Jet-driven outflows of ionized gas in the nearby radio galaxy 3C 293

Abstract

Fast outflows of gas, driven by the interaction between the radio jets and interstellar medium (ISM) of the host galaxy, are being observed in an increasing number of galaxies. One such example is the nearby radio galaxy 3C 293. In this paper we present integral field unit observations taken with OASIS on the William Herschel Telescope, enabling us to map the spatial extent of the ionized gas outflows across the central regions of the galaxy. The jet-driven outflow in 3C 293 is detected along the inner radio lobes with a mass outflow rate ranging from ∼0.05 to 0.17 M_⊙ yr⁻¹ (in ionized gas) and corresponding kinetic power of ∼0.5–3.5 × 10⁴⁰ erg s⁻¹. Investigating the kinematics of the gas surrounding the radio jets (i.e. not directly associated with the outflow), we find linewidths broader than 300 km s⁻¹ up to 5 kpc in the radial direction from the nucleus (corresponding to 3.5 kpc in the direction perpendicular to the radio axis at maximum extent). Along the axis of the radio jet linewidths >400 km s⁻¹ are detected out to 7 kpc from the nucleus and linewidths of >500 km s⁻¹ at a distance of 12 kpc from the nucleus, indicating that the disturbed kinematics clearly extend well beyond the high surface brightness radio structures of the jets. This is suggestive of the cocoon structure seen in simulations of jet–ISM interaction and implies that the radio jets are capable of disturbing the gas throughout the central regions of the host galaxy in all directions.

ISM: jets and outflows, galaxies: individual: 3C 293, galaxies: ISM, galaxies: jets, radio lines: galaxies, radio lines: ISM

1 INTRODUCTION

In order to reproduce the tight correlations observed between the supermassive black hole (SMBH) and the host galaxy, active galactic nuclei (AGN)-driven outflows have become a key ingredient in galaxy formation and evolution models (see e.g. Silk & Rees 1998; Di Matteo, Springel & Hernquist 2005; Bower et al. 2006; Croton et al. 2006; Ciotti, Ostriker & Proga 2010; Fabian 2012). Fast outflows are generally invoked to regulate the growth of the galaxy by halting the accretion of gas on to the central AGN and prohibiting star formation. There are a number of physical mechanisms thought to be responsible for driving these outflows such as starburst-driven winds (see review by Veilleux, Cecil & Bland-Hawthorn 2005) or radiation pressure from the accretion disc (e.g. King 2005; Zubovas & King 2014). However, there is increasing evidence that in radio-loud AGN the interaction between the radio jets and interstellar medium (ISM) of the host galaxy can also be responsible for large outflows. Massive and fast outflows of neutral or molecular gas driven by the radio jets have been observed in a growing number of objects (Morganti et al. 2005, 2013, 2015; Lehnert et al. 2011; Dasyra & Combes 2012; McNamara et al. 2014).

In the optical, jet-induced outflows of ionized gas have also been observed (Emonts et al. 2005; Morganti et al. 2005, 2007; Holt et al. 2011). While many of these studies used long-slit observations, the increasing availability of integral field units (IFUs) means we can now trace the impact of these outflows in 2D (Davis et al. 2012; Harrison et al. 2014). Pinpointing the location and spatial extent of these outflows enables us to derive crucial parameters, such as the mass outflow rates and kinetic energy involved, which we can compare to predictions from galaxy evolution simulations. Current theoretical models of jet–ISM interaction (Wagner & Bicknell 2011; Wagner, Bicknell & Umemura 2012) predict that these fast outflows are produced by shocks, which ionize the surrounding gas. As the gas then cools in the post-shock region, outflows in H i and CO are observed. Determining whether ionized and neutral gas outflows originate from the same location in the galaxy will therefore provide important constraints for these models.

