1: \documentclass{aa}
2: \usepackage{graphics,latexsym,amssymb,times,psfig}
3: \def\gsim{ \lower .75ex \hbox{$\sim$} \llap{\raise .27ex \hbox{$>$}} }
4: \def\lsim{ \ls\csiower .75ex\hbox{$\sim$} \llap{\raise .27ex \hbox{$<$}} }
5:
6: \begin{document}
7:
8: \title{Internal shocks and the blazar sequence}
9: \subtitle{Low and intermediate power BL Lac objects}
10:
11: \author{
12: Dafne Guetta \inst{1,}\inst{2},
13: Gabriele Ghisellini \inst{3},
14: Davide Lazzati \inst{4} and
15: Annalisa Celotti \inst{5}
16: }
17:
18: \offprints{D. Guetta; dafne@arcetri.astro.it}
19: \institute{ INAF--Osservatorio Astrofisico di Arcetri, Largo E. Fermi
20: 5, I--50125 Firenze, Italy; \and JILA, University of Colorado,
21: Boulder, CO 80309, USA; \and INAF--Osservatorio Astronomico di Brera,
22: via Bianchi 46, I--23807 Merate, Italy; \and Institute of Astronomy,
23: Madingley Road, CB3 0HA, Cambridge, UK; \and SISSA/ISAS, via Beirut
24: 2-4, I--34014 Trieste, Italy.}
25:
26:
27: \titlerunning{Internal shocks in blazars}
28: \authorrunning{Guetta et al.}
29:
30:
31: \abstract{ We consider internal shocks as the main dissipation
32: mechanism responsible for the emission in blazars and show that it can
33: satisfactorily account for the properties of all blazars. In
34: particular, we extend previous work (Spada et al. 2001) on powerful
35: objects, to intermediate (BL Lac) and low power sources (Mkn 421), in
36: order to reproduce the whole of the blazar sequence. The model
37: self-consistently treats the dynamics, spectral emission and its
38: variability. The key parameters driving the phenomenological sequence
39: are the jet power and the properties of the broad line region, which
40: regulate the cooling efficiency of the emitting particles and thus the
41: shape of the spectral energy distribution. By assuming that the
42: remaining parameters are similar for all objects it has been possible
43: to reproduce the full range of the observed spectral ``states". A more
44: detailed comparison of the variability properties shows (for Mkn 421)
45: a good agreement in the X--ray band, while in the optical the
46: simulated flux appears to be too variable. For BL Lac lags ($\sim$ 10
47: days) are predicted between the $\gamma$--rays and the infrared
48: emission.
49: %
50: \keywords{Galaxies: jets --- Galaxies: nuclei --- Radio continuum:
51: galaxies -- Objects: Mkn 421, BL Lac.}
52: }
53: \maketitle
54:
55: \section{Introduction}
56: The discovery that blazars are strong $\gamma$--ray emitters together
57: with the results of multiwavelength campaigns have allowed to deepen
58: our knowledge on these objects. Their Spectral Energy Distribution
59: (SED) is characterized by two broad peaks (Fossati et al. 1998)
60: strongly variable on different timescales (Wagner \& Witzel 1995; Ulrich,
61: Maraschi \& Urry 1997).
62:
63: Two main radiation processes dominate the emission, namely synchrotron
64: at low frequencies and -- plausibly -- inverse Compton at high
65: energies (see e.g. Sikora 1994 for a review). The relative luminosity
66: in the two peaks and their peak frequency appear to be functions of
67: the total power (Fossati et al 1998, Ghisellini et al. 1998),
68: resulting into a sequence for the whole of the blazar population
69: properties, ranging from powerful, low frequency peak, through
70: intermediate, to low power, high frequency peak (blue) blazars (see
71: however Padovani et al. 2003). The blazar emission is variable on
72: energy dependent timescales, typically of weeks--months in the
73: radio and of the order of a day in the $\gamma$--ray band.
74:
75: Several studies, mainly based on the modeling of the SED, consistently
76: derived the physical parameters of the emitting region. However key
77: issues in the understanding of relativistic jets remain open, most
78: notably the jet energetics and the particle acceleration process(es).
79: In order to explore these issues and their relationship we have
80: (quantitatively) considered a scenario in which the plasma conditions
81: and their variability are not treated as free parameters, but follow
82: from the jet dynamics, thus relating the observed emission properties
83: with the energy transport along the jet. Such a scenario assumes that
84: internal shocks are responsible for the dissipation within jets (Rees
85: 1978 and then mostly explored for Gamma--Ray Bursts, Rees \&
86: M\'esz\'aros 1994).
87:
88: The key assumption of the model is that the energy is channeled into
89: jets in an intermittent way by the central engine, though such a time
90: dependent process cannot be inferred from first principles. Different
91: parts of the jet (`shells') moving at different speeds can collide
92: giving rise to shocks and dissipation as non--thermal radiation. The
93: mechanism has a limited efficiency (unless the contrast in Lorentz
94: factors between different shells is extremely large, see Beloborodov
95: 2000; Guetta, Spada \& Waxman 2001, but also Ghisellini 2002), which
96: has to be indeed the case for blazars since most of the energy
97: propagates up to the extended radio lobes.
98:
99: Beside the low efficiency, the internal shock scenario can naturally
100: account for other blazar properties. It predicts that jets become
101: radiative at $\gsim 10^{16}- 10^{17}$ cm from the central engine,
102: implying a minimum distance from the accretion disk and a minimum
103: dimension for the $\gamma$--ray source, as required by observations
104: (in order to avoid copious pair production; see Ghisellini \& Madau
105: 1996). Furthermore, successive collisions taking place at larger
106: distances have reduced efficiency -- since the Lorentz factor contrast
107: of the colliding shells decreases -- explaining why the jet luminosity
108: decreases with distance. Finally, the internal shock scenario appears
109: a promising non steady state model to account for the observed large
110: amplitude variability.
111:
112: A detail study of the predictions of this model via numerical
113: simulations has been carried out by Spada et al. (2001; hereafter S01)
114: for powerful blazars (specifically 3C 279). As the SED of this object
115: does not represent the whole of the blazar family mentioned above, we
116: focus here on two well studied sources representative of extremely
117: blue and intermediate blazars, namely Mkn 421 and BL Lac itself.
118: Rather than reproducing in detail particular spectra of specific
119: objects, the aim of this work is to determine whether the proposed
120: scenario, including the dynamics and emission properties, i) can
121: account for the different SED along the blazar sequence, ii) under
122: which hypothesis this can occur and iii) whether the internal shock
123: model can reproduce the observed variability behavior.
