JITTER RADIATION AS A POSSIBLE MECHANISM FOR GAMMA-RAY BURST AFTERGLOWS

:
SPECTRA AND LIGHT CURVES
Mikhail V. Medvedev
1
Department of Physics and Astronomy, University of Kansas, Lawrence, KS 66045
and
Davide Lazzati, Brian C. Morsony, and Jared C. Workman
JILA, University of Colorado, Campus Box 440, Boulder, CO 80309-0440
Received 2007 March 9; accepted 2007 June 6
ABSTRACT
The standard model of gamma-ray burst afterglows assumes that the radiation observed as a delayed emission is
of synchrotron origin, which requires the shock magnetic field to be relatively homogeneous on small scales. An
alternative mechanism—jitter radiation, which traditionally has been applied to the prompt emission—substitutes
for synchrotron when the magnetic field is tangled on a microscopic scale. Such are the fields produced at relativistic
shocks by the Weibel instability. Here we explore the possibility that small-scale fields populate afterglowshocks. We
derive the spectrumof jitter radiation under the afterglowconditions. We also derive the afterglowlight curves for the
interstellar mediumand wind profiles of the ambient density. Jitter self-absorption is calculated here for the first time.
We find that jitter radiation can produce afterglows similar to synchrotron-generated ones, but with some important
differences. We compare the predictions of the two emission mechanisms. With future observational data, one may be
able to discriminate between the synchrotron and jitter afterglow light curves, and, hence, between the small-scale
versus large-scale magnetic field models in afterglow shocks.
Subject headinqs: gamma rays: bursts — magnetic fields — shock waves
1. INTRODUCTION
The general framework for the interpretation of the long-
wavelength radiation of gamma-ray burst (GRB) afterglows is
the external shock synchrotron model ( Meszaros & Rees 1997;
Waxman 1997; Piran 1999). In that scenario, a blast wave is
generated by the interaction of the GRB ejecta with the inter-
stellar medium ( ISM). At the shock front, electrons are accel-
erated in a power-lawdistribution with energy (or with Lorentz
factor ¸), and a strong magnetic field is generated by some
mechanism. The model assumes that the magnetic field is co-
herent on the Larmor scale of the emitting electron, hence al-
lowing for synchrotron emission. Both the relativistic electron
population and the magnetic field are originally thought to share
a sizable fraction of ~10% of the internal energy of the blast
wave; those fractions are being called c
e
and c
B
, respectively.
Afterglow spectral fits yield typical values, c
e
~ 0.1Y0.01 and
c
B
~ 0.01Y0.0001, with relatively large scatter (Panaitescu &
Kumar 2001; Panaitescu 2005).
No mechanism or instability capable of generating a sub-
equipartition magnetic field in GRBs has been identified for a
while, until Medvedev & Loeb (1999) suggested that the field
can be generated through the Weibel instability. This prediction
has been extended to nonrelativistic shocks, e.g., in supernovae
and galaxy clusters ( Medvedev et al. 2006), and confirmed via
numerical particle-in-cell simulations (Silva et al. 2003; Nishikawa
et al. 2003; Frederiksen et al. 2004; Medvedev et al. 2005;
Spitkovsky 2005; Chang et al. 2007). The volume-averaged
value of c
B
deduced from the simulation is indeed ~0.01Y
0.0001 (depending on the location with respect to the main
shock compression). An intriguing relation,
c
e
’ c
1,2
B
. (1)
recently found by Medvedev (2006b) can allow one to reduce
the number of fit parameters in afterglow studies.
We emphasize here that we do not address the question of
whether the small-scale Weibel-generated fields can survive far
downstream, through distances many orders of magnitude larger
than the field coherence length. Instead, we develop a framework
that will allow us to explore this possibility for GRB afterglow
shocks using future observational data.
Weibel-generated fields have a very short coherence length
scale (smaller than 1/¸
2
times the electron Larmor radius), and
standard synchrotron equations cannot be adopted. The theory of
jitter radiation has been proposed by Medvedev (2000) and fur-
ther developed in subsequent works (Medvedev 2006a; Fleishman
2006).
Unlike synchrotron, jitter radiation is sensitive to the statis-
tical properties of the magnetic field in the shock, that is, to the
spectrum of magnetic fluctuations, and not just to its ‘‘global
property’’—the strength (Medvedev 2000). In addition, the spec-
trum of jitter radiation depends on the shock viewing angle, i.e.,
the angle 0
/
between the shock velocity (propagation direction)
and the line of sight (0
/
is measured in the shock comoving frame).
The two extreme cases are characterized by the emissivity func-
tion being a power law at frequencies below the spectral peak,
P(.) · .
c
. (2)
with c = 1 for 0
/
= 0—a shock viewed face-on (this is also the
case of the ‘‘effective’’ one-dimensional magnetic turbulence
considered in Medvedev 2000)—and c = 0 for 0
/
= ¬/2—an
edge-on shock (this is also the case in isotropic two- and three-
dimensional turbulence). These asymptotes, along with a gen-
