Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.atnf.csiro.au/people/Matthew.Whiting/thesis/ch7.pdf Äàòà èçìåíåíèÿ: Wed Mar 29 06:51:54 2006 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 13:52:38 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: m 87 jet

Chapter 7

BL Lacs in the PHFS
7.1 Intro duction

BL LacertÔ ob jects, or BL Lacs, are a class of AGN that are defined by their lack of any prominent emission lines in their optical spectra. Most theories to explain this lack involve the presence of an extra continuum component that helps to mask emission lines (although these lines may already be weak). From the shape of the radio­IR­optical SED (Impey and Neugebauer 1988), as well as high optical and radio polarisation, this extra component is thought to be synchrotron emission, which comes from the relativistic jet (such as those seen in the optical that are discussed in Chapter 2). Due to the relativistic nature of the jet, Doppler beaming will be important (see Section 3.5), an effect strongly dependent on the viewing angle, or the orientation angle of the jet. Hence, since the orientation to the line of sight is important for the appearance of most of these flat-spectrum radio quasars and BL Lacs, it follows that there must be ob jects with similar properties but a different appearance due to a different orientation. If the angle at which we view the radio quasars or BL Lacs is small (as seems likely), this other group of ob jects must be much more numerous. Such a group is termed the "parent population" of the closely aligned ob jects.

7.1.1

Radio galaxies

Clear candidates for the parent populations of radio quasars and BL Lacs are the radio galaxies. These are galaxies that exhibit large amounts of radio


106

BL Lacs in the PHFS emission, most often in the form of jets or lobes that can extend far beyond the boundaries of the (optical) host galaxy. Radio galaxies are separated into two distinct classes based on their radio luminosity: Fanaroff-Riley type I and type II (Fanaroff and Riley 1974). FR Is, the low luminosity galaxies, have their radio emission peaking closer to the central nucleus, and exhibit fainter, more diffuse lobes. FR IIs, always at high radio luminosities, have their dominant radio emission from the bright parts of their lobes, which are usually connected to the nucleus by a thin collimated jet. The division in luminosity is strongest at lower radio frequencies (e.g. 178 MHz, where the dividing luminosity is 2 â 10
25

W Hz

-1

). At higher

frequencies, however, there is considerable overlap in the luminosities of the two classes. The low frequency emission is dominated by the lobes, so this indicates that the lobes of the two types of radio galaxies are significantly different. The higher frequency emission, on the other hand, comes from the inner jet and core region, indicating that these regions in FR Is and FR IIs are more similar. Aside from the differences in their radio properties, the two classes of radio galaxies differ in a number of ways. The division between the two classes appears to also be a function of optical luminosity. If one plots the radio luminosity against the optical luminosity (that is, the optical magnitude down to a fixed surface brightness in the rest frame), then one finds that the two classes are well separated (with mixing only occurring over a factor of 10 in radio luminosity), and that the dividing radio luminosity increases roughly as L
2 opt

(Owen and Ledlow 1994).

The emission line properties of the two classes are also different. It has been shown that, in FR IIs, the narrow emission line (particularly [O iii] 4959,5007) luminosities are correlated with that of the radio (Baum and Heckman 1989; Rawlings et al. 1989). This may be related to the optical luminosity relationship, in the sense that a bigger, brighter galaxy will have more emission-line gas. This gas is found to be extended, usually spatially located near the radio source axis (in both radio galaxies and steep-spectrum radio quasars) (Baum and Heckman 1989). The total emission line luminosities (that is, broad plus narrow) are less correlated with the radio power, indicating either that only the narrow line gas is associated with the radio source, or that the viewing-angle dependence of the broad lines is important


7.1 Intro duction (not all sources of a given power will have the same range of line luminosities due to their different viewing angles). Meanwhile, the emission line luminosities of FR Is are also correlated, but are typically 5­30 times less than those of FR IIs with the same total radio power (Zirbel and Baum 1995). The correlation for the FR Is is dominated by a correlation between line luminosity and host galaxy optical magnitude ­ no similar correlation exists for FR IIs. Finally, optical spectra of FR IIs are reasonably heterogeneous, as some show quite strong emission lines, whereas others only show weak, low-excitation lines, similar to FR Is (Laing et al. 1994).

107

7.1.2

Unification Schemes

FR I and FR II radio galaxies have been proposed (Urry and Padovani 1995, and references therein) as the parent populations of radio-loud AGN. Under this scheme, radio quasars are aligned versions of FR II galaxies (with flatspectrum quasars possibly more aligned than steep-spectrum quasars). BL Lacs, on the other hand, have lower luminosities than quasars (particularly when the effects of beaming are removed), and so are believed to be aligned versions of FR I galaxies. The supporting evidence for these unified schemes generally involve radiation (or some property) that is assumed to be isotropic. Good evidence for this is found in the study by Antonucci and Ulvestad (1985), who studied the extended radio emission of a sample of blazars (including both BL Lacs and OVV quasars). They found that they had similar extended radio powers to radio galaxies from the 3C (mainly FR IIs with some FR Is) and B2 (predominantly FR Is) samples. It is uncertain, however, whether the clear morphological dichotomy between FR Is and FR IIs is reflected in the extended emission of quasars and BL Lacs. If one assumes that the narrow emission lines are emitted isotropically, then it is found that quasars and FR IIs have [O ii] luminosities that span the same range of values (Hes et al. 1993). BL Lacs, on the other hand, show somewhat larger [O iii] luminosities than FR Is. This may, however be due to either underestimation of the FR I fluxes due to a strong stellar continuum, or a (relatively) small slit that may not have included all the emission line region (Tadhunter et al. 1993). Also, there is some evidence that [O iii] in quasars and FR IIs is emitted anisotropically (at least in comparison to [O ii]) (Hes et al. 1993; Urry and Padovani 1995), and this