124:
125: The outline of the paper is the following. In \S 2 we describe the
126: hypothesis on the wind dynamics and on the radiative properties of the
127: shocked plasma. The results -- specifically referring to the sources
128: Mkn421 and BL Lac -- are presented in \S 3. In \S 4 we draw our
129: conclusions.
130:
131:
132: \section{Internal shocks and radiated spectra}
133:
134: \subsection{Physical scenario and dynamics}
135:
136: In this work we use the approximate model of an unsteady wind
137: described in detail in S01 and in the following we report only its
138: most relevant assumptions.
139:
140: We consider a compact source, of typical dimension $R_0\sim 10^{14}$
141: cm, which produces an unstable relativistic wind characterized by an
142: average luminosity $L_{\rm w}$. The emission from the wind is
143: obtained by adding pulses radiated in a series of internal shocks that
144: occur in the outflow (Daigne \& Mochkovitch 1998; Spada, Panaitescu \&
145: M\'esz\'aros 2000; Guetta, Spada \& Waxman 2001).
146:
147: The wind flow is described as a sequence of $N=t_{\rm w}/t_{\rm v}$
148: shells, where $t_{\rm w}$ is the overall duration of the wind emission
149: and $t_{\rm v}\ll t_{\rm w}$ is the average interval between two
150: consecutive shells. Each shell is characterized by four parameters:
151:
152: \begin{enumerate}
153:
154: \item the ejection time $t_{\rm j}$, where the subscript j denotes the
155: $j^{\rm th}$ shell,
156:
157: \item the Lorentz factor $\Gamma_{\rm j}$,
158:
159: \item the mass $M_{\rm j}$,
160:
161: \item the width $\Delta_{\rm j}$ as measured in the lab frame.
162:
163: \end{enumerate}
164:
165: The Lorentz factors of the shells are randomly extracted from a
166: uniform distribution between $\Gamma_m$ and $\Gamma_M$. The masses
167: are also randomly extracted from a uniform distribution with an
168: average value $M_t/N$, where the total mass in the wind $M_t$ is
169: normalized to the jet energy budget. The time intervals $t_{j+1}-t_j$
170: are drawn randomly from a uniform distribution with an average value
171: $t_v$. The initial shell width $\Delta_j$ is of the same order of
172: $R_0$ and the initial internal energy $U_j$ is negligible (all of the
173: energy is in the form of bulk kinetic energy).
174:
175: The dynamics of the wind expansion is characterized by a series of
176: two--shell collisions in which the faster shells catch up with the
177: slower ones. The model computes the collision time of two shells and
178: replaces them with a new shell having mass, Lorentz factor, and
179: internal energy given by the conservation laws for inelastic
180: collisions (for details see S01). Before colliding, the shells are
181: assumed cold with constant thickness $\Delta=R_0$ but after collision
182: they get heated and thus assumed to expand at the (comoving) sound
183: velocity. Adiabatic losses during the expansion are taken into
184: account but the bulk Lorentz factor is kept constant since the
185: internal energy is always much smaller than the bulk kinetic one. The
186: magnetic field is generated at each collision without keeping memory
187: of any field generated in previous collisions.
188:
189: In each collision a forward (FS) and a reverse (RS) shock are formed
190: and propagate into the front and the back shell, respectively. The
191: plasma parameters behind each shock are determined by the jump
192: conditions in mildly relativistic shocks, namely continuity in the
193: energy density and velocity across the contact discontinuity
194: separating the two shells (Panaitescu \& M\'esz\'aros 1999). We also take
195: into account the compression of the merged shell width due to the
196: propagation of the FS and RS.
197:
198: The energy released in each shock is assumed to be distributed among
199: protons, electrons and magnetic field with fractions $\epsilon_p$,
200: $\epsilon_{\rm e}$ and $\epsilon_{\rm B}$, respectively. The resulting
201: magnetic field strength allows us to evaluate the synchrotron and the
202: inverse Compton emission.
203:
204: Electrons are assumed to be accelerated at relativistic energies with
205: a power law spectrum, $N(\gamma)$, above a (random) Lorentz factor
206: $\gamma_{\rm b}$. $\gamma_{\rm b}$ is determined by requiring that
207: $\int N(\gamma) \gamma m_e c^2 d\gamma$ equals the fraction
208: $\epsilon_{\rm e}$ of the total available internal energy and
209: accounting for the fact that only a fraction, $\zeta_{\rm e}$, of
210: electrons is effectively accelerated. We assume that below
211: $\gamma_{\rm b}$ electrons are accelerated with smaller efficiency,
212: corresponding to a flatter energy distribution (with respect to the
213: slope above $\gamma_{\rm b}$), assumed to be $\propto \gamma^{-1}$.
214:
215: In conclusion, by considering the wind dynamics and shock
216: hydrodynamics, the energy density in non--thermal electrons and in
217: magnetic field and the typical random Lorentz factor $\gamma_{\rm b}$
218: can be determined for each collision.
219:
220: \subsection{Local spectra}
221:
222: We followed the same basic assumptions of S01, which we briefly
223: recall:
224:
225: \begin{itemize}
226:
227: \item
228: The emitting zone is homogeneous with comoving volume $V=\pi \psi^2
229: R^2\Delta R^\prime$, where $\psi$ is the half--opening angle of a
230: conical jet and $\Delta R^\prime$ is the comoving thickness of the
231: shell at the collision time, whose expansion while emitting is
232: neglected for simplicity.
233:
234: \item
235: The relativistic particles, embedded in a tangled magnetic field, have
236: the same energy distribution throughout the region corresponding to
237: the shell--shell interaction. This simplification is justified as we
238: are interested in the ``average" spectrum, but does not account for
239: spectral details occurring on a timescale shorter than the light
240: crossing time.
241:
242: \item
243: Even in the case of low power BL Lacs, we consider the presence of
244: soft photons external to the jet (identified with the emission from
245: the broad line region, BLR). The external photon
246: luminosity $L_{\rm ext}=a L_{\rm disk}$ is produced within $R_{\rm
247: BLR}$ and abruptly vanishes beyond, and corresponds to a comoving
248: radiation energy density $U_{\rm ext} = (17/12) aL_{\rm disk} \Gamma^2
249: /(4 \pi R_{\rm BLR}^2 c)$ (e.g. Ghisellini \& Madau 1996).
250:
251: High and low power blazars have different line/photoionizing disk
252: luminosities. Following Kaspi et al. (2000), $R_{\rm BLR}$ and $L_{\rm
253: disk}$ are related by $R_{\rm BLR}\propto L_{\rm disk}^b$ with $b\sim
254: 0.7$. Thus blazars with weaker broad emission lines should have
255: smaller BLR, implying that the first collisions occur preferentially
256: outside $R_{\rm BLR}$.
