Circumnuclear Multi-Phase Gas in the Circinus Galaxy Revealed with ALMA

Dec 10-14, 2018 Torus 2018@Puerto Varas

Circumnuclear *Multi-phase* Gas in the Circinus Galaxy Revealed with ALMA

Takuma Izumi (NAOJ Fellow), K.Wada, R.Fukushige, S.Hamamura (Kagoshima Univ.), K.Kohno (IoA/Univ. of Tokyo)

1 Radiation-driven “fountain” 2

Radiation pressure + X-ray heating → Outﬂow → Failed wind

8 Gas density + Velocity vector MBH = 1.3×10 Msun, ER = 0.3 → see Keiichi’s talk

Wada 2012, ApJ, 758, 66 Radiation-driven “fountain” 2

Radiation pressure + X-ray heating → Outﬂow → Failed wind

8 Gas density + Velocity vector MBH = 1.3×10 Msun, ER = 0.3 → see Keiichi’s talk

Wada 2012, ApJ, 758, 66 IR emission from central obscuring structures 3883

Dust rad. transfer 3

One snapshot from RHD → 3D Monte Carlo

12μm N

Polar elongation! 100 1 pc 100 E − m A&A 563, A82 (2014) µ 50 .5 50 −

m, 10

] ] µ as as .0 0 Tristram et al. (2014) 0

RA [m

er- [m DEC δ

ﬀ δ et

et et s

s off )conﬁdenceintervals. off 50 50 σ −

.3% (1 ), which cannot be 100 100 4 − • Successfully reproduced 32 pc the MIR polar elongation!False-colour image of the three-component model for the mid- m, respectively. The colour scaling is logarithmic in order to Fig. 7. µ and the surface filling 0 i infrared emission. of the nucleus of the Circinus galaxy (fit 3). The directly estimated from the bootstrap distribution of the respec- T (Schartmann etcolours al. red, 2014) green and blue correspond to the model at 13 tive parameterOur model and can mark fit the the 68 data on the shortest baselines very and 8 show both bright and faint features. Clearlyset of the the mid-infrared colour gradient emission of the with respect to the extended component due to the increaseff in the silicate depth towards Schartmannwell, which means that itet reproduces al. (2014) the low spatial frequen- And observed SEDthe can south-east isbe visible. This colour gradient leads to a chromatic pho- Downloaded from cies of the source adequately. On longer baselines, however, the • tocentre shift towards the north-west. Despite the lower surface bright- data is not well reproduced by our model. This is predominantly reproduced relativelyness, well 80% of the emission comes from the extended component. Also ). This in- due to small scale variations of the correlated fluxes and di plotted is the trace of the water maser disk: the blue and red parts trace ential phases at longer baselines (cf. Fig. the approaching and receding sides of the maser disk respectively. Note (2007 in reproduced by our smooth model. We interpret these variations that the relative o ). r m), the temperatures of our dust components as signatures for small scale structuresµ that our model obviously maser disk is not known (see text5.3 for details). cannotFinally, replicate. a few remarks on degeneracies: several parameters 3° at 13 1pc.Duetothestrongpositionangle .Becausewearefittinganarrowwavelengthrange< . f i 1 ± of our model areλ not independent. The clearest example is the < 46°, our disk- degeneracy between the temperature material only contributes insignificantly2 ∼ to the infrared emissionTristram et al. m )andourmodelling ∆ factorµ 3 (see also Sect. .4pc.Themaserdiskis2 ∼ (8 0 ψ are not well constrained. A small change in temperature has a ∼ direct influence on the brightness of the source, which can be 5.1.1. The disk-like component r out compensated by changing the surface filling factor. Similar de- The disk-like component3°. The is highly strong elongated elongation and of this has component a major with an generacies are present between the size and the axis ratio of the axis FWHM of source, which all change the emitted flux density. Depending on ). Primarily, how- ± ). The.1pcto masers were modelled by a thin disk ). dependency46 of the correlated fluxes for the0 longest baselines, how well these parameters are constrained by the interferometric 3.1 erent orientations of the position= angle of the major2003 axis is very well constrained: measurements, these parameters can become degenerate. ff 2 ( ∼ ψ r in axis ratio of more than 6 : 1 at first suggests an interpretation as a highly inclined disk, as in .Withapositionangleof 5. Discussion )and terpretation is supported byr theout close agreement in orientation and size of this component with the warped maser disk from ), and a two-component 2009 5.1. Morphology4)confirmthatthemid-infraredemissioninthenucleus ( Greenhill et al. 6° at Burtscher et al. 2013 The direct analysis of the data (Sect. ; extending from± http://mnras.oxfordjournals.org/ (Sect. warped with the position angle changing from 29°). By consequence, the relative position of the Circinus galaxy comes from at least two distinct compo- to 56° 2.2 Wada et al. like component now matches this orientation much better than nents: a highly elongated,erent regimes compact of the “disk-like” correlated component fluxes as a and function of ), amoderatelyelongated,extendedcomponent.Tosomedegree,ff previously. The larger size of the mid-infrared disk as compared Hönig et al. 2013 to the maser disk could be evidence of the disk extending out the distinction between the two components; is suggested by the (2009 m two di µ to larger radii than is probed by the maser emission. We em- .5 the projected baseline length (see Sect. phasise that the agreement is only in orientation and size, not in ever, the distinction is suggested byKishimoto the di et al. 2011b the absolute position. With MIDI alone, no absolute astrometry m, 10 is possible because the absolute phase signal is destroyed by the the two components: the two components are elongated roughly µ perpendicular to one another. Two clearly separated emission .0 atmosphere (see Sect. components have also been found in NGC 1068 and NGC 3783 between the maser disk and our disk-like component cannot be (Raban et al. 2009 Schartmann et al. Figure 3. Wavelength dependence of model ER20 at a time 2.0 Myr after the central radiation has been switched on. The variousmorphology in the columns infrared appears to be common correspond to a large to numberWe of AGN interpret ( the). mid-infrared These models emission find a relatively as emission cold, geometrically from 2012 warm dust in( the context of the hydrodynamic models of dusty inclination angles of 0 ,30 ,60 and 90 (from left to right). The rows correspond to wavelengths of 0.1 (upper row), 12 (middletori row) in AGN by and 500 µm(lower ◦ ◦ ◦ ◦ Wada thin and very turbulent disk in the mid-plane of the torus, sur- rounded by a filamentary6 structure. The latter consists of long radial filaments with a hot tenuous medium in between. We as- row). The dynamic range is chosen to scale from the maximum intensity of dust emission (excluding the central source) down to the 10 th fraction7,withthemodelimagesat13 of it. sociate the− central, highly elongated component in the Circinus nucleus with the dense disk in these simulations, and we interpret the extended mid-infrared emission in the context of the fil- Afalse-colourimageofourbestfittingmodel(fit3)ismmappedtothered,greenandbluechannelsofthe amentary torus structureµ seen in these models. Labels are given in parsec. 0 . shown in Fig. and 8 image,When respectively. interpreting our observations, we have to take into ac- 30 count that the emission is dominated by the warmest dust at a certain location, which normally comes from the dust clouds directly illuminated by the central UV source. There are probably also considerable amounts of cooler dust. However, the cooler

A82, page 12 of at University of Tokyo Library on March 11, 2015

Figure 4. Ratio of intensity maps at a wavelength of 10 and 20 µm – indicative of dust temperature – of model ER20 at a time 2.0 Myr after the central radiation has been switched on, where brighter colours correspond to higher temperatures. The panels correspond to inclination angles of 30◦,60◦ and 90◦ 3 (from left to right). The dynamic range is chosen to scale from the maximum ratio down to the 10− rd fraction of it. Labels are given in parsec. appearance of the dust configuration. An example can be seen in the line of sight and slightly altered by scattering off dust grains. the middle panel. Heating the surrounding dust to temperatures close to the sublimation temperature results in re-emission of the grains in the IR wavelength regime, visible in the form of the ‘IR bump’. Overlaid 4.3 Time-resolved spectral energy distributions over the two bumps are spectral features: most prominent and also Fig. 6 shows a compilation of SEDs of all models and time steps most important for our analysis is the 10 µmfeature(peakingmore discussed in this paper. The different panels refer to the three Ed- accurately at 9.7µm, see discussion in Section 5.3), arising due dington ratios taken into account: from 1 per cent (upper panel, to stretching modes in silicate tetrahedra (Si–O stretching modes). ER01) to 10 per cent (middle panel, ER10) and 20 per cent (lower Only slightly visible is the second resonance of silicate dust, namely panel, ER20). Inclinations are colour coded (black solid – i 0◦, the 18.5 µm feature (attributed to O–Si–O bending modes). The very = face-on; blue dotted – i 30◦;greendashed–60◦;reddash–dotted pronounced absorption feature at around 0.08 µmhascontributions = –90◦, edge-on), where the thin lines in light colour refer to the both from the small graphite grains and small silicate grains. At individual SEDs from the time series and thick lines show the SEDs 0.2175 µm, graphite shows a significant feature. Given the temper- averaged over the available time steps. In general two bumps are ature constraints at these wavelengths, the latter only show up as visible: (i) the ‘big blue bump’ – peaking at roughly 0.1 µm – and absorption features against the background of the flux from the cen- (ii) the ‘IR bump’ – peaking at around 10 µm. The first represents tral source SED. As the dust extinction properties and the central the SED of the central source, which acts as the only source of source SED peak at roughly 0.1 µm, dust gets heated very efficiently energy in our simulations (see Section 3 and fig. 3 b in Schart- to temperatures up to the sublimation temperature (approximately mann et al. 2005 for details). It is partially absorbed due to dust on 1500 K for graphite grains and 1000 K for silicate grains). When

