a

r
X
i
v
:
p
h
y
s
i
c
s
/
0
3
1
2
1
4
6

v
1



2
7

D
e
c

2
0
0
3
Dependence of the MHD shock thickness on the nite electrical conductivity
Alejandra Kandus and Reuven Opher
Departamento de Astronoma, IAG-USP, Rua do Matao 1226,
Cidade Universitaria, CEP: 05508-900, Sao Paulo, SP, Brazil.
The results of MHD plane shock waves with innite electrical conductivity are generalized for
a plasma with a nite conductivity. We derive the adiabatic curves that describe the evolution of
the shocked gas as well as the change in the entropy density. For a parallel shock (i.e., in which
the magnetic eld is parallel to the normal to the shock front) we nd an expression for the shock
thickness which is a function of the ambient magnetic eld and of the nite electrical conductivity of
the plasma. We give numerical estimates of the physical parameters for which the shock thickness is
of the order of, or greater than, the mean free path of the plasma particles in a strongly magnetized
plasma.
I. INTRODUCTION
Previously, shock waves in plasmas, both relativistic and non-relativistic, were studied assuming ideal MHD, [13].
Although this theory is suitable for studying most astrophysical shock waves, such as those in hot raried astrophysical
plasmas, where the electrical conductivity is extremely high and the magnetic eld is weak, it is interesting to study
the eect of the simplest dissipative process in non-ideal MHD, that due to a nite value of the electrical conductivity
. We generalize here the results for planar shock waves to non-ideal, non-relativistic MHD. The junction conditions
that must be satised across a shock wave in a non-ideal plasma are given in Sec. II. We then derive the adiabatic
curve with corrections due to and nd the corresponding change in the entropy density across the shock. We
perform this analysis for an oblique shock (i.e., in which

B is neither parallel nor perpendicular to the normal to the
shock surface) as well as for a parallel one (i.e., in which

B is parallel to the normal to the shock surface). In Sec. III,
we nd a closed expression for the shock thickness for a parallel shock in a strongly magnetized plasma, and estimate
its value for some physical situations. Our conclusions are discussed in Sec. IV.
II. JUNCTION CONDITIONS AND ADIABATIC CURVE
For non ideal MHD, the uxes of mass, energy, and momentum are given by Eqs. (1-4), respectively, [1,4],

M = v, (1)
q = v
_
1
2
v
2
+ w
_
+
1
4

B
_
v

B
_

c
2
16
2

B
_

B
_
, (2)
and

ik
= v
i
v
k
+ p
ik

1
4
_
B
i
B
k

1
2
B
2

ik
_
, (3)
where is the uid density, v the velocity, w the enthalpy per unit mass,

B the ambient magnetic eld, and p is the
uid pressure. The electric eld is

E =
c
4
_

B
_

v
c


B, (4)
where Ohms law in its simplest form [2,5] was used.
A. Junction conditions
We assume a two dimensional, planar, shock wave in the y z plane. The normal to the transition surface is in the
x direction. The velocity eld can be decomposed into perpendicular and tangential components to the surface of
1
transition, v = (v
x
, v
t
). It is assumed that all quantities vary as a function of x . Let n
x
be a unit vector normal to
the transition surface. We then have the hydrodynamical junction conditions, [4],
[v
x
n
x
] = 0 (5)
[q
x
n
x
] = 0 (6)
[
ix
n
x
] = 0 (7)
where i is x (t) [normal (tangential) to the shock surface] and [] means the dierence between the value of the
corresponding quantities far upstream (which we denote by subscript 1) and the value at some point in the shock
(no subscript). It is also assumed that both far upstream and far downstream, all gradients vanish (i.e., the elds and
ows are uniform).
In the MHD case that we are considering, Eqs. (5), (6), and (7) must be suplemented with the electromagnetic
junction conditions, i.e., that the normal component of the magnetic eld and the tangential component of the electric
eld must be constant across the shock surface:
[B
n
] = 0, (8)
_

E
t
_
=
_
c
4
_

B
_
t

v
x
c

B
t
+
B
n
c
v
t
_
= 0. (9)
Eq. (5) states that the mass ux along x is conserved, i.e., v
x
= j = const. We write = 1/V , where V is the
specic volume, and replace v
x
= jV in the other junction conditions, obtaining
j
_
1
2
j
2
V
2
+
1
2
v
2
t
+ w
_
+
1
4
j
_
V B
2
t

1
4
B
x
_
v
t
.