eral case of 0 _ 0
/
_ ¬/2, are considered elsewhere ( Medvedev
2006a; Workman et al. 2007). These works also support an
early prediction ( Medvedev 2000) of hard spectra, F
i
· i
1
in
GRBs, in contrast to some recent claims ( Fleishman 2006; but
1
Also at: Institute for Nuclear Fusion, Russian Research Center Kurchatov
Institute, Moscow 123182, Russia.
339
The Astrophysical Journal, 666:339Y345, 2007 September 1
# 2007. The American Astronomical Society. All rights reserved. Printed in U.S.A.
see Medvedev 2005). For reference, synchrotron radiation has
c = 1/3 at low energies.
Let us first consider a spherical outflow. Most of the radiation
observed from a relativistic external shock is beamed within a
cone with the opening angle 0
0
~ 1/À
sh
around the line of sight.
However, the photons emitted from the shock along the line of
sight, 0 ~0, have shorter light travel time to the observer than
those from the limb, 0 ~ 0
0
. Therefore, in order to be observed
simultaneously with the limb photons, the line-of-sight photons
must be emitted later, that is, when the shock radius is larger and
the shock Lorentz factor is proportionally smaller. Hence, the
line-of-sight emission is intrinsically dimmer than the limb
(Granot et al. 1999). Because of relativistic aberration, the pho-
tons emitted from the limb, at 0 ~ 0
0
~ 1/ À
sh
with respect to the
line of sight in the lab (observer’s) frame, have been emitted at
angle 0
/
in the shock comoving frame,
cos 0
/
=
cos (1,À
sh
) ÷u
sh
1 ÷u
sh
cos (1,À
sh
)
= 0. (3)
where u
sh
is the dimensionless shock speed. That is, these pho-
tons are emitted at the angle 0
/
= 90
·
away from the shock
normal. Thus, a shock is essentially seen ‘‘edge-on’’ in the co-
moving frame and c ~0 in this case.
Now, consider a jetted outflowwith the jet opening angle 0
jet
.
For the early afterglow, before the jet break, 0
jet
0
0
~1/ À
sh
,
the afterglow is essentially indistinguishable from the spherical
ejecta, hence c ~ 0. After the jet break, 0
jet
< 0
0
, the emission
is still limb-brightened, but most of the emission is coming from
angles 0 ~ 0
jet
< 0
0
, which corresponds to the comoving 0
/
<
90
·
, thus 0Pc P1 is expected. Even later, when the ejecta
becomes weakly relativistic, À
sh
~ few, emission comes from
0 ~0
jet
T0
0
; therefore, the comoving angle is approaching
zero, 0
/
~0, for which a hard spectrum with c ~1 is expected.
At last, at very late times, when the external shock becomes non-
relativistic, À
sh
~1, sideways expansion of the ejecta makes the
external shock nearly spherical. Emission is nearly uniformacross
such a nonrelativistic shock, so that the overall spectrum is a con-
volution of individual ones from different patches, hence we ex-
pect 0 Pc P1.
In fact, jitter radiation is anisotropic (depends on 0
/
) and for
intermediate angles 0 < 0
/
< ¬/2 it cannot be characterized by
a single power law at low energies, below the peak. There is
a break i
b
(0
/
) in the jitter spectrum ( Medvedev 2006a) below
the jitter peak, i
b
< i
jp
, such that c = 1 above the break (at
i
b
< i < i
jp
) and c = 0 below it (at i < i
b
). The dependence
of the break frequency on 0
/
is not easy to parameterize. Since
the break is quite smooth, it is also not easy to establish its
precise position. One characteristic point is i
b
~ 0.01i
jp
at
0
/
= ¬/10 = 18
·
.
To summarize, for spherical outflows and for jets seen before
the jet break, the reasonable approximation will be the ‘‘edge-
on’’ shock, with c = 0. At later times, especially when the jet
becomes weakly relativistic, the ‘‘face-on’’ case with c = 1
should be a better approximation. Finally, for a nonrelativistic
external shock, 0Pc P1.
We have one cautious remark. The above consideration as-
sumes strong anisotropy of the magnetic turbulence in the shock
( Medvedev & Loeb 1999), which is likely true in the internal
shocks. We do not understand well the properties of magnetic
turbulence far downstream the external shock. If the magnetic
turbulence will become nearly isotropic, then a three-dimensional
jitter regime gives c = 0, independent of the viewing angle.
Below, we consider both the c = 0 and 1 cases in calculating
the jitter self-absorption frequency. This paper is organized as
follows. In ¸ 2 we summarize the shock kinematics, in ¸ 3 we
derive the jitter self-absorption frequency in various regimes,
while in ¸ 4 we compute all the observable properties of the
spectrum. We summarize and discuss our results in ¸ 5.
2. SHOCK KINEMATICS
Here we summarize, for future use, the kinematic relations
of several blast wave parameters. For a highly relativistic blast
wave with Lorentz factor À
sh
, the density jump condition relates
the preshock ISM density to the density downstream, n
/
, mea-
sured in the shock frame,
n
/
’ 4À
sh
n
ISM
. (4)
The magnetic field strength downstream
B
/
= 32¬À
2
sh
n
ISM
m
p
c
2
c
B
À Á
1,2
. (5)
where c
B
is the magnetic field equipartition parameter. The min-
imum Lorentz factor of the accelerated electrons, which share a
fraction c
e
< 1 of the total energy of the shock (outflow), is
¸
m
=
s ÷2
s ÷1

