Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

Cambridge Books Online

http://ebooks.cambridge.org/

Plasma Physics

An Introduction to the Theory of Astrophysical, Geophysical and Labora

tory Plasmas

Edited by Peter Andrew Sturrock

Book DOI: http://dx.doi.org/10.1017/CBO9781139170598

Online ISBN: 9781139170598

Hardback ISBN: 9780521443500

Paperback ISBN: 9780521448109

Chapter

4 - Adiabatic invariants pp. 32-48

Chapter DOI: http://dx.doi.org/10.1017/CBO9781139170598.005

Cambridge University Press


4
Adiabatic invariants

4.1 General adiabatic invariants


We have already encountered the first adiabatic invariant p, the magnetic
moment, that arose in a nonrelativistic treatment of particle motion in a
magnetic field. It is useful to consider a more general development of the
concept of adiabatic invariance for application to other situations, including
relativistic particle motion.
Consider a system described by dynamical variables qx and a Lagrangian
function L(qx,qx,t). The canonical momentum variables are defined by

and the dynamical equations are given by

Ms-
We consider a closed family of solutions enumerated by the variable K that
we assume runs from 0 to 2TT:

(413)

If we now consider the quantity defined by

we find that the total time derivative of J is given by

32
Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
General adiabatic invariants 33
where we have changed the order of differentiation in the second term. In
view of (4.1.1) and (4.1.2), this may be expressed as

where it is to be understood that the actual values of q{ and qx are Qx and

Since L itself does not depend upon K, we see that

so that 7 is a strict invariant of the system. It is known as the Poincare


invariant.
Note that we made no assumption about the significance of the parameter
K. One possibility is that the system exhibits periodic motion with frequency
co. Then K could be a phase factor that enters a description of the form

tfi = G(arf + *). (4.1.8)


Normally we expect to have strictly periodic motion only if L is time indepen-
dent. Then the constancy of J (which, as we shall see, is related to energy)
would not be particularly significant.
However, now suppose that L depends on time but in such a way that it
varies slowly and aperiodically with respect to the oscillatory behavior under
consideration. If we consider that the motion begins at time / = 0, then, for

where K0 determines the initial phase at which the system begins its motion.
However, for given functional form Qit0 and given phase K0, the initial con-
ditions of the system are completely determined so that K0 enumerates a
closed set of dynamical evolutions of the system for all time.
Hence Jo, defined by

^ ^ - , (4.1.10)

is a Poincare invariant, so that it is a strict invariant of the system.


Now suppose that, as a result of the slow and aperiodic variation of L,
the system slowly evolves through a sequence of oscillatory states, so that the
evolution of qx may be written as
(/),/). (4.1.11)

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
34 Adiabatic invariants
We suppose that the frequency co( /) and the phase factor K (/) are slowly vary-
ing functions of time, and that the dependence of qx upon time, as indicated
by the last term in parentheses, is also a slow variation - representing, for
instance, the slow variation in the amplitude of the oscillation. Then, clearly,
we could at any time form the quantity

^ . (4.1.12)

This is not a Poincare invariant, so there is no obvious reason why this


should be an invariant, but it is normally found that *(/) is simply related
to the initial phase K0:

*(O=«o+/('), (4.1.13)
where/(/) is a slowly varying function of time. In particular, if the variation
of L is slow and aperiodic, we will normally find that the instantaneous phase
at time / is related to the initial phase at time / = 0 by (4.1.13) where/(0 is
a slowly varying function of /. In this case, we see from (4.1.7) that

^ (4.1.14)

