4. Spectrum of calm emission
We first extracted a spectrum for the average emission over the whole NFI observation. The spectral channels were rebinned so as to sample the spectral full-width at half-maximum resolution by three bins and to accumulate at least 20 photons per bin. The bandpasses were limited to 0.4-4.0 keV (LECS), 1.8-10.5 keV (MECS) and 15-120 keV (PDS) to avoid photon energies where either the spectral calibration of the instruments is not complete or no flux was measured above the statistical noise. We tried to model the spectrum with various descriptions. In all models, an allowance was made to leave free - within reasonable limits - the relative normalization of the spectra from LECS and PDS to that of the MECS spectrum, to accommodate cross-calibration uncertainties in this respect. Publicly available instrument response functions and software were used (version November 1998).
The continuum could best be fitted ( for 96 dof) with a Comptonization model (Titarchuk 1994) plus black body radiation, see Table 1. Next to that there is a strong emission line at 7 keV. A fit with a single narrow line results in a centroid energy of 6.85 keV. We identify this as K fluorescence in strongly ionized iron. The centroid energy is between the expected Fe-K lines for helium-like (6.68 keV) and hydrogen-like Fe (6.96 keV). We included in the model narrow lines at these fixed energies. The best-fit parameter values are given in Table 1 and a graph of the spectrum and the model fit in Fig. 4. The fits with other continuum models, in combination with a black body and two narrow line components results in fit qualities of for 98 dof (thermal bremsstrahlung), for 96 dof (broken power law) and for 96 dof (power law with high-energy cut off).
Table 1. Best-fit parameter values of the Comptonized model to the NFI spectrum. The last line specifies the value for the fit without a bb (black body) component. This value applies after re-fitting the remaining parameters. EW means equivalent width.
Terada et al. (1999) reported about the 1996 transient AX J1842.8-0423 which exhibited an Fe-K line at 6.80 keV with a large equivalent width of 4 keV. The 0.5-10 keV continuum plus line spectrum was successfully fitted with a thin hot thermal plasma emission model of temperature 5.1 keV. We fitted such a model according to the MEKAL code implementation (Mewe et al. 1995). The fit was reasonable, provided two additional components were included. With a power law and black body as additional components, (98 dof), which is a worse fit than the Comptonization model in Table 1. The resulting plasma temperature is keV. The fitted contribution of the thin plasma to the flux is of order 10%. The emission measure of the thermal plasma is cm-3.
E(B-V)=+0.24 (Wagner 1999) implies cm-2 according to the relationship defined by Predehl & Schmitt (1995). If we assume that the uncertainty in E(B-V) is 0.10, where most of the uncertainty comes from the uncertainty in the calibration of the relationship used by Wagner (1999) between the equivalent width of the 578.0 nm interstellar absorption line to E(B-V) (see Herbig 1975), then the error in is 0.05 cm-2. If we fix to cm-2 and leave free the remaining parameters of the Comptonized model in Table 1, is 1.31 (97 dof). We conclude that as determined from the X-ray spectrum is consistent with E(B-V)=+0.24.
To determine whether the variability as illustrated in Fig. 3 is accompanied by strong spectral changes, we extracted a spectrum for times when the source was relatively faint and one for times when the source was relatively bright. These times are indicated by hatched areas in Fig. 3. Subsequently, we employed the same Comptonized model as for the whole observation, leaving free only the normalizations of the different contributions. The resulting values for are 1.36 for the bright data (104 dof) and 1.22 (103 dof) for the faint data. The 44% difference in the 0.4-10 keV flux between the faint and bright data is due to approximately equal changes in blackbody and Comptonized components (i.e., 25 and 19% respectively). The flux of the emission lines scales with the integral flux: the combined equivalent widths of both lines is identical in both cases at 0.26 keV. This spectral behavior is illustrated in Fig. 5 which zooms in on the MECS part of the spectrum (including the emission lines) for the two extremes. In Fig. 6 the ratio between both spectra is presented. This is consistent with a constant ratio of throughout the spectrum ( for 65 dof). The apparent bump between 4 and 5 keV is statistically not significant.
© European Southern Observatory (ESO) 2000
Online publication: June 5, 2000