Using Megamaser Disks to Probe Black Hole Accretion
Jenny E. Greene, Anil Seth, Mark den Brok, James A. Braatz, Christian Henkel, Ai-Lei Sun, Chien Y. Peng, Cheng-Yu Kuo, C. M. Violette Impellizzeri, K. Y. Lo
DD RAFT VERSION O CTOBER
31, 2018Preprint typeset using L A TEX style emulateapj v. 11/12/01
USING MEGAMASER DISKS TO PROBE BLACK HOLE ACCRETION J ENNY
E. G
REENE , A
NIL S ETH , M ARK DEN B ROK , J AMES
A. B
RAATZ , C HRISTIAN H ENKEL ,A I -L EI S UN , C HIEN
Y. P
ENG , C HENG -Y U K UO , C. M. V IOLETTE I MPELLIZZERI , K. Y. L O Draft version October 31, 2018
ABSTRACTWe examine the alignment between H O megamaser disks on sub-pc scales with circumnuclear disks and barson <
500 pc scales observed with
HST /WFC3. The
HST imaging reveals young stars, indicating the presenceof gas. The megamaser disks are not well aligned with the circumnuclear bars or disks as traced by stars in the
HST images. We speculate on the implications of the observed misalignments for fueling supermassive blackholes in gas-rich spiral galaxies. In contrast, we find a strong preference for the rotation axes of the megamaserdisks to align with radio continuum jets observed on ∼ >
50 pc scales, in those galaxies for which radio continuumdetections are available. Sub-arcsecond observations of molecular gas with ALMA will enable a more completeunderstanding of the interplay between circumnuclear structures. INTRODUCTION
Active galactic nuclei have always posed a basic and funda-mental problem – how to cram gas that is happily rotating onkpc scales onto an accretion disk on AU scales (e.g., Balick &Heckman 1982). We do not know the mechanism that dissi-pates angular momentum and allows gas to accrete. There areno shortage of ideas, including major or minor mergers (e.g.,Sanders et al. 1988; Hopkins et al. 2006), bars or bars withinbars (e.g., Shlosman et al. 1990; Maciejewski et al. 2002; Huntet al. 2008; Kim et al. 2012), or nuclear spirals (Englmaier &Shlosman 2000; Maciejewski 2004; Martini et al. 2003; Ann &Thakur 2005). In a couple of nearby cases, inflows are directlyobserved along circumnuclear spirals in ionized gas on hun-dreds of pc scales (e.g., Storchi-Bergmann et al. 2007; Davieset al. 2009).One intriguing clue to the origin of accreting gas comes fromthe well-known misalignment between the rotation axes of ac-cretion disks and the galaxy-scale disk, typically measured us-ing radio jets as a tracer of the angular momentum on sub-pcscales (e.g., Ulvestad & Wilson 1984). In this paper we use 22GHz water megamaser emission as a tracer of sub-pc–scale ac-cretion disks (e.g., Miyoshi et al. 1995; Lo 2005). Early exam-ples of megamaser disks revealed the same misalignment, thistime between the outer galaxy disk and the ∼ . Hubble Space Telescope observations of nine megamaser host galaxies.Combining
HST /WFC3 imaging from F336W to F160W, weare able to identify disk-like structures on (cid:46)
500 pc scales in themajority of the sample galaxies. Even on these circumnuclearscales, we find no evidence for alignment with the megamaserdisks. CIRCUMNUCLEAR STRUCTURES
We observed nine megamaser disk galaxies with
HST in Cy-cle 18 under GO-12185. We have examined the M BH − σ ∗ rela-tion of these galaxies already (Greene et al. 2010), and proposedto study the correlations between galaxy bulge properties and M BH with HST . The galaxies are all ∼ L ∗ spirals with M B ≈ UBIJH ) images of each galaxywith integration times of 1320 , , , ,
420 sec respec-tively. Two galaxies (NGC 6264 and NGC 6323) are more dis-tant than 100 Mpc, which prohibits the analysis presented here.Thus we limit our attention to the seven nearest galaxies in oursample (Table 1), including five galaxies from Kuo et al. (2011),NGC 3393 (Kondratko et al. 2008), and IC 2560 (Ishihara et al.2001). Published VLBI maps reveal the orientation of the mas-ing disk very precisely in all cases but IC 2560, where we relyon unpublished information (Wagner et al. in prep). We fol-low the convention of Kuo et al. and refer to the position angle(PA) of the megamaser disk as the angle East of North to theblue-shifted side of the disk.2.1.
