We first discuss the X-ray properties of the individual objects, in the context of other object-specific observations. Then, a more general discussion on NLSy1s is given.
5.1. QSO 0117-2837
Three models were found to fit the ROSAT X-ray spectrum of QSO 0117-2837 successfully. A single steep powerlaw with , a flat powerlaw plus soft excess, and a warm-absorbed flat powerlaw. In the latter case a rather large column density of the warm absorber is inferred, . This would make the ionized absorber in QSO 0117-2837 the one with the largest column density known, and suggests that other spectral components contribute to, or dominate, its X-ray spectral steepness. Given the reported relations between X-ray and UV absorption, we note that we do not predict any UV absorption from NV and CIV for our best-fit warm absorber model. Both elements are more highly ionized; i.e., the absence of UV absorption alone could not be used as argument against a warm absorber.
Given the very steep rise towards the blue of QSO 0117-2837's optical spectrum (Grupe et al. 1999a; Fig. 1a of Komossa et al. 2000), it is tempting to speculate that a giant soft-excess dominates the optical-to-X-ray spectrum. We strongly caution, though, that simultaneous optical-X-ray variability studies in other Seyferts and NLSy1s (e.g., Done et al. 1995) do not favor a direct relation between optical and X-ray spectral components. Furthermore, such a giant optical-to-X-ray bump in QSO 0117-2837 and NLSy1s in general, would be inconsistent with the finding of Rodriguez-Pascual et al. (1997) that NLSy1s tend to be underluminous in the UV.
A most interesting peculiarity of QSO 0117-2837 is revealed, when combining its X-ray and optical properties: Whereas its X-ray spectrum is among the steepest observed among NLSy1s, its H emission line is fairly broad. Fitting this line with Gaussian components it is best described by a two component Gaussian with a narrow component of similar width as [OIII] plus a broad component of 4000 km/s (Komossa et al. 2000). This combination of and places QSO 0117-2837 in a region in the popular versus diagram (e.g., Fig. 8 of BBF96) which is barely populated by objects, therefore occasionally referred to as `zone of avoidance'.
Its peculiar properties make QSO 0117-2837 a good target for follow-up X-ray spectral observations with XMM and AXAF , as well as for high-resolution optical observations of the H complex.
Some NLy1s have been reported to show rapid X-ray variability. QSO 0117-2837 shows constant source flux during the observation, though.
5.2. RX J0134-4258
5.2.1. Comparison with similar objects
The drastic spectral variability of RX J0134-4258 is rather peculiar. Often, in AGN, the X-ray flux is variable but the spectral slope remains constant. Cases were a strong change in hard X-ray spectral shape was reported, are IRAS 13224 (Otani et al. 1996), NGC 4051 (Guainazzi et al. 1996), and Mrk 766 (Leighly et al. 1996). However, in all cases the spectral index varied mainly between and flatter values. Further, these sources changed countrate when changing spectral shape. Recently, Guainazzi et al. (1998b) presented observations of the Seyfert galaxy 1H0419-577 which underwent a spectral transition from steep ( =-2.5) to flat ( =-1.6) between a ROSAT and a SAX observation. The most similar case to RX J0134-4258 we are aware of is the observation reported by Fink et al. (1997) who detected changes in spectral index of the NL quasar TON S180 that were not accompanied by a noticeable change of the total soft X-ray emission.
5.2.2. Variability mechanisms
One natural mechanism to produce the spectral variability in RX J0134-4258 is warm absorption because this is an efficient method to produce steep X-ray spectra (e.g., by a change in ionization state of the warm absorber). Note, that Grupe (1996) argued against the presence of the warm absorber based on the erroneous statement that a warm absorber could not produce a steep soft X-ray spectrum.
Examination whether, and under which conditions, a warm absorber is indeed a viable description of the X-ray spectrum, and whether it is the only one, has to be based upon detailed modeling and careful consideration of alternatives.
