Not So Windy After All: MUSE Disentangles AGN-driven Winds from Merger-induced Flows in Galaxies along the Starburst Sequence

Poststarburst galaxies are believed to be in a rapid transition between major merger starbursts and quiescent ellipticals, where active galactic nucleus (AGN) feedback is suggested as one of the processes responsible for the quenching. To study the role of AGN feedback, we constructed a sample of poststarburst candidates with AGN and indications of ionized outflows in optical. We use MUSE/VLT observations to spatially resolve the properties of the stars and multiphase gas in five of them. All galaxies show signatures of interaction/merger in their stellar or gas properties, with some at an early stage of interaction with companions ∼50 kpc, suggesting that optical poststarburst signatures may be present well before the final starburst and coalescence. We detect narrow and broad kinematic components in multiple transitions in all the galaxies. Our detailed analysis of their kinematics and morphology suggests that, contrary to our expectation, the properties of the broad kinematic components are inconsistent with AGN-driven winds in three out of five galaxies. The two exceptions are also the only galaxies in which spatially resolved NaID P-Cygni profiles are detected. In some cases, the observations are more consistent with interaction-induced galactic-scale flows, an often overlooked process. These observations raise the question of how to interpret broad kinematic components in interacting and perhaps also in active galaxies, in particular when spatially resolved observations are not available or cannot rule out merger-induced galactic-scale motions. We suggest that NaID P-Cygni profiles are more effective outflow tracers, and use them to estimate the energy that is carried by the outflow.


Introduction
Active galactic nucleus (AGN) feedback is now a key process in models of galaxy evolution.It is invoked to explain different properties of local massive galaxies, including their stellar mass function, in particular its higher mass end, the observed correlation between their stellar velocity dispersion and the mass of their supermassive black hole, and the enrichment of their circumgalactic medium (e.g., Silk & Rees 1998;Fabian 1999;King 2003;Springel et al. 2005a;Kormendy & Ho 2013;Nelson et al. 2019).AGN feedback couples the energy that is released by the accreting black hole with the gas in the interstellar medium (ISM) of the host galaxy.
The two main AGN feedback modes, which are related to the type of nuclear activity, are (i) the radiative/quasar mode, which is associated with black holes accreting close to the Eddington limit, and (ii) the kinetic/jet mode, which is associated with lower-power AGN with Eddington ratios of 0.1 (e.g., Croton et al. 2006;Alexander & Hickox 2012;Fabian 2012;Harrison et al. 2018).In the radiative feedback mode, the quasar is believed to drive gas outflows that can reach galactic scales, destroy, push, or even compress molecular clouds, and escape the host galaxy (e.g., Faucher-Giguère & Quataert 2012; Zubovas & King 2012;Zubovas & Nayakshin 2014).While some simulations suggest that these outflows can have a dramatic impact on their host, by violently quenching its star formation and transforming it into a red and dead elliptical (e.g., Springel et al. 2005aSpringel et al. , 2005b;;Di Matteo et al. 2005;Hopkins et al. 2006), others suggest a more limited impact on the ISM, with these outflows escaping along the path of least resistance (e.g., Gabor & Bournaud 2014;Hartwig et al. 2018).At the current stage, cosmological hydrodynamical simulations of galaxy formation can reproduce the observed properties of local massive galaxies only when AGN feedback is invoked, and only if AGN-driven winds carry a significant fraction of the AGN energy (1%-10% of L AGN ; e.g., Crain et al. 2015;Pillepich et al. 2018;Davé et al. 2019;Nelson et al. 2019;Villaescusa-Navarro et al. 2021).
During the past decade, studies used large and public surveys (e.g., SDSS: York et al. 2000;SAMI: Croom et al. 2021, andMANGA: Bundy et al. 2015), along with dedicated deeper multiwavelength observations to study the occurrence rate, extent, kinematics, excitation/ionization, and density of multiphase outflows in typical AGN hosts in the local Universe.Using these properties, studies estimated the energy that is carried out by the multiphased winds, typically finding values that are several orders of magnitude lower than the theoretical coupling efficiency of 1%-10% (e.g., Villar-Martín et al. 2016;Fiore et al. 2017;Rupke et al. 2017;Baron & Netzer 2019;Fluetsch et al. 2019Fluetsch et al. , 2021;;Lutz et al. 2020;Shangguan et al. 2020;Revalski et al. 2021;Ruschel-Dutra et al. 2021;Baron et al. 2022).It is unclear whether the observations are in tension with the theoretical requirement, as simulations require a coupling efficiency of 1%-10% at the wind injection scale (<< kpc), while the observed winds are typically detected on kpc scales.According to one scenario, the wind evolves from injection scale to galactic scales, where it is shock-heated when encountering the host ISM, creating a bubble of extremely hot and ionized gas that dominates the energetics of the flow (e.g., Faucher-Giguère & Quataert 2012; Richings & Faucher-Giguère 2018).In such a case, observations that trace T < 10 5 K gas will underestimate the total energy that is carried out by the wind.
At higher redshifts of z ∼ 2, the mass and energetics of outflows in AGN hosts are still highly uncertain.Earlier studies focused on the most extreme objects or outflow cases and were limited to small sample sizes and/or spatially integrated observations.(e.g., Nesvadba et al. 2008;Cano-Díaz et al. 2012;Brusa et al. 2015;Perna et al. 2015).Later works studied larger samples, including more typical high-redshift galaxies, using integral field unit (IFU) observations with adaptive optics in some cases (e.g., Newman et al. 2012;Genzel et al. 2014;Harrison et al. 2016;Förster Schreiber et al. 2018, 2019;Leung et al. 2019;Kakkad et al. 2020; see review by Förster Schreiber & Wuyts 2020).Despite these advances, the resulting sensitivity and spatial resolution do not allow for the detection and resolving of the weaker transitions of the outflow, making its reddening, ionization state, and density, still uncertain.As a result, the energy that is carried out by outflows at z ∼ 2 remains largely unconstrained.This is expected to change in the near future, with first observations by JWST/NIRSpec detecting and resolving z ∼ 2-4 outflows even in weak transitions such as [S II] and Hβ emission lines and NaID absorption (e.g., Rupke et al. 2023;Veilleux et al. 2023;Davies et al. 2024;Wang et al. 2024).
Other possible sites for significant AGN feedback are galaxies on the merger sequence.During this short phase, the galaxies reach a peak in star formation rate (SFR) and black hole accretion rate (e.g., Sanders et al. 1988;Sanders & Mirabel 1996), leading to powerful supernova-(SN) and AGNdriven winds (e.g., Springel et al. 2005a;Hopkins et al. 2006).
A particular short phase within this sequence is the poststarburst phase (see review by French 2021).Poststarburst galaxies are believed to be galaxies in a rapid transition from starburst to quiescence, with optical spectra that are dominated by A-type stars, suggesting a significant burst of star formation that ended abruptly <1 Gyr ago (Dressler et al. 1999;Poggianti et al. 1999;Dressler et al. 2004;Goto 2004).Some of these systems are bulge dominated with tidal features, suggesting that they are merger remnants (Canalizo et al. 2000;Goto 2004;Yang et al. 2004;Cales et al. 2011).Stellar population synthesis modeling of their optical spectra suggests high peak SFRs, ranging from 50 to 300 M e yr −1 (Kaviraj et al. 2007), with estimated mass fractions forming in the burst of 10%-80% of the total stellar mass (Liu & Green 1996;Norton et al. 2001;Yang et al. 2004;Kaviraj et al. 2007;French et al. 2018).
Although AGN feedback is believed to be one of the processes responsible for the sudden quenching of star formation in poststarbursts, little is known observationally about AGN-driven winds in this short-lived phase.Highvelocity gas flows have been detected in poststarburst galaxies (e.g., Tremonti et al. 2007;Coil et al. 2011;Maltby et al. 2019), though in some cases were later attributed to winds driven by obscured starbursts in the systems (e.g., Diamond-Stanic et al. 2012).To study the properties of AGN-driven winds in poststarburst galaxies, in Baron et al. (2022) we constructed a sample of local poststarburst candidates with AGN and evidence for ionized outflows.In Baron et al. (2018) and Baron et al. (2020) we used optical IFUs to spatially resolve the multiphased gas in two such galaxies, finding massive ionized +neutral outflows with kinetic powers that are 10-100 times larger than those observed in typical active galaxies in the local Universe.We therefore suggested that AGN feedback, in the form of galactic-scale outflows, may be significant in the poststarburst phase.
In this paper we present follow-up MUSE/VLT observations of five additional poststarburst galaxies with AGN and ionized outflows.Together with the two already-published poststarbursts (Baron et al. 2018(Baron et al. , 2020)), we perform a detailed analysis of their stellar population and multiphased gas, paying particular attention to the detection and characterization of galactic-scale flows.The paper is organized as follows.In Section 2 we describe our methods, including the sample selection (2.1), MUSE observations (2.2), stellar population analysis (2.3), ionized (2.4) and neutral (2.5) gas analysis, collection of ancillary properties (2.6), and characterization of outflows (2.7).In Section 3 we describe our results and discuss their broader context in Section 4. We conclude and summarize in Section 5. Readers who are interested only in the results may skip directly to Section 3. Throughout this paper we use a Chabrier initial mass function (IMF; Chabrier 2003) and assume a standard ΛCDM cosmology with Ω M = 0.3, Ω Λ = 0.7, and h = 0.7.

