Skip to main content

Full text of "Seyfert Activity and Nuclear Star Formation in the Circinus Galaxy"

See other formats


On 
On 



Seyfert Activity and Nuclear Star Formation in the Circinus 

Galaxy 

R. Maiolino 1 ' 2 , A. Krabbe 1 ' 3 , N. Thatte 1 , R. Genzel 1 
ABSTRACT 

We present high angular resolution (0.15" — 0.5") near infrared images and 
P-r spectroscopy of the Circinus galaxy, the closest Seyfert 2 galaxy known. The 

C/} ■ data reveal a non-stellar nuclear source at 2.2/xm whose size is smaller than 

CNj ■ 1.5 pc in radius. The coronal line region and the hot molecular gas emission 

extend for 20-50 pc in the ionization cone. The data do not show evidence for 
CN ■ a point-like concentration of dark mass; we set an upper limit of 4 x 10 6 M Q to 

the mass of a putative black hole. 

We find evidence for a young nuclear stellar population, with typical ages 
Q\ ■ between 4 x 10 7 and 1.5 x 10 8 yrs. The luminosity of the starburst inside a 

few hundred pc is comparable to the intrinsic luminosity of the Seyfert nucleus, 
and the two of them together account for most of the observed bolometric 
luminosity of the galaxy. Within the central 12 pc the starburst has an age of 
about 7 x 10 7 yrs and radiates ~ 2% of the luminosity of the active nucleus. We 
discuss the implications of these results for models that have been proposed for 



Or 



o 

£ ■ 

the starburst-AGN connection. 



Subject headings: galaxies: individual (Circinus) - galaxies: active - galaxies: 
starburst — galaxies: kinematics and dynamics — galaxies: nuclei — infrared: 
galaxies 



1. Introduction 

The connection between starburst and Seyfert activity is one of the most hotly debated 
issues in the context of AGNs. A variety of data have provided growing evidence that 



1 Max-Plank-Institut fur Extraterrestrische Physik, Postfach 1603, D-85740 Garching, Germany 
2 Osservatorio Astrofisico di Arcetri, L.go E.Fermi 5, 1-50125 Firenze, Italy 
3 DLR-Institute for Space Sensor Technology, Rudower Chaussee 5, 12489 Berlin, Germany 



- 2 - 



starburst and Seyfert phenomenon often coexist (Cid Fernandes & Terlevich 1995, Delgado 
and Perez 1993, Genzel et al. 1995, Heckman et al. 1989, 1995, Maiolino et al. 1995, 
Moorwood et al. 1996c, Neff et al. 1994, Oliva et al. 1995, Rodriguez-Espinosa et al. 
1987, Storchi-Bergmann et al. 1996, Cid Fernandes et al. 1997). However, most of these 
studies have detected star forming activity, and/or a young stellar population in Seyfert 
galaxies (especially type 2) on the kpc scale, which might not be directly relevant to 
the active nucleus. Indeed, Maiolino et al. (1997) speculate that star formation in the 
host galaxy of Seyfert 2 nuclei is probably related only indirectly to the active nucleus: 
non-axisymmetric morphologies (bars/distortions/interactions) are likely to be the prime 
cause of the relationship, as they both enhance star formation in the galaxy and drive 
gas into the nuclear region to obscure the active nucleus. Terlevich and collaborators 
(Terlevich & Melnick 1985, Terlevich et al. 1992, 1995, Terlevich 1994) have proposed a 
much closer link between nuclear star formation and Seyfert activity. They claim that 
the whole phenomenology observed in AGN can be accounted for by assuming that these 
nuclei are powered by a massive nuclear starburst evolving in the metal rich environment of 
early type galaxies: the ionizing radiation emitted by an extreme population of Wolf-Rayet 
stars, along with fast shocks generated by SNe, would produce the spectrum observed 
in Sy galaxies. So far, very few observational tests of this theory have been made, due 
to difficulties in probing star formation and stellar populations in the innermost region 
of AGNs. However, recently Heckman et al. (1997) have discovered a powerful, young 
(in the Wolf-Rayet phase) nuclear starburst in the Sy2 galaxy Mkn477; such a nuclear 
starburst might dominate the bolometric luminosity of the central region, therefore partly 
supporting Terlevich's scenario, at least in this object. Norman & Scoville (1988), following 
Bailey (1980) and David (1987a,b), have proposed a different link between the nuclear stellar 
population and AGN. According to their model, a young nuclear (< 10 pc) star cluster 
would return mass into the interstellar medium, by means of post-main-sequence stars, that 
would feed a nuclear black hole. An analogous model has been proposed more recently by 
Murphy et al. (1991), who also include effects of tidial disruption of stars close to the black 
hole. Jenkins & Binney (1994) have proposed a similar mechanism for the fueling of the 
Galactic center. Again, these models have not been tested, so far, due to the shortage of 
observations that can probe the nuclear stellar population. 

The Circinus galaxy is a nearby (4 Mpc), edge-on (incl.~65°), Sb-d system that is 
seen through a low interstellar extinction window near the Galactic plane {Ay = 1.5 mag, 
Freeman et al. 1977). The nuclear optical line ratios (Oliva et al. 1994) are typical of a 
Seyfert 2 galaxy. The Seyfert 2 classification is supported also by the detection of intense 
coronal lines (Oliva et al. 1994, Moorwood et al. 1996a), the discovery of an intense 
X-ray iron 6.7 keV line (Matt et al. 1996), rapid variation of the powerful H 2 maser 



- 3- 



emission (Greenhill et al. 1997) and a prominent ionization cone in the [OIII]A5007 maps 
(Marconi et al. 1995), with filamentary supersonic outflows (Veilleux & Bland-Hawthorn 
1997). The large equivalent width of the Fe 6.7 keV line detected by ASCA (EW ~ 2 keV) 
indicates that the nucleus is heavily absorbed along the line of sight: Nh > 10 25 cm -2 , i.e 
it is Compton thick. The Circinus galaxy is also characterized by enhanced star forming 
activity: Ha and [SII] narrow band images (Marconi et al. 1995) have revealed the presence 
of a star forming ring at a radius of ~10"=200pc. Given its proximity and the coexistence 
of both star formation and an active nucleus, the Circinus galaxy appears to be one of the 
most suitable objects to tackle the starburst-AGN connection issue. 

In this paper we present high angular resolution near-IR images and integral field 
spectroscopic data that probe the physics of the nuclear region of the Circinus galaxy, on 
scales from a few parsecs to ~ 100 pc. Using these data, we investigate the properties of 
the active nucleus and its interaction with the circumnuclear environment, with emphasis 
on the connection between Seyfert and stellar activity. 

2. Observations 

2.1. 3D integral field spectroscopy 

The nucleus of the Circinus galaxy was observed in March and April 1996 with 3D, 
the MPE near-IR imaging spectrometer (Weitzel et al. 1996), assisted by ROGUE, a first 
order adaptive optics system (Thatte et al. 1995), at the 2.2m MPE-ESO telescope. 3D 
slices the focal plane into 16 slits and disperses their light in wavelength; the spectrum of 
the slices is then imaged onto a NICMOS3 detector (256x256 pixels). The resulting "cube" 
provides simultaneous spectra of an area 4.8" x 4.8", as projected on the sky, divided into 
16x16 pixels (i.e. 0.3"/pix). We observed Circinus in two different spectral configurations: 
in March (two nights) we observed the whole K band at a spectral resolution of R=1000, 
while in April (four nights) we observed the spectral region around the CO stellar bands 
(2.2-2.4/im) at a spectral resolution of R=2000. The FWHM of the point spread function 
(PSF) during the observations, after the tip tilt correction, was about 0.5" — 0.6" in the K 
band. One of the nights in April had an optical seeing of 0.4", while the seeing in the K 
band, after tip tilt correction, was about 0.35", though we could not determine it accurately 
because of heavy undersampling of the PSF. We observed a nearby O star to correct the 
atmospheric transmission features in the R=1000 data, as O stars are almost featureless 
in the K band, while we observed an A star to correct the R=2000 data, as A stars are 
featureless in the corresponding spectral region. Data reduction was performed as described 
in Weitzel et al. (1996) and Thatte et al. (1997). 



-4- 



2.2. Speckle observations 

We also observed the Circinus galaxy by means of SHARP, the MPE speckle near-IR 
imaging camera at the NTT ESO telescope, in April 1996. This camera uses a NICM0S3 
array with a pixel scale of 0.05"/pix (providing a field of view of 12.8" x 12.8"). We observed 
the central region of Circinus in H and K bands at a frame rate of 1 Hz. The observations of 
Circinus were interleaved with observations of a reference star (about the same magnitude 
as the Circinus nucleus) to monitor the PSF. The individual images were background 
subtracted, flat fielded, shifted to center the brightest speckle and then co-added. Details 
on the data reduction are given in Eckart et al. (1993, 1995). The K band images have a 
PSF of 0.2" FWHM; however, if the diffuse, seeing limited PSF is subtracted, the central 
spike is close to the diffraction limit (~ 0.15" FWHM). In H, the PSF is slightly worse: 
0.27" FWHM. 



3. Results 

This section is aimed at showing general results extracted from the data, to provide 
an overview of the features observed in the central region of the Circinus galaxy. A more 
detailed analysis is given in Sect. 4 & 5. 

3.1. Near IR images 

Fig. la shows the K-band SHARP image of the central region of Circinus, smoothed 
with a 3 pixel (0.15") FWHM gaussian to improve the S/N on the low surface brightness 
features; while Fig. lb shows the central 3" at full resolution. The K band image shows a 
two armed spiral-like morphology. The western arm is also apparent in the 7000 A image 
(Marconi et al. 1995), but the southern arm is less evident in the optical image, which 
might be partly due to obscuration towards the South-East (see below) and partly to lack 
of angular resolution in the optical images. In the central region (R<80 pc) the surface 
brightness distribution is smoother and rises slowly toward the K peak, with a shallow 
power law (surface brightness oc R~ a with a ~ 0.5-0.7). This central region (R< 80 pc) bay 
be considered as the inner part of a bulge (a "nuclear bulge" , in analogy with our Galaxy, 
Mezger et al. 1996), though in this galaxy an outer bulge seems to be absent or very faint. 
The nuclear K band spike is unresolved and is most likely due to emission by hot dust, 
as we will discuss in sect. 4.1. The hot dust emission might also affect the H band nuclear 
peak, though this is more difficult to test. 