108 Name 3C 15 3C 66B 3C 78 3C 120 3C 200 0521-365 3C 212 3C 245 3C 264 3C 273 M 87 3C 346 3C 371 2201+044 Jet flux (µJy) 5.9 5.9 33 14 18 60 0.19 0.17 33 57 1960 11 15 0.57 Source magnitude V=15.8 V=15.0 R=12.5 V=14.8 V=20.0 R=14.6 V=19.1 V=17.3 R=12.5 V=12.8 B=9.6 V=17.5 V=14.4 V=15.2

BL Lacs in the PHFS Source flux (mJy) 1.75 3.66 35.63 4.40 0.04 5.15 0.08 0.44 35.63 27.79 589.88 0.37 6.37 3.05 Flux ratio (Jet/Source) 3.37â10-3 1.61â10-3 0.93â10-3 3.18â10-3 4.91â10-1 1.17â10-2 2.26â10-3 0.38â10-3 0.93â10-3 2.05â10-3 3.32â10-3 3.00â10-2 2.36â10-3 0.19â10-3

Table 7.1: Ratios of jet flux to central source flux at optical wavelengths for known
optical jets. Jet fluxes taken from Table 1 of Scarpa and Urry (2001), while central source fluxes are converted from magnitudes taken from NED. Where possible, the magnitudes are taken to be at approximately the same wavelength as the jet observation.

could also account for the difference between BL Lacs and FR Is.

7.2

Orientation of the PHFS quasars

The integral part of the Urry and Padovani (1995) unification scheme is orientation, in that BL Lacs are beamed FR Is and quasars are beamed FR IIs, due to the viewing angle being close to the direction of the relativistic jet. Can we estimate the amount of beaming that is present in the PHFS sources, and thereby deduce their orientation? The synchrotron emission fit by the modelling in Chapter 5 provides the means for doing this. From the model fitting, we found that many of the PHFS quasars have significant synchrotron emission at optical wavelengths. This synchrotron emission is assumed to come from the relativistic jet that provides the radio emission ­ that is, from an optical jet. It can be noted that if the synchrotron component in the fit is taken to be representative of the optical jet emission, and the blue power law component representative of the central source (i.e. accretion disk) emission, then the ratio of jet flux to central source flux takes


7.2 Orientation of the PHFS quasars a wide range of values (see Fig. 5.7). This can be compared to those sources that have been observed to have optical jets. Scarpa and Urry (2001) have compiled a list of optical fluxes of the optical jets in each of these sources, and we have used these values, together with optical magnitudes of the sources taken from the literature
1

109

to compile a list of jet-to-central-source flux ratios for the optical jet sources. Where available, the magnitudes are at approximately the same wavelength as the observation of the jet. When just a single magnitude was given, it has been taken to be V band. These data are listed in Table 7.1, and the ratios are plotted as a histogram in Fig. 7.1, where they are compared with the synchrotron to power law flux ratio for the PHFS quasars (calculated at an observed wavelength of 0.5µm). Although there are three optical jet sources that are in the PHFS (3C 273, 0521-365 and 3C 120), none of these sources are in the sub-sample with optical/NIR photometry, and so there is no overlap between the two plots. It can be seen in Fig. 7.1 that the PHFS quasars have much higher ratios of jet to central source flux. If, as discussed previously, the PHFS quasars are viewed at a smaller angle to the jet axis, then this difference in flux ratios will be due to different amounts of Doppler beaming. Recall from Chapter 3, for a relativistic source viewed at an angle to the direction of motion, the observed flux is enhanced by the Doppler factor = 1 (1 - cos ) (7.1)

where 2 = 1 - 1/2 . The degree of enhancement depends on the nature of the source, but in general takes the form F ( ) = p F ( ) where a (7.2)

indicates a quantity is measured in the rest frame. (Note that
-

= .) If F

, then the values for p are either p = 2 + for a

smooth, continuous jet, or p = 3 + for discrete moving blobs (or spheres). The amount of enhancement as a function of for two different p values is shown in Fig. 3.7. (Note that here we are simply using the standard Doppler formulae, not the versions altered to account for an ordered magnetic field
1 The magnitudes were found using NED, the NASA/IPAC Extragalactic Database, http://nedwww.ipac.caltech.edu/


110

BL Lacs in the PHFS

Figure 7.1: Histogram of jet-to-central-source flux ratios for both sources that have
observed optical jets, and the PHFS sources based on the fitting. In the PHFS plot, the open histogram shows all sources that were able to be fit by the combined model, while the green hatched histogram shows those sources best fit by the combined model (in preference to the power law model ­ see Chapter 5 for details on the fits). The red hatched histogram shows the BL Lac ob jects.