m
p
m
e
c
e
À
sh
~ 6.12 ; 10
2
c
e
À
sh
. (6)
Here the last expression is calculated for a typical electron power-
lawindex s = 2.5 (see definition belowin eq. [12]). We have two
models.
Constant density ISM model.—The radius and Lorentz factor
of a blast wave propagating in a constant density ISM depend
on the observed time as (Granot et al. 1999)
À
sh
~ 3.65
E
52
n
ISM.0

1,8
t
days
1 ÷z

÷3,8
. (7)
R ~ 5.53 ; 10
17
E
52
t
days
n
ISM.0
(1 ÷z)
!
1,4
cm. (8)
where E
52
= E
explosion
/ 10
52
ergs ( ) and n
ISM.0
= n
ISM
/(1 cm
÷3
).
Wind model.—If the blast wave is propagating in the wind
environment with the density decreasing with distance, n · r
÷2
,
the Lorentz factor and the radius of the blast wave are (Chevalier
& Li 2000)
À
sh
~ 4.96E
1,2
52
A
+
t
days
1 ÷z

÷1,4
. (9)
R ~ 1.56 ; 10
17
E
52
t
days
A
+
(1 ÷z)
!
1,2
cm. (10)
Here the wind parameter A
+
= [
˙
M
w
/(10
÷5
M
¸
yr
÷1
)[/
[V
w
/(10
3
km s
÷1
)[, where
˙
M
w
is the mass loss rate and V
w
is the
wind velocity. Since the ambient density is no longer constant,
one needs to substitute n
ISM
with the wind density
n
wind
= AR
÷2
~ (3.00 ; 10
35
A
+
cm
÷1
) R
÷2
. (11)
3. THEORY
Here we work in the shock comoving frame. Thus, all the fre-
quencies are expressed in this frame. In addition, all the shock
MEDVEDEV ET AL. 340 Vol. 666
parameters, such as the particle number density n
/
and magnetic
field B
/
, are those in the shocked region and are measured in the
comoving frame as well, unless stated otherwise. We use a prime
to denote quantities measured in the shock frame. Sometimes we
omit the prime when this does not cause any confusion.
We introduce the accelerated electron distribution function
N(¸) = (s ÷1) N
e
¸
s÷1
m
¸
÷s
. ¸ _ ¸
m
. (12)
where s is the power-law index, ¸
m
is the minimum Lorentz
factor (low-energy cutoff ), N
e
= 4¬R
2
Á
/
n
ISM
is the total num-
ber of nonthermal (emitting) electrons, R is the radius of the
blast wave, Á
/
is its thickness, and n
ISM
is the number density of
the electrons in the ambient medium.
The absorption coefficient at comoving frequency i
/
is (Dermer
et al. 2000)
i
i
/ =
1
8¬m
e
V
bw
i
/ 2
Z
·
1
d¸ P i
/
. ¸ ( )¸
2
0

N(¸)
¸
2
!
. (13)
where V
bw
= 4¬r
2
Á
/
is the volume of the blast wave of the
comoving thickness Á
/
and P(i
/
. ¸) is the emissivity function.
Straightforwardly,
¸
2
0

N(¸)
¸
2
!
= ÷(s ÷2)(s ÷1)N
e
¸
s÷1
m
¸
÷s÷1
. (14)
3.1. Emissivity Functions
We first discuss synchrotron, for reference. The synchrotron
comoving peak frequency is at
i
/
s
= (3,2)i
B
¸
2
. (15)
where i
B
= eB
/
/ 2¬m
e
c ( ) is the cyclotron ( Larmor) frequency
in a homogeneous magnetic field of strength B
/
and the pitch
angle is assumed to be ¬/2, for simplicity. At i
/
Ti
/
s
, the syn-
chrotron emissivity is
P
synch
(i
/
. ¸) ’
e
2
c
ffiffiffi
3
_
i
B

