The Close AGN Reference Survey (CARS). A massive multi-phase outflow impacting the edge-on galaxy HE1353-1917

[Abridged] We combine extensive spatially-resolved multi-wavelength observations, taken as part of the Close AGN Reference Survey (CARS), for the edge-on disc galaxy HE1353-1917 to characterize the impact of the AGN on its host galaxy via outflows and radiation. Multi-color broad-band photometry is combined with spatially-resolved optical, NIR and sub-mm and radio observations taken with VLT/MUSE, Gemini-N/NIFS, ALMA and the VLA to map the physical properties and kinematics of the multi-phase inter-stellar medium (ISM). We detect a biconical extended narrow-line region (ENLR) ionized by the luminous AGN oriented nearly parallel to the galaxy disc, extending out to at least 25kpc. The extra-planar gas originates from galactic fountains initiated by star formation processes in the disc, rather than an AGN outflow, as shown by the kinematics and the metallicity of the gas. Nevertheless, a fast multi-phase AGN-driven outflow with speeds up to 1000km/s is detected close to the nucleus at 1kpc distance. A radio jet, in connection with the AGN radiation field, is likely responsible for driving the outflow as confirmed by the energetics and the spatial alignment of the jet and multi-phase outflow. Evidence for negative AGN feedback suppressing the star formation rate (SFR) is mild and restricted to the central kpc. But while any SFR suppression must have happened recently, the outflow has the potential to greatly impact the future evolution of the galaxy disc due to its geometrical orientation. Our observations reveal that low-power radio jets can play a major role in driving fast multi-phase galaxy-scale outflows even in radio-quiet AGN. Since the outflow energetics for HE1353-1917 are consistent with literature scaling relations of AGN-driven outflows the contribution of radio jets as the driving mechanisms still needs to be systematically explored.


Introduction
Quantifying the direct impact of active galactic nuclei (AGN) for galaxy evolution across cosmic time remains one of the big observational challenges.The enormous energy released by ties at the high-mass end (e.g.Benson et al. 2003;Bower et al. 2006;Cattaneo et al. 2006;Croton et al. 2006;Somerville et al. 2008;Davé et al. 2011;Genel et al. 2014;Crain et al. 2015).In those models, AGN feedback is often separated into a "radio mode" (also known as "maintenance mode") and a "quasarmode" ("ejective mode").The radio-mode AGN feedback describes the dissipation of mechanical energy from powerful radio jets.The energy heats the hot gas in halos of galaxies or galaxy groups/clusters such that condensation of gas is prevented and only a low cold gas content is maintained, limiting the ability for ongoing star formation.This mode of AGN feedback has been well established observationally in massive galaxy clusters and groups that host radio galaxies at their centers (see reviews by Fabian 2012;Gaspari et al. 2013).The quasar-mode feedback instead is thought to remove a significant fraction of the gas from host galaxies of luminous AGN through fast radiatively-driven winds released from the accretion disc (e.g.Feruglio et al. 2015;Tombesi et al. 2015;Bieri et al. 2017).
Evidence for fast winds exceeding 10 000 km s −1 have been observed close to the AGN accretion disc through X-ray and UV spectroscopy (e.g Pounds et al. 2003;Gibson et al. 2009;Tombesi et al. 2010a,b).Whether those radiatively-driven AGN winds are energetic enough to expand out to galaxy scale is still a matter of great debate and intensive ongoing investigation.In particular, the optical emission lines of the ionized gas phase have been extensively explored because they systematically show blue-shifted broad wings in their line shapes which are most pronounced in the [O III] λλ4960, 5007 optical emission lines (e.g.Heckman et al. 1981;Whittle 1985;Mullaney et al. 2013;Zakamska et al. 2016;Woo et al. 2016;Harrison et al. 2016;Bischetti et al. 2017).These asymmetric optical line shapes have been interpreted as a common signature of fast bi-conical AGN-driven outflows where the receding side is obscured by dust from the host galaxy (e.g.Pogge 1988;Bae & Woo 2016).Such biconical outflows have indeed been spatially resolved in several nearby AGN host galaxies with Hubble (e.g.Schmitt et al. 2003;Crenshaw & Kraemer 2000;Das et al. 2006;Fischer et al. 2013;Revalski et al. 2018) and ground-based spectroscopy (e.g.Pogge 1988;Storchi-Bergmann et al. 1992;Riffel et al. 2013;Cresci et al. 2015b;Mingozzi et al. 2019).
An increasing number of AGN have been targeted either with long-slit or integral-field unit (IFU) spectrographs (e.g.Humphrey et al. 2010;Greene et al. 2011;Husemann et al. 2013;Liu et al. 2013b;Harrison et al. 2014;Husemann et al. 2014;Liu et al. 2014;Carniani et al. 2015;McElroy et al. 2015;Perna et al. 2015;Carniani et al. 2016;Kakkad et al. 2016;Karouzos et al. 2016;Leung et al. 2017;Rupke et al. 2017;Sun et al. 2017;Kang & Woo 2018) to spatially map the ionized gas kinematics using the rest-frame optical emission lines.Although there is a clear trend emerging that the [O III] outflow velocity is systematically increasing with AGN luminosity (Fiore et al. 2017;Perna et al. 2017;Woo et al. 2017;DiPompeo et al. 2018), the associated ionized gas kinetic energies can be uncertain by a few orders of magnitude because electron densities n e and detailed outflow geometries are usually unconstrained (Harrison et al. 2018, and references therein).Another uncertain aspect of ionized gas outflows is their physical size.Some studies reported ubiquitous outflow sizes of several kpc scales (e.g.Harrison et al. 2014;McElroy et al. 2015;Liu et al. 2013b,a), but recently much lower outflow sizes of <1 kpc were reported in several studies (Husemann et al. 2016b;Villar-Martín et al. 2016;Tadhunter et al. 2018;Baron & Netzer 2019).The difference may partially be explained due to the effect of beam smearing in seeing-limited observations (Husemann et al. 2016b).
Furthermore, it has been realized that the cold gas phase may actually dominate the total mass outflow rate budget (e.g.Morganti et al. 2005;Feruglio et al. 2010;Cicone et al. 2014;Morganti et al. 2015;Fiore et al. 2017;Veilleux et al. 2017).While many observations have been naturally biased towards gas-rich starburst galaxies hosting AGN (e.g.Sturm et al. 2011;Spoon et al. 2013;Veilleux et al. 2013), a more systematic search on cold gas outflows are currently being conducted (Fluetsch et al. 2019).Nevertheless, the investigation of the cold gas phase in AGN host galaxies still lags behind the ionized gas phase as argued in Cicone et al. (2018a), because extinsive multiwavelength observations are required to obtain the full outflow rate budget in the different gas phases.This limits a robust assessment on the nature of AGN-driven outflows and prevents to distinguish different theoretical predictions of energy-driven or momentum-driven outflows (e.g.Ostriker et al. 2010;King et al. 2011;Faucher-Giguère & Quataert 2012;Costa et al. 2014).
Despite the radio-quiet classification of most AGN being studied so far, they are not radio silent and may host lowluminosity radio jets on various scales from several tens of pc to a few kpc (e.g.Kukula et al. 1998;Blundell & Beasley 1998;Ulvestad et al. 2005;Leipski et al. 2006;Doi et al. 2013).High turbulent velocities in all gas phases have often been associated even with those low-luminosity jets (e.g.Ferruit et al. 1999;Leipski & Bennert 2006;Fu & Stockton 2009;Husemann et al. 2013;Tadhunter et al. 2014;Villar Martín et al. 2014;Cresci et al. 2015a;Harrison et al. 2015;Villar-Martín et al. 2017) and numerical simulation have shown that even the highly collimated jets can actually lead to large scale outflows when propagating through a clumpy high-density ISM (e.g.Sutherland & Bicknell 2007;Wagner et al. 2012;Mukherjee et al. 2018).Usually, outflow energetics are compared to AGN luminosity, jet power, or even star formation rate to determine the powering mechanisms in individual sources.All of these comparisons suffer from systematic measurement uncertainties as summarized by Wylezalek & Morganti (2018).Another complication is that a fast-moving plasma is necessarily creating radio emission itself (Zakamska & Greene 2014;Hwang et al. 2018) which may mimic lowluminosity radio jet characteristics.
It is clear that only extensive multi-wavelength observations of AGN host galaxies will improve our understanding about the nature of AGN-driven outflows and the impact on their host galaxies.We therefore started the Close AGN Reference Survey (CARS, www.cars-survey.org, Husemann et al. 2017) which aims to map the host galaxies of almost 40 nearby (0.01 < z < 0.06) unobscured (type 1) AGN host galaxies at all wavelength from radio to X-rays.The CARS sample of AGN is drawn from the Hamburg/ESO survey (HES, Wisotzki et al. 2000) and focuses on the sub-sample of the most luminous optically-selected AGN at those redshift which has previously targeted for single-dish observations of CO(1-0) as described in Bertram et al. (2007).By combining optical and near-IR IFU spectroscopy, radio and sub-mm interferometry and multi-band imaging of the entire galaxies we will characterize many aspects of the AGN feedback paradigm, enabling a better understanding of the conditions of star formation in AGN host galaxies, quenching via AGN-driven outflows, and the galaxy-scale (im)balance of AGN feeding vs. feedback as described in the chaotic cold accretion model (Gaspari et al. 2017(Gaspari et al. , 2018)).
In this paper we present an in-depth multi-wavelength study of the CARS galaxy HE 1353−1917 (ESO 578-G009, 13 h 56 m 36 s .7 −19 • 31 45 ) at redshift z = 0.035.This AGN host galaxy appears as an edge-on disc-like galaxy where the AGN ionization cones are directly intercepting the galaxy disc.It is therefore a unique test case to study the process of AGN feedback in detail and to highlight the power of our multi-wavelength approach in constraining the AGN outflow energetics and its putative connection to its host galaxy.