7.2 Orientation of the PHFS quasars ­ Chapter 3.) Now, the case of the optical jets can be considered in this light. By comparing jets from both those sources with known optical jets and the PHFS quasars (based on the modelling from Chapter 5), the viewing angle for the PHFS quasars can be determined by making a number of assumptions: < The jets in both types of ob jects are, on average, the same, particularly the bulk Lorentz factors (). > The intrinsic (unbeamed) jet flux scales the same way as the core flux. This means that the ratio of intrinsic jet flux to core flux is a constant. In other words, the apparent differences in this ratio are purely due to Doppler amplification (or de-amplification) of the jet flux. fi The viewing angle and the Lorentz factor are known for the sources with known optical jets. These values are needed to derive the parameter values for the PHFS sources. Since the ratios of jet-to-core flux have been measured (Fig. 7.1), one can define R as the ratio of the jet-to-core ratios in two different sources. Now, let Fj and Fc represent the unbeamed jet and core fluxes respectively, and Fj = Fj and Fc = Fc the observed jet and core fluxes (note that the core flux is assumed to be unbeamed). Then, assuming that the Doppler
p q beaming in sources 1 and 2 is given by 1 and 2 , R can be defined by

111

R=

Fj 1 /F Fj 2 /F

c1 c2

=

p 1 Fj 1 /F

c1



q 2 Fj 2

/F

=

q -p

c2

(1 - cos 2 )q , (1 - cos 1 )p

(7.3)

since the ratio of unbeamed jet flux to core flux is assumed to be constant across all sources. So, given that the value of R is determined observationally, and that and 2 are known, then 1 can be calculated: cos 1 = R
1/p

-

(q -p)/p

(1 - cos 2 ) R1/p

q /p

(7.4)

This is not defined for all values of R, however. If R is sufficiently large, then even making 1 as small as possible will not provide the necessary boosting. Similarly, R cannot be too small either. These limits can be calculated from Eqn. 7.4, setting 1 = 0 and 1 = 180 respectively, and


112 are defined as
q -p

BL Lacs in the PHFS

(1 - cos 2 )q R (1 + )p

q -p

(1 - cos 2 ) (1 - )p

q

(7.5)

To use this method to estimate viewing angles for the PHFS quasars, the values of and 2 need to be defined for the sources with known optical jets (i.e. point fi above). We use the two results detailed in Section 2.2.2: those of Scarpa and Urry (2001) and Sparks et al. (2000). The two sets of (, ) values used were (20 , 7.5) (from Scarpa and Urry (2001)), and (30 , 3) (since the angles and factors given by Sparks et al. (2000) are limiting values, we choose a slightly more extreme case, taking their lower angle limit and a slightly higher value). To calculate the viewing angle for the PHFS quasars, one must also assume a jet-to-core ratio for the optical jet sources. This was taken to be the median value of the ratios measured in Table 7.1. In calculating this value, however, we have excluded those sources that are either in the PHFS or have a blazar nature: 3C 120, 0521-365, 3C 273, 3C 371, and 2201+044. The median jet-to-source ratio of the remaining ob jects was found to be 2.26 â 10
-3

. (The median value in this case is slightly more robust than

the average, as it is not affected greatly by the outliers.) For each PHFS source, the value of R is found by taking the ratio, at 0.5µm, of the fitted synchrotron component (representing the emission from the optical jet) to the fitted blue power law (representing the emission from the central source), and dividing that by the above median value. (Note that finding this 0.5µm ratio is equivalent to finding the ratio of the total jet flux to total central source flux at 0.5µm, since both are unresolved in our images.) This value of R is then used in Eqn. 7.4 to find the PHFS quasars' viewing angles. This is done for each of the two different sets of (, ) parameters detailed above. We can make a brief comment here about the effect of the host galaxy emission. The optical jet sources are generally at low redshift, and the host galaxy is well resolved. The source fluxes listed in Table 7.1 for these cases are the integrated fluxes over the entire galaxy. Thus, the jet-to-source ratios are really fjet /(f
core

+f

gal

) (since the central source flux will be included in

the integrated flux). Now, for the PHFS sources, the host galaxy is generally unresolved and at a much lower flux level than the central continuum (Masci et al. 1998). This can be expressed as f
gal

f

core

, which implies that


7.2 Orientation of the PHFS quasars

113

Figure 7.2: Viewing angles for the PHFS quasars, for different and assumptions,
as well as different Doppler boosting dependencies. with combined model fits, while the green hatched by the combined model (in preference to the power on the fitting). The red hatched histograms are the The open histograms are all sources histograms are those sources best fit law model ­ see Chapter 5 for details BL Lac ob jects.

fjet /f

core

fjet /(f

core

+f

gal

). Hence, the ratios used are consistent with

each other, assuming the host galaxy emission in the PHFS sources is much less than the emission from the central source. Also needed to calculate the viewing angle is the value of the Doppler index p. The calculations are performed for the two cases of p = 2 + and p = 3 + . For the optical jet sources, is taken to be 1.4 (the average value for optical jets as calculated from published values ­ see Section 2.2.1). For the PHFS sources, is calculated from the observed fitted synchrotron component by taking the average slope of just the synchrotron component (since it is just the jet emission we are concerned with here), between the


114 p 2 3 2 3 + + + + (2 , ) (20 (20 (35 (35 ,7.5) ,7.5) ,3) ,3) 1 ( ) (All) Average Median 14.0 10.6 13.4 10.6 30.8 20.8 26.5 16.8