MNRAS 445, 3878–3891 (2014) Multi-phase & multi-species gas distribution 4 The Astrophysical Journal Letters, 828:L19 (7pp), 2016 September 10 Wada, Schartmann, & Meijerink

The Astrophysical Journal, 852:88 (11pp), 2018 January 10 Wada et al. Wada, …, TI, et al. (2018)

CO(1-0)

• Target: The Circinus galaxy • MBH, λEdd, CND-scale Mgas: matched to the values of Circinus • Hydrodynamic simulation + XDR Wada et al. 2016, chemistry + radiative transfer ApJ, 828, L19

Figure 2. Integrated intensity map of CO(1−0) and line ratio distributions (ICO() 2-- 1I CO () 1 0 , ICO() 3-- 2I CO () 1 0 , and ICO() 4-- 3I CO () 1 0 ) for qv =0 (face-on).

ratio of 0.2 and the bolometric luminosity of Lbol =´5 has been attained (e.g., Nenkova et al. 2008); therefore, Tdust is 43 −1 10 erg s are fixed during the calculation. not necessarily equivalent to Tgas. Here, we assume three independent radiation fields: (1) non- For simplicity, the third component of the radiation field, i.e.,

0 + FUV, is assumed to be uniform. Instead of solving the radiation Figure 1. Distributions of (a) H , (b) H2, (c) H , and (d) gas temperature on the equatorial plane and x−z plane. The arrows represent velocityspherical vectors of the gas.ultraviolet The (UV) radiation emitted from a thin average supernova rate is 0.014 yr−1. accretion disk; (2) spherically symmetric X-ray radiation transfer for FUV in the inhomogeneous media with multi- emitted from the corona of the accretion disk (Netzer 1987; radiation source—which is beyond our numerical treatment— 4 3 Xu 2015); and (3) uniform far-UV (FUV) radiation arising we change its strength from G0 = 100 to G0 = 10 , where G0 is the incident FUV field normalized to the local interstellar from the star-forming regions that exist in the circumnuclear -3 −2 −1 disk. All three components are assumed to be time-indepen- value (1.6´ 10 erg cm s ) and check whether it affects dent. The first component plays an important role in under- the results (see Section 4.2). The 3D hydrodynamic grid data (i.e., density, temperature, standing the radiation pressure effects in the dust, whereas the abundances, and three components of velocity) of the 2563 grid second contributes primarily toward the heating of the gas and 3 8 cells are averaged to produce 128 grid cells (i.e., the spatial non-equilibrium XDR chemistry. The SED of the AGN and resolution is 0.25 pc) in order to reduce the computational cost. the dust absorption cross-section are taken from Laor & Draine This is passed to the 3D non-LTE line transfer code, as fl (1993). The UV ux is assumed to be angle-dependent, i.e., described in the next section, to derive the line intensities. FUV ()qqµ+cos ( 1 2 cos q) (Netzer 1987), where θ denotes the angle from the rotational axis (z-axis), which is calculated for all grid cells using 2563 rays. The UV and X-ray fluxes 2.2. Non-LTE Line Transfer are calculated from the bolometric luminosity (Marconi The numerical code of the 3D line transfer calculations is the et al. 2004). The total X-ray luminosity (2–10 keV) is 42 −1 same as that used by Wada & Tomisaka (2005) and Yamada LX =´2.8 10 erg s . The temperature of the interstellar et al. (2007), which is based on the Monte–Carlo and long- dust Tdust at a given position that is irradiated by the central UV characteristic transfer code (Hogerheijde & van der Tak 2000). radiation is calculated by assuming that thermal equilibrium The statistical equilibrium rate equations and the transfer

8 equations are iteratively solved using photon packages We use a selection of reactions from the chemical network described by (sampling rays) propagating into each grid cell. We solve the Meijerink & Spaans (2005) and Ádámkovics et al. (2011) for 25 species: H, 12 + + + − − + + + + + rate equations for the energy levels of CO from J=0 to 15. H2,H , H2 , H3 ,H ,e , O, O2,O , O2 ,O2H , OH, OH ,H2O, H2O , + + + + + H3O , C, C , CO, Na, Na , He, He , and HCO . Once the radiation ﬁeld and optical depth are determined for all

3 Multi-phase Obscuring Structures 5

Izumi et al. (2018) 6 Multi-phase circumnuclear obscuring structures studied with ALMA → T. Izumi et al. 2018, ApJ, 867, 48

• Atomic vs molecular gas dynamics? → Can we see different dynamical/ geometrical structures? • Is that difference due to outflows? ALMA Cycle 4 Observations (Band 7 + 8) 7 66 S. J. Curran, B. S. Koribalski and I. Bains Circinus Galaxy (4.2 Mpc) polar elongation N HI 21cm 18 kpc • Compton-thick AGN (Arevalo+2014) 100 1 pc 100 E − 24 -2 m A&A 563, A82 (2014) µ - NH ~(6-10) × 10 cm 50 .5 50 −

m, 10

] ] µ as as .0 42 0 0 - L2-10keV ~ (2-5) × 10 erg/s

RA [m

er- [m DEC δ

ﬀ δ et

et et s

s off )conﬁdenceintervals. off 50 50 σ −

.3% (1 Clearest MIR polar elongation 100 • ), which cannot be 100 4 − → Test the fountain scheme False-colour image of the three-component model for the mid-

m, respectively. The colour scaling is logarithmic in order to Fig. 7. µ and the surface filling 0 i infrared emission. of the nucleus of the Circinus galaxy (fit 3). The directly estimated from the bootstrap distribution of the respec- T colours red, green and blue correspond to the model at 13 tive parameterOur model andcan mark fit the the 68 data on the shortest baselines very and 8 Low SFR show both bright and faint features. Clearlyset of the the mid-infrared colour gradient emission of the with respect to the extended component due to the increaseff in the silicate depth towards • well, which means that it reproduces the low spatial frequen- the south-east is visible. This colour gradient leads to a chromatic pho- cies of the source adequately. On longer baselines, however, the tocentre shift towards the north-west. Despite the lower surface bright- data is not well reproduced by our model. This is predominantly ness, 80% of the emission comes from the extended component. Also → Weak). This in-stellar feedback due to small scale variations of the correlated fluxes and di plotted is the trace of the water maser disk: the blue and red parts trace ential phases at longer baselines (cf. Fig. the approaching and receding sides of the maser disk respectively. Note (2007 in reproduced by our smooth model. We interpret these variations that the relative o ). r m), the temperatures of our dust components as signatures for small scale structuresµ that our model obviously maser disk is not known (see text5.3 for details). cannotFinally, replicate. a few remarks on degeneracies: several parameters 3° at 13 1pc.Duetothestrongpositionangle .Becausewearefittinganarrowwavelengthrange< . f i 1 ± of our model areλ not independent. The clearest example is the < High resolution46°, our disk- CO(3-2) + [CI](1-0) degeneracy between the temperature material only contributes insignificantly2 ∼ to the infrared emissionTristram et al. m )andourmodelling ∆ • factorµ 3 (see also Sect. .4pc.Themaserdiskis2 ∼ (8 0 ψ are not well constrained. A small change in temperature has a ∼ direct influence on the brightness of the source, which can be 5.1.1. The disk-like component @ALMAr out Cycle 4 compensated by changing the surface filling factor. Similar de- The disk-like component3°. The is highly strong elongated elongation and of this has component a major with an generacies are present between the size and the axis ratio of the axis FWHM of source, which all change the emitted flux density. Depending on ). Primarily, how-Curran et al. 2008,± ). The.1pcto masers were modelled by a thin disk ). dependency46 of the correlated fluxes for the0 longest baselines, how well these parameters are constrained by the interferometric 3.1 erent orientations of the position= angle of the major2003 axis is very well constrained: measurements, these parameters can become degenerate. ff MNRAS, 389,2 63 ( ∼ ψ r in axis ratio of more than 6 : 1θ at first~ suggests 5 an interpretationpc (CO) & ~15 pc (CI) as a highly inclined disk, as in .Withapositionangleof )and • r out 5. Discussion terpretation is supported by the close agreement in orientation and size of this component with the warped maser disk from ), and a two-component 2009 5.1. Morphology4)confirmthatthemid-infraredemissioninthenucleus ( Greenhill et al. 6° at Burtscher et al. 2013 The direct analysis of the data (Sect. ; extending from± (Sect. warped with the position angle changing from 29°). By consequence, the relative position of the Circinus galaxy comes from at least two distinct compo- to 56° 2.2 Wada et al. like component now matches this orientation much better than nents: a highly elongated,erent regimes compact of the “disk-like” correlated component fluxes as a and function of ), amoderatelyelongated,extendedcomponent.Tosomedegree,ff I previously.I The larger size of the mid-infrared disk as compared Figure 1. High-resolution H momentHönig maps et al. 2013 of the Circinus galaxy. Left: the H todistribution; the maser disk could be evidence the contour of the disk extending levels out are 0.4, 1, 2, 3, 4, 6, 8, 10, 12, 18 and the distinction between the two components; is suggested by the (2009 m 1 1two di µ to larger radii than is probed by the maser emission.1 We em- 1 20 Jy km s beam . Right: the mean H I velocity field; the contour levels range.5 from 300 (NE) to 570 km s (SW) in steps of 10 km s . The synthesized − − the projected baseline length (see Sect. phasise that the agreement is only in orientation− and size, not in − ever,2 the distinction is suggested byKishimoto the di et al. 2011b the absolute position. With MIDI alone, no absolute astrometry m, 10 is possible because the absolute phase signal is destroyed by the beam (124 107 arcsec the) two is components: shown the in two the components bottom are elongated left-hand roughly corner. µ × perpendicular to one another. Two clearly separated emission .0 atmosphere (see Sect. components have also been found in NGC 1068 and NGC 3783 between the maser disk and our disk-like component cannot be (Raban et al. 2009 Schartmann et al. morphology in the infrared appears to be common to a large numberWe of AGNinterpret ( the). mid-infrared These models emission find a relatively as emission cold, geometrically from 2012 warm dust in( the context of the hydrodynamic models of dusty tori in AGN by Wada thin and very turbulent disk in the mid-plane of the torus, sur- rounded by a filamentary structure. The latter consists of long radial filaments with a hot tenuous,withthemodelimagesat13 medium in between. We as- sociate the central, highly elongated7 component in the Circinus nucleus with the dense disk in these simulations, and we interpret the extended mid-infrared emission in the context of the fil- Afalse-colourimageofourbestfittingmodel(fit3)ismmappedtothered,greenandbluechannelsofthe amentary torus structureµ seen in these models. .0 shown in Fig. and 8 image,When respectively. interpreting our observations, we have to take into ac- 30 count that the emission is dominated by the warmest dust at a certain location, which normally comes from the dust clouds directly illuminated by the central UV source. There are probably also considerable amounts of cooler dust. However, the cooler