Identifying circumnuclear structures
We use a combination of ellipse-fitting and color maps toidentify organized circumnuclear structures such as nuclearbars (e.g., Maciejewski et al. 2002; Erwin & Sparke 2003),nuclear rings (e.g., Buta 1986), or nuclear spirals (e.g., Mar-tini et al. 2003). The ellipse fits are performed on the F160W Department of Astrophysics, Princeton University, Princeton, NJ 08540 University of Utah, Salt Lake City, UT 84112 National Radio Astronomy Observatory, 520 Edgemont Road, Charlottesville, VA 22903, USA Max-Planck-Institut für Radioastronomie, Auf dem Hügel 69, 53121 Bonn, Germany GMTO Corporation. 251 S. Lake Ave., Suite 300. Pasadena, C 91101 ASIAA Joint Alma Office, Alsonso de Cordova 3107, Vitacura, Santiago, Chile Alfred P. Sloan Fellow Astron. Dept., Kind Abdulaziz University, P.O. 80203, Jeddah, Saudi Arabia a r X i v : . [ a s t r o - ph . C O ] M a y Table 1. Megamaser Galaxies
Galaxy D HT Diam P.A. Diam CN P.A. CN Diam maser
P.A. maser S Jet λ Jet L Jet PA Jet
Ref
Jet (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)NGC1194 52 SA0 24 143 ···
150 1.3 160 0.6 3.6 70 56 1NGC2273 26 SB(r)a: 24 62 0.25 30 0.1 150 19 6 170 90 2UGC3789 50 (R)SA(r)ab 23 152 0.73 180 0.3 40 14.4 20 2900 145 FIRSTNGC2960 71 S0a 37 48 1.0 50 0.4 −
130 1.6 20 7000 145 3IC2560 42 (R’)SB(r)b 32 45 0.40 40 0.3 135 ··· ··· ··· ··· ···
NGC3393 53 (R’)SB(rs) 34 161 0.52 150 0.4 −
30 13 3.6 680 56 1NGC4388 19 SA(s)b: 26 90 ···
90 0.3 110 190 3.6 ···
24 4Note. — Col. (1): Galaxy. Col. (2): Distance to the galaxy (Mpc) from Kuo et al. (2011). Col. (3): Hubble Type, as modified from NEDbased on our
HST images. Col. (4): Galaxy diameter (kpc) from NED. Col. (5): Position angle of the galaxy as a whole measured E of N. Thesemeasurements have an uncertainty of ∼ ◦ . Our measurements agree with those from NED to within 10 ◦ in all cases. Col. (6): Size (kpc) ofcircumnuclear disks or bars. Col. (7): Position angle of structure in Col. (6) E of N. Col. (8): Outer radius of the megamaser disks (pc), as measuredby Kuo et al. (2012), except for NGC 3393 that was measured by Kondratko et al. (2008). Col. (9): Position angle of megamaser disk measuredeast of north. P.A. equals zero when the blueshifted side of the disk plane has zero east offset and positive north offset. Note that all megamaserdisks except for NGC 4388 have a measured inclination > ◦ , and are thus effectively edge-on. All are published but IC 2560, where we rely onunpublished information (Wagner et al. in prep). Col. (10): The flux of a detected elongated radio structure (mJy). When there are multiple jetstructures on different scales, we tabulate the most compact imaged jet. Col. (11): Wavelength of detected jet (cm). Col. (12): The extent of theradio jet (pc). Col. (13): Position angle (East of North) of the radio jet ( ◦ ). Col. (14): Source of the radio observations: (1) Schmitt et al. (2001)(2); Ulvestad & Wilson (1984); (3) A. Sun et al. in preparation; (4) Falcke et al. (1998); FIRST: Becker et al. (1994). images using the algorithm of Jedrzejewski (1987) as imple-mented by the IRAF code ellipse . The code fits ellipses tothe galaxy isophotes, allowing the center, position angle (PA),and ellipticity ( e = 1 − b / a , where a , b are the major and minoraxis respectively) to vary as a function of radius. Circumnuclearstructures are identified based on peaks and valleys in elliptic-ity with corresponding changes in PA