BL Lacs in the PHFS 1 ( ) (Combined fits) Average Median 13.8 12.7 12.5 11.9 22.4 19.5 21.4 20.2

Table 7.2: Table showing the average and median viewing angles for the PHFS quasars,
for a range of different (, ) values and options for the Doppler index p.

observed wavelengths of 0.4µm to 0.8µm (simulating a power law index calculated from two measurements made at these wavelengths). The resulting viewing angles are plotted in Fig. 7.2, for all sources that have a combined fit (recall that some sources are not able to be fit by the combined model as their SED is bluer than the assumed blue power law), and also for those sources that are best fit by the combined model when compared to the power law model. The angles for the BL Lacs (which are all fit best by the combined model) are also shown. The average and median angles are shown in Table 7.2. Not all sources are plotted on any graph. This is due to some sources having an R value outside the range specified in Eqn. 7.5. Either Doppler beaming alone cannot explain these large (or small) ratios of jet flux to core flux, or the assumed values of or 2 do not apply. Note that for the smaller angle ­ high energetic case for the optical jets, the PHFS angles are smaller, with the distribution concentrated below 15 . For the less energetic case, the distribution is more spread out, and there are fewer sources for which the beaming model works (i.e. for which the value of R is within the allowed range). An interesting point to draw out of Fig. 7.2 is that the BL Lacs are not preferentially at the low angles, as might be expected from the pure beaming hypothesis (i.e. that the BL Lac phenomenon is due to a relativistic jet viewed at a small angle to the line of sight). Instead, they are consistent with having the same distribution as the non-BL Lac synchrotron sources. This is largely due to the spread in values of R seen in Fig. 5.7. This raises an interesting point: are the BL Lacs significantly different from the non-BL Lacs in the PHFS, for any of their properties? The model fitting performed in Chapter 5 provides an excellent basis to do this, and


7.3 Mo delling results for the PHFS BL Lacs this will be discussed in the following section.

115

7.3

Mo delling results for the PHFS BL Lacs

We can use the modelling results to explore the differences (and similarities) between the quasars and BL Lacs in the PHFS, since the sample contains both types of ob jects. In determining which sources in the PHFS are BL Lacs, we have used two particular catalogues. The first is that of Padovani and Giommi (1995), which was used as the primary source of classification as either a BL Lac or a BL Lac candidate (more on this classification below). The second is Table 2 of V´ eron-Cetty and V´ eron (2000) ­ this was used if a source was not listed in Padovani and Giommi (1995). It should be noted that whether a source is a BL Lac or not is sub ject to some uncertainty, as it depends on the quality of the optical spectra used for the determination, and also on the source itself. Some BL Lacs have been shown to exhibit broad emission lines when they are at a low overall flux level, including BL LacertÔ itself ! (Corbett et al. 1996; Vermeulen et al. 1995) (We do note, however, that these lines, when seen, do have quite small equivalent widths ­ albeit greater than the 5° cutoff for BL Lacs.) Another A example is the BL Lac 0537-441 (see the spectrum, taken by Wilkes et al. (1983), in the plot in Appendix D, which exhibits a significant Mg ii line). Often sources that may be BL Lacs are classed as "BL Lac Candidates", which generally means that the spectrum used shows no significant emission lines, but that the spectrum itself was not of sufficient quality to confidently declare the ob ject a BL Lac.

7.3.1

Parameters from mo delling

Are the BL Lac ob jects physically different from the other PHFS quasars, or do they represent just one extreme of a continuum of variations (in this case, in emission line strength)? The distributions for a number of key parameters from the fitting (the power law slope, the peak synchrotron wavelength and the relative strength of the synchrotron component at 0.5µm) are shown in Fig. 7.3, along with the distribution of both the absolute magnitudes and the equivalent widths of the Mg ii 2798 and H lines. These equivalent widths are taken from Francis et al. (2001) ­ details of the measurements can be found therein. Note that the spectra are not simultaneous with the


116

BL Lacs in the PHFS

Figure 7.3: A summary of the modelling (and other) results, showing where BL Lacs are in relation to the rest of the PHFS. For all plots, the open histogram shows all sources with a good power law fit (with the exception of (a), where it is all sources that are used in the model fitting ­ i.e. from Table B.1), the green shows those sources best fit by the combined model, while the red shows the BL Lacs. (a) Absolute V magnitudes of the sources, calculated from the photometry using H0 = 75 km s-1 Mpc-1 and q0 = 0.5. (b) Power law index from the power law fit of Section 5.1. (c) The peak wavelength p from the combined fit of Section 5.2 (for sources best fit by the combined model). (d) The ratio of synchrotron to power law flux for the combined model at 0.5µm (again, for sources best fit by the combined model). (e) Equivalent widths of Mg ii. (f ) Equivalent widths of H .