A82, page 12 of

Figure 2. Low-resolution H I moment maps of the Circinus galaxy as obtained from the deep Parkes H I survey of the Zone of Avoidance (Juraszek et al. 1 1 2000; Henning et al., in preparation). Left: the H I distribution; the contour levels are (0.5, 1, 2, 5, 10, 20, 30, 40, 50, 60 and 66) 7Jykms− beam− , where 1 1 19 2 × 7Jykms− beam− correspond to an H I column density of 10 cm− .ThecontourlevelswerechosentomatchthosebyFreemanetal.(1977,theirﬁg.6) 19 2 above their detection limit of NH I 5 10 cm− ;ourdetectionlimitisafactorof 10 lower. Right: the mean H I velocity ﬁeld (masked at NH I 5 18 2 = × 1 ∼ 1 = × 10 cm− ); the contour levels range from 324 (NE) to 534 km s− (SW) in steps of 10 km s− . The cross marks the centre of the Circinus galaxy as given in Table 1. The gridded beam of 15.5 arcmin is shown in the bottom left-hand corner.

C C ⃝ 2008 The Authors. Journal compilation ⃝ 2008 RAS, MNRAS 389, 63–74 Downloaded from https://academic.oup.com/mnras/article-abstract/389/1/63/992361 by guest on 20 February 2018 Why we observed CO & CI? 8

Wada et al. 2016, ApJ, 828, L19

50° 90°

0°

Face-on Face-on (Thicker volume) (Thicker volume) Edge-on The Astrophysical Journal, 867:48 (20pp), 2018 November 1 Izumi et al.

Figure 3. Overlay of the Band 7 continuum map (as shown in Figure 2; green contours) on the Ks/F814 ﬂux ratio map (color scale; brighter regions denote Figure 5. Schematic picture of the nuclear obscuring structures in the Circinus those with higher ratios, Mezcua et al. 2016). The Ks/F814 ratio distribution galaxy. Components 1 and 2 revealed by our ALMA Band 7 continuum around the AGN position delineates the boundary of the V-shaped Hα emission observations are denoted by the blue ellipses. The central ∼1 pc region is line distribution (white contours; 0.1, 0.2, 0.4, 0.6, L, and 1.0 in units of zoomed-in on the right side. Also plotted are the MIR disk (brown short-dashed countsCold s−1 pixel−1). Thedust spiral-like structure emission seen in the Band 7 continuum (ALMA:line ellipse), the MIR 860µ polar elongationm(red dashed cont.) line ellipse, Tristram et9 al. emission traces the region that is radially farther than the very nuclear warm 2014), and the 22 GHz H2O maser disk (central solid line, colors represent the The Astrophysical Journal,dust structure867:48 seen in(20pp the Ks),/F814 2018 ratio November map (black 1 dashed lines). line-of-sight velocity structure, Greenhill et al. 2003), respectively.Izumi et al. Schematic view Visibility modeling (UVMULTIFIT) Table 3 Results of the Visibility Modeling (Band 7)

Component-1 Component-2 Major axis (mas) 80±13 1634±34 Minor axis (mas) 60±6 951±21 Major~30 pc axis (pc) 1.6±0.3 33.3±0.7 Minor axis (pc) 1.2±0.1 19.4±0.4 PA (°) 298±21 218±1.4 Integrated ﬂux (mJy) 20.7±1.3 73.5±1.6

Note. The two-dimensional double Gaussian ﬁt shown here was performed with the CASA task UVMULTIFIT. Both components have a centroid consistent with the AGN position.

(;65°) possibly because the spatially extended/elongated emission along the northeast–southwest direction, particularly at the southwestern side of the nucleus, is faint (see also Figure 2(a)). This will decrease the inclination angle with our simple method. Such an extended structure is clearly recognized in, e.g., the CO(3–2) integrated intensity map (Section 4), which indeed yields a better agreeing inclination angle with the above-mentioned galactic-scale one. Meanwhile, Component-1 appears elongated along the polar direction of the H2O maser disk. This immediately brings to mind the parsec-scale MIR polar elongation (Tristram et al. 2014). The derived source size (1.6±0.3 pc×1.2±0.1 pc) and PA (298 21) of Component-1 are roughly consistent with those of the MIR polar elongation (∼1.9 pc and ∼287°, Figure 4. (a) ALMA uv distance vs. amplitude plot of the Band 7 continuum respectively). Therefore, we suggest that Component-1 traces data of the Circinus galaxy (black). The model visibility data of our double the same physical structure as the MIR polar elongation. Gaussian ﬁt, returned by the UVMULTIFIT task, is also shown (red). Although we need higher-resolution direct imaging to robustly (b) Residual visibility amplitude of the Band 7 continuum data, after reveal this compact structure, our ﬁnding here will support the subtracting the model component, shown as a function of uv distance. actual existence of such elongated structures at the hearts of AGNs (e.g., López-Gonzaga et al. 2016), which created a new challenge to the classic torus paradigm. The spatial distributions of several (circum)nuclear major components that we are interested in are illustrated in Figure 5. 4. Gas Distributions and Line Ratios The axis ratio of Component-2 suggests that this structure is moderately inclined if it is a circular disk (;55°). This In this section, we analyze the unprecedented high-resolution inclination angle is slightly smaller than the galactic-scale one CO(3–2) and [C I](1–0) emission line data of Circinus and then

6 The Astrophysical Journal, 867:48 (20pp), 2018 November 1 Izumi et al.

The AstrophysicalFigure 3. Overlay Journal, of the Band867:48 7 continuum(20pp), 2018 map ( Novemberas shown in 1 Figure 2; green Izumi et al. contours) on the Ks/F814 ﬂux ratio map (color scale; brighter regions denote Figure 5. Schematic picture of the nuclear obscuring structures in the Circinus those with higher ratios, Mezcua et al. 2016). The Ks/F814 ratio distribution galaxy. Components 1 and 2 revealed by our ALMA Band 7 continuum around the AGN position delineates the boundary of the V-shaped Hα emission observations are denoted by the blue ellipses. The central ∼1 pc region is line distribution (white contours; 0.1, 0.2, 0.4, 0.6, L, and 1.0 in units of zoomed-in on the right side. Also plotted are the MIR disk (brown short-dashed countsCold s−1 pixel−1). Thedust spiral-like structure emission seen in the Band 7 continuum (ALMA:line ellipse), the MIR 860µ polar elongationm(red dashed cont.) line ellipse, Tristram et9 al. emission traces the region that is radially farther than the very nuclear warm 2014), and the 22 GHz H2O maser disk (central solid line, colors represent the The Astrophysical Journal,dust structure867:48 seen in(20pp the Ks),/F814 2018 ratio November map (black 1 dashed lines). line-of-sight velocity structure, Greenhill et al. 2003), respectively.Izumi et al. Schematic view Visibility modeling (UVMULTIFIT) Table 3 Results of the Visibility Modeling (Band 7)

Note. The two-dimensional double Gaussian ﬁt shown here was performed with the CASA task UVMULTIFIT. Both components have a centroid consistent with the AGN position.