257:
258: \item
259: Relativistic electrons are injected with a broken power--law energy
260: distribution, $\propto \gamma^{-1}$ below $\gamma_{\rm b}$ and
261: $\propto \gamma^{-s}$ between $\gamma_{\rm b}$ and $\gamma_{\rm max}$,
262: with a power $L_{\rm e}$. The equilibrium particle distribution
263: $N(\gamma)$, resulting from the continuous injection and cooling
264: processes, is determined through the following procedure.
265:
266: \noindent
267: Above the (comoving) random Lorentz factor $\gamma_{\rm cool}$,
268: corresponding to a cooling time comparable with the shock crossing
269: time $t_{\rm cross}$, the distribution is assumed to have a power law
270: index $p=s+1$. Below $\gamma_{\rm cool}$ there are two different
271: behaviors:
272:
273: i) if $\gamma_{\rm cool} > \gamma_{\rm b}$, we assume $N(\gamma) \propto
274: \gamma^{\rm -s}$ between $\gamma_{\rm b}$ and $\gamma_{\rm cool}$, and
275: $N(\gamma)\propto \gamma^{-1}$ for $\gamma < \gamma_{\rm b}$;
276:
277: ii) if $\gamma_{\rm cool} < \gamma_{\rm b}$, $N(\gamma) \propto
278: \gamma^{-2}$ between $\gamma_{\rm cool}$ and $\gamma_{\rm b}$, and
279: $N(\gamma)\propto \gamma^{-1}$ for $\gamma <\gamma_{\rm cool}$.
280:
281: \item
282: The normalization of $N(\gamma)$ is determined according to whether
283: particles with $\gamma\sim \gamma_{\rm b}$ can or cannot cool in the
284: timescale $t_{\rm cross}$ (fast and slow cooling regime,
285: respectively):
286:
287: {\bf Fast cooling regime ---} If most of $L_{\rm e}$ is radiated in a
288: timescale $<t_{\rm cross}$ (i.e. $\gamma_{\rm b} > \gamma_{\rm cool}$)
289: we apply a luminosity balance condition:
290: %
291: \begin{equation}
292: L_{\rm e} \, =\, V m_{\rm e}c^2 \int \dot\gamma N(\gamma) d \gamma,
293: \end{equation}
294: %
295: where $\dot \gamma$ is the radiative cooling rate, including
296: synchrotron, synchrotron self--Compton (SSC) and external Compton (EC)
297: losses. In order to allow for multiple Compton scatterings we a
298: priori calculate how many scattering orders $n_{\rm IC}$ can take
299: place before the Klein--Nishina regime is reached. By considering
300: $\gamma_{\rm b}$ as the relevant energy
301: %
302: \begin{equation}
303: n_{\rm IC}\, =\, {\ln (\gamma_{\rm b}/x_{\rm B}) \over \ln (4\gamma_{\rm
304: b}^2/3)} \, -1,
305: \end{equation}
306: %
307: where $x_{\rm B}\equiv h\nu_{\rm B}/(m_{\rm e}c^2)$ and $\nu_{\rm
308: B}=eB/(2\pi m_{\rm e} c)$ is the (non--relativistic) Larmor frequency.
309: We define $U_{\rm e} \equiv L_{\rm e}/(\pi R^2 c)$ and introduce the
310: Comptonization parameter $y$:
311: %
312: \begin{equation}
313: y\, \equiv \, {3\over 4} \sigma_{\rm T} \Delta R^\prime
314: \int \gamma^2 N(\gamma) d \gamma.
315: \end{equation}
316: %
317:
318: The equality between the injected and the radiated power in the fast
319: cooling regime translates into an equality between the electron and
320: the radiation energy densities, where the latter comprises the
321: synchrotron and the $n_{\rm IC}$ inverse Compton orders energy
322: density, i.e.
323:
324: %
325: \begin{equation}
326: U_{\rm e}\, = y U_{\rm B} \left( 1+{U_{\rm ext} \over U_{\rm B}} + y +
327: y^2 + \, ....\, y^{n_{\rm IC}}\right).
328: \end{equation}
329: %
330: This provides the normalization of $N(\gamma)$, assuming the above
331: slopes.
332:
333: {\bf Slow cooling regime ---} If most of the power is not radiated in
334: $t_{\rm cross}$, the electrons retain their energy and
335: %
336: \begin{equation}
337: E_{\rm e} \, =\, L_{\rm e} t_{\rm cross}\, = \, m_{\rm e}c^2 V \int
338: N(\gamma)\gamma d\gamma.
339: \end{equation}
340: %
341: Assuming the same distribution for $N(\gamma)$, its normalization can
342: be determined.
343:
344: \item
345: The slopes of the $N(\gamma)$ distribution are fixed, but the cooling
346: energy is found by iteration as $\gamma_{\rm cool}$ depends on the
347: amount of SSC radiation, which in turn depends on the exact shape of
348: $N(\gamma)$ (because of Klein--Nishina effects). Since $\gamma_{\rm
349: cool}$ determines the normalization of $N(\gamma)$, we re--iterate the
350: procedure described above until it converges (usually 3--4 iterations
351: are enough).
352:
353: \item
354: The synchrotron--self--absorption process, the inverse Compton
355: emission and the beaming of the radiation are treated as in S01.
356:
357: \end{itemize}
358:
359: \subsection{Observed spectrum}
360:
361: Both the particle and radiation local spectra thus determined are not
362: steady, but roughly correspond to those attained after a time $t_{\rm
363: cross}$ after the shell--shell interaction, at the maximum bolometric
364: radiative output. This spectrum -- adopted for the entire duration of
365: the emission -- has a normalization modulated as a function of time.
366:
367: Each photon pulse -- with its corresponding spectrum -- starts at the
368: time of a collision and lasts for a duration set by the combination of
369: the radiative, shock crossing and angular spread timescales. The
370: shape of the pulses (raise and decay in the light curves) is assumed
371: symmetric and linear with time if its duration is determined by
372: geometrical effects and/or by the shock crossing time, as generally
373: occurs for collisions inside the BLR (as typical of powerful blazar
374: like 3C279). If the pulse duration is instead set by cooling, the
375: pulse still rises linearly, but decays exponentially [$\propto
376: \exp(-t/3 t_{\rm cool}(\gamma_{\rm b}) )$], implying that the rise and
377: decay times can be different. This usually occurs in weaker blazars,
378: like Mkn 421. The pulses of intermediate BL Lac objects can be
379: characterized by either of these pulse shapes, depending on the broad
380: lines intensity and thus on whether collisions take place within
381: $R_{\rm BLR}$.