One such example of jet-induced outflows is observed in 3C 293, a nearby radio galaxy at a redshift of z = 0.045 (Sandage 1966). 3C 293 is a ‘double–double’ radio galaxy, where the older, outer lobes have a projected linear size of ∼190 kpc and the younger, inner radio lobes are about 1.7 kpc across (Akujor, Leahy & Garrington 1996; Beswick et al. 2004; Joshi et al. 2011). Fast outflows in 3C 293, up to 1400 km s⁻¹, were first detected in neutral hydrogen using the Westerbork Synthesis Radio Telescope (Morganti et al. 2003). However, the spatial resolution was not high enough to determine where along the inner ∼1 kpc jets the outflows originated from. To investigate this further, long-slit spectroscopy was taken along the radio jet axis where outflows of ionized gas were detected associated with the eastern radio lobe. The profile of the H i outflows and the ionized gas outflows was remarkably similar and it was therefore concluded that the H i outflows were associated with the same jet–ISM interaction (Emonts et al. 2005, hereafter EM05).

Recent observations using the Karl G. Jansky Very Large Array (VLA) provided higher spatial resolution data needed to localize the H i outflows on arcsecond scales. Unexpectedly, these data show that the H i outflows are coming from the western radio lobe, opposite to where the most extreme ionized outflows are detected (Mahony et al. 2013).

In addition to the detection of outflows of cool gas, 3C 293 also contains a large amount of warm molecular hydrogen, possibly heated by shocks driven by the interaction between the radio jets and the ISM (Ogle et al. 2010). To confirm this, follow-up X-ray observations revealed 10⁷ K shock-heated gas in both the nucleus and the inner radio jets, again driven by jet-induced shocks (Lanz et al. 2015). Studies of the cold molecular gas in this galaxy indicate a cold H₂ mass of 2.2 × 10¹⁰ M_⊙ and a star formation rate of 4 M_⊙ yr⁻¹ (Evans et al. 1999; Labiano et al. 2014). The outflows detected in H i and ionized gas have not yet been detected in CO, however, Labiano et al. (2014) derive a 3σ upper limit of the molecular gas outflow of 3.2 per cent of the total H₂ mass. This limit can rule out an extreme molecular outflow like that observed in Mrk 231 (Feruglio et al. 2010), but it is within the range observed in other objects with molecular outflows (e.g. Alatalo et al. 2011; Morganti et al. 2015).

In this paper, we present IFU observations of 3C 293 in order to confirm the location of the fast outflow of ionized gas and unambiguously determine its location with respect to the fast outflow seen in H i. IFU observations allow us, for the first time, to map the properties of the ionized gas outflow across the inner regions of 3C 293. Section 2 describes the observations and data reduction and the results are presented in Section 3. In Section 4 we discuss the properties of the outflow and how these new observations compare with previous results before concluding in Section 5.

2 OBSERVATIONS AND DATA REDUCTION

Observations were carried out using the OASIS IFU on the William Herschel Telescope (WHT) on 2013 April 9. Two different spectral configurations were used to target the Hα and [N ii] lines (MR661) and [S ii] lines (MR735) separately. For each configuration, six exposures of 900 s were taken with four of the exposures offset by 2 arcsec in a rectangular pattern. Fig. 1 shows the region covered by the OASIS observations (the blue boxes show the OASIS 10 × 7 arcsec² field of view) overlaid on an optical V-band image (Emonts et al., in preparation) with the 1.4 GHz radio contours of the inner radio jet shown in red. This exposure pattern was chosen to slightly expand the area observed whilst still maintaining the highest signal-to-noise ratio (S/N) for the central 8 × 3 arcsec² region, coincident with the radio jets. Since the angular size of 3C 293 is much larger than the field of view, each exposure was followed by a separate sky exposure of the same duration (900 s) in order to perform adequate sky subtraction. The average seeing throughout the observations was 0.8 arcsec.

Figure 1.

Optical V-band image of 3C 293 observed with the Hiltner 2.4-m telescope of the Michigan–Dartmouth–MIT (MDM) observatory (Emonts et al., in preparation). The blue boxes show the position of each exposure; two were centred on the nucleus of 3C 293 and four exposures were offset by 2 arcsec in a rectangular pattern. The red contours show the inner radio lobes observed at 1.4 GHz with the Karl G. Jansky VLA. The contour levels range from 0.1 to 1.25 Jy in steps of 0.25 Jy. North is up and east to the left as indicated by the arrows in the top left-hand corner.

