4.2. Soft X-ray Surveys
The Einstein and, especially, the ROSAT surveys have provided very large samples of soft X-ray-selected AGN. There have been extensive discussions of samples obtained by these missions (see Puchnarewicz et al. 1996 for a summary of the pointed observations, and Fischer et al. 2000 and Zickgraf et al. 2003 for the survey data). At moderate X-ray fluxes (> 10-13 ergs cm-2 s-1), there is only ~ 1 X-ray source per square degree, and the ROSAT PSPC 6 had sufficiently small positional errors that unique identifications could be made for the sources brighter than m ~ 20 mag in the B or V bands on the basis of the X-ray positions alone (see Fig. 4). However, at lower fluxes, more accurate positions and better angular resolution are often required, and many of the identifications were made on the basis of finding a broad-line object inside the X-ray error circle. Historically, this is how the previous large soft X-ray survey with Einstein (the EMSS; Gioia et al. 1990) obtained optical counterparts. At faint X-ray fluxes (< 10-14 ergs cm-2 s-1), very long observations were required to detect the sources (the ROSAT Deep Surveys; Hasinger et al. 1998). It is historically interesting that the identifications of most of the ROSAT Deep Survey sources were just at the limit of the capabilities of the largest ground-based optical telescopes, with the optical identification of the survey being almost complete at R ~ 23 mag (Hasinger et al. 1999).
Figure 4. X-ray contours from the ROSAT All-Sky Survey superimposed on a field from the Palomar sky survey for the quasar S51027+74. Note that the ROSAT contours are small enough for a high-probability optical identification of the source.
Soft X-ray surveys find that there is a moderate correlation of optical and X-ray properties (see Fig. 5) with a relatively narrow (± 1 order of magnitude) range in X-ray-to-optical flux ratios, that the X-ray evolution of AGN is similar to that of optically-selected AGN (see Miyaji, Hasinger, & Schmidt 2000), and that most of the objects are broad-line AGN (see Hasinger et al. 1999 for a review). In the shallow ROSAT All-Sky Survey (Appenzeller et al. 2000), the median redshift is z ~ 0.2, and there are only two objects at z > 2. This is to be contrasted with the "shallow" optical Bright Quasar Survey (BQS; Schmidt & Green 1983), which has a much flatter redshift distribution out to z ~ 2. However, there is a significant fraction of unusual AGN, some of which have rather red continuum colors and broad optical emission lines (Puchnarewicz & Mason 1998).
Figure 5. Correlation of the ROSAT All-Sky Survey counting rate and optical magnitude for a large sample of active galaxies. Note that the majority of objects lie in a narrow band in magnitude vs. X-ray counts space.
As discussed in Section 3, there are significant selection effects in soft X-ray surveys caused by obscuration in the line of sight. Perhaps the most direct example of this is the analysis of ROSAT observations of the Piccinotti et al. (1982) hard X-ray survey (Schartel et al. 1997), which shows that approximately half of all the hard X-ray sources are absorbed and thus have significantly lower soft X-ray fluxes - up to 300 times less - than expected if the spectra were not absorbed. A comparison of the hard X-ray properties of a soft X-ray-selected sample has not been done because of the limited sensitivities of large solid angle hard X-ray surveys.
As one goes to higher redshifts, the effect of obscuration in the X-ray band decreases as the spectrum redshifts (Barger et al. 2001), so the observed 0.5 - 2 keV band is much less affected by obscuration effects. This seems to be a rather important effect for the Chandra sources, and thus the ROSAT soft-band data need to be corrected for the changing effective bandpass.
ROSAT also performed a survey in the softest X-ray bands at 0.25 keV (Vaughan et al. 2001). Because of the effect of obscuration by the galactic column density in this low-energy range, only ~ 0.6 sr could be covered. The sample appears to be complete. The identifications are rather different than in other AGN surveys, with one-third of the AGN identifications being BL Lacertae objects and one-quarter being narrow-line Seyfert 1 galaxies. These are relatively large percentages for these two classes of objects compared with surveys in any other wavelength band. Such a survey has a bias towards these objects because of their steep X-ray spectra. The almost disjoint nature of the objects contained in a hard (see Section 4.3) and extremely soft X-ray selection is a warning to the completeness of any survey.
6 The PSPC is the Position Sensitive Proportional Counter, the workhorse instrument on ROSAT with a 15" spatial resolution and 3" - 5" positional errors. Back.