Astron. Astrophys. 343, 33-40 (1999)

4. Discussion

The detection of the iron line and of the reflection component in the BeppoSAX observation of 3C 390.3 indicates that beamed non-thermal radiation does not contribute significantly to the X-ray continuum. This is probably true independently of the brightness of the X-ray source, because a strong iron line () was also detected by ASCA in 1995 when the source flux was about 1.5 times larger than the BeppoSAX value (Leighly et al. 1997).

In 1995, 3C 390.3 was the object of a multifrequency campaign which included IUE, ROSAT and ASCA observations (Leighly et al. 1997, O'Brien et al. 1998). The UV and X-ray light curves, covering a period of about 8 months with a regular 3 day sampling, showed similar forms and variability amplitudes. As pointed out by O'Brien & Leighly (1997), if the UV were a direct extension of the X-ray emission, the two light curves should show different variability amplitudes, because the ASCA spectral slopes from two observations during the monitoring differed by .

It is therefore likely that (at least part of) the UV is due to reprocessing of X-rays. Indeed, an excess of UV emission above the X-ray power law extrapolation (the blue bump) was noted by Walter et al. (1994) using simultaneous ROSAT-IUE observations performed during the ROSAT all-sky survey. However, the blue bump component, if present, is weak, as also indicated by the historical compilation of non-simultaneous ultraviolet and X-ray data of Wamsteker et al. (1997). The lack of a soft X-ray excess attested by several satellites (Walter et al. 1994, Eracleous et al. 1996, Leighly et al. 1997) and confirmed by our data (and by the re-analysis of the EXOSAT observations) further strengthens this conclusion.

In Fig. 6 the radio to -ray energy distribution of the 3C 390.3 is shown. Data from the literature (Rudnick et al. 1986, Steppe et al. 1988, Knapp et al. 1990, Poggioli 1991) are combined with the simultaneous optical-UV-X-ray data collected on 1995 January 14-15 during a 3C 390.3 multifrequency campaign (Leighly et al. 1997, Dietrich et al. 1998, O'Brien et al. 1988). The radio points correspond to the core flux only. Optical and UV measurements collected on 1995 January 14-15 refer to the continuum emission dereddened with the extinction curve of Seaton (1979) assuming Av=0.708. The visual extinction was deduced by the  column density measured by a simultaneous ASCA observation performed on 1995 January 15 (Leighly et al. 1997). On that occasion 3C 390.3 was in a state of brightness very similar to that observed by BeppoSAX later, as can be seen in Fig. 6, where the ASCA  flux at 1 keV is plotted together with the MECS and PDS data. For the sake of clarity, the MECS and PDS data have been rebinned in order to have a signal to noise ratio of about 50 and 10 per each bin, respectively.

Fig. 6. Spectral energy distribution of 3C 390.3, from non-simultaneous observations. Radio-mm-infrared data (filled circles) are taken from literature (see text). Simultaneous optical UV and X-ray data (open circles) refer to the observation on 1995 January 14-15. The MECS and PDS data (filled squares) have been rebinned.

Fig. 6 shows that the energetics are dominated by the high energy end of the power-law, and by the large IR emission. The power emitted in the UV is definitely smaller than that in the X-ray to hard X-ray component. This is consistent with the results of Wozniak et al. (1998), which show that during spectral variations the total energy output in X-rays does not change (see their Fig. 5).

All these facts indicate that the UV-emitting, optically thick gas must subtend a relatively small solid angle to the X-ray source, otherwise strong reprocessing would give rise to a thermal UV component that is energetically important and efficient cooling would steepen the X-ray power-law. On the other hand, as discussed in Sect. 3, the observed reflection hump and iron line need a fairly large covering factor of reprocessing material in order to be accounted for. How can these two, apparently contradictory, constraints be matched?

There are basically two models under discussion to account for X-ray emission from accretion disks in AGN:

1. A Seyfert-like model that assumes a geometrically thin accretion disk (Shakura & Sunyaev 1973), which is responsible for the UV thermal emission and for the reprocessing of (part of) the X-ray photons. The high energy photons are produced by an active corona embedding the inner portion of the cold accretion disk (Liang 1979, Haardt & Maraschi 1991, 1993). In this class of models, the Compton parameter is kept fixed by the energetic feedback linking the disk and the corona. UV photons in the disk are produced by thermalization of the absorbed X-rays, and X-rays in the corona are produced by inverse Comptonization of the UV disk radiation;

2. A hot accretion flow model, such as the original two-temperature solution introduced by SLE or the ion-supported torus proposed by Ichimaru (1977), and Rees et al. (1982), or its modern version, the ADAF (see NMQ for a review and all the relevant references). In the ion-supported torus and in the ADAF, relevant for low to modest accretion rates, the small gas density makes Coulomb collisions very ineffective in transferring energy from the ions (which are supposed to be directly energized by viscous stresses) to the electrons which bear the ultimate responsibility of radiating away the heat and cooling the gas. The direct consequence is the formation of a hot, two-temperature plasma in the inner region of the flow. In contrast, in the SLE solution the energy deposited in the gas is assumed to be locally radiated. In both classes of models, at larger distances from the black hole, the flow is thought to be described by a standard cooling-dominated thin disk (e.g., Mahadevan 1977).

