Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 362, 75-96 (2000)

Previous Section Next Section Title Page Table of Contents

5. Similarities and differences in the SED of RLQ and RQQ

The present sample contains a range of radio source classifications, with which we can elucidate the dependence of broad-band spectral features on radio properties, thereby testing some unification scenario predictions. The limitations of these tests lie in the sample's relatively small size and heterogeneous nature.

5.1. Average SED

A quick look at the main spectral differences between the different kinds of quasars is provided by the comparison of the average SED of RQQ, RIQ, FSRQ, and SSRQ (including the RG 3C 405). The SEDs are shown in [FORMULA]-[FORMULA] and [FORMULA]-[FORMULA] spaces separately for each class over the radio/soft X-ray frequency range in Fig. 7. The broad spectra of two typical host galaxies (a giant elliptical and a spiral galaxy), in their rest frames and without any normalization, are also plotted in Fig. 7, as in Fig. 2. The average SEDs have been computed using the conventional mean, excluding upper limits. The width of each frequency bin is equal to 0.5 in Log([FORMULA]). The reported uncertainties correspond to the standard deviation of the mean of the data per frequency bin. All the data have been connected by straight lines. At soft X-ray energies we indicate the average power law [FORMULA] 1[FORMULA] computed from the distribution of best fit soft X-ray power law models of all objects of the same class (photon index [FORMULA] = 2.73[FORMULA]0.61 (RQQ), 2.39[FORMULA]0.19 (RIQ), 2.39[FORMULA]0.23 (SSRQ), and 2.25[FORMULA]0.12 (FSRQ)).

[FIGURE] Fig. 7. The average quasar energy distribution for RQQ (solid line), RIQ (dotted line), FSRQ (dashed line), and SSRQ (dash-dot line). The number of objects that were used in each class is reported on the figure with the class name. A symbol is reported for each frequency bin, if at least one data point is available. The two high redshift objects of the sample: HS 1946+7658, and PG 1718+481, were excluded since they are characterized by very high luminosities that modify the average SEDs at a few frequencies producing an irregular spectral shape. Two host galaxy templates are also reported: a typical spiral galaxy (dash-dot-dot-dot line), and a giant elliptical galaxy (long dashed line). No normalization was applied to the reported curves.

As expected, the largest difference in luminosity among the different classes appears at radio wavelengths. A smaller difference is observed at soft X-ray energies, and in the near-IR ([FORMULA] 1014 Hz corresponding to [FORMULA]µm), while the luminosity and the spectral shape in the mid- and far-IR are remarkably similar (see Fig. 7). The large difference in the IR spectral shape between quasars and the host galaxy templates indicates that the contribution from the host galaxy is negligible also at radio and soft X-ray energies, and not only in the far/mid-IR. This result is in agreement with previous studies on the broad SED of quasars (Sanders et al. 1989; Elvis et al. 1994). A quantitative comparison of the luminosity emitted at different frequencies by each quasar class is presented in the next section.

5.2. Multi band luminosities

The IR component was also modeled by fitting a parabola in Log [FORMULA]-Log [FORMULA] space (see Fig. 2). The parabola model gives a rough estimate of the strength and shape of the IR component, even if the spectral coverage is not complete. For several objects upper limits were also used in the fit. This model has by itself no physical meaning, however, it describes the IR component relatively well, it can easily be traced even with poor spectral coverage, and can take into account the whole IR emission of most of the objects in a larger wavelength range than the detailed grey body models. The parabola is too narrow to satisfy the observed IR SED in a few cases, e.g., in 3C 405 and PG 1543+489. In these cases we fitted only the far/mid-IR data where the IR emission usually peaks. The parabola parameters are its width, the frequency of maximum luminosity density ([FORMULA]) and the maximum luminosity density ([FORMULA]). The parabola fit to the IR component was applied to all objects of the sample, except PKS 0135-247, for which no IR data are available (see Fig. 2 a). The distribution of the peak frequencies values ([FORMULA]) observed in the four different classes of objects is reported in Fig. 8. The distribution is quite similar for SSRQ and RQQ, ranging from 2.6[FORMULA]1012 Hz (114 µm) to 3.6[FORMULA]1013 Hz (8 µm), while it is shifted to higher frequencies for FSRQ and RIQ, ranging from 9.0[FORMULA]1012 Hz (33 µm) to 2.8[FORMULA]1013 Hz (11 µm). This difference may be due to the flat radio non-thermal component extending to high frequencies in FSRQ and RIQ, and dominating the dust emission.