7(0 is an invariant, not a strict invariant, but an 'adiabatic invariant.' For


further discussion of this point, see Sturrock (1955).
In order to see the difference between the Poincare invariant and the
adiabatic invariant, we might consider the simple case of a system with only
one degree of freedom, that is then described by the variables q and /?.
Suppose that, at time / = 0, a closed family of solutions is that shown in
Fig.4.1(#). Some time later, for a complex system, this set of states of the
system may have evolved to that shown in Fig. 4.1(6). The shape is quite dif-
ferent, but the area enclosed by the contour has necessarily remained
constant.
Now suppose that, at /==0, the system is going through simple periodic
motion, as indicated by the dotted lines in Fig. 4.2(#). Suppose also that the
initial closed family of solutions is one that corresponds to a particular
amplitude of oscillation (and therefore to a particular energy) as indicated
by the solid line in Fig. 4.2(#). If the system evolves in time and then settles
down to a steady state, the system may once again exhibit periodic behavior
as indicated by the broken lines in Fig. 4.2(6). In general, there is no reason
to expect that the initial family of solutions corresponds to a new family lying
on a constant-amplitude contour. However, if the system evolves slowly and
aperiodically, then it is a good approximation that a constant-amplitude

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
General adiabatic invariants 35

(a) (b)

Fig. 4.1. We consider a closed family of trajectories in p, qy t space: (a) shows the
intersections of those trajectories with the plane / = 0, and (b) shows the intersections
of the trajectories with a plane representing a later time.

A P

(a) (b)

Fig. 4.2. We now consider a single trajectory. In (a), the system is in an almost steady
state so that the representative point in phase space maps out a closed contour. In
(b), the system has evolved to another almost steady state; the representative point
still maps out a closed contour, and the area of the closed contour in (b) is the same
as that of the closed contour in (a).

family of solutions will remain a constant-amplitude family of solutions.


Then the new family will be represented by the heavy contour in Fig. 4.2(6),
namely, the constant-amplitude contour that embraces the same area as the
original contour in Fig. 4.2(a).
A comparison of Fig. 4.1 and Fig. 4.2 is a comparison of the concepts of
the Poincare invariant and the adiabatic invariant.

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
36 Adiabatic invariants

Fig. 4.3. Schematic representation of a pendulum, the length of which that varies
slowly in time. The string passes through a ring, and the ring is moved up or down
at a rate that is slow in comparison with the oscillation frequency.

It is interesting also to consider a classical problem, that of the motion


of a pendulum when the length of the string is changed. Fig. 4.3 indicates
a pendulum, the length of which is determined by a small ring that may be
moved vertically up or down. If the ring moves slowly and aperiodically, then
there is an average upward force on the ring due to the string. Hence if the
ring is allowed to move upward slowly, that motion is carried out partly in
response to work done by the string. Hence the pendulum must be losing
energy. Calculation of the rate of change of energy shows that the fractional
rate of change of the oscillatory energy is the same as the fractional rate of
change of the frequency. Hence the 'action,' defined by

(4.1.15)
CO

where U is the oscillatory energy and co is the frequency, is an adiabatic


invariant.
It is easy to see that this quantity would not be an invariant if the motion
of the ring were arbitrarily rapid. If the ring were to move up rapidly at
the instant the string is vertical, there will be no change in the energy of the
pendulum. Hence, if the motion of the ring were composed of many very
small sudden movements, each movement occurring when the string is ver-
tical, there would be no change in the energy of the pendulum. Hence we

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
The first adiabatic invariant: magnetic moment 37
see why, in this problem, it is necessary that the motion be slow and aperiodic
in order for the action to remain approximately constant.

4.2 The first adiabatic invariant: magnetic moment


T h e nonrelativistic motion of a particle in a magnetic field m a y be described
by t h e Lagrangian function

—2 2 c0'

if we adopt polar coordinates z, /*, 0, ignore motion in the z direction, and


assume that the magnetic field is symmetric about the origin. If the field is
uniform, then
A4) = ±B(t)r, (4.2.2)
where we allow for a slow variation of the strength of the field with time.
The concept of cylindrical symmetry of a uniform magnetic field may seem
odd. If the field is static, the concept makes no sense. However, if the field
is time-dependent, the concept is significant. We know that a time-dependent
magnetic field necessarily leads to the development of an electric field. Hence
we are in fact assuming that the electric field is of cylindrical symmetry.
Combining (4.2.1) and (4.2.2) into

we see that

prz=zmr, p<t> = mr2<j> + - -Br2. (4.2.4)