The majority of the data reduction was carried out using the xoasis data reduction pipeline with the exception of the sky subtraction, flux calibration and mosaicking which were done using standard procedures in idl. The data were then resampled to 0.3 arcsec pixels and continuum subtracted. In order to detect the faint, broad signatures indicative of an outflow, the data were binned using a Voronoi binning algorithm (Cappellari & Copin 2003) to ensure there was a minimum S/N of 10 for every bin. This results in bin sizes which vary from 0.3 × 0.3 arcsec² in the central regions to 3.6 × 2.7 arcsec² in the outer regions.

In order to remove the effects of instrumental broadening, the widths of the sky lines were measured prior to the sky subtraction and this value (6.8 Å for MR661 and 7.1 Å for MR735) used to deconvolve the measured linewidths. The astrometry was checked by cross-matching with Two Micron All Sky Survey (2MASS) images (Skrutskie et al. 2006) and we estimate that the positional error between the radio and optical data is less than 0.5 arcsec.

2.1 Gaussian fitting

The binned, continuum-subtracted cubes of each grism were stitched together and run through a Gaussian line-fitting routine.¹ Combining the data allowed us to fit the Hα, [N ii] and [S ii] lines simultaneously providing better constraints on the fitting. The linewidths of all the lines were required to be the same and the ratios of the two [S ii] lines were required to fall between the high- and low-density limits (i.e. 0.44–1.44).

Many of the spectra exhibited broad wings that were not well fit by a single Gaussian component. In these cases, an additional Gaussian component was added to the model. Fig. 2 shows the regions in which a two-Gaussian fit was required and some example spectra are shown in Fig. 3.

Figure 2.

The number of Gaussian components needed to fit the data; the blue bins represent pixels where a single Gaussian component was sufficient to fit the data and the red denotes regions where two Gaussian components were required. The black contours show the inner radio lobes observed at 1.4 GHz with the VLA (Mahony et al. 2013). The contour levels range from 0.1 to 1.25 Jy in steps of 0.25 Jy and are the same contours shown in all the following figures. North is up and east to the left as indicated by the arrows in the top left-hand corner.

Figure 3.

Example spectra extracted from regions associated with the two hotspots in 3C 293. The top row shows the [S ii] λ6716 and [S ii] λ6737 lines observed with OASIS (the [S ii] λ6716 line was used as the rest wavelength when converting to velocity). The observed data are denoted by the black histogram while the solid magenta line shows the total of the Gaussian components fit to the data. In both instances a second, broader Gaussian component (shown by the dashed blue line) was required to fit the broad wings observed. The bottom row shows the corresponding H i absorption spectrum (inverted for comparison) from previous VLA observations (Mahony et al. 2013).

3 MAPPING THE IONIZED GAS OUTFLOW IN 3C 293

In regions where two Gaussian components were needed to fit the data, the second Gaussian component was much broader than the first component fitted. As such, the two components were separated to distinguish between the gas associated with the galaxy disc, what we term the ‘narrow’ component, and the ‘broad’ component indicative of more extreme kinematics. We treat these two components separately in the following analysis.

3.1 Ionized gas kinematics

Fig. 4 shows the velocity maps for the narrow components in the left-hand panel, and the broad components on the right. The velocity map of the narrow components shows a regularly rotating disc in agreement with previous studies (Haschick & Baan 1985; Evans et al. 1999; Beswick et al. 2004, EM05; Labiano et al. 2014), providing confidence that the Gaussian line-fitting routine produces sensible fits to the data.

Figure 4.

Velocity maps of the narrow Gaussian components, representing the gas associated with the disc of the galaxy (left), and broad components, representative of the outflow (right). The broad component associated with the western jet is close to the systemic velocity in this region, similar to the central velocity of the H i outflow which is also detected here (Mahony et al. 2013). At the eastern jet the broad component is significantly blueshifted, in agreement with previous long-slit observations (EM05). For the colour plots we refer the reader to the online version of this journal.

