TE1VIPERA'TURE il.ND STABILITY PHOPERTIES OF A RADIALLY CONSTRICTED STEADY -STATE PLASMA BETWEEN ELECTRODES
A Thesis submitted to The Australian National University for the Degree of
Doctor of Philosophy in the Department of Theoretical Physics
by
Patrick William Seymour
Research School of Physical Sciences, Institute of Advanced Studies, The Australian National University,
PREFACE
Each chapter of this thesis describes original work carried out by the candidate at The Australian
National University during an effective period of nine terms from April
1958
to July1961
,
Vfuere it has been necessary to relate the candidate's workv
to the work of others, proper acknowledgment has been made in the form of detailed, specific references.
Signed:
July
1961
iii ACKNOWLEDGNlENTS
In various ways many people have contributed to the completion of this thesis, and it is my pleasure to set down the follov'ling expressions of appreciation.
Within the Department of Theoretical Physics, I first sincerely thank my Supervisor, Professor K. J. Le Couteur, who at intervals suggested the problems considered in these pages, then constructively criti-cized my attempts at solution, and finally, when at certain stages of advancement my difficulties seemed insurmountable, provided timely advice and encourage-ment.
I run also very much indebted to Dr. F. C. Barker, Senior Fellow, who, in the earlier stages of my
researches, spent much time patiently providing me with some insight into the mathematical techniques of the theoretical physicist, and who assisted me, via medium of discussion, to clarify my views on the
topics discussed in Chapters 1 and 2 of this thesis. In connection with Chapter 2, I am also indebted to Dr. D. C. Peaslee, Reader, who 'perused an earlier form of Chapter 2, and offered helpful suggestions.
The presentation of results in convenient graphical form through the chapters involved me in the rather extensive use of an electric desk
calculator, and it is with gratitude that I acknow-ledge the many spot-check calculations carried out by Mrs. B. Nerdal, Departmental Assistant, in
iv
Within the Department of Particle Physics exists the Experimental Plasma Physics Group, led by Dr. A. H. Morton, Research Fellow. As I have produced for
publication manuscripts corresponding approximately to the four chapters of this thesis, Dr. Morton has kindly read them, and offered useful suggestions based on his knowledge and practical experience in the plasma
physics field. In voicing my appreciation to
Dr. Morton for this assistance, I should particularly like to stress my pleasure at his pronouncement, after study of my principal results, that the laboratory experiment finally proposed (pp. 128-131) is feasible. This is also a good opportunity for me to express my
gratitude to Mr. G. F. Cawsey, Principal Scientific Officer, Department of Supply, Victoria, who , during an earlier period of association with the Plasma Physics Group, taught me the meaning and use of
norder-of-magnitude" or "good!! physics, and provided me with much valuable food for thought in connection with the stability topic of Chapter
4.
Here myindebtedness to the Plasma Physics Group would not be complete without a note of thanks to the Plasma
Physics Research Scholars, Messrs. I. S. Falconer and R. H. Hosking, with whom many pleasant and inform-ative discussions have been held.
The Australian National University possesses
a comprehensive Institute of Advanced Studies Library, associated with which is a growing specialised
particular I gratefully acknowledge the pleasant and efficient manner in which my numerous requests have been handled by Miss N. G. Cook, Assistant Librarian, and Mr. L. A. Delpratt, Library Assistant, at the former location; and by NIrs. E. Lennon, Library Assistant, at the latter location.
The four chapters of this thesis correspond approximately to the following papers by the writer:
(i) Drift of a Charged Particle in a Magnetic Field of Constant Gradient. Aust. J. Phys., Vol. 12, No.4, December 1959. (ii) Estimation of the Maximwn Temperature in a
Radially Constricted Gas Discharge Between Electrodes. Aust. J. Phys., Vol. 14, No.1, March 1961.
(iii) The Influence of Thermoelectric Effects on the Maximum Temperature in a Radially Constricted Gas Discharge Between Electrodes. Aust. J. Phys., Vol. 14, No.2, June 1961.
(iv) A Stability Criterion for a Radially Constricted Gas Discharge Between Electrodes. Aust. J. Phys., Vol. 14, No.3, September 1961.
This is a convenient opportunity for me to express my sincere appreciation to the Board of
Standards and the Advisory Co®nittee of the Australian Journal of Physics for the extremely cooperative
manner in which these papers were accepted for publication.
Solution of the problems considered in this
vi thesis was made possible by the award to the writer of a Research Scholarship, and I am much indebted to the Council of The Australian National University for making this opportunity available to me. In this connection, I am also indebted to lVIr. J. Vi. J. Byrne, Assi stant Secretary, Department of Supply, N.S.W., who kindly released me from my duties as Supervising Engineer, Telecommunications Branch, N.S.Vv., to permit me to
undertake my doctoral studies; and to the Commonwealth public Service Inspector, N.S.W., who approved leave of absence for this purpose under the provisions of
Section 71(1) (b) (1) of the Public Service Act.
Pursuit of this course of study was considerably aided by my wife , Gladys Joyce Seymour, who carried out the full-time duties of Secretary, Department of History, Institute of Advanced Studies, during my
studies here, typed the manuscripts of my four papers on an evening basis, and then, very recently, retired from official life - to type the stencils for the entire thesis! All who read this appreciation will understand how very much I am indebted to Gladys Joyce.
There are others who have helped me in general ways in connection with completion of this work, and here I place on record my appreciation of their
efforts.
To conclude these acknowledgments, the pleasant period from April 1958 to July 1961 at The Australian National University is epitomised for me in the
Edward FitzGerald's first version of the Rubaiyat of
Omar Khayyaro:
Myself when young did eagerly frequent Doctor and Saint, and heard great Argument About it and about; and in the end
Came out a wiser man than in I went.
Canberra, A.C.T. July 1961.
viii
TEMPERATURE AND STABILI'rY PROPERTIES OF A RADIALLY CONSTRICTED STEADY-STATE PLA9~ . BETV'lEEN ELECTRODES
SUMMARY
The above subject is discussed in four chapters through this thesis. The first chapter bears on charged particle orbit theory and a simple case of
plasma instability, while the second and third chapters are concerned with the flows of heat and electricity in a centrally-constricted plasma geometry under assumed steady-state conditions. In the fourth chapter a sufficient stability criterion is derived for this particular plasma configuration.
A table of contents, giving the section titles within each chapter and the corresponding page numbers, is included for ease of reference. To facilitate
cross referencing, combined chapter and section numbers are included at the top left-hand side of each page. Specific references given in the text correspond to alphabetical reference lists provided at the close of each chapter. The fourth chapter is followed by a review of topics covered in the thesis, and some
remarks on possible future experimental and theoretical studies. Additional references, relevant to the
research undertaken, appear at the end of the thesis. A detailed swnmary of the work covered now follows.
