5 Broad band source properties

5.1 Comparison with shallower radio-X-ray samples

Papers I and II constructed a large sample of radio-loud X-ray objects from the correlation of the RASS survey with the 5 GHz Green Bank survey. The resulting RGB sample is an order of magnitude less sensitive in both spectral bands than the NEP sample obtained here, although it covers a much larger area of the sky. The relationship between the two surveys is illustrated in Fig. 6; the RGB histograms are drawn from Fig. 4 of Paper II. Two features can be noted. First, while the X-ray fluxes of the NEP sources are confined to about 1.5 orders of magnitude, the radio flux densities are spread over 3 orders of magnitude. Any correlations between the quantities must show considerable scatter. Second, no bimodality is seen in the radio flux density distribution as is seen, for example, in the optically-selected quasar samples of Strittmatter et al.(1980) and Kellerman et al.(1989). This result is consistent with the findings of Papers I and II which, however, did not significantly sample the faint end of the radio-loud AGN population in either radio luminosity or the ratio of radio-to-optical emission. Bimodality is evident in the NEP sample in the distribution of the ratio of radio-to-optical emission, as measured by the radio-to-optical spectral index, $\alpha_{\rm ro}$. As discussed in the next section, the observed bimodality is consistent with optically-selected quasar samples (e.g. Kellerman et al.1989, 1994, and Stocke et al.1992) and the X-ray-selected AGN sample of Della Ceca et al.(1994).

  

Figure 7: X-ray flux distribution of all the NEP sources in the inner $3^\circ \times 3^\circ$ field (open diagram) and the radio-loud sub-sample (hatched area)

Figure 7 shows the integrated 0.1 - 2.4 keV flux distribution of all RASS sources in the central 3$^\circ \times$3$^\circ$ field of the NEP and the distribution of the radio-detected subgroup. The radio-loud group has a slightly higher average X-ray flux than the parent population, but radio-loud sources are found at all X-ray flux levels. At the higher flux levels of the RGB samples, a similar uniform distribution of radio-loud objects is seen (Paper I, Fig. 6). The absolute detection rate in the NEP field is similar to that of other deep X-ray observations (de Ruiter et al.1994; de Ruiter et al.1997; Warwick & Barber 1992) and considerably higher than for the RGB survey (Papers I and II).

5.2 The $\alpha_{ro}-\alpha_{ox}$ diagram

The broad-band $\alpha_{\rm ro}-\alpha_{\rm ox}$ color-color diagram has been extensively used to display broad-band spectral energy distributions and to classify extragalactic objects. The radio-to-optical index is defined as $\alpha_{\rm ro} \equiv$ $\log( S_{\nu_{{\rm rad}}} / S_{\nu_{{\rm opt}}}) / \log(\nu_{{\rm opt}} /$ $\nu_{{\rm rad}})$ and the optical-to-X-ray index as $\alpha_{\rm ox} \equiv$ $-\log( S_{\nu_{{\rm x}}} / S_{\nu_{{\rm opt}}}) / \log(\nu\rm _x / \nu_{{\rm opt}})$.Here we use the radio fluxes at $\nu_{{\rm rad}} = 1.5$ GHz, the optical data at 4500 Å and the X-ray data at 2 keV. While it has been widely asserted that different classes of objects occupy distinct locations in this diagram, we note that the boundaries can become less distinct as surveys become deeper in all bands (see Fig. 10 of Paper II). Nonetheless, the diagram is an effective tool for examining the spectral energy distributions of objects without redshifts such as the NEP population discussed here.

