2 Analysis of ASCA data

We searched the ASCA public archive at HEASARC, finding 12 observations of 7 blazars that have not been analyzed and published before: 0405-123 and PKS 0420-014 (classified as Flat Spectrum Radio Quasars, FSRQs); B2 1308+326 and 1807+698 (classified as Low Peaked BL Lac objects, LBLs); 1ES 1028+511, 1553+511 and 1ES 2344+514 (classified as High Peaked BL Lac objects, HBLs).

1ES 1028+511 and 1ES 2344+514 have 3 separate observations each, while B2 1308+326 was observed twice. The search for unpublished observations is updated to November 1999. During the preparation of this work the "Tartarus" data base became available, presenting results of an automatic spectral and temporal analysis for the AGNs observed by ASCA.

2.1 Data reduction and spectral analysis

We extracted the spectra of all sources from the files produced by the Revision-2 data release and included data transmitted in all 3 modes (High, Medium and Low) to increase the signal/noise ratio. The event files are obtained from all four instruments on board ASCA: the Solid-State Imaging Spectrometers (SIS0 and SIS1) and the Gas Imaging Spectrometers (GIS2 and GIS3). For the SIS we used the event files converted into BRIGHT mode. For a description of the ASCA observatory see e.g. Tanaka et al. (1994).

To screen the SIS and GIS data we follow the criteria given in the ABC ASCA reduction guide, rejecting the data taken during the passage of the South Atlantic Anomaly, or with geomagnetic cutoff rigidity lower than 8 GeV/c, or with angles between the targets and the day/night terminator smaller than 20$^\circ$ or for Elevation angles smaller than 5$^\circ$.

The source spectra were extracted from circular regions centered on the sources, with radii of 6 arcmin for the GIS and 4 arcmin for the SIS0, while for the SIS1 the source is normally nearer to the detector border and we had to use a smaller radius of ${\sim} 3.3$ arcmin. For the GIS we extracted the background in circular regions with the same dimensions used for the sources but centered on a symmetric point with respect to the optical axis, where the contribution of the source to the counts was negligible. For the SIS, instead, the background was extracted from blank field files because the sources occupied a large area of the detector. On these blank fields we chose circular regions with the same radii and positions used for the sources.

For the GIS spectra we used the 1994 May response matrices, while for the SIS spectra we generated the matrices with the SISRMG program of the FTOOLS V3.6 package. The ARF files for both SIS and GIS were derived with the ASCAARF V2.62 program. The GIS and SIS data were fitted in the channel ranges 69-1020 and 15-510, corresponding to the energy ranges 0.7-10 and 0.4-10 keV, respectively. The spectra were rebinned in order to have at least 25 counts in each new bin.

The ASCA spectra were fitted using XSPEC V10 with four models: single or broken power law with free or fixed (Galactic) absorption. The cross section for photoelectric absorption is calculated following Morrison & McCammon (1983), while the Galactic column density in the direction of the sources was estimated from the 21 cm radio maps of neutral hydrogen (Brinkmann & Siebert 1994; Danly et al. 1992; Dickey & Lockman 1990; Elvis et al. 1989; Lamer et al. 1996; Lockman & Savage 1995; Murphy et al. 1996). The data were fitted simultaneously from all the instruments with the same model. However, the normalizations were left as independent parameters for each data set to account for the cross-calibration uncertainties between the four detectors, estimated to be of the order of 6%. The differences found between the various normalizations were always consistent with these uncertainties.

 

 

Table 1: Best fits of the 12 observations. ^a day/month/year; ^b 10^-12 erg cm^-2 s^-1.

Source	Obs. datea	$N_{\rm H}$	$\Gamma$	$\chi_{\rm r}^2/{\rm d.o.f.}$	F[2-10]b	$F_{\rm 1\,keV}$
		1021 cm-2				$\rm\mu Jy$
0405-123	09/08/1998	0.72 +0.51-0.63	1.76 +0.09-0.10	1.0/160	5.2	0.9
0420-014	31/08/1997	0.90 +1.04-0.81	1.86 +0.19-0.18	0.8/73	1.4	0.3
1028+511	28/04/1995	1.01 +0.23-0.21	2.53 +0.06-0.05	1.0/287	6.2	3.4
	29/04/1995	1.33 +0.18-0.17	2.59 +0.05-0.05	0.9/310	7.4	4.5
	08/05/1995	1.12 +0.12-0.12	2.49 +0.04-0.04	0.9/201	7.8	4.1
1308+326	10/06/1996	1.97 +2.96-1.97	1.99 +0.47-0.36	1.5/29	0.5	0.1
	11/06/1996	0.97 +1.01-0.97	1.74 +0.32-0.23	1.4/78	0.6	0.1
1553+113	16/08/1995	1.30 +0.62-0.61	2.47 +0.19-0.18	1.3/294	29.4	14.9
1807+698	05/11/1996	0.50 +0.28-0.27	1.75 +0.07-0.06	1.0/122	3.1	0.5
2344+514	10/01/1997	2.71 +0.17-0.17	2.13 +0.03-0.03	1.1/216	17.2	5.3
	23/01/1997	2.91 +0.34-0.33	2.39 +0.08-0.06	0.9/72	10.4	5.2
	10/12/1997	2.93 +0.37-0.35	2.31 +0.07-0.07	1.2/202	10.4	4.2