- 5 - 



Fig. 2 shows the H— K color map. To generate such a map we had to center the K band 
image relative to the H image. Since the nucleus might be heavily reddened, the peak of 
the H light might be offset with respect to the K peak. Therefore, we did not determine 
the relative position of the two maps by aligning their peaks. Instead, we used both the 
cross-correlation of the two maps (excluding the peaks) and the location of a point source 
situated north of the nucleus. The H-K color of an unreddened stellar population does not 
change significantly as it evolves; indeed, both spiral and ellipticals show a fairly narrow 
distribution of the H-K color, which clusters at 0.22 mags, with a standard deviation of 
about 0.1 mag (Frogel 1985, Giovanardi & Hunt 1996, Hunt et al. 1996). As a consequence, 
the H-K excess relative to the standard value traces regions affected by reddening. This is 
not the case for the Seyfert nucleus, because it has intrinsically redder colors. Moreover, 
PSF differences between H and K make color gradients around the nuclear spike unreliable. 
However, that is not a problem for the extended emission, where small variations of the 
PSF do not significantly affect the colors. Fig. 2 indicates a color gradient towards the SE. 
It is unlikely that this gradient is an instrumental effect due to imperfect flat fielding, as we 
found the same result by using different types of flat fields (night sky, twilight, and even 
without any flat field). Such a gradient very likely reflects an increase in the extinction 
towards the SE; this variation of the reddening is expected, as the SE is the near side of the 
galactic disk and therefore the dusty screen is expected to be more effective in this region 
(Quillen et al. 1995). However, the most interesting feature is the dust lane that enters 
the nuclear region from East toward South and twists toward the nucleus by 90 degrees at 
about 80 pc from the center. Such a dust lane (A v ~7.8) can account almost entirely for 
the interarm depression near the southern spiral arm observed in the K image (Fig. la), 
though a stellar spiral component is probably also present. 

3.2. Spectroscopic data: emission lines distribution 

In Fig. la the solid line box indicates the region sampled by 3D during the R=1000 
observations. Fig.3a shows the 3D, R=1000, spectrum extracted from an aperture of 
0".75 (=15pc) centered on the nucleus. The spectrum has been rebinned to R~700 to 
improve the signal to noise. Most prominent in this spectrum are the coronal lines: [SiVI] 
(ionization energy 167 eV)f], [CaVIII] (128 eV) 4 , and [MIX] (285 eV) 4 ; the latter is the 
first detection in an extragalactic object. Tab.l lists the fluxes of the lines detected in the 
nuclear region. The continuum rising toward longer wavelengths, and the stellar features 



4 This is the ionization energy of the lower ionization stage, i.e. the minimum energy required to obtain 
the observed ion. 



- 6- 



being much shallower than in normal galaxies, indicates that the non-stellar emission from 
the active nucleus contributes significantly to the continuum, diluting the stellar features. 
We extracted line maps by interpolating the underlying continuum with a first order fit in 
the AA ~ 0.2/xm around each line (by excluding regions affected by other lines and stellar 
features). An accurate continuum subtraction is one of the most critical issues, as the 
continuum slope and shape changes significantly over the field of view. Fig.4 shows maps 
of some of the lines detected in the K band spectrum. The last contour in each map is at 
3o" above the noise. The [SiVI] map extends in the direction of the ionization cone traced 
by the [OIII] maps, as indicated by means of yellow dashed lines (Marconi et al. 1995, 
Veilleux & Bland-Hawtorn 1997). In contrast, the Br7 emission is more diffuse and extends 
also along the major axis of the galaxy (P.A.~ 25°, see note 5). The [A1IX] shows the most 
compact emission. The nuclear H 2 emission is elongated towards the West, i.e. along the 
ionization cone, as [SiVI] does, but at lower surface brightness levels it also extends towards 
the NE, along with the Bvy emission. We will discuss these line maps in sections 4 and 5. 

3.3. Spectroscopic data: stellar kinematics 

The dashed rectangle in Fig. la indicates the region of the nucleus sampled by the 3D 
observations at R=2000. Fig. 3b shows the R=2000 nuclear spectrum, extracted from an 
aperture of 0".75. For comparison, the 3D spectrum of a template star (M3I) is shown in 
Fig. 3c (note the expanded scale on the y axis). 

We have used such spectra to map the stellar velocity field in the nuclear region of the 
Circinus galaxy. Both the galaxy spectra and the stellar template spectra were continuum 
subtracted. Then we used two different methods to determine both the velocity of the 
stellar system and its velocity dispersion. We first used a method similar to that described 
in Tonry & Davis (1979): the cross-correlation of the galaxy spectrum with the template 
star was fitted by the autocorrelation function of the star convolved with a gaussian, whose 
cr and peak velocity were free parameters. The second method consisted of fitting the 
galaxy spectrum in real space with a template stellar spectrum convolved with a gaussian: 
in this case the fit parameters were the a of the gaussian profile, the velocity shift applied 
to the stellar spectrum, and a multiplicative factor to fit the depth of the stellar features. 
Fitting was performed by using different stellar templates (K-M giants and supergiants) to 
monitor possible uncertainties introduced by a mismatch between the stellar template and 
the observed galaxy's stellar population. To minimize these mismatch effects, when working 
in real space, we separately fitted each stellar feature, so that the minimization of the x 2 
was not limited by differences in the relative depths of the features between the stellar 



- 7- 



template and the galaxy spectrum. Both fitting methods, i.e. fitting of the cross-correlation 
function and fitting in real space, gave consistent results. As the velocity dispersion turns 
out to be quite low (~80 km/s), i.e close to our resolution element, fitting the higher order 
moments instead of a simple gaussian (i.e. the kurtosis I13 and I14) was found not to be 
provide additional information when compared to uncertainties. Errors were estimated by 
determining the variation of the fit parameters that decreased the x 2 probability by 68% . 
Not all the stellar features in the spectrum could be used: the 12 CO(3-l) band is corrupted 
by the [CaVIII] emission line; the 13 CO(2-0) band is affected by a helium emission line; the 
12 CO(4-2) band is affected by some atmospheric features or by some unidentified emission 
line; the Nal doublet is corrupted by some instrumental effect in the nuclear spectrum (see 
Fig. 3b), but could be used in other regions. Therefore, we used only the 12 CO(2-0) band 
and the Cal triplet over the whole field of view, and the Nal doublet only in the off nuclear 
regions (though the latter does not affect the results significantly). 

Fig. 5 shows the map of the stellar peak velocity as derived by fitting a cubic spline 
to a grid of 21 points obtained by rebinning the field of view of 3D, to improve the 
signal-to-noise of each spectrum. The errorbars in the outer regions of the map are 
±15 km/s. Fig. 6 shows the stellar peak velocity and velocity dispersion along the major 
and minor axis of the galaxy (P.A. 25° and 155° respectively^ as measured by means of the 
3D spectra (filled dots). The stellar velocity dispersion is roughly constant in the central 
100 pc. In Fig. 6 we also plot, with hollow squares, the peak velocity determined by using 
the [Nil] emission line data from Oliva et al. (1994)0. The velocity field along the major 
axis shows a noticeable rotation pattern. 

4. The Seyfert nucleus 



5 The major axis of the galaxy is subject to some uncertainty. Freeman et al. (1977) quote P.A. 30°. The 
RC3 catalog gives P.A. 40°. Elmouttie et al. (1995) give a P.A. ~ 30° based on an AAT optical image. The 
near-IR images in Marconi et al. (1995) should be less affected by extinction and light from the young stellar 
population than optical images are. The major axis of the isophotes of these IR images is at P.A. ~ 25°. 
We suspect that the optical images are affected by patchy extinction in our Galactic plane, especially at the 
low surface brightness level. Indeed, the optical images show a weird tilt towards the East of the northern 
extremity of the galactic disk, which might be due to larger Galactic absorption in the northern region. 
Also, the crowding of Galactic sources makes it difficult to identify of the low surface brightness isophotes, 
probably used to measure the inclination of the galaxy. In this paper we assume the P.A. of the major axis 
to be that determined by means of the NIR images of Marconi et al. (1995), i.e. 25°. However, results do 
not change significantly if the actual major axis is 30° as determined by Freeman et al. and Elmouttie et al.. 

6 Their long slit was oriented almost in the same direction as the major axis: P.A. 20°. 



- 8- 



4.1. The nuclear dust emission 

The shallow stellar features in the nuclear K band spectrum, the slope of the spectrum 
(Fig. 3) and the extremely red colors of the nucleus indicate that the non-stellar emission 
from the Seyfert nucleus contributes significantly to the radiation observed in the nuclear 
region and that it dilutes the stellar features. As discussed by several authors (Efstathiou 
& Rowan-Robinson 1995 EF95, Granato & Danese 1994 GD94, Oliva et al. 1995, Thatte et 
al. 1997, Pier & Krolik 1993 PK93, McAlary & Rieke 1988), such K band radiation from 
Seyfert nuclei is very likely emitted by hot dust, at temperatures close to the sublimation 
limit (~ 1500 K), heated by the intense UV radiation field emitted by the Seyfert source. 
As a consequence, the K band images cannot be used to trace the stellar light in the nuclear 
region. The H band nuclear peak is also likely to be affected by hot dust emission (Thatte 
et al. 1997, Oliva et al. 1995). 

However, the CO stellar bands provide information about the fraction of the nuclear 
flux that is stellar. By assuming that the intrinsic equivalent width of the CO bands of 
the stellar population in the nuclear region is the same as in the circumnuclear region (R 
~ 50-70 pc), where the contribution of the nuclear hot dust source is absent, we can then 
use the measured EW(CO) on the nucleus to provide the amount of non-stellar dilution, or 
the fraction of the observed radiation that actually comes from stars. As the mass-to-light 
ratio turns out not to change significantly within the central ~200 pc (sect. 5. 2 and Fig.ll), 
the intrinsic EW of the CO stellar bands is not expected to change much within the central 
region. Therefore, the fraction of the stellar light in the nuclear region can be reliably 
estimated in the way we just described, with an accuracy of about 20% . 

