The above results have shown that the broad-band spectrum of Mkn 3 is best described by the sum of a soft power-law, an unabsorbed reflection component, an iron line and a strongly absorbed hard power-law component. The most remarkable result of the present analysis is that the hard (E 20 keV) X-ray spectrum is steep with 1.8 and that there is no evidence of a spectral break in the data for energies up to at least 150 keV. The steep intrinsic spectrum is consistent with the canonical value found for Seyfert 1 galaxies (Nandra & Pounds 1994, Perola et al. 1999). The lack of a cutoff in our data is consistent with the large average value ( 0.7 MeV) found for a sample of 5 radio-quiet Seyfert 1 galaxies detected both by OSSE and Ginga (Gondek et al. 1996, Zdziarski et al. 1997). It is also consistent with the lower-limits ( 150 keV, Perola et al. 1999) and detections ( 150-200 keV, Piro et al. 1999) obtained with BeppoSAX for several Seyfert 1-1.5 galaxies. Both measurements thus support a unified model of Seyfert galaxies.
The very high absorption column density measured here makes Mkn 3 very similar to other known buried Seyfert 2 galaxies: NGC 4945 (Done et al. 1996), Circinus galaxy (Matt et al., in preparation) and possibly NGC 4941 (Salvati et al. 1997). As pointed out by Salvati et al. (1997), a better knowledge of the number of such heavily absorbed sources is extremely important because it has direct consequences on synthesis models of the X-ray background (Comastri et al. 1995, Gilli et al. 1999). In the 3-6 keV band, the spectrum of Mkn 3 requires a relevant contribution from reprocessed emission, namely an unabsorbed reflection component with associated iron emission line. The best-fit parameters obtained (R 0.95, E(Fe ) 6.4 keV, and EW(Fe ) 650 eV with respect to the reflection component) are comparable to the theoretical values expected for a cold reflector covering a solid angle of 2 at the source (George & Fabian 1991, Matt et al. 1991). The long-term variability study suggests a distance of the reprocessor 7 light years (Sect. 3.5), that could thus be identified with an obscuring torus. Similar results have been found for example in NGC 2992 (Weaver et al. 1996), NGC 4151 (Piro et al. 1999) and NGC 4051 (Guainazzi et al. 1998). Moreover, a 2 covering of the reprocessor is also consistent with a torus interpretation, provided that its half-opening angle is 45%.
In conclusion, both the strong absorption and unabsorbed reflection can be explained in the framework of unified models (i.e. assuming the existence of an optically thick torus). The absorption resulting from the transmission of the direct component through the rim of the torus. The reflection component (observed directly) resulting from reprocessing of the (same) direct component by the inner surface of the torus.
The origin of the second narrow line at higher energies is instead puzzling. Its energy (E 7.0 keV) and intensity (EW 235 160 eV) are roughly consistent with fluorescence iron K emission from either the reflection component or the absorption component. Theoretical models would in fact predict (E 7.06 keV and EW 72 eV, assuming a K/K ratio of 1:9, Matt et al. 1996). Alternatively, it could be interpreted as H- and/or He-like iron emission produced by the scattered soft component itself. Indeed, as shown by Matt et al. (1996), very strong resonantly scattered H- and He-like lines (with EW between 2-4 keV with respect to the scattered component, depending on the material optical depth and the line being strongest in the optically thin regime) are expected if warm material is responsible for the scattering. Given the large uncertainties in our measurement of its equivalent width, especially when calculated with respect to the soft scattered component (Sect. 3.4), such possibility cannot be excluded. There could also be a contribution from both K fluorescence and resonant scattering. Moreover, we cannot exclude that both the weak red and blue wings of the Fe line (Fig. 6) are produced by an additional reflection component (with associated diskline) from an accretion disk, as commonly observed in Seyfert 1 galaxies (Nandra & Pounds 1994, Perola et al. 1999) and, possibly, in some Narrow Emission Line Galaxies (Weaver & Reynolds 1998). Reflection from a highly inclined relativistic accretion disk would indeed produce extra emission at energies below and above the transmitted 6.4 keV Fe line. In the case of Mkn 3, the effects of strong absorption and poor statistics hamper, however, a detailed modeling of such additional component that will require the use of more sensitive instruments like AXAF or XMM.
© European Southern Observatory (ESO) 1999
Online publication: March 29, 1999