(;65°) possibly because the spatially extended/elongated emission along the northeast–southwest direction, particularly at the southwestern side of the nucleus, is faint (see also Figure 2(a)). This will decrease the inclination angle with our Figure 3. Overlay of the Band 7 continuum map (as shown in Figure 2; green simple method. Such an extended structure is clearly contours) on the Ks/F814 flux ratio map (color scale; brighter regions denote Figure 5.recognizedSchematic picture in, e.g., of the the nuclear CO(3 obscuring–2) integrated structures intensity in the Circinus map those with higher ratios, Mezcua et al. 2016). The Ks/F814 ratio distribution galaxy. Components(Section 4), 1 which and 2 indeed revealed yields by our a better ALMA agreeing Band inclination 7 continuum around the AGN position delineates the boundary of the V-shaped Hα emission observationsangleDouble-Gaussian are with denoted the above-mentioned by the blue ellipses. galactic-scale fit Theto centralthe one. ∼ Meanwhile,1 pc region is line distribution (white contours; 0.1, 0.2, 0.4, 0.6, L, and 1.0 in units of zoomed-in• Component-1 on the right side. appears Also plotted elongated are the along MIR the diskpolar(brown direction short-dashedof −1 −1 counts s pixel ). The spiral-like structure seen in the Band 7 continuum line ellipsethevisibility), the H2 MIRO maser polar data disk. elongation This →(red immediately CND dashed line+ brings ellipse,core to Tristram mind the et al. emission traces the region that is radially farther than the very nuclear warm 2014), andparsec-scale the 22 GHz MIR H2O maser polar diskelongation(central(Tristram solid line, et colors al. 2014 represent). The the dust structure seen in the Ks/F814 ratio map (black dashed lines). line-of-sightderivedConsistent velocity source structure, size Greenhillwith(1.6± the0.3 et al. pc2003 ×MIR1.2), respectively.± 0.1 pc) and PA • (298 21) of Component-1 are roughly consistent with thoseelongation?? of the MIR polar Or elongation just a(∼ jet?1.9 pc and ∼287°, respectively). Therefore,Table we suggest 3 that Component-1 traces Figure 4. (a) ALMA uv distance vs. amplitude plot of the Band 7 continuum Results of the Visibility Modeling (Band 7) data of the Circinus galaxy (black). The model visibility data of our double the same physical structure as the MIR polar elongation. fi Gaussian t, returned by the UVMULTIFIT task, is also shown (red). Although we need higher-resolutionComponent-1 direct imaging toComponent-2 robustly (b) Residual visibility amplitude of the Band 7 continuum data, after reveal″ this″ compact structure, our finding here will support the Figuresubtracting 2. (a) Band the 7 model(νrest= component,351 GHz, shownλrest= as854 a functionμm) continuum of uv distance. emission in the central 10 ×10 region of the Circinus galaxy, shown in both color scale (mJy −1 (n − 1)/2 Major axis (mas) −1 80 13 1634 34 beam ) and contours (5×2 σ with n=1, 2, ..., and 12, where 1σ=0.075actual mJy beam existence). (b) Band of such 8 (ν elongatedrest=±485 GHz, structuresλrest=618 at theμm) heartscontinuum± of emission in the same region as shown in (a). The contours indicate 5×2(n − 1)Minor/2σ with axisnAGNs=(mas1, 2,)( ...,e.g., and López-Gonzaga 9, where 1σ=0.8060 et mJy± al.6 beam2016−1).,( whichc) Zoomed-in created 951view a± ofnew the21 central 2″×2″ of (a). (d) Zoomed-in view of the central 2″×2″ of (b). In eachMajor panel, axis thechallenge(pc central) plus to signthe classicmarks the torus AGN1.6 paradigm.± position0.3 and the synthesized 33.3 beams± are0.7 shown asThe a white spatial ellipse. distributions These maps were of several constructed(circum with Briggs)nuclear weighting major(robust=0.5), without correcting the primary beam attenuations. components that we are interested in are illustrated in FigureMinor5. axis (pc) 1.2±0.1 19.4±0.4 4. Gas Distributions and Line Ratios The axis ratio of Component-2 suggests that this structurePA is (°) 298±21 218±1.4 fl moderately inclined3.4. The if Polar it is Elongation a circular disk (;55°). ThisIntegratedMIRuxIn polar this(mJy section, elongation) we analyze(Tristram20.7 the± et unprecedented1.3 al. 2007, 2014 high-resolution), in 73.5 addition±1.6 to the extended CND seen in Figure 2. The results of the fit are Toinclination further investigate angle is slightly the detailed smaller spatial than the distribution galactic-scale of the one CO(3–2) and [C I](1–0) emission line data of Circinus and then summarized in Table 3 and thefi corresponding model is submillimeter continuum emission, we applied the UVMULTI-Note. The two-dimensional double Gaussian t shown here was performed 6displayed in Figure 4(a) as well. As expected in the above, FIT procedure (Martí-Vidal et al. 2014) to the Band 7with the CASA task UVMULTIFIT. Both components have a centroid consistentour with best thefi AGNt indeed position. supports the existence of a compact continuum visibility data, in which we have sampled long uv component at the AGN position (Component-1 in Table 3), in distance visibilities denser than the case of the Band 8 data. addition to the apparently extended structure (Component-2). This model fitting method to the direct interferometric(;65°) possibly because the spatially extended/elongated emissionThe along latter would the northeast correspond–southwest to a bright direction, part of the particularly CND (see observable (visibility) is preferable to avoid systematic also Figure 2). The fact that Component-2 is brighter than fi at the southwestern side of the nucleus, is faint (see also uncertainty in the image plane tting due to, e.g., a nonlinear Component-1 is consistent with the prediction of our multi- fi deconvolution algorithm, particularly when we t beam-Figurephase2(a)) dynamic. This will torus decrease model that the cold inclination dust visible angle at longer with FIR our unresolved components. simpleto submillimeter method. Such wavelengths an extended is predominantly structure located is clearly in an The achieved visibility data is shown in Figure 4(a) as arecognizedextended in, disk, e.g., while the more CO( centrally3–2) integrated concentrated intensity components map function of uv distance (UVD), after averaging the visibilities(Sectionincluding4), which warm indeed dusty out yieldsflows are a better prominent agreeing at shorter inclination NIR to over the observation time and frequency (representativeangleMIR with the wavelengths above-mentioned(Schartmann galactic-scale et al. 2014 one.). Meanwhile, Note that, frequency=350.2 GHz). The almost constant amplitude atComponent-1however, appears we also found elongated that the along residual the polar amplitude direction remainsof the UVD600 kλ manifests the existence of a compact high at the short UVD (400 kλ; Figure 4(b)). This is due to fl the H2O maser disk. This immediately brings to mind the component, while the short UVD components re ect spatiallyparsec-scalethe fact MIR that there polar are elongation some other(Tristram spatially etextended al. 2014 circum-). The extended structures. Here, we performed a two-dimensional derivednuclear source dusty size structures(1.6 in± Circinus0.3 pc(×e.g.,1.2 spiral±0.1 arms; pc) Figureand2 PA). double Gaussian fit to this visibility data: the use of double As we would like to focus on the very nuclear structures in this components is motivated by the existence of a parsec-scale(298subsection, 21) offitting Component-1 all components are is roughly beyond our consistent scope. with those of the MIR polar elongation (∼1.9 pc and ∼287°, Figure 4. (a) ALMA uv distance vs. amplitude plot of the Band 7 continuum respectively5 ). Therefore, we suggest that Component-1 traces data of the Circinus galaxy (black). The model visibility data of our double the same physical structure as the MIR polar elongation. Gaussian fit, returned by the UVMULTIFIT task, is also shown (red). Although we need higher-resolution direct imaging to robustly (b) Residual visibility amplitude of the Band 7 continuum data, after reveal this compact structure, our finding here will support the subtracting the model component, shown as a function of uv distance. actual existence of such elongated structures at the hearts of AGNs (e.g., López-Gonzaga et al. 2016), which created a new challenge to the classic torus paradigm. The spatial distributions of several (circum)nuclear major components that we are interested in are illustrated in Figure 5. 4. Gas Distributions and Line Ratios The axis ratio of Component-2 suggests that this structure is moderately inclined if it is a circular disk (;55°). This In this section, we analyze the unprecedented high-resolution inclination angle is slightly smaller than the galactic-scale one CO(3–2) and [C I](1–0) emission line data of Circinus and then

6 Molecular & Atomic gas distributions 10

12CO(3-2) • CND (D ~70 pc) + Spirals 6 • MH2 ~ 3×10 Msun (CND) CND 6 - c.f., MBH ~ 2×10 Msun

23 -2 • NH2 ~ 5×10 cm (AGN) + - beam-averaged 24 -2 - NH (Xray) ~ 6×10 cm Spiral arm • Cold gas (CND-scale) contributes signiﬁcantly to the nuclear obscuration! Molecular & Atomic gas distributions 11

[CI](1-0) • First extragalactic high spatial resolution imaging (~15 pc) + • Similar 2D distribution to the CO(3-2) • But we don’t know 3D structures! 16 Decomposition of Cold Gas DynamicsT. Izumi et12 al. CO(3-2) (a) (b) (c)

• 3D Barolo code • Global motion is dominated by rotation (Di Teodoro & Fraternali 2015) (d)• We decomposed the dynamics with (e) (f) tilted-ring models - Vrot, σ, inclination, P.A.