Sample Selection
Our sample was selected from our parent sample of poststarburst galaxy candidates with AGN and ionized outflows, and is described in detail in Baron et al. (2022).For completion, we give a brief overview of its main properties.The parent sample was drawn from the 14th data release of the Sloan Digital Sky Survey (SDSS; York et al. 2000).To select poststarburst systems, we selected galaxies with strong Hδ absorption lines, in particular requiring EW(Hδ) > 5Å, where EW is the absorption equivalent width.To select poststarbursts with AGN, we performed emission line decomposition on the stellar continuum-subtracted spectra, fitting narrow and broad kinematic components to the emission lines: Hα, Hβ, [O III], [N II], [S II], and [O I].We selected systems with narrow-line ratios that are consistent with AGN photoionization using standard line diagnostic diagrams (including LINERs; Kewley et al. 2001;Cid Fernandes et al. 2010).We used the broad Balmer emission lines to filter out type I AGN.To select poststarbursts with AGN and ionized outflows, we selected systems in which broad kinematic components are detected in both the Balmer and forbidden lines.We found a total of 215 poststarburst candidates with AGN and ionized outflows, out of which 144 show evidence of an ionized outflow in multiple lines.
We selected a subset of 32 systems for follow-up observations.These galaxies show the highest S/Ns in the broad kinematic components of the [O III] and Hα emission lines.Such a selection is biased toward galaxies with more luminous emission lines, and thus favors systems with higher AGN luminosity and SFR.In addition, as we show throughout the paper, it may also favor tidally interacting systems with significant interaction-induced flows.We followed up a subset of 15 galaxies with NOEMA and reported the result in Baron et al. (2023).One of the galaxies was observed with KCWI/ Keck and results were reported in Baron et al. (2018).Another galaxy was observed with MUSE/VLT and results were reported in Baron et al. (2020).In this work we present MUSE/ VLT observations of five additional galaxies from this subset, four of which are part of the NOEMA sample from Baron et al. (2023).

MUSE Observations
MUSE is a second-generation integral field spectrograph on the VLT (Bacon et al. 2010).It consists of 24 integral field units that cover the wavelength range 4650-9300 Å, achieving a spectral resolution of 1750 at 4650 Å to 3750 at 9300 Å.In its Wide Field Mode (WFM), MUSE splits the 1 × 1 arcmin 2 field of view (FOV) into 24 channels which are further sliced into 48 15″ × 0 2 slitlets.Since period 101, MUSE has used the GALACSI adaptive optics (AO) module to offer AO-corrected WFM, WFM-AO.
The five galaxies were observed as part of our program "Mapping AGN-driven outflows in quiescent poststarburst E +A galaxies" (0100.B-0517(A) and 0102.B-0228(A)).For the earlier program, 0100.B-0517(A), observations were carried out in a seeing-limited WFM with a pixel scale of 0 2 and a spatial resolution of ∼0 8.For the later program 0102.B-0228 (A), observations were carried out with WFM-AO, reaching a spatial resolution of 0 4. We downloaded the data from the ESO phase 3 online interface, which provides fully reduced, calibrated, and combined MUSE data for all targets with multiple observing blocks.Table 1 lists the galaxies' properties and follow-up observations, including the instruments, modes, and exposure times for each source.In Figure 1 we show the stellar continuum and ionized gas emission for the five galaxies in our sample.

Stellar Properties
To accurately determine the stellar kinematics and stellar population properties from full spectral fitting, a minimum S/N is required (Johansson et al. 2012;Westfall et al. 2019).We therefore binned the spaxels using VORBIN, which is a broadly used package to bin data cubes along the two spatial dimensions using Voronoi tessellations (Cappellari & Copin 2003).The method uses an iterative process to find the optimal binning solution, given the S/N of each individual spaxel and the target S/N as inputs.We concentrated on the wavelength range 5300-5700 Å, which is free from strong emission or absorption lines.We defined the S/N of an individual spaxel to be S/σ, where S is the sum of the flux in this wavelength range, and σ is the square root of the sum of squared residuals of a 2°polynomial fit in this range.The fit to a 2°polynomial is required to account for large-scale continuum variations between 5300 and 5700 Å.
Prior to applying VORBIN to the data cubes, we manually inspected the cubes and identified all the objects that were not associated with the primary galaxy (i.e., high-z galaxies or stars in the field), and masked them out.Galaxies that show a similar redshift to the primary galaxy, or large-scale tidal tails, were not masked out.We then manually inspected the spectra of spaxels with different S/Ns of the collapsed cubes and found that spaxels with S/N <10 generally do not show any signal that can be associated with the galaxy, and are instead dominated by sky lines.We therefore masked these spaxels out.As shown in Figure 2, this step only masks out spaxels that are far away (30 kpc) from the galaxy (or pair).We then applied VORBIN to the masked data cubes, experimenting with different target S/Ns.We found that requiring a target S/N of 250 of the collapsed cubes allows us to keep high spatial resolution in the centers of the galaxies, while also allowing us to trace the stellar properties accurately (best-fitting stellar parameters are estimated with >3σ) in their outskirts.The resulting stellar properties do not change significantly for target S/Ns between 150 and 350.Table 4 in Appendix A summarizes the binning parameters and number of bins per galaxy.The top row of Figure 2 illustrates the binning process for J022912.The figure shows that the Voronoi bins include only single spaxels in the center of the galaxy, while including tens of spaxels in the outskirts.
To estimate the stellar kinematics and population properties, we used PPXF (Cappellari 2012) to fit stellar population synthesis models to the binned spectra.PPXF is a widely used code for the extraction of stellar kinematics and stellar population from absorption line spectra of galaxies (Cappellari & Emsellem 2004).We used the MILES library, which contains single stellar population synthesis models that cover the entire optical wavelength range with an FWHM resolution of 2.3 Å (Vazdekis et al. 2010).We used models produced with the Padova 2000 stellar isochrones assuming a Chabrier IMF (Chabrier 2003).The stellar ages of the MILES models range between 0.03 and 14 Gyr.Since we are also interested in the dust reddening toward the stars, we applied PPXF without any additive or multiplicative Legendre polynomials.The output of PPXF includes the best-fitting stellar model and χ 2 , the contribution of stars with different ages (i.e., a nonparametric star formation history), the stellar kinematics, and the reddening toward the stars, assuming a Calzetti et al. (2000) extinction law.

Ionized Gas
To study the properties of the ionized gas, we first need to fit and subtract the stellar continuum emission.The Voronoi bins obtained in Section 2.3 for the stellar population are optimized for stellar continuum emission rather than for line emission.For example, Figure 1 shows significant line emission in regions where no stellar continuum emission is detected for J022912 and J111943.Such spaxels are masked out in Section 2.3 prior to applying the Voronoi binning and thus will be missed.To achieve optimal spatial resolution and line S/N, we ran VORBIN again, using a different definition for spaxel S/N as described below.We concentrated on the wavelength range 6500-6640 Å, which includes the Hα and [N II] lines.We defined the "line region" to be 6520-6620 Å and the "continuum region" to be 6500-6520 Å and 6620-6640 Å.We fitted the continuum region with a 1°polynomial and subtracted it from the line region spectrum.The signal, S, is then the sum over the continuum-subtracted line region, and the noise, σ, is the square root of the sum of squared residuals, multiplied by a factor N line /N cont , where N line is the length of the line region and N cont is the length of the continuum region.Therefore, this S/N definition is proportional to the integrated Hα+[N II] line flux.
Prior to applying VORBIN, we masked out spaxels that were not spatially associated with the galaxy (or galaxy pair; see all the scattered orange pixels throughout the FOV in Figure 1).We experimented with different minimum S/Ns of the integrated flux, below which spaxels were masked out, and target S/Ns.The selected minimum and target S/Ns are different for the different galaxies and are listed in Table 4 in Appendix A. The minimum S/Ns are not significantly different for different galaxies, and range between 5 and 8. Changing the minimum S/N mostly affects the number of spaxels included within bins in the outskirts of the galaxies, where tens of spaxels are binned together.As for the target S/N, there is a tradeoff between setting a lower and a higher value.For a higher value, the resulting line S/N is higher, allowing the detection of broad kinematic components of weak lines such as Hβ and [S II].However, a higher value also results in larger bins, often mixing separate gas kinematic components into a single bin.The latter, for example, results in double or triplepeaked line profiles which are more challenging to interpret.Our selected target S/Ns, which range between 30 and 100, were chosen as the maximal S/Ns for which different kinematic components remain not blended.Therefore, most of the binned spectra exhibit either one (narrow) or two (narrow + broad) kinematic components. 5The bottom row of Figure 2 illustrates the binning process for J022912, and shows that the bins include single spaxels in the galaxy center and tens of spaxels in the outskirts.
We then used PPXF to fit and subtract the stellar continuum from the binned spectra.Our emission line decomposition follows the method outlined in Baron et al. (2020), which is briefly summarized below.We fit the emission lines Hβ, Hα, and [S II] in each binned spectrum.The amplitude ratios of the [N II] and [O III] doublet lines are set to their theoretical values.The [S II]λ6717 Å/[S II]λ6731 Å intensity ratio is allowed to vary between 0.44 and 1.44.Similarly to the poststarburst studied in Baron et al. (2020), the galaxies in our sample are dusty, and thus their Hα+[N II] lines are significantly stronger than their Hβ+[O III].We therefore started by fitting the Hα and [N II] lines.We modeled each of the emission lines using one or two Gaussians, where the first represents the narrow kinematic component and the second the broader kinematic component. 7We tied the central wavelengths and widths of the narrow Gaussians to have the same velocity and velocity dispersion and did the same for the broader Gaussians.The broad kinematic component was kept if the reduced χ 2 was improved compared to a fit with only a narrow kinematic component, and only if its flux in Hα and [N II]6584 Å was detected to more than 3σ.
Once we obtained a fit for the Hα+[N II] complex, we used the best-fitting kinematic parameters to fit the other ionized emission lines.We fitted Hβ, [O III], [O I], and [S II] with narrow and broad (if existing in the Hα+[N II] fit) kinematic components, locking their central wavelengths and line widths to the best-fitting values obtained for the Hα+[N II].A broad kinematic component was considered detected only if the amplitude of the broad Gaussian was larger than 3 times its uncertainty.Otherwise, we ran the fit again with a narrow kinematic component only.
To estimate the line fluxes, we integrated over the best-fitting profiles.We assume that the narrow kinematic component originates from nonoutflowing gas in these galaxies.We therefore integrated over the best-fitting narrow profiles to obtain the line fluxes associated with the nonoutflowing gas.As discussed in Section 3.1, most of the galaxies in our sample exhibit signs of ongoing mergers and/or disturbed gas kinematics.Broad kinematic components are detected in a large fraction of the spaxels, and it is not clear whether these components originate from outflowing gas or from gas disturbed by the merger.We therefore followed the conservative approach by Lutz et al. (2020), and estimated the flux in the wings of the broad component, where we integrated the broad component only over wavelengths in which it contributes more than 50% of the total flux density of the line (i.e., wavelengths in which the flux density of the broad component is larger than the flux density of the narrow component).The broad component dominates the line profile for velocities in the range ± (600-1200) km s −1 with respect to the narrow core.
We then used the measured Hα/Hβ flux ratios to estimate the dust reddening toward the line-emitting gas, once for the nonoutflowing gas and once for the broad wings.Assuming case-B recombination, a gas temperature of 10 4 K, a dusty screen, and the Cardelli et al. (1989) extinction law, the color excess is given by: where (Hα/Hβ) obs is the observed line ratio.We then corrected all the observed fluxes for dust extinction using the derived E (B − V ) values.We used the [S II]λ6717 Å/[S II]λ6731 Å intensity ratio to estimate the electron density in the gas, once for the narrow component and once for the broad wings.Assuming a gas temperature of 10 4 K, the electron density is given by (e.g., Fluetsch et al. 2021): Å, and a = 0.4315, b = 2107, c = 627.1.Due to the critical densities of the two [S II] transitions, the intensity ratio is sensitive to electron densities in the range 50-5000 cm −3 , and its value ranges between 0.44 and 1.44.For electron densities outside this range, the intensity ratio is constant and cannot be used to infer the density.In addition, in Baron & Netzer (2019) and Davies et al. (2020) we used the ionization parameter of the gas to estimate the electron density and found that the [S II] lines can underestimate the electron density in the ionized outflow.By comparing the observed line ratios to photoionization models, we suggested that this is because the [S II] lines are emitted close to the ionization front of the cloud rather than the mean over the ionized region.Since the electron density drops rapidly near the ionization front, the [S II] lines trace low electron density regions.Therefore, we also use the ionization parameter method presented in Baron & Netzer (2019) to estimate the electron density.

