MNRAS 000, 1–20 (2015) Preprint 8 June 2023 Compiled using MNRAS LATEX style file v3.

Outflow densities and ionisation mechanisms in the NLRs of the

prototypical Seyfert galaxies NGC 1068 and NGC 4151
Luke R. Holden,1★ Clive N. Tadhunter,1
1 Department of Physics & Astronomy, University of Sheffield, S6 3TG Sheffield, UK.

Accepted XXX. Received YYY; in original form ZZZ

arXiv:2306.03920v1 [astro-ph.GA] 6 Jun 2023

ABSTRACT
Despite being thought to play an important role in galaxy evolution, the true impact of outflows driven by active galactic nuclei
(AGN) on their host galaxies is unclear. In part, this may be because electron densities of outflowing gas are often underestimated:
recent studies that use alternative diagnostics have measured much higher densities than those from commonly used techniques,
and consequently find modest outflow masses and kinetic powers. Furthermore, outflow ionisation mechanisms — which are
often used to probe acceleration mechanisms — are also uncertain. To address these issues, we have analysed archival HST/STIS
spectra of the inner regions (r < 160 pc) of the nearby prototypical Seyfert galaxies NGC 1068 and NGC 4151, which show
evidence of warm-ionised outflows driven by the central AGN. We derive high electron densities (103.6 < 𝑛𝑒 < 104.8 cm−3 ) using
the transauroral [OII] and [SII] emission lines ratios for the first time with spatially-resolved observations. Moreover, we find
evidence that the gas along the radio axis in NGC 1068 has a significant AGN-photoionised matter-bounded component, and
there is evidence for shock-ionisation and/or radiation-bounded AGN-photoionisation along the radio axis in NGC 4151. We
also note that the outflow extents are similar to those of the radio structures, consistent with acceleration by jet-induced shocks.
Taken together, our investigation demonstrates the diversity of physical and ionisation conditions in the narrow line regions of
Seyfert galaxies, and hence reinforces the need for robust diagnostics of outflowing gas densities and ionisation mechanisms.
Key words: galaxies: active – galaxies: evolution – galaxies: individual: NGC 1068 – galaxies: individual: NGC 4151 – galaxies:
Seyfert – ISM: jets and outflows

1 INTRODUCTION et al. 2014; Harrison et al. 2014; Rose et al. 2018; Riffel 2021). How-
ever, many key outflow properties are highly uncertain, leading to a
Active galactic nuclei (AGN) can drive gas outflows through
wide range of observationally-derived coupling efficiencies (Harri-
radiation-pressure driven winds from their accretion disks (Di Mat-
son et al. 2018). For the warm ionised outflow phase (i.e. traced by
teo et al. 2005; Hopkins & Elvis 2010) and/or radio jets (Axon et al.
[OIII] and H𝛽; 10,000 < 𝑇𝑒 < 25,000 K), the largest source of uncer-
1998; Wagner & Bicknell 2011; Mukherjee et al. 2018). These out-
tainty is likely to be the electron density of the outflowing gas, which
flows, as well as the heating and ionising of near-nuclear gas, may
is often estimated or assumed to be in the range 𝑛𝑒 ∼100–1000 cm −3
constitute an important part of ‘AGN feedback’, which now routinely
(e.g. Kraemer & Crenshaw 2000a; Nesvadba et al. 2006; Fiore et al.
plays a crucial role in theoretical models of galaxy evolution. AGN-
2017). This is because the commonly used ‘traditional’ density diag-
feedback is required to explain observed galaxy properties (e.g. Di
nostics — the [SII](6717/6731) and [OII](3726/3729) emission-line
Matteo et al. 2005; Somerville et al. 2008; Schaye et al. 2015; Dubois
doublet ratios — are only sensitive up to 𝑛e ∼ 103.5 cm −3 , and are
et al. 2016; Davé et al. 2019 and empirical scaling relations between
often blended in the case of complex outflow kinematics. However,
super massive black holes and host galaxy properties (e.g. Magorrian
in recent years, alternative density diagnostics have been developed
et al. 1998; Silk & Rees 1998; Fabian 1999; Gebhardt et al. 2000; Fer-
and used, such as detailed photoionisation modelling that makes use
rarese & Merritt 2000). Models often require that the kinetic power
of a wide range of emission lines (Collins et al. 2009; Crenshaw
(E¤ kin ) of the outflowing gas is above a certain fraction of the AGN
et al. 2015; Revalski et al. 2021, 2022), and a technique involv-
bolometric luminosity (Lbol ): this is characterised by a ratio known
ing ionisation parameter measurements with infrared estimates of
as the ‘coupling factor’ (𝜖 𝑓 =E¤ kin /Lbol ), and is typically required to
outflow radii (Baron & Netzer 2019). Such methods have measured
be in the range 0.5 <𝜖 f <10 per cent (Di Matteo et al. 2005; Springel
higher electron densities for the warm ionised phase than commonly-
et al. 2005; Hopkins & Elvis 2010).
used traditional techniques, up to 𝑛𝑒 ∼ 105.5 cm −3 . Studies which
Observational studies commonly attempt to quantify the impact
make use of the higher critical density ‘transauroral’ (‘TR’; Boyce
of outflows on their host galaxies by comparing measured coupling
et al. 1933) [OII](3726 + 3729)/(7319 + 7331) and [SII](4068 +
efficiencies to those required by models (e.g. Liu et al. 2013; Cicone
4076)/(6717 + 6731) diagnostic ratios have similarly found densities
in the range of 103 < 𝑛e < 105.5 cm −3 (Holt et al. 2011; Rose et al.
★
2018; Santoro et al. 2018; Spence et al. 2018; Ramos Almeida et al.
E-mail: [email protected]

© 2015 The Authors

2 L.R. Holden et al.
2019; Santoro et al. 2020; Davies et al. 2020; Speranza et al. 2022; using Very Large Telescope (VLT) / Xshooter ultraviolet (UV), opti-
Holden et al. 2023). Considering that the derived outflow kinetic cal and near-infrared (NIR) spectroscopy: we found electron densities
power is inversely proportional to the electron density, if electron just above the critical density of the traditional [SII] ratio, and evi-
densities are truly orders of magnitude higher than are commonly dence for a post-shock cooling sequence and reionisation via AGN
assumed or estimated, the resulting kinetic powers and coupling fac- photoionisation.
tors for the warm ionised phase will be orders of magnitude lower. There is a clear need to determine whether the conditions found
This could significantly change our understanding of the importance in the narrow line region (NLR) of IC 5063 are similar in other
of AGN feedback in galaxy evolution. Seyfert galaxies, specifically to further investigate the true outflow
Moreover, where possible, it is important to use spatially-resolved gas density, kinetic powers, and ionisation mechanisms present on
observations when deriving electron densities, since global electron different spatial scales. Therefore, here we analyse archival Hubble
densities may significantly underestimate the values at small radial Space Telescope (HST) / Space Telescope Imaging Spectrograph
distances from the nucleus, where the outflows are the most extreme (STIS) spectra of the inner NLRs (𝑟 < 160 kpc) of the prototypical
(Revalski et al. 2022; but see also Kakkad et al. 2018). Thus, detailed Seyfert galaxies NGC 1068 and NGC 4151, and apply and expand
spatially-resolved observations are needed to robustly assess electron upon many of the techniques presented in Holden et al. (2023). We
densities in different types of AGN, as well as to compare and verify take the distances to NGC 1068 and NGC 4151 to be 𝐷 = 13.0 Mpc
different density diagnostics. (Revalski et al. 2021) and 𝐷 = 15.8 Mpc (Yuan et al. 2020), respec-
Investigations into the impact of outflows on their host galaxies tively, which correspond to spatial scales of 0.067 kpc/arcseconds
are further complicated by the fact that the dominant acceleration for NGC 1068 and 0.078 kpc/arcseconds for NGC 4151.
and ionisation mechanisms are unclear: while it is thought that out- The structure of the paper is as follows: in Section 2, we intro-
flows may be accelerated by radiation pressure from the AGN (either duce the prototypical Seyfert galaxies NGC 1068 and NGC 4151;
‘in situ’: e.g. Crenshaw et al. 2015; Fischer et al. 2017; Revalski in Section 3, we detail the archival HST/STIS observations and our
et al. 2018; Meena et al. 2023, or from the nucleus: e.g. Hopkins data reduction and handling processes; in Section 4, we present our
& Elvis 2010; Meena et al. 2021), a study of a large sample of analysis of the STIS data; in Section 5, we discuss the implications
local AGN found a link between intermediate radio power AGN of our findings, and in Section 6 we give our conclusions.
(L1.4 GHz = 1023−25 W Hz −1 ) and outflow kinematics (Mullaney
et al. 2013), suggesting that feedback from jets is also important in
AGN that are classified as radio-quiet. Indeed, hydrodynamic simu- 2 TWO PROTOTYPICAL SEYFERTS: NGC 1068 &
lations have shown that jets interacting with the ISM on kpc-scales NGC 4151
can explain observed gas kinematics in some objects (e.g. Mukherjee
et al. 2018; Audibert et al. 2023), and may have both a positive and NGC 1068 and NGC 4151 appeared in Carl Seyfert’s original paper
negative impact on local star formation rates (Mandal et al. 2021). that established the Seyfert class (Seyfert 1943), and are respectively
Therefore, determining dominant acceleration mechanisms is crucial the prototypical Seyfert 2 (Sey2) and Seyfert 1 (Sey1) galaxies. In
for facilitating proper comparisons between observations and predic- consequence, they are perhaps the most well-studied AGN of their
tions from theoretical modelling, which are needed to interpret the respective types. Their close proximity to Earth, and the previous,
role of outflows in AGN feedback. extensive multi-wavelength studies of their properties, make them
The ionisation and excitation mechanisms of the outflowing gas ideal objects for our project: the outflows in their central regions
may provide clues as to the acceleration mechanism(s) present. For can be spatially resolved, and we can compare our results to those
example, shock-ionised gas must have passed through (and been ac- obtained using other methods. Principally, this allows us to assess
celerated by) a shock. However, AGN-photoionised gas may have the validity of the different density diagnostic techniques, as well as
been previously accelerated by another mechanism, and reionised by investigate the ionisation of the gas.
photons from the AGN after cooling (Holden et al. 2023). Hence,
the true nature of the relationship between outflow acceleration and
2.1 NGC 1068
ionisation mechanisms is complex, and requires further careful anal-
ysis. NGC 1068 is one of the closest and brightest (in terms of observed
Regardless of how outflows are accelerated, understanding the flux) Seyfert 2 galaxies, allowing detailed spatially-resolved obser-
dominant ionisation mechanisms impacts our ability to extract key vations, and thus making it the target for extensive studies that cover
diagnostic information for the warm outflowing gas. Specifically, the a range of spatial scales in the optical (e.g. Cecil et al. 1990; Evans
techniques presented by Holt et al. (2011) and Revalski et al. (2021) et al. 1991; Axon et al. 1998; Crenshaw & Kraemer 2000b; Krae-
(see also Collins et al. 2009) both rely on photoionisation models, mer & Crenshaw 2000b; Das et al. 2006), NIR (e.g. Raban et al.
and the transauroral lines (in the case of the Holt et al. 2011 method) 2009; Müller Sánchez et al. 2009; May & Steiner 2017) and ra-
cannot be emitted by a matter-bounded component. If, in reality, dio (e.g. Wilson & Ulvestad 1983; Gallimore et al. 1996; García-
a gas outflow is shock-ionised or has a large contribution from a Burillo et al. 2014, 2019). NGC 1068 has a radio luminosity of
matter-bounded component, this may have a significant impact on L1.4 GHz = 2.3 × 1023 W Hz −1 (Ulvestad & Wilson 1984), placing
the validity of these methods. Thus, it is important to investigate it in the upper end of the radio luminosity range for Seyfert galaxies,
the ionisation mechanisms present in active galaxies for which these and its high bolometric luminosity (0.4 < 𝐿 bol < 4.7×1038 W: Woo
techniques have been applied in the past, as well the potential im- & Urry 2002; Alonso-Herrero et al. 2011; Lopez-Rodriguez et al.
pact of matter-bounded components or shock ionisation on derived 2018; Gravity Collaboration et al. 2020) is close to the lower bound-
densities. ary of the luminosity range for quasars (Lbol > 1038 W). The galaxy
In order to address these issues, we are undertaking detailed also has an important historical role, as it was the first object used to
spatially-resolved studies of nearby AGN that show clear evidence of verify the orientation-based unified scheme for AGN (Antonucci &
outflows on pc to kpc scales. In Holden et al. (2023), we presented a Miller 1985).
detailed study of the central regions of the nearby Seyfert 2 IC 5063 The NLR of NGC 1068 presents as an ‘hourglass’-shaped bicone

MNRAS 000, 1–20 (2015)

 NLR warm-ionised outflows in NGC 1068 and NGC 4151 3
(Riffel et al. 2014; Barbosa et al. 2014; May & Steiner 2017) with NGC 1068 (May & Steiner 2017) — albeit on smaller spatial scales
an opening angle of 𝜃 ∼ 40◦ along PA= 30 ± 2◦ at an inclination with less extreme kinematics — to explain the NLR and outflow
of 𝑖 = 5◦ , placing the bicone axis close to the plane of the sky and structure in NGC 4151: the radio jet impacts a molecular cloud near
inclined ∼ 45◦ out of the galaxy’s disk (Das et al. 2006; but see the nucleus (potentially due to misalignment between the jet and
also Crenshaw & Kraemer 2000b). Outflows of warm-ionised gas torus/disk: Storchi-Bergmann et al. 2010; May et al. 2020), driving
with velocities up to ∼1500 km s −1 have been detected in the bicone fragmented, shock-accelerated gas into the cones and contributing to
(Crenshaw & Kraemer 2000c; Das et al. 2006). In the NE cone, the the NLR morphology.
radio axis is closely aligned with the bicone axis — interpreted as a
radio jet propagating within the hollowed-out cone — with a radio
lobe that extends just beyond the maximum extent of the cone (Wilson 2.3 Previous photoionisation modelling of NGC 1068 and
& Ulvestad 1987; shown in Figure 1). Lower velocity cold molecular NGC 4151
CO(3-2) outflows have been detected at this position, indicating that Crenshaw et al. (2015) and Revalski et al. (2021, 2022) performed
the lobe may represent the termination of the AGN-driven outflows detailed, multi-ionisation component photoionisation modelling of
García-Burillo et al. (2014). the warm ionised outflows in NGC 1068 and NGC 4151, finding den-
The outflows in the NLR of NGC 1068 have been argued to be ra- sities in the range 103.0 cm −3 < 𝑛𝑒 < 107.2 cm −3 for the NLR gas in
diatively accelerated by some authors (Kraemer & Crenshaw 2000b; both objects, and coupling efficiencies above the lower limit required
Das et al. 2006; Revalski et al. 2021; Meena et al. 2023), while others by galaxy evolution models (0.5 per cent: Hopkins & Elvis 2010) in
have proposed they are driven by jet-induced shocks (Capetti et al. the case of NGC 1068. In order to further investigate the electron
1997; Axon et al. 1998). May & Steiner (2017) propose a scenario in densities of the outflowing gas in the NLR of these two important
which the radio jet impacts molecular clouds on small radial scales objects, and to attempt to clarify the uncertainties regarding the ac-
near the central AGN, accelerating high-velocity ‘bullets’ of gas that celeration and ionisation mechanisms of the gas, we require high
propagate within the bicone but constitute only a small fraction of spatial resolution, wide wavelength-coverage long-slit spectroscopy
the total outflowing mass. with the slit aligned along the radio axes (which is approximately
along the bicone axes).