The CARS multi-wavelength data set
As part of the CARS survey we have acquired a large number of multi-wavelength observations for HE 1353−1917 from Xrays to radio which are complemented by archival and literature data.Specifically, we obtained optical integral-field spectroscopy with the Multi Unit Spectroscopic Explorer (MUSE, Bacon et al. 2010Bacon et al. , 2014) ) at the Very Large Telescope under programme 095.B-0015(A) (PI: Husemann) and near-IR (NIR) integral-field spectroscopy with the Near-Infrared Integral Field Spectrometer (NIFS, McGregor et al. 2003) at Gemini North as a Fast Turnaround project (Program ID GN-2017A-FT-10, PI: J. Scharwächter).Deep optical (B, V and R bands) and NIR images (J, H and K s bands) were taken with the 4.1 m Southern Astrophysical Research (SOAR) Telescope using the SOAR Optical Imager (SOI; Walker et al. 2003) and with the 2.2m telescope at the Calar Alto Observatory using the Panoramic Near-Infrared Camera (PANIC, Baumeister et al. 2008), respectively, under programmes SOAR/NOAO 2016A-0006 (PI: G. Tremblay) and CAHA F15-2.2-014(PI: B. Husemann). 12CO(1-0) line mapping was obtained with ALMA for HE 1353−1917 in the two configurations C40-5 and C40-2 as part of programme 2016.1.00952.S (PI: G. Tremblay).Radio observations were taken with the Karl G. Jansky Very Large Array (VLA) at X-band (10 GHz) in A configuration and at L-band (1.4 GHz) in CnB configuration under programmes 16B-084 (PI: Pérez-Torres) and 16A-102 (PI: Pérez-Torres), respectively.Finally, a deep X-ray spectrum with 62 ksec integration time was taken with XMM-Newton (ObsID: 0803510201, PI: Krumpe) and a high-resolution X-ray image with Chandra X-ray Observatory as part of the Guaranteed Time Observation (GTO) program 19700783 (PI: Kraft).
The basic characteristics of our multi-wavelength observations in terms of field-of-view, spatial resolution and depth are provided in Table 1.All details of the observations and the data reduction are fully described in the appendix from Sects.A.1 to A.9. Furthermore, we collected archival data from the Wide-field Infrared Survey Explorer (WISE, Wright et al. 2010)

Broad-band SED fitting
We first measure a reliable stellar mass and star formation rate (SFR) for HE 1353−1917 based on our multi-wavelength photometry.The overall spectral-energy distribution from the FIR to the UV bands is listed in Table 2.We employ a template fitting approach to model the broad-band SED.A Bayesian approach is used byAGNFITTER (Calistro Rivera et al. 2016) to decompose the total SED into the stellar, AGN and dust emission contribution.We ran AGNFITTER with 150 walkers, 3 burn-in sets, of 3000 samples followed by 12000 MCMC samples to infer the distribution of parameters.The broad-band SED and 100 representative MCMC samples are shown in Fig. 2.
From the posterior probability distributions derived by AGNFITTER we infer a stellar mass of M * = (2.5±0.5)×10 10 M and a star formation rate of SFR = 2.3 ± 0.1M yr −1 .Due to the strong hot dust component from the AGN torus, AGNFITTER infers the SFR by integrating only the starburst component over the 8-1000µm FIR band as described in Calistro Rivera et al. (2016).Considering the inferred stellar mass we compute a specific star formation rate (sSFR) of log(sSFR/[yr −1 ]) = −10.0± 0.1.Given the molecular gas content, the derived SFR leads to a gas depletion time scale of t dep = (1.6 ± 0.2) × 10 9 yr as expected from normal star forming galaxies (e.g.Leroy et al. 2008;Bigiel et al. 2011).

IFU spectroscopy
We also have to decompose the optical and NIR IFU spectroscopic data from MUSE and NIFS into an AGN and a host galaxy component.To do that we use the fact that the emission from the AGN appears as a point source at all wavelength while the host galaxy necessarily is an extended component.Thus, the first step in the decomposition process is to estimate the PSF for the data.While MUSE covers a large wavelength range leading to wavelength dependent PSF, the PSF for the separate J and K band NIFS observations will be assumed to be constant for the respective band.Following the prescription of Jahnke et al. (2004) we empirically obtain the PSF from the intensity distribution of the broad AGN emission lines using our own software QDEBLEND 3D (Husemann et al. 2013(Husemann et al. , 2014)).We reconstruct the PSF in the optical MUSE data from the broad Hβ, Hα and the O I λ4448, which we interpolate in wavelength pixel-by-pixel using a 2 order polynomial after normalizing the brightest pixel to 1.The J and K band observations with NIFS cover the prominent broad Paβ and Brγ emission, respectively, which are used to construct the PSF for each band without the need to interpolate.Following the iterative AGN-host galaxy deblending scheme presented in Husemann et al. (2013) we initially construct an AGN cube from the brightest AGN spectrum and the wavelength depended PSF.After subtracting the AGN from the original cube, the residual cube contains the host galaxy signal.However, even the brightest AGN spectrum in the initial cubes has a host galaxy contribution which causes an AGN over-subtraction at the center of the galaxy.We mitigate this effect by iteratively subtracting the residual host galaxy spectrum from the initial AGN spectrum taking into account the surface brightness distribution.Given the large field-of-view (FoV) and relatively low spatial resolution of the MUSE data, we use the measured surface brightness profile from the broad-band photometry to interpolate the brightness of the galaxy spectrum towards the nucleus, but assume a constant surface brightness of the host on the very small scales probed by the NIFS data.
We stop the iterative process after 6 and 4 iterations for MUSE and NIFS, respectively, at which the changes are less than 1%.In Fig. 3 we show the decomposed spectra from MUSE and NIFS within an aperture of 3 and 1 , respectively, which confirm that the host galaxy spectra are free from any remaining broad line components.Based on the recovered PSF we estimate a spatial resolution due to the seeing of ∼0 .7 full-width-at-halfmaximum (FWHM) at the Hα wavelength for the MUSE observations and ∼0. 4 for the NIFS observations.

AGN parameters
Basic AGN parameters are determined by fitting the emission lines in the nuclear spectrum of HE 1353−1917 (Fig. 4).The nuclear spectrum was obtained from the MUSE datacube by integrating an aperture of 2 radius centered on the AGN position.A stellar continuum datacube was subtracted beforehand which we constructed by fitting stellar population synthesis model spectra to the QSO-deblended datacube as described below in Sect.3.4.The remaining pure AGN continuum was subtracted by fitting a power-law to the continuum sampled at four different rest wavelength ranges (4700-4732 Å, 5070-5100 Å, 5630-5650 Å, and 6810-6870 Å).Emission lines in the nuclear spectrum were modeled by simultaneously fitting the two wavelength regions around Hβ and Hα with a set of Gaussian line profiles together with an Fe II template from Kovačević et al. (2010).During the fit, all narrow permitted and forbidden lines were kinematically tied to each other in velocity and line width, assuming instrumental spectral resolutions of R S = 1800 and 2500 in the wavelength regions around Hβ and Hα, respectively.
The broad Hβ and Hα lines were modeled using the superposition of two broad Gaussian components.Two components are needed to capture the extended blue wings of the broad Hα line.Both components for the broad Hβ and Hα line were forced to Fig. 3.The total AGN and host galaxy spectrum within a 3 aperture centred on the AGN after applying the AGN-host galaxy deblending procedure as described in the main text.Important emission lines are clearly marked and wavelength regions prone to residuals from telluric absorption correction are highlighted with the dark gray shaded areas.
share the same kinematics.A two component model can, however, not capture the full line asymmetry of the broad line as seen in the residuals of Fig. 4.Those deviations do however not affect the FWHM estimates of the full line profile beyond the systematic errors.For Fe II, the line width and velocity shift were tied to the broad Balmer lines during the fit.We used the narrowest Fe II templates from Kovačević et al. (2010) and convolved these with Gaussian kernels to adjust the line width during the fit.To further constrain the fit, we used fixed flux ratios of 3. From the best-fit model we infer a broad Balmer line width of FWHM BLR = (5820 ± 80) km s −1 and σ BLR = (3990 ± 180) km s −1 .The measured continuum flux at the rest-frame wavelength of 5100Å is f 5100 = (10.5 ± 0.1) × 10 −16 erg s −1 cm −2 Å −1 and the broad Balmer emission-line fluxes are f Hβ = (4.1 ± 0.2) × 10 −14 erg s −1 cm −2 and f Hα = (29.5 ± 0.2) × 10 −14 erg s −1 cm −2 .Here, we estimate the singleepoch BH mass using the broad Hα line luminosity and FWHM as calibrated by Woo et al. (2015) M BH = 1.12 FWHM Hα 1000 km s The bolometric AGN luminosities can be estimated from various indicators.Adopting a bolometric correction factor of ∼ 10 for the continuum luminosity at 5100Å (Richards et al. 2006) and the conversion from the broad Hα to continuum luminosity reported by Greene & Ho (2005) we estimate a bolometric luminosity of L bol = (2.0 ± 0.2) × 10 44 erg s −1 .This estimate from the optical line emission of the BLR is fully consistent with the estimated bolometric luminosity from our deep X-ray observations.Considering our BH mass estimate we computed an Eddington ratio of log(L bol /L Edd ) = −1.9± 0.4, so the BH is accreting close to ∼1% of its Eddington limit.

