2. The QSOs and Seyfert 1 galaxies
2.1. General properties
According to Krautter et al. (1999) 228 sources in our Catalog (about 34% of the total) were identified as Seyfert 1s or QSOs. Apart from stars (37%) these objects form the largest subgroup in our sample. In the Catalog we list the X-ray flux , the visual magnitude V, the X-ray/visual index and the redshift z. From these data we calculated approximate absolute visual magnitudes (assuming , = 0.5, = 0 and negligible interstellar extinction). The distribution of the resulting values is plotted in Fig. 1. The redshift distribution of our Seyfert 1s and QSOs is given in Fig. 2. As shown by the figures, our sample is dominated by moderate-luminosity objects () with an average redshift = 0.40. Only 8% of our Seyfert 1s and QSOs have redshifts 1.0. Only one QSO (RX J1028.6-0844, z = 4.28, cf. Zickgraf et al. 1997b) has a redshift 2.2. The shape of the distributions in Fig. 1 and Fig. 2 indicate that our sample provides an essentially complete inventory for the local () AGN population while for higher redshifts only the progressively rarer objects with high (X-ray) luminosity are detected by the RASS.
In Fig. 3 we present the distribution of the X-ray-visual index , which provides a measure for the relative strength of the X-ray and visual emission of an object. For AGN is approximately linearly related to the index, which is used for the same purpose in part of the literature (see e.g. Stocke et al. 1991). For our ROSAT data we have with good approximation . The accuracy of this relation for our data can be estimated from Fig. 4. Our values are of the same order as those found in other AGN surveys (e.g. Stocke et al. 1991).
Since our sample is X-ray flux-limited, we have to expect a selection effect in the sense that apparently faint objects with low will not be included in our survey and only very high objects are observed at the faint end. As shown in Fig. 5, this selection effect is clearly present. However, as shown by Fig. 6, no such effect is conspicuous in our -redshift relation, since (apart from the very bright objects, which are all at small redshifts) there exists no strong correlation between redshift and apparent brightness in our sample. Therefore, the observed distribution is probably characteristic for the RASS AGN independent of the redshift. (The presence of 3 low-redshift objects with in Fig. 6 is probably caused by an overestimate of the luminosity of these relatively faint AGN due to a contamination of the photometry by their host galaxy light. Hence, these 3 values are probably lower limits only).
The characteristic property of Seyfert 1s and QSOs is the presence of broad (BLR) emission lines. While the instrumental resolution (corresponding to about 750 kms-1) did not allow us to resolve the forbidden line profiles, the BLR profiles were usually well resolved and intrinsic line widths 500 kms-1 could normally be detected from the broadened profiles. In the Standard Model of AGN the widths of the BLR lines are assumed to be caused by the motions of the line emitting plasma in the potential of the central black hole. Since the distance of the BLRs to the central continuum source can (in principle) be estimated using reverberation techniques, the line widths and their distributions provide important information on the central masses and mass distributions. A direct comparison of all BLR line widths in our sample is complicated by the redshift range of our spectra. In order to allow a direct and unbiased comparison, we, therefore, had to restrict our analysis to a spectral region which is common to at least most of our spectra. Best suited for this purpose turned out to be the region of the H line.
Depending on the position of the object in the field and the of the spectra this line falls into our observed spectral range for redshifts of about 0 z 0.8. For all objects in this range with spectra of adequate we measured the FWHM and FWZI line widths of H. Since the quality of our spectra is not sufficient to allow a reliable decomposition of the profiles into various components, the FWHM measurements refer to the full line profiles, including broad and narrow components. (The FWZI widths measure the BLR components only, but, due to difficulties defining the continuum level, FWZI values are normally less reliable). The resulting FWHM distribution is plotted in Fig. 7. (The FWZI distribution is broader by about a factor of 2 but qualitatively similar).
The distribution in Fig. 7 shows a relatively large fraction (18% 3%) of objects with FWHM 2000 km-1. However, a comparison with the literature indicates that this fraction is not unusual for AGN. Stephens (1989) finds for a small () X-ray selected sample 24%. In the optically selected sample of Boroson & Green (1992) the corresponding fraction is 23% 5%, which agrees within the error limits well with our X-ray selected sample. If the 6 narrow-line objects with [O III ]/H mentioned above are added to our Seyfert 1 sample, our FWHM 2000 kms-1 fraction increases to 21% 3%, providing an even better agreement with Boroson and Green. The fact that Puchnarewicz et al. (1992) find in a sample of 17 Seyfert 1s with ultra-soft X-ray spectra 9 objects (53%) with H FWHM 2000 kms-1 may indicate a relation with the X-ray spectral index. However, because of the limited X-ray spectral information for our objects (see below) this relation cannot be tested with our data. Most of the objects with FWHM 2000 km-1 also have BLR components with larger FWHM. Only three objects classified as Seyfert 1s in our Catalog are NLS1 galaxies without detectable Balmer line components of FWHM 2000 km-1, but with strong Fe II emission. The presence of many objects with strong narrow Balmer components and weak broad components argues for a smooth transition between the NLS1s and other Seyfert types. (The fact that Engels & Keil (2000) in their analysis of a different sample of X-ray selected AGN find a higher fraction of NLS1s is probably due to different classification or selection criteria).