ffiffiffi
3
_
À(1,3)
i
/
3i
B
¸
2

1,3
" #
. (16)
where À(1/3) ’ 2.68 is a gamma function. This regime of emis-
sion occurs when the magnetic field coherence length is much
larger than the gyroradius of the emitting particles.
We now consider emission from magnetic fields of the same
energy density, but whose coherence length is much smaller than
the electron’s gyroradius, that is, ¸
˜
B
/
) = 0 and ¸
˜
B
/2
) ,= 0 and in
equation (5) one substitutes B
/
with ¸
˜
B
/2
)
1/2
. The jitter peak
frequency is determined by the magnetic field spectrum in the
postshock medium. Recent numerical simulations ( Frederiksen
et al. 2004; Medvedev et al. 2006) demonstrate that the field
generation by both the electrons and the protons occurs well
before the main shock compression. The wavevector of the fast-
est growing mode of the Weibel instability in the linear regime is
( Medvedev & Loeb 1999)
k
W
~ (.
p.e
,c) ¯ ¸
÷1,2
e
. (17)
where .
p.e
= 4¬e
2
n/m
e
( )
1/2
~ 5.64 ; 10
4
n
1/2
s
÷1
is the plasma
frequency. Here we omitted a factor 2
÷1/4
, which is nearly
unity. Because .
p.e
is a Lorentz invariant, k
W
is determined by
the parameters of the ambient, unshocked medium alone. Thus,
the density n = n
ISM
is the ambient medium density, and the
mean Lorentz factor of the electrons in the ambient medium,
¯ ¸
e
, is close to unity (the ISM is cold and nonrelativistic). The
shock compression and the proton thermalization occur far down-
stream, significantly behind the region where the field has been
generated. Therefore, the correlation length of the field decreases
(primarily in the parallel direction) because of the compression.
Hence, the characteristic wavevector of the downstream random
fields is
k
rand
’ (4À
sh
) k
W
. (18)
Numerical simulations also indicate that the correlation scale of the
field k
B
= 2¬/k
rand
varies with distance from the shock front be-
cause of the highly nonlinear dynamics of the Weibel-generated
currents and fields (Frederiksen et al. 2004; Medvedev et al. 2005;
Chang et al. 2007). We therefore parameterize this as
k
rand
= j(.
p.e
,c). (19)
where the parameter j incorporates all relativistic effects, the
shock compression, and the nonlinear evolution of the Weibel
turbulence. Hereafter, we use
j ’(4À
sh
) ¯ ¸
÷1,2
e
’ 4À
sh
. (20)
where we assumed that ¯ ¸
e
~1.
The characteristic frequency of the electron’s jitter while it
moves at roughly the speed of light through these magnetic fields
is i ~ c/k
B
. More precisely,
i
r
= k
rand
c,2¬ = j.
p.e
,2¬ = ji
p. ISM
. (21)
where we introduced the plasma frequency of the ISM, i
p. ISM
=
(4¬e
2
n
ISM
/m
e
)
1/2
/2¬ = 8.98 ; 10
3
n
1/2
ISM
Hz.
The peak of the emitted jitter radiation is i
/
j
~ i
r
¸
2
, but its
exact position slightly depends on the magnetic field spectrum,
defined as (B
/2
)
k
· k
2j
. The magnetic spectrum affects the ob-
served radiation spectrum at frequencies above the peak fre-
quency. For monoenergetic electrons, F
i
· i
2j
for the ‘‘face-on’’
(0
/
~ 0) and F
i
· i
2j÷1
for the ‘‘edge-on’’ (0
/
~¬/2) case
( Medvedev 2006a). Thus, for sufficiently small [j[, j < 0, the
spectral index of radiation could be determined by the mag-
netic field, rather than by the electron distribution index s.
Simulations of electron-proton plasmas by Frederiksen et al.
(2004) show that j ~÷2 in the foreshock region, whereas it
seems to be exponentially decaying at very short wavelengths
k k.
p
/c, as is indicated by electron-positron simulations of the
shock downstream (Chang et al. 2007). Unlike Frederiksen
et al. (2004) however, larger scales, kT.
p
/c, are poorly resolved
in Chang et al. (2007), so it is premature to make any conclu-
sions about the actual value of j in the downstreamregion. The
foreshock emission is likely relevant to the prompt emission, for
which we get F
i
· i
÷4
and · i
÷3
for 0
/
~ 0 and ~¬/2, re-
spectively, provided ÷(s ÷1)/2 2j ÷1. This requires a rather
steep electron spectrum with s 4[j[ ÷1 ~ 7. Fermi acceler-
ation yields s ~2.2, at least for afterglows. Thus, the electron
distribution is likely to determine the spectral slope, and one can
safely assume [j[ 31. This greatly simplifies further calculations.
JITTER RADIATION AND GRB AFTERGLOWS 341 No. 1, 2007
For steep spectra, [j[ 31, the peak is at, roughly, twice the
jitter frequency,
i
/
j
’ 2i
r
¸
2
. (22)
The jitter and synchrotron peak frequencies in the magnetic
fields of identical strengths are related to each other via the
identities
i
r
= i
B
,c and i
/
j
= i
/
s
,(3c,4). (23)
These identities define the parameter c P1, which is the ratio
of the deflection angle of the particle path in chaotic fields to
the beaming angle 1/¸ ( Medvedev 2000). It is expressed via the
magnetic field equipartition parameter in the shock, c
B
, as
c =
m
p
m
e

sh
j
2
c
B

1,2

m
p
m
e
c
B

1,2
~ 43
ffiffiffiffiffi
c
B
_
. (24)
The jitter emissivity function below the peak is
P
jitter
(i
/
. ¸) ’
e
2
c
¬f (j)c
2
i
r
i
/
2i
r
¸
2

c
. (25)
Later on, we neglected a factor f (j) = (2j ÷1)/(2j ÷1) (cal-
culated for c = 1), because it is of order unity when [j[ 31.
Thus, for both emission mechanisms,
P(i
/
. ¸) ’ a
i
/