Figure 12. (Top row) (a) Intensity-weighted mean velocity map of CO(3–2) in the central 10′′ 10′′ (200 pc 200 pc) box × × of the Circinus galaxy. The central plus marks the AGN position, which applies to the other panels too. Contours indicate the 1 1 line-of-sight velocity of 300–600 km s− in steps of 15 km s− . (b) Same contours as in (a), but are superposed on the CO(3–2) integrated intensity map. (c) Intensity-weighted velocity dispersion (Gaussian sigma) map of CO(3–2) shown by the contours 1 1 (10–60 km s− in steps of 10 km s− ), superposed on the CO(3–2) integrated intensity map. (Bottom row)(d)Intensity-weighted mean velocity map of [C I](1–0) in the same region and with the same contour levels as in (a). (e) Superposition of (d) in contours on the [C I](1–0) integrated intensity map. (f) Intensity-weighted velocity dispersion (Gaussian sigma) map of [C I](1–0) shown 1 in contours (same levels as in (c)) overlaid on the [C I](1–0) integrated intensity map. The systemic velocity (445.8 km s− ) estimated from our single Gaussian ﬁt to the CO(3–2) spectrum (Figure 10) is highlighted by the red line in both (a) and (d). 1 1 The integrated intensities are shown in the unit of Jy beam− km s− .

912 are formed in situ within the AGN outﬂow, as recently 931 non-circular motions, those can be revealed by subtract- 913 indicated by the hydrodynamic plus chemical simula- 932 ing the modelled rotation map from the observed one 914 tion (Richings & Faucher-Gigu`ere 2018). Further high 933 (e.g., Salak et al. 2016).

915 resolution multi-phase gas observations of AGNs with 934 Through our experiments, however, we noticed that 916 molecular outﬂows are desired to reveal their physical 935 the results of the ﬁtting are sensitive to the spatial reso- 917 and chemical origin. 936 lution of the data. Given that the spatial resolutions are

937 significantly different between the CO(3–2) cube and the 918 5.2. Tilted-ring models 938 [C I](1–0) cube, we first present the results with using 939 the full-resolution CO(3–2) cube in 5.2.1 to describe 919 In order to extract basic beam-deconvolved dynami- § 940 the details of the gas dynamics as much as possible, then 920 cal information, particularly the rotation velocity (Vrot) 941 do with using the beam-matched CO(3–2) and [C I](1– 921 and dispersion (σdisp), we fitted titled rings to the 3D 942 0) cubes in 5.2.2. The latter results are used to delin- 922 observed velocity structures with the Barolo code § 943 923 (Di Teodoro & Fraternali 2015). The main parame- eate qualitative difference of molecular and atomic gas 944 dynamics and/or geometry. 924 ters are dynamical center, systemic velocity (Vsys), Vrot, 925 σdisp, galactic inclination (i), and P.A., all of which can 945 5.2.1. The CO(3–2) full-resolution data 926 be varied for each ring. Here we fix the dynamical center 1 927 to the AGN position and Vsys to 446 km s− based on 946 The full-resolution CO(3–2) cube (θ =0′′.29 0′′.24, 1 × 928 our Gaussian fits to the observed nuclear spectra (Table 947 dV = 10 km s− ) is used here. We modelled 50 con- 929 5): Vrot, σdisp, i, and P.A. are thus the free parameters 948 centric ring with the separation ∆r =0′′.15, which is 930 in our study. Note that the code itself will not account 949 roughly the half size of the major axis of its synthe- 16 Decomposition of Cold Gas DynamicsT. Izumi et12 al. 16 T. Izumi et al.

CO(3-2) P.A. = 216° (a) 700 (a) (a) (b) 200 NE (c) SW 600 100 500 (km/s) 0 (km/s) LOS 400 LOS V V Δ −100 300 −200 200 −10 −5 0 5 10 Offset (arcsec) P.A. = 306° (b) NWSE NWSE

(b) StreamingStreaming motion motion

• 3D Barolo code (Di Teodoro & Fraternali 2015) Global motion is dominated by rotation Streaming motion • Streaming motion (d)• We decomposed the dynamics with (e) (f) tilted-ring models - Vrot, σ, inclination, P.A. Figure 15. (a) The observed position-velocity diagram (PVD) of the CO(3–2) line along the major axis (P.A. = 216◦) shown in the blue contours. The overlaid red con- Figure 14. (a) Model velocity ﬁeld (moment 1) of the tours indicate the PVD produced by our tilted-ring model. Both contours are plotted at 25, 50, 75, 100, 200, ,and Circinus galaxy. Contours indicate the same ones as in Fig- 1 ··· ure 13(a). (b) Residual velocity image after subtracting the 700σ, where 1σ =0.37mJybeam− . The full-resolution data (θ =0′′.29 0′′.24) is used. (b) The PVD along the model from the observed data (color scale). Overlaid con- × tours indicate the velocity-integrated CO(3–2) intensity as minor axis (P.A. = 306◦). Same contours as in the panel-(a) shown in Figure 7(a). The central plus marks the AGN lo- are shown. Due to the inclined geometry of the concentric cation. rings, no model component exists at oﬀset ! 5′′ (see also Figure 14a).

851 a slight increasing trend of the i toward the AGN po- 866 pc increases to 0.3 0.4 (except for the innermost 852 sition. The inferred i ! 70◦ at r " 10 pc is consis- ∼ − 867 r = 3 pc ring), i.e., the dense molecular disk becomes 853 tent with the i = 75◦ estimated for the 1 pc scale nu- 868 geometrically thicker. Within the qualitative framework 854 clear MIR disk (Tristram et al. 2014), as well as with 869 of our model, this moderate thickness at the central 855 the i 75◦ required to well-reproduce the nuclear IR- " ∼ 870 pc can be interpreted such that the gas and dust are 856 SED of this AGN based on our torus model (Wada et al. 871 puffed-up due to the turbulence induced by the AGN- 857 2016). This increasing i will eventually reach 90◦ as ∼ 872 driven failed winds (Wada 2012; Wada et al. 2016,see 858 Circinus hosts the 22 GHz H2O maser disk at the center 873 also Figure 1). 859 (Greenhill et al. 2003). Figure 12. (Top row) (a) Intensity-weighted mean velocity map of CO(3–2)874 It is very in noteworthy the central that the innermost 10′′ r =3pcring10′′ (200 pc 200 pc) box 860 The radial profile of the σdisp/Vrot ratio is shown in 875 shows a very low σ /V 0.1, which× is not seen at × of the Circinus galaxy. The central plus marks861 Figure 16 the(d). This AGN ratio can position, be regarded as which a scale- applies to thedisp otherrot ∼ panels too. Contours indicate the 876 the other radii. The implied very thin disk geometry, as 1 862 height (H) to scale-radius (R) ratio1 (H/R, i.e., an aspect 877 well as the high i, would suggest that we finally start line-of-sight velocity of 300–600 km s− in863 ratio steps of a disk), of under 15 km the hydrostatic s− . equilibrium (b) Same condi- contours as in (a), but are superposed on the CO(3–2) 878 to capture the geometrically thin Keplerian disk (i.e., 864 tion. From the figure, one can find that the σdisp/Vrot at 879 the gas motion is governed by the gravity of the cen- integrated intensity map. (c) Intensity-weighted865 r 15 pc is velocity0.25, while the dispersion ratio at the central r (Gaussian10 sigma) map of CO(3–2) shown by the contours ! ∼ " 1 1 880 tral SMBH) with this thermal emission line. Indeed, in (10–60 km s− in steps of 10 km s− ), superposed on the CO(3–2) integrated intensity map. (Bottom row)(d)Intensity-weighted mean velocity map of [C I](1–0) in the same region and with the same contour levels as in (a). (e) Superposition of (d) in contours on the [C I](1–0) integrated intensity map. (f) Intensity-weighted velocity dispersion (Gaussian sigma) map of [C I](1–0) shown 1 in contours (same levels as in (c)) overlaid on the [C I](1–0) integrated intensity map. The systemic velocity (445.8 km s− ) estimated from our single Gaussian fit to the CO(3–2) spectrum (Figure 10) is highlighted by the red line in both (a) and (d). 1 1 The integrated intensities are shown in the unit of Jy beam− km s− .

