Astron. Astrophys. 334, 969-975 (1998)

4. Discussion

New and surprising information resulted from `binary polarimetry' since we found a remarkable polarization for both binary components of the Z CMa system. Whitney et al. (1993) argued that only the (IR) primary is a significant component of polarized radiation that is intrinsic to Z CMa in the optical. In the K band, we found the secondary as a strong polarized source too.

Degree and orientation of the K band total linear polarization of Z CMa (, ) agree very well with optical measurements (in 'normal' stage: up to and ). Assuming dust scattering as polarization mechanism, the wavelength independence of the polarization orientation indicates that the 'light scattering geometry' is the same in the optical range as well as in the NIR. The high-velocity outflow emanating from the Z CMa system (Poetzel et al. 1989) at the position angle of is indicative for a geometry represented by bipolar cavities with walls of scattering dust particles. These cavities focus the light from the central source into the outflow directions while the direct light is blocked by an optically dense (disk-like) envelope. Model calculations show that such focusing configurations are able to produce large polarization degrees (Whitney & Hartmann 1993, Fischer et al. 1996).

In spite of the small amount of light which comes to us in the optical range from the primary in comparison to the secondary (at R: / 1/7, Barth et al. 1994), the primary (, ) dominates the total polarization. Whitney et al. (1993) explained the appearance of larger polarization in the emission lines than in the continuum by this dominance. Whereas the polarized continuum light of the primary is diluted by less polarized light of the secondary, the emission lines originate only from the primary and remain strongly polarized. In this context, Whitney et al. identified the primary as the Herbig Ae/Be object which gives us one face of the Z CMa system. In the K band, the primary contributes most of the light ( / 2.5). Now direct light from the Herbig Ae/Be star itself can dilute the polarized radiation. The orientation of the primary polarization gives motivation to identify this object as driving source of the outflow. High degrees of polarization can be expected for viewing angles (angle between line of sight and disk normal) nearly perpendicular to the outflow axis. Such a tilt is also supposed from the large extension of the outflow seen in projection. However, because of the large radial velocities observed in Z CMa's jet, some authors (e.g. Koresko et al. 1991) expected nearly no tilt.

From the assessment of the individual luminosities, Whitney et al. (1993) argued that the secondary component of the Z CMa system is the FU Orionis star. We found an unexpected high polarization for this object () which has to be discussed in connection with its nature - a luminous disk. Our reasoning for this interpretation comes from the following considerations. In case of dust scattering (the most likely polarization process) we need a thick flared disk to give the `accretion light' a strong anisotropy for the illumination of the circumstellar region. A thick disk, which is mainly observable by accretion luminosity was used as input for polarization simulation calculations (Fischer et al. 1994). The disk resulted from 2D hydrodynamical models of the protostellar collapse of a rotating 1 cloud. The resulting spatially unresolved polarization degrees show that a disk with a small `opening angle' is able to produce polarization degrees large in the NIR and small in the optical. This is possible for viewing angles , which still agrees with the demand that we see the secondary (optical primary) more pole-on than edge-on. Moreover, this geometry yields a quite large total linear polarization at 2.2 µm while the optical polarization remains small (a condition set by the results of the Whitney et al. (1993) investigations).

The orientation of the plane of the optically thick accretion disk is parallel to the polarization orientation (). Such a disk orientation requires us to think about changes in the position angle of the polarization observed in 1987/1988 (see Fig. 3). A possible explanation for the polarization rotations could be that they appear if the flux of the primary is reduced and the secondary polarization becomes significant for the total polarization. This idea is supported by the fact that the polarization rotation is connected with a decrease of the polarization degree. Since the components may vary independently of each other, any interpretation must solve the problem which component is responsible for variations in the polarization as well in the spectrum and the brightness.

Fig. 3. Polarization observations of Z CMa from different authors during the last 25 years. Note that measurements were made through apertures in diameter, i. e. polarisation data shown here are different `contaminated' due to the large-scale reflection nebula .