Neutral Gas
To study the neutral gas properties, we used the same Voronoi bins defined in Section 2.4 for the emission lines.The binned spectra show evidence of NaID absorption, emission, or  and 2.5).The first column shows the integrated flux, which is used as the signal S when defining the spaxel S/N.For the stellar properties, the wavelength range is 5300-5700 Å, and for the gas properties, the summed flux is the continuum-subtracted 6520-6620 Å range, which includes the Hα and [N II] lines.The second column shows the bin indices assigned to each spaxel, and the third shows the combined S/N of each bin.By construction, the minimal combined S/N is 250 for the stellar binning and 100 for the line binning, since these are the target S/Ns we selected.The fourth column shows the number of spaxels in every bin.The Voronoi bins include only single spaxels in the centers of the galaxies, while having tens of spaxels in the outskirts.
a combination of the two.Interstellar NaID absorption, in particular blueshifted absorption, has been detected in numerous starburst and AGN-dominated systems.It has been widely used to constrain the neutral phase of galactic winds (see a review by Veilleux et al. 2020).Redshifted NaID emission, until recently, has only been detected in a handful of sources (Rupke & Veilleux 2015;Perna et al. 2019;Baron et al. 2020).However, more recent studies using 1D spectroscopy (Baron et al. 2022) and IFU observations (Fluetsch et al. 2021) of large samples of starburst and AGN-dominated galaxies detect NaID emission in a large fraction of the sources and/or spaxels (30%), suggesting that NaID emission is not a rare phenomenon.
NaID absorption is the result of the absorption of continuum photons along the line of sight.On the other hand, NaID emission is the result of absorption by neutral sodium outside the line of sight and an isotropic reemission.As we discussed extensively in Baron et al. (2020) and Baron et al. (2022), a neutral outflow may produce a P-Cygni profile in the NaID lines, with the approaching part of the outflow producing blueshifted absorption, and the receding part of the outflow producing redshifted emission (see also Prochaska et al. 2011).Neglecting the redshifted NaID emission component may result in an underestimation of the neutral outflowing gas mass, similar to neglecting the red wing of broad emission lines originating in ionized outflows.
Our modeling of the NaID absorption and emission profile follows the methodology outlined by Baron et al. (2022).We refer the reader to Appendix B in that paper for the full model, discussion of assumptions, and possible degeneracies between the parameters, while here we only briefly summarize the main components of the model.We considered three different models: (i) NaID absorption only, (ii) NaID emission only, and (iii) a combination of blueshifted NaID absorption and redshifted emission.To properly model the NaID profile, we must also include a model for the HeIλ5876 Å emission.We modeled the HeI profile using the best-fitting parameters of the Hα line, where the central wavelength and line width were locked to those of the Hα, and the HeI/Hα amplitude ratio was allowed to vary between 0.01 and 0.05.In the most general case of NaID absorption and emission, we modeled the observed flux as: where f (λ) is the observed flux, f stars (λ) is the stellar continuum obtained using PPXF, f HeI (λ) represents the HeI emission, f NaID emis (λ) represents the redshifted NaID emission, and I NaID abs (λ) represents the NaID absorption.Since the NaID emission is additive, while the NaID absorption is multiplicative, one must model the observed spectrum rather than the normalized one.We fitted all three models and selected the model with the lowest reduced χ 2 .In Figure C1 in Appendix C we show examples of the best-fitting NaID profile in cases of absorption only, emission only, and a combination of the two.
In case of NaID absorption, we used the best-fitting optical depth τ 0 (NaID K ) to estimate the neutral sodium column density via (e.g., Draine 2011): where f lu = 0.32, λ lu = 5897 Å, and b is the Doppler parameter, which is related to the velocity dispersion via s = b 2 .We then estimated the hydrogen column density using (Shih & Rupke 2010): where (1 − y) is the sodium neutral fraction which we assume to be 0.1, A is the sodium abundance term, B is the sodium depletion term, and C is the gas metallicity term.Following Shih & Rupke (2010), we took . For the stellar masses of the systems in our sample, the mass-metallicity relation (e.g., Tremonti et al. 2004) suggests that the metallicity is roughly twice solar.We therefore used In the case of redshifted NaID emission, we integrated the emission line profile to obtain the flux.The NaID-emitting gas has a comparable spatial extent to the broad wings-emitting gas (see details in Section 3.3).We therefore assume that it is affected by roughly similar dust columns, and correct the NaID flux for dust extinction using the E(B − V ) derived using the Hα/Hβ flux ratio in the broad wings.

Ancillary Properties
We extracted the stellar masses reported in the MPA-JHU value added catalog for the galaxies in our sample (Kauffmann et al. 2003a;Tremonti et al. 2004).For the AGN bolometric luminosity, we estimated the total dust-corrected Hβ luminosity by integrating the luminosities of all the spaxels that are associated with the primary galaxies and with line ratios consistent with AGN ionization (see Figure B1).We then used the bolometric correction factor from Netzer (2019) to convert the dust-corrected Hβ luminosity to bolometric luminosity.
We use several different estimates for the SFR.In Baron et al. (2022) we used IRAS 60 μm observations to show that many systems selected to have poststarburst optical signatures host in fact significant obscured star formation.In particular, for our parent sample of poststarbursts with AGN and ionized outflows, we found that 45% are >0.3 dex above the starforming main sequence (confidence intervals 36%-56%), and 32% are >0.6 dex above (confidence intervals 24%-41%).We found a significant correlation between the far-infrared SFR and the AGN bolometric luminosity, which is in line with the relation observed in active starbursts.In Baron et al. (2023) we used NOEMA observations to study the star formation and molecular gas properties in a subset of the galaxies.In particular, we combined the millimeter continuum emission from NOEMA with the IRAS 60 μm observations to estimate the SFR.Four out of the five galaxies were observed with NOEMA and we use the SFR estimates (or upper limits) reported by Baron et al. (2023).The fifth galaxy has only an IRAS-based upper limit on the SFR.In total, three of the five galaxies have SFR estimates, while two have only upper limits.We also used the derived AGN bolometric luminosities and our best-fitting relation between L(AGN) and L(SF) from (Baron et al. 2022, Figure 5) to estimate the far-infrared SFR.
We list the AGN luminosities and the SFRs in Table 1.