Focusing now on the broad components (right-hand panel of Fig. 4) we see that the gas is significantly blueshifted (with outflow velocities ranging from −300 to −500 km s⁻¹) in regions associated with the eastern radio jet, indicative of an outflow. This is in agreement with the results obtained from the previous long-slit observations of 3C 293 where a clear blueshifted outflow of ionized gas was detected at the eastern jet (EM05). Regions coincident with the western radio jet also show evidence for blueshifted velocities, but these are less extreme than those associated with the eastern jet, with outflow velocities ranging from −50 to −150 km s⁻¹.

The linewidths measured for the broad components indicate very high velocity dispersions of the outflowing material all along the inner radio jet. The broadest line is detected at the eastern hotspot which has a full width at half-maximum (FWHM) of 1755 ± 141 km s⁻¹ while for the western jet the FWHM of the outflow ranges from 1080 ± 70 to 1354 ± 79 km s⁻¹. Fig. 3 shows the comparison between the broad components detected in the ionized gas compared to the corresponding H i absorption spectrum associated with each hotspot. While the broad components associated with the eastern jet have larger velocity shifts and linewidths, the broad components fit to the spectrum of the western jet have a much higher peak flux.

Given the blueshifted velocities and the broad linewidths observed, we interpret the broad component as representing a fast outflow, in line with previous studies of 3C 293 (EM05; Mahony et al. 2013). These earlier investigations have led to seemingly conflicting results on where the outflow is located. The outflow detected in the neutral gas has only been detected in front of the western inner radio lobe, but previous long-slit observations have detected the most extreme ionized gas outflow associated with the eastern radio jet. As suggested in EM05, Figs 2 and 4 show that the ionized gas outflow is detected all along the radio jet axis, providing further evidence that this outflow is being driven by the radio jet. It appears that the outflow is detected preferentially in the northern regions of the radio lobes, however, this is most likely due to the higher S/N observed here. The southern regions are more obscured by the dust lane of 3C 293 making it more difficult to detect the faint, broad wings.

The linewidths of the narrow components are roughly uniform across the field of view, ranging from ∼300 up to 450 km s⁻¹. These values are higher than usually seen in quiescent gas, which suggests that we are tracing unsettled gas out to a few kpc from the nucleus. However, to further investigate this claim higher spectral resolution data are needed. To this effect we have compared the IFU data with higher spectral resolution long-slit data.

3.2 Comparison with long-slit data

The long-slit data were taken using the Intermediate dispersion Spectrograph and Imaging system (ISIS), mounted on the WHT on the nights of 2006 May 27 and June 29. These data were taken with the primary aim of measuring the kinematics of the Na i absorption doublet in order to further investigate the neutral gas kinematics (see Bessiere et al., in preparation), but they are also useful for checking the IFU results on the ionized gas kinematics, since their spectral resolution is a factor >3.5 times higher. ISIS allows for the use of a dichroic filter, thus facilitating simultaneous observing in both the blue and red. The standard 5300 dichroic was used with the R300B grating for the blue arm, the R600R for the red arm and a 1-arcsec slit. We observed 3C 293 at two position angles (PAs), PA of 93 comprises 6 × 900 s exposures whilst PA of 225 consists of 6 × 1200 s exposures. The average seeing throughout the observations made at PA of 93 was 0.5 arcsec, whilst those made at PA of 225 had an average seeing of 0.8 arcsec.

These data were reduced using packages within the iraf environment and also the starlink package figaro. The data were bias subtracted, flat-fielded and wavelength calibrated in the usual manner. The arc exposures were taken immediately before and after the observations at the same positions as the target. To ensure an accurate relative flux calibration between the blue and the red arm, several standard stars were taken on each night and individual flux calibration curves in both the blue and the red were produced for each star. For each night, the flux calibration curves were compared with one another, and any obvious outliers were removed. The individual curves were then combined into a master blue and a master red flux calibration curve for that night. The final reduced data cover the wavelength range 3300–7070 Å in the observed frame, with a spectral resolution of 3.7 Å in the blue and 1.8 Å in the red.