The equation for pr becomes

(4.2.5)

and the equation for p^ becomes

-^f = 0- (4.2.6)

If B varies only slowly in time then, at any instant, (4.2.5) is satisfied


approximately if /"^constant and
<j>=-eQ, (4.2.7)

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
38 Adiabatic invariants
where we now use fl, in place of cog, for the gyrofrequency defined by

0 =-^?. (4.2.8)
me
Since the system varies only slowly with time, we expect that initially circular
motion about the origin will remain circular, so that
</>(/) = - e ( Q ( t ) t + K ( t ) ) , (4.2.9)
where the phase function will vary only slowly with time. Hence we can form
an adiabatic invariant from
dr l
^ (4.2.10)

Since the motion is almost circular, we may neglect the term involving p r .
Hence we obtain
J=-2irep<t). (4.2.11)
On using (4.2.7) in (4.2.4), we obtain
J=Trmr2Q. (4.2.12)
Since the transverse energy is given by
U±=±mr2Q\ (4.2.13)
we see that

/=2TT^. (4.2.14)

This expression for the adiabatic invariant is related to the magnetic moment
obtained in Chapter 3 by

7=2ir-py M . (4.2.15)

4.3 Relativistic form of the first adiabatic invariant


In relativistic theory, we must replace (4.2.1) by
, z.2 , 2
+), (4.3.1)

where we now allow for the possibility of motion in the z direction. Hence
we find that

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
Relativistic form of the first adiabatic invariant 39

(4.3.2)

If we now repeat the argument of the previous section for the present
relativistic case, we find that the equation of motion for the radial coordinate
leads to

kl* (4.3.3)
ymc
for the case that the motion is almost circular. If we introduce the phase
factor by writing

(4.3.4)
we find that (4.2.12) is now replaced by
J=irymr2Q. (4.3.5)
The two terms of p^ in (4.3.2) in fact give rise to two contributions to the
adiabatic invariant. We find that we may write

J=2TTH-^1<I>, (4.3.6)

where
H=ymr2Q, (4.3.7)
so that H is the kinetic angular momentum, and
<i> = i r r 2 B . (4.3.8)
However, these two terms are related by

7r//=M$. (4.3.9)

Hence the fact that J is an adiabatic invariant also guarantees the fact
that H and $ also are adiabatic invariants. We see that

J=TTH and 7 = - ^ - 4>. (4.3.10)

In the nonrelativistic case, we saw that the transverse kinetic energy varies

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
40 Adiabatic invariants
in proportion to the gyrofrequency, and therefore in proportion to the
magnetic field strength. This is not true in the relativistic case. If we write

0,=^, (4.3.11)

we find from (4.3.5) that


W2
y/3±ocBW2. (4.3.12)
2
In the nonrelativistic case, this gives us once more the result that (3 ± is pro-
portional to B. However, in the ultra-relativistic case, if there is no z motion
so that j8± ~ 1, we find that the energy is proportional to fi1/2, not to B.

4.4 The second adiabatic invariant: the bounce invariant


Consider, to start with, the motion of a charged particle in a static magnetic
field, allowing for a slow spatial variation in the strength of the field. We
adopt relativistic theory, since it is no more difficult than nonrelativistic
theory. Since, in a static field, 7 = constant, we see from (4.3.12) that

/3x2 = / 3 2 - ^ , (4.4.1)

where BR is a constant determined by the initial conditions of the orbit.