B
t
_

c
2
16
2

x
B
2
t

= 0, (10)
j
2
[V ] + [p] +
1
8
_
B
2
t

= 0, (11)
j [v
t
]
1
4
B
n
_

B
t
_
= 0, (12)
c
4
__

B
t
__

j
c
_
V

B
t
_
+
B
n
c
[v
t
] = 0. (13)
B. Adiabatic curve and entropy density change
We derive the expression for the adiabatic curve and the corresponding entropy density change. From Eqs. (12)
and (13), we obtain
1
4
B
2
n
_

B
t
_
= j
2
_
V

B
t
_

jc
2
4
__

B
t
_
t
_
. (14)
From Eq. (12), we have [v
t
] = B
n
_

B
t
_
/4j. We can therefore complete the squares in Eq. (10 ), obtaining
[w] +
1
2
j
2
_
V
2

+
1
2
_
_
v
t

1
4
B
n
j

B
t
_
2
_

1
32
2
B
2
n
j
2
_
B
2
t

+
1
4j
B
n
_
v
t
.

B
t
_
+
1
4
_
V B
2
t

1
4j
B
n
_
v
t
.

B
t
_

c
2
16
2
j
_

x
_
B
2
t
_
= 0. (15)
2
From Eq. (12), the third term of Eq. (15) is zero and we are left with
[w] +
1
2
j
2
_
V
2

1
32
2
B
2
n
j
2
_
B
2
t

+
1
4
_
V B
2
t

c
2
16
2
j
_

x
_
B
2
t
_
= 0. (16)
Using Eq. (14), we can write the third term of Eq. (16) as
1
32
2
B
2
n
j
2
_
B
2
t

=
1
8
_
V B
2
t

+
1
8
(V V
1
)

B.

B
1

c
2
32
2
j
_

B
_
.
_

B
t
+

B
1t
_
. (17)
From momentum conservation, we obtain
j
2
=
(p p
1
)
(V V
1
)

1
8
_
B
2
t
B
2
1t
_
(V V
1
)
. (18)
Using Eqs. (17) and (18) in Eq. (16) and w = + pV , we have

1
+
1
2
(p + p
1
) (V V
1
) +
1
16
_

B
t

B
1t
_
2
(V V
1
) (19)
+
c
2
32
2
j
__

B
_
.
_

B
t
+

B
1t
_
2
x
_
B
2
t
_
_
= 0.
The rst four terms are found in the equation for ideal MHD (e.g. Ref. [1]), while the last two are due to the
nite electrical conductivity. The rst term in the square brackets is due to the fact that B
n
= 0; the second is the
contribution from the tangential component of the magnetic eld.
To obtain the entropy density change, we follow the procedure found in the standard literature, [4] and develop
V V
1
in powers of (p p
1
). We also expand (w w
1
) in powers of (p p
1
) and to rst order in powers of (s s
1
).
The resulting expression is
T (s s
1
) =
1
12
_

2
V
p
2
_
s
(p p
1
)
3

1
16
_
V
p
_
s
(B
t
B
1t
)
2
(p p
1
)

c
2
32
2
j
__

B
_
.
_

B
t
+

B
1t
_
2
x
_
B
2
t
_
_
. (20)
The rst two terms are found in the equation for ideal MHD and the last two are the corrections due to a nite .
In the following section, we repeat the calculations for a perpendicular shock and nd that in the expressions for the
adiabatic curve and the entropy density change, the only term present which depends on the conductivity is the last
one.
C. Perpendicular shock
Shock waves in a plasma permeated with a magnetic eld show several features, of which the most well known is
related to the orientation of the magnetic eld with respect to the shock plane. Although perpendicular shocks can be
considered to be a special case of oblique shocks, it is interesting to write the simplied expressions for the junction
conditions explicitly, and re-derive the adiabatic curve and the entropy density change for this case.
1. Hydrodynamical and electromagnetic junction conditions
For perpendicular shocks B
n
= 0, so that the junction conditions now read
[v
x
] = 0, (21)
_
v
x
_
1
2
v
2
+ w
_
+
1
4
v
x
B
2
t

c
2
16
2

x
_
B
2
_
_
= 0, (22)
3
_
v
2
x
+ p +
1
8
B
2
t
_
= 0, (23)
[v
t
v
x
] = 0 v
t
= v
t1
, (24)
c
4
__

