GASP XIX: AGN and their outflows at the center of jellyfish galaxies
Mario Radovich, Bianca Poggianti, Yara L. Jaffe', Alessia Moretti, Daniela Bettoni, Marco Gullieuszik, Benedetta Vulcani, Jacopo Fritz
MMNRAS , 1–20 (2019) Preprint 23 May 2019 Compiled using MNRAS L A TEX style file v3.0
GASP XIX: AGN and their outflows at the center ofjellyfish galaxies
Mario Radovich (cid:63) , Bianca Poggianti , Yara L. Jaff´e , Alessia Moretti ,Daniela Bettoni , Marco Gullieuszik , Benedetta Vulcani , and Jacopo Fritz INAF- Osservatorio astronomico di Padova, Vicolo Osservatorio 5, IT-35122 Padova, Italy Instituto de F´ısica y Astronom´ıa, Facultad de Ciencias, Universidad de Valpara´ıso, Avda. Gran Breta˜na 1111, Casilla 5030, Valpara´ıso, Chile Instituto de Radioastronom´ıa y Astrof´ısica, UNAM, Campus Morelia, A.P. 3-72, C.P. 58089, Mexico
Accepted 2019 March 15. Received 2019 March 15; in original form 2019 February 5
ABSTRACT
The GASP survey, based on MUSE data, is unveiling the properties of the gas inthe so-called ”jellyfish” galaxies: these are cluster galaxies with spectacular evidenceof gas stripping by ram pressure. In a previous paper, we selected the seven GASPgalaxies with the most extended tentacles of ionized gas, and based on individual di-agnostic diagrams concluded that at least five of them present clear evidence for anActive Galactic Nucleus. Here we present a more detailed analysis of the emissionlines properties in these galaxies. Our comparison of several emission line ratios withboth AGN and shock models show that photoionization by the AGN is the dominantionization mechanism. This conclusion is strengthened by the analysis of H β luminosi-ties, the presence of nuclear iron coronal lines and extended ( >
10 kpc) emission lineregions ionized by the AGN in some of these galaxies. From emission line profiles, wefind the presence of outflows in four galaxies, and derive mass outflow rates, timescalesand kinetic energy of the outflows.
Key words: galaxies: clusters: general – galaxies: active
It is now widely accepted that there is a strong connectionbetween the presence of an Active Galactic Nucleus (AGN)and the host galaxy properties, based both on cosmologicalmodels and observational results from wide-field surveys (seee.g. Heckman & Best 2014, and refs for a review). However,the way this interaction occurs is still unclear, and it mayactually be the outcome of a wide range of different physi-cal processes (e.g. merging, bars). A major improvement inour understanding of the complex environment around AGNis given by the availability of Integral Field Spectroscopy(IFU), allowing to map emission line fluxes and kinematicstracing the AGN and its surroundings (see e.g. Venturi et al.2018; Ilha et al. 2019; Mingozzi et al. 2019).An important issue is the effect of the environment onthe presence of the AGN: it is still debated (see e.g. Marzianiet al. 2017, and refs) whether or not a dense galaxy environ-ment such as in galaxy clusters has any effect on the presenceof AGN. Early spectroscopic studies (Dressler et al. 1985)suggested that the fraction of AGN in clusters ( ∼ (cid:63) E-mail: [email protected] nificantly lower than in a field environment ( ∼ c (cid:13) a r X i v : . [ a s t r o - ph . GA ] M a y M. Radovich et al. id cluster z cl scale RA DEC class seeing r NLR r AGN log L [OIII] log L corr[OIII] kpc/ (cid:48)(cid:48) (cid:48)(cid:48) kpc kpc erg s − erg s − JO135 A3530 0.05480 1.07 12 57 04.2 -30 22 30.0 AGN 0.73 0.8 3.7 41.24 ± ± ± ± ± ± ± ± ± ± ± ± Table 1.
The table shows: the galaxy ID, the host cluster name, redshift and scale, the coordinates of the central spaxel, the seeing, theclassification (AGN/LINER/SF) assigned in Poggianti et al. (2017b) and, for galaxies classified as AGN or LINER, the estimated AGNsizes and the observed and dereddened [OIII] λ r NLR . et al. 2019) for the less luminous AGN to tens of kpc for thebrightest AGN (Harrison et al. 2014).The GAs Stripping Phenomena (GASP) survey (Pog-gianti et al. 2017a, P17a hereafter) is aimed at studying withthe MUSE Integral Field spectrograph on VLT the prop-erties of the so-called jellyfish galaxies in clusters, whose tentacles of UV and optically bright material that makethem similar to a jellyfish (Smith et al. 2010) are thoughtto originate via ram-pressure stripping by the intra-clustermedium (Ebeling et al. 2014; Fumagalli et al. 2014; Rawleet al. 2014; Fossati et al. 2016). Poggianti et al. (2017b)(P17b hereafter) showed that at least five and possibly sixof seven galaxies with the strongest evidence of gas strippingand the most favourable conditions for ram pressure (Jaff´eet al. 2018) host an AGN, suggesting a connection betweenram pressure stripping and AGN triggering. In P17b the[NII] λ α vs. [OIII] λ β line ratios were used toselect the most likely mechanism that ionized the gas: radi-ation from hot young stars in star-forming regions, from anAGN, a combination of them ( composite ), and either low-luminosity AGN or shocks (LINERs), using as reference theclassification by Kewley et al. (2006). As already shown inPoggianti et al. (2019), adding other line ratio diagnosticdiagrams such as [OI] λ α and [SII] λλ α vs. [OIII] λ β (Veilleux & Osterbrock 1987) can pro-vide a more detailed description of the physical processesat work. In this paper we expand the work by P17b andcritically scrutinize those results using all three main diag-nostic diagrams simultaneously and comparing observed lineratios with photoionization and shock models. Moreover, weinspect additional features such as coronal Fe lines and an-alyze separately the extended extranuclear AGN-poweredemission regions. Finally, we discuss the presence and prop-erties of outflows.The paper is structured as follows. A short summary ofthe data and how they were analyzed is given in Sect.2. InSect.3, observed emission line ratios are compared with bothphotoionization and shock models, to confirm that photoion-ization from the AGN is required to reproduce the line ratiosand derive the best-fit model parameters. As further probesof the AGN, we estimate the maximum contribute to theobserved H β luminosity from shock models; in some cases,we detect the presence of high-ionization iron coronal lines(JO201 and JO135) and of extended ( >
10 kpc) AGN-likeemission lines (JO204 and JO135): this is used in Sect.3.4 toderive the number of ionizing photons that should be emit- ted by the AGN. In Sect.4 we analyze the [OIII] λ H =70 km s − Mpc − , Ω m = 0 .