Ionized gas phase
To obtain clean measurements of the ionized-gas emission lines in the optical MUSE IFU spectra, we first have to subtract the stellar continuum properly avoiding any influence of absorption lines on emission-line fluxes.Modelling the stellar continuum and subsequent emission-line analysis is implemented in PY-PARADISE (see Husemann et al. 2016a;Weaver et al. 2018), which is a Python version of the stellar population synthesis fitting code PARADISE (Walcher et al. 2015) to which emissionline fitting capabilities are added.The stellar and emission-line modelling is done in two subsequent steps.First, PYPARADISE fits the continuum spectrum as a linear super-position of a template library of stellar or stellar population spectra after convolving them with a Gaussian line-of-sight velocity distribution (LOSVD).Second, the best-fit stellar continuum models are subtracted from the original spectra and the emission-lines are modelled as Gaussian functions.A key feature of PYPARADISE is that the input spectra and the template library spectra are initially normalized by a running mean where emission lines or strong sky line residuals are masked and linearly interpolated.In this way, we minimize systematic effects caused by wavelength-dependent flux calibration issues or low-frequency residuals after the AGN subtraction.Below we briefly describe the underlying algorithm and its application to our MUSE data after AGN-host galaxy deblending.
PYPARADISE determines the best-fit LOSVD and linear combination of normalized spectral templates independently based on an MCMC algorithm and a non-negative linear least square fitting, respectively, in an iterative scheme.First, the algorithm infers an initial guess on the LOSVD based on a single spectrum drawn from the template library.Second, the previously determined best-fit LOSVD is applied to all spectra in the template library to obtain the best-fit non-negative linear combination of template spectra excluding masked spectral regions contaminated by emission lines or sky line residuals.This bestfit linear combination of spectra is used as the input spectrum for further iterations and the two step process of estimating LOSVD and linear combination is repeated.The final best-fit spectrum is then denormalized and subtracted from the original spectrum.λλ6717, 6731 emission lines.All the emission lines are modelled as a system of Gaussians with a common LOS velocity and velocity dispersion taking into account the wavelength-dependent spectral resolution of MUSE as derived by Bacon et al. (2017).In addition, the line ratios of the [O III] λλ4960, 5007 and [N II] doublets λλ6545, 6583 are fixed to their theoretical values (Storey & Zeippen 2000).
Since the S/N of the stellar continuum is much lower than that of the emission-lines in the MUSE cube of HE 1353−1917, we apply a Voronoi binning (Cappellari & Copin 2003) to increase the S/N to a minimum of 20 in the stellar continuum in the wavelength range 7000Å < λ < 7100Å.The emission lines are then modelled again per spaxel after repeating the spectral synthesis modelling with PYPARADISE but now fixing the previously estimated LOSVD parameters of the corresponding Voronoi bin used for the continuum fitting.Given that some of the important emission lines, e.g.Hβ, are often too weak in a single spaxel, we also perform the emission-line fitting after binning the original cube by 8 × 8 spaxel (1 .6 × 1 .6).Uncertainties for all parameters are estimated using a Monte-Carlo approach where the stellar continuum and emission lines fitting are repeated 40 times after each spectrum is modulated randomly within the error of each spectral pixel.
The resulting LOS velocity v LOS , velocity dispersion σ and flux maps are shown in the first two rows of Fig. 5.As expected the stellar and ionized gas kinematics reveal a clear rotational pattern along the major axis of the nearly edge-on disc galaxy.The unexpected features for this galaxy are the quite extended filamentary structures of ionized gas outside of the disc.The LOS velocity of the filaments shows that the gas rotates in the same sense as the galaxy disc, which already suggests that the ionized gas is linked to the galaxy disc.Furthermore, we detect exceptionally high velocity dispersion in the ionized gas about 1 .3 (900 pc) south-west of the AGN position.Understanding the origin of both features is the main topic and driver for the following analysis of this paper.

Cold molecular and atomic gas phase
From the cleaned ALMA datacube we created a zeroth moment (integrated intensity) map, a first moment (mean velocity) map, and a second moment (velocity dispersion) map of the detected CO(1-0) line emission using a masked moment technique (Dame 2011).A copy of the clean data cube was Gaussian-smoothed spatially (with a FWHM equal to that of the synthesized beam), and then smoothed (using a 4 channel FWHM Gaussian) in velocity.A three-dimensional mask was then defined by selecting all pixels above a fixed flux threshold of 3 times the RMS noise in this new cube.The moment maps were then created using the original un-smoothed cubes within the masked regions only, without any threshold.
The resulting moment maps are shown in the third row of Fig. 5. Strong molecular gas emission is detected from the disc of the galaxy.The velocity dispersion (second moment) map of this source also shows the typical X-shaped structure of a beamsmeared gas rotating regularly, apart from two hot spots in the south-west and north-east of the nucleus of the galaxy, at the same location as that found in the ionised gas.The origin of these features will be discussed more in Sect.3.6.
After integrating all the CO(1-0) emission, we obtain a cold gas mass of M H2 = (3.6 ± 0.3) × 10 9 M adopting a Galactic conversion factor of α CO = 4.35 M (K km s −1 pc −2 ) −1 .Our estimate is slightly higher than the single-dish measurements from the IRAM-30m telescope of M H2 = (2.7 ± 0.2) × 10 9 M reported by Bertram et al. (2007).The discrepancy is mainly caused by the large extent of the source with respect to the IRAM beam (∼22 ) so that the integrated signal would need to be corrected by the primary beam response.This correction can only be applied to the ALMA data where the CO(1-0) flux distribution is properly resolved with two different baselines to cover the diffuse extended emission with maximum recoverable scale of 20 .
Based on the integrated H I line flux (see Sect.A.7) we estimate a at total atomic gas mass of M H I = (1.3 ± 0.2) × 10 10 M following the prescription of König et al. (2009).The inferred atomic gas mass is fully consistent with the upper limit previ- ously reported by König et al. (2009) for this source using the single-dish Effelsberg telescope.Unfortunately, the S/N is not sufficient to make proper moment maps as in the case of the CO data, but at least we present the peak flux morphology in Fig. 1 which confirms that the atomic gas is aligned with disc.

Warm-molecular gas phase
Warm-molecular gas emission in the NIFS ∼ 3 × 3 FoV is traced via the near-infrared rovibrational H 2 [1-0 S(1)] λ21218Å line.Based on the QSO-deblended NIFS data cubes, we derived emission-line maps for the H 2 [1-0 S(1)] and narrow Brγ component in the K-band as well as Paβ and [Fe II] λ12570Å in the J-band as shown in Fig. 6.Those emission lines were fitted with a Gaussian line profile together with a constant continuum.We also convolved the line widths with the instrumental resolutions of R S ∼ 5290 and R S ∼ 6040 in the K-and J-bands, respectively.Fig. 6 shows that all emission lines in the central ∼ 3 × 3 of HE1353-1917 originate mainly from two emission regions located in the galaxy disc at about 0 .8 (∼ 0.6 kpc) to the north-east and south-west of the nucleus.The emission from the narrow component of the hydrogen recombination lines (i.e.Brγ in the K-band and Paβ in the J-band) is found to be co-spatial with the H 2 emission.The NIFS data also reveal a deficiency of H 2 or narrow-line hydrogen recombination emission within a 0 .3 (0.2 kpc) region (radius) around the nucleus.The nuclear region is instead dominated by the PSF from the broad Brγ and Paβ emission associated with the AGN broad-line region (see Fig. 3) and decomposed during the QSO-host deblending step.
Interestingly, a comparison with the radio continuum map shows that the southern emission peak is nearly co-spatial with the putative radio jet.Also the kinematics reveal a velocity gradient along the jet axis which is not following a pure rotational pattern of the galaxy disc (Fig. 6).Hence, the excitation of H 2 and [Fe II] together with potentially non-gravitational motions of the gas may be directly linked to the impact of an AGN-driven ra-dio jet on the cold gas disc.The enhanced radio emission on the southern approaching side may be explained either by Doppler boosting of a relativistic jet (e.g.Laing 1988;Cohen et al. 2007) or by an intrinsic asymmetry of a bi-polar jet.