As shown by Fig. 7, the line widths cover a range exceeding a factor of 10. The broadest H line was observed for the object RX J1021.6-0327 = Akn 241 (FWHM = 9600 km-1, FWZI = 17 900 km-1). The spectrum of this object seems to show some other spectral peculiarities, which have to be studied with better and higher resolution, however.
According to the AGN Standard Model the line width distribution of the broad lines can be caused (a) by variations of the depth of the gravitational potential of the line forming region or (b) by variations of the orientation of the rotation axis relative to the line of sight to the observer. For disklike rotating emission regions with uniform velocities the theory predicts for (b) a distribution with a minimum at low velocities and a maximum and cutoff at the high velocity limit. The distribution in Fig. 7 is obviously very different, indicating that the line widths variations are probably dominated by intrinsic orbital velocity differences (i.e. variations of the potential) of the BLRs.
Except for the hydrogen and helium lines Fe II multiplets are normally the most conspicuous emission features in the visual spectra of the Seyfert 1s and QSOs. Their strength is normally measured by the Fe II index (flux ratio) = Fe 4570Å /H. Unfortunately our spectra were normally not of sufficient quality to measure the Fe II 4570Å blend directly. On the other hand, for 63 objects it was possible to derive the total strength of the Fe II (37,38) blends near 4570Å and the Fe II (48,49) blend near 5300Å. Assuming that the relative strength of the Fe II multiplets is constant, we converted these measurements to approximate values using the well observed (strong-Fe II ) Seyfert 1 galaxy I Zw I (Phillips 1977; Boroson & Green 1992) for calibration. As in other Seyfert 1 samples the great majority (80%) of our values fall into the interval 0.1 - 1.0. Our mean value 0.7 is somewhat higher than the normally quoted average for Seyfert 1s (0.4, Osterbrock 1977; Bergeron & Kunth 1984), although this difference is not significant in view of our approximate method and the size of our sample. Nevertheless, the fact that the average Fe II emission is certainly not lower in our X-ray selected sample than in normal Seyfert 1 galaxies seems to argue against the result of Lawrence et al. (1997), who (on the basis of a smaller sample) find the Fe II emission to be anticorrelated to the X-ray emission in Seyferts. As pointed out below, we also found no anticorrelation (or correlation) between and and for our sample.
One object in our sample (RX J0757.0+5832) shows (as already noted in the Catalog) exceptionally strong Fe II emission ( 2.05).
As pointed out e.g. by Dahari & De Robertis (1988) there are few strong correlations between different AGN properties. Hence it was no surprise that we found (apart from trivial relations, such as between FWHM and FWZI) few correlations in our data. In particular we find no significant correlation between the absolute visual brightness and the H line width, although such a correlation seems to be present in other AGN samples (e.g. Miller et al. 1992). Our Fig. 8, showing the observed relation, present essentially a random scatter apart from the fact that the few objects with H FWHM 8000 km-1 all have luminosities below , while the most luminous QSOs show moderate line widths. Within the Unified AGN Model this could perhaps be explained assuming that the high-FWHM objects are likely seen edge-on with the central light source partially obscured by a circumnuclear dust torus. However, in this case we may expect to find redder than average values and different values for the high line width objects. Since this is not observed, we conclude that the high line width of low-luminosity objects in Fig. 8 is probably not caused by an inclination effect.
In Fig. 9 we plotted the Fe II emission strength (expressed in ) as a function of the H FWHM line widths for all those objects where both these quantities could be measured. Our plot confirms the well known anticorrelation between Fe II emission and BLR line widths for Seyfert 1s and QSOs (see e.g. Zheng & Keel 1991; Wang et al. 1996; Lawrence et al. 1997). On the other hand, in contrast to Lawrence et al. (1997), we found in our sample no indication of any correlation between the Fe II emission strength and the X-ray loudness (as expressed by ).
As pointed out e.g. by Mushotzky et al. (1993), luminous AGN normally tend to show lower (or steeper ) values than low-luminosity objects. As shown in Fig. 10 this correlation is also indicated in our data. However, apart from the large scatter in our data, the relation derived here is probably affected (i.e. weakened) by the selection effect demonstrated in Fig. 4.
Boller et al. (1996), Laor et al. (1994), Grupe et al. (1999), and others pointed out a correlation between the ROSAT spectral index and the H line width for Seyfert galaxies and QSOs. Our relatively large sample of Seyfert galaxies with ROSAT X-ray data provide in principle a possibility to study this correlation. Therefore, we calculated the ROSAT photon index for all our objects. Unfortunately for most of our objects the photon counts turned out to be much too low to derive with an acceptable accuracy. For only 13 objects with good H data we were able to determine with a mean error 1.0. These data, plotted in Fig. 11, are consistent with the known anticorrelation between the BLR line widths and .
Most of our spectra are not of sufficient quality to detect the weak forbidden high ionization lines (FHILs) or "coronal" lines of the Seyfert spectra. In only 5 objects in our sample the [Fe X ] lines were strong enough to be visible on our spectra. Interestingly, two of these 5 objects (RX J0707.2+6435 and RX J1218.4+2948) are also among the 3 objects with , supporting the existence of a correlation between and the FHIL strength, as proposed by Erkens et al. (1997).
© European Southern Observatory (ESO) 2000
Online publication: January 29, 2001