3, Ω Λ = 0 . . In this paper we analyze the seven galaxies in P17b (Table 1and Fig. 1), which are all characterized by tails of ionized gasat least as long as the stellar galaxy diameter. These galax-ies represent extreme cases where the cluster environmentstrongly acts on the gas and possibly on the AGN. Spectraand individual emission lines for the central spaxel of eachgalaxy, selected as discussed later, are displayed in Fig. 2.In the following description, a reference is given in squarebrackets after the galaxy name for those galaxies studiedindividually in a GASP paper.
JO201 , or Kaz 364 [P17a, Bellhouse et al. (2017),Georgeet al. (2018)] was also classified by Arnold et al. (2009) asan AGN based on XMM observations. Two components arepresent in the nuclear emission lines (P17b, see also Fig. 2):a narrow, stronger component, and a broader one, slightlyblueshifted.
JO204 [Gullieuszik et al. (2017)] was included by Nevinet al. (2016) in a sample of 71 double-peaked AGN selectedfrom the SDSS and classified as an AGN with an outflow.At least two components are visible in the MUSE spectra;an extended emission with AGN-like line ratios is detectedup to ∼
20 kpc from the nucleus, see below Sect. 3.4.
JO135
Complex emission line profiles are observed in thenucleus, with a strong redshifted wing (Fig. 2). AGN-likeemission is present up to ∼
10 kpc (P17b), see Sect. 3.4.
JW100 , or IC 5337, was classified as an AGN by Wonget al. (2008), based on X–ray Chandra observations. It wasclassified as a head-tail radio source by Gitti (2013), whodetected radio emission in Very Large Array radio measure-ments at 1.4 and 4.8 GHz: the peak of the radio emissioncoincides with the MUSE center. Double-peaked profiles aredetected in the region around the nucleus (P17b).
JO175 , JO206 [P17a],
JO194 : in these galaxies emissionlines appear as single-component Gaussians.We refer to P17a for a detailed description of the GASP
MNRAS000
MNRAS000 , 1–20 (2019)
GN and outflows in jellyfish galaxies JO135 JO201JO204 JO206JW100 JO194 JO175
Figure 1.
VRI images built from the MUSE cubes for the seven galaxies analyzed in the paper. survey, data and adopted reduction techniques. Observa-tions were obtained with the MUSE spectrograph in wide-field mode with natural seeing (Bacon et al. 2010). One ortwo MUSE pointings per galaxy, each with a 2700sec ex-posure and covering a 1’x1’ field of view, are sampled with0.2”x0.2” pixels over the spectral range 4800-9300 ˚A with aspectral resolution FHWM ∼ . <
1” seeing (Table 1).We remind the reader that the fitting of emission lines inGASP was done using
KubeViz (Fossati et al. 2016). Ve-locity and velocity dispersion were derived from the fit ofthe lineset consisting of H α and the [NII] λλ MNRAS , 1–20 (2019)
M. Radovich et al.
Figure 2.
The plots show for the central spaxel of each galaxy: left - the uncorrected spectra with the identification of the main emissionlines; right - the normalized line profiles after the subtraction of the stellar underlying spectrum as described in the text: the dashedlines indicate the assumed galaxy velocity. A meaningful detection (SN >
3) of [FeVII] λ000
3) of [FeVII] λ000 , 1–20 (2019) GN and outflows in jellyfish galaxies Figure 2 – continued ity dispersions ( σ obs ) were corrected for the instrumentalcomponent: σ = (cid:112) σ obs − σ inst ; σ inst was derived at eachwavelength using a third order polynomial fit of the MUSEresolution curve (Fumagalli et al. 2014). In the following,we will use the data cube average filtered with a 5x5 pixelkernel in the spatial direction, unless otherwise stated, hav- ing subtracted the stellar underlying spectrum fitted withthe SINOPSIS code (Fritz et al. 2017). The stellar kinemat-ics is derived with the pPXF code (Cappellari & Emsellem2004) using stellar population templates from Vazdekis et al.(2010), see P17a for details. The galaxy center was defined as
MNRAS , 1–20 (2019)
M. Radovich et al.
Figure 2 – continued the centroid of the continuum map obtained by the KubeViz best fit model in the H α region.All the line fluxes, selected to have a signal to noise ratioSN >
3, were corrected for dust extinction using the Balmerdecrement as in P17a. Fig. 3 displays the extinction map( A v ) in each galaxy, as well as the electron density ( n e ): thiswas derived using the relations in Proxauf et al. (2014), as inP17a. We detect increased values of the extinction ( A v > A v < A v > .