2.2 NGC 4151

NGC 4151 is the prototypical Seyfert 1 (Sey1) galaxy1 and is also 3 OBSERVATIONS AND DATA REDUCTION
one of the closest and brightest (in terms of observed flux) of its
3.1 Archival HST/STIS observations
class, leading to its NLR outflows being the target of extensive stud-
ies of the coronal (e.g. Storchi-Bergmann et al. 2009, 2010), warm To achieve our science goals, suitable archival HST/STIS long-slit
ionised (e.g. Winge et al. 1997; Hutchings et al. 1999; Crenshaw spectra were downloaded from the Hubble Legacy Archive (https:
et al. 2000; Das et al. 2005; May et al. 2020) and warm molecu- //hla.stsci.edu/hlaview.html). We required data taken using
lar (H2 ; 𝑇 ∼2000 K, e.g. May et al. 2020) gas phases, which have both the G430L and G750L gratings in order to ensure sufficient
distinct flux distributions (Storchi-Bergmann et al. 2009). Similar to wavelength coverage, namely that the spectra contained the blue
NGC 1068, the bicone-shaped NLR also has an hourglass morphol- [SII]𝜆𝜆4068, 4076 and red [OII]𝜆𝜆7319, 7331 transauroral doublets.
ogy (May et al. 2020), with PA=22◦ at an inclination of 𝑖 = 21◦ Both gratings have a spatial pixel scale of 0.051 arcseconds per pixel,
(Pedlar et al. 1992; 36◦ to the galactic disk) and an opening angle and the dispersions of the two gratings are 2.72 Å/pixel (G430L;
of 33◦ (Das et al. 2005). However, the bolometric luminosity of the 2900–5700 Å) and 4.92 Å/pixel (G750L; 5240–10270 Å). We also
AGN in NGC 4151 (𝐿 bol = 1.4 × 1037 W) is approximately an order required that these data were taken along (or close to) the PA of
of magnitude below that of NGC 1068. the radio/bicone structures to ensure we are tracing the the gas that
The radio source (of luminosity L1.4 GHZ =1.6×1022 W Hz −1 ; Ul- is impacted most by the jet. The data for NGC 1068 were taken as
vestad & Wilson 1984) consists of a double sided jet (PA ∼ 77◦ ) part of the Cycle 7 HST Proposal GTO:7573 (PI Kraemer), with a
originating from the nucleus. High-resolution radio imaging (Carral 52×0.1 arcsecond slit along PA=202◦ , centred on a bright emission-
et al. 1990; Pedlar et al. 1993; Williams et al. 2017) shows several line knot close (<0.4′′ ) to the nucleus (see Crenshaw & Kraemer
radio knots along this structure within the central few arcseconds, 2000a and Kraemer & Crenshaw 2000a). Data for NGC 4151 were
whereas lower-resolution radio observations (Johnston et al. 1982; taken with a 52×0.1 arcsecond slit along PA=70◦ , offset to the south
Pedlar et al. 1993) reveal a larger-scale lower surface brightness struc- by 0.1 arcsecond to reduce contamination from the bright Sey1 nu-
ture with a radio lobe in the NE cone extending to 6.3 arcseconds cleus, and were taken in Cycle 7 as part of HST Proposal GTO:7569
from the nucleus along the radio axis. It has been argued that the radio (PI Hutchings) — a full description of the NGC 4151 observations
jet has little connection to the NLR outflow kinematics in NGC 4151 is given by Nelson et al. (2000). We show the positions of the STIS
(Hutchings et al. 1999; Crenshaw et al. 2000; Das et al. 2005). How- slits over the central regions of the two Seyferts in Figure 1.
ever, enhanced line fluxes from the warm ionised gas, high electron
temperatures (Te >16,000 K) and high [FeII]/[PII] ratios have been
spatially associated with the radio structure (Mundell et al. 2003; 3.2 Reduction and handling of STIS data
Storchi-Bergmann et al. 2009, 2010), indicating that jet-ISM inter- 3.2.1 Data reduction
actions may still drive shocks into the gas at certain locations within
the bicone (see also Wang et al. 2011a,b; Williams et al. 2017). The first step in the data reduction was performed with the standard
May et al. (2020) propose a similar model as they proposed for CALSTIS pipeline. For the NGC 1068, only a single exposure for
each grating was available, while for NGC 4151 we took the aver-
age of two exposures for each grating using Python scripts which
1 NGC 4151 was later classified as an intermediate ‘Seyfert 1.5’ (Osterbrock made use of the Numpy (Harris et al. 2020) and AstroPy (Astropy
& Koski 1976; Robinson et al. 1994). Collaboration et al. 2013, 2018) modules. In order to ensure that

MNRAS 000, 1–20 (2015)

4 L.R. Holden et al.

Figure 1. The STIS slits of our archival observations (red) shown plotted over archival HST/WFPC2 [OIII] emission-line images of the inner regions of
NGC 1068 and NGC 4151, taken with the F502N filter (NGC 1068: GTO:5754, PI Ford; NGC 4151: GTO:5124; PI Ford). The extents of our apertures (Section
3.2.2) along the slits are shown in blue. Left: The STIS slit shown over the [OIII] emission-line image of the near-nuclear regions of NGC 1068. VLA 22 GHz
contours from Gallimore et al. (1996) are presented in green, showing the radio structure near the core and an extended lobe to the NE. Right: the STIS slit
shown over the [OIII] emission-line image of the near-nuclear regions of NGC 4151; the green contours are from high-resolution eMERLIN 1.5 GHz imaging
presented by Williams et al. (2017), and show a string of radio knots near the nucleus. We note that, while the narrow-band images are not continuum-subtracted,
the brighter parts of the NLR emission are dominated by [OIII] emission in the filter bandpass, and so the images provide a good representation of the main
NLR structures.

the individual exposures for each grating were aligned, we first ex- 3.2.2 Aperture selection and extraction
tracted spatial slices along the slit direction in a line-free region
of the continuum covering the wavelength range 5480–5600 Å for The STIS long-slit spectra of NGC 1068 and NGC 4151 show dis-
the G430L grating and 6795—6890 Å for the G750L grating. The turbed kinematics (indicating outflows) and several bright emission-
centroids of the spatial peaks — determined with Gaussian profile line knots in the central few hundred parsecs, as noted by previous
fits — were consistent within better than 0.4 pixels, confirming that studies (Crenshaw & Kraemer 2000b; Kraemer & Crenshaw 2000a;
each exposure was taken with the same telescope pointing within Das et al. 2005, 2006; Meena et al. 2023). We extracted several aper-
0.02 arcseconds. We also checked that the spectra taken with the tures (integrated groupings of pixel rows) from the two-dimensional
G430L and G750L gratings for the each object were aligned, using G430L and G750L spectra, with each aperture forming an integrated
the same method of Gaussian fits to the spatial flux profiles. Again, one-dimensional spectrum that corresponds to a certain spatial posi-
the spatial positions of the peak flux between gratings were consis- tion along the slit. We selected the apertures to cover the locations of
tent to within better than 0.4 pixels, indicating that the observations the bright emission knots seen in our two-dimensional spectra (Fig-
with different gratings were closely spatially aligned. ure 2). The widths of the apertures (6–15 pixels; 0.3–0.8 arcseconds)
were set to contain sufficient signal in the fainter emission lines that
are used for diagnostics in our analysis, namely the fainter transauro-
ral [OII]𝜆𝜆7319,7331 and [SII]𝜆𝜆4068,4076 doublets. We extracted
the same apertures from the G430L and G750L spectra for each
Residual hot pixels and cosmic rays were removed from the object, as we previously determined that the spectra were closely
spectra using the CLEAN command from the STARLINK FI- spatially aligned (Section 3.2.1). Flux errors were determined by
GARO software package (Currie et al. 2014). We then corrected adding the flux errors from individual pixel rows (which constitute
for extinction due to dust in the Milky Way using the Galactic a given aperture) in quadrature. As an example, we present part of
extinction maps presented by Schlegel et al. (1998) and recali- the spectrum of Aperture 2 for NGC 1068 in Figure 3. The chosen
brated by Schlafly & Finkbeiner (2011). Using the NASA/IPAC apertures extended out to a maximum radial distance of 139 pc for
Infrared Science Archive reddening lookup tool (https://irsa. NGC 1068, and 151 pc in the case of NGC 4151.
ipac.caltech.edu/applications/DUST/) with these maps, we Aperture 3 for NGC 1068 was placed over a bright emission knot
find that there are mean colour excesses in the directions of that corresponds to a previously detected radio source at the likely
NGC 1068 and NGC 4151 of E(B-V)mean = 0.0289 ± 0.0004 and position of the galaxy’s nucleus (see discussion in Kraemer & Cren-
E(B-V)mean = 0.0237 ± 0.0011 respectively. The 𝑅 𝑣 = 3.1 extinc- shaw 2000a), while Aperture 4 for NGC 4151 corresponds to the
tion law presented by Cardelli et al. (1989) (hereafter CCM89) was location along the slit that is closest to the nucleus. We note that the
then used to correct for Galactic extinction. spectra for NGC 4151 do not directly cover the nucleus, due to the

MNRAS 000, 1–20 (2015)

 NLR warm-ionised outflows in NGC 1068 and NGC 4151 5

Figure 2. Selected apertures for NGC 1068 (left) and NGC 4151 (right), positioned over the [OIII]𝜆𝜆4959,5007 doublet in the two-dimensional STIS G430L
spectra. The spectral direction is horizontal (left = bluewards; right = redwards) and the vertical direction is spatial along the slit (with the direction shown by the
labelled arrows); velocity scale bars are shown in green, and the spatial extents in arcseconds and parsecs are shown to the right of each spectrum. The apertures
are shown as regions bounded by dashed lines, and are labelled on the left of each image — they were chosen to contain enough signal for the measurement of
faint lines in distinct kinematic regions within the central few hundred parsecs of each galaxy.

0.1 arcsecond slit offset to the south to avoid nuclear contamination. In order to verify whether stellar continuum modelling was needed
Unfortunately, the south-west part of the slit for NGC 1068 (seen in this study, we measured the equivalent widths (EWs) for the H𝛽
above Aperture 4 in Figure 2) did not contain enough signal for the recombination line. We find 36 < EW < 148 Å in our NGC 1068
measurement of the faint [OII]𝜆𝜆7319, 7331 transauroral doublet, apertures and 30 < EW < 151 Å in the NGC 4151 apertures. The
even when integrated as a single aperture. Therefore, we omit this lowest emission-line equivalent width we measure (EW= 30 Å for
region from our analysis. Aperture 3 in NGC 4151) is a factor of three higher than that of
Following aperture extraction, we ensured that the flux calibra- the H𝛽 absorption feature as modelled for a ∼400 Myr old stellar
tion was consistent between the two gratings for each aperture by population (which gives the highest EWs in modelling by González
overplotting the spectra in the region where the wavelength ranges Delgado et al. 1999). Thus, underlying stellar absorption features
of the gratings overlap (5275–5705 Å). We found that all apertures may affect our measured H𝛽 luminosities by a maximum factor of
for NGC 1068 are closely matched in flux. However, for apertures 2 1.3 (for a stellar EW= 10 Å). However, this is very much an upper
and 4 of NGC 4151, the flux in the overlap region was >8 per cent limit since we do not detect a Balmer break in the continuum in any
higher in the G430L grating than the G750L grating, potentially due of our apertures, as would be expected for intermediate age stellar
to internal reflections within the instrument caused by the bright Type populations that have strong Balmer absorption lines.
1 nucleus (see Nelson et al. 2000). Therefore, we do not use these
apertures in further analysis.
3.2.4 Fits to key emission lines
The NLR kinematics in NGC 1068 and NGC 4151 are complex,
3.2.3 The contribution of stellar continua to the spectra
and have been previously modelled in detail as biconical outflows
We did not model and subtract the underlying stellar continuum based on higher resolution STIS spectra than those used here (Das
in detail using a template-fitting approach (as was done for similar et al. 2006 and Das et al. 2005 respectively; but see also Cren-
analyses of other objects by Rose et al. 2018 and Holden et al. shaw & Kraemer 2000c and Crenshaw et al. 2000). In those studies,
2023) for various reasons. First, our archival STIS G430L and G750L the [OIII]𝜆𝜆4959,5007 doublet line profiles were fit with multiple
spectra did not have sufficient spectral resolution to clearly resolve Gaussians for each pixel row of the 2D spectra. Here, we perform a
absorption features that could be used to verify the robustness of similar procedure for our extracted apertures by simultaneously fit-
the continuum fits. Second, there may be substantial contamination ting a 1st or 2nd order polynomial to the continuum surrounding the
by direct and scattered AGN continuum (Antonucci & Miller 1985) [OIII]𝜆𝜆4959,5007 doublet, and one or two Gaussian profiles to each
and nebular continuum (Tadhunter 2016) that precludes accurate of the lines in the doublet itself. We set the wavelength separation
stellar continuum modelling. Finally, the emission lines in our spectra of the lines in the doublet, as well as the intensity ratio of the lines
have relatively high equivalent widths, which fill in various stellar (1:2.99), to those defined by atomic physics (Osterbrock & Ferland
absorption features. 2006). Furthermore, we constrained the widths of a given Gaussian

MNRAS 000, 1–20 (2015)

6 L.R. Holden et al.

[OIII] 5007,4959
3.0

[OII] 4068,4076
[OII] 3726,3729
[FeVII] 3759
[NeIII] 3869
[NeV] 3426

[OIII] 4363

HeII 4686
2.5
(erg s 1 cm 2 Å 1)

H
2.0

1.5
12
Flux / 10

1.0

0.5

0.0
3400 3600 3800 4000 4200 4400 4600 4800 5000
Observed wavelength (Å)

Figure 3. The G430L grating spectrum for Aperture 2 of NGC 1068 (Figure 2). Key emission lines that are used in our analysis are labelled with dotted lines.
Note that, for presentation reasons, the limit on the flux axis has been chosen so that fainter lines can be clearly seen; as a result, the peak of the [OIII]𝜆5007 line
is not visible.

component to be the same for each line in the doublet. We present and FWHMinst = 12.3 Å (560 km s −1 at 6575 Å) for the G430L and
the model parameters for each aperture in Table 1. G750L gratings, respectively.
Once we had established [OIII]𝜆𝜆4959,5007 doublet fits in each
aperture, we calculated the difference between the mean wave-
length of each Gaussian component and the rest [OIII] wave-
length in the reference frame of the galaxy, using redshifts2 of
𝑧 = 0.00381 for NGC 1068 and 𝑧 = 0.003262 for NGC 4151.
We also determined the intrinsic width of each component by In subsequent analysis, we only consider total line fluxes — in-
subtracting the instrumental width of the STIS G430L grating in cluding all Gaussians components used — rather than fluxes from
quadrature from the measured widths. According to the STIS man- individual components (i.e. potentially representing outflowing and
ual, for a slit of width 0.1 arcseconds, the instrumental broaden- quiescent gas). This was done because of the low spectral reso-
ing in the spectral direction is in the range 2 < FWHM < 3 pix- lutions of the G430L and G750L gratings, which made it chal-
els, corresponding to 5.5 < FWHM < 8.2 Å for the G430L grating lenging to separate different kinematic components in cases where
and 9.8 < FWHM < 14.8 Å for the G750L grating. By fitting sin- lines are heavily blended. Nonetheless, in order to improve the
gle Gaussians to the [OIII]𝜆𝜆4959,5007 emission-line doublet at a accuracy of the fits to the weaker emission lines and blends in
radial distance of 4 arcseconds from the nucleus of NGC 4151 in the spectra, we used the kinematics (velocity shifts and widths)
the G430L spectra (where the lowest line widths are measured), we derived from fits to the [OIII] doublet in each aperture to con-
measure a line width of FWHMinst = 6.0 ± 0.4 Å; similarly, mea- strain the fits to the other key diagnostic lines used in our analysis,
suring the [SII]𝜆9531 line in the G750L spectra with this method such as H𝛽, H𝛾, [OIII]𝜆4363, [OII]𝜆3726,3729, [OII]𝜆𝜆7319,7331,
resulted in a line width of FWHMinst = 12.3 ± 2.4 Å. Thus, we adopt [SII]𝜆𝜆4068,4076, [SII]𝜆𝜆6717,6731, [ArIV]𝜆𝜆4711,4740 and
instrumental widths of FWHMinst = 6.0 Å (360 km s −1 at 5007 Å) HeII𝜆4686. We found that this procedure produced acceptable fits
to these lines, including the transauroral [SII]𝜆𝜆4068,4076 and
[OII]𝜆𝜆7319,7331 doublets. However, for closely spaced doublets
such as [OII]𝜆3726,3729, the low spectral resolution meant that we
2 21 cm redshifts from the NASA/IPAC Extragalactic Database (https: did not resolve individual lines, and so we modelled the total doublet
//ned.ipac.caltech.edu/). profile as a single emission line during the fitting process.

