Classical Electrodynamics (Jackson) (3rd Ed. 1999)

Vector Formulas
a· (b x c) = b · (c x a) = c ·(a x b)
a x (b x c) = (a· c)b - (a· b)c
r.,. v h 1 • r~ v 111 - r
., '. "Vh • rll (,. • riVh ...1q •
vX
VI/!= 0
V · (V X a) = 0
.... ,..., ·' _,
..,. ...,,..., .,,\ u2
' (fa)
v. =a. v.p + .pv. a
'
V x (l{la) VI{! X a -t- lf!V x a
v "~a """UJ \""
_,,
""Vju
/L
\" ',.
...,,
.. ,.,..., ._,
"!
•.
'
,..., " ,\
~,
V ·(a X b) = b • (V a) - a· (V X b)
X
V X (a X b) a(V · b) b(V ·a) + (b · V)a (a· V)b

If x is the coordinate of a point with respect to some origin, with magnitude
r = jxj, n = x!r is a unit radial vector, and f(r) is a well-behaved function of r,
then
V • X= 3 Vxx=O
.... '- r l .. \1 2 .c af ..., \l
L..J \' /J
{\
l"J \' /J r J
ar
f(r) at
(a· v )nJlr) r La n~a · DJJ -r n~a · DJ
ar
'afv
,- n\ " + vfV • "\ + ;(t I
'>l " \
'
1
where L i (x x v) IS me angmar-momemum operator.
Theorems from Vector Calculus
In the following </J, .p, and A are well behaved scalar or vector functions, V is a
three-dimensional volume with volume element d 3 x, S is a closed two-
dimensional surface bounding V, with area element da and unit outward normal
n a a.
Jv V · A d 3x = js A • n da (Divergence theorem)
L Vif! d 3x = L if!n da
r
.,
v X A d 3x = Tn
vS
X A da
( 2
(
Jv ((/)V ljf + V(/). vif) d 3 x- Js (/)n. Vljf da (Green's first Identity)
(
Jv
(rpr:z2r~,-
r ,,,V 2 3
r ¢) d JC - ( (¢'V•''- ·"'V¢) . u da
Js r r
(Green's theorem)
In/the following S is an open surface and C is the contour bounding it, with line
element dl. The normal n to Sis defined by the right-hand-screw rule in relation
to the sense of the lme mtegral around C.
Js
( (V X A) • n da - t. c
A • dl (Stokes's theorem)
€1assical
Electrodynamics
€1assical
Electrodynamics
Third Edition
John David Jackson

Professor Emeritus of Physics,
I Iniversity of California, Berkeley
JOHN WILEY & SUNS, INC.

This book was set in 10 on 12 Times Ten by UG and printed and hound by Hamilton Printing
Company.
This book is printed on acid~free paper. (OOi
The paper in this book was manufactured by a mill whose forest management programs include
sustained yield harvesting of its timberlands Sns1ained yield hanresting principles ensure that the
numbers of trees cut each year clgos ngt e*ceod tho ammmt of new grmvtll.
Copyright© 1999 John David Jackson. All nghts reserved.
No part oi this publication may be reproduced, stored in a retrieval system or transmitted

in any form or by any means, electronic, mechanical, photocopying, recording, scanning
or otherwi.se, except as permitted under Sections 107 or 108 of the 1976 United States
Copyright Act, without either the prior written permission of the Publisher, or
authorization through payment of the appropriate per-copy fee to the Copyright
Clearance Center, 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax
(978) 750-4470. Requests to the Publisher for permission should be addressed to the
Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030,
(201) 748-6011, fax (201) 748-6008, E-Mail: PERMREQ@WILEY.COM.
To order books or for customer service please, call I (800)-CALL-WILEY (225-5945).
Library of Congress Cataloging-in-Publication Data
Jackson, John David, 1925-

Classical electrodynamics I John David .Jackson.-3rd ed.
97-46873
Printed in the United States of America

10 9
To the memory of my father,
Walter David Jackson
Preface
n nas oeen -'O years smce me appearance 01 Lne HfSL emuon 01 LIIIS oOOK, anu LJ
years since the second. Such intervals may be appropriate for a subject whose
Ud>I> Wd> , "_'"Y ''"' tllt;Vl' iJ'T Y""" a,;u uy
Maxwell and experimentally 110 years ago by Hertz. Still, there are changes in
rl ,1: ••: -'T'l-' ·'-· ·" "' . rlrl • -·· .• ~,
any significant increase in. size. Inevitably, some t~pics present in the second
~rJ;t;A h~rJ fA h~ p];m;n >f• ,rJ fA m~]..-p rAAm- fAr nPm ;~] {')np m~;Ar Am;OO;AM
- '-
is the chapter on plasma physics, although some pieces appear elsewhere. Read-
ers who miss · •l:.r · mav. I hone. be able to avail th ,Jves of the
~' .
The most visible change is the use of SI units in the first 10 chapters. Gaussian
;,, ~ . ... l .. t. -\, ,], ;,, ~
relativity and relativistic electrody,namics than SI. As a reminder of the sys-

t.em of units h.eimr -' ~ th.e nmnimr helle! on eHch left-hHnd nHP"e ·
"-SI" or "-G" depending on the chapter.
M v tardv adontion of the universallY accented SI svstem is a reco11:nition that
almost all undergraduate physics texts, as well as engineering books at all levels,
employ SI units throughout. For many years Ed Purcell and I had a pact to
support each other in the use of Gaussian units. Now I have betrayed him! AI-
though this book is formally dedicated to the memory of my father, I dedicate
this third edition informally to the memory of Edward Mills Purcell ( 1912-1997 ),
a marvelous physicist with deep understanding, a great teacher, and a wonderful
man.
Because of the increasing use of personal computers to supplement analytical
work or to attack problems not amenable to analytic solution, I have included
some new sections on the principles of some numerical techniques for electro-
statics and magnetostatics, as well as some elementary problems. Instructors may
use tnetr mgenmty to create more cnauengmg ones. 1 ne a1m 1s to prov10e an
understanding of such methods before blindly using canned software or even
m~,,~,n~dcu VI ··r
There has been some rearrangement of topics-Faraday's law and quasi-
O<UU~
' c.u.
H~<UO U<~ HV W
f"L
~HUtH~< "'
J
"<L
•au '.' .~ U HW<~
I.
. ''
discussion of energy and inductances. Another major change is the consolidation
of the discussion of radiation by charge-current sources, in both elementary and
. .
exact mUILtpo,e 10nus, m '-'uapter "· rtu "'" LV >t;dll a11u uu-
fraction are in Chapter 10.
HI<- p1 VI V]JU'-<ll 'auu ·'"'- ~av<- 0 uiuvo aiv 1 ""~"
new sections in Chapter 8. In Chapter 13 the treatment of energy loss has been
,], ~ rl •L rl Q, ,f th ,f -h
·r -,
Iron raaiaiion as a researcn IOOI, me aiscussion in \.-napLer 1'+ nas oeen aug-
mPntPc1 hu ~ c1Pb;]p,j op,-t;An An thP .L · nf · <lnc1 nnc1nhtnr< for sun.
.L • .L ~L ' ' . -' . f"L ~~ ~:
. U
0 Ht OVU<~~O. <H~<~ >0 H~" lW> U> ::-uupt~> '"
>V VH" '<
and models of classical char11:ed narticles as well as chan11:ed emphasis.