382:
383: The observed spectrum is determined by all of the photons reaching the
384: observer simultaneously: since these are produced at different
385: distances, their different light propagations times should be taken
386: into account. In other words, the resulting spectrum is obtained as
387: the convolution of the emission produced by shocks simultaneously
388: active in the observer frame, taking into account the radius at which
389: collisions occur and the photon travel path.
390:
391: \begin{figure}
392: \psfig{figure=1101f2.ps,angle=0,width=9cm}
393: \vskip -0.5 true cm
394: \caption{The SED of Mkn421 during different observational campaigns,
395: as labeled. For the data points see Costamante \& Ghisellini (2002)
396: and references therein.}
397: %\end{figure}
398:
399: %\begin{figure}
400: \psfig{figure=2200f.ps,angle=0,width=9cm}
401: \vskip -0.5 true cm
402: \caption{The SED of BL Lac during different observational campaigns,
403: as labeled. For the data points see Ravasio et al. (2002) and
404: references therein.}
405: \end{figure}
406:
407: \section{Results}
408:
409: The main aim of this work is to explore whether the internal shock
410: scenario can account for the difference of the SED along the blazar
411: sequence. We thus a) concentrate on the global properties, rather
412: than to explain details of the spectrum of a specific object; b) apply
413: the model to Mkn 421 and BL Lac itself which, besides being among the
414: best studied BL Lac objects, are representative of extremely blue and
415: intermediate blazars, respectively, complementing the study already
416: performed for the powerful blazar 3C 279 (S01).
417:
418: {\it Mkn 421 --- } This is one example of bright, extremely blue and
419: strong TeV emitting blazar. The peaks of its synchrotron and inverse
420: Compton spectra are at $\sim$ keV and a few hundreds GeV energies,
421: respectively. The bolometric (observed) power is of order $8\times
422: 10^{45}$ erg s$^{-1}$. The SED of coordinated simultaneous
423: observational campaigns of Mkn 421 are reported in Fig.~1 (Costamante
424: \& Ghisellini 2002).
425:
426: {\it BL Lac --- } The prototype of the BL Lacertae class is an
427: intermediate blazars (e.g. Fossati et al. 1998). It is a 1 Jy source,
428: with broad band peaks in the near IR and MeV--GeV band and its total
429: bolometric power is of order $\sim 10^{46}$ erg s$^{-1}$. Despite of
430: BL Lac being the prototype of the BL Lac class, it sporadically showed
431: optical emission lines (with maximum observed EW $\sim 6$ \AA;
432: Vermeulen et al. 1995; Corbett et al. 2000). In July 1997 it
433: underwent a major outburst, followed in the optical and by the CGRO
434: (EGRET), {\it Rossi}XTE and ASCA satellites. During this flare the
435: entire SED dramatically changed, as illustrated in Fig. 2, with EGRET
436: detecting a flux ($>$100 MeV) a factor 3.5 times higher than that
437: observed in 1995 (Bloom et al. 1997).
438:
439: \begin{figure}
440: \vskip -0.5 true cm
441: \psfig{figure=effi_421.ps,width=9cm}
442: \vskip -0.5 true cm
443: \caption{Radiative efficiency as a function of the collision radius
444: for Mkn 421. The solid line refers to the global efficiency, i.e.
445: the fraction of the total wind kinetic energy ($E_{\rm w}=L_{\rm
446: w}\times t_{\rm w}$) radiated on scales smaller than any given radius;
447: the shaded histogram shows the differential efficiency, namely the
448: fraction of $E_{\rm w}$ radiated for a given radius interval. The
449: cone--like insert in the upper part of the figure shows a grey--tone
450: representation of the differential efficiency of the jet: the darker
451: the color the higher the efficiency.}
452: % \end{figure}
453: %
454: %
455: % \begin{figure}
456: \psfig{figure=effi_bllac.ps,width=9cm}
457: \vskip -0.5 true cm
458: \caption{Radiative efficiency versus the collision radius for BL Lac.
459: The solid line refers to the global efficiency, i.e. the fraction of
460: the total wind kinetic energy ($E_{\rm w}=L_{\rm w}\times t_{\rm w}$)
461: radiated on scales smaller than a given radius; the shaded histogram
462: shows instead the differential efficiency, namely the fraction of
463: $E_{\rm w}$ radiated for a given radius interval. The cone--like
464: insert in the upper part of the figure shows a grey--tone
465: representation of the differential efficiency of the jet: the darker
466: the color the higher the efficiency. The vertical line indicates
467: $R_{\rm BLR}$.}
468: \end{figure}
469:
470: In order to reproduce the SED and variability of these two sources, we
471: took as an initial set of physical parameters that determined for 3C
472: 279 (see S01), and changed as few as possible of them to account for
473: the different broad band SED and 'spectral states' of these sources.
474:
475: It turned out that the global properties of the blazar ``sequence" can
476: be reproduced by only changing two key quantities, namely the jet
477: luminosity $L_{\rm jet}$ and the luminosity $L_{\rm BLR}$ and
478: extension $R_{\rm BLR}$ of the external photon field. The other
479: parameters [$\epsilon_{\rm B}$, $\zeta _{\rm e}$, the slope $n$ of the
480: (high energy) electron distribution, the observing angle $\theta$] are
481: only constrained to fit the specific SED. The remaining quantities
482: are at most slightly varied just to optimize the fit. In particular
483: $\epsilon_{\rm e}$ has been fixed for all sources to 0.5.
484: % at the equipartition value (with the protons).
485: In Table 1 we list all of the input parameters adopted to reproduce
486: the time dependent behavior of the simulated sources (for convenience
487: of the reader we also report the same parameters for 3C 279 from S01).
488:
489: Let us now examine the physical quantities inferred from the modelling
490: and their dependence on $R$.
491:
492: \vskip 0.3 true cm
493: \noindent
494: {\it Efficiency ---} The radiative efficiency represents the fraction
495: of the dissipated energy at each location channeled into relativistic
496: electrons. In absence of non--radiative losses this energy should be
497: all radiated. As shown in Fig. 3 and 4 for Mkn 421 and BL Lac,
498: respectively, the efficiency decreases simply due to the decrease in
499: the contrast of bulk Lorentz factors of colliding shells. The
500: apparent different efficiencies of BL Lac and Mkn 421 reflect only the
501: different viewing angle (chosen to reproduce the detailed properties
502: of the two sources).