2

c
. (26)
Here, for synchrotron
a =
e
2
c

À(1,3)
i
B
. b = 3i
B
. c = 1,3. (27)
and for jitter
a =
e
2
2c
¬c
2
i
r
. b = i
r
. c = 0 or 1. (28)
3.2. Self-Absorption Frequencies for i
a
< i
m
Here we assume that the self-absorption frequency is below
the jitter peak in the spectrumfromthe ensemble of the electrons,
i
/
a
Pi
/
m
= 2i
j. m
= 2i
r
¸
2
m
. The self-absorption coefficient is eas-
ily calculated from equations (14) and (26) to yield
i
i
/ =
(s ÷2)(s ÷1)N
e
ab
÷c
(i
/
)
c÷2
8¬m
e
V
bw
(s ÷2c)¸
2c÷1
m
. (29)
The self-absorption frequency is that at which the optical
thickness of the blast wave shell is unity,
i
i
/
a
Á
/
~1. (30)
This condition gives
i
/ 2÷c
a

(s ÷2)(s ÷1)
(s ÷2c)
Rn
ISM
ab
÷c
24¬m
e
¸
2c÷1
m
. (31)
This equation works for both synchrotron and jitter radiation.
One just need to replace the general parameters a, b, and c with
those above, for a process of interest.
For jitter radiation, we have
ab
÷c
=
e
2
c
¬
2
c
c
2
i
1÷c
r
. (32)
Therefore,
i
/ 2÷c
a

(s ÷2)(s ÷1)
(s ÷2c)
¬
24
(R,c)i
2
p. ISM
i
1÷c
r
c
2
¸
2c÷1
m
. (33)
In the case of c = 0,
i
/
a

¬
24
(s ÷2)(s ÷1)
s
!
1,2
(R,c)
1,2
i
p. ISM
i
1,2
r
c
¸
1,2
m
~ 2.9 Hz R
1,2
n
3,4
ISM
¸
÷1,2
m
j
1,2
c.
~ 5.4 Hz R
1,2
n
3,4
ISM
¸
÷1,2
m
À
1,2
sh
c. (34)
Hereafter, we use a typical value, s = 2.5, in numerical estimates.
The case of c = 1 may be relevant to the late afterglow from
a jet. We have
i
/
a
’ (s ÷1)
¬
24
(R,c)i
2
p. ISM
c
2
¸
3
m
~5.3 ; 10
÷4
Hz Rn
ISM
¸
÷3
m
c
2
. (35)
3.3. Self-Absorption Frequency for i
a
i
m
In order to calculate the self-absorption frequency in this re-
gime, we need the full expression for the emissivity function
P(i
/
. ¸) =
e
2
2c
c2¬i
r
J
i
/
i
r
¸
2

. (36)
where the function J(¸) for [j[ 31 and cT1 is
J(¸) = (2j ÷1)¸
2j
I(2) ÷I(¸) [ [. (37)
I(¸) = ÷
¸
÷2j÷1
2j ÷1
÷
¸
÷2j÷2
2j ÷2
÷
1
2
¸
÷2j÷3
2j ÷3

. (38)
This function corresponds to the c = 1 case. The expression for
the c = 0 case is more complicated. Our analysis indicates,
however, that the jitter emissivity function is approximated by a
sharply broken power law very well (much better than the syn-
chrotron one does). Therefore, we use the approximate expres-
sion for J(¸), which describes well the c = 1 and 0 spectra. We
use
J(¸) =
¸
c
. ¸ < 2.
0. ¸ _ 2.
&
(39)
This function mimics a power-lawspectrumup to the peak jitter
frequency i
j
’ 2i
r
¸
2
, with the sharp cutoff above it.
MEDVEDEV ET AL. 342 Vol. 666
The self-absorption coefficient is
i
i
/ =
(s ÷2)(s ÷1)N
e
¸
s÷1
m
8¬m
e
V
bw
i
/ 2
e
2
2c
; c2¬i
r
Z
·
¸
m
J
i
/
i
r
¸
2

¸
÷s÷1
d¸. (40)
Upon substitution of J(¸), the integral becomes
1
2
i
i
r

÷s,2
Z
min i
/
,(i
r
¸
2
m
).2 [ [
0
z
c÷s,2÷1
dz. (41)
We are interested in the high-energy part of the spectrum, above
i
/
3i
/
m
’ 2i
r
¸
2
m
. Therefore, the upper limit in the integral is
equal to 2. Thus, we have for the opacity at frequencies above
the spectral peak
i
i
/ =
(s ÷2)(s ÷1)
(s ÷2c)
2
s,2÷c÷3
12¬
(R,c).
2
p.e. ISM
Á
/
¸
3
m
c
2
i
/
m
i
/
i
/
m