_._ ,11 th ,],
' ' T ],
. "
th
'] 'r
.-L
"'
vii
viii Preface
not notice, or will notice only greater clarity. To mention but a few minor addi-
rions. esLimaLing seu-inuucLances, l:'oyntmg s theorem m lossy matenals, polar-
ization potentials (Hertz vectors), Goos-Hi.inchen effect, attenuation in optical
<:~ T rl. " -' ·' •
' y- -r IH ' - r >U'-<VJO. /"l.HU wwv I·"' S, 01
course! Over 110 new problems, a 40% increase, all aimed at educating, not
r1 ior-cmro crin n
In p;ep;~ing this new edition and making corrections, I have benefited from
ouestions SUQ:Q:eStions. cri . . onrl o.-hrirf> frnm momr otn.-IP.nto rnll ~nrl
newfound friends. I am in debt to all. Particular thanks for help in vari';;us ways
go to Myron Bander, David F. Bartlett, Robert N. Cahn, John Cooper, John L.
GommPI novi.-1 I G iffitho T Pr~" T TT T v· 1'1 ~K v -"
' -.J • ' ·~o ' ' . '
Michael A Lee, Harry J. Lipkin, William Mendoza, Gerald A Miller, William
A. Newcomb Ivan Otero Alan M. Porti< Frit7 Rohrlirh M " _,
Chris Schmid, Kevin E. Schmidt, and George H. Trilling.
.. • .., . . . JU
Berkeley, California, 1998, 2001

Preface to the Second Edition
In the thirteen years since the appearance of the fiist edition, my interest in
classical electromagnetism has waxed and waned, but never fallen to zero. The
subject is ever fresh. There are always important new applications and examples.
The present edition reflects two efforts on my part: the refinement and improve-
ment of material already jn the first edjtjon; the addition of new topics (and the
omission of a few).
The major purposes and emphasis are still the same, but there are extensive
changes and additions. A major augmentation is the "Introduction and Survey"
at the beginning. Topics such as the present experimental limits on the mass of
the photon and the status of linear superposition are treated there The aim is to
provide a survey of those basics that are often assumed to be well known when
one writes down the Maxwell equations and begins to solve specific examples.
Other major changes in the first half of the book include a new treatment of the
derivation of the equations of macroscopic electromagnetism from the micro-
scopic description; a discussion of symmetry properties of mechanical and elec-
tromagnetic quantities; sections on magnetic monopoles and the quantization
condition of Dirac; Stokes's polarization parameters; a unified discussion of the
frequency dispersion characteristics of dielectrics, conductors, and plasmas; a dis-
cusston of causahty and the Kramers-Kromg d1sperswn relaflons; a simplified,
but still extensive, version of the classic Sommerfeld-Brillouin problem of the
arrival of a signal in a dispersive medium (recently verified expeiimentally), an
unusual example of a resonant cavity; the normal-mode expansion of an arbitrary
field in a wave guide; and related discussions of sources in a guide or cavity and
the transmtsswn and reffectwn coefficients of Hat obstacles m wave gmdes.
Chapter 9, on simple radiating systems and diffraction, has been enlarged to
include scattering at long wavelengths (the blue sky, for example) and the optical
theorem. The sections on scalar and vectorial diffraction have been improved.
Chaplets 11 and 12, on specialrelathity, have been rewritten almost eom-
pletely. The old pseudo-Euclidean metric with x 4 = ict has been replaced by
g"" (with g 00 -tl, g" 1, i - 1, 2, 3). The change of metric necessitated a
complete revision and thus permitted substitution of modern experiments and
concerns about the experimental basis of the special theory for the time-honored
aberration of starlight and the Michelson-Morley experiment. Other aspects
have been modernized, too. The extensive treatment of relativistic kinematics of
the first edition has been relegated to the problems. In its stead is a discussion
of the Lagrangian for the electromagnetic fields, the canonical and symmetric
stress-energy tensor, and the Proca I agrangian for massive photons
Significant alterations in the remaining chapters include a new section on
transition radiation, a completely revised (and much more satisfactory) semi-
classical treatment of radiation emitted in collisions that stresses momentum
transfer instead of impact parameter, and a better derivation of the coupling of
maltipole fields to their sources The collection of formulas and page references
to special functions on the front and back flyleaves is a much requested addition.
Of the 278 problems, 117 (more than 40 per cent) are new.
IX
A Clla"~" tu " ' "
..
The one area that remains ·" comnlPtPiv nnrl ' i< thP. r.h,ntf'r on
magnetohydrodynamics and plasma physics. I regret this. But the book obviously
has grown tremendously, and there are avmlable many books devoted exclusively
+ .,: ••• •+ ,j' ,], .I. _, ... •
"J r . "J "J .
Of minor note IS the change from Maxwell's eouations and a Green's func-
tion to the Maxwell equations and a Green function. The latter boggles some
. . . . . .
, uu1, u1 !Ly W11u 01uer usage ~oesse1 runcuon, ror example). 1t IS
still Green's theorem, however, because that's whose theorem it is.
Work on this edition began in earnest during the first half of 1970 on the
occasion of a sabbatical leave spent at Clare Hall and the Cavendish Laboratory
in Cambridge. I am grateful to the University of California for the leave and
indebted toN. F. Mott for welcoming me as a visitor to the Cavendish Laboratory
and to R. J. Eden and A B. Pippard for my appointment as a Visiting Fellow of
Clare Hall. Tangible and intangible evidence at the Cavendish of Maxwell, Ray-
leigh and Thomson provided inspiration for my task; the stimulation of everyday
activities there provided necessary diversion.
Thi< nP.w <"clition h'" • ·" ·" from m '< ,c
;Anc 'tc ~nrl
criticism from many students, colleagues, and strangers. Among those to whom
I owe some specific debt of gratitude are A. M. Bincer, L. S. Brown, R. W. Brown,
~ n f"' ~ n TT ~ ~ r ·
~. ~. , "· u. , ~. , "'-· J. L ld!\L, V. L. rllCll, lVl. D. r ·r ru,
A. Hobson J.P. Hurlev. D. L. Judd L. T. Kerth E. Marx M. Nauenh,.ro- A R
Pippard, A M. Portis, R. K. Sachs, W. M. Saslow, R. Schleif, V. L. Telegdi, T.
1reaon, t . r. uyon, v. t<. w eJssKopr, ana vudley Williams. bspecJally helpful
mPrP n n Rm·' , R 1\.T r~ hn , T tt n. . , T. ., J.r " •r . .I
, " T7 r. .
ham, E. M. Purcell, and E. H. Wichmann. I send mv thanks and fraternal greet-
ings to all of these people, to the other readers who have written to me, and the
cuuuut"' '" wuu u<tve MI wi111 we pr; s~sometimes written
askimr for solutions to he fii<n'ltched I. .c. s.om.e_fip,fiJi.nPII To mv min.-1 thP
book is better than ever. May each reader benefit and enjoy!
J. D. Jackson
Berkeley, California, 1974
Dvn-fFJ,-.n +r.. +1/1 n v;vc-.+ c;' ;-/;+; ro."J/1
.L I l[....f L..f-'...-1[.... t-V t-1 l-1[.... .L t-1 Ll 1- ~Ut-I-t-V I 1-
v
,
;uc; llltoUly, lU)!,ClllCl Wllll <tHU 41
forms the core of present-day theoretical training for undergraduate and grad-
uate physicists. A thorough grounding in these subjects is a requirement for more
aavancea or specJanzea trammg.
Typically the undergraduate program in electricity and magnetism involves
two or perhaps three semesters beyond elementary physics, with the emphasis
on the fundament~! l~ws l~horMorv verific~tion ~nd el~borMion of their con-
. 1. .1. -'-
_, -l." crl.
••
0~'"1' , ~H~~H UHU>JOW, < OUO~ pu~H~H<~HU, UH~" , <''~ '"U<H~"
matical tools utilized include vector calculus, ordinary differential equations with
constant coefficients, Fourier series, and perhaps Fourier or Laplace transforms,
partial differential equations, Legendre polynomials, and Bessel functions.
As a general rule, a two-semester course in electromagnetic theory is given
to beginning graduate students. It is for such a course that my book is designed.
My aim in teaching a graduate course in electromagnetism is at least threefold.
Th~ firct ~;~ " t.-. th 1- " ,]-.. •t ,.1. .L ;tl-
emphasis on the uni;y of electric and ma~~etic phenomena, both in their physical
basis and in the mode of mathematical description. The second, concurrent aim
is to develop and utilize a number of topics in mathematical physics which are
usetul m both electromagnetic theory and wave mechamcs. These mclude
Green's theorems and Green's functions, orthonormal expansions, spherical har-
monies, cylindrical and spherical Bessel functions. A third and perhaps most
· t>mt nurno"'· ;, thP •htion of nPw m"tPri"l i"ll" on thP intPr-
•. of re'1 atlvJstlc
actwn : . . ch arge•d partie
. les w1t . ft e ld s. In th"1s last area
. h e Iectromagnet1c
personal preferences and prejudices enter strongly. My choice of topics is gov-
erned by what I feel is important and useful for students interested in theoretical
pnys1cs, expenmema1 nuClear anu rngn-energy pnys1cs, anu Lnm as ye1w-uenneu
field of plasma physics.
The book begins in the traditional manner with electrostatics. The first six
chanters are devoted to the develooment of Maxwell's theorv of electroma!!ne-
... l\,f, ,)., .+tl.
'J •• .1
-rr
. t . ..l .1
·o
tJ.,
'J•
especially in Chapter 2 and 3, where boundary-value problems are discussed
thoroughly. The treatment is initially in terms of the electric field E and the
..
v, Wllll lllC UCllVIOU '1' , 1./ <lllU H , 111ll u-
duced by suitable averaging over ensembles of atoms or molecules. In the dis-
cussion of dielectrics, simple classical models for atomic polarizability are de-
scribed, but for magnetic materials no such attempt to made. Partly this omission
"'"' " '"'. .-.f '"""" , h,.t trHlu ,.].,,,;,..,] -' •lc .-.f m<>·
·o
. " ;h;];tu <>r~
~J ~J
llUl , . l"Ull , 101Ul01U<tl1Ull Ul ll!C lllliOll Ul 110110
r
magnetism needs almost a book in itself.