In Chapter 1 charged particle motions are considered. Using the Lorentz transformation of special relativity to relate the electric field, E'
-.1. ' and the magnetic field, ordinate system, to and H -.1.
co-ordinate system (where the subscript ~ on a field vector signifies that it is perpendicular to the constant relative velocity, ~, between the two co-ordinate systems), the familia.r E x H drift velocity expression for a charged particle in crossed electric and magnetic fields is obtained by a method not
ix
usually employed in the textbooks. This result, which is independent of the sign of the charge, is then
generalised to obtain the usual sign-dependent drift velocity result for charged particle motion in crossed force a.nd magnetic fields.
To obtain physical insight into the mechanism of plasma instability in simple cases, the Rosenbluth-Lon~aire particle orbit approach to the hydromagnetic analogy of the classical Rayleigh-Taylor instability of fluid dynamiCS is reviewed non-mathematically, using as principal factors the sign-independent and sign-dependent drifts mentioned above. Since in this case initiation of instability is found to arise from a sign-dependent g x H type of drift (g the gravitation acceleration vector), mention is also made of the Alfven
sign-dependent drift velocity for a charged particle moving in an inhomogeneous magnetic field as a possible source of instability.
Partly as an exercise in itself, and partly because of its possible aggravating influence on instability, an extension of Alfven's first-order perturbation theory is included in the form of an
exact solution, in terms of complete elliptic integrals of the first and second kind, for the motion of a
x having constant gradient perpendicular to the direction
of the main field. This exact solution yields as
approximations Alfven's result and the case of circular orbit, and includes the somewhat novel case in which the magnetic field vanishes at points in the particle's orbit, for which perturbation methods are inappropriate. In this latter case larger drift velocities, of the
order of the particle velocities, are expected in the regions of vanishing magnetic field, and hence larger charge separation effects than in the Alfven region may be encoID1tered. One likely interpretation of this result is that plasma instabilities may be increased in growth rate.
In Chapter 2 a steady-state deuterium discharge between two electrodes is considered and the free boundary surface of the plasma is assumed thermally
insulated when pinched away from the walls of the discharge tube. Cooling is therefore by heat conduction to the electrodes , compared to which
bremsstrahlung loss is shown to be negligible if the dischargo is not too long. The main question exruained is how much the plasma maximum temperature T
ill can be raised by constricting the cross section of the dis-charge near the centre.
The analysis is confined to substantially-ionized deuterium, and curves based on the Saha equation are provided to show the minimum gas temperatures required for various particle densities. With neglect of
thermoelectric effects, there exists a median plane, normal to the longitudinal axis of the discharge,
xi
are symmetrical. Defining we as the electron gyro-frequency and ~e as the electron self-collision time, the analysis is carried through both by specialising the current density and heat flux vectors, j and q,
to cater for an isotropic plasma (We~e«1), and for
a plasma made anisotropic by a strong, external magnetic field H (w~»1) o
e e Prior to detailed mathematical analyses, however, a simple continuity
argument yields the important relationship,
q + V j = o ,
where V is the electric potential, provided that everywhere within the discharge q is parallel to j.
This type of flow is for convenience termed lon gi-tudinal flow.
The detailed axi-symmetric analyses for slightly and greatly constricted discharges show that
where
a
o and Ko are related to Spitzer's formulaefor the electrical and thermal conductivities of a highly-ionized gas, and T
m is the temperature on the median plane, where V
=
0. For strictly longitudinal flow, with j and q parallel to Hat every point within the discharge, this result is
valid for all values of we~e ' and hence for all
values of the radial compression ratio
v
from unity upward, where, in t erms of the maximum and central radii of the discharge, P1 and Po respectively,xii
are held at zero temperature, we obtain for the
constricted discharge the useful numerical relationship
where V* e
*
* - 1
T ~ 5899 V volt deg. ,
ill e
is the semi-voltage across the discharge, and the star superscript is used when necessary to indicate that the discharge is no longer Itlinear".
It is emphasized that this result is independent of v. For these detailed analyses, initial use of an electric stream function 'f and the electric potential V as orthogonal curvilinear coordinates simplifies the plasma energy equation. An analytic solution of this equation if the temperature T is a function of ~ only shows that the heat flow can be everywhere perpendicular to the flow of electricity only when the streamlines are straight, and parallel to the longitudinal axis of symmetry. Analytic solutions if T is a function of
V, representing the longitudinal flow, are given for (1) straight strerualines parallel to the longitudinal axis of symmetry, (2) hyperbolic streamlines to
represent a discharge constricted at the median plane. A curve giving the variation of T with axial distance
along the linear discharge is included.
expressions for tne central temperature
Upper-limit
T a r e m
obtained in terms of the total current carried by the longitudinally stabilized discharge and its character-istic dimensions, and a curve gives the dependence on constriction of T and of the resistance R between
m
semi-xiii
length. If large radial constriction at the median plane is achieved by use of a strong guiding magnetic field which makes the conductivities anisotropic, a tensorial analysis is required, but leads to the same results for all values of we're when the flow is
strictly longitudinal. Vfilere the thermal insulation and neglect of bremsstrahlung approximations apply, the direction of heat flow is not expected to depart
significantly anywhere from that of the flow of electricity when we't e« 1, and so the above curve should provide a useful guide to the increase of Tm
and R due to constriction. For an area constriction of 400:1, Tm and R are increased by a factor of
about
4
.
field
Characteristics relating T and the magnetic H when W 't
=
1e e are given for various values of the total particle density. In terms of the known central temperature and total central confining
magnetic field ~, the total concentration can be
estimated from the pressure balance as
14 2
*
-3n
=
2.88 x 10 HIT
particles Cill. m mand so the characteristics given can be used to obtain an order of magnitude estimate of we'te' and hence of
the state of the plasma.
Chapter J has for its topic a generalisation of the cases examined in Chapter 2. The same a pproxi-mations are made, but the analysis is extended by
including in the expressions for j and q vector
xiv
effects of Seebeck and Peltier. This results in the temperature and voltage distributions no longer having symmetry about a median plane. For simplicity the analysis is limited to strictly longitudinal flow, so that a vector method is sufficient.
Again a simple continuity argument shows that
q + V j = o ,
where V is the electric potential, but now detailed analyses reveal that the equipotential surface on
which q
=
V=
°
is displaced from midway between theelectrodes nearly to the cathode, and that the maximum temperature is displaced somewhat from the midway
position towards the anode by amounts which decrease as central constriction increases. The important influence of inclusion of thermoelectric effects is
•
that the maximum temperature is increased by approxi-mately
14%
for about the same total applied voltageproducing a given current in a particular discharge geometry.