Outflow Properties
In this section we describe our methods to derive different outflow properties, starting with the ionized outflow.Assuming that the broad wings of the emission lines originate from an ionized outflow (see, however, Section 3.2), we used the bestfitting line profiles to derive the outflow extent, velocity, ionization parameter, electron density, outflowing gas mass, mass outflow rate, and outflow kinetic power.
As described in Section 2.4, broad kinematic components are detected in a large fraction of the spaxels in each galaxy.Since the galaxies are undergoing mergers and/or show signatures of disturbed gas kinematics, we followed the conservative approach by (Lutz et al. 2020, see Figure 1 therein) and considered only the wings of the broad profiles as originating from an outflow.The wings are defined as the wavelengths in which the broad component contributes more than 50% of the total flux density.In each spaxel, we considered a red/blue wing detected if its integrated flux was larger than 3 times its uncertainty, which was estimated by propagating the uncertainties of the best-fitting parameters of the broad kinematic component.As a result, in some spaxels we detected only the red or only the blue wing.
For the outflow extent, we defined the brightest spaxel in the primary galaxy as the center, and estimated the distance of all the spaxels in which broad wings in Hα were detected from the center.We considered two definitions for the outflow extent.The first is r outflow = r 95 , where r 95 is the 95th percentile of the distance distribution.According to this definition, the outflow extent is close to the maximal distance in which broad wings are detected.The second definition is the Hα flux-weighted average distance, which is smaller than the maximal distance by a factor of 1-2.5.As discussed in Section 3.4, our main conclusions do not change when adopting one definition versus the other.
For the outflow velocity, we defined the maximal outflow velocity in each spaxel as Δv + 2σ for the red wing and Δv − 2σ for the blue wing, where Δv and σ are the centroid velocity and velocity dispersion of the broad kinematic component.We used these velocities when estimating the mass outflow rate and kinetic power in each spaxel.We also estimated the global outflow velocity in each galaxy as the Hα flux-weighted average of the outflow velocities of the individual spaxels.We show in Section 3.2 that the red and blue wings are roughly symmetric in terms of dust-corrected Hα flux, maximal velocity, and spatial extent.Therefore, for spaxels where both the red and blue wings are detected, the outflow velocity was defined as the average of the two.Otherwise, we adopted the velocity of the detected wing only.
We used the ionization parameter method presented in Baron & Netzer (2019) to estimate the average electron density in the ionized wind independently from the [S II] method (Equation (4) there).The ionization parameter method assumes AGN-photoionized gas and is based on the simple relation between the AGN luminosity, the gas distance from the AGN, and its ionization state.It requires knowledge of the AGN bolometric luminosity, the outflow extent, and its ionization parameter.For each galaxy, we used the median emission line ratios [N II]/Hα and [O III]/Hβ in the broad wings to estimate the ionization parameter in the outflowing gas (Equation (2) in Baron & Netzer 2019).Using the estimated AGN bolometric luminosity (Section 2.6) and the outflow extent r outflow = r 95 , we estimated the electron density in the outflow.
We estimated the outflowing ionized gas mass M ion , mass outflow rate  M ion , and kinetic power  E ion using Equation (7) and the related text from Baron & Netzer (2019).These estimates are standard and have been used extensively in the literature (e.g., Harrison et al. 2014;Fiore et al. 2017;Rupke et al. 2017;Fluetsch et al. 2021;Ruschel-Dutra et al. 2021).They require the knowledge of the dust-corrected Hα luminosity, the outflow extent, the electron density, and the effective outflow velocity.We estimated these properties for the red and blue wings in each spaxel separately, and then obtained the global M ion ,  M ion , and  E ion by summing over all spaxels.
For the neutral outflow, the best-fitting parameters of the NaID profile (see Section 2.5) include the absorption optical depth τ 0 (NaID K ), covering factor C f , centroid velocity, and velocity dispersion.If redshifted NaID emission is detected, then the best-fitting parameters also include the amplitude of the NaID K emission component, the doublet amplitude ratio, centroid velocity, and velocity dispersion.To distinguish between absorption/emission that originate from a neutral outflow versus from the nonoutflowing interstellar medium, we used the best-fitting centroid velocities.We considered the absorption to originate from an outflow if its centroid velocity is blueshifted by more than 100 km s −1 from the centroid velocity of the stars.Similarly, NaID emission was considered to originate from an outflow if its centroid velocity is redshifted by more than 100 km s −1 from the stars (see details in Section 3.3).
We defined the maximal velocity of the neutral outflow to be Δv − 2σ for the blueshifted absorption and Δv + 2σ for the redshifted emission.Similarly to the ionized outflows, for each spaxel where an outflow is detected, we estimated its distance from the brightest spaxel.For the blueshifted absorption, to estimate the outflowing neutral gas mass M neut , mass outflow rate  M neut , and kinetic power  E neut , we assumed the thin shell model by Rupke et al. (2005a), and used Equations (6) and (7) from Shih & Rupke (2010).These equations are the standard method to estimate the mass and energetics of the neutral outflow using spatially resolved observations (see review by Veilleux et al. 2020).
To the best of our knowledge, the redshifted NaID emission has not been included in estimates of the mass and energetics of neutral outflows so far.To estimate the neutral gas mass that is associated with the NaID emission, we estimated the dustcorrected NaID luminosity in each spaxel where a redshifted outflow has been identified.The NaID emission line luminosity can be expressed as: L emis (NaID) = EW emis (NaID) × L λ,stars , where L λ,stars is the stellar continuum at the NaID wavelength.We used the observed NaID luminosity and stellar continuum to estimate EW emis (NaID).Assuming absorption on the linear part of the curve of growth (optically thin absorption henceforth), the NaID column density associated with the redshifted emission is given by Draine (2011): where f lu = 0.32 and λ lu = 5897.In Section 3.3 we show that for the NaID emission that is associated with the outflow, the NaID amplitude ratio is close to 2, suggesting optically thin gas.We then used Equation (5) to convert the NaI column density to hydrogen column density N H, emis .We consider N H, emis as the hydrogen column of the NaID-emitting gas, and use Equations ( 6) and (7) from Shih & Rupke (2010) to estimate the gas mass, mass outflow rate, and kinetic power, of the neutral gas that produces the redshifted NaID emission.

Results
In Figure 3 we show the SFR versus stellar mass of the objects in our combined sample.The shape of the markers indicates the merger stage as derived from the IFU observations and described in Section 3.1 (see Table 2 for a summary).The color of the markers represents our adopted dynamical origin of the observed winds as described in Section 3.2, where we distinguish between the case of nuclear outflows that are consistent with AGN-driven winds, and nonnuclear flows that are inconsistent with AGN-driven winds (see Table 3 for a summary).The figure shows that five out of the seven galaxies in our combined sample are above the star-forming main sequence, with four galaxies being ∼1 order of magnitude above it.As we describe in Sections 3.1 and 3.2, these sources resemble local (ultra)luminous infrared galaxies (ULIRGs) more than their putative poststarburst descendants.

Ongoing Tidal Interactions and/or Mergers
The galaxies studied here were selected from our parent sample of poststarburst galaxies with AGN and evidence for ionized outflows (Baron et al. 2022).In particular, all the galaxies in our sample are Hδ-strong, with EW(H δ) > 5 Å, and with optical spectra that are dominated by A-type stars, suggesting a recent burst of star formation that was terminated abruptly ∼1 Gyr ago (e.g., Dressler & Gunn 1983;Couch & Sharples 1987;Poggianti et al. 1999).However, using IRAS 60 μm observations to estimate the SFR, in Baron et al. (2022) we showed that many systems selected as poststarbursts in the optical band host in fact obscured star formation, with some showing infrared luminosities comparable to local (U)LIRGs (see Calabrò et al. 2018 for similar results at z ∼ 0.7).We found that 45% of galaxies in our parent sample are >0.3 dex above the star-forming main sequence (36%-56% at 95% confidence), and 32% are >0.6 dex above (24%-41%).For our combined sample of seven galaxies observed with IFUs, Table 1 shows that six galaxies have far-infrared luminosities , while the remaining one has  » L L log 10.5 SF .These observations call into question the traditional interpretation of these sources as galaxies that started their transition to quiescence.In this section we use the spatially resolved properties of the stars and gas in these galaxies to further address this question.The main properties we discuss are summarized in Table 2.
In Figure 4 we show the spatially resolved properties of the stars.In particular, we show the stellar continuum emission in the range 5300-5700 Å, the stellar velocity and velocity dispersion, the reddening toward the stars, and the fraction of young stars.The latter is defined using the nonparametric star formation history obtained with PPXF as the sum of the weights of stellar templates younger than 1 Gyr, divided by the sum of all the weights.A fraction f young = 1 represents spaxels where the spectrum is completely dominated by stars younger than 1 Gyrs, while f young = 0 represents spaxels that are dominated by stars older than 1 Gyr.
All five galaxies show evidence of a recent or ongoing interaction, with J022912, J020022, and J111943 showing companion galaxies at the same redshift, J112023 showing at least two bright centers suggestive of a later-stage merger, and J080427 showing a tidal feature that extends to distances of ∼30 kpc.For the two previously published poststarbursts, J003443 shows a companion galaxy (Baron et al. 2018) and J124754 shows disturbed gas kinematics (Baron et al. 2020).Thus, among the seven galaxies selected as poststarbursts, six show clear signatures of an early interaction or ongoing merger in their stellar continuum emission.
Interestingly, three out of the five galaxies presented here (and four out of the combined seven), are at an early stage of an interaction, with visible companions at distances >10 kpc.This suggests that poststarburst optical signatures may appear well before the final coalescence and starburst.It is consistent with the idea that these galaxies have already had their first close passage, leading to the increased star formation seen in far-infrared wavelengths, but with enough time that has passed so that some regions have already experienced rapid quenching and are traced by poststarburst signatures (e.g., Hopkins et al. 2013a).
Figure 4 shows that four out of the five galaxies presented here (and five out of the combined seven) show ordered disklike motions in their stellar kinematics, consistent with observations of other major mergers (e.g., Engel et al. 2010;Perna et al. 2021).The exception is the ongoing merger J112023 which shows nonordered stellar velocities, very highvelocity dispersions (σ * ∼ 300 km s −1 ), and significant dust Figure 3. SFR vs. stellar mass for the galaxies in the combined sample.The SFRs and stellar masses are described in Section 2.6.The shape of the markers represents the merger stage as derived from our MUSE observations (see Section 3.1 and Table 2).The color of the markers indicates the most probable dynamical origin of the observed flow (see Section 3.2 and Table 3), with magenta markers representing nuclear winds (consistent with AGN-driven outflows) and gray markers with nonnuclear winds (inconsistent with AGNdriven outflows).Markers with black edges represent sources from this study, and the others represent objects presented in previous works by our group (see Table 1 for details).For comparison, the gray contours represent local SDSS galaxies, and the blue dashed line represents the star-forming main sequence at z = 0 by Whitaker et al. (2012) with shaded regions showing ± 0.3 dex.
reddening at the outskirts of the galaxy.We do not find a strong correspondence between the fraction of young stars and the reddening toward the stars.Some spaxels show high fractions of young stars and high reddening values, as expected.However, other spaxels show high reddening values and little young stars.This may be due to the significant obscuration of the young stellar population in these spaxels.
In Figure 5 we show the spatially resolved properties of the nonoutflowing ionized gas that is traced by the narrow kinematic component.We show the gas velocity and velocity dispersion, reddening toward the line-emitting gas, and the surface brightness of the narrow Hα.In three out of the five galaxies (four out of the combined seven), we find disturbed gas kinematics.The figure also shows significant reddening values, E(B − V ) of 0.5-1 mag, in a large fraction of the spaxels.The reddening toward the narrow-line-emitting gas is larger than toward the stellar continuum in most of the spaxels, which is in line with previous studies (e.g., Calzetti et al. 2000;Charlot & Fall 2000).
To summarize: all seven galaxies from our combined sample show signatures of tidal interaction/merger in their morphology, stellar kinematics, or gas morphology and kinematics.The galaxies are at different stages of interaction, including a pair with a separation of ∼50 kpc with no visible tidal tails or bridges, an ongoing merger with two nuclei, and a single visible nucleus with a ∼30 kpc tidal feature.These observations suggest that poststarburst signatures in optical (i.e., strong Hδ absorption) do not necessarily trace postmerger systems.Combined with our far-infrared and millimeter-based results (Baron et al. 2022(Baron et al. , 2023)), these observations suggest that systems selected as Hδ-strong with AGN and ionized outflows are more likely interacting pairs of dust-obscured starbursts than their postmerger poststarburst descendants.Their farinfrared luminosities, , are lower than those of z < 0.15 ULIRGs from the PUMA survey (which is designed to spatially resolve the ionized gas in local ULIRGs; e.g., Perna et al. 2021).At this stage, it is not clear whether our galaxies are less luminous versions of the PUMA ULIRGs, or at an earlier stage of the evolution.We leave a thorough comparison between the samples to a future publication.