From the point of view of the formation of the radiation spectrum, the main difference between the two pictures is the presence (in the disk-corona system) or the absence (in the ion-supported torus) of optically thick cold matter close to the X-ray source providing (or not) soft photons for the Comptonization mechanism. In the absence of a soft photon input from thermal optically thick gas, the seed photons are provided by cyclo-synchrotron radiation by the electrons themselves, yielding a power law which extends from the far IR up to hard X-rays with spectral index fixed by the accretion rate.

In the case of 3C 390.3, although a disk-corona model could explain the strong correlation between the IUE and X-ray light curves, the absence of a soft excess and the weakness of a possible blue bump argues against a large fraction of reprocessed radiation. An optically thick corona radiating all the available gravitational power could in principle scatter off all the black body photons from the accretion disk and hence produce a unique Compton-scattered power law (Haardt & Maraschi, 1993), with weak or absent signature of thermal emission. In this case, however, such a disk-corona system would give rise to a power law spectrum steeper than observed (). In order to produce an X-ray power-law as flat as observed, one has to assume a photon-starved corona. Thus the required geometry is one in which the UV-emitting layer is at least partly external with respect to the region containing the hot electrons.

A completely hot inner flow, on the other hand, is a plausible description of the nuclear region. A hot inner region can in fact explain the lack of soft excess and the weak UV bump, which might still be the signature of an external standard cold thin disk. The ion-supported torus, or the ADAF, is one of the possible stable configurations of gas accreting onto a black hole. The optical to X-ray radiation is due to Compton cooling of the hot thermal electrons (with temperature  K), scattering off soft free-free and cyclo-synchrotron photons. If the accretion rate is high (but still below the ADAF critical accretion rate, , see NMQ), the bremsstrahlung contribution to the X-ray spectrum is negligible, and the X-ray continuum is hard (). In the case of 3C 390.3, the bolometric luminosity estimated from Fig. 6 is erg sec^-1. If we assume a central mass of 1-4 10⁸ (Wamsteker et al. 1997), the luminosity in Eddington units is ., which would place 3C 390.3 in the range of high accretion rate ADAFs, consistent with the hard X-ray continuum. We note nevertheless that with this model the similarity of the spectral shape of 3C 390.3 to that of Seyfert galaxies would be coincidental.

If the accretion flow at larger radii is in the form of a standard thin disk, the weak blue bump could be also explained, as due to local energy release, and reprocessing of the (small) fraction of the X-rays intercepted and reprocessed by the cold matter. A flat infinite disk illuminated by a central hot ion-supported torus cannot intercept more than 25 of the primary continuum (Chen & Halpern 1989), adequate for the observed UV emission in this case, but not for the observed reflection component and for the iron line EW. It is then necessary that further cold material, encircling the central source, is shaped like a warped disk or a thick dusty torus at parsec distances. These geometries can ensure large covering factors, and produce a broad reflection hump practically indistinguishable (at this sensitivity level) from that arising from an infinite plane parallel medium (Ghisellini, Haardt & Matt, 1994; Krolik, Madau & Zycki, 1994). In particular, the feature is expected to be narrow, in agreement with the iron line profiles observed by BeppoSAX ( eV) and ASCA (Eracleous et al. 1997). The energetics would not be a problem in this picture, as most of the X-rays are absorbed at large distances from the source, and then re-emitted as IR radiation, rather than in the form of a UV bump as in the case of a Seyfert-like geometry. A simple test of this model would be the absence of short term variability both in the intensity of the Fe Line (Wozniak et al. 1998) and of the reflection component. For the latter, further BeppoSAX observations would be valuable.

An inner hot torus surrounded by an outer cold thin disk was already proposed by Chen and Halpern (1989) to explain the optical properties of BLRGs with double peaked emission lines and in particular of 3C 390.3. Our observations independently strengthen this picture. Whether this configuration can be consistent with the SLE solution, with the ion-supported torus, and/or with the ADAF accretion models is a matter of future investigations, beyond the scope of the present paper.

Finally, an important result of our studies is the discovery in 3C 390.3 of temporal variations of the local column density. The origin of the  variability is unknown. The presence of a warm absorber, usually invoked to explain modification of the column density in Seyfert galaxies, seems unlikely in 3C 390.3. The lack of features in absorption/emission in the soft X-ray spectrum and the absence of any correlation between the  values and the X-ray flux argue against this possibility. The long term variations can be better explained by geometrical modifications of a cold absorber.

© European Southern Observatory (ESO) 1999

Online publication: March 1, 1999
helpdesk.link@springer.de