5) in an extendedregion of radius ∼ (cid:48)(cid:48) around the central spaxel. JO201,JO204 and JO135 show a steep increase in density ( n e > . cm − ) in the nucleus: lower densities are measured inJO206 and JW100 ( n e ∼ cm − ), and JO194, JO175 ( n e < cm − ) . The left panels of Fig. 4 present the classificationin HII-regions, Composite, AGN and Liners based on[NII] λ α vs. [OIII] λ β as in P17b, whose mainconclusions are summarized here. In some cases (JW100,JO135 , JO201, JO204) one Gaussian was not enough tofit the emission line profiles and a double-Gaussian fit wasadopted: in these cases, the classification displayed in Fig. 4refers to the narrow component.Based on the P17b analysis, JO201, JO204, JO206,JO135 and JW100 present AGN-like line ratios in the in-ner kpcs. None of them shows broad ( > − ), per-mitted lines typical of the Broad Line Region in AGN; theobserved emission lines are therefore produced in the Nar-row Line Region (NLR). Extended AGN-like emission overseveral kpcs is observed in JO204 and JO135, and it canbe attributed to anisotropic ionization from the AGN (theso-called ionization cones). In JO175 the emission is mostly Compared to P17b, we improved the fitting of the lines of thecentral spaxels of JO135. due to star formation, while in JO194, composite line ratiosare detected throughout the galaxy.We define the size of the Narrow Line Region in theAGN candidates as in Bae et al. (2017), that is weighting theprojected distances from central spaxel on the [OIII] λ r NLR = (cid:88) r < rf [OIII] ( r ) / (cid:88) r < f [OIII] ( r ) , (1) r being the distance from the central spaxel and f [OIII] theline flux at that distance. The NLR size so defined is of theorder of 1 kpc for all galaxies: as the typical seeing was ∼ r AGN in Table 1), keeping in mind that thesevalues may be biased by spaxels with fainter [OIII] fluxes,where the measurement uncertainties may produce a wrongclassification. The AGN emission can be therefore describedas the sum of a pointlike source producing the bulk of theemission and a fainter, extended ( r >
We now complement the classification done in P17bwith a more detailed analysis: we simultaneously considerthe lines ( Hα , H β , [OI] λ λ λ λλ NebulaBayes (Thomas et al. 2018), apython code that adopts a Bayesian approach to select themodel optimally fitting the target emission line fluxes.
NebulaBayes includes grids where constant gas pres-sure photoionization models are computed with the
MAP-PINGS V code, for HII regions and AGN. A full discussion of
MNRAS000
MNRAS000 , 1–20 (2019)
GN and outflows in jellyfish galaxies JO135 JO201JO204 JO206JW100 JO175JO194
Figure 3.
For each galaxy the plots show the spatial distribution in the central 10 (cid:48)(cid:48) x 10 (cid:48)(cid:48) of left: extinction ( A v ), right: log n e . For allgalaxies, 1 arcsec is ∼ P/k log E peak
12 + log
O/H log U JO135 6.6 -1.2 9.16 -2.49JO201 7.0 -1.5 8.99 -2.77JO204 7.0 -1.5 8.99 -3.06JO206 6.6 -1.5 8.87 -3.06JW100 7.0 -1.5 9.15 -3.06JO194 6.2 -1.7 9.30 -3.34
Table 2.
Best-fit parameters for the nuclear AGN photoioniza-tion models. the assumptions and parameters of these models is given inThomas et al. (2018), we summarize here the main aspects.For HII regions, the ionizing continuum is defined bythe
SLUG2 (Krumholz et al. 2015) stellar population synthe-sis code, with five metallicities ( Z = 0.0004, 0.004, 0.008,0.02, 0.05). For AGN, the ionizing continuum is describedin Thomas et al. (2016), and is parametrized by the energyof the peak of the accretion disk emission ( E peak ), the pho- ton index of the inverse Compton scattered power-law tail(Γ), and the proportion of the total flux in the non-thermaltail ( p NT ). In the grid models, the latter two parameters arefixed (Γ=2, p NT =0.15).For both HII regions and AGN, the other model freeparameters are: the metallicity (12 + log O/H), the ion-ization parameter ( U ) and the gas pressure (log P/k ), with
P/k ∼ . n e T , see e.g. Kakkad et al. (2018).Considering the environment of these galaxies and thepresence of outflows in the nuclear regions, it is importantto understand what may be the contribution from shocks,and if shocks alone can produce the observed line ratios.As extensively discussed by Allen et al. (2008), in the so-called fast shock models the cooling of the hot gas behindthe shock front produces high energy photons which ionizethe pre–shocked gas (precursor). When the shock velocityis >
170 km s − , the contribution from the photoionizedgas in the precursor starts to become increasingly impor-tant and both high and low ionization lines are present inthe observed spectrum. Varying the input model parame- MNRAS , 1–20 (2019)
M. Radovich et al.
JO135JO201JO204JO206
Figure 4.
Color coded maps in the a region of 100x100 spaxels around the nucleus.
Left : classification from P17b (HII, composite, AGN,LINER).
Right:
NB models: HII; shock with n = 0 . − , solar abundances; shock with n = 1 cm − , solar abundances (M); shock with n = 1 cm − , 2x solar abundances (R); AGN. Spaxels classified as HII in P17b were not fitted with NB. ters, that is the pre–shock density, n , the shock velocity, v s , the pre-shock transverse magnetic field, B , and the gasatomic abundances, it is possible to produce a wide rangeof emission line ratios, from HII-like regions to Liners andAGN.Since shock model libraries are not directly available in NebulaBayes , we adapted the Allen et al. (2008) fast shockgrids so that they could be used in
NebulaBayes . From these grids, we selected the models with n =0.1, 1, 10 cm − , forwhich solar ( n =0.1, 1, 10 cm − ) and 2x solar ( n =1 cm − )abundances are available.For each model type (shock, HII and AGN), Nebula- We remind that the pre–shock density is not directly related tothe electron density measured e.g. by the [SII] λλ000
NebulaBayes . From these grids, we selected the models with n =0.1, 1, 10 cm − , forwhich solar ( n =0.1, 1, 10 cm − ) and 2x solar ( n =1 cm − )abundances are available.For each model type (shock, HII and AGN), Nebula- We remind that the pre–shock density is not directly related tothe electron density measured e.g. by the [SII] λλ000 , 1–20 (2019) GN and outflows in jellyfish galaxies JW100JO194JO175
Figure 4 – continued Bayes was run spaxel by spaxel, providing as output thebest-fit model line ratios and the χ : the optimal model wasselected as the one giving the lowest χ . We stress that dif-ferent reasons may contribute to produce the wrong classi-fication for a given spaxel, as for instance the uncertaintieson the line measurements, the limited number of parame-ters in the models, and the fact that in many cases we mayhave at the same time a contribution from different ionizingmechanisms. Fig. 5 presents AGN and shock model grids overlaid on theobserved spaxel line ratios; the latter are color coded withthe distance from the galaxy center. For AGN models, wedisplay those with log E peak = [ − , , , +0 .
