The impact of low luminosity AGN on their host galaxies: A radio and optical investigation of the kpc-scale outflow in MaNGA 1$-$166919

The integrated flux densities of the SE and NW lobes, as measured from the wideband images discussed above, differ significantly between the three observed bands. To better measure how the flux density of these components changes with frequency, we first imaged contiguous subsets of the spectral windows (SPWs) within each band, and then – as in §2.1, and fit the resultant image with two Gaussians to measure the integrated flux density of each lobe. The SPWs were grouped such that there would be a $\gtrsim 3\sigma$ change in the flux density of the fainter NW lobe assuming its continuum radio spectrum in this frequency radio is well-described by a single power law with spectral index $\alpha\sim-0.9$, the value resulting from fitting a power-law the flux densities derived from the wideband images (Table 4). These images were also produced using the casa task tclean, as in §2.1, again using Natural weighting and the same pixel and image size as before, using the “multiscale” deconvolved algorithm (Cornwell, 2008) since the decreased fractional bandwidth of the dataset made an additional spectral term unnecessary. We again used the miriad task imfit to fit the central $12\arcsec\times 12\arcsec$ region of each image with two Gaussians. In these fits, the peak flux, size, and orientation of both ellipses were allowed to vary, The resultant integrated flux densities of both the SE and NW lobes are given in Table 5.

The resultant radio spectrum is shown in Figure 5, with the parameters derived from fitting a single power-law to the integrated flux densities of both the NW and SE radio lobes given in Table 6. As shown in Figure 5, this model does a good job of reproducing the observed flux densities. We also attempted to fit these flux densities with both a broken power-law (as expected if synchrotron cooling is important at higher frequencies) and a power-law with exponential cutoff at lower frequencies (as expected from free-free absorption along the line of sight), but these more complicated models did not produce significantly improved fits to the data.

The technique used to get the spectral index maps assumes that the flux density $S$ at a particular frequency $\nu$ and sky position $(\alpha,\delta)$ can be accurately expressed as:

and then iteratively solves for the value of $S_{\nu_{0}}$ and $\frac{\Delta S_{\nu_{0}}}{\Delta\nu}$ at each location on the sky. As implemented in the casa command widebandpbcor, the derived value of $\frac{\Delta S_{\nu_{0}}}{\Delta\nu}$ is used to calculate the spectral index $\alpha$ within the frequency range of the input data in each pixel of the resultant image.

As shown in Figure 6, in all three bands the spectral index of pixels in the SE and NW are $\alpha\lesssim-0.5$, consistent with the value derived from the analysis described in §2.2.1 (Table 6). However, in all three bands the values of $\alpha$ in the SE lobe are, in general, steeper (more negative) than those in the NW lobe, with a difference in spectral index $\Delta\alpha\sim 0.1-0.2$ (Figure 6) which may not be statistically significant. However, in C-band, this analysis indicates the presence of flat spectrum ($\alpha\gtrsim 0$) radio emission. For example, such a spectral index is measured for the $\sim 3-5\sigma$ peak located W of the two lobes. This suggests this component has a different physical origin than the two lobes which will be discussed in §4.3.

We used the NBursts full spectral fitting package (Chilingarian et al., 2007a, b) to both derive the properties of the stellar population of this galaxy, as well as determine the stellar contribution to its spectrum. This method uses a $\chi^{2}$ minimization algorithm to fit the spectrum in each (spatial) pixel with a model derived from broadening the spectrum predicted from a stellar population model with a Gauss-Hermite parameterized distribution (van der Marel & Franx, 1993) of the line-of-sight velocity at this position. To avoid systematically biasing the resultant parameters, we masked wavelengths corresponding to strong emission lines (e.g. Figure 7). The stellar spectra were chosen from a grid of PEGASE.HR high-resolution simple stellar population (SSP) models (Le Borgne et al., 2004) based on the ELODIE3.1 empirical stellar library (Prugniel et al., 2007) assuming a Salpeter initial mass function (Salpeter, 1955), pre-convolved with the line spread function provided within the MaNGA datacube in order to account for instrumental broadening. While the derived properties of the ionized gas does depend on the choice of stellar models, the high signal-to-noise of our data suggests this will be a small effect (Chen et al., 2018). An example of the results of this procedure is shown in Figure 7.