[FIGURE] Fig. 8. Histogram of the IR peak frequency of the parabola model for the different classes: RQQ, SSRQ (including the RG 3C 405), FSRQ, and RIQ.

We define the IR luminosity [FORMULA] as the product of the luminosity value at which each parabola peaks and the corresponding frequency (L(IR) = [FORMULA]). Note that this parameter does not depend on the width of the parabola. Only upper limits for [FORMULA] could be derived for PKS 0408-65 and PG 1040-090. The distribution of [FORMULA] is reported in Fig. 9. In this, and in the following histograms upper limits are shown with arrows, one per object. The similarity in the IR luminosities and spectra (see also Fig. 7) in all quasars suggest a similar origin.

[FIGURE] Fig. 9. Histogram of the peak luminosity of the IR parabola model for the different classes: RQQ, SSRQ (including the RG 3C 405 whose position is indicated), FSRQ, and RIQ.

The radio emission in the RLQ arises from two very different spatial scales, the core and extended components. We calculated the average of [FORMULA] over the rest-frame interval 5-9 GHz for each spatial component in all of the RLQ, except PG 1354+213 and HS 1946+7658, which were undetected at these frequencies. Fig. 10 displays histograms for the two components separately, and Fig. 11 shows the distribution of the median of all measured [FORMULA] over the same frequency range, without component distinction. The distribution of the median radio luminosity is bi-modal (Fig. 11). However, if we consider only core radio luminosities (top panel of Fig. 10), the SSRQ radio luminosity distribution shifts towards lower values, making a continuous distribution, rather than a bi-modal one, but without overlapping. The contribution from the extended components are very similar in FSRQ and SSRQ (bottom panel of Fig. 10). In the following analysis we will consider only the core luminosity [FORMULA]. When the core luminosity is not available (PKS 0408-65, B2 1721+34, 3C 405, and PG 2308+098), we report an upper limit corresponding to the average radio luminosity relative to the extended component.

[FIGURE] Fig. 10. Histogram of the core and extended components radio luminosity for the different classes: RQQ, SSRQ (including the RG 3C 405 whose position is indicated), FSRQ, and RIQ.

[FIGURE] Fig. 11. Histogram of the median radio luminosity for the different classes: RQQ, SSRQ (including the RG 3C 405 whose position is indicated), FSRQ, and RIQ.

In the soft X-ray, we define L(SX) as [FORMULA] with [FORMULA] corresponding to 1 keV in the observer's rest-frame. The distribution of L(SX) for each class is reported in Fig. 12. In the soft X-ray, no data are available for 3C 405, and PKS 0408-65, and only an upper limit is available for PG 1004+130.

[FIGURE] Fig. 12. Histogram of the soft X-ray luminosity at 1 keV for the different classes: RQQ, SSRQ, FSRQ, and RIQ.

5.3. Origin of the observed luminosities

The main factors determining the observed luminosities are: the energy emitted by the central engine (AGN); the amplification due to Doppler boosting in a relativistic jet; and the contribution from a starburst. We will estimate the role of each of these parameters in producing the SEDs through the comparison of the observed radio, IR, and soft X-ray luminosities, represented by L(Radio), L(IR) and L(SX), respectively (see Sect. 5.2 for their definition).