Ongoing star formation and extended NLR
A key analysis from the IFU spectroscopy is to discriminate various ionization sources of the ionized gas throughout the galaxy and beyond.In the case of AGN host galaxies, we do expect a mix of ionization from ongoing star formation, AGN photoionization and potentially from shocks due to AGN-driven or starburst-driven outflows.Various optical emission-line ratios are usually employed to discriminate between different ionization mechanisms (e.g.Baldwin et al. 1981;Veilleux & Osterbrock 1987;Kewley et al. 2006;Stasińska et al. 2008).We show the [O III]/Hβ vs. [N II]/Hα diagnostic diagram for HE 1353−1917 in Fig. 7 (top left panel) with the distance from the nucleus indicated as a color gradient.While the QSO-host deblending algorithm already takes out artificial mixing of light from the bright nucleus with the surrounding galaxy, there is an additional mixing between ionization from star formation and the extended NLR.
To disentangle the varying contribution of ionization by young stars and the AGN we use a modified version of the algorithm presented in Davies et al. (2016).The basic idea is that all measured extinction-corrected emission-line fluxes can be modelled as a linear superposition of a characteristic emission-line basis spectrum for the AGN photoionized and classical H II regions.While two basis spectra are sufficient to disentangle the ionization close to the nucleus, the line ratios change across the galaxy due to a radial gradient in gas-phase metallicity.Hence, we expand the method of Davies et al. (2016) by considering a sample of H II basis spectra across the H II sequence from SDSS and an empirical set of basis spectra for the AGN ionization taken from the MUSE data itself.The optimal combination for each spaxel is then determined through a Markov Chain Monte Carlo approach which also provides estimates on the uncertainty on the SF fraction due to the possible combination of basis spectra combined with the intrinsic errors on the emission-line ratios.
In Fig. 7 (top right) we present the estimated SF fraction across the classical BPT diagram.While we compute the AGN fraction for the mixing region within the disc of galaxy ranging from 0 to 1, we set all spaxel regions to an AGN faction of 1 that are well outside the edge-on disc galaxy area.Those spaxels correspond to the AGN ionization cones which are illuminating gas above the main body of the galaxy where a contribution from H II region can be neglected.Those regions indeed keep a high [O III]/Hβ ratio while the [N II]/Hα ratio is decreasing.This is characteristic for a decreasing metallicity (e.g.Groves et al. 2006;Husemann et al. 2011) rather than a mixing with H II regions.
Combining the spatially-resolved extinction-corrected Hα flux with the SF fraction estimates, we obtain a total Hα flux of f Hα = (5.5±0.1)×10−14 erg s −1 cm −2 .This Hα flux corresponds to a luminosity of L Hα = (1.55 ± 0.03) × 10 41 erg s −1 which leads to a Hα-based SFR estimate of SFR Hα = 1.23 ± 0.03 M yr −1 .Similarly, we estimate an integrated extinction-corrected [O III] luminosity from the ENLR as L [O III] = (1.15 ± 0.01) × 10 42 erg s −1 , with an extension that seems to be limited by the MUSE FoV and exceeds 35 or 25 kpc in projection.Hence, the AGN phase has already been lasting for at least 0.1 Myr given the light travel time across the entire ENLR.
3.5.2.Gas-phase metallicity along the galaxy disc and extended NLR The gas-phase metallicity is an important diagnostic as it provides a record of the metal enrichment history of galaxies.1998;Groves et al. 2006), there may also be a dependence on AGN luminosity as highlighted by Stern & Laor (2013).
An alternative metallicity diagnostic is the N2S2 index (N2S2=log([N II] λ6583/[S II] λλ6717, 6731)) which is a proxy for the N/O abundance (e.g Sabbadin et al. 1977;Viironen et al. 2007) which in turn is also correlated with the O/H abundance (e.g.Alloin et al. 1979;Pérez-Montero & Contini 2009).The advantage of the N2S2 index is that it correlates well with the metallicity even for the AGN photo-ionized ENLR, similar to the N2 index, but does not exhibit a strong dependence on the AGN luminosity as discussed by Stern & Laor (2013).While Stern & Laor (2013) provides a calibration of N2S2 to the oxygen abundance scale established by Tremonti et al. (2004), we independently correlate the N2S2 index with the oxygen abundance as inferred by the O3N2 index of Pettini & Pagel (2004) based on the line ratios of 2300 star-forming galaxies from the SDSS DR7 value-added catalog (Brinchmann et al. 2004), This exercise is necessary to compare our results for HE 1353−1917 with those of Sánchez et al. (2014) as the absolute calibration of the O/H abundance is significantly dependent on the used empirical calibration (Kewley & Ellison 2008).
In Fig. 8 we show the O/H abundance based on the N2S2 index as a function of galacto-centric distance in units of kpc and normalized to the effective radius of the galaxy disc.We extract pseudo long-slits along the major axis of the galaxy and the ENLR of the putative AGN-ionization cone (see Fig. 7), which allow us to measure the gas-phase metallicity out to 15 kpc or 2.5 effective radii.We find an error-weighted slope of α for our measured stellar mass of HE 1353−1917 following the calibration of Kewley & Ellison (2008).The consistency checks above shows that our metallicity estimates based on the N2S2 index provide reliable measurements even in the case of dominating AGN-ionization.Most importantly, it highlights the fact that the gas illuminated by the AGN above the galaxy's disc plane shares the same radial metallicity dependence seen in the ionized gas within the galaxy's disc.This is an important piece of evidence which suggests that the gas illuminated by the AGN has been driven out vertically from the disc plane likely by stellar feedback through stellar mass loss and supernova explosions rather than originating from the galaxy centre through AGN outflows.
3.6.Multi-phase gas kinematics 3.6.1.CO(1-0) kinematic modelling and dynamical mass The cold gas kinematics as shown in Fig. 5 seems to be dominated by classical rotational motion of a disc galaxy.Hence, we model the ALMA CO observation directly with the KINematic Molecular Simulation (KinMS, Davis et al. 2013a,b) software tool1 taking the beam size and velocity binning into account.As an input for the model we perform a multi-Gaussian expansion (MGE, Emsellem et al. 1994;Cappellari 2002) to our K s band image observed with PANIC to determine the deprojected mass model assuming a linear mass-to-light ratio gradient (M/L) which is a free parameter for the fitting.The MGE parameters are listed in Table 3.We neglect the unresolved MGE component at the galaxy centre, which arises from the AGN.We assume a truncated exponential disc for the CO surface brightness distribution with the scale radius (r exp ) and the inner truncation radius (r trunc ) as free parameters together with the position angle (PA disc ), inclination (i disc ), the intrinsic velocity dispersion (σ cold ), systemic velocity (v sys ) and the kinematic center (x kin , y kin ) of the disc.
In Fig. 9 we show the CO input maps, model maps and residuals for the best-fit model parameters listed in Table 4.This dark matter fraction is fairly typical for spiral galaxies of this mass (e.g.Salucci et al. 1991;Martinsson et al. 2013).
A striking feature in the cold gas kinematics are two welldefined regions on both sides of the nucleus which exhibit a significantly higher velocity dispersion than our kinematic model predicts.To understand the nature and properties of this region Within the central 3 we detect significant deviation from the PVD model (left) which may be related either to a galaxy ring structure of a bipolar outflow."Hot spots" with an excess in velocity dispersion (right) are found on both sides of the nucleus in the same region.Notes. (a) K s surface brightness of Gaussian component. (b) Standard deviation (width) of Gaussian component. (c) Axis ratio of Gaussian components (*) The central unresolved Gaussian is dominated by AGN light, and is thus subtracted to minimize the effect of the AGN on our kinematic fitting.
we reconstruct the multi-phase position-velocity diagram (PVD) along the major axis of the galaxy (Fig. 10) based on the CO(1-0), H 2 [1 − 0S(1)] and ionized gas kinematic maps.We find that the LOS velocity profile of the ionized and molecular gas matches exactly along the disc at nearly all radii.This shows that also the ionized gas kinematics are determined pre-dominantly by gravitational motions.However, the LOS velocity and, in particular, the velocity dispersion most strongly deviates from the kinematical model in all gas phases about 1 .5-2 .0 away from the center on both sides.The warm molecular gas, which typically arises in strong shocks, is only detected at the location of those two "hot spots" supporting the notion that it is indeed a special region within the galaxy.Most importantly, we detect an extra-ordinary high velocity dispersion in the ionized gas phase at the same location, but only on one side of the galaxy centre.
Given the high inclination and dust content of the disc, we assume that we cannot see the optical emission lines of the ionized gas on the other side of the nucleus due to the high obscuration.
It might be possible to explain the elevated velocity dispersion in the molecular gas by adding an additional point-like mass in the centre or a circum-nuclear bar.However, the significantly stronger deviations in the warm and ionized gas phase, as clearly visible in the PVD and velocity dispersion diagram (Fig. 10) and the kinematic maps (Fig. 5), cannot be explained by gravitational motion and appear symmetrically offset from the nucleus.Hence, we disfavor a simple gravitational induced scenario and conclude that we are detecting the signature of a fast bipolar multi-phase outflow driven by the AGN.

Multi-phase kinematics of the bipolar outflow
The velocity dispersion maps shown in Fig. 11 (left panels) reveal the exact location of regions with a high velocity dispersion.Those kinematic features have already been seen in PVD along the major axis of the disc (Fig. 10).All the different gas phases show a higher dispersion about ∼1 . 2 or ∼900 pc south-west of the nucleus.A similar region is on the opposite side north-east of the nucleus.It appears much weaker in the optical [OIII] than in the Hα emission line, because it is the receding side of the outflow and obscured by the foreground dust screen of the disc.We extract a co-added spectrum within an aperture of 0.8 radius for both high dispersion regions.The spectra for the south-west and north-east region are shown in the right panels of Fig. 11.All optical lines as well as the CO(1-0) line exhibit a prominent broad wing on the blue and red side, respectively.The lack of a broad wing in the NIR lines is likely caused by the limited sensitivity given the peak flux ratio of the narrow-to-broad components seen in the CO(1-0) lines.We modelled the emission lines as a superposition of two Gaussian profiles when applicable.We force doublet lines to be fixed in their theoretical flux ratios and coupled the velocity dispersion of all lines for the two distinct kinematic systems to avoid degeneracies in the solution.The parameters for the best-fit model are listed in Table 5 for the south-west and north-east nucleus.The velocity of the core component is in all cases very close to the corresponding LOS velocity of the stars that we used as a reference.The velocity dispersion of the core component varies between σ core ∼60-140 km s −1 (corrected for spectral resolution) for the ionized and warm-molecular gas phases, but much lower for the cold molecular gas as expected.The wing components are systematically blueshifted by about −300 km s −1 and −160 km s −1 on the south-west side and redshifted by about 140 km s −1 and 100 km s −1 on the north-east side for the ionized and molecular gas, respectively.The ionized gas velocity dispersion of 365±27 km s −1 on the south-west side is significantly higher than the stellar velocity dispersion of σ * = 239 ± 19 km s −1 (in the same aperture) and the broad component of cold molecular gas with 67 ± 25 km s −1 .Those extreme ionized gas kinematics cannot be easily explained solely by gravitational motions and are a rather clear signature of a kpcscale AGN-driven outflow.
Strikingly, the different gas phases are not exactly co-spatial after properly co-registering point-like AGN continuum emission of the various observations as seen in Fig. 11.The warmmolecular gas as probed by H 2 [1-0 S(1)] is mainly excited at the location of the radio emission of the jet in the south-west region and the putative counter-jet in the north-east region.The highdispersion molecular gas peaks are located at a slightly larger distance from the nucleus whereas the very fast ionized gas appears closer to the nucleus and slightly offset vertically from the disc plane.This suggests that the AGN-driven outflow is pushing on the cold gas in the disc with a stratification of excitation and  Fig. 12.A simple cartoon like diagram to highlight the interaction of the AGN with its host galaxy in HE 1353−1917 seen fully edge-on but perpendicular to the ionization cones.In this orientation our observing position in reality would be about 5 • above the disc plane and an unknown azimuth angle to the right so that we can look down the ionization cone into the nucleus.The AGN ionization cones and radio jet are nearly perpendicular to the rotation axis of this galaxy disc in our proposed scenario.The AGN ionization is then able to illuminate extra-planar material above the disc plane.This material was likely expelled from the disc due to mass-loss from star formation and supernova explosions.The gas becomes visible only on one side of the galaxy due the shadowing of the ionization cone given the intrinsic thickness of the dusty galaxy disc.
A jet is hitting the inner wall of a molecular gas ring which significantly enhances the cloud turbulence and disrupts the clouds in the vertical direction.The fast outflow in the ionized gas is then accelerated, either by the radiation pressure or the hot outflow from the jet, and moving around the dense material at the inner edge of the dense gas ring in the disc plane.
speed as the shock front is passing the cold gas clouds which disrupts the front side of the clouds as sketched in Fig. 12.The gas inside the shock front is much warmer than the cold gas outside where the ionized gas is flowing around the thin and dense cold gas disc and diverted outside of the plane of the galaxy following the path of less resistance.In the following we will estimate the outflow energetics of the different gas phases to quantify the driving mechanisms of the outflow and to predict the potential damage to the galaxy.

