3.1. Optical morphology
The F606W image of Frosty Leo (Fig. 1 & Fig. 2) shows the previously known, extended bipolar nebula in scattered light. Two large lobes ( and ) of radial extent separated by a relatively dark waist can be seen. The lobes are significantly limb-brightened, indicating that they are optically-thin bubble-like structures with relatively dense walls and a tenuous interior. The bright ansae, and , seen at the tips of these lobes, appear knotty, and their shapes follow the curvature of the periphery of the lobes. has a higher surface brightness than by a factor which decreases from about 4 at a radial offset of 6" to 1.5 at the tips (i.e. in the ansae).
The complex inner region of Frosty Leo has several distinctive geometrical features. Two bright blobs ( & , saturated in image), lie north and south of a narrow waist with a flaring disk-like structure, although it departs significantly from a smooth, symmetrical geometry - e.g. the W side of the waist is significantly narrower than the E side. A bright point-like source, S, presumably the central star, is located in the waist at the center of the nebula, and additional compact knots appear attached to the inner edge of blob (inset, Fig. 1). The binary found by Roddier et al. (1995) at the center of the nebula has a separation of , and would therefore not be well resolved in our resolution HST images. However, the identical spectral type of both binary members is cause for concern that one of these may be due to light scattered off a dust condensation or an artifact of the adaptive optics technique.
Three radial jet-like structures, J1,J2, and J3, emanate from blob . The brightest of these, J1, shows three substructures - a very bright tip, and two bright limb-brightened lobes. J2 is similar in overall shape and size to J1, but its substructure is less well defined. J3 consists of a chain of knots lying on the periphery of an elongated structure, roughly similar in size to J1 and J2, but with a less sharply tapered shape. Faint counterparts to J1 and J2 are marginally visible on the northern side of the nebula (labelled J1', J2'); their faintness is probably due to their location on the far-side of the nebula resulting in significant attenuation by intervening nebular dust. The collimated structures J1, J2, J3, J1' and J2' probably signify the presence of low-latitude jets. The substructures in J1 and J3 seem to show a sequence of shock fronts, that are most likely due to temporal variations in the momentum flux of these jets. Two protrusions ( and ) appear at the periphery of the waist, and additional faint lobes ( and ) extend to the NW and E.
The bipolar nebula is surrounded by a faint, roughly round halo (Langill et al. 1994). Averaging the radial intensity over large (20o-35o) angular wedges with their apex at the center, we can trace the halo out to r23", where it becomes limited by background sky noise. The halo intensity, S, is well-fit by a power-law (S ), with 3.9-4.1, significantly different from the value (3) expected for scattered light in a spherical envelope characterised by (i) a constant expansion velocity and (ii) a constant mass-loss rate, . Therefore, either (i) and/or (ii) above are not true (e.g. due to an increase in over the last 30,000 yrs), or the halo density distribution has been modified by strong shock interactions.
The ansae have peak intensities, =0.27 () & 0.18 () mJy arcsec-2, comparable to the maximum possible intensity of scattered light at an offset = from the central star [given by =/(4=0.32 mJy arcsec-2, where =0.69 mJy is the stellar flux density at 0.6µm]. Similarly, the two bright knots near the tip of jet J1 have peak intensities of 0.8 & 1.5 mJy arcsec-2, which are only a factor 1.4-1.8 less than the theoretical maximum. Since the central region (radius 4") is quite optically thick even at 1 µm (Robinson et al. 1992), we conclude that the jets which have produced the ansae and J1, have carved out holes in this region, allowing the starlight to escape with very little attenuation. Line-emission from e.g. shocked gas, is not likely to be responsible for the excess brightness because (i) Morris & Reipurth (1990) found no line-emission from the ansae, and (ii) we compute similar excesses in the brightness of the ansae in the V and I-band images obtained by Langill et al. (1994).