5.3.1. Orientation effects in RLQ

The orientation of the beamed emission can be estimated from the radio core fraction R. This quantity, defined as the ratio between the core radio luminosity and the luminosity of the extended radio emission at 5-9 GHz in the rest frame, serves as an orientation indicator of the radio source with respect to the observer, measuring the relative strength of the core component (Hes et al. 1995). The core flux was not available for three SSRQ and one RG, thus the parameter R was not computed. In the case of FSRQ we computed the luminosity of the extended component in the frequency range 5-9 GHz by extrapolating the power law observed at low frequencies (power law index [FORMULA] in Table 7).

The FSRQ are well separated from SSRQ in the distribution of the ratio R (see Fig. 13). This difference permits us to estimate the enhancement factor of the beamed emission after a few considerations. First, the observed radio emission in RLQ is mainly produced by the jet and its core rather than a starburst, since star-emitting ULIRG have much lower radio luminosity than RLQ (Colina 1995). Second, we assume that the radio source is intrinsically identical in FSRQ and SSRQ, and that the difference in their radio emission is due only to the orientation of the beamed emission. After these approximations we can write


where A is the amplification factor of the beamed emission. Since the luminosity of the extended components are the same for the flat and steep radio quasars (see above), using Eq. (7) we can derive a relation between the parameters [FORMULA] and [FORMULA]: [FORMULA] = [FORMULA]. Replacing the observed values of [FORMULA] ([FORMULA] 0.05-0.15), and [FORMULA] ([FORMULA] 3-4) (see Fig. 13) in the above relation yields A [FORMULA]20-80.

[FIGURE] Fig. 13. Luminosity emitted in the soft X-ray, radio and IR domains compared to the parameter R given by the ratio between the radio luminosity of the core and of the extended components. The different symbols correspond to different classes of objects: squares for SSRQ, and circles for FSRQ. The dashed lines are the best fit lines.

Fig. 13 displays L(Radio), L(SX), and L(IR) versus the core fraction R. Linear correlation results for these relations are reported in Table 9, where the parameter pairs are reported in the first two columns, the number of data pairs in the third, in column 4 the linear correlation rank ([FORMULA]), and in column 5 the associated probability to have such a correlation rank from uncorrelated values ([FORMULA]).


Table 8. Best fit parameters of grey body models


Table 9. Correlation test results for FSRQ and SSRQ

Higher radio and soft X-ray luminosities are observed in objects with higher values of the radio core fraction R (when the jet points towards us). The orientation effect is more important in the radio domain, as shown by the stronger correlation, than in the soft X-ray, and negligible in the IR. This implies that the radio core and a fraction of the total emitted soft X-ray luminosities are emitted anisotropically. We furthermore verified that R is not correlated with the redshift and thus that the above result is not an artifact of distance related biases in the measurement of R.

Assuming that the soft X-ray source is intrinsically identical in FSRQ and SSRQ, the observed difference in L(SX) arises from the orientation of the fraction, f, that is beamed. If the fraction of emitted radiation that is beamed is enhanced by a factor A, identical to that of the radio emission, the following relation between the soft X-ray luminosity in FSRQ and that in SSRQ will be valid:


Using the average value of the ratio [FORMULA] that is [FORMULA], and the range of values obtained for the factor A, we derive a fraction [FORMULA] 3-12% for the beamed fraction of the soft X-ray component.

5.3.2. SSRQ versus RQQ

The radio and the soft X-ray luminosities are mainly produced by the AGN component (see Sects. 5.1 and 5.3.1). The comparison between the luminosities emitted in the radio and soft X-ray domains is then equivalent to a comparison of the AGN power in the two types of quasars. The radio core emission of SSRQ is on average 200 times higher than that of RQQ (see Fig. 10) and the soft X-ray luminosity is on average 8 times higher in SSRQ than in RQQ (see Fig. 12). Since the SSRQ show luminosities higher than RQQ not only at radio and soft X-ray energies, but also in the hard X-ray domain (Lawson & Turner 1997), we argue that the bolometric AGN luminosity is much higher in SSRQ than in RQQ. The difference in the AGN power should be observable at all frequencies where the AGN emission dominates. We have already pointed out the similarity in IR luminosities and spectra of SSRQ and RQQ (see Fig. 7 and Fig. 9). This similarity suggests that the origin of the dominant IR component is not AGN-related. The candidate is then a starburst.