B
t
__

1
c
_
v
x

B
t
_
= 0. (25)
2. Adiabatic curve and entropy density change
Proceeding as for an oblique shock and dening

= +
B
2
t
V
8
, p

= p +
1
8
B
2
t
, (26)
we obtain
w w
1

1
2
[p

] (V + V
1
) +
1
4
_
V B
2
t

c
2
16j
2

x
_
B
2
t
_
= 0 (27)
and

1
+
1
2
(p

+ p

1
) (V V
1
)
c
2
16j
2

x
_
B
2
t
_
= 0. (28)
There is now only one term that depends on the dissipative properties of the plasma, while for the oblique case, we
had two such terms. The missing term is related to the normal component of the magnetic eld.
III. THICKNESS OF THE SHOCK WAVE
The calculation of a general expression for the shock thickness is very dicult, if not impossible. However for a
perpendicular shock it is possible to calculate the shock thickness exactly. We then have B
n
= 0, v
t
= 0 and consider
a coordinate system in which the only non-zero component of the magnetic eld is B
y
= B, [1]. In this case, the
equation

.

B = 0 is satised identically. The unidimensional ideal MHD equations are

B
t
=

x
(v
x
B) , (29)

t
+

x
(v
x
) = 0, (30)
v
x
t
+ v
x
v
x
x
+
1
8
B
2
x
=
1

p
x
. (31)
From the rst two equations, it is easy to see that the ratio B/ satises the equation /t + v
x
/x = 0 or
d/dt = 0 [1]. Hence, if the uid is homogeneous at some initial instant, so that = const., then it will remain so at
all subsequent times. Substituting B = in the third equation, we obtain
v
x
t
+ v
x
v
x
x
=
1

x
_
p +

2

2
8
_
(32)
Thus, the magnetic eld has been eliminated from the equations. The equation for the velocity eld, Eq. (32), is
formally identical to that for the ideal uid case, provided we dene the true pressure as p

= p +
2

2
/8. We can
now proceed to evaluate the thickness, following Ref. [4]. We write
4

t
p

x
p

p
p

x
p

= cL

2
x
2
p

, (33)
where v
2
s
(p

/)
s
and L

is the damping length, given by Eq. (61) of the Appendix,

L

=
c
8
B
2
0y
_
B
2
0y
+ 4
0
v
2
s
_. (34)
From Ref. [4] we have

p
=
1
2
v
3
s

2
_

2
p
2
_
1

__
s
. (35)
Equation (33) can be solved using the procedure in Ref. [4] , obtaining the thickness of the shock wave as

=
4cL

p
(p

2
p

1
)
, (36)
where p

2
and p

1
are the true pressures far downstream and far upstream respectively. Equation (36) is quantitatively
valid for suciently small dierences (p

2
p

1
). However we can use it qualitatively to estimate the order of magnitude
of the thickness in cases where the dierence (p

2
p

1
) is of the same order of magnitude as p

2
and p

1
themselves.
The velocity of sound in the gas, v
s
(not v

s
) is of the same order of magnitude as the thermal velocity v. Let be
the mean free path of the atoms in the plasma. Then from dimensional analysis, the electric conductivity can be
estimated as, v/ v
s
/, where takes into account anomalous eects and can have a value 10
6
1
1
.
In Eq. (34), we take B
2
0y
/
_
B
2
0y
+ 4
0
v
2
s
_
B
2
0y
/p

and in Eq. (35)

p
(p

2
p

1
) v
2
s
p

/. Using these
relations in Eq. (36) we obtain
c
2

2
B
2
0y
p
2
p
. (37)
The shock thickness is larger than the mean free path when c
2

2
B
2
0y
p
2
p. Let us rst assume that B
2
0y
p.
We then have p
2
B
4
0y
and p B
2
0y
c
2

2
/p. As a specic numerical example, consider 10
2
gr/cm
3
and
T 10
8
K (characteristic parameters at the center of a massive star before collapse). Assuming hydrogen gas, we
have n 10
26
cm
3
and p nT 10
18
erg/cm
3
(for iron nuclei, the pressure would be two orders of magnitude
smaller). For the above densities and pressures, the shock thickness is larger than the mean free path of the particles
if the magnetic eld is in the interval 10
9
G B
0y

1/2
10
14
G. If we now assume that B
2
0y
p, the shock thickness
is larger than the mean free path if p
3
/c
2

