Near-Infrared Coronal Line Observations of Dwarf Galaxies hosting AGN-driven Outflows
Thomas Bohn, Gabriela Canalizo, Sylvain Veilleux, Weizhe Liu
DDraft version February 18, 2021
Typeset using L A TEX twocolumn style in AASTeX63
Near-Infrared Coronal Line Observations of Dwarf Galaxies hosting AGN-driven Outflows
Thomas Bohn , Gabriela Canalizo , Sylvain Veilleux ,
2, 3 and Weizhe Liu University of California, Riverside, Department of Physics & Astronomy900 University Ave., Riverside, CA 92521 Department of Astronomy, University of Maryland, College Park, MD 20742, USA Joint Space-Science Institute, University of Maryland, College Park, MD 20742, USA (Received January 4, 2021; Revised January 22, 2021; Accepted February 16, 2021)
Submitted to ApJABSTRACTWe have obtained Keck NIR spectroscopy of a sample of nine M (cid:63) < M (cid:12) dwarf galaxies toconfirm AGN activity and the presence of galaxy-wide, AGN-driven outflows through coronal line(CL) emission. We find strong CL detections in 5/9 galaxies (55%) with line ratios incompatible withshocks, confirming the presence of AGN in these galaxies. Similar CL detection rates are found inlarger samples of more massive galaxies hosting type 1 and 2 AGN. We investigate the connectionbetween the CLs and galaxy-wide outflows by analyzing the kinematics of the CL region, as well asthe scaling of gas velocity with ionization potential of different CLs. In addition, using complementaryKeck KCWI observations of these objects, we find that the outflow velocities measured in [Si VI ] aregenerally faster than those seen in [O III ]. The galaxies with the fastest outflows seen in [O
III ] also havethe highest [Si VI ] luminosity. The lack of J -band CN absorption features, which are often associatedwith younger stellar populations, provides further evidence that these outflows are driven by AGN inlow mass galaxies. Keywords: galaxies: active — galaxies: dwarf — galaxies: evolution — galaxies: Seyfert — infrared:galaxies — AGN: low-mass INTRODUCTIONIt is now well accepted that supermassive black holes(SMBHs) lie at the center of most massive galaxies.Strong correlations have supported the idea that theyevolve with their host galaxies and feedback from theAGN has been shown to regulate star formation (fora review, see Kormendy & Ho 2013). Powerful, AGN-driven outflows are believed to suppress star formation(Rupke et al. 2017; U et al. 2019) in high mass galaxies,leading their hosts to the well-defined red sequence.The general consensus regarding feedback in dwarfgalaxies (M (cid:63) < M (cid:12) ) is that stellar processes, suchas starbursts and supernovae, provide the main sourceof quenching (eg., Veilleux et al. 2005, 2020; Heckman& Thompson 2017). However, the growing rate of AGNdetections in dwarf galaxies (Reines et al. 2013; Moran [email protected] et al. 2014; Sartori et al. 2015) necessitates a closer lookat AGN feedback in the low-mass regime. Evidence ofAGN-driven feedback in dwarf galaxies is already start-ing to emerge. Penny et al. (2018) present a sample ofdwarf galaxies with AGN line ratios and kinematicallydisturbed gas at their centers, possibly due to AGN feed-back. In addition, Bradford et al. (2018) found a sampleof isolated ( > a r X i v : . [ a s t r o - ph . GA ] F e b Bohn et al. dwarf galaxies with optical and IR signatures suggestiveof AGN activity. Nine of these show fast outflows withmedian speed of ∼
645 km s − , higher than the medianhost escape velocities of ∼
300 km s − . If the outflows areof AGN origin, this would suggest that AGN feedbackcould eject material beyond the dark matter halo andhave a substantial impact on the star formation rate.Furthermore, follow-up integral field spectroscopy doneby Liu et al. (2020, hereafter L20) of this sample showspatially extended outflows up to 3 kpc in a number ofthese targets. In addition, they detect outflow velocitiesgreater than 500 km s − in six of these galaxies. In-deed, a small but non-negligible fraction (up to 6%) ofthe ionized outflowing gas has the necessary speeds toescape their galaxy.However, line ratios from star-forming processes cansometimes mimic those of AGN and some contaminationexists between the AGN and star-forming regions of theBaldwin, Phillips & Terlevich (hereafter BPT; Baldwinet al. 1981) diagram (eg., Allen et al. 2008; Rich et al.2011). Moreover, L20 found core-collapse supernovaeto be energetic enough to drive the mass outflow rates(3 × − — 0.3 M (cid:12) yr − ) seen in their sample andthus the contribution of stellar processes in driving theoutflows can not be formally ruled out.To further characterize the kinematics of outflows,coronal lines (CLs) have been used as an additionaltracer of outflows. CLs are forbidden transitions fromhighly ionized ( >
100 eV) species with widths suggestingthe coronal line region lies between the broad and nar-row line regions (De Robertis & Osterbrock 1984, 1986;Penston et al. 1984; Erkens et al. 1997; Rodr´ıguez-Ardilaet al. 2002, 2006). Due to the high energies needed fortheir ionization, CLs are excellent indicators of AGNactivity. They are often observed blueshifted relative tothe systemic velocity of the host galaxy and thus are be-lieved to be linked with outflows. M¨uller-S´anchez et al.(2011) have measured outflow speeds upwards of 1500km s − through CL emission and the velocity fields sug-gest the outflows are of biconical shape, with collimationlikely due to the torus (see Standard Model, Antonucci1993).In this article, we present follow-up NIR spectroscopyof the nine dwarf galaxies from MK19 and L20 that showoptical AGN line ratios and fast outflows. Through NIRdiagnostics, we aim to confirm the presence of AGN ac-tivity and characterize the outflows through NIR emis-sion lines. Details of the sample selection, observations,and data reduction are summarized in Section 2. Anal-ysis of the data, including spectra fitting and AGN di-agnostics are covered in Section 3. In Section 4, we dis-cuss the NIR emission lines and outflow characteristics. Throughout this article, we adopt a standard ΛCDMcosmology with H = 70 km s − Mpc − , Ω M = 0.3,and Ω Λ = 0.7. DATA AND OBSERVATIONS2.1.
Sample Selection
Our sample of nine dwarf galaxies comes from theAGN sample of MK19. Briefly, Reines et al. (2013),Moran et al. (2014), and Sartori et al. (2015) have iden-tified hundreds of dwarf galaxies with optical and IRsignatures indicative of AGN activity. MK19 createda subsample of candidate AGN whose optical line ra-tios place them above the star-forming region of theBPT and Veilleux & Osterbrock 1987 (hereafter VO87;Veilleux & Osterbrock 1987) line ratio diagrams or thathave He II λ III ] λ < log(M ∗ / M (cid:12) ) < < Observations and Reductions
NIR spectroscopy was obtained on three separatedates: on 2017-10-29 using Keck II NIRSPEC (McLeanet al. 1998), and on 2018-10-25 and 2019-01-25 withKeck II NIRES (Wilson et al. 2004). NIRSPEC is a NIRechelle spectrograph with a wavelength coverage from0.9 — 5.5 µ m. The NIRSPEC-7 filter was used in low-resolution mode with a cross-dispersion angle of 35.31degrees. This resulted in a wavelength coverage of ∼ µ m. The 42 (cid:48)(cid:48) × . (cid:48)(cid:48) slit was used and a spectralresolution of 196 km s − (R ≈ µ m was mea-sured with a seeing of ∼ (cid:48)(cid:48) . Observations throughoutthe night were done under mostly clear conditions. Notethat these observations were done before the NIRSPECupgrade. NIRES is a NIR echelette spectrograph withthe slit being 18 (cid:48)(cid:48) × . (cid:48)(cid:48) and the wavelength coverageset from 0.94 — 2.45 µ m across five orders. There isa small gap in coverage between 1.85 and 1.88 µ m, butthis is a region of low atmospheric transmission. Theaverage spectral resolution of the five orders range be-tween 84 — 89 km s − (R ≈ GN Outflows in Dwarf Galaxies Table 1.