Apertures of either 0.2 or 0.4 arcsec were extracted along the slits. The extracted spectrum was then continuum subtracted and Gaussian fits were made using the starlink dipso package. The best fit for both the [S ii] and the Hα, [N ii] lines were used to determine the velocity shifts and linewidths for each aperture. These are shown in Fig. 5. In some cases it was possible to fit the profiles with a single Gaussian, but for most of the spectroscopic apertures a double Gaussian was required. Consistent with the results of EM05, along the inner radio axis (PA93) there is evidence for extreme kinematics in the form of large linewidths (FWHM > 400 km s⁻¹) found out to a radial distance of 7 kpc to the east and west of the nucleus – significantly beyond the high surface brightness inner radio structures. The results also show evidence that, where double Gaussians were required to fit the lines, the broad components are strongly blueshifted (ΔV > 200 km s⁻¹) relative to the narrow components. In addition, these observations, which have a higher sensitivity and spectral resolution than those of EM05, detect a region with disturbed kinematics at a radial distance of 12 kpc to the west. In this region, the linewidths reach FWHM > 500 km s⁻¹.

Figure 5.

Linewidths of the Gaussian components fitted in the long-slit data as a function of distance from the nucleus of 3C 293. The left-hand figure shows the data extracted along the axis of the disc of the galaxy (position angle of 225°) and the right-hand figure shows the linewidths measured along the axis of the radio jets (position angle of 93°). For apertures where a single Gaussian component fit was adequate to fit the emission lines the measured linewidths are shown by the triangles. In regions where two Gaussian components were fit to the data, these have been separated into narrow components (marked by the crosses) and broad components (circles). For the slit placed along the radio jet axis we have also marked features of the inner radio jet structure using the same notation as EM05. The brightest hotspots correspond to E0 and C while E1 and W1 refer to diffuse jet emission detected beyond the hotspots (see Beswick, Pedlar & Holloway 2002). Positive offsets are west of the nucleus for both slits.

Based on the long-slit data, the evidence for disturbed emission line kinematics is not only confined to the radio axis. The results for PA225 presented in Fig. 5 demonstrate that relatively broad lines (FWHM > 300 km s⁻¹) are detected out to a radial distance from the nucleus of 3.5 kpc to the north-east and 5.0 kpc to the south-west; at maximum extent, the broad lines are detected out to a distance of 3.5 kpc in the direction perpendicular to the radio axis.² Such large linewidths in the off-nuclear regions are inconsistent with the normal gravitation motions of gas in the disc of a galaxy (i.e. Labiano et al. 2014 show that the kinematics of the inner 7 kpc CO disc are consistent with rotation around the core). Thus, the long-slit results confirm what was seen in the IFU observations in the sense that they show that the disturbed emission line kinematics have a significant extent perpendicular to the radio axis.

One possible explanation for this is that we are seeing the cocoon structure described in simulations of jet-driven feedback (Wagner & Bicknell 2011; Wagner et al. 2012). As the jets propagate through the inhomogeneous ISM, they not only impact the ISM along the radio jet axis, but also create a spherical bubble which drives dense clouds outwards in all directions as the bubble expands. Although this mechanism cannot impart the gas with the extreme velocities required to escape the galaxy, it does disturb and heat the gas sufficiently to inhibit star formation (Nesvadba et al. 2010; Guillard et al. 2012). This turbulent behaviour is similar to what is observed in a growing number of objects exhibiting jet-driven outflows such as IC 5063, NGC 1266 and 3C 326 (Alatalo et al. 2014; Guillard et al. 2015; Morganti et al. 2015).