Outflow energetics and mass outflow rates
As a first step we compute the outflowing ionized gas mass from the extinction-corrected Hα luminosity (L Hα ) and the electron density (n e ) with as described in Husemann et al. (2016a) following Osterbrock & Ferland (2006).From the Hα/Hβ Balmer decrement we infer an optical LOS attenuation, adopting a Milky Way-like attenuation curve (Cardelli et al. 1989), of A V,core = 2.3 ± 0.1 mag A V,wing = 1.8 ± 0.2 mag in the south-west side for the core and outflow component, respectively.The corresponding values for the north-east side are A V,core = 1.2 ± 0. λλ6717,6731 doublet line (Osterbrock & Ferland 2006), which is often the greatest obstacle to robustly measure the outflowing ionized gas mass as discussed in Harrison et al. (2018), Kakkad et al. (2018) and Rose et al. (2018).Here, we are able to robustly deblend the [S II] doublet into a narrow and broad component for both the south-west and north-east region.We measure an S2=[S II]λ6717/[S II]λ6731 ratio of S2 core = 1.21 ± 0.04 and S2 wing = 0.73 ± 0.12 for the core and wing components in the south-west region, respectively, and S2 core = 1.20 ± 0.14 and S2 wing = 1.19 ± 0.15 for the north-east region.Assuming an electron temperature of T e ∼ 10 000 K, we estimate electron densities of n e,core = 198 +45 −41 cm −2 and n e,wing = 1500 +1700 −670 cm −2 for the south-west region and n e,core = 185 +172 −111 cm −2 and n e,wing = 214 +175 −130 cm −2 for the north-east region.Hence, we obtain ionized gas masses of M ion,core = 1.1 +0.35 −0.2 × 10 6 M and M ion,wing = 0.5 +0.4  −0.3 × 10 5 M in the south-west region and M ion,core = 1.9 +2.5 −0.9 × 10 5 M in the north-east region.Since the wing component of Hβ remains undetected due to extreme obscuration towards the north-east region, we can only compute a lower limit on the ionized gas mass of M ion,wing > 1.1 × 10 6 M by applying the same extinction correction to the Hα wing component as for the south-west region.
Based on the ro-vibrational H 2 [1-0 S(1)] NIR emission line we estimate the warm molecular gas mass following the prescription of Mezcua et al. (2015), which assumes a temperature of 2000 K, a transition probability of A = 3.47 × 10 −7 s −1 and population fraction of f = 0.0122.The leads to a mass conversion of Although the H 2 [1-0 S( 1)] shows only one kinematic component in our NIFS observations, it is most likely associated with the outflow as the kinematics are close to those of the molecular gas wing component.We estimate warm-molecular gas outflow masses of M warm = 370 ± 37M for the south-west region and M warm = 250 ± 25M under the assumption stated above.Since those gas masses are orders of magnitude less than the outflowing ionized and molecular gas masses, the warm-molecular gas does not significantly contribute to outflow energetics and mass outflow rate budgets.We therefore ignore this gas-phase for the discussion of the outflow energy budget in the following.
We estimate the cold molecular gas from the CO luminosity following the prescription of Davis et al. (2015): for which we choose a conversion factor of X CO = 2 × 10 20 cm −2 (K km s −1 pc 2 ) −1 as a typical value for massive spiral galaxies like our Milky Way in the nearby Universe.This yields molecular gas masses of M mol,core = (1.8 ± 0.11) × 10 8 M and M mol,winge = (0.84 ± 0.25) × 10 8 M for the south-west region and M mol,core = (0.9 ± 0.2) × 10 8 M and M mol,wing = (1.0 ± 0.3) × 10 8 M for the north-east region.However, the conversion factor is the greatest source of systematic uncertainties and the conditions of the gas in the fast moving clouds is still matter of debate with a lower limit provided by optically-thin gas at a temperature of T ex = 30 K corresponding to X CO = 1.6×10 19 cm −2 (K km s −1 pc 2 ) −1 .While several studies have found significantly higher conversion factors for the outflowing molecular gas (Cicone et al. 2018b;Walter et al. 2017;Zschaechner et al. 2018) also optically-thin conditions have been reported (Dasyra et al. 2016;Oosterloo et al. 2017).Hence, our molecular gas estimates for the putative outflows could be at most an order of magnitude lower and represent a hard lower limit.
Mass outflow rates have often be computed using the assumption of a homogeneously filled cone (e.g.Fiore et al. 2017) with a size R out which leads to where v out is the outflow velocity for which different measurements have been assumed in the literature, which strongly depends on which part of the line shape is assumed to be participating in the outflow.A maximum velocity was defined by Rupke & Veilleux (2013) to be v max = v broad + 2 × σ broad instead of v broad alone for the bulk motion of the broad component.Alternatively, non-parametric line shape measurements of the entire line shape have been used to define v 10 , the velocity containing 10% of the line flux, and W 80 , the width of the line between 10 and 90% of the line flux.We list all those outflow velocity parameters in Table 4.While for v max and v broad only the mass in the broad component is considered, v 10 and W 80 are often used with the gas mass inferred from the entire line shape.
In the specific case of HE 1353−1917 it seems that a filled cone geometry is not a good assumption because our spatiallyresolved kinematic mapping suggests an expanding shell-like Notes.Units for mass outflow rates ( Ṁ) are M yr −1 , for the momentum injection rates ( ṗ) are dyne, and for energy injection rates ( Ė) are erg s −1 .We do not provide the formal measurement errors here since they are significantly smaller than the spread implied by the different asummptions.Combined molecular+ionized gas measurements are shown as shaded (shell assumption) or hatched (cone assumption) areas highlighting the one order of magnitude uncertainty of X CO , while pure ionized gas mass computations results are represented by solid lines (shell assumption) or dashed lines (cone assumption).For comparison, the magenta lines highlight the SFR of the entire galaxy from the FIR measurements in the left panel, the point where L bol /c = ṗout in the middle panel, and the fraction of L bol at 10% (solid line), 1% (dashed line) and 0.1% (dotted line) as well as the jet power P jet (yellow dashed line) in the right panel.
shock front scenario that is propagating through the galaxy.In this case the mass outflow rate can be described as which implicitly represents the localized outflow rate at the shell distance from the nucleus independent of the outflow opening angle.The shell-thickness ∆R cannot be directly measured from our observations of HE 1353-1917 at our given spatial resolution.For the outflow energy in Table 6 we adopt a thickness of ∆R = 200 pc, but parametrize the thickness within a range of 20 and 500 pc in Fig. 13.In addition to the mass outflow rate we also compute the momentum injection rate ṗout = v out Ṁout and kinetic energy injection rate as Ėkin = 0.5v 2 out Ṁout .The turbulent motion is implicitly considered in all outflow velocity definitions except for v br for which we add σ br in quadrature to be comparable.
We visualize the results in Fig. 13 for Ṁout , ṗout and Ėkin for the shell and cone geometry, various outflow velocities and the ionized and total gas phases.Since there is a one order of magnitude systematic uncertainty on the molecular gas mass depending on the actual X CO factor in the outflow we show them as bands in all panels.To separate the ionized gas and total gas mass is important for various reasons, 1) most observations of AGN are only able to measure the ionized gas energetics, 2) the molecular and ionized gas have different kinematics, and 3) it is unclear what fraction of the broad component in the molecular gas is caused by an outflow or governed by internal galaxy structure, such as a circumnuclear ring.This already tells us that even with our high-quality measurements computation of outflow en-ergetics is hard and strongly model dependent.While we have taken out the uncertainty on the electron density for the ionized gas, the definition of outflow velocity, geometry and X CO still leads to order of magnitude uncertainties.
Nevertheless, we can draw some solid conclusions from the data.The total mass outflow rate already exceeds the total SFR independent of X CO or geometry.The total mass-loading factor η = Ṁ/SFR therefore ranges between ∼1-30, if the molecular gas is indeed outflowing.Given that the outflow is still confined to the central kpc, the mass-loading factor would further increase by at least an order of magnitude if the SFR would be taken from the same region.Compared to the ionized gas the molecular phase clearly dominates Ṁout even at much lower speed, which implies η < 1 (for the total SFR) for the ionized gas phase alone for all cases when the shell thickness is ∆R > 100 pc.The difference in the momentum injection rate ṗout and the kinetic energy injection Ėout is significantly reduced between the ionized and molecular outflow due to the higher outflow velocity of the ionized gas phase.Our observations are consistent with ṗout ∼ L bol /c only for the cone geometry, if X CO is on the low side, and for the ionized gas phase if ∆R > 300 pc.For the shell geometry we find ṗout > L bol /c if ∆R < 100 pc independent of X CO .A caveat here is that the covering factor of the spherical AGN luminosity is significantly smaller than 1.Assuming a vertical height of ±250 pc for the outflow at 1 kpc distance means that the covering factor is only 25%, which would imply a momentum boost for radiation pressure only of > 1 for the total ṗout in all cases.
Only ∼ 1% of the total bolometric luminosity is required to explain the estimated kinetic energy injection rates Ėkin in the outflow for all cases where ∆R > 300 pc.Taking into account the covering factor would bring this to 4%.Hence, the AGN radiation may indeed be able to drive the outflow if such a fraction of the luminosity can be efficiently transferred to the gas in the disc.However, we also see a radio jet impinging on the gas disc.We convert the inferred radio luminosity at 1.4 GHz from our VLA observations at 10 GHz to a kinetic jet power of log(P jet /[erg s −1 ]) = 43.3based on the calibration by Cavagnolo et al. (2010), which is nearly equal to 0.1L bol .The radio jet alone would therefore also be able to mechanically drive the outflow with just 10% of coupling efficiency as seen in Fig. 13.

The origin of the large extended narrow-line region
One of the most prominent feature seen in the MUSE data is the extended narrow-line region (ENLR) where AGN photoionized gas can be traced out several tens of kpc to the limits of the MUSE FoV.The two-side geometry combined with a prior on the AGN orientation through the type 1 AGN classification implies that the geometry of the ENLR is driven by the opening angle and orientation of the AGN ionization cone fully consistent in line with the unification model of AGN (Antonucci 1993;Urry & Padovani 1995).However, various ENLR studies usually find that the ENLR is bound to the galaxy in the majority of cases as the AGN illuminates primarily the ISM (e.g.Bennert et al. 2002;Schmitt et al. 2003;Netzer et al. 2004;Husemann et al. 2013;Hainline et al. 2013;Husemann et al. 2014).There is only a minority of cases reported where the ENLR significantly reaches beyond the galaxy which have been explained by the illumination of gas accretion flows from the environment (e.g.Husemann et al. 2011;Johnson et al. 2018), gas debris or filaments due to past or ongoing galaxy interaction (e.g.Villar- Martín et al. 2010;da Silva et al. 2011;Husemann et al. 2016a;Storchi-Bergmann et al. 2018;Villar-Martín et al. 2018), or large-scale gas outflows (e.g.Greene et al. 2012;Liu et al. 2013a).
For HE 1353−1917 we have presented a number of measurements that let us draw a conclusive picture for the origin of the ionized gas a few kpc above the galaxy disc.The gas kinematics show that the ENLR shares the same direction of the angular momentum vector as the gas and stars of the disc without any signatures of counter-rotation.The LOS velocity amplitude on both sides of the ENLR is slightly smaller than that of the gas in the disc at the same radial distance along the plane.This could in principle be a signature of an additional velocity component (like an outflow) or simply an inclination effect since the ionization cone cannot be fully perpendicular to our line-of-sight given that we observe an unobscured AGN nucleus.
Additionally, the metallicity gradients along the galaxy disc and along the ENLR ionization cone axis are in good agreement.This is evidence against an outflow scenario because the metal enriched gas from the center would be ejected so that we would expect a significant increase of metals along the ENLR with respect to the galaxy disc.That the ENLR follows the metallicity gradient of the disc strongly suggests that the material in the ENLR was expelled nearly vertically from the disc and thereby retains a similar metallicity pattern as well as the same angular momentum as a function of radius.Hence, the AGN-illuminated gas above the plane is tightly connected to the disc properties as a function of radius.Stellar mass loss and supernova-driven winds is a natural explanation for the coupled similar kinematics and metal enrichment.However, the velocity dispersion in the extra-planar gas is much smaller than in typical starburst-driven winds (e.g.Ho et al. 2014Ho et al. , 2016)).This means that the outflow is significantly weaker or that the gas is already falling back after a short burst in star formation a few Myr ago.Both scenarios agree with the small amount of illuminated ionized gas 10 5 M and the unusual star formation activity in the disc.

