37th EPS Conference on Plasma Physics P4.

181

The Bennett Pinch revisited

J.E. Allen

University College, Oxford, OX1 4BH, UK

OCIAM, Mathematical Institute, Oxford, OX1 3LB,UK
Imperial College, London, SW7 2BW, UK

Abstract

The original derivation of the well-known Bennett relation is presented. Willard H.

Bennett developed a theory, considering both electric and magnetic fields within a pinched
column, which is completely different from that found in the textbooks. The latter theory is
based on simple magnetohydrodynamics which ignores the electric field.

1. An elementary theory of the pinch effect The pinch

The configuration of the pinch effect is illustrated in Fig. 1.
effect:
A current flowing in the axial direction interacts with a self-
!!
produced magnetic field to produce an inwardly directed j × B
force. If we adopt a magnetohydrodynamic model for the equi-
librium case we can write

∇p = j × B (1)

and I Z
B · dl = µ0 j.dS (2) B!
The first equation describes the force balance, and the second
equation is Ampères Law, the integral form of the Maxwell
equation
∇ × B = µ0 j.
Figure 1: The pinch effect
In the present case equations 1 and 2 take the form
the Bennett r
dp
= − jB (1a)
dr
and
2πrB = µ0 I (r) (2a)

where I (r) is the current flowing within a radius r. It can readily be shown that these two
equations lead to the following result , whatever the current distribution I (r).

2Nk (Te + Ti ) = (µ0 /4π) I 2 (3)

37th EPS Conference on Plasma Physics P4.181

This result is known as the Bennett relation. N is the number of electrons or ions per unit length
of the plasma column, I is the total current, k is Boltzmann‘s constant and µ0 is the permeability
of free space, µ0 = 4πx10−7 H/m.

2. The theory developed by Bennett (1934)

The model studied by Bennett [1], illustrated in Fig. 2, was The theory develop
entirely different from the MHD model described in the pre-
electrons!
vious paragraph. Bennett assumed that the electrons have a
u !
Maxwellian distribution with a drift u in the z direction, the drift
being independent of the radius r. The positive ions have a ve-
locity v in the −z direction.
electrons
In a co-ordinate system moving with the electrons they do not
experience a j ×B force, since there is no electron current in this
system. Their radial distribution will be given by the Boltzmann
B!
relation,  
eV1
n11 = n110 exp . (4)
kT11 ions
The notation used here is that subscript (1) refers to electrons
and (2) refers to ions. A second subscript refers to the co-
v !
ions !
ordinate system in which the density is measured. In a co-
ordinate system moving with the ions the ion density is given by Figure 2: The Bennett pinch.
 
eV2
n22 = n220 exp − . (5)
kT22

In the first system of co-ordinates, that moving with the electron drift velocity, Poisson‘s equa-
tion takes the form
e
∇2V1 = − [n21 − n11 ] ,
ε0
where n21 is the ion density in that system. Using eqn. 4 we can write

e2
∇2 log n11 = [n11 − n21 ] . (6)
ε0 kT11

Analogously
e
∇2V2 = − [n22 − n12 ] ,
ε0
so that
2 e2
∇ log n22 = [n22 − n12 ] . (7)
ε0 kT22
37th EPS Conference on Plasma Physics P4.181

Let α1 = e2 /ε0 kT11 and α2 = e2 /ε0 kT22 . The equations can be transformed into equations re-
lating to quantities measured in the rest system of co-ordinates, using well-known results from
the special theory of relativity.

n1 h uv i h uv i n2
n11 = ; n12 = γ2 1 + 2 n1 ; n21 = γ1 1 + 2 n2 ; n22 =
γ1 c c γ2

−1/2
−1/2
where γ1 = 1 − u2 /c2 and γ2 = 1 − v2 /c2 . Equations 6 and 7 then become
  h
2 u2 uv i
∇ log n1 = α1 γ1 1 − 2 n1 − α1 γ1 1 + 2 n2 (8)
c c
and  
v2 h uv i
2
∇ log n2 = α2 γ2 1 − 2 n2 − α2 γ2 1 + 2 n1 . (9)
c c
We can note that the temperatures are also relativistically transformed, T1 = γ1 T11 and T2 =
γ2 T22 . Strictly speaking the temperatures employed by Bennett are “two-dimensional” tempera-
tures, the relevant velocities are those transverse to the drift velocities of the electrons and ions.
The following simple solution to the pair of equations 8 and 9 was found by Bennett

n0
n1 = 2
(10a)
[1 + bn0 r2 ]
and h i h i
2
α1 γ1 1 − uc2 + α2 γ2 1 + uv
c2
n2 = h 2
i h i n1 , (10b)
α2 γ2 1 − vc2 + α1 γ1 1 + uv
c2

where
α1 α2 γ 1 γ 2 (u + v)2
b= h i h i. (10c)
8c2 α γ 1 + uv + α γ 1 − v2
1 1 c2 2 2 c2

We can note that the quantity b has the dimensions of length. When the electron drift velocity
becomes comparable with the velocity of light we no longer have a plasma in that the electron
and ion densities are no longer closely equal. One consequence of this is that we can no longer
employ a single fluid model as in magnetohydrodynamics.
The solution found is relativistically invariant. It is of interest, however, to find a very good
approximate solution for the case where the velocities u and v are small compared with the
velocity of light c. The solution then becomes

n0
n1 = n2 = 2
, (11)
[1 + bn0 r2 ]
where
µ0 e2 (u + v)2
b= ,
8k (T1 + T2 )
37th EPS Conference on Plasma Physics P4.181

We can note that n1 and n2 are very nearly equal, but there must be a small difference to produce
the electric field given by Poisson’s equation. The Lorentz factors γ1 and γ2 introduced in order
to obtain equations 8 and 9 are not identically equal to unity. The total number of electrons, or
ions, per unit length is readily found by integration to be N = π/b so that

8πk (T1 + T2 )
N=
µ0 e2 (u + v)2

and
µ0 I 2
= 2Nk (T1 + T2 ) , (12)
4π
which is the well-known Bennett relation.
In practice the current will be mainly carried by electrons, i.e. v is very small. We can note that
the combination of perpendicular electric and magnetic fields does not lead to an E × B drift of
the ions of magnitude E/B as stated in some textbooks.
If v is negligible, then in the rest system of coordinates we can write
 
eV
n2 = n20 exp − . (13)
kT2

The conclusion is that the positive ions are confined by the electrostatic field, and they are
not confined by orbiting around magnetic field lines. This conclusion will no doubt apply to
all “magnetic confinement” systems. This leads to the interesting question as to whether the
possibility of purely electrostatic confinement should be seriously reconsidered. It is known that
such a system can exist in thermal equilibrium, a natural state of matter, whereas magnetically
confined plasmas are inherently unstable.

References
[1] W.H. Bennett, Magnetically Self-focusing Streams, Phys. Rev. 45, 890 (1934)