The characteristic relating the maximum
temperature and resistance ratio e and the radial compression ratio v obtained in Chapter 2 is not changed by thermoelectric effects. Comparison of voltage and also temperature versus distance
characteristics for linear and constricted discharges without and with thermoelectric effects is given by means of graphs. For the temperature characteristics
simple physical interpretations are provided.
In Chapter
4
the stability of thexv
electrodes is considered. Ini t ially, the noriIlal lllode
and energy principles for the examinat ion of pl asma
stability are reviewed and the results contrasted, and
it is recalled that the diffusion equat ion,
1 2
:::: \I H
a
t
4 7C (5 f..lsuggests that the magnet ic field l eaks through the plasma with a characteristic time of decay
't:d
=
4 1t f..l (J L 2where f..l is the permeability, (J the electrical
conductivity, and L a l ength comparable with the
dimensions of the system. Teller's power ful stability
criterion is next obtained in a useful integral form by
means of a thermodynamic dnalysis of interchange
instability. This general geometrical result is then applied to the centrally-constrict ed plasma between
electrodes, and a sufficient stability criterion
d.erived. For this type of system, initially
stabilized by a strong, external guiding magnet i c
field, it is found that the onset of instability occurs for a discharge current
where ¢E froln the external sol enoid is the total
flux through any discharge equipotential surface, is the discharge semi-length, and practical simpli
-z e
fying asswnptions of high radial compression ratio v and maximum discharge radius P
xvi In terms of the temperature, the decay time can be written as
-13 2 3/2 -2 -3/2 'L" d ~ 2 x 1 0 L T cm . de g . sec 0 ,
a result which confirms, for practical values of L and T, that the plasma !?steady state1l is in the region of milliseconds.
To conclude this detailed sUflhuary, it is thought that the procedure outlined in Section
4
and discussed in Section 5 of Chapter 4 could form the basis of an interesting laboratory experiment for the observation of transition from stability to instability in axvii
CON'rENTS
PREFACE.
ACKNOWLEDGI\lIENTS , SUMMARY.
Page ii
Section 1.
Chapter 1
CHARGED PARTICLE MOTIONS
Introduction.
iii viii
1
2. Charged Particle Drift in Crossed
3.
4.
Electric and Magnetic Fields. Charged Particle Drift in Crossed
Force and Magnetic Fields. Physical Mechanism of Plasma Instability. Charged Particle Drift in an
Inhomogeneous Magnetic Field;
Introductory Remarks.
2
5
8
5.
Exact Solution for Charged Particle6.
Drift in a Magnetic Field of
Constant Gradient. 9
(a) Derivation of Basic Equations. 9 (b) Comparison with Alfven's Result. 12 (c) Exact Solution in Terms of
Elliptic Integrals.
Case 1. Electron does not enter Region of Reversed
13
Magnetic Field. 13 Case 2. Electron enters Region of
Reversed Magnetic Field. 15 (d) Discussion of Results. 18
Section
1.
2.
Chapter 2
ESrrDIATIOH OF 'rHE IJlA.XIlviUM TEMPERATURE IN A RADIALLY CONSTRICTED
P:r.A'SMA-BE'rWEEN ELEC'EHODES Introduction.
Application of the Saha Equation.
25
25
3.
Preliminary Discussion of the CurrentDensity and Heat Flux Vectors. 26
4.
j and q for Conditions of Isotropyand Extreme Anisotropy (Hall and
Righi-Leduc effects only included) .
30
(a) j and q for IsotropicConditions (we're < < 1 ) .
35
(b) j and q for Conditions of
Extreme Anisotropy (we'r e > > 1).
36
5.
Establishment of the PlasmaEnergy Equation. 38
6.
Solutions of the PlasmaEnergy Equation. 42
6
.1.
Curved Stream Tube of aConstricted Discharge. 43 6.2. Use of the Stream Function '¥
and the Electric Potential V as Orthogonal Curvilinear
Coordinates (we're «1).
46
6
.3.
Solution of the Energy Equationwhen the Heat Flow is every-where Perpendicular to the
Flow of Electricity (We 're«1). 50
6
.4
.
Solution of the Energy Equationfor Longitudinal Flow (we're«1).
51
Case 1. Streamlines Parallel tothe Axis of Symmetry. 52
xix Section
Case 2. Streamlines Curved. 57
6.5.
Solution of the Energy Equationfor Longitudinal Flow (we't"e»1). 62 7 . Compari son of Maximum Temperatures
8.
1.
and Discussion of Results. -Keferences.
Chapter
3
THE INFLUENCE OF THERMOELECTRIC EFFECTS ON THE IVLAXINiUM TEMPb;H.ATURE IN A RADIALLY
CONSTRICrrED PLASMA BE'rvVEEN ELECTRODES Introduction.
2. Isotropic Forms of j and q (Seebeck
71
and Peltier effects only included). 72
3.
Solution of the Plasma Energy Equation1.
for Strictly Longitudinal Flow. 3.1. Curved Stream 'rube of a
ConstricteQ Discharge.
3.2. Solution of a SL~plified Form of the Plasma Energy Equation using Complex Variables.
3.3.
Specialisation of Results for aLinear Discharge.
3.4. Specialisation of Results for a Discharge having Hyperbolic Strear1l1ines.
Discussion of Results. References.
Chapter 4
A STABILITY CRITERION FOR A RADIALLY CONSTRICTED PLASNIA l3EThEEN ELECTRODES Introduction.
75
76
78
85
88
96
99
xx
Se ction .fage
2. Stability Analyses of Cylindrical
Gaseous Cqnductors: TellerYs
Stabilit~ Criterion, 102
(a) Examples of Normal Mode Analysis,
( b) EX8J.Jlples of Energy Principle
1
Analysis.
102
106
3. Thermodynamic Derivation of Teller's Stability Criterion.
(a) Calcula~ion of OW.
(b) Calculation of OU,
(c) The Flute Instability and
Teller's Criterion.
I I I
I I I
112
114 4. .Application of Teller's Criterion to
5.
6.
a Radially Constricted Gas Discharge. 118
Discussion of Results.
References.