503:
504: \vskip 0.3 true cm
505: \noindent
506: {\it Parameters profiles: $\Gamma$, $B$, $\gamma_{\rm b}$ ---} Fig. 5
507: and Fig. 6 (top three panels) show how the bulk Lorentz factor, the
508: magnetic field, the Lorentz factor $\gamma_{\rm b}$ of the injected
509: electrons evolve with distance from the nucleus, for Mkn 421 and BL
510: Lac, respectively. Different symbols correspond to shells having
511: experienced a different number of collisions. The bulk Lorentz
512: factors tend toward average values as result of successive collisions
513: and merging of the shells. The range and the asymptotic bulk Lorentz
514: factor $\Gamma$ are fully consistent with the typical values inferred
515: from observations. The magnetic field decreases roughly as $R^{-3/2}$
516: for both sources: it is worth remarking that we assume that a constant
517: fraction ($\epsilon_{\rm B}$) of the available energy is converted
518: into magnetic field and no field component survives between successive
519: collisions. The minimum energy of the injected electrons, $\gamma_b
520: m_e c^2$, is also a decreasing function of $R$, due to the decrease in
521: the bulk Lorentz factors contrast, implying decreasing efficiency and
522: mean energy per particle.
523:
524: In general the only difference between Mkn 421 and BL Lac is the
525: smaller spread of the parameters for the former, due to the fact that
526: in Mkn421 all the collisions occur in the same region.
527:
528: \vskip 0.3 true cm
529: \noindent
530: {\it Parameters profiles: $\gamma_{\rm peak}$, $\nu_{\rm peak}$ ---}
531: The two bottom panels of Figs. 5, 6 show the spatial profile of the
532: energy of the electrons emitting at the peak of the SED, $m_{\rm e}
533: c^2 \gamma_{\rm peak}$, and the corresponding observed synchrotron
534: peak frequency, $\nu_{\rm peak}$.
535:
536: A significantly different behavior characterizes the two sources. In
537: Mkn 421 $\gamma_{\rm peak}$ is a continuous decreasing function of $R$
538: and attains large values. This follows from the fact that in low
539: power blazars only high energy electrons can radiatively cool in a
540: finite injection time, thus determining high values of $\gamma_{\rm
541: peak}$. Specifically, its value is determined by $\gamma_{\rm max}$
542: at each collision radius, as set by the spectral fit. The behavior of
543: $\nu_{\rm peak}$ simply follows that of $\gamma_{\rm peak}$ and $B$.
544: On the contrary for BL Lac $\gamma_{\rm peak}$ (and consequently
545: $\nu_{\rm peak}$) shows a discontinuity. This occurs at $R_{\rm
546: BLR}$: inside the BLR electrons are in the fast cooling regime and
547: thus $\gamma_{\rm peak}=\gamma_b$ while outside the BLR the behavior
548: is similar to that of Mkn 421, namely the radiative cooling time
549: exceeds the shock crossing time, implying $\gamma_{\rm
550: peak}=\gamma_{\rm max}$. Note that for 3C 279 the BLR is even larger
551: and $\gamma_{\rm peak}=\gamma_b$ for almost all of the collisions.
552:
553: \begin{figure}
554: \psfig{figure=vpeak.ps,angle=0,width=10.3cm}
555: \caption{ Evolution of the post--shock shell parameters as a function
556: of the collision radius for Mkn 421. The upper panel shows the bulk
557: Lorentz factor of the merged shell after each collision; the second
558: panel represents the value of the generated magnetic field
559: ($\epsilon_{\rm B} =6\times 10^{-4}$). The third panel shows the
560: minimum random Lorentz factor of the injected electrons $\gamma_{\rm
561: b}$ (for $\epsilon_{\rm e} = 0.5$; $\zeta_{\rm e} = 0.08$). In the
562: fourth and last panels $\gamma_{\rm peak}$ and $\nu^{\rm obs}_{\rm
563: peak}$ are plotted. Only one tenth of the points is displayed for
564: clarity. Circles refer to collisions in which none of the shells has
565: ever collided before, stars to collisions in which one of the shells
566: has collided, while triangles indicate collisions between shells that
567: have both collided before. }
568: \end{figure}
569:
570: \begin{figure}
571: \psfig{figure=vpeak2200.ps,angle=0,width=10.3cm}
572: \caption{ Evolution of the post--shock shell parameters as a function
573: of the collision radius for BL Lac. The upper panel shows the bulk
574: Lorentz factor of the merged shell after each collision; the second
575: panel represents the value of the generated magnetic field
576: ($\epsilon_{\rm B} =10^{-2}$). The third panel shows the minimum
577: random Lorentz factor of the injected electrons $\gamma_{\rm b}$ (for
578: $\epsilon_{\rm e} = 0.5$; $\zeta_{\rm e} = 10^{-2}$). In the fourth
579: and last panels $\gamma_{\rm peak}$ and $\nu^{\rm obs}_{\rm peak}$ are
580: plotted. Only one tenth of the points is displayed for clarity.
581: Circles refer to collisions in which none of the shells has ever
582: collided before, stars to collisions in which one of the shells has
583: collided, while triangles indicate collisions between shells that have
584: both collided before. }
585: \end{figure}
586:
587: \vskip 0.3 true cm
588: \noindent
589: {\it Spectra at various $R$ ---} In order to understand how and where
590: the overall spectrum is generated, the spectra produced at different
591: radii are shown in Fig. 7 and Fig. 8: in the top panel the average
592: ``instantaneous" spectrum of a single shell at $R$ is shown while in
593: the bottom panel the spectra are integrated for all of the shells
594: emitting at a certain time (the flux is thus weighted for the duration
595: of the emission in each shell at a given frequency). While the single
596: shell emission is qualitatively similar for the two sources, the
597: integrated spectra of BL Lac do not simply reproduce the single shell
598: behavior. This is due to the different (shorter) duration of each
599: single shell emission, due to the fast cooling of the electrons
600: (caused by the external Compton). The end result is that in Mkn 421
601: most of the emission is produced in the inner jet, while in BL Lac the
602: inner part only dominates at the highest frequencies of the two
603: spectral peaks,
604: %The lower energies are instead produced mostly at larger radii,
605: with interesting consequences on the correlations between variability
606: in different bands.
607:
608: It is also worth commenting on the extreme differences in the SED of
609: BL Lac at different epochs (as shown in Fig. 2). This has been
610: already interpreted as due to variable broad emission lines in the
611: spectrum (mainly $H_\alpha$ and $H_\beta$), which would cause a
612: variable contribution of seed photons for the external Compton
613: process. The maximum observed total luminosity from the BLR is
614: 5--8$\times 10^{42}$ erg s$^{-1}$ (see e.g. Vermulen et al., for
615: $H_0=65$ km s$^{-1}$ Mpc$^{-1}$). If one adopts the correlation
616: (Kaspi et al. 2000) between $R_{\rm BLR}$ and the ionizing continuum
617: luminosity (here assumed to be $\sim$ 10 times $L_{\rm BLR}$), $R_{\rm
618: BLR}\sim$ 5--7 $\times 10^{16}$ cm. This inferred distance is
619: indeed similar to
620: % comparison of $R_{\rm BLR}$
621: the typical distance of the first shell--shell collisions,
622: %(i.e. $z\sim \Gamma^2 \Delta$)
623: suggesting that the collisions sometimes occurs within $R_{\rm BLR}$,
624: sometimes outside it (see also Ravasio et al. 2002, 2003). Dramatic
625: spectral changes, such as those between the 1995 and the 1997 SED (see
626: Fig. 2) can be therefore accounted for in this scenario.