Observation LogGalaxy Date Redshift Exp. Time a Slit PA Ext. Ap. S/N b Airmass Telluric(YYYY-mm-dd) (degrees) (arcsec)SDSS J010005.93-011058.89 c d d d d d a Exposures were typically done in ABBA nodding. b Average continuum signal-to-noise ratio across all orders. c SDSS J0100-0110 was observed with NIRSPEC. All other targets were observed with NIRES. d Indicates galaxies with coronal line detections. See Section 3.3. for all sets of observations were four minutes each andwere done using the standard ABBA nodding. A telluricstandard star, typically of A0 spectral class with mea-sured magnitudes in J , H , and K -bands, was observedeither directly before or after the target galaxy to cor-rect for the atmospheric absorption features. Typicalairmass differences with the target were below 0.10. Asummary of the NIR observations is shown in Table 1.In addition to the Keck NIRES and LRIS observa-tions, follow-up optical IFU observations with KeckKCWI (Morrissey et al. 2018) and Gemini GMOS(Allington-Smith et al. 2002; Gimeno et al. 2016) weredone to obtain high spatial resolution of the outflows.The details of this analysis are discussed in L20.Four of our targets (J0811+2328, J0906+5610,J0954+4717, and J1005+1257) were observed with the Chandra X-ray Observatory . Baldassare et al. (2017) re-port hard X-ray emission that is likely originating fromthe AGN in J0906+5610 and J0954+4717. Additionally,Wang et al. (2016) provide fluxes, corrected for galac-tic absorption, for J0811+2328 and J1005+1257, fromwhich we calculated a luminosity using the cosmologylisted above. We discuss these results in Section 4.1.The data were reduced using two modified pipelines.The first provided flat fielding and a robust backgroundsubtraction by using techniques described in Kelson(2003) and Becker et al. (2009). In short, this routinemaps the 2D science frame and models the sky back-ground before rectification, thus reducing the possibilityof artifacts appearing due to the binning of sharp fea- tures. The sky subtraction attained with this procedureis excellent, despite the strong OH lines present in theNIR; the procedure is also quite insensitive to cosmicrays and hot pixels, and is reliable regardless of skylineintensity.Rectification, telluric correction, wavelength calibra-tion, and extraction were all done with a slightly mod-ified version of
REDSPEC . Telluric correction wasdone by dividing by the spectrum of the telluric stan-dard star and multiplying by a blackbody curve of thesame temperature. Strong OH skylines were used forwavelength calibration and the 1D spectra were thenmedian combined. Flux calibration of individual expo-sures was done using the telluric star and the SpitzerScience center unit converter to convert the magnitudeof the star to the associated flux in that band. A smallcorrective factor ( < ANALYSIS3.1.
Spectral Fitting
NIR Fitting
We fit all NIR spectra using emcee , an affine invari-ant Markov Chain Monte Carlo (MCMC) ensemble sam-pler (Foreman-Mackey et al. 2013). A narrow Gaussian http://ssc.spitzer.caltech.edu/warmmission/propkit/pet/magtojy/ Bohn et al. component, along with a second order polynomial forthe continuum, were fit simultaneously for each emissionline. We determined whether a second Gaussian compo-nent was needed to fit the emission lines using the fol-lowing F -test: F = ( σ single ) / ( σ double ) , where σ is thestandard deviation of the residuals using either single ordouble Gaussian components. If F > σ detectionsand those with a FWHM greater than the resolutionelement.As discussed in Section 4.1, we also use NIR absorp-tion features in our analysis. These were also fit withMCMC in a similar fashion, the main difference beingthe amplitude was restricted to be negative. The fittingwas done simultaneously with that of emission lines inorder to keep the continuum level consistent across allmeasurements. Although we ran the F -test as definedabove, only one gaussian was needed for all the absorp-tion fits. The widths and depth of the absorption fea-tures used in our analysis are also listed in AppendixA. 3.1.2. SDSS Fitting
SDSS spectra are available for our entire sample andthese provide full wavelength coverage from 4000 ˚A to9000 ˚A, which includes the [O II ] λλ BADASS, Sexton et al. 2020), a spec-tral analysis tool that fits the stellar and Fe II features.Absorption features were fit using the penalized PixelFitting ( pPXF ; Cappellari & Emsellem 2004) and Fe II emission was fit using Fe II templates. All of these com-ponents were fit simultaneously, allowing for a detailedand robust analysis of the spectrum.The code allows the user to test for the presence ofoutflows by setting various constraints on parameterssuch as minimum amplitude, minimum width, and ve-locity offset. The profile of [O III ] λ III ] https://github.com/remingtonsexton/BADASS2 ∼ mxc/software/ λ > F -test model comparison in-cluded in the code. J1009+2656 is a little more uncer-tain with a 89% confidence. We do not detect outflows(confidence < Extinction
We note that L20 reports extinction values calcu-lated from the H γ /H β Balmer decrement. However, forJ0100-0110 and J0811+2328, they used H α /H β mea-surements from SDSS due to weak H γ emission in theirspectra. In addition, they did not observe J1442+2054.We thus opted to use the H α /H β Balmer decrementmeasured from SDSS for our entire sample.To quantify the extinction, we used the intrinsic lineratio of H α /H β = 3.1, typically used for AGN (eg.,Veilleux & Osterbrock 1987; Osterbrock et al. 1992), anda Cardelli reddening law (Cardelli et al. 1989) with anextinction factor of R V = 3.1. We used the narrow-line flux measurements from the SDSS data (see Sec-tion 3.1.2), where we have decomposed the H α and H β emission into narrow and broad (outflow) components.For the two galaxies, J0840+1818 and J1442+2054,where no outflow component was detected, we used thefull emission line to obtain a flux. These values ofthe Balmer decrement and E(B-V) for each galaxy arelisted in Table 2 and all flux values in Appendix A re-flect extinction corrected fluxes. Note that three galax-ies, J0840+1818, J0842+0319, and J0906+5610, haveBalmer decrements slightly below the intrinsic ratio sowe did not apply any extinction correction to them.Our results are slightly different but generally agreewith those calculated by L20. The discrepancies arelikely caused by the different line ratios (H γ /H β versusH α /H β ) and line profiles used, where we only used thenarrow profile while L20 used the full (narrow + broad)profile. When comparing the effect of these two meth-ods, the difference in flux measurements amounts to <
5% for the majority of our sample. For the rest of thisarticle, we use these extinction corrected flux values un-less otherwise specified.3.3.
Coronal Line Detections
55% (5/9) of our sample have NIR CL emission withinthe spectral window of 0.94 — 2.45 µ m (see Table 1).For a similar spectral window and galaxy type, Sy1 and GN Outflows in Dwarf Galaxies N o r m a li z e d F l u x J09541.001.25 J10050.985 0.990 0.995 1.0001.001.25 J1009 1.245 1.250 1.255 1.260 1.42 1.43 1.44
Rest-frame Wavelength ( m)
Figure 1.
Zoom-in plots of the spectral regions around the most prominent NIR CLs, where the flux has been normalizedto unity and the systemic redshift was used to shift the spectra to rest-frame wavelength. Dotted blue (detections) and red(non-detections) indicate the rest-frame wavelength of CLs. MCMC fits to the emission and absorption line profiles are shownas solid green lines.
Table 2.
Measured Extinction Values of the SampleGalaxy H α/ H β Balmer Decrement E(B-V)J0100–0110 3.47 ± ± ± ± ± ± ± ± ± Note —An intrinsic ratio of H α/ H β = 3.1 and aCardelli reddening law were used. Sy2, this rate is consistent with others found in the liter-ature: 66% (36 out of 54, Rodr´ıguez-Ardila et al. 2011),25%(5 out of 20, Mason et al. 2015), and 43% (44 outof 102, Lamperti et al. 2017). Our detections predom-inately come from sulfur and silicon species: [S
VIII ]0.9913 µ m, [S IX ] 1.2523 µ m, [Si X ] 1.4305 µ m, and[Si VI ] 1.9630 µ m. We also detect [Ca VIII ] 2.3214 µ mbut it falls within the CO(3-1) absorption band at 2.3226 µ m, making their measurements more uncertain. All of the detected NIR CLs are shown in Figure 1, where weinclude our MCMC fits to the emission line profiles. Wedo not detect other common NIR CLs, such as [Fe XIII ]1.0747 µ m, [S XI ] 1.9196 µ m, and [Al IX ] 2.0450 µ m.The most prominent CL is [Si VI ] and is detected infour of the five galaxies with CL detections. This is notsurprising since it has a lower IP level than other CLsand is not near any absorption features. In three of thesefour galaxies, we find broad [Si VI ] emission and thus usea two component fit to the profile, in accordance withthe F -test mentioned in Section 3.1.1. Details of themulti-component fits to [Si VI ] and plots of the entirespectra are presented in Appendix A.No NIR CL emission is detected in the other fourgalaxies. To investigate this, we plot total [Si VI ] fluxversus WISE W µ m) flux in Figure 2. Note thatone galaxy, J0842+0319, has a CL detection, [Ca VIII ],but does not show any [Si VI ] emission. Upper limitfluxes to the non-[Si VI ] detections were calculated byintegrating over a Gaussian with a width equaling theresolution element and amplitude equalling the 1 σ noiselevel where [Si VI ] should appear. This value was multi-plied by three to obtain the 3 σ upper limit that is plottedin Figure 2. We also plot the AGN sample (black dots)from M¨uller-S´anchez et al. (2018), from which we de-rive the plotted line of best fit. This relation (J. Cann,private communication) provides an expected value for Bohn et al. log ((W2 4.6 m) l o g (([ S i V I ] . m ) J0906+56J0954+47 J1005+12J1009+26J0100-01 J0811+23J0840+18J0842+03 J1442+20
CL and [Si VI] detectionCL detectionnon-CL detectionMS'18
Figure 2. [Si VI ] flux vs W2 (4.6 µ m) flux. Galaxies inour sample with detected [Si VI ] emission are shown as bluestars. J0842+0319, the one object with a CL detection butno [Si VI ] emission, is shown as a purple cross. Objectswithout any CL detections are shown as red crosses. Forthese latter two, the crosses represent 3 σ upper limit fluxesto [Si VI ]. The solid line is the best-fit line to the data foundin M¨uller-S´anchez et al. (2018). [Si VI ] based on W VI ] measurements, suggesting thatdeeper observations may be required to detect any CLemission. Further discussion on our non-detections iscovered in Section 4.1.The optical data also reveal a number of CLs, par-ticularly in those with NIR CL detections. 55% (5/9)of our sample have strong optical CL emission. Themost common detections in our sample include [Ne V ] λ VII ] λ X ] λ V ] λ VII ] λ V ] λ AGN Diagnostic Plots
Although there is evidence for AGN activity in ourdwarf galaxy sample, including optical BPT/VO87AGN line ratios, the addition of NIR lines allows usto run additional AGN diagnostics. With the inclu- sion of [S
III ] λ III ] λλ α versus[S II ] λλ α , [S III ] λλ α versus[O II ] λλ α , and [S II ] λλ α versus [O II ] λλ α (their Figures 4, 5 and7). All three of these relations are plotted in Figure3, where we include the AGN and star-forming samplesfrom Osterbrock et al. (1992). We overplot narrow-linefluxes of our sample as stars and exclude the galaxieswith no measurable [O II ] λλ III ] λ III ] λ III ] λ λ III ] λ DISCUSSION4.1.
