Astron. Astrophys. 317, L47-L50 (1997)

3. Results

Firstly, we consider simple models. In the case of the PSPC non-dip spectra of X 1755-338, a simple absorbed power law model gives an acceptable fit, with however a value of = 1.9 0.1. In our previous work on the Exosat ME data, we found that a two-component model was necessary to describe the source, consisting of a blackbody with of 0.88 keV, and a power law with photon index of 2.67. Having re-examined the ME data we find that a value of 1.9 can be rejected for a single spectrum with certainty greater than 93%, and in our Exosat work, we were able to fit the complete observation divided into a sequence of 314 spectra with values between 2.6 and 2.8. Thus the simple power law fit is not acceptable. It can however be used to test whether the low energy cut-off (LECO) of the spectrum changed between non-dip and dip data (minimising model dependency). There was no detectable change in , but dipping could be seen as a decrease in the spectral flux in the higher part of the PSPC band (0.5 - 2 keV).

Use of a two-component model is clearly indicated. It is shown below (see Fig. 2) that the LECO is determined by the power law, and so the blackbody can not have a smaller than the power law, and a model of the form (where AB, BB and PL are the absorption, blackbody and power law terms) should be used. Very good fits were obtained for both non-dip and dip data using this model, showing that the PSPC data are fully consistent with the two-component model. If and are allowed to be free, low values of are again obtained 1.5, inconsistent with the values we obtained from the ME data. The Rosat PSPC is not able to constrain power law index well in the presence of a second component, or for a blackbody peaking above the PSPC band. However the PSPC data can be used in conjunction with parameter values already established for a source to determine the lower energy part of the spectrum. It is clearly more sensible to fix both and at the ME values.

Results are shown in Fig. 2 (plotted with primitive channels, see below). In Fig. 3, the spectra analysed with 24 channels are shown, plotted together with typical Exosat ME spectra. Again, there was little or no change in the LECO between non-dip and dip spectra. The energy range of the Rosat PSPC is ideal for accurate determination of the cut-off, and for the power law, which can be seen to determine the cut-off, did not change within the errors. The lack of change in the LECO is model-independent and is primary proof that dipping is absent at low energies, since the non-dip and dip spectra have to converge towards the cut-off. Moreover, it is clear from Fig. 2 that dipping occurs preferentially at higher energies in the PSPC band, and is seen to be due to increased absorption of the blackbody. At 1 keV the non-dip and dip spectra exhibit the energy-independence well known from Exosat. The blackbody contribution (dashed line) shows large changes in indicating that dipping is primarily due to absorption of this component. Thus below 0.5 keV, the contribution of the blackbody is very small and so the extent of dipping is markedly reduced compared with the 0.5 - 2.0 keV band.

Typical spectral fitting results for non-dip data have for the power law = , normalisation I (at 1 keV) = photons , for the blackbody = H atom and I = photons (90% confidence errors). There was very good consistency of the results from different sections of data with only small variations of parameter values, eg varied by about 0.016 photons . The corresponding blackbody normalisation was non-zero at the level, showing clearly the necessity for this component in fitting the spectra of this source. The additional absorption term for the blackbody was in all cases close to zero, showing that that during persistent emission both spectral components are subject to the same absorption. The dip data were fitted by the same model with the normalisations fixed at non-dip values. Allowing the dips to be modelled by normalisation changes would be to assume that some change in the emission processes takes place in the source coincident with the absorption during dipping, which can be discounted as unphysical. For dip data, the total for the blackbody increased to H atom , with for the power law equal to H atom .

Fig. 2. Spectral fits to non-dip and to dip emission. The solid lines show the total fit, and the dotted line and dashed lines show the power law and blackbody components respectively.

Fig. 2 shows that below 0.5 keV the blackbody is very small and so implies that below 0.5 keV dipping should be effectively absent since in this source, it appears that absorption of the power law in dipping is very small. To demonstrate this we have obtained the light curve in the band 0.1 - 0.5 keV shown in Fig. 1b. Although the count rate is of course low in this band, there is little or no sign of dipping, and this is confirmed by examining the mean count rates in all of the sections of data in this figure. In the band 0.5 to 2.0 keV, the decrease is 3.6 c/s, ie % which in terms of the standard deviation of the mean of non-dip means 0.49 c/s is 7.2 , and so is highly significant. for the power law did not change in dipping within the errors of fitting, and we estimate possible residual dipping in the band 0.1 - 0.5 keV as % in count rate, as indicated by the bar in Fig. 1b at the position of dipping (which is 3% deep). There may be residual dipping at a low level; however the data of Fig. 1b do not constitute a significant detection of this.

From the results, absorbed photon fluxes of the spectral components were calculated. In the band 0.1 - 2.0 keV, the blackbody comprises 17% of the total, whereas in the band 1.0 - 10.0 keV it is 38% of the total. This is in good agreement with a typical blackbody percentage of 39% taken from our previous Exosat ME work. The lower percentage in the PSPC band corresponds to a blackbody contribution to the count rate of 16% in the band 0.1 - 2.0 keV. Thus since the depth of dipping in the Exosat observation indicated only partial absorption of the blackbody, a level of dipping 10% in the PSPC band might be expected, as is seen.

Fig. 3. Comparison of the Rosat PSPC spectra with typical Exosat ME results for non-dip and dip data.

© European Southern Observatory (ESO) 1997

helpdesk.link@springer.de