pointing to the presenceof nuclear bars or spirals (e.g., Erwin & Sparke 2003). Colorinformation provides secondary clues as to the nature of thestructures that we identify. For instance, nuclear rings are iden-tified as red or blue rings (e.g., Buta & Crocker 1993). Note thatwhile we do not have direct observations of gas yet, blue col-ors that identify recent star formation identify gas-rich regionsby inference. Our primary interest here is not in identifyingall galactic structures. Instead, we are most interested in theprojected PA of the circumnuclear ( (cid:46)
500 pc) structures, forcomparison with the PA of the megamaser disks (Figure 1).We must be careful in interpreting blue colors in the circum-nuclear region, because the active galactic nuclei (AGNs) areexciting narrow-line region emission that also falls into ourbroad-band filters. We use a pixel-by-pixel fitting method toisolate pixels that are dominated in the blue by narrow emis-sion lines. For each pixel in the three optical bands (F336W,F438W, F814W), we explore a grid of single stellar population(SSP) models (Bressan et al. 2012), with metallicities in therange [Fe/H]= − U − B and B − I . Dust is present in almost all of our galaxies, but weare only interested in identifying emission-line gas. Thus weuse, for each pixel and grid point, a range of extinction values(A V = 0 −
50) to modify the SSP colors according to a stan-dard extinction law (Girardi et al. 2008). We then determinethe maximum likelihood value for each pixel by comparing thelikelihoods of all grid points and extinction values. We conser-vatively mask all pixels with a low maximum likelihood (corre-sponding to χ > Circumnuclear structures and megamaser disks
In three galaxies (NGC 2273, NGC 2960 or Mrk 1419, andUGC 3789) we see evidence for a circumnuclear disk. In thecase of NGC 2273 and UGC 3789, there is an inner ring whichmanifests as a spike in the ellipticity profile, and then dust-lanesinterior to the ring most readily identified with spiral arms. Inthe case of NGC 2273, kinematic data confirm this interpreta-tion (Barbosa et al. 2006; Falcón-Barroso et al. 2006). We find ablue structure in the inner ∼ (cid:48)(cid:48) of NGC 2960 with constant PAand ellipticity as the outer disk. Within the inner ∼ (cid:48)(cid:48) , there isa PA change likely associated with the narrow-line region gas,as shown (Figure 1).NGC 3393 also shows a bar and maybe a ring on ∼ (cid:48)(cid:48) scales, and then has what appears to be a nuclear bar on ∼ (cid:48)(cid:48) scales. At the position of the bar, we see an ellipticity maxi-mum at a constant position angle, the classic signature of a bar(e.g., Wozniak et al. 1995; Menéndez-Delmestre et al. 2007).IC 2560 is a difficult case, because the central blue light isdominated by narrow-line emission. On kpc-scales ( > (cid:48)(cid:48) ),the galaxy shows the classic signature of an X-shaped bar (e.g.,Li et al. 2011). However, in the nucleus, dust obscuration andnarrow-line emission conspire to make identification of any or-ganized structure quite difficult. Since the PA profile is flat forradii r > . (cid:48)(cid:48) , we assign IC 2560 an inner PA of 45 ◦ , but cau-tion that the identification in this case is uncertain. Finally, wehave no information for NGC 4388 or NGC 1194, which areedge-on, such that dust from the galaxies on large scales ob-scures the nucleus. In these two cases we adopt the PA of thelarge-scale disk (Figure 2).