627:
628: \begin{figure}
629: \vskip 0.5 true cm \psfig{figure=spettri_421.ps,width=8cm}
630: \caption{ Average spectra of collisions for Mkn 421. The upper panel
631: shows the average spectrum of the shells without taking into account
632: the duration of the emission.
633: % i.e. the average spectrum of a single
634: %shell in the given range of radii.
635: In the lower panel, instead, the duration of the emission of each
636: shell has been taken into account.}
637: % \end{figure}
638: %
639: % \begin{figure}
640: \vskip 0.5 true cm \psfig{figure=spettri_bl.ps,width=8cm}
641: \caption{ Average spectra of collisions for BL Lac. The upper panel
642: shows the average spectrum of the shells without taking into account
643: the duration of the emission.
644: %The figure shows the average spectrum of
645: %a single shell in the given range of radii.
646: In the lower panel, instead, the duration of the emission of each
647: shell has been taken into account.}
648: \end{figure}
649:
650: \begin{figure}
651: \vskip 0.5 true cm
652: {\hskip 1.3 truecm
653: \psfig{figure=lcur_421.ps,width=7cm,height=9.5true cm} }
654: \vskip 0.7 true cm
655: \caption{Simulated light curves for Mkn 421 in different bands, as
656: labeled. }
657: % \end{figure}
658: %
659: % \begin{figure}
660: \vskip 1 true cm
661: {\hskip 1.3 true cm
662: \psfig{figure=lcur_bl.ps,width=7cm,height=9.5true cm} }
663: \vskip 0.7 true cm
664: \caption{Simulated light curves for BL Lac in different bands, as
665: labeled. }
666: \end{figure}
667:
668: \begin{figure}[b]
669: \centerline{ \psfig{figure=croco_421.ps,width=8cm,height=7 true cm} }
670: \vskip 0.3 true cm
671: \caption{ Results of the cross correlations analysis for Mkn 421,
672: between the $\gamma$--ray and X--ray (solid grey line), optical (solid
673: dark line) and infrared (dashed line) light curves. There are no
674: delays between the different frequencies. }
675: \end{figure}
676:
677: \begin{figure}[b]
678: \centerline{ \psfig{figure=croco_bl.ps,width=8cm,height=7 true cm} }
679: \vskip 0.3 true cm
680: \caption{ Results of the cross correlations analysis for BL Lac,
681: between the $\gamma$--ray and X--ray (solid grey line), optical (solid
682: dark line) and infrared (dashed line) light curves. There is a delay
683: between the $\gamma$--ray and the infrared emission of roughly 10 days.
684: }
685: \end{figure}
686:
687: \vskip 0.3 true cm
688: \noindent {\it Light curves ---} The simulations performed aimed at
689: reproducing (and successfully did) all of the flux range of the SED
690: observed in the various multiwavelength campaigns. However the model
691: also predicts the detailed variability behavior at all frequencies
692: (see also Tanihata et al. 2003) which has not been considered a priori
693: to constrain the model. Let us then examine the light curves in
694: different bands, reported in Fig. 9 and Fig. 10 for the two sources,
695: and simply compare their variability properties with observations. A
696: movie showing the spectral variability during the whole simulation for
697: Mkn 421 (and 3C 279) can be found at the URL
698: http://ares.merate.mi.astro.it/$\sim$gabriele/421/index.html. The
699: long term 'periodicity' that appears in the light curves is due to the
700: sinusoidal modulation chosen for the mass distributions of the shells.
701:
702: In order to assess the potential of the model to reproduce the data,
703: we focused on comparing the simulated light curves with the observed
704: ones, by calculating basic indicators, namely the r.m.s., the
705: cumulative distribution of the variance [(Flux -- $\langle$ Flux
706: $\rangle)^2$], and the minimum variability doubling timescales,
707: defined as $\langle$ Flux $\rangle \Delta t /(\Delta$ Flux). In
708: particular we focused on the X--ray and optical bands for Mkn 421. For
709: the optical variability we sampled the simulated curves with a
710: frequency of about once a day for all the 300 days covered by the
711: simulation and compared the results with the photometric light curves
712: in the $V$ band presented by Tosti et al. (1998). For the X--rays, we
713: considered the results presented by Tanihata et al. (2001) on 6 day
714: continuous observations of the source with ASCA and binned the
715: observed 0.2--2 keV light curve over the same time bins (0.125 days)
716: of the simulations.
717:
718: In the optical band the simulated light curves show significantly
719: larger variability than the observed one (as from Tosti et al. 1998):
720: the r.m.s. from the observations is about 4 times smaller than that
721: derived from the model, and the Kolmogorov--Smirnoff test (KS) on the
722: variance distributions rejects the null hypothesis. On the contrary,
723: the comparison reveals promising results for the variability in the
724: X--ray band. Here, the observed and simulated r.m.s. are comparable
725: (0.2 vs 0.3) and the KS test shows that the variance distributions are
726: marginally compatible (5 per cent level). The X--ray variability
727: properties of the simulated and observed light curves appear to be
728: similar also in terms of the variability timescale, which is of the
729: order of 0.3--0.4 day.
730:
731: While a proper comparison of the variability properties would require
732: a more sophisticated analysis and better sampled data, these tentative
733: results show that the variability in the X--rays is reasonably
734: reproduced by the model. A significant discrepancy appears instead in
735: the optical, where the model predicts larger variability: this could
736: be in principle improved by modifying the model parameters
737: (e.g. reducing the Lorentz factor contrast and modifying its
738: distribution), although beyond the aim of this paper. We however
739: notice that in the case of Mkn 421 the possible role of the host
740: galaxy emission in reducing the optical variability amplitude should
741: be assessed.
742:
743: Finally, we computed the cross--correlations between the variability
744: in different energy bands. The results are presented in Fig. 11 and 12
745: for Mkn 421 and BL Lac, respectively. As discussed above for Mkn 421
746: $\gamma_{\rm peak}$ is determined, at all radii, by $\gamma_{\rm
747: max}$, accounting for the absence of lags between the different
748: frequencies (Fig. 11). On the contrary, in BL Lac lags ($\sim$ 10
749: days) are present between the infrared and the $\gamma$--ray band. In
750: fact, in this source the $\gamma$--ray emission is dominated by the
751: external Compton within $R_{\rm BLR}$, while the infrared one by
752: synchrotron radiation from collisions occurring outside the broad line
753: region. We remind that in the case of 3C 279 (S01) the infrared
754: emission is also determined by collisions outside $R_{\rm BLR}$,
755: resulting in lags between the IR and the $\gamma$--rays of $\sim$ 40
756: days.
