∼4 × 10⁻¹⁰ dynes cm⁻². From the ALMA CO (3−2) observations, we measure a molecular gas surface density of 490 ${M}_{\odot }$ pc⁻¹, indicating a molecular gas pressure ($P=(\pi /2)\,G{{\rm{\Sigma }}}_{\mathrm{molecular}}^{2}$) of ∼1 × 10⁻⁹ dynes cm⁻². Both the BAL wind and quasar jet pressures are 2–3 orders of magnitude higher than the current pressure of the ionized and molecular ISM. The fact that the ionized gas pressure is comparable to or lower than the weight per unit area of the molecular gas shows that the jet is not currently producing an overpressure (which would be reflected in the thermal gas pressure of the ionized gas in the ISM) relative to the overburden of the molecular gas. This suggests that the jet is now venting out of the galaxy. The interaction between the ISM and the jet and/or BAL wind in the past was likely in a denser environment, and thus confined to a smaller volume with a higher pressure, to initially generate the outflows.

Both a BAL quasar wind and jet have enough kinetic energy to drive galactic-scale winds. These large outflows in 3C 298 supply the necessary energy and momentum to hinder star formation along their path and expel large amounts of gas from the galaxy. If the sources of turbulence were to halt at the present time, the shocks would dissipate on a dynamical timescale of 3–6 Myr. By that time, the majority of the gas in the ionized outflow will have already escaped into the intergalactic medium (IGM), with most of the molecular gas in the disk removed from the inner few kiloparsec and potentially swept up into the ionized outflow. Any left-over gas in the galaxy would quickly cool back into a molecular state. Taken together, this shows both negative and ejective feedback occurring along the outflow region, thereby impacting the stellar mass growth of the quasar host.