3. Pulse phase resolved spectroscopy

The data were divided into twenty pulse-phase resolved bins. The minimum number of counts per bin for dataset 1 was 2.6 x 10⁴ which ensured adequate counting statistics. It was found that the column density, N_H, was not well constrained by either the GSPC or the ME. Thus in the pulse-phase resolved fits N_H was fixed at the best-fit value from the fits to the pulse-phase averaged data. Takalo et al. (1990) do not state which continuum parameters (if any) they held fixed in their pulse-phase resolved fits. In each case the iron line energy has no significant variation with pulse phase. Takalo et al. (1990) also report that the iron line energy is constant with pulse phase. Thus the iron line energy was fixed at the phase-average value. Takalo et al. (1990) fixed the iron line width at 0.1 keV for pulse-phase resolved spectroscopy. In the present analysis it is clear from the pulse-phase average fits that the iron line has a finite physical width and it is not appropriate to fit it with a narrow Gaussian. Thus the iron line physical width was fixed at the pulse-phase average value as was the cyclotron line width.

The pulse-phase dependence of the fitted parameters is shown in Figures 3.3, 3.3, and 3.3. The pulse-phase dependence of the iron line intensity is consistent with the results of Takalo et al. (1990).

Figure: Variation of spectral parameters with pulse phase. The data from dataset 1 were fitted to four different models. Clockwise from upper left: power law with interstellar absorption and an exponential cut-off, black body and thermal bremsstrahlung, non-relativistic Comptonization model, power law with cyclotron absorption. The iron line energy and width and the column density were fixed at the pulse-phase average values.

Figure: Variation of spectral parameters with pulse phase. The data from dataset 3 were fitted to four different models. Clockwise from upper left: power law with interstellar absorption and an exponential cut-off, black body and thermal bremsstrahlung, non-relativistic Comptonization model, power law with cyclotron absorption. The iron line energy and width and the column density were fixed at the pulse-phase average values.

Figure: Variation of spectral parameters with pulse phase. The data from dataset 4 were fitted to four different models. Clockwise from upper left: power law with interstellar absorption and an exponential cut-off, black body and thermal bremsstrahlung, non-relativistic Comptonization model, power law with cyclotron absorption. The iron line energy and width and the column density were fixed at the pulse-phase average values.