A molecular ring in the Circinus galaxy

Astron. Astrophys. 338, 863–873 (1998)

ASTRONOMY AND ASTROPHYSICS

A molecular ring in the Circinus galaxy? S.J. Curran, L.E.B. Johansson, G. Rydbeck, and R.S. Booth Onsala Space Observatory, Chalmers University of Technology, S-439 92 Onsala, Sweden ([email protected]) Received 16 March 1998 / Accepted 21 August 1998

Abstract. The Circinus galaxy, a type 2 Seyfert, is suspected to have a molecular gas ring structure with a diameter of ∼ 500 pc. We have mapped the lowest three rotational transitions of 12 CO in the central region of the galaxy, and upon detailed examination, the data suggest that a fraction of the molecular gas is orbiting the galaxy’s nucleus in the form of a ring. There also appears to exist an associated molecular outflow which is coincident with the previously observed ionisation cone. In order to determine the geometry and kinematics of the ring, we have modelled various configurations, obtaining an initial estimate from the MEM de-convolved data. The model takes defined ring parameters, i.e. velocities across the ring at various radii, the inclination angle and the intensity of emission, and convolves these with a synthesised telescope beam in order to regenerate the observed spectra. The results suggest that the ring, which has the same position angle as the main galaxy body, has an outer radius of > ∼ 600 pc and orbits with non-Keplerian rotation around a central dynamical mass of ≈ 3 × 108 M . We also estimate the dynamical mass within the central 560 pc to be ≈ 3 × 109 M i.e. only double the total molecular mass calculated from the intensity of CO emission, indicating that use of the Galactic NH2 /ICO conversion ratio may possibly cause us to overestimate the molecular mass in this nucleus. Key words: galaxies: active – galaxies: individual: Circinus – galaxies: kinematics and dynamics – galaxies: ISM – radio lines: galaxies

which are at least an order of magnitude more luminous than any known Galactic H2 O source Gardner & Whiteoak (1982) thus indicating an accretion disk, on which the masers originate Greenhill et al. (1995), Greenhill et al. (1997), and consequently an active galactic nucleus (AGN). Furthermore, because the presence of visible and near infra-red coronal lines Olivia et al. (1994), an X-ray reflection dominated spectrum Matt et al. (1996) and broad polarised Hα Oliva et al. (1998) are characteristic of a Seyfert nucleus, this unusual galaxy is generally considered to be the closest example of a type 2 Seyfert which is experiencing ongoing star burst activity within its nucleus Harnett et al. (1990), Moorwood et al. (1996a). Previous observations of Circinus suggested, upon the deconvolution of the 12 CO J = 2 → 1 data, an apparent face-on ring structure Johansson et al. (1991) possibly associated with the circumnuclear star-burst ring Marconi et al. (1994). In order to confirm this, we mapped the J = 2 → 1 and J = 3 → 2 rotational transitions of 12 CO in the central region of the galaxy. Since Johansson et al. (1991) estimated the diameter of the ring to be ∼ 500 pc, we expected that the ring would be visible in the raw data of the 3 → 2 transition; corresponding to a linear resolution of ∼ 300 pc. In this paper we present the results of the CO J = 2 → 1, J = 3 → 2 and the previous J = 1 → 0 observations. In Sect. 3 we discuss how by means of a MEM analysis we selected the CO J = 2 → 1 data for further investigation, leading us to a determination of masses and other parameters of the molecular gas.

1. Introduction

2. Observations and results

The Circinus galaxy is a nearby (4 Mpc distant) spiral (type 11 Sb-Sd) galaxy with a mass of > ∼ 10 M and an inclination of ≈ 65◦ Freeman et al. (1977). This isolated galaxy is a powerful radio source with a flux density of 1.5 Jy at 1400 MHz and although it lies close to the galactic plane (b = −4◦ ), it is in a region of the sky which does not suffer too seriously from interstellar extinction Freeman et al. (1977). The nucleus of Circinus galaxy contains luminous H2 O masers

2.1. Observations

Send offprint requests to: S. Curran ? Based on results collected at the European Southern Observatory, La Silla, Chile

The observations were done in June 1988 and June 1993 with the 15 m SEST1 at La Silla, Chile. In 1988, the CO 1 → 0 and 2 → 1 transitions were observed with the Schottky mixers; in 1993, SIS receivers were used for the CO 2 → 1 and 3 → 2 transitions. All receivers were tuned to the single-band mode and typical system temperatures, on the TA∗ -scale, were 500 K 1

The Swedish-ESO Sub-millimetre Telescope is operated jointly by ESO and the Swedish National Facility for Radio Astronomy, Onsala Space Observatory, Chalmers University of Technology.