For each spectrum, this model returns the stellar line-of-sight velocity $V_{\star}$, velocity dispersion $\sigma_{\star}$, and the equivalent stellar age $T_{\rm SSP}$ and metallicity [Fe/H]_SSP of the best-fit SSP. To ensure the derived parameters are reliable, we binned all spaxels with a signal-to-noise ratio (estimated in the stellar continuum spectrum in a narrow 10Å spectral window centered on 5100 Å in the galaxy’s restframe) ${\rm SNR}>1$ into spatial regions with ${\rm SNR}\geq 20$ using the adaptive Voronoi algorithm developed by Cappellari & Copin (2003). As shown in Figure 8, the spatial distribution of stellar velocity $V_{\star}$ is suggestive of a regularly rotating stellar disk – consistent with the spiral morphology inferred from its optical morphology (e.g., 88% of Galaxy Zoo users classified this source as a spiral galaxy; Lintott et al. 2008, 2011). Furthermore, the increased stellar velocity dispersion $\sigma_{\star}$ and stellar age $T_{\rm SSP}$ observed towards the center of galaxy is indicative of a stellar bulge, with a peak velocity dispersion of $\sigma_{\star}\sim 170~{}{\rm km~{}s}^{-1}$ and light-weighted velocity distribution (here we averaged $\sqrt{v_{\star}^{2}+\sigma_{\star}^{2}}$ values within an elliptical aperture of 4″ size using an ellipticity $\epsilon=1-b/a=0.12$ ($b$ and $a$ - minor and major semi-axes) estimated from the optical image isophotes) within the central 4″ of $\sigma_{\star}=161.8\pm 0.4~{}{\rm km~{}s}^{-1}$.

As shown in Figure 8, the distribution of the stellar line-of-sight velocity of the stellar component $V_{\star}$ is indicative of a regular rotating stellar population. We modelled this stellar velocity field by assuming that, for a spaxel located at a particular $(x,y)$ measured relative to the center of the galaxy, the emitting stars have a line-of-sight velocity:

where the azimuthal rotational velocity in the center of the disk (its “galactic plane”) is:

with $V_{\rm sys}$ is the systemic velocity of the galaxy, $g=\sqrt{\sec^{2}i-\cos^{2}\phi\tan^{2}i}$ is a geometrical factor converting the projected sky distance $r\equiv\sqrt{x^{2}+y^{2}}$ between a spaxel at the center of the galaxy to the the distance along the galactic plane $R=gr$, and $i$ is the inclination angle of the disk, $R_{0}$ is a radius where velocity riches a constant maximum value $V_{0}$ in case of $c=0$, and $c$ describes the growth ($c>0$) or decline ($c<0$) of $V_{\phi}$ for $R>R_{0}$. We then determined the values of $V_{\rm sys}$, $V_{0}$, $R_{0}$, $c$, $i$ and the orientation of the galactic disk on the plane of the sky using a $\chi^{2}$ minimization routine.

To measure the properties of emission lines produced by ionized gas in this galaxy, we first substracted the stellar continuum, as derived in §3.1.1, from the observed spectrum in each region, weighting appropriately the contribution of the consituent spaxels. An example of the resulting emission line spectrum is shown in Figure 10. We then estimate the SNR of the resultant emission line spectra in each spaxel using the total flux in the ${\rm H}\alpha$+[N ii] lines, and removed from further analyses all spaxels with ${\rm SNR}<30$. As shown in Figure 11, this requirement primarily excluded spaxels in the outer regions of this galaxy – beyond the observed extent of its radio emission (Figure 3). We then used the Voronoi algorithm developed by Cappellari & Copin (2003) to spatially bin the remaining spaxels into regions with ${\rm SNR}\geq 50$.

As shown in Figure 10, the profiles of the emission lines in a particular spaxel were not always well-described by a single Gaussian. As a result, we modelled the emission line spectrum in each spaxel assuming two Gaussian components. Unfortunately, using a $\chi^{2}$ minimization routine to model the emission line spectrum in each spaxel with two independent Gaussians yielded unreliable results due – in large part – to the degeneracies inherent in this model. As a result, we developed a procedure to fit the emission line spectra in all of the spaxels as the sum of two components:

• •

  a “main” component dominated by regularly rotating gas in the disk of this galaxy, and

• •

  an “outflow” component.