.
CT'l.
"~ "~"'
'1.
"" ~~
,).,
·r '" \
/'"7 (l\
mostly ot a macroscomc sort. Plane waves m different media mcludmg plasmas

/ J
:11.
"'~OHUW •uu~~O
.1
,..~
-· '
as well as dispersion and the propagation of pulses, are treated in Chapter 7. The
discussion of wave guides and cavities in Chapter 8 is developed for systems of
arhitrarv cross section ~ncl the of IMion in P"llicles ~ncl the n of
xi
xii Preface to the First Edition
a cavity ate haudled in a vety genetal way which emphasizes the physical pro-
cesses involved. The elementary theory of multipole radiation from a localized
source and diffraction occupy Chapter 9. Since the simple scalar theory of dif
fraction is covered in many optics textbooks, as well as undergraduate books on
electricity and magnetism, I have presented an improved, although still approx
imate, theory of diffraction based on vector rather than scalar Green's theorems.
The subject of magnetohydrodynamics and plasmas receives increasingly
more attention from physicists and astrophysicists. Chapter 10 represents a sur-
vey of this complex field with an introduction to the main physical ideas involved.
The first nine or ten chapters constitute the basic material of classical elec-
tricity and magnetism. /', graduate student in physics may be expected to have
been exposed to much of this material, perhaps at a somewhat lower level, as an
undergraduate But he obtains a more mature view of it, understands it more
deeply, and gains a considerable technical ability in analytic methods of solution
when he studies the subject at the level of this book. He is then prepared to go
on to more advanced topics. The advanced topics presented here are predomi-
nantly those involving the interaction of charged particles with each other and
with electromagnetic fields, especially when moving relativistically.
The special theory of relativity had its migins in classical electrodynamics.
And even after almost 60 years, classical electrodynamics still impresses and de-
lights as a beautiful example of the covariance of physical laws under Lorentz
transformations. The special theory of relativity is discussed in Chapter 11, where
all the necessary formal apparatus is developed, various kinematic consequences
are explored, and the covariance of electrodynamics is established. The next
chapter is devoted to relativistic particle kinematics and dynamics. Although the
dynamics of charged particles in electromagnetic fields can properly be consid-
ered electrodynamics, the teadet may wonder whether such things as kinematic
transformations of collision problems can. My reply is that these examples occur
naturally once one has established the four vector character of a pmtide's rna
mentum and energy, that they serve as useful practice in manipulating Lorentz
transformations, and that the end results are valuable and often hard to find
elsewhere.
Chapter 13 on collisions between charged particles emphasizes energy loss
and scattering and develops concepts of use in later chapters. Here for the first
time in the book I use semiclassical arguments based on the uncertamty pnnctple
to obtain approximate quantum-mechanical expressions for energy loss, etc.,
from the classical results. This approach, so fmitful in the hands of Niels Boln
and E. J. Williams, allows one to see clearly how and when quantum-mechanical
effects enter to modify classical considerations.
The important subject of emission of radiation by accelerated point charges
is discussed in detail in Chapters 14 and 15 Relativistic effects are stressed, and
expressions for the frequency and angular dependence of the emitted radiation
are developed in sufficient generality for all applications. The examples treated
range from synchrotron radiation to bremsstrahlung and radiative beta processes.
Cherenkov radiation and the 'Neizsacket Vv'illiams method of virtual quanta are
also discussed. In the atomtc and nuclear collision processes semiclassical argu-
ments are again employed to obtain approximate quantum mechanical results. I
lay considerable stress on this point because I feel that it is important for the
student to see that radiative effects such as bremsstrahlung are almo~t entirely
Preface to the First Edition xiii
classical in nature. even though involving smalJ-scale col!jsions A student who

meets bremsstrahlung for the first time as an example of a calculation in quantum
field theory will not understand its physical basis.
Multipole fields form the subject matter of Chapter 16. The expansion of
scalar and vector fields in spherical waves is developed from first principles with
no 1estiictions as to the 1 elative dimensions of somce and wavelength. Then the
properties of electric and magnetic multipole radiation fields are considered.
Once the connection to the multiple moments of the source has been made,
()Xampl()s of atomic and nucl()ar multipol() radiation ar() discuss()d, as well as a
macroscopic source whose dimensions are comparable to a wavelength. The scat-
tering of a plane electromagnetic wave by a spherical object is treated in some
detail in order to illustrate a boundary-value problem with vector spherical
waves.
In the last chapter the difficult problem of radiative reaction is discussed.
The treatment is physical, rather than mathematical, with the emphasis on delim-
iting the areas where approximate radiative corrections are adequate and on
finding where and why existing theories fail. The original Abraham-Lorentz the-
ory of the self-force is presented, as well as more recent classical considerations.
The book ends with an appendix on units and dimensions and a bibliography.
In the appendix I have attempted to show the logical steps involved in setting up
a system of units, without haranguing the reader as to the obvious virtues of my
chmce of umts. I have provided two tables which I hope will be useful, one for
converting equations and symbols and the other for converting a given quantity
of som()thing from so many Gaussian units to so many mks units, and vice versa.
The bibliography lists books which I think the reader may find pertinent and
useful for reference or additional study. These books are referred to by author's
name in the reading lists at the end of each chapter.
This book is the outgrowth of a graduate course in classical electrodynamics
which I have taught off and on ove1 the past eleven years, at both the University
of Illinois and McGill University. I wish to thank my colleagues and students at
both institutions for countless helpful remarks and discussions Special mention
must be made of Professor P. R. Wallace of McGill, who gave me the opportunity
and encouragement to teach what was then a rather unorthodox course in elec-
tromagnetism, and Professors H. W. Wyld and G. Ascoli of Illinois, who have
been particularly free with many helpful suggestions on the treatment of various
topics. My thanks are also extended to Dr. A. N. Kaufman for reading and com-
mentmg on a prehmmary versiOn of the manuscnpt, and to Mr. G. L. Kane for
his zealous help in preparing the index.
J. D. Jackson
Urbana, Illinois, January, 1962
Contents
Introduction and Survey 1

1.1 Maxwell Equations in Vacuum, Fields, and Sources 2
1.2 Inverse Square Law, or the Mass of the Photon 5
I.3 Linear Superposition 9
1.4 Maxwell Equations in Macroscopic Media 13
1.5 Boundary Conditions at Interfaces Between Different Media 16
1.6 Some Remarks on Idealizations in Electromagnetism 19
References and Suggested Reading 22
Chapter 1/ Introduction to Electrostatics 24