864

S.J. Curran et al.: A molecular ring in the Circinus galaxy

at 115 GHz, 1500 K (Schottky) and 800 K (SIS) at 230 GHz, and 700 K at 345 GHz. The HPBWs are 4500 , 2200 and 1500 , respectively. The backends were acousto-optical spectrometers with 1440 or 1600 channels and a channel width of 0.7 MHz. We used dual-beam switching with a throw of about 120 in azimuth, with pointing errors being typically 300 r.m.s. on each axis. The intensity was calibrated using the chopper-wheel method. Between 1988 and 1993 several attempts were made to improve the surface accuracy of the telescope, so we expect the beam parameters to have changed between the two epochs. In order to account for this discrepancy, we have used the CO 2 → 1 data, taken in 1988 and 1993, to get a consistent intensity scale over the whole frequency range. The main beam efficiencies applied to the data are 0.72, 0.60 and 0.33 for the CO 1 → 0, 2 → 1 and 3 → 2 transitions, respectively. For a more detailed discussion of SEST beam efficiencies see Johansson et al. (1991). 2.2. Results In Figs. 1 and 2 we show contour maps of the integrated emission and central spectra, respectively, of the three lowest rotational transitions of CO. In Fig. 1 the contour map shows the integrated CO 1 → 0 emission (the contour levels range from 20 to 96 K km s−1 in steps of 19 K km s−1 ) observed with a position angle of 210◦ Freeman et al. (1977), the middle and the bottom figures are corresponding maps for the 2 → 1 (the contour levels range from 28 to 128 K km s−1 in steps of 25 K km s−1 ) and 3 → 2 (the contour levels range from 16 to 80 K km s−1 in steps of 16 K km s−1 ) transitions, respectively. The contour levels are in the TA∗ scale and the observed points are indicated. The contour maps indicate slightly different peak positions. However the differences are within the pointing errors and are thus not considered to be significant, although excitation effects cannot be ruled out. Assuming that the CO emission follows a Gaussian distribution in radius and correcting for the beam smearing, the emission extents are 4000 , 3500 and 2000 for the CO 1 → 0, 2 → 1 and 3 → 2 transitions, respectively. This shows that the excitation of the CO varies with distance from the centre; whether the kinetic temperature or the hydrogen density or both follow this decrease cannot be determined from the CO data alone. Since the CO 3 → 2 beam width of 1500 corresponds to a linear resolution of ∼ 300 pc, compared with > ∼ 400 pc for the CO 2 → 1 beam, we expected the ring apparent in the deconvolved 2 → 1 map of Johansson et al. (1991) to be visible in a map of this data. As seen from Fig. 1 any complex structure apparent in the map of Johansson et al. (1991) is not seen in the CO 3 → 2 data. Again varying excitation conditions may provide an explanation to the observed discrepancy. 3. Discussion 3.1. General Figs. 3 and 4 (left) show the observed radial velocity distributions of the CO 3 → 2 and 2 → 1 emission. In these figures

Fig. 1. Contour maps of the integrated CO emission; 1 → 0 (top), 2 → 1 (middle) and 3 → 2 (bottom). See the main text for details

the fuller contours represent emission integrated over velocities


865

the figures it is apparent that a major part of the observed velocity and intensity structure of the near nucleus molecular gas in Circinus agrees with that of a more or less edge on rotating disk or ring with an outer radius of about 300 pc and a rotational speed of about 150 km s−1 . In addition there appears to be gas of near systemic velocity drawn out along the rotational axis, i.e. along the south-east (SE) north-west (NW) direction. We note that the NW portion of this structure appears in the same general direction as the ionised gas emission cone observed by Marconi et al. (1994). This interpretation of the CO data contrasts with the nearly face-on ring structure indicated by the de-convolved map of Johansson et al. (1991). However, that map shows the emission integrated over the total velocity range, implying that the apparent structure may be a mixture of kinematically different gas components, e.g. an edge-on ring and, possibly, an outflow component (see Sect. 3.2.1). In order to investigate the circumnuclear structure more in detail, we have used a MEM routine to deconvolve the molecular emission to a resolution of half a HPBW (Figs. 3 and 4, right), as well as a routine fitting a model distribution to the observed data. Tests, in particular on model distributions of the kind discussed in this text, indicate that the MEM results are reliable. When applying the model fitting procedure to the observed data we used the MEM results to set initial parameter values. There are clear differences between the CO 3 → 2 and 2 → 1 distributions, which become more pronounced in the deconvolved data. It should be noted here that the MEM routine includes a procedure which checks the consistency of the input data. This procedure is particularly effective if the input data is densely spaced, as are ours (1/3 HPBW) since the redundancy is then high. The CO 2 → 1 data was taken under the best possible observational condition so the quality was expected to be very high. Also the CO 3 → 2 data were expected to be of high quality, but at these frequencies the observations are more susceptible to the effects of the atmosphere, and the sensitivity to pointing errors are more marked. Our analysis showed that the CO 2 → 1 data have extremely high quality, and the CO 3 → 2 quite high quality, but have been subjected to some variations in either gain or pointing. Thus while the large scale differences between the CO 3 → 2 and 2 → 1 emission distributions are most likely due to real spatial differences in excitation conditions, the small scale features in the right part of Fig. 4 (right) should be viewed with caution. 3.2. Modelling the nuclear feature Fig. 2. The central spectra; 1 → 0 (top), 2 → 1 (middle) and 3 → 2 (bottom). The intensity scale is TA∗ and the velocities (of resolution 10 km s−1 ) are relative to l.s.r.

