2005A&A...439..287V

We report XMM-Newton observations of the intermediate polar (IP) GK Per on the rise to the 2002 outburst and compare them to Chandra observations during quiescence. The asymmetric spin light curve implies an asymmetric shape of a semi-transparent accretion curtain and we propose a model for its shape. A low FeXVII (λλ15.01/15.26Å) line flux ratio confirms the need for an asymmetric geometry and significant effects of resonant line scattering. Medium resolution PN spectra in outburst and ACIS-S spectra in quiescence can both be fitted with a leaky absorber model for the post shock hard X-ray emission, a black body (outburst) for the thermalized X-ray emission from the white dwarf and an optically thin spectrum. The difference in the leaky absorber emission between high and low spin as well as quasi-periodic oscillation (QPO) or flares states can be fully explained by a variation in the absorbing column density. For the explanation of the difference between outburst and quiescence a combination of the variation of the column density and the electron and ion densities is necessary. The Fe fluorescence at 6.4keV with an equivalent width of 447eV and a possible Compton scattering contribution in the red wing of the line is not significantly variable during spin cycle or on QPO periods, i.e. a significant portion of the line originates in the wide accretion curtains. High-resolution RGS spectra reveal a number of emission lines from H-like and He-like elements. The lines are broader than the instrumental response with a roughly constant velocity dispersion for different lines, indicating identical origin. He-like emission lines are used to give values for the electron densities of logn_e∼12. We do not detect any variation in the emission lines during the spin cycle, implying that the lines are not noticeably obscured or absorbed. We conclude that they originate in the accretion curtains and that accretion might take place from all azimuths.