1.1 Coulomb's Law 24
1.2 Electric Field 24
1.3 Gauss's Law 27
1.4 Differential Form of Gauss's Law 28
1.5 Another Equation of Electrostatics and the Scalar Potential 29
1.6 Surface Distributions of Charges and Dipoles and Discontinuities
in the Electric Field and Potential 31
1.7 Poisson and Laplace Equations 34
1.8 Green's Theorem 35
1.9 Uniqueness of the Solution with Dirichlet or Neumann Boundary
Conditions 37
1.10 Formal Solution of Electrostatic Boundary-Value Problem
with Green Function 38
1.11 Electrostatic Potential Energy and Energy Density; Capacitance 40
1.12 Variational Approach to the Solution of the Laplace and Poisson
Equations 43
1.13 Relaxation Method for Two-Dimensional Electrostatic Problems 47
Problems 50
Chapter 2/ Boundary- Value Problems in Electrostatics: I 57

2.1 Method of Images 57
2.2 Point Charge in the Presence of a Grounded Conducting Sphere 58
2.3 Point Charge in the Presence of a Charged, Insulated, Conducting
Sphere 60
2.4 Point Charge Near a Conducting Sphere at Fixed Potential 61
2.5 Conducting Sphere in a Uniform Electric Field by Method
of Images 62
2.6 Green Function for the Sphere; General Solution
for the Potential 64
2.7 Conducting Sphere with Hemispheres at Different Potentials 65
XV
xvi Contents
2.8 Orthogonal Functions and Expansions 67

.
')(\ n .~\!. ',)...1 .1 ,1 TI . ~ _,
u
"t' .
'
Coordmates 70
7111 ,1. Tmn_n;mPn<;nnol ~ ,t;ol TJ, ,J...l . "'· ..•.
'
of Fourier Series 72
? 11 Fitekls ;c~nrl l h~HYP n .. ·in Two-ni~~~";nnol
and Along Edges 75
2.12 Introduction to Finite Element Analvsis for F' "'·~~., 79
Problems 85
Chapter 3 I Boundary- Value Problems in Electrostatics: II 95

1
_, ...,
' • n-'T ~
J.~ --r ~'iuauvu"' "1-'"'-'H'-'"' ---r.J
3.2 Legendre Equation and Legendre Polynomials 96

'2'2 D ,r1 \!nl .D."'J...l "<<A" "'-,-<n
~ cry ry ,_v,_
""
3.4 Behavior of Fields in a Conical Hole or Near a Sharp Point 104
1 <; ,1. ~ T P<TPndrP p, ,< ond th • "'1
Y,,.(e. q,) l07

3.6 Addition Theorem for SnhP.rir~l H~rmnn;r< 11 (l
3.7 Laplace Equation in Cylindrical Coordinates; Bessel Functions 111
3.8 Boundarv-Value Problems in Cvlindrical Coordinates 117
3.9 Expansion of Green Functions in Spherical Coordinates 119
3.10 Solution of Potential Problems with the Spherical Green Function
Expansion 112
~ . . . .
J.~~ -r ut u1coc11 r 111 "J neal Loormnaws JL.)
3.12 Eigenfunction Expansions for Green Functions 127
1 , 1
-. 'K: _, Tl. _, "- _,, • "- -'- ~- .. - ~·
-
Hole 129 - , ~vuu• I ld11C W ll11 d ~11 C,Uldl
n -•· no -/ (", ,. -1."

'J
'""'"
,/ 1).,
u ""'
-
Problems 135
Chapter 4 I Multipoles, Electrostatics of Macroscopic Media,

uietecrncs 145
4.1 Multinole Exnansion 14S
4.2 Multipole Expansion of the Energy of a Charge Distribution
in an External Field 150
4.3 Elementary Treatment of Electrostatics with Ponderable Media 151
'+.'+ nounuary- v ame rro01ems wiLn uie1ecrrics 1)4
4.5 Molecular Polarizability and Electric Susceptibility 159
;."L
' r 'K o p ~ • • • •
... v
4.7 El lVl. E ''. D' I ''_Y M

ectrostatJc nergy m 1e ectnc e 1a 165
D r
~}
.J ('
~66 ~~.
_J D _J·
~
.
1 rn
Problems 169
r" -r
c; I l4. -~- .~: Jl, J.
'.- T nw n • ~-~·:~ D:nlrf~
,..,_,
' • '=·
5.1 Introduction and Definitions 174
).L l:Siot and :savart 1 aw 1/)
Contents xvii
5.3 Differential Equations of Magnetostatics and Ampere's Law 178

5.4 Vector Potential 180
5.5 Vector Potential and Magnetic Induction for a Circular Current
5.6 Magnetic Fields of a Localized Current Distribution, Magnetic

Moment 184
5.7 Force and Torgue on and Energy of a Localized Current Distribution
in an External Magnetic Induction 188
5.8 Macroscopic Equations, Boundary Conditions on B and II 191
5.9 Methods of Solving Boundary-Value Problems
in Magnetostaties 194
5.10 Uniformly Magnetized Sphere 198
5.11 Magnetized Sphere in an External Field; Permanent Magnets 199
5.12 Magnetic Shielding, Spherical Shell of Permeable Material
in a Uniform Field 201
5.13 Effect of a Circular Hole in a Perfectly Conducting Plane
with an Asymptotically Uniform Tangential Magnetic Field
on One Side 203
5.14 Numerical Methods for Two-Dimensional Magnetic Fields 206
5.15 Faraday's Law of Induction 208
5.16 Energy m the Magnetic Field 212
5.17 Energy and Self- and Mutual Inductances 215
5.18 Quasi Static Magnetic Fields in Conductors, Eddy Cnnents, Magnetic

Problems 225
Chapter 6/ Maxwell Equations, Macroscopic Electromagnetism,

Conservation Laws 237
61 Maxwell's Displacement Current; Maxwell Equations 237
6.2 Vector and Scalar Potentials 239
6.3 Gauge l'ransformatwns, Lorenz Gauge, Coulomb Gauge 240
6.4 Green Functions for the Wave Equation 243
6.5 Retarded Solutwns for the Fields: Jefimenko"s Generalizations
of the Coulomb and Biot Savart I aws: Heaviside Feynman
Expressions for Fields of Point Charge 246
6.6 Derivation of the Equations of Macroscopic Electromagnetism 248
6.7 Poynting's Theorem and Conservation of Energy and Momentum
for a System of Charged Parllcles and Electromagnetlc Frelds 258
6.8 Poynting's Theorem in Linear Dissipative Media with Losses 262
6.9 Poynting's Theorem for Harmonic Fields; Field Definitions
of Impedance and Admittance 264
6.10 Transformation Properties of Electromagnetic Fields and Sources
Under Rotations, Spatial Reflections, and Time Reversal 267
6.11 On the Question of Mag11etic Monopoles 273
6.12 Drscusswn of the Drrac QuantrzatJon Condrtron 275
6.13 Polarization Potentials (Hertz Vllctors) 280
Problems 283
xviii Contents
Chapter 7/ Plane Electromagnetic Waves and Wave Propagation 295

7 1 Dl UJ. "-T. _, ~ -1:
K
""'
7.2 Linear and Circular Polariza;i~; Stokes Param:~;rs 299
7.3 Reflection and Refraction of Electromagnetic Waves at a Plane
. .
1meu"'"" otoLwtotoH 1 wo uteJecrncs .1UL
7.4 Polarization by Reflection, Total Internal Reflection; Goos-Hiinchen
Effect 306
7.5 Frequency Dispersion Characteristics of Dielectrics Cond
and Plasmas 309
7.6 Simplified Model of Propagation in the Ionosphere
~nrl M~ ~1 •rP -,,(;
7.7 Magnet~hydr~dynamic Waves 319

7.8 Superposition of Waves in One Dimension; Group Velocity 322
Til • c <L
"" .J
·~ . . ~1 '""'
n
'-'}'I'
•
VI d r UI>t;; ,._., H r I'
•
•llidL, "1ve
JVleamm jLtJ
7.10 Causality in the Connection Between D and E; Kramers-Kronig
Relations 330
7.11 Arrival of a Signal After Propagation Through a Dispersive
Medium 335
References and Su ·~
, . J(7
iN
n .u ?A/l
1 'vuoc " " J7V
f'l lil I U!. . -1. n ,, r. .. nnrl fl ,,,- rl v:J. '210")