3.3 Line ratios

Having fit the Hα, [N ii] and [S ii] lines simultaneously we can also investigate some line ratio diagnostics. Combining the [N ii]/Hα, [S ii]/Hα line ratios from the OASIS data with the [O iii]/Hβ ratios from EM05, we find that 3C 293 can be classified as a LINER (see also EM05). The excitation mechanism for LINERs is still not well understood, but in the case of 3C 293 our observations can be explained both by dusty AGN and by (pure) shock models (Groves, Dopita & Sutherland 2004; Allen et al. 2008). Given the morphological association of the higher excitation region with the inner radio jets in 3C 293 we favour the shock models over the AGN photoionization models. In Fig. 6 we show the [N ii]/Hα, [S ii]/Hα line ratios for regions where only a narrow component was fit to the data compared to regions where two Gaussian components were needed. There is a clear increase in the line ratios in the regions where we detect outflowing material (i.e. where both a narrow and broad component were fit to the data). In terms of shock models this increase reflects the increase in shock velocity (Dopita & Sutherland 1995; Allen et al. 2008). This picture is further supported by the increased linewidths observed in the radio region (Dopita & Sutherland 1995).

Figure 6.

Comparison of the [S ii]/Hα line ratio against the [N ii]/Hα line ratio. To avoid errors associated with fitting multiple Gaussian components to blended emission lines, we have calculated the ratios using the total [S ii] λ6716 + λ6731 and [N ii] λ6548 + λ6583 fluxes. The line ratios for regions where only narrow components were fitted are shown in the black circles while the red diamonds show the line ratios calculated by adding both the narrow and broad components.

4 PROPERTIES OF THE OUTFLOW

4.1 Geometry of the outflow

With the addition of the IFU data we can now present a more unified picture of the kinematics of the outflow in 3C 293. Previous studies have concluded that, for the inner radio jets, the eastern jet is approaching (Akujor et al. 1996; Beswick et al. 2004; Floyd et al. 2006). This agrees with the fact that we observe the outflow at the eastern jet to be significantly blueshifted. At the western jet, although the outflow velocities are still blueshifted (as shown in Fig. 4), they are closer to the systemic velocity, similar to the outflow detected in H i (Mahony et al. 2013). Since the western jet is more embedded in the disc of the galaxy it is likely that any redshifted emission associated with this jet is obscured. However, EM05 reported a redshifted outflow component detected in the region beyond the western hotspot consistent with this orientation.³

In Fig. 7 we show a 3D cartoon representation of what the inner regions of 3C 293 may resemble. The disc of the galaxy is orientated such that the eastern side of the disc is towards us (Floyd et al. 2006; Labiano et al. 2014). Since the eastern inner radio jet is also approaching this implies that the jet is crossing through the disc. The jets also create a cocoon which disturbs the gas in all directions from the core.

Figure 7.

Cartoon representation of the orientation and geometry of the nearby radio galaxy 3C 293. The disc of the galaxy and the radio jets are roughly orientated the same way (eastern side approaching us) meaning that the jets pass through the galaxy disc. On the eastern side this results is blueshifted outflows as observed in the ionized gas. On the western side we detect broad lines with velocities close to the systemic velocity. We also detect the neutral gas outflow in front of the western radio lobe due to the larger amount of gas along that sightline. As the radio jets propagate through the disc they also create a spherical bubble which disturbs the gas in all directions shown by the cocoon-type structure surrounding the jets.

This orientation can also explain why the neutral gas phase of the outflow is only detected in the western jet. The H i outflows are detected in absorption, meaning that we are only sensitive to the gas that lies along our line of sight in front of the radio continuum. The fact that the outflow is detected all along the radio jet axis suggests that the outflow is not produced only at the hotspots, but could also indicate that there is fast-moving gas entrained by the radio jets. This, along with the presence of the cocoon, suggests that there is a lot of turbulent gas in the central regions of 3C 293. If the western jet is receding, it is likely that there is a much higher column density of gas in front of the jet to detect the broad, shallow component in H i whereas on the eastern side the jet is in front of a lot of this gas. As the western hotspot is close to the core we observe an outflow velocity slightly blueshifted compared to the systemic rather than redshifted as would be the case in more extended radio jets.