In Fig. 7 we show the K-band surface brightness profile around the nuclear peak as 
derived from the SHARP image (filled dots and solid line), normalized to its peak intensity, 
compared with the fraction of stellar light (hollow circles) as derived by means of the CO 
stellar bands. The innermost point for the stellar light fraction is determined by extracting 
the 3D spectrum from an aperture of 0".3 (6 pc) from the data of the night with best 
seeing (~ 0".3). The dotted line indicates a fit to the stellar data using a modified King 
profile. It is noticeable that most of the nuclear emission is indeed non-stellar. If such a 
nuclear non-stellar source were unresolved, then the observed K-band profile should be the 
sum of the shallow stellar profile and a nuclear PSF, the latter being normalized to the 
peak value of the non stellar light (the PSF with this normalization is plotted with hollow 
squares connected by a long-dashed line). The sum of these components is indicated by the 
short-dashed line in Fig. 7 and, within uncertainties, it is identical to the observed K profile. 
Therefore, the non-stellar K-band source is point like at our resolution, i.e. it has a radius 
smaller than 1.5 pc. 



- 9- 



If Circinus were a type 1 Seyfert such a result would not be surprising, as the size of the 
region containing dust close to the sublimation limit is much smaller than 1 pc. Moorwood 
et al. (1996a) estimate an intrinsic luminosity of the Seyfert nucleus of ~ 10 10 L Q , therefore, 
by assuming the grain mixtures given in Rowan- Robinson (1986), we derive a size of the 
hot (1500 K) dust region R su b — 0.03 pc. In case of type 2 Seyferts the modelling is more 
complex. According to the unified model (Antonucci 1993) the nuclear hot dust should 
be partly obscured by the dense circumnuclear molecular torus. Therefore, the observed 
non-stellar near IR light should be partly associated with the short-wavelength tail of the 
mid-IR emission, and partly to radiation from the nuclear hot dust transmitted through 
the torus, the relative contribution of the two components being dependent on the optical 
thickness of the torus. Various authors have modelled the radiative transfer processes 
through the putative torus to determine the expected IR emission from type 2 Seyferts 
(PK93, GD94, ER95). In the model used by Granato et al. (1997) to fit the IR spectrum 
of NGC 1068, the size of the 2.2/zm source is expected to be ~ 70 x R su b (at a brightness 
level of 30% of the peak). If we scale the same model to Circinus, the size of the 2.2/xm 
source is expected to be ~ 1 pc in radius, consistent with our upper limit. Siebenmorgen et 
al. (1997) have proposed a spherically symmetric model to specifically fit the IR emission 
from Circinus nucleus. However, their model predicts a size of 4-4.5 pc (in radius) for 
the emitting region in the near-IR: such a value is 3 times larger than our upper limit at 
2.2/j.m. Therefore, the axisymmetric geometry expected from the unified model better fits 
the observational constraints in terms of size of the emitting region in the near-IR. 

In Tab. 2 we compare the luminosity of the non stellar nuclear K component, as 
determined from our data, along with the luminosity in the L' (3.8/im), M (4.8/xm) and 
N(10.3/im) bands from Siebenmorgen et al. (1997); the latter data should be little affected 
by stellar light within the small aperture listed in Tab. 2. The continuum rises steeply 
toward longer wavelengths. When compared to models, either PK93 or GD94 or ER95, 
such a steeply rising continuum indicates that the putative torus should be observed nearly 
edge-on. More specifically, given that the light cone in Circinus is quite wide (opening angle 
~ 100°), ER95's model can reproduce the observed spectral energy distribution only if the 
torus is highly inclined, so that it can significantly self-absorb the near-IR emitting region. 

4.2. The coronal lines 

As discussed by several authors (Korista & Ferland 1989, Oliva et al. 1994, Moorwood 
et al. 1996a, Oliva 1997, Marconi et al. 1996), the coronal lines observed in AGNs are most 
likely emitted by gas highly ionized by the hard UV-X ray nuclear radiation. Our [SiVI] 



-10- 



map (Fig.4b), showing the coronal line emission extending into the narrow line region, 
supports such a picture. However, the scenario is more complex. 

The [SiVI] (167 eV) coronal line emission map extends preferentially in the western 
direction, up to ~ 50 pc from the nucleus. Instead, the emission region of the [A1IX] (285 
eV) line is much more compact (Fig.4c) and, if any, it is only barely resolved. This is 
better illustrated by Fig. 8, where we plot the radial surface brightness of [SiVI], [A1IX] 
and of the PSF, within the sector —60° < P. A. < —95°. The [A1IX] radial profile departs 
from the PSF only by la, while [SiVI] is clearly resolved. Also, Fig. 8 indicates that the 
ionization state of the Narrow Line Region changes with radius in the western direction: at 
about 20 pc the [AlIX]/[SiVI] ratio is 2.7 times lower than on the nucleus at a confidence 
level of 5a. The difference is less pronunced in the northern part of the ionization cone, 
where the [SiVI] surface brightness drops faster and whose radial profile is closer to that 
of [A1IX]. Possibly, the western part of the radiation cone (P. A. ~ —90°) intercepts the 
dense, circumnuclear molecular clouds in the galactic plane, thereby lowering the ionization 
parameter lower[] and increasing the emissivity of the coronal lines (e oc n 2 ). This would 
explain the lower [AlIX]/[SiVI] ratio in this region and the enhanced [SiVI] emission with 
respect to the northern part of the ionization cone. Another possibility is that the ionization 
mechanism is different in the western extension. Collisional ionization, due to fast shocks, 
could contribute in this region. However, Oliva (1997), Oliva et al. (1994) and Marconi et 
al. (1996) show that ionization of the coronal gas as a consequence of fast shocks is very 
unlikely in AGNs, and specifically in Circinus. 

Finally, it is interesting to note that the peak of the [SiVI] emission is slightly shifted, 
by 0".15 (=3pc), with respect to the K band nucleus and to the peak of the Br7 and [A1IX] 
emission. Although higher resolution data are required to confirm such a finding, it is 
worthwhile to compare this result with the ionization structure expected to occur in the 
(inner) coronal line region (CLR). As discussed in Marconi et al. (1994), if the CLR is 
"thick" for Si +5 , i.e. the He + continuum opacity is responsible for determining its ionization 
structure, the peak of the [SiVI] emission should occur at the He +2 Stromgren radius, i.e. 

_ [ Q(He+) 1 1/3 

RlSiVI^eak) - [ nlaB{He+) _ 

(assuming a filling factor close to unity). By using the parameters for the Circinus 



7 01iva (1997) warns that the ionization parameter at the inner face of the cloud might be meaningless 
when dealing with the coronal line region (CLR), while a quasi-spherical photoionization model is more 
appropriate given the estimated pc-scale size of the CLR. However, at the distances (>20 pc) that we are 
considering here, the plane parallel slab model is a good enough approximation for the clouds in the galactic 
disk that are illuminated by the AGN continuum. 



- 11 - 



CLR given in Moorwood et al. (1996a), i.e. Q(He + ) ~ Q(H)/2 ~ 10 53 phot/sec and 
n e ~ 5000cm~ 3 , we estimate R\sivi](peak) ~ 5 pc, in agreement with the valuie observed in 
the 3D map. Such gas, responsible for the nuclear coronal emission lines, must intercept 
only a fraction of the nuclear solid angle, allowing the UV-X radiation to ionize gas in the 
outer light cone. 

A more detailed discussion on the coronal emission lines, including the [CaVIII] line 
(which requires a careful subtraction of the underlying stellar features) and their kinematical 
properties, will be presented in a forthcoming paper. 

4.3. The molecular gas 

There are three main mechanisms that may be responsible for the emission of the H 2 
lines observed in the near IR. These lines can be excited thermally in hot gas, either heated 
by shocks (Hollenbach & Shull 1977, Draine 1980, Draine et al. 1983) or by intense X-ray 
radiation (Lepp & McCray 1983, Maloney & Hollenbach 1996, Tine et al. 1997), or can be 
emitted under fluorescent excitation conditions if the UV radiation field is strong enough 
(Black & van Dishoeck 1988, Sternberg 1988). Near IR spectroscopy of AGN and starburst 
galaxies has indicated that UV fluorescence pumping is unlikely to play a major role in 
most of these objects (Moorwood & Oliva 1990, Sugai et al. 1997), and therefore thermal 
excitation must prevail. However, Sternberg & Dalgarno (1989), have studied the UV 
excitation with a more detailed model and found that i) the UV field can contribute to the 
gas heating and ii) when the gas densities are higher than 10 4 cm -3 collisional de-excitation 
of the H 2 levels becomes important, and the emitted spectrum resembles the thermal case. 

The maps of the H 2 emission from the Circinus galaxy (Figs.4.d-e) indicate that the 
nuclear emission is elongated 10-20 pc towards the West, i.e. in the same direction as the 
X-ray excited [SiVI] line. However, this is not sufficient evidence to support X-ray heating, 
as the UV field in this direction is also intense. Moreover, at lower surface brightness levels 
the H 2 emission also extends in the NE direction, i.e. it extends in the same direction as 
the Br7 emission (oc UV radiation), which probably traces star forming activity (sect. 5.1). 

The H 2 line ratios help to better understand between various excitation mechanisms. 
The curves in the diagrams on the left side of Fig. 9 show the expected fluxes of some of the 
H 2 lines, relative to H 2 (1-0)S(1), as a function of temperature, for pure thermal excitation. 
The dots in the diagrams on the right side show the same ratios according to the combined 
UV pumping+thermal models of Sternberg & Dalgarno (1989), as a function of gas density, 
for their x parameter (proportional to the intensity of the UV radiation) ranging from 10 



- 12 - 



to 10 4 ; except for the case of (l-O)Q(l), for which only the case x — 10 2 is given in their 
paper. 

