Astron. Astrophys. 356, 445-462 (2000)

3. The radio to X-ray properties of the objects

The sample of 843 "singly matched" objects contains 241 objects with existing spectroscopic classifications while classifications for an additional 217 sources were obtained through follow-up observations of the FIRST survey or specifically as part of this program. The remaining 385 sources remain spectroscopically unclassified. A more detailed break down into the various classes is given in Table 3.

Table 3. Source contents of the sample

According to the classification scheme given above, most of the new identifications listed as `others' are starburst galaxies.

In Fig. 5 we plot the 1.4 GHz radio luminosity of all spectroscopically classified sources as function of redshift. The open circles denote previously known objects, the full dots the newly classified sources from the sample. Many of the objects are close to the lower radio luminosity limit as can be seen from a comparison with Fig. 1 and many of the new identifications cluster around a redshift of 0.15. Further, there is a substantial number of `known' objects close to the luminosity limit which are not present in Fig. 1. These are previously known objects for which radio counterparts had not previously been identified.

Fig. 5. 1.4 GHz radio luminosity as function of redshift for the spectroscopically classified sources. Open circles are previously known objects, solid dots the newly classified objects.

Fig. 6a-f. Sample new spectra for the RASS-FIRST correlation. One object from each major class is shown. From top, a Quasar, b BL Lac, c Broad-line AGN d Narrow-line AGN, e Starburst, f Galaxy.

3.1. Flux distributions

In Fig. 7 we show the distribution of the objects as function of their fluxes in the different wavelength bands; (top: peak 1.4 GHz flux density (in mJy); middle: soft X-ray flux; bottom: optical magnitude m_E, obtained from from the POSS plates).

Fig. 7. Distribution of RASS/FIRST sources as function of their fluxes in different wavelength bands. From top to bottom: number of objects per radio flux density, soft 0.1 - 2.4 keV X-ray flux and optical magnitude. The open histograms denote the total sample, the grey-shaded areas indicate previously optically classified objects; the hatched areas represent the newly classified objects.

The top line represents all objects in the sample, the grey shaded areas are the previously known objects, and the hatched regions represent the newly classified objects. The sharp decline in the number of objects at low radio and X-ray fluxes is a direct consequence of the sensitivity limit of the RASS and the corresponding distribution (Fig. 2 of B95) which shows that the X-ray sensitivity is insufficient to detect radio sources at all flux densities at the same rate. Clearly visible are the identification biases with respect to the observed fluxes. This effect is particularly strong for the radio fluxes as most of the sources with fluxes above a few hundred mJy have been identified previously. In the optical band, more than half of the unclassified objects are fainter than 18th magnitude. The figure thus directly reflects previous biases towards identification of stronger sources and demonstrates the importance of sensitive large scale sky surveys for the study and characterization of multi-wavelength class properties of sources.

3.2. X-ray properties

Because the average Survey exposure on a source is rather low ( s), formal spectral fits can be attempted for only the strongest sources. Fortunately, the low background of the PSPC detector allows an approximate determination of the X-ray spectral parameters from a relatively small number of photons. Spectral fits were therefore obtained for the majority of sources using the hardness ratio method. This method uses the the hardness ratios provided by the SASS processing, and maps them onto power law slopes, assuming either free or Galactic absorption (for details of the method see B95). This allows an approximate spectral determination for objects with count rates as low as 0.03 cts/s or even less if the exposure is correspondingly higher. If Galactic absorption is assumed, this leaves only one free parameter in the procedure which is then equivalent to a least squares fit of the underlying power-law slope.

3.2.1. Spectral properties

The results of a maximum-likelihood analysis for the distribution of power-law slopes assuming Galactic for the previously known and the newly classified sources are given in Table 4 (for details of the analysis see Maccacaro et al. 1988, Worrall & Wilkes 1990). The 90% confidence contours for the various classes as a function of spectral index and intrinsic dispersion are shown in Fig. 8.

Fig. 8. Best-fit mean spectral index from 0.1 - 2.4 keV and Gaussian standard deviation for power-law fits to different classes of objects assuming Galactic absorption. Contours correspond to 90% confidence levels. Upper panel: previously known objects; lower panel: newly classified sources.

Table 4.

A significant intrinsic dispersion is an indicator for either the inhomogeneity of the sample; i.e., it consists of different subclasses with different spectral properties grouped together into one larger class (for example for the `Low Luminosity AGN' group which contains Seyferts and Broad and Narrow line Radio galaxies), or that the individual sources show an intrinsically large dispersion of their spectral properties, perhaps showing intrinsic absorbers, Compton scatterers or other components. Due to the limited photon statistics for most of the sources, we cannot apply more complex spectral models to the data.

The photon indices of the previously known and newly classified sources appear the same to within the uncertainties. The average index of the unclassified objects indicates that these sources are mostly a mixture of quasars and BL Lacs. The Seyfert class includes all Seyfert sub-classifications in NED and, for the newly classified sources, all broad and narrow-line objects, as given in Table 1. This nonspecific classification of all types into one group likely leads to the large dispersion of their spectral index distributions.

