Astron. Astrophys. 342, 57-68 (1999)

2. X-ray spectral analysis

2.1. ROSAT PSPC observations

3C 219 has been observed with the ROSAT PSPC (Pfeffermann et al. 1986) on May 3, 1992.

We have analyzed the archive data finding that the spatial profile in the 0.1-2.4 keV range is consistent with that of a point source convolved with the PSPC point spread function (PSF), taking into account the source spectral properties and the background level. However, the maximum likelihood detection algorithm suggests some possible evidence of extended emission (see Sect. 3). There is no evidence of significant time variability.

The source spectrum was extracted from a circular region with radius. Different background spectra have been extracted either from annuli centered on the source or from circular regions uncontaminated by nearby sources. In all cases the background appears stable without any appreciable variations within the statistical errors. Corrections were included for vignetting and PSPC dead time. Source spectra were extracted in the pulse invariant (PI) channels in the range 11-240 ( 0.1-2.4 keV). The photon event files were analysed using the EXSAS/MIDAS software (version 94NOV, Zimmermann et al. 1993) and the extracted spectra were analysed using version 9.0 of XSPEC (Shafer et al. 1991) with the appropriate response matrix. The resulting exposure time is 4349 s and the background subtracted source count rate is 0.127 0.006 cts s^-1. The source spectrum was rebinned in order to obtain a significant signal to noise ratio () for each bin and fitted either with a power law or with a thermal plasma spectrum (Raymond-Smith model) plus absorption arising from cold material with solar abundance (Morrison & McCammon 1983). The derived spectral parameters are given in Table 1, where the reported errors are at 90% confidence level (Lampton et al. 1976); the value of the Galactic column density towards 3C 219, cm^-2, has been retrieved from the 21 cm radio survey of Dickey & Lockman (1990). A single power law with the absorption either fixed at the Galactic value, or free to vary, provides an acceptable description of the observed spectrum (Table 1). Formally, also a Raymond-Smith thermal model with solar abundance fits well the observed counts, however the derived temperature is extremely high and almost unconstrained.

Table 1. Spectral fits with ROSAT and ASCA data (f indicates a frozen parameter)

The best fit spectral photon index is unusually flat in agreement with the results of Prieto (1996). The flat power law slope can be due to the effect of absorption on a steeper continuum. In order to test this possibility we have tried a partial covering model. Assuming a typical AGN spectrum with photon index =1.8 plus Galactic absorption, a good description of the data can be obtained if about 75% of the nuclear radiation is absorbed by a column density of the order of 2 cm^-2 (Table 1). The X-ray flux in the 0.1-2.4 keV energy range, corrected for Galactic absorption, is 1.8 10^-12 ergs cm^-2 s^{- 1} corresponding to an isotropic luminosity of 1.1 10⁴⁴ ergs s^-1 (rest frame), while the unabsorbed nuclear luminosity would be erg s^-1.

2.2. ASCA observations

3C 219 was observed with ASCA (Tanaka, et al. 1994) on April 12, 1995 and on November 13, 1994 with the Gas Imaging Spectrometers (GIS2/GIS3) and with the Solid-state Imaging Spectrometers (SIS0/SIS1). Both observations were analyzed by us using standard calibration and data reduction methods (FTOOLS) provided by the ASCA Guest Observer Facility at Goddard Space Flight Center. The net exposure time for the 1995 observation was 18 Ks in the GIS detectors and 16.5 Ks in the SIS. Slightly lower exposure times were obtained for the 1994 observation.

Source photons were extracted from a circular region centered on the source with radius for GIS and radius for SIS. The background was estimated using both source-free regions and blank-sky GIS and SIS observations available in the calibration area. Different background estimates give consistent results.

Data preparation and spectral analysis were performed using version 1.3 of the XSELECT package and version 9.0 of the XSPEC program. The light curves from each instrument do not show any significant flux variation over the whole observation. GIS and SIS spectra were binned with more than 20 cts/bin in the 0.7-10 keV and 0.6-10 keV energy ranges respectively. The lowest SIS energy channels have been excluded because of the uncertain calibrations (Dotani et al. 1996; Cappi et al. 1997). Since the spectral parameters obtained by fitting the four detectors separately were all consistent within the errors, data from both pair of SIS and GIS were fitted simultaneously to the same model, but with the normalization of each dataset allowed to vary relative to the others in order to account for the small discrepancies in the absolute flux calibrations of the detectors.

Both observations gave very similar results; in the following we discuss only the results obtained from the 1995 observation characterized by a better counting statistics.

A single power law model clearly provides an acceptable fit to the data (Table 1). Absorption in excess of the Galactic value is required. There is no need of more complex models. In particular the addition of a narrow emission line at 6.4 keV does not improve the quality of the fit, while a thermal component is not required by the data. The absorption corrected X-ray flux in the 2-10 keV energy band is 3.6 erg cm^{- 2} s^-1, which corresponds to a luminosity of 2.4 erg s^-1 in the source frame.

2.3. ROSAT and ASCA joint fits

In the overlapping 0.6-2.0 keV energy range the observed PSPC flux is about 15% lower than the ASCA flux with a weak dependence on the assumed spectral parameters. Moreover the value of the column density derived from the PSPC data assuming a partial covering model (Table 1) is in good agreement with the absorption observed in the ASCA data (Table 1) suggesting that this model is viable or that two different spectral components are present. Therefore, in order to make full use of the available spectral band (0.1-10 keV) we have performed joint fits to the ROSAT PSPC and ASCA data leaving the relative normalizations free to vary to account for the residual absolute flux uncertainties among the different instruments.

Not surprisingly a partial covering model provides a good description of the observed 0.1-10 keV spectrum: the resulting power law spectrum () is partially (%) absorbed by cold gas with a column density cm^-2 (Fig. 1, Table 1). This implies that in the soft 0.1-2.4 keV band about 38% of the observed flux may be due to a scattered component. A description of the soft X-ray spectrum in terms of thermal emission is not viable, while the addition of a thermal component to the best fit partial covering model does not significantly improve the fit. Any thermal component (if present) cannot contribute more than 8-10% (90% confidence limit) to the observed X-ray flux.

Fig. 1. 3C 219 spectrum from ROSAT PSPC, ASCA GIS and SIS joint fit. The fit has been obtained by leaving the relative normalizations free to vary (see text). The best fit is a partial covering model with , covering fraction, and cm-2. This model accounts for both the high energy (ASCA) part and the more flat low energy (ROSAT) part of the X-ray spectrum. The normalized cts s-1 keV-1 are shown in the upper panel , while the ratio between the observed points and the model is shown in the lower panel ; no iron line is requested by the fit.

We conclude that the 0.1-10 keV spectrum of the radio galaxy 3C 219 can be explained in terms of an obscured central source characterized by a power law with a slope typical of radio loud quasars in the 2-10 keV energy range (Lawson & Turner 1997). An additional unabsorbed spectral component is present in the soft X-ray band. In principle this component may be due to scattering of the soft X-ray nuclear radiation by circumnuclear clouds or thermal electrons, but we note that the unabsorbed power law (Table 1) has a slope close to that of the synchrotron radio spectrum (), so that the possibility that it might originate from IC scattering of the IR-optical nuclear and/or CMB radiation can also be taken into account. A crucial test to decide which one of these two possibilities should be adopted is the study of the spatial distribution of the X-ray flux.

© European Southern Observatory (ESO) 1999

Online publication: December 22, 1998
helpdesk.link@springer.de