Astron. Astrophys. 339, 327-336 (1998)

5. ROSAT PSPC observation

5.1. ROSAT data analysis

In order to cope with the lack of information at energies lower than , we have analyzed a still unpublished ROSAT PSPC observation of the same target. 1H0419-577 was observed by ROSAT on 1992 April 7, for a total exposure time of 4094 seconds. Source spectra have been extracted from a circular region of 2'45" radius around the centroid of the source, while background spectra have been extracted from an annulus of internal and external radii 6'20" and 8'30" respectively. The radii have been chosen in order to avoid six serendipitous sources that have been detected in the field of view. To perform spectral fitting, an appropriate PSPC-B redistribution matrix for the observation date (pspcb_gain2.rmf ) has been used and the effective area calculated with the software tool pcarf included in FTOOLS 3.6 . Spectral PI channels lower than 11 ( keV) and higher than 32 ( keV) in the rebinned spectrum have been excluded from the spectral analysis. The resulting net count rate is  s^-1; background contributes less than 1/30 for each spectral channel.

The broadband 0.1-2.4 keV light curve shows a slight (i.e. ) tendency to increase with time. We have investigated whether such a trend could be associated to spectral variability by studying the light curves in the 0.1-0.5 keV and 0.5-2.0 keV energy bands (see Fig. 4). The suggestion of a wider dynamical range of the soft light curve variation is only marginal. In the following we will therefore focus on the average spectral behavior only.

Fig. 4. PSPC 1H0419-577 light curves in the 0.1-0.5 keV (upper panel ) and 0.5-2.0 keV (central panel ) bands, and their hardness ratio (lower panel ). Binning time is  s. The panels are displayed in the same relative ordinate scale to allow direct comparison of the dynamical ranges.

A simple power-law with photoelectric absorption yields a formally acceptable fit (). and the 1 keV normalization is  photons cm^-2 s^-1. The residuals show however a regular wavy trend (see Fig. 5). This trend cannot be removed with a linear adjustment of the instrumental gain, as achievable with the command gain in XSPEC . We checked therefore the statistical significance of spectral complexity. An absorption edge (with energy broadly consistent with the photoionization threshold energy for O VII ), a thermal optically thin or thick emission or a break in the power-law shape yield a comparable improvement of the ( for two further degrees of freedom, corresponding to a confidence level 99.97-99.99%). There is therefore a suggestion of soft X-ray spectral complexity, with an ultrasoft component emerging below , or, alternatively, an absorption edge due to highly ionized interposed matter.

Fig. 5. Spectrum (upper panel ) and residuals (lower panel ) in units of standard deviations when a simple absorbed power-law model is applied to the ROSAT PSPC data of 1H0419-577.

In some fits, the best-fit value is significantly lower than the Galactic in the direction of 1H0419-577 ( cm^-2), however not inconsistent with it if we take into account a conservative estimate of the uncertainties on the nominal Galactic values of 20-30% (i.e. :  cm^-2, Dickey & Lockman 1990). An investigation of the spectral complexity in the soft X-ray spectrum would require much better spectroscopic quality and/or simultaneous monitoring of the soft and hard X-ray. However, we stress that the ROSAT/PSPC spectrum is always much steeper than the MECS one , regardless of the specific model adopted. (see Sect. 4) ranges from 0.8 to 1.3.

We investigated also whether we could get comparably good fits if an underlying MECS-like power-law is assumed. We than fitted the PSPC spectrum with two-components models, where one of the components is a power-law with photon index held fixed at 1.55 and free normalization. The only case for which an acceptable is obtained is the double power-law model, which yields:  d.o.f., and  cm^-2. It is better than the single power-law model at the 99.6% level of confidence (cfr. first row in Table 3). The ratio between the "hard" power-law normalization as measured by the MECS and PSPC is . On the other hand, adding a MECS-like power-law to the other models listed in Table 3 yields a negligible (i.e. : ) improvement of the quality of the fit. In particular, a double blackbody + power-law model with photon index held fixed to the high-energy measured value - which have often been used to approximate the superposition of a high-energy non-thermal continuum and a disc emission - is not a viable model ( d.o.f.).

Table 3. Best-fit results of the ROSAT PSPC spectral analysis. PO=power-law, ED=absorption edge, BKNPO=broken power-law, BB=blackbody, BREMS=thermal bremsstrahlung.

5.2. ROSAT PSPC/Beppo-SAX MECS comparison

Beppo-SAX observed 1H0419-577 about four years later than ROSAT. The source was caught with very different spectral properties; namely the ROSAT PSPC 1992 spectrum is much steeper than the 1996 Beppo-SAX MECS one.

Recently, several authors have pointed out that the spectral indices as measured by the PSPC are systematically steeper than measured by other missions with a sensitivity bandpass extending at higher energies. The amount of such a difference has been estimated between 0.2 (Turner 1993, Fiore et al. 1994) and 0.4 (Iwasawa et al. 1998). It cannot therefore account completely for the observed difference. Moreover, If we fit the PSPC data with a power-law of index and a blackbody component, the turns out to be unacceptably high (439/16 d.o.f.). However, a double power-law model with one of the photon index held fixed to 2.0 yields /16 d.o.f., with  cm^-2 and .

The energy interval where the two PSPC and the MECS are well calibrated and overlap (1.8-2.0 keV) is too narrow to allow a direct comparison of the spectra via standard fitting techniques.

In order to partly overcome such problems, we studied the ratio between the PSPC and MECS 1H0419-577 spectra and the corresponding spectra of the featureless radio-loud quasar 3C273, whose X-ray spectrum is known to be well represented by a simple power-law in the whole 1-200 keV range (Grandi et al. 1997). The usage of the ratio washes out in principle any problem regarding instrumental calibration and the - until now not deeply studied - relative calibration between ROSAT and Beppo-SAX instruments. Moreover, it allows the use of an extended energy range of both instruments, where the response matrices are currently not well calibrated yet, but where a significant amount of counts still exists. A PSPC 3C273 spectrum has been extracted from a pointing observation included in the public archive. The spectral ratios are shown in Fig. 6. MECS spectrum remains pretty flat till  keV, being inconsistent with the PSPC one also in the extended energy range, where the PSPC flux is times higher. Such a difference cannot be explained in terms of spectral pivoting; in such a case the pivot should lie at  keV, while we found no evidence of spectral steepening in the Beppo-SAX data.

Fig. 6. Ratio of the PSPC (open circles ) and MECS (filled circles spectra with the relevant 3C273 spectra. Each point corresponds to a S/N

In Fig. 7 the X-ray SED of the PSPC (0.1-2.4 keV) and MECS (1,5-10.5 keV) data is plotted, together with one ASCA (0.5-10 keV) observation performed during summer 1996 (a complete analysis of the ASCA data is included in Turner et al. 1998). ASCA spectrum is consistent with and rather flat () power-law in the whole 0.7-10 keV band and a soft excess at lower energies, although with a much lower flux than the PSPC spectrum (Turner et al. 1998). The 0.7-2 keV spectrum had undergone a transition from a steep () to a flat () state. Two possible explanations are viable to account for the observed behavior. 1H0419-577 could display a remarkably variable soft excess. Alternatively, ROSAT and Beppo-SAX could have monitored two different states of the nuclear primary continuum source.

Fig. 7. X-ray SED ( in units of erg cm-2 s-1 Hz-1) of the ROSAT/PSPC, Beppo-SAX/MECS and 1996 July ASCA/SIS0 observations

© European Southern Observatory (ESO) 1998

Online publication: October 21, 1998
helpdesk.link@springer.de