from -200 to -73 km s−1 , the broken contours represent emission integrated over velocities from 73 to 200 km s−1 and the grey-scale shows the emission integrated between -73 km s−1 and 73 km s−1 , where (and from now on) the velocities are quoted relative to the systemic velocity of 439 km s−1 . From

3.2.1. The model In order to better determine its parameters, the ring was modelled using the routine, mentioned in the previous section, to produce spectra which can then be compared with the observed spectra. A more complete description of the modelling is given in Curran (1998), but basically the CO distribution is modelled as a ring across which a rotation curve and a relative intensity distribution are defined. The ring is then projected onto the sky, according to a specified position and inclination angle, and “ob-

866


Fig. 3. Left: The observed CO 2 → 1 radial velocity map. The peak intensities are 70 K km s−1 (full contours), 60 K km s−1 (broken contours) and 80 K km s−1 (central dark region). Right: The de-convolved CO 2 → 1 radial velocity map. The peak intensities are 153 K km s−1 (full contours), 62 K km s−1 (broken contours) and 149 K km s−1 (central dark region). In this and Fig. 4 the contour increment is 10% of the peak value in each case and the contour coding is explained in the main text

Fig. 4. Left: The observed CO 3 → 2 radial velocity map. The peak intensities are 32 K km s−1 (full contours), 30 K km s−1 (broken contours) and 47 K km s−1 (central dark region). Right: The de-convolved CO 3 → 2 radial velocity map. The peak intensities are 38 K km s−1 (both full contour peaks), 40 K km s−1 (broken contours) and 68 K km s−1 (central dark region)


867

Fig. 5. The optimum model (broken-dotted) spectra superimposed upon the observed spectra (full). The velocity increment is 200 km s−1 (the resolution is 10 km s−1 ) and the antenna temperature increment is 0.5 K

served” with the required HPBW. As well as defining the extent of the ring, the defining points, in conjunction with the rotation direction Rots (1975), also define the position angle (PA) of the ring’s major axis. Upon the testing of many models Curran (1998), we obtained an optimum fit to the observed data when we applied the model described in Table 1. This model observed with a 2200 (CO 2 → 1) beam gives the spectra which are shown superimposed upon the observed CO 2 → 1 spectra in Fig. 5. Since the observed spectra were not the same in the NE and SW regions it was necessary to apply two different relative intensity distributions. Although their intensity distributions differ, the NE and SW regions were found to share a common rotation curve.

Table 1. The model parameters which generate the model spectra shown in Fig. 5. αn and δn are the points defining the major axis. vn and In are the velocity and the relative intensity of emission, respectively, at each corresponding point. The ring is inclined at an angle of 78◦

3.2.2. The NW and SE regions

in all of the tested models which led us to believe that this emission was not part of the ring structure. When we remove the ring model from the observed data, Fig. 6, we see a structure indicative of a conical outflow corresponding to the low velocity emission regions seen in the de-convolved map (Fig. 3, right).

In Fig. 5 we see a lack of model emission in the NW (−3000 < ∼ 00 00 00 00 ∆α < ∼ ∆δ < ∼ ∆α < ∼ −10 , 10 < ∼ 30 ) and SE (10 < ∼ 00 3000 , −3000 < ∼ ∆δ < ∼ 0 ) regions. This deficit was apparent

00

|αn |[ ] |δn |[00 ] |vn | [km s−1 ] In (NE) [K] In (SW) [K]

Point 1

Point 2

Point 3

Point 4

4 6 100 10 22

8 12 170 10 4

12 18 180 4 4

16 24 150 5 2

868


Fig. 6. A grey-scale plot of the residuals remaining after subtracting the observed from the model spectra (Fig. 5). The black regions are where the simple ring model has insufficient emission to account for the observed data. The NW lobe has a peak intensity of ≈ 50 K km s−1 and the SE lobe ≈ 60 K km s−1 . The region surrounding the lobes has an intensity of ≈ 20 K km s−1 . These residuals (i.e. the outflow plus the surrounding gas) constitute 50% of the total 1.38 × 108 K km s−1 pc2 emitted from the central ≈ 6000 × 6000

Fig. 7. The central model spectrum compared with the central observed spectrum. Top: In the case of the disc model with I0 = 5 K. Bottom: In the case of the ring/disc with I0 = 0 K model

Marconi et al. (1994), Veilleux & Bland-Hawthorn (1997) most probably arises from the galactic disk optically obscuring this portion. This agrees with our model in which the inclination of the outflow (NW approaching and SE receding) shows that the ring is inclined in such a way as to obscure the SE portion.