5] around thebest-fit value of the central spaxel, and for different valuesof log U . The abundances in the nucleus derived from theNB fits are super-solar (12 + log O/H >
9) in the AGN-dominated nuclei, and are Solar (12 + log
O/H = 8 .
76) out-side. For fast shock models, we plot a grid of varying veloci- ties and magnetic field values, fixing the best fit density andmetallicity.The maps of the best fit classification from
Nebula-Bayes , compared with the classification derived as in P17b,are displayed in Fig. 4.Fast shock models in individual diagnostic diagrams canproduce a wide range of line ratios, covering both the HIIand AGN regions of the diagrams. For the most extremecases like JO135 and JO201 that have log [OIII]/H β ∼ v sh >
500 km s − while in our case the widthof the strongest component is σ v ∼
100 km s − . Whenlog [OIII] λ β is low ( < λ α ratios or too low[NII] λ α ratios, while AGN photoionization modelsreproduce all diagnostic ratios, as found for typical AGN byAllen et al. (2008).For JO194 the emission line ratios fall in the LINERside of the diagrams, with log [OIII] λ β ≤ .
2, whereoptical lines alone are not able to clearly separate between
MNRAS , 1–20 (2019) M. Radovich et al.
JO135: observed vs. AGN modelsJO135: observed vs. shock modelsJO201: observed vs. AGN modelsJO201: observed vs. shock models
Figure 5.
Observed emission line ratios color coded with the projected distance from the center; the empty square displays the valuemeasured in the central spaxel. The black solid and dashed ([NII]/H α panel) curves indicate the empirical SF/Composite/AGN classifi-cation by Kewley et al. (2006). For each galaxy, overlaid are best-fit AGN (not for JO175) and shock models. AGN models – The redlines display models for different values of log U , adopting the best-fit values of log P/k , 12 + log
O/H and E peak in the central spaxels;models with an offset ± E peak are displayed in cyan and blue respectively. For JO204 and JO135, the green line shows AGNmodels in the EENLR. Shock models – The lines display models for v sh = 100-700 km s − , pre-shock density n = 0 . − , solarabundances (JO135, JO201, JO206); n = 1 cm − , 2x solar abundances (JO204, JW100, JO175, JO194), and magnetic field B as in thelegend. MNRAS000
O/H and E peak in the central spaxels;models with an offset ± E peak are displayed in cyan and blue respectively. For JO204 and JO135, the green line shows AGNmodels in the EENLR. Shock models – The lines display models for v sh = 100-700 km s − , pre-shock density n = 0 . − , solarabundances (JO135, JO201, JO206); n = 1 cm − , 2x solar abundances (JO204, JW100, JO175, JO194), and magnetic field B as in thelegend. MNRAS000 , 1–20 (2019) GN and outflows in jellyfish galaxies JO204: observed vs. AGN modelsJO204: observed vs. shock modelsJO206: observed vs. AGN modelsJO206: observed vs. shock models
Figure 5 – continued MNRAS , 1–20 (2019) M. Radovich et al.
JW100: observed vs. AGN modelsJW100: observed vs. shock modelsJO194: observed vs. AGN modelsJO194: observed vs.shock models
Figure 5 – continued MNRAS000
Figure 5 – continued MNRAS000 , 1–20 (2019)
GN and outflows in jellyfish galaxies JO175: observed vs. shock models
Figure 5 – continued AGN and other ionization mechanisms (see e.g. Belfioreet al. 2016). Around the central spaxel there is a very smallregion, whose size is few spaxels, where AGN models pro-duce a lower χ than shock models. However, as displayed inFig. 4 the line ratios move to the SF region of the diagramsalready within 2 kpc; line ratios similar to what observedin the nucleus are also present up to ∼ L . − = 1 . × erg s − ). Since we de-tect a strong extinction ( A v > .
5) in the nuclear regions,it is possible that AGN emission is obscured by dust in theoptical.For JO175, line ratios are consistent with star forma-tion, as well as with some of the shock models as suggestedby the high [OI] λ α ratio, but not with an AGN.As a further test, we compare the observed, dust cor-rected L(H β ) luminosity within r NLR with the value derivedfrom the Allen et al. (2008) library, selecting models with n = 1 cm − and best-fit values in the central spaxel for B and v sh . We estimate the maximum contribution fromshocks to be negligible ( < <
20% for JO194,JO201, JO204 and JO206, <
40% for JW100.We conclude that, in agreement with P17b, the
Nebula-Bayes results confirm that the central spaxels of all galaxiesexcept JO175 are best fitted by AGN models, whose param-eters are given in Table 2. For JO175, nuclear line ratios canbe fitted either by SF or by shocks: HII-like [NII] λ α and [SII] λλ α , but high [OI] λ α , agreewell with shock models (either fast or slow).Finally, four galaxies (JO201, JO204, JO206 andJW100) also show AGN line ratios in the circumnuclear re-gions ( < E peak consistent with an in-creasing contribution from HII ( composite ) regions. In fact,as discussed in Thomas et al. (2018), variations in E peak may be due either to screening by gas and dust (harden-ing the ionizing continuum and thus increasing E peak ), orto contamination from shock or HII regions (softening thecontinuum and thus decreasing E peak ). The high ionization (coronal) line [Fe VII] λ ∼
20, corresponding to [FeVII] λ α ∼ λ λ λ λ < λ ∼ r NLR (FeVII) < r AGN (FeVII) ∼ NebulaBayes grids, since
MAPPINGS does not accuratelymodel these lines (Davies et al. 2016). Different models werepresented to reproduce coronal lines in AGN. Mingozzi et al.(2019) reported the presence of Fe coronal lines in a sam-ple of AGN with outflows, observed with MUSE: they at-tributed them to the inner, optically thin and highly ionizedregions of the outflows. Korista & Ferland (1989) attributedthem to a low-density ( n e ∼ − ) ISM heated by theAGN radiation, in a region whose size is similar or largerthan the NLR ( ∼ T ∼ K) gas heated by the AGN in thehigh-ionization inner regions (see also Mazzalay et al. 2010).Thomas et al. (2017) reported the existence of a correlationbetween [Fe VII] λ α and [OIII] λ β in Seyfertgalaxies in their sample, that they interpreted as an effect ofa radiation pressure dominated environment, where Comp-ton heating in the central regions triggers the production of MNRAS , 1–20 (2019) M. Radovich et al.