20 40 60 80
10 30 50 70
20 60 100
20 40 60 80100
20 40 60 80100 F IG . 1.— Images of the five megamaser disk galaxies from our HST program that appear to contain circumnuclear disks or bars on <
500 pc scales. North isup and East to the left. For each galaxy we show, from left to right, the F814W image on large scales, a three-color image made with F336W, F438W, and F814W,and the F160W image. The central and right-most images are shown on the same scale, as indicated by the yellow box in the left-most image. The angular scale(in arcsec) is indicated in the upper right of the color image, while the red scale bar in the right-most image indicates 200 pc. In the central three-color image,red light indicates both reddened regions and older stellar populations, while blue light indicates recent star formation. The grey regions are pixels that have beenmasked because the colors are dominated by narrow line emission rather than starlight. In the right-most F160W image the red arrows show the orientation of thekpc-scale galaxy (G), the megamaser disk (M) and a jet if known (J; see also Table 1). Recall that the megamaser disks are effectively edge-on in all cases, unlikethe large-scale galaxies, and the arrow points towards the blue-shifted component of the disk. Finally, we show the radial profile from ellipse , along with the radialdistribution in ellipticity and PA. All identified structures are marked as B=bar, D=disk, N=narrow-line emission, and R=ring. F IG . 2.— Three color images for the edge-on galaxies NGC 1194 (left)and NGC 4388 (right), constructed as above but without masking. The yellowscale bar indicates 1 kpc while the red arrows indicate the jet (J) and the direc-tion towards the blue-shifted side of the megamaser disk (M). The blue arrowindicates the extended jet (Ext J) directions. As an alternate way to look at the data, we also gener-ate structure maps from the F814W images (Pogge & Martini2002) shown in Figure 3. Structure maps use deconvolutionto remove large-scale power and highlight fine structures fromdust (dark in our maps) and emission-line regions excited eitherby the AGN or star formation (bright in the structure maps).These emphasize the dust structures in NGC 2273, NGC 2960,and IC 2560. There is dust structure apparent in UGC 3789, al-though it is subtle. However, the structure in NGC 3393 is dom-inated by the narrow-line region emission (see also the mask inFigure 1). In summary, for the five megamaser disk galaxiesstudied here, whose large-scale galaxies are not edge-on (Fig-ure 1), we find clear evidence for circumnuclear disks or barsin four cases. We previously suggested that the majority of themegamaser disks are found in pseudobulge galaxies (Greeneet al. 2010). Now, with the high spatial resolution of
HST , wehave confirmed that there is evidence for ongoing secular evolu-tion in the circumnuclear regions of these galaxies, as expectedfor pseudobulges (e.g., Kormendy & Kennicutt 2004).Having identified these circumnuclear structures, we nowmust assign a projected position angle to each. We use the PAfrom the ellipse fitting at the scale of the feature, often corre-sponding to a local extreme value in ellipticity. As describedin Erwin & Sparke (2003), projection effects (e.g., the super-position of a bright bulge and a nuclear bar) can lead to offsetsbetween the isophotes and the measured PA. However, basedon visual inspection of our galaxies, we believe that our valuesare good to ∼ ◦ . Table 1 contains the orientation and linearsize of each circumnuclear structure. The nuclear disk in NGC2273 (radius of 2 (cid:48)(cid:48) or 260 pc) is well-studied (Erwin & Sparke2003), and our PA agrees with the literature value within ouruncertainty of 10 ◦ . For the remainder of the paper, we willfocus on comparisons between the PA of these circumnuclearstructures on 100-500 pc scales with the megamaser disks onsub-pc scales. If we need to refer to larger scales in the galaxy,we will refer to the kpc-scale galaxy.In principle, we are also interested in the distribution of in-clinations for these circumnuclear structures. Taking the ellip-ticity measurements at the same radius as the PA measurementsand assuming thin disks, we find that all of the targets have highinclinations ( i > ◦ ). However, it is not at all clear that we areidentifying thin disks in all cases.We show the distribution of ∆ PA ≡ PA disk − PA maser in Figure4 a . The first main result of this paper, highlighted in Figure 4 a ,is that the megamaser disks do not tend to align with the cir-cumnuclear structures ( (cid:46)
500 pc) traced by
HST . We have alsoincluded NGC 1068, Circinus, and NGC 3079, where the cir-cumnuclear disk PA is measured from molecular disks. In the case of Circinus, the megamaser disk is warped, and we takethe PA of the inner disk. The molecular disk in NGC 1068 iswarped, but in inclination rather than in PA.Of course, we are interested in the orientation of the starsand the gas. With the