7.3 Mo delling results for the PHFS BL Lacs photometry observations. The median values for each parameter are listed in Table 7.3, for BL Lacs as well as the non-BL Lacs (including all the non-BL Lacs and those best fit with a synchrotron component). Firstly, how do the luminosities of the BL Lacs compare to the rest of the sample? Fig. 7.3a shows the values of MV (calculated in the same way as in Section 4.4, with H0 = 75 km s
-1

117

Mpc

-1

and q0 = 0.5), for all sources that

have been modelled. Also shown are the distributions of the synchrotrondominated sources (i.e. those best fitted with the combined model), and the BL Lac ob jects. There is no statistical difference between the BL Lac ob jects and the rest of the synchrotron-dominated sources, and minimal difference (still not statistically significant) between the BL Lacs and the rest of the sample as a whole. The next point to note is that, in comparison to the whole PHFS, the BL Lacs are redder (Fig. 7.3b). They are also, in fact, redder as a group than the ob jects that are best fitted by the combined model, although the difference here is significant only at the 95% level ­ see Table 7.3. If just the synchrotron fits for the combined model sources are considered, there is very little difference between BL Lacs and non-BL Lacs. The median p (Fig. 7.3c) is slightly shorter for BL Lacs (0.44µm compared to 0.67µm), but this is not significant ­ the two distributions are consistent with being from the same parent distribution. Similarly, there is no statistical difference between the two synchrotron ratio distributions (Fig. 7.3d), although the BL Lacs do have a larger median ratio (15.7 compared to 6.1 for the non-BL Lacs). In Table 7.3 are also listed the Kolmogorov-Smirnov probabilities that the BL Lac and non-BL Lac distributions come from the same parent distribution ­ a low probability indicates that the parent distributions are likely to be different.

7.3.2

Emission lines and BL Lacs

If one looks at the equivalent widths of both the Mg ii and H emission lines (Fig. 7.3e,f ), it can be seen that the BL Lac ob jects are at the lower extreme of the distribution, consistent with their definition. Note also that the non-BL Lac synchrotron sources have a large range of equivalent widths: some have very small emission lines, while others have quite large ones. Does the synchrotron component that has been fitted to the optical/NIR continuum decrease the equivalent width of the emission lines, as expected


118 Parameter (mag) P L p (µm) R(0.5µm) W (H ) (° A) W (Mg ii) (° A)
V

BL Lacs in the PHFS Median values Comb. BL Lacs -24.07 -24.34 -0.78 -0.42 0.67 0.44 6.10 15.67 45.59 2.59 30.33 5.33 K.-S. Prob. (%) (All) (Comb.) 20.25 81.92 0.01 4.60 -- 24.21 -- 97.82 2.73 4.26 0.46 2.54

M

All -25.30 -1.22 -- -- 46.67 31.79

Table 7.3: This table shows the median values of the distributions in Fig. 7.3 for BL Lacs compared to the non-BL Lac sources (All = all non-BL Lacs with a good power law fit; Comb. = all non-BL Lacs best fit with the combined model). Also shown are the Kolmogorov-Smirnov probabilities that the BL Lac and non-BL Lac distributions are from the same parent distribution.

under the "swamping hypothesis"? One would expect that adding a nonionising synchrotron component to the continuum will have the effect of reducing the equivalent width of the emission lines from the BLR, due to the fact that the flux in the emission lines is not changed but the continuum flux is increased. To test this prediction, we compared the equivalent widths of five emission lines (C iv 1549, C iii] 1909, Mg ii 2798, H , and the doublet [O iii] 4959,5007) with the ratio of synchrotron to continuum flux at the line wavelength, to see if some form of an anti-correlation is present. The details of the spectral dataset will presented elsewhere (Francis et al. 2001). Objects that had spectra taken were essentially a random sample of the PHFS (sub ject to visibility during the observing runs). Fig. 7.4 shows the results of this comparison for the Mg ii and H lines, which are the two broad lines with the longest wavelength (and hence the two lines most likely to show a reduction in equivalent width). Those sources best fit with the power law model are also shown: the value of the ratio used for these sources was taken from fitting the combined model, and so are upper limits to the ratio. The Mg ii line does not show much relationship to the synchrotron ratio, while the H line does show a reduction in equivalent width with increasing amount of synchrotron. This lends some support to the hypothesis that excess synchrotron light is present. The difference in the two plots is likely due to the presence of the turnover in the synchrotron flux, so that it has less


7.4 Implications for the nature of BL Lacs

119

Figure 7.4: The equivalent width of a) the Mg ii and b) H lines as a function of the ratio of synchrotron to power law flux at the emitted wavelengths. The sources are given different symbols according to the nature of their best fit model (the arrow indicates an upper limit for a source that was best fit with the combined model). Ratios for sources best fit by the power law model are calculated from the combined model fits, and so are upper limits to the ratios (for the crosses and stars). The dashed lines represent the expected change in equivalent width with increasing amount of synchrotron, for an emission line with intrinsic equivalent width of 100° A.

effect at the shorter wavelength of Mg ii. There are, however, several sources that have high synchrotron ratios together with high equivalent widths.

7.4

Implications for the nature of BL Lacs

What do the results of the modelling for the BL Lacs tell us? Firstly, the fitted values of p and R indicate that the BL Lacs, as a group, are not statistically different from the rest of the sources with synchrotron-dominated spectra. They are generally redder than the population as a whole, and slightly (but not significantly) redder than the synchrotron-dominated sources. The bulk of the difference in this latter case seems to be made up by the sources with a blue optical SED and a redder NIR ­ i.e. where the synchrotron component turns over in the optical range. The main difference between BL Lacs and non-BL Lacs is, not surprisingly, the equivalent widths of the emission lines. By selection, the BL Lacs are preferentially the ob jects with low equivalent widths. The median H ° equivalent widths for the BL Lacs and non-BL Lacs are 2.6A and 46.7° A ° respectively (or 45.6A if just the synchrotron-dominated non-BL Lacs are used).