MNRAS 000, 1–20 (2015)

 NLR warm-ionised outflows in NGC 1068 and NGC 4151 7

Distance Distance 𝑣c,a FWHMc,a 𝑣c,b FWHMc,b

Aperture (km s −1 ) (km s −1 ) log10 (𝑛𝑒 [cm −3 ]) E(B-V)TR 𝑇𝑒 (K)
(arcseconds) (pc) (km s −1 ) (km s −1 )

NGC 1068

1 −1.45 −97 −828±4 572±25 295±40 1078±96 4.06+0.05

−0.06
0.16+0.04
−0.05
14300+2100
−1300
2 −0.74 −50 −184±7 1017±31 — — 4.65+0.05
−0.04
0.05+0.04
−0.05
14400+1500
−1100
3 −0.23 −15 −306±3 662±26 −5±20 1770±43 4.74+0.05
−0.05
0.16+0.05
−0.04
16100+1400
−1000
4 0.10 7 95±3 367±26 235±8 1684±34 4.45+0.09
−0.09
0.17+0.08
−0.08
16000+1600
−1100

NGC 4151

1 −1.58 −123 −172±2 420±25 −263±29 1261±102 3.68+0.08

−0.10
0.11+0.05
−0.07
16300+3400
−1800
3 −0.38 −30 −392±6 0±28 𝑎 −356±3 1065±26 4.04+0.07
−0.15
0.13+0.05
−0.06
21000+3300
−2100
5 0.48 37 34±1 234±24 121±11 1013±44 3.94+0.10
−0.10
0.15+0.08
−0.08
17300+5200
−2400
6 1.02 80 40±1 307±25 126±8 768±35 3.75+0.8
−0.10
0.23+0.07
−0.07
15200+3400
−1800
𝑎 The measured width of the component is consistent with the instrumental width, and hence unresolved.
Table 1. [OIII] model parameters (galaxy rest-frame component velocity shift: 𝑣c ; instrumentally-corrected component velocity width: FWHMc ), distances
from the nucleus (in arcseconds and pc), electron densities, reddening values and electron temperatures for each of our apertures for NGC 1068 and NGC 4151.
In apertures where there are multiple Gaussian components for the [OIII] models, we label the kinematic parameters for the two components with the subscripts
‘a’ and ‘b’. The densities and reddenings were determined simultaneously using the transauroral line technique (Section 4.1; Figure 4), and the temperatures
were determined using the [OIII](5007+4959)/4363 ratio (Section 4.2.1).

4 ANALYSIS OF THE STIS SPECTRA spectral index measured in radio-quiet AGN (Zamorani et al. 1981;
Miller et al. 2011), and is consistent with photoionisation modelling
4.1 Transauroral line diagnostics of the emission-line ratios of the extended and nuclear NLRs in var-
In order to provide estimates of the electron densities of the warm ious samples of AGN (e.g. Ferland & Netzer 1983, Robinson et al.
ionised gas in NGC 1068 and NGC 4151, we make use of a technique 1987). We note, however, that the TR ratios are relatively insensitive
first described by Holt et al. (2011) which requires measurement of to the shape of the ionising continuum (see Appendix B in Santoro
the transauroral [SII] and [OII] ratios: et al. 2020). We selected an ionisation parameter of log 𝑈 = −3 (the
highest value that reproduced the measured TR ratios) and varied
𝑇 𝑅([𝑂𝐼 𝐼]) = 𝐹 (3726 + 3729)/𝐹 (7319 + 7331), the electron density of the modelled gas in 0.01 dex steps between
2.00 < log10 (𝑛𝑒 [cm −3 ]) < 5.00. We then reddened the modelled TR
𝑇 𝑅([𝑆𝐼 𝐼]) = 𝐹 (4068 + 4076)/𝐹 (6717 + 6731). ratios produced for each electron density value with the 𝑅 𝑣 = 3.1
CCM89 law, producing a grid of values that we compared to our
In this technique, measured TR([OII]) and TR([SII]) ratios are com- measured ratios in order to provide simultaneous values of electron
pared to those expected from photoionisation modelling in order density and reddening. The resulting TR grid is shown in Figure 4,
to simultaneously derive electron densities and reddenings. This has and the derived values are given in Table 1.
several important advantages as a density diagnostic over commonly-
used, traditional methods. First, these lines have higher critical den-
sities (Appendix A), meaning that the TR ratios are sensitive to The electron densities measured in this way for NGC 1068 have
higher electron densities (𝑛e ∼ 105.5 cm −3 ) than the traditional values in the range 4.00 < log10 (𝑛𝑒 [cm −3 ]) < 4.75, while those
[OII](3726/3729) and [SII](6717/6731) density diagnostics, which for NGC 4151 are approximately an order of magnitude lower
are only sensitive up to 𝑛e ∼ 103.5 cm −3 . Furthermore, the TR (3.60 < log10 (𝑛𝑒 [cm −3 ]) < 4.10). This is the first time that den-
method uses the ratios of the total line fluxes of widely-separated sities above 𝑛𝑒 = 103.5 cm −3 have been found using the transauroral
emission-line doublets, unlike the traditional [SII] and [OII] tech- lines with spatially-resolved observations, and agree with similarly
niques, which rely on the flux ratios of lines within the doublets. This high electron densities derived using this technique for non-spatially
means that the TR ratios are less susceptible to uncertainties from fit resolved observations of other AGN (e.g. Holt et al. 2011; Rose
degeneracy resulting from the larger velocity widths (as often seen et al. 2018; Santoro et al. 2018; Spence et al. 2018; Davies et al.
for outflowing gas) and low spectral resolutions (as for our STIS 2020; Speranza et al. 2022). Importantly, the densities we find here
spectra) that lead to blending of line profiles within the doublets. are above the critical densities of the traditional [OII](3726/3729)
We used the [OIII] model fits to the TR lines to measure line and [SII](6717/6731) line ratios (Appendix A), and since we do not
fluxes, which were then used to calculate measured TR ratios. The separate broad (outflowing) and narrow (quiescent; non-outflowing)
CLOUDY code (version C17.02: Ferland et al. 2017) was then used components, are likely to be underestimates for the outflowing gas
to generate plane-parallel, single-slab, radiation-bounded models of (which is expected to be denser than the quiescent gas: e.g. Villar-
solar-composition gas with no dust depletion, photoionised by a cen- Martín et al. 1999; Holden et al. 2023). The reddenings that we mea-
tral source. We set the ionising continuum of this source to follow sure are relatively modest and in the range 0.05 < E(B-V)TR < 0.25
a power-law of shape 𝐹𝑣 ∝ 𝑣 − 𝛼 between 10 µm and 50 keV, with a for both objects — these values were used to deredden our spectra
spectral index of 𝛼 = 1.5. This is close to the average optical to X-ray for all further analysis.

MNRAS 000, 1–20 (2015)

 8 L.R. Holden et al.
NGC 4151. The high electron temperatures that we find in our aper-
tures for both objects may not be fully explainable as being due to
2.0 AGN-photoionisation of radiation-bounded gas (Fosbury et al. 1978;
E(B-V) 0 2.02.5 Binette et al. 1996; Villar-Martín et al. 1999; Holden et al. 2023).
0.25 3.0
1.5 4.2.2 Shock-ionisation vs matter-bounded AGN photoionisation
0.5
3.5 log10nH In order to investigate the cause of the high electron temperatures
further, we produced the [OIII](5007/4363) vs HeII/H𝛽 diagnostic
1.0
log10[OII]

4.0
diagram developed by Villar-Martín et al. (1999), as shown in Fig-
ure 5. The radiation-bounded photoionisation models shown here
are the same as those used for the TR ratio grid in Section 4.1
0.5 4.5
(Figure 4), albeit for an electron density of 𝑛𝑒 = 104 cm −3 , vary-
ing ionisation parameters (between -3.5 < log10𝑈 < -2.0), and two
values of spectral index (𝛼 = 1.0, 1.5). The pure shock and precur-
0.0
sor (pre-shock) models are taken from the MAPPINGS III library
5.0 presented by Allen et al. (2008), with varying shock velocities in
the range 100 < 𝑣 shock < 1000 km s −1 and magnetic parameters of
√
𝐵/ 𝑛 = 2,4 𝜇G cm3/2 for a solar-composition pre-shock gas with a
0.5 1.5 1.0 0.5 0.0 density of 𝑛 = 102 cm −3 . The magnetic parameters were chosen to
log10[SII] cover a reasonable range of values expected in the ISM (Dopita &
Sutherland 1995), in addition to being close to the magnetic param-
√
eters near equipartition (𝐵/ 𝑛 ∼ 3.23 𝜇G cm3/2 : Allen et al. 2008).
Note that we do not use the standard ‘BPT’ diagrams (Baldwin et al.
Figure 4. Grid of modelled transauroral (TR) [SII] and [OII] line ratios for
1981) to investigate the ionisation of the gas, because some of the
radiation-bounded gas at different electron densities and reddenings, (black
joined squares; as modelled with the Cloudy code and the CCM89 extinction
lines involved in those diagrams (such as H𝛼 and [NII]𝜆𝜆6548,6583)
curve) and measured line ratios for NGC 1068 (orange circles) and NGC 4151 are strongly blended in our apertures due to the outflow kinematics
(blue circles). and relatively low spectral resolution, and therefore are affected by
major fit degeneracies.
In Figure 5, we also plot [OIII](5007/4363) and HeII/H𝛽 as func-
4.2 Ionisation states and mechanisms of the warm gas tions of AM/I : the ratio of the solid angles subtended by matter-
bounded clouds and radiation-bounded clouds, from modelling by
The relatively low-ionisation transauroral lines must be emitted by
Binette et al. (1996). This ratio allows us to estimate the rela-
radiation-bounded clouds. Therefore it is uncertain how well den-
tive contribution of matter-bounded clouds and radiation-bounded
sities derived from the transauroral ratios would represent the den-
clouds in our apertures. The modelling by Binette et al. (1996) as-
sities of clouds or cloud complexes that have been shock-ionised
sumes solar-metallicity gas, with an ionising source spectral index
or have significant matter-bounded components. Furthermore, the
of 𝛼 = −1.3, an ionisation parameter of log 𝑈 = −1.4, and a density
model used in the transauroral ratio method assumes radiation-
of 𝑛MB = 50 cm −3 . The radiation-bounded clouds are ionised by
bounded AGN-photoionised clouds, with no contribution from a
UV photons which have passed through the matter-bounded compo-
matter-bounded component or shock-ionisation. Similarly, the multi-
nent, thus the shape of the ionising spectrum reaching the radiation-
component ionisation modelling by Revalski et al. (2021) — which
bounded clouds has changed relative to that from the source — the
has previously been applied to NGC 1068 and NGC 4151 — uses
parameters of the radiation-bounded clouds are determined using
AGN photoionisation models. Therefore, it is important to investi-
the resulting ionising spectrum and by assuming that the clouds have
gate the ionisation mechanisms for the gas detected in our STIS slits,
fixed pressures.
which potentially can also give information regarding the outflow
Due to the continuum underlying the H𝛽, [HeII]𝜆4686, and
acceleration mechanism(s) present.
[OIII]𝜆4363 lines being more complex than that which underlies
the [OIII]𝜆𝜆4959,5007 doublet and transauroral lines, we used a
MCMC (Markov Chain Monte Carlo) fitting routine to fit the lines
4.2.1 Electron temperatures
involved in the HeII𝜆4686/H𝛽 and [OIII](5007/4363) ratios in each
Electron temperatures of the warm ionised phase are expected to aperture for both objects — this was done to ensure that we were not
be higher for shocked gas than AGN-photoionised gas (e.g. Fosbury significantly overestimating line flux uncertainties due to blending
et al. 1978; Villar-Martín et al. 1999). Therefore, to provide a first of spectral lines and the continuum. We used the results of the Gaus-
indication of the ionisation mechanisms of the warm ionised gas ob- sian fits described in Section 3.2.4 (determined using least squares
served in our apertures, we measured electron temperatures using the optimisation) to these lines as initial starting points for the MCMC
(dereddened) [OIII](5007+4959)/4363 emission-line ratio and the routine, which fit the same models (namely one or two Gaussians
PyNeb Python module (Luridiana et al. 2015), taking the electron and a low order polynomial) to the spectra — taking into account
densities for the apertures to be those derived using the transauroral the observational the flux uncertainty of the HST data — with priors
line technique for both objects (3.75 < log10 (𝑛𝑒 [cm −3 ]) < 4.75: see chosen to ensure the resulting models were physical (i.e. the line
Table 1 and Section 4.1). We present the measured electron tempera- fluxes, mean wavelengths, and line widths must have been positive).
tures in Table 1, which are found to be high (14,300 < Te < 21,000 K) For each fit, we initialised 500 walkers in a Gaussian distribution
for every aperture in both objects, with particularly high tempera- around the starting parameters, and used a total of 5000 iterations
tures (up to Te = 21, 000 K) being found in the central apertures of (including a 1000 iteration ‘burn-in’ phase). The MCMC fits them-

MNRAS 000, 1–20 (2015)

 NLR warm-ionised outflows in NGC 1068 and NGC 4151 9

2.5
100 -3.0
AGN 2
-2.5 -3.5 2.0 AM/I

[NeV] 3426 / [NeIII] 3869

80 -2.0 Precursor
[OIII](5007/4363)

1.5 1
-3.0 1 AM/I
60 2
-2.5 1.0
-2.0
40 -2.5
Shock 0.5 AGN Precursor
20 Shock
0
0.0 -3.5
0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5
HeII 4686 / H HeII 4686 / H