REVIEW OF TOPICS COVERED ~~D POSSIBLE FUTURE EXPERIMENTAL Al\JD THEORETICAL
123 125
STUDIES. 128
Chapter 1
CHARGED PARTICLE MOTIONS
1. Introduction.
The motion of a charged particle under various
magnetic and electric field conditions has been discussed by authors such as Alfven (1950), Spitzer
(1952, 1956), Allis (1956), Post (1956), Simon (1959), Chandrasekhar and Trehan (1960), Delcroix (1960), and
Linhart (1960). The subject is important because,
although the macroscopic equations describing the
behaviour of an ionized gas can, in various
approxi-mations, facilitate certain plasma calculations (see,
for example, Chapman and Cowling 1953; Brueckner and
Watson 1956; Chew, Goldberger and Low 1956;
Chandrasekhar, Kaufman and Watson 1957; Green 1959;
and several Geneva 1958 papers), it becomes increasingly obvious that the subtleties of topics such as plasma
instability can, in simple cases, be better understood physically via the microscopic or particle orbit theory.
In this chapter we first dwell briefly upon the
derivation of two important charged particle "drift
velocity"results. Application of these results to a
simple case of plasma instability, following Rosenbluth
and Longmire, is then quickly reviewed, and, after some
remarks on Alfven's approximate drift velocity solution
for a charged particle moving in an inhomogeneous
magnetic field, an exact solution is obtained in terms
of elliptic integrals for the motion of a charged particle in a magnetic field having constant gradient perpendicular to the field direction.
Ch .1-2 2
In this work we adopt the symbols E and H . respectively for the space-time averaged Lorentz
electric and magnetic field vectors (Rosenfeld 1951) , and, by considering the current and charge densities explicitly in the averaged form of Maxwell's field equations for a vacuwn, draw no distinction between either E and the electric displacement, D, or Hand the magnetic induction, B (See also Spitzer 1956, p.23). The space-time averages are evaluated for regions
which, although small on the macroscopic scale, are nevertheless large compared to regions in which the microscopic electric and magnetic fields can be
expected to vary significantly; that is, for regions which must be large enough to contain many charged particles.
2. Charged Particle Drift in Crossed Electric and Magnetic Fields.
lf ~ is the constant relative velocity between chosen moving and fixed frames of r eference, and the subscript 1 on a field vector indicates that it is perpendicular to
y,
then, from the special theory ofrelativity (well discussed by Stratton 1941, McRea 1954) it is possible by means of the Lorentz
trans-,
formation to relate the electric field, Eo , - ..1- and
,
the magnetic field, H 1.' in the moving frame of reference, to E.l and H.l in the fixed frame of reference, as follows:
E +
Y
X g.l,
-.1EJ. :: (2.1)
J
2Ch.1-2 3
and
1
=
H1. -
c
__
V x E c2 - -1.(2.2)
J
1 - (V/C)2where c is the light velocity, and electromagnetic
*
units are employed .
The components of the total electric and magnetic field parallel to v are unchanged by the Lorentz transformation.
As pointed out by Alfven (p.6) and Post (p.348), when the el ectrical conductivity of an ionized gas is high, electrostatic fields are not readily supported. Hence if we can assume that the el ectric field is
small compared to the magnetic fi eld, and that v < < c, equations (2.1) and (2.2) reduce to the simpl e forms
E' = E + v x H
(2.))
and
H'
=
Hwhere the components of the field vectors parallel to v have been included in these results.
Alfven remarks that equations (2.3) and (2.4),
which indicate the independence of H and the
dependence of E on the coordinate syste~, are omitted in most treatises on electromagnetism.
The relativity of the el ectric field can l ead to a familiar drift velocity result as will now be shown.
Using the superscripts 1. and II to denote respectively the electric field components perpendicular and
parallel to the total magnetic field,
g,
it is*Electromagnetic units will be employed throughout this work in conformity with those of Spitzer
Ch.1-2 4
evident from (2.3) that the term v x H cannot cancel
E~
but that it can cancelE~,
so thatE'~
can bemade to vanish. From E' 1
=
E.l + V x H == 0 we obtain by operation with H x the equationH x E.l + H x (,y x H) =:: 0, w.hich l eaves the component of v parallel to H undefined. Expansion of the
vector triple product gives
v
.
H El x Hv
-
H-
=
(2.5)
2 2
H H
Without loss of generality we can now impose the condition v ' H =:: 0, to obtain the simple form
v
E.l X H
= : : -2
H
E x H
=
- - -
2H
If E and H are
asslli~ed constant in space and time, this result may be
displayed as
v =
-e
E x H
Since, for this velocity, there is no electric
field perpendicular to H in the moving frame of
reference, it follows that a charged particle will
(2
.6)
perform either circular orbital motion about a magnetic line of force, in a plane perpendicular to the direction of H'
-'
or helical motion of constant or increasing pitch about a line of force , depending on whether there is a uniform or accelerating component of the motion parallel to H--
'
i . e., depending on whether Ell is zero or finite, Here we neglect the effect ofcollisions on the motion parallel to H.
From the fixed frrune of reference, plasma in the moving frame of reference will appear to be moving with
Ch.1-3 5
Wi th E,II pr esent, the resultant charged particle
motion in the moving frame as viewed from the fixed
frame will be a sideways-drifting helical motion of
increasing pitch. From these r emarks it follows that
ve must bo the electric drift velocity particularly
well discussed by Spitzer
(1
956
,
pp.3-5), and alsotreat ed by Alfv~n (pp.
47-48),
Spitzer(1
9
52,
pp. 301-307),
Allis (pp.3
8
4-3
8
5),
Post (pp.353
-
354)
,
Simon(
p
.26),
Chandrasekhar and Trehan(
pp
.
1
8
-1
9)
,
Delcroix(
p
.
45)
and Linhart (pp.21-24).
result
(
2
.
6
)
for particles of theConsideration of the
having charge q.
1 and mass n1i ,
·th
1 species ,
l eads to the
important conclusion that ve(i) is independent of
q.
1m
..
1 13.
Charged Particle Drift in Crossed Force and Magnetic Fields. Physi cal-Me"chaIiTsm of .Plasma Ins ta bili ty .The electric drift veloci ty ,Ye can be
generalized by replacing E with ,F -i
Iq
i ' where - 1 F·is a constant external force field. 'rhen, wi th H
again constant,
v
=
-d
F. x H
-1
dependent on the sign of qio
( 3 .1)
Various drift velocities (polarization, curvature,
magnetic gradient and those given by
(
2
.6)
and (J.l)above) are discussed and tabulated by Chandrasekhar
and Trehan (pp.
14-36,
6
5-
96
).
Here we note particularly that the qi have the
direction of motion E x H in constant, crossed
Ch.1-3
whereas for constant, crossed ext ernal force and
magnetic fields, the q.
l of opposite sign proceed in opposite directions.
Hence, for a plasma having characteristic
dimensions much in excess of the Debye length (Post ,
pp.
349
-
350;
Spitzer1
956,
pp.16
-
17;
Francis1960)
,
these results indicate that where an uninhibited drift
of charged particles can form, the E x H drift will
produce, under steady-state conditions, zero el ectric
current, while the F. x H drift can produce an - l '
-electric current of finite value.