If this emission is indeed the result of an outflow, we may postulate the following: 1. From Figs. 3 (right) and 6, the outflow appears to have a position angle approximately perpendicular to that of the ring (214±4◦ ; see Table 4), i.e. ≈ −60◦ towards the NW and ≈ 120◦ towards the SE. This is the same position angle as the outflows observed in continuum by Harnett et al. (1990) and in Hα by Veilleux & Bland-Hawthorn (1997). 2. The outflow probably has a low inclination since it’s maximum observed radial speed is 73 km s−1 (Sect. 3.1). Assuming an inclination angle of −12◦ , i.e. perpendicular to the ring, we can assign a maximum possible outflow velocity of ≈ 350 km s−1 along this inclination. This agrees with the outflow observed in CO 1 → 0 by Elmouttie et al. (1997), in which the outflow speed is estimated to be 320 km s−1 . ◦ 3. The opening angle is > ∼ 90 . This is seen in Fig. 5 and in other sets of model spectra. A wide outflow is also apparent in Figs. 3 and 6. This width is approximately equal to that of the ionisation cone observed by Veilleux & Bland-Hawthorn (1997). The low inclination also implies that the outflow extends to ≈ 3000 in each direction, Figs. 3 (right), 5 and 6. 4. Finally, although the outflow appears to have the orientation and opening angle of the observed ionisation cone, it is dissimilar in that the SE component is the most dominant. The absence of a SE component in the ionisation cone

3.2.3. Digression; a disc model In order to demonstrate the results obtained from a different model2 , we show the results of a disc version, i.e. setting the inner radius to ≈zero. To do this, we set the velocity to 100 km s−1 at a distance of 0.4 pc from the centre, i.e. according to the outer edge of the Keplerian disk expected by Greenhill et al. (1997). Since the addition of an intensity at this point increases the intensities in the non-central spectra3 (thus indicating at least a depletion of gas in the centre), we initially applied the modest value of 5 K to the intensity distribution describing the NE portion of the ring. The differences between the non-central spectra and those in the model were too fine to be displayed as in Fig. 5, and so we show the observed and model central spectra, the most affected by this alteration, in Fig. 7. Comparing the result of this centrally depleted model with the model/observed spectrum we see that the central antenna temperature is too high. We gradually decreased the value of the 2

The synthesised spectra are extremely sensitive to the choice of model parameters. 3 Even applying the (NE) ring’s inner edge value of 10 K produces inferior model spectra.

S.J. Curran et al.: A molecular ring in the Circinus galaxy Table 2. Integrated intensities and intensity ratios observed towards the centre position. The intensities are convolved to the CO 1 → 0 beam response and are given in the main-beam brightness temperature scale Transition CO 1 → 0 CO 1 → 0 CO 2 → 1 13 CO 2 → 1 CO 3 → 2 13

I [K km s−1 ] 156 13.0 125 12.5 70

I/ICO 1→0 0.083 ± 0.07 0.80 ± 0.20 0.45 ± 0.15

I/I13 CO 1→0

0.95 ± 0.20

central intensity and, not surprisingly, obtained an exact match to all of the model ring spectra with a disc in which I0 = 0 K, thus giving marginal evidence for zero centre emission. Also, since a disc model gives an inferior match to all of the spectra, the ring model (or disc with significantly depleted near-centre emission) provides a better representation of the nuclear feature than a simple disc. 3.3. Molecular gas properties from the CO data To compare the line intensities of the different CO transitions we used the main-beam brightness temperature scale. This scale is appropriate when the extent of the source is of the same order as the beam size, a condition approximately satisfied for the CO transitions observed in Circinus. However, at the higher frequencies the contribution from the error beam can not be neglected in the case of an extended source. Johansson et al. (1991) have recently presented estimates of the error beam contributions for SEST, and we adopt their beam efficiencies (η) used to convert TA∗ to TMB (TMB = TA∗ /η) : 0.33 ± 30%, 0.60 ± 20% and 0.72 at 345, 230 and 115 GHz, respectively. The errors indicated represent the uncertainties in intensities normalised to that at 115 GHz. The uncertainty in the main-beam efficiency at 115 GHz is estimated to be less than ±10% when an absoluteintensity scale is considered. Table 2 gives the observed CO intensities towards the centre positions in the main-beam brightness temperature scale. All intensities refer to the CO 1 → 0 beam response, i.e. the higher transition data have been convolved to respond to this resolution. We applied a statisticalequilibrium excitation and radiative transfer code to the CO intensities in Table 2. The model assumes a spherical gas distribution with constant densities and kinetic temperatures and treats the radiative transfer in the mean escape probability approximation (see, e.g. Jansen 1995). The solutions are largely defined by the intensity ratios presented in Table 2. However, a further condition was applied; the radiation temperature in the 12 CO 1 → 0 line is greater than 1 K. The errors in intensity ratios are dominated by the uncertainties in the beam efficiencies with the exception of the 13 CO 1 → 0/12 CO 1 → 0 ratio, where the error is due to noise and baseline uncertainties. It turns out that our CO data do not constrain the physical properties of the gas well. The results are summarised in Table 3, and indicate that the bulk of the molecular gas can be either cold