Our modelling (Sect. 4.2) leads to the following results: The ultra-soft state is well fit by a warm absorber with column density . This is a factor of about 2-3 larger than that of the well-studied warm absorber in the NLSy1 galaxy NGC 4051 (e.g., Pounds et al. 1994, Komossa & Fink 1997a, and our Table 3).
The most suggestive scenario within the framework of warm absorbers then is a change in the ionization state of ionized material along the line of sight, caused by varying irradiation by a central ionizing source. One problem arises immediately, though: In the simplest case, lower intrinsic luminosity would be expected, to cause the deeper observed absorption, in 1990. However, the source is somewhat brighter in the RASS observation. (Fig. 4). Some variability seems to be usual, though. The countrate changes by about a factor of 2 during the pointed observation (Fig. 3). If one wishes to keep this scenario, one would have to assume that the ionization state of the absorber still reflects a preceding (unobserved) low-state in intrinsic flux.
Alternatively, and more likely, gas heated by the central continuum source may have crossed the line of sight , producing the steep RASS spectrum, and has (nearly) disappeared in the 1992 observation. This scenario explains most naturally the nearly constant countrate from RASS to pointed observation, because the countrate is dominated by the soft energy part of the spectrum (below 0.7 keV) which is essentially unaffected by warm absorption. The transient passage of a BLR cloudlet would be consistent with the scenario proposed by Rodriguez-Pascual et al. (1997) who suggested a matter-bounded BLR in NLSy1s on the basis of emission line profiles and strengths. They derive a lower column density for the BLR clouds but since our best-fit X-ray warm absorber also has a higher ionization parameter, the hydrogen Strömgren sphere is shifted further out and thus the clouds remain matter-bounded. 4
The short duration of the RASS observation has to be kept in mind, and both, an intrinsically steep powerlaw and a strong soft excess fit the X-ray spectrum as well. Variability in only one component seems to be problematic, though, since the nearly constant countrate has to be accounted for.
A spectral change with constant countrate is reminiscent of one class of Galactic black hole transients (the one in which both spectral components change simultaneously as to mimic constant countrate; e.g., Tanaka 1997). In fact, the potential similarity of NLSy1s with Galactic black hole candidates has been repeatedly mentioned (starting with Pounds et al. 1995), but has never been explored in more detail. We do not follow this one further, since the analogy between NLSy1s and Galactic black hole candidates does not seem to go very far (e.g., p. 411 of Brandt & Boller 1999).
Finally, it is also possible that the constant countrate is pure coincidence: Both, variable soft excesses (see, e.g., 1E1615+061 for an extreme example) and variable powerlaws (often of constant shape) have been observed in AGN and these two might have compensated each other to produce nearly constant total countrate (this seems to be the model favored by Grupe 1996; their Fig. 8.11).
5.3. NGC 4051
The variability amplitude of NGC 4051, a factor 30 over the measured time interval, is fairly large. During all individual ROSAT observations, the source is variable. No long-term very low state as recently reported by Guainazzi et al. (1998a) occurred.
The warm absorber properties, averaged over individual observations, are found to be quite constant, with changes less than about a factor of two in column density and ionization parameter. This is consistent with the finding of Komossa & Fink (1997a) that the bulk of the warm material in NCC 4051 does not react to short-timescale changes in the ionizing luminosity and thus the bulk of the ionized absorber must be of low density, or alternatively, the warm material is out of photoionization equilibrium. The latter possibility was also suggested by Nicastro et al. (1999) based on shorter-timescale variability behavior of NGC 4051. Recently, Contini & Viegas (1999) presented detailed multi-wavelength modelling of NGC 4051, including in their models the presence of shocks besides photoionization.
5.4. Mrk 1298
It is interesting to note that Mrk 1298 exhibits all characteristics of a NLSy1 galaxy except that its observed FWHM of H, 2200 km/s, just escapes the criterion of Goodrich (1989).
A warm absorber fits well the X-ray spectrum of this galaxy, whereas a powerlaw plus black-body-like soft excess does not. This also holds for further possible shapes of the soft excess.