These sign-independent and sign-dependent drifts
can lead to a clear appreciation of the detailed
physical mechanism of plasma instability where the
geometry is simple, as has been emphasised by
Rosenbluth and Longnire
(1
957).
To illustrate this 6point we sketch the argwuent given by the above authors
for the hydromagnetic anology of the classical Rayleigh
-'Taylor instability of non-conducting fluid dynamics.
This type of instability was first analysed hydro
-magnetically by Kruskal and Schwarzscnild
(1954),
whoused a macroscopic approacn; the hydrodynamic case is
treated by Lamb
(1
945)
.
A horizon tal plasma (heavy fluid) is Lllagined
supported against gravity by a uniform horizontal
magnetic field (weightless fluid), the resulting
in-terface being plane horizontal, as in Fig. l(a).
The uniform magnetic field that penetrates the
plaSIlla is asslJIlled attenuated with J~espect to the
oxternal field. The combined effect of the change
Ch.l 7
j
!
J
k
· .
.
.
. . . .
.
0.!=L jj
,k
unit vectors·
·
.
.
.
.
.
.
.
. . . .
.
.
.
/
H
urw~rcLs, il1 x.-d..irec.t,·oh.eeL)
PLa..ne.,
hori:z.ont"-a..L in"terfo-ce- be.tween.t
La..
sma- a.. n cL ma...j
net i c... fieL
~}
H·
z
o
J
.
.
Pos·,rive ·,on. bouncLa.r
J /
E Lec.tron. bouncLa.rJ
H
ufwa.rols, in )(- oLif'€.c.ti 0 I'\....Fi
J
.
i
I
LLustra..tiovv
of+he.
~Jd..roW\.o...Jn~tic
A
no...Lo~LJ
ot the..
R-CL
C11.1-4 8
pressure gradient in the plasma initially balances the
gravitational force. It is then assumed that the
in-t erface is perturbed as shown by the solid line boundary in Fi g. 1 ( b) . Here the q. (ions and el ectrons) in the
l
electrically neutral plasma are under the influence of crossed F. - l = m. l _ g and H fields (g the gravitational acceleration vector), and so proceed in opposite
directions by equation (3.1). Since charged particle drifts are soon inhibited in the perturbed boundary region of this plasma configuration, charge separation results within a surface layer, and consequently charge appears on the undul ating boundary surface shown in Fig. 1 (b) . Attendantly, el ectric field vectors ' E
~ Y and k Ez are creat ed, and, from equation (2.6) , mass motions in the direct ions j Ey x
li
and k Ez x Htake place. It is at once seen that the
1
Ey x H drift is in the z-direction, and, as the detailedmathematical treatment of Rosenbluth and Longmire shows, the corresponding mass motion t ends to encourage growth of the perturbation at A and B, Fig. l(b), and the
syst em is unstable , with a rate of growth identical
with that found in the conventional Raylei gh-'raylor case,
An interesting practical identification of R ayleigh-Taylor instability in a stabilized linear pinched plasma between electrodes is described by Albares et al (1961). However, it is not appropriate to take up a more
general study of plasma stabil ity her e, and so this is conveniently deferred until Chapter
4.
4.
Charged Particle Drift in an Inhomogeneous Magnetic Field; Introductory Remarks.A sDnple expr ession for the sign-dependent drift
Ch.1-5 9
magnetic field has been obtained by Alfv~n
(1
9
4
0;
1
950
,
pp. 13-23), who, in his first-order theory, considered
the inhomogeneity as a small perturbation of a uniform illagnetic field. Spit zer
(1
956
,
pp.6
-7)
di scussesAlfven's drift result, and emphasises that, in contrast
to the cases of E X Hand g X ~ drifts , the magnetic gradient drift vel ocity, of direction H X V H for a
positively-charged particle, can now only be found by
means of an approximat e theory. While this r emark is probably true for rather general magnetic fiel ds, it i s neverthel ess possi ble to obtain an exact solution
for the drift vel ocity of a charged particle in a magnetic field having constant gradient perpendicular
to the direction of the main fi eld (Seymour
1
9
5
9
).
This exact solution, which is developed in Section 5,yields as approximations Alfv~n's result and the case
of circular or bit, and includes the case where the
magnetic field vanishes at points in the particle's
orbit , for which perturbation illethods are inappropriate.
5. Exact Solution for Charged Particle Drift ~~~agnetic Field of Constant Gradient.
(a) Derivation of Basic Equations.
Consider f irst the motion of an electron, and suppose the i.nagnetic field is in the z-dir ection and
is represented by Hz
=
"'A x. Then, from the non-relativistic equation of motion we have, in the absence of el ectric field,
e
ill = vXH ( 5 .l)
d t c
Ch
.1-
5
10is - e/c (Spitzer
1956,
p.l).The Z-coillponent of v is constant, and need not
be considered explicitly, The other components vary
as
V2 == v2 +
X
d Vx e )
ill == A x Vy )
d t c )
)
(
5
.2)
)d Vy e )
ill _ _
== + A x Vx )
dt c )
)
is constant, and in the notation of
Fig. 2, Vx == v cos W, v y == v sin W •
these relations in
(5
.
2)
givesSubstitution of
d W e A x
=
( 5 . J)d t ill C
Since v == d s/ d t, where s denotes distance along
the electron trajectory,
dlf eAX
==
-d S ill C V
TI1us
e A clx eA
- - - -
= - - -- cos If ,d S ill C V
and so
(::)
2 e A== sin W + constant. (5.6) ill c v
From (5.4) and ( 5.6)
;t
Jx~
2 ill C VX == + sin W
,
(5.7)e A
where x
Ch.1 11
H::::.o
_ _ O~ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ~ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ~j
Ma..jne..-t"'G
+ieLL
uFwa..rcLsX,.-
-
- -
HI
X1.- -
~-- ~-- ~-- ~-- ~-- ~-- ~-- ~-- H
I 0D.j
- - -
-
-
--
- -
-
- - H
1.
x - - -o
-:O"lrec. t,"OV"l...-
of
d..l"ift v
-x
Ch.1-5 12
For x positive,
(
5
.
])
givesm c d'¥
( 5 .8)
t
-e t.. 2 2 ill C V
Xo + sin '¥
e t..
Also , since
dy
sin '¥ ( 5 .9)
=
ds( 5 • 4) gives
m c v sin '¥ d '¥
(5.10)
y=
e A. x2
0 +
2 m c v sin '¥
e A.
(b) Comparison with Alfven' s Result .