÷s,2÷2
.
(42)
Since F
i
· i
5
,
2
for i
m
< i < i
a
, rather than · i
2
, the absorp-
tion frequency should be defined as the frequency at which
dF
i
/di = 0, where F
i
· i
5
,
2
(1 ÷e
÷i
i
/ Á
/
). For a simple esti-
mate, we can still use the condition i
i
/
a
Á
/
’ 1. We have
i
/
a
’ 2
s,2÷c÷3
¬
3
(s ÷2)(s ÷1)
(s ÷2c)
!
2,(s÷4)
;
(R,c)i
2
p.ISM
i
/s,2÷1
m
c
2
¸
3
m
" #
2,(s÷4)
. (43)
The peak frequency i
/
m
is calculated in ¸ 3.4 (eq. [45]). To
proceed further, we again use the typical value of the electron
power-law exponent, s = 2.5. For such s, the value of the nu-
merical factor (the first term) in equation (43) is equal to 0.95
and 0.99 for c = 0 and 1, respectively. We, therefore use the
representative value of 0.97 for both cases, which gives an error
in the i
/
a
value to within a couple of percents. Thus,
i
/
a
~ 320 Hz R
0.31
n
0.65
ISM
À
0.69
sh
¸
0.46
m
c
0.62
. (44)
3.4. Peak Frequency
The peak frequency of the shock spectrum (from an ensemble
of electrons) is determined by the jitter frequency at ¸
m
,
i
/
m
’ 2i
r
¸
2
m
~ 1.8 ; 10
4
Hz n
1,2
ISM
¸
2
m
j
~ 7.2 ; 10
4
Hz n
1,2
ISM
À
sh
¸
2
m
. (45)
3.5. Coolinq Break Frequency
The total (integrated over frequencies) emitted power by
an electron is identical in the jitter and synchrotron regimes
( Medvedev 2000). Thus, the cooling break is unchanged. We
quote results from Sari et al. (1998),
i
/
c
= (3,2)i
B
¸
2
c
. (46)
where the cooling Lorentz factor (neglecting Compton losses)
is
¸
c
=
3m
e
16m
p
co
T
c
B
1
n
ISM
À
sh
t
loc
(47)
where t
loc
= t/(1 ÷z) is the cosmologically local time for a GRB.
Thus,
i
/
c
~ 4.3 ; 10
24
Hz n
÷3,2
ISM
À
÷5
sh
c
÷3,2
B
t
÷2
loc
. (48)
4. OBSERVABLES
The power-lawsegments in the i
a
< i
m
and i
a
i
m
regimes
are, respectively,
F
(i
a
<i
m
)
i
·
i
2
. i < i
a
.
i
c
. i
a
< i < i
m
.
i
÷(s÷1),2
. i
m
< i < i
c
.
i
÷s,2
. i
c
< i.
8
>
>
>
<
>
>
>
:
(49)
F
(i
a
i
m
)
i
·
i
2
. i < i
m
.
i
5,2
. i
m
< i < i
a
.
i
÷(s÷1)
,
2
. i
a
< i < i
c
.
i
÷s,2
. i
c
< i.
8
>
>
>
<
>
>
>
:
(50)
We now calculate the frequencies in the observer’s frame
[thus, all frequencies are boosted by À
sh
/(1 ÷z)] and their de-
pendencies on the burst parameters for s = 2.5. For the ISM
model of external medium, we have
i
(c=0)
a
~ 5.9 ; 10
8
Hz (1 ÷z)
÷3,4
E
1,4
52
c
÷1,2
e
c n
1,2
ISM.0
t
÷1,4
days
. (51)
i
(c=1)
a
~ 9.6 ; 10
4
Hz (1 ÷z)
÷2
E
0
52
c
÷3
e
c
2
n
ISM.0
t
days
. (52)
i
(i
m
)
a
~ 3.1 ; 10
10
Hz (1 ÷z)
÷0.27
; E
0.35
52
c
0.46
e
c
0.62
n
0.30
ISM.0
t
÷0.73
days
. (53)
i
m
~ 4.8 ; 10
12
Hz (1 ÷z)
1,2
E
1,2
52
c
2
e
t
÷3,2
days
. (54)
i
c
~ 8.5 ; 10
16
Hz (1 ÷z)
÷1,2
E
÷1,2
52
c
÷3,2
B.÷3
n
÷1
ISM.0
t
÷1,2
days
. (55)
Here we also quoted the result from Sari et al. (1998) for the
cooling frequency in the adiabatic regime of blast wave evolu-
tion in the constant ISM density environment, corrected for the
redshift.
Finally, we compute the peak flux of jitter afterglows by us-
ing the relation F
i. max
= c
2
F
i. max. synch
i
m
/i
m. synch
and the syn-
chrotron peak flux from Sari et al. (1998),
F
i. max. synch
=10
3
E
52
c
B.÷3
n
ISM.0
D
÷2
28
jJy. (56)
We remind that the jitter deflection parameter is related to the
magnetic field equipartition parameter as
c ~ 43c
1,2
B
~1.3c
1,2
B.÷3
. (57)
Figure 1 shows the jitter and synchrotron spectra for an after-
glow with E =10
53
ergs, c
e
= 0.1, c
B
= 0.0001, n
ISM
=1, and
s = 2.5, computed at t = 0.1, 1, and 10 days.
JITTER RADIATION AND GRB AFTERGLOWS 343 No. 1, 2007
For the wind model of external medium, the frequencies in
the observer’s frame are
i
(c=0)
a
~ 2.8 ; 10
9
Hz (1 ÷z)
÷1,4
E
0
52
c
÷1,2
e
c A
+
t
÷3,4
days
. (58)
i
(c=1)
a
~ 1.8 ; 10
5
Hz (1 ÷z)
÷1
E
÷3,2
52
c
÷3
e
c
2
A
2
+
t
0
days
. (59)
i
(i
m
)
a
~ 1.4 ; 10
11
Hz (1 ÷z)
0.04
E
0.58
52
c