Figure 5. [OIII](5007/4363) vs HeII/H𝛽 diagnostic diagram (Villar- Figure 6. [NeV]3426/[NeIII]3869 vs HeII/H𝛽 diagnostic diagram — both
Martín et al. 1999), used to distinguish between radiation-bounded AGN- ratios are sensitive to the presence of significant matter-bounded components.
photoionisation, matter-bounded AGN-photoionisation and shock-ionisation. The line and marker scheme is the same as Figure 5. The line ratios measured
The black markers show the predicted line ratios from radiation-bounded in our NGC 1068 apertures are located in the matter-bounded photoionisation
Cloudy modelling (see Section 4.1) for solar-composition gas with a den- region of the diagram (corresponding to 1 < AM/I < 2; consistent with Figure
sity of 𝑛𝑒 = 104 cm −3 and varying ionisation parameters (log 𝑈; labelled) 6) whereas the NGC 4151 line ratios fall in the shock/precursor/radiation-
and spectral indices (squares: 𝛼=1.5; diamonds: 𝛼=1.0). The solid red grid bounded AGN photoionisation region.
shows the line ratios predicted from shock modelling (Allen et al. 2008) for
solar-composition gas with a pre-shock density of 𝑛 = 102 cm −3 and mag-
√
netic parameters of 𝐵/ 𝑛 = 2,4 𝜇G cm3/2 , with lighter regions on the plot
corresponding to lower shock velocities. The purple dashed lines show the (centered slightly above the nucleus: Figure 2), we measure HeII/H𝛽
predicted emission from the precursor gas, which has not yet passed through
ratios consistent with both matter-bounded AGN-photoionisation and
(but is photoionised by) the shock, and the dotted green line shows the line
ratios expected for different ratios of matter-bounded and radiation-bounded
shock-ionisation.
clouds (AM/I , labelled and marked with green circles) from modelling by Bi- To further probe the ionisation mechanism of the gas, we also
nette et al. (1996). Observed line ratios for each aperture are shown in orange measured the [NeV]𝜆3426/[NeIII]𝜆3869 ratio — which is sen-
for NGC 1068 and blue for NGC 4151, with the aperture number annotated. sitive to higher ionisation gas — using the same MCMC fit-
ting routine described earlier. We produced a diagnostic dia-
gram of [NeV]𝜆3426/[NeIII]𝜆3869 vs HeII/H𝛽 using the same
selves were run using the emcee Python module (Foreman-Mackey radiation-bounded photoionisation, matter-bounded photoionisation,
et al. 2013). and shock-ionisation models as used for the [OIII](5007/4363) and
From Figure 5, we find clear evidence for significant matter- HeII/H𝛽 diagram (Figure 5), and present this in Figure 6. We find
bounded emission in Apertures 1, 2, and 3 in NGC 1068, implied that the values for all of the NGC 1068 apertures are consistent with
by high electron temperatures and HeII/H𝛽 > 0.4 (similar ratio val- matter-bounded AGN-photoionisation with 1 < AM/I < 2. This fur-
ues were also measured by Kraemer & Crenshaw 2000b); the ap- ther indicates that the gas in these apertures is matter-bounded and
proximate ratio of matter-bounded to radiation-bounded clouds is AGN-photoionised, including Aperture 4.
AM/I ∼ 2. The difference between the [OIII](5007/4363) ratios mea- With the exception of Aperture 3, the [OIII](5007/4363) vs
sured in the NGC 1068 apertures and those predicted from the Bi- HeII/H𝛽 ratios measured in our NGC 4151 apertures (Figure 5), are
nette et al. (1996) modelling can be explained as due to the mod- consistent with both shock ionisation and radiation-bounded AGN
els only representing one combination of parameters: it is possible photoionisation (assuming a relatively flat spectral index of 𝛼 = 1.0
for matter-bounded clouds with different parameters to have similar and log 𝑈 ∼ −2.0). However, from the [NeV]𝜆3426/[NeIII]𝜆3869 vs
[OIII](5007/4363 ratios to those found for NGC 1068. Specifically, HeII/H𝛽 diagram (Figure 6), it can be seen that the measured ratios
this ratio would be smaller for higher electron densities than the for NGC 4151 are not consistent with pure shock-ionisation alone: if
low density assumed by Binette et al. (1996). Moreover, the pres- the gas is shock ionised, then a contribution from the precursor com-
ence of matter-bounded emission in these apertures is supported by ponent is required. Alternatively, the gas in these apertures may have
the strength of high-ionisation emission lines (Eion > 100 eV), such pure radiation-bounded AGN photoionisation, however we highlight
as [NeV]𝜆3426, [FeVII]𝜆3759, and [FeVII]𝜆6087, relative to lower- that this requires a relatively flat spectral index (𝛼 = 1.0), and/or
ionisation lines (such as [OIII]) in our STIS spectra. These and other higher ionisation parameters (−3.0 < log 𝑈 < −2.0) and densities
high-ionisation lines were previously identified in the same dataset (𝑛𝑒 >105 cm −3 ) than can explain our transauroral line ratios (Sec-
by Kraemer & Crenshaw (2000a). For Aperture 4 of NGC 1068 tion 4.1). Ultimately, it is not possible to determine unambiguously

MNRAS 000, 1–20 (2015)

10 L.R. Holden et al.
the true, dominant ionisation mechanism of the gas in our NGC 4151
apertures with the diagnostic features that are available in our data. 10 log10ne
6.5 6.0 5.5 5.0
9 7.5 7.0
8.0

[NeV] 3426 / [FeVII] 6086

4.2.3 The viability of shock-ionisation
In order to further investigate the viability of shocks as the dominant 8
ionisation mechanism along our slits for NGC 1068 and NGC 4151, log U
7
we compared our measured H𝛽 fluxes to those expected from shock
models — a technique presented by Baron et al. (2017). First, we
converted our measured (and dereddened) H𝛽 fluxes (𝐹H𝛽 ) into H𝛽
luminosities using the luminosity distances (𝐷 𝐿 ) for each galaxy. 6
The resulting luminosities were then converted into luminosities per
surface area using the aperture sizes in arcseconds (i.e. the aperture
width multiplied by the slit width) and the spatial scales for each 5
object (0.067 kpc/arcsecond and 0.078 kpc/arcsecond, respectively).
We then compared the measured luminosities per surface 4
area to those expected from the MAPPINGS III shock mod-
els of pre-shock density 𝑛 = 102 cm −3 (corresponding the den-
sities measured in our apertures, assuming a compression fac-
3
tor of 100: Sutherland & Dopita 2017) and magnetic parameters 1.0 1.5 2.0 2.5 3.0 3.5
√
𝐵/ 𝑛 = 2, 4 𝜇G cm3/2 . From this comparison, we find that the [FeVII](6086/3759)
H𝛽 luminosities per surface area, as measured in each aperture
for NGC 1068 (4.9 × 10 −3 < LH𝛽 < 2.2 × 10 −2 erg s −1 cm −2 ) and
NGC 4151 (2.2 × 10 −3 < LH𝛽 < 4.6 × 10 −3 erg s −1 cm −2 ), can be Figure 7. [FeVII](6087/3759) vs [NeV]𝜆3426/[FeVII]𝜆6086 diagnostic dia-
accounted for by shocks with velocities 𝑣 shock > 425 km s −1 and gram, sensitive to the spectral index, ionisation parameter, and electron den-
𝑣 shock > 225 km s −1 respectively. In both cases, the outflow veloc-
sity of the gas. The grey grid was generated using CLOUDY for radiation-
bounded AGN-photoionisation with a spectral index of 𝛼 = 1.5 (see Ap-
ities for our apertures (Section 4.4.1; Table 2) are above these re- pendix B) and varying electron densities and ionisation parameters: dashed
quired velocities. This demonstrates that shock-ionisation could fea- lines connect points of constant density (labelled), and dotted lines connect
sibly produce the recombination line fluxes measured in both objects, points of constant ionisation parameter (also labelled). The measured ratio
however this alone does not necessarily confirm the ionisation mech- values in our NGC 1068 apertures are shown in orange and labelled. From
anism. this diagram, the high ionisation gas is found to have densities in the range
Note that here we assumed a gas covering factor of unity relative 6.45 < log10 (𝑛𝑒 [cm −3 ]) < 8.00.
to the shock (i.e. that the emitting-gas covers the entire area of the
shock within each aperture), which may not be the case in reality.
If this covering factor is in fact much lower than unity, then a larger than the gas traced by the lower critical-density [OII] and [SII] lines.
shock area or higher shock velocities would be needed to produce We discuss the implications of this for the gas structures within our
the same H𝛽 luminosity. apertures in Section 5.2.1.

4.3 The high-ionisation gas in NGC 1068 4.4 Energetics of the outflowing gas

The relative strengths of the high ionisation (Eion > 100 eV) lines de- 4.4.1 Outflow kinematics
tected in several of our apertures for NGC 1068 indicate the presence In order to determine the mass outflow rates, kinetic powers and cou-
of matter-bounded clouds, and therefore may play an important role pling efficiencies of the gas outflows detected in our STIS spectra, we
in the structure of the cloud complexes present in the NLR. Deter- required measurements of the kinematics of the outflowing gas3 . For
mining the physical conditions of this high-ionisation component is this purpose, we used the results from detailed kinematic modelling
therefore necessary. To this end, we measured the [FeVII](6087/3759) (based the same HST/STIS spectra used here) of NGC 1068 and
and [NeV]𝜆3426/[FeVII]𝜆6086 emission-line ratios, which are sen- NGC 4151 presented by Crenshaw & Kraemer (2000c) and Cren-
sitive to the density and ionisation parameter of the high-ionisation shaw et al. (2000) (hereafter CKN1068 and CKN4151), respectively.
gas. These ratios were calculated using the measured line fluxes of We note that, due to the different PAs used and the fact that the
the lines in the ratios, which were themselves determined using the outflow geometry likely depends greatly on PA, we do not use the
same MCMC fitting method described in Section 4.2. updated kinematic models from Das et al. (2005) and Das et al.
We present the [FeVII](6087/3759) vs [NeV]𝜆3426/[FeVII]𝜆6086 (2006).
diagnostic diagram (see Rose et al. 2011) with our mea- To calculate deprojected velocities, we first derived a universal
sured line ratios for the NGC 1068 apertures in Figure 7; ‘deprojection factor’ by dividing the maximum observed velocities
a CLOUDY radiation-bounded photoionisation grid for a so- (located at the velocity ‘turnover’ position - see Crenshaw & Kraemer
lar metallicity, plane-parallel, single-slab cloud of varying ion- 2000c and Crenshaw et al. 2000) by the maximum model-deprojected
isation parameters (-3.5 < log 𝑈 < 2.0) and electron densities velocities from the CKN1068 and CKN4151 bicone models. We then
(5.0 < log(𝑛𝑒 [cm −3 ]) < 8.0), and a central ionising source with spec-
tral index 𝛼 = 1.5 (see Appendix B), is shown. From this grid, we
determine the densities of the high-ionisation gas to be in the range 3 We do not use kinematics derived from our [OIII] models due to the
6.45 < log10 (𝑛𝑒 [cm −3 ]) < 8.00: several orders of magnitude higher relatively low spectral resolution and high instrumental widths of our spectra.

MNRAS 000, 1–20 (2015)

 NLR warm-ionised outflows in NGC 1068 and NGC 4151 11
took the the highest observed (projected) velocity at the position of For NGC 1068, the mass outflow rates for the warm-ionised phase
each aperture, and divided these velocities by our determined depro- are much below that of the cold molecular gas at a similar extent from
jection factor to give the maximum deprojected outflow velocity in the nucleus (i.e. traced by CO, HCN; 𝑇 ∼100 K): García-Burillo
each aperture. We label the deprojected outflow velocities as 𝑣 out , et al. (2014) derive a mass outflow rate of 𝑀¤ out = 63+21
−37
M ⊙ yr −1
and give their values in Table 2. within the 𝑟 ∼200 pc circumnuclear disk (CND) of NGC 1068. This
indicates that most of the outflowing mass may be present in the
colder gas phases, as has been found for other objects (see Ramos
4.4.2 Mass outflow rates, kinetic powers and coupling efficiencies Almeida et al. 2019 and Holden et al. 2023).
We used the H𝛽 luminosities to determine masses for the warm
ionised gas in each aperture with
5 DISCUSSION
𝐿 (𝐻 𝛽)𝑚 p
𝑀ion = , (1)
eff ℎ𝑣
𝛼H𝛽 H𝛽 𝑛e
From our analysis of archival STIS spectra of the central regions
(𝑟 < 160 pc) of NGC 1068 and NGC 4151, we find evidence for
where 𝑀ion is the total mass of the warm ionised gas, 𝑚 p is the proton dense (103.6 cm −3 < 𝑛𝑒 < 104.8 cm −3 ) gas that shows line ratios
𝑒𝑓 𝑓 consistent with matter-bounded AGN-photoionisation in the case of
mass, 𝛼 𝐻 𝛽 is the Case B recombination coefficient for H𝛽 (taken
NGC 1068, and shock-ionisation (with precursor gas ionisation) or
to be 1.61×10 −14 cm3 s −1 for a gas of density 𝑛e = 104 cm −3 and
radiation-bounded AGN-photoionisation in the case of NGC 4151.
temperature 𝑇e = 20, 000 K; Osterbrock & Ferland 2006) and 𝑣 H𝛽 is
Furthermore, we find that the measured H𝛽 luminosities could be ex-
the frequency of the H𝛽 line.
plained as being due to shock-ionisation for both objects, assuming
Assuming that the derived masses (estimated using the total line
a shock covering factor of unity. In both objects, we find coupling
fluxes) are dominated by outflowing gas, we combined them with the
efficiencies that are close to the lowest value required by models of
aperture crossing time to calculate mass outflow rates
galaxy evolution, however these are likely underestimates. In this
𝑀 𝑣 out section, we discuss the implication of these results on the dominant
𝑀¤ out = ion , (2)
Δ𝑅 ionisation and acceleration mechanisms of the gas seen in our slits,
where 𝑣 out is the outflow velocity from the CKN1068 and CKN4151 compare our results to past work on these two well-studied objects,
models, and Δ𝑅 is the aperture width. and investigate the impact on the density diagnostic techniques used.
Kinetic powers were estimated from the mass outflow rates using Finally, we place our results in a broader context by comparing with
those from a similar study of the nearby Seyfert 2 galaxy IC 5063.
1
𝐸¤ kin = 𝑀out 𝑣 2out . (3)
2
Finally, the ratio of the kinetic power to the bolometric AGN 5.1 The outflow ionisation and acceleration mechanisms in the
luminosity (𝐿 bol ) was taken to estimate coupling efficiencies for NLRs of NGC 1068 and NGC 4151
each aperture: To determine the true impact of the outflowing gas on the host galax-
𝐸¤ ies, quantitative comparison of observations to theoretical modelling
𝜖 f = kin . (4) is needed. However, both modelling of jet-ISM interactions (e.g.
𝐿 bol
Mukherjee et al. 2018; Audibert et al. 2023) and AGN radiation-
NGC 1068 is estimated to have a bolometric luminosity in the range pressure-driven outflows (e.g. Crenshaw & Kraemer 2000c; Meena
0.4 < 𝐿 bol < 4.7×1038 W (Woo & Urry 2002; Alonso-Herrero et al. et al. 2023) is able to explain outflow kinematics in different objects.
2011; Lopez-Rodriguez et al. 2018; Gravity Collaboration et al. In order to enable accurate future comparisons to theoretical mod-
2020), of which we take the lowest value to ensure higher estimates els and therefore accurately quantify the impact of the outflows in
of coupling efficiencies and thus determine the maximum potential NGC 1068 and NGC 4151 — which have been conversely argued to
impact of the outflowing gas on the host galaxy. For NGC 4151, we be radiatively-accelerated (Crenshaw et al. 2000; Crenshaw & Krae-
took the bolometric luminosity to be 𝐿 bol = 1.4 × 1037 W (Kraemer mer 2000c; Das et al. 2005, 2006; Revalski et al. 2021; Meena et al.
et al. 2020). 2023) and jet-accelerated (Capetti et al. 1997; Axon et al. 1998; May
We present our derived mass outflow rates, kinetic powers and & Steiner 2017; May et al. 2020) — the dominant outflow accelera-
coupling efficiencies for both cases in Table 2. For NGC 1068, tion mechanisms in these objects need to be robustly identified.
our estimates are less than the maximum values determined from
photoionisation modelling by Revalski et al. (2021) ( 𝑀¤ out =
9.0 ± 1.13 M ⊙ yr −1 ; 𝐸¤ kin = (5.4 ± 0.5) × 1035 W, 𝜖 f = 0.54 ± 5.1.1 Matter-bounded ionisation and the acceleration mechanism
0.05 per cent)4 . For NGC 4151, our derived values are similar to in NGC 1068
the results of photoionisation modelling by Crenshaw et al. (2015)
It has been previously proposed that the outflows in NGC 1068
( 𝑀¤ out ∼ 3.01 ± 0.45 M ⊙ yr −1 ; 𝐸¤ kin = (4.3 ± 1.0) × 1034 W,
are driven via radiation pressure (Kraemer & Crenshaw 2000a;
𝜖 f = 0.54 ± 0.11 per cent). Our calculated mass outflow rates
Das et al. 2006; Revalski et al. 2021; Meena et al. 2023), in-
for NGC 4151 are also consistent with previous values derived
stead of via shocks induced by the radio jet colliding with the
for the warm ionised phase by Storchi-Bergmann et al. (2010)
ISM within the bicone. While we do not separate the outflowing
(𝑀out ≈ 2.4 M ⊙ ) and the X-ray emitting gas (𝑀out ≈ 2 M ⊙ yr −1 :
gas from the quiescent gas in this work, our results are consistent
Wang et al. 2011b and Kraemer et al. 2020).
with this mechanism: we find evidence for matter-bounded AGN-
photoionisation of the warm-ionised gas in the form of simultaneous
4 Crenshaw et al. (2015) and Revalski et al. (2021) assume bolometric lumi- high [OIII] temperatures (Table 1: 𝑇𝑒 ∼15,000 K; corresponding to
nosities of Lbol = 1 × 1038 W for NGC 1068 and Lbol = 7.9 × 1036 W for [OIII](5007/4363) < 60) and line ratios (HeII𝜆4686 / H𝛽 > 0.4: Fig-
NGC 4151 when calculating coupling efficiencies. ure 5; [NeV]𝜆3426 / [NeIII]𝜆3869 > 1.0: Figure 6) within a 134 pc