2
B
2
0y
. For the above parameters we have
1/2
10
5
G B
0y
10
10
G
2
. If neither of the two conditions above are fullled, the shock thickness is smaller than the mean free path. This
means that the MHD approach breaks down and kinetic theory is needed to study the structure of the shock.
IV. CONCLUSIONS
In this article, we extended the results of shock waves treated in ideal MHD to the non-ideal case, in which the
electrical conductivity is nite. We considered Ohms law in its simplest form, [5], but took into account phenomeno-
logically (through the parameter ) plasma eects that can modify the classical Spitzer electrical conductivity (e.g.,
1
We know that in accretion disks, protostars, galactic nuclei and neutron X-ray sources, for example, the plasma cannot
have ideal Spitzer values for the conductivity and viscosity in order to obtain the observed accretion rates. Therefore it is
generally assumed that these quantities are highly anomalous (due to turbulence, for example). Another example where the
assumption of anomalous resistivity is used is in the treatment of solar ares, which are generally assumed to be due to magnetic
reconnection. If ideal Spitzer values are used for the plasma in solar ares, reconnection times are 10
6
times longer than the
observed time scales for the ares. In general, plasmas near shocks are expected to be highly anomalous (i.e., 1) due to
turbulence.
2
For 10
6
, the magnetic eld range is 10
2
G B0y 10
10
G. A magnetic eld B 10
2
G can be easily be present at the
center of a massive star. The magnetic eld increases in the collapse of the core of a massive star to B 10
6
(for the collapse
to a white dwarf) and to B 10
9
(for the collapse to a neutron star). Thus, at the center of a massive star, we may expect
that the magnetic eld varie from 10
2
G to 10
10
G during the collapse of its core and the start of a supernova explosion.
5
turbulence). The expressions for the adiabatic curve and the entropy density change across the shock were generalized.
Finally, we derived the expression for the shock thickness for a nite conductivity in the case of a parallel shock in
strongly magnetized plasmas. The conditions that the ambient magnetic eld must satisfy for the thickness to be
of the order of the particle mean free path were estimated. We found that these conditions can be fullled for the
plasma expected in the origin of a supernova explosion, [6,7]. Extensions of the results presented in this paper, using
a more general Ohms law, as well as to relativistic shocks, are presently under investigation.
ACKNOWLEDGMENTS
This work was partially supported by the Brazilian nancing agency FAPESP (00/06770-2). A.K. acknowledges
the FAPESP fellowship (01/07748-3). R. O. acknowledges partial support from the Brazilian nancing agency CNPq
(300414/82-0).
V. APPENDIX
In this appendix, we sketch the derivation of the expression for L

, the damping length used to calculate the shock

thickness. Neglecting the displacement current (which is a good approximation in non-relativistic electrodynamics),
the evolution equation for the magnetic eld in a medium with electrical conductivity moving with a velocity v is

B
t

_
v

B
_
=
c
2
4

B. (38)
Adding the equations for the uid, which we assume has neither viscosity nor thermal conduction, we have

t
+

. (v) = 0, (39)
v
t
+
_
v.

_
v =
1

p
1
4

B
_

B
_
. (40)
A. Hydromagnetic waves
Let us assume that B =

B
0
+

b, =
0
+, p = p
0
+p, and v = v. Replacing these terms in the above equations,
keeping only terms to rst order in the perturbations, expanding the density in powers of the perturbation in the
pressure (i.e., = p/v
2
s
+
_

2
/p
2
_
s
p
2
, where v
s
is the sound velocity of the medium) and taking the Fourier
transform of the equations, we obtain
p
0
+ v
2
s

k.v
0
= 0, (41)

_
+ i
c
2
4
k
2
_

b
0
=

k
_
v
0

B
0
_
, (42)
v
0
=

0
p
0

1
4
0

B
0

b
0
_
. (43)
From Eq. (41), we nd p
0
= v
2
s

0
_

k.v
0
_
/ and using this in Eq. (43 ), we obtain
v
0
=

v
2
s
_

.v
0
_

1
4
0

B
0

b
0
_
. (44)
We dene the scalar phase velocity as u = /k, assuming that

k is along the x-axis, (i.e.,

k = k x) and that

B
0
is
in the x y plane. Writing the previous equations in its components, we have
6
p
0
=
v
2
s