120

BL Lacs in the PHFS

7.4.1

Emission lines

Why do the emission lines have such small equivalent widths? The equivalent width is calculated by dividing the integrated line flux that is above the continuum by the average continuum level over the extent of the line. Thus, a comparatively low equivalent width means that either the continuum level is comparatively high, or the flux in the line is comparatively low. We note first that there is evidence from the literature that the level of the continuum is important, as emission lines have been seen in BL Lacs when they are at a low flux level (e.g. BL LacertÔ). However, even in these cases, the lines are still relatively weak (BL LacertÔ's H equivalent width ° was measured by Vermeulen et al. (1995) to be 15.3A, and only 5.6° by A Corbett et al. (1996)). So it seems that a large continuum cannot be the whole story, although it is undoubtedly important in many sources. If a large continuum is important for the PHFS BL Lacs, could this be accounted for by the synchrotron component? Fig. 7.4b partially supports this hypothesis, as there is a trend for the equivalent widths of H to decrease with increasing amount of synchrotron. However, there does appear to be a large dispersion in the distribution of equivalent widths, particularly at high synchrotron ratios, with sources present that have relatively high ratios at H , as well as large equivalent widths. Also, there is little evidence that the level of the continuum in BL Lacs is larger than that in the non-BL Lac sources. Certainly the ratios of synchrotron to power law appear to be larger (although statistically they are consistent with being the same as the rest of the synchrotron-dominated sample), and this accounts for the slightly redder colour of the BL Lacs when compared to the rest of the synchrotron-dominated sources. However, the continuum levels are not very much larger than those of the non-BL Lacs. This is seen in Fig. 7.5, where the equivalent widths of the Mg ii and H lines are compared with the luminosity of the fitted SED (i.e. approximately the continuum level) at the line wavelength. The luminosities of the BL Lacs have the same range of values as the non-BL Lacs, and there is no correlation between the luminosity of the continuum, represented by the value of the fitted SED, and the equivalent width of the emission lines. We do add the standard caveat here that the spectra and the photometry (to which the models are fitted) are non-simultaneous, so there is always the possibility that the source has varied in between the two observations.


7.4 Implications for the nature of BL Lacs

121

Figure 7.5: Equivalent width of emission lines as a function of the luminosity of the
fitted model at the wavelength of the line. The different symbols show either if the source is a BL Lac or, if not, whether it was best fit by the combined model or the power law model.

Figure 7.6: A histogram of luminosities of the fitted power law components, evaluated at a rest wavelength of 0.5µm (with H0 = 75 km s-1 Mpc-1 and q0 = 0.5). The open histogram is all sources with a good power law fit, the green is those sources best fit by the combined model, while the red is the BL Lacs.


122

BL Lacs in the PHFS An alternative explanation for the small equivalent widths is that the line fluxes in the BL Lacs are intrinsically small compared to the non-BL Lacs. This could be due to one of two reasons: there are no (or not enough) ionising photons present to provide the necessary excitation (for example, due to a small, or absent, accretion disk); or there is not enough emission line gas. Do the BL Lacs have accretion disks? If so, are they as powerful as those seen in the non-BL Lac ob jects? These questions can be investigated by examining the strength of the power law component in the model fits. The luminosities of the power law components (either the BBB power law in the combined fit or the single power law from the power law fit) are shown in Fig 7.6, calculated at a rest wavelength of 0.5µm, with H0 = 75 km s
-1

Mpc

-1

and q0 = 0.5.

The BL Lac distribution is clearly different from that of the rest of the sample (and is significant at the 99% level). However, there is no difference between the BL Lac distribution and that of the rest of the sources best fit by the combined model, indicating that the BL Lacs do not have significantly smaller blue power laws than the non-BL Lac synchrotrondominated sources. We get a similar result if the power law luminosity is calculated at a wavelength of 0.1µm (this is closer to the likely energies of the ionising photons). We do note, however, that visual inspection of the fits shows the presence of a significant power law component in only two sources (0537-441, which, as was noted in Section 4.5, is a highly variable source and may have an SED affected by short-term variability; and 2131-021). The rest of the BL Lacs have very small contributions from the power law component, even at the blue end of the SED. This is, of course, also the case for many of the non-BL Lac sources (although many of these have red SEDs that may be due to dust absorption rather than synchrotron emission), but is indicative of the fact that the BL Lacs have their optical SEDs dominated by synchrotron. Nevertheless, the similarity of the power law strengths in the BL Lacs and some of the non-BL Lacs may indicate that the cause of the small emission lines in BL Lacs, rather than being a lack of ionising photons, is instead a simple lack of emission line gas.