• u
' ·r
8.1 Fields at the Surface of and Within a Conductor 352

8.2 Cylindrical Cavities and Waveguides 356
8.3 Waveguides 359
8.4 Modes in a Rectangular Waveguide 361
R.'i " ·Flow >~nrl A H ,,· in "'· ~ rl .
~ '
u.v • C.I CUI vauvu VI
,
LUIIUillUll'
,
v
.JUU
'"'
8.7 Resonant Cavities 368
8.8 Power Losses in a Cavity; Q of a Cavity 371
{).':} tann anu wnospnere as a Kesonant Lav1ty:
Schumann Resonances 374
8.10 Multimode Propagation in Optical Fibers 378
R. 11 . in Die! ' w~uPmoidPO ':!R<;
8.12 Expansion in Normal Modes; Fields Generated by a Localized

Source in a Hollow Metallic Guide :liN
-1 (', ,.
·- ':':'bOLoHCU·~
D on~
'1 'b J7J
Problems 396
Chapter 9 I Radiating Systems, Multipole Fields and Radiation 407

9.1 Fields and Radiation of a Localized Oscillating Source 407
0? 1= . T . Fi olrlc o~rl D orl;,,,;n A H1
9.3 Magnetic Dipole and Electric Quadrupole Fields 413

9.4 Center-Fed Linear AntPnn~ 411'>
C\C ~ K .1. , .1. n.
/,J lUI
~no
1 .:>VUII.-C VI 1"\.}'CllUIC
m Wavegm'd e 419
lJ.n Snheric~l Solutions of the Scalar Wave Eouation 425
n ~ 'K. _' ~- £ ,,_ ~' • ~·.'" ,.-,n
-'· •vn.. " ' v• "''-' ~''-''-'" .. w .. • H.-<uo ~~-'
9.tl Properties of Multipole Fields Energy and Angular Momentum
~-~ Angular Uistnbution ot Mult!po!e KaOlation 45/

9.10 Sources of Multipole Radiation; Multipole Moments 439
9.11 Multipole Radiation in Atoms and Nuclei 442
9.12 Multipole Radiation from a Linear, Center-Fed Antenna 444
Problems 449
Chapter 10 I Scattering and Diffraction 456

10.1 Scattering at Long Wavelengths 456
10.2 Perturbation Theory of Scattering, Rayleigh's Explanation
of the Blue Sky, Scattering by Gases and Liquids, Attenuation
in Optical Fibers 462
10.3 Spherical Wave Expansion of a Vector Plane Wave 471
10.4 Scattering of Electromagnetic Waves by a Sphere 473
10.5 Scalar Diffraction Theory 478
10.6 Vector Equivalents of the Kirchhoff Integral 482
10.7 Vectorial Diffraction Theory 485
10.8 Babinet's Principle of Complementary Screens 488
10.9 Diffraction by a Circular Aperture; Remarks on Small
Apertures 490
10.10 Scattering in the Short-Wavelength Limit 495
10.11 Optical Theorem and Related Matters 500
Problems 507
Chapter 11/ Special Theory of Relativity 514

11.1 The Situation Before 1900, Einstein's Two Postulates 515
11.2 Some Recent Experiments 518
11.3 Lorentz Transformations and Basic Kinematic Results of Special
Relativity 524
11.4 Addition of Velocities; 4-Velocity 530
11.5 Relativistic Momentum and Energy of a Particle 533
11.6 Mathematical Properties of the Space-Time of Special
Relativity 539
11.7 Matrix Representation of Lorentz Transformations, Infinitesimal
Generators 543
11.8 Thomas Precession 548
11.9 Invariance of Electric Charge; Covariance of Electrodynamics 553
11.10 Transformation of Electromagnetic Fields 558
11.11 Relativistic Equation of Motion for Spin in Uniform or Slowly Varying
External Fields 561
11.12 Note on Notation and Units in Relativistic Kinematics 565
Problems 568
XX Contents
Chapter 12 I Dynamics of Relativistic Particles

~ .rJ V 1.
~-~~-· ~ ·~··
"
~ ........,
D." .·I ..1.
12.1 Lagrangian and Hamiltonian for a Relativistic Charged Particle

m External Electromagnetic Fields 579
Fl ') ~A. ' ' , TT. ' 0 C' • ~< • ~· • • ~o~
, ""~ L .JO.J
12.3 Motion in Combined, Uniform, Static Electric and Magnetic

FiP]rl, ">RI>
12.4 Particle Drifts in Nonuniform, Static Magnetic Fields 588
12.5 Adiabatic Invariance of Flux Throw:>:h r lrhil nf P~rtid~e 'iQ?
12.6 Lowest Order Relativistic Corrections to the Lagrangian for Interacting
Charged Particles: The Darwin Lagrangian 596
12.7 Lagrangian for the Electromagnetic Field 598
12.8 Proca Lagrangian; Photon Mass Effects 600
12.9 Effective "Photon" Mass in Superconductivity; London Penetration
Depth 603
lL.lU Lanomcal ana :symmetnc :Stress 1 ensors; Conservation Laws 605
12.11 Solution of the Wave Equation in Covariant Form; Invariant Green
> UJ.L

..I2 ,_j,l .:77
Chapter 13 I Collisions, Energy Loss, and Scattering of Charged Particles,

rh.nPn ,,,~ nnrl 'r. • • D -1;,.~;~" L"Jd
~-~
13.1 Energy Transfer in Coulomb Collision Between Heavy Incident Particle

ana t<ree l::,Jectron; bnergy Loss m Hard Collisions 625
13.2 Energy Loss from Soft Collisions; Total Energy Loss 627
'~ ~
. . .
LJ,J "y . l1l ~Ullb!Onat energy LOSS O.:ll
13.4 Cherenkov Radiation 637
1 '1 <: Dl • o • c • ~
L~.~ ..,""""u"1S v• , "" '"'""'1S"'u , "' """'" uy fiLum~

• U'+ll
13.6 Mean Square Angle of Scatteril}g: Angular Distribution of Mul~
Scattering 643
13.7 Transition Radiation 646
Problems 655
£".. .. ~ I Jn .l.• • • • I o ,.
.... "'• _..~ 1 ·~·· uy II'.Iuvutg \-tturges 00.1
14.1 Lienard-Wiechert Potentials and Fields for a Point Charge 661

14.2 Total Power Radiated by an Accelerated Charge: Larmor's Formula
and Its Relativistic Generalization 665
14.3 Angular Distribution of Radiation Emitted by an Accelerated
Charge 668
14.4 KaQlatwn bmittecl by a Charge m Arbitrary, Extremely Relativistic
l\lf, . t:'71
'"'·-' · · JuLion in rrequency anu fingw 01 l::,nergy KaciJatea

14.6 Frequency Spectrum of Radiation Emitted by a Relativistic Charged
Particle in Instantaneously Circular Motion 676
14.7 Undulators and Wigglers for Synchrotron Light Sources 683
14.8 Thomson Scattering of Radiation 694
Problems 698
Chapter 15 I Bremsstrahlung, Method of Virtual Quanta,

Radiative Beta Processes 708
15.1 Radiation Emitted During Collisions 709
15.2 Bremsstrahlung in Coulomb Collisions 714
I 53 Screening Effects; Relativistic Radiative Energy I ass 721
1.5. "+4 .. k et ,vvx
,we1zsac
.. . ... ,,.xants •we
f tlro d uf ""
vn t ua I Q
A uan ta "'""'
, L<-+
15.5 Bremsstrahlung as the Scattering of Virtual Quanta 729
15.6 Radiation Emitted During Beta Decay 730
15.7 Radiation Emitted During Orbital Electron Capture: Disappearance
of Charge and Magnetic Moment 732
Problems 737
Chapter 16/ Radiation Damping, Classical Models

of Charged Particles 745
16.1 Introductory Considerations 745
16.2 Radiative Reaction Force from Conservation of Energy 747
16.3 Abraham Lorentz Evaluation of the Self Force 750
16.4 Relaflvisflc Covanance; Stability and Pomcare Stresses 755
16.5 Covariant Definitions of Electromagnetic Energy
and Momentum 757
16 6 Covariant Stable Charged Particle 759
16.7 Level Breadth and Level Shift of a Radiating Oscillator 763
16.8 Scattering and Absorption of Radiation by an Oscillator 766
Problems 769
Appendix on Units and Dimensions 775