The hatched areas indicate the ratios observed on the nucleus, and their thickness give 
the ±lcr range. The lightly-shaded areas indicate the ratios observed in an aperture ~ 
20 pc NE of the nucleus. The observed (2-l)S(l)/(l-0)S(l) ratio clearly rules out pure 
UV pumping (the low density regime of Sternberg & Dalgarno's models) as the excitation 
mechanism in both regions. Thermal excitation fits the data of the H 2 nuclear emission, 
within ~ 1.2 a, if the gas has a temperature of about 2500 K. For the UV+thermal models 
the agreement with the nuclear line ratios is worse, this is because in the very high density, 
quasi-thermal regime the molecular gas reaches temperatures that are too low (~1000 K). 
However, we cannot exclude that by varying the parameters in Sternberg & Dalgarno's 
model a better agreement can be obtained. The similar western extension of the H 2 and 
[SiVI] lines, together with these results suggests that the nuclear H 2 lines are mostly emitted 
thermally in gas heated by the X-ray, and possibly UV, radiation emitted by the AGN. If 
this were also the case for other AGNs (e.g. NGC4945, Moorwood et al. 1996b), then the 
H 2 emission line distribution in active nuclei would not necessarily trace the distribution of 
molecular gas, but it would also reflect the distribution of the X-ray radiation field. 

In the NE off-nuclear region, where extended Br7 emission is seen, the thermal model 
reproduces poorly the observed H 2 line intensities, as different line ratios measure different 
temperatures. In this region UV pumping might contribute to the gas excitation. More 
specifically, Sternberg & Dalgarno's model could fit the observed ratios if the gas density is 
close to the transition between UV-fluorescence and thermal regimes, i.e. ~ 10 4 ' 5 cm -3 . 

4.4. The nuclear dynamical mass 

As illustrated in Fig. 6, within the central 100 pc the bulge is isothermal within 
uncertainties. The smallest aperture over which we can sample the velocity dispersion is 
given by the seeing during the observations, i.e. ~ 0".5 — 0".6 (= 10-12 pc). Though on 
one night the seeing was as good as ~ 0".3, data from that night alone do not have a high 
enough signal-to-noise to sample the stellar velocity dispersion within the innermost 0".3, 
also due to higher dilution from the hot dust component in the central aperture. Following 
Kormendy (1988) and Kormendy & Richstone (1995), we extracted the spectrum of the 
nuclear stellar cluster (R<6pc), cleaned of the bulge contribution, by subtracting the mean 
of the spectra at ±17 pc along the minor axis (which have the same velocity shift as the 
nucleus). The resulting stellar spectrum has a velocity dispersion of 55 ± 15 km/ s. The 
velocity dispersion is lower in the nuclear region indicating that we do not detect a nuclear 



13 



point-like concentration of mass (Binney & Tremaine, 1987). By assuming the nuclear 
stellar cluster to have a virialized, isotropic distribution and a King core radius of 10 pc, as 
derived from Fig. 7, we estimate an upper limit of 4 x 1O 6 M to any compact concentration 
of mass. The crucial point, as we will show in sect. 5. 2, is that such a nuclear mass is not 
associated with a mass-to-light ratio higher than that of the circumnuclear region, as would 
be expected in the case of a nuclear concentration of dark mass significant with respect to 
the stellar mass. 

If the AGN is powered by an accreting black hole, then our upper limit on the black 
hole mass indicates that the AGN is radiating at a rate (1O 1O L , Moorwood et al. 1996a) 
that is higher than 0.1 times its Eddington limit, i.e. I^agn I^em >0.1. In NGC 1068, 
another Compton thick Seyfert 2 nucleus, ^agn I^em — 0.2 (Greenhill et al. 1996, Pier et 
al. 1994). Such high values of the L/Lem ratio might be common in these highly obscured 
AGNs. Indeed, as pointed out by Granato et al. (1997), the L/LEdd ratio is expected to 
increase with the circumnuclear column density (in their model L/LEdd oc Nfj). 

5. Star formation and stellar population 

As discussed in the introduction, the Ha and [SII] maps (Marconi et al. 1995) show 
evidence for a star forming ring at R = 200 pc, however Ha and [SII] emission is also 
observed inside the ring. The 3D Br7 map (Fig.4a) shows evidence for star forming activity 
in the close vicinity of the Seyfert nucleus: indeed, the Br7 emission extends outside the 
ionization cone, along the galaxy major axis, at 20-40 pc from the Seyfert nucleus. 

In this section we discuss the properties of the star formation and of the stellar 
population in the nuclear region of Circinus. We constrain age and luminosity of the stellar 
population by comparing the observed Br7 equivalent width (EW(Br7)) and mass-to-light 
ratio with values predicted by the stellar population synthesis code described in Sternberg 
& Kovo (1997)0. We assume a Salpeter initial mass function between 1 M and 60 M , and 
a broken power law for lower masses (Miller & Scalo, 1977). The star formation history is 
assumed to be exponentially decaying: 

SFR oc exp(—t/t ). 

When dealing with the nuclear stellar population (R < 6 pc) we consider two different 
models. Model "A" assumes the AGN and the starburst to be two different entities, and 



8 This code uses the stellar evolutionary tracks of Meynet et al. (1994) for solar metallicity, though in 
Sect. 5. 3. 3 we also consider the case of Z = 2Zq. 



-14- 



that the Br7 emission in the NLR is accounted for by the AGN. Model "B" assumes that 
the NLR is mostly ionized by a nuclear starburst (Terlevich's theory). 

5.1. Br7 equivalent width 

The ratio between Lyman continuum luminosity {L* Lyc ) and luminosity emitted in the 
K band (£Jf)P| by the stellar photospheres is a function of the ratio between the number of 
massive hot stars and number of giant-supergiant red stars. Therefore, L* L / L* K provides 
information on the star formation history. EW(Br7) oc L* Lyc /L* K if: 

i) there is no significant contribution from non-stellar sources to the ionizing photon flux; 

ii) no ionizing photons escape out of the plane of the galaxy; 

Hi) there is no differential extinction (in the K band) between the HII regions and the 
red giant-supergiant population (in this case EW(Br7) is independent of foreground 
extinction); 

iv) there is no contribution from non-stellar reprocessed light (e.g. hot dust emission) in 
the K band. 

Fig. 10a shows the evolution of EW(Br7) expected for continuous star formation 
(to = I0 9 yrs), an instantaneous burst (to = I0 6 yrs), and the intermediate case of 
t = 10 7 yrs, as derived by means of Sternberg & Kovo's code, and assuming that hypotheses 
ir-iv are true. 

Within the central ~20 pc, the Br7 emission is dominated by the PSF of the innermost 
part of the Narrow Line Region (NLR). When considering model A, regions contaminated 
by the NLR must be avoided. The lightly shaded area in Fig. 10a indicates the EW that we 
observe in the circumnuclear region, at 30-40 pc radius, outside the ionization cone (i.e. 
North and North-East). In this region we assume negligible deviations from conditions i-iv. 
The age of the burst derived from Fig. 10a is naturally quite sensitive to the assumed length 
of the burst (to). 

When dealing with model B (Terlevich's model) the Br7 emission from the NLR must 
be included in the parameters constraining the stellar population, i.e we should use the 
EW(Br7) measured on the nucleus (R < 6 pc). However, as discussed in Sect. 4.1, the 



ha is defined as the luminosity emitted in the 1.9-2.5 /im spectral range. 



- 15 - 



nuclear stellar light is significantly diluted by hot dust emission in the K band. Therefore, 
we corrected the nuclear EW(Br7) by subtracting the continuum contribution from hot 
dust. Moreover, the nuclear Lyman continuum flux is known to escape out of the plane of 
the galaxy through the ionization cone. Thus, we corrected for this effect by dividing the 
EW(Br7) by the covering factor. Moorwood et al. (1996a) estimate the nuclear covering 
factor to be 0.05 by comparing the Lyman continuum flux, as measured by means of the 
Ha flux from a spatially resolved cloud within the ionization cone, and the nuclear Bra 
flux (dereddened according to the nuclear Balmer decrement). The hatched area in Fig. 10a 
indicates the nuclear (R < 6 pc) EW(Br7) corrected for hot dust emission and covering 
factor. We will discuss the effect of possible differential extinction in Sect. 5. 3. 3. 

5.2. The mass— to— light ratio 

The mass-to-light ratio in the K band, M/L* K (see note 9), of the stellar population 
can better constrain both the age of the stellar population and the length of the burst t . 

The K-band (projected) stellar luminosity was determined for various apertures by 
means of the SHARP K image at radii 1".5 < R < 5" and by using photometric data given 
in Moorwood & Glass (1984) at larger radii. Similar to what is required for the EW(Br7), 
we are interested only in the K band light emitted directly by stellar photospheres and not 
in the contribution from reprocessed light. Therefore, when measuring L* K the contribution 
from the nuclear non stellar spike was subtracted. At radii < 1".5, we estimated the fraction 
of the K-luminosity that is stellar by means of the stellar CO map, as described in sect. 4.1. 

The extinction affecting the K-band stellar light was estimated by means of the H-K 
color excess with respect to the unreddened value of 0.22 (sect. 3.1). By assuming the 
extinction law given in Koorneef (1983), a foreground screen model gives Ax = 1.4 • Eh-k- 
The estimated extinction ranges from an average value of A^=0.3, within 18", to A^=0.63 
at l."5. The latter value is consistent with that derived by Oliva et al. (1995) using stellar 
spectra in both H and K bands, and it is also close to the extinction Ak = 0.52 derived for 
the innermost (~ 1") part of the NLR by means of the hydrogen emission lines (Oliva et al. 
1994). Colors involving shorter wavelength bands have larger intrinsic scatter amongst field 
galaxies and are more sensitive to age and metallicity effects, therefore they are less suitable 
to derive the K-band extinction. However, it is possible to perform a consistency check by 
comparing the V-K color with that expected for young and evolved stellar populations. 
Within the central 18" V-K=4.3 (Moorwood & Glass 1984), when compared to the average 
colors of ellipticals (V-K~3.3, Frogel et al. 1978) it yields ~ 0.1, while if compared to 
intrinsic colors expected for young stellar populations (V-K~ 1.7 for a 10 s yrs old stellar 



-16- 



population, Leitherer & Heckman 1995) the observed V-K color implies an A^=0.3. For 
apertures smaller than ~ 1".5 the H-K color cannot be used to derive the extinction 
because of the contribution from the Seyfert nucleus. Therefore, we assume the nuclear 
extinction to be the same as that derived from the H-K color at ~ 1".5. In sect. 5. 3. 2 we 
will discuss the effects of nuclear extinction (within the nuclear R < 1".5) higher than that 
assumed here. 