  

Figure 8: Broad band energy distribution of all sources. The axes are the radio-to-optical and optical-to-X-ray spectral indices as defined in the text. Arrows indicate objects with upper limits in the optical magnitude

The $\alpha_{\rm ro}-\alpha_{\rm ox}$ diagram for the 74 sources in Tables 1 and 2 is shown in Fig. 8. It was constructed in a similar manner to Fig. 10 in Paper II with the following exception. In cases where the radio source appears to be multiple but the peaks were individually cataloged in Kollgaard et al.(1994), only the integrated flux density of the dominant component is used. The monochromatic X-ray fluxes at 2 keV were computed from the observed count rate assuming a typical power law photon index $\Gamma \rm _x = 2.1$ and absorption by the Galactic neutral hydrogen column density in the direction to the source. Optical magnitudes are primarily $B \rm _J$-magnitudes from the POSS-II plates and, as most objects lack redshift information, no K-corrections are applied. The optically identified objects are denoted by different symbols as given in the insert. Sources labelled "No ID'' include those identified with faint sources classified as unresolved by COSMOS or the APM. The cross indicates the ranges of the indices corresponding to an optical error of $\delta m = \pm 1$ magnitude.

The bulk of the objects are found along the diagonal swath from high-$\alpha_{\rm ro}$ and low-$\alpha_{\rm ox}$ to low-$\alpha_{\rm ro}$and high-$\alpha_{\rm ox}$. This swath is the region generally occupied by radio-loud quasars and blazars (see Fig. 9 in Paper I). Only four of the objects in Tables 1 and 2 are spectroscopically identified as quasars and one is the well known BL Lac object S5 1749+70. We therefore predict that most of the remaining objects are quasars or related species such as flat spectrum radio sources (upper left of the $\alpha_{\rm ro}-\alpha_{\rm ox}$ diagram), BL Lac objects (throughout most of the swath; Laurent-Muehleisen et al. 1998), and Seyfert galaxies (lower right of the $\alpha_{\rm ro}-\alpha_{\rm ox}$ diagram).

A considerable fraction, $\approx\!35 \%$, of the RASS-VLA sources do not have optical counterparts brighter than B = 22.5. These are indicated by circles with arrows pointed toward the upper left of the $\alpha_{\rm ro}-\alpha_{\rm ox}$ diagram. Many of these have X-ray fluxes above $1 \ 10^{-13}$ erg s^-1 cm^-2 and thus do not represent the large population of extremely faint sources thought to constitute a significant fraction of the X-ray background radiation. Those that lie in the middle or lower-right portion of the diagram are likely to be normal quasars. But those lying in the upper left portion have unusual spectral energy distributions with $\alpha_{\rm ro} \simeq \alpha_{\rm ox}$ or even inverted spectra with $\alpha_{\rm ro} \gt \alpha_{\rm ox}$. These sources can arise in three plausible ways: (a) they are distant clusters whose X-ray emission is elevated due to an intra cluster medium; (b) they are members of a subclass of optically-quiet quasars or red quasars (e.g. Webster et al.1995; Kollgaard et al.1995); or (c) they are spurious X-ray/radio correlations. Option (c) is likely correct for those sources with very weak radio emission, but over half (11 out of 21, see Fig. 11) have S_1.5 > 5 mJy and almost a third have S_1.5 > 10 mJy, for which spurious coincidences are unlikely. It is unclear whether the quasars in option (b) constitute a rare or a major class of AGN. We looked at the three most radio-bright NEP examples in a search for optically-quiet quasars (Kollgaard et al.1995) and concluded that they were more or less typical radio galaxies.

The source 1747.2+6532 in the upper left in Fig. 8 at $\alpha_{\rm ro} \sim$ 0.78 is identified as 4C+65.22. Lacy et al.(1993) mark it as "Q?" based on CCD observations. Kollgaard et al.(1995) argue that 4C+65.22 is not a quasar, but a radio galaxy, based on the two-component radio morphology at 8.4 GHz, associated spectral indices, and marginally-resolved optical (R = 21.2 mag) image. It is undetected on the blue POSS II plate. However, its position in the diagram, and even more that in the flux-ratio plot (Fig. 10) where it is found close to left boundary, at $\log (f_{\rm x}/f\rm _r)$ $\sim -6.7$, would be very unusual for a galaxy and thus the object may be a quasar instead. The RASS-VLA source 1743.4+6342 which we classify as a group/cluster occupies an extreme position in the diagram as well, and it may be spurious in some sense. The radio source, also known as 8C 1743+637, arises from a faint galaxy behind the cluster Abell 2280 (Lacy et al.1993), while the X-ray emission is likely due to the Abell cluster's gaseous medium. In any case, the extended X-ray RASS source represents the detection of a galaxy grouping.

