Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 347, 194-202 (1999)

Previous Section Next Section Title Page Table of Contents

4. Interpretation

4.1. The structure of the molecular envelope

We are most likely witnessing a former AGB-envelope where the bulk of the gas has been accelerated to (much) higher velocities by a high-velocity post-AGB wind that is inherently bipolar or that has been channelled by a dense disk, as e.g. in the post-AGB object M1-92 (Bujarrabal et al. 1994, 1997, 1998). The HVWs and the HVFs would then correspond to the gas in the polar directions that has been accelerated to the highest velocities, as e.g. in M1-92 and another (probable) post-AGB object OH231.8+4.2 (Alcolea et al. 1996). This is certainly consistent with the brightness distributions, Figs. 2, 3 and 4, where IVF emission still comes from a region close to the star, while the high-velocity gas seem to outline a bipolar outflow with a velocity increasing linearly with distance from the centre ([FORMULA]30[FORMULA] arcsec-1; the results for M1-92 and OH231.8+4.2 are [FORMULA]8 and [FORMULA]7[FORMULA] arcsec-1, respectively). The IVF probably originates from the equtorial regions of the former AGB-envelope. Below we will discuss the interpretation of the individual features and Fig. 5 shows a cartoon of a possible model for the source.

[FIGURE] Fig. 5. Cartoon of a possible model for the molecular envelope of HD101584

The LVF in the 13CO spectra, for which we lack spatial information, can be fitted with a parabolic line shape with [FORMULA]14[FORMULA] full width at the base, i.e., an expansion velocity of [FORMULA]7[FORMULA] if there are no projection effects. The 12CO/13CO line intensity in this velocity range is so low, [FORMULA]1.5, that the optical depths must be very high. It is not clear why this feature is not clearly seen in the 12CO lines. The most reasonable explanation is high optical depths in these lines (see Sect. 4.2). This suggests that the LVF originates in very dense gas, probably close to the star. Very similar features are present in the CO spectra of M1-92 (Bujarrabal et al. 1994, 1997), and the (probable) post-AGB object HR3126 (Nyman et al. 1998), where also the 12CO/13CO line intensity ratios are low, [FORMULA]2. Similar, but much narrower, features are seen towards the post-AGB objects 89 Her (Alcolea & Bujarrabal 1991; also here the 12CO/13CO line intensity ratio is low, [FORMULA]2) and the Red Rectangle (Jura et al. 1995). Except for 89 Her, a disk structure is regarded as the most likely explanation in these cases, and in M1-92 the emission is actually resolved (Bujarrabal et al. 1998). In fact, also towards normal AGB-stars such (very narrow) features in the CO spectra are not uncommon (Knapp et al. 1998), and they are interpreted as being due to varying wind characteristics with time, perhaps connected to thermal pulses (Knapp et al. 1998), or a disk channelling a bipolar outflow in the one case where data with sufficient spatial resolution exist (Kahane & Jura 1996). We cannot at this point exclude that the LVF originate from an AGB-envelope gas that has not experienced additional acceleration.

The IVF may originate in the outer parts of the disk, whose inner, denser parts are traced by the LVF, or in a separate component. Since this part of the CO emission is unresolved in our observations we do not have any spatial information, but we can use the high-resolution OH maser observations by te Lintel Hekkert et al. (1992) to discuss the structure of this component. We believe that the OH masers and the IVF of the CO emission originate in the same region because their velocity ranges agree very well, and there is no OH emission at velocities corresponding to LVF, possibly indicating that the densities in this region are high enough to collisionally quench the OH maser. The OH emission traces an apparent bipolar structure whose expansion velocity increases almost linearly with the distance from the centre. However, we do not believe that the OH masers are associated with the high-velocity CO outflow because their bipolar structure is not precisely aligned with the high-velocity CO emission, it has a smaller velocity gradient ([FORMULA]20[FORMULA] arcsec-1 compared to [FORMULA]30[FORMULA] arcsec-1), and, in particular, it expands in the opposite direction. The most attractive explanation to this is that the OH emission originates in the plane (or at low latitudes) of an expanding disk perpendicular to the CO outflow [an interpretation supported by the OH observations of M1-92 (Seaquist et al. 1991) and the planetary nebula Roberts 22 (Sahai et al. 1999)]. In this way the opposite expansion direction may be explained by a purely geometrical effect (see Fig. 5). If the OH masers are located in a disk it is required that the disk has a velocity gradient (there may also be a latitudinal velocity dependence), and that it is seen more edge-on than face-on (with the major axis approximately in the N-S direction). The latter provides a welldefined axi-symmetry, and may explain the alignment of the OH maser spots in terms of larger amplification paths in the radial direction than in the tangential direction, while the former results in the (linear) increase in the velocity of the OH masers with angular distance from the star. An upper limit to the inclination angle of the disk (with respect to the line-of-sight) can be obtained by noting that the true size of the OH emission (the apparent size is [FORMULA]4") is very likely smaller than the size of the central CO emission ([FORMULA]15" as estimated from our map data). This results in an inclination angle [FORMULA]75o, but this estimate must be regarded as very uncertain. If the disk and the outflow are perpendicular, the maximum velocity in the outflow lies in the range 150-550[FORMULA]. Another, but less likely, explanation to the apparent different outflow directions of the OH and CO emissions is that the outflow is directed almost along the line of sight, and that some precessional motion is present.