Figure 6.
The contour maps showing emission at SN of ∼ λ λ β mapfor the nuclear AGN regions of JO135 ( left ) and JO201 ( right ). The cross displays the position of the peak in [OIII] λ Figure 7. [OIII] λ β maps showing the AGN-like extranuclear emission in JO135 ( left ) and JO204 ( right ). The coordinates arecentered on the position of the central spaxel. coronal lines (Davies et al. 2016). Consistently, both JO201and JO135 show a high [OIII] λ β ( >
10) ratio in theinner regions where the emission of [FeVII] is observed. Theabsence of this line in the other galaxies may be due eitherto orientation effects preventing us to see the inner regions,or to an intrinsic difference in the properties of the ionizedgas.
AGN-like regions are detected in JO135 and JO204 (Gul-lieuszik et al. 2017) up to ∼
20 kpc from the nucleus (Fig. 7),with high values of [OIII]/H β ∼
10 (Fig. 5). This is typicalof the so-called Extended Emission Line Regions (EELR)(see e.g. Yoshida et al. 2004; Maddox 2018), where the gasionized by the AGN extends over scales of tens of kilopar-secs.In these regions the [SII] λ λ ∼ .
4, indicates low values of the electron density ( n e < − ), and the emission line widths are close to the in-strumental value. The line ratios could be reproduced byshock models with n ∼ .
1, but the required shock velocity, V sh >
400 km s − , is too high compared to the observed val-ues: this rules out the presence of fast shocks (see also Fu &Stockton 2009), and favors photoionization from the AGN(Fig. 5).In JO204, the required ionization parameter in the outerregions ( ∼
15 kpc) is close to the value derived in the nu-clear ( r < U ∼ − r − . This is consistentif there is a coupling between the radiation and the illumi-nated gas, as in the case of radiation pressure mechanisms(Thomas et al. 2018). The best-fit AGN models indicate forJO204 log P/k ∼ r ∼ P/k ∼ n H ∼ r − , suggest- MNRAS000
15 kpc) is close to the value derived in the nu-clear ( r < U ∼ − r − . This is consistentif there is a coupling between the radiation and the illumi-nated gas, as in the case of radiation pressure mechanisms(Thomas et al. 2018). The best-fit AGN models indicate forJO204 log P/k ∼ r ∼ P/k ∼ n H ∼ r − , suggest- MNRAS000 , 1–20 (2019)
GN and outflows in jellyfish galaxies ing that the low-density gas in the ISM may be ionized byanisotropic radiation from the AGN. In order to verify if wecan reproduce the observed H β luminosities, we proceed asfollows. For a gaseous cloud, the rate (photons s − ) of ion-izing photons required to produce the observed, dereddenedH β luminosity (erg s − ) is (Osterbrock & Ferland 2006): Q ( H ) = L ( Hβ ) hν Hβ α B ( H , T ) α eff Hβ ( H , T ) ∼ . × L ( Hβ ) (2)where L ( Hβ ) is the observed, dereddened H β lumi-nosity (erg/s); in the Case B approximation α B ( H , T ) =2 . × cm s − , α eff Hβ ( H , T ) = 3 . × cm s − ( T ∼ K).We can thus derive the rate of ionizing photons thatshould be emitted by the nucleus: Q ( H ) nuc = Q ( H ) (cid:18) Ω4 π (cid:19) − (3)where Ω is the solid angle covered by the extra-nuclear re-gion, that is Ω = A/d , for a region of area A and projecteddistance d from the nucleus. From the Hβ fluxes measuredin the EENLR of JO204 at 15 kpc, we obtain Q ( H ) nuc ∼ ph/s: using this value to compute the ionization param-eter, we obtain log U ∼ − n H ∼ cm − , in agreementwith the value expected from photoionization models. Sim-ilar results are obtained for JO135. As discussed before, in four galaxies (JO201, JO204, JW100and JO135) emission lines in the circumnuclear regions arecharacterized by complex profiles, that require at least twoGaussian components to be fitted. In order to disentanglethe contribution to the emission lines from gas in disk andother components (e.g. outflows), we focus on the [OIII] λ α +[NII] lineset used in P17b,[OIII] is more suitable for this analysis as it is not affectedby the presence of other nearby emission lines (the [NII]doublet in the case of Hα ) or by possible residual broadcomponents in the inner AGN regions, as it may be the casefor permitted lines, and better traces the ionized gas in theoutflows (see e.g. Bae & Woo 2014, and refs.). To this end,we selected a region of ∼ (cid:48) x 10 (cid:48) around the center of eachgalaxy and fitted the [OIII] λ lmfit . For each spaxel, wemade the fit adopting both one and two Gaussian compo-nents. The two component solution was chosen if it gavean appreciable improvement to the fit compared to the onecomponent solution: based on visual inspection, we definedthis condition as χ n =1 > χ n =2 , χ being the chi-squareof the fit (see Davis et al. 2012, for a similar approach), andin addition we requested that the flux in each componentmust be at least 10% of the summed flux. We also discardedthose fits where S/N([OIII]) <
5, where the signal S is thetotal line flux and the noise N is the standard deviation ofthe fitting residuals. We neglect the factor due to the unknown projection, but herewe are interested in order of magnitudes.
