Astron. Astrophys. 321, 907-920 (1997)

3. Distribution and kinematics of H in AFGL 2688

AFGL 2688 (the "Egg nebula") is one of the rare sources known to be in the rapidly evolving transition from the AGB to the PN phase. First reported by Ney et al. (1975), AFGL 2688 has been since the subject of numerous observational and theoretical investigations (Latter et al., 1993 and references therein). It is a bright infrared source with a bipolar optical and near-infrared nebula scattering the light of the cool central star (F5 supergiant, ). Using photon scattering models, Yusef-Zadeh et al. (1984) convincingly reproduced the optical and near-infrared appearance of AFGL 2688 (see also Latter et al. 1993) with the equatorial torus lying in the east-west direction and the bipolar axis aligned north-south along the reflection nebula. Single-dish measurements in millimeter lines reveal a nearly circular, cold and dense component tracing the slowly expanding AGB envelope, which is shocked by a warm, optically thin, fast wind (Young et al. 1992; Yamamura et al. 1995). The fast wind presents an asymmetric morphology, with the blue and red-shifted gas aligned along two axes instead of one as expected for a bipolar structure (Jaminet et al. 1992).

Molecular hydrogen was first observed toward AFGL 2688 by Beckwith et al. (1978) and subsequently by Thronson (1982). Beckwith et al. (1984) detected two H₂ emission peaks towards the visible lobes which are aligned along the poles of the nebula. This detection strengthened the idea of shock-excited emission along the bipolar axis. By using a Fabry-Perot spectrometer, Smith et al. (1990) were able to study both the morphology and the dynamics of the molecular hydrogen emission. Besides the northern and southern clumps already observed by Beckwith et al. (1984), they discovered two additional clumps to the east and to the west, which are also seen in the image published by Gatley et al. (1988). Although a significant velocity spread is present in the observed line profiles, the dynamics of these H₂ clumps indicate that the north and the east lobes have a peak velocity of , whereas the south and west clumps are red-shifted at . Such dynamics are clearly more complex than a simple bipolar flow along the north-south direction leading Smith et al. to suggest a large cone-opening angle. These findings are in agreement with the distribution of the high-velocity CO gas derived by Jaminet et al. (1992). The four H₂ clumps are also seen in the broad band images of Latter et al. (1993) after a careful subtraction of the continuum. They interpret the outflow as bipolar along the north-south axis, and explain the E-W velocities as a result of rotation of the entire nebula along the polar axis (Bieging & Nguyen-Quang-Rieu 1988, 1996). More recently, Skinner et al. (1997) have presented new near-infrared images which convincingly show the four clumps of emission, with perhaps some emission linking them, and, in addition, two weaker clumps of molecular hydrogen lying at on each side of the center along an axis with a position angle of .

Our observations of the distribution of the 1-0 S(1) and 1-0 S(0) transitions in AFGL 2688 measured with BEAR are shown in Fig. 1. The third panel shows the continuum emission measured in the K'-band. The molecular hydrogen emission in both transitions reveals four bright clumps forming a remarkable cross-like pattern. The clumps are at about the same distance () from the central (hidden) star. No continuum emission or line emission other than is associated with these structures. Whereas the clumps to the south and west (hereafter S and W) are relatively weak and somewhat compact, the two other clumps to the north and east (hereafter N and E) are clearly more intense and elongated. Weaker emission (best seen in the more intense 1-0 S(1) line) also links both clumps S and W, and clumps N and E, respectively. No H₂ emission links clumps E and W or is detected in the central part of the nebula. The K'-band continuum emission is slightly elongated along the north-south direction and is mainly detected in between clumps N and S, i.e., nearer to the central star than the emitting gas. No continuum emission is detected in the direction of clumps E and W. The clear decrease of the near-infrared continuum and the H₂ line emission towards the central regions is probably due to opacity effects (see below). The continuum emission is stellar light scattered by the dust of the inner cavity.

Fig. 1. Images of AFGL 2688 in the 1-0 S(1) and 1-0 S(0) transitions of H2 and in the K'-band continuum (left, middle, and right panels, respectively). In the line images the continuum has been subtracted. The position of the field center (which is assumed to be the position of the hidden central star) is shown by an asterisk. The lowest contour and contour intervals are 10% of the peak intensities which (in units of ) are: a (left) 6.4 10-3, b (middle) 1.2 10-3 and c (right) 1.3 10-3.

The kinematics of H₂ in AFGL 2688 are shown in in three velocity channel maps for the 1-0 S(1) line in Fig. 2. The data were obtained with the narrow H₂ filter and the continuum has been subtracted. Two points are immediately clear from Fig. 2. First, the four clumps are detected at all velocities, indicating a significant velocity dispersion in these structures, and second, the S and W clumps are red-shifted with respect to the N and E clumps with a velocity difference of the order or less than (a velocity channel). These results are in accord with the velocity information derived by Cohen & Kuhi (1980) from optical observations of the two nebular lobes, where a velocity difference of about is found with the northern lobe approaching us and the southern lobe receding from us. The present results also confirm the earlier H₂ observations by Smith et al. (1990), and provide a much more detailed view of the H₂ morphology and velocity field.

