2006ApJ...650.1082D

The X-ray spectra of anomalous X-ray pulsars have long been fit by smooth, empirical models such as the sum of a blackbody plus a power law. These reproduce the ∼0.5-10 keV range well, but fail at lower and higher energies, grossly overpredicting the optical and underpredicting the hard X-ray emission. A poorly constrained source of uncertainty in determining the true, intrinsic spectra, in particular at lower energies, is the amount of interstellar extinction. In previous studies, extinction column densities with small statistical errors were derived as part of the fits of the spectra to simple continuum models. Different choices of model, however, each produced statistically acceptable fits but a wide range of columns. Here we attempt to measure the interstellar extinction in a model-independent way, using individual absorption edges of the elements O, Fe, Ne, Mg, and Si in X-ray grating spectra taken with XMM-Newton. We find that our inferred equivalent hydrogen column density N_Hfor 4U 0142+61 is a factor of 1.4 lower than the typically quoted value from blackbody plus power-law fits, and is now consistent with estimates based on the dust scattering halo and visual extinction. For three other sources, we find column densities consistent with earlier estimates. We use our measurements to recover the intrinsic spectra of the AXPs empirically, without making assumptions on what the intrinsic spectral shapes ought to be. We find that the power-law components that dominate at higher energies do not extend below the thermal peak.