Our interpretation of the polarization of the Z CMa secondary in terms of a circumstellar disk requires the discussion of major implications. First of all, the angular separation of the binary corresponds to a minimum linear distance of about 100 AU. Thus, if both disks would be co-planar (see below), their average maximum size is restricted to this distance. A circumbinary disk does not explain our results since no differences in the polarization properties of the individual components would result (although is seems likely that the dust envelope of Z CMa may influence the overall polarization). Disks with diameters less than 100 AU are smaller than disks for which the size has been measured directly (e.g in the Orion Trapezium Cluster, McCaughrean & O'Dell 1996) or inferred by other measurements (Beckwith & Sargent 1996). However, the gravitational interaction of the binary components will reduce the extent of the circumstellar disks and lead to their truncation by gravitational torques and resonances (Artymowicz & Lubow 1994). Evidence for this effect has been gained since binary systems tend to have smaller disks compared those of single stars (Jensen et al. 1994, Jensen & Mathieu 1997). Thus, this mechanism could explain the minimum sizes which might be valid for the Z CMa circumstellar disks.

Of course, the true linear separation of the components can be much larger reducing the necessity of relatively small disks (and the interaction efficiency of the binary system as well).

To strengthen our views on the Z CMa system, we also looked at the results for distance and position angle at different analyzer orientations. We expected a smaller position angle but larger distance for analyzer orientations around the polarization orientation of the primary because for this embedded object the scattering light patch is more shifted from the unpolarized source position in the direction of the blue shifted outflow than in the case of the secondary. Indeed, the values listed below suggest such a behaviour. Statistical tests of the linear variation of both distance and position angle in dependence on the analyzer position suggest that the suspected trend cannot be ruled out at the 90 per cent significance level.

Jain & Bhatt (1995) found most of their observed Herbig Ae/Be stars to be polarimetrically variable. From Z CMa (one object of their list), we know that such characteristics can be `complicated' by the existence of other objects in the vicinity of the young stars. Our high angular resolution polarimetry allows us to attach the observed features to their origins.

The overall illumination of the vicinity of Z CMa seems to be also affected by the newly discovered PMS sources (Nakajima & Golimowski 1995). Large scale polarization maps taken by the help of coronographic techniques should bring new insights to this.

Which conclusions can be drawn from our results on the formation history of the Z CMa binary? At first glance, the polarization orientations of both components are marginally different (at the level). However, keeping in mind that the position angle of the polarization indicates the orientation of the projected disk normal onto the plane of the sky, we have also to constrain the viewing angles in order to decide whether the disks are aligned or not. As discussed above, the explanation of the large polarisation degree of the secondary requires viewing angles . On the other hand, a circumstellar disk seen pole-on has no net polarization. So, we consider a possible range of viewing angles of ... . The large radial velocities of the Herbig-Haro objects associated with the jet as well as the small displacement of the blue- and red-shifted CO molecular line peaks suggest that the disk of the primary is not seen edge-on but inclined with respect to the line of sight. The axes ratio of the dust disk reported by Malbet et al. (1993) points to a value of (however, no evidence for this feature was found by Tessier et al. 1994). The best estimate for the viewing angle can be obtained by relating the highest absolute radial velocity of the Herbig-Haro objects (620 km/s) to the absolute upper limit of the velocities observed in the P Cygni line profiles (1000 km/s, Finkenzeller & Mundt 1984). This yields a viewing angle of degrees with the possible variation corresponding to a change of 100 km/s in velocity. Thus, we conclude that the space orientation of the disks of the Z CMa binary system are only marginally different. A graphical representation of the range of possible disk orientations is displayed in Fig. 4. Here, the line of sight corresponds to the -z direction.

Fig. 4. Cones indicating the 1 boundaries of the space orientation for the normal of the circumstellar disks of the Z CMa primary (at coordinate origin) and the secondary (units are in milliseconds of arc - mas). The ellipses in the plane of the sky illustrate the polarisation degree independence on the position angle.

Some caution is required in interpreting these results. Although co-planar disks would strongly support the view that both components had a common origin from the same molecular cloud fragment, we cannot firmly establish the co-planarity from our measurements. Another possibility to explain the observed polarization behaviour is the individual formation of the stars from fragments that were aligned by roughly parallel magnetic field lines but with some longitudinal displacement with respect to the field.

© European Southern Observatory (ESO) 1998

Online publication: June 2, 1998

helpdesk.link@springer.de