In the following, we give some predictions made by the warm absorber scenario, in terms of line emission and absorption. Depending on the covering factor of the warm absorber, the ionized material might contribute to high-ionization emission lines in the optical-EUV spectral region (e.g., Komossa & Fink 1997a,d,e). Among the strongest predicted lines are [FeXIV]5303/H = 3, (NV1240+FeXII)/H = 13 and OVI1035/H = 284. However, the warm absorber is matter bounded and the total emissivity in H is fairly small when compared to the observed H-luminosity of erg/s (Rafanelli & Bonoli 1984; the value given in Miller et al. 1992 is a factor 1.6 lower). Scaled to observed H, the strongest predicted line is 0.4.
WBB96 mention the presence of UV absorption lines in an IUE spectrum of Mrk 1298. The following equivalent widths of UV absorption lines are predicted for the best-fit warm absorber model (see also the discussion in Wang et al. 1999): / in CIV and Ly and / in NV (adopting a velocity parameter b = 60 km/s). This is assuming all spectral steepness is indeed caused by the warm absorber. If an additional soft excess is present (note that just a powerlaw plus black-body-like soft excess does not fit the ROSAT spectrum; but in case more than two spectral components are allowed, fits are not well constrained due to the limited PSPC spectral resolution) the contribution from the warm absorber would be less, since a lower column density would be inferred from spectral fits.
5.5. 4C +74.26
There is growing evidence that several (but not all ) warm absorbers contain dust. The reported individual cases are IRAS 13349+2438 (Brandt et al. 1996, Komossa & Greiner 1999, Komossa et al. 1999b), NGC 3227 (Komossa & Fink 1997b, George et al. 1998), NGC 3786 (Komossa & Fink 1997c), MCG 6-30-15 (Reynolds et al. 1997), IRAS 17020+4544 (Leighly et al. 1997, Komossa & Bade 1998).
The advantage of invoking dust mixed with the warm absorber in 4C +74.26 is the steeper intrinsic powerlaw spectrum ( ) required to compensate for the `flattening effect' (Komossa & Fink 1997b) of dust, as compared to a single powerlaw fit which gives a peculiarly flat spectrum ( ).
A steep intrinsic spectrum has the advantage of being close to the ASCAhard-energy value of derived for this source (Brinkmann et al. 1998) and the general expectations for nearby radio-loud quasars in the ROSAT band ( ; e.g., Schartel et al. 1996a, Brinkmann et al. 1997, Yuan 1998). Better spectral resolution soft X-ray data are needed to distinguish between both possibilities, an intrinsically flat powerlaw, or a dusty warm absorber.
5.6. X-ray spectral complexity in NLSy1 galaxies
Based on limited spectral resolution in the soft X-ray band, early models attempted to explain the X-ray spectral steepness of NLSy1s with one component only; either (i) a single steep powerlaw, or (ii) a strong soft excess on top of a flat powerlaw (e.g., Puchnarewicz et al. 1992, BBF96), in analogy to Seyferts (e.g., Walter et al. 1994) and quasars (e.g., Schartel et al. 1996a,b, 1997b) which were believed to have soft X-ray excesses, or (iii) heavily warm-absorbed flat powerlaws (e.g., Komossa & Greiner 1995, Komossa & Fink 1997a,d,e). One of the comfortable properties of both, the soft excess plus flat powerlaw and the warm-absorbed flat powerlaw interpretation, is the presence of enough X-ray photons to account for the strong observed FeII emission in NLSy1s. It does not immediately explain the occasionally observed trend of stronger FeII in objects with steeper X-ray spectra, but it is interesting to note that Wang et al. (1996) find a trend for stronger FeII to preferentially occur in objects whose X-ray spectra show the presence of absorption edges.