The included mass has been calculated by using the [Nil] emission line data at radii 
> 5", and assuming axisymmetric rotation. At radii < 2" we used the stellar velocity 
maps from 3D. We used the Boltzmann equation for a spherically symmetric space density 
distribution whose velocity dispersion is isotropic (Binney & Tremaine, 1987). Initially, 
we have also assumed the mass tracers to have a constant mass-to-light ratio within 
the central region, in order to estimate the deprojected volume density of mass tracers 
from the observed surface brightness profile (such a volume density p enters in the Jeans 
equation as a multiplicative factor in the form dln(p) / dln(r)) . The latter bit of reasoning 
might sound circular, as we will use it to estimate the mass-to-light ratio of the stellar 
population; however, it is possible to iterate the procedure and, within uncertainties, check 
for self-consistency. Due to uncertainties both in the measurements and in the model, the 
errorbars range from 40% up to a factor of 2 of the estimated enclosed mass. The mass 
derived in such a way is the dynamical mass, which is actually an upper limit to the stellar 
mass. Indeed, the gas mass might contribute significantly to the dynamical mass. There are 
no CO millimetric interferometric measurements of the Circinus galaxy, but CO(2-l) data 
obtained with a 22" beam (Aalto et al. 1995) indicate that the gas mass enclosed within 
the star forming ring is about 6 x 1O 8 M , if the Galactic conversion factor is assumed 
(Kenney & Young 1989), i.e. about half of the dynamical mass in the same region. On the 
other hand, the galactic conversion factor might not be appropriate in the central active 
region of the galaxy (Aalto et al. 1995, Casoli et al. 1992) and, therefore, the amount of 
molecular gas determined by means of the CO (2-1) line might be overestimated. Finally, 
the old stellar bulge population might contribute significantly to the dynamical mass, thus 
affecting the measure of mass-to-light ratio for the starbursting population. 

Fig.ll shows the estimated mass-to-light ratio M/L* K of the stellar population within 
various apertures. Though uncertainties are large, the M/L* K is clearly larger outside 
the star forming ring than inside it, indicating a younger stellar population in the inner 
region. In the central 30 pc the M/L* K ratio increases slightly, probably due to the mass 
contribution from an older stellar population, though it is still sustantially lower than 
typical M/L* K in bulges of normal galaxies and in ellipticals (~ 15 and ~ 35 respectively)^] 



These values are from Oliva et al. 1995, converted to our units, where we have deduced from their 



-17- 



characterized by a fairly old stellar population. In particular, we do not find a nuclear peak 
in the M/L as would be expected from the presence of a massive dark object. 

Fig. 10b shows the expected M/L* K ratio vs. age of the stellar population, as determined 
by using the Sternberg & Kovo code, for the same burst models as in Fig. 10a. The 
dark-shaded area corresponds to the M/L* K determined just inside the star forming ring, 
the lightly-shaded area corresponds to the value determined within 40 pc from the nucleus, 
while the hatched area gives the M/L* K on the nucleus (R< 6 pc). For a given M/L* K the 
age of the stellar population depends on the burst duration to a lesser extent than the Br7 
equivalent width. 

5.3. Starburst models and implications for the Starburst-AGN connection 

5.3.1. Starburst activity within R < 200 pc (model A) 

We initially consider model A, i.e. the AGN and nuclear star formation are two 
different entities and the AGN accounts for the Br7 emission in the NLR. By comparing 
Figs. 10a and b, we note that within the central R<40 pc (lightly shaded areas) the constant 
star formation model (to = 10 9 ) does not fit both observational constraints, i.e. EW(Br7) 
and M/L* K . In other words, in case of constant star formation the burst would have 
converted into stars a mass that is a factor of 100 higher than the observed dynamical mass. 
The 5-burst model (to = 10 6 ) also fits poorly both M/L* K and EW(Br7) constraints, though 
modelling of the 8 burst is more uncertain as it is more sensitive to the discrete sampling 
of the stellar evolutionary tracks. The parameters that best fit the observations within the 
central 40 pc are a moderate duration of the burst, t ~ 10 7 yrs, and an age of the stellar 
population of 5 x 10 7 yrs. Similar parameters are also typical of other starburst nuclei (e.g. 
Lutz et al. 1996, Rieke et al. 1993). 

If we assume that the same burst model (to = 10 7 yrs) applies to the whole region 
within the central 400 pc, then the observed distribution of M/L* K (Fig. 11) indicates that 
the age of the stellar population within this region ranges from 4 x 10 7 yrs to 1.5 x 10 8 yrs. 

It is interesting to note that by including the star forming ring, i.e. the annulus at 200 
pc traced by Ha and [SII] (Marconi et al. 1995), the M/L* K increases by about a factor of 2 
(Fig. 11). This might indicate that such a ring has undergone a more recent burst and that it 
has not yet entered in the red-supergiants phase. Circinus could be experiencing an outward 



Lh by assuming H-K=0.22. 



-18- 



propagating starburst: the nuclear region (R < 100-150 pc) experienced a burst about 
0.4 — 1.5 x 10 s yrs ago, and its near-IR continuum is now dominated by red supergiants, 
while the starburst activity has propagated to R~200 pc where it is currently forming 
young hot stars. Such a phenomenology is common to other starburst galaxies, such as 
M82 and IC342, where evidence has been found for a nuclear stellar population dominated 
by red supergiants few times 10 7 yrs old, while a circumnuclear ring is undergoing a more 
recent burst that is producing young hot stars (Rieke et al. 1980, Satyapal et al. 1997, 
Fdrster et al. 1996, Bdker et al. 1996, 1997). 

Fig. 10c shows the evolution of the bolometric-to-K band luminosity ratio, Ll ol /L* K , 
as a function of age, for the same models as in Figs. 10a and b. Given an average age 
of ~ 10 8 yrs for the stellar population inside R=200 pc, we get L* hol jL* K ~ 50 for the 
to = I0 7 yrs model. Note that the derived L* hol jL* K ratio is not very sensitive to the adopted 
burst model (to), if the M./~L* K is used as the age indicator, as the various models in Fig. 10b 
and Fig. 10c have opposite trends versus age. In the extreme case of constant star formation 
(which we ruled out) the Ll ol /L* K ratio increases only by a factor of 1.5, while a 5-burst 
would give the same Ll ol /L* K ratio as the to = 10 7 yrs model. 

The stellar bolometric luminosity can be derived from the observed K-band stellar 
luminosity by using the following relation: 



where L* K (obs) is the observed K-band stellar luminosity corrected for extinction. 

Given the K band luminosity within R < 200 pc, the total bolometric luminosity 
emitted by the stellar population within the same aperture is L^ ^ ~ 1.1 x 1O 1O L . This 
luminosity has to be compared with the intrinsic AGN luminosity, 10 10 L Q (Moorwood et 
al. 1996a), and with the bolometric luminosity of the galaxy, ~ 1.7 x 1O 1O L (corrected for 
Galactic absorption, but not for internal extinction)^, i.e. Seyfert nucleus and starburst 
contribute to a similar extent to the total bolometric luminosity. Tab. 3 summarizes the 
luminosity budget of the various sources in Circinus. It is not surprising that the sum of the 
starburst and AGN luminosities is slightly larger than the observed bolometric luminosity: 
in this edge-on system a fraction of the light escapes perpendicular to the galactic plane 
and, therefore, is not reprocessed into IR radiation nor is directly observed; in particular, a 



11 The IR luminosity from IRAS data is 1.2 x 1Q 10 Lq (Siebenmorgen et al. 1997). By using the photometric 
data in Moorwood & Glass (1983) along with the B^ in the RC3 catalog, corrected for a Galactic extinction 
Ay = 1.5, we estimate a luminosity of ~ 5 x 1O 9 L in the K to U wavelength range. 




model 




(1) 



-19- 



significant fraction of the radiation from the AGN is probably lost through its light cones. 
From the energetics point of view Circinus is similar to Mrk 477, the most powerful Seyfert 
2 in the local Universe, for which Heckman et al. (1997) estimate that the nuclear (few 100 
pc) starburst and the Seyfert nucleus provide most of the total luminosity of the galaxy, in 
about equal proportions. However, the starburst in Mrk 477 is at a much earlier stage (6 
Myr) than the starburst in Circinus. 

5.3.2. Starburst activity within R < 6 pc (model A) 

In the innermost 12 pc the Br7 emission is affected by the emission from the NLR. If 
we assume that the intrinsic EW(Br7) of the starburst within the central 12 pc is the same 
as that measured at ~ 30-40 pc, then the nuclear M/L* K ratio implies an age of the nuclear 
star cluster of about 7(±3) x 10 7 yrs, and a burst length t = 10 7 yrs. 

The L\ ol jL* K derived for this starburst model (Fig. 10c) implies a bolometric luminosity 
(eq.l) of the nuclear stellar population of L^(R<6pc) = 2(±0.6) x 10 8 L Q . Therefore, the 
nuclear stellar luminosity accounts for only ~ 2% of the Seyfert luminosity estimated by 
Moorwood et al. (1996a). 

As discussed in the introduction, David (1987b), Norman & Scoville (1988) and Murphy 
et al. (1991) proposed models that ascribe the AGN fuelling to mass loss from post-main 
sequence stars. The gas ejected from post-main sequence red giants and supergiants is 
kinematically hot, as its velocity dispersion reflects the velocity dispersion of bulge stars 
(~ 80 km/s for Circinus, Fig.6), and therefore it is characterized by the large turbulent 
viscosity that is required for the gas to inflow. More specifically, Norman & Scoville's (1988) 
model predicts the ratio between the luminosity of the active nucleus and the luminosity of 
the evolving stellar cluster, within the central 10 pc, to be ~50 when the nuclear cluster is 
70 Myrs old. This number is in fair agreement with the ratio between the intrinsic Seyfert 
luminosity and the luminosity derived for the star cluster in the nuclear 12 pc of Circinus: 
Lagn 1 ' L*(R < 6pc) ~ 46. This result does not necessarily prove these models to be correct, 
but it is the first time that observational data provide a consistency check for the theory of 
AGN fuelling through stellar mass loss. Finally, we note that Thatte et al. (1997) and Oliva 
et al. (1995) have found, though on larger scales, that other Seyfert nuclei are characterized 
by similar moderately young stellar populations (few 10 8 yrs). It is tempting to speculate 
that such intermediate age stellar populations are closely linked to Seyfert activity in the 
way we just depicted. 