1 Units and Dimensions, Basic Units and Derived Units 775
2 Electromagnetic Units and Equations 777
3 Various Systems of Electromagnetic Units 779
4 ConversiOn of EquatiOns and Amounts Between SI Umts
and Gaussian I Inits 782
Introduction and Survey
Although amber and lodestone were known to the ancient Greeks, electro-
dynamics developed as a quantitative subject in less than a hundred years.
Cavendish's remarkable experiments in electrostatics were done from 1771 to
1773. Coulomb's monumental researches began to be published in 1785. This
marked the beginning of quantitative research in electricity and magnetism on a
worldwide scale. Fifty years later Faraday was studying the effects of time-varying
currents and magnetic fields. By 1864 Maxwell had published his famous paper
on a dynamical theory of the electromagnetic field. Twenty-four years later
(1888) Hertz published his discovery of transverse electromagnetic waves, which
propagated at the same speed as light, and placed Maxwell's theory on a firm
experimental footing.
The story of the development of our understanding of electricity and mag-
netism and of light is, of course, much longer and richer than the mention of a
few names from one century would indicate. For a detailed account of the fas-
cinating history, the reader should consult the authoritative volumes by
Whittaker.* A briefer account, with emphasis on optical phenomena, appears at
the beginning of Born and Wolf.
<;:;n thP 1Qfll), thPrP h~' hPPn ~ trnP rP.vnlntinn in nnr lmclPr<t~nclincr nf thP
U<"l'-' lVl <.,c;o dllU v1 111<111101. nvvv "
\177VO)
' C'lC'-'11 VU y lldiiiiC,O
rests in a sector of the umhed descnotwn ot oart1cles and mteractwns known as
'" "· " •L 1 'T'"C .1. •• .L.l
b'
' ,],
'1'
.1- . 'I
aescnpuon ot electromagnetic, weak, ana strong mteractwns based on tunda-

ol <-nnot;tw>ntc. ·""Qrlrc Qnrl · r ,,_ ;nrr ,;~ fnr<-o _nhn.
T ·o. r
lUll,, VV allU L LJU,UI", allU giUUI". lllt: tneureucai u; IUIK " gener-
ated throu!'h nrin, · of continuous (rP.:<llv nh~cp) in· of thP

~ ~' £
LV<~~O UHU U<O~<~O~ OJ v< pm "~'~ !-" vp~< u~o.
From the nomt of v1ew of the standard model. classical electrodvnam1cS IS a

]; ,j'
,. -'- L
' ,.Nr .11 ..1 ,, •<'<
• '
ana 1arge average numoers or v1nuaJ or rea1 pnotons J. vuamum eJecrroaynam1cs,
'n t wo ..., >>nn"o " j ' ~ cnAnt~nonnclu •tru ;n ~ thonru ;n
-: .. T ·o ·o ·-: '"
vv lllC,ll lllC WCaJ\. dllU C1CC,ll' llllCldU1UllO d1C dllU lllC
fnrctc · · nf hnth :<rf'. m:<"l"" ThP. <Vml hrP:<kino- ]P:<VPS thP. p]pctrn-
1VlC,C C,dlUCl qJ VVlUld
~- ' ' ; .. " .
0 ldVV Vl lllllllllC l_<lll];;C,
1)
while the weak torce earners acamre masses ot the order ot I:SU-'!U ve V /c With
.. r•. '1. ·'- ·• f'"> v 1 n· 18
L
0 ·' -o 'b' ,.
-\
'/'
tiecause or the ongms m a unwea rneory, rne range ana strength or rne weak
;ntPro<-tinn :<rP rPhtPrl tn thP ir rno
r ·o
fthP finp orP """-
1/LJ/).
""'" a
nccoc
"· . L • 1. •L ·" r .. n .
·o·
"" -~
.l
2 Introduction and Survey
Despite the presence of a rather large number of quantities that must be

taken from experiment, the standard model (together with general relativity at
large scales) provides a highly accurate description of nature in all its aspects,
from far inside the nucleus, to microelectronics, to tables and chairs, to the most
remote galaxy. Many of the phenomena are classical or explicable with nonrel-
ativistic quantum mechanics, of course, but the precision of the agreement of the
standard model with experiment in atomic and particle physics where relativistic
quantum mechanics rules is truly astounding. Classical mechanics and classical
electrodynamics served as progenitors of our current understanding, and still play
1mportant roles m practical life and at the research frontier.
This book is self-contained in that, though some mathematical background
(vector calculus, differennal equatwns) IS assumed, the subject of electrodynam-
ics is developed from its beginnings in electrostatics. Most readers are not coming
to the subject for the fi1 st time, however. The purpose of this mtroductwn IS
therefore not to set the stage for a discussion of Coulomb's law and other basics,
but rather to present a review and a snney of classical electromagnetism. Ques
tions such as the current accuracy of the inverse square law of force (mass of the
photon), the limits of validity of the principle of linear superposition, and the
effects of discreteness of charge and of energy differences are discussed. "Bread
and butter" topics snch as the boandary conditions for macroscopic fields at
surfaces between different media and at conductors are also treated. The aim is
to set classical electromagnetism in context, to indicate its domain of validity,
and to elucidate some of the idealizations that it contains. Some results from later
in the book and some nonclassical ideas are used in the course of the discussion
Certainly a reader beginning electromagnetism for the first time will not follow
all the arguments or see their significance. For others however this introduction
will serve as a springboard into the later parts of the book, beyond Chapter 5,
and will remind them of how the subject stands as an experimental science.
1.1 Maxwell Equations in Vacuum, Fields, and Sources