0
u
A
v
0x
, (45)
_
u
v
2
s
u
_
v
0x
=
1
4
0
b
0y
B
0y
, (46)
uv
0y
=
1
4
0
b
0y
B
0x
, (47)
_
u + i
c
2
4
k
_
b
0y
= v
0x
B
0y
v
0y
B
0x
, (48)
uv
0z
=
1
4
0
b
0z
B
0x
, (49)
_
u + i
c
2
4
k
_
b
0z
= v
0z
B
0x
. (50)
B. Generalized Alfven waves
Using Eqs. (49) and (50) we obtain the compatibility relationship
u
2
+ i
c
2
k
4
u
B
2
0x
4
0
= 0, (51)
from which we obtain
u =
1
2

B
2
0x

c
4
k
2
16
2

2
i
c
2
k
8
. (52)
From Eq. (52), the phase velocity is a complex number if is nite. Rewriting u
A
in terms of , we obtain the
dispersion relationship:
=
1
2
B
0x
k

1

0
B
2
0x
c
4
k
2
16
2
i
c
2
k
2
8
. (53)
We take the plus sign in eq. (52) since the frequency is a positive quantity. The fact that the imaginary part is
non-linear in k means that the Alfven waves are damped and dissipated as a function of k. For we recover
the known dispersion relationship for ideal MHD.
Assuming that the second term in the square root in Eq. (53) is much smaller than unity, the group velocity is
v
A
=

k

B
0x

4
0
i
c
2
k
4
. (54)
When the electrical conductivity is innite, we recover the known ideal MHD result, v
AI
= B
0x
/2

0
.
7
C. Generalized magnetosonic waves
From Eqs. (46), (47) and (48), we obtain the generalized dispersion relationship for magnetosonic waves,

_
B
2
0
4
0
+ v
2
s
_
k
2

2
+
B
2
0x
4
0
v
2
s
k
4
+ i
c
2
k
2
4

3
i
c
2
v
2
s
k
4
4
= 0. (55)
This relationship can be inverted to obtain = (k) . However, for our purposes, it suces to consider the dissipative
terms as a correction to the ideal dispersion relationship,

2
I
=
1
2
_
_
_
B
2
0
4
0
+ v
2
s
_

_
B
2
0
4
0
+ v
2
s
_
2

B
2
0x

0
v
2
s
_
_
k
2
v
2
g0
k
2
. (56)
The plus sign corresponds to fast magnetosonic waves, while the minus sign to slow magnetosonic waves. Replacing

I
in the last two terms of Eq. (55), we obtain

_
B
2
0
4
0
+ v
2
s
_
k
2

2
+
B
2
0x
4
0
v
2
s
k
4
+ i
c
2
v
g0
4
_
v
2
g0
v
2
s
_
k
5
0. (57)
If the electrical conductivity is large (but not innite), we have

2

2
I

i
4
c
2
v
g0

_
v
2
g0
v
2
s
_
(v
4
B0
v
2
B0x
v
2
s
)
1/2
k
3
, (58)
where v
2
B0
=
_
B
2
0
/4
0
+ v
2
s
_
and v
2
B0x
= B
2
0x
/
0
. We thus have
v
g0
k icL

k
2
, (59)
with
L

=
c
8
_
v
2
g0
v
2
s
_
(v
4
B0
v
2
B0x
v
2
s
)
1/2
, (60)
where L

is the damping length. For a perpendicular shock, (i.e., B

0x
= 0), we have
L

=
c
8
B
2
0t
(B
2
0t
+ 4
0
v
2
s
)
. (61)
[1] Landau, LD., Lifshitz, E. M. and Pitaevskii, L. P. 1999 Electrodynamics of Continuous Media. Second edition, Butterworth-
Heinemann.
[2] Priest, E. R. 1982 Solar Magnetohydrodynamics. Reidel.
[3] Anile, A. M. 1989 Relativistic Hydrodynamics and Magnetohydrodynamics. Cambridge University Press.
[4] Landau, L. D. and Lifshitz, E. M. 1997 Fluid Mechanics. Second edition, Butterworth-Heinemann.
[5] Spitzer Jr., L. 1962 Physics of Fully Ionized Gases. Second edition, John Wiley & Sons.
[6] Thompson, C. and Murray, N. 2001 Astrophys. J. 560, 339.
[7] Akiyama, S. and Wheeler, J. C. 2002 astro-ph/0211458.
8