CO(3-2) P.A. = 216° P.A. = 216° (a) (a) 700 700 (a) (a) (a) 200 NE(b) SW 200 NE (c) SW 600 600 100 100 500 500 (km/s) (km/s)

(km/s) 0 0 (km/s) LOS LOS LOS LOS 400 V 400 V V V Δ Δ −100 −100 300 300 −200 −200 200 200 −10 −5 0 5 10 −10 −5 0 5 10 Offset (arcsec) Offset (arcsec) P.A. = 306° P.A. = 306° (b) NW SE (b) NWSE NWSE SE NW (b) StreamingStreaming motion motion (b) StreamingStreaming motion motion

• 3D Barolo code (Di Teodoro & Fraternali 2015) Global motion is dominated by rotation Streaming motion • Streaming motion Streaming motion Streaming motion (d)• We decomposed the dynamics with (e) (f) tilted-ring models - Vrot, σ, inclination, P.A. Figure 15. (a) The observed position-velocity diagram Figure 15. (a) The observed position-velocity diagram (PVD) of the CO(3–2) line along the major axis (P.A. = (PVD) of the CO(3–2) line along the major axis (P.A. = 216◦) shown in the blue contours. The overlaid red con- 216◦) shown in the blue contours. The overlaid red con- Figure 14. (a) Model velocity field (moment 1) of the tours indicate the PVD produced by our tilted-ring model. Figure 14. (a) Model velocity field (moment 1) of the tours indicate the PVD produced by our tilted-ring model. Both contours are plotted at 25, 50, 75, 100, 200, ,and Circinus galaxy. Contours indicate the same ones as in Fig- 1 ··· Both contours are plotted at 25, 50, 75, 100, 200, ,and 700σ, where 1σ =0.37mJybeam− . The full-resolution Circinus galaxy. Contours indicate the same ones as in Fig- ure 13(a). (b) Residual velocity image1 after subtracting··· the 700σ, where 1σ =0.37mJybeam− . The full-resolution data (θ =0′′.29 0′′.24) is used. (b) The PVD along the ure 13(a). (b) Residual velocity image after subtracting the model from the observed data (color scale). Overlaid con- × data (θ =0′′.29 0′′.24) is used. (b) The PVD along the minor axis (P.A. = 306◦). Same contours as in the panel-(a) model from the observed data (color scale). Overlaid contours indicate the× velocity-integrated CO(3–2) intensity as minor axis (P.A. = 306◦). Same contours as in the panel-(a) tours indicate the velocity-integrated CO(3–2) intensity as shown in Figure 7(a). The central plus marks the AGN lo- are shown. Due to the inclined geometry of the concentric are shown. Due to the inclined geometry of the concentric shown in Figure 7(a). The central plus marks the AGN location. rings, no model component exists at offset ! 5′′ (see also cation. rings, no model component exists at offset ! 5′′ (see also Figure 14a). Figure 14a).

851 a slight increasing trend of the i toward the AGN po- 851 a slight increasing trend of the i toward the AGN po- 866 pc increases to 0.3 0.4 (except for the innermost 852 sition. The inferred i ! 70◦ at r " 10 pc is consis- ∼ − 866 pc increases to 0.3 0.4 (except for the innermost 867 r = 3 pc ring), i.e., the dense molecular disk becomes 852 sition. The inferred i ! 70◦ at r " 10 pc is consis- 853 tent with the i∼= 75◦−estimated for the 1 pc scale nu- 867 r = 3 pc ring), i.e., the dense molecular disk becomes 868 geometrically thicker. Within the qualitative framework 853 tent with the i = 75◦ estimated for the 1 pc scale nu- 854 clear MIR disk (Tristram et al. 2014), as well as with 868 geometrically thicker. Within the qualitative framework 869 of our model, this moderate thickness at the central 854 clear MIR disk (Tristram et al. 2014), as well as with 855 the i 75◦ required to well-reproduce the nuclear IR- " 869 of our model,∼ this moderate thickness at the central 870 pc can be interpreted such that the gas and dust are 855 the i 75◦ required to well-reproduce the nuclear IR- 856 SED of this AGN based on our torus model (Wada et" al. ∼ 870 pc can be interpreted such that the gas and dust are 871 puffed-up due to the turbulence induced by the AGN- 856 SED of this AGN based on our torus model (Wada et al. 857 2016). This increasing i will eventually reach 90◦ as 871 puffed-up due to the turbulence induced by the∼ AGN- 872 driven failed winds (Wada 2012; Wada et al. 2016,see 857 2016). This increasing i will eventually reach 90◦ as 858 Circinus hosts the 22 GHz H2O maser disk at the center ∼ 872 driven failed winds (Wada 2012; Wada et al. 2016,see 873 also Figure 1). 858 Circinus hosts the 22 GHz H2O maser disk at the center 859 (Greenhill et al. 2003). ′′ ′′ Figure 12. (Top row) (a) Intensity-weighted873 also Figure mean1). velocity map of CO(3–2)874 It is very in noteworthy the central that the innermost 10 r =3pcring10 (200 pc 200 pc) box 859 (Greenhill et al. 2003). 860 The radial profile of the σdisp/Vrot ratio is shown in 874 It is very noteworthy that the innermost r =3pcring 875 shows a very low σdisp/Vrot 0.1, which× is not seen at × of the860 CircinusThe radial profile galaxy. of the σ Thedisp/Vrot centralratio is shown plus in marks861 Figure 16 the(d). This AGN ratio can position, be regarded as which a scale- applies to the other∼ panels too. Contours indicate the 875 shows a very low σ /V 0.1, which is not seen at 876 the other radii. The implied very thin disk geometry, as 861 Figure 16(d). This ratio can be regarded as a scale- 1 862 height (H) to scale-radiusdisp rot (R∼) ratio1 (H/R, i.e., an aspect 876 the other radii. The implied very thin disk geometry, as 877 well as the high i, would suggest that we finally start line-of-sight862 height (H velocity) to scale-radius of (R) 300–600 ratio (H/R, i.e., km an aspect s− in863 ratio steps of a disk), of under 15 km the hydrostatic s− . equilibrium (b) Same condi- contours as in (a), but are superposed on the CO(3–2) 877 well as the high i, would suggest that we finally start 878 to capture the geometrically thin Keplerian disk (i.e., 863 ratio of a disk), under the hydrostatic equilibrium condi- 864 tion. From the figure, one can find that the σdisp/Vrot at 878 to capture the geometrically thin Keplerian disk (i.e., 879 the gas motion is governed by the gravity of the cen- integrated864 tion. From intensity the figure, one map. can find that(c) the Intensity-weightedσ /V at 865 r 15 pc is velocity0.25, while the dispersion ratio at the central r (Gaussian10 sigma) map of CO(3–2) shown by the contours disp rot ! ∼ " 1 1 879 the gas motion is governed by the gravity of the cen- 880 tral SMBH) with this thermal emission line. Indeed, in 865 r ! 15 pc is 0.25, while the ratio at the central r " 10 (10–60 km s− ∼in steps of 10 km s− ), superposed880 tral SMBH) with on this the thermal CO(3–2) emission line. integrated Indeed, in intensity map. (Bottom row)(d)Intensity-weighted mean velocity map of [C I](1–0) in the same region and with the same contour levels as in (a). (e) Superposition of (d) in contours on the [C I](1–0) integrated intensity map. (f) Intensity-weighted velocity dispersion (Gaussian sigma) map of [C I](1–0) shown 1 in contours (same levels as in (c)) overlaid on the [C I](1–0) integrated intensity map. The systemic velocity (445.8 km s− ) estimated from our single Gaussian fit to the CO(3–2) spectrum (Figure 10) is highlighted by the red line in both (a) and (d). 1 1 The integrated intensities are shown in the unit of Jy beam− km s− .

937 significantly different between the CO(3–2) cube and the 918 5.2. Tilted-ring models 938 [C I](1–0) cube, we first present the results with using 939 the full-resolution CO(3–2) cube in 5.2.1 to describe 919 In order to extract basic beam-deconvolved dynami- § 940 the details of the gas dynamics as much as possible, then 920 cal information, particularly the rotation velocity (Vrot) 941 do with using the beam-matched CO(3–2) and [C I](1– 921 and dispersion (σdisp), we fitted titled rings to the 3D 942 0) cubes in 5.2.2. The latter results are used to delin- 922 observed velocity structures with the Barolo code § 943 923 (Di Teodoro & Fraternali 2015). The main parame- eate qualitative difference of molecular and atomic gas 944 dynamics and/or geometry. 924 ters are dynamical center, systemic velocity (Vsys), Vrot, 925 σdisp, galactic inclination (i), and P.A., all of which can 945 5.2.1. The CO(3–2) full-resolution data 926 be varied for each ring. Here we fix the dynamical center 1 927 to the AGN position and Vsys to 446 km s− based on 946 The full-resolution CO(3–2) cube (θ =0′′.29 0′′.24, 1 × 928 our Gaussian fits to the observed nuclear spectra (Table 947 dV = 10 km s− ) is used here. We modelled 50 con- 929 5): Vrot, σdisp, i, and P.A. are thus the free parameters 948 centric ring with the separation ∆r =0′′.15, which is 930 in our study. Note that the code itself will not account 949 roughly the half size of the major axis of its synthe- CI vs CO: Multi-phase Torus Structures 13 18 T. Izumi et al.