2.2 Results of the fits

The results of the 12 spectral fits of the 7 blazars observed by ASCA are reported in Table 1. The uncertainties for the spectral parameters are at the 90% confidence errors for two parameters of interest ( $\Delta \chi^2= 4.6$). The unabsorbed integrated 2-10 keV and monochromatic 1 keV fluxes are obtained using only the SIS0 data. As the observations presented in this paper were all taken after 1994, they are most likely affected by the so-called "excess $N_{\rm H}$" problem. This is due to a degradation of the SIS efficiency below 1 keV which can give incorrect results for the column density and/or other parameters. As suggested in the ASCA Web site, to avoid the calibration uncertainties we considered only the SIS data above 1 keV in the SIS+GIS combined fit. Of the four models considered, according to the F-test, the one that better represents the data in all twelve cases is the single power law with free $N_{\rm H}$. In some cases (detailed below) we obtain a value for the absorbing column greater (by a factor 3-10) than the Galactic value. Note that the results obtained by the automatic analysis presented in the "Tartarus" database are in excellent agreement with ours (not surprisingly, since the same model is adopted). What is somewhat surprising is that the broken power law model (either with free or fixed $N_{\rm H}$) did not significantly improve the fits. This may be indicative of true extra-absorption or a spectral behavior more complex than those here examined (one possibility being a gradual but continuous steepening of the spectrum).

In the following we will compare the spectral properties of blazars in the soft and hard X-ray bands considering only the single power law model. To this end, the results of this model allow a more uniform comparison.

The results of spectral fit for the 7 sources are discussed below, grouped in the three subclasses (FSRQ, LBL and HBL).

2.2.1 FSRQ

For 0405-123, the best fit gives a flat photon spectral index $\Gamma = 1.76 \pm 0.1$, indicating the dominance of the inverse Compton component. The derived $N_{\rm H}$ value is consistent with the Galactic value, $N_{\rm H}^{\rm Gal} = 0.37 \times 10^{21}$ cm^-2 (Danly et al. 1992).

Similar results are obtained for PKS0420-014, with a photon spectral index $\Gamma = 1.86 \pm 0.19$. Also in this case the $N_{\rm H}$ value is consistent with the Galactic one ( $N_{\rm H}^{\rm Gal} = 0.94 \times 10^{21}$ cm^-2, Elvis et al. 1989).

2.2.2 LBL

For B21308+326 there are two observations on two consecutive days. The spectra are quite noisy and also the reduced $\chi_{\rm r}^2$ are not very good. In both cases the best fits are obtained with a spectral index $\Gamma \simeq 1.75$-2.0. Due to the large error bars (see Table 1), the derived absorption column density can be consistent with the Galactic one ( $N_{\rm H}^{\rm Gal} = 0.11 \times 10^{21}$ cm^-2, Lockman & Savage 1995).

For 1807+698 we obtained $\Gamma = 1.75\pm0.07$ and a value of $N_{\rm H}$consistent with the Galactic one ( $N_{\rm H}^{\rm Gal} = 0.44 \times 10^{21}$ cm^-2, Murphy et al. 1996).

2.2.3 HBL

1ES1028+511 has been observed three times over a time span of two weeks. The source did not significantly vary in flux nor in shape ( $\Delta\Gamma \simeq 0.12$). In all cases the best fits are obtained for a value of $N_{\rm H}$ a factor 10 larger than the Galactic one ( $N_{\rm H}^{\rm Gal} = 0.12 \times 10^{21}$ cm^-2; Lamer et al. 1996).

For 1553+113 the best fit is obtained with a value of $N_{\rm H}\sim 3$times larger than the Galactic one ( $N_{\rm H}^{\rm Gal} = 0.37 \times 10^{21}$ cm^-2; Brinkmann & Siebert 1994).

The source 1ES2344+514 has been observed three times, twice in Jan. 1997 and one in Dec. 1997. Both the fluxes and the spectral indices show some variability. The best fit value for the $N_{\rm H}$ is about 1.5 times the Galactic value ( $N_{\rm H}^{\rm Gal} = 1.67 \times 10^{21}$ cm^-2Dickey & Lockman 1990).