HST images, we infer the presence of gasbased on either blue colors corresponding to recent star forma-tion or dust lanes. We have neither gas or stellar kinematicobservations at these sub-arcsecond scales yet. However, inthe megamaser galaxies NGC 1068, NGC 3079, Circinus, andNGC 2273, detailed gas observations do exist. The detailed in-terpretation is different in each case. In NGC 1068, Schinnereret al. (2000) probe ∼
20 pc scales with CO, and show that themolecular gas disk has a warp that starts at ∼
70 pc and ex-tends all the way to the maser disk on 1 pc scales. NGC 3079contains a kpc-scale molecular disk that is aligned with the ma-jor axis of the galaxy, and then a 600-pc scale molecular ovalthat is aligned in PA both with the kpc-scale disk and the pc-scale masing disk (Koda et al. 2002; Kondratko et al. 2005). InCircinus, the maser disk is nearly perpendicular to the bipolarradio jet and a CO outflow (Greenhill et al. 2003). Finally, thekinematic major and minor axes of the disk in NGC 2273 arenot aligned, and the disk is likely warped (e.g., Barbosa et al.2006). The CO disk appears to align with the warped disk on200 pc scales (Petitpas & Wilson 2002), while the megamaserdisk is misaligned in PA by 60 ◦ . Existing observations of gasand kinematics thus suggest that the circumnuclear structuresobserved here are manifestations of torques that facilitate ac-cretion onto smaller scales.In conclusion, for objects where (cid:46) ∼ > ◦ with the masing disks. We see no tendency foralignment between the stellar structures traced by HST and themegamaser disks, although the gas structures do tend to align.Note that because we focus here on projected position angles,we are observing a lower limit on the misalignments. NUCLEAR JET TO DISK ANGLE
We have seen that stellar structures on (cid:46)
500 pc are fre-quently misaligned with the sub-pc scale accretion disk. Westill expect that the nuclear jet should emerge along the rotationaxis of the sub-pc scale disk (e.g., Pringle et al. 1999). Since themegamaser disks must be close to edge-on in order for ampli-fication to occur, the jets should be perpendicular to the mega-maser disk positions on the sky. Indeed, Henkel et al. (2005)found a high rate of megamaser detections in galaxies wherethe jet was likely in the plane of the sky, suggesting that jets areusually aligned with the rotation axis of the maser disk. We areable to observe the relative orientation of the megamaser diskrotation axis and the jet directly.Most objects in our sample have been observed in the ra-dio continuum at relatively high angular resolution (Table 1),including NGC 1194, NGC 3393 (Schmitt et al. 2001), NGC2273 (Ulvestad & Wilson 1984), IC 2560 (Morganti et al.1999), and NGC 4388 (Falcke et al. 1998). In the case of UGC3789, we simply quote the FIRST fit on kpc scales since nohigher resolution data are yet available. Finally, for NGC 2960we measured a marginally resolved structure at 20 cm from ourEVLA observations with a size of 20 ± (cid:48)(cid:48) (A. Sun et al. inpreparation). Only IC 2560 is unresolved (on the 1 (cid:48)(cid:48) or 0.2 kpcscales measured by Morganti et al.). We caution that these ob-servations probe jets on tens of pc scales and in some cases,VLBI observations do reveal PA changes on pc scales (e.g., Ul- F IG . 3.— Structure maps (Pogge & Martini 2002) made in with F814W to emphasize the circumnuclear dust structure. Dark regions are dust lanes while brightregions correspond to emission line knots due to either star formation or narrow-line region gas. We have zoomed in on comparable regions as seen in the colormaps in Figure 1, with the black scale bars again indicating 200 pc. In the case of NGC 2273, NGC 2960, and IC 2560, the dust structure is clear. UGC 3789appears to show two arms reaching towards the center inside the inner ring. The structure in NGC 3393 is dominated by the narrow-line gas. vestad et al. 1998; Middelberg et al. 2004). In fact, the pc-scalejet in NGC 3079 is misaligned (Trotter et al. 1998), while thediffuse radio emission on pc scales is aligned with the rotationaxis of the disk (Duric & Seaquist 1988). Given the alignmentthat we report on tens of pc scales, it would be quite interestingto see different behavior on pc scales for more of our sources,in light of the possibility that the jet axis may be determinedon small scales by the black hole spin but on large (tens ofpc scales) by electromagnetric forces from the disk (McKinneyet al. 2013).In Figure 4 b , we show the distribution of ∆ PA for the pro-jected jet and megamaser disk angles. Note that the megamaserdisks are edge-on, so the PA is determined by measuring the an-gle on the sky of the observed linear alignment of maser spotsand is 90 ◦ from the rotation axis of the disk. As Figure 4 shows,the vast majority of the jets are aligned with the rotation axis ofthe megamaser disk, as expected. We more than double