7.4 Implications for the nature of BL Lacs

123

7.4.2

UV sp ectra

A further way to differentiate between different causes of the small emission lines in BL Lacs would be to examine their ultra-violet spectra. If the emission lines are being swamped by a non-thermal (i.e. synchrotron) emission component, then, as is shown in the modelling in this thesis, this component will turn over at some wavelength. It is therefore likely to be far less important at the shorter wavelengths, near for instance the emission lines of Ly and C iv. These lines need to be observed in the UV, as most BL Lacs are not of sufficient redshift for the lines to move into the optical region. If the UV lines are strongly visible (i.e. do not seem to be reduced or swamped in any way), then this would indicate that the lack of emission lines at optical wavelengths can be explained by the swamping by a continuum that turns at longer wavelengths that the UV lines. If, however, the UV lines are either weak or absent, then either the swamping continuum extends past the UV, or the emission line region is intrinsically weak. A large resource with which to perform this test is the catalogue of IUE spectra of quasars and BL Lacs by Lanzetta, Turnshek, and Sandoval (1993). This study contained 24 sources from the PHFS, including 7 sources classed as BL Lacs (namely 0048-097, 0118-272, 0301-243, 0422+004, 0537-441, 0829+046 and 1253-055 (3C 279)). None of the BL Lacs showed any significant emission lines in their spectra (that generally ranged from 1200° to 3200° while many of the other PHFS quasars showed strong A A), emission lines (typically Ly, C iv, and sometimes C iii]). This indicates that the lack of emission lines seen in BL Lacs in the optical extends to shorter wavelengths as well. This is confirmed by Kinney et al. (1991), who presented IUE spectra of 69 quasars and "blazars". The ma jority of the blazar ob jects presented showed no significant emission lines in their UV spectra. We do note that Wills et al. (1995), who obtained UV spectra of 31 radioloud quasars with HST, did observe emission lines (Ly, C iv and C iii]) in 1253-055. However, in each case these lines had the smallest equivalent widths of all ob jects observed. Quasars are very active in the ultraviolet (Kinney et al. 1991), and so it is likely that the synchrotron continuum (should that be present and responsible for diluting the emission lines) had varied in this source in the period between the two observations (1253-055 is seen to be very variable in other frequency bands).


124

BL Lacs in the PHFS

7.4.3

Unification mo dels

Orientation-based What do these results tell us about possible unification models for radio-loud AGN? One of the main unification models is that describing BL Lacs and quasars respectively as beamed versions of FR I and FR II radio galaxies (i.e. viewed at small angles to the direction of propagation of the relativistic jet). Can we explain the observed differences (or similarities) between the BL Lacs and quasars in the PHFS in terms of this model? One difference between the two classes is their optical­NIR colour, with BL Lacs being redder. This is almost certainly due to the presence of a strong synchrotron component in all the BL Lacs (while there is not one in all the non-BL Lac quasars), and so does not relate to the nature of the parent population. The second, and ma jor, difference is the distribution of equivalent widths, with BL Lacs having substantially weaker emission lines than the non-BL Lacs, due to a relative lack of emission line flux. Is this compatible with the unification model? FR Is, the putative parent population of BL Lacs, do have significantly lower emission line luminosities than FR IIs of the same total radio power (Zirbel and Baum 1995), which is in accord with our finding here. We do note also that some FR IIs only show weak, low-excitation lines in their spectra (Laing et al. 1994), and so may also be included in the BL Lac parent population. Other than this finding (which is consistent with previous results), the modelling does not shed much light on the veracity or otherwise of this unification model. Synchrotron-based Another popular unification model is that of Fossati et al. (1998), which unifies all "flavours" of BL Lacs (from LBL to HBL) as well as FSRQs, on the basis of the peak of the synchrotron component. Under this scheme, the multi-wavelength SEDs of blazars are made up primarily of two broad peaks, representing the synchrotron emission (the low-frequency peak ­ roughly IR to soft X-rays) and the inverse-Compton emission (the high-frequency peak ­ roughly hard X-rays to gamma rays). The frequency of the synchrotron peak is related to the overall luminosity of the source, in the sense that the more luminous ob jects (which are generally the FSRQs) have lower frequency peaks, while the less luminous ob jects (HBLs) have higher frequency peaks.


7.4 Implications for the nature of BL Lacs

125

Figure 7.7: Comparison of the PHFS sources with the data from Fossati et al. (1998)
and Fossati et al. (1997). a) Radio luminosity versus peak frequency. b) Peak luminosity versus peak frequency. c) Radio­optical slope versus peak frequency. d) Radio­optical slope versus optical­X-ray slope. The red points are data from Fossati et al. (1998) (for (a), (b) & (c)) and Fossati et al. (1997) ((d) only): star = FSRQs; circle = 1 Jy BL Lacs; triangle = Slew BL Lacs; square (in (d)) = EMSS BL Lacs. Green plus signs in (d) are from best fitting SED parametrisations of Fossati et al. (1997). The other points are PHFS sources, with symbols relating to their best fit from the model fitting: cross ­ best fit by power law; circle ­ best fit by combined model; circle with plus ­ BL Lac.


126

BL Lacs in the PHFS The ob jects on which this analysis was based were taken from three complete samples of blazars: the Slew survey (an X-ray survey made with the Einstein satellite (Elvis et al. 1992), from which was selected a sample of BL Lacs (Perlman et al. 1996)); the 1 Jy sample of BL Lacs (a radio survey, (Stickel et al. 1994)); and the 2 Jy survey of FSRQs (another radio survey ­ the sample is that of Padovani and Urry (1992), drawn from Wall and Peacock (1985)). The result of this sampling is that the sources that are analogous to the PHFS sources are all bright (either brighter than 1 Jy or 2 Jy depending on the survey, whereas the PHFS sources go down to 0.5 Jy), and the data used is non-simultaneous. We can therefore use the PHFS sources to explore whether the conclusions hold as the radio flux is reduced, and use the simultaneous photometry to get a better idea of where the synchrotron peak is. Fossati et al. (1998) find a number of correlations between luminosity, synchrotron peak frequency, and radio­optical spectral slope. We show these in Fig. 7.7, and compare their data with that of the PHFS sources. We also compare the PHFS data with the model SEDs of Fossati et al. (1997) in a colour-colour diagram. The peak frequencies used for the PHFS sources are defined as
peak

= c/5p , which gives the approximate value of the
peak

peak of the f SED (this matches the

values found by Fossati et al.