869

Table 3. Results from the radiative transfer model (see text) n(H2 ) [cm−3 ]

Tkin [K]

102 103 104 − 106

50-150 6-15

NCO [cm−2 ] no solutions (2 − 200) × 1017 > 2 × 1017

(Tkin ≈ 10 K, n(H2 ) > 104 cm−3 ) or warm (Tkin ≈ 100 K, n(H2 ) ≈ 103 cm−3 ). The former solution is more in agreement with Galactic cloud properties, possibly justifying the use of the canonical NH2 /ICO conversion ratio Strong et al. (1988), valid for the inner Galactic disk, to estimate molecular gas masses in Circinus. This is more questionable if the warm gas solution applies. The solutions in Table 3 give, at the very best, integrated properties of the molecular gas in the central few hundred parsecs; density gradients and kinetic temperature variations are certainly present within this volume. This is evident in Fig. 8, which shows variations in the CO 3 → 2/CO 2 → 1 integrated intensity ratio within an area comparable to the size of the CO 1 → 0 beam. Here the resolution is determined by the CO 2 → 1 beam and the velocity range of the emission is limited to the central ±70 km s−1 , i.e. the emission from the tangential points of the ring is omitted. This ratio shows a clear gradient perpendicular to the ring, with the highest ratios along the ring (the small offset relative to the position of the ring is not significant). The lower ratios along the direction perpendicular to the ring, i.e. along the conical outflow, could be due to decreasing H2 densities or lower kinetic temperatures. It should be noted that the possible secondary antenna lobe response is more significant at the CO 3 → 2 frequency, implying that the ratio variation should be larger than indicated by Fig. 8. 3.4. Ring properties and masses If the gas is in molecular cloud complexes 00 Krolik & Begelman (1988) of < ∼ 10 in extent, then the 00 flat ring model convolved with a 22 beam should provide a realistic representation of such a structure. This allows us to infer the properties of the ring from the results of our model. These are summarised in Table 4 and Figs. 9 and 10. In Table 4 it should be noted that the value of inner radius quoted is somewhat arbitrary, although it is the radius within which the model suggest the onset of a significant depletion in the molecular gas. We suggest that interferometric observations are necessary in order to satisfactorily constrain the conditions within the central 140 pc, Figs. 9 and 10. The ring shares the same position angle as the main galaxy body; 210◦ ±5◦ but has a slightly higher inclination c.f. 65◦ ±2◦ Freeman et al. (1977). The extent of the CO ring is consistent with those found in other H2 O masering type 2 Seyferts, i.e. NGC 1068 Myers & Scoville (1987), NGC 3079 Irwin & Sofue (1992), NGC 4258 Plante et al. (1991) and

870

S.J. Curran et al.: A molecular ring in the Circinus galaxy |v | (km s -1 )

200

150

100

50

r (pc) 0

140

280

420

560

Fig. 9. The rotation curve of the model ring. This applies to both the NE and SW portions

Fig. 8. Map of the CO 3 → 2/CO 2 → 1 intensity ratio integrated over the central ±70 km s−1 . The resolution is defined by the CO 2 → 1 beam, i.e. 2200 . The contour levels range from 0.5 to 0.8 in steps of 0.05 in the main beam efficiency temperature scale Table 4. Summary of the ring properties. The rotation curve and the relative intensity distributions are shown in Figs. 9 and 10. The radii are calculated at a distance of 4 Mpc and the position angle is defined relative to the receding half of the ring Rots (1975) Quantity Right ascension (1950) Declination (1950) Inner radius Outer radius Position angle Inclination

Value h

m

14 09 17s .5 ± 1s −65◦ 060 1900 ± 400 < ∼ 140 pc > ∼ 600 pc 214◦ ± 4◦ 78◦ ± 1◦

NGC 4945 Bergman et al. (1992), Dahlem et al. (1993) of which NGC 1068 Schild, Tresch-Fienberg & Huchra (1985), Ichikawa et al. (1987), Neff et al. (1994), NGC 3079 (Baan & Irwin 1995, and references within) and NGC 4945 Moorwood & Glass (1984), Moorwood & Oliva (1994), Nakai et al. (1995), Moorwood et al. (1996b) are experiencing current vigorous star-formation activity. Fast rotating gas rings have also been found in the star-burst galaxies NGC 253 Koribalski (1996), NGC 660 Gottesman & Mahon (1990), NGC 1365 Koribalski (1996)4 , NGC 1808 Véron-Cetty & Véron (1985)4 , M82 Weliachew, Fomalont & Greisen (1984), Yun (1992), NGC 3828 Schmelz, Baan & Haschick (1987), (possibly) NGC 4

These also exhibit Seyfert characteristics.