Since

we see that
2
V (!} (4-4.3)
This indicates that j8|j=0 wherever B = BR. Furthermore, it is obviously not
allowable that the particle should move into a region where B>BR. This
indicates that the charged particle will be reflected wherever B attains the
value BR. That is, the particle will behave as if it had been reflected by a
'mirror.' Such a field configuration is therefore termed a 'magnetic mirror.'
It is clear that the motion along the magnetic field will cease when the parti-
cle arrives at the point where B = BR, but it is perhaps not so obvious that
the particle will be reflected. It may therefore be helpful to look at the prob-
lem in a slightly different way. The motion described by (4.4.3) is the same
as that of a fictitious particle for which the total energy is expressible in
the form

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
The second (bounce) adiabatic invariant 41

(4.4.4)

where V is the 'potential energy' defined by


l
1/ 2B
(4.4.5)

We use U{ for the total energy, vt for the total speed, and we are adopting
a fictitious nonrelativistic model, even if the actual motion is relativistic.
Equation (4.4.4) is clearly the energy equation of a particle moving in a poten-
tial well described by the potential energy V(s). We know that the form
(4.4.4) of the total energy leads to the equation of motion
dV
(4.4.6)

Hence, as long as dB/ds is nonzero at the reflection point, we expect that


particles will be reflected from that point. That is, the field in that region
will indeed behave as a 'magnetic mirror.'
Let us now consider the case that the magnetic field, as a function of
distance s along a field line, varies in strength as shown in Fig. 4.4. We assume
that B has its minimum value Bm at s = s0, and that it has maxima Z?MJ and
#M,2 at positions s = sx and s = s2, respectively. Then the trapping properties
of this configuration are determined by the lesser value of Z? MJ , BM 2 , that
we refer to as BM. If we write
(3±=(3 sine, j3ii=j8cos0, (4.4.7)

Bm

S\ So S2 S

Fig. 4.4. Example of magnetic field strength, as a function of position, that leads
to a 'magnetic bottle.'

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
42 Adiabatic invariants
w e see
and if 0O is the value of 6 at s =s 0 , that a particle will be trapped
if and only if 00>6L, where

so that do = dL may be said to define the the 'loss cone' of the system.
If we consider that an isotropic particle distribution is suddenly introduced
at s=s0, the fraction of particles that will be lost, FL, is given by

FL = j - [^27rsin(9dl9=l-cosl? L , (4.4.9)
2TT JO

that is, by
/2
(4A10)

It is clear that any particles that are trapped in a magnetic mirror will
undergo a second type of oscillatory motion, in addition to their gyromotion,
namely their 'bouncing' motion between the reflection points. Hence we can
assign a second adiabatic invariant to the motion of these particles by using
(4.1.12).
We can, for convenience, re-write this expression as an integral over time,
as in (4.4.11), where the integral is taken by following the particle through
one oscillation:

^idgi. (4.4.11)

However, if any confusion arises, one should remember that the integral is
really defined as an integral over phase. For instance, where a system is
multiply periodic (as is true in the present case, since the particle also exhibits
gyromotion about the magnetic field lines), an integral over phase provides
a simple expression for each adiabatic invariant, whereas an integral over
time would lead to a confusing mixture of contributions related to all the
periodicities of the system.
Since, in rectangular Cartesian coordinates,

\--Ar9 (4.4.12)

(4.4.11) may be expressed as

J2=® \ymvyi+-Ai\ ds, (4.4.13)

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
Magnetic traps 43
where V\\ and A\\ are the components of v and A in the direction of the
element of arc length ds. It is clear that the second term involving A l{ will
vanish, since the sign of A^ changes with the direction of motion along the
field line. We can also see this result by noting that the second term really
represents the quantity given by

A-dx = ^<i>, (4.4.14)


c
where * is the flux included in the contour mapped out as the particle moves
through a complete bounce motion. However, since the particle is moving
to and fro along the same field line, the area embraced by the contour is zero
so that $ = 0.
Furthermore, we see that the first term in (4.4.13) includes two equal con-
tributions, one from motion from s} to s2, and the other from the reverse
motion. Hence we may write (4.4.13) as

= 2ym (%||ds. (4.4.15)


J
On using (4.4.3), we see that this is expressible as
J2 = 2mcyt3I (4.4.16)
where

I=\ 1-nr1 cb. (4.4.17)

If the magnetic field is static, so that y and 0 are constants, our calculation
shows that the geometrical quantity / i s an adiabatic invariant. However, if
the magnetic field varies slowly in time, so that the energy of particles may
change, then the quantity given by (4.4.16) is an adiabatic invariant, but the
quantity given by (4.4.17) is not.
The existence of the first and second adiabatic invariants have important
consequences concerning the trapping of particles in complex magnetic fields,
as we shall now see.