The equations governing electromagnetic phenomena are the Maxwell
equations,
V·D = p
aD
VxH--=J
at
aB
VxE+-=0
V·B = 0
where for external sources in vacuum, D = EoE and B ~-t 0 H. The first two
equations then become
V · E = piE0
Sect. 1.1 Maxwell Equations in Vacuum, Fields, and Sources 3
Implicit in the Maxwell equations is the continuity equation for charge density
and current density,
I h1s follows from combmmg the time denvaflve of the first equat10n m (I.la)
with the divergence of the second equation. Also essential for consideration of
charged particle motion is the Lorentz force equation,
F = q(E + v x B) (1.3)
which gives the force acting on a pomt charge q m the presence of electromag-
netic fields.
These equations have been written in SI units, the system of electromagnetic
units used in the first 10 chapters of this book. (Units and dimensions are dis-
cussed in the Appendix.) The Maxwell equations are displayed in the commoner
systems of units in Table 2 of the Appendix. Essential to electrodynamics is the
speed of light in vacuum, given in SI units by c = (fLoEo)- 112 . As discussed in the
Appendix, the meter is now defined in terms of the second (based on a hyperfine
transition in cesium-133) and the speed of light (c = 299 792 458 m/s, exactly).
These definitions assume that the speed of light is a universal constant, consistent
with evidence (see Section 11.2.C) indicating that to a high accuracy the speed
of light in vacuum is independent of frequency from very low frequencies to at
least v = 1024 Hz (4 GeV photons). For most practical purposes we can approx-
imate c = 3 X 108 m/s or to be considerably more accurate, c = 2.998 X 108 m/s.
The electric and magnetic fields E and Bin (1.1) were originally introduced
by means of the force equation (1.3). In Coulomb's experiments forces acting
between localized distributions of charge were observed. There it is found useful
(see Section 1.2) to introduce the electric field E as the force per unit charge.
Similarly, in Ampere's experiments the mutual forces of current-carrying loops
were studied (see Section 5.2). With the identification of NAqv as a cmrent in a
conductor of cross-sect10nal area A with N charge earners per umt volume mov-
ing at velocity v we see that Bin (I 3) is defined in magnitude as a force per unit
current. Although E and B thus first appear _just as convenient replacements for
fmces pwduced by distributions of charge and current, they have other important
aspects. First, their mtroduchon decouples conceptually the sources from the test
bodies experiencing electromagnetic forces If the fields E and B from two somce
distributions are the same at a given point in space, the force acting on a test
charge or cunent at that point will be the same, regardless of how different the
source d1stnbut10ns are. Th1s g1ves E and B m (1.3) meanmg m their own nght,
independent of the somces Second, electromagnetic fields can exist in regions
of space where there are no sources. They can carry energy, momentum, and
angular momentum and so have an existence totally independent of charges and
currents. In fact, though there are recurnng attempts to ehmmate exphc1t ref-
erence to the fields in favor of action-at-a-distance descriptions of the interaction
of charged particles, the concept of the electromagnetic field is one of the most
fruitful ideas of physics, both classically and quantum mechanically.
The concept of E and B as ordinary fields is a classical notion. It can be
thought of as the classical limit (limit of large quantum numbers) of a quantum-
mechanical description in terms of real or virtual photons. In the domain of
aspect of the electromagnetic field can usually be ignored or at least glossed over.
. .
field is of the order of 50 V/m and there are of the order of 1015 visible photons/
cm2 • s. Similar! an isotro i 8
produces an rms electric field of only 0.5 mV/m at a distance of 100 kilometers,
but this still corres onds to a flux of 1012 hotons/cm 2 ·s or abou 1 9 ·
3
a volume of 1 wavelength cubed (27m ) at that distance. Ordinarily an apparatus
wi no e sens1 e o e m lVI ua p otons; t e cumu atJve e ect o many
photons emitted or absorbed will appear as a continuous, macroscopically ob-
e
netic fields is adequate? Some sophistication is occasionally needed, but the fol-
. . . . . .
be taken as large but the momentum carried by an individual photon is small
compared to the momentum of the material system, then the response of the
material s stem can be determined ade uatel from a classical descri tion of the
electromagnetic fields. For example, each 108 Hz photon emitted by our FM
antenna gives it an impulse of only 2.2 X 10- 34 N •s. A classical treatment is sure!
adequate. Again, the scattering of light by a free electron is governed by the
classical Thomson formula (Section 14.8) at low frequencies, but by the laws of
the Compton effect as the momentum liw!c of the incident photon becomes sig-
nificant compared to me. The photoelectric effect is nonclassical for the matter
current can be calculated quantum mechanically for the electrons using a classical
The quantum nature of the electromagnetic fields must, on the other hand,
. .
other system that initially lacks photons and has only a small number of photons
. .
sica! terms, basically because of conservation of energy and momentum. An ex-
amp e IS t e classical treatment (Section 16.2) of the cascading of a charged
particle down through the orbits of an attractive potential. At high particle quan-
urn num ers, a c ass1ca escnpt10n of particle motion is adequate, and the sec-
ular changes in energy and angular momentum can be calculated classically from
continuous distributions in x, although we consider from time to time localized

distributions that can be ·
charges are assumed to be completely arbitrary, but are known to be restricted
in realit to discrete values. The basic unit of h r
charge on the electron,
Sect. 1.2 Inverse Square Law or the Mass of the Photon 5
where the errors in the last two decimal places are shown in parentheses. The
. .
are integral multiples of this basic unit.* The experimental accuracy with which
. .
part in 1020 ). The experiments are discussed in Section 11.9, where the question
of the Lorentz invariance of charge is also treated.
course, some e 1cate macroscopic or a most macroscopic expenments m w 1c

the discreteness of charge enters. Millikan's famous oil drop experiment is one.
His dro lets were t icall 10- 4 em in radius and had a few or few tens of ele-
mentary charges on them.
There is a lack of symmetry in the appearance of the source terms in the
Maxwell equations (I.la). The first two equations have sources; the second two
do not. This reflects the experimental absence of magnetic charges and currents.
Actuall as is shown in Section 6.11 articles could have rna netic as well as
electric charge. If all particles in nature had the same ratio of magnetic to electric
charge, the fields and sources could be redefined in such a way that the usual
. .
vention to say that no magnetic charges or currents exist. Throughout most of
· n rr n t in the axwell
equations, but some consequences of the existence of a particle with a different
magnetic to electric charge ratio, for example, a magnetic monopole, are de-
tively by Cavendish and Coulomb to be an inverse square law. Through Gauss's

law and the divergence theorem (see Sections 1.3 and 1.4) this leads to the first
formed over the years. It is now customary to quote the tests of the inverse square
law in one of two ways:
(b) Assume that the electrostatic potential has the "Yukawa" form (see Section
. . .
w IS some 1mes p r se 1 r .,..
experiments usually give t: and perhaps JL or m.,.; geomagnetic experiments
give JL or m.,..
*Quarks have charges %and - Va in these units, but are never (so far) seen individually.
Figure 1.1 Cavendish's apparatus for establishing the inverse square law of
make them the more perfect conductors of electricity." (Figures reproduced by

Sect. 1.2 Inverse Square Law or the Mass of the Photon 7
The original experiment with concentric spheres by Cavendish* in 1772 gave

an upper limit on E of IEl < 0.02. His apparatus is shown in Fig. 1.1: About 100
years later Maxwell performed a very similar experiment at Cambridget and set
an uppet limit of Iel "'5 X 10 '.Two othet notewmthy labotatOiy experiments
based on Gauss's law are those of Plimpton and Lawton,* which gave IEl < 2 X
10- 9 • and the recent one of Williams. Faller. and Hill.* A schematic drawing of
the apparatus of the latter experiment is shown in Fig. 1.2. Though not a static
experiment (v = 4 X 106 Hz), the basic idea is almost the same as Cavendish's.
He looked for a charge on the mner sphere after 1t had been brought mto elec-
trical contact with the charged outer sphere and then disconnected; he found
none. Williams, Faller, and Hill looked for a voltage difference between two
concentric shells when the outer one was subjected to an alternating voltage of
±10 kv with respect to ground. Ihe1r sensltivlty was such that a voltage differ-
ence of less than 10- 12 V could have been detected. Their null result, when
interpreted by mean~ of the Proca eqnations (Section 12 8), gives a limit of
16
E = (2.7 ± 3.1) X 10- •
Measurements of the earth's magnetic field, both on the surface and out from
the surface by satellite observation, permit the best direct limits to be set on E or
equivalently the photon rna~~ mY" The geophysical and also the laboratory ob-
servations are discussed in the reviews by Kobzarev and Okun' and by Goldhaber
and Nieto, listed at the end of this introduction. The surface measurements of
the earth's magnetic field give slightly the best value (see Problem 12.15), namely,
or
For comparison, the electron mass is m" = 9.1 X 10- 31 kg. The laboratory
experiment of Williams, Faller, and Hill can be interpreted as setting a limit
my < 1.6 X 10-so kg, only a factor of 4 poorer than the geomagnetic limit.
A rough IImlf on the photon mass can be set quite easily by noting the ex-
istence of very low frequency modes in the earth-ionosphere resonant cavity
(Schumann resonances, discussed in Section 8.9). The double Einstein relation,
hv = myc 2 , suggests that the photon mass must satisfy an inequality, my <
hv0 /c 2 , where v0 IS any electromagnetic resonant frequency. The lowest Schumann
resonance has v0 - 8 Hz. From this we calculate my < 6 X 10 50 kg, a very small
valne only one order of magnitude above the best limit While this argument has
crude validity, more careful consideration (see Section 12.8 and the references
given there) shows that the limit is roughly (R/H) 112 = 10 times larger, R = 6400
km being the radius of the earth, and H = 60 km being the height of the iono-
*H. Cavendish, Electncal Researches, ed. J. C. Maxwell, Cambndge Umverstty Press, Cambndge
(1879), pp. 104-113.
'Ibid., see note 19.
'S. J. Plimpton and W. E. Lawton, Phys. Rev. 50, 1066 (1936).
'E. R. Vfilliams, J. E. FalteJ, and H:. A. Hill, f'izys. Rev. Let!. 26, 721 (1971).
L Transmitter I
Crystal
and
oocur """ J oven
L buffer amp
I
rooase snmer reterence
U!Jj_
shifts ltnearly 360°
per l:2 hour A High Q coil
(IG\";:'·v."D'""' water cooled
I
l
I
Oscilloscope
Iff/~/ ' ~ \
/~
~\
I f M". •••crronics
Inside are battery
powered
"' \
1 MIXer 1--IAmj)>-- pass