757:
758: \begin{table*}
759: \begin{center}
760: \begin{tabular}{lllllllllll}
761: \hline
762: Source
763: & $\langle L_{\rm jet}\rangle$
764: & $\Gamma_{\rm m} $--$\Gamma_{\rm M}$
765: & $ t_{\rm v}$
766: & $ \epsilon_{\rm e}$
767: & $ \epsilon_{\rm B}$
768: & $ \zeta_{\rm e}$
769: & $ n$
770: & $ L_{\rm BLR}$
771: & $ R_{\rm BLR}$
772: & $\theta$ \\
773: &erg s$^{-1}$ & & s & & & & &erg s$^{-1}$ & cm & degree\\
774: \hline
775: Mkn 421 &8e45 &10--20 &1.3e4 &0.5 &6e-4 &8e-2 &3.1 &1e42 &1e16 &2 \\
776: BL Lac &4e46 &8--20 &1.6e4 &0.5 &1e-2 &1e-2 &4.3 &8e42 &7e16 &4 \\
777: 3C 279 &1e48 &10--25 &1.0e4 &0.5 &4e-3 &4e-2 &2.1 &1e45 &5e17 &2 \\
778: \hline
779: \end{tabular}
780: \caption{Model input parameters for Mkn 421, BL Lac and 3C279}
781: \end{center}
782: \label{tab:fits}
783: \end{table*}
784:
785: \section{Discussion}
786:
787: In this work we have considered internal shocks as the dissipation
788: mechanism responsible for the emission in blazars. As mentioned in
789: the Introduction, this scenario is currently the most accredited to
790: explain the prompt gamma-ray emission in Gamma-Ray Bursts, despite of
791: an efficiency problem (Lazzati, Ghisellini \& Celotti 1999). In
792: blazars instead the relatively low efficiency is in fact required as,
793: at least in powerful objects, most of the jet power is carried to the
794: large scales.
795:
796: In S01 it was found that the internal shock model was successful in
797: reproducing the observed SED and variability properties of a powerful
798: blazar, namely 3C~279. However, the SED of this object does not
799: represent the whole of the blazar class, which covers a wide range of
800: spectral characteristics (luminosity, frequency of the peaks of the
801: emission and their relative intensity). Interestingly these parameters
802: appear to be correlated and the whole class can be seen as a sequence:
803: the frequency and the ratio of the low vs high energy peak intensity
804: increase with decreasing luminosity (Fossati et al. 1998).
805:
806: We show here that the internal shock model can satisfactorily account
807: also for the properties of the lower power blazars. The key parameters
808: driving the phenomenological sequence are the jet power (proportional
809: to the radiated one) and the intensity of the broad lines. These
810: parameters in turn regulate the SED shape, as they control the cooling
811: efficiency of the emitting particles. The global radiative efficiency
812: appears instead to be similar for all of the sources examined.
813:
814: The internal shock scenario determines a characteristic time interval
815: for the injection of relativistic electrons of the order of the
816: dynamical timescale, when the intensity of the spectrum from each
817: collision is maximized. Two regimes are relevant: {\it fast} and {\it
818: slow} cooling, corresponding to whether electrons of energy
819: $\gamma_{\rm b}$ can or cannot radiatively cool in the dynamical time.
820: In highly powerful blazars the fast cooling regime dominates in the
821: inner regions (within the BLR) where also most of the power is
822: dissipated. Consequently the peak frequencies are produced by
823: electrons of energy $\gamma_{\rm b} m_{\rm e} c^2$. In the weakest
824: blazars instead the slow cooling regime prevails over the whole
825: jet. Consequently only the highest energy electrons can cool over such
826: timescale: the peak frequencies thus shift to high values.
827:
828: Between these two extremes there are intermediate sources, like BL
829: Lac, with broad lines of intermediate intensity, produced at a
830: distance within which a few shell-shell collisions can occasionally
831: take place. Observationally this corresponds to SED with moderate
832: Compton to synchrotron luminosity ratio, as the scattered seed photons
833: are only the synchrotron ones (in collisions outside the BLR). The
834: rare collisions within the BLR give rise to dramatic changes in the
835: SED characterized by a large increase of the Compton component
836: (e.g. BL Lac itself, see Fig.~2).
837:
838: The model considered here, which considers self-consistently the
839: dynamics and spectral emission, predicts also the time dependent
840: spectral properties. In general, the selected parameters allow to
841: reproduce the full range of spectral 'states' observed in both BL Lac
842: and Mkn 421. A more detailed comparison performed for Mkn 421, and
843: based on the r.m.s. and variability timescales, shows good agreement
844: for the X--ray variability properties, while the simulated optical
845: variations appear to be too large with respect to the considered
846: observed light curve.
847:
848: An analysis of the predicted cross-correlated variability between the
849: $\gamma$--ray and other bands reveals that only for BL Lac lags
850: ($\sim$ 10 days) are expected between the $\gamma$--rays and the
851: infrared emission.
852:
853: We conclude that the internal shock scenario can account for the the
854: main properties of blazars and the `blazar sequence'.
855:
856: While it has been previously pointed out that the key quantity in
857: reproducing the different characteristics of blazars is the ratio
858: between the jet and the disc (i.e. broad lines) luminosities, the
859: internal shock scenario discussed here provides more physical
860: insights, namely i) directly connects the radiated jet luminosity with
861: the jet effective power and ii) accounts for the `preferred' distance
862: where most of the luminosity is dissipated. While the latter is
863: similar for high and low power blazars, the BLR is instead located at
864: different distances in the different sub--classes of objects (as
865: determined by the ionizing luminosity).
866:
867: From a more theoretical point of view, the model does not (yet)
868: address the role of a seed magnetic field amplified by the shell-shell
869: collisions. This would probably lead to a faster synchrotron cooling
870: on the large scales and a flatter dependence of the $B$ field from $R$
871: (Fig.~3). We intend to further explore this issue. Furthermore, only
872: internal shocks have been considered in this scenario. It is
873: conceivable that at large jet scales some entrainment may occur
874: causing an interaction of the jet with the external medium, possibly
875: leading to external shocks. As a result the radiative efficiency and
876: large scales synchrotron and inverse Compton emission could be
877: enhanced. Clearly, the inverse Compton emission could be also enhanced
878: if a significant photons field is present externally to the jet even
879: on large scales (such as microwave background and/or beamed nuclear
880: radiation and/or dust emission, e.g. Celotti, Ghisellini \& Chiaberge
881: 2001; Sikora et al. 2002).