Fig. 2. Velocity channel maps of AFGL 2688 in the H2 1-0 S(1) line. The continuum emission has been subtracted from the data. The panels are displayed in order of decreasing velocity (red-shifted to blue-shifted) with a velocity step between each frame of 26.9  . The systemic velocity is

The H₂ clumps in AFGL 2688 show structure on the scale of one arcsecond with systematic variations in position and morphology with velocity. The northern clump, which is brighter than the southern clump, appears elongated towards the SE at systemic and blue-shifted velocities. At red-shifted velocities, a comparable elongation (from S to NW) is seen towards clump S. In clump N, the peak emission shifts by about from north to south when going from red- to blue-shifted velocities. Although less contrasted, the same shift is observed toward clump S. Such a velocity dispersion along the south-north axis is consistent with gas outflowing in a small-opening cone along the polar regions, the bipolar axis being inclined to the plane of the sky with the northern lobe towards the observer (from an analysis of the scattered light in the optical, the inclination angle is found to be , Yusef-Zadeh et al. 1984).

The interpretation of the H₂ velocity field in the equatorial plane is less straightforward. The individual channel maps show that the E and W clumps are elongated with weaker emission extending towards the inner regions of the nebula. The E clump (which is brighter at blue-shifted than at red-shifted velocities) is stretched along the north-south direction with a bright peak lying to the south. A weak, curved H₂ filament extends from this bright peak towards the center and then bends towards the north. The W clump, which is somewhat more compact than clump E and is mostly detected at red-shifted velocities, has one bright core with a V-like extension pointing towards the center. The bow-like morphology of clumps E and W and the asymmetrical distribution of the H₂ emission (the emission being more extended to the west than to the east) suggest that the H₂ near-infrared lines trace the edges of a tilted equatorial torus.

At this point it is of interest to compare the morphology of the molecular hydrogen emission with other observable features of the envelope. Fig. 3 shows contours of the H₂  1-0 S(1) emission superposed on the optical image in the V-band from Crabtree & Rogers (1993). Besides the two polar lobes of reflected light, the optical image shows horn-like features which extend more than 20 arcsec from the central parts of the nebula, and a series of concentric rings which are centered on the obscured exciting star. These rings probably trace the episodic mass-loss events which took place during the AGB phase. The distance between each ring is typically about 4 arcsec. Adopting an expansion velocity of 20  , this corresponds to time intervals of about 900 years (for an adopted distance of 1 kpc) which could be compatible with the helium flash cycle of the central star. The H₂ clumps N and S are found towards the brightest regions of the optical emission. For both clumps the peak H₂ emission is located in between the bright plateau of emission in the optical image and the innermost horn-like structures. The H₂ clumps E and W are distributed along the outer regions of the optical emission with their peak emission being exactly aligned along the axis defined by the waist of obscuration. We note that the weak extension of clump E towards the west is also directed towards the central waist. Recently, Lucas (1994) presented high-resolution maps of AFGL 2688 in HC₅ N and SiS obtained with the Plateau de Bure interferometer (Lucas et al., in preparation). The SiS emission which probes the densest gas is compact and peaked on the center, whereas the HC₅ N distribution is clearly more extended, delineating a hollow, fragmented shell. A comparison of the near-infrared H₂ maps with these new interferometric maps shows that the E and W H₂ clumps exactly delineate the east and west outer boundaries of the dense, central regions probed in HC₅ N and in SiS. Similar conclusions are reached when comparing the H₂ distribution with high-resolution maps of the mid-infrared dust continuum (Jaye et al. 1989; Skinner et al. 1997) or with the optically thin dust emission at  3-mm presented by Lucas et al. As for the HC₅ N and SiS distributions, the molecular hydrogen exactly follows the eastern and western edges of the inner dust/molecular shell. The filaments of emission connecting the N-E and S-W clumps also follow the outer contours of the mid-infrared/millimeter emission. But the northern and southern H₂ clumps, which lay outside the volume defined by the molecular gas and dust emission, are not morphologically correlated with the outer boundaries of the molecular/dust torus as in the case of the eastern and western H₂ clumps.

Fig. 3. Comparison of the H2 1-0 S(1) emission in AFGL 2688 with the optical image in the V-band from Crabtree & Rogers (1993)

The blue-to-red velocity shift between the E and W H₂ clumps is also observed in the molecular data presented by Lucas (1994) and Bieging & Nguyen-Quang-Rieu (1988, 1996). The latter authors suggested that the equatorial regions are rotating (they measured a velocity difference of about 4 km s^-1 along the optical dark lane). However, the velocity difference found in the molecular hydrogen is about an order of magnitude higher. Accounting for this difference in terms of rotation alone would imply unphysically large orbital velocities. Therefore, we suggest that the kinematics of the inner regions of AFGL 2688 are dominated by expansion and that the molecular hydrogen traces the outer, shocked (see below) regions of a dense equatorial torus. It is important to note that extinction almost certainly hides part of the H₂ emission. From Knapp et al. (1993), the flux density of AFGL 2688 at 1.3 mm, which is due to optically thin dust emission and originates in the inner regions of the nebula (Lucas 1994), is 2.1 Jy in a  beam. Taking the usual conversion factors (e.g., Mezger et al. 1990) and adopting a dust temperature of 100 K, this flux yields a beam averaged column density of hydrogen . This corresponds to A  160 mag or A  20 mag, and is consistent with the large equatorial absorption seen in the infrared continuum. Thus much of the H₂ emission from the inner torus is likely to be obscured, and our view is determined by the disk structure. It may be incomplete or warped, which could explain the differences between the observed E and W clumps of H₂ emission. The detailed structure of the inner regions may also allow the fast wind to break through near the equatorial plane in two opposite direction, forming the two main regions of interaction where the strongest H₂ emission is seen. Future millimeter observations at high spatial resolution are needed to study in detail the relation in the equatorial plane between the dense, central torus which is probably rotating (Bieging & Nguyen-Quang-Rieu 1996) and the high-velocity molecular gas traced in H₂ (this paper) and in CO (Young et al. 1992).

© European Southern Observatory (ESO) 1997

Online publication: June 30, 1998
helpdesk.link@springer.de