Note that this orientation of the inner radio jets (eastern jet approaching) is the opposite of what is seen in the larger scale radio lobes. There, the north-western lobe is significantly brighter than the south-eastern lobe and therefore assumed to be the approaching jet (Joshi et al. 2011). However, since the orientation of the jet axis has clearly changed between the outer and inner jets (Beswick et al. 2004), it is possible that the axis has also shifted along the line of sight.

4.2 Outflow rates

Having now confirmed that the ionized gas outflow is evident all along the radio jet axis we can calculate more accurately the properties of the outflow. Since the outflow is resolved we calculate the total mass (M_gas) for the eastern and western jets separately based on the Hα luminosities using the following relation (Osterbrock 1989; EM05):

\begin{equation} M_{{\rm gas}} = \frac{m_{\rm p} L({\rm H}\alpha )}{N \alpha ^{{\rm eff}}_{{\rm H}\alpha } h \nu _{{\rm H}\alpha }}. \end{equation}

(1)

Here m_p is the mass of a proton (in kg), L(Hα) is the Hα luminosity (in erg s⁻¹), N the electron density (in cm⁻³), h the Planck constant and ν_Hα the frequency of Hα. The effective recombination coefficient, |$\alpha ^{{\rm eff}}_{{\rm H}\alpha }$|⁠, is calculated to be |$2.87 \times 1.24 \times 10^{-25}/ h \nu _{{\rm H}\alpha }$| assuming case B recombination theory and a temperature of 10 000 K (Osterbrock 1989). The density was estimated using the [S ii] λ6716/λ6731 ratio, based on single Gaussian fits to each of the two lines in the [S ii] blend in each of the spectroscopic apertures of the long-slit data.⁴ Significant density measurements (i.e. the [S ii] ratio measurement was >3σ from the low-density limit) were obtained in two apertures: one 1.4 arcsec north-east of the nucleus along PA225 (N_e = 368 ± 107 cm⁻³), and the other 1.4 arcsec to the east of the nucleus along PA93 (N_e = 110 ± 31 cm⁻³). However, all the [S ii] ratio estimates are consistent within the errors for all the spectroscopic apertures. Therefore we adopted the density derived⁵ from the error weighted mean of all the [S ii] estimates (⁠|$N_{\rm e} = 204^{+49 }_{ -44}$| cm⁻³) as a representative density for all the apertures.

Using Hα luminosities of L(Hα) = 3.7 × 10⁴⁰ for the eastern hotspot and L(Hα) = 2.5 × 10⁴⁰ erg s⁻¹ for the western hotspot, we calculate the total mass of ionized gas to be 2.9 × 10⁵ and 4.4 × 10⁵ M_⊙for the eastern and western jets, respectively.

To calculate the mass outflow rate associated with each hotspot we used the measured outflow velocities of v_out = 304 and 90 km s⁻¹ and divided this by the width of each bin in the direction along the jet (0.8 and 0.5 kpc, respectively) to derive the time taken for gas to travel that distance. This gives mass outflow rates of 0.17 M_⊙ yr⁻¹ for the eastern jet and 0.05 M_⊙ yr⁻¹ for the western jet. From these mass outflow rates we can then calculate the kinetic power, including both the radial and turbulent components of the gas, from the following relation (Holt et al. 2006):

\begin{equation} \dot{E} = 6.34 \times 10^{35} \frac{\dot{M}}{2} \left(v^{2}_{{\rm out}} + 3\sigma ^{2}\right)\,{\rm erg}\,{\rm s}^{-1}, \end{equation}