Broad Wings of the Ionized Lines: Outflows or Interactioninduced Motions?
Broad kinematic components are detected in the majority of spaxels in each of the galaxies in our sample.Visual inspection reveals high-S/N broad features, with peak fluxes that are 10%-100% of the peak fluxes of the narrow lines.In Section 3.1 we found that all the galaxies show signs of interactions/mergers, with some of the galaxies showing disturbed gas morphologies and kinematics.In such systems, broad kinematic components do not necessarily trace AGN or SN-driven outflows, and may originate from galactic-scale gas flows caused by the interaction (see discussion in Section 4).In this section we study the properties of the ionized gas traced by the broad kinematic component, with the goal of constraining its origin.We summarize the different properties in Table 3, where we also list our suggested dominant origin for each of the components.
We derive the kinematics, spatial extent, fluxes and reddening, and dominant ionizing source for the blue and red wings separately, and use these, across many spaxels, to identify the most probable dynamical origin of the gas flows.In our classification, we distinguish between the case of a nuclear outflow (driven by the AGN, SN, or a combination of the two) and a nonnuclear flow that could be caused by SN at the closer edge of the galaxy or due to interaction-induced flows.Importantly, we make the distinction between the dynamical origin of the wind and the dominant ionization source of the gas in the wind.For example, it is possible that nonnuclear flows are primarily ionized by the AGN and show Seyfert-like optical line ratios.
For a nuclear outflow, we expect to find an asymmetry in either the kinematics, spatial extents, or fluxes and reddening between the red and the blue wing, when considering all the spaxels of a single galaxy.For example, a nuclear outflow might show a double-cone-like structure in kinematics and/or line extents (e.g., J022912 in Figure B2), with the red wing dominating in one part of the galaxy and the blue wing in the other.For a nuclear wind viewed face-on, the red and blue wing kinematics and extents may be symmetric on galactic scales, but we expect the flux of the red wing to be lower than that of the blue wing, or to show significantly larger reddening values.This is because a larger fraction of the receding part of the outflow will be behind the stellar disk, compared to the approaching side of the flow (see Figure 14 in Davies et al. 2014 and the discussion in Section 4).In case the red and blue wings show comparable velocities, symmetric extents on kpc scales, and comparable fluxes and reddening values (see, e.g., J080427 in Figure B3), we rule out the nuclear outflow origin.
To study the dominant ionization source of the gas, we use the line ratios of the broad wings and their relation to the velocity dispersion.The galaxies in our sample were selected to have narrow emission line ratios consistent with AGN ionization (Seyfert or LINER) in their 1D SDSS spectrum.The spatially resolved MUSE observations allow us to investigate the dominant ionization source in different regions within each galaxy, and study the contribution of shock excitation.We use   normalized stellar continuum emission in the range 5300-5700 Å, stellar velocity with respect to systemic, stellar velocity dispersion, dust reddening toward the stars, and the fraction of young stars, which is defined as the sum of the weights of the stellar templates younger than 1 Gyr, divided by the sum of all the weights.Perna et al. 2017), and they indicate a coupling between the gas ionization and kinematics.Since such coupling is not predicted by photoionization models, studies attributed this relation to shock excitation.However, as discussed by Laor (1998), shocks are inefficient in converting mass to radiation, and as a result, even significant mass outflow rates will produce very little line radiation.Therefore, it is necessary to compare not only the line ratios, but also the line luminosities, to those predicted by shock models.We consider shock excitation as the dominant ionization mechanism if the following criteria are met: (i) there is a correlation between σ broad and at least one of the line ratios [N II]/Hα and [S II]/Hα, (ii) the observed line ratios are consistent with the predicted line ratios from fast radiative shocks by Allen et al. (2008) for densities in the range 10-1000 cm −3 , and most importantly, (iii) the observed Hα surface brightness is consistent, to a factor of 3, with the Hα luminosity predicted by Allen et al. (2008).As discussed later in this section, these criteria are met for the low-velocity dispersion, low luminosity, and extended broad lines in two galaxies.They are not met for the inner spaxels that show highvelocity dispersions and high line luminosity.
Table 3 summarizes our results, which include a comparison of the geometry of the red and blue wings; a comparison of the reddening values and electron densities derived for the narrow and wing components; our adopted dominant source of ionizing radiation; and our adopted dynamical origin of the observed line wings.In Appendix B.1 we describe the individual properties of each source and present various diagnostic diagrams in Figures B2, B3, B4, B5, and B6.Our detailed analysis suggests that the observed broad emission lines are inconsistent with AGN-driven outflows in three out of five galaxies.This is in stark contrast to our initial expectation and the default assumption by most studies that high-velocity (v > 500 km s −1 ) ionized gas in active galaxies originates from AGN-driven outflows.
In Figure 6 we show the distribution of reddening and electron density values for different kinematic components (narrow, broad, red or blue wings, red+blue wings) across all the Voronoi bins of the different galaxies.We find comparable reddening values for the broad lines and red/blue wings and find them to be significantly lower than the reddening values derived using the narrow lines.This is in line with the results by Mingozzi et al. (2019) andFluetsch et al. (2021) for local AGN and (U)LIRGs respectively, where the studies classified all broad components as originating from an outflow.However, the individual diagnostic diagrams (Figures B2, B3, B4, B5, and B6) show a diversity in the reddening properties, with some spaxels showing significantly higher reddening values in their broad components and wings compared to the narrow lines.This diversity in reddening can be used to place additional constraints on the outflow geometry and its origin, and by considering only the combined distributions from all the sources, important differences may be averaged out.
The right panel of Figure 6 compares the electron densities derived for the narrow lines, broad lines, and red+blue wings, using the [S II] doublet.When considering all the bins, we find that the electron density of the wings (∼300 cm −3 ) is higher than the electron density of the broad component (∼200 cm −3 ), which is higher than that of the narrow component (∼100 cm −3 ).This too, is in line with the results by Mingozzi et al. (2019) and Fluetsch et al. (2021).However, inspecting the individual diagnostic diagrams reveals that this is not the case for J080427, where the electron density in the wings/broad component is lower than that of the narrow lines.
We used the ionization parameter method to derive the electron density in spaxels that are dominated by AGN photoionization (Baron & Netzer 2019; see Table 5).The resulting electron densities are comparable to those derived using the [S II] for two sources (J022912 and J020022) and are 10-100 times larger than the [S II]-based ones for three galaxies (J080427, J112023, and J111943).For the estimates of mass and energetics of the ionized outflows, we use the [S II]-based estimates to allow for a straightforward comparison with other studies, and since the ionization parameter method can only be applied to spaxels dominated by AGN photoionization.