The working hypothesis of this model is that the extinction in the K band toward 



-20- 



the nucleus is the same as that estimated by means of the H-K color at 30 pc from the 
nucleus (i.e. A K {nuc) = 0.63 mag, sect. 5. 2). However, the nucleus might be affected by 
an extinction higher than in the circumnuclear 30 pc. Correcting the nuclear K band 
luminosity for higher extinction results in a lower M/L* K ratio, i.e. a younger stellar 
population. The M/L* K ratio cannot be lower than the minimum value predicted by the 
model (Fig. 10b), this constrains the maximum possible excess nuclear extinction to be 
AA = 21.5 mag. If L^(nuc) is corrected for excess nuclear extinction, then the estimated 
bolometric luminosity of the stellar population, L* bol , increases both because of the increased 
Ll ol /L* K derived from the model (as the lower M/L* K indicates a younger population) and 
because of the increased L* K after correcting for extinction (eq.l). Fig. 12 indicates that if 
the extinction toward the nucleus is higher than what we derived from the circumnuclear 
region, the nuclear stellar cluster could be younger (as young as 10 7 yrs) and more luminous 
(as luminous as L| oi = 2 x 1O 9 L ). With regard to the implications for Norman & Scoville's 
model, a younger star cluster would provide an even larger mass loss available to fuel the 
accreting black hole. 



5.3.3. Starburst activity within R < 6 pc (model B) 

In model B (Terlevich's model) most of the luminosity of the active nucleus is due 
to starburst activity. In this discussed in sect. 5.1, the nuclear starburst also 

contributes to the ionization of the Narrow Line Region. Therefore, we should use the 
nuclear EW(Br7), corrected as described in sect. 5.1 (hatched area in Fig.lOa), to model 
the nuclear stellar population. A 5 x 10 6 yrs old <5-burst fits both the EW(Br7) and M/L^ 
ratio. The age derived for the nuclear star cluster is close to that expected in Terlevich's 
model during the Sy2 phase. However, the derived bolometric luminosity is ~ 20 x 1O 8 L , 
which is only 20% of the luminosity of the Seyfert nucleus^. 

However, the main problem is not the luminosity deficit, but the Spectral Energy 
Distribution (SED). Our starburst model predicts a SED peaking close to the Lyman 
edge, while the AGN radiates most of its energy around 100 eV and in the hard X-rays 
(Moorwood et al. 1996a). On the other hand, our model does not take into account the 
emission from SNR radiative shocks, expected to play a role at high energies in Terlevich's 
model. 



12 We should mention that the starburst models were also run by assuming the metallicity to be twice the 
solar value, to account for Terlevich's claim that AGN-like starbursts should be evolving in a metal rich 
environment, and the results were found not to change significantly. 



- 21 - 



There are other factors that might affect our estimate of the bolometric luminosity 
of the nuclear star cluster. If the foreground extinction is not uniform, but the nuclear 
star cluster is more obscured than the Narrow Line Region (differential extinction), then 
the observed EW(Br7) is higher than that expected from L* Lyc /L* K . In this case, the 
extinction-corrected EW(Br7) is lower than the observed value and the derived age of 
the burst is older, as shown in Fig. 13. The corrected M/L* K ratio cannot be lower than 
that expected by the model, and this constrains the maximum differential extinction 
to be Ay(stars)-Ay(NLR) < 30 mag. The derived bolometric luminosity changes as a 
consequence of two competitive factors: i) L^ (corrected) increases, but ii) (L* bol / L* K ) rnode i 
decreases as a consequence of the older age. The net behaviour is an increase of L£ oZ , 
as shown in Fig. 13. The corrected L^ could be as high as 7 x 10 9 L Q , for a differential 
extinction of 30 mags. 

Moorwood et al. (1996a) estimate the Lyman continuum luminosity by using the 
Ha flux from a "resolved" cloud in the NLR, whose projected distance from the nucleus 
is known. However, there are a few factors which lead to an underestimate of the real 
L>Ly C '- i) a fraction of the Ha flux from the cloud might be lost as a consequence of dust 
extinction within the cloud, ii) a fraction of the ionizing photon flux might be absorbed by 
dust in front of or within the cloud, Hi) the distance of the cloud from the nucleus might 
be larger than the observed projection, iv) the cloud might be made up of subclumps or 
filaments, i.e. the effective area of the cloud might be lower than observed. If the real 
L Lyc is higher than that estimated by Moorwood et al. (1996a), then in sect. 5.1 we have 
undercorrected the nuclear EW(Br7) when accounting for the covering factor. Therefore, 
correcting the Ll vc for effects i-iv makes the nuclear EW(Br7) larger and makes the derived 
age for the cluster younger, as shown in Fig. 14. The maximum value of the EW(Br7) 
predicted by the model imposes a maximum correction to the Lyman continuum luminosity: 

(corrected) /L^^c (observed) < 13. Correcting Li yc affects both the derived Ll ol (because 
of the younger age of the cluster) and the bolometric luminosity of the AGN (because the 
total L Lyc is the normalizing factor of the SED derived by Moorwood et al. 1996a). The net 
effect is a decrease of the ratio L^/L^AGN). The most extreme correction would imply 
a very young nuclear cluster (~ 10 6 yrs) which radiates about 8% of the luminosity of the 
Seyfert nucleus. 

Summarizing, if we assume that a compact nuclear starburst is responsible for most of 
the Br7 emission of the Seyfert nucleus, the starburst must be as young as 5 x 10 6 yrs, in 
fair agreement with the expectation of Terlevich's model, but the bolometric luminosity of 
the starburst is only 20% of the observed luminosity of the active nucleus. However, if the 
nucleus is affected by differential extinction the luminosity derived for the nuclear starburst 
could be higher by up to a factoor of 4. Yet, the main problem of fitting the observed SED 



-22 - 



in Circinus at high energies with the softer emission expected from the nuclear starburst 
remains unsolved. 

Finally, it is interesting to note that Veilleux & Bland-Hawthorn (1997) estimated 
that the filamentary ejecta, extending radially from the nucleus of Circinus, were probably 
expelled by an explosive nuclear event that occurred a few Myr ago. Such a time scale fits 
with the age of the nuclear burst estimated in this section. 

6. Summary 

We have studied the nuclear region of the Circinus galaxy by means of near IR images 
which have an angular resolution ~ 0".15 (= 3 pc as projected on the source), and K-band 
integral field spectroscopy with an angular resolution ~ 0".5 and spectral resolution of 1000 
and 2000. 

The K-band image reveals a nuclear non-stellar source that is unresolved, i.e. whose 
size is smaller than 1.5 pc in radius. Such nuclear non-stellar light is most likely emitted 
by dust heated by the Seyfert nucleus. When compared to models of the IR emission from 
obscuring torii, our upper limit fits the expected size of the K-band emission in Sy2s. 

The Coronal Line Region traced by the [SiVI] line extends for « 50 pc into the Narrow 
Line Region, but the higher excitation [A1IX] emission line is much more compact. The 
variation of the [AlIX]/[SiVI] ratio in the light cone might indicate that the ionization 
parameter of the NLR clouds decreases radially due to photoionization of high density 
clouds in the galactic plane. 

The nuclear emission from the H2 lines observed in the K-band extends in the same 
direction as the [SiVI] emission. The H 2 line ratios indicate that such lines are emitted by 
thermally excited gas, probably heated by the same X-ray radiation that excites [SiVI]. At 
lower surface brightness levels the H2 emission extends along the plane of the galaxy: in 
these regions UV-fluorescence might contribute to the excitation mechanism. 

The stellar velocity field traced by the stellar features in the K-band does not show 
evidence for a point-like mass concentration in the nuclear region. Also, there are no hints 
of an increase in the mass-to-light ratio toward the nucleus. We set an upper limit of 
4 x 1O 6 M for the mass of a putative black hole. As a consequence, the Lagn /^Edd ratio 
must be larger than 0.1. 

The Br7 map indicates evidence for ongoing star forming activity within a few tens of 
pc from the active nucleus. The mass-to-light ratio, from the 100 pc scale to the innermost 



-23- 



10 pc, is quite low and indicates that the stellar population is relatively young, with an age 
ranging between 4 x 10 7 and 1.5 x 10 8 yrs. There are indications of an outward propagation 
of the starburst. The bolo metric luminosity derived for the young stellar population inside 
the central 400 pc is comparable to the (intrinsic) luminosity of the active nucleus, and 
both of them together contribute most of the bolometric luminosity of the galaxy. The 
stellar population within R < 6 pc has an age of 7 x 10 7 yrs and contributes only 2% of the 
intrinsic luminosity of the active nucleus. 

We have discussed models that ascribe AGN fuelling to mass loss from a young nuclear 
star cluster in the light of our observational constraints. Our results fit predictions of these 
models in terms of expected nuclear stellar luminosity relative to the AGN luminosity. 

We have also considered the case of a nuclear starburst contributing significantly to the 
Br7 emission in the Narrow Line Region (Terlevich's hypothesis). In this case the nuclear 
(~ 10 pc) stellar population could be as young as 5 x 10 6 yrs, but the implied bolometric 
luminosity of the nuclear star cluster would be only 20% of the intrinsic luminosity of the 
active nucleus. However, differential extinction effects could lead to an underestimate of the 
luminosity of the nuclear star cluster by up to a factor of 4. 

We appreciate comments and helpful discussions with A. Moorwood, T. Storchi- 
Bergmann and G.L.Granato. We are grateful to S. Anders, A. Eckart, J. Gallimore and H. 
Kroker for help during the observations. We thank A. Sternberg for making available his 
code. R.M. acknowledge the support of the ASI grant ASI-95-RS-120. 