For motion which does not cross the line H
=
0,as in Fig. 2, (5.10) gives the y-drift per cycle
exactly as 271:
sin'¥ d '¥ (5.11)
o
where Ho == t.. x o ' and ~o== m c v i s the orbitys
eHo
radius of curvature for field strength H o'
When Po « xo' which implies electron orbital
motion far from the y-axis, a first-or~er result for
the drift velocity in the y-direction can be readily
obtained. After expansion of the denominator in
(5.11),
integration gives approxlinately!:. Y
=
2
71: Po ==
2
7C t.. Po
Ch,l-5
13
Similarly, from (5.8) the periodic time in this
case is
r
r
=
2 7C Po and so the corresponding driftv
velocity in the y-direction is given by
6 y
T
=
(5.12)
This is the result obtained by Alfven, and
discussed by Spitzer and Post (p.354).
When Ho has a fixed value and A approaches
zero, we obtain the circular orbit result for a
homogeneous magnetic field, i.e. va
2 7C Po 27C e
T = = - -
,
where w c =v Wc m
called cyclotron angular frequency.
(c) Exact Solution in Terms of
Elliptic Integrals.
= 0 and
Ho i s the so-c
To complete the exact solution, two cases require
consideration:
Case 1. Electron does not enter Regiqn of
Reversed Magnetic Field.
For an electron motion which does not cross the
line H
=
0, as shown in Fig. 2, the limits of xar e, from (5.7)
2 m c v
e A
for '¥ = 3 7C /2,
(5.13 )
for 'If = 7C/2 ,
so that
°
~ x < x < X1 0 2
Using the sUbstitution '¥
=
7C _ 2¢,
we 0 btainCh.1
-
5
14
from
(5
.
10)
the drift per cycle in the y-direction as'lC
2m c
J
(1
= e H2 v
J
1 o!::'y - 2 sin
2
¢)
d¢
, 2 . 2 r / .
- K Sln '{J 1
where A x2 = H2 and k~ = .L~~:
::
~,
if P;2 ise A x2 x2
the orbit' s radius of curvature
r::
:
'
Jd:s:r:n
g
:hJx~2~~
,
\iV'hen x
1 ~ 0,
(5
.
13)
givesso that the upper l imit of is unity.
Reduct ion of
(
5
.1
4)
to standard form for ellipticintegrals results in
t
~
/
2
7C/2
2
J
1-
k~
• 2¢
d¢-(2
-
k~)J
d ¢l1y=x Sln
2
J1
-k~
. 2¢
Sl!la
a
(5
.
15)
where K and E are complete elliptic integrals of the
first and second kind respectively, of modulus k1
(Dwi ght
1957)
.
The periodic time is derived from
( 5
.
8)
as'lC/
24 ill C
f
J
1
d
¢
k2
1
(
5
.
1
6
)
T =
=
x -2 - K ·e H2 v
-
k1 2 sin2 ¢The exact drift velocity i s, therefore, from
(5
.
15)
and
(
5
.
16)
l1y
_V(
~(1
-
E)
-)
(5
.
17)
v --_.-a - =
T k2 Ie
Ch.1-5 15
Since(~
for the driftk2
1
velocity is always in the negative y-direction.
2
v1;2
k «1, Alfven's result va =
-1 8
lifhen is
again obtained.
l~lfhen k~ = 1 , which occurs when x1 == 0, we have
va = - v, and the electron moves along the line H == O.
For x < 0, the electron drift pattern is precisely the mirror b"Uage in the y-axis of that shown in Fig. 2.
Again the drift is always in the negative y-direction.
Case 2, Electron enters Region of Reversed
Magnetic Field.
Consider now a motion in which the electron crosses
the line H
=
0, and tllUS enters a reversed magneticfield, as shown in Fig.
3.
when x
=
0 and Vx > O.Let ¥o be the value of ¥ Then in place of
(5.7)
v .
(sln ¥ - sin ¥o )
and so the limits of x are now
~
J
2 In C v (1 - sin ¥o ) for ¥= 71./2. (5.19)e f..
In this case various drift patterns may occur, as
shown in Fig. 4, but each pattern possesses s~llinetry about the ,V-axis, on which H = O.
¥
o =
Introducing as before W = 71. 2
2
2 rI.
'P o' From Figs. 3 and 4,
2
¢,
then71. 71.
< ¥ «
+-2 " 0 " 2
0<00 < 71. and
the above limits thus reduce to
,
"
"
2x
Ch.1
H==-o
x
Mo...Jne.tic..
fie.Lc:L.
d..owYlWo..rd....s:Dire.c.tio\'"\
of
eLY-'l f tv
-
- -
x
~
16
Ch.1-5
17
Utilising the properties of symmetry exhibited by
the drift pattern of, say, Fig. 3, then for positive x
in
(5.1
8
)
we obtain from(
5
.4)
and(5.
9
)
/:;y= AC
=
2 ... \B - 2 TIl C V sin '1' d '1'e f...
'1'0
J
2
ill C v (sin 'l' - sin 'l'ole t..
0
00
0
cos2
\1
d0
-
~
h Sin2d0
o
-
sin20
0
0
-0
0
If another variable of integration G is defined by
sin
¢
=
sin0
0 sin 8,(5.21)
may be transformed to:::
8
~
e A.
where k2 sin
0
0=
=
x2
Similarly
T
=
4(
E
-
~
K)
2
F
4
ill C V 1~
J~
Kusing
and the drift velocity in this case is
E
(2 _ - 1 ) K
( 5 • 20) •
(5.21)
Ch.1-5
E k2
When k2 «1, ~ 1
-
2 andK ~ 2
18
( 1 2
Va ~ V
-k2 ) N V
N (5.25)
as expected from Fig. 3.
For ElK
==~,
corresponding to0
0~
650, thedrift ve10ci ty becomes zero. As ¢o increases
beyond 65 0 (ElK < ~) the drift veloci ty becomes
negative, as in Case 1. Fig.
4
gives typical driftpa tterns for ElK < ~, ElK == ~ and ElK > ~. Since in
this Case va becomes + v and - v for k2 == 0
and k2 == 1 respectively, it follows that at the limit
¢
=
0 the electron moves along the line H == 0 ino
the positive y-direction, while at the limit ~o== ~/2
it moves along the same line in the negative y-direction,
Thus the drift patterns of Cases 1 and 2 coincide in
the limits k == 1
1 and k2 == 1 •
For positively-charged particles, the principal
results are that the formulae for k1 and k2 relilain
unchanged, if the particle charge is + elc, whereas
the va are changed only in sign.
Since dx
ds
=
cos '¥ ,it can be readily
ascertained that the drift velocity in the x-direction
is zero in all cases.
(d) Dis cussion of Results.