Complex Picture of the Neutral Gas as Traced by NaID Absorption and Emission
In Section 3.2 above we performed a detailed analysis of the broad wings of the ionized lines, finding them to be inconsistent with AGN-driven outflows in three out of five of the cases.Furthermore, in some galaxies, the observations suggested that the broad emission lines originate from galacticscale interaction-induced motions, rather than from outflows.These observations raise the question of how to interpret broad kinematic emission components, given that both inflows and outflows can produce redshifted and blueshifted wings.Contrary to the ionized lines, the resonant NaID absorption does not suffer from the inflow-outflow degeneracy.In the case of an outflow, the NaID profile will resemble a P-Cygni profile, with blueshifted absorption and redshifted emission.In the case of an inflow, the profile is expected to be reversed, showing redshifted absorption and blueshifted emission.Therefore, NaID emission and absorption may be more straightforward to Figure 6.Reddening and electron density distributions for different kinematic components.The left panel compares the distributions of reddening measured using the narrow lines (gray), broad lines (black), and red and blue wings (red and blue respectively), in every Voronoi bin, for all the galaxies we consider.The dashed lines represent the median values.Similarly to other studies, when considering all bins from all galaxies together, the reddening for the broad lines and red/blue wings is lower than those of the narrow lines.The right panel compares the distribution of electron density measured for the narrow lines (gray), broad lines (black), and red +blue wings (purple), using the [S II] doublet, in every Voronoi bin and for all the galaxies we consider.Similarly to other studies, we find higher electron densities in the broad lines/ wings.This simple picture of lower reddening and higher electron density of the outflow can only be reproduced when considering all the spaxels from all the galaxies combined.Examining each source individually reveals a significant diversity in reddening and density properties.These properties can be used to place constraints on the outflow origin, and considering only the combined distributions may average out important differences.
interpret in interacting and/or merging systems, in particular in cases where spatial information (e.g., IFU) is not available.
Table 2 summarizes the NaID properties of the galaxies in our sample.NaID absorption is detected in the three galaxies: J022912, J080427, and J112023.Interestingly, NaID emission is detected in two galaxies, which are the only two galaxies for which a nuclear outflow origin (either AGN-driven wind or central SN-driven wind) is our favored interpretation for the broad ionized lines.NaID was neither detected in absorption nor in emission for J020022 and J111943.For these two galaxies, the observations suggested that the broad ionized lines originate from either interaction-induced flows or SNdriven winds at the closer edge of the galaxy.
We now focus on the NaID profile of J022912, which is the most complex out of the three.A similar analysis has been applied to J080427 and J112023 (see Figure C3), and we report the mass and energetics of the neutral outflows for all three galaxies in Section 3.4.In Figure 7 we show the spatially resolved properties of the NaID absorption for J022912.In particular, we show the best-fitting absorption optical depth, covering factor, and centroid velocity with respect to systemic.The observations suggest optically thin absorption, with a unity covering factor in the center of the galaxy and low covering factors outside.The spatially resolved centroid velocity is consistent with an outflowing gas cone with a large opening angle.The figure also compares the centroid velocity of the NaID absorption and the stars.Our conservative approach is to consider the blueshifted NaID absorption as an outflow only if it is blueshifted with respect to the stars by more than 100 km s −1 , which is indicated in the diagram.
In Figure 8 we show the spatially resolved properties of the NaID emission for J022912.We show the derived NaID emission flux, the doublet amplitude ratio NaID H /NaID K , where a ratio of 1 (2) suggests optically thick (optically thin) gas, and the centroid velocity of the NaID emission with respect to systemic.Contrary to the NaID absorption that shows only blueshifted velocities, the NaID emission shows both blueshifted and redshifted NaID emission.Since this gas is located at the farther side of the galaxy, behind the stars, the blueshifted emission can be easily interpreted as an inflow and the redshifted emission as an outflow.These two components show distinct NaID H /NaID K ratios, where the inflow is optically thick and the outflow is optically thin.Similarly to the absorption case, we consider redshifted NaID emission as an outflow only if it is redshifted with respect to the stars by more than 100 km s −1 .
In Figure 9 we compare the extents of the narrow-lineemitting gas, the nuclear-ionized outflow, and the blueshifted NaID absorption and redshifted emission that are associated with the neutral outflow.The extent of the neutral outflow is comparable to that of the nuclear-ionized outflow.Interestingly, the blueshifted NaID absorption and redshifted NaID emission are most dominant in different regions in the galaxy.This seems to be a generic property of neutral outflows, where similar behavior is seen for J112023 (see Figure C3), J124754 (Baron et al. 2020), and F05189-2524 (Rupke & Veilleux 2015).It may be partially due to absorption-emission line filling and velocity projections.It further emphasizes the importance of taking into account the NaID emission when estimating the mass and energetics of neutral outflows, as it traces separate regions of the outflow.
We do not find a significant connection between the neutral and ionized outflow phases (see Figure C2 in Appendix C for J022912).In particular, we do not find a significant relation Figure 8. Spatially resolved properties of the NaID emission in J022912.From left to right: the integrated flux of the NaID emission, the amplitude ratio of the H and K components, where NaID H /NaID K = 1 for optically thick gas, and NaID H /NaID K = 2 for optically thin gas, the centroid velocity of the NaID emission with respect to systemic, and a comparison between the centroid velocity of the NaID emission and the stars.Redshifted NaID emission is considered as an outflow if it is redshifted with respect to the stellar velocity by more than 100 km s −1 .
between the redshifted NaID velocity and the velocity of the red Hα wing, and similarly for the blueshifted NaID absorption and the blue Hα wing.In addition, we do not find a significant relation between the NaID EW or flux and the flux of the Hα wings.Nevertheless, similarly to J124754 (Baron et al. 2020), we find that the NaID emission to Hα flux ratio is about 0.1.

Multiphased Wind Energetics
In Tables 5 and 6 in Appendix D we list the derived mass and energetics of the ionized and neutral winds for the galaxies in our sample.As noted in Sections 3.2 and 3.3, the derived red and blue wing kinematics, spatial extents, fluxes, and reddening, for the ionized lines are consistent with a nuclear outflow origin (AGN, SN, or a combination of the two), in two out of the five galaxies.These galaxies, J022912 and J112023, are also the only two systems for which NaID has been detected in emission.These two galaxies have AGN bolometric luminosities of = L log 45 bol .For these galaxies, the mass outflow rates of the ionized gas are 10 and 3 M e yr −1 respectively, compared to the mass outflow rate of the neutral outflow which is 7 and 21 M e yr −1 respectively.Assuming that the multiphased outflows are driven solely by the AGN, the wind coupling efficiency is estimated to be ∼0.5%.These systems are also luminous infrared galaxies with the highest star formation luminosities within our sample ( L log SF of 11.3 and 11.5 L e , equivalent to SFRs of 20 and 30 M e yr −1 , respectively).In such systems, the observed winds are probably driven by a combination of SN and AGN feedback (e.g., Nelson et al. 2019), making the derived coupling efficiencies upper limits.
Table 6 suggests that the mass and energetics of the neutral outflow as traced by NaID emission are comparable to those traced by NaID absorption.Therefore, we suggest that, when detected, NaID emission should be taken into account when estimating the mass and energetics of neutral outflows.The derived mass and energetics of the neutral outflows are within the ranges observed in (U)LIRGs+AGN by Rupke et al. (2005a) and by Fluetsch et al. (2021) for AGN of comparable luminosities.Similarly, the derived mass and energetics of the ionized outflows are comparable to those reported by Fluetsch et al. (2021) for (U)LIRGs+AGN with comparable luminosities.