MNRAS 000, 1–20 (2015)

12 L.R. Holden et al.

Aperture 𝑣out (km s −1 ) Mout (M ⊙ ) ¤ out (M ⊙ yr −1 )

𝑀 𝐸¤ kin (W) 𝜖f (per cent)

NGC 1068

1 -1300 (1.7 ± 0.3) × 105 3.7 ± 0.6 (2.0 ± 0.3) × 1035 (4.9 ± 0.7) × 10 −1
2 -1100 (6.3 ± 0.8) × 104 1.6 ± 0.2 (6.1 ± 0.7) × 1034 (1.5 ± 0.2) × 10 −1
3 -450 (7.3 ± 0.9) × 104 1.2 ± 0.1 (7.6 ± 0.9) × 1033 (1.9 ± 0.2) × 10 −2
4 -150 (9.4 ± 2.2) × 104 (6.0 ± 1.4) × 10 −1 (4.2 ± 1.0) × 1032 (1.1 ± 0.2) × 10 −3

NGC 4151

1 -700 (2.5 ± 0.6) ×105 3.7 ± 1.0 (5.8 ± 0.2) × 1034 (4.1 ± 1.1) × 10 −1
3 -800 (7.2 ± 3.0) × 104 3.4 ± 1.4 (6.9 ± 2.9) × 1034 (4.9 ± 2.0) × 10 −1
5 800 (1.7 ± 0.4) × 105 4.5 ± 1.2 (9.2 ± 2.4) × 1034 (6.5 ± 1.7) × 10 −1
6 800 (2.9 ± 0.7) × 105 6.9 ± 1.8 (1.4 ± 0.4) × 1035 (9.9 ± 2.6) × 10 −1

Table 2. Outflow velocities, outflow masses, mass outflow rates, kinetic powers and coupling efficiencies for the apertures of our STIS spectra of NGC 1068
and NGC 4151. The outflow velocities used to calculate the mass outflow rates and kinetic powers presented here are from the CKN1068 and CKN4151 models
(see Section 4.4.1).

radius from the nucleus in the NE cone along the radio axis, con- axis. Moreover, the NLR molecular CO(3–2) outflows (as seen in
sistent with radiative acceleration. However, it is possible that the ALMA imaging by García-Burillo et al. 2014) decelerate within the
outflowing gas has been shock-ionised and accelerated by the jet, but radio lobe, at a distance of ∼ 400 pc (∼5.7 arcseconds) from the nu-
has subsequently cooled and then been reionised by the AGN (e.g. cleus. Taken together, this shows that the NE cone outflows have a
as in Holden et al. 2023). Spatially-resolved, high spectral-resolution similar extent to the NE radio lobe. This is evidence for the outflows
observations are needed to further investigate this situation by sepa- being accelerated by the radio jet, although it does not entirely rule
rating the emission from the outflowing and quiescent gas, and then out radiative acceleration.
determining the ionisation and excitation mechanisms of each kine-
matic component. In addition, comparing the electron densities of
the outflowing and non-outflowing gas may reveal signs of shock
compression, which is expected to be a factor of ∼4–100 (Sutherland
& Dopita 2017).
We note that the outflowing gas appears to be spatially confined to
the extent of the radio structure: the broad (FWHMv > 250 km s −1 ) 5.1.2 Shock-ionisation and acceleration in NGC 4151
[OIII]𝜆𝜆4959,5007 emission in our spectra is seen to a maximum
Our results for NGC 4151 indicate that the near-nuclear gas
radius of ∼ 4.8 arcseconds from the nucleus in the NE cone (as mea-
along the radio axis may be shock-ionised, since the measured
sured from the line profiles of the [OIII] emission that extends beyond
[OIII](5007/4363), HeII/H𝛽, and [NeV]𝜆3426/[NeIII]𝜆3869 line ra-
the regions covered by our apertures), similar to the maximum ra-
tios and H𝛽 luminosities are consistent with those expected from a
dial extent of the NE radio lobe (6.18 arcseconds; 420 pc) measured
mixture of shock and shock-precursor ionisation (Figures 5 and 6;
from radio imaging (e.g. 15 GHz: Wilson & Ulvestad 1987 5 GHz:
Section 4.2.3).
Wilson & Ulvestad 1983, Gallimore et al. 1996; 22 GHz: Gallimore
The radio structure in the NLR of NGC 4151, as seen in low-
et al. 1996; 1.4 GHz: Gallimore et al. 1996, García-Burillo et al.
resolution 1.5–5 GHz VLA radio imaging by Johnston et al. (1982),
2014). This is also in agreement with ground-based Fabry-Pérot in-
has a lobe-like component with a centroid 6.43 arcseconds from
tegral field spectroscopy by Cecil et al. (1990) — which finds no
the nucleus along the radio axis in the NE cone. This structure lies
significant velocity deviation from the systematic velocity beyond
beyond the maximum ∼4 arcseconds extent of the warm-ionised out-
the radio lobe — and kinematic modelling by Crenshaw & Krae-
flows (Meena et al. 2023; see also Das et al. 2005), and — as we have
mer (2000c), Das et al. (2006) and Meena et al. (2023), which find
argued for the situation in NGC 1068 — is consistent with the out-
outflows extended up to ∼5.1 arcseconds from the nucleus5 . Further-
flows being launched by the radio jet. From HST/PC + HST/WFPC2
more, VLT/MUSE spectroscopy presented by Venturi et al. (2021)
imaging, Williams et al. (2017) found higher [OIII]/H𝛼 ratios close
shows that the measured [OIII] W70 velocity parameter6 has high
to the string of radio knots that are seen in their higher-resolution
values (400 km s −1 < [OIII] W70 < 1200 km s −1 ) between the nu-
1.51 GHz observations (shown here in Figure 1), with the values of
cleus and the lobe, out to a radius of 3.6 arcseconds along the bicone
this ratio decreasing beyond ∼4 arcseconds from the nucleus along
the radio axis. The authors interpreted this as the radio jet having
a contribution to the ionisation of the gas close to the nucleus, but
5 An [OIII] emission knot in the NLR of NGC 1068, labelled ‘A’ by Meena AGN-photoionisation being dominant further out. This is also in
et al. (2023) and located 7.3 arcseconds from the nucleus (i.e. beyond the radio agreement with the results from X-ray and optical imaging by Wang
source), has outflow-like kinematics (200 < FWHM < 1000 kms −3 ; vout = et al. (2011a), who propose a mixture of shock-ionisation and AGN-
863 kms −3 ). As noted by Meena et al. (2023), this knot lies beyond the
photoionisation in the NLR of NGC 4151.
expected extent of radiatively-driven outflows. Regardless, we highlight that
the vast majority of the outflows along the radio axis are located at lower radii Taken together with the findings of these previous investigations,
than the maximum extent of the NE radio lobe. the results presented here may indicate that the outflows in NGC 4151
6 W70 is defined as the difference between the velocities that contain have been shock-accelerated and then re-ionised by photons from the
85 per cent and 15 per cent of the total flux of the fits to the line profile AGN, with AGN-photoionisation being dominant further from the
(see Venturi et al. 2021). nucleus.

MNRAS 000, 1–20 (2015)

 NLR warm-ionised outflows in NGC 1068 and NGC 4151 13
5.2 The effect of ionisation mechanisms on density diagnostics orders of magnitude, and the effect on derived reddenings is E(B-
V)±0.13. Crucially, we note that this is much less than the impact of
The ionisation mechanisms (Section 4.2), electron temperatures
using lower-critical-density techniques (such as the [SII](6717/6731)
(Section 4.2.1), and densities (Sections 4.1 and 4.3) of the warm
ratio), and is similar to the effect of varying the parameters of the
gas detected in our STIS slits allows us to investigate the structures
photoionisation model (Santoro et al. 2020: log(𝑛𝑒 [cm −3 ])±(0.1–
and conditions of the line-emitting clouds, and therefore verify the
0.7); E(B-V)±(0.1–0.2)). In summary, while using the transauroral
origin of different emission lines and thus the precision of diagnos-
line method presented by Holt et al. (2011) as a density and reddening
tics which make use of them. For example, the TR density diagnostic
diagnostic for shock-ionised gas does incur some uncertainty on the
(Section 4.1) relies on AGN-photoionisation being dominant, with
derived densities, the derived densities are still likely more accurate
no significant contribution from a matter-bounded component or
than those derived from commonly used, traditional methods.
shock-ionisation. Since we find evidence for matter-bounded emis-
Gas that has been shock-ionised by jet-ISM interactions presents a
sion in NGC 1068 and potential shock-ionisation in NGC 4151, it is
problem for the photoionisation modelling method used by Revalski
important to investigate the effect of this on derived densities.
et al. (2021), as the technique relies on assuming that the material at a
given distance from the nucleus is being photoionised by the central
5.2.1 The impact of matter-bounded photoionisation AGN engine. In the case of shock-ionisation, the outflows are instead
being shock-ionised locally by the jet within the bicone at any given
If the higher ionisation lines are indeed emitted by matter- distance from the nucleus, and so any electron densities derived us-
bounded gas structures in the outflow (as shown by the ing an assumed ionisation parameter and distance will be incorrect.
[OIII] temperatures, HeII/H𝛽 ratios, and [FeVII](6086/3759) vs Revalski et al. (2021) used the standard BPT diagrams (Baldwin et al.
[NeV]𝜆3426/[FeVII]𝜆6086 diagram: Sections 4.2 and 4.3), then the 1981) in an attempt to ensure all of the measured line ratios were
transauroral lines cannot be emitted by the same structures. How- consistent with AGN-photoionisation. However, the regions of AGN
ever, it is possible that they are emitted by different clouds within shock and photoionisation in these diagrams overlap considerably,
the same cloud complexes, considering that we see these lines with thus further diagnostics should also be used in order to disentan-
similar profiles in each of our apertures. Alternatively, or perhaps gle the contribution from shocks and photoionisation, such as the
in addition, it is possible that the outer layers of a single cloud are [OIII](5007/4363) vs HeII𝜆4686/H𝛽 (Villar-Martín et al. 1999) and
matter-bounded, while the denser core is radiation-bounded (one of [FeII]𝜆12570/Pa𝛽 vs H2 𝜆21218/Br𝛾 (Rodríguez-Ardila et al. 2005;
the scenarios presented by Binette et al. 1996). In this scenario, Riffel et al. 2013; Colina et al. 2015; Riffel et al. 2021; Holden et al.
the matter-bounded layers may represent lower density gas that was 2023) diagnostic diagrams, and/or the three-dimensional diagram
driven away from the ionisation front by the increase in pressure (which makes use of line ratios and velocity dispersion) presented
that occurred when the gas structure was first photoionised by the by D’Agostino et al. (2019).
AGN. However, this is not consistent with our findings: in Section Overall, despite the challenges that shock-ionisation and signif-
4.3, we use the [FeVII](6087/3759) and [NeV]𝜆3426/[FeVII]𝜆6086 icant matter-bounded photoionisation components present to the
emission-line ratios to determine high-ionisation gas densities of transauroral line technique and the Revalski et al. (2021) photoioni-
6.45 < log10 (𝑛𝑒 [cm −3 ]) < 8.00 in our NGC 1068 apertures: signif- sation modelling, we argue that these methods are nonetheless more
icantly higher than that of the lower-ionisation gas. A potential ex- robust density diagnostics than the commonly used [SII](6717/6731)
planation is that the gas that is emitting the high-ionisation [FeVII] and [OII](3726/3729) ratios. In the case of matter-bounded photoion-
and [NeV] lines represents dense fragments of the expanding matter- isation, the [SII]𝜆𝜆6717,6731 and [OII]𝜆𝜆3726,3729 lines arise from
bounded component: since these lines have high critical densities the same part of the ionisation structure of the cloud as the transauro-
(7.1 < log10 (𝑛crit [cm −3 ]) < 8.5; Appendix A), they would only be ral lines, meaning they face the same issues as the TR method, while
emitted strongly by such dense cloud components. the Revalski et al. (2021) modelling allows for higher-ionisation
Therefore, given the ionisation energies of the lines (Appendix A), components, and therefore is a more accurate diagnostic of the over-
we propose that the [FeVII] and [NeV] lines trace matter-bounded, all cloud density. Furthermore, we have established here that using
higher ionisation clouds within the complexes (or edges of individual radiation-bounded photoionisation grids to measure the TR densities
clouds), and the [OII] and [SII] lines are emitted from radiation- of shock-ionised gas incurs an uncertainty on the overall density that
bounded clouds (or cores of individual clouds). In this scenario, is much less than using lower-critical density line ratios for high-
much of the [OIII] emission must arise from the matter-bounded density (𝑛𝑒 > 103 ) gas: in the case of NGC 4151 (where there may
regime in order to explain the high electron temperatures that we be some contribution from shock-ionisation), the TR-derived den-
measure in our NGC 1068 apertures (Section 4.2.1). Hence, given sities are similar to those reported by Revalski et al. (2022) (see
the high density of the high-ionisation gas, it is likely that the gas also Crenshaw et al. 2015), indicating that both methods still give
emitting the [OIII] lines is denser than the gas that is emitting the more precise density determinations than traditional methods, despite
transauroral lines. This reinforces the need for outflow diagnostics some of their underlying assumptions potentially being incorrect.
that are sensitive to high (>103.5 cm −3 ) densities.

