Soft X-ray excesses, i.e. excess emission in the soft ( keV)
X-ray band with respect to the extrapolation of the spectrum at higher
energies, are fairly common in radio-quiet Active Galactic
Nuclei (AGN). Turner & Pounds (1989) found that about 50% of the
sources in their EXOSAT sample exhibits a significant low energy spectral
steepening. An analysis of
ROSAT All Sky Survey sources by
Walter & Fink (1993) has shown that the ROSAT PSPC
(0.1-2.4 keV) spectral index is
significantly steeper than the typical one
at higher energies for about 90% of the objects.
The spectral steepening in the soft X-ray band has been
originally attributed to a real emission component, i.e.
thermal emission from an accretion disk (i.e. Arnaud et al. 1985).
Later on, the relevance of
absorption features due to ionized matter (i.e. the "warm absorber",
Halpern 1984) in modifying the spectral shape in soft
X-rays
has also been recognized in some sources (e.g. Nandra & Pounds 1992).
More recently an alternative explanation based on
reflection by ionized media has been successfully applied to a few
AGN (Piro et al. 1997; Życki et al. 1994;
Ross & Fabian 1993).
On the other hand, the systematic analysis of Walter & Fink indicates that the shape of this component is similar in all the sources of their sample, suggesting a common origin. Unfortunately, while the ROSAT-PSPC has been of valuable importance in establishing the presence of soft excesses, its limited bandwidth and energy resolution do not permit to study in detail their shape and strength. To this aim, an independent knowledge of the underlying hard X-ray power law would be extremely important, this component being not easily measurable by ROSAT. For a few sources, simultaneous ROSAT-Ginga observations were available. Most of them have been obtained during the ROSAT All Sky Survey (RASS); owing to the limited sensitivity of the RASS data, the conclusions proposed by two different groups (Walter et al. 1994; Pounds et al. 1994) on the origin of the soft X-ray excess were somewhat different.
We have therefore
performed a systematic study of the spectra of the pointed observations
contained in the ROSAT archive
of a sample of hard X-ray selected sources,
whose 2-10 keV spectrum is well known from not simultaneous
Ginga observations
(Nandra & Pounds 1994). In our analysis we have assumed that
the slope of the intrinsic power law remains constant, leaving its
normalization as a free parameter to account for flux variations.
In fact, convincing evidence for spectral variability of the
intrinsic power law
has been found so far only in two objects,
namely NGC 4151 (Perola et al. 1986; Yaqoob & Warwick 1991),
which is not part of our sample, and
NGC 4051 (Guainazzi et al. 1997).
We will comment specifically on the latter source in Sect.4 (click here); we anticipate here that for this object
the results are not significantly affected by the
choice of the intrinsic slope.
The paper is organized as follows: the sample selection, ROSAT observations and data reduction are discussed in Sect.2 (click here) . The results are described in Sect.3 (click here) , while Sect.4 (click here) is devoted to individual sources and Sect. 5 (click here) to the discussion and the conclusions.