Origin of the compact ring-like gas structure
Another striking feature in our data is the enhancement of velocity dispersion in the cold gas and shocked-heated H 2 gas in two symmetric regions around the nucleus (Fig. 11), which also resembles a ring-like structure seen edge-on.Such a ring-like structure is tentatively supported by a deficit of ionized gas (Brγ, Paα, or [FeII]) seen in the high-angular resolution Gemini observations (Fig. 6).This is not affacted by the AGN subtraction as the PSF is significantly narrower than the apparent diameter of the ring.Given the orientation of the ionized gas outflow and the edge-on nature of this galaxy, there are several scenarios that may cause such a ring structure.HE 1353−1917 is one of the few cases, such as NGC 1068 (García-Burillo et al. 2014;Gallimore et al. 2016) or NGC 5643 (Alonso-Herrero et al. 2018;Cresci et al. 2015b), where the ionization cones and jet axis are directly intercepting the cold gas of the galaxy.Hence, one potential scenario is that the AGNdriven outflow has been pushing the gas outwards for some time.This scenario is attractive to support the impact of AGN outflow, since a multi-phase outflow and a radio jet within the depleted gas disc is clearly detected.However, this scenario assumes that the cold gas disc was initially homogenous and reaching down to the cricumnuclear scales of a few pc.
Alternatively, the outflow is impinging on a pre-existing nuclear ring with a diameter of 2 .3 or 1.6 kpc seen edge-on.The position-velocity diagram of Fig. 10 indeed can be interpreted as a typical signature for a nuclear ring rather than a pure out-flow.The velocity signature of a ring is a straight and steep line in the center, as can be seen in our own Milky-Way central molecular zone for example (e.g.Yusef-Zadeh et al. 2009), with a ring of 200 pc radius.Other examples of edge-on rings are M82 (Neininger et al. 1998) or NGC 4565 (Yim et al. 2014).Obviously, it is more easy to identify nuclear rings when not edge-on, like in NGC 1433 (Combes et al. 2013), NGC 1068(García-Burillo et al. 2014) or NGC 1097 (Martín et al. 2015).
These rings are thought to be due to the gravity torques of a bar, driving gas toward the centre, where it accumulates transiently into a ring at the inner Lindbald resonance of the bar (e.g.Buta & Combes 1996).The ring is then massive, and produces some kinematic decoupling with a characteristic feature in PVDs.When the barred galaxies are associated to a lowluminosity AGN (Seyfert, LINER), it is frequent to see also a molecular outflow, as revealed in the references above.Weak outflows are usually not the origin of such ring structure, which pre-existed the nuclear activity.However, we cannot detect a bar due to the inclination of the galaxy so we cannot test if the ring is matching with the expected Lindbald resonance.Alternatively, the turbulent regime at the galaxy centre with a turbulent Taylor number of Ta t ≡ v rot /σ v < 1 will lead to inelastic collisions of gas clouds enhancing their condensation fostering rapid feeding of the BH (e.g.Gaspari et al. 2017).Whether the molecular gas in the ring has already being pushed outwards by the outflow or acts as a stiff barrier for the outflow given its mass and density is still unclear at this point.In the following section we assume nevertheless that most of the high dispersion in the molecular phase is associated with an outflow to compute the associated outflow energetics.

The powering source of the multi-phase gas outflow
In contrast to the large ENLR on tens of kpc scales, the prominent multi-phase gas outflow in HE 1353−1917 is confined to the inner ∼1 kpc from the nucleus.This is a large size for the outflow compared to what we find in other CARS targets (Singha et al. in prep.).HE 1353−1917 is therefore an ideal laboratory to understand the driving mechanisms for this outflow as we could make relatively precise estimations of the outflow energetics from our multi-wavelength observations.One big question is whether these outflows are radiatively driven by the AGN radiation or mechanically driven by a jet even in radio-quiet AGN (Wylezalek & Morganti 2018).
Considering the energetics of the outflow in comparison to the AGN bolometric luminosity and the jet power as shown in Fig. 13 it seems that both the AGN and the radio jet are capable of powering the outflow.Yet, the jet power is already highly directional and the size of the jet exactly matches with the location of the warm molecular gas and the turbulent molecular gas.Since only a fraction of the AGN luminosity would impact the disc, the dissipation of the radiation into kinetic energy would need to be much more efficient than 1% to power the entire multi-phase outflow.Indeed, the orientation of the jet axis such that it almost exactly intercepts the gas disc is uncommon which makes HE 1353−1917 a rare case.However, similar examples have been reported before such as IC 5063 (e.g.Kulkarni et al. 1998;Morganti et al. 2013;Tadhunter et al. 2014;Morganti et al. 2015;Dasyra et al. 2015Dasyra et al. , 2016;;Oosterloo et al. 2017), 3C 293 (e.g. Emonts et al. 2005;Mahony et al. 2013Mahony et al. , 2016) ) or NGC 3079 (e.g.Duric & Seaquist 1988;Irwin & Sofue 1992;Veilleux et al. 1994;Cecil et al. 2001;Middelberg et al. 2007).In all cases, high turbulent velocities have been found at the hot spots of radio jets.A spatial coincidence of radio jet morphology and velocity dispersion of the ionized gas has already been reported for spatially-resolved spectroscopy of more luminous radio-quiet AGN (e.g.Husemann et al. 2013;Villar-Martín et al. 2017) and powerful compact radio sources (e.g.Roche et al. 2016), but it has been correctly proposed that the fast moving plasma itself can lead to radio emission that mimics jet activity (Zakamska & Greene 2014;Hwang et al. 2018).In the case of HE 1353−1917 we can rule out that the ionized plasma is creating the radio emission because the high-velocity ionized gas traced by [O III] is significantly displaced compared to the observed jet-like radio emission.Hence, we think that the radio jet is transferring its energy and momentum to the ambient medium through an extended shock front which creates turbulence in a dense clumpy interstellar medium.Such a great impact of the radio jet has been observationally shown in many cases (e.g.Villar-Martín et al. 1999;O'Dea et al. 2002;Nesvadba et al. 2006;Holt et al. 2008;Guillard et al. 2012;Villar Martín et al. 2014;Harrison et al. 2015;Santoro et al. 2018;Tremblay et al. 2018) and theoretically supported through detailed hydrodynamic simulations (e.g.Krause & Alexander 2007;Sutherland & Bicknell 2007;Wagner & Bicknell 2011;Wagner et al. 2012;Cielo et al. 2018;Mukherjee et al. 2018).As we discussed in Sect.3.6, the jet power alone is sufficient to energetically drive the outflow because only a small fraction of the AGN luminosity would impact the thin disc of the galaxy implying conversion efficiencies of more than 10% of L bol .Hopkins & Elvis (2010) proposed a two-stage process for efficient radiation-driven outflows.They describe a scenario in which an initial weak wind in the hot gas phase, possibly initiated by an accretion disc wind or a radio jet, creates additional turbulence in the surrounding medium so that massive gas clouds will subsequently expand and disperse.This expansion of gas clouds would significantly increase their apparent cross-section with respect to incident radiation field of the AGN.Such a twostage process may increase the coupling efficiency by an order of magnitude.While we cannot directly confirm this process with our observations, the close alignment of the jet axis and the ionization cone greatly suggest that the outflow is driven jointly by both mechanical and radiative energy with an unknown ratio of the two.The open question is whether the same powerful outflow could have developed without the fast radio jet impacting the cold gas directly given its unique orientation.
Recent compilations of ionized (Fiore et al. 2017) and molecular (Fluetsch et al. 2019) outflows found a significant correlation with outflow kinetic power and AGN luminosity.Fiore et al. (2017) assumed a conical outflow geometry with v max as the outflow velocity, while Fluetsch et al. (2019) used the average thin shell approximation that is systematically smaller by a factor of 3. In Fig. 14 we compare the literature measurements re-computed for a conical outflow geometry with our conical outflow measurements for HE 1353−1917 using v max as in Table 6.We find that the mass outflow rates and kinetic power of the outflow measured for HE 1353−1917 is fully consistent with the clear trends between outflow energetics and AGN luminosity as found in previous studies.While we argue that a conical geometry is an inappropriate description for the outflow it is still valid for a homogenous relative comparison between different sample measurements where spatial information is limited.Although this could be taken as evidence that the outflow must be driven by the AGN luminosity, the fact that the radio jet clearly contributes to the AGN outflow cautions against this simple interpretation.Since the radio luminosity correlates with optical AGN luminosity even for radio-quiet AGN the relative contribution to the outflow is not clear and is not sufficiently probed in the current literature.Statistically, Mullaney et al. (2013) and Villar Martín et al. (2014) argued for a significant role of radio jets in driving ionized gas outflows whereas Woo et al. (2016) attributes this to simple sample selection effects.Only spatiallyresolved mapping of the multi-phase outflows in relation to radio jet morphology and AGN ionization cones as presented here will be able to disentangle the relative importance of both mechanisms as discussed in Wylezalek & Morganti (2018).