0.46
e
c
0.62
A
0.62
+
t
÷1.04
days
.
(60)
i
m
~ 2.9 ; 10
13
Hz (1 ÷z)
1,2
E
3,2
52
c
2
e
t
÷3,2
days
. (61)
i
c
~ 5.7 ; 10
15
Hz (1 ÷z)
÷3,2
E
1,2
52
c
3,2
B.÷3
A
÷2
+
t
1,2
days
. (62)
Here we also quoted the result from Chevalier & Li (2000) for
the cooling frequency in the adiabatic regime of blast wave
evolution in the wind environment.
5. DISCUSSION
We considered in this work the properties of GRB afterglows
with radiation produced by jitter radiation instead of synchro-
tron. For the first time we evaluate the self-absorption frequency
in various regimes and for blast waves propagating in different
ambient media.
Within the present framework, we analyzed two possible
regimes of the jitter mechanism, c = 0 and 1. If the postshock
magnetic turbulence is isotropic (which is very likely in the far
downstream region), then c = 0. If the turbulence remains an-
isotropic throughout the shell, then two possibilities arise. First,
for a spherical outflow and a jetted outflow observed before the
jet break, one has c = 0. Second, for late times, long after the jet
break, one can expect c 0 and even c ~ 1. Since the prop-
erties of turbulence likely vary downstream as a function of the
distance from the shock front, it is likely that one can have an
intermediate regime for most of the time, 0 Pc P1.
Note that in the jitter regime, the peak frequency is indepen-
dent of the magnetic field strength. In general, it depends on the
ambient density. However, for the assumed parameter j · À
sh
,
which incorporates all details (not so well known) of the mag-
netic field evolution far downstream, the jitter peak is indepen-
dent of the density, either n
ISM
or A
+
. Note also that the jitter
self-absorption frequencies and the peak frequency strongly de-
pend on the electron equipartition parameter. It may be helpful
to remember that the jitter peak frequency is higher than the syn-
chrotron peak frequency in the field of the same strength (same
c
B
) and the same electron energy distribution (same c
e
and elec-
tron index s) by a factor ~c
÷1
, which, in turn, is · c
1/2
B
.
Finally, we estimate the times when the emission becomes
optically thick, i
a
~i
m
for c = 0 and 1, for the ISM case,
t
(c=0)
a
~ 1300 days (1 ÷z)E
1,5
52
c
2
e
c
÷4,5
n
÷2,5
ISM.0
. (63)
t
(c=1)
a
~ 1200 days (1 ÷z)E
1,5
52
c
2
e
c
÷4,5
n
÷2,5
ISM.0
. (64)
t
(i
m
<i
a
)
a
~ 690 days (1 ÷z)E
0.20
52
c
2.00
e
c
÷0.80
n
0.40
ISM.0
. (65)
Ideally, these times should coincide. The discrepancies are due
to the approximations made in our analysis. In particular, the
self-absorption frequency for i
m
< i
a
is overestimated by a fac-
tor of 2 (as is explained above); hence, the thin-to-thick transi-
tion time is earlier. A more detailed treatment of the self-ab-
sorption frequencies will be presented elsewhere ( Workman et
al. 2007; B. C. Morsony et al. 2007, in preparation). In addition,
the time when the cooling break is equal to the jitter peak,
i
c
~ i
m
, is
t
c
~ 5.6 ; 10
÷5
days (1 ÷z)E
52
c
2
e
c
3,2
B.÷3
n
ISM.0
. (66)
The numerical factor in the above equation is equal to 4.9 s.
Similarly, we calculate different values for t
a
and t
c
for the
wind case,
t
(c=0)
a
~ 2.3 ; 10
5
days (1 ÷z)E
2
52
c
10,3
e
c
÷4,3
A
÷4,3
+
. (67)
t
(c=1)
a
~ 3.0 ; 10
5
days (1 ÷z)E
2
52
c
10,3
e
c
÷4,3
A
÷4,3
+
. (68)
t
(i
m
<i
a
)
a
~ 1.1 ; 10
5
days (1 ÷z)E
2.00
52
c
÷3.35
e
c
÷1.34
A
÷1.34
+
. (69)
t
c
~6.5 ; 10
÷2
days (1 ÷z)E
1,2
52
c
e
c
÷3,4
B.÷3
A
+
. (70)
The discrepancies in the values of t
a
are, again, due to the ap-
proximations made in calculating i
a
.
Figure 1 allows us to comment on the differences between
synchrotron and jitter afterglows. The high-energy part of the
spectrum (mainly the optical and X-ray regimes) are hardly