Detection Rate of Coronal Lines
Although CLs are excellent tracers of AGN activity,they do not always appear in AGN spectra and a num-ber of reasons have been proposed to explain this. Ingeneral, CL detections decrease with increasing IP andour results agree with this; we do not detect any of thehigh-IP CLs, [Fe
XIII ] 1.0747 µ m (330.8 eV) and [S XI ]1.9196 µ m (447.1 eV). Stellar absorption features canalso affect detection rates by attenuating the CL emis-sion profile. Such is the case with [Ca VIII ] which fallswithin the CO (3-1) absorption feature. In addition, Caand Al species can be affected by metallicity and deple-tion onto dust. Moreover, telluric absorption from theatmosphere always has to be contended with in ground-based observations.In nearby low-luminosity AGN, circumnuclear stellarpopulations can dominate the NIR continuum and thusdrown out any CL emission. Bright AGN, particularlyat high redshift, can also hamper CL emission due totheir strong continuum. These effects can be seen inthe results of Rodr´ıguez-Ardila et al. (2011) who foundthat many galaxies with stars contributing ∼
90% of the
GN Outflows in Dwarf Galaxies log ([S II] 6718 + 6732/H ) l o g ([ S III ] + / H ) Sy2Sy1.5NLSy1LINERSBHIICL Detectionnon-CL Detection log [O II] 7320 + 7330/H ) l o g ([ S III ] + / H ) Sy2Sy1.5NLSy1LINERSBHIICL Detection log ([O II] 7320 + 7330/H ) l o g ([ S II ] + / H ) Sy2Sy1.5NLSy1LINERSBHIICL Detection
Figure 3.
AGN diagnostic plots derived by Osterbrock et al. (1992). We plot our sample as blue (CL detection) and red(non-CL detection) stars. Our sample falls within the scatter of other galaxies hosting AGN, further confirming the presence ofAGN. continuum do not show CL emission. They also foundtheir CL detection rate decreased by 17% when selectinggalaxies with a redshift > J , H , and K -bands allows us to es-timate the contribution of circumnuclear stellar popu-lations. The CO(6-3) absorption at 1.62 µ m can pro-vide an estimate to the flux contribution of red giantsto the H -band continuum (Martins et al. 2010), wherethe continuum level arises due to stellar and AGN con-tribution. For a population of GKM giants, the typicalobserved depth of the absorption is ∼
20% of the contin-uum (Schinnerer et al. 1998). Our sample ranges from7% - 14% (median of 11%), suggesting a large contri-bution to the H -band continuum from red giants, upto 70%. The weighted average of the depth of galaxieswith no CL detections is 28% deeper than those withCL emission. One possibility is that a larger populationof red giants near the center could be the cause of thedeeper absorption, and are thus directly increasing thecontinuum level. Alternatively, a shallower CO absorp-tion may be indicative of a stronger AGN contributionto the continuum, which would likely lead to strongerCL emission that is more easily detectable.Baldassare et al. (2017) report 0.5-7 keV X-ray lumi-nosities for J0906+5610 and J0954+4717, from whichwe convert to L − using PIMMS . This results in a L − of 2.89 × erg s − and 6.12 × erg s − for J0906+5610 and J0954+4717, respectively. Addi-tionally, using 0.3-8 keV fluxes from Wang et al. (2016),we calculated 2-10 keV luminosities for J0811+2328 https://cxc.harvard.edu/toolkit/pimms.jsp (1.33 × erg s − ) and J1005+1257 (5.20 × ergs − ). Although J0811+2328 does have the lowest X-rayluminosity, we find no clear distinctions between galax-ies with CL detections and those without. We do note,however, that these X-ray luminosities are about twoorders of magnitude lower than the L − – L W2 re-lation discussed in Secrest et al. (2015). Indeed, all fourtargets have L − / L W2 < -2.1. Similar low L X havebeen reported in low-mass galaxies (Dong et al. 2012;Simmonds et al. 2016; Cann et al. 2020), suggesting ob-scuration of the X-ray source emission or that AGN indwarf galaxies are X-ray weak compared to their MIRemission.4.2. Coronal Line and Outflow Kinematics
Photoionization is widely considered as the mainexcitation mechanism behind CL emission, althoughRodr´ıguez-Ardila et al. (2006) found that their modelsmore precisely matched emission line ratios from theirdata when shocks were included. If, however, photoion-ization is the principle excitation mechanism behind CLsthen we expect a correlation between the FWHM andIP. Emission from a high IP line would suggest thatthe emitting gas is located closer to the central ioniz-ing source and thus be deeper in the gravitational well,causing a broadening of its emission line profile. Be-cause of their high range of IPs ( ∼
100 eV – 500 eV),CLs are ideal in investigating the gas kinematics near theAGN. Indeed, positive correlations between line widthand IP have been found in past studies (eg., De Robertis& Osterbrock 1984; De Robertis & Shaw 1990; Veilleux1991a) for optical high ionization lines. A positive trendhas also been found between line width and critical den-
Bohn et al.
Table 3.
Emission Lines with IP and Critical DensitiesEmission Line IP log( n e ) Emission Line IP log( n e ) µ m eV cm − µ m eV cm − H α sity in these studies. Similar trends between line widthand IP for NIR lines have been seen in some galaxies(Rodr´ıguez-Ardila et al. 2002) but larger samples arestarting to show more varied CL widths (eg., Rodr´ıguez-Ardila et al. 2011; Villar Mart´ın et al. 2015; Cerqueira-Campos et al. 2020), and in many cases no trends arefound at all. Rodr´ıguez-Ardila et al. (2011) find a pos-itive slope up to ∼
300 eV, after which the slope turnsnegative (i.e. higher IP, lower FWHM). They attributethis to the increase in electron density when approachingthe central AGN. Due to densities exceeding the criti-cal densities of the high IP CL ions, these lines may besuppressed. Specifically, collisional de-excitation couldreduce emission associated with the broader, high veloc-ity components, thus causing us to only see the narrowemission and explaining the decrease in FWHM at highIPs.We run a similar analysis by plotting narrow-component FWHM versus IP and critical density for theCLs detected in our sample in Figure 4. We have alsoincluded optical (open circles) and NIR low-ionizationlines to draw comparisons (see Table 3). In two galaxies,J0906+5610 and J1005+1257, the widths peak around250 – 300 eV and subsequently decrease, consistent withcollisional de-excitation of the broader components ofhigh IP ions. The CL emission in the other three galax-ies have lower S/N so it is difficult to evaluate anytrends. Aside from these low S/N measurements (asindicated by their error bars), these CLs have widthsconsistent with the lower IP lines. However, some uncer-tainty may arise from overestimating/underestimatingthe broad/narrow component of the line profile. Forinstance, the relatively small width of [Si VI ] (166.8eV) in J0954+4717, where we have added a secondarybroad component, could be due to overestimating thebroad component. Regardless, our results are consistentwith the recent reports finding no clear trends betweenFWHM and IP. Critical density, defined as the density when the col-lision rate matches the radiative de-excitation rate, isplotted against FWHM in the bottom panels of Figure 4.Consistent with past works (De Robertis & Osterbrock1986; Veilleux 1991b; Ferguson et al. 1997), we find lin-ear correlation in some galaxies, with varying degreesof slope. Regardless, we find a stronger correlation ofFWHM increasing with critical density, consistent withpast studies.Table 4 lists the kinematic properties of the narrowand broad (outflow) components of [O III ] and [Si VI ],where the [O III ] values come from L20. The higherspectral resolution of the IFU data of L20 allowed forthe decomposition of the emission line profile into two orthree components. Generally, the C1 component tracesthe gas in the narrow-line region while C2 and C3 rep-resent the broader and bluer components to the multi-component fit, and thus likely trace the outflow. ForJ0811+2328, only one component (C1) was fit but it isblueshifted relative to the stellar velocity and broaderthan the stellar velocity dispersion, and thus L20 sug-gest that it is likely part of the outflow. For v , themedian velocity offset of the profile relative to systemicvelocity, we take the minimum values (i.e. maximumblue offset) for [O III ].The velocity of an outflow is often calculated throughthe use of W , the width containing 80% of the fluxof an emission line (Harrison et al. 2014). For a singleGaussian profile, W = 1.09 × FWHM. To measure W for [Si VI ] , we use the full profile due to the relativelylarge uncertainties in the broad component. We candefine the outflow velocity as, v out = − v + W out calculations forall [O III ] components. In addition, for [Si VI ] v out , weuse v of the full profile if a multi-component fit wasused. This is because of the large uncertainties associ-ated with the broad component fits.Our data show the outflow velocities seen in [Si VI ],measured through either W or v out , are generallyfaster than those seen in [O III ]. The main exception tothis are the velocities seen in the C3 components. Thisis likely due to combining the narrow and broad com-ponents of [Si VI ] when calculating W and v . Usingonly the broad component, we find velocities to be con-sistent or higher than those of C3, albeit with muchhigher uncertainty. This overall trend of higher veloc- GN Outflows in Dwarf Galaxies IP (eV) F W H M ( k m s ) J0842
IP (eV)
J0906
IP (eV)
J0954
IP (eV)
J1005
IP (eV)
J1009 log( n e ) (cm ) F W H M ( k m s ) J0842 log( n e ) (cm ) J0906 log( n e ) (cm ) J0954 log( n e ) (cm ) J1005 log( n e ) (cm ) J1009
Figure 4.
FWHM versus IP (top panels) and critical density (bottom panels) for our sample with NIR CL detections. Low-ionization optical emission lines are plotted as open circles while all NIR emission lines are plotted as filled circles. ities seen in [Si VI ] than [O III ] implies a deceleratingoutflow, one where a high velocity wind originates nearthe AGN and slows down as it approaches the outer,lower ionization gas.We also note that the objects with the fastest out-flows and broadest profiles seen in [O
III ], J0906+5610,J0954+4717, and J1005+1257, have the broadest [Si VI ]emission. This is perhaps unsurprising since we are ob-serving the same outflow, just at different locations.We also list the bolometric AGN luminosities (L AGN )in Table 4. These values were derived from the observed[O
III ] λ AGN = 87 L [O III] for 38 < log(L [O III] ) <
40 and L
AGN = 142 L [O III] for 40 < log(L [O III] ) <
42. To correctfor extinction, we applied the extinction values listed inTable 2. We find no strong correlation between AGNluminosity and outflow speeds but there is a trend forthe more luminous AGN to have a higher number andmore luminous CL detections. If we assume a simple bi-conical outflow starting close to the AGN (for a detailedanalysis, see M¨uller-S´anchez et al. 2011), highly ionizedgas (i.e. CLs) could more easily be sent out to farther,possibly less obscured regions due to a fast outflow. Al-though it is difficult to make any firm conclusions due tothe CL region being unresolved in our data, it appearsthat faster outflows may result in stronger and broaderCL emission.4.3.
Ionization and Origin of the Outflows
In this section, we investigate whether AGN or stellarprocesses (or both) are the primary source of ionizationand the driving mechanism for the outflows. 4.3.1.
AGN or Stellar Ionization?
The analysis of line ratios by L20 indicates that AGNare the primary source of ionization in our sample.However, they note that ionization from shocks, pos-sibly originating from starburst driven-winds (Sharp &Bland-Hawthorn 2010), can not be ruled out. To inves-tigate this further, we compare line ratios in our sam-ple to ionization models found in the literature. Riffelet al. (2013) and Colina et al. (2015) plot [Fe II ] 1.64 µ m/Br γ versus H µ m/Br γ to separate AGN fromstar-forming samples. However, the five galaxies withobtainable line ratios fall within the overlap betweenthe AGN and SNe-dominated distributions (see Figure5 from Colina et al. 2015). We instead use the flux ratiosof [Si VI ]/Br γ and [Fe II ] 1.64 µ m/Br γ , where [Fe II ] isa sensitive indicator of shocks (U et al. 2013) and Br γ serves as an indicator of stellar activity and the ionizingradiation field. We compare these ratios to the AGN ion-ization models from Groves et al. (2004a,b) and shockmodels from Allen et al. (2008). We extracted thesemodels from the ITERA library (Groves & Allen 2010)and plotted them in Figure 5. The shock model (Figure5, left) takes into account both the shocked gas and pre-cursor gas, which lies ahead of the shock front. Quick in-spection of shock-only model (not plotted) shows a shifttowards higher [Fe II ]/Br γ ratios, farther away from ourmeasured line ratios. We thus focus our analysis onthe shock + precursor model (hereafter simply shockmodel) since the individual regions cannot be resolved inour data. Free parameters for the shock model includethe shock velocity v shock (10 – 1000 km s − ) and themagnetic field parameter B/n / (10 − – 10 µ G cm / ),where B is the transverse magnetic field. The AGN0 Bohn et al.
Table 4.
Outflow PropertiesGalaxy [O III] [O III] v [Si VI] v [O III] W (v out ) [Si VI] W (v out ) log(L [SiVI] ) log(L AGN )Component (km s − ) (km s − ) (km s − ) (km s − ) (erg s − ) (erg s − )(1) (2) (3) (4) (5) (6) (7) (8)J0100–0110 Total -130 — 440 (350) — < < < < Note —Columns: (1) Galaxy name. (2) Components of the [O
III ] fit according to L20. In general, the C3 componenttraces the faster, broader outflow component while C2 traces the more narrow outflow component. C1 generally tracesthe gas of the narrow-line region. (3) Minimum values (i.e. maximum blue offset from the systemic velocity) of v basedon [O III ] λ values for [Si VI ]. The first value is for the narrow component, followedby the broad component. (5) Maximum W values based on [O III ] λ out values as defined in Equation 1. (6) W of the outflow based on the full [Si VI ] profile. v out values are inparentheses. (7) Total luminosity of [Si VI ]. Upper limits are included for galaxies without [Si VI ] detections (see Section3.3 for details). (8) Extinction-corrected AGN luminosity as derived from [O III ] λ model free parameters include the power law index α and the ionization parameter U , where U ≡ n ion /n e ,where n ion is the density of the ionizing photons and n e is the electron density. The model uses a simple powerlaw, F ν ∝ ν α , where 5eV < ν < VII ] λλλ VII ]emission lines that are measurable. For these targets, we calculated an upper limit flux to the third [Fe
VII ]line and obtained a rough estimate of the gas density.These values are consistent with a density of 1000 cm − .In addition, L20 used the S II λ II λ ∼ − for most of their sample. They thus use a den-sity of 1000 cm − in their models. Therefore, in both ofour models, we use a gas density of 1000 cm − . Lastly,metallicity was set to solar.We use the total flux for each emission line and plotthe ratios over the shock model (Figure 5, left) and AGN GN Outflows in Dwarf Galaxies J0906+56J0954+47J1005+12J1009+26 log (([Fe II] 1.644 m)/Br ) l o g (([ S i V I ] . m ) / B r ) (a) J0906+56J0954+47J1005+12J1009+26 log (([Fe II] 1.644 m)/Br ) l o g (([ S i V I ] . m ) / B r ) (b) Figure 5.
Log ([Si VI ]/Br γ ) versus log (Fe II /Br γ ) for (a) shock + precursor and (b) AGN model plots. Free parametersfor the shock + precursor gridlines are the shock velocity v shock and magnetic field parameter B. Free parameters for the AGNgridlines are the power law index α and the ionization parameter U . The four galaxies with detected [Si VI ] emission are plottedas black circles with error bars. Metallicity was set to solar and a gas density of n = 1000 cm − was assumed. model (Figure 5, right). It is clear that the line ratiosare well within the AGN model parameters (-2.0 (cid:46) α (cid:46) -1.2 and -3.3 (cid:46) log( U ) (cid:46) -2.0). As for the shockmodels, the line ratios are offset by more than 0.5 dexand all have systematically larger [Fe II ]/Br γ ratios. Inorder to match the measured data, it appears a relativelylarge magnetic parameter ( > ) is required. Thus it islikely that AGN are the dominant ionizing source in oursample, though a small contribution from shocks cannotbe formally ruled out.4.3.2. AGN or Stellar Driven Winds?
The matter of identifying the driving force behind theoutflows is a difficult but important issue to investi-gate. Outflowing winds are often cited coming from ei-ther AGN or starburst activity (eg., Fabian 2012; Rupke2018; Veilleux et al. 2020), though distinguishing be-tween the two is often difficult. L20 investigated theenergies associated with the outflows in our sample andfound that the AGN are more than powerful enough todrive them. However, they found that typical core col-lapse supernovae can also provide the necessary energyoutput needed.One avenue of checking the likelihood of starburstdriven-winds in our sample is to estimate the age ofa stellar population. The CN absorption features at1.1 µ m have been shown to serve as an indicator ofan intermediate-aged stellar population (Maraston 2005;Riffel et al. 2007). This band most notably arises fromcarbon stars where there is an excess of carbon that is not bonded in CO molecules. To form a carbon star, athird ’dredge-up’ during the thermally pulsing asymp-totic giant branch (TP-AGB) is necessary and this con-strains the stellar population age to be within 0.3 – 2Gyr. Thus CN absorption would identify a younger tointermediate aged stellar population and signify an oc-currence of starburst activity.The absence of CN absorption, however, may not nec-essarily imply an absence of a young to intermediate-aged stellar population. Riffel et al. (2009) and Martinset al. (2013) report cases with no CN absorption butwith known intermediate populations and vice versa.But as they point out, this is not to say that CN does nottrace an intermediate population. Overlap with strongemission from He I µ m and Pa γ can partially orfully obstruct the CN absorption. Telluric absorption,particularly in galaxies with low redshift (z < Bohn et al. deficit of young stars would suggest a lack of starburstthat could drive the outflows we see, making the out-flows more likely to be driven by AGN. CONCLUSIONWe have presented Keck NIR spectroscopy of a sampleof dwarf galaxies with strong evidence of AGN activityand have shown the outflows detected are likely AGN-driven. Eight galaxies were observed with Keck NIRES,providing full wavelength coverage from 0.9 – 2.4 µ m,and one galaxy with Keck NIRSPEC, with wavelengthcoverage from 1.97 – 2.39 µ m. The main results aresummarized below. • NIR CLs (IP >
100 eV) are detected in 5/9 (55%)galaxies in our sample, consistent with detection ratesfound in larger mass studies. Due to their high ioniza-tion potentials, CL emission is highly indicative of AGNactivity. Coupled with optical and other NIR AGN di-agnostics, there is strong evidence for AGN activity inthese dwarf galaxies. • For the four galaxies without NIR CL emission,we suspect a strong contribution from a population ofred giants that could be dominating the continuum leveland hamper any weak CL emission. The deeper CO(6-3) absorption at 1.62 µ m in these galaxies indicates alarger population of old stars. Alternatively, the deeperCO absorption may be indicative of weaker AGN thatwould result in more elusive CL emission. • No clear trends are found between the widths ofthe CLs and their IPs. Two galaxies show a decreasein width after peaking around 250 – 300 eV. As notedin previous works, this is likely due to collisional de-excitation caused by the high density environments nearthe central ionizing source. In addition, we see a positivetrend between the FWHM and the critical densities inthese two galaxies. The other galaxies show CL widthsthat are consistent with lower IP lines and their widthsare not as tightly correlated with critical densities. • The outflow velocities measured from [Si VI ] emis-sion are generally faster than those measured from[O III ]. This indicates the presence of a decelerating out-flow. We also find that the galaxies with the highest[Si VI ] luminosity also have the fastest outflows mea-sured in [O III ]. • Examination of ionization models reveals that NIRemission line ratios of our sample are more consistentwith AGN models than with shock + precursor models.This indicates that AGN are the main ionizing source,though a smaller contribution from shocks cannot beformally dismissed. • The lack of CN absorption at 1.1 µ m suggests alack of young/intermediate (0.3 – 2.0 Gyr) circumnu-clear stars in our sample. This goes against the scenariowhere the outflows are produced by starburst activity,suggesting AGN as the main driving force of the out-flows. ACKNOWLEDGMENTSWe thank the referee for their time and helpful com-ments on this work. We also thank Lisa Prato for herassistance with REDSPEC
GN Outflows in Dwarf Galaxies
Software:
BADASS (Sexton et al. 2020) https://github.com/remingtonsexton/BADASS2), pPXF (Penget al. 2002, 2010),
PyRAF (PyRAF is a product ofthe Space Telescope Science Institute, which is operatedby AURA for NASA),
REDSPEC A. DETAILS OF CORONAL LINE EMISSIONIn this section, we cover the NIR spectra in our sample of dwarf galaxies, with particular detail to the CL emissionand important absorption bands. The following subsections are in order of increasing RA of the galaxy. In the figuresshowing the full wavelength coverage of our data, we provide subplots that give additional detail to key features, such asabsorption features and emission lines (or lack thereof). The measured properties of every NIR emission line detectedat the 3 σ level are listed in Tables 5 through 13. The listed uncertainties come from the random error estimates duringthe fitting process. A.1. J010005.92-011058.89
J0100-0110 is one of only two galaxies in the composite region of the BPT diagram. Based on optical emission linemeasurements, Manzano-King & Canalizo (2020, hereafter MC20) report disturbed gas that is offset from the stellarrotation curve derived from a NFW dark matter density profile (Navarro et al. 1996). Indeed, L20 report strong,blueshifted [O
III ] emission in the southern portion of J0100-0110, seen predominately in the C2 component. Theysuggest this could be emission from the near side of a biconical outflow.The NIR spectrum of J0100-0110 is shown in Figure 6 and line measurements are listed in Table 5. It is the onlygalaxy in our sample observed with NIRSPEC (wavelength coverage from ∼ µ m). Because of this, wecannot make estimates to the stellar age as we have done with the rest of the sample. The spectrum shows strong Pa α and Br γ ; it shows no CL emission in K -band. No significant optical CL emission is detected in this galaxy.A.2. J081145.29+232825.72
Moran et al. (2014) report a lower limit to the BH mass of log( M BH /M (cid:12) ) > Chandra × erg s − . MC20 report both disturbed and stratified gas in J0811+2328. Here, the gas is both offset from thestellar rotation curve, and the Balmer and forbidden lines are kinematically distinct from one another. In L20, only asingle component was needed to fit the [O III ] profile but the larger line widths (relative to the velocity dispersion ofthe stellar components) and velocity offsets (up to -60 km s − ) indicate that the outflow could be traced by this singleprofile.The NIR spectrum of J0811+2328 is shown in Figure 7 and line measurements are listed in Table 6. We detect noNIR CL emission at the 2 σ level, though deeper observations could reveal faint [S VIII ], [Fe
XIII ], [S IX ], and [Ca VIII ].These lines either fall on skylines or their S/N is consistent with the noise level so we do not include them in ouranalysis. The rest of the spectrum is relatively featureless aside from a handful of typically strong NIR lines, suchas [S
III ] 0.9531 µ m and He I µ m. Unfortunately, the CN absorption at 1.1 µ m falls within heavy telluricabsorption so we cannot make an estimate on the contribution of a younger stellar population.A.3. J084025.54+181858.99
Like J0811+2328, Moran et al. (2014) also report a lower limit to the BH mass for J0840+1818, log( M BH /M (cid:12) ) > III ].Data from L20, however, show velocity offsets ranging from -30 km s − to +20 km s − and only a single Gaussian wasused in the fit. In addition, the line widths are smaller than the velocity dispersions of the stellar component. This4 Bohn et al.
Rest-frame Wavelength ( m) F ( e r g s c m s m ) J0100-0110
Figure 6.
Spectrum of J0100-0110 taken with NIRSPEC, shifted to rest-frame wavelength using the systemic redshift. Dashedred lines represent CLs that we do not detect at the 2 σ level. J0811+2328
Rest-frame Wavelength ( m) F ( e r g s c m s m ) Figure 7.
Spectrum of J0811+2328 taken with NIRES, shifted to rest-frame wavelength using the systemic redshift. Dashedred and green lines represent CLs that we do not detect at the 2 σ level and the CO (6-3) absorption at 1.62 µ m, respectively.The shaded grey regions indicate areas of significant telluric absorption. GN Outflows in Dwarf Galaxies J0840+1818
Rest-frame Wavelength ( m) F ( e r g s c m s m ) Figure 8.
Same as Figure 7 but for J0840+1818. all suggests that the gas is rotating in the same direction as the stars and they find no clear evidence of outflows intheir data.The NIR spectrum of J0840+1818 is shown in Figure 8 and line measurements are listed in Table 7. We find no CLsat the 2 σ detection level, though faint [Ca VIII ] may be present. Similar to J0811+2328, we only detect a handful ofemission lines and the 1.1 µ m CN absorption falls within significant telluric absorption. MK19 do report the opticalCL [Ne V ] λ J084234.50+031930.68
MK19 report broad H α J0842+0319, from which they calculated a BH mass of log( M BH /M (cid:12) ) = 5.84. MC20 reportboth disturbed and stratified gas, which is consistent with the findings of L20 where they find blueshifted gas up to-160 km s − .The NIR spectrum of J0842+0319 is shown in Figure 9 and line measurements are listed in Table 8. We detect[Ca VIII ] as the only NIR CL in this object. To provide a more accurate measurement of the flux and width, we includesimultaneous fits to the CO band absorption. Of the objects that show NIR CL emission, J0842+0319 has the lowestL
AGN and deepest CO(6-3) band absorption feature. Deeper observations may reveal other, more faint CLs. The restof the spectrum shows a number of hydrogen recombination, H , and He I lines, and significant CN absorption is seen,suggesting an older stellar population.6 Bohn et al.
J0842+0319
Rest-frame Wavelength ( m)
25 123060 4555 F ( e r g s c m s m ) Figure 9.
Same as Figure 7 but for J0842+1818. The dashed blue line indicates a CL with a 2 σ detection. A.5.
J090613.75+561015.5
Reines & Volonteri (2015) report a BH mass of log( M BH /M (cid:12) ) = 5.40, based on virial mass estimates. Marleauet al. (2017) also report a BH mass using bolometric luminosities from WISE IR data, from which they calculatelog( M BH /M (cid:12) ) = 6.93. Baldassare et al. (2017) report bright X-ray emission, L . − = 4.47 × erg s − , in Chandra data which is higher that what is expected from high-mass X-ray binaries. They conclude that AGN activityis the likely source of this X-ray emission. MC20 report disturbed gas and three components were used in the fitsdone by L20. Maximum widths of the third component reach up 1250 km s − , the largest seen in the sample and isindicative of a fast outflow.We also note that the C2 component of [O III ] measured in L20 is redshifted 60 km s − relative to the systemicvelocity. One explanation of this is that they are observing the far side of the outflow. Interestingly, this redshiftedcomponent is also seen in [Si VI ], where it is offset by ∼
90 km s − . Since the two values are consistent with each other,we are both likely observing the far side of the outflow.The NIR spectrum of J0906+5610 is shown in Figure 10 and line measurements are listed in Table 9. The spectrumhas an abundant number of emission lines, including strong hydrogen recombination, H , and He I lines. The CLemission is also strong, with significant emission from [Si VI ], [S IX ], and [Si X ]. We also detect weaker emission from[S VIII ] and [Ca
VIII ]. [Si VI ] is best fit with two Gaussian components (see Figure 11) while only one Gaussian wasnecessary to fit the other CLs properly. It is noteworthy that [Si VI ] falls within a region of telluric absorption andskylines. However, on closer inspection of the telluric standard star used, the degree of absorption is comparable to GN Outflows in Dwarf Galaxies J0906+5610
Rest-frame Wavelength ( m)
10 81015 1215 F ( e r g s c m s m ) Figure 10.
Same as Figure 7 but for J0906+5610. F ( e r g s c m s m ) H Rest-frame Wavelength ( m) F noise = 0.21 Figure 11.
MCMC fit to [Si VI ]1.9630 µ m and H µ m emission for J0906. The spectrum is shifted to rest-framewavelength using the systemic redshift. The purple line represents the fit to the H emission while the dotted green and bluefits represent the narrow and broad components of [Si VI ]. The bottom panels show residuals to the fits and 1 σ noise level(represented as dotted lines). The location of skylines and regions of telluric absorption near [Si VI ] are represented as grey andand yellow lines, respectively. Note that running a F test on these fits suggests that a two component fit is needed. Bohn et al. that seen redward at 1.98 µ m. Thus we expect the noise level around [Si VI ] to be at similar levels. The skylinesare also similar to those blueward and the effects can be seen in the residuals of the fit. Lastly, we do not detect CNabsorption in J -band.Inspection of the available optical data from SDSS, LRIS, GMOS, and KCWI yields a number of CLs. These include[Ne V ] λ VII ] λλλ X ] λ J095418.16+471725.1
Both Reines & Volonteri (2015) and Marleau et al. (2017) report a BH mass for J0954+4717. Using the methodsdescribed in Section A.5, they report log( M BH /M (cid:12) ) equal to 5.00 and 6.46, respectively. Similar to J0906+5610,Baldassare et al. (2017) also report strong X-ray emission, L . − = 9.77 × erg s − that is likely coming fromAGN activity. MC20 report disturbed gas and, as shown in L20, three components were needed to properly describethe profile. Both of these indicate the presence of outflows.The NIR spectrum of J0954+4717 is shown in Figure 12 and line measurements are listed in Table 10. Strongemission is seen in various hydrogen recombination, H , and He I lines. A large number of these emission lines requirea multi-component fit to properly match their profile, possibly due to the outflow. [Si VI ], plotted in Figure 13, isthe only strong CL. Inspecting the skyline immediately on top of [Si VI ], we find it to be of similar strength to thoseredward. As shown in the bottom panel, the residuals at the locations of the skylines are comparable, leading us tobelieve our fit is representative of the emission. Evaluating the emission profile of [Si VI ], we obtain a F -test value of ∼ VI ], we also detect two sulfur lines, [S VIII ] and [S IX ]. Similar to the other galaxies in our sample,we do not detect any CN absorption in J -band.Like J0906+5610, inspection of the available optical data from SDSS, LRIS, GMOS, and KCWI yields a number ofCLs. These include [Ne V ] λλ V ] λ VII ] λλ J100551.19+125740.6
Marleau et al. (2017) report a BH mass of log( M BH /M (cid:12) ) = 6.47 and MC20 find disturbed gas in their optical data.Like J0811+2328, Wang et al. (2016) report a L . − flux for J1005+1257, from which we obtain a L − of5.20 × erg s − . L20 fit three components to the [O III ] emission line profile, and the broadest component, C3,have widths up to 1200 km s − . This is the second broadest they see in the sample and suggests the presence of a fastoutflow.The NIR spectrum of J1005+1257 is shown in Figure 14 and line measurements are listed in Table 11. Like theother galaxies with CL emission, its spectrum shows a wealth of strong emission lines. Of particular note is [Si VI ],which shows a strong broad component (see Figure 15, right). Unfortunately, the emission falls in a number of smalltelluric absorption features but they don’t drastically alter the fit. The region that was most affected was the largetelluric feature redward of the [Si VI ] line. In order to not overestimate the [Si VI ] emission, particularly the broadcomponent, we masked out the entire telluric feature (as represented as the large yellow block) from the fit.We also detect strong emission of [Si X ] and [S IX ], and weaker emission of [S VIII ] and [Ca
VIII ]. We detect [Si X ]in two NIRES orders and list both measurements. We select the values from order 5 since the S/N is higher and themeasurements are less uncertain. Also, we do not detect any CN absorption in J -band.Optical data reveal [Fe X ] λ VII ] λ J100935.66+265648.9
Moran et al. (2014) provide an upper limit to the BH mass through Eddington Luminosity arguments, log( M BH /M (cid:12) ) > VI ] (see Figure 17), this being the only case in our sample. Although there GN Outflows in Dwarf Galaxies J0954+4717
Rest-frame Wavelength ( m) F ( e r g s c m s m ) Figure 12.
Same as Figure 7 but for J0954+4717. F ( e r g s c m s m ) Br-H Rest-frame Wavelength ( m) F noise = 0.28 Figure 13.
Same as Figure 11 but for J0954+4717. We additionally fit Br δ , represented as a cobalt line. Bohn et al.
J1005+1257
Rest-frame Wavelength ( m)
45 45 F ( e r g s c m s m ) Figure 14.
Same as Figure 7 but for J1005+1257. F ( e r g s c m s m ) H Rest-frame Wavelength ( m) F noise = 1.18 (a) F ( e r g s c m s m ) H Rest-frame Wavelength ( m) F noise = 0.69 (b) Figure 15.
Same as Figure 11 but for J1005+1257. We plot a single component fit (left) and a double component fit (right).Note that running a F -test on these fits suggests that a two component fit is needed. GN Outflows in Dwarf Galaxies J1009+2656
Rest-frame Wavelength ( m) F ( e r g s c m s m ) Figure 16.
Same as Figure 7 but for J1009+2656. F ( e r g s c m s m ) Br-H Rest-frame Wavelength ( m) F noise = 0.13 Figure 17.
Same as Figure 13 but for J1009+2656. No broad component could be fit. Bohn et al.
J1442+2054
Rest-frame Wavelength ( m) F ( e r g s c m s m ) Figure 18.
Same as Figure 7 but for J1442+2054. appears to be a red broad component, closer inspection of the standard star reveals telluric absorption in that regionwhich leads us to believe that it is due to a poorly corrected telluric feature. We also detect weak emission in [S
VIII ],[S IX ], and [Ca VIII ]. Lastly, we detect no CN absorption.We also detect the optical CL [Ne V ] λ VII ] λλ J144252.78+205451.67
J1442+2054 is one of two galaxies in our sample that falls in the composite region of the BPT diagram (see Manzano-King et al. (2019)). MC20 find both disturbed and stratified gas in their optical data. This object was not observedin L20 and no BH mass has been reported to date.The NIR spectrum of J1442+2054 is shown in Figure 18 and line measurements are listed in Table 13. The NIRis relatively featureless, with only a handful of typically strong emission lines appearing. We detect no CL emission(optical nor NIR) in this galaxy. We also do not detect any CN absorption.
GN Outflows in Dwarf Galaxies Table 5 . J0100 NIR Emission Line ParametersEmission FWHM Flux Notes(km s − ) (10 − erg cm − s − )Pa α µ m 62 ± ± γ µ m 97 ±
46 11 ± Note —Only NIRSPEC (K-band) observations were done.Columns: (1) Emission line. (2) FWHM of the emission line and its error after instrumentcorrection. Uncertainties are only representative of random error in the fit. (3) Fluxand error of the emission line, where uncertainties are from random errors. (4) Notesregarding the fitting.
Table 6 . J0811 NIR Emission Line ParametersEmission FWHM Flux Notes(km s − ) (10 − erg cm − s − )[S III] 0.9531 µ m 173 ± ±
26 Low transmissionHe I 1.0830 µ m Total 137 ±
16 98 ± µ m 19 +42 − ± µ m 265 ±
44 61 ±
13 On skylinePa β µ m 51 ±
48 19 ± µ m 110 ±
19 39 ± µ m 66 +32 − ± µ m 386 ±
81 — Absorption Feature
Note —Same as Table 5. The CO(6-3) absorption feature at 1.62 µ m is added atthe bottom, along with the full-width half-minimum. The depth of the absorptionfeature measures about 13% of the continuum, suggesting roughly 65% of the H -bandcontinuum comes from GMK red giants. Table 7 . J0840 Emission Line ParametersEmission FWHM Flux Notes(km s − ) (10 − erg cm − s − )[S III] 0.9531 µ m 82 ± ± µ m 89 ± ± γ µ m 10 +19 − ± µ m 43 ±
31 28 ± β µ m 3 +34 − ± γ µ m 152 ±
140 11 ± Table 7 continued Bohn et al.
Table 7 (continued)
Emission FWHM Flux Notes(km s − ) (10 − erg cm − s − )CO(6-3) 1.62 µ m 518 ±
126 — Absorption Feature, on skyline
Note —Same as Table 6. The depth of the absorption feature measures about 14% of thecontinuum, suggesting roughly 70% of the H -band continuum comes from GMK red giants. Table 8 . J0842 Emission Line ParametersEmission FWHM Flux Notes(km s − ) (10 − erg cm − s − )[S III] 0.9531 µ m Total — 950 ±
14 Order 7, F > ± ± ± ±
11 Order 7[S III] 0.9531 µ m Total — 986 ±
19 Order 6, low transmission, F = 2Narrow 62 ± ±
12 Order 6Broad 350 ± ±
15 Order 6Pa (cid:15) µ m 155 ±
16 76 ± λ (cid:15) µ m 171 ±
15 103 ± λ δ µ m 48 ±
34 24 ± δ µ m 77 ±
60 24 ± µ m 299 ±
40 71 ± µ m 71 ±
40 20 ± µ m 200 ± ± µ m 280 ±
46 159 ± β µ m 127 ±
12 152 ±
28 Blue asymmetric tailFe II 1.6435 µ m 269 ±
27 152 ±
12 On skylinePa α µ m 192 ± ± µ m 165 ±
21 84 ± µ m 129 ±
74 17 ± µ m 64 ±
34 25 ± µ m 82 ±
13 53 ± γ µ m 136 ±
29 30 ± µ m 50 ±
35 14 ± µ m 90 ±
53 29 ± µ m 723 ±
136 — Absorption Feature
Note —Same as Table 6. The depth of the absorption feature measures about 13% of the continuum,suggesting roughly 65% of the H -band continuum comes from GMK red giants. GN Outflows in Dwarf Galaxies Table 9 . J0906 Emission Line ParametersEmission FWHM Flux Notes(km s − ) (10 − erg cm − s − )[S III] 0.9531 µ m Total — 710 ± F > ± ± ± ± µ m Total — 694 ±
16 Order 6, Low transmission, F > ± ±
10 Order 6Broad 630 ±
25 531 ±
12 Order 6[S VIII] 0.9913 µ m 277 ±
20 32 ± µ m 381 ±
77 29 ± δ µ m 92 ±
29 15 ± µ m 306 ±
73 42 ± µ m 206 ±
71 21 ± µ m 350 ±
122 35 ± µ m[S II] 10339 µ m 235.16 ± ± µ m[N I] 1.0400 µ m 151 ±
77 15 ± µ m Total — 1660 ±
45 On skyline, within telluric absorption, F > ± ± ±
81 401 ± µ m 299 ±
53 31 ± µ m 199 ±
23 44 ± β µ m 282 ±
27 102 ± µ m 309 ±
58 46 ± µ m 344 ±
57 44 ± α µ m Total — 300 ±
19 On skyline, F > ±
30 66 ± ±
37 234 ± µ m 63 ±
36 15 ± µ m Total — 127 ±
15 On skyline, within telluric absorption, F > ±
29 92 ± ±
479 35 ± µ m 97 ±
44 7 ± µ m 81 ±
42 14 ± γ µ m 174 ±
34 13 ± µ m 397 ±
131 33 ± µ m 508 ±
135 — Absorption Feature, on skyline
Note —Same as Table 6. The depth of the absorption feature measures about 10% of the continuum, suggesting roughly50% of the H -band continuum comes from GMK red giants. Bohn et al.
Table 10 . J0954 Emission Line ParametersEmission FWHM Flux Notes(km s − ) (10 − erg cm − s − )[S III] 0.9531 µ m Total — 1719 ±
75 Order 7, F > ± ±
56 Order 7Broad 530 ±
36 468 ±
46 Order 7[S III] 0.9531 µ m Total — 1736 ±
74 Order 6, F > ± ±
60 Order 6Broad 401 ±
21 590 ±
43 Order 6Pa (cid:15) µ m 75 ±
24 52 ±
10 Order 7, blend with [S III] λ (cid:15) µ m 129 ±
42 64 ±
13 Order 6[C I] 0.9853 µ m 75 ±
45 24 ± µ m 109 ±
30 30 ± δ µ m 119 ±
10 108 ± δ µ m 102 ±
12 89 ± µ m 194 ±
23 56 ± µ m 128 ±
44 28 ± µ m 80 ±
28 38 ± µ m 128 ±
29 37 ± µ m Total — 1391 ±
44 red asymmetric tail, F > ± ± ±
61 325 ± γ µ m 96 ± ±
10 On skyline[S IX] 1.2523 µ m 387 ±
116 24 ± µ m Total — 197 ±
57 On skyline, F = 2Narrow 125 ±
19 102 ± ±
124 95 ± β µ m Total — 342 ±
53 On skyline, F > ±
18 186 ± ±
64 157 ± µ m Total — 150 ±
21 On skyline, F > ±
16 90 ± ±
115 61 ± α µ m Total — 952 ±
36 Within telluric absorption, F > ± ± ±
30 298 ± δ µ m 83 ±
11 33 ± µ m 64 ±
28 25 ± µ m Total — 53 ±
11 On skyline, F ∼ ±
54 19 ± ±
164 35 ± µ m 117 ±
29 11 ± µ m 111 ± ± Table 10 continued
GN Outflows in Dwarf Galaxies Table 10 (continued)
Emission FWHM Flux Notes(km s − ) (10 − erg cm − s − )H µ m 105 ±
60 4 ± µ m 71 ±
12 28 ± γ µ m Total — 79 ± F > ±
19 45 ± ±
94 35 ± µ m 160 ±
39 12 ± µ m 553 ±
63 — Absorption Feature
Note —Same as Table 6. The depth of the absorption feature measures about 11% of the continuum,suggesting roughly 55% of the H -band continuum comes from GMK red giants. Table 11 . J1005 Emission Line ParametersEmission FWHM Flux Notes(km s − ) (10 − erg cm − s − )[S III] 0.9531 µ m 144 ± ±
220 Low transmission, red asymmetric tail[C I] 0.9853 µ m 215 ±
18 134 ± µ m 299 ± ±
14 On skylinePa δ µ m 77 ± ± µ m 173 ±
53 105 ±
30 Order 7He II 1.0126 µ m 202 ±
36 102 ±
22 Order 6[S II] 1.0289 µ m 185 ±
45 66 ± µ m 145 ±
20 134 ±
19 On skyline[S II] 1.0339 µ m 109 ±
36 59 ± µ m Total — 2433 ±
194 On skyline, F > ± ± ±
89 976 ± γ µ m 130 ±
11 179 ± µ m 256 ±
38 160 ± µ m 169 ±
17 251 ±
26 On skylinePa β µ m 133 ± ± µ m 143 ± ± µ m 162 ± ± µ m 156 ±
18 168 ±
20 Order 4, within telluric absorptionFe II 1.6435 µ m 157 ±
17 188 ± µ m 114 ±
11 135 ±
11 Within telluric absorption[Si VI] 1.9630 µ m Total — 491 ±
31 On skyline, within telluric absorption, F > ±
14 193 ± ±
114 298 ± Table 11 continued Bohn et al.