5.3 Comparison of the TR electron densities to other techniques

5.2.2 The impact of shock-ionisation
Using the TR method, we find high electron densities in
Since the gas in our NGC 4151 apertures may be shock-ionised, it both objects: 4.00 < log(𝑛𝑒 [cm −3 ]) < 4.75 in NGC 1068 and
is essential to quantify the effect of this on the transauroral ratio 3.50 < log(𝑛𝑒 [cm −3 ]) < 4.10 in NGC 4151 (Section 4.1; Table 1).
density diagnostic. In Appendix C, we plot the TR ratios from shock This agrees with the similarly high densities (> 103 cm −3 ) derived
models over the TR photoionisation diagnostic grid used in Section from multi-component photoionisation modelling of both objects
4.1, and quantify the impact of shock-ionisation on the TR electron presented in Crenshaw et al. (2015) and Revalski et al. (2021) (see
density and reddening values derived from the photoionisation grid. also Collins et al. (2009) and Revalski et al. 2022). Crucially, the
We find that, overall, the effect on the derived density is ±0.38 derived densities from both techniques lie above the sensitivity range

MNRAS 000, 1–20 (2015)

14 L.R. Holden et al.
of the traditional [SII](6717/6731) and [OII](3726/2739) techniques,
which are commonly used (either directly or as a basis for assump-
tion) to derive electron densities in studies of the warm-ionised phase 140
(e.g. Nesvadba et al. 2006; Liu et al. 2013; Harrison et al. 2014; Fiore
-3.5
et al. 2017), thus further supporting the need for robust warm-ionised
gas electron density diagnostics such as the transauroral line tech-
120
-3.0
nique and multi-component photoionisation modelling.
Considering the traditional [SII](6717/6731) ratio, Kraemer & 100 AGN

[OIII](5007/4363)
Crenshaw (2000a) (using the same STIS dataset as used in this work)
-2.5 -3.5
and Kakkad et al. (2018) and Mingozzi et al. (2019) (using IFU
80 -2.0 Precursor
data) derived electron densities of 𝑛𝑒 ∼ 103 cm −3 for the outflows
in the NLR of NGC 1068. These [SII]-derived densities are 1–1.5 -3.0
orders of magnitude lower than those we find using the TR method, 60
-2.5
and are close to the upper limit of the density range for the [SII]
ratio technique (Appendix A: 𝑛crit ∼ 103.5 cm −3 ). This provides
-2.0
40
further evidence that, for gas of electron density 𝑛𝑒 > 103.5 cm −3 , the Shock
[SII](6717/6731) ratio may underestimate the true electron density
by more than an order of magnitude. 20

0
5.4 The impact of the outflowing gas on the host galaxies 0.0 0.1 0.2 0.3 0.4 0.5 0.6
HeII 4686 / H
Using densities derived from the transauroral line ratios, reddening-
corrected recombination line fluxes and kinematics taken from
previous modelling, we find mass outflow rates in the range Figure 8. [OIII](5007/4363) vs HeII4686/H𝛽 diagnostic diagram (as in Fig-
0.6 < 𝑀¤ out < 6.9 M ⊙ yr −1 , and coupling efficiencies in the range ure 5) with line ratios measured from the STIS spectra of NGC 1068 and
1.1×10 −3 <𝜖kin < 0.99 per cent (Table 2). In many cases, our cal- NGC 4151 (presented in this work) and from Xshooter spectra of IC 5063
culated coupling efficiencies are just above the lower limit required (presented in Holden et al. 2023). The AGN, shock and precursor models
by models of the co-evolution of galaxies and their supermassive are the same as those described in Section 4.2. The three objects each show
black holes (e.g. ∼0.5–10 per cent: Di Matteo et al. 2005; Springel distinct ionisation conditions, hinting at the complex nature of the NLRs of
Seyfert galaxies.
et al. 2005; Hopkins & Elvis 2010). It is important to note that there
is likely more outflowing material within the bicones that is not
covered by our slits (which are only 0.1 arcseconds wide), and that
2021; Holden et al. 2023); neutral (Morganti et al. 1998; Oosterloo
comparisons between coupling efficiencies from models and obser-
et al. 2000); warm molecular (Tadhunter et al. 2014; Holden et al.
vations are not straightforward (see Harrison et al. 2018 for further
2023) and cold molecular (Morganti et al. 2013, 2015; Dasyra et al.
discussion). To properly account for the impact of the warm ionised
2016; Oosterloo et al. 2017).
outflows, detailed studies that make use of robust density diagnostics,
In Holden et al. (2023), we presented evidence that both the out-
separate emission from the outflowing and quiescent gas and, impor-
flowing and quiescent warm ionised gas in IC 5063 has dominant
tantly, cover the entire NLRs of both objects, are needed. Moreover,
AGN-photoionisation — even though the outflows show clear signa-
we highlight that assessments of all gas phases — not just the warm
tures of shock acceleration — and that the different outflow phases
ionised phase — are needed to robustly assess the total impact of
may represent a post-shock cooling sequence. We interpreted this sit-
the AGN-driven outflows (Cicone et al. 2018), as the warm ionised
uation as the pre-shock gas being AGN-photoionised, and the closest
gas may represent just a fraction of the total outflowing gas mass
post-shock gas to the AGN kept in an ionised state by photoionisa-
at a given radius (e.g. Ramos Almeida et al. 2019; Holden et al.
tion. In Figure 8, we add the [OIII](5007/4363) and HeII4686/H𝛽
2023). Therefore, it is likely the the true coupling efficiencies of the
ratios for IC 5063 from Holden et al. (2023) to the diagnostic di-
total NLR outflows in NGC 1068 and NGC 4151 are higher than we
agram presented in this work (Figure 5). Furthermore, we present
calculate here.
[OII](7319+7331)/[OIII]𝜆5007 and [SII](4068+4076)/H𝛽 ratios for
IC 5063 (alongside NGC 1068 and NGC 4151) in Appendix D (Fig-
ure D1) — determined using the dataset from Holden et al. (2023)
5.5 A tale of three Seyferts: NGC 1068, NGC 4151 and IC 5063
— and find that they are consistent with radiation-bounded AGN-
Finally, using the results for the nearby Seyfert 2 IC 5063 presented photoionisation with gas densities 103 cm −3 < 𝑛𝑒 < 104 cm −3 and
in Holden et al. (2023) along with the results for NGC 1068 and ionisation parameters in the range −3 < log 𝑈 < −2, in agreement
NGC 4151 that we present here, we can begin to construct a sample with the values determined in Holden et al. (2023).
of nearby Seyferts with spatially-resolved, detailed studies of their It is interesting that the overall differences in ionisation condi-
NLR outflows. tions between the three galaxies are significantly larger than the
IC 5063 is a nearby (𝑧 = 0.01131) early-type Seyfert 2 galaxy range of ionisation conditions within the galaxies. Our small sam-
that is seen close to edge-on, with a radio jet propagating almost in ple thus shows three distinct cases in the three objects: radiation-
the plane of the disk which drives fast (𝑣 out > 700 km s −1 ) outflows bounded AGN-photoionisation in IC 5063, matter-bounded AGN-
(Morganti et al. 1998; Oosterloo et al. 2000; Morganti et al. 2015; photoionisation in NGC 1068, and shock-ionisation or radiation-
Mukherjee et al. 2018; Holden et al. 2023). These outflows are seen bounded AGN photoionisation with a relatively flat spectral index
in multiple gas phases, including warm ionised (Morganti et al. 2007; and higher ionisation parameters in NGC 4151 — despite all being
Sharp & Bland-Hawthorn 2010; Congiu et al. 2017; Venturi et al. classified as Seyferts, the details of the ionisation mechanisms in the

MNRAS 000, 1–20 (2015)

 NLR warm-ionised outflows in NGC 1068 and NGC 4151 15

Object Lbol (W) L1.4 GHz (W Hz −1 ) Pjet (W) 𝜃jet Ionisation mechanism𝑎

NGC 1068 0.4–4.7×1038 2.3×1023 1.8×1036 (García-Burillo et al. 2014) ∼ 45◦ Matter-bounded photoionisation
NGC 4151 1.4×1037 1.6×1022 ∼ 1035 (Wang et al. 2011b) ∼ 36◦ Photo- and/or shock-ionisation
IC 5063 7.6×1037 3×1023 1037−38 (Mukherjee et al. 2018) ∼ 5◦ Radiation-bounded photoionisation
𝑎 Determined with line ratios detected in slits along PA=202◦ (NGC 1068), PA=70◦ (NGC 4151) and PA=115◦ (IC 5063).
Table 3. Bolometric luminosities, 1.4 GHz radio luminosities, jet powers (Pjet ), jet orientations with respect to the disk (𝜃jet ) and ionisation mechanisms detected
along the radio axis for NGC 1068, NGC 4151 and IC 5063.

objects vary greatly. This is particularly interesting considering that jets on the star-forming gas in their NLRs, as well as the impact of
in all three objects, the outflows detected along the radio axes appear inclined kpc-scale jets in general.
to be spatially-confined to the radio structures (i.e. the outflows do Ultimately, further observations of NGC 1068 and NGC 4151 are
not extend beyond the radio lobes in the NLRs). As argued in Section required to decisively determine the outflow acceleration mecha-
5.1, this is consistent with shock-acceleration, although it does not nism(s). Namely, wide wavelength coverage spectroscopy (to make
rule out radiative-acceleration. If the outflows in IC 5063, NGC 1068 available a range of diagnostics) with sufficient velocity resolution
and NGC 4151 are shock-accelerated, then this would highlight the to kinematically discriminate between outflowing (post-shock?) and
importance of not deriving information regarding the outflow accel- quiescent (pre-shock?) gas.
eration mechanisms based solely on the ionisation/excitation mech-
anisms or kinematics of the gas in NLRs: a full account, involving
detailed multi-wavelength observations with multiple diagnostics, is
required to properly evaluate the relative contributions of different 6 CONCLUSIONS
mechanisms.
By analysing archival HST/STIS spectra taken along the radio axes of
Despite evidence that the outflows in all three objects are being the inner few hundred parsecs of the NLR of the prototypical Seyfert
driven by the radio jet, the densities of the outflowing gas differ by galaxies NGC 1068 and NGC 4151, we have found the following.
more than an order of magnitude: for IC 5063, we found that the out-
flowing gas has densities in the range 3.17 < log(𝑛𝑒 [cm −3 ]) < 3.43, • Using the transauroral line ratio technique, we derive spatially-
while for NGC 1068 and NGC 4151 we find densities in the ranges resolved electron densities of 4.00 < log10 𝑛𝑒 [cm −3 ]) < 4.75 for
4.00 < log(𝑛𝑒 [cm −3 ]) < 4.75 and 3.50 < log(𝑛𝑒 [cm −3 ]) < 4.10 re- NGC 1068 and 3.60 < log10 𝑛𝑒 [cm −3 ]) < 4.10 for NGC 4151. These
spectively (Table 1). The reason for this may simply be due to different values are an order of magnitude above those commonly reported and
pre-shock gas densities in the different objects (assuming the outflows assumed based on traditional density estimates, but are in agreement
in all three are shock-accelerated): for IC 5063 the pre-shock density with the results from alternative diagnostics such as multi-component
is 2.1 < log(𝑛𝑒 [cm −3 ]) < 2.7, however without higher velocity reso- photoionisation modelling. Overall, our results provide further mo-
lution spectra, we are unable to determine the quiescent gas densities tivation for the use of the transauroral lines in deriving electron
in NGC 1068 and NGC 4151. In addition, the differing post-shock densities of AGN-driven outflows.
densities in the three Seyferts may be due to different cooling condi- • The measured emission-line ratios for the warm ionised gas are
tions behind the shock front. Standard shock-jump conditions predict consistent with the dominant ionisation mechanisms being matter-
a compression factor of ∼4, however this may be much higher (∼100) bounded AGN-photoionisation in NGC 1068, and shock-ionisation
if the post-shock gas has cooled in pressure equilibrium (Sutherland and/or radiation-bounded AGN-photoionisation with a relatively flat
& Dopita 2017; Santoro et al. 2018). spectral index (and/or higher ionisation parameters and lower metal-
licities) in NGC 4151.
Moreover, all three objects have low-to-intermediate radio lu- • Along the radio axes, the outflows in the northeastern cones of
minosities (1.6×1022 < L1.4 GHz < 3.0×1023 W Hz −1 ; Table 3) — both objects have similar spatial extents to the radio structures — this
again, if the outflows in these Seyferts are shock-accelerated, then is consistent with the outflows in their NLRs being shock-accelerated
this would reinforce the importance of jet-driven shocks as a feed- by the radio jets and reionised by radiation from the AGN, although
back mechanism in the inner regions of galaxies, even at lower radio it does not rule out radiative acceleration.
luminosities, in agreement with a statistical study of nearby AGN • Applying the transauroral line technique to gas that has domi-
presented by Mullaney et al. (2013). Furthermore, the radio jets in nant shock-ionisation may incur an uncertainty on the derived elec-
NGC 1068 and NGC 4151 are oriented out of galactic disks by ∼ 45◦ tron densities by up to ±0.38 orders of magnitude, which is still far
and ∼ 36◦ respectively, unlike IC 5063 in which the jet propagates below the potential order-of-magnitude error incurred when using
almost directly into the plane of the disk. Therefore, at least within the techniques which are not sensitive to higher density gas. However,
central few hundred parsec of the AGN, this would show that inclined care must still be taken when using detailed density diagnostic tech-
jets can still have an impact on the kinematics and ionisation of the niques, as the ionisation mechanism of the gas may alter the results.
NLR, as predicted by recent relativistic hydrodynamic simulations Therefore, robust ionisation-mechanism diagnostics should be used
(Mukherjee et al. 2018; Meenakshi et al. 2022), which show that a to verify the validity of the density measurements.
jet inclined 𝜃 jet ∼ 45◦ to the galaxy’s disk may have a significant • Finally, by combining our findings with those for the nearby
effect on the kinematics, density and temperature of the gas within Seyfert 2 galaxy IC 5063, we find that the ionisation mechanisms and
the central few kpc (albeit less so than a jet inclined in the plane outflow conditions along the radio axes in the central few hundred
of the disk, such as is the case in IC 5063). Similar hydrodynamic parsecs vary significantly between the different objects. Thus overall,
simulations, specifically tailored to the situations in NGC 1068 and our study highlights the necessity of care when deriving information
NGC 4151, could thus be used to quantify the impact of their radio about outflow acceleration mechanisms from the ionisation of the

MNRAS 000, 1–20 (2015)