(a) 160 Blue = CO(3-2) Red = [CI](1-0) 120 Rotation 80

40 Dispersion Velocity (km/s) Velocity 0 0 20 80 100 40 60 Figure 18. Line proﬁle of the [C I](1–0) measured at Radius (pc) the AGN position (with the single 0′′.71 0′′.66 beam) of × 0.5 the Circinus galaxy. Three-component Gaussian ﬁt was per- (b) formed: each component is indicated by the black-dashed

) 0.4 Geometricallylines, and the thin total mol. model & is shownatomic by the disks red-solid at line, re- • spectively. The derived line properties are summarized in r > 20 Tablepc (c.f.,6. low-SFR in Circinus)

H/R 0.3

(~ Geometrically thicker atomic disk than rot 0.2 • Table 6. Results of the three-component Gaussian fit to the /V the dense molecular disk (r < 10 pc) σ 0.1 [C I](1–0) spectrum → Multi-phase structures!Peak Centroid FWHM 0 1 1 1 0 20 40 60 80 (mJy beam− )(kms− )(kms− ) Central 1107.9 1.6 453.7 5.6 97.7 0.6 Radius (pc) ± ± ± Blueshifted 466.4 5.6 379.1 0.2 56.7 0.5 Figure 17. (a) Radial profiles of the rotation velocity (V ; ± ± ± rot Redshifted 365.9 4.4 512.3 0.1 28.3 0.4 circles) and the velocity dispersion (σdisp; diamonds) derived ± ± ± with the CO(3–2) (blue) and [C I](1–0) (red) emission lines, Note—These fits were performed for the spectra at the in the central r ! 100 pc of the Circunus galaxy. (b) Radial AGN position, measured with the single synthesized beam. profiles of the σdisp/Vrot ratios traced by the CO(3–2) and Results of the single Gaussian fit are listed in Table 5. the [C I](1–0) lines, at the same region as in (a). The uv- ranges and the beam sizes are matched between the CO(3–2) cube and the [C I](1–0) cube here. 2 945 (reduced χ = 656.2 for the single Gaussian, but 64.3 for 8 946 the triple Gaussians) as shown in Figure 18 (see also 947 Table 6). The central Gaussian component has almost 927 to puff-up the disk) should be weak given the low SFR 1 948 the same centroid velocity and FWHM as the CO(3– 928 ( 0.1 M yr− ; Esquej et al. 2014). ∼ ⊙ 949 2) spectrum (Table 5), suggesting that this traces the 929 Meanwhile, the discrepancy of the σdisp/Vrot ratios at 950 same region as the CO(3–2) line does. We also found 930 r ! 10 pc ( 0.15 in CO(3–2) and 0.4in[CI](1–0) , ∼ ∼ 951 that the blueshifted and redshifted components appear 931 respectively) is more evident, which indicates the exis- 952 almost symmetrically with respect to the central com- 932 tence of multi-phase dynamical structures in the torus. 1 953 ponent, with the velocity offset of 75 km s− (blue) 933 Note that it is hard to distinguish coherent outflow mo- 1 ∼ 954 and 60 km s− (red), respectively: this is sugges- 934 tion that can widen the line profile from isotropic tur- ∼ 955 tive of a coherent atomic gas motion around the AGN. 935 bulent motion due to failed winds, based simply on the 956 These results therefore support the view that the diffuse 936 observed σdisp that mixes up these motions. However, 957 atomic gas is distributed in a geometrically thicker vol- 937 as the outflows are preferentially seen at ionized and/or 958 ume than the dense molecular gas at around the AGN, 938 atomic gas at the very nuclear region in our model 959 due to atomic outflows, as is expected in our multi-phase 939 (Wada et al. 2016), we expect that such outflows are 960 dynamic torus model. As this argument is a rather qual- 940 the prime source of the geometrically thicker nature of 961 itative, however, we will perform a more quantitative 941 the C I disk than the CO disk at this very central re-

942 gion around the AGN. Indeed, we found that adding 943 two more components (i.e., triple Gaussians) can fit the 8 The goodness of the fit was evaluated at the velocity range of 1 944 observed nuclear [C I](1–0) spectrum significantly better 250 to 650 km s− . Circumnuclear Multi-phase Gas in the Circinus Galaxy 13

Table 4. The [C I](1–0)/CO(3–2) line ratio (a) AGN CND 10′′ box 1 Jy km s− unit 1.80 0.25 1.03 0.15 0.97 0.14 1 ± ± ± Kkms− unit 0.89 0.13 0.51 0.07 0.48 0.07 ± ± ± Note—The “AGN” indicates that the ratio is measured at the exact AGN location with the single [C I](1–0) synthesized beam (0′′.71 0′′.66). The CND × refers to the 3′′.62 1′′.66 (P.A. = 212◦) region deﬁned in × the CO(3–2) map ( 4.1.2). The 10′′ box contains both the § CND and the spiral arms. The 10% absolute ﬂux uncertainties are included here. What makes the geometrical thickness? 14

Nuclear-scale (15 pc) spectrum 695 CND + spiral arms), the [C I](1–0) emission clearly be- 1200 (b) 696 comes brighter than the CO(3–2) emission at the AGN FWHM ~ 320 km/s 1000 [CI](1-0) 697 position. The relevant RCI/CO measured at several spa- 698 tial scales are listed in Table 4: this high RCI/CO at 800 FWHM ~ 240 km/s 699 the AGN position is uncommon in star-forming galax- 700 ies, suggesting an eﬃcient CO dissociation due to hard 600

727 ture of the XDR needs careful consideration: it does 732 has a wider FWHM than the CO(3–2) spectrum, as well 728 not necessarily form a simple one-zone structure. In- 733 as deviated components from the single Gaussian pro- 1 729 deed, single Gaussian fits to the observed CO(3–2) and 734 file at about 70 km s− to the systemic velocity (446 1 ± 730 [C I](1–0) spectra measured at the AGN position (Figure 735 km s− ). Both of these imply that the [C I](1–0) emis- 731 11b and Table 5) revealed that the [C I](1–0) spectrum 736 sion also traces a different gas volume that the CO(3–2) Circumnuclear Multi-phase Gas in the Circinus Galaxy 13

(a) Nuclear-scale (15 pc) spectrum 695 CND + spiral arms), the [C I](1–0) emission clearly be- 1200 (b) 696 comes brighter than the CO(3–2) emission at the AGN FWHM ~ 320[CI](1-0) km/s 1000 [CI](1-0) 697 position. The relevant RCI/CO measured at several spa- (observation) 698 tial scales are listed in Table 4: this high RCI/CO at 800 FWHM ~ 240 km/s 699 the AGN position is uncommon in star-forming galax- 700 ies, suggesting an eﬃcient CO dissociation due to hard 600