Implications for AGN feedback
The direct suppression of star formation caused by large-scale AGN-driven winds is one potential mechanism to regulate the growth of massive galaxies.While the cold gas in disc galax- ies is hard to destroy when the outflow expands perpendicular to the disc (e.g.Gabor & Bournaud 2014), the outflow in HE 1353−1917 is orientated such that it is directly impacting the cold gas disc.Hence, we would expect the efficiency of negative AGN feedback, i.e. suppression of star formation, to be maximized at the given outflow energy.In Fig. 15 we compare the sSFR and the associated cold gas depletion time (t dep ) of HE 1353−1917 with the overall distribution of local non-AGN galaxies from SDSS (Brinchmann et al. 2004) and the COLDGASS survey (Saintonge et al. 2011).We find that HE 1353−1917 is nearly matching galaxies in the star-forming main sequence pointing against negative AGN feedback caused by the outflow.However, we also find a significant difference between the sSFR based on the FIR and Hα-based SFR tracers where the FIR-based sSFR is closer to the star-forming main sequence than the Hα-based SFR.The difference is a factor of 2, which is significant considering the measurement errors.As pointed out by Hayward et al. (2014), the FIR-based SFRs may over-predict the SFR in galaxies undergoing star formation suppression, because the FIR emission is excited by stars with life times of up to ∼100 Myr and may not react on SFR changes on much smaller timescales.Here, we also compute the SFR based on the AGN core subtracted 1.4GHz radio luminosity and the radio-SFR calibration by Bell et al. (2003).The SFR range between 2.8 M yr −1 and 10 M yr −1 for an assumed range in spectral index of −1.2 < α < −0.5.This suggest that the SFR has remained nearly constant till a few tens of Myr as probed by radio emission.Since the Hα line can only be excited by stars with life times of up to ∼Myr it is much more sensitive to more rapid SFR variations than the FIR or the radio.This timescale effect has been detected in merging galaxies where the SFR is enhanced on small timescale at close galaxy separation which is only recovered in Hα and not in the longer timescales SFR tracers (Davies et al. 2015).Considering a time scale of at most a few Myr years for the AGN phase in HE 1353−1917 and the associated impact on the surrounding medium, it is plausible that the difference in the SFR is caused by the same timescale effect.On the contrary, the Hα luminosity may be particularly prone to the systematics of dust extinction given the edge-on view we have on HE1353−1917, which may also cause the discrepancy in the SFRs.
In any case, the impact of the current powerful AGN-driven outflow in HE 1353−1917 on the total SFR and star formation efficiency is small and nearly indistinguishable from the parent population of non-AGN galaxies.This is in agreement with several works finding that AGN host galaxies have similar star formation properties as their non-AGN counterparts (e.g.Harrison et al. 2012;Husemann et al. 2014;Xu et al. 2015;Balmaverde et al. 2016;Vaddi et al. 2016;Zhang et al. 2016;Suh et al. 2017;Rosario et al. 2018;Shangguan et al. 2018), while others report a significant decrease in the SFR (e.g.Farrah et al. 2012;Mullaney et al. 2015;Shimizu et al. 2015;Wylezalek & Zakamska 2016) for different kind of AGN samples.

Summary and conclusions
In this paper we presented an extensive spatially-resolved multi-wavelength investigation of the edge-on disc galaxy HE 1353−1917 hosting a luminous AGN.The observations offer unique insights into the interaction of an AGN with its host galaxy.Our main results can be summarized as follows: -HE 1353−1817 reveals a large biconical ENLR which extends about 25 kpc nearly parallel to the galaxy disc.Considering the kinematics and the metallicity of the ionized gas we conclude that the gas likely originated from the galaxy disc and was vertically lifted by stellar mass loss and supernovae on several Myr timescales.Without the illumination by the AGN such low-density diffuse gas would be a neutral gas phase and undetectable at optical wavelengths.-A fast multi-phase outflow about ∼1 kpc away from the nucleus is detected in the ionized gas (v max ∼ 1000km/s) and molecular gas (v max ∼ 300km/s).This region of the outflow appears as a distinct feature in the PVD and is coincident with jet-like radio emission at 10 GHz as detected with the VLA.Warm molecular emission is also directly associated with the radio jet location.This co-location implies that the ISM is heated directly through a fast moving shock, which is hitting the dense, cold gas disc of the galaxy.The cold gas has possibly formed a ring structure dynamically at this radius, created through a bar or other internal mechanisms, such as chaotic cold accretion.-Despite the great uncertainties in computing outflow energetics, we can confirm that the molecular gas phase would dominate the mass outflow rate if a ring structure is not dominating the observed cold gas kinematics.In the most pessimistic case the mass outflow rate is Ṁ > 2M yr −1 and may be as high as Ṁ ∼ 200M yr −1 implying a mass loading factor greater than unity.-It is unclear whether the momentum injection rate of the outflow is consistent with ṗ = L bol /c or requires a momentum boost as suggested for many other AGN-driven outflows.However, the power in the radio jet is sufficient to drive the outflow alone, so that the outflow is not necessarily radiatively driven.While indeed 5% of the total AGN luminosity would be sufficient to power the outflow as well, the geometric covering factor of the AGN radiation field intercepting the gas disc is almost an order of magnitude lower which implies that the radiation alone may not be able to power the outflow.-The ionized and molecular mass outflow rates and the associated kinetic outflow power for HE 1353−1917 is fully in line with previous trends with AGN luminosity as found in larger samples of AGN host galaxies.-The host galaxy of HE 1353−1917 is still on the star-forming main sequence based on FIR and radio continuum diagnostics which would not imply strong negative or positive feedback by the outflow.However, we detect a mild reduction in the SFR by a factor of 2 when estimated from the extinctioncorrected and ionization-corrected Hα luminosity.It is unclear at this point if this implies a recent reduction of the SFR in the last 5 Myr (Hα) compared to 100 Myr (FIR) or if a large fraction of the Hα luminosity is missed by the high attenuation of dust due to edge-on orientation of the galaxy.
Our study of HE 1353−1917 demonstrates the unique potential of CARS to unravel the details of multi-phase AGNdriven outflows, their driving mechanism and impact on the host galaxies.This single-object study shows how the different observations can be combined to provide a complete picture of the galaxy characteristics.While it is difficult to make general conclusions for the overall AGN population from a single object study, there are a few important points to be drawn from this initial investigation, which will drive further studies from the overall CARS sample.
A striking feature is the orientation of the AGN ionization cones and the radio jet axis which directly intercept the cold gas disc of this galaxy.This orientation effect certainly maximizes the efficiency of the AGN to drive a massive outflow that already reaches out to distance of 1 kpc from the nucleus.The direct spatial coincidence of the radio jet with the fast outflow suggests a dominant role of the jet power in driving the outflow which is confirmed based on the outflow energetics.Nevertheless, the impact of the outflow on the SFR is still very mild and may at most affect the central kpc at very recent times.Whether AGN feedback can significantly reduce the molecular gas content and SFR of the entire disc on longer timescales is unclear at this point.
Considering that the outflow rates and energetics of HE 1353−1917 are in agreement with estimates for larger AGN samples from the literature, we may pose the question whether the previous trends with AGN luminosity can be considered as unambiguous evidence for radiatively-driven AGN outflows.Systematic investigations of the connection between small-scale low-power radio jets and the multi-phase outflows are still lacking for many existing samples.One of the primary goals of CARS is therefore to systematically explore the relative role of jets and radiation as the powering mechanisms of AGN-driven outflows and its impact on star formation.versity) and by NASA Chandra grants GO7-18121X/GO8-19104X.SMC acknowledges support from the Australian Research Council (DP190102714).We thank Alex Markowitz for helpful discussions on the RGS data in the context of warm absorbers.Herschel is an ESA space observatory with science instruments provided by European-led Principal Investigator consortia and with important participation from NASA.This publication makes use of data products from the Wide-field Infrared Survey Explorer, which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory/California Institute of Technology, funded by the National Aeronautics and Space Administration.This research has made use of the NASA/IPAC Infrared Science Archive, which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.The Pan-STARRS1 Surveys (PS1) and the PS1 public science archive have been made possible through contributions by the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max-Planck Society and its participating institutes, the Max Planck Institute for Astronomy, Heidelberg and the Max Planck Institute for Extraterrestrial Physics, Garching, The Johns Hopkins University, Durham University, the University of Edinburgh, the Queen's University Belfast, the Harvard-Smithsonian Center for Astrophysics, the Las Cumbres Observatory Global Telescope Network Incorporated, the National Central University of Taiwan, the Space Telescope Science Institute, the National Aeronautics and Space Administration under Grant No. NNX08AR22G issued through the Planetary Science Division of the NASA Science Mission Directorate, the National Science Foundation Grant No. AST-1238877, the University of Maryland, Eotvos Lorand University (ELTE), the Los Alamos National Laboratory, and the Gordon and Betty Moore Foundation.This work is based in part on observations made with the Galaxy Evolution Explorer (GALEX).GALEX is a NASA Small Explorer, whose mission was developed in cooperation with the Centre National d'Etudes Spatiales (CNES) of France and the Korean Ministry of Science and Technology.GALEX is operated for NASA by the California Institute of Technology under NASA contract NAS5-98034.

Fig. 1 .Fig. 2 .
Fig. 1.A multi-wavelength overview of HE 1353−1917.Top panels: SOAR optical BVR image composite image, PANIC near-IR K s band image and a VLA L-band (1.4GHz) continuum as contours covering a FoV of 3 × 3 (∼ 130 kpc × 130 kpc) centered on the galaxy.For the L-band we display 1, 3, 5, and 25 σ levels as positive (solid black lines) and negative contours (dashed golden lines).We also show the peak flux density in the H I frequency range at 1.3 and 2 mJy levels indicating roughly the location of the atomic gas along the galaxy disc.The black dashed box indicates the orientation and position of the MUSE field-of-view (FoV).Middle panels: A color composite image is shown on the left side based on a narrow-band [O III], Hα filter and an I band filter reconstructed from the 1 × 1 MUSE observation.The CO(1-0) flux map observed with ALMA is shown on the right side where the 10 GHz VLA continuum image is overplotted as 1, 3 and 5σ positive (blue lines) and negative (golden lines) contours.Bottom panels: Mid-to far-IR archival images from WISE and Herschel space observatories for the 3.4 µm to 160 µm wavelength range.Beam sizes are shown when applicable in the lower left corner for guidance in each panel.