The relation between the putative disk, responsible for the LVF, and the putative disk, responsible for the OH emission and the IVF, is unclear. If the major fraction of the IVF emission emanates from the former AGB envelope one would expect this region to have the structure of a (thick) toroid. Since the outflow velocities in the IVF are much higher than that normally found in AGB-envelopes, a large fraction of the gas in the former AGB-envelope must have been substantially accelerated.

The IVF of the 12CO(2[FORMULA]1) emission also shows two horns, giving the line profile a double-peaked appearance. The horns are not visible in the 13CO(2[FORMULA]1) profile. In velocity they are located almost exactly at the edges of the main component of the 13CO(2[FORMULA]1) emission. We note here that the HD101584 12CO line profiles to a large extent qualitatively resemble those of the carbon star V Hya, except for the LVF and the HVFs (Kahane et al. 1996; Knapp et al. 1997). It is interesting though that Kahane et al. and Knapp et al. differ in their interpretations. The Knapp et al. scenario is the one most similar to ours, and they interpret the horns as arising preferentially from regions close to the minor axis of a flattened structure (with roughly the same expansion velocity in all directions) inclined towards the line of sight. In our case the horns lie close to the edge of the IVF suggesting that the flattened structure is seen almost edge-on. The geometry/kinematics combined with optical depth effects could possibly explain why the horns are not seen in the 13CO emission.

The HVWs most likely correspond to gas participating in a bipolar high-velocity outflow since the red- and blueshifted wings are clearly spatially separated (Figs. 3 and 4). The linear relation between velocity and displacement from the centre is most easily explained as due to a relatively short period of acceleration after which the gas has been expanding freely (i.e., the highest velocities reaches the largest distances in a given time), but it can also be due to the interaction between a fast bipolar wind running into a slow wind (e.g., Icke 1988). The HVFs at the edges of the line wings suggest that a fair fraction of the high-velocity gas somehow has reached a well defined terminal velocity, which to within a few% is the same on the opposite sides of the star. Such marked features have not been seen in any other post-AGB object with high-velocity winds, although similar, but much weaker, features may be present in the 12CO(2[FORMULA]1) data obtained towards OH231.8+4.2 (Alcolea et al. 1996). It is not obvious how gas being so strongly accelerated (probably by an order of magnitude) in two opposite directions will reach close to exactly the same terminal velocity. It is tempting to assign these features to emission from gas in interaction zones, between a fast outflow and a slowly moving AGB-wind, in the polar regions. The very similar expansion velocities of the HVFs could, as an alternative, suggest that this matter has been ejected at high velocities by some mechanism [we note here that in their spectral appearance these features resemble the "bullet"-features observed in bipolar outflows of young stars, that may be effects of episodic mass ejection (Bachiller & Cernicharo 1990)]. Despite not knowing whether one or several mass loss events created the high-velocity outflow we may crudely estimate size and time scales by adopting a very uncertain distance of 1 kpc (assuming a post-AGB object, Bakker et al. 1996b). The HVFs lie at a projected distance of [FORMULA]7[FORMULA]1016 cm from the centre. The dynamical time scale is 170[FORMULA]cot [FORMULA] yr (where [FORMULA] is the angle between the line-of-sight and the outflow axis, i.e., the same as the inclination angle of the putative disk introduced above; we assume here that the velocity does not change with radius and that the gas starts close to the star), i.e., the high-velocity outflow could be as young as [FORMULA]50 yr (the disk is seen almost edge-on), but also an age approaching 103 yr (the disk is seen almost face-on) is possible.

A high-velocity outflow is traced also in the broad P Cygni-profiles of the Balmer lines (Bakker et al. 1996b). These indicate expansion velocities of [FORMULA]100[FORMULA], suggesting that the CO high-velocity outflow is an effect of the present mass loss. Low-excitation optical emission and absorption lines are interpreted as arising in a circumsystem disk that is viewed almost edge-on, i.e., the same orientation that we infer for the region responsible for the CO IVF and the OH emission. It is tempting to assign the LVF to the inner regions of this disk.

4.2. The 12CO/13CO-ratio

There are significant, and systematic, variations in the 12CO/13CO line intensity ratios within the line profiles. The IVF, and in particular the LVF, and probably the HVFs have low ratios, [FORMULA]1.5-3, while the ratios in the HVWs are higher by at least a factor of [FORMULA]2-3. A change in the stellar 12C/13C-ratio as a function of time may in principle play a rôle, but this seems less likely in this case. We believe that the variation mainly reflects changes in optical depth. This makes the estimate of the 12CO/13CO-ratio very uncertain, but we expect it to be [FORMULA]10 (the maximum line intensity ratio found in the line wings, see Table 3).