distinguishable between the two mechanisms, especially if,
as expected from simulation, jitter fields are created such that
c P1. The two main differences are in the low-energy branches,
in the radio band. First, the spectral slope at frequencies below
the peak frequency is not canonical i
1/3
; it is flat, i
0
, at times
before the jet break and can, possibly, be as steep as i
c
, with
c P1, well after the jet break ( but before the shock becomes
nonrelativistic). Second, the location of the self-absorption break
Fig. 1.—Jitter and synchrotron spectra for a typical afterglow running into a
uniform medium. The parameter set is E =10
53
ergs, c
e
= 0.1, c
B
= 0.0001,
n
ISM
= 1, and the electron energy distribution power-law index s = 2.5, com-
puted at t = 0.1, 1, and 10 days. Note that the jitter and synchrotron regimes do
not occur simultaneously. Instead, they correspond to the emission from mag-
netic fields with the coherence length much smaller and much larger than the
electron gyroradius, respectively.
MEDVEDEV ET AL. 344 Vol. 666
is different and evolves in time differently than in the synchrotron
afterglow scenario.
The best observational strategy can be to perform radio ob-
servations between the self-absorption and peak frequencies,
roughly from GHz to THz sufficiently early on, at times smaller
than a day after the burst. The flat spectrum in this frequency
range will indicate the jitter afterglow. Radio observations at
times of about a few days to 10 days are also interesting. The
steeper than synchrotron spectrum i
c
, with c P1, above the
self-absorption range i
2
will not only indicate jitter radiation,
but it will also hint at the presence of highly anisotropic small-
scale magnetic turbulence in the external shock. Unfortunately,
the spectral interval where such an unusual spectrum can exist is
rather narrowat late times, which makes such observations some-
what difficult.
M. M. gratefully acknowledges support from the Institute for
Advanced Study. This work was supported by NASA grants
NNG 04-GM41G ( M. M.) and NNG 06-GI06G ( B. M. and
D. L.), Swift Guest Investigator grant NNX 07-AJ50G ( M. M.)
and NNX 06-AB69G ( B. M. and D. L.), DOE grant DE-FG02-
04ER54790 ( M. M.), and NSF grant AST 03-07502 ( B. M.
and D. L.).
REFERENCES
Chang, P., Spitkovsky, A., & Arons, J. 2007, ApJ, submitted (arXiv:0704.3832)
Chevalier, R. A., & Li, Z.-Y. 2000, ApJ, 536, 195
Dermer, C. D., Bo¨ttcher, M., & Chiang, J. 2000, ApJ, 537, 255
Fleishman, G. 2006, ApJ, 638, 348
Frederiksen, J. T., Hededal, C. B., Haugbølle, T., & Nordlund, 8. 2004, ApJ,
608, L13
Granot, J., Piran, T., & Sari, R. 1999, ApJ, 527, 236
Medvedev, M. V. 2000, ApJ, 540, 704
———. 2005, preprint (astro-ph/0503463)
———. 2006a, ApJ, 637, 869
———. 2006b, ApJ, 651, L9
Medvedev, M. V., Fiore, M., Fonseca, R. A., Silva, L O., & Mori, W. B. 2005,
ApJ, 618, L75
Medvedev, M. V., & Loeb, A. 1999, ApJ, 526, 697
Medvedev, M. V., Silva, L. O., & Kamionkowski, M. 2006, ApJ, 642, L1
Meszaros, P., & Rees, M. J. 1997, ApJ, 476, 232
Nishikawa, K.-I., Hardee, P., Richardson, G., Preece, R., Sol, H., & Fishman,
G. J. 2003, ApJ, 595, 555
Panaitescu, A. 2005, MNRAS, 363, 1409
Panaitescu, A., & Kumar, P. 2001, ApJ, 560, L49
Piran, T. 1999, Phys. Rep., 314, 575
Sari, R., Piran, T., & Narayan, R. 1998, ApJ, 497, L17
Silva, L. O., Fonseca, R. A., Tonge, J. W., Dawson, J. M., Mori, W. B., &
Medvedev, M. V. 2003, ApJ, 596, L121
Spitkovsky, A. 2005, in AIP Conf. Proc. 801, Astrophysical Sources of High
Energy Particles and Radiation, ed. T. Bulik & B. Rudak ( Melville: AIP),
345
Waxman, E. 1997, ApJ, 491, L19
Workman, J., Morsony, B., Lazzati, D., & Medvedev, M. V. 2007, MNRAS,
submitted (astro-ph/0703420)
JITTER RADIATION AND GRB AFTERGLOWS 345 No. 1, 2007