-24- 



REFERENCES 

Aalto, S., Booth, R.S., Black, J.H., & Johansson, L.E.B., 1995, A&A, 300, 369 
Antonucci, R., 1993, ARA&A, 31, 473 
Bailey, M.E., 1980, MNRAS 191, 195 

Binney, J.J, Tremaine, S., 1987, Galactic Dynamics, Princeton Univ. Press, Princeton, NJ 
Black, J.H., van Dishoeck, E.F., 1987, ApJ, 322, 412 

Boker, T., Forster, N., Krabbe, A., Genzel, R., 1996, in From Stars to Galaxies, ASP Conf. 
Ser. Vol. 98, Eds. C. Leitherer, U. Frits-v. Alvensleben, and J. Hucra, p. 455 

Boker, T., Forster, N., Genzel, R., 1997, ApJ submitted 

Casoli, F., Dupraz, C, Combes, F., 1992, A&A, 264, 55 

Cid Fernandes, R., & Terlevich, R., 1995, MNRAS 272, 423 

Cid Fernandes, R., Schmitt, Storchi-Bergmann, T., 1997, MNRAS, submitt. 

David, L.P, Durisen, R.H., Cohn, H.N., 1987a, ApJ 313, 556 

David, L.P, Durisen, R.H., Cohn, H.N., 1987b, ApJ 316, 505 

Delgado, R.M.G., Perez, E., 1993, Astroph. Sp. Sci 205, 127 

Draine, B.T., 1980, ApJ, 241, 1021 

Drainc, B.T., Roberge, W.G., & Dalgarno, A., 1983, ApJ, 264, 485 

Eckart, A., Genzel, R., Hofmann, R., Sams, B.J., Tacconi-Garman, L.E., ApJ, 407, L77 

Eckart, A., Genzel, R., Hofmann, R., Sams, B.J., Tacconi-Garman, L.E., ApJ, 445, L26 

Efstathiou, A, Rowan-Robinson, M., 1995, MNRAS, 273, 649 (ER95) 

Elmouttie, M., et al., 1995, MNRAS, 275, L53 

Forster, N., Boker, T., Krabbe, A., Genzel, R., 1996, in From Stars to Galaxies, ASP Conf. 
Ser. Vol. 98, Eds. C. Leitherer, U. Frits-v. Alvensleben, and J. Hucra, p. 417 

Freeman, C, Karlsson, B., Lyng, G., Burrell, J.F., van Woerden, H., Goss, W.M., Mebold, 
U., 1977, A&A, 55, 445 



-25 - 

Frogel, J.A., Persson, S.E., Matthews, K., Aaronson, M., 1978, ApJ, 220, 75 

Frogel, J. A., 1985, ApJ, 298, 528 

Genzel, R., et al, 1995, ApJ 444, 129 

Giovanardi, C, Hunt, L.K., 1996, ApJ 

Granato, G.L., Danese, L., 1994, MNRAS, 268, 235 (GD94) 

Granato, G.L., Danese, L., Franceschini, A., 1997, ApJ, in press 

Greenhill, L.J., Gwinn, C.R., Antonucci, R., Barvainis, R., 1996, ApJ 172, L21 

Greenhill, L.J., Ellingsen, S.P., Norris, R.P., Gough, R.G., Sinclair, M.W., Moran, J.M., 
Mushotsky, R., 1997, ApJ 474, L103 

Heckman, T.M., Blitz, L., Wilson, A.S., Armus, L., & Miley, G.K. 1989, ApJ, 342, 735 

Heckman, T.M., et al., 1995, ApJ 452, 549 

Heckman, T.M., et al., 1997, ApJ, in press 

Hollenbach, D.J, & Shull, J.M., 1977, ApJ 216, 419 

Hunt, I.K., et al., 1997, ApJS, in press 

Jenkins, A., Binney, J., 1994, MNRAS 270, 703 

Kenney, J.D., & Young, J.S., 1989, ApJ, 344, 171 

Koornneef, J., 1983, A&A, 128, 84 

Korista, K.T., Ferland, G.J., 1989, ApJ, 343, 678 

Kormendy, J., 1988, ApJ, 335, 40 

Kormendy, J., Richstone, D., 1995, ARA&A, 33, 581 

Leitherer, C, Heckman, T.M., 1995, ApJS, 96, 9 

Lepp, S., & McCray, R., 1983, ApJ, 269, 560 

Lin, D.N.C., Pringle, J.E., 1987, MNRAS, 225, 607 

Lutz, D., et al., 1996, A&A, 315, L137 



-26- 



Maiolino, R., Ruiz., M., Rieke, G.H., Keller, L.D., 1995, ApJ 446, 561 
Maiolino, R., Ruiz., M., Rieke, G.H., Papadopoulos, P., 1997, ApJ, 485, 522 
Maloney, P.R., Hollenbach, D.J., Tielens, A.G.G.M., 1996, ApJ, 472, 582 
Marconi, A., Moorwood, A.F.M., Salvati, M., & Oliva, 1994, A&A, 291, 18 
Marconi, A., Moorwood, A.F.M., Origlia, L., Oliva, E., 1995, ESO Mess, 78, 20 
Marconi, A., van der Werf, P.P., Moorwood, A.F.M., Oliva, E., 1996, A&A, 
Matt, G., et al, 1996, MNRAS, 208, 253 
McAlary, C.W., Rieke, G.H., 1988, ApJ, 333, 1 

Meynet, G., Meader, A., Schaller, G., Schaerer, D., & Charbonnel, C, 1994, A&AS, 103, 97 
Moorwood, A.F.M., & Oliva, E., 1990, A&A, 329, 78 
Moorwood, A.F.M., & Glass, I.S. 1984, A&A 135, 281 
Moorwood, A.F.M., 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, LI 

Moorwood, A.F.M., Origlia, L., Kotilainen, J.K., and Oliva, E., 1996c, in Proceedings of 
the ESO Workshop on Spiral Galaxies in the near IR, eds. D.Minniti and H.-W.Rix, 
p. 299 

Murphy, B.W., Cohn, H.N., Durisen, R.H., 1991, ApJ, 370, 60 

Neff, S., et al., 1994, ApJ 430, 545 

Norman, C, Scoville, N., 1988, ApJ 332, 124 (NS) 

Oliva, E., Salvati, M., Moorwood, A.F.M., Marconi, A., 1994, A&A, 288, 457 

Oliva, E., Origlia, L., Kotilainen, J.K., Moorwood, A.F.M., 1995, A&A, 301, 55 

Oliva, E., 1997, in Emission Lines in Active Galaxies: New Methods and Techniques, 
B.M.Peterson, F.-Z., Cheng, and A.S.Wilson (eds.), p.228 

Pier, E.A., Krolik, J.H., 1993, ApJ, 418, 673 (PK93) 



-27- 



Pier, E.A., et al, 1994, ApJ, 428, 124 

Quillen, A.C., Frogel, J.A., Kuchinski, L.E., Terndrup, D.M., 1995, AJ, 110, 205 

Rieke, G.H., Lebofsky, M.J., Thompson, R.I., Low, F.J., Tokunaga, A.T., 1980, ApJ 238, 
24 

Rieke, G.H., Loken, K., Rieke, M.J., Tamblyn, P., 1993, ApJ, 412, 111 
Rodriguez-Espinosa, J., Rudy, R., Jones, B., 1987, ApJ 312, 555 
Rowan-Robinson, M., 1986, MNRAS, 219, 737 
Satyapal et al. 1997, ApJ submitted 

Siebenmorgen, R., Moorwood, A., Freudling, W., Kaufl, H-U., A& A, in press 

Sternberg, A., 1988, ApJ, 332, 400 

Sternberg, A., Dalgarno, A., 1989, ApJ 338, 197 

Sternberg, A., Kovo, 0., 1997, in preparation 

Storchi-Bergmann, T., Rodriguez-Ardila, A., Schmitt, H.R., Wilson, A.S. Baldwin, J. A., 
1996, ApJ, 472, 83 

Suchkov, A.A., et al., 1994, ApJ, 430, 511 

Sugai, H., Malkan, M.A., Ward, M., Davies, R.I., 1997, ApJ, 481, in press 
Terlevich, R., Melnick, J., 1985, MNRAS 213, 841 

Terlevich, R., Tenorio-Tagle, G., Franco, J., and Melnick, J., 1992, MNRAS 255, 713 

Terlevich, R., 1994, in Violent Star Formation from 30 Doradus to QSOs, ed. G. 
Tenorio-Tagle, p. 329 MNRAS 255, 713 

Terlevich, R., Tenorio-Tagle, G., Rozycka, M., Franco, J., and Melnick, J., 1995, MNRAS 
272, 198 

Thatte, N., Kroker, H., Weitzel, L., Tacconi-Garman, L.E., Tecza, M., Krabbe, A., Genzel, 
R., 1995, Proceedings of the SPIE, 2475, 228 

Thatte, N., et al., 1997, ApJ, in press 

Tine, S., Lepp, S., Gredel, R., Dalgarno, A., 1997, ApJ, in press 



-28- 



Tonry, J., Davis, M., 1979, AJ, 84, 1511 

Veilleux, S., Bland-Hawtorn, J., 1997, ApJ, in press 

Weitzel, L., Krabbe, A., Kroker, H., Thatte, N., Tacconi-Garman, L.E., Cameron, M., 
Genzel, R., 1996, A&AS, 119, 531 



This preprint was prepared with the AAS IATeX macros v4.0. 



-29- 



Fig. 1.— a) K-band SHARP image, smoothed with a 3 pixel (=0.15") FWHM gaussian. 
Contours give Log(F^), in units of 10~ 24 W cvrc' 1 ' [inrr x arcsec~ x , at the following levels: 0.6, 
0.75, 0.8, 0.9, 1.0, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8. The green solid-line and dashed-line 
boxes show the regions observed with 3D at a spectral resolution of R=1000 and R=2000 
respectively, b) Enlargement of the central 3" of the unsmoothed K-band SHARP image. 
Contours are in the same units as in Fig.la at the levels: 1.2, 1.3, 1.4, 1.5, 1.6, 1.8, 2.0, 2.2, 
2.4. 

Fig. 2. — H-K color map, smoothed as Fig.la. Contours are at 0.25, 0.35, 0.45, 0.55, 0.65, 
0.75 mags. Regions affected by low signal-to-noise have been blanked. 

Fig. 3. — a) Nuclear spectrum of Circinus, extracted from an aperture of 0.75", rebinned 
to a spectral resolution of 700. Labels marked with a "*" show the position of absorption 
stellar features, while other labels identify emission lines. 6) The long wavelength part of the 
nuclear spectrum (same aperture as a) at a spectral resolution of 2000. c) Spectrum of one 
of the template stars (HD94613, shifted to the redshift of Circinus) used to fit the profile of 
the stellar features in Circinus' spectra. 