16 L.R. Holden et al.
gas, and the need for robust ionisation-mechanism diagnostics with Crenshaw D. M., Kraemer S. B., 2000c, ApJ, 532, L101
detailed observations. Crenshaw D. M., et al., 2000, AJ, 120, 1731
Crenshaw D. M., Fischer T. C., Kraemer S. B., Schmitt H. R., 2015, ApJ,
799, 83
Currie M. J., Berry D. S., Jenness T., Gibb A. G., Bell G. S., Draper P. W.,
ACKNOWLEDGEMENTS 2014, in Manset N., Forshay P., eds, Astronomical Society of the Pacific
Conference Series Vol. 485, Astronomical Data Analysis Software and
We thank the anonymous referee for their helpful comments and
Systems XXIII. p. 391
suggestions, which improved the clarity of this manuscript. LRH and D’Agostino J. J., Kewley L. J., Groves B. A., Medling A., Dopita M. A.,
CNT acknowledge support from STFC. Based on observations made Thomas A. D., 2019, MNRAS, 485, L38
with the NASA/ESA Hubble Space Telescope, and obtained from the Das V., et al., 2005, AJ, 130, 945
Hubble Legacy Archive, which is a collaboration between the Space Das V., Crenshaw D. M., Kraemer S. B., Deo R. P., 2006, AJ, 132, 620
Telescope Science Institute (STScI/NASA), the Space Telescope Eu- Dasyra K. M., Combes F., Oosterloo T., Oonk J. B. R., Morganti R., Salomé
ropean Coordinating Facility (ST-ECF/ESA) and the Canadian As- P., Vlahakis N., 2016, A&A, 595, L7
tronomy Data Centre (CADC/NRC/CSA). This research has made Davé R., Anglés-Alcázar D., Narayanan D., Li Q., Rafieferantsoa M. H.,
use of the NASA/IPAC Infrared Science Archive, which is funded by Appleby S., 2019, MNRAS, 486, 2827
the National Aeronautics and Space Administration and operated by Davies R., et al., 2020, MNRAS, 498, 4150
the California Institute of Technology. This work makes use of the Di Matteo T., Springel V., Hernquist L., 2005, Nat, 433, 604
Starlink software (Currie et al. 2014), which is currently supported Dopita M. A., Sutherland R. S., 1995, ApJ, 455, 468
Dubois Y., Peirani S., Pichon C., Devriendt J., Gavazzi R., Welker C., Volon-
by the East Asian Observatory. For the purposes of open access,
teri M., 2016, MNRAS, 463, 3948
the authors have applied a Creative Commons Attribution (CC BY)
Evans I. N., Ford H. C., Kinney A. L., Antonucci R. R. J., Armus L., Caganoff
licence to any Author Accepted Manuscript Arising. S., 1991, ApJ, 369, L27
Fabian A. C., 1999, MNRAS, 308, L39
Ferland G. J., Netzer H., 1983, ApJ, 264, 105
DATA AVAILABILITY Ferland G. J., et al., 2017, RMXAA, 53, 385
Ferrarese L., Merritt D., 2000, ApJ, 539, L9
The data used in this report is available from the Hubble Legacy Fiore F., et al., 2017, A&A, 601, A143
Archive (HLA) (https://hla.stsci.edu/hlaview.html) with Fischer T. C., et al., 2017, ApJ, 834, 30
proposal IDs GTO:5754 (PI Ford) and GTO:5124 (PI Ford) for the Foreman-Mackey D., Hogg D. W., Lang D., Goodman J., 2013, PASP, 125,
HST/WFPC2 [OIII] imaging, and proposal IDs GTO:7573 (PI Krae- 306
mer) and GTO:7569 (PI Hutchings) for the HST/STIS spectra. Fosbury R. A. E., Mebold U., Goss W. M., Dopita M. A., 1978, MNRAS,
183, 549
Gallimore J. F., Baum S. A., O’Dea C. P., Pedlar A., 1996, ApJ, 458, 136
García-Burillo S., et al., 2014, A&A, 567, A125
REFERENCES García-Burillo S., et al., 2019, A&A, 632, A61
Gebhardt K., et al., 2000, ApJ, 539, L13
Allen M. G., Groves B. A., Dopita M. A., Sutherland R. S., Kewley L. J.,
González Delgado R. M., Leitherer C., Heckman T. M., 1999, ApJS, 125,
2008, ApJS, 178, 20
489
Alonso-Herrero A., et al., 2011, ApJ, 736, 82
Gravity Collaboration et al., 2020, A&A, 634, A1
Antonucci R. R. J., Miller J. S., 1985, ApJ, 297, 621
Harris C. R., et al., 2020, Nat., 585, 357
Astropy Collaboration et al., 2013, A&A, 558, A33
Harrison C. M., Alexander D. M., Mullaney J. R., Swinbank A. M., 2014,
Astropy Collaboration et al., 2018, ApJ, 156, 123
MNRAS, 441, 3306
Audibert A., et al., 2023, A&A, 671, L12
Harrison C. M., Costa T., Tadhunter C. N., Flütsch A., Kakkad D., Perna M.,
Axon D. J., Marconi A., Capetti A., Macchetto F. D., Schreier E., Robinson
Vietri G., 2018, Nat. Astron, 2, 198
A., 1998, ApJ, 496, L75
Baldwin J. A., Phillips M. M., Terlevich R., 1981, PASP, 93, 5 Holden L. R., Tadhunter C. N., Morganti R., Oosterloo T., 2023, MNRAS,
Barbosa F. K. B., Storchi-Bergmann T., McGregor P., Vale T. B., Rogemar 520, 1848
Riffel A., 2014, MNRAS, 445, 2353 Holt J., Tadhunter C. N., Morganti R., Emonts B. H. C., 2011, MNRAS, 410,
Baron D., Netzer H., 2019, MNRAS, 486, 4290 1527
Baron D., Netzer H., Poznanski D., Prochaska J. X., Förster Schreiber N. M., Hopkins P. F., Elvis M., 2010, MNRAS, 401, 7
2017, MNRAS, 470, 1687 Hutchings J. B., et al., 1999, AJ, 118, 2101
Binette L., Wilson A. S., Storchi-Bergmann T., 1996, A&A, 312, 365 Johnston K. J., Elvis M., Kjer D., Shen B. S. P., 1982, ApJ, 262, 61
Boyce J. C., Menzel D. H., Payne C. H., 1933, Proceedings of the National Kakkad D., et al., 2018, A&A, 618, A6
Academy of Science, 19, 581 Kraemer S. B., Crenshaw D. M., 2000a, ApJ, 532, 256
Capetti A., Macchetto F. D., Lattanzi M. G., 1997, ApJ, 476, L67 Kraemer S. B., Crenshaw D. M., 2000b, ApJ, 544, 763
Cardelli J. A., Clayton G. C., Mathis J. S., 1989, ApJ, 345, 245 Kraemer S. B., Turner T. J., Couto J. D., Crenshaw D. M., Schmitt H. R.,
Carral P., Turner J. L., Ho P. T. P., 1990, ApJ, 362, 434 Revalski M., Fischer T. C., 2020, MNRAS, 493, 3893
Cecil G., Bland J., Tully R. B., 1990, ApJ, 355, 70 Liu G., Zakamska N. L., Greene J. E., Nesvadba N. P. H., Liu X., 2013,
Cicone C., et al., 2014, A&A, 562, A21 MNRAS, 436, 2576
Cicone C., Brusa M., Ramos Almeida C., Cresci G., Husemann B., Mainieri Lopez-Rodriguez E., et al., 2018, ApJ, 859, 99
V., 2018, Nat. Astron., 2, 176 Luridiana V., Morisset C., Shaw R. A., 2015, A&A, 573, A42
Colina L., et al., 2015, A&A, 578, A48 Magorrian J., et al., 1998, ApJ, 115, 2285
Collins N. R., Kraemer S. B., Crenshaw D. M., Bruhweiler F. C., Meléndez Mandal A., Mukherjee D., Federrath C., Nesvadba N. P. H., Bicknell G. V.,
M., 2009, ApJ, 694, 765 Wagner A. Y., Meenakshi M., 2021, MNRAS, 508, 4738
Congiu E., et al., 2017, MNRAS, 471, 562 May D., Steiner J. E., 2017, MNRAS, 469, 994
Crenshaw D. M., Kraemer S. B., 2000a, ApJ, 532, 247 May D., Steiner J. E., Menezes R. B., Williams D. R. A., Wang J., 2020,
Crenshaw D. M., Kraemer S. B., 2000b, ApJ, 532, L101 MNRAS, 496, 1488

MNRAS 000, 1–20 (2015)

 NLR warm-ionised outflows in NGC 1068 and NGC 4151 17
Meena B., Crenshaw D. M., Schmitt H. R., Revalski M., Fischer T. C., Polack Seyfert C. K., 1943, ApJ, 97, 28
G. E., Kraemer S. B., Dashtamirova D., 2021, ApJ, 916, 31 Sharp R. G., Bland-Hawthorn J., 2010, ApJ, 711, 818
Meena B., et al., 2023, ApJ, 943, 98 Silk J., Rees M. J., 1998, A&A, 331, L1
Meenakshi M., Mukherjee D., Wagner A. Y., Nesvadba N. P. H., Morganti Somerville R. S., Hopkins P. F., Cox T. J., Robertson B. E., Hernquist L.,
R., Janssen R. M. J., Bicknell G. V., 2022, MNRAS, 511, 1622 2008, MNRAS, 391, 481
Miller B. P., Brandt W. N., Schneider D. P., Gibson R. R., Steffen A. T., Wu Spence R. A. W., 2018, PhD thesis, University of Sheffield, https:
J., 2011, ApJ, 726, 20 //etheses.whiterose.ac.uk/23431/
Mingozzi M., et al., 2019, A&A, 622, A146 Spence R. A. W., Tadhunter C. N., Rose M., Rodríguez Zaurín J., 2018,
Morganti R., Oosterloo T., Tsvetanov Z., 1998, ApJ, 115, 915 MNRAS, 478, 2438
Morganti R., Holt J., Saripalli L., Oosterloo T. A., Tadhunter C. N., 2007, Speranza G., et al., 2022, A&A, 665, A55
A&A, 476, 735 Springel V., Di Matteo T., Hernquist L., 2005, MNRAS, 361, 776
Morganti R., Frieswijk W., Oonk R. J. B., Oosterloo T., Tadhunter C., 2013, Storchi-Bergmann T., McGregor P. J., Riffel R. A., Simões Lopes R., Beck
A&A, 552, L4 T., Dopita M., 2009, MNRAS, 394, 1148
Morganti R., Oosterloo T., Oonk J. B. R., Frieswijk W., Tadhunter C., 2015, Storchi-Bergmann T., Lopes R. D. S., McGregor P. J., Riffel R. A., Beck T.,
A&A, 580, A1 Martini P., 2010, MNRAS, 402, 819
Mukherjee D., Wagner A. Y., Bicknell G. V., Morganti R., Oosterloo T., Sun A.-L., Greene J. E., Zakamska N. L., 2017, ApJ, 835, 222
Nesvadba N., Sutherland R. S., 2018, MNRAS, 476, 80 Sutherland R. S., Dopita M. A., 2017, ApJS, 229, 34
Mullaney J. R., Alexander D. M., Fine S., Goulding A. D., Harrison C. M., Tadhunter C., 2016, A&ARv, 24, 10
Hickox R. C., 2013, MNRAS, 433, 622 Tadhunter C., Morganti R., Rose M., Oonk J. B. R., Oosterloo T., 2014, Nat.,
Müller Sánchez F., Davies R. I., Genzel R., Tacconi L. J., Eisenhauer F., 511, 440
Hicks E. K. S., Friedrich S., Sternberg A., 2009, ApJ, 691, 749 Ulvestad J. S., Wilson A. S., 1984, ApJ, 285, 439
Mundell C. G., Wrobel J. M., Pedlar A., Gallimore J. F., 2003, ApJ, 583, 192 Venturi G., et al., 2021, A&A, 648, A17
Nelson C. H., Weistrop D., Hutchings J. B., Crenshaw D. M., Gull T. R., Villar-Martín M., Tadhunter C., Morganti R., Axon D., Koekemoer A., 1999,
Kaiser M. E., Kraemer S. B., Lindler D., 2000, ApJ, 531, 257 MNRAS, 307, 24
Nesvadba N. P. H., Lehnert M. D., Eisenhauer F., Gilbert A., Tecza M., Abuter Wagner A. Y., Bicknell G. V., 2011, ApJ, 728, 29
R., 2006, ApJ, 650, 693 Wang J., Fabbiano G., Elvis M., Risaliti G., Mundell C. G., Karovska M.,
Oosterloo T. A., Morganti R., Tzioumis A., Reynolds J., King E., McCulloch Zezas A., 2011a, ApJ, 736, 62
P., Tsvetanov Z., 2000, ApJ, 119, 2085 Wang J., et al., 2011b, ApJ, 742, 23
Oosterloo T., Raymond Oonk J. B., Morganti R., Combes F., Dasyra K., Williams D. R. A., et al., 2017, MNRAS, 472, 3842
Salomé P., Vlahakis N., Tadhunter C., 2017, A&A, 608, A38 Wilson A. S., Ulvestad J. S., 1983, ApJ, 275, 8
Osterbrock D. E., Ferland G. J., 2006, Astrophysics of gaseous nebulae and Wilson A. S., Ulvestad J. S., 1987, ApJ, 319, 105
active galactic nuclei Winge C., Axon D. J., Macchetto F. D., Capetti A., 1997, ApJ, 487, L121
Osterbrock D. E., Koski A. T., 1976, MNRAS, 176, 61P Woo J.-H., Urry C. M., 2002, ApJ, 579, 530
Pedlar A., Howley P., Axon D. J., Unger S. W., 1992, MNRAS, 259, 369 Yuan W., et al., 2020, ApJ, 902, 26
Pedlar A., Kukula M. J., Longley D. P. T., Muxlow T. W. B., Axon D. J., Zamorani G., et al., 1981, ApJ, 245, 357
Baum S., O’Dea C., Unger S. W., 1993, MNRAS, 263, 471
Raban D., Jaffe W., Röttgering H., Meisenheimer K., Tristram K. R. W., 2009,
MNRAS, 394, 1325
Ramos Almeida C., Acosta-Pulido J. A., Tadhunter C. N., González-
Fernández C., Cicone C., Fernández-Torreiro M., 2019, MNRAS, 487,
L18 APPENDIX A: CRITICAL DENSITIES AND IONISATION
Revalski M., Crenshaw D. M., Kraemer S. B., Fischer T. C., Schmitt H. R., ENERGIES FOR THE EMISSION LINES USED IN OUR
Machuca C., 2018, ApJ, 856, 46 ANALYSIS
Revalski M., et al., 2021, ApJ, 910, 139
Revalski M., et al., 2022, ApJ, 930, 14 In Table A1, we present critical densities and ionisation energies for
Riffel R. A., 2021, MNRAS, 506, 2950 the lines used in our analysis, calculated using the PyNeb Python
Riffel R., Rodríguez-Ardila A., Aleman I., Brotherton M. S., Pastoriza M. G., module for a gas of temperature Te = 15,000 K.
Bonatto C., Dors O. L., 2013, MNRAS, 430, 2002 There have previously been concerns that the transauroral lines —
Riffel R. A., Vale T. B., Storchi-Bergmann T., McGregor P. J., 2014, MNRAS, used to derive electron densities and reddenings in Section 4.1 —
442, 656
do not trace the same gas that is emitting other key diagnostic lines
Riffel R. A., Bianchin M., Riffel R., Storchi-Bergmann T., Schönell A. J.,
Dahmer-Hahn L. G., Dametto N. Z., Diniz M. R., 2021, MNRAS, 503,
such as H𝛽 and [OIII]𝜆𝜆4959,5007 (see Sun et al. 2017, Rose et al.
5161 2018, Spence et al. 2018 and Holden et al. 2023). We note that, if
Robinson A., Binette L., Fosbury R. A. E., Tadhunter C. N., 1987, MNRAS, the transauroral lines originate from denser clumps of gas within the
227, 97 same cloud complexes as the clouds emitting other lines (Sun et al.
Robinson A., et al., 1994, A&A, 291, 351 2017), we would also expect those clumps to also radiate strongly in
Rodríguez-Ardila A., Riffel R., Pastoriza M. G., 2005, MNRAS, 364, 1041 H𝛽 since the recombination line emissivity scales as 𝑛2 . Furthermore,
Rose M., Tadhunter C. N., Holt J., Ramos Almeida C., Littlefair S. P., 2011, we note that the transauroral lines have critical densities that are
MNRAS, 414, 3360 closer to the critical density of the [OIII]𝜆𝜆4959,5007 lines than the
Rose M., Tadhunter C., Ramos Almeida C., Rodríguez Zaurín J., Santoro F., traditional [SII] and [OII] lines (Table A1), so they are more likely
Spence R., 2018, MNRAS, 474, 128
to trace the [OIII]-emitting clouds than the traditional lines (Rose
Santoro F., Rose M., Morganti R., Tadhunter C., Oosterloo T. A., Holt J.,
et al. 2018; Spence et al. 2018). Furthermore, the transauroral ratios
2018, A&A, 617, A139
Santoro F., Tadhunter C., Baron D., Morganti R., Holt J., 2020, A&A, 644, involve emission lines that arise from transitions within the [OII]
A54 ion, which has an ionisation energy that is closer to the ionisation
Schaye J., et al., 2015, MNRAS, 446, 521 energy of [OIII] than [SII]. This highlights that the transauroral lines
Schlafly E. F., Finkbeiner D. P., 2011, ApJ, 737, 103 are likely better tracers of the [OIII]-emitting gas than the commonly
Schlegel D. J., Finkbeiner D. P., Davis M., 1998, ApJ, 500, 525 used [SII](6717/6731) ratio.