4.5 Magnetic traps


Consider the magnetic field configuration shown in Fig. 4.5, a configuration
that is not assumed to have any particular type of symmetry. Consider the
motion of a particle that starts out with a given velocity vector so that we
know the initial value of the first adiabatic invariant, that is approximately

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
44 Adiabatic invariants

= BR

Fig. 4.5. A magnetic field configuration that does not have cylindrical symmetry but
leads to the trapping of charged particles.

constant. If the field is static, so that the particle energy is a constant, we


know that the particle is reflected at points where the magnetic field strength
takes the value BR.
As we shall see in the next chapter, a particle tends to drift in an
inhomogeneous magnetic field, so that it will migrate to other magnetic-field
lines. If we now construct two surfaces on which B = BR, and between
which B<BR, we know that the particle will be reflected at these two
surfaces and trapped between the surfaces.
However, since we are assuming that the magnetic field is static, we can
go further. We know, from the properties of the second adiabatic invariant,
that the particle is constrained to move in such a way that / = constant, where
/ i s defined by (4.4.16).
Hence, for each field line bounded by the two surfaces B = BR, we may
calculate the value of /. We may then identify the shell of magnetic-field lines
for which 7=/ 0 , where Io is the initial value of /. Hence we know that the
particle will move in such a way that it is constrained to remain on the shell
and that its motion is bounded by the two reflection surfaces. If the shell
forms a closed surface, as indicated in Fig. 4.5, then the particle is constrained
to remain on that surface, no matter what type of drift the magnetic-field
gradients may imply.
Another example of a magnetic-field configuration that leads to trapping
is that shown in Fig. 4.6, that of the magnetosphere of a body with a dipole
magnetic field, such as the Earth. In this case, particles bounce from one
reflection point to another in helical trajectories. In addition to this motion,
there is (as we shall see in the next chapter) a drift around the earth due to

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
Magnetic traps 45
Reflection point
- Gyromotion
and
Bounce motion

Reflection point

Fig. 4.6. A dipole-type magnetic field, such as that of the Earth, leads to particle
trapping. Particles exhibit gyromotion around magnetic field lines, bounce motion
along field lines between reflection points, and drift motion around the Earth.

the curvature of the magnetic field lines (curvature drift) and to the spatial
variation of B with radius (gradient drift). Hence, for the particles trapped
in the Earth's magnetosphere, there are three types of oscillations due to
(a) the gyromotion, (b) the 'bounce' motion, and (c) the drift motion.
Another example is the trapping of particles in the Sun's magnetic field,
that is much more complex than the approximately dipole field of the earth.
In an active region, that inevitably contains surface magnetic fields of
opposite polarities, and typically contains at least one pair of sunspots of
opposite polarities, part of the magnetic field will be as shown in Fig. 4.7.
The Sun produces some radio bursts (stationary Type IV microwave radio
bursts) that are initiated by flares and are believed to be due to gyrosyn-
chrotron radiation. Hence, such flares indicate that mildly relativistic elec-
trons are somehow trapped at coronal heights in an active region. We see
from Fig. 4.7 that, here again, if the field is static it is reasonable that particles
should be trapped in the region above the two reflection surfaces.

Trapping shell

Fig. 4.7. A magnetic flux tube in a solar active region also provides for particle
trapping. A: positive-polarity sunspot; B: negative-polarity sunspot.