882:
883: Observationally the improved detection sensitivity of the planned TeV
884: Cherenkov telescopes -- such as VERITAS, HESS, MAGIC -- will allow in
885: the near future to measure emission from BL Lacs in low luminosity
886: states, and thus to estimate their flare activity duty cycles. Since
887: in the internal shock scenario the TeV emission is largely produced by
888: the few powerful internal collisions, it will be then possible to
889: further constrain model parameters such as the range of bulk Lorentz
890: factors and the initial separation of the shells. Analogous results
891: will be likely achieved for the more powerful blazars in the GeV band
892: thanks to the AGILE and GLAST satellites.
893:
894: \begin{acknowledgements}
895: DG thanks the Observatory of Brera in Merate for kind hospitality and
896: acknowledges the NSF grant AST-0307502 for financial support. AC
897: thanks the JILA fellows, University of Colorado, for the warm
898: hospitality and the Italian MIUR and INAF for financial support.
899: \end{acknowledgements}
900:
901:
902: \begin{thebibliography}{}
903: \bibitem[]{B00} Beloborodov A.M., 2000, ApJ, 539, L25
904: \bibitem[]{} Bloom S. D., Bertsch D. L., Hartman R. C. et al.,
905: 1997, ApJ, 490, L145
906: \bibitem[]{} Celotti A., Ghisellini G. \& Chiaberge M.,
907: 2001, MNRAS 321, L1
908: \bibitem[]{} Corbett E.A.,Robinson A., Axon D.J., Hough J.H.,
909: 2000, MNRAS, 311, 485
910: \bibitem[]{}Costamante L. \& Ghisellini G., 2002, A\&A, 384, 56
911: \bibitem[]{} Daigne F. \& Mochkovitch R., 1998, MNRAS, 296, 275
912: \bibitem[]{F98} Fossati G., Maraschi L., Celotti A., Comastri A. \&
913: Ghisellini G., 1998, MNRAS, 299, 433
914: %\bibitem[]{} Fossati, G., Celotti A., Chiaberge M. et al., 2000a,
915: %ApJ, 541, 153 % 421
916: %\bibitem[]{} Fossati, G., Celotti A., Chiaberge M. et al., 2000b,
917: %ApJ, 541, 166 % 421
918: \bibitem[]{} Ghisellini G. \& Madau P., 1996, MNRAS, 280, 67
919: \bibitem[]{} Ghisellini G., Celotti A., Fossati G., Maraschi L.
920: \& Comastri A., 1998, MNRAS, 301, 451
921: %\bibitem[]{} Ghisellini G., 1999. 4th ASCA symp.,
922: % Astronomische Nachrichten.
923: % Editors: H. Inoue, T.Ohashi \& T.Takahashi, 320, 232
924: \bibitem{G02} Ghisellini G., 2002, in 25th John Hopkins Workshop,
925: astro-ph/0111584
926: \bibitem[]{G01} Guetta D., Spada M., \& Waxman E., 2001, ApJ 557, 399
927: \bibitem[]{} Kaspi S., Smith P.S., Netzer H., Maotz D., Iannuzi
928: B.T. \& Giveon U., 2000, ApJ, 533, 631
929: \bibitem[]{} Lazzati, D., Ghisellini G. \& Celotti A., 1999,
930: MNRAS 309 , L13
931: %\bibitem[]{} Maraschi L., Fossati G., Tavecchio F. et al., 1999,
932: %ApJ, 526, L81 %(421)
933: \bibitem[]{} Padovani P., Perlman E.S., Landt H.,
934: Giommi P., \& Perri M., 2003, ApJ, 588, 128
935: \bibitem[]{} Panaitescu A. \& M\'esz\'aros P., 1999, ApJ, 526, 707.
936: %\bibitem[]{} Pian E., Vacanti G., Tagliaferri G., et al., 1998,
937: %ApJ, 491, L17 %(501)
938: \bibitem[]{} Ravasio M., Tagliaferri G., Ghisellini G.
939: et al. 2002, A\&A, 383, 763
940: \bibitem[]{} Ravasio M., Tagliaferri G., Ghisellini G., Tavecchio F.,
941: Bottcher M. \& Sikora M., 2003, A\&A, 408, 479
942: \bibitem[]{R78} Rees M.J., 1978, MNRAS, 184, P61
943: \bibitem[]{} Rees M.J. \& M\'esz\'aros P., 1994, ApJ, 430, L93
944: %\bibitem[]{} Sambruna R.M., Ghisellini G., Hooper E., Kollgaard R.I.,
945: % Pesce J.E. \& Urry C.M., 1999, ApJ, 515, 140
946: \bibitem[]{} Sikora M., Blazejowski M.,
947: Moderski R. \& Madejski G.M., 2002, ApJ, 577, 78
948: \bibitem[]{S94} Sikora, M., 1994, ApJS, 90, 923
949: \bibitem[]{} Spada M., Ghisellini G., Lazzati D. \& Celotti A.,
950: 2001, MNRAS, 325, 1559 (S01)
951: \bibitem[]{} Spada M, Panaitescu A. \& M\'esz\'aros P., 2000, ApJ 537, 824
952: \bibitem[]{}Tanihata C., Takahashi T., Kataoka J. \&
953: Madejski G.M., 2003, ApJ, 584, 153
954: \bibitem[]{}Tanihata C., Urry M., Takahashi T., Kataoka J., Wagner S.J.,
955: Madejski G.M., Tashiro M. \& Kouda M., 2001, ApJ, 563, 569
956: \bibitem[]{} Tosti G., Fiorucci M., Luciani M. et al.,
957: 1998, A\&A, 339, 41
958: \bibitem[]{} Ravasio M., Tagliaferri G., Ghi
959:
960: \bibitem[]{} Ulrich M.--H., Maraschi L. \& Urry C.M., 1997, ARA\&A, 35, 445
961: \bibitem[]{WW95} Wagner S.J. \& Witzel A., 1995, ARA\&A, 33,163
962: \bibitem[]{} Vermeulen R.C., Ogle P.M., Tran H.D. Browne I.W.A., Cohen M.H.,
963: Readhead A.C.S. \& Taylor G.B., 1995, ApJ, 452, L5
964:
965: % \bibitem[]{} Maraschi L. et al., 1994, ApJ, 435, L91;
966: % \bibitem[]{} Hartman R.C. et al. 1996, ApJ, 461, 698.
967:
968:
969: \end{thebibliography}
970:
971: %\clearpage
972: %\clearpage
973:
974: \end{document}
975:
976:
977:
978: