Secularly powered outflows from AGN: the dominance of non-merger driven supermassive black hole growth
Rebecca Smethurst, Brooke Simmons, Chris Lintott, Jesse Shanahan
MMon. Not. R. Astron. Soc. , 000–000 (0000) Printed 5 September 2019 (MN L A TEX style file v2.2)
Secularly powered outflows from AGN: the dominance ofnon-merger driven supermassive black hole growth
R. J. Smethurst, B. D. Simmons, , C. J. Lintott, J. Shanahan Oxford Astrophysics, Department of Physics, University of Oxford, Denys Wilkinson Building, Keble Road, Oxford, OX1 3RH, UK Physics Department, Lancaster University, Lancaster, LA1 4YB, UK Center for Astrophysics and Space Sciences (CASS), Department of Physics, University of California, San Diego, CA 92093, USA
ABSTRACT
Recent observations and simulations have revealed the dominance of secular pro-cesses over mergers in driving the growth of both supermassive black holes (SMBH)and galaxy evolution. Here we obtain narrowband imaging of AGN powered outflowsin a sample of 12 galaxies with disk-dominated morphologies, whose history is assumedto be merger-free. We detect outflows in 10 /
12 sources in narrow band imaging of the[O iii ] 5007 ˚A emission using filters on the Shane-3m telescope. We calculate a meanoutflow rate for these AGN of 0 . ± .
14 M (cid:12) yr − . This exceeds the mean accretionrate of their SMBHs (0 . ± .
039 M (cid:12) yr − ) by a factor of ∼
18. Assuming that thegalaxy must provide at least enough material to power both the AGN and the outflow,this gives a lower limit on the average inflow rate of ∼ . ± .
14 M (cid:12) yr − , a ratewhich simulations show can be achieved by bars, spiral arms and cold accretion. Wecompare our disk dominated sample to a sample of nearby AGN with merger domi-nated histories and show that the black hole accretion rates in our sample are 5 timeshigher (4 . σ ) and the outflow rates are 5 times lower (2 . σ ). We suggest that thiscould be a result of the geometry of the smooth, planar inflow in a secular dominatedsystem, which is both spinning up the black hole to increase accretion efficiency andless affected by feedback from the outflow, than in a merger-driven system with chaoticquasi-spherical inflows. This work provides further evidence that secular processes aresufficient to fuel SMBH growth. Key words: galaxies, outflows, AGN, black holes, co-evolution
Understanding the co-evolution of galaxies and their centralsupermassive black holes (SMBHs) is integral to modern as-trophysics. Determining the physical processes which drivethe growth of both the galaxy and the SMBH is a key goalof current observational and theoretical work. An increasingbody of evidence shows that galaxy growth mainly occursthrough ‘secular processes’ rather than by mergers. Kavi-raj et al. (2013) for example, show that only 27% of starformation is triggered by major or minor mergers at z ∼ c (cid:13) a r X i v : . [ a s t r o - ph . GA ] S e p Smethurst et al. 2019 have extremely disk-dominated morphologies must thereforehave had their evolution, and therefore SMBH growth, dom-inated by merger free processes, at least since z < ∼ z ∼ . (cid:54) ˙m (cid:54) .
37 M (cid:12) yr − . SSL17 then made the simplifying assumptionthat any process driving this accretion must provide gas ata rate which is at least the calculated SMBH accretion rate.Whilst this assumption put a lower limit on this inflow rate,there is a growing body of evidence suggesting that molec-ular outflows are ubiquitous in both star forming galaxiesand AGN (Feruglio et al. 2010; Alatalo et al. 2011; Aaltoet al. 2012; Cicone et al. 2014; Alatalo 2015; Gallagher et al.2019), inlcuding those in dwarf galaxies (Penny et al. 2018;Manzano-King et al. 2019). Indeed Bae et al. (2017) haveshown that the flux of material in outflows in a sample of20 nearby AGN exceeds the rate of accretion of the SMBHby a factor of ∼ and the outflow rate.In this work, we aim to measure the outflow rates inthe sample of 101 disk-dominated AGN studied by SSL17.The outflow measurements are enabled with narrow-bandimaging centered on [O iii ] 5007˚A to measure both the ex-tent of the outflow and the gas mass present. By combiningthis measurement with existing measurements of the blackhole accretion rate from SSL17, we can constrain the totalinflow rate to the centre of these systems driven by secularprocesses. By using a sample of galaxies where we can besure that secular processes dominate we can, for the firsttime, understand what the limits to merger-free black holegrowth are.The results for this particular disk-dominated samplewill be compared with the more typical AGN systems of Baeet al. (2017) which have morphologies indicative of an evo-lutionary history containing (at least) minor mergers. Dif-ferences or similarities between the properties of the twosamples of galaxies will have important implications for thefeeding of the SMBHs in these samples.In the rest of this work we adopt the Planck 2015(Planck Collaboration et al. 2016) cosmological parameterswith (Ω m , Ω λ , h ) = (0 . , . , .
68) and any emission or ab-sorption features referred to are in the Lick system. All mea-sured values are quoted to 2 decimal places. In Section 2.1 wesummarise our selection and observations of disk-dominatedsystems hosting AGN before detailing the data reduction in Section 3. In Section 4 we present our results and discusstheir implications in Section 5.
Here we utilise a well-studied sample of 101 disk-dominatedgalaxies with unobscured Type 1 AGN first identified inSSL17. The sample comprises galaxies in the SDSS (Yorket al. 2000) Data Release 8 (Aihara et al. 2011) imagingsample cross-matched with sources identified by Edelson &Malkan (2012) using multi-wavelength data from the Wide-field Infrared Survey Explorer (WISE; Wright et al. 2010),Two Micron All-Sky Survey (2MASS; Skrutskie et al. 2006),and ROSAT all-sky survey (RASS; Voges et al. 1999). Thedisk-dominated morphologies were assigned by expert re-view of the SDSS imaging (see Simmons et al. 2013 andSSL17), and later confirmed using images from an
HST snapshot survey with broadband imaging using ACS WFC(programme ID HST-GO-14606, PI: Simmons).
HST im-ages were reduced using the standard pipeline. Black holemasses for this sample were calculated by SSL17 using therelation between black hole mass and the FWHM and lu-minosity in the broadened Hα emission line from Greene& Ho (2005). SSL17 consequently used a bootstrap methodwhilst fitting the width of the Hα line to estimate the un-certainty on the black hole mass measurement. Bolometricluminosities for this sample were also calculated by SSL17using the WISE W3 band at 12 µm , by applying a correctionfrom Richards et al. (2006). It is possible that the W3 fluxdensities could be contaminated by star formation, howeverRichards et al. (2006) concluded that since there were min-imal differences between their composite SEDs of Type 1AGN around ∼ µm this suggested minimal host galaxycontamination. Once again uncertainties were derived bySSL17 using a bootstrap method. SSL17 used their calcu-lated bolometric luminosities to calculate both Eddingtonratios, λ Edd , and black hole mass accretion rates, ˙ m , usinga simple energy to matter conversion (see Equation 5).96 of these sources had spectra available in the SDSSData Release 9 spectroscopic sample (Ahn et al. 2012) and5 spectra were obtained using the Intermediate DispersionSpectrograph on the Isaac Newton Telescope, La Palmafrom 21st-23rd May 2014. The spectra of these 5 sourcesdid not have sufficient wavelength range to probe the [O iii ]emission. The spectra of the 96 systems with SDSS spectrawere fitted using the spectral fitting code GANDALF (Sarziet al. 2006) which fits multiple simultaneous lines as wellas the continuum.
GANDALF is optimised for use with SDSSspectra and allows for the identification of multiple compo-nent emission lines. Initially all emission lines are modelledas a single Gaussian, with the same width for all lines. Moreinformation is then available from inspection of the [O iii ](4959, 5007) doublet emission line shape. The main emis-sion line is identified from the expected wavelength relativeto the emission lines in the rest of the spectrum. A secondrun of
GANDALF was performed allowing all emission lines tohave two Gaussian components; if
GANDALF returned a non-zero flux value for a blueshifted [O iii ] component this wasconsidered to be an outflow. We made no cuts based on the c (cid:13) , 000–000 ecularly powered accretion and outflows F l u x d e n s i t y Ron [OIII] regionFWHM of wing: 1069 km/sS/N of wing: 28.4
Neville [OIII] regionFWHM of wings: 376, 1302 km/sS/N of wings: 41.5, 21.8
Object spectrumGANDALF fit to spectrumGANDALF fits to individual[OIII] components5100 5120 5140 5160 5180 5200 52200200400 F l u x d e n s i t y Hermione [OIII] regionFWHM of wing: 819 km/sS/N of wing: 6.9
Harry [OIII] regionFWHM of wing: 612 km/sS/N of wing: 6.8 F l u x d e n s i t y Theodore [OIII] regionFWHM of wing: 343 km/sS/N of wing: 17.7
Snape [OIII] regionFWHM of wing: 939 km/sS/N of wing: 1.1 F l u x d e n s i t y Regulus [OIII] regionFWHM of wing: 543 km/sS/N of wing: 35.9
Goyle [OIII] regionFWHM of wing: 699 km/sS/N of wing: 11.6 F l u x d e n s i t y Crabbe [OIII] regionFWHM of wing: 1188 km/sS/N of wing: 3.0
Padma [OIII] regionFWHM of wing: 1355 km/sS/N of wing: 35.4 λ obs [˚ A ] F l u x d e n s i t y Cho [OIII] regionFWHM of wing: 1628 km/sS/N of wing: 2.4
Figure 1.
GANDALF (Sarzi et al. 2006) fits of the SDSS spectra of the 12 AGN in the disk-dom-outflow sample observed using theShane-3m telescope at the Lick Observatory, each showing a blueshifted wing component in the [O iii ] emission lines. In each panel, thesolid black line shows the SDSS spectrum, with the corresponding error on the spectrum showed by the grey shaded region. Note thatthe errors on the spectrum are small. The thicker, red solid line shows the fit to this spectrum. The grey solid lines show the separatecomponents of the fitted emission for the [O iii ] (4959, 5007) doublet emission. Grey points indicate the residuals between the observed andfitted spectra. In each panel we note the full width half maximum (FWHM) and signal-to-noise ratio (S/N) of the blueshifted [O iii ] 5007˚Awing component(s). Note that we did not make any cuts on the signal-to-noise ratio of the wing (see Section 2.1). Corresponding SDSSimages are shown in Figure 3 and coordinates are listed in Table 1. Note that the emission at ∼ ii line, rather than a redshifted [O iii ] component. signal-to-noise ratio returned by GANDALF as we wanted toretain anything that may be a detection for further inves-tigation, since these detections are limited by the 3” sizeof the SDSS spectral fibre. A further run of
GANDALF with three Gaussian components was performed for those sourceswhich a two Gaussian component fit was unsatisfactory.Of these 96 galaxy spectra, 58 required 2 (or more)components in
GANDALF fits to their [O iii ] emission with bothnarrow and blueshifted broadened wing components. From c (cid:13)000
GANDALF fits to their [O iii ] emission with bothnarrow and blueshifted broadened wing components. From c (cid:13)000 , 000–000 Smethurst et al. 2019 F l u x d e n s i t y Scorpius [OIII] region
Pansy [OIII] region
Object spectrumGANDALF fit to spectrumGANDALF fits to individual[OIII] components5280 5300 5320 5340 5360 5380 5400 5420 5440050100150200 F l u x d e n s i t y Millicent [OIII] region
Blaize [OIII] region F l u x d e n s i t y Bellatrix [OIII] region
Penelope [OIII] region λ obs [˚ A ] F l u x d e n s i t y Luna [OIII] region
Figure 2.
GANDALF (Sarzi et al. 2006) fits of the SDSS spectra of the 7 AGN in the disk-dom-none sample observed using the Shane-3mtelescope at the Lick Observatory, each without a blueshifted wing component in the [O iii ] emission lines. In each panel, the solid blackline shows the SDSS spectrum, with the corresponding error on the spectrum showed by the grey shaded region. Note that the errorson the spectrum are small. The thicker, red solid line shows the fit to this spectrum. The grey solid lines show the separate componentsof the fitted emission for the [O iii ] (4959, 5007) doublet emission. Grey points indicate the residuals between the observed and fittedspectra. Corresponding SDSS images are shown in Figure 4 and coordinates are listed in Table 1. this detection of a blueshifted component in the the spectrawe know that there is some outflowing material from theAGN within the 3” diameter central SDSS fibre, howeverthis may not capture the full extent of the outflow.We selected the 12 brightest galaxies in the blushifted[O iii ] 5007˚A spectral component which had coordinates ap-propriate for the 2018A semester, for which we were awarded3 nights on the Shane-3m telescope from 12-14th May 2018at the Lick Observatory, California, USA . We shall refer tothis sample as the disk-dom-outflow sample (0 . < z < . The availability of a variety of narrow-band filters makes Lickan optimal facility for imaging the outflows of relatively nearbyAGN. with their SDSS IDs and labels . We shall refer to this sam-ple as the disk-dom-outflow sample. In addition to these12 galaxies showing outflowing components in their SDSSspectra, we also observed 7 of our 43 disk-dominated AGNwhich had no detected second component in [O iii ] in theirSDSS spectra fits. We shall refer to this sample as the disk-dom-none sample (0 . < z < . disk-dom-outflow and disk-dom-none galaxies are shown in Figures 1 & 2 respec-tively, while SDSS postage stamp images of both samplesare shown in Figures 3 & 4. Each source had been highlighted in either red, blue or greendepending on which filter pair we planned to observe it with.Sources were then named after characters in the Harry Potter nov-els (Rowling 1997), with red, blue and green highlighted sourcesnamed after Gryffindors, Ravenclaws and Syltherins respectively.Alas, Hufflepuffs were overlooked due to the first author’s lack ofa yellow highlighter pen. c (cid:13) , 000–000 ecularly powered accretion and outflows Table 1.
Coordinates of the 12 disk-dominated AGN with outflows present in their SDSS spectra, the disk-dom-outflow sample (top)and the 7 without, the disk-dom-none sample (bottom), observed with narrow band filters on the Shane-3m telescope at the LickObservatory, Mt. Hamilton, using the instrument in imaging mode with the gratings removed.SDSS ID Label RA Dec z Outflowsurfacebrightness[magarcsec − ] Totalexposuretime [s] Outflownarrowband filtercentral λ [˚ A ] Continuumnarrowband filtercentral λ [˚ A ]1237654880205209621 Ron 226.486 3.707 0.036 30.10 499 5199 55031237662195054673929 Neville 158.661 39.641 0.043 28.33 91 5232 55031237663531870191740 Harry 123.350 54.377 0.031 29.12 190 5232 55031237662504293892158 Hermione 239.790 35.030 0.043 31.85 9734 5199 55031237661967971385403 Theodore 198.715 42.305 0.073 28.45 102 5382 56821237659149386973357 Snape 226.969 51.853 0.075 32.67 5023 5382 56821237651736294195257 Regulus 180.887 2.493 0.077 29.02 184 5382 56821237661966353104933 Goyle 175.014 41.251 0.071 30.84 934 5382 56821237667911661649935 Crabbe 171.515 24.554 0.069 28.96 165 5382 56821237660669817061494 Padma 153.161 10.289 0.069 27.88 64 5382 56821237662662680051805 Flitwick 239.578 25.857 0.070 27.05 29 5382 56821237667486470176778 Cho 171.904 24.823 0.059 31.91 2217 5298 56101237665548886081560 Scorpius 205.986 25.647 0.086 29.53 278 5434 57191237658424619696216 Pansy 141.839 6.166 0.078 26.49 18 5382 56821237654879118557355 Millicent 197.013 3.854 0.070 28.29 94 5382 56821237663655882064127 Blaize 112.861 45.372 0.092 26.53 47 5434 57191237662500012949661 Bellatrix 252.763 26.296 0.079 30.86 1051 5434 57191237667782293979150 Penelope 175.317 21.939 0.063 27.36 36 5332 56101237661852549709862 Luna 206.110 44.272 0.055 33.19 7731 5265 5610 We observed the 19 disk-dominated AGN host galaxies listedin Table 1 using narrow-band imaging centered on the ob-served wavelength of the [O iii ] 5007˚A emission with the Kastspectrograph on the Shane 3-m telescope at the Lick Obser-vatory using the instrument in imaging mode with the grat-ings removed. The SDSS spectra provided the observed peakwavelength of the [O iii ] 5007˚A component for all 19 sources,giving observed wavelengths in the range of 5165 − ∼ ∼ iii ] 5007˚A emission and reduce contamination fromnearby [feii] , [ni] and H β emission.We also observed the continuum of each galaxy, near tothe [O iii ] 5007˚A emission, in an appropriate narrow bandfilter ensuring that there was no overlap between the wave-length ranges of the chosen filters. Continuum measurementsare required to separate the flux from [O iii ] 5007˚A ionisationfrom the continuum at that wavelength. The exposures timeslisted in Table 1 were calculated using the throughput of thenarrow band filters convolved with the SDSS spectra, in or-der to give a predicted signal-to-noise ratio (SNR) (cid:62) iii ] emission for the disk-dom-none sample listed in the bottom half ofTable 1). Each image of a source, in either the [O iii ] 5007˚A or con-tinuum centered filters, was reduced using the functions inthe ccdproc package written for python (Craig et al. 2015).A basic reduction was performed first, subtracting the bias,removing the overscan region and flat fielding the images.Cosmic rays were removed before performing a backgroundsubtraction. Images were then flux calibrated using the stan-dard star observed closest in time to the exposure in the ap-propriate narrow band filter. The standard stars observedover the 3 nights of observations were Feige 34, Feige 67,G193-74, BD+33d2642, HZ21, HZ44 and HD93521.Images of each source were then combined to give a mas-ter [O iii ] 5007˚A and master continuum image which werenormalised by their total exposure times so that the twoimages could be directly compared. The master continuumimage was then subtracted from the master [O iii ] 5007˚Aimage, leaving only the flux from the [O iii ] 5007˚A narrowemission (ionised either by the AGN or due to star forma- c (cid:13) , 000–000 Smethurst et al. 2019
5” Ron Neville, F775WHermione Harry, F775WTheodore, F814W Snape, F814WRegulus, F814W GoyleCrabbe Padma, F814WFlitwick Cho
Figure 3.
SDSS gri or HST
ACS WFC (where available, withWFC filters stated) postage stamp images of the 12 AGN in the disk-dom-outflow sample. The AGN can be seen as a brightpoint source in the centre of each image, which we assume ispowered by merger-free processes due to the disk-dominated mor-phology of these sources. The scale for each image is ∼ .
15 arc-sec/pixel, resulting in images approximately 63” across. Labelsand coordinates are listed in Table 1.
5” Scorpius Pansy, F814WMillicent, F814W Blaize, F850LPBellatrix, F814W Penelope, F814WLuna
Figure 4.
SDSS gri or HST
ACS WFC (where available, withWFC filters stated) postage stamp images of the 7 AGN in the disk-dom-none sample. The AGN can be seen as a bright pointsource in the centre of each image. The scale for each image is ∼ .
15 arcsec/pixel, resulting in images approximately 63” across.Labels and coordinates are listed in Table 1. tion producing extended emission) and the flux from theblueshifted wing component (ionised by the outflow). In or-der to remove the [O iii ] 5007˚A flux ionised by the centralAGN, we extracted the PSF from the standard star image inthe [O iii ] 5007˚A centered filter (previously used to flux cal-ibrate), and subtracted this from the continuum subtractedimage. We modelled the PSF of the standard star images(using the
EPSFBuilder module from the astropy affiliatedpackage photutils ; Bradley et al. 2019), rather than mod-elling as a Gaussian, as the reduced standard star imagesclearly show that the PSF was not Gaussian and that theshape of the PSF varied across different narrow band fil-ters due to the differing refraction properties of each filter.Note that whilst different atmospheric conditions on differ-ent observing nights could account for variations in the PSF,these variations are noticeable even in images of the samestar taken a few minutes apart in the different filters. Fig-ure 5 shows this with reduced images of the same standard c (cid:13) , 000–000 ecularly powered accretion and outflows HZ445265˚A HZ445382˚A HZ445434˚A HZ445610˚A HZ445682˚A HZ445719˚A
Figure 5.
Example images of reduced standard star HZ44 observed across six different narrow band filters used in this study. The centralwavelength of each filter is stated in each panel. These images show how the PSF through each of these filters is not Gaussian, and isnot a consistent shape due to the way that the light refracted differently through each filter. A PSF for each galaxy image was thereforeextracted from a standard star image taken through the [O iii ] centered filter in order to remove flux ionised by the central AGN fromthe continuum subtracted [O iii ] image( see Section 3.1). star, HZ44, imaged across the different narrow band filtersused in this study.An example of the [O iii ] 5007˚A centered, continuumcentered, continuum subtracted and PSF subtracted imagesfor two sources, Hermione and Padma are shown in Fig-ure 6. At this point we would ideally sum the flux within1 r petro (as calculated by the SDSS pipeline) in order to de-rive the total flux in the outflow. However, whilst Padmadoes not have any extended emission due to star formationionisation left in the final PSF subtracted image, it is clearthat Hermione does. We therefore also show the one dimen-sional traces for each source across the final PSF subtractedimage in Figure 7 to show how we determined an empir-ical limit for Hermione, above which the total flux in the[O iii ] 5007˚A outflow was summed. This method of empiri-cally determining the limit above which to sum the flux wasemployed for Hermione (4 . σ ), Neville (2 . σ ) and Cho (5 σ ),so that the calculated outflow rates are lower limits. For allother sources, the total flux in the [O iii ] outflow was summedabove 3 σ , where σ is the standard deviation of the image.The total luminosity of the outflow, L [O iii ], is then calcu-lated using the luminosity distance of the source (calculatedfrom the redshift using the astropy cosmology module, seeAstropy Collaboration et al. 2013, 2018). This measurement of the outflow luminosity, L [O iii ], can beused to calculate a gas mass in the outflow following themethod outlined in Carniani et al. (2015): M gas = 0 . × M (cid:12) × (cid:18) C [ O/H ] − [ O/H ] (cid:12) (cid:19) (cid:18) L [O iii ]10 erg s − (cid:19) (cid:16) n e
500 cm − (cid:17) − (1)where n e is the electron density, [ O/H ] − [ O/H ] (cid:12) is themetallicity relative to solar, and C = < n e > / < n e > .Here < n e > is the volume averaged electron densitysquared and < n e > is the volume averaged squared electrondensity. This method requires some simplifying assumptionson the nature of the outflowing gas, particularly on the tem-perature, metallicity and density of the gas, however thesecaveats affect all such measurements in the literature whichwe intend to compare to. We therefore assume typical valuesfrom the literature; a gas solar metallicity, [ O/H ] = [
O/H ] (cid:12) and an electron density, n e = 500 cm − . Note that there isno general agreement on the best value of n e , with conflict-ing estimates across the literature. The long assumed value of n e = 100 cm − has recently been challenged by (Pernaet al. 2017, 700 < n e < − ) and (Villar Mart´ın et al.2015, n e ∼ cm − ). Assuming a smaller value of n e canlead to an overestimate of the gas mass present. We chose touse n e = 500 cm − in order to be consistent with Carnianiet al. (2015).In order to measure the mass loss rate due to the out-flow, we then combined this measurement of the gas massof a source with the timescale of the visible outflow, t outflow .The timescale for the outflow is calculated using the veloc-ity of the outflow, measured between the peak of the narrowand blueshifted [O iii ] 5007˚A components in each spectrum,v outflow and the most distant spatial extent of the outflowaway from the central AGN, r max (assumed to be the bright-est pixel in the continuum subtracted image): t outflow [yr] = (cid:18) r max km (cid:19)(cid:18) v [OIII] km yr − (cid:19) − . (2)The extent of the outflow was measured on the image itselfin arcseconds and converted to kilometers using the redshiftof the source and the angular diameter distance functionin the astropy.cosmology module (Astropy Collaborationet al. 2013, 2018). The outflow rate is then calculated in thefollowing way: (cid:18) ˙M outflow M (cid:12) yr − (cid:19) = (cid:18) M [OIII] M (cid:12) (cid:19)(cid:18) t outflow yr (cid:19) − . (3)Since energy is conserved across the galaxy-AGN system,knowing the outflow rate precipitates the calculation of theinflow rate by the following assumption: (cid:18) ˙M inflow M (cid:12) yr − (cid:19) = (cid:18) ˙ m M (cid:12) yr − (cid:19) + f (cid:18) ˙M outflow M (cid:12) yr − (cid:19) , (4)where ˙ m is the accretion rate of the black hole. These valueswere measured previously for each source by SSL17 by usingthe bolometric luminosity, L bol ;˙ m = L bol /ηc , (5)where the radiative efficiency, η = 0 .
15 (see Elvis et al.2002). Since Equation 4 stems from a conservation of en-ergy assumption, f is therefore an unknown factor propor-tional to ( v [OIII] /v wind ) , where v wind is the velocity of thefeedback driven wind from the AGN accretion disk whichimpacts with the surrounding medium. The wind dumps en-ergy into the surrounding medium, both ionizing it and driv-ing it out from the centre in a gas mass outflow. This energyexchange will cause v [OIII] < v wind . Whilst we can measure c (cid:13) , 000–000 Smethurst et al. 2019 . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec
PSF subtracted . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec R A [OIII] filter . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec
Continuum filter . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec
Continuum subtracted . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec
PadmaPSF subtracted . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec R A Padma[OIII] filter . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec
Continuum filterPadma . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec
Continuum subtractedPadma
Figure 6.
From left to right; the [O iii ] filter centered, continuum filter centered, continuum subtracted and PSF subtracted images forHermione (top) and Padma (bottom). All images have a square root stretch applied. In the PSF subtracted images (far right), the redcross denotes the position of the brightest pixel in the continuum subtracted image, assumed to be the central AGN. Over-subtractionof the central AGN is expected given the size and shape of the PSF through the narrow band filters on the Shane-3m. Hermione is aclear example of a galaxy with extended [O iii ] emission, ionised due to star formation in the bar and spiral arms, which is still presentin the PSF subtracted emission. Padma however, does not have any extended [O iii ] emission so we assume that any remaining flux inthe PSF subtracted image is purely ionised by the outflow (see Section 3.1). v [OIII] spectroscopically, we cannot probe v wind . Therefore,throughout the rest of this work we assume that f = 1,i.e. v [OIII] = v wind , and therefore derive upper limits on theinflow rates to the AGN in the disk-dom-outflow sample. The reduced images for each source in the disk-dom-outflow sample are shown in Figure 8, sorted by the cal-culated outflow rates, and those in the disk-dom-none areshown in Figure 9.The calculated mass loss rates from the flux in the[O iii ] outflow (see Equation 1) remaining in the PSF sub-tracted images are quoted in each of the panels of Figure 8and in Table 2 along with the assumed inflow rates andtimescale of the outflow, t outflow (see Equation 2).The mean outflow rate for the disk-dom-outflow sample is 0 . ± .
14 M (cid:12) yr − . This exceeds themean accretion rate of their SMBHs (0 . ± .
039 M (cid:12) yr − )by a factor of ∼
18. The mean (median) outflow timescaleof the sample is 121 Myr (23 Myr).For the disk-dom-outflow sample we have detected4 asymmetric and 5 bi-symmetric outflows within the Pet-rosian radius of the galaxy (see Figure 8). There are alsoarguably 3 non-detections of outflows (Snape, Crabbe andNeville) which we discuss below.For the galaxies in the disk-dom-none sample, wemust consider whether the blueshifted outflow componentin [O iii ] was not detected in the original SDSS spectra be-cause it lay outside of the 3” SDSS fiber. In each panel of Figure 9, the extent of the fiber is shown by the blue circles,similarly the Petrosian radius of the galaxy is also shown bythe red circle. For the majority of the sources in the disk-dom-none sample, some of the flux leftover in the PSF sub-tracted image lies within the SDSS fiber, therefore we canassume that this [O iii ] flux is due to star formation ionisa-tion and will contribute to the narrow [O iii ] 5007˚A emissionin the spectra (see for example Pansy and Luna). Similarlyfor Penelope the remaining flux is found within and outsideof the SDSS fibre, suggesting that this gas is ionised due tostar formation. We can therefore take Penelope’s calculatedgas mass of 7 . × M (cid:12) as an upper limit on the amountof gas that can be confused for an outflow.Blaize is the only source where [O iii ] flux is not detectedinside the diameter of the 3” SDSS fiber (the blue circleon Figure 9). Given the amount of noise in the rest of theimage, we can assume that no emission in [O iii ] remainedin the PSF subtracted image, meaning we can take Blaize’scalculated gas mass of 3 . × M (cid:12) (see Figure 9) as alower limit for detection of gas mass in an outflow. In thisinstance, we can therefore assign this as an upper limit tothose galaxies with detected outflows in the spectra whichhad calculated gas masses less than this value, includingSnape and Crabbe, which we earlier identified as possiblenon-detections and which have calculated gas masses lowerthan 3 . × M (cid:12) . This is also the case for Neville, howeversince Neville also had contamination from star formationionised [O iii ] 5007˚A emission the calculated value is alreadyknown to be a lower limit, so we retain this value. c (cid:13) , 000–000 ecularly powered accretion and outflows . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec [deg] R A [ d e g ] HermionePSF subtracted . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec [deg] R A [ d e g ] PadmaPSF subtracted
Figure 7.
The PSF subtracted images for Hermione (left) and Padma (right), showing the maximum value across each of the RA andDec axes. For Padma, the standard limit value of 3 σ , where σ is the standard deviation of the image, is shown by the blue dashed line. Ineach image, the red cross denotes the position of the brightest pixel in the continuum subtracted image, assumed to be the central AGN.Only positive flux values are shown in the image, hence over-subtraction of the central AGN can be seen in the black regions either sideof the red cross for Hermione (left panel). This is expected given the size and shape of the PSF through the narrow band filters on theShane-3m. For Hermione, the PSF subtracted image still shows extended [O iii ] emission due to ionisation from star formation. The limitabove which to sum the [O iii ] flux is therefore determined empirically by inspection of the one dimensional traces of the maximum valueacross each axis of the image (blue dashed line). The Petrosian radius is shown by the red circles in each image and the correspondingred dashed lines. In each case, the [O iii ] flux in the outflow is determined by summing the flux within the Petrosian radius and abovethe derived limit shown by the blue dashed lines (see Section 3.1). We calculate the inflow rates of the disk-dom-outflow sample using Equation 4; consequently eachinflow rate calculated is an upper limit on the rate neededto account for both the accretion rate of the black holeand mass outflow rate from the AGN in the disk-dom-outflow sample. The mean value of these inflow ratesis ∼ . ± .
14 M (cid:12) yr − (the median inflow rate is ∼ .
57 M (cid:12) yr − ). We therefore must consider what pro-cesses would be able to drive such an inflow and whether therate could be sustained over (cid:38)
921 Myr (the maximum timeover which the outflows in the disk-dom-outflow samplehave been active, see Table 2 and Equation 2).Simulations have shown that bars (0 . − few M (cid:12) yr − ,Sakamoto 1996; Maciejewski et al. 2002; Regan & Teuben2004; Lin et al. 2013), spiral arms (0 . − . (cid:12) yr − , Ma-ciejewski 2004; Davies et al. 2009; Schnorr-M¨uller et al. 2014;Slater et al. 2019) and smooth accretion of cold gas onto iso-lated galaxies (0 . − . (cid:12) yr − , Kereˇs et al. 2005; Sancisiet al. 2008) can all sustain at least this level of inflow rate.Bars and spiral arms are also thought to be long lived mor-phological features of a galaxy (Miller & Smith 1979; Sparke& Sellwood 1987; Donner & Thomasson 1994; D’Onghiaet al. 2013; Hunt et al. 2018) and so could feasibly drivean inflow over such an extended period of time. Similarly,accretion of cold gas from the cosmic web is a long livedprocesses unless it is interrupted by some feedback process to the inter galactic medium (van de Voort et al. 2011).Since the outflows in the disk-dom-outflow sample have v [OIII] < v esc , where v esc is the escape velocity of the galaxy,we can assume that these outflows will only feedback inter-nally on the galaxy rather than the inter-galactic medium.Given our calculated upper limits on the inflow rates in the disk-dom-outflow sample and the timescales of the out-flows, it is feasible that all of these mechanisms could drivethe growth of these AGN. This suggests that secular pro-cesses are more than sufficient at fuelling black hole growth,at least at z ∼
0. This supports the findings of SSL17 andof Martin et al. (2018) who show that 65% of SMBH growthsince z ∼ Bars are often cited as the most common mechanism fordriving inflows to feed AGN in disk-dominated galaxies,however many studies have struggled to find any correlationsbetween the presence of a bar with either black hole mass(Oh et al. 2012), accretion rate (Goulding et al. 2017) orEddington rate (Lee et al. 2012). Galloway et al. (2015) didfind a weak correlation between the presence of a bar and thepresence of an AGN after controlling for mass and colour,but no correlation between AGN strength (L [OIII] / M BH ) andthe presence of a bar. However, the majority of the materialinflowed to an AGN will be ejected in an outflow ( ∼ c (cid:13) , 000–000 Smethurst et al. 2019 . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ R A Snape0.007 ± (cid:12) yr − ± (cid:12) yr − . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Crabbe0.059 ± (cid:12) yr − ± (cid:12) yr − . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Neville0.065 ± (cid:12) yr − ± (cid:12) yr − . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ R A Goyle0.16 ± (cid:12) yr − ± (cid:12) yr − . ◦ . ◦ . ◦ . ◦ . ◦ Ron0.21 ± (cid:12) yr − ± (cid:12) yr − . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Theodore0.28 ± (cid:12) yr − ± (cid:12) yr − . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ R A Cho0.78 ± (cid:12) yr − ± (cid:12) yr − . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Flitwick0.97 ± (cid:12) yr − ± (cid:12) yr − . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Harry1.80 ± (cid:12) yr − ± (cid:12) yr − . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec R A Regulus1.95 ± (cid:12) yr − ± (cid:12) yr − . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec
Hermione2.24 ± (cid:12) yr − ± (cid:12) yr − . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec
Padma2.92 ± (cid:12) yr − ± (cid:12) yr − Figure 8.
Continuum and PSF subtracted images for sources in the disk-dom-outflow sample. Only flux above either 3 σ , or theempirically determined value to isolate the outflow from star formation ionised emission, is shown in each image. In each panel we showthe name of the source, the outflow rate (see Equation 3) and the inflow rate (which combines the outflow rate with the accretion rate ofthe black hole; see Equation 4 and Section 3.2). These values are also listed in Table 2. We show the Petrosian radius, r petro , calculatedfrom the SDSS imaging, within which the total flux in [O iii ] was summed, with a red circle. The size of the SDSS fibre that the originalspectra were taken with is also shown by the blue circle. Sources are ordered from left to right by the measured outflow rate. Labels andcoordinates for these sources are listed in Table 1. c (cid:13) , 000–000 ecularly powered accretion and outflows . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ R A Blaizelog [ M [ OIII ] /M (cid:12) ] = 6.55 ± . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Pansylog [ M [ OIII ] /M (cid:12) ] = 7.75 ± . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Penelopelog [ M [ OIII ] /M (cid:12) ] = 8.87 ± . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ R A Millicentlog [ M [ OIII ] /M (cid:12) ] = 7.22 ± . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec
Scorpiuslog [ M [ OIII ] /M (cid:12) ] = 7.68 ± . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec
Lunalog [ M [ OIII ] /M (cid:12) ] = 6.57 ± . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ . ◦ Dec R A Bellatrixlog [ M [ OIII ] /M (cid:12) ] = 7.46 ± Figure 9.
Continuum and PSF subtracted images for sources in the disk-dom-none sample. Only flux above 3 σ is shown in each image.In each panel we show the name of the source and the gas mass of [O iii ] (see Equation 1) measured. This provides a limit for the amountof emission from star formation ionisation that could be confused for an outflow (see Section 4). We show the Petrosian radius, r petro ,calculated from the SDSS imaging, by the red circle. The size of the SDSS fibre that the original spectra were taken with is also shownby the blue circle. Labels and coordinates for these sources are listed in Table 1. on average in the disk-dom-outflow sample in this work)which these studies do not account for. In addition, theseprevious works selected barred galaxies with a range of bulgesizes, indicating that (at least minor) mergers will also haveaffected the evolutionary history of the galaxies in thesesamples. These studies will therefore not have probed theeffect of the bar alone.However, in the disk-dom-outflow sample, we can bepositive that our galaxies have not undergone a significantmerger since at least z ∼ disk-dom-outflow sample, all butone (Snape) host either a strong or a weak bar. An exceptionto this is Flitwick as it is at a higher redshift than the othersources and no HST imaging is available, so it is possible that a bar is present but is unresolved in the SDSS imaging.In contrast, there are only two bars present in the disk-dom-none sample that can be identified in either the SDSSor
HST imaging.It is therefore tempting to speculate that the outflows inthe disk-dom-outflow sample are the result of inflows ofgas driven by bars. We must first consider that this sampleis by no means complete, as it is drawn from a larger sampleof 101 disk-dominated AGN from SSL17 (0 . < z < . (cid:38)
30% (again,a lower limit due to the edge-on nature of some galax-ies in the sample) consistent with typical fractions of barsfound across the population of SDSS galaxies (29 . ± . . < z < .
06; Masters et al. 2011). However, we detected58 of these AGN with spectroscopically confirmed outflows, c (cid:13) , 000–000 Smethurst et al. 2019
Table 2.
Properties of the 12 disk-dom-outflow galaxies, with outflow rates calculated from the extent and flux of [O iii ] 5007˚A innarrow band imaging taken with the Shane-3m telescope at the Lick Observatory. Neville, Hermione and Cho have lower limits ontheir calculated [O iii ] gas masses since the narrow band image was contaminated by gas ionised by star formation (see Figure 7 andSection 3.1). Note that the uncertainties are not included in the upper and lower limits, we simply state the uncertainties alongside thelimits. Snape and Crabbe have upper limits set by the measured gas mass in Blaize from the disk-dom-none sample (see Section 4).Name log [M BH /M (cid:12) ] ∗ λ Edd * ˙ m *[M (cid:12) yr − ] log [M OIII /M (cid:12) ] Outflow Rate[M (cid:12) yr − ] Inflow Rate[M (cid:12) yr − ] % accreted t outflow [Myr]Ron 8 . ± .
11 0 . ± .
55 0 . ± .
19 8 . ± .
04 0 . ± .
03 0 . ± .
20 19 ±
4% 921 ± . ± .
12 0 . ± .
22 0 . ± . > . ± . > . ± . > . ± . < ±
3% 6 ± . ± .
14 0 . ± .
385 0 . ± . > . ± . > . ± . > . ± . < ±
5% 9 ± . ± .
10 0 . ± .
29 0 . ± .
04 7 . ± .
03 1 . ± .
25 1 . ± .
25 1 ±
3% 16 ± . ± .
28 0 . ± .
19 0 . ± .
04 6 . ± .
03 0 . ± .
12 0 . ± .
16 18 ±
2% 25 ± . ± .
11 0 . ± .
12 0 . ± .
24 6 . ± . < < . ± . < . ± . > ±
2% 228 ± . ± .
13 0 . ± .
62 0 . ± .
16 7 . ± .
10 1 . ± .
35 2 . ± .
37 4 ±
2% 35 ± . ± .
22 0 . ± .
01 0 . ± .
10 7 . ± .
17 0 . ± .
04 0 . ± .
13 31 ±
2% 101 ± . ± .
11 0 . ± .
32 0 . ± .
06 6 . ± . < < . ± . < . ± . > ±
2% 22 ± . ± .
10 0 . ± .
46 0 . ± .
04 7 . ± .
74 2 . ± .
42 2 . ± .
43 2 ±
2% 4 ± . ± .
37 0 . ± .
60 0 . ± .
13 6 . ± .
03 0 . ± .
42 1 . ± .
49 11 ±
2% 8 ± . ± .
09 0 . ± .
080 0 . ± . > . ± . > . ± . > . ± . < ±
2% 86 ±
10* Measurements from SSL17. Black hole masses are calculated using a virial assumption by measuring the full width half maximum ofthe broadened H α component. Eddington ratios and black hole mass accretion rates are calculated using the bolometric luminosity ofthe AGN, inferred from the WISE W3 band at 12 µm , applying a correction from (Richards et al. 2006, see Section 2.1). The large errorson ˙ m and λ Edd are due to the propagation of uncertainties from the WISE W3 magnitudes. of which 33% hosted a bar, compared with 26% of the 43AGN without spectroscopically confirmed outflows. A χ contingency test ( scipy.stats.chi2 contingency ) revealsthat this difference is not statistically significant ( p = 0 . . σ ).However, the 12 galaxies in the disk-dom-outflow sample were specifically selected to havethe brightest blueshifted [O iii ] 5007˚A emission, and there-fore the brightest outflows. Once again it is thereforetempting to postulate that it is specifically these bright-est outflows which are powered by the inflow of gas tothe AGN by a bar (which simulations have shown caninflow material at a higher rate than spiral arms or coldaccretion; see Section 5.1). However, a χ contingencytest ( scipy.stats.chi2 contingency ) reveals that thefraction of bars in the disk-dom-outflow sample (66%;0 . < z < .
08) compared to the overall parent sample of101 disk-dominated AGN ( ∼ . < z < . σ level ( p = 0 . . σ ).Although a 2 . σ result is promising, it does not allow us tomake definitive conclusions. This is in part due to the small number statistics we are working with, but we must also bewary of the fact that the presence of a bar is correlated withstellar mass (Nair & Abraham 2010), colour (Masters et al.2011) and environment (Noguchi 1988; Moore et al. 1996;Skibba et al. 2012). Future work therefore needs to controlfor these effects in a larger sample of disk-dominated AGN,with and without outflows, before making conclusionson whether the bars in these systems are responsible forfuelling AGN. We now compare the properties of our disk-dom-outflow sample with a sample from Bae et al. (2017, here-after B17) of 20 nearby (0 . < z < . Note that we are comparing the Type 2 AGN of B17 withthe Type 1 AGN of the disk-dom-outflow sample under theassumption of AGN unification theory Urry & Padovani (1995).c (cid:13) , 000–000 ecularly powered accretion and outflows log ( M BH /M (cid:12) ) . . . . . . n o r m a li s e dd e n s i t y Bae et al . (2017) diskdomoutflow p = 0 .
31 (1 . σ ) 43 44 45 log ( L bol [erg s − ]) . . . . . . σ ) − − log ( λ Edd ) . . . . . .
17 (1 . σ )0 .
000 0 .
025 0 .
050 0 .
075 0 .
100 0 . ˙m [M (cid:12) yr − ] n o r m a li s e dd e n s i t y p = 0 . . σ ) 0 2 4 6 8 10 ˙M outflow [M (cid:12) yr − ] . . . . . . . . .
009 (2 . σ ) − − − −
100 0 100 200 v [OIII] [km s − ] . . . . . . .
006 p = 0 . . σ ) Figure 10.
Comparison between the properties of the disk-dom-outflow sample (solid histograms) and the Bae et al. (2017) sample of20 AGN with merger histories (dashed lines). The secularly fueled AGN of the disk-dom-outflow sample have SMBHs with statisticallysimilar masses and Eddington ratios. Similarly, they launch outflows at the same velocity. However, the disk-dom-outflow sample havestatistically significant higher bolometric luminosities (and consequently higher black hole accretion rates), but lower mass outflow rates.We argue that this is a product of both the smooth planar inflow spinning up the black hole and the geometry in a secularly fuelledAGN system (see Section 5.3 and Figure 11). with merger dominated histories. B17 use integral field spec-troscopy to spatially resolve the outflow properties of theirAGN, rather than narrow band imaging. This allowed B17to empirically determine the column densities of the ionisedgas, n e , using the [S ii ] line ratio and separate star forma-tion ionisation from the outflow component spatially acrossthe galaxy. Such a method is therefore advantageous overnarrow band imaging (and one we intend to exploit in thefuture, see Section 5.4) as it is more precise and requiresless assumptions about the nature of the outflowing gas.B17 also used the M − σ ∗ relation of Park et al. (2012) toderive black hole masses (rather than the virial assumptionof Greene & Ho 2005 as implemented by SSL17) and cal-culated bolometric luminosities from the luminosity of the[O iii ] emission (see Heckman et al. 2004). Unfortunately B17do not provide their measured uncertainties when quotingtheir derived properties.With these caveats in mind, we examine the differencein the distributions of the calculated black hole masses, bolo-metric luminosities, Eddington ratios, black hole accretionrates, velocities of the outflowing [O iii ] gas and outflow ratesin Figure 10. In each panel we provide the p -value, and corre-sponding σ value, of a 2D Kolmogorov-Smirnov (KS) test todetermine if the properties of the disk-dom-outflow andthe B17 samples are statistically distinguishable.Whilst the black hole masses, Eddington ratios and out-flow velocities of the two samples are not significantly differ-ent, the KS tests revealed that the bolometric luminosities,black hole mass accretion rates and outflow rates are statis- tically distinguishable (see Figure 10). The outflow rates ofthe disk-dom-outflow sample exceed the accretion ratesof their SMBHs by a factor of ∼
18 on average. In compari-son, the outflow rates of the B17 AGN exceed the black holeaccretion rates by a factor of ∼
260 on average. From expertvisual inspection of the SDSS images for the B17 sample wehave determined that there are 8 barred galaxies (40%) and2 mergers (10%). However, all of the B17 sources have ob-vious bulges suggesting their evolutionary history has beenimpacted by the effects of minor and major mergers. Thiscould manifest as a recent effect driving the SMBH accre-tion and outflows, or if the merger occurred earlier than theAGN lifetime, by changing the dynamical structures aroundthe SMBH so that it operates in a different potential.Merger driven inflows are thought to occur quasi-spherically; accretion will be chaotic, with gas fed into thecentre from all angles (Sanders 1981; Moderski et al. 1998;King & Pringle 2006). This is in contrast to secularly drivengrowth where smooth accretion is thought to occur via a pla-nar inflow (Nayakshin et al. 2012; Reynolds 2013). We invokethese different accretion mechanisms to explain the differentproperties of the disk-dom-outflow and B17 samples.The galaxies of the disk-dom-outflow sample havesignificantly higher (4 . σ ) black hole accretion rates (mean˙ m ∼ .
05 M (cid:12) yr − ) than the B17 sample (mean ˙ m ∼ c (cid:13)000
05 M (cid:12) yr − ) than the B17 sample (mean ˙ m ∼ c (cid:13)000 , 000–000 Smethurst et al. 2019
Figure 11.
Toy model of accretion for a secularly (left) and merger (right) fed supermassive black hole to account for the results of thiswork discussed in Section 5.3, adapted from Nayakshin et al. (2012). The black hole accretion rates of the disk-dom-outflow sampleare 5 times higher (4 . σ ) and the outflow rates are 5 times lower (2 . σ ) than for a sample of 20 AGN from Bae et al. (2017) with mergerdominated histories. We account for these differences by considering the effect of the planar accretion spinning up the black hole toincrease the accretion efficiency and how the geometry of the two systems causes a larger feedback effect in the merger scenario resultingin a more massive outflow (see Section 5.3). .
01 M (cid:12) yr − ). The accretion rates in the secularly fuelled disk-dom-outflow sample are therefore a factor of ∼ η = 0 . Due to the large uncertainties on ˙ m propagated from the WISEW3 magnitude uncertainties (see Table 2), we performed a boot-strap test to determine the robustness of this 4 . σ result. Foreach source, we shifted the ˙ m value by a randomly sampled valuefrom a Gaussian distribution centred on zero and ranging to ± the uncertainty on the accretion rate, and recalculated the p -value and significance. We repeated this 10000 times and foundthat for 6364 of the iterations, the KS-test had a greater than 3 σ significance. The minimum value found was 1 . σ . We thereforebelieve that our statement that the accretion rates in our disk-dom-outflow sample are statistically significantly higher thanin the B17 sample is robust. whereas we assumed a value of η = 0 .
15 (see Equation 5).We therefore corrected the ˙ m values given by B17 to use η = 0 .
15 so that they are directly comparable to the onesderived in SSL17. Since ˙ m is inversely proportional to η (seeEquation 5), this correction resulted in a decrease of the ac-cretion rates from those quoted in B17. If we do not correctthe value of radiative efficiency, η = 0 .
1, used to calculate theaccretion rates in the B17 sample, this causes the differencein accretion rates compared to the disk-dom-outflow sam-ple to become slightly less pronounced; a factor of ∼ . σ ) on average. By correcting the B17 accretion rates touse the same value of η as the disk-dom-outflow sam-ple, as in Figure 10, we are making the most conservativeassumption we can about the radiative efficiencies of thesesources, yet without that correction we still find a significant(3 . σ ) difference in their calculated accretion rates.This difference in accretion rates between the disk-dom-outflow and B17 samples is not unsurprising if one c (cid:13) , 000–000 ecularly powered accretion and outflows considers basic accretion physics. If material with a con-stant angular momentum is fed to the black hole, this willspin up the black hole (King et al. 2008; Davis & Laor 2011;Reynolds 2013), increasing the temperature of the accretiondisk whilst reducing the radius of the inner stable circularorbit (ISCO) and consequently increasing the accretion effi-ciency of the black hole (Thorne 1974; Reynolds 2013).Since the inflows in the disk-dom-outflow sample aredriven by secular processes, resulting in smooth planar ac-cretion of gas with the same angular momentum direction,this will indeed cause the spin of these black holes to in-crease. Conversely, in the merger grown AGN of the B17sample, the chaotic, sporadic accretion of gas from a quasi-spherical inflow should eventually spin down the black hole,reducing its accretion efficiency. This is true of both cur-rent and past AGN episodes, so even if the AGN of theB17 sample have not had their current episode triggeredby a merger, their SMBH spin should still be lower due tothe spin down effects of previous merger accretion eventsin their evolutionary history. Therefore this theorised differ-ence in black hole spin due to feeding mechanism could bethe cause of the higher black hole accretion rates seen forthe disk-dom-outflow sample. This increase in the spinwill cause the radiative efficiency, η , to increase (Shakura& Sunyaev 1973), suggesting that using a higher value of η compared to the B17 sample may be more appropriate toestimate the black hole accretion rates of the disk-dom-outflow sample. However, as discussed above, this stillleads to a significant (3 . σ ) difference in their calculatedaccretion rates.Unfortunately, simulations are currently not able to si-multaneously cover the range of scales involved in the largescale galactic inflows and small scale black hole accretion.Similarly, it is incredibly difficult to estimate a timescalefor observable effects on the galaxy itself, as the simulationsdo not span large enough scales to simulate the effects ofboth the chaotic or smooth accretion and the outflow onthe entire galaxy. Whilst many works have studied the rela-tion between the spin of the black hole and the powering ofcollimated radio jets through the tangling of magnetic fieldlines (e.g. Ruffini & Wilson 1975; Blandford & Znajek 1977;Koide et al. 2002; Benson & Babul 2009; Tchekhovskoy et al.2011; Gong & Jiang 2014), there is little consensus about thelink, if any, between spin and outflow rates, as outflows arenot accounted for in any of the main accretion models (seereview by Abramowicz & Fragile 2013). We are reluctanttherefore to attribute the statistical difference between theoutflow rates of the disk-dom-outflow and B17 samplesto the spin of the black holes. The mean outflow rates inthe secularly fuelled disk-dom-outflow sample are a fac-tor of ∼ . σ ) than the merger fuelled sample ofB17. Instead we consider whether this difference is due to acombination of two effects; feeding and geometry.Firstly mergers, both minor and major, are known todrive gas to the centres of galaxies at a larger rate than in-ternal, calm, secular processes (e.g. Hernquist 1989; Barnes& Hernquist 1991, 1996; Hopkins et al. 2005; Kazantzidiset al. 2005; Mayer et al. 2010; Naab & Ostriker 2017). By asimple consideration of conservation of energy, if the galax-ies in the B17 sample have large amounts of gas fed to theircentres where their black hole accretion efficiency is limitedby their spin, then a massive outflow of gas must result. These outflows are often modelled as quasi-spherical,essentially averaging over the many orientations of the ac-cretion disk due to chaotic accretion (Sanders 1981; King &Nixon 2015). In the case of planar inflows, it is thought abiconical outflow will be set up out of the plane of the accre-tion disc (Shlosman & Vitello 1993). By a simple considera-tion of geometry, it is clear that a quasi-spherical inflow willexperience a greater feedback effect from a quasi-sphericaloutflow than a planar inflow will from a biconical outflow(Nayakshin & Power 2010; Nayakshin et al. 2012; Angl´es-Alc´azar et al. 2015). This feedback effect in a merger drivengrowth scenario will cause some of the inflowing material tobe picked up, adding mass to the outflow whilst prevent-ing gas in the inflow from reaching the central regions ofthe galaxy. We summarise the effects of the different feedingmechanisms in Figure 11.The scenario summarised in Figure 11 has previouslybeen considered in Nayakshin et al. (2012) with a simpletheoretical consideration of the effects of accretion via thetwo different mechanisms. Nayakshin et al. (2012) then con-cluded that under this hypothesis, secularly grown super-massive black holes should be over-massive in comparisonto merger grown black holes, whereas measurements of theblack hole masses of galaxy with psuedo-bulges had shownthat this was not the case (Hu 2008; Graham 2008; Kor-mendy et al. 2011). However, the work of Nayakshin et al.was published prior to the results of SSL17 and Martin et al.(2018) which showed that SMBHs in disk-dominated ’bulge-less’ galaxies were over-massive (by 2dex) given their bulgesize (or lack thereof). This along with the differences foundin this work between the black hole accretion rates andoutflow rates of the disk-dom-outflow and the B17 sam-ples, therefore suggests that the hypothesis first outlined inNayakshin et al. (2012) and summarised in Figure 11, mayaccount for their differences in black hole accretion rate andoutflow rates. Such a scenario would naturally give rise tothe dominance of secular mechanisms in the growth of su-permassive black holes. Whilst the narrow band imaging method used in this studydoes allow us to constrain the inflow rates to the AGN ofthe disk-dom-outflow sample, it is limited in a number ofways: (i) by the PSF of the narrow band filters and (ii) bycontamination from star formation ionisation emission.Firstly, the inability to resolve the shape and extent ofthe outflow with increasing redshift is a significant limita-tion to this study. This is in part due to the limiting PSFof the Shane-3m telescope but also due to the aberrationsof the PSF caused by refraction through the different nar-row band filters. These limitations particularly affect thecentral PSF subtraction to remove the narrow [O iii ] emis-sion ionised by the AGN. It is likely that the AGN has beenover-subtracted in these images, leading to an underestimateof the outflow rates in these sources (albeit not enough tocause the differences observed between the outflow rates inthe disk-dom-outflow and B17 samples). The limited PSFalso affects our ability to accurately determine the largestradial extent and morphology of the outflow. All sourcesin the disk-dom-outflow sample would therefore bene-fit from future observations with sub-arcsecond resolution c (cid:13) , 000–000 Smethurst et al. 2019 provided by space based optical observatories. For example,see the [O iii ] narrow band Hubble Space Telescope obser-vations of AGN driven outflows in a sample of ULIRGs byTadhunter et al. (2018).The greatest limitation to a complete study of outflowsin disk-dominated AGN is the inability to remove narrowband emission ionised by star formation from narrow bandimaging. In order to tackle this problem, the narrow andbroad emission needs to be spatially decomposed across thegalaxy using Integral Field Spectroscopy (IFS). Dependingon the resolution of the unit employed, such observationswill also allow for a more accurate determination of the ex-tent of the outflow. Keck Cosmic Web Imager (KCWI) datafor four sources in the disk-dom-outflow sample (Harry,Neville, Padma, Theodore) has therefore been acquired dur-ing a 2018B observing run and will be analysed in futurework.
We have observed a sample of 12 disk-dominated AGN,with spectroscopically confirmed outflows, in narrow bandfilters with the Shane-3m telescope at the Lick Observa-tory. This is the disk-dom-outflow sample. By studyinggalaxies with disk-dominated morphologies, we can isolatethose systems with evolutionary histories dominated by sec-ular mechanisms. Images were obtained in filters centred onthe [O iii ] 5007˚A emission and nearby continuum for eachsource. The images were reduced using the ccdproc pack-age, flux calibrated using standard stars and normalised bytheir exposure times. The continuum image was then sub-tracted from the [O iii ] 5007˚A image.From this continuum subtracted image, the central PSFof the AGN [O iii ] emission was removed by extracting a PSFfrom a standard star image in the same narrow band filter.The remaining flux within the Petrosian radius of the sourcewas then summed to give the [O iii ] flux in the outflow. Byassuming density and metallicity properties of the outflow-ing gas, the [O iii ] flux was converted to give the amount ofmass in the outflow. Combining this measurement with theextent of the outflow in the PSF subtracted image and thevelocity shift from the narrow [O iii ] emission in the SDSSspectra, the outflow rate from the AGN was calculated. Theinflow rate was then calculated assuming that this equatedto at most the outflow rate plus the accretion rate of theblack hole (previously derived by SSL17). Our findings canbe summarised as follows:(i) The mean outflow rate from the AGN in the disk-dom-outflow sample is 0 . ± .
14 M (cid:12) yr − . This ex-ceeds the mean accretion rate of their SMBHs (0 . ± .
039 M (cid:12) yr − ) by a factor of ∼
18, giving an average inflowrate of ∼ . ± .
14 M (cid:12) yr − .(ii) Bars, spiral arms and cold accretion of gas have allbeen shown in simulations to be capable of providing over ∼ . (cid:12) yr − to the central regions of a galaxy. Thesemechanisms can be sustained over long periods of time, inexcess of the maximum outflow timescale derived for the disk-dom-outflow sample of 920 Myr. This suggests thatsecular processes are more than sufficient at fuelling blackhole growth, at least at z ∼
0. (iii) The majority (66%; a lower limit due to the edge-onnature of some of the galaxies in the sample) of the disk-dom-outflow sample host a strong bar, an excess whichis marginally significant ( p = 0 . σ = 2 .
1) comparedto a parent sample of 101 disk-dominated AGN studied inSimmons, Smethurst & Lintott (2017). Although temptingto speculate that the outflows and SMBH growth of theAGN in this study are powered by gas inflows driven bybars, this result does not account for the dependence of barfraction on stellar mass, colour or environment. Future workis therefore still needed.(iv) We compare our disk-dom-outflow sample of sec-ularly fuelled AGN with a sample of 20 nearby AGN fromBae et al. (2017, B17) with merger histories. We find thatthe accretion rates of the black holes in the disk-dom-outflow sample are ∼ . σ ) than thosein the B17 sample. This is further evidence that secu-lar processes are sufficient to fuel black hole growth at z ∼
0. We consider how the black holes of the disk-dom-outflow sample are spun up due to their smooth accre-tion of gas occurring in a single plane of the galaxy disc; asopposed to the chaotic quasi-spherical accretion occurringin the merger fuelled B17 sample, which results in a lowerspin black hole. Since the spin of a black hole is known tocorrelate with accretion efficiency, we attribute the differ-ence in black hole accretion rates between the disk-dom-outflow and B17 sample to a possible difference in spin.(v) In contrast to the accretion rates, the outflow rates inthe disk-dom-outflow sample are ∼ . σ )than those in the B17 sample. We suggest that the differ-ent geometry of the accretion discs and outflows in the twosystems causes this difference. For the secularly fed blackholes of the disk-dom-outflow sample, a steady biconi-cal outflow from the accretion disc will result from the pla-nar accretion. For the merger grown black holes of the B17sample a quasi-spherical outflow will result from the chaoticquasi-spherical accretion. By a simple consideration of geom-etry, it is clear that a quasi-spherical inflow will experiencea greater feedback effect from a quasi-spherical outflow thana planar inflow will from a biconical outflow. This results ina much larger outflow rate from a merger fed AGN, possi-bly explaining the differences in outflow rates between the disk-dom-outflow and B17 samples.(vi) Future work is still needed to address the limitationswith the PSF aberrations from the narrow band imagingand to disentangle the [O iii ] emission from star formationionisation. This can be achieved by utilising space basedobservatories and integral field spectroscopy for future ob-servations.The results in this work suggest that the hypothesisfirst outlined in Nayakshin et al. (2012) and summarised inFigure 11, combining the effects of black hole spin and accre-tion geometry, may account for the differences in the growthrate of supermassive black holes and outflows in AGN, givingrise to the dominance (65%; Martin et al. 2018) of secularmechanisms in the growth of supermassive black holes. ACKNOWLEDGEMENTS
The authors would like to thank Roger Davies, JamesMatthews, Adam Ingram and Sam Vaughan for many useful c (cid:13) , 000–000 ecularly powered accretion and outflows discussions which contributed to the interpretation of theseresults.RJS gratefully acknowledges funding from ChristChurch, Oxford.BDS gratefully acknowledges support from the NationalAeronautics and Space Administration (NASA) throughEinstein Postdoctoral Fellowship Award Number PF5-160143 issued by the Chandra X-ray Observatory Center,which is operated by the Smithsonian Astrophysical Obser-vatory for and on behalf of NASA under contract NAS8-03060.This research made use of Astropy, a community-developed core Python package for Astronomy (AstropyCollaboration et al. 2013, 2018) and the affiliated ccdproc package (Craig et al. 2015).Funding for the Sloan Digital Sky Survey IV has beenprovided by the Alfred P. Sloan Foundation, the U.S. De-partment of Energy Office of Science, and the ParticipatingInstitutions. SDSS acknowledges support and resources fromthe Center for High-Performance Computing at the Univer-sity of Utah. The SDSS web site is .SDSS is managed by the Astrophysical Research Con-sortium for the Participating Institutions of the SDSS Col-laboration including the Brazilian Participation Group, theCarnegie Institution for Science, Carnegie Mellon Univer-sity, the Chilean Participation Group, the French Par-ticipation Group, Harvard-Smithsonian Center for Astro-physics, Instituto de Astrof´ısica de Canarias, The JohnsHopkins University, Kavli Institute for the Physics andMathematics of the Universe (IPMU) / University of Tokyo,Lawrence Berkeley National Laboratory, Leibniz Institut frAstrophysik Potsdam (AIP), Max-Planck-Institut f¨ur As-tronomie (MPIA Heidelberg), Max-Planck-Institut fr Astro-physik (MPA Garching), Max-Planck-Institut f¨ur Extrater-restrische Physik (MPE), National Astronomical Observato-ries of China, New Mexico State University, New York Uni-versity, University of Notre Dame, Observat´orio Nacional /MCTI, The Ohio State University, Pennsylvania State Uni-versity, Shanghai Astronomical Observatory, United King-dom Participation Group, Universidad Nacional Aut´onomade M´exico, University of Arizona, University of ColoradoBoulder, University of Oxford, University of Portsmouth,University of Utah, University of Virginia, University ofWashington, University of Wisconsin, Vanderbilt Universityand Yale University. REFERENCES
Aalto S., Muller S., Sakamoto K., Gallagher J. S., Mart´ınS., Costagliola F., 2012, A&A, 546, A68Abramowicz M. A., Fragile P. C., 2013, Living Reviews inRelativity, 16, 1Ahn C. P., Alexandroff R., Allende Prieto C., AndersonS. F., Anderton T., Andrews B. H., Aubourg ´E., BaileyS., Balbinot E., Barnes R., et al. 2012, ApJS, 203, 21Aihara H., Allende Prieto C., An D., Anderson S. F.,Aubourg ´E., Balbinot E., Beers T. C., Berlind A. A., Bick- erton S. J., Bizyaev D., Blanton M. R., Bochanski J. J.,Bolton A. S., SDSS Collaboration 2011, ApJS, 193, 29Alatalo K., 2015, ApJ, 801, L17Alatalo K., Blitz L., Young L. M., Davis T. A., Bureau M.,Lopez L. A., Cappellari M., 2011, ApJ, 735, 88Angl´es-Alc´azar D., ¨Ozel F., Dav´e R., Katz N., KollmeierJ. A., Oppenheimer B. D., 2015, ApJ, 800, 127Astropy Collaboration Price-Whelan A. M., Sip˝ocz B. M.,G¨unther H. M., Lim P. L., Crawford S. M., Astropy Con-tributors 2018, AJ, 156, 123Astropy Collaboration Robitaille T. P., Tollerud E. J.,Greenfield P., Droettboom M., Bray E., Aldcroft T., DavisM., Ginsburg A., Price-Whelan A. M., Kerzendorf W. E.,Conley A., Crighton N., 2013, A&A, 558, A33Bae H.-J., Woo J.-H., Karouzos M., Gallo E., Flohic H.,Shen Y., Yoon S.-J., 2017, ApJ, 837, 91Barnes J. E., Hernquist L., 1996, ApJ, 471, 115Barnes J. E., Hernquist L. E., 1991, ApJ, 370, L65Benson A. J., Babul A., 2009, MNRAS, 397, 1302Blandford R. D., Znajek R. L., 1977, MNRAS, 179, 433Bradley L., Sipocz B., Robitaille T., Tollerud E., , 2019,astropy/photutils: v0.6Carniani S., Marconi A., Maiolino R., Balmaverde B.,Brusa M., Cano-D´ıaz M., Cicone C., Comastri A., CresciG., Fiore F., Feruglio C., La Franca F., Mainieri V., 2015,A&A, 580, A102Cicone C., Maiolino R., Sturm E., Graci´a-Carpio J., Fer-uglio C., 2014, A&A, 562, A21Craig M. W., Crawford S. M., Deil C., Gomez C., G¨untherH. M., Heidt N., Horton A., ccdproc contributors, 2015,ccdproc: CCD data reduction software, AstrophysicsSource Code LibraryDavies R. I., Maciejewski W., Hicks E. K. S., Tacconi L. J.,Genzel R., Engel H., 2009, ApJ, 702, 114Davis S. W., Laor A., 2011, ApJ, 728, 98D’Onghia E., Vogelsberger M., Hernquist L., 2013, ApJ,766, 34Donner K. J., Thomasson M., 1994, A&A, 290, 785Edelson R., Malkan M., 2012, ApJ, 751, 52Elvis M., Risaliti G., Zamorani G., 2002, ApJ, 565, L75Feruglio C., Maiolino R., Piconcelli E., Menci N., AusselH., Lamastra A., Fiore F., 2010, A&A, 518, L155Gallagher R., Maiolino R., Belfiore F., Drory N., Riffel R.,Riffel R. A., 2019, MNRAS, 485, 3409Galloway M. A., Willett K. W., Fortson L. F., CardamoneC. N., Schawinski K., Cheung E., Lintott C. J., MastersK. L., Melvin T., Simmons B. D., 2015, MNRAS, 448,3442Gong X.-L., Jiang Z.-X., 2014, Chinese Physics Letters, 31,089801Goulding A. D., Matthaey E., Greene J. E., Hickox R. C.,Alexander D. M., Forman W. R., Jones C., Lehmer B. D.,Griffis S., Kanek S., Oulmakki M., 2017, ApJ, 843, 135Graham A. W., 2008, ApJ, 680, 143Greene J. E., Ho L. C., 2005, ApJ, 630, 122H¨aring N., Rix H.-W., 2004, ApJ, 604, L89Heckman T. M., Kauffmann G., Brinchmann J., CharlotS., Tremonti C., White S. D. M., 2004, ApJ, 613, 109Hernquist L., 1989, Nature, 340, 687Hopkins P. F., Cox T. J., Younger J. D., Hernquist L.,2009, ApJ, 691, 1168Hopkins P. F., Hernquist L., Cox T. J., Di Matteo T., Mar- c (cid:13)000
Aalto S., Muller S., Sakamoto K., Gallagher J. S., Mart´ınS., Costagliola F., 2012, A&A, 546, A68Abramowicz M. A., Fragile P. C., 2013, Living Reviews inRelativity, 16, 1Ahn C. P., Alexandroff R., Allende Prieto C., AndersonS. F., Anderton T., Andrews B. H., Aubourg ´E., BaileyS., Balbinot E., Barnes R., et al. 2012, ApJS, 203, 21Aihara H., Allende Prieto C., An D., Anderson S. F.,Aubourg ´E., Balbinot E., Beers T. C., Berlind A. A., Bick- erton S. J., Bizyaev D., Blanton M. R., Bochanski J. J.,Bolton A. S., SDSS Collaboration 2011, ApJS, 193, 29Alatalo K., 2015, ApJ, 801, L17Alatalo K., Blitz L., Young L. M., Davis T. A., Bureau M.,Lopez L. A., Cappellari M., 2011, ApJ, 735, 88Angl´es-Alc´azar D., ¨Ozel F., Dav´e R., Katz N., KollmeierJ. A., Oppenheimer B. D., 2015, ApJ, 800, 127Astropy Collaboration Price-Whelan A. M., Sip˝ocz B. M.,G¨unther H. M., Lim P. L., Crawford S. M., Astropy Con-tributors 2018, AJ, 156, 123Astropy Collaboration Robitaille T. P., Tollerud E. J.,Greenfield P., Droettboom M., Bray E., Aldcroft T., DavisM., Ginsburg A., Price-Whelan A. M., Kerzendorf W. E.,Conley A., Crighton N., 2013, A&A, 558, A33Bae H.-J., Woo J.-H., Karouzos M., Gallo E., Flohic H.,Shen Y., Yoon S.-J., 2017, ApJ, 837, 91Barnes J. E., Hernquist L., 1996, ApJ, 471, 115Barnes J. E., Hernquist L. E., 1991, ApJ, 370, L65Benson A. J., Babul A., 2009, MNRAS, 397, 1302Blandford R. D., Znajek R. L., 1977, MNRAS, 179, 433Bradley L., Sipocz B., Robitaille T., Tollerud E., , 2019,astropy/photutils: v0.6Carniani S., Marconi A., Maiolino R., Balmaverde B.,Brusa M., Cano-D´ıaz M., Cicone C., Comastri A., CresciG., Fiore F., Feruglio C., La Franca F., Mainieri V., 2015,A&A, 580, A102Cicone C., Maiolino R., Sturm E., Graci´a-Carpio J., Fer-uglio C., 2014, A&A, 562, A21Craig M. W., Crawford S. M., Deil C., Gomez C., G¨untherH. M., Heidt N., Horton A., ccdproc contributors, 2015,ccdproc: CCD data reduction software, AstrophysicsSource Code LibraryDavies R. I., Maciejewski W., Hicks E. K. S., Tacconi L. J.,Genzel R., Engel H., 2009, ApJ, 702, 114Davis S. W., Laor A., 2011, ApJ, 728, 98D’Onghia E., Vogelsberger M., Hernquist L., 2013, ApJ,766, 34Donner K. J., Thomasson M., 1994, A&A, 290, 785Edelson R., Malkan M., 2012, ApJ, 751, 52Elvis M., Risaliti G., Zamorani G., 2002, ApJ, 565, L75Feruglio C., Maiolino R., Piconcelli E., Menci N., AusselH., Lamastra A., Fiore F., 2010, A&A, 518, L155Gallagher R., Maiolino R., Belfiore F., Drory N., Riffel R.,Riffel R. A., 2019, MNRAS, 485, 3409Galloway M. A., Willett K. W., Fortson L. F., CardamoneC. N., Schawinski K., Cheung E., Lintott C. J., MastersK. L., Melvin T., Simmons B. D., 2015, MNRAS, 448,3442Gong X.-L., Jiang Z.-X., 2014, Chinese Physics Letters, 31,089801Goulding A. D., Matthaey E., Greene J. E., Hickox R. C.,Alexander D. M., Forman W. R., Jones C., Lehmer B. D.,Griffis S., Kanek S., Oulmakki M., 2017, ApJ, 843, 135Graham A. W., 2008, ApJ, 680, 143Greene J. E., Ho L. C., 2005, ApJ, 630, 122H¨aring N., Rix H.-W., 2004, ApJ, 604, L89Heckman T. M., Kauffmann G., Brinchmann J., CharlotS., Tremonti C., White S. D. M., 2004, ApJ, 613, 109Hernquist L., 1989, Nature, 340, 687Hopkins P. F., Cox T. J., Younger J. D., Hernquist L.,2009, ApJ, 691, 1168Hopkins P. F., Hernquist L., Cox T. J., Di Matteo T., Mar- c (cid:13)000 , 000–000 Smethurst et al. 2019 tini P., Robertson B., Springel V., 2005, ApJ, 630, 705Hopkins P. F., Quataert E., Murray N., 2012, MNRAS,p. 2655Hu J., 2008, MNRAS, 386, 2242Hunt J. A. S., Hong J., Bovy J., Kawata D., Grand R. J. J.,2018, MNRAS, 481, 3794Kaviraj S., Cohen S., Ellis R. S., Peirani S., WindhorstR. A., O’Connell R. W., Silk J., Whitmore B. C., HathiN. P., Ryan R. E., Dopita M. A., Frogel J. A., Dekel A.,2013, MNRAS, 428, 925Kazantzidis S., Mayer L., Colpi M., Madau P., DebattistaV. P., Wadsley J., Stadel J., Quinn T., Moore B., 2005,ApJ, 623, L67Kereˇs D., Katz N., Weinberg D. H., Dav´e R., 2005, MN-RAS, 363, 2King A., Nixon C., 2015, MNRAS, 453, L46King A. R., Pringle J. E., 2006, MNRAS, 373, L90King A. R., Pringle J. E., Hofmann J. A., 2008, MNRAS,385, 1621Koide S., Shibata K., Kudoh T., Meier D. L., 2002, Science,295, 1688Kormendy J., Bender R., Cornell M. E., 2011, Nature, 469,374Lee G.-H., Woo J.-H., Lee M. G., Hwang H. S., Lee J. C.,Sohn J., Lee J. H., 2012, ApJ, 750, 141Lin L.-H., Wang H.-H., Hsieh P.-Y., Taam R. E., Yang C.-C., Yen D. C. C., 2013, ApJ, 771, 8Maciejewski W., 2004, MNRAS, 354, 892Maciejewski W., Teuben P. J., Sparke L. S., Stone J. M.,2002, MNRAS, 329, 502Manzano-King C., Canalizo G., Sales L., 2019, arXiv e-prints, p. arXiv:1905.09287Marconi A., Hunt L. K., 2003, ApJ, 589, L21Martig M., Bournaud F., Croton D. J., Dekel A., TeyssierR., 2012, ApJ, 756, 26Martin G., Kaviraj S., Volonteri M., Simmons B. D., De-vriendt J. E. G., Lintott C. J., Smethurst R. J., DuboisY., Pichon C., 2018, MNRAS, 476, 2801Masters K. L., Nichol R. C., Hoyle B., Lintott C., BamfordS. P., Edmondson E. M., Fortson L., Keel W. C., Schaw-inski K., Smith A. M., Thomas D., 2011, MNRAS, 411,2026Mayer L., Kazantzidis S., Escala A., Callegari S., 2010,Nature, 466, 1082Miller R. H., Smith B. F., 1979, ApJ, 227, 785Moderski R., Sikora M., Lasota J. P., 1998, MNRAS, 301,142Moore B., Katz N., Lake G., Dressler A., Oemler A., 1996,Nature, 379, 613616Naab T., Ostriker J. P., 2017, ARA&A, 55, 59Nair P. B., Abraham R. G., 2010, ApJL, 714, L260L264Nayakshin S., Power C., 2010, MNRAS, 402, 789Nayakshin S., Power C., King A. R., 2012, ApJ, 753, 15Noguchi M., 1988, A&A, 203, 259Oh S., Oh K., Yi S. K., 2012, The Astrophysical JournalSupplement Series, 198, 4Park D., Kelly B. C., Woo J.-H., Treu T., 2012, ApJS, 203,6Penny S. J., Masters K. L., Smethurst R., Nichol R. C.,Krawczyk C. M., Bizyaev D., Greene O., Liu C., MarinelliM., Rembold S. B., Riffel R. A., Ilha G. d. S., WylezalekD., Andrews B. H., Bundy K., Drory N., Oravetz D., Pan K., 2018, MNRAS, 476, 979Perna M., Lanzuisi G., Brusa M., Cresci G., Mignoli M.,2017, A&A, 606, A96Planck Collaboration Ade P. A. R., Aghanim N., ArnaudM., Ashdown M., Aumont J., Baccigalupi C., BandayA. J., Barreiro R. B., Bartlett J. G., et al. 2016, A&A,594, A13Pontzen A., Tremmel M., Roth N., Peiris H. V., SaintongeA., Volonteri M., Quinn T., Governato F., 2016, ArXive-printsRegan M. W., Teuben P. J., 2004, ApJ, 600, 595Reynolds C. S., 2013, Classical and Quantum Gravity, 30,244004Richards G. T., Lacy M., Storrie-Lombardi L. J., Hall P. B.,Gallagher S. C., Hines D. C., Fan X., Papovich C., VandenBerk D. E., Trammell G. B., Schneider D. P., VestergaardM., York D. G., Jester S., Anderson S. F., Budav´ari T.,Szalay A. S., 2006, ApJS, 166, 470Rowling J. K., 1997, Harry Potter and the Philosopher’sStone, 1 edn. Vol. 1, Bloomsbury PublishingRuffini R., Wilson J. R., 1975, Phys. Rev. D, 12, 2959Sakamoto K., 1996, ApJ, 471, 173Sancisi R., Fraternali F., Oosterloo T., van der Hulst T.,2008, Astronomy and Astrophysics Review, 15, 189Sanders R. H., 1981, Nature, 294, 427Sarzi M., Falc´on-Barroso J., Davies R. L., Bacon R., Bu-reau M., Cappellari M., de Zeeuw P. T., Emsellem E.,Fathi K., Krajnovi´c D., Kuntschner H., McDermid R. M.,Peletier R. F., 2006, MNRAS, 366, 1151Schnorr-M¨uller A., Storchi-Bergmann T., Nagar N. M.,Ferrari F., 2014, MNRAS, 438, 3322Shakura N. I., Sunyaev R. A., 1973, A&A, 24, 337Shlosman I., Vitello P., 1993, ApJ, 409, 372Simmons B. D., Lintott C., Schawinski K., Moran E. C.,Han A., Kaviraj S., Masters K. L., Urry C. M., WillettK. W., Bamford S. P., Nichol R. C., 2013, MNRAS, 429,2199Simmons B. D., Smethurst R. J., Lintott C., 2017, MN-RAS, 470, 1559Skibba R. A., Masters K. L., Nichol R. C., Zehavi I., HoyleB., Edmondson E. M., Bamford S. P., Cardamone C. N.,Keel W. C., Lintott C., Schawinski K., 2012, MNRAS,423, 1485Skrutskie M. F., Cutri R. M., Stiening R., Weinberg M. D.,Schneider S., Carpenter J. M., Beichman C., Capps R.,Chester T., Elias J., Huchra J., Liebert J., 2006, AJ, 131,1163Slater R., Nagar N. M., Schnorr-M¨uller A., Storchi-Bergmann T., Finlez C., Lena D., Ramakrishnan V.,Mundell C. G., Riffel R. A., Peterson B., Robinson A.,Orellana G., 2019, A&A, 621, A83Sparke L. S., Sellwood J. A., 1987, MNRAS, 225, 653Sparre M., Springel V., 2016, ArXiv e-printsTadhunter C., Rodr´ıguez Zaur´ın J., Rose M., SpenceR. A. W., Batcheldor D., Berg M. A., Ramos AlmeidaC., Spoon H. W. W., Sparks W., Chiaberge M., 2018,MNRAS, 478, 1558Tchekhovskoy A., Narayan R., McKinney J. C., 2011, MN-RAS, 418, L79Thorne K. S., 1974, ApJ, 191, 507Tonini C., Mutch S. J., Croton D. J., Wyithe J. S. B., 2016,MNRAS, 459, 4109 c (cid:13) , 000–000 ecularly powered accretion and outflows Toomre A., 1977, in B. M. Tinsley & R. B. G. LarsonD. Campbell ed., Evolution of Galaxies and Stellar Pop-ulations Mergers and Some Consequences. p. 401Urry C. M., Padovani P., 1995, PASP, 107, 803van de Voort F., Schaye J., Booth C. M., Haas M. R., DallaVecchia C., 2011, MNRAS, 414, 2458Villar Mart´ın M., Bellocchi E., Stern J., Ramos AlmeidaC., Tadhunter C., Gonz´alez Delgado R., 2015, MNRAS,454, 439Voges W., Aschenbach B., Boller T., Br¨auninger H., BrielU., Burkert W., Dennerl K., Englhauser J., Gruber R.,Haberl F., Hartner G., 1999, A&A, 349, 389Walker I. R., Mihos J. C., Hernquist L., 1996, ApJ, 460,121Welker C., Dubois Y., Pichon C., Devriendt J., ChisariE. N., 2015, arXiv e-prints, p. arXiv:1512.00400Wright E. L., Eisenhardt P. R. M., Mainzer A. K., ResslerM. E., Cutri R. M., Jarrett T., Kirkpatrick J. D., PadgettD., McMillan R. S., Skrutskie M., 2010, AJ, 140, 1868York D. G., Adelman J., Anderson Jr. J. E., Anderson S. F.,Annis J., Bahcall N. A., Bakken J. A., Barkhouser R.,Bastian S., Berman E., Boroski W. N., Bracker S., BriegelC., Briggs J. W., Brinkmann J., SDSS Collaboration 2000,AJ, 120, 1579 c (cid:13)000