Fig. 4 .
Fig. 4. Fit to the nuclear spectrum of HE 1353-1917 in the wavelength regions around Hβ and Hα after continuum subtraction.The upper panels show an overlay of the nuclear spectrum (black) and the total fit (red).The residuals from the fit are indicated in gray (offset from the nuclear spectrum for better visibility).The lower panels show the individual Gaussian components fitted to the nuclear spectrum together with an Fe II template from Kovačević et al. (2010).All components shown with the same color were kinematically tied to each other during the fit.This includes the narrow components (Hβ n , [O III] λ4959 and λ5007, [O I] λ6300 and λ6364, [N II] λ6548 and λ6583, and Hα n and [S II] λλ6717, 6731) shown in blue, the broad components (Hβ b and Hα b ) shown in yellow, and the very broad components (Hβ vb and Hα vb ) shown in red.
Afterwards emission lines are then measured subsequently in the residual spectrum.Given the spectral coverage of MUSE and redshift of HE 1353−1917, we focus on modelling the Hβ, [O III] λλ4960, 5007, [O II] λ6300, Hα, [N II] λλ6545, 6583 and [S II]

Fig. 5 .
Fig.5.Maps of the stellar (first row), ionized gas (second row) and cold molecular gas (third row) kinematics and flux distribution.The first column shows the line-of-sight LOS velocity with respect to the systemic velocity of the galaxy, the second column is the velocity dispersion and third/forth column highlights the surface brightness of the respective tracer.The beam size for the ALMA observations is indicated as a red symbol in the respective panels.

Fig. 6 .
Fig. 6.Top panels: Surface brightness maps for the H 2 [1-0 S(1)] λ21218Å, Brγ, Paβ and [Fe II] λ12570Å emission lines in our Gemini-NIFS NIR observations after subtracting the point-like AGN contribution.The red contours correspond to the distribution of 10GHz radio continuum emission as resolved with our VLA radio interferometric observations.Lower panels: Radial velocity and velocity dispersion for the two brightest lines, H 2 and Paβ respectively.

Fig. 7 .
Fig. 7. Top left panel: Classical BPT [O III]/Hβ vs. [N II]/Hα diagnostic plot after binning the MUSE data to 1 .2 spatial sampling.The distance from the nucleus is color coded and the distribution of line ratios from the SDSS MPA/JHU catalog is shown as contours for comparison.Demarcation curves from Kauffmann et al. (2003) (dashed blue line), Kewley & Dopita (2002) (red solid line) and Stasińska et al. (2008) (solid magenta line) are displayed for reference.Top right panel: Same as the left panel where the color coding represent the SF fraction to the emission line measured as described in the main text.The SF fraction is set to zero for emission regions above the galaxy disc and set to one for regions below the Kauffmann et al. demarcation line.Bottom left panel: 2D excitation classification map based on the boundaries set by the demarcation lines.The solid and dashed line represent the primary disc galaxy axis (∼ 28 • with respect to North) and an ENLR axis along (∼ 10 • with respect to North) the strongest emission, respectively.The black shaded area around the axis indicates the 2 -wide pseudo long-slit extraction region for the metallicity gradients shown in Fig. 8. Bottom right panel: Spatial distribution of the SF fraction as defined from the emission-line ratios.

Fig. 8 .
Fig.8.Radial metallicity gradient as a function of radius based on the N2S2 index for 2 -wide pseudo long-slits along galaxy disc (black points) and along the ENLR bicone (red points) as defined in Fig.7.The best-fit linear relations are shown as solid lines in the respective color.

Fig. 9 .
Fig. 9. Kinematic modelling of the molecular gas traced by CO(1-0).Upper panels: Moment maps of the CO(1-0) line from the ALMA observations.The beam size is indicated in all panels as the filled blue elliptical in the lower right for comparison.Middle panels: Best-fit kinematic model determined with the KinMS software tool (Davis et al. 2013a).Lower panels: Residuals of the best-fit model maps compared to the original data.

Fig. 10 .
Fig.10.Position-velocity (left panel) and position-dispersion (right panel) diagrams along the major axis of HE1353-1917.The black contours are for the cold molecular gas and the colored data points are the curves for the ionized and the warm molecular phase where the latter is limited by the much smaller NIFS FoV.The maximum velocity curve from the best-fit MGE model prediction is shown in the PV diagram on the left.Within the central 3 we detect significant deviation from the PVD model (left) which may be related either to a galaxy ring structure of a bipolar outflow."Hot spots" with an excess in velocity dispersion (right) are found on both sides of the nucleus in the same region.

Fig. 11 .
Fig. 11.Left panels: The velocity dispersion maps for the [O III], H 2 [1-0 S(2)], and CO(1-0) emission lines (assuming a single Gaussian LOSVD) in the central 5 × 5 (∼ 3.5 kpc × 3.5 kpc) for [O III], H 2 and CO(1-0).The black contours indicate the [OIII] surface brightness distribution as a reference in all maps.The green cross marks the position of the AGN and the light blue contours represent the distribution of 10 GHz radio continuum emission as resolved with our VLA radio interferometric observations.The red circles indicates a circular aperture of 0.8 in radius which we use to extract spectra covering the two high dispersion region located south west and north east of the nucleus.Middle and right panels: Spectra of the high dispersion regions in various emission lines from the optical, NIR and sub-mm observations for the south-west (middle) and north-east region (right).Most emission lines reveal an asymmetry towards the blue or red side that we model with two Gaussian line components which we refer to as the core and wing component.Only Paβ and H 2 [1-0 S(2)] can be well-described by only a single Gaussian component.
3 mag A V,wing > 3.5 mag, which highlights the severe extinction along the LOS for the broad component at optical wavelengths and weak signature in the [OIII] line on the north-east side of the nucleus.Based on the measured attenuations, we obtained extinction-corrected Hα luminosities of log(L Hα,core /[erg s −1 ]) = 40.79± 0.04 and log(L Hα,wing /[erg s −1 ]) = 40.34± 0.06 on the south-west side and log(L Hα,core /[erg s −1 ]) = 40.05± 0.08 and log(L Hα,wing /[erg s −1 ]) > 40.85 on the north-east side.Here, we face the issue that the luminosity of the wing component is highly uncertain on the North side due to the severe extinction.The electron density can be estimated from the [S II]

Fig. 13 .
Fig. 13.Mass outflow rate (left panel), momentum injection rate (middle panel) and kinetic energy injection rate (right panel) through a sphere at 1 kpc radius as a function of shell thickness.Measurements based on v br and v max are shown in top panels and bottom panels, respectively.Combined molecular+ionized gas measurements are shown as shaded (shell assumption) or hatched (cone assumption) areas highlighting the one order of magnitude uncertainty of X CO , while pure ionized gas mass computations results are represented by solid lines (shell assumption) or dashed lines (cone assumption).For comparison, the magenta lines highlight the SFR of the entire galaxy from the FIR measurements in the left panel, the point where L bol /c = ṗout in the middle panel, and the fraction of L bol at 10% (solid line), 1% (dashed line) and 0.1% (dotted line) as well as the jet power P jet (yellow dashed line) in the right panel.

Fig. 15 .
Fig. 15.Upper panel: Specific star formation rate against stellar mass.Contours highlight the distribution of local galaxies as probed by SDSS DR7 (Brinchmann et al. 2004).The position of HE 1353−1917 is denoted by the filled symbols, where the SFR is traced by the FIR continuum (blue symbol) and by the Hα emission lines (red symbol) after deblending with the ENLR contribution as discussed in the text.The open symbols represent a local sample of luminous AGN with Hα-based SFR measurements from Husemann et al. (2014) for comparison.Lower panel: Star formation rate (SFR) against molecular gas mass (M H 2 ).The gray symbols represent the properties of a local galaxy sample from the COLDGASS survey (Saintonge et al. 2011).The dashed diagonal lines represent constant value of the gas depletion time.The colored symbols are the measurements for HE 1353−1917 as in the upper panel.
The work of SAB, CPO and MS was supported by a generous grant from the Natural Sciences and Engineering Research Council of Canada.Based on observations collected at the European Organization for Astronomical Research in the Southern Hemisphere under ESO programme 095.B-0015(A).Based on observations obtained at the Gemini Observatory, which is operated by the Association of Universities for Research in Astronomy, Inc., under a cooperative agreement with the NSF on behalf of the Gemini partnership: the National Science Foundation (United States), National Research Council (Canada), CONICYT (Chile), Ministerio de Ciencia, Tecnología e Innovación Productiva (Argentina), Ministério da Ciência, Tecnologia e Inovação (Brazil), and Korea Astronomy and Space Science Institute (Republic of Korea).Based on observations obtained at the Southern Astrophysical Research (SOAR) telescope, which is a joint project of the Ministério da Ciência, Tecnologia, Inovações e Comunicações (MCTIC) do Brasil, the U.S. National Optical Astronomy Observatory (NOAO), the University of North Carolina at Chapel Hill (UNC), and Michigan State University (MSU).Based on observations collected at the German-Spanish Astronomical Center, Calar Alto, jointly operated by the Max-Planck-Institut für Astronomie Heidelberg and the Instituto de Astrofísica de Andalucía (CSIC).This paper makes use of the following ALMA data: ADS/JAO.ALMA#2016.1.00952.S. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada), MOST and ASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Republic of Chile.The Joint ALMA Observatory is operated by ESO, AUI/NRAO and NAOJ.This work is based on observations obtained with XMM-Newton, an ESA science mission with instruments and contributions directly funded by ESA Member States and NASA.The VLA is operated by the National Radio Astronomy Observatory, a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.

Table 1 .
Summary of observational characteristics and depth

Table 2 .
Spatially-integrated multi-band SED of HE 1353-1917 Marino et al. 2013oral [O III] λ4363 emission line is outside the wavelength range covered by MUSE we can only use strong-line methods to infer the gas-phase metallicity of the ionized gas.Marino et al. 2013).In the case of HE 1353−1917 we cannot directly apply those metallicity indicators, because a significant part of the galaxy is contaminated by strong AGN ionization due to the orientation of the ionization cone.Although variations in the [N II] λ6583/Hα line ratio have been attributed to changes in the metallicity even for AGN-ionized gas clouds based on photo-ionization models (e.g.Storchi-Bergmann et al.

Table 3 .
MGE parametrization of the galaxies K s band light profile.

Table 5 .
Emission-line measurements for the southern outflow region

Table 6 .
Mass outflow rate, momentum injection rate and kinetic energy injection at a distance of 1 kpc from the nucleus