701 X-ray irradiation from the AGN (XDR, Maloney et al. 400 702 1996; Meijerink & Spaans 2005). 703 Based on the one-zone XDR model of Maloney et al. 200 CO(3-2) 704 (1996), we estimated the X-ray energy deposition rate Flux density (mJy/beam) 705 per particle (H /n)with 0 X 250 300 350 400 450 500 550 600Figure650• 18.NuclearLine [CI](1-0) profile of spectrum the [C I](1–0) clearly measured shows at 22 2 1 1 Velocity (km/s) the AGN position (with the single 0′′.71 0′′.66 beam) of HX 7 10− L44r2− N22− erg s− , (2) a larger FWHM than CO(3-2).× ∼ × the Circinus galaxy. Three-component Gaussian fit was per- (b) Figure 11. Line profiles of the [C I](1–0) and the CO(3–2) 706 formed: each component is indicated by the black-dashed where L44 is the 1–100 keV X-ray luminosity in the unit measured at (a) the inner 10′′ 10′′ box and (b) the AGN[CI](1-0) profile = Multiple Gaussians 44 1 × lines, and• the total model is shown by the red-solid line, re- 707 of 10 erg s− , r2 is the distance from the AGN to the location (with the single 0′′.71 0′′.66 beam) of the Circinus × spectively. The derived line properties are summarized in 708 point of our interest in the unit of 100 pc, and N22 is galaxy. The uv-range and the beam size of the CO(3–2) dataAlmost symmetric (coherent) blue- and 22 Table •6. 709 the X-ray attenuating column density in the unit of 10 are matched to those of the [C I](1–0) data. The multiplered-components → Outflows 2 horn-line profile in (a) is due to the spiral arms. Results of 710 cm− , respectively. Here we adopted L44 =0.13 (with single Gaussian fits to the CO(3–2) and [C I](1–0) spectra 711 the photon index = 1.31 and the cutoff energy = 160 Vwind < Vesc: will fall back to the disk 5 at the AGN position (indicated by the dashed line and• dot- 712 keV) from Arévalo et al. (2014), as well as n = 10 Table 6. Results of the three-component Gaussian fit to the H2 dashed line, respectively) are summarized in Table 5. 3 I (Failed wind)→ Fountain 713 cm− as a typical value in the molecular part of CNDs of [C ](1–0) spectrum 714 Seyfert galaxies (e.g., Izumi et al. 2013; Viti et al. 2014). Peak Centroid FWHM 715 Then, we estimated the HX /n at r = 7 pc from the cen- 1 1 1 (mJy beam− )(kms− )(kms− ) 716 ter (i.e., half-major axis of the [C I](1–0) synthesized Table 5. Results of the single Gaussian fit Central 1107.9 1.6 453.7 5.6 97.7 0.6 717 beam) as log(HX /n)= 27.3. According to the mod- ± ± ± − Peak Centroid FWHMBlueshifted 466.4 5.6 379.1 0.2 56.7 0.5 718 els in Maloney et al. (1996), this rate is almost exactly Figure 17. (a) Radial profiles of the rotation velocity1 (Vrot;1 1 ± ± ± (mJy beam− )(kms− )(kmsRedshifted− ) 365.9 4.4 512.3 0.1 28.3 0.4 719 I the one at which the fractional abundancecircles) of C and(X theCI) velocity dispersion (σdisp; diamonds) derived ± ± ± CO(3–2) 807.9 0.9 445.8 0.1 100.3 0.1 720 becomes equal to that of CO (XCO).with At the the regions CO(3–2) (blue) and [C I](1–0) (red)± emission lines,± Note± —These fits were performed for the spectra at the [C I](1–0) 1115.1 4.6 446.1 0.1 134.3 0.2 721 with log(HX /n) > 27.3, we can expectinXCI the/X centralCO > 1.r 100 pc of the Circunus galaxy. (b) Radial AGN position, measured with the single synthesized beam. − ! ± ± ± 722 Thus, we argue that the molecular dissociationprofiles of due the toσdisp/VrotNoteratios—These traced fits by were the performed CO(3–2) and for the spectraResults at the of the single Gaussian fit are listed in Table 5. 723 X-ray irradiation is significant at the closethe [C vicinityI](1–0) of lines, at the sameAGN region position, as in measured (a). The withuv- the single [C I](1–0) synthesized beam (Figure 11b). 724 this Circinus AGN, which can cause theranges high andRCI the/CO beam sizes are matched between the CO(3–2) 725 there. cube and the [C I](1–0) cube here. 2 945 (reduced χ = 656.2 for the single Gaussian, but 64.3 for 726 However, the actual 3-dimensional geometrical struc- 8 946 the triple Gaussians) as shown in Figure 18 (see also 727 ture of the XDR needs careful consideration: it does 732 has a wider FWHM than the CO(3–2) spectrum, as well 947 Table 6). The central Gaussian component has almost 728 not necessarily form a simple one-zone927 to structure. puff-up the In- disk)733 as should deviated be weak components given the from low the SFR single Gaussian pro- 1 1 948 the same centroid velocity and FWHM as the CO(3– 729 deed, single Gaussian fits to the observed928 ( CO(3–2)0.1 M andyr− ; 734Esquejfile at et about al. 201470). km s− to the systemic velocity (446 ∼ ⊙ 1 ± 949 2) spectrum (Table 5), suggesting that this traces the 730 [C I](1–0) spectra measured at the AGN929 positionMeanwhile, (Figure the735 discrepancykm s− ). Both of the ofσ thesedisp/V implyrot ratios that at the [C I](1–0) emis- 950 same region as the CO(3–2) line does. We also found 731 11b and Table 5) revealed that the [C I](1–0) spectrum 736 sion also traces a different gas volume that the CO(3–2) 930 r ! 10 pc ( 0.15 in CO(3–2) and 0.4in[CI](1–0) , ∼ ∼ 951 that the blueshifted and redshifted components appear 931 respectively) is more evident, which indicates the exis- 952 almost symmetrically with respect to the central com- 932 tence of multi-phase dynamical structures in the torus. 1 953 ponent, with the velocity offset of 75 km s− (blue) 933 Note that it is hard to distinguish coherent outflow mo- 1 ∼ 954 and 60 km s− (red), respectively: this is sugges- 934 tion that can widen the line profile from isotropic tur- ∼ 955 tive of a coherent atomic gas motion around the AGN. 935 bulent motion due to failed winds, based simply on the 956 These results therefore support the view that the diffuse 936 observed σdisp that mixes up these motions. However, 957 atomic gas is distributed in a geometrically thicker vol- 937 as the outflows are preferentially seen at ionized and/or 958 ume than the dense molecular gas at around the AGN, 938 atomic gas at the very nuclear region in our model 959 due to atomic outflows, as is expected in our multi-phase 939 (Wada et al. 2016), we expect that such outflows are 960 dynamic torus model. As this argument is a rather qual- 940 the prime source of the geometrically thicker nature of 961 itative, however, we will perform a more quantitative 941 the C I disk than the CO disk at this very central re-

CO(3-2): TB [CI](1-0): TB

(inclination = 75°) • Hydrodynamic simulation + XDR chemistry + rad. transfer • CO(3-2) → mid-plane of the CND • [CI](1-0) → mid-plane + puﬀed-up component due to outﬂows 18 Comparison withT. Izumi our et al. model 16

(a) [CI](1-0) Model (observation) [CI](1-0) (model)

Figure 18. Line profile of the [C I](1–0) measured at the AGN position (with the single 0′′.71 0′′.66 beam) of × (b) the Circinus galaxy. Three-component Gaussian fit was per- Different line profiles for the simulatedformed: each CO(3-2) component and is [CI](1-0) indicated by the black-dashed • lines, and the total model is shown by the red-solid line, re- Triple-Gaussians can well fit thespectively. profile The→ outflows derived line stand properties out! are summarized in • Table 6. • Good consistency between ALMA obs. and our simulation → Support the fountain scheme → Physical origin of the “torus” !? Table 6. Results of the three-component Gaussian fit to the [C I](1–0) spectrum Peak Centroid FWHM 1 1 1 (mJy beam− )(kms− )(kms− ) Central 1107.9 1.6 453.7 5.6 97.7 0.6 ± ± ± Blueshifted 466.4 5.6 379.1 0.2 56.7 0.5 Figure 17. (a) Radial profiles of the rotation velocity (V ; ± ± ± rot Redshifted 365.9 4.4 512.3 0.1 28.3 0.4 circles) and the velocity dispersion (σdisp; diamonds) derived ± ± ± with the CO(3–2) (blue) and [C I](1–0) (red) emission lines, Note—These fits were performed for the spectra at the in the central r ! 100 pc of the Circunus galaxy. (b) Radial AGN position, measured with the single synthesized beam. profiles of the σdisp/Vrot ratios traced by the CO(3–2) and Results of the single Gaussian fit are listed in Table 5. the [C I](1–0) lines, at the same region as in (a). The uv- ranges and the beam sizes are matched between the CO(3–2) cube and the [C I](1–0) cube here. 2 945 (reduced χ = 656.2 for the single Gaussian, but 64.3 for 8 946 the triple Gaussians) as shown in Figure 18 (see also 947 Table 6). The central Gaussian component has almost 927 to puff-up the disk) should be weak given the low SFR 1 948 the same centroid velocity and FWHM as the CO(3– 928 ( 0.1 M yr− ; Esquej et al. 2014). ∼ ⊙ 949 2) spectrum (Table 5), suggesting that this traces the 929 Meanwhile, the discrepancy of the σdisp/Vrot ratios at 950 same region as the CO(3–2) line does. We also found 930 r ! 10 pc ( 0.15 in CO(3–2) and 0.4in[CI](1–0) , ∼ ∼ 951 that the blueshifted and redshifted components appear 931 respectively) is more evident, which indicates the exis- 952 almost symmetrically with respect to the central com- 932 tence of multi-phase dynamical structures in the torus. 1 953 ponent, with the velocity offset of 75 km s− (blue) 933 Note that it is hard to distinguish coherent outflow mo- 1 ∼ 954 and 60 km s− (red), respectively: this is sugges- 934 tion that can widen the line profile from isotropic tur- ∼ 955 tive of a coherent atomic gas motion around the AGN. 935 bulent motion due to failed winds, based simply on the 956 These results therefore support the view that the diffuse 936 observed σdisp that mixes up these motions. However, 957 atomic gas is distributed in a geometrically thicker vol- 937 as the outflows are preferentially seen at ionized and/or 958 ume than the dense molecular gas at around the AGN, 938 atomic gas at the very nuclear region in our model 959 due to atomic outflows, as is expected in our multi-phase 939 (Wada et al. 2016), we expect that such outflows are 960 dynamic torus model. As this argument is a rather qual- 940 the prime source of the geometrically thicker nature of 961 itative, however, we will perform a more quantitative 941 the C I disk than the CO disk at this very central re-

Circumnuclear *Multi-Phase* Gas in the Circinus Galaxy Revealed with ALMA

Circumnuclear Multi-Phase Gas in the Circinus Galaxy Revealed with ALMA