Discussion
Figure 10 depicts the emerging picture of the warm ionized gas in the sources in our sample.The observed broad blueshifted and redshifted wings may be a complex combination of nuclear outflows due to AGN and/or SN, nonnuclear SN-driven outflows (e.g., at the edge of the galaxy), and merger-induced flows.The gas velocities and line fluxes were arbitrarily chosen and are not based on a physically motivated model for the gas kinematics expected from each of the dynamical processes.However, simulations of galaxy major mergers find that the warm ionized gas in merger-induced flows can reach velocities of 500-750 km s −1 on tens of kpc scales (e.g., Hopkins et al. 2013b).The cartoon illustrates the challenge of disentangling the different processes even with spatially resolved observations.Importantly, a distinction should be made between the process that ionizes/excites the gas and the dynamical origin of the flow.For example, tidally stripped gas may be primarily ionized by the AGN, showing significant blueshifted and redshifted emission lines with Seyfert-like line ratios.Therefore, line ratios alone cannot be used to distinguish between different processes and isolate the contribution of AGN feedback.In this paper we used the spatial distribution of the flux in the redshifted and blueshifted wings, the derived reddening, and kinematics to place constraints on the dynamical origin of the observed flows.
The sample presented in this paper was selected from our parent sample of poststarburst galaxies with AGN and indications for an ionized outflow (Baron et al. 2022).In our selection, we assumed that systems that show a combination of narrow and broad kinematic components in their recombination and forbidden lines host ionized outflows.Therefore, we formed our parent sample by selecting galaxies that show broad kinematic components in their optical emission lines.This is a typical assumption and choice in studies of outflows in active galaxies, in particular in cases where only a small subset of objects can be followed up with IFU observations (e.g., Mullaney et al. 2013;Harrison et al. 2014;Bae et al. 2017;Fiore et al. 2017;Mingozzi et al. 2019;Fluetsch et al. 2021).Despite this selection, our analysis suggests that the broad emission lines originate from nuclear outflows (AGN or SN driven) only in two out of five objects.In the rest, the broad emission line properties are more consistent with a nonnuclear dynamical origin, for example, interaction-induced galacticscale flows.This suggests that selecting galaxies with broad emission lines may bias the sample to include a larger fraction of interacting systems.Estimates of SFR, stellar mass, and morphological class can be used to remove interacting galaxies, thus minimizing this bias.
While this does not pose a particular problem for studies of AGN in low-z and high-z galaxies that show no dynamical disturbance (e.g., low-z: Davies et al. 2014;Mingozzi et al. 2019;high-z: Genzel et al. 2014;Förster Schreiber et al. 2019), it may be a more significant challenge for infrared-bright galaxies, which are more likely to be interacting systems.In such systems, the close interaction may produce high gas velocities on galactic scales, which appear as broadened kinematic components in the emission lines and may mistakenly be classified as galactic-scale outflows (see Puglisi et al. 2021 for an example with molecular gas).In particular, the largest compilations of multiphase outflows in active galaxies (Fiore et al. 2017;Fluetsch et al. 2019Fluetsch et al. , 2021) ) include a large fraction of infrared-bright galaxies in which the outflows have been detected primarily in emission (in optical or millimeter wavelengths), and it is not clear what is the contribution of merger-induced motions to the observed broadened lines.In such cases, absorption lines may be more straightforward to interpret as outflows (e.g., Rupke et al. 2005cRupke et al. , 2017)).For example, spatially integrated and resolved spectroscopic observations of NaID in nearby (U)LIRGs show an excess number of blueshifted relative to redshifted NaID absorption systems.This favors outflows over merger-induced streams, since the latter case will result in an equal number of blueshifted and redshifted systems (Figure 2 in Rupke et al. 2002; Figure 1 in Rupke et al. 2005a, Rupke et al. 2017).
Since quasar activity is linked to galaxy interactions (e.g., Sanders & Mirabel 1996;Genzel et al. 1998;Hopkins et al. 2006;Veilleux et al. 2009;and more recently, e.g., Hernández-Toledo et al. 2023;Pierce et al. 2023;Dougherty et al. 2024), our results raise the question of how to interpret broadened kinematic components observed in quasar spectra, in particular where spatially resolved information is not available, and where the Balmer and the lower ionization [N II] and [S II] lines that may trace the outflow are often blended with the broad Hα and Hβ lines originating in the broad-line region.The detection of a blueshifted broad wing without a redshifted broad wing in a large sample of sources may indicate an outflow origin, rather than merger-induced flows, on average (e.g., Zakamska & Greene 2014;Perna et al. 2015;Zakamska et al. 2016).However, a nonnegligible fraction of these quasars shows evidence of a redshifted broad wing as well (see, e.g., Figure 2 in Zakamska & Greene 2014).Since the red and blue broadened wings of the Hα and Hβ lines are blended with the broad-line region Hα and Hβ, one cannot derive the reddening of the kinematic components to test whether these cases are consistent with a nuclear outflow origin.

Summary and Conclusions
Poststarburst E+A galaxies are believed to be the evolutionary link between major merger (ultra)luminous infrared galaxies and quenched ellipticals.Both observations and simulations suggest that this transition is rapid, with the starburst quenching abruptly over a timescale of a few hundred Myrs.Although simulations invoke AGN feedback as one of the processes responsible for the rapid quenching of star formation, little is known observationally about AGN feedback, in particular AGN-driven winds, in this stage.To study the role of AGN feedback in the transition from Figure 10.Emerging picture of the warm ionized gas in interacting galaxies.The cartoon on the right depicts two interacting galaxies, where gas is stripped during the interaction to form tidal features.The upper galaxy hosts AGN-driven winds in a cone-like structure and SN-driven winds shown as stars.The field of view is marked with the horizontal dashed lines.The left panels represent the warm ionized gas line emission of the different components, showing that the total line emission is a combination of emission from the stationary gas in the primary galaxy, nuclear outflows due to AGN and/or SN, nonnuclear SN-driven winds (e.g., at the edge of the galaxy), and merger-induced gas flows.The line fluxes and gas velocity scales were arbitrarily chosen and are not based on a physically motivated model of the expected gas kinematics of the different processes.The cartoon illustrates that the blueshifted and redshifted wings that are typically attributed to AGN winds can be in fact a complex combination of multiple dynamical processes.Spatially resolved observations of multiple lines tracing the flow may be used to place constraints on the dominant dynamical process.
starburst to quiescence, we constructed a sample of galaxies with poststarburst signatures in optical (strong Hδ absorption), evidence for an AGN (using narrow-line ratios), and evidence for ionized outflows (presence of broad kinematic components in Hα and [O III]).We presented the full sample in Baron et al. (2022), where we found that a large fraction of the poststarburst galaxies host obscured star formation, with some systems showing infrared luminosities comparable to those of local (ultra)luminous infrared galaxies.In Baron et al. (2018) and Baron et al. (2020) we used optical IFUs to spatially resolve the stars and gas in two such galaxies.In this work, we used MUSE/VLT observations of five additional galaxies to study the spatial distribution of the stars and multiphased gas, and in particular, constrain the properties of the multiphased outflows.Our results and their broader implications are summarized below.
(I) Tidal interactions and/or mergers.All seven galaxies from our combined IFU sample show signatures of interaction or merger in their stellar or gas morphology or kinematics.In addition, five out of seven galaxies show infrared luminosities of The galaxies in our sample are at different stages of interaction, including a pair of interacting galaxies at a distance of ∼50 kpc from each other, ongoing mergers with two visible nuclei at a distance <5 kpc, and a galaxy with no visible companion but with a tidal tail extending to a distance of 30 kpc.Interestingly, four out of our combined sample of seven are at an early stage of the interaction, with visible companions at distances of >10 kpc.This suggests that poststarburst signatures in optical (strong Hδ absorption) are not necessarily associated with postmerger systems.The observations are consistent with the idea that these galaxies have already had their first close passage, which led to the elevated SFR seen in infrared wavelengths and the poststarburst signatures seen in optical.Importantly, our observations suggest that Hδ-strong galaxies selected to have signatures of AGN and ionized outflows are more likely interacting starburst galaxies, rather than postmerger poststarburst galaxies.
(II) Broad kinematic components in optical emission lines do not necessarily trace outflows.Using the MUSE observations, we performed a detailed analysis of the morphology, kinematics, flux distribution, and reddening of the broad kinematic components in each of the galaxies.Contrary to our initial expectation, the observations are consistent with nuclear (AGN or SN-driven) outflows only in two out of the five galaxies (four out of the seven galaxies in the combined sample).For some of the galaxies, the observations are more consistent with galacticscale motions induced by the interaction/merger, a process that is often overlooked in studies of outflows.It is possible that our selection of galaxies with broad components in their optical emission lines favors interacting systems.This has significant implications for studies of ionized outflows in active galaxies, where it is a common practice to select systems with broader kinematic components in Hα or [O III] and classify the broad component as originating from an AGN-driven outflow.This poses a particular challenge for studies of higher redshift quasars, where spatially resolved information is not available, and since quasar activity is linked to mergers in the local Universe.Our results question the common assumption that broad kinematic components in the ionized emission lines trace primarily galactic-scale AGN or SN-driven outflows.
(III) NaID emission and absorption are effective outflow tracers.We detect NaID absorption in three out of the five galaxies.We detect a combination of blueshifted NaID absorption and redshifted NaID emission (classical P-Cygni profile) in two systems, which are also the only two systems whose ionized lines are consistent with a nuclear (AGN or SN-driven) outflow.Contrary to the ionized emission lines, where it is not clear whether the blue/red wings trace inflows or outflows, the NaID P-Cygni profile does not suffer from this degeneracy.In the case of an outflow, we expect to find blueshifted NaID absorption and redshifted NaID emission, while in the case of an inflow, the profile will be reversed with redshifted absorption and blueshifted emission.The blueshifted NaID absorption and redshifted NaID emission tend to trace separate regions within the galaxies, suggesting that, when detected, the NaID emission should be taken into account when estimating the mass and energetics of neutral outflows.We estimated the mass of the outflow, mass outflow rate, and kinetic power of the neutral gas that is traced by the NaID emission, and found them to be comparable to those derived from the absorption.We did not find a significant connection between the neutral and ionized outflows but generally found L NaID ∼ 0.1L Hα .
(IV) Properties of multiphased outflows.For the two galaxies where the observations are consistent with an ionized nuclear (AGN or SN-driven) outflow, we found mass outflow rates of 10 and 3 M e yr −1 .These two galaxies also show a combination of NaID emission and absorption, with a total mass outflow rate of 7 and 21 M e yr −1 .Assuming that the multiphased outflows are driven solely by the AGN, the wind coupling efficiency is estimated to be ∼0.5%.However, both of these systems are infrared luminous galaxies ( L log SF of 11.3 and 11.5 L e ), where we expect some contribution from SN to the observed winds.Therefore, the reported coupling efficiencies are upper limits.
The present study, together with the earlier papers published by our group (Baron et al. 2022(Baron et al. , 2023) ) highlights the importance of using IFU observations combined with FIR-based estimates of SFRs in studies of galactic outflows.The IFU observations can be used to distinguish between different types of flows in galaxies (inflows versus outflows), different types of ionization mechanisms (AGN and star formation [SF] ionization, shock excitation), and to map the kinematics and morphologies of neutral and ionized gas clouds.To form a more complete picture of the flows in these transitioning galaxies, it is also necessary to study the cold and warm molecular phases, traced by millimeter carbon monoxide lines and infrared H 2 lines respectively.We are currently involved in follow-up observations to trace these phases, and results will be reported in a forthcoming publication.
J111943: Figure B6 summarizes the broad-line diagnostics.The broad wings show similar extents to that of the narrow lines, with the red and blue wings having similar extents and comparable flux and reddening values.This rules out nuclear outflow origin for this source.Similarly to J020022, the [O III] is masked out by the reduction pipeline and we use the integrated [O III]/Hβ line ratio from the SDSS.The line ratios are consistent with either LINER or Seyfert radiation, and they show no correlation with the velocity dispersion.We therefore rule out shock excitation as the main source of ionizing radiation.(Baldwin et al. 1981;Veilleux & Osterbrock 1987).We mark the two separating criteria that are used to separate star-forming from AGN-dominated galaxies (Kewley et al. 2001;Kauffmann et al. 2003b

Appendix D Multiphase Outflow Properties
Table 5 summarizes the ionized outflow properties in the galaxies of our sample, and Table 6 the neutral outflow properties.