6221 Koribalski (1996)4 , NGC 7469 Genzel et al. (1995)4 and NGC 7582 Morris et al. (1985)4 . These structures are believed to evolve from large scale bars colliding with the gas causing the clouds to lose momentum and thus forming a smaller scale ring Schwarz (1981), Buta (1986) within which intense star formation commences5 . In the case of Circinus, a bar has been observed in the HI structure by Jones et al. (1998) and the star-burst ring by Marconi et al. (1994). From the model we believe we have derived the properties of the intermittent gas ring. From the velocity of the inner edge of the ring, we calculate the dynamical mass inside this radius to be 3.2 ± 0.8 × 108 M , with the uncertainty arising from the uncertainties in the model and angular to linear distance conversion. The dynamical mass within the inner 560 pc may likewise be determined to be 3.3 ± 0.3 × 109 M 6 . The non-Keplerian shape of the rotation curve, Fig. 9, could be due mainly to a central stellar bulge of ∼ 109 M . The asymmetry in the distribution of gas between the NE and SW regions of the ring is shown in Fig. 10. As mentioned previously, these regions share a common rotation curve, implying that the dynamics are not dominated by the unevenly distributed CO gas. This suggests that the rotation curve provides an accurate description of the kinematics of the hydrogen gas. It is clear that our data say little about a possible massive central object and that one must observe the kinematics closer to the 5

Ultimately this gas may be responsible for the formation and sustenance of an AGN Simkin, Su & Schwarz (1980), Weedman (1983), Norman & Scoville (1988), Hernquist (1989). 6 It should be emphasised that our dynamical mass calculations assume a spherical distribution. However, in the case of disk distributions the corrections are small, of the order 10% Lequeux (1983). More significant errors may arise (e.g. depending on the orientation relative to the observer) if a bar structure is present.

S.J. Curran et al.: A molecular ring in the Circinus galaxy I (K)

20

15

10

NE

5

SW r (pc) 0

140

280

420

560

Fig. 10. The relative intensity profiles of the NE and SW portions. Approximately 50% and ∼ 70% of the emission is from within 280 pc in the NE and SW portions, respectively. This accounts for the fact that the ring has an observed outer radius of ∼ 300 pc (Fig. 3, right)

centre in order to see its dynamical (Keplerian) effects. Maiolino et al. (1998) estimate that the non-stellar nuclear source has a mass of < 4 × 106 M confined within 1.5 pc. The integrated central H2 mass itself may however account for a large portion of the dynamical mass. An estimate based on the CO 2 → 1 data and a NH2 /ICO conversion ratio of 2.3 × 1020 cm−2 Strong et al. (1988) gives a total molecular gas mass of around 1.6 × 109 M (the conversion factor is de2→1 fined for the CO 1 → 0 transition, however, the CO CO 1→0 intensity ratios are close to unity in Circinus, Johansson et al. 1991). This is of the same order of magnitude as the value of > 7.5±4.1×108 M for the total molecular gas mass in Circinus calculated (from CO 1 → 0 intensity, using the same conversion factor) by Elmouttie et al. (1997), but, even allowing a possible overestimation by a factor of two, this is a very large molecular mass. Upon comparison with the calculated dynamical mass, this value suggests that either the gas constitutes half of the total mass7 within the central 560 pc, and/or that our choice of conversion factor is at fault, causing us to overestimate the molecular mass by a factor of up to ≈ 5 (provided that typical gas mass fractions apply to Circinus). This then gives a value 8 of > ∼ 3 × 10 M for the molecular gas mass which is comparable with the gas mass of 1.6 × 108 M determined from dust emission at 1.3 mm by Siebenmorgen et al. (1997)8 . Although this value is somewhat uncertain, when compared with the dynamical mass calculated here, it suggests a gas mass fraction of only 5%, which gives a conversion ratio of 0.2 × 1020 cm−2 . As This is considerably greater than the gas mass fraction of < ∼ 10% expected globally in a typical galaxy Mihalas & Binney (1981). 8 This of course depends on the atomic gas abundance, although Johansson et al. (1991) suggest that the gas within the central 500 pc is mostly molecular.