thenumber of systems with a direct disk-jet comparison (Herrn-stein et al. 1999; Greenhill et al. 2003; Kondratko et al. 2005).The largest outliers are are NGC 3079 ( ∆ PA= 44 ◦ ) and NGC2273 ( ∆ PA= 60 ◦ ). In summary, in all but one of the systems,the jets on tens of pc to kpc scales are found to align within < ◦ of the megamaser rotation axis.In addition to compact jets, we would eventually like to in-vestigate possible kpc-scales jets to seek evidence of a changein direction of the accretion disk on longer timescales. Forinstance, radiation-pressure–driven warping may drive sub-pcscale disk precession on ∼ year timescales (e.g., Pringle1997). Furthermore, it is interesting to ask whether ioniza-tion cones align with the jet/megamaser disk rotation axis. Wedo not really have complete information on the orientation ofthe narrow-line regions for the full sample. However, from themasked (grey) pixels in Figure 1, we can see that the spectac-ular S-shaped narrow-line region in NGC 3393 does align withthe radio jet and the megamaser rotation axis (e.g., Cooke et al.2000; Kondratko et al. 2008), and this is seen in Circinus aswell (Greenhill et al. 2003). In other cases, we do not have ad-equate information yet to address the relative orientation of anyionization cones. WARPS AND KINEMATIC MISALIGNMENTS FACILITATEACCRETION
We have seen that nuclear (and in some cases extended) ra-dio jets align with the rotation axis of the megamaser disks.In contrast, the sub-pc scale accretion disk does not align withthe kpc-scale galactic disk (see §1). Using
HST /WFC3 imag-ing, we have found no strong tendency for alignment betweenthe megamaser disks and the circumnuclear structures traced bystars on (cid:46)
500 pc scales. Here we review what these misalign-ments may tell us about the accretion process.
Disk warping on pc scales causes the observed misalign-ments.
One route to disk misalignment is warping of the pc-scale accretion disk. According to Pringle (e.g., 1997), the sub-pc scale disk may change position angle by a significant fractiondue to radiation pressure. For the megamaser disks, the typicaltimescale for disk PA changes due to radiation-pressure warp-ing would be ∼ × yr (see also Maloney et al. 1996; Gam-mie et al. 2000) with only a linear dependence on BH mass.AGN lifetimes are very uncertain, but Martini et al. (2003)suggest that accretion episodes in low-luminosity sources mayonly last millions of years, perhaps just long enough for somePA change over the AGN lifetime. Note that radiation pres-sure warping cannot cause the warps observed on circumnu-clear ( ∼
200 pc) scales in NGC 2273 and NGC 1068.Another possible source of sub-pc–scale torques comes fromstars in a cusp around the BH; Bregman & Alexander (2012)find that resonant relaxation with a stellar cusp will drive warp-ing of order 10 ◦ on sub-pc scales, comparable to the warp inNGC 4258. The expected timescale for evolution is ∼ yrfor NGC 4258, perhaps in some tension with the million-yearlifetimes from Martini et al. Alternatively, Kartje et al. (1999)suggest that masing clumps are accelerated out of the disk planeby magnetic pressure to a height where the balance of shieldingand pumping is optimal. Since the optimal height will dependon the AGN luminosity, they predict more warping at higher lu-minosities. Both radiation pressure and stellar resonance warp-ing act on sub-pc scales, i.e. on the scales of the megamaserdisks. Warping does not appear to be a compelling explana-tion in general for the observed misalignments for three rea-sons. First, in contrast to NGC 4258, the observed warps in thenew megamaser disks are < ◦ (Impellizzeri et al. 2012; Reidet al. 2013; Kuo et al. 2013) and thus smaller than predictedby either warping model. Second, the observed warps are toosmall to explain the broad range of PA differences that we ob- F IG . 4.— Histogram of the difference in projected position angle between the megamaser disk and ( left ): the circumnuclear structures on (cid:46)
500 pc scales identifiedin the
HST images and ( right ): the most compact jet structure we could find in the literature, as summarized in Table 1 (note – this is the PA of the megamaserdisk, 90 ◦ from the rotation axis). Left : In the case of NGC 1194 and NGC 4388 we simply use the large-scale PA of the galaxy. The filled histograms indicate theliterature megamasers, where we use the difference in projected position angle between known CO disks in NGC 1068 (Schinnerer et al. 2000), Circinus (Curranet al. 1998), and NGC 3079 (Kondratko et al. 2005). Thus, these measurements are not strictly comparable (ours probes recent star formation while the other directlyprobes molecular gas).
Right : In this case the filled histograms include NGC 4258 as well. serve. Third, the timescales for these warps to operate may belong compared with typical lifetimes.
Changes in angular momentum as a function of scale resultnaturally in accretion events.