The lower right region of the diagram, which is usually populated by optically prominent galaxies with low X-ray and radio emission, remains relatively empty. We suggest that another order of magnitude in X-ray sensitivity is needed before normal spiral galaxies are commonly found in all-sky surveys. Instead, we find optical counterparts that are morphologically classified as galaxies all along the swath occupied in the $\alpha_{\rm ro}-\alpha_{\rm ox}$ diagram. This is equivalent to the statement that galaxy magnitudes are roughly uniformly distributed over the range $14 \leq B \leq 21$. These objects include two IRAS-selected Seyfert 1 galaxies at $z \simeq 0.05-0.08$, a nearby Seyfert 2 galaxy at $z \simeq 0.02$, and several distant unstudied (presumably elliptical) radio galaxies. In the latter cases, the X-ray emission may arise either from the active nucleus or from a surrounding intracluster medium.

Finally, we note that the galaxy groups and clusters are spread throughout the diagram, illustrating the lack of tight correlations between optical, radio and X-ray luminosities from these objects (Burns et al.1994). The $\alpha_{\rm ro}-\alpha_{\rm ox}$ diagram is thus a poor diagnostic tool for discriminating groups and clusters from AGN.

5.3 The $\alpha_{ro}$ distribution and the radio-loud criterion

We use the radio-to-optical spectral index, $\alpha_{\rm ro}$, to analyze the NEP sample for evidence of the apparent dichotomy between radio-quiet and radio-loud AGN. Both $\alpha_{\rm ro}$ and its equivalent, the ratio of radio to optical flux densities (or luminosities), R, have been extensively used for this purpose. The commonly accepted dividing line between the two populations, established by studies of optically-selected AGN, is $\log R \sim 1$(Kellerman et al.1989; Stocke et al.1992), or $\alpha_{\rm ro} \sim 0.2$. The flux densities are typically taken at 5 GHz and an observed wavelength of 4400 Å, taken from the B magnitude. A second criterion is the radio luminosity; the division occurs $\sim L_{\rm 5\ GHz} = 10^{24}-10^{26}$ [W Hz^-1] (Miller et al. 1990; Stocke et al.1992), with some dependence on redshift (Padovani 1993). Miller et al.(1990) note that the luminosity criterion is to be preferred if the optical and radio emission of AGN are not correlated. The two definitions overlap by approximately 90% (Kellerman et al.1994; Della Ceca et al.1994). Falcke et al.(1996) have proposed separate radio-loud - radio-quiet criteria for flat- and steep-spectrum quasars: $R \simeq 25$ ($\alpha_{\rm ro} \simeq 0.24$) for steep-spectrum sources and $R \sim 250$ ($\alpha_{\rm ro} \sim 0.44$) for flat-spectrum.

The radio-loud - radio-quiet dichotomy is most evident as bimodality in the distributions of the measures R, $\alpha_{\rm ro}$, or $L\rm _R$. Bimodality has been reported for both optically-selected samples (e.g. Sramek & Weedman 1980; Kellerman et al. 1989, 1994; Miller et al.1990; Visnovsky et al. 1992; Stocke et al.1992) and the X-ray-selected sample of AGN from the Einstein Medium Sensitivity Survey (EMSS; Della Ceca et al.1994). Bimodality is sometimes evident in the distribution of radio flux densities, at least for optically-selected samples (Strittmatter et al.1980; Kellerman et al.1989). Evidence in terms of other observed AGN properties, such as redshift and optical luminosity, has been more difficult to establish.