Fig. 4. — Maps of emission lines in the K band spectrum. The cross marks the position 
of the continuum peak. The last contour in each map is at 3a above the noise, a) Br7; 
contours are at 80% ,60% , 40% , 30% , 20%, 15% and 10% of the peak. 6) [SiVI] (the yellow 
dashed lines indicate the position of the ionization cone traced by the [OIII] line maps from 
Marconi et al. 1995 and Veilleux & Bland-Hawthorn 1997); contours are at 90% ,60% , 40% 
, 30% , 20%, 15% , 7.5% and 5% of the peak, c) [MIX]; contours are at 80% , 60% , 40% , 
30% , 20%, 17% and 12% of the peak, d) H 2 (1-0)S(1); contours are at 80% , 60% , 40% , 
30% , 20%, 15% of the peak, e) H 2 (1-0)S(2); contours are at 80% , 60% , 50% , 40% , 28% 
of the peak. In d and e two areas affected by bad pixels have been blanked. 

Fig. 5. — Map of the stellar velocity field as measured by means of the IR stellar features 
in the 3D spectra. Contour levels are spaced by 10 km/s, from 224 to 294 km/s. Errorbars 
in the outer parts of the map are ±15 km/s. 

Fig. 6. — Stellar peak velocity and velocity dispersion as measured by means of the stellar 
features in the 3D spectra (filled dots). Hollow squares indicate the velocity derived by 
means of the [Nil] emission line data in Oliva et al. (1994). 

Fig. 7. — Radial profile of the nuclear K-band surface brightness as derived from the SHARP 
image (filled circles and solid line). Hollow circles indicate the fraction of the K-band light 
emitted by stars as determined by means of the 3D maps of the CO stellar bands. The 
dotted line indicates a fit of the stellar fluxes by using a modified King profile. The hollow 



-30- 



squares connected by a long dashed line indicated the PSF profile normalized to the peak 
of the non stellar light. The short-dashed line is the sum of the stellar component (dotted 
line) and of the PSF (long-dashed line). 

Fig. 8. — Radial profile of the surface brightness of the [SiVI] and [A1IX] emission lines, 
compared to the PSF, in the sector -60° < P. A. < -95°. 

Fig. 9. — Flux of some of the H 2 lines detected in the K band relative to the H 2 (1-0)S(1) 
line. The hatched areas indicate the values observed on the nucleus (0.6" aperture), while 
the lightly-shaded areas indicate the ratios observed in an aperture at about 20 pc NE of the 
nucleus. The thickness of the shaded areas gives the ±cr range. The curves in the diagrams 
on the left side indicate the ratios expected in case of emission from thermally excited gas, 
as a function of temperature. The dots in the diagrams on the right side give the ratios 
predicted by Sternberg & Dalgarno's models (see text), as a function of gas density, for their 
parameter x (that is proportional to the UV radiation field) ranging from 10 to 10 4 ; except 
for the (1-0)Q(1) line where only the x — 10 2 case is given. 

Fig. 10. — a) Expected quivalent width of the Br7 versus age of a starburst in case of a 
5 burst (dotted line), a burst duration of 10 7 yrs (solid line) and continuous star formation 
(dashed line). The lightly-shaded area indicates the observed range of EW(Br7) at 30-40 
pc from the nucleus, outside the ionization cone. The hatched area indicates the nuclear 
EW(Br7) after correcting the Br7 flux for the covering factor of the NLR and the K band 
light for contamination by hot dust, b) Expected mass-to-light ratio for the same models as 
in a. The darkly-shaded area gives the M/L^ determined inside the star forming ring, the 
lightly-shaded area gives the same ratio inside R = 40 pc, while the hatched area gives the 
M/L^ on the nucleus (R< 6pc). The solid-line and dashed-line boxes indicate the range 
of values observed in bulges of spirals and in ellipticals (from data in Oliva et al. 1995). c) 
Expected bolometric-to-K band luminosity for the same models as in a. 

Fig. 11. — Mass-to-light ratio within various regions centered on the nucleus. 

Fig. 12. — Effects of correcting the nuclear K-band light for excess nuclear extinction with 
respect to the circumnuclear region, in model A. Top, effect on the age derived for the nuclear 
star cluster. Bottom, effect on the bolometric luminosity derived for the nuclear star cluster. 
See text about estimates of the maximum excess extinction. 

Fig. 13.— Effects of correcting the nuclear K-band light and EW(Br7) for differential 
extinction between nuclear stellar population and Narrow Line Region, in model B. Top, 
effect on the age derived for the nuclear star cluster. Bottom, effect on the bolometric 
luminosity derived for the nuclear star cluster. See text about estimates of the maximum 



differential extinction. 



Fig. 14. — Effects of correcting the nuclear EW(Br7) for asborption of UV and Ha photons 
in the NLR and other factors possibly responsible for reducing the L Lyc derived from the 
observations (see text), in model B. Top, effect on the age derived for the nuclear star cluster. 
Bottom, effect on the ratio between the bolometric luminosity derived for the nuclear star 
cluster and the bolometric luminosity of the AGN. See text about estimates of the maximum 
L Lyc correction. 



-32 - 



Table 1. Nuclear emission lines (aperture = 0.75") 



Line 




Flux 6 




H 


(10~ 15 erg cm" 2 s _1 ) 


H 2 (1-0)S(3) 


1.9576 


14.4±5 


[SiVI] 


1.9635 


70.2±5 


H 2 (1-0)S(2) 


2.0332 


6.09±1.2 


[MIX] 


2.043 


15.6±1.3 


Hel 


2.0581 


5.5±0.7 


H 2 (1-0)S(1) 


2.1218 


11.5±0.6 


Br7 


2.1655 


15.5±0.6 


Hell 


2.1885 


2.4±0.6 


H 2 (1-0)S(0) 


2.2235 


2.07±0.46 


H 2 (2-1)S(1) 


2.2471 


1.13±0.37 


[CaVIII] 


2.3213 


47.5±5 


H 2 (1-0)Q(1) 


2.4066 


9.5±1.9 



a Wavelengths are in the rest frame 

b Fluxes are corrected for a Galactic extinction 
of Ay = 1.5 mag, but uncorrected for internal 
reddenning. 



-33- 



Table 2. Luminosities and sizes of the nuclear non stellar source at various wavelengths in 

the near-to-mid IR spectral region. 



A (//m) 


i/L„ (xlO 7 L ) 


Radius (pc) 


2.2 (K) a 


1.5 


<1.5 


3.8 (L') b 


30 


3 


4.8 (M) 6 


56 


3 


10.3 (N) 6 


90 


13 



a From the SHARP image, corrected for the 
contribution of the stellar light as derived 
from the stellar CO map (Fig. 6) 

b From Siebenmorgen et al. (1997). 



Table 3. Starburst models and luminosity budget of the various sources in Circinus. 



Source 


Luminosity 


Age 


Burst duration 


Observed" 


1.7 x 1O 1O L 






AGN intrinsic 6 


1O 1O L 






Starburst (R<200pc) 


1.1 x 1O 1O L 


10 s yrs 


W 7 yrs 


Starburst (R<6pc), mod. A 


2 x 1O 8 L 


7 x 10 7 yrs 


10 7 yrs 


Starburst (R<6pc), mod. B 


2 x 1O 9 L 


5 x 10 6 yrs 


10 6 yrs 



a Bolometric luminosity corrected for Galactic extinction, but not for internal 
reddening (see note 10). 

b From Moorwood et al. (1996a). 



-34- 



a 



1.3 
1.2 
1.1 
1 

0.9 
0.8 - 



[SiVl] 



_u'i 
o 
I 



Nuclear spectrum 
(R=1000) 



[CaVIII] 
I 



^ r n H,(l-0)S(l) I 11= 
S [A1IX] aV ; Hell 

O I 

Hel 



H a (l-0)S(0) 





o 

w 

CD 
U 



1/ 



Mg'Al* Si * 



II ' 1 Cal" ^u. ^> ™ 
Ti'Na'Fel o 7 - 

Ti ' co 



o 



o o — 
u u 



_l I I l_ 



_l I I I 1_ 



2.1 



2.2 



2.3 



1.2 
1.1 
1 

0.9 

1.2 

1 

fc* 0.8 
0.6 
0.4 



o 

cT m 



Na* ~ 



Nucleus 
(R = 2000) 

Car la CO(2-0)- 



V 




Fe* 



H 2 (1-0)Q(1)- 

> 

cj 13 C0(2-0)* 
1/ 12 C0(4-2)' 

k 7 




j 1 1 i_ 



/ 12 C0(3-1)* 13 C0(5-3)* _; 

I I I I I I I L 




Template star 
(M3I) 



_L 





2.2 



2.25 



2.3 

X(/xm) 



2.35 



2.4 



Fig. 3.- 



-35 - 



10 pc 




1 -1 

Relative R.A. (arcsec) 



-36- 



(S) Major axis (N) 



600 -' 

















500 




>> 






-+-> 






'a 
o 


400 










> 







300 

ra 90 
60 
30 




150 300 



-40 -20 



20 40 




-20 20 
radius (pc) 



40 



Fig. 6. 



-37- 



radius (arcsec) 



0.25 0.5 0.75 1 




5 10 15 



radius (pc) 

Fig. 7.- 



-38- 




Fig. 8.- 



-39- 



I— Relative to H a (l-0)S(1) W/A Nucleus 
20 pc NE 



Thermal UV pump. + therm. 




1000 2000 3000 4000 2 3 4 5 6 7 
Temperature (K) Log density (cm -3 ) 



Fig. 9.- 



-40- 



Starburst Models versus Observations 
4 i 1 




6 6.5 7 7.5 8 8.5 9 9.5 10 



log(age) [yrs] 



10.- 



-41 - 



6 



~~\ i i i i i i i i i i i i i i 



J 4 



o 



1-J 



bulge i 



ring 



1 - 



J I I I I I I I I I I I I I I I I L 

10 100 

Radius encircled region (pc) 



Fig. 11.- 



-42- 



Effect of excess nuclear extinction 
(model A) 




J I I I I I I I I I I I I I I I I I I I I I L 

5 10 15 20 

A v (nuc)-A v (R = 30pc) [mag] 



-43- 




Fig. 13.- 



-44-