The drift velocity results obtained in Cases 1
and 2 for an electron are shown plotted against the
in Fig.
5.
This [image:38.571.12.554.14.734.2]x
E
I- <
K
2.,M
~ n~t,·e,..
fl·e.LL
cLow'I\wof"J..sA
rrovvs sho""cL i ,.. e.c
t
"0 If\. ofFart icLe cLrift
-E _
I---"v
" E I
->"2-
K
- -
-K
2..M
0-5
VI e.tic
.f
I'e.L
d.- ()
F
\IV Gl. r"' c::L S
"""-- """--
-F
i
9
.
'
+_-,
j
tic
a.L
E
L
e
c.t
ro
n. ] ) (' ;f
t
P
a.
t
t-
e
r
V)s
i
VL.-re
r
VV' S 0f
V VIfor Ca..se
~.
J
I
E
lK
(")
~
f-'
f-'
Ch.l 20
~
--d
- - - - c
0 d
~ ~ V) Q) V> d
C>
U
u;
~
0
L\->
0
(.
>
~ ~ ~r<-eV \I) d U 0
("r) .-C
+l
.
-$
>
()~
N ::> Y-OC
0.-N
-t.J
CV
<n d
d
U i..
.t.
~
lJ) lI) It)
~ ~
0 r- It, N
...,
r- 00 0 0 0 0 0 0
-
.
lr)I
,
ICh.1-6 21
the region of Case 1. For 1 /k1'~ 00 , we have the
well-known region of Alfven drift velocities. In spite
of the smallness of Alfven drift velocities r elative
to particle velocities, it has been suggested that
charge separation effects could be obtained in a plasma ,
and that these might lead to motion of the plasma
towards the region of weakest magnetic field (Post,
p. 354). Larger drift velocities, of the order of the
particle velocities, are expected, however, in the
region x
J
eX2 4 m c v < 1 •
Thus if the magnetic
field within a plasma varies in a direction normal
to the field, and somewhere changes sign, the motion
of the charged particles in the neighbourhood where
it changes sign may lead to larger charge separation
effects than in the Alfv~n region.
Recalling the briGf comments given in Section 3
on the physical mechanism of instability in a simple
plasma geometry, we see that if charge separation is
produced in a plasma by non-uniformity of the
magnetic field, instability could similarly be
initiated or au@aented. In fact , if larger charge
separation effects than produced by the Alfven drift
are encountered, it is likely that the instability
growth rate will be increasod.
6. References.
Albares, D. J. , Krall, N. A., Oxley, C. L. (1961).
-Rayleigh-Taylor Instability in a Stabilized
Linear Pinch Tube. General Atomic Report
Ch.1-6
Alfv~n, H. (1940) . - On the Motion of a Charged
Particle in a Magnetic Field. Ark. Mat. Astr.
Fys. 27A, No. 22: 1-20.
Alfven,
n
.
(1950). - (ICosmical Electrodynamics."pp. 13-23. (Clarendon Press: Oxford.)
22
Allis, vv. P. (1956). - "Handbuch der l)hysik.1i pp. 38
4-392, Vol. 21. Edited by S. Flugge. (Springer:
Berlin. )
Brueckner, K. A., ~ivatson, K. M. (1956). - Use of the
Boltzmann Equation for the Study of Ionized Gases
of Low Density. II. Phys. Rev. 102 : 19-27.
Chandrasekhar, S., Kaufman, A. N., vl/atson, K. M.
(1957). - Properties of an Ionized Gas of Low
Density in a Magnetic Field. III. Ann. Phys. 2:
435-470.
Chandrasekhar, S., Trehan, S. K. (1960). - "Plasma
Physics. I? pp. 14-36. ('rhe University of Chicago
Press. )
Chapman, S., Cowling, T. G. (1953). - "The Ivlathematica1
Theory of Non-uniform Gases." pp. 107-133, 134-150,
319-358. (Cambridge University Press: London.)
Chew, G. F., Goldberger, M. L., Low, F. E. (1956).
-The Boltzmann Equation and the One-fluid Hyd
ro-magnetic Equations in the Absence of Particle
Collisions. Proc. Royal Soc. A, 236: 112-118.
Cowling, T. G. (1957). - !?I'Jlagnetohydrodynamics.1! p. 3.
(Interscience Publishers: New York and London.)
Delcroix, J . L. (1960). - "Introduction to the Theory
of Ionized Gases. 11 pp. 42-56. (Interscience
cnol-6 23
Dwight, Ho B. (1957). - "Tables of Integrals and Other
:Mathematical Data.11 pp. 272-274. (The Macmillan
Company: New York.)
Francis, G. F. (1960). - :'Ioniza tion Phenomena in
Gases .l i pp. 253-254. (Butterworths Scientific
Publications: London.)
Geneva Papers. (1958). - "Proceedings of the Second
United Nations International Conference on the
Peaceful Uses of Atomic Energy.1I P/2300 (pp.
99
-Ill), p/1307 (pp. 134-136), P/365 (pp. 137-143),
P/349 (pp. 144-150), P/2214 (pp. 151-156), Vol.31.
(United Nations Publication: Geneva.)
Green, H. S. (1959). - Ionic Theory of Plasmas and
Magnetohydrodynamics. Phys. Fluids 2: 341-349.
Kruskal, M., Schwarzs child, M. (1954). - Some
Instabilities of a Completely Ionized Plasma.
Proc. Royal Soc. A, ,223: 348-354.
Lamb, H. (1945). - lIHydrodynamics.!1 pp. 370-371.
(Dover Publications: New York.)
Linhart, J . G. (1960) . - ItPlasma Physics.?? pp. 6-53.
(North-Holland Publishing Company: Amsterdam.)
McRea, W. H. (1954). - l?Relativity Physics. it See
particularly pp. 50-51, Ch. VI. (Methuen and
Company: London.)
Post, R. F. (1956) . - Controlled Fusion Research
-An Application of the Physics of High
Tempera ture Plasmas. Rev . .lvlod. Phys. 28:
353-355.
Rosenbluth, M. N. , Longmire, C. L. (1957).
-Stability of Plasmas Confined by Magnetic
Ch.1-6 24
Rosenfeld, L. (1951). - "Theory of Electrons.1i
pp.
13-27. (No~th-Holland Publishing Company:
Ams t erdam. )
Seymour, P. W. (1959) . - Drift of a Charged Particle
in a Magnetic Field of Constant Gradient. Aust.
J. Phys. 12: 309-314.
Simon, A. (1959) . - IIAn Introduction to Thermonuclear
Research.?? pp. 24-26. (Pergamon Press: London.)
Spitzer, L., Jr. (1952). - Equations of Motion for an
Ideal Plasrr-a. Astrophys. J. 116: 299-316.
