Astron. Astrophys. 331, 925-933 (1998)

4. Discussion

RX J0947.0+4721 is denoted as a QSO (rather than a Seyfert) by its optical luminosity of . This is supported by the overall averaged X-ray luminosity of ; Seyferts do not normally have above several . Otherwise, RX J0947.0+4721 is very similar to NLS1 such as IRAS 13224-3809 (Boller et al. 1993). The photon index and FWHM of place it well inside the range obtained by the NLS1 sample in Boller et al. (1996; their fig. 8). The ratio FeII is comparable with other values found for these objects. X-ray variability as observed is a common feature in NLS1 (Boller et al. 1996). As for IRAS 13224-3809, IZw1 and other NLS1, RX J0947.0+4721 has a high far infrared flux. It has been detected by IRAS at 60 with a flux of (Moshir et al. 1990), corresponding to an IR-luminosity of .

Table 4. Parameters of the narrow line QSOs.

A comparison of RX J0947.0+4721 with the narrow line quasars shows that their X-ray parameters are quite similar: the luminosities in the ROSAT band are of the same order of magnitude, and the photon indices are practically identical. This latter observation may be taken as an indication that the steepness extends towards fairly high rest frame energies. However, the indication must not be overstressed. The value of for E1346 266 is not well determined; Puchnarewicz et al. (1994) find smaller values () than Boller et al. (1996).

A better description of the observed spectrum can be achieved by thermal models, however, both the blackbody and thermal bremsstrahlung models underestimate the flux above 1 keV. The analysis in Sect. 2.1has shown that the hard component of the spectrum cannot be totally explained with the neighbouring harder source which is too weak. It must therefore be intrinsic to RX J0947.0+4721.

A good fit is obtained with a blackbody modeling the soft excess plus a power law as hard component, modified by Galactic absorption. For an accretion disk model this means that the X-rays originate from a small region with little variations in temperature.

This interpretation assumes a thermal origin for the X-ray emission in the inner parts of the accretion disk. There is, however, an indication against this assumption. No change of the hardness ratio with increasing count rate is observed, which means that the temperature of the emitting region remains constant over nearly one order of magnitude in luminosity. Over this luminosity range temperature variations should be easily detectable from the correlation .

Model calculations for a spectrum with fixed power law component, and all flux changes attributed to a temperature change in the blackbody component, show that only small deviations from the average flux (table 3, last row) would be compatible with the data.

This is illustrated in Fig. 9, which shows the ratio of high to low count rate spectra for the data (crosses with error bars) and six simulations (lines) with ratios between 0.4 and 2. The curves for ratios 2, 1.75, and 0.4, show significant deviations from the observed ratio, while the others agree with the data below channel 100 (roughly 1 keV). Only changes less than a factor 1.75-2 can be caused by a change in T. The large changes observed (factors 5.7, 2.7) must be caused by another process.

Fig. 9. Ratio of high to low count rate spectra versus channel number. Crosses: data; solid line: , dotted line: , short dashed line: , dash-dotted line: , long dashed line: , dash-dot-dotted line:

A luminosity change in dependency of the radius (pulsations with constant temperature, ) is difficult to model for an accreting black hole because the X-ray emitting region should have always the same distance (in Schwarzschild radii) from the central engine. As a consequence, models involving reprocessing of harder X-ray into softer photons could be a solution.

In Guilbert & Rees (1988), the central engine produces a non-thermal spectrum of hard X-rays and -rays of sufficient energy to produce electron-positron pairs which in turn can produce and maintain a secondary plasma with optical depth . The large optical depth is responsible for the existence of a cool component in the gas close to the central engine. Incident radiation is reflected and reprocessed by cold material. More recent works (e.g. Maraschi & Haardt 1997, Haardt et al. 1997) have shown that the disk corona may have only electrons instead of pairs, and that it is not necessarily optically thick. The fact that the iron line is not Comptonized to invisibility supports for the corona. However, although these models can explain the shape of the soft X-ray spectrum, the lack of spectral variations may cause problems. In Haardt et al. (1997), noticeable spectral changes are predicted for the ROSAT band, none of which are observed for RX J0947.0+4721.

Models including warm absorbers may be another possible explanation of NLS1 phenomena. A detailed description of such a model applied to Mrk 766 can be found in Leighly et al. (1996). Unfortunately, Mrk 766 is more a Sy 1.5 than a NLS1 (Osterbrock & Pogge 1985), and it may be inappropriate to generalize the results for this special object to the whole NLS1 class. The application of such a model to RX J0947.0+4721 would conflict with the lack of spectral variability, anyway. Changes of ionisation parameter and/or column density of the warm absorber would cause changes in spectral shape. Spectral changes are expected even if the variations arise via changes in the central source itself since photoionisation contributes to the ionisation structure of the absorber. The lack of spectral changes for RX J0947.0+4721 is a strong argument against warm absorber models.

Further model constraints can in principle be derived from timing analysis. The shortest variation significantly detected for RX J0947.0+4721 corresponds to a decrease (bb + pl model) in . Following Fabian & Rees (1979), this value implies an efficiency -far below the limit of 0.057 for accretion onto a Schwarzschild black hole. If the single power law model is used instead, we have , and subsequently , still more than a factor 4 below the limit.

Blazar-like activity, known to produce rapid X-ray variations, is usually not applied to NLS1 galaxies because NLS1 objects do not show properties which are typical for jet activities, as there are flat spectrum radio emission, strong polarisation, and nearly featureless multifrequency spectra. RX J0947.0+4721 may be different in that the possibility of a flat radio spectrum (i.e. , which would make it also radio loud) can presently not be ruled out.

If the short time variations are confirmed, a Schwarzschild black hole will no longer be a proper model. The Schwarzschild limit is passed if large amplitude variations () in 16000 s are detected. These would suggest either the presence of a Kerr black hole or relativistic X-ray beaming effects.

The large amplitude variation of a factor hints at models like a filamentary or spot-like emission region (IRAS 13224-3809; Otani et al. 1996) or tidal disruption of stars (IC 3599; Brandt et al. 1995, Grupe et al. 1995) which have been suggested to explain giant amplitude variations in these ultrasoft NLS1 galaxies.

© European Southern Observatory (ESO) 1998

Online publication: March 3, 1998
helpdesk.link@springer.de