In this decomposition, we accounted for the per-locus disperson, and assumed that the spatial distribution of the kinematic properties of gas in the “main” component are well described by the prescription for a regularly rotating disk given in Equations 2 & 3.

Initial parameters were derived assuming the main component dominated the emission in each spaxels, but final values resulted from simultaneously fitting the emission line spectra for the “main” and “outflow” component, as described below.

We then refit the emission line spectrum in every spatial bin, assuming the profile of each spectral line is described by two Gaussians. The free parameters in this model are the:

• •

  line-of-sight velocity $V_{\rm los}$ and intrinsic velocity dispersion (the observed velocity dispersion of a line $\sigma_{\rm obs}^{2}=\sigma_{\rm gas}^{2}+\sigma_{\rm inst}^{2}$ where $\sigma_{\rm gas}$ is the intrinsic velocity dispersion and $\sigma_{\rm inst}$ is the instrumental resolution) $\sigma_{\rm gas}$ of the emitting gas,

• •

  $H\alpha$ flux,

• •

  Balmer decrement $H\alpha/H\beta$,

• •

  $\log$ [N ii]6584/$H\alpha$,

• •

  $\log$ ([S ii]6717+[S ii]6731)/$H\alpha$,

• •

  $\log$ [O iii]5007/$H\beta$, and

• •

  $\log n_{e}$, as determined from [S ii]6717/[S ii]6731 ratio using the methods described by Osterbrock & Ferland 2006; Proxauf et al. 2014,

of both the “main” and “outflow” components. To determine the values of these quantities in each spatial region, we used the Levenberg-Marquardt minimization method as implemented by the Python-based lmfit package (Newville et al., 2016) to determine the combination of values which minimized the $\chi^{2}$. Furthermore, we required that – for both components – our fits returned values within the following ranges:

• •

  $2.75\leq\frac{\mathrm{H}\alpha}{\mathrm{H}\beta}\leq 10$

• •

  $-1.2\leq\log\left(\frac{[\mathrm{O}\textsc{iii}]5007}{\mathrm{H}\beta}\right)\leq 1.3$

• •

  $-1.0\leq\log\left(\frac{[\mathrm{N}\textsc{ii}]6584}{\mathrm{H}\alpha}\right)\leq 0.4$

• •

  $-0.9\leq\log\left(\frac{[\mathrm{S}\textsc{ii}]6717+[\mathrm{S}\textsc{ii}]6731}{\mathrm{H}\alpha}\right)\leq 0.25$

• •

  $0.475\leq\left(\frac{[\mathrm{S}\textsc{ii}]6717}{[\mathrm{S}\textsc{ii}]6731}\right)\leq 1.425$,

as expected from the physical processes governing these emission lines and observations of large samples of other galaxies (e.g, Baldwin et al. 1981; Osterbrock & Ferland 2006; Proxauf et al. 2014). We further required that, for each spatial region, the fitted value $V_{\rm LOS}$ of the “main” was within $50~{}{\rm km}~{}{\rm s}^{-1}$ of the value for the “main” component derived from the initial analysis described above. Using this procedure, we simultaneously fit for the properties of the “main” and “outflow” contribution to the emission line spectrum in each spaxel. An example of the results from this fitting procedure is shown in Figure 10.

To assess the statistical signficance of the “outflow” component in a given spaxel, we calculated the Bayesian Information Criterion (BIC) statistic (Schwarz, 1978; Liddle, 2007):