Spi tzor, L" Jr 0 (1956). - l1Physi cs of Fully Ionized
Gases,n pp. 1-7. (Interscience Publishers:
New York and London.)
Stratton, J. A. (1941). - l1Electromagnetic Theory."
pp. 78-81. (McGraw-Hill Book Company: New York
Chapter 2
ESTU1ATION OF THE MAXJJVIUM TEMPERATURE IN A RADIALLY CONSTRICTED PLASMA BETWEEN ELECTRODES
1. Introduction.
In this chapter the flows of heat and electricity
in a steady non-equilibrium state gas discharge between
a pair of electrodes are examined, with the principal
object of determining theoretically the maximum plasma
temperature Tm' For simplicity, the treatment is
confined to discharges of highly-ionized gases having
low bremsstrahlung loss and free boundary surfaces
perfectly insulated thermally by a high vacuwn. Non-constricted and Non-constricted discharges are treated.
2. Application of the Saha Equation.
The Saha equation (Saha
1920,
Saha and Saha1934,
Allis
1956,
Cobine1
9
58)
applies to a gaseous system inthermal equilibrium, v:hereas we shall consider sys tems which depart from this condition by virtue of an excess
of the electron temperature over the ion temperature, as discussed by Allis in the above reference, and
treated in greater detail by Alfven
(1950).
However,as Allis points out, the temperature can never become
less than that given by the Saha equation, and
accordingly we apply it here to obtain some idea of
the minimum. temperatures at which a gas may be regarded as substantially ionized as the particle density varies
over an appropriate range. Thus, applying Cobine's form of the Saha equation to a plasma, having equal concentrations of singly ionized atoms and electrons
(ne = n i ) we obtain
2
5,050
V.log (_(1_)
=
1.5 log T - (N- 15.385) _.
l ,(2.1)
1-(1 T
Ch.2-J
26
where, if na atoms/cm~ is the original concentration
of the gas,
Q ~ ni/na is the degree of ionization,
T is the gas temperature in degrees Kelvin,
N ~ log n a ,
v· is the ionization potential of the gas in volts,
l
Here our main interest will be in the hydrogen
isotope, deuterium, and so the temperature dependence
of Q for hydrogen has been calculated from equation
(2.1) for N-values varying from 14 to 20, and presented
in Fig. 1.
For substantially-ionized hydrogen gas (Q - 0.9 say)
the minimum temperature for a given value of N can now
be readily obtained. It is of interest to note from
each of these curves that this minimum temperature lies
well below the temperature corresponding to the
ionization potential of the gas concerned, particularly
for the smaller values of N.
J.
Prelim~narz Di.scussion of thELQurrent Density and Heat Flux Vectors.In the absence of an external magnetic field H,
the flows of electricity and heat in a system mutually
interfere, and as Callen (1948) has pointed out, the
thermoelectric effects of Peltier and Seebeck may be
viewed as the result of this interference. Further,
if an external magnetic field is impressed on the
system, thermomagnetic and galvanomagnetic effects
(Ettingshausen, Hall, Nernst and Righi-Leduc effects)
appear. To account for these effects, we shall now
follow the procedure adopted by Marshall in his
\·0
0·9
o·s
0·7
cC 0·61
://;/
:z
0·;
0'4
0·3
0·2,
0·\
o
104I
0'\I.
2,..104 3-xI04- 4)(10'"
h~ o..tow\~ /c.t'f\~
(
Vl"=
ION)n
~ 5'''~\'J ~Ol'\i~e.cl ()..'corn'>/c.~~
_/ ni. j r . . L '
' " " - - L h Q . ) o~,re(... 01' lOnl,C).t'IOI'\
T}
~o..S kMpU-e"lUI'l-- ~ dC-jree.s Ke.lvI·~.5~104- '"./0'" 7)(.104 8}( /04- 9,clo4 /05
T°k. ~
Fill
Sa.ha... Curves -For
HJcLroJ-en...-.
o P'
N
Ch.2-3
anisotropic by the presence of a magnetic field, write
the conduction current density vector, j, and the
heat flux vector, q, as
j==OI~" + OIIE.L + OIIIh~.L + ¢I (9T)lt +0'11 (9T}.L +¢IIIhx( 9T).L,
\ I \ \ ---J
No special Hall Seebeck Nernst
28
name
().l) where
and
E is the electric field,
Ell is the component of E parallel to H,
E.L is the component of E perpendicular to H,
h = H/H is the unit vector in the direction
of H,
VT is the temperature gradient,
( VT)II is the component of 9T parallel to H
-'
(VT)J. is the component of VT perpendicular
to H,
cj I, cj II, and (J III are coeffi cients of
electrical conductivity,
¢ I, ¢ II, and ¢III are thermal diffusion
coefficients,
q=_r...I(9T)"_r...II(9T).L_rtIIhx(9rr).L+1 jll+~I j.L+ '1'IIIhxj J. ,
.
-
-().2)
where
jll is the component of j parallel to H,
ji is the component of j perpendicular to H,
f...I, f...II and r...III are thermal conductivity
coefficients, '1'1, 'I'll, and '1'111 are coefficients
giving the heat flux due to electric currents.
Ch.2-J 29
q = _KI (VT)" _ KII(VT).L _KIII£x (VT) .L+-f:IEIl + r;IIE.L+ t;IIIhxE.L ,
-
\.
-
-
I \-
I \ -1 \ INo special name Righi-Leduc Peltier
Ettings-hausen
( J. J)
where are thermal conductivity
coefficients when electric currents are allowed to
flow, given by
KI =
,,.1 _
\][1 ¢I ,KII= All _ \][11 ¢II + \][111 ¢III ,
KIII = AlII _ \][11 ¢III _ \][111 ¢II
and r;I, r;II, and r;III are coefficients accounting
for the contribution to the heat flux from the electric
field E, given by
1;;1 = \][1 0 I ,
r;II \][11
6'
II _ ,¥III ~III(J .5)
=
o
,
t;III = \][11
0
111 + rII (j IIThe results (J.4) and (J.5) are not given explicitly
by Marshall: they can alternatively be obtained from
results given in Part J of his report (pp. 28-29) by
neglecting the electron mass me relative to the ion
mass mi' and by noting that the K's here correspond
to Marshall's 8's.
Further information on the terms representing the
effects discovered by Ettingshausen, Hall, Nernst,
Peltier, Righi and Leduc, and Seebeck in the expressions
(J.l) and (J.J) for j and q is provided by Chapman
and Cowling (195J), Hix and Alley (1958), Linhart (1960)
and the above article by Callen.
An attempt to base the followin~ analysis on