  

Figure 9: Distribution of the radio-to-optical spectral index $\alpha_{\rm ro}$. All sources in Tables 1 and 2 are plotted, including those with upper limits on their optical flux

Figure 9 shows the distribution of $\alpha_{\rm ro}$values for the reliable NEP sources (Tables 1 and 2). A bimodal distribution is strongly suggested, with a minimum $\sim\!0.2$. This is in good agreement with the minimum value of $\alpha_{\rm ro}$ found for optically-selected samples and the value of $\alpha_{\rm ro} = 0.35$obtained by Della Ceca et al.'s results for EMSS AGN. (The different radio frequencies and the lack of K-corrections for the NEP sample would change $\alpha_{\rm ro}$ by approximately $\pm 0.05$.) It should be noted that the NEP sources include non-AGN (such as the apparent cD galaxies of several group candidates). As the NEP and EMSS samples have roughly similar sensitivities in the radio ($\sim$ 1 mJy at 1.5 and 5 GHz, respectively), optical ($\sim$23 mag in V and B), and X-ray ($\lower.5ex\hbox{$\; \buildrel \gt \over \sim \;$}5 \ 10^{-14}$ erg s^-1 for 0.1-2.4 keV and $\lower.5ex\hbox{$\; \buildrel \gt \over \sim \;$}8 \ 10^{-14}$ erg s^-1 for 0.3-3.5 keV), the NEP sample's results provide further confirmation of bimodality but do not extend it to significantly fainter limits. The minimum in the $\alpha_{\rm ro}$ distribution is also evident as a sparsely-populated band ("gap'') in the $\alpha_{\rm ro}-\alpha_{\rm ox}$ and $f_{\rm x}/f_{\rm r} - f_{\rm o}/f\rm _r$ (Sect. 5.4) plots. The gap is most apparent if different morphological classes are considered separately. Although bimodality is not evident in RGB sample (its $\alpha_{\rm ro}$ distribution cuts off at the lower limit of radio-loudness), it is quite evident in the sample of radio-selected, ROSAT-detected quasars drawn from the Véron-Cetty - Véron (1993) catalog (VV93) studied by Brinkmann et al.(1997; see Fig. 16).

5.4 The f_x/f_r - f_o/f_r diagram

While AGN classes often overlay in the $\alpha_{\rm ro}-\alpha_{\rm ox}$diagram, they can occupy more distinct regions in the $f_{\rm x}/f_{\rm r} - f_{\rm o}/f\rm _r$ (or equivalently, $\alpha_{\rm rx}-\alpha_{\rm ro}$)diagram. For example, X-ray-selected BL Lacs and radio-selected BL Lacs are widely separated in this diagram with BL Lacs from the RGB survey filling the gap between these populations (Fig. 11 in Paper II; Laurent-Muehleisen et al.1998). Figure 10 shows the flux ratio diagram for objects in Tables 1 and 2. Objects on the top right should be mainly galaxies, X-ray-selected BL Lacs reside in the top central region, while quasars and radio-selected BL Lac objects congregate at the lower left. The diagram is similar to that of the optically unidentified objects from the RGB survey (Fig. 11 in Paper II), indicating that the populations have not changed dramatically between the brighter RGB and fainter NEP samples. The optically faint and unclassified NEP sources are spread throughout the region of radio- and X-ray-selected AGN. However, few are present among the Seyferts and other galaxies in the radio-quiet regime, $\log [f_{\rm o}/f_{\rm r}] \gt -1$ ($R \lower.5ex\hbox{$\; \buildrel < \over \sim \;$}10$). The $f_{\rm x}/f_{\rm r} - f_{\rm o}/f\rm _r$ diagram is consistent with Fig. 8 in suggesting that several source identifications may be incorrect.

  

Figure 10: Flux ratios logf$_{\rm x}$/$f\rm _r$   versus logf$\rm _o$/$f\rm _r$   for all sources. Arrows indicate sources with upper limits for the optical fluxes. The cross marks the position of the planetary nebula

  

Figure 11: Distribution of various object types as function of the optical B - magnitude. Open histogram: all objects; shaded region: specific object class