2011ApJ...730..131F

In the magnetar model, the quiescent non-thermal soft X-ray emission from anomalous X-ray pulsars and soft gamma repeaters is thought to arise from resonant Comptonization of thermal photons by charges moving in a twisted magnetosphere. Robust inference of physical quantities from observations is difficult, because the process depends strongly on geometry, and current understanding of the magnetosphere is not very deep. The polarization of soft X-ray photons is an independent source of information, and its magnetospheric imprint remains only partially explored. In this paper, we calculate how resonant cyclotron scattering would modify the observed polarization signal relative to the surface emission, using a multidimensional Monte Carlo radiative transfer code that accounts for the gradual coupling of polarization eigenmodes as photons leave the magnetosphere. We employ a globally twisted, self-similar, force-free magnetosphere with a power-law momentum distribution, assume a blackbody spectrum for the seed photons, account for general relativistic light deflection close to the star, and assume that vacuum polarization dominates the dielectric properties of the magnetosphere. The latter is a good approximation if the pair multiplicity is not much larger than unity. Phase-averaged polarimetry is able to provide a clear signature of the magnetospheric reprocessing of thermal photons and to constrain mechanisms generating the thermal emission. Phase-resolved polarimetry, in addition, can characterize the spatial extent and magnitude of the magnetospheric twist angle at ∼100 stellar radii, and discern between uni- or bidirectional particle energy distributions, almost independently of every other parameter in the system. We discuss prospects for detectability with the Gravity and Extreme Magnetism (GEMS) mission.