Some indication of the dominant IR emission mechanism can be gleaned from the shape of the SED. An AGN can emit a significant fraction, often a majority, of its infrared luminosity at shorter wavelengths, [FORMULA] [FORMULA] 60 µm, as long as the obscuring columns are not so large as to be optically thick at these wavelengths. Starburst dominated galaxies, on the other hand, produce the bulk of their infrared emission at [FORMULA] [FORMULA] 60 µm. The ratio of the luminosities in these two wavelength regimes, L(60-200 µm)/L(3-60 µm), thus provides a rough estimate of the primary driver of the infrared component. A histogram of this ratio is presented in Fig. 14, in which only the sources having at least two grey body components with T [FORMULA] 1000 K are included. For comparison, the luminosity ratio from an average SED of low reddening starburst galaxies (Schmitt et al. 1997) is also indicated. All of the AGN in the sample have infrared luminosity ratios less than the starburst fiducial value (=0.76) by a factor or four or more (the maximum ratio is 0.20 corresponding to 27% of starburst contribution), suggesting that the infrared in these sources is dominated by the central engine. The RQQ and SSRQ have similar average ratios, 8% of the total IR emission is produced by a starburst.

[FIGURE] Fig. 14. Histogram of the ratio L(60-200 µm)/L(3-60 µm) for the different classes: RQQ, SSRQ, FSRQ, and RIQ. The dashed line represents the value derived from the average SED of a sample of low reddening starburst galaxies.

Since the dominant IR source is the AGN, which differs in SSRQ and RQQ, other factors have to be taken into account to explain the similarity of their IR luminosities and average spectra. These factors can be the dust covering factor, its amount and geometric distribution. A higher dust covering factor in RQQ, due to larger total dust mass and/or a particular geometric distribution, may account for the similar IR radiation. Aspects of this hypothesis are attractive, since the presence of more circumnuclear material may be related to the physical conditions that hinder the jet formation and/or development in RQQ. However, if larger dust covering factors were present in RQQ, a higher probability to observe a RQQ through optically thick dust would be expected and a large fraction of RQQ with absorption features in the soft X-ray would be observed. Observations do not support these predictions (see Sect. 2). Moreover, the similarity in the observed IR luminosities and spectra requires a fine tuning between the dust heating source, the dust amount and its geometric distribution, which makes this explanation improbable. We propose another scenario in which the dust properties (amount and distribution) and the heating source are similar in SSRQ and RQQ. The dust distribution contributing in the far-IR probably extends from the more external regions of the AGN, and is predominantly heated by the optical and UV radiation fields filling these external regions which are similar in both classes. At relatively large distances from the centre, the AGN components are similar in both types of quasars, towards the innermost regions, where the soft X-rays are emitted and a jet is formed, SSRQ and RQQ become different. The high energy photons escape from the centre without significant dust absorption and provide an important probe of the central radiation source. Observations indicate that the soft X-ray radiation is higher in SSRQ (see Fig. 7) in agreement with the proposed scenario.

From this analysis it is suggested that the main difference between RQQ, and SSRQ takes place in the innermost nuclear regions where the emitted power is higher in SSRQ than in RQQ, while the AGN external regions (dust distribution and optical/UV source) have similar properties in both types of quasars.

Since FSRQ and SSRQ show similar properties in the IR, once the non-thermal contribution is subtracted, this conclusion can be also extended to FSRQ.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 2000

Online publication: October 30, 19100