We have studied the properties of the soft excesses for the Nandra & Pounds (1994) sample of hard X-ray selected Seyfert 1 galaxies observed by Ginga. After excluding the objects with line-of-sight absorption in excess of 1021 cm-2 and/or with a total number of counts less than 2000, we obtained for 13 out of the 14 remaining sources a ROSAT power law index significantly steeper than the Ginga power law index, indicating the presence of a soft excess. We then fitted these 13 sources with three different models for the soft excess: a black body, an absorption edge and a reflection model (in all these fits the power law index of the hard component has been kept fixed to the Ginga value, implicitely assuming lack of spectral variability).
The main results can be summarized as follows:
We therefore conclude that the soft excess phenomenon is complex and probably different from source to source. In some sources, more than one of the models discussed here are likely to occur simultaneously. This is not surprising, because the models are not mutually exclusive. Regarding the black body model, which is the most successful one, there is no strong evidence of correlation between temperature and luminosity. This could argue against simple accretion disc models, even if, after excluding NGC 4051, the range of luminosities spanned by our sample does not permit any definitive conclusion.
The analysis presented here is based on the assumption of absence of spectral variability. Even if this assumption appears realistic, as spectral variability has been observed up to now only in a handful of sources, it would obviously be desirable to determine simultaneously the shapes of the soft excess and the hard component. Besides, simultaneous observation would give informations on correlated variability, and therefore would further constrain models. Broad band observations, like those that will be performed by BeppoSAX, are expected to shed new light on the soft excess problem.
Acknowledgements
We thank G.C. Perola for a critical reading of the paper and suggestions and E. Massaro for useful discussions.