However, there were early indications of spectral complexity of NLSy1s (e.g., Brandt et al. 1994, Komossa & Fink 1997a,d; see also Vaughan et al. 1999). E.g., a detailed study of NGC 4051, a bright source for which several deep ROSAT observations were performed, revealed all three of the spectral components to be simultaneously present: The spectral steepness is dominated by the warm absorber, but the index of the underlying powerlaw can become as steep as 5 and an additional soft excess ist present in source high-states (Pounds et al. 1994, Komossa & Fink 1997a; the latter authors additionally provided evidence for an EUV bump component based on photon counting arguments). Further complications concerning the X-ray spectra of NLSy1s have emerged recently, via the suggestion of dusty warm absorbers in some NLSy1-like galaxies (Brandt et al. 1996, Leighly et al. 1997, Komossa & Bade 1998).
We have examined two further NLSy1s observed with ROSAT, 1ZwI and PHL 1092, and find that neither a flat powerlaw with soft excess nor a warm absorber can account for most of the spectral steepness. The sources are best described by a single steep spectral component.
As suggested by Komossa (1997) these spectral components of NLSy1s may be linked in the sense that a more polar view on the accretion disk (e.g., Fig. 3 of Madau 1988) causes the soft excess component to be more pronounced, while along the funnels of the disk outflows are driven which cause the characteristic absorption edges of warm absorbers if viewed along the line-of-sight against the continuum source.
Except for NGC 4051 the faintness of the objects of the present study (and many other ROSAT observed ultrasoft sources) does not allow to perform n-component spectral fits. However, the one-component models presented here still provide upper limits on the contribution of each single component (steepness of powerlaw, strength of soft excess, column density of warm absorber).
Given the few cases that have been observed with high X-ray spectral resolution and sufficient countrates, other approaches to distinguish between different EUV-X-ray spectral shapes are important. Such an approach is described in the following section.
5.7. EUV - soft X-ray spectral shape and stability of broad line clouds
One suggestion to link the apparently steep soft X-ray spectra with the small FWHM of the broad lines in NLSy1s was the influence of the X-ray spectral shape on multi-phase BLR cloud equilibrium, especially the hindrance of BLR formation due to illumination by a steep X-ray spectrum (Brandt et al. 1994). Here, we test this suggestion based on calculations carried out with the code Cloudy . In particular, we investigate how different EUV-X-ray spectral shapes change the range in which a multi-phase equilibrium is possible and attempt to distinguish, within the limits of this scenario, between different suggested spectral models.
The thermal stability of broad line clouds can be examined by studying the behavior of temperature T in dependence of pressure, i.e., (e.g. Guilbert et al. 1983, their Fig. 1; for a general discussion see also Krolik et al. 1981, Netzer 1990, Reynolds & Fabian 1995, Komossa & Fink 1997a,d). In case T is multi-valued for constant , and the gradient of the equilibrium curve is positive, several phases may exist in pressure balance.
To test this idea, we have calculated equilibrium curves for the BLR gas for intrinsically steep X-ray spectra or spectra with a black-body-like soft excess. Additionally, the metal abundances were varied, which affect the cooling of the gas. Results are shown in Figs. 7, 8. The parts of the equilibrium curve with negative gradient correspond to thermally unstable equilibria. It is well known that the original Krolik-McKee-Tarter models face problems when a more realistic continuum shape is used, in the sense that a pressure balance between a cold, photoionization heated, and a hot, Compton heated phase no longer exists (see Fig. 7, e.g. = -1.9 compared to = -1.5). Reynolds & Fabian (1995) point to the existence of an intermediate-temperature stable region where is multi-valued. We find that this intermediate region disappears for steep X-ray spectra. 6 Due to the relatively weaker X-ray flux and the fact, that the value of U is dominated by the EUV flux near the Lyman limit, the gas remains longer in the `photoionization-heating, collisional-deexcitation-cooling' phase. The same holds for a continuum with a hot soft X-ray excess (Fig. 8).
These studies reveal that several spectral shapes and gas metal abundances lead to similar results, but show that the mechanism suggested by Brandt et al. (1994) works in general. More detailed models should then await better knowledge of the 0.1-10 keV X-ray spectral shape.
© European Southern Observatory (ESO) 2000
Online publication: February 9, 2000