Figure 1 .
Figure 1.The MUSE view of the stellar and gas emission of the galaxies in our sample.Each row represents a galaxy from our sample.The left column shows the integrated emission within the rest-frame wavelength range 5300-5700 Å, assuming the redshift of the target galaxy.The right column shows the integrated continuum-subtracted flux in the rest-frame wavelength range 6520-6620 Å, which includes the Hα and [N II] lines.The 3" SDSS fiber is marked as a green circle.

Figure 2 .
Figure2.Illustration of the Voronoi binning applied to the data cubes.The top row represents the binning done for the stellar properties (Section 2.3), and the bottom for the emission and absorption lines (Sections 2.4 and 2.5).The first column shows the integrated flux, which is used as the signal S when defining the spaxel S/N.For the stellar properties, the wavelength range is 5300-5700 Å, and for the gas properties, the summed flux is the continuum-subtracted 6520-6620 Å range, which includes the Hα and [N II] lines.The second column shows the bin indices assigned to each spaxel, and the third shows the combined S/N of each bin.By construction, the minimal combined S/N is 250 for the stellar binning and 100 for the line binning, since these are the target S/Ns we selected.The fourth column shows the number of spaxels in every bin.The Voronoi bins include only single spaxels in the centers of the galaxies, while having tens of spaxels in the outskirts.

Figure 4 .
Figure 4. Spatially resolved stellar properties.Each row represents a galaxy in our MUSE sample.Within each row, the different panels show, from left to right: normalized stellar continuum emission in the range 5300-5700 Å, stellar velocity with respect to systemic, stellar velocity dispersion, dust reddening toward the stars, and the fraction of young stars, which is defined as the sum of the weights of the stellar templates younger than 1 Gyr, divided by the sum of all the weights.

Figure 5 .
Figure 5. Spatially resolved properties of the nonoutflowing ionized gas.Each row represents a galaxy in our MUSE sample.Within each row, the different panels show, from left to right: the centroid velocity of the narrow kinematic component, the velocity dispersion, dust reddening toward the line-emitting region, and the surface brightness of the narrow Hα flux.

Figure 7 .
Figure 7. Spatially resolved properties of the NaID absorption in J022912.From left to right: the best-fitting absorption optical depth of the NaID k doublet component, the best-fitting gas covering factor, the centroid velocity of the NaID absorption with respect to systemic, and a comparison between the centroid velocity of the NaID absorption and the stars.Blueshifted NaID absorption is considered an outflow if it is blueshifted with respect to the stellar velocity by more than 100 km s −1 .

Figure 9 .
Figure 9. Neutral and ionized outflow extents in J022912.The figure shows the distribution of the narrow-line-emitting gas (gray contours), the nuclear-ionized outflow (black contours), the blueshifted NaID absorption (blue contours), and redshifted emission (red contours) that are associated with the neutral outflow.The blueshifted NaID absorption and redshifted NaID emission show up in distinct regions in the galaxy.

Figure B1 .
Figure B1.Line diagnostic diagrams for the ionized gas.Each row represents a different galaxy.The first two columns show the narrow kinematic component, and the last two show the wings of the broad kinematic component.The first panel shows the location of different spaxels on the line diagnostic diagram [N II]/Hα vs. [O III]/ Hβ (Baldwin et al. 1981; Veilleux & Osterbrock 1987).We mark the two separating criteria that are used to separate star-forming from AGN-dominated galaxies (Kewley et al. 2001; Kauffmann et al. 2003b; Ke01 and Ka03 respectively), and the LINER-Seyfert separating line from (Cid Fernandes et al. 2010, CF10).The second panel shows the spatial distribution of the classification.For J020022 and J111943, the [O III] is masked out by the reduction pipeline since it coincides with the WFM-AO lasers sodium lines.We therefore use the SDSS [O III]/Hβ and show the variation of the [N II]/Hα ratio.
Figure B1.Line diagnostic diagrams for the ionized gas.Each row represents a different galaxy.The first two columns show the narrow kinematic component, and the last two show the wings of the broad kinematic component.The first panel shows the location of different spaxels on the line diagnostic diagram [N II]/Hα vs. [O III]/ Hβ (Baldwin et al. 1981; Veilleux & Osterbrock 1987).We mark the two separating criteria that are used to separate star-forming from AGN-dominated galaxies (Kewley et al. 2001; Kauffmann et al. 2003b; Ke01 and Ka03 respectively), and the LINER-Seyfert separating line from (Cid Fernandes et al. 2010, CF10).The second panel shows the spatial distribution of the classification.For J020022 and J111943, the [O III] is masked out by the reduction pipeline since it coincides with the WFM-AO lasers sodium lines.We therefore use the SDSS [O III]/Hβ and show the variation of the [N II]/Hα ratio.

Figure B2 .
Figure B2.Broad-line diagnostic diagrams for J022912.Top row from left to right: (i) surface brightness of the broad Hα wings (gray), with red and blue wings shown separately with contours, (ii) distribution of the flux in the wings, (iii) distribution of E B−V measured toward the narrow lines (gray), broad wings (black), and red and blue wings, with dashed lines showing median values, and (iv) distribution of the electron density.Second row from left to right: (i) velocity of the broad kinematic component with respect to systemic, (ii) + (iii) maximum outflow velocity in the red and blue wings, and (iv) a comparison between the two.Third row: (i) the location of different spaxels on the line diagnostic diagram [N II]/Hα vs. [O III]/Hβ (Baldwin et al. 1981; Veilleux & Osterbrock 1987; separating criteria: Kewley et al. 2001; Kauffmann et al. 2003b; Ke01 and Ka03 respectively), (ii) and spatial distribution of the classification.Fourth row: (i) velocity dispersion of the broad kinematic component, (ii) distance of each spaxel vs. the velocity dispersion, and (iii) + (iv) the [N II]/Hα and [S II]/Hα broad-line ratios vs. the velocity dispersion.The black and blue points represent two classes of spaxels showing different behavior in wind velocity and velocity dispersion.

Figure B3 .
Figure B3.Broad-line diagnostic diagrams for J080427.Top row from left to right: (i) surface brightness of the broad Hα wings (gray), with red and blue wings shown separately with contours, (ii) distribution of the flux in the wings, (iii) distribution of E B−V measured toward the narrow lines (gray), broad wings (black), and red and blue wings, with dashed lines showing median values, and (iv) distribution of the electron density.Second row from left to right: (i) velocity of the broad kinematic component with respect to systemic, (ii) + (iii) maximum outflow velocity in the red and blue wings, and (iv) a comparison between the two.Third row: (i) the location of different spaxels on the line diagnostic diagram [N II]/Hα vs. [O III]/Hβ (Baldwin et al. 1981; Veilleux & Osterbrock 1987; separating criteria: Kewley et al. 2001; Kauffmann et al. 2003b; Ke01 and Ka03 respectively), (ii) and spatial distribution of the classification.Fourth row: (i) velocity dispersion of the broad kinematic component, (ii) distance of each spaxel vs. the velocity dispersion, and (iii) + (iv) the [N II]/Hα and [S II]/Hα broad-line ratios vs. the velocity dispersion.

Figure B4 .
Figure B4.Broad-line diagnostic diagrams for J112023.See the caption of Figure B3 for details about the panels.The stellar rotation axis marked in the leftmost panel in the first row is based on the stellar velocities measured within ∼7 kpc of the primary galaxy.Outside this region, J112023 shows disturbed stellar kinematics.

Figure C1 .
Figure C1.Examples of NaID profile fitting for J022912.Each row represents a different binned spectrum.The left column shows the observed flux (black) and the best-fitting stellar model by PPXF around the stellar MgIb absorption complex.The good correspondence between the observed spectrum and the stellar model suggests that the excess NaID absorption/emission is not due to a bad stellar fit nor of a stellar origin.The right column shows the observed flux (black) around the NaID region, along with the best-fitting model.The full fit is shown in red.The separate model components are also shown: HeI emission (dotted blue), NaID emission (dashed purple), and NaID absorption (dashed orange).The full fit also depends on the best-fitting stellar model, shown in green.

Figure C2 .
Figure C2.Comparison between the ionized and neutral outflow of J022912.The top row compares the blueshifted absorption EW(NaID) to the blue wing of the broad Hα emission (left) and the redshifted NaID flux to the red wing of the broad Hα emission (right).The bottom panel compares the derived outflow velocities, where the left panel compares the blueshifted NaID absorption velocity to the velocity of the blue Hα wing, and the right panel compares the redshifted NaID emission velocity to the velocity of the red Hα wing.

Figure C3 .
Figure C3.Properties of the neutral outflow.Each row represents a different galaxy.The first two columns show the hydrogen column density and the adopted velocity of the outflow traced by blueshifted NaID absorption.The second two columns show the hydrogen column density and the adopted velocity of the outflow traced by the redshifted NaID emission.The gray contours represent the extent of the narrow-line-emitting ionized gas, and the black represents the ionized outflow.For J020022 and J111943, neither NaID absorption nor emission is detected, and show the upper limits on these properties.

Table 2
Stellar and Gas Morphologies