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
46 Adiabatic invariants
4.6 The third adiabatic invariant
Let us now consider the possibility that particles are trapped in configurations
such as those shown in Figs. 4.5 through 4.7, and that the magnetic field is
not static but slowly varying in time. Some of the details in the arguments
concerning trapping must be changed, since particle energy changes and
therefore the reflection point of a particle will change, and we must consider
the bounce invariant given by (4.4.16) rather than that given by (4.4.17). We
note that, in such situations, there is a third periodicity associated with the
motion, namely the periodic motion around the flux tube. In the case of the
magnetosphere of the Earth, this corresponds to the periodic drifting of
particles around the Earth. Once again, we may use (4.1.12) to associate an
adiabatic invariant with this periodic motion. The equation for this invariant
may be written as

j[ f J s , (4.6.1)
where the contour is now a phase contour related to this drifting motion, that
is, a closed contour that lies on the surface / = / 0 and is directed in the same
way as the drift motion.
The ratio of the first and second terms in (4.6.1) may be written approx-
imately as

*™** (4.6.2)

where vD is the magnitude of the drift velocity around the tube. As we shall
see in the next chapter, the magnitude of vD is given approximately by
1 me v±2

considering only the gyromotion and ignoring the motion along the magnetic
field, so that

However,
v±=Slr±, (4.6.5)
so that (4.6.4) becomes

fe]2 (4.6.6)

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
Problems 41
Since it is usually the case that r±<R, we see that the first term in the
integral (4.6.1) may be neglected in comparison with the second term. Hence,
to good approximation, the third adiabatic invariant may be written as

/3=f*. (4.6.7)
where $ is the magnetic flux embraced by the closed contour associated with
the periodic drift motion around the trapping shell.
It is worth reminding ourselves of the conditions for the existence of these
three invariants. The first adiabatic invariant is valid if the magnetic field
varies on a time-scale long compared with the gyroperiod, and if the spatial
gradients are characterized by lengths that are large compared with the
gyroradius. (Throughout these discussions, we also include the condition that
the variation should be aperiodic.) The second adiabatic invariant is valid if
the time scale of variation is long compared with the bounce period, and the
third adiabatic invariant is valid if the time scale for variation is long com-
pared with the period of drift motion around the trapping shell.
For simplicity of discussion, we have been considering the motion and
trapping properties of particles of given energy and given magnetic moment.
In discussing a real plasma machine, or a real astrophysical situation, one
would of course be concerned with a range of particle energy and a range
of magnetic moment, and also perhaps with particles of more than one
species. In these cases, all particles will not be confined. As shown in Section
4.4, particles with direction vectors inside the 'loss cone' will not be trapped.
For a specific magnetic-field configuration, there may also be a similar
restriction on the range of values of the second adiabatic invariant / for which
particles are trapped. Such cases need to be investigated on a case-by-case
basis.

Problems
Problem 4.1. Consider a mirror machine of length 2L with a mirror ratio
of 10, so that B(L) =B( -L) = 105(0). A group of N(N> 1) electrons with
an isotropic velocity distribution is released at the center of the machine.
Ignoring collisions and the effect of space charge, how many electrons
escape?
Problem 4.2. An electron with speed v0 moves along the axis of a tube of
length 2L, in which the axial magnetic field has the form
B(z)=Boe K\z\

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013
48 Problems
(a) Find an equation for da/d/, where a is the pitch angle of the trajectory.
(b) Find the minimum value of a 0 , where a0 is the value of a at z = 0, that is
needed to ensure that the electron is trapped.
(c) Sketch a(t) and vz(t), where vz is the component of velocity parallel to the
axis.
(d) Find the bounce period as a function of a0.
Problem 4.3. Suppose that electrons are injected at the center of the con-
figuration described in Problem 4.2 with a pitch angle distribution f(a) of
the form

Ignoring collisions, what fraction of the electrons escape?

Downloaded from Cambridge Books Online by IP 210.212.187.78 on Wed Aug 14 06:45:06 WEST 2013.
http://dx.doi.org/10.1017/CBO9781139170598.005
Cambridge Books Online © Cambridge University Press, 2013

You might also like