(1998)). The radio fluxes used are either the contemporaneous fluxes from the ATCA, or, if these are not available, the fluxes from PKSCAT90 (Wright and Otrupcek 1990). The X-ray fluxes used in Fig 7.7d are from Siebert et al. (1998). (Note of course that neither the radio fluxes nor the X-ray fluxes are simultaneous with the optical photometry.) The two luminosity­peak plots (Figs 7.7a&b) show that the sources from the PHFS are somewhat consistent with the data of Fossati et al. (1998), although the fact that the PHFS sources are of lower luminosity has resulted in lower luminosities at both the radio and peak frequencies. The correlation of peak luminosity with peak frequency is consequently reduced (Fig. 7.7b). The ma jor difference is apparent in Fig. 7.7c. The PHFS sources have a greater range in optical fluxes and are generally lower in flux (see Fig. 7.8), and so even though the radio fluxes are generally less than the 1 Jy and 2 Jy sources, the radio­optical slope is still steeper for many sources (the slopes here are defined such that F
-

). This moves the PHFS points

away from the Fossati et al. points (and also away from the locus of points


7.4 Implications for the nature of BL Lacs

127

Figure 7.8: Comparison of optical fluxes (at V band) of PHFS sources (the red histogram) and sources from Fossati et al. (1998) (the blue histogram). The PHFS fluxes are calculated from the quasi-simultaneous photometry (Chapter 4).

predicted by their "synthetic" SEDs, which, although not shown in Fig. 7.7c, form a smooth curve through the middle of the red points). Finally, Fig. 7.7d shows the radio­optical slope as a function of the optical­X-ray slope, for data from Fossati et al. (1997) (a slightly different data set from that of Fossati et al. (1998)), along with predictions from their SED parametrisation. The PHFS sources are also shown. They tend to fall in approximately the same region of the plot as the FSRQs and 1 Jy BL Lacs, as one would expect. However, on closer examination, we see that the ma jority of the points lie below the predicted points (in green), based on the "two-peak" SED model. This has two possible explanations, and both are important for the PHFS sources. Firstly, the radio flux will be slightly lower for many of the PHFS sources, since the radio flux limit is less, which will reduce the value of . Secondly, from the fitting in Chapter 5, many of the PHFS sources have some power law emission present at optical wavelengths, which is not due to synchrotron emission. This component, which is not included in the Fossati et al. models, will contribute a substantial amount of the flux, thereby reducing the value of from that expected of a pure synchrotron model.


128

BL Lacs in the PHFS

7.5

Summary

The model-fitting performed in Chapter 5 provides an excellent way to compare the BL Lacs with the non-BL Lacs in the PHFS. We find that the BL Lacs are generally redder than the rest of the sample, particularly when compared with the sources best fit with a power law. However, this is the only significant different to arise out of the modelling. In all other model parameters, the BL Lacs are consistent with having the same distribution as the non-BL Lacs. The main difference between the two classes is, not surprisingly, the equivalent widths of the emission lines. The BL Lacs have much smaller equivalent widths ­ on average an order of magnitude smaller. This seems to be caused by the emission lines having intrinsically low fluxes, as there is little difference between the continuum luminosities of the BL Lacs and the non-BL Lac sources, and no correlation between the continuum luminosity and the equivalent widths. The low emission line fluxes in the BL Lacs are more likely to be due to a relative lack of emission line gas, since the luminosities of the power law component of the combined fit (which has been taken to represent the emission from the accretion disk, or big blue bump) are no different between the BL Lacs and non-BL Lacs. This suggests that both classes of sources have the same amount of ionising photons, but the BL Lacs have less gas to get ionised, resulting in lower emission line fluxes. Finally, what can we learn about unification schemes from this analysis? The BL Lacs do tend to have relatively low emission line luminosities, which is consistent with them having FR I radio galaxies as their parent population. However, the fact that we find little difference (apart from emission line properties) between the BL Lacs and the other synchrotron-dominated quasars indicates that the region from which the optical synchrotron emission originates is similar in both types of ob jects. In the context of unification schemes (Urry and Padovani 1995), this probably indicates that the jets in FR Is and FR IIs have similar structure, particularly in the inner regions (where the optical synchrotron emission is more likely to originate). The analysis of the "two-peak" unification model (Fossati et al. 1998) produced a couple of interesting conclusions. We have seen that by using less luminous ob jects (the PHFS has a lower flux cut-off than the samples


7.5 Summary used by Fossati et al. (1998), and the ob jects are often much fainter in the optical), the correlations found in the model are not as strong. Also, the Fossati et al. model does not account for the presence of the power law emission that is present (or indeed dominant) in many of the PHFS sources. This shows the need to account for all types of emission mechanisms when building models for radio-loud AGN.

129


130

BL Lacs in the PHFS