From the velocities measured at 10%, 50% and 90% ofthe cumulative flux percentiles, we used the definitions inLiu et al. (2013) to estimate the parameters introduced byWhittle (1985): the peak velocity of the [OIII] line ( v pk ), themedian velocity ( v ), the width W = v − v and theasymmetry A sym = ( v − v ) − ( v − v ) W . In this definition,positive/negative values of A sym indicate red/blue asymmet-ric lines. We used the fit to model the line profile and com-pute these parameters. In this way we do not assign anyphysical meaning to the decomposition, using the fit only toreduce the effect of the noise on the estimate of the profileparameters (Liu et al. 2013; Harrison et al. 2014; Balmaverdeet al. 2016).The spatial distribution of these parameters for thegalaxies showing two emission line components (JO135,JO201, JO204 and JW100) is displayed in Fig. 8, where thevelocity measured for stars is also displayed as reference.For a more quantitative analysis, we then analyzed(Fig. 9) the velocity and velocity dispersion of the two fittedcomponents, bearing in mind that the fit may be degenerate,in particular when the components are close, or in the outerregions where the line is fainter and noise may introducespurious features.In the presence of an outflow we expect to see a clearseparation in both velocity and velocity dispersion between a(primary) component, kinematically dominated by the grav-itational potential traced by the stellar component, and asecondary (slightly broader) component with a velocity off-set showing its non-gravitational origin (Woo et al. 2016;Karouzos et al. 2016a). To check if this is the case, Fig. 10shows the radially binned values of the deviations from thestellar values of the velocity and velocity dispersion.As discussed in Crenshaw et al. (2010), the observedline profiles can be explained by a combination of biconi-cal outflows and extinction from dust in the inner galaxydisk. Biconical outflows are most often observed as broad,blueshifted components in the emission lines as the red-shifted, receding part of the outflow more likely lies behindthe galaxy disk and is thus suppressed by dust. In somecases, instead, red asymmetric lines are observed: this canstill happen when the inclination of the disk is such to hidethe approaching part of the outflow. In addition, Lena et al.(2015, see their Fig. 15) proposed a model where dust isembedded in the outflowing clouds: in this case, blueshiftedclouds at small distances from the center preferentially showtheir non ionized face and are thus fainter compared to red-shifted clouds, showing instead their ionized face. At largeprojected distances, an increasing fraction of the ionized faceis visible in the blueshifted clouds, which are then brighter.Fig. 8, Fig. 9 and Fig. 10 demonstrate that the fourgalaxies with double components have an outflow, and thattheir outflow properties are quite diverse, as we discuss inthe following. JO135 : within a radius ∼ λ W ∼ − ), which isasymmetric in the red (Fig. 8): in the same region, there isan increased dust extinction ( A v ∼ MNRAS , 1–20 (2019) M. Radovich et al.
JO135JO201JO204JW100
Figure 8.
For each galaxy hosting an AGN, the plots display the spatial distribution of the stellar velocity [km s − ] ( first panel ) and ofthe following parameters (see text for details) derived from the [OIII] line: peak and median velocity [km s − ]; W [km s − ]; asymmetry( right panel ). In each plot, North is up and East is at left.id r r log E kin ¯ v out log t out log M out log ˙ M out log ˙ E kin log L AGN kpc kpc erg km s − yr M (cid:12) M (cid:12) yr − erg s − erg s − JW100 1.1 2.1 50.5 248 6.9 4.40 -2.53 36.0 43.9JO201 1.1 2.4 53.7 261 6.9 5.99 -0.95 39.2 45.4JO204 0.7 1.2 52.2 317 6.6 4.27 -2.31 38.1 43.8JO135 0.9 1.5 53.7 544 6.4 5.27 -1.16 39.7 44.8
Table 3.
Outflow properties derived as described in the text. to the model proposed by Lena et al. (2015), to explain thedominant redshifted component in the outflow. Outside thisregion, the velocity pattern follows the stellar one. This canalso be seen in the radial plots of the fitted components(Fig. 10), where at r > W ∼
600 km s − ) is de- tected in the inner kpc, with a small blue asymmetry. Inthe same region, we measure an increase of the dust extinc-tion ( A v ∼ n e ∼ cm − ). The velocity and velocity dispersion of the narrowcomponent agree with the stellar values: we therefore iden-tify the narrow component with gas in the disk and the broadblueshifted component with the outflow. The low asymme-try in [OIII] may imply a spherical or wide-angle outflow MNRAS000
600 km s − ) is de- tected in the inner kpc, with a small blue asymmetry. Inthe same region, we measure an increase of the dust extinc-tion ( A v ∼ n e ∼ cm − ). The velocity and velocity dispersion of the narrowcomponent agree with the stellar values: we therefore iden-tify the narrow component with gas in the disk and the broadblueshifted component with the outflow. The low asymme-try in [OIII] may imply a spherical or wide-angle outflow MNRAS000 , 1–20 (2019)
GN and outflows in jellyfish galaxies JO135JO201JO204JW100
Figure 9.
Spatial distribution of velocity and velocity dispersion from [OIII] λ (Liu et al. 2013), allowing to see the inner outflow regionsand hence the [FeVII] λ ∼ W ( ∼
900 km s − ). Out-side this strip, we identify two regions, one (NW) with blueasymmetric lines and one (SE) with red asymmetric lines.From the two-component fits, we find a narrow component,where σ v is close to the stellar values, and a significantlybroader, but fainter, one ( σ v ≤
500 km s − ). Since up toa distance of 1 kpc both components have a velocity thatis significantly different than the stellar value, we interpretboth of them as produced by the two sides of a biconicaloutflow; at larger distances ( > σ v <
200 km s − ), blue and redshifted, with avelocity offset compared to the stellar velocity of v ∼ ± − up to a radius r ∼ ∼ From the secondary (broader) components, we computedthe radius containing a fraction f (f = 50% and 90%)of the total broad [OIII] flux: (cid:80) r
1. A typical density (cid:104) n e (cid:105) = 500 cm − wasassumed. The outflow kinetic energy is E kin = M out v , v out being the bulk velocity of the outflow. As discussed e.g.in Karouzos et al. (2016b), different choices to estimate v out were made in the literature, reflecting different strategies totake into account geometrical effects (e.g. projection, open-ing angle of the outflow). Here we adopt the approximation(Karouzos et al. 2016b; Harrison et al. 2018) v = v + σ ,where v rad is the measured radial velocity and σ the [OIII]velocity dispersion, corrected (Karouzos et al. 2016b) for thecontribution from the gravitational potential by subtract-ing in quadrature the stellar velocity dispersion. The massoutflow was derived summing on all spaxels within r the MNRAS , 1–20 (2019) M. Radovich et al.
