3. The standard model For the AFGL2688 nebula
From the very extensive observational database available, a fairly detailed model for AFGL2688 has evolved, which is discussed in some detail by Latter et al. (1993), for example. According to this "standard model", the progenitor C-star left the AGB a few hundred years ago, at which time the star had a very massive, and equatorially concentrated wind. This generated a dense equatorial torus of dust, which currently totally obscures the central star. Visible and near-IR radiation from the central F5Ia star escapes along the polar axes and is scattered off dust, forming the spectacular bipolar reflection nebula. The current fast wind from the warm central star is focussed along the polar axes by the dense equatorial torus, generating a very fast bipolar flow, which punches its way through the surrounding, more or less spherical and slower AGB wind. The axis of the bipolar flow lies at position angle 15 , and is inclined at about 5-15 to the plane of the sky. Despite its wide acceptence, we find that there are a number of observations that the standard model fails to explain, and we outline these in this section. In the next section we will propose an alternative model which we suggest accounts better for AFGL2688's observational properties.
Foremost is the velocity structure revealed by the H2 images of Smith et al. (1990). The bipolar flow is supposed to be aligned roughly N-S, yet the N and E H2 blobs have the same velocity to within a few km/sec, as do the W and S blobs. The velocity difference between the two pairs (N-E and S-W) of blobs is of order 30km/sec. Latter et al. ascribed the velocities of the E and W blobs to the rotation of the torus claimed by Bieging & Rieu (1988). However, the rotation velocity 4" E and 4" W of the central object in Bieging & Rieu's data is only 1km/sec, an order of magnitude too small. Additionally it is exceedingly unlikely that the projected velocities of the N and S bipolar lobes would coincidentally be identical to those of the E and W limbs of the rotating torus. It is much more likely that the N and E blobs are physically related to one another, and likewise the S and W blobs.
Second, Young et al. (1992) reported velocities up to 100km/sec in the outflow from AFGL2688. If the bipolar axis is inclined at 15 to the plane of the sky, then the bipolar flow velocity, by the arguments presented by Young et al., is 350km/sec. If the inclination angle is close to 5 , as required by the "standard model" (e.g. Latter et al. 1993), then the bipolar flow speed is 1000km/sec. This velocity is extraordinarily high. The similar, and possibly related, C-rich bipolar nebula GL618 has a high velocity wind component with velocities reaching no more than 270km/sec (Cernicharo et al., 1989, assuming an inclination angle for the AFGL618 nebula of 45 ), despite having a hotter central star and more highly excited bipolar nebula. Another evolved star, OH231.8+4.2, has a bipolar outflow with a velocity of 140 km/sec (Morris et al. 1987, assuming an inclination angle for the OH231.8+4.2 nebula of 47 ). No known post-AGB star or PN exhibits an outflow with so high a veolcity. However, if the bipolar axis of AFGL2688 were inclined at a much larger angle to the plane of the sky than 5 , the deprojected bipolar flow speed becomes much smaller, and comparable to those of other nebulae.
Third, whilst Latter et al. suggest the E-W structure seen in their broadband K image is an equatorial torus, close examination of the observed K-band morphology reveals an inconsistency and suggests another possibility. If the bipolar axis is really inclined at an angle of 15 to the sky, and the equatorial torus lies in a plane orthogonal to the bipolar axis, as expected in the standard model, then the ratio of the semi-major to semi-minor diameters of the projected torus should be 3.9. If the inclination angle is actually 5 , as Latter et al.'s model suggests, then the ratio should be 11.5. The ratio we measure from the K image of Latter et al. is 2 demonstrating that the standard model is not consistent with the observations We have already mentioned the filaments which appear to link the N to E blobs and the W to S blobs, and we will suggest in the next section that these pairs of blobs are really physically connected, each a part of a single structure much larger than the individual blobs.