3.2. Millimeter-wave emission
The observed CO spectra, with linear baselines subtracted are shown in Fig 3. The resolution is 2.7 (1.5) km s-1 for the 3 (1) mm lines. Small maps of the CO emission show that its angular extent is only marginally larger than the 12-13" telescope beam, and we estimate that the full size at half maximum of the intrinsic CO brightness distribution is smaller than 10", probably 5". The CO spectra show a narrow central component centered at =-11 km s-1, with a width of 20 km s-1. We assume that this strong feature comes from a compact, spherical component, expanding at 10 km s-1, although it may have a flattened disk-like geometry, as found in other PPNe from direct observations (e.g. Bujarrabal et al 1998, Sánchez Contreras et al 1997). This component probably represents the remnant of the AGB envelope that has not been significantly accelerated by interaction with a fast post-AGB wind. We find intense wings at both sides of this spectral component; comparison with other PPNe leads us to believe that the wing emission probably arises from material in the nebular lobes expanding parallel to the long axis of the nebula. Assuming an inclination of the nebular axis to the sky plane, =15o (Roddier et al. 1995), the maximum deprojected expansion velocity of the lobes is 190 km s-1, where , and 50 km s-1.
The intensity ratio 12CO/13CO for both transitions shows that the 13CO transitions are optically thin (optical depth 0.5), assuming a similar rotational excitation temperature () for both species. The mass, momentum and energy of the material emitting in each velocity channel can then be obtained from the 13CO line profiles, taking into account that the source is unresolved (Bujarrabal et al. 1997, 2000). is given for every frequency by
where is the main-beam solid angle, , is the absorption coefficient, dl is the length increment along the line of sight, and A and D are, respectively, the projected area of, and distance to, the source. This equation implies that , where is the mass moving at velocity v in the line of sight. The total mass of a kinematic component comes from integrating over the appropriate velocity range.
For the spherical component expanding with constant (10 km s-1), we get the "momentum" and the energy , where and the integral extends over the central spectral component's velocity range. For the fast component, each mass element is assumed to be in axial expansion with (); hence and , for every channel in the line wings (i.e. for every parcel of gas at high velocity). The scalar addition of all these momenta and energies gives the total scalar momentum , , and the energy, , of the fast outflow.
In our calculations we have used the 13CO J=1-0 data, since for this line the optical depth is lowest and is the largest (ensuring that the source is unresolved). We have assumed a 13CO/H2 abundance ratio of (Bujarrabal et al. 1997, Sánchez Contreras et al. 1997) and a distance of 3 kpc (see x1). The excitation temperature, is taken to be 10 K similar to that found in well studied PPNe, and consistent with the intensity ratio of the optically-thin 13CO transitions. In addition, if we reasonably assume that the 12CO lines are optically thick in the peak of the profile and that the source size is 5", then the peak implies 11 K. We have also carried out our computations using 15 K. In order to evaluate the uncertainty due to the poorly known geometry of the molecular outflow, we have also estimated and assuming that the high velocity emission comes from a spherical shell or a disk perpendicular to the nebular axis of the nebula. The results from all these calculations depend only slightly on the assumed geometry and excitation. The mass of the low- and high-velocity components are respectively 0.15 and 0.25 , the momenta of the fast outflow ranges between 2 and g cm s-1, and the kinetic energy ranges between and erg (for details see Bujarrabal et al 2000). We can compare these results with the energy and momentum that the stellar luminosity can contribute per yr, L = erg yr-1, = g cm s-1 yr-1. The time taken by Frosty Leo to evolve from the AGB stage to its current state must be 1000 yr, since 1000 yr is the ratio between the radius of ansae and the maximum velocity (190 km s-1), and the ratio between the radius of the CO emitting region and this velocity is even smaller. Moreover, the wind interaction time is thought to be still much smaller. Therefore, we conclude that the maximum momentum that the stellar luminosity could have contributed during this period is a factor 500 smaller than that carried by the fast outflow.
Finally, we estimate from the 12CO to 13CO intensity ratio in the line wings (where opacity effects are likely to be small) that the 12C/13C isotope ratio is 4. This value lies at the low end of the range for O-rich circumstellar envelopes around AGB stars.
© European Southern Observatory (ESO) 2000
Online publication: August 17, 2000