871

noted in Sect. 3.2.2., about 50% of the total CO 2 → 1 emission within the central 6000 × 6000 comes from the ring, which, provided the conversion factor does not vary significantly between the ring and the residual gas, indicates a similar gas fraction. Thus, if the conversion factor is overestimated by about a factor of 5 the mass of the ring should be some 1.5 × 108 M . The inapplicability of the Galactic conversion ratio to other galaxies has been previously noted by, for example, Maloney & Black (1988) who suggest that the abundance of molecular gas in IRAS star-burst galaxies have been overestimated by factors of 4-5, Maloney (1990) who notes such an overestimation (up to a factor of 8) in galactic nuclei, star-burst and ultra-luminous galaxies, and by Shier et al. (1994) who find that the Galactic conversion ratio causes an overestimation of the molecular mass when applied to ultra-luminous galaxies. Also, by comparing CO data with those of other tracers, Dahmen (1995) suggests that the conversion ratio could be a factor of ≈ 9 less than that suggested by Strong et al. (1988) when applied to the Galactic bulge. Perhaps most relevant though is the fact that the conversion ratio within the central 800 pc of the Seyfert1/star-burst NGC 7469 is believed to be significantly lower than the canonical Galactic value Genzel et al. (1995). This may be a result of the high star formation rate on the interstellar medium which may consequently be warmer (up to 80 K, Harris et al. 1991, Genzel 1986, Güsten et al. 1993) and denser Harris et al. (1991), Solomon, Downes & Radford (1992) than in the Galactic disk. This lends support to the warm gas solution of the gas properties in Circinus (Sect. 3.3.). It should be noted, however, that there exists support for the use of the Galactic conversion ratio in external galaxies, e.g. Young & Scoville (1982) suggest that the gas mass can constitute up to 25% of the dynamical mass in late type spiral galaxies and Scoville et al. (1997) state that in star-burst and high infra-red luminous galactic nuclei, the gas may constitute over 50% of the total mass. However, if a major fraction of the mass inside a certain radius is in the form of gas, and therefore in a disk, it would be dynamically unstable. It has been known since early N -body simulations (e.g. Hohl 1971; Ostriker & Peebles 1973), that galactic disks are very unstable unless the velocity dispersion is large (which it is not for a gas disk) or there is a stabilising potential from a spherical distribution of matter (e.g. a bulge or a halo) which is itself more massive than the disk. It is clear from the above argument, that whatever the case (high gas mass fraction, low conversion ratio or both), that the molecular gas properties in Circinus differ from those of Galactic disk clouds and thus favour the warm gas solution (Table 3), i.e. the dominant gas component has the parameters Tkin ≈ 100 K and n(H2 ) ≈ 103 cm−3 . 4. Summary

7

We have mapped the Circinus galaxy in the three lowest rotational transitions of CO. The CO 2 → 1 and 3 → 2 show clean kinematic evidence of a ring/disk distribution of the molecular gas in the central region. The comparison of a model to the

872


CO 2 → 1 data suggests a ring with an outer radius of > ∼ 600 pc orbiting the nucleus with non-Keplerian rotation. The ring has the same position angle and a slightly higher inclination angle (≈ 78◦ ) than the main galaxy body and its molecular mass is estimated to be of the order of 108 M . We have reason to believe that in addition to the ring, there also exists a molecular outflow of low inclination. This outflow extends to > ∼ 600 pc from the centre of −1 the nucleus with a velocity < ∼ 350 km s . The outflow appears to be coincident with a previously detected molecular outflow Elmouttie, Haynes & Jones (1997) and ionisation cone which have been observed in the galaxy Harnett et al. (1990), Marconi et al. (1994), Elmouttie et al. (1995), Veilleux & Bland-Hawthorn (1997). The dynamical mass within the central 560 pc is ≈ 3 × 109 M , based on the rotation curve defined by the model. The bulk of the molecular gas in this region seems to be warm (Tkin ≈ 100 K) although our excitation and radiative transfer analysis does not exclude a colder gas (Tkin ≈ 10 K). Comparing the molecular mass calculated from the CO emission with the dynamical mass, we find that while the NH2 /ICO conversion ratio for the Galactic plane Strong et al. (1988) may be applicable, there is a possibility that it is lower by a factor of about 5 in the central regions of Circinus. Acknowledgements. We acknowledge the help of the SEST team. We would also like to thank Per Bergman for the prompt modifications to the xspec program, and Magnus Thomasson and John Conway for their comments.

References Baan W.A., Irwin J.A., 1995, ApJ 446, 602 Bergman P., Aalto S., Black J.H., Rydbeck G., 1992, A&A 265, 403 Buta R., 1986, ApJS 61, 609 Curran S.J., 1998, Licentiate thesis, Chalmers University of Technology Dahlem M., Golla G., Whiteoak J.B., et al., 1993, A&A 270, 29 Dahmen G., 1995, PhD thesis, Max-Planck-Institut für Radioastronomie Elmouttie M., Haynes R.F., Jones K.L., et al., 1995, MNRAS 275, L53 Elmouttie M., Haynes R.F., Jones K.L., 1997, Proc. Astron. Soc. Aust. 14, 140 Freeman K.C., Karlsson B., Lyng˚a G., et al., 1977, A&A 55, 445 Gardner F.F., Whiteoak J.B., 1982, MNRAS 201, 13p Genzel R., Weitzel L., Tacconi-Garman L.E., et al., 1995, ApJ 444, 129 Gottesman S.T., Mahon M.E., 1990, HI Observations of the Peculiar Galaxy NGC 660. In: Sulentic J., Keel W., Telesco C. (eds.) IAU Colloquium No.124. Paired and Interacting Galaxies. NASA, Washington, p. 209 Greenhill L.J., Jiang D.R., Moran J.M., et al., 1995, ApJ 440, 619 Greenhill L.J., Ellingsen S.P., Norris R.P., et al., 1997, ApJ 474, L103 Güsten R., Serabyn E., Kasemann C., et al., 1993, ApJ 402, 537 Harnett J.I., Whiteoak J.B., Reynolds J.E., Gardner F.F., Tzioumis A., 1990, MNRAS, 244 130 Harris A.I., Hills R.E., Stutzki J., et al., 1991, ApJ 382, L75 Hernquist L., 1989, Nat 340, 687 Hohl F., 1971, ApJ 168, 343