where $N_{\rm data}$ is the number of data points, $N_{\rm vars}$ is the number of free parameters in the model, and $\chi^{2}$ is the result from fitting the data with said model, resulting from fitting the emission line spectrum of a given region with a single Gaussian (BIC₁) and two Gaussian (BIC₂). As shown in Figure 11, BIC₁ is substantially higher than BIC₂ in the innermost spaxels, strongly implying that the “outflow” component is significant in these regions. These spaxels are also coincident with the radio emission detected from this galaxy, suggesting a physical connection between the ionized gas “outflow” and radio-emitting plasma. In the spaxels beyond the radio emission, BIC₁ is either slightly larger or smaller than BIC₂ – implying that either the “outflow” component is not present or a marginal fraction of the ionized gas at these locations. As a result, in 2D maps of the parameters of the “outflow” component, we mask spaxels with $\mathrm{BIC}_{1}<{\rm BIC}_{2}$, while those with ${\rm BIC}_{1}-{\rm BIC}_{2}<50$ are shown in transparent color. Furthermore, in the parameter maps of main component, we present values from the one-component fit for spaxels with ${\rm BIC}_{1}\leq{\rm BIC}_{2}$, and values from the two-component fit for those spaxels where ${\rm BIC}_{1}>{\rm BIC}_{2}$.

The difference between the “main” and “outflow” components manifests themselves not only in the statistical significance of the fits, but also in the derived properties of the ionized gas. As shown in Figure 12, the velocity dispersion $\sigma_{\rm gas}$ of the “outflow” component is in general higher than that of the “main” component – even for spaxels with similar line-of-sight velocities. The “V” shape of the “outflow” component on the $V_{\rm LOS}-\sigma_{\rm gas}$ diagram shown in Figure 12 is suggestive of a biconal geometry (e.g., Bae & Woo 2016). Furthermore, as shown in Figure 13, the line ratios measured for the “main” and “outflow” occupy very different regions on the Baldwin, Philips, and Telervich (BPT) diagrams – indicating they are ionized by different mechanisms (Baldwin et al., 1981). The physical implications of both results will be discussed further in §4.1.2.

As described in §2.2, the spectrum of the both the SE and NW radio lobes is well described by a power-law with spectral index $\alpha\approx-0.7$, consistent with optically thin synchrotron emission resulting from relativistic electrons (and positrons) interacting with a magnetic field (e.g., Pacholczyk 1970; Condon & Ransom 2016). In this case, the observed radio luminosity $L_{\rm rad}$ depends on the size $R$, the relativistic electron $u_{\rm e}$, and the magnetic $u_{\rm B}$ energy density of the emitting region. However, since the synchrotron power $P_{\rm syn}$ radiated by an electron of energy $E$ in a magnetic field of strength $B$ is (e.g., Pacholczyk 1970; Rybicki & Lightman 1986):

where $e$ and $m_{e}$ are, respectively, the charge and mass of the electron and $c$ is the speed of light, for a given radio luminosity and size there is not a unique solution for $u_{\rm e}$ and $u_{\rm b}$. However, there is a minimum in the combined (relativistic electron + magnetic field) energy required to power such a source when $u_{e}\equiv\frac{4}{3}u_{\rm B}$ (e.g., Pacholczyk 1970; Rybicki & Lightman 1986; Condon & Ransom 2016. The magnetic field strength $B_{\rm min}$ is (e.g., Pacholczyk 1970; Condon & Ransom 2016):

where $\eta$ is the ion to electron energy ratio, $c_{\rm 12}$ is a “constant” whose value depends on $\nu_{\rm min}$, $\nu_{\rm max}$, and $\alpha$ (for $\nu_{\rm min}=10^{6}~{}{\rm Hz}$, $\nu_{\rm max}=10^{10}~{}{\rm Hz}$, and $\alpha\approx-0.7$ as derived for both the SE and NW radio lobes (Table 6), $c_{12}\approx 10^{8}$ in cgs units (Condon & Ransom, 2016)), and $R$ is the radius of the assumed spherical emitting region, and relativistic particle (electrons + ions) energy (e.g., Pacholczyk 1970; Condon & Ransom 2016):

where $c_{13}$ is another constant whose value depends on $\nu_{\rm min}$, $\nu_{\rm max}$, and $\alpha$ (for $\nu_{\rm min}=10^{6}~{}{\rm Hz}$, $\nu_{\rm max}=10^{10}~{}{\rm Hz}$, and $\alpha\approx-0.7$ as derived for both the SE and NW radio lobes (Table 6), $c_{13}\approx 3\times 10^{4}$ in cgs units (Condon & Ransom, 2016)). Since the 3D geometry of the radio emitting regions are not known, we assume that $R$ for a particular lobe is between physical radius inferred by the smallest deconvolved semi-minor axis and the largest deconvolved semi-major axis derived from our modeling of the wideband radio images of this source (§2.1), as reported in Table 4. The “minimum energy” magnetic field strengths and relativistic particle energies of both the SE and NW lobes inferred from our measurements of their radio morphology (§2.1) and spectrum (§2.2) are reported in Table 8.