JO135 JO201JO204 JW100
Figure 10.
Velocity and velocity dispersion of the narrow (green) and broad (red) components of [OIII] λ σ v is the velocity dispersion from which the stellar velocity dispersion was subtracted in quadrature.For the velocity dispersion, a value of 0 was assigned if the velocity dispersion was lower than the stellar velocity dispersion. contributes from both line components since, with the possi-ble exception of JO201, we were not able to unambiguouslyseparate the disk and outflow contributions to [OIII].From the mean bulk velocity and the size of the outflowwe can then compute the outflow lifetime, t out = r out / ¯ v out ,the outflow mass rate, ˙ M out = M out /t out , the energetic rate,˙ E kin = E kin /t out and the outflow efficiency, η = ˙ E kin /L AGN where L AGN is the bolometric luminosity. In Karouzos et al.(2016b) the bolometric AGN luminosity was computed as L AGN = 3500 L [OIII] erg/s (Heckman et al. 2004), L [OIII] being the dust uncorrected luminosity, for comparison withother literature samples. We adopted the same choice andused the [OIII] luminosities in Table 1, that include bothemission line components.The results obtained from the above analysis are dis-played in Table 3. We emphasize all the uncertainties relatedboth to the outflow kinetic energy and the estimate of thebolometric luminosity. Nevertheless, the outflow mass ratesand kinetic energies that we obtain are comparable with thevalues obtained by Karouzos et al. (2016b) in a sample of AGN having similar [OIII] luminosities ( L [OIII] < ergs − ). Consistently with these results, we derive low efficien-cies, η (cid:28) . t out ∼ yr and the outflow velocity is v out ∼
300 km s − : this corresponds to a distance of ∼ r in Table 3). Evi-dence for star formation suppression around the AGN inJO201, based on NUV and CO data, will be presented inGeorge et al. (2019, submitted). For comparison, from lit-erature the mass outflow and energy rates in the brightestAGN ( L [OIII] > erg s − ) can be as high as 10 M (cid:12) yr − and log ˙ E kin ∼ erg s − (Liu et al. 2013), thus be-ing able to impact on much larger scales in the host galaxyenvironment. MNRAS000
300 km s − : this corresponds to a distance of ∼ r in Table 3). Evi-dence for star formation suppression around the AGN inJO201, based on NUV and CO data, will be presented inGeorge et al. (2019, submitted). For comparison, from lit-erature the mass outflow and energy rates in the brightestAGN ( L [OIII] > erg s − ) can be as high as 10 M (cid:12) yr − and log ˙ E kin ∼ erg s − (Liu et al. 2013), thus be-ing able to impact on much larger scales in the host galaxyenvironment. MNRAS000 , 1–20 (2019)
GN and outflows in jellyfish galaxies In this paper we have carried out a detailed investigation ofthe seven jellyfish galaxies presented in P17b, where basedon the [NII]/H α ratio it was found that at least five of them(JO201, JO204, JO206, JO135 and JW100) host an AGN.We first performed a detailed comparison with photoioniza-tion and shock model taking into account several diagnos-tic diagrams simultaneously. We concluded that while shockmodels can play a role in the ionization of the gas, AGNmodels are required to explain the line ratios observed inthe nuclear regions of these five galaxies: this conclusion iscorroborated by an analysis of the H β luminosity. The pres-ence of iron coronal lines in the nuclei of JO201 and JO135indicates the existence of hot ( T ∼ K) gas heated bythe AGN. JO204 and JO135 also present Extended Emis-sion Line Regions of >
10 kpc that are ionized by the AGN.In JO194, that was classified as a LINER in P17b, line ra-tios in the central spaxels are better reproduced by an AGNmodel, though shock models may also marginally reproducethe observed ratios. Finally, in JO175 the [NII] λ α ratio is typical of star forming regions, but we still observea high [OI] λ α ratio that could point to the presenceof shocks.We then focused on the [OIII] λ L [OIII] < erg s − ), with outflows.Finally, we derived conclusions on possible AGN feedbackeffects on the circumnuclear regions ( ∼ ACKNOWLEDGEMENTS
This work is based on observations collected at the Euro-pean Organisation for Astronomical Research in the South-ern Hemisphere under ESO program 196.B-0578. We ac-knowledge financial support from PRIN-SKA ESKAPE-HI (PI L.Hunt). Y.J. acknowledges financial support fromCONICYT PAI (Concurso Nacional de Inserci´on en laAcademia 2017) No. 79170132 and FONDECYT Iniciaci´on2018 No. 11180558. We acknowledge the usage of thefollowing Python libraries:
