Astron. Astrophys. 355, 113-120 (2000)

2. X-ray data reduction and analysis

IRAS12393+3520 was observed by ROSAT/PSPC and ASCA on 1993, 16-17 December, and 1997, 4-5 June, respectively. The data have been retrieved from the HEASARC archive, as cleaned event lists. The ROSAT PSPC is a proportional counter sensitive in the 0.1-2 keV band, with moderate energy resolution ( at 1 keV) and a spatial resolution of about 20" for on-axis sources. The ASCA scientific payload comprises two Solid Imaging Spectrometers (SIS0 and SIS1, sensitive in the 0.6-9 keV band, Gendreau 1995) and two Gas Imaging Spectrometers (GIS2 and GIS3, sensitive in the 0.7-10 keV band, Ohashi et al. 1996). SIS grades 0,1,2 and 4 and 1-CCD mode were used.

The best ROSAT/PSPC centroid position (= and =03´46") is within 22" from the optical position (Santagata et al. 1987) and 40" from the ASCA centroid. The latter quantity is comparable with the typical ASCA positional uncertainties (Gotthelf & Ishibashi 1997). Several serendipitous sources are present in the PSPC field of view (see Fig. 2).

Fig. 2. PSPC field of view of IRAS12393+3520 (the central source) superimposed to the GIS2 intensity contours of the ASCA observation

Two of them could in principle contaminate the ASCA source spots. Their properties are reported in Table 1.

Table 1. Properties of the serendipitous sources #1 and #2 in the PSPC field. Spectra parameters refer to a power-law plus photoelectric absorption model of the PSPC spectrum.

We expect that any contamination to the IRAS12393+3520 ASCA flux is lower than 10% and 5% in the 0.5-2 and 2-10 keV bands, respectively, well within the statistical uncertainties of the ASCA measurement. PSPC scientific products have been extracted from a circular region of about 2´ around the apparent centroid of the source. Background spectra have been extracted in an annulus of radii 3´20" and 5´20", after removing circular areas of 2´ radius around any serendipitous source included in the annulus.

In the SIS0 or SIS1 images no source is detected other than IRAS12393+3520, the 3 upper limit at the position of source #1 in PSPC field being  s^-1. The closest serendipitous source in the GIS field of view (= and =03´46"; distance 5.8´) is probably associated with the foreground galaxy NGP9 F268-1081523. ASCA source scientific products have been then extracted from circles of radii 4´, 3´.25 and 3´.75, in the SIS0, SIS1 and GIS, respectively. Background scientific products have been extracted from regions of the field of view free from contaminating sources. None of the presented results changes significantly if the background is extracted from blank sky fields. Total exposure times are about 33, 32 and 13 ks for the SIS, GIS and PSPC, respectively. Full band count rates were , , , and s^-1, for the SIS0, SIS1, GIS2, GIS3 and PSPC, respectively.

2.1. The X-ray light curve

In Fig. 3 the light curves in the 0.1-2 keV (PSPC) and

Fig. 3. 0.1-2 keV background-subtracted (ROSAT/PSPC, upper panel ) and 0.5-4 keV (ASCA/SIS, lower panel ) light curves. Binning time is 5760 s. Only bins with an exposure fraction higher than 15% are shown. If a linear fit is performed on the ASCA light curve in the 5-35 and 35-50 ks intervals, one obtains a rate of flux change of and , respectively

0.5-4 keV bands (SIS) are shown. The latter shows a gentle rise and decay before about 50 ks from the start of the observation. A similar trend is observed in the GIS light curves. If a fit with a constant line is performed on the light curve, degrees of freedom (dof). From the flux change rate of the light curve in the 5-35 and 35-50 ks intervals, we estimate a minimum doubling/halving time in the range 30-75 ks. Variability on lower timescales cannot be ruled out, but the limited available statistics prevented us to reach a firm conclusion on its existence. This variability is not associated with any of the serendipitous sources detected in the PSPC and laying at the border of the ASCA extraction regions. We have extracted SIS0 images in the time intervals between 10-40 and 50-80 ks after the beginning of the ASCA observation. The source spot does not show sign of elongation and the centroid best-fit positions agree within 6". Although an eye inspection of the PSPC light curve might suggest that a similar variability was observed by ROSAT as well, the evidence is not statistically that clear ( dof). We have searched for spectral dynamics associated with the ASCA flux changes, without success. We will therefore focus in the next section on the time averaged spectrum of IRAS12393+3520.