Hopkins & Quataert (2010) studythe progression of gas from galaxy-wide to pc scales. Becausethe gas is dissipational, the kinematics of the gas and stars de-couple. As a result, the stars are able to torque the gas, leadingto warps and ultimately shifts in the angular momentum vectorof the gas as a function of scale. Observations of the gas kine-matics on ∼
100 pc scales in nearby active galaxies reveal thatfueling is indeed facilitated by decoupled dynamical compo-nents such as inner bars, ovals or spirals (e.g., Hunt et al. 2008;García-Burillo et al. 2009), but that these fueling episodes arestochastic and short-lived (e.g., Dumas et al. 2007; Haan et al.2009), in keeping with the observed kinematic misalignment onall spatial scales that we can probe.In the Hopkins & Quataert models a massive bulge compo-nent will suppress these torques, thus requiring different feed-ing mechanisms (e.g., merging) in bulge-dominated systems(see also their analytic model in Hopkins & Quataert 2011).It is intriguing to note that the megamaser aligns with the large-scale disk in the two bulge-dominated galaxies NGC 1194 andNGC 2960. Our statistics are obviously quite limited. We canuse our strong confirmation that nuclear jets align with the ro-tation axis of megamaser disks (and by extension sub-pc scaledisks in general). We look for morphology-dependent align-ment in the Kinney et al. (2000) Seyfert sample, with morpho-logical types from the RC3. There is no preference for jet align-ment with the rotation axis of the kpc-scale galaxy disk amongS0 or S0/a galaxies. However, it would be interesting to inves-tigate possible jet alignment with the rotation axes of circum-nuclear ∼
500 pc-scale structures in the Kinney et al. sample(e.g., using
HST ). Disk misalignments boost accretion rate.
Nixon et al. (2012)show that if there are strong misalignments between inner andouter disks, then the gas angular momentum is dissipated wherethe disks meet, facilitating accretion. As a result, if there is amajor misalignment as a function of scale, then the accretionrate onto the BH can be boosted by an order of magnitude. We may preferentially observe AGN activity at times of mis-alignment. With a much larger sample, we could look for ananti-correlation between AGN luminosity and PA alignment aspredicted by this model. The gas has an external origin, and thus knows nothing aboutthe angular momentum of the disk.
It would be very challeng-ing to completely rule out that the gas is supplied at randomangles via accretion of small satellites. In fact, in the case ofNGC 3393 there is tentative evidence for an accretion event inthe putative detection of two AGN separated by ∼
150 pc inprojection (Fabbiano et al. 2011). Nevertheless, none of the tar-gets shows evidence for morphological disturbance. Given thehigh incidence of circumnuclear disks and bars in the sample,an internal origin for the accreted gas, with accretion facilitatedby secular processes, seems more natural for the bulk of thesample (e.g., Kormendy & Kennicutt 2004). SUMMARY
Using
HST /WFC3 observations of megamaser disk galaxies,we have shown that the majority of the megamaser disk galax-ies contain clear evidence for circumnuclear structures (disksor bars) often with associated star formation. We find that theaccretion disk on sub-pc scales (as traced by the megamaserdisks) shows no strong tendency to align with circumnuclear(often star-forming) structures such as bars, spirals, or rings on <
500 pc scales. In contrast, we find that the rotation axis ofthe megamaser disk is usually aligned with the compact radiojet axis. We review possible explanations for the misalignmentbetween circumnuclear and sub-pc scale disks, and favor a sce-nario in which accretion naturally requires changes in angu-lar momentum as a function of scale. Given the good correla-tion between the jet and megamaser rotation axis, in the futurewe can boost the sample statistics using
HST observations ofSeyfert galaxies with radio jets. Furthermore, future direct ob-servations of circumnuclear gas in the megamaser disk galaxies(e.g., with ALMA) will help clarify the processes that feed thecentral monster.We would like to thank Paul Martini, Jeremy Goodman, EliotQuataert, Jim Ulvestad, and John Kormendy for very usefulconversations. We thank the anonymous referee for a timely and thorough report that improved this manuscript.observations ofSeyfert galaxies with radio jets. Furthermore, future direct ob-servations of circumnuclear gas in the megamaser disk galaxies(e.g., with ALMA) will help clarify the processes that feed thecentral monster.We would like to thank Paul Martini, Jeremy Goodman, EliotQuataert, Jim Ulvestad, and John Kormendy for very usefulconversations. We thank the anonymous referee for a timely and thorough report that improved this manuscript.