This paper extends results presented in Paper II on anisotropic resonance line scattering profiles to show the polarized emission from such scattering. The main results are that (a) as in Paper II, expanding disks yield line profiles that are asymmetric about line center, whereas those from rotating disks are symmetric, (b) the integrated polarized flux across a resonance line profile will generally be non-zero, so that interesting information about the disk can be derived even from low resolution (i.e., narrow band pass) spectropolarimetry, (c) both and Stokes profiles will be observed in rotating disks, whereas for expanding disks, the Stokes component will be zero (under the assumptions of axisymmetry) for a suitable orientation of the observer's measurement axes.
Although these results are quite instructive and may have application to certain restricted cases, numerical simulations for more realistic situations are needed. Of key importance is dropping the assumption of optically thin profiles. To do so will require more sophisticated radiative transfer techniques, as for example the Sobolev method for polarized radiation transport by Jeffery (1990) or techniques based on Monte Carlo simulations.
Since resonance line scattering in its phase scattering properties is so similar to that of Thomson scattering, already one might "guess" at some of the expected effects from scattering in optically thick disks. One naturally expects that at sufficiently large optical depths, multiple scattering will tend to destroy the line polarization. However, Wood et al. (1996) have shown that for Thomson scattering in equatorial disks, the continuum polarization actually increases when the disk moves from optically thin to optically thick, and then decreases at larger optical depths. A similar effect should likewise be observed for the polarization from resonance line scattering.
An advantage of resonance lines over Thomson scattering is that the polarized emission is spread out across the line profile according to the isovelocity topology, which affords a better opportunity of probing the disk geometry and velocity field. One difficulty in relation to "hot" disks where a substantial fraction of the gas is ionized is that Wood & Bjorkman (1995) have shown that electron scattering will greatly "smear out" the polarized line emission. They were not specifically modelling the effect for resonance line scattering but rather a continuum absorption line. Nonetheless, scattering of polarized resonance line radiation by hot thermal electrons can severely affect the profiles. But note that Thomson scattering has a quite low cross-section of cm-2, whereas resonance lines have cross-sections that can be larger by several orders of magnitude. Hence in disks of relatively low density, the effects of electron scattering may be entirely negligible, yet the polarized emission in suitably chosen lines should be within the realm of current detection thresholds. For example, dedicated monitoring of the polarizations in several Be stars at the Pine Bluff Observatory (Univ. Wisconsin) is easily capable of detecting variations at the 0.1% level (Bjorkman et al. 1997). In general, since resonance line scattering is similar to Thomson scattering in its angular distribution, a given aspherical envelope will yield a similar overall polarization level from line scattering as for electron scattering, the main difference being the scale of the line optical depth as compared to that of the free electrons.
Even if thermal smearing by electron scattering can be ignored, there is the issue of separating the line polarization from that of the continuum. Fortunately, many strong doublets of Li-like atoms that are commonly observed in winds (e.g., CIV 1550) have one component (shorter wavelength) with , but the other (longer wavelength) with , the latter producing no polarization from resonance scattering. Thus the Q and U values at the longer wavelength line can be used to disentangle the continuum polarization from that of the line scattering that does contribute at the shorter wavelength component. Indeed in this way, one can correct not only for the electron scattering polarization by the envelope, but also for the ISM contribution. Optical depth effects and the influence of electron scattering are the kinds of issues that must be addressed quantitatively in future studies if resonance line scattering polarization is to be of practical diagnostic value in observations of circumstellar disks.
© European Southern Observatory (ESO) 2000
Online publication: December 5, 2000