AstroPy , lmfit , NebulaBayes , lineid_plot , mpdaf . REFERENCES
Allen M. G., Groves B. A., Dopita M. A., Sutherland R. S., Kew-ley L. J., 2008, ApJS, 178, 20Arnold T. J., Martini P., Mulchaey J. S., Berti A., Jeltema T. E.,2009, ApJ, 707, 1691Bacon R., et al., 2010, in Ground-based and Airborne Instrumen-tation for Astronomy III. p. 773508, doi:10.1117/12.856027Bae H.-J., Woo J.-H., 2014, ApJ, 795, 30Bae H.-J., Woo J.-H., Karouzos M., Gallo E., Flohic H., Shen Y.,Yoon S.-J., 2017, ApJ, 837, 91Balmaverde B., et al., 2016, A&A, 585, A148Belfiore F., et al., 2016, MNRAS, 461, 3111Bellhouse C., et al., 2017, ApJ, 844, 49Bing L., et al., 2019, MNRAS, 482, 194Cappellari M., Emsellem E., 2004, Publications of the Astronom-ical Society of the Pacific, 116, 138Carniani S., et al., 2015, A&A, 580, A102Crenshaw D. M., Schmitt H. R., Kraemer S. B., Mushotzky R. F.,Dunn J. P., 2010, ApJ, 708, 419Davies R. L., et al., 2016, ApJ, 824, 50Davis T. A., et al., 2012, MNRAS, 426, 1574Dopita M. A., Sutherland R. S., 1995, ApJ, 455, 468Dressler A., Thompson I. B., Shectman S. A., 1985, ApJ, 288,481Ebeling H., Stephenson L. N., Edge A. C., 2014, ApJ, 781, L40Ehlert S., et al., 2014, MNRAS, 437, 1942Fabian A. C., 2012, Annual Review of Astronomy and Astro-physics, 50, 455Ferguson J. W., Korista K. T., Baldwin J. A., Ferland G. J., 1997,ApJ, 487, 122Fossati M., Fumagalli M., Boselli A., Gavazzi G., Sun M., WilmanD. J., 2016, MNRAS, 455, 2028Fritz J., et al., 2017, ApJ, 848, 132Fu H., Stockton A., 2009, ApJ, 690, 953Fumagalli M., Fossati M., Hau G. K. T., Gavazzi G., Bower R.,Sun M., Boselli A., 2014, MNRAS, 445, 4335George K., et al., 2018, MNRAS, 479, 4126Gitti M., 2013, MNRAS, 436, L84Gullieuszik M., et al., 2017, ApJ, 846, 27Harrison C. M., Alexander D. M., Mullaney J. R., SwinbankA. M., 2014, MNRAS, 441, 3306Harrison C. M., Costa T., Tadhunter C. N., Fl¨utsch A., KakkadD., Perna M., Vietri G., 2018, Nature Astronomy, 2, 198Heckman T. M., Best P. N., 2014, Annual Review of Astronomyand Astrophysics, 52, 589Heckman T. M., Kauffmann G., Brinchmann J., Charlot S.,Tremonti C., White S. D. M., 2004, ApJ, 613, 109Ilha G. S., et al., 2019, MNRAS, 484, 252Jaff´e Y. L., et al., 2018, MNRAS, 476, 4753Kakkad D., et al., 2018, A&A, 618, A6Karouzos M., Woo J.-H., Bae H.-J., 2016a, ApJ, 819, 148Karouzos M., Woo J.-H., Bae H.-J., 2016b, ApJ, 833, 171Kewley L. J., Groves B., Kauffmann G., Heckman T., 2006, MN-RAS, 372, 961Komossa S., Schulz H., 1997, A&A, 323, 31Korista K. T., Ferland G. J., 1989, ApJ, 343, 678Krumholz M. R., Fumagalli M., da Silva R. L., Rendahl T., ParraJ., 2015, MNRAS, 452, 1447Lena D., et al., 2015, ApJ, 806, 84Liu G., Zakamska N. L., Greene J. E., Nesvadba N. P. H., Liu X.,2013, MNRAS, 436, 2576Maddox N., 2018, MNRAS, 480, 5203Marshall M. A., Shabala S. S., Krause M. G. H., Pimbblet K. A.,Croton D. J., Owers M. S., 2018, MNRAS, 474, 3615Marziani P., et al., 2017, A&A, 599, A83Mazzalay X., Rodr´ıguez-Ardila A., Komossa S., 2010, MNRAS,405, 1315MNRAS , 1–20 (2019) M. Radovich et al.
Mingozzi M., et al., 2019, A&A, 622, A146Nevin R., Comerford J., M¨uller-S´anchez F., Barrows R., CooperM., 2016, ApJ, 832, 67Osterbrock D. E., Ferland G. J., 2006, Astrophysics of gaseousnebulae and active galactic nucleiPimbblet K. A., Shabala S. S., Haines C. P., Fraser- McKelvie A.,Floyd D. J. E., 2013, MNRAS, 429, 1827Poggianti B. M., et al., 2017a, ApJ, 844, 48Poggianti B. M., et al., 2017b, Nature, 548, 304Poggianti B. M., et al., 2019, MNRAS, 482, 4466Proxauf B., ¨Ottl S., Kimeswenger S., 2014, A&A, 561, A10Ramos-Mart´ınez M., G´omez G. C., P´erez-Villegas ´A., 2018, MN-RAS, 476, 3781Rawle T. D., et al., 2014, MNRAS, 442, 196Smith R. J., et al., 2010, MNRAS, 408, 1417Thomas A. D., Groves B. A., Sutherland R. S., Dopita M. A.,Kewley L. J., Jin C., 2016, ApJ, 833, 266Thomas A. D., et al., 2017, ApJS, 232, 11Thomas A. D., Dopita M. A., Kewley L. J., Groves B. A., Suther-land R. S., Hopkins A. M., Blanc G. A., 2018, ApJ, 856, 89Vazdekis A., S´anchez-Bl´azquez P., Falc´on-Barroso J., CenarroA. J., Beasley M. A., Cardiel N., Gorgas J., Peletier R. F.,2010, MNRAS, 404, 1639Veilleux S., Osterbrock D. E., 1987, The Astrophysical JournalSupplement Series, 63, 295Venturi G., et al., 2018, A&A, 619, A74Whittle M., 1985, MNRAS, 213, 1Wong K.-W., Sarazin C. L., Blanton E. L., Reiprich T. H., 2008,ApJ, 682, 155Woo J.-H., Bae H.-J., Son D., Karouzos M., 2016, ApJ, 817, 108Yoshida M., et al., 2004, AJ, 127, 90This paper has been typeset from a TEX/L A TEX file prepared bythe author. MNRAS000