With this information, we can estimate the energy of the radio-emitting electrons. The synchrotron emission from an electron with energy $E$ in a magnetic field of strength $B$ peaks at a frequency (e.g., Pacholczyk 1970; Rybicki & Lightman 1986):

As a result, for a particular $\nu_{\rm peak}$ and $B$, the energy of the emitting electron is:

For the observed frequency range of $\nu=1-8~{}{\rm GHz}$ and range of $B_{\rm min}$ given in Table 8, for both lobes the radio emission is dominated by $E\sim 1-4~{}{\rm GeV}$ electrons. The synchrotron cooling time for such particles $t_{\rm cool}$, is $t_{\rm cool}\equiv\frac{E}{P_{\rm synch}}\sim 100-200~{}{\rm Gyr}$ in both lobes for the estimated particle energies and magnetic field strengths. This suggests that radiative cooling plays a minor role in the evolution of the radio emission from the relativistic particles in this outflow.

Furthermore, the synchrotron spectrum of the observed radio emission can be used to determine the spectrum and origin of the GeV-emitting electrons. Optically-thin synchrotron radiation with a power-law spectrum $S_{\nu}\propto\nu^{\alpha}$ is the result of emission from particles with a power-law energy spectrum:

where $\frac{dN}{dE}(E)$ is the number of particles per unit energy at particle energy $E$ and $p$ is the particle index ($p=1-2\alpha$) (e.g., Rybicki & Lightman 1986; Condon & Ransom 2016). For the spectral index $\alpha\sim-0.85$ measured for both the SE and NW lobes (§2.2, Table 6), this suggests $p\sim 2.7$ – the value expected from first order Fermi or Diffusive Shock Acceleration (DSA; e.g. Fermi 1949, 1954). DSA requires that particles cross a shock multiple times (e.g., Bell 1978a, b; Blandford & Ostriker 1978), gaining energy in each shock crossing. The particle spectrum, and the spectral index $\alpha$ of its resultant synchrotron emission, generated from this process is dependent on the Mach number ${\mathcal{M}}$ of the shock, where (e.g., Berezhko & Ellison 1999; Guo et al. 2014; Di Gennaro et al. 2018):

The spectral index $\alpha\sim-0.8-(-0.9)$ observed from both the SE and NW lobes (§2.2, Table 6) suggests the emitting particles in both components are accelerated in shocks with ${\mathcal{M}}\sim 2-3.5$.

However, such shocks should also efficiently accelerate ions to high energies (e.g., Guo et al. 2014). Recent simulations (e.g., Park et al. 2015) and observations of particles accelerated in shock with similar Mach numbers ${\mathcal{M}}$ (e.g., SNR 5.7$-$0.1; Joubert et al. 2016) suggest the ion to electron energy ratio $\eta>100$. If that also occurs in this galaxy, then minimum total relativistic particle energies in the SE and NW lobes is $\sim 10\times$ higher than the values given in Table 8, or on the order of $\sim 10^{55}~{}{\rm ergs}$. Regardless of the true value of $\alpha$, the larger energy in the relativistic component of this outflow suggests it can have a significant impact on the host galaxy (e.g., Mao & Ostriker (2018); Hopkins et al. (2020b)).

As described in §4.1.1, the radio emission observed from MaNGA 1-166919 is believed to be produced by electrons accelerated by a shock propagating through this galaxy. However, theoretical work suggests that $\lesssim 10\%$ of the shocked material is accelerated to relativistic energies (e.g., Caprioli & Spitkovsky 2014; Caprioli et al. 2015), with the bulk of the material heated to a temperature $T_{\rm shock}$ (e.g., Faucher-Giguère & Quataert 2012; Caprioli & Spitkovsky 2014):