MNRAS 000, 1–20 (2015)

18 L.R. Holden et al.

Emission line ncrit (cm −3 ) Eion (eV)

10 log10ne
[FeVII]𝜆6087 1.9×107 125.0
[FeVII]𝜆3759 3.5×108
1.8×107
125.0
9

[NeV] 3426 / [FeVII] 6086

[NeV]𝜆3426 126.2
[NeIII]𝜆3869 1.3×107 63.4
8
[OIII]𝜆4959,5007 7.8×105 54.9
log U
Transauroral [OII] and [SII] lines
[OII]𝜆7320,7331 6.9×106 35.1 7
[OII]𝜆7319,7330 4.7×106 35.1
[SII]𝜆4069 3.2×106 23.3 6
[SII]𝜆4076 1.6×106 23.3

Traditional [OII] and [SII] lines 5

[OII]𝜆3726 4.8×103 35.1
1.4×103
4
[OII]𝜆3729 35.1
[SII]𝜆6716 1.9×103 23.3
[SII]𝜆6731 5.1×103 23.3
3
Table A1. Critical densities and ionisation energies at Te = 15,000 K for
several key diagnostic lines that trace the warm ionised outflow phase: the 1.0 1.5 2.0 2.5 3.0 3.5
[NeV]𝜆3426 and [NeIII]𝜆3869 lines are used to investigate the presence
of matter-bounded gas (Section 4.2); the lines in the [FeVII](6087/3759)
[FeVII](6086/3759)
ratio are used to determine densities for high ionisation gas (Section 4.3);
the [OIII]𝜆4959,5007 doublet is used for kinematics (Section 4.4.1); the
Figure B1. [FeVII](6087/3759) vs [NeV]𝜆3426/[FeVII]𝜆6086 diagnostic di-
[SII](6717/6731) and [SII](3726/3729) are ‘traditional’ density ratios, and the
agram, as in Figure 7, with radiation-bounded AGN-photoionisation grids
transauroral lines are used in the TR method to derive electron densities and
generated with CLOUDY for two values of spectral index: 𝛼 = 1.0 (green)
reddenings (Section 4.1). The PyNeb Python module was used to produce
and 𝛼 = 1.5 (purple). Otherwise, the labelling, colour, marker and line
the values in this table.
scheme is the same as in Figure 7.

APPENDIX C: MODELLING THE EFFECT OF

SHOCK-IONISATION ON THE TRANSAURORAL LINE
RATIO DENSITY DIAGNOSTIC GRID
In order to investigate the effect of shock-ionisation on the TR tech-
nique7 , in Figure C1 we plot the expected TR([OII]) and TR([SII])
APPENDIX B: VARIATION OF THE [FEVII](6086/3759) VS line ratios with the radiation-bounded diagnostic grid for pho-
[NEV]𝜆3426/[FEVII]𝜆6086 DIAGNOSTIC DIAGRAM WITH toionised gas that we previously presented in Section 4.1 (Figure
SPECTRAL INDEX 4). The shock models shown on this grid were taken from the library
presented by Allen et al. (2008) (generated with the MAPPINGS III
The ratios used in the [FeVII](6086/3759) vs code), and are for solar-composition gas.
[NeV]𝜆3426/[FeVII]𝜆6086 diagnostic diagram (Section 4.3; We first investigate the effect of varying the gas veloc-
Figure 7) are sensitive to both electron density and temperature, and ity between 0 < 𝑣 shock < 1000 km s −1 with a constant mag-
thus the position of the AGN-photoionisation grid on this diagram √
netic parameter of 𝐵/ 𝑛=2 𝜇G cm3/2 and pre-shock densities of
— which we use to determine the density of the high ionisation 𝑛 = 1, 10, 100, 1000 cm −3 . Assuming a compression factor of ∼ 100
gas — changes with the assumed ionising source spectral index (𝛼) if the gas cools in pressure equilibrium behind the shock (Sutherland
and ionisation parameter (𝑈) of the gas. To further investigate this & Dopita 2017; Santoro et al. 2018), these correspond to post-shock
beyond only varying the ionisation parameter (as is shown in Figure densities of 𝑛 = 102 , 103 , 104 , 105 cm −3 . From Figure C1, it can be
7 for 𝛼 = 1.5), here we show the effect of assuming a lower spectral seen that for high shock velocities (⪆500 km s −1 ) at a given density,
index. We used the same CLOUDY model as described in Section the modelled line ratios are similar to those predicted by photoion-
4.3, but instead took the spectral index to be 𝛼 = 1.0. We present isation modelling. However, for lower shock velocities (a few hun-
the resulting grid (along with the grid for 𝛼 = 1.5 for comparison dred km s −1 ), the predicted densities may differ by ±0.22 orders of
purposes) in Figure B1. magnitude. Similarly, shock-ionisation may effect the derived color
From Figure B1, it can be seen that a flatter spectral index produces excesses by E(B-V)TR ± 0.13.
lower [NeV]𝜆3426/[FeVII]𝜆6086 ratios, with little effect on the [Fe- Secondly, we investigate the effect of varying the magnetic param-
VII](6086/3759) ratios: for low values of [NeV]𝜆3426/[FeVII]𝜆6086 √
eter between typical values for the ISM (2 < 𝐵/ 𝑛 < 4 𝜇G cm3/2 :
(i.e. as measured in Aperture 4 for NGC 1068), the effect on de- Dopita & Sutherland 1995; Allen et al. 2008). We did this for three
rived density is small (∼0.1 dex). However, a shallower spectral in- values of shock velocity (𝑣 shock = 400, 600, 800 km s −1 ), which we
dex cannot reproduce higher values of [NeV]𝜆3426/[FeVII]𝜆6086 show in Figure C2. The impact on derived electron densities is greater
(as measured in NGC 1068 apertures 2 and 3) without very
low (log 𝑈 < −3.0) ionisation parameters. Therefore, we use the
CLOUDY grid with 𝛼 = 1.5 to derive electron densities for the 7 The effect of shock-ionisation on transauroral line density and reddening
high ionisation gas in Section 4.3. diagnostic was investigated in a preliminary fashion by Spence (2018).

MNRAS 000, 1–20 (2015)

 NLR warm-ionised outflows in NGC 1068 and NGC 4151 19

2.0 0 2.0 2.0 0 2.0

0.25 2.5 0.25 2.5
E(B-V) 3.0 E(B-V) 3.0
1.5 0.5 1.5 0.5
3.5 log10nH 0.75 3.5 log10nH
0.75
1.0 1 1.0 1
4.0 4.0
log10[OII]

log10[OII]
0.5 4.5 0.5 4.5
0.0 5.0 0.0 5.0

0.5 0.5

1.0 1.0
2.0 1.5 1.0 0.5 0.0 2.0 1.5 1.0 0.5 0.0
log10[SII] log10[SII]

Figure C1. TR([SII]) vs TR([OII]) grid for radiation-bounded AGN- Figure C2. TR grid (as in Figure 4) with line ratios predicted by shock
photoionised gas (black with red E(B-V) labels; as in Figure 4) with the modelling for three values of shock velocity 𝑣shock =400, 600 and 800 km s −1
√
line ratios expected from modelled shocks (blue crosses) of fixed magnetic and two values of magnetic parameter (2 < 𝐵/ 𝑛 < 4 𝜇G cm3/2 ) which are
√
parameter (𝐵/ 𝑛=2 𝜇G cm3/2 ) at different post-shock densities (labelled in typical of the ISM. Line ratios generated using the same post-shock density
blue), and for a range of shock velocities (0 < 𝑣shock < 1000 km s −1 ). Line are show as crosses with the same color (labelled), and points of the same
ratios from shock modelling at a single post-shock density (assuming a com- density and velocity are joined by a dashed line.
pression factor of 100) are grouped by blue ellipses, and points of the same
shock velocity (vshock = 400, 600, 800 km s −1 ) are joined by dashed blue
lines.

at higher densities (±0.25 orders of magnitude) than at lower den-

2.0 0 2.02.5
E(B-V) 0.25
sities (±0.10 orders of magnitude), with little effect on the derived
3.0
reddening value. 1.5 0.5
Finally, we quantify the effect of simultaneously varying the veloc- 3.5 log10nH
ity (between 0 km s −1 < 𝑣 shock < 1000 km s −1 ) and the magnetic 0.75
√
parameter (2 < 𝐵/ 𝑛 < 4 𝜇G cm3/2 ) on the TR-derived electron 1.0 1 4.0
densities and reddenings (Figure C3). We find that the effect on
log10[OII]

the derived density is ±0.38 orders of magnitude, regardless of the 0.5 4.5
density of the modelled gas, and that the effect on derived reddening
value is the same as varying the velocity (E(B-V)±0.13).
0.0 5.0
APPENDIX D: THE ORIGIN OF THE TRANSAURORAL 0.5
LINES IN NGC 1068, NGC 4151 AND IC 5063
In order to investigate the origin of the transauroral lines in the 1.0
archetypal Seyfert galaxies (and the nearby Sey 2 IC 5063: Holden
et al. 2023), we plotted the measured [OII](7319+7331)/[OIII]𝜆5007 2.0 1.5 1.0 0.5 0.0
vs [SII](4068+4076)/H𝛽 ratios with overlaid grids from radiation- log10[SII]
bounded photoionisation modelling (as first presented by Spence
et al. 2018) and shock modelling, which we present in Figures D1a
and D1b, respectively.
Figure C3. TR grid (as in Figure 4) with line ratios predicted by shock
The photoionisation models in Figure D1a were generated using √
modelling for two magnetic parameters (𝐵/ 𝑛 =2,4 𝜇G cm3/2 ) and a range
CLOUDY for a radiation-bounded cloud with varying values of den- of shock velocities (0 < 𝑣shock < 1000 km s −1 ) at each value of post-shock
sity, ionisation parameter and spectral index. The measured ratios density. Predicted ratios with the same post-shock density are shown with
in our apertures for NGC 1068 and NGC 4151 are consistent with crosses of a colour for each density (labelled) and are grouped with ellipses.
this grid; however, the corresponding ionisation parameters are a
half-an-order-of-magnitude higher than that which was required to
reproduce the measured transauroral line ratios (log 𝑈 = −3; Figure

MNRAS 000, 1–20 (2015)

20 L.R. Holden et al.
4). This is further evidence that radiation-bounded AGN photoion-
isation is not the dominant ionisation mechanism in our apertures,
and in the case of NGC 1068 can be explained as the [OIII] and
H𝛽 emission being dominated by matter-bounded components (as
indicated by the measured HeII/H𝛽 ratios; Figure 5). This is be-
cause matter-bounded emission will increase the relative strength of
the [OIII] and H𝛽 lines, reducing the [OII](7319+7331)/[OIII]𝜆5007
and [SII](4068+4076)/H𝛽 ratios used in Figure D1a and thus giving a
higher corresponding ionisation parameter on the radiation-bounded
grid.
In Figure D1b, we present the same measured ratios, but with the
expected ratios from shock-ionisation modelling. The shock mod-
els here are those presented by Allen et al. (2008), and are for two
values of pre-shock density (101 cm −3 and 102 cm −3 ), velocities
of 𝑣 shock = 400, 600, 800 km s −1 , and magnetic parameters in the
√
range 2 < 𝐵/ 𝑛 < 10 𝜇G cm3/2 . In all apertures for NGC 4151, the
measured line ratios are consistent with shock ionisation, albeit with
√
relatively high magnetic parameters (4 < 𝐵/ 𝑛 < 10 𝜇G cm3/2 ).
These magnetic parameters are higher than those typical for the
√
ISM (2 < 𝐵/ 𝑛 < 4 𝜇G cm3/2 ; Dopita & Sutherland 1995; Allen
et al. 2008), potentially indicating higher magnetic fields associated
with the shocked material. If the gas detected in our NGC 4151
data is indeed shock-ionised, then the position of the shock model
grid on the diagram would explain why the ionisation parameter de-
duced from the radiation-bounded photoionisation grid (Figure D1a)
for the NGC 4151 apertures is higher than expected from the TR
diagnostic grid (Figure 4): shock ionisation produces lower values
of [OII](7319+7331)/[OIII]𝜆5007 and [SII](4068+4076)/H𝛽, corre-
sponding to higher ionisation parameters on the photoionisation grid.
We also present [OII](7319+7331)/[OIII]𝜆5007 and
[SII](4068+4076)/H𝛽 ratios for the Seyfert 2 galaxy IC 5063,
as measured from the dataset described in Holden et al. (2023).
In agreement with our previous findings (Holden et al. 2023), we
find that the ratios are consistent with radiation-bounded AGN-
photoionisation for an ionisation parameter of −3 < log 𝑈 < −2 and
densities in the range 3 < log10 (𝑛𝑒 [cm −3 ]) < 4.

This paper has been typeset from a TEX/LATEX file prepared by the author.

MNRAS 000, 1–20 (2015)

 NLR warm-ionised outflows in NGC 1068 and NGC 4151 21

CLOUDY radiation-bounded MAPPINGS III

photoionisation models shock models
0.5 105cm 3
log10([OII](7319+7331)) / [OIII] 5007)

a) 104cm 3 b)
0.0 103cm 3 logU = -4
=2.0,1.5,1.0 104cm 3
0.5 2.0 G cm3/2
103cm 3

1.0 2.0 G cm3/2 4.0 G cm3/2

logU = -3
1.5 4.0 G cm3/2 10.0 G cm3/2
2.0 9.5 G cm3/2
logU = -2 NGC 1068
NGC 4151
2.5 IC 5063
1.5 1.0 0.5 0.0 0.5 1.5 1.0 0.5 0.0 0.5
log10([SII](4068+4076) / H ) log10([SII](4068+4076) / H )
Figure D1. TR([SII])/H𝛽 vs TR([OII])/[OIII]𝜆5007 ratio grids from CLOUDY radiation-bounded photoionisation modelling (a) and the Allen et al. (2008)
MAPPINGS III shock-ionisation models (b), along with measured values for each aperture in NGC 1068 (orange crosses), NGC 4151 (blue crosses) and
IC 5063 (green crosses; see Holden et al. 2023). The modelled photoionisation line ratio grid shown in a) is for three spectral indices (𝛼 = 1.0, 1.5, 2.0;
labelled) at different electron densities (squares: 𝑛𝑒 =103 cm −3 ; triangles: 104 cm −3 ; stars: 105 cm −3 ) and ionisation parameters (log 𝑈; labelled), while the
shock-ionisation grids in b) are for post-shock densities of 𝑛=103 cm −3 and 𝑛=104 cm −3 (labelled; assuming a shock compression factor of 100), magnetic
√
parameters of 𝐵/ 𝑛 = 2, 4, 10 𝜇G cm3/2 . (labelled), and velocities of 𝑣shock = 400, 600, 800 km s −1 .

MNRAS 000, 1–20 (2015)