Astron. Astrophys. 319, 413-429 (1997)

3. Non - detections, variability

Before discussing the ROSAT detected objects in more detail we want to compare the sample with those objects, for which no X-ray data are available.

There are 1033 radio detected objects in the VV93 catalogue which were not seen in the ROSAT All-Sky Survey . 126 of these objects were later discovered in deeper pointed observations. A comparison of the properties of these less X-ray loud catalogued sources with those detected in the ROSAT All-Sky Survey can give valuable information about the reasons for a source not to be detected in X-rays as well as on the inherent biases and uncertainties of the selected sample.

In Fig. 1 we plot the histogram of the count rates of all sources detected in the RASS and, as shaded area, of those not seen in the RASS but in pointed observations only. It appears that the non detections are in general below or close to the sensitivity limit of the Survey. We checked the RASS data of all not detected sources with count rates greater than 0.025 counts s^-1 in pointed observations. A small fraction of the objects either falls onto the strip boundaries in the SASS processing where the detection efficiency is reduced, or they are in regions of exceptionally low Survey exposure, or in regions with enhanced diffuse emission. The majority of the objects had a rather low signal-to-noise ratio in the Survey and was thus not regarded as a statistically significant detection by the SASS. Only a small number of quasars (of the order of  10%) might have been missed due to their intrinsic variability. Thus we have to conclude that most of the 'non-detections' are sources with soft X-ray fluxes below the Survey's limiting sensitivity.

Fig. 1. Histogram of the count rates of objects seen in the RASS (open line) and those not detected in the Survey but in pointed observations.

More than 82 quasars have been observed repeatedly in pointed observations and thus provide a good test sample for an evaluation of the variability of the sources. In Fig. 2 we plot a histogram of the maximal variability of the objects, i.e., the ratio of the higher count rate divided by the lower count rate. If a source has been observed more than twice only the maximum and the minimum values have been used.

Fig. 2. Histogram of the variability of quasars seen more than once in pointed observations.

Fig. 2 clearly shows that a large fraction of the quasar population is variable, however mostly by less than a factor of two. For weak sources low variability cannot be distinguished from statistical fluctuations. Only 3 objects vary by more than a factor of 5: S4 1050+54 by a factor of , MS 09584+6913 by more than a factor of 7, and 1E 1640+401 by a factor . Unfortunately, as most of the individual observations on a source have been performed in different  ROSAT  pointed observation periods, i.e., at least half a year apart, we cannot make any reliable statistical estimates about a relation between the time scale and the amplitude of the variability of a source.

3.1. Detection biases

The ROSAT All-Sky Survey has a relatively uniform limiting sensitivity of a few erg cm with the exact value depending slightly on the amount of intervening Galactic absorption, on the exact shape of the X-ray spectrum of the quasar, and on the local Survey exposure.

In Fig. 3 we show the detection rate in percent of the quasars as function of redshift, i.e., the number of objects from the radio selected sample detected in the RASS in a redshift bin divided by the total number of catalogued radio-loud quasars for that redshift. The dashed line represents a curve , where is the luminosity distance (Schmidt & Green 1986) assuming a Friedman cosmology with H₀ = 50 km s^-1  Mpc^-1 and = 0.5 ¹. The redshift term accounts approximately for the quasars' luminosity evolution (Boyle et al. (1993) derive a value of for the X-ray luminosity function, however, for a universe). The detection rate roughly follows this simple law, with positive deviations around (as noted for  ROSAT   detections of quasars in general, Boyle et al. 1993) and large scatter at high redshifts. Details like luminosity K-corrections and variations seem to play only a minor role.

Fig. 3. Detection probability in percent of radio selected quasars in the RASS as a function of redshift.

The total average detection rate of all radio-loud quasars in the RASS is about 33.2%, markedly higher than the % of radio-quiet objects (Yuan et al. 1996) and at highest redshifts more than 50% of the catalogued objects are detected. This high detection rate indicates that the currently known high z quasars are 'special' in a certain sense.

As the sample is drawn from existing flux limited radio surveys, changes of the detection probability as a function of radio flux density give insight into the relative strengths of the X-ray and radio emission of the objects.

In Fig. 4 we show the detection probability (in percent) as function of the 5 GHz radio flux density. The probability increases from about 10% at low radio fluxes to more than 75% at the highest radio fluxes. It should be noted that, at even lower radio fluxes not shown here this value increases again due to the inclusion of X-ray loud Seyfert I type QSOs like the PG sample which are nearby objects but, as mentioned above, don't qualify properly as radio-loud quasars.

Fig. 4. Detection probability in percent of quasars in the RASS as function of the 5 GHz radio flux.

The decrease of the X-ray detection rate with radio flux clearly shows the X-ray detection bias caused by the sensitivity limit of the RASS. The radio surveys are thus, relative to the X-ray observations, more sensitive. Additional changes of the source population or luminosity dependent variations of intrinsic source properties cannot be ruled out either.

© European Southern Observatory (ESO) 1997

Online publication: July 3, 1998
helpdesk@link.springer.de