Fourth, the torus of H2 emission proposed in the standard model has an implausibly large radius of 5" ( 1017 cm at 1.2 kpc), which is inconsistent with other observations, and suggests a dynamical structure for the nebula that is unstable. If we believe the well substantiated suggestion that the H2 emission of the torus arises from the shock excited gas of the colliding fast post-AGB wind and slow AGB wind, then we would expect this excitation on the inner edge of the torus. However, all models of AFGL2688's emission strongly suggest an inner radius of 1" or less (e.g. Latter et al. 1993; Jura& Kroto 1990; this work), at least five times smaller than the observed H2 radius. If one maintains that the E-W H2 emission regions are the outer edges of a very large torus that is rotating at the 15 km/s velocity measured by Smith et al. (1990) then we find inconsistencies with other observations. According to the mm-wave observations of AFGL2688 by Truong-Bach et al. (1990) the circumstellar envelope is approximately spherically symmetric with a FWHM of 16" (resolution of 6"). Bieging & Rieu (1988) imaged AFGL2688 in the HCN line at 5" resolution and found a moderately flattened structure rotating at 1.2 km/s aligned with the dust lane; however, the angular resolution of those observations is insufficient to clearly resolve a toroidal structure leaving the interpretation of a torus uncertain. If the rotation is a result of mass loss induced by binary interaction, the angular momentum in Bieging & Rieu's torus rotating at 1.2km/sec is already implausibly large (they indicate that it is about two orders of magnitude larger than could be sustained by any binary system with components in the mass-range of AGB star progenitors): Latter et al.'s suggested 15km/sec rotation velocity and larger radius makes the problem almost two orders of magnitude worse. Neither can the toroidal rotation be induced by magnetic corotation of the wind and central star. If we follow the argument presented by Bieging & Rieu (1988), the stellar magnetic field strength necessary to generate a 15km/sec rotation velocity at 1017 cm is 106 G. There is no known way to sustain a torus of such dimensions rotating so rapidly. Additionally, a structure of this size rotating so rapidly should have been strikingly obvious in the observations by Bieging & Rieu (1988) or Truong-Bach et al. (1990). To explain the kinematics of the AFGL2688 nebula, we require some explanation other than a rotating torus.
Fifth, the scattering models presented by Latter et al. (1993) and Yusef-Zadeh, Morris & White (1984) for AFGL2688 differ with the data in two important ways, noted by Latter et al. (1993). First, at all wavelengths the model's intensity gradient outward along the bipolar lobes rises too steeply at the inner edges of the scattering lobes in the models when compared with the observed images. Second, as the wavelength increases, from I to K, the lobe intensity peaks of the AFGL2688 images move spatially closer together, while in the models of Latter et al. (1993) the intensity peaks remain almost fixed in position. Latter et al. (1993) suggest that the differences between the observations and the model arises because they used only single scattering models, while multiple scatterings may increase at shorter wavelengths. However, Yusef-Zadeh, Morris & White (1984) showed quite clearly that the effect of changing from single scattering to multiple scattering is like placing a diffusing screen in front of the nebula. The structure is simply blurred by the increased scattering, but the structure does not change. In summary, while individual images can be modelled moderately well with these scattering models, the detailed morphologies and the wavelength dependence are not matched in the case of AFGL2688.
Sixth, the standard model assumes that the bipolar outflow from AFGL2688 lies along an axis delineated by the N and S reflection lobes, at PA 15 . However, close examination of the images reveal that not only does the PA of the N and S lobes change with wavelength, but also the N and S lobes do not even lie on the same axis. We can determine the PA of the lobes from our IR images, either from the position of peak brightness or by determining the flux averaged center of brightness of the nebula along the RA direction defined as
Here x is the R.A. position in arcseconds, X is the center of brightness position in arcseconds, and the surface brightness at x. This function weights the center of brightness towards the region(s) of brightest emission. The RA center of brightness, plotted as a function of declination (DEC), then traces the axes of the lobes. In Fig. 7a-d for the J, 2.122 m (0.31" pixel scale), nbL and 10.0 m images, we plot the RA center of brightness position against DEC (solid line), and the R.A. position of the brightest point against DEC (dashed line). At each wavelength the results from the two methods are very similar. At J and 2.122 m both the N and S lobes apparently lie on an axis at PA 5 , but there is a significant E-W offset between the symmetry axes of the N and S lobes - 2" at J dropping to about 1" at nbL. At nbL the axis of the S lobe is tilted about 7 with respect to the N lobe. In all the mid-IR images the N lobe is oriented at PA 0 , and the S lobe at PA 210 , but the axes of the two lobes join at the centre of the nebula. In the scattered near-IR continuum the two lobes are oriented more or less identically, but are offset tangentially from one another, whilst in the mid-IR thermal emission the two lobes point from a common origin but in entirely different directions.
© European Southern Observatory (ESO) 1997
Online publication: March 24, 1998