3. Spectral analysis
The spectral analysis has been performed by means of the XSPEC 9.0 package, and using the instrument response matrices released by the BeppoSAX Science Data Centre in January 1997. Figure 1 shows the ratio of the broad band spectrum between 0.1 and 100 keV to a simple power law () model. Residuals indicate the presence of a significant soft excess below 2 keV, a strong line emission around 6.4 keV, and excess emission in the 13-30 keV band. These clear spectral features make NGC 7674 an interesting case for investigating and testing reflection models. All the quoted errors correspond to 90% confidence intervals for one interesting parameter ( of 2.71).
3.1. The 2-10 keV spectrum
The MECS spectral data in the 2-10 keV energy band were first fitted with an absorbed power law model, plus a Gaussian narrow ( fixed) line to take into account the 6.4 keV feature, with all other parameters left free to vary. The best fit (model #1 in Table 1) parameters are a photon index , an absorption column density cm-2, and a gaussian line centered (in the reference frame of the emitting source) around 6.4 keV with an equivalent width of 1 keV. The fit improves significantly ( 98% via F-test) if the line is allowed to broaden and gives a width of keV and an equivalent width of 2 keV (model #2). Unless extreme iron abundances are assumed (about an order of magnitude larger than the cosmic value), the observed iron emission line intensity cannot be explained by transmission through the measured absorption column density which would predict an EW 100 eV (Makishima 1986, Ptak et al. 1996). Such a strong line, together with the unusual flatness of the continuum, is therefore a very strong evidence that the observed hard X-ray spectrum is reflection-dominated.
Table 1. Upper panel: MECS fit with a power law plus different models; lower panel: pure reflection fits.
3.2. The soft excess
The low statistics at energies below 2 keV do not allow to firmly discriminate among different models tested. In fact, the excess detected in the LECS data at keV is well modeled both by a steep () power law absorbed by the galactic column density in the direction of the source ( cm-2 ; Dickey and Lockman ) and by a Raymond-Smith model with solar abundance and kT 0.9 keV (models #3a and #3b). Moreover, an acceptable fit is also obtained assuming an electron scattering model with (model #4). The possible implications of the scattering model are discussed in Sect. 4.
3.3. The broad-band reflection-dominated spectrum
The broad band (0.1-100 keV) spectrum has been fitted with a soft component (see Sect. 3.2) plus a pure reflection continuum resulting from a power law illumination of cold and thick material plus a gaussian line. The use of this model is justified by the fact that the emission line centroid energy is consistent with K shell fluorescence from neutral Iron and has an EW of the order of what is expected in the case of a pure reflection continuum (e.g. Matt, Perola, and Piro 1991; Ghisellini, Haardt, and Matt 1994; Krolik, Madau, and Zycki 1994). In this reflection model (PLREFL in XSPEC) the only free parameter is the relative normalization of the reflected component to the direct one. If this parameter is left free to vary, only a lower limit () is obtained, thus confirming that the observed spectrum is dominated by reflection. Therefore a purely reflected spectrum (i.e. no direct component) was chosen. The best fit gives in this case (model #3a) an intrinsic photon index consistent with the average values of Seyfert 1 galaxies (Nandra and Pounds 1994). The 2-10 keV observed flux is erg cm-2 s-1 corresponding to an observed luminosity of erg s-1. The observed keV emission feature can also be fitted with a more complex model. An acceptable fit (model #5) is in fact obtained by adding to the Fe neutral line a second emission feature. The best-fit energy centroid of this second line (6.90 keV) is consistent both with H-like Fe (6.97 keV) or with a blend of He- (6.70 keV) plus H-like Fe. This indicates that, certainly, reflection from thick cold matter plays an important role. Nonetheless, contribution from reflection caused by ionized matter cannot be ruled out on the basis of the present data. Finally, an acceptable fit (model #6) is also obtained assuming both and emission to be present in the form of narrow lines. The best-fit equivalent widths are consistent with the ratio expected for neutral iron (0.135 [Weast 1987]).
© European Southern Observatory (ESO) 1998
Online publication: February 16, 1998