Assuming that a 12CO/13CO abundance ratio of 10 applies also to the gas in the LVF and the IVF, and that there is no difference in the excitation of the two isotopic variants, we estimate optical depths for these components of [FORMULA]0.5 and [FORMULA]5 in the 13CO(2[FORMULA]1) and 12CO(2[FORMULA]1) lines, respectively, from a line intensity ratio of [FORMULA]2.

We note here that HD101584 shares the characteristic of a low 12CO/13CO line intensity ratio with a number of other (candidate) post-AGB objects: the O-rich objects 89 Her (Alcolea & Bujarrabal 1991), OH231.8+4.2 (Alcolea et al. 1996), M1-92 (Bujarrabal et al. 1997), and HR3126 (Nyman et al. 1998), and the C-rich objects M1-16 (Sahai et al. 1994), and the Boomerang nebula (Sahai & Nyman 1997). In M1-92 and OH231.8+4.2 the 12CO/13CO line intensity ratio also markedly decrease towards the line centre, and in the case of M1-92 the interferometric mapping shows that the low intensity ratio arises from material at the centre of the nebula (as seems to be the case also in HD101584, see Sect. 3.5), and that it is most likely due to a higher optical depth in this gas.

4.3. The mass of the molecular envelope

We can make a crude estimate of a lower limit to the mass of the molecular material, assuming LTE-excitation and optically thin and unresolved emission, using the formula,


where the normal symbols have been used for the constants, and [FORMULA] is the abundance of CO with respect to H2, D is the distance, [FORMULA] is the effective telescope area, I=[FORMULA][FORMULA]dv, Q is the partition function, [FORMULA] is the excitation temperature, and [FORMULA] is the energy of the upper level. Using the 13CO(2[FORMULA]1) line, [FORMULA]=0.1[FORMULA] (where [FORMULA]=3[FORMULA]10-4, a value appropriate for an O-rich envelope), and [FORMULA]=15 K (see the analysis of the CO emission from M1-92, Bujarrabal et al. 1997; furthermore if the (2[FORMULA]1)/(1[FORMULA]0) intensity-ratio of the 13CO-lines is mainly attributed to beam filling the temperature must be low since the 13CO optical depths appears to be less than one), we obtain M[FORMULA]0.1[FORMULA]. This is much less than the envelope mass estimated from the IR emission (2[FORMULA]), suggesting that either the dust mass estimate is severely in error, or the dust and CO emission come from different regions. We regard it as unlikely that the CO mass is severely underestimated, unless a fair fraction of the CO molecules are photodissociated (if so, predominantly by the interstellar UV radiation).

We can also make a crude estimate of a lower limit to the final AGB mass loss rate. At least half of the emission (corresponding to [FORMULA]0.05[FORMULA]) comes from a region smaller than [FORMULA]5" in radius (corresponding to a time scale of 1000 years if we assume an AGB wind velocity of 20[FORMULA]). Thus, the AGB mass loss rate must have been well above [FORMULA] during the final evolution, and maybe considerably more if a major fraction of the original CO is photodissociated.

4.4. The wind momentum

A crude lower limit (e.g., no correction for inclination) to the (excess) momentum of the gas that has been accelerated to velocities above that of the AGB wind is obtained by applying Eq. (1) to the gas outside the central [FORMULA]20[FORMULA] (of the systemic velocity) in 10[FORMULA]-intervals and by multipying with the velocity increase (i.e., the average velocity in the 10[FORMULA]-interval minus 20[FORMULA]). The result is [FORMULA]5[FORMULA]1038 g cm s- 1. This corresponds to the total momentum supplied by a radiation luminosity of [FORMULA]104[FORMULA] in 104 yr [i.e., an order of magnitude longer than the (rough) estimate of the maximum dynamical age of the high-velocity gas]. Even though the estimate is crude, it clearly points towards a mechanism different than that of a radiation-driven wind. Bujarrabal et al. (1998) found a similar situation for M1-92, and they suggest an accretion-driven wind.

4.5. The SiO and HCN data

Neither SiO nor HCN emission was detected despite a relatively sensitive search. The line intensity ratios are: SiO(3[FORMULA]2)/CO(1[FORMULA]0) [FORMULA] 0.07 and HCN(1[FORMULA]0)/CO(1[FORMULA]0) [FORMULA] 0.06. The SiO/CO-ratio is very low for a normal O-rich AGB-envelope, and the HCN/CO-ratio is very low for a normal C-rich AGB-envelope [Olofsson et al. 1998; note that we expect the SiO(3[FORMULA]2)/CO(1[FORMULA]0)-ratio to be higher than the SiO(2[FORMULA]1)/CO(1[FORMULA]0)-ratio]. Considering the detection of an OH maser and a possible detection of a 10 µm dust feature, the most reasonable explanation to our results is that the envelope is O-rich but that presently there is very little gas left in the inner regions of the envelope where SiO is normally excited. We note here also the marked difference with OH231.8+4.2 towards which a large number of molecular species have been detected (Morris et al. 1987; Lindqvist et al. 1992; Sánchez-Contreras et al. 1997).

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1999

Online publication: June 18, 1999