where $m$ is the mass of the particle, $k$ is Boltzmann’s Constant, and $v_{\rm shock}$ is the velocity of the shock relative to the surroundings. If $v_{\rm shock}$ is high enough, a copious amount of UV and soft X-ray photons will be generated at the shock front (e.g., Raymond 1976; Allen et al. 2008). The spectra from material photoionized by this radiation are expected to have emission line ratios (e.g., Dopita & Sutherland 1995a; Allen et al. 2008) which lie within the Low Ionization Excitation Region (LIER) of the [S ii] BPT diagram (Kewley et al., 2006) (in the literature, this emission is often referred to arise from a “Low Ionization Nuclear Excitation Region” (LINER) since they were first and primarily identified in the centers of galaxies (e.g., Heckman 1980; Heckman et al. 1981). However, subsequent work has found that such emission can be detected throughout a galaxy (e.g., Belfiore et al. 2016), and therefore use the more general term). Indeed, as shown in Figures 13 & 16, the emission line ratios of the “outflow” material largely fall within the LIER region of such a diagram. Furthermore, as shown in Figure 16, LIER-like emission is only detected in the center of this galaxy, and predominantly found in the “outflow” component of the ionized gas. Therefore, it seems likely the ionized gas “outflow”, as inferred from our analysis of the MaNGA (§3.1.2) and GMOS (§3.2) data, is dominated by material photo-ionized by the shock which also accelerates the relativistic electrons responsible for the observed radio emission. However, in many galaxies, such line ratios are instead thought to result from photo-ioninzation by post-AGB stars (e.g., Belfiore et al. 2016; Singh et al. 2013; Yan & Blanton 2012). Such stars are expected to be prevalent in older ($\gtrsim 1~{}{\rm Gyr}$) stellar populations – as inferred for the central regions of this galaxy ($T_{\rm SSP}\sim 3$ Gyr; Figure 8) from our derivation of its stellar population as described in §3.1.1.

It is possible to distinguish between these models by measuring the kinematics of the putative “outflow” component of the ionized gas in this galaxy For the GMOS data, we estimated this by subtracting the line-of-sight velocity measured in a particular spaxel (left panel of Figure 15) with that predicted by our model for the regular rotating gas in this galaxy (right panel of Figure LABEL:fig:main_rot) while, for the MaNGA data, we calculated the difference in line-of-sight velocity between the “main” and “outflow” components ($\Delta V_{\rm gas,OF}$) as derived from the modelling described in §3.1.2. As shown in Figure 17, the relative line-of-sight velocities show a clear spatial separation of “red” and “blue” components – strongly suggesting a biconical “outflow”. This geometry is consistent with the $V_{\rm LOS}-\sigma_{\rm gas}$ of the “outflow” component (Figure 12). Furthermore, the correspondence between the kinematics of the “outflow” ionized gas and the SE and NW radio “lobes”, which is particularly evident in the higher angular resolution GMOS data (left panel, Figure 17), strongly suggests a physical connection between the two. Since the radio emission is produced by shock-accelerated particles, we therefore conclude these shocks are indeed responsible for producing the LIER-like emission observed from the ionized gas. This conclusions is further supported by the observed dependence between $|\Delta V_{\rm gas,OF}|$ and the line-ratios of the “outflow” gas. As shown in Figure 16, spaxels with higher values of $|\Delta V_{\rm gas,OF}|$ typically falling above and to the right of spaxels with lower $|\Delta V_{\rm gas,OF}|$ on the [S ii] BPT diagram. This trend is similar to that predicted by models for the emission of material photoionized by shock heated gas, which find that their location on the [S ii] BPT diagrams moves up and to right as the shock velocity $v_{\rm shock}$ increases (e.g., Allen et al. 2008).