Table 11 (continued)
Emission FWHM Flux Notes(km s − ) (10 − erg cm − s − )H µ m 106 ±
35 43 ± µ m 202 ±
56 47 ± µ m 121 ± ± γ µ m 208 ±
19 95 ± µ m 55 ±
35 28 ± µ m 291 ±
129 48 ± µ m 104 ±
18 104 ± µ m 778 ±
232 — Absorption Feature
Note —Same as Table 6. The depth of the absorption feature measures about 7% of the continuum, suggestingroughly 35% of the H -band continuum comes from GMK red giants. Table 12 . J1009 Emission Line ParametersEmission FWHM Flux Notes(km s − ) (10 − erg cm − s − )[S III] 0.9531 µ m 83 ± ±
20 Order 7[S III] 0.9531 µ m 128 ± ±
26 Order 6, Low transmissionPa (cid:15) µ m 125 ± ± µ m 145 ±
40 17 ± δ µ m 44 ±
17 14 ± δ µ m 26 ±
21 23 ± µ m 67 ±
22 39 ± µ m 114 ±
13 44 ± µ m 92 ±
37 9 ± µ m 116 ± ± F < γ µ m 71 ±
30 35 ± µ m 106 ±
45 22 ± µ m 269 ±
90 11 ± µ m 71 ±
16 22 ± β µ m 83 ± ± µ m 83 ±
22 18 ± α µ m 86 ± ± δ µ m INDEF 6 ± < telescope spectral resolutionH µ m 76 ±
34 5 ± µ m 99 ±
15 17 ± µ m 48 +26 − ± µ m 117 ±
70 4 ± µ m 128 ±
26 8 ± Table 12 continued
GN Outflows in Dwarf Galaxies Table 12 (continued)
Emission FWHM Flux Notes(km s − ) (10 − erg cm − s − )Br γ µ m 63 ±
22 12 ± µ m 323 ±
79 12 ± µ m 237 ±
39 — Absorption Feature
Note —Same as Table 6. The depth of the absorption feature measures about 11% of the continuum, suggesting roughly 55%of the H -band continuum comes from GMK red giants. Table 13 . J1442 Emission Line ParametersEmission FWHM Flux Notes(km s − ) (10 − erg cm − s − )[S III] 0.9531 µ m 88 ± ±
29 Order 7[S III] 0.9531 µ m 115 ± ±
20 Order 6, Low transmissionHe I 1.0830 µ m 115 ± ±
12 In telluric absorptionPa γ µ m 24 ± ± µ m 134 ±
51 33 ± β µ m 86 ±
15 83 ± µ m 137 ±
65 17 ± α µ m 93 ± ± µ m 103 ±
60 8 ± γ µ m 75 ±
26 21 ± µ m 115 ±
43 12 ± µ m 446 ±
133 — Absorption Feature, on skyline
Note —Same as Table 6. The depth of the absorption feature measures about 11% of the con-tinuum, suggesting roughly 55% of the H -band continuum comes from GMK red giants. REFERENCES
Allen, M. G., Groves, B. A., Dopita, M. A., Sutherland,R. S., & Kewley, L. J. 2008, ApJS, 178, 20,doi: 10.1086/589652Allington-Smith, J., Murray, G., Content, R., et al. 2002,PASP, 114, 892, doi: 10.1086/341712Antonucci, R. 1993, ARA&A, 31, 473,doi: 10.1146/annurev.aa.31.090193.002353Baldassare, V. F., Reines, A. E., Gallo, E., & Greene, J. E.2017, ApJ, 836, 20, doi: 10.3847/1538-4357/836/1/20Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981,PASP, 93, 5, doi: 10.1086/130766 Becker, G. D., Rauch, M., & Sargent, W. L. W. 2009, ApJ,698, 1010, doi: 10.1088/0004-637X/698/2/1010Bradford, J. D., Geha, M. C., Greene, J. E., Reines, A. E.,& Dickey, C. M. 2018, ApJ, 861, 50,doi: 10.3847/1538-4357/aac88dCann, J. M., Satyapal, S., Bohn, T., et al. 2020, ApJ, 895,147, doi: 10.3847/1538-4357/ab8b64Cappellari, M., & Emsellem, E. 2004, PASP, 116, 138,doi: 10.1086/381875Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ,345, 245, doi: 10.1086/167900 Bohn et al.
Cerqueira-Campos, F. C., Rodr´ıguez-Ardila, A., Riffell, R.,et al. 2020, arXiv e-prints, arXiv:2010.12595.https://arxiv.org/abs/2010.12595Colina, L., Piqueras L´opez, J., Arribas, S., et al. 2015,A&A, 578, A48, doi: 10.1051/0004-6361/201425567De Robertis, M. M., & Osterbrock, D. E. 1984, ApJ, 286,171, doi: 10.1086/162585—. 1986, ApJ, 301, 727, doi: 10.1086/163939De Robertis, M. M., & Shaw, R. A. 1990, ApJ, 348, 421,doi: 10.1086/168251Dong, R., Greene, J. E., & Ho, L. C. 2012, ApJ, 761, 73,doi: 10.1088/0004-637X/761/1/73Erkens, U., Appenzeller, I., & Wagner, S. 1997, A&A, 323,707Fabian, A. C. 2012, ARA&A, 50, 455,doi: 10.1146/annurev-astro-081811-125521Ferguson, J. W., Korista, K. T., & Ferland, G. J. 1997,ApJS, 110, 287, doi: 10.1086/312998Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman,J. 2013, PASP, 125, 306, doi: 10.1086/670067Gimeno, G., Roth, K., Chiboucas, K., et al. 2016, inSociety of Photo-Optical Instrumentation Engineers(SPIE) Conference Series, Vol. 9908, Proc. SPIE, 99082S,doi: 10.1117/12.2233883Groves, B. A., & Allen, M. G. 2010, NewA, 15, 614,doi: 10.1016/j.newast.2010.02.005Groves, B. A., Dopita, M. A., & Sutherland, R. S. 2004a,ApJS, 153, 9, doi: 10.1086/421113—. 2004b, ApJS, 153, 75, doi: 10.1086/421114Harrison, C. M., Alexander, D. M., Mullaney, J. R., &Swinbank, A. M. 2014, MNRAS, 441, 3306,doi: 10.1093/mnras/stu515Heckman, T. M., & Thompson, T. A. 2017, arXiv e-prints,arXiv:1701.09062. https://arxiv.org/abs/1701.09062Kelson, D. D. 2003, PASP, 115, 688, doi: 10.1086/375502Kormendy, J., & Ho, L. C. 2013, ARA&A, 51, 511,doi: 10.1146/annurev-astro-082708-101811Lamastra, A., Bianchi, S., Matt, G., et al. 2009, A&A, 504,73, doi: 10.1051/0004-6361/200912023Lamperti, I., Koss, M., Trakhtenbrot, B., et al. 2017,MNRAS, 467, 540, doi: 10.1093/mnras/stx055Landt, H., Ward, M. J., Steenbrugge, K. C., & Ferland,G. J. 2015a, MNRAS, 449, 3795,doi: 10.1093/mnras/stv062—. 2015b, MNRAS, 454, 3688, doi: 10.1093/mnras/stv2176Liu, W., Veilleux, S., Canalizo, G., et al. 2020, arXive-prints, arXiv:2010.09008.https://arxiv.org/abs/2010.09008Manzano King, C. 2020, PhD thesis, Physics, University ofCalifornia, Riverside Manzano-King, C. M., & Canalizo, G. 2020, MNRAS, 498,4562, doi: 10.1093/mnras/staa2654Manzano-King, C. M., Canalizo, G., & Sales, L. V. 2019,ApJ, 884, 54, doi: 10.3847/1538-4357/ab4197Maraston, C. 2005, MNRAS, 362, 799,doi: 10.1111/j.1365-2966.2005.09270.xMarleau, F. R., Clancy, D., Habas, R., & Bianconi, M.2017, A&A, 602, A28, doi: 10.1051/0004-6361/201629832Martins, L. P., Rodr´ıguez-Ardila, A., de Souza, R., &Gruenwald, R. 2010, MNRAS, 406, 2168,doi: 10.1111/j.1365-2966.2010.17042.xMartins, L. P., Rodr´ıguez-Ardila, A., Diniz, S., Riffel, R., &de Souza, R. 2013, MNRAS, 435, 2861,doi: 10.1093/mnras/stt1482Mason, R. E., Rodr´ıguez-Ardila, A., Martins, L., et al.2015, ApJS, 217, 13, doi: 10.1088/0067-0049/217/1/13McLean, I. S., Becklin, E. E., Bendiksen, O., et al. 1998, inProc. SPIE, Vol. 3354, Infrared AstronomicalInstrumentation, ed. A. M. Fowler, 566–578,doi: 10.1117/12.317283Moran, E. C., Shahinyan, K., Sugarman, H. R., V´elez,D. O., & Eracleous, M. 2014, AJ, 148, 136,doi: 10.1088/0004-6256/148/6/136Morrissey, P., Matuszewski, M., Martin, D. C., et al. 2018,ApJ, 864, 93, doi: 10.3847/1538-4357/aad597M¨uller-S´anchez, F., Hicks, E. K. S., Malkan, M., et al.2018, ApJ, 858, 48, doi: 10.3847/1538-4357/aab9adM¨uller-S´anchez, F., Prieto, M. A., Hicks, E. K. S., et al.2011, ApJ, 739, 69, doi: 10.1088/0004-637X/739/2/69Navarro, J. F., Frenk, C. S., & White, S. D. M. 1996, ApJ,462, 563, doi: 10.1086/177173Oke, J. B., Cohen, J. G., Carr, M., et al. 1995, PASP, 107,375, doi: 10.1086/133562Osterbrock, D. E., Tran, H. D., & Veilleux, S. 1992, ApJ,389, 196, doi: 10.1086/171197Peng, C. Y., Ho, L. C., Impey, C. D., & Rix, H.-W. 2002,AJ, 124, 266, doi: 10.1086/340952—. 2010, AJ, 139, 2097, doi: 10.1088/0004-6256/139/6/2097Penny, S. J., Masters, K. L., Smethurst, R., et al. 2018,MNRAS, 476, 979, doi: 10.1093/mnras/sty202Penston, M. V., Fosbury, R. A. E., Boksenberg, A., Ward,M. J., & Wilson, A. S. 1984, MNRAS, 208, 347,doi: 10.1093/mnras/208.2.347Reines, A. E., Greene, J. E., & Geha, M. 2013, ApJ, 775,116, doi: 10.1088/0004-637X/775/2/116Reines, A. E., & Volonteri, M. 2015, ApJ, 813, 82,doi: 10.1088/0004-637X/813/2/82Rich, J. A., Kewley, L. J., & Dopita, M. A. 2011, ApJ, 734,87, doi: 10.1088/0004-637X/734/2/87
GN Outflows in Dwarf Galaxies31