(2)

where σ = FWHM/2.35. Using FWHM values of 1750 and 1250 km s⁻¹ for the eastern and western outflows, respectively, this gives us kinetic energy rates ranging from 4.6 × 10³⁹ erg s⁻¹ to 3.5 × 10⁴⁰ erg s⁻¹. This represents only a small fraction of the Eddington luminosity (⁠|$\dot{E}/ L_{{\rm edd}} \sim 0.4\hbox{--}3\times 10^{-6}$|⁠) and is significantly less than the associated outflow of neutral gas (8–50 M_⊙ yr⁻¹), even though this was only detected in front of one of the radio jets (Mahony et al. 2013). This is similar to what is found in other sources where the majority of the mass is in the neutral and molecular gas phases and only a small fraction of the total mass outflow rate is detected in the warm, ionized gas (e.g. Feruglio et al. 2010; Dasyra & Combes 2012; Morganti et al. 2013; García-Burillo, Combes & Usero 2014). Combining the total energy rate for both the ionized and neutral gas we calculate the kinetic power injected into the ISM to be of order 0.01–0.08 per cent of the Eddington luminosity (Mahony et al. 2013).

5 CONCLUSIONS

In this paper we presented IFU observations of the nearby radio galaxy 3C 293 observed with OASIS on the WHT. Previous long-slit observations of this galaxy had detected outflows of ionized gas associated with the eastern inner radio lobe (EM05), but H i absorption studies had located an outflow of neutral gas in front of the western radio lobe (Mahony et al. 2013). The IFU observations allowed us to spatially map out the outflow and resolve these seemingly contradictory results.

To measure the kinematics of the gas, either one or two Gaussian components were fit to each spectrum. Where two Gaussian components were needed to fit the data, the lines were separated into a ‘narrow’ and ‘broad’ component representing the gas associated with the disc of the galaxy and the outflow, respectively. Using this method we detected an outflow of ionized gas all along the radio jet axis. In addition to detecting fast outflows associated with the radio jet, we also detected broad linewidths (up to 500 km s⁻¹) in regions up to 12 kpc from the nucleus. We suggest that this could be evidence of the cocoon structure seen in simulations of Wagner & Bicknell (2011) and Wagner et al. (2012), which show that as the jet propagates through a clumpy ISM it also inflates a spherical-like bubble capable of disturbing the gas in all directions. In the case of 3C 293, this region of disturbed gas extends up to 12 kpc from the nucleus along the radio jet axis and up to 3.5 kpc in the direction perpendicular to the jets.

We calculate mass outflow rates of 0.05–0.17 M_⊙ yr⁻¹ and corresponding kinetic powers of ∼0.5–3.5 × 10⁴⁰ erg s⁻¹. This represents a tiny fraction of the Eddington luminosity (⁠|$\dot{E}/ L_{{\rm edd}} \sim 1\times 10^{-6}$|⁠), two orders of magnitude less than the kinetic power of the corresponding neutral gas outflow.

To further investigate the detailed kinematics of the central regions of 3C 293, higher spatial resolution data are required to further map the cocoon structure. Higher S/N across a larger area would also allow us to more readily detect the faint, broad wings indicative of outflows. In addition, IFU observations covering the Hβ and [O iii] lines would enable us to study the emission line diagnostics to better understand the properties of both the regularly rotating gas and the outflow. Finally, studying the molecular gas properties of the outflow with deeper CO observations at higher spatial resolution would provide a comprehensive view of the jet–ISM interaction in 3C 293.

The William Herschel Telescope is operated on the island of La Palma by the Isaac Newton Group in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofísica de Canarias. We thank Lilian Dominguez for her support when carrying out the observations and data reduction, Tom McCavana for his help in creating Fig. 7 and the anonymous referee for their useful comments. The research leading to these results has received funding from the European Research Council under the European Union's Seventh Framework Programme (FP/2007-2013)/ERC Advanced Grant RADIOLIFE-320745. BHCE acknowledges funding through the European Union FP7-PEOPLE-2013-IEF grant no. 624351.

1

This used the pythonmpfit package written by Mark Rivers (http://cars.uchicago.edu/software/python/mpfit.html) which was based on the idlmpfit package (Markwardt 2009).

2

The linewidths have been corrected for the instrumental width.

3

This redshifted emission is also confirmed in the long-slit data presented in Section 3.2.

4

Note that there was not sufficient S/N to allow the [S ii] ratio to be determined for broad and narrow components separately. Therefore the estimated density is an average for both components.

5

Assuming a temperature of 10 000 K.