In contrast to previous studies of ROSAT - radio correlations (Brinkmann et al. 1994, B95), the galaxy class is characterized by flat spectral indices. This is likely caused by a number of galaxies being members of an X-ray emitting cluster. While the BL Lac objects have a spectral index distribution very similar to those of the RASS-Green Bank (RGB) sample (see Laurent-Muehleisen et al. 1998, Brinkmann et al. 1997b), the quasar distribution is shifted to steeper slopes and shows a larger dispersion. It appears that, due to the low detection limit of the radio fluxes, we see a smooth transition of the quasar population between radio-loud objects with flat spectral indices (Brinkmann et al. 1997a, B97) and radio-quiet objects with on average considerably steeper spectra (Yuan et al. 1998). The narrow 90% confidence contour (the small error of ) for the unclassified sources must be primarily related to the large number of objects.

3.3. The diagram

Flux ratios combining data from the radio to the X-ray have been extensively used for classification of extragalactic objects (e.g., Tananbaum et al. 1979, Stocke et al. 1991). Based on two-point spectral indices, the radio-to-optical and optical-to-X-ray , it has been shown that different classes of objects typically populate different regions of the diagram.

For the construction of the diagram, the 1.4 GHz VLA flux densities were converted to 5 GHz flux densities by assuming a power law , with slope . For the monochromatic optical fluxes at 2500 Å we took the E-magnitudes from the POSS plates and an optical power-law slope . The monochromatic X-ray fluxes at 2 keV were computed from the (0.1 - 2.4 keV) fluxes assuming a power law with an average photon index of . The adoption of this mean value allows us to avoid the large scatter of the spectral indices introduced by the limited photon statistics, although in some cases it may result in an incorrect flux determination if the actual spectral index of a source truly differs from the mean value.

In contrast to similar diagrams (B95, Brinkmann et al. 1994), the fluxes are not K-corrected as the types and the redshifts of many of the objects remain unknown. The shift in phase space expected from the K-corrections for, e.g., an object at z = 1 with the above typical quasar power-law slopes is, however, rather small: , while the value of does not change for the assumed spectral indices.

The majority of known objects are found along the diagonal swath from high- and low- to low- and high-, the region generally occupied by radio-loud quasars and blazars, and, at large values, by bright galaxies with low X-ray and radio emission. Many of the unclassified objects are found in the region of classical X-ray selected BL Lacs, i.e., at and , although a large fraction of them have indicating a more intermediate nature. In contrast to the RGB sample, the upper left of the diagram remains empty, i.e., there are no objects with unusually low optical fluxes but with strong radio- and X-ray emission (so-called "Optically Quiet Quasars", see Kollgaard et al. 1995).

Most of the classified objects are found in the region of phase space typical for their class; few sources appear to be misplaced, possibly due to questionable classification, incorrect association of the radio/optical/X-ray sources, or spectral properties largely different from the average values for a particular subclass.

3.4. Flux ratios

Because we lack redshifts for many sources, we cannot employ luminosity correlations to study the bulk emission properties of the sample. Instead, flux ratios provide distance-independent measures of emission characteristics, neglecting K-corrections. In Fig. 10 we show from top to bottom the flux ratios versus for quasars, BL Lacs, galaxies, and the spectroscopically unclassified objects, respectively. The open symbols represent objects which have been classified previously, although not necessarily identified with a radio counterpart. For comparison with previous papers, we have used the monochromatic X-ray fluxes at 2 keV, optical fluxes at 2500 Å, and radio fluxes at 5 GHz.

Fig. 9. Broad band energy distribution of all sources. Spectroscopically unclassified objects are marked as bullets, others (plus signs) are all objects in the sample not belonging to one of the indicated classes.

Fig. 10. Logarithmic flux ratios versus for quasars (top panel), BL Lac objects, galaxies, and spectroscopically unclassified objects (bottom panel). Open and filled symbols represent previously and newly classified sources, respectively.

While many of the previously known quasars occupy the phase space of classical radio-loud objects (see Fig. 16 of B97), most of the newly classified quasars are found in the `radio-intermediate' transition region between radio-loud and radio-quiet quasars (see W00). A similar situation holds for the BL Lac objects where the newly classified sources reside primarily in an intermediate region between the HBL (High energy peaked BL Lacs) and LBL (Low energy peaked, Giommi & Padovani 1994) classes. These Intermediate BL Lacs have been found previously in the RGB survey (Laurent-Muehleisen et al. 1999), as well as in the DXRBS (Deep X-ray Radio Blazar Survey; Perlman et al. 1996) and REX (Radio Emitting X-ray; Caccianiga et al. 1999) surveys. A large number of unclassified objects inhabit the same region of phase space as the intermediate BL Lacs and thus it is expected that many of these objects are BL Lac objects or extreme flat-spectrum quasars.

Most of the galaxies were previously known. Three objects with extreme flux ratios at the left at low ratios might be misidentifications: a classification for one (RXS J1504.5+2854) is based on a low signal-to-noise spectrum while the other two (RXS J1317.3+3925 and RXS J1625.5+2705) have X-ray luminosities of erg and erg s^-1, respectively, far in excess of that expected from a galaxy. These "optically passive X-ray galaxies" have been seen in a number of other surveys including the RGB (Laurent-Muehleisen et al. 1998), the Einstein Two-Sigma catalog (Moran et al. 1996) and some deep ROSAT PSPC fields (Griffiths et al. 1995). These sources are good candidates for new clusters, since that could easily account for the large X-ray luminosity associated with these optically unremarkable sources, although other explanations including hidden AGN or early type galaxies with an extraordinarily hot ISM are also possible.

© European Southern Observatory (ESO) 2000

Online publication: April 10, 2000
helpdesk.link@springer.de