Whether the observed polarimetric features of pulsar radio emission are intrinsic properties of the emission mechanism or if they are a result of propagational effects is one of the open questions for the understanding of the polarization of radio pulsars. Early observations of the `S'-type swing of the polarization position angle through the pulse profile of the Vela pulsar led to an geometrical interpretation by Radhakrishnan & Cooke (1969). In their model the swing reflects the geometry of the magnetic field lines projected on the plane of the sky. In some cases the expected swing is matched so well that a non-geometrical origin for the position angle seems rather improbable. Ruderman & Sutherland (1975) gave an explanation for this phenomenon as an intrinsic property of emission by coherent curvature radiation. The particles experience an acceleration perpendicular to their motion in the plane of curvature of the field line. The resulting radiation has its electric field vector within this plane, thus producing the geometrical signature.
Alternatively, it was proposed that the observed polarization originates from propagational effects or is at least influenced by them. Various authors calculated the properties of the propagational modes in a pulsar magnetosphere (e.g. Allan & Melrose 1982, Barnard & Arons 1986, Lyutikov 1998 and references therein). It is widely agreed among those authors, that two independent, generally elliptical polarization modes propagate, which are oriented parallel and perpendicular to the plane of curvature of the magnetic field line. The observed polarization then corresponds to the shape of these modes at the distance from the neutron star, where the radiation decouples from the plasma. This distance is called the polarization limiting radius (hereafter PLR) and does not necessarily coincide with the place of emission. The magnitude of the PLR has been a subject of theoretical debates. Barnard (1986) and Beskin et al. (1993) place the PLR at the cyclotron resonance. Melrose (1979) defines a coupling ratio where denotes the difference in wavenumber of the two modes and L is a characteristic length scale. The PLR is at the position where .
In a previous paper (von Hoensbroech et al. 1998b) we have shown that many of the complex variety of radio pulsar polarization states can be understood qualitatively if some propagational effects in the pulsar magnetosphere are taken into account. This was done by using a simple approximation for the properties of natural polarization modes as they propagate through the magnetosphere. The model is based on the following assumptions: 1. The temperature of the background plasma is assumed to be zero (distribution function , hereafter is the -factor of the background plasma). 2. Furthermore no strong pair production in a sense, that the -plasma is not approximately neutral, is assumed. Considering all relevant angles under which the propagating wave and the magnetic field lines intersect, the shape of the natural polarization modes can be calculated at any given point between the emission height and the light cylinder radius (hereafter , ). 3. Assuming a PLR and a certain Lorentz-factor for the background plasma, the following qualitative statements can be made: a. The polarization modes propagate independently with their position angle parallel and orthogonal to the local plane of field line curvature, b. the radiation depolarizes towards high frequencies, c. the degree of polarization correlates with the pulsars loss of rotational energy and d. the polarization smoothly changes from linear to circular with increasing frequency.
This last point was met by recent observations at the relatively high radio frequencies of GHz (von Hoensbroech et al. 1998a) as it was already suspected by early measurements of Morris et al. (1981). For a few pulsars the degree of circular polarization increases strongly with frequency. For three of those objects, the circular polarization reaches values of more than 50%, even surpassing the linear polarization. This rather unusual behavior appears smooth and is thus likely to be systematic, as an inspection of lower frequency data showed. We therefore use the data of those three clear examples to quantitatively compare the rate of change from linear to circular to the predicted one in statement d .
© European Southern Observatory (ESO) 1999
Online publication: February 23, 1999