If correct, then the relative line-of-sight velocity between the “main” and “outflow” component provides a lower limit on $v_{\rm shock}$, since this quantity is not sensitive to differences in velocity between these components in the plane of the sky. As a result, we estimate $v_{\rm shock}\gtrsim 100~{}{\rm km~{}s}^{-1}$ for the NW lobe and $v_{\rm shock}\gtrsim 200~{}{\rm km~{}s}^{-1}$ for the SE lobe – sufficient to photoionize substantial amounts of material both “downstream” (post-shock) and “upstream” (pre-shock) of the shock (e.g., Dopita & Sutherland 1995b; Dopita et al. 1996; Raymond 1976; Wilson & Raymond 1999). Furthermore, the observed geometry, and differences in extent and shock velocity between the two components of the outflow, are consistent with that expected from outflows resulting from high-velocity material ejected from an AGN interacting with a clumpy interstellar medium (e.g., Mukherjee et al. 2018; Nelson et al. 2019). In fact, simulations suggest that the bi-conical geometry of this outflow is the natural consequence of a central outflow being confined by the disk of a galaxy (e.g., Wagner et al. 2012). In summary, both the relativistic and ionized component of this outflow appears to be the result of shocks driven into the surrounding ISM by a central engine.

In §4.1.1, we presented our measurements for the energy contained in the relativistic component of this outflow. However, this outflow also consists of several non-relativistic components, including ionized gas, atomic gas, and molecular material. In this section, we use the emission line spectra derived from the MaNGA (§3.1.2) and GMOS (§3.2) to measure the properties of its ionized component. While studies of similar outflows in other galaxies suggest that atomic and molecular material may constitute the bulk of the entrained mass (e.g., Oosterloo et al. 2017), currently the observational data needed to measure the properties of these components in this galaxy are not available.

The mass of the ionized gas in this outflow $M_{\rm out}$, can be estimated as (e.g., Soto et al. 2012; Baron et al. 2017):

where $m_{\rm H}$ is mass of the Hydrogen atom, $\mu$ is the average atomic number of the emitting material (assumed to be Solar, such that $\mu\equiv 1.4$), $V$ is volume of the emitting region, $f$ is the filling factor, and $n_{\rm e}$ is the number density of electrons. The H$\alpha$ luminosity of this region is equal to (e.g. Baron et al. 2017):

where $\gamma_{\rm H\alpha}$ is the H$\alpha$ emissivity of the ionized plasma. In the case of highly ionized material with an electron temperature $T_{e}\approx 10^{4}$ K and optically thick to Lyman line emission (“Case B”; e.g. Baker & Menzel 1938; Burgess 1958), $\gamma=3.56\times 10^{-25}$  erg cm³ s^-1 (Osterbrock & Ferland, 2006). As a result, we can calculate $M_{\rm out}$ by evaluating:

Therefore, to calculate this quantity, we first need to determine $n_{e}$ in the outflow as well as correct the observed H$\alpha$ emission for extinction along the line of sight.

We estimate $n_{\rm e}$ using the observed ratio of the [S ii]6717/[S ii]6731 emission lines (Osterbrock & Ferland, 2006; Proxauf et al., 2014) separately for the “outflow” component of the MaNGA emission line spectrum and the GMOS data. As shown in Figure 18 for the MaNGA “outflow”, there are considerable variations in this parameter between adjacent spaxels due to the weakness of these lines and/or complexity of the line decomposition in many spaxels. We therefore calculated an average $n_{\rm e}$ for the outflow in each dataset, weighting the value in each spaxel by the total H$\alpha$+[N ii] flux of its outflow component. This yields a weighted average of $\log n_{e}\approx 1.9$ ($\approx 80$ cm^-3) in the MaNGA data and $\log n_{e}\approx 2.0$ ($\approx 100$ cm^-3) in the GMOS data, similar to the value inferred for outflows in other galaxies (e.g., Harrison et al. 2014; Karouzos et al. 2016) as well as in the previous analysis of this galaxy by Wylezalek et al. (2017) (Table 9). However, this method for estimating the electron density preferentially returns values $10~{}{\rm cm}^{-3}\lesssim n_{e}\lesssim 10^{4}~{}{\rm cm}^{-3}$ (e.g., Proxauf et al. 2014). Therefore, the central regions of the outflow where we estimate $n_{\rm e}\sim 10~{}{\rm cm}^{-3}$ may have a true density below this value. Furthermore, the low ionization of the emitting gas suggests there is a significant neutral component to this material, whose density is not measured using this technique. As a result, the total (neutral and ionized) gas density is likely to be significantly higher than the estimated value of $n_{e}$ (e.g., Dempsey & Zakamska 2018). Lastly, recent results suggests that the electron density estimated using [S ii] are systematically lower than those using other emission lines (e.g., Davies et al. 2020) unfortunately not detected with sufficient spectral resolution or low signal-to-noise in our data.