Ichikawa S., Okamura S., Kaneko N., Nishimura M., Toyama K., 1987, PASJ 39, 411 Irwin J.A., Sofue Y., 1992, ApJ 396, L75 Jansen D., 1995, PhD thesis, Leiden, The Netherlands Johansson L.E.B., Aalto S., Booth R.S., Rydbeck G., 1991, Molecular Gas in the Circinus Galaxy. In: Sundelius B., (ed.) Dynamics of Disc Galaxies. University of Göteborg and Chalmers University of Technology, Göteborg, p. 249 Johansson L.E.B., Greve A., Booth R.S., et al., 1998, A&A 331, 857 Jones K.L., Koribalski B.S., Elmouttie M., Haynes R.F., 1998, MNRAS, submitted Koribalski B., 1996, The Relation Between the Large-Scale Gas Dynamics, Nuclear Kinematics, and Activity in Spiral Galaxies. In: Skillman E. (ed.) Minnesota Lectures on Extragalactic HI. ASP Conference Series, p. 238 Krolik J.H., Begelman M.C., 1988, ApJ 329, 702 Lequeux J., 1983, A&A 125, 394 Maiolino R., Krabbe A., Thatte N., Genzel R., 1998, ApJ 493, 650 Maloney P., Black J.H., 1988, ApJ 325, 389 Maloney P., 1990, Mass Determinations from CO Observations. In: H.A.Thronson J., Shull J.M. (eds.) The Interstellar Medium in Galaxies. p. 493 Marconi A., Moorwood A.F.M., Origlia L., Oliva E., 1994, ESO messenger 78, 20 Matt G., Fiore F., Perola., et al., 1996, MNRAS 281, L69 Mihalas D., Binney J., 1981, Galactic Astronomy. W. H. Freeman and Company, San Francisco Moorwood A.F.M., Glass I.S., 1984, A&A 135, 281 Moorwood A.F.M., Oliva E., 1994, ApJ 429, 602 Moorwood A.F.M., Lutz D., Oliva E., et al., 1996a, A&A 315, L109 Moorwood A.F.M., Van Der Werf P.P., Kotilainen J.K., Marconi A., Oliva E., 1996b, A&A 308, L1 Morris S., Ward M., Whittle M., Wison A.S., Taylor K., 1985, MNRAS 216, 193 Myers S.T., Scoville N.Z., 1987, ApJ 312, L39 Nakai N., Inoue M., Miyazawa K., Miyoshi M., Hall P., 1995, PASJ 47, 771 Neff S.G., Fanelli M.N., Roberts L.J., et al., 1994, ApJ 430, 545 Norman C.A., Scoville N.Z., 1988, ApJ 332, 124 Oliva E., Marconi A., Cimatti A., Alighieri S.D.S., 1998, A&A 329, L21 Olivia E., Salvati M., Moorwood A.F.M., Marconi A., 1994, A&A 288, 457 Ostriker J.P., Peebles P.J.E., 1973, ApJ 186, 467 Plante R.L., Lo K.Y., Roy J.-R., Martin P., Noreau L., 1991, ApJ 381, 110 Rots A.H., 1975, A&A 45, 43 Schild R., Tresch-Fienberg R., Huchra J., 1985, AJ 90, 441 Schmelz J.T., Baan W.A., Haschick A.D., 1987, ApJ 315, 492 Schwarz M.P., 1981, ApJ 247, 77 Scoville N.Z., Yun M.S., Bryant P.M., 1997, Molecular Gas in Galactic Nuclei. In: Latter W.B., Radford S. J.E., Jewell P.R., Jeffrey J.G., Bally J. (eds.) CO: Twenty-Five Years of Millimeter-wave Spectroscopy. Kluwer, Dordrecht, p. 305 Shier L.M., Rieke M.J., Rieke G.H., 1994, ApJ 433, L9 Siebenmorgen R., Moorwwod A., Freudling W., Käufl H.A.1997, A&A 325, 450 456 Simkin S.M., Su H.J., Schwarz M.P., 1980, ApJ 237, 404 Solomon P.M., Downes D., Radford S.J.E., 1992, ApJ 398, L29 Strong A.W., Bloemen J.B.G.M., Dame T.M., et al., 1988, A&A 207, 1

S.J. Curran et al.: A molecular ring in the Circinus galaxy Veilleux S., Bland-Hawthorn J., 1997, ApJ 479, L105 Véron-Cetty M.P., Véron P., 1985, A&A 145, 425 Weedman D., 1983, ApJ, 266, 479 Weliachew L., Fomalont E.B., Greisen E.W., 1984, A&A 137, 335 Young J.S., Scoville N., 1982, ApJ 258, 467 Yun M.S., 1992, PhD thesis, Harvard Univ., Cambridge, MA

873