2.2. Spectral analysis

All spectra have been rebinned in order to have at least 20 counts per energy channel, in order to ensure the applicability of the statistics. Proper matrices for the date of the observations have been retrieved from the HEASARC archive or built with the software available in the FTOOLS 4.0 version. First, we fitted the PSPC and ASCA spectra in the overlapping 0.8-2 keV band with a simple photoelectric absorbed power-law model, leaving all parameters free (except the , which has been tight to be the same for all instruments), to check if significant spectral variability arose between the epochs of the two observations. The spectral indices turn out to be well consistent within the statistical uncertainties (; ), despite of an increase in the flux of . In the following, we will therefore fit the spectra of all detectors simultaneously, only allowing a relative normalization factor as a free parameter among all the instruments.

In Fig. 4, the results of a spectral fit with a simple

Fig. 4. PSPC, SIS0 and GIS2 spectra (upper panel ) and residuals in units of standard deviations (lower panel ) when a photoelectric absorbed power-law model is applied to the data of all detector simultaneously. Each data points has a signal-to-noise ratio

power-law with photoelectric absorption are shown. The fit is rather good ( dof), with and  cm^-2. Further spectral complexity is not strongly required by the data. The only feature, whose addition is required at 99% level, is an absorption edge ( for two more parameters), with threshold energy , formally consistent with the K-shell photoionization energy of OVII . The edge is detected with comparable significance in both the ROSAT and ASCA spectra separately. Similar features have been commonly observed in the spectra of Seyfert 1 galaxies (Reynolds 1997; George et al. 1998). The addition of this feature makes the spectrum steeper and therefore consistent with that typically observed in Seyfert 1 galaxies (Nandra & Pounds 1994; Nandra et al. 1997b; see Table 2).

Table 2. Best-fit parameters and results. wa  = photoelectric absorption; ed  = photoelectric absorption edge; px  = Compton reflection; mk  = optically thin plasma; ga  = Gaussian emission line

We have therefore performed a fit with a Seyfert-like model, constituted by a Compton-reflected power-law (model pexrav in XSPEC , Magdziarz & Zdziarski 1995), an absorption edge and a fluorescent broad (i.e.:  keV, Nandra et al. 1997b) iron line from neutral iron (i.e: centroid energy held fixed to 6.4 keV). The line is actually not required by the fit, but we have included it because it is expected on theoretical grounds (George & Fabian 1991; Matt et al. 1992). The fit is comparably good as in the simple power-law case. The amount of reflection, parameterized through the ratio between the reflected and the transmitted components, R ¹, is 1.3, which is not inconsistent with the upper limit on the EW of the iron line (600 eV), if standard cosmic abundances are assumed (Matt et al. 1992). If we add to this model a 1 keV thermal emission from an optically thin plasma, its 0.1-10 keV luminosity is constrained to be lower than  erg s^-1 at the 90% confidence level. The cold absorbing column density is broadly consistent with the Galactic contribution along the IRAS12393+3520 line of sight ( cm^-2, Dickey & Lockman 1990). The fluxes in the 0.5-2, 0.5-4.5 and 2-10 keV energy bands are , 1.00 and  erg cm^-2 s^-1, implying rest frame unabsorbed luminosities of 1.29, 2.4 and  erg s^-1, respectively.

Alternatively, a good fit can be formally obtained also with a double-temperature optically thin plasma (code mekal in XSPEC ), with temperatures  eV and  keV and abundances consistent with solar. In this scenario, the contribution of a (1.0) power-law to the 0.5-4 keV flux is lower than 7% (6%).

© European Southern Observatory (ESO) 2000

Online publication: March 17, 2000
helpdesk.link@springer.de