To estimate the extinction along the line-of sight towards the outflowing material in both datasets, we use the Balmer Decrement (the H$\alpha$/$H\beta$ flux ratio) of the outflow component measured in the MaNGA data. This is because, as described in §3.2, H$\beta$ is not detected in the GMOS data. As shown in Figure 18, this quantity varies significantly, and for the MaNGA outflow corrected the H$\alpha$ flux of the outflow in each spaxel with the corresponding Balmer Decrement measured for this component. The differing angular resolutions of the MaNGA and GMOS data precludes us from making a similar spaxel by spaxel correction. As a result, we estimate the average extinction towards the outflow in the GMOS data from the MaNGA data in two ways: the mean value, $\langle H\alpha/H\beta\rangle=6.8$, and the H$\alpha$+[N ii] flux weighted averaged value $\langle H\alpha/H\beta\rangle_{\mathrm{H\alpha+NII}}=8.6$. In all cases, we use Balmer Decrement to correct the H$\alpha$ flux using the extinction law derived by Cardelli et al. (1989), with the results shown in Figure 19.

With these measurements of $\log n_{e}$ and extinction corrected H$\alpha$ flux in hand, it is now possible to measure the total mass of outflowing ionized gas in this galaxy. For the MaNGA data, we do so by adding together the mass estimated in each spaxel (Figure 18), deriving a total mass of $M_{\rm out}\sim 2.4\times 10^{7}~{}{\rm M}_{\odot}$. For the GMOS data, we use the total H$\alpha$ flux measured in the “outflow” region – which, as described in §3.2, corresponds to those spaxels with $\sigma_{\rm gas}>100~{}{\rm km~{}s}^{-1}$, or the central $r_{\mathrm{OF}}\approx 1\arcsec\approx 1.4$ kpc (Figure 15) of this galaxy. In this case, we estimate $M_{\rm out}\sim(2.2-3.9)\times 10^{7}~{}{\rm M}_{\odot}$ – in good agreement with the value derived from the MaNGA data alone.

Furthermore, we can estimate the total velocity of the outflow $v_{\rm out}$ as (e.g., Karouzos et al. 2016):

where the value of these quantities as measured from the GMOS data is shown in Figure 15 and as derived for the “outflow” component in the MaNGA emission line spectrum is shown in Figure 20. Both data sets give similar values of $v_{\rm out}$, with the ionized gas moving near $\sim 400$ km s^-1 near the center of the galaxy and slowing to values of $\sim 200-300$ km s^-1 near the edge of the radio emission.

With this velocity information, we can calculate the kinetic energy and “age” of the ionized gas in this outflow. We determine the kinetic energy $K_{\rm ion}$ of the ionized gas in the MaNGA data by evaluating $K=\frac{1}{2}M_{\rm ion}v_{\rm out}^{2}$ in each spaxel (Figure 21), and add together the values to measure a total $K_{\rm ion}=2.4\times 10^{55}~{}{\rm ergs}$ in the MaNGA data and $K_{\rm ion}=(4.6-8)\times 10^{54}$ erg for the GMOS data. These energies are comparable to the minimum energy estimated for the relativistic content of this outflow (Table 8, §4.1.1), suggesting these two components are in rough equipartition. Furthermore, we estimate the age of the MaNGA outflow in each spaxel as:

where $R$ is the projected physical separation between the spaxel and the center of the galaxy. As shown in Figure 21, this suggests the outflow is $\sim 6$ Myr old. For the GMOS data, the $\sim 1.4$ kpc extent of the outflow coupled with the H$\alpha$+[N ii] flux weighted average outflow velocity $v_{\mathrm{out}}$ value of 222 km s^-1 suggests $t_{\mathrm{age}}=6.2$ Myr – consistent with the results derived from the MaNGA data. As shown in Table 9, this suggests a mass outflow rate of $\sim 4~{}{\rm M}_{\odot}\,{\rm yr}^{-1}$ and a kinetic power of $\dot{E}_{\rm kin}\sim(0.2-1)\times 10^{41}~{}{\rm ergs~{}s}^{-1}$.