I V filter ) )
~ c;;;,:t:l Lr--,. J
\
\
,..-
1
\
'I(-'
~ I 4 MHz
Voltage to I
J
IJ
'\,._ / ~ opti~~'j/ Calibration JA /

t Coppe;ons~ ~ J~~ ~ltor _g' ~--~~~minum
Calibration
,,.,.,
_j-- 1 ---
Scaler 50 Fourier analyze

sec. countingf---1 for signal with
cvclo
L <"
diode
Fi2ure 1.2 Schematic diagram of the "Cavendish" exoeriment of Williams Faller and
Hill. The concentric icosahedrons are conducting shells. A 4 MHz voltage of 10 kV
peak is applied between shells 5 and 4. Shell 4 and its contiguous shells 2 and 3 are
rougnty 1.:> meters m~ameter and contam shell 1 mside. The voltage difference
between shells 1 and 2 (if any) appears across the inductor indicated at about 8 o'clock
"'" ~11 1 crt. ~
· ~ .. ~ "t' 'J"<O>U auc IV ICAUaU uoc., ::"'
intormation tO tue ou,siue wona. 1 ney are eqmva1ent to L.avenmsn s system o1 stnngs
that automatically opened the hinged hemispheres and brought up the pith balls to test
fAr th . _.. ,.,; .,.. . . cL..J. •• '
·o ·r 'b r · r ... "·)
"·
~~~ ...
T ~ T'
~up~
.. 0
so here.* In soite of this dilution factor the limit of 10_ 4 , kg set by the mere
existence of Schumann resonances is quite respectable.
The laboratory and geophysical tests show that on length scales of order 10- 2
to 107 m, the inverse square law holds with extreme precision. At smaller dis-
tances we must turn to less direct evidence often involving additional assump-
tinn' For ""~mnlP R 11th . J•, h' · 11 ~n~lvsi' nf thP .£. nf ~lnh~
particles by thin foils substantiates the Coulomb law of force down to distances
of the order of 10- 13 m. orovided the aloha particle and the nucleus can be treated
as classical point charges interacting statically and the charge cloud of the elec-
trons can be ignored. All these assumptions can be, and have been, tested, of
h11t nnhr · · thP ~- ·'· nf thP ·'·'J' nf n11~nt11m mPrh~nirs
linear superposition (see below), and other (very reasonable) assumptions. At
still smaller distances relativistic auantum mechanics is necessarY. and strong
interaction effects enter to obscure the questions as well as the answers. Never-
theless, elastic scattering experiments with positive and negative electrons at cen-
t. ,f · ,f t. 1(111 r:.,u h thgt mgntnm .1 ..l. •
(the relativistic-theory;[point electrons interacting with massless photo;;-'s) holds

tn rli• nf thP nf 1n· 18 m WP · 1 nrlP th~t thP nhntnn m~'' r~n hP
taken to be zero (the inverse square force law holds) over the whole classical
ran2:e of distances and deen into the auantum domain as well. The inverse sauare
. 1. ..l . • .,_ 1
L
>U o W LU>V • U .V
L
UVL~
.1..1
V vL u•
1
LvUO.
""'
~~ VL
..l.
~VL 0 '
VL •u~v w cuv
v
.1 • 1
JvU<vo
1.3 Linear Superposition

The Maxwell equations in vacuum are linear in the fields E and B. This linearity
is exploited so often, for example, with hundreds of different telephone conver-
sations on a single microwave link, that it is taken for granted. There are, of
course, circumstances where nonlinear effects occur-in magnetic materials, in
crystals responding to intense laser beams, even in the devices used to put those
telephone conversations on and off the microwave beam. But here we are con-
cerned with fields in vacuum or the microscopic fields inside atoms and nuclei.
What evidence do we have to support the idea of linear superposition? At
the macroscopic level, all sorts of experiments test linear superposition at the
level of 0.1% accuracy-groups of charges and currents produce electric and
magnetic forces calculable by linear superposition, transformers perform as ex-
" "ino are ".on ission lines -the reader can make
a list. In optics, slit systems show diffraction patterns; x-ray diffraction tells us
about crystal structure· white light is refracted by a prism into the colors of the
rainbow and recombined into white light again. At the macroscopic and even at
the atomic level, linear superposition is remarkably valid.
Tt i< in thP <nl ;,. . th~t frnm linP~r .. r~n
be legitimately sought. As charged parti~les approach each oth~r ~ery closely,

electric field stren2:ths '- enormous. If we think of a chamed narticle as a
*The basic point is that, to the extent that HIR is negligible, the extremely low frequency (ELF)
i< the. ""me.. "" in " nl"te.. i«inn line.. in the. fnnrl"me.nt"l TFM mnrle. Thi<
propagation is unaffected by a finite photon mass, except through changes in the static capacitance
. . .. . 7
UHU · pv• uu" 'v"5"'· . . . . UOO ~HWCO VWUL HL VoUVL \L<OHJ 1"',
localized distribution of charge we see that its electromaJLnetic ener~>v

~nrll~ ·~ •r ~ t~ ~]., · ' '' ~' --1 T. · '
·o 'b' ~.... 'Y''"E>'va·•wu
infinite self-energies of point particles, it is natural to speculate that some sort of
saturation occurs, that field strengths have some upper bound. Such classical
nonlinear theories have been studied in the past. One well-known examole is
the of _Ram_ _and _lnfp].-f * ThP 1m i~ crivPn p~ m~~not;~
permeabilities,
-1/2
E
{T A\
I
E
'
where b is a maximum field strength. Equation (1.4) is actually a simplification
-' p:<rlie.r__hlLRom ~ l~no. Tt onfh~oo t~ ;11 t} ol ·-' . v: ~ U
<Ht: 'Y utOuiueu at suon msrances; au electromagnetic energres are hmte.

But such theories suffer from arbitrariness in the manner of how the nonlinearity
occurs and also from grave problems with a transition to a quantum theory.
Furthermore there is no . ' nf thio bnrl nf ,-.]"ooir'" 1 nnn linP"ritv ThP ~"~"-
-'~ ,±_ J . ' . . "1'
~• ·:~•~]- ~0 U<OO\.,HUc-U IV ~H/5l~ }'~' '":~
quantum theory with the interactions between nucleus and electrons and between
electrons and electrons given by a linear superposition of pairwise potentials (or
retarded relativistic interactions for fine effects). Field strengths of the order of
<>~>H0 0 H HH~ >»~ » · ll<OlU a1
21
the edge of a heavy nucleus is of the order of 10 V/m. Energy level differences
in light atoms like helium, calculated on the basis of linear superposition of elec-
"""'' ,--""" iu agtccmt:Ht wiw experiment tO accuracies tnat
approach 1 part in 10". And Coulomb energies of heavy nuclei are consistent
with linear superposition of electromagnetic effects. It is possible, of course, that
for field strengths greater than 1021 V/m nonlinear effects could occur. One place
1 ~ l • ~t . . ' . . .
·~ ·~ >~~" w > OUjJ<OHl<OUV J ~Lo - ~~Vj, UVUl 111 111<0 a
energy tevets anu m tne nuclear Coulomb energy. At the present time there
is no evidence for any classical nonlinear behavior of vacuum fields at short
distances.
i< " nunnturn. ' ,/; ,f ,1. ".1.1. •'-
: • . • .''J
a"''"' tHe rtarmy pnncrp1e permits me momentary creanon ot an
electron-positron pair by two photons and the subsequent disappearance of the
pair with the emission of two different photons, as indicated schematically in Fig.
1.3. This process is called th,; o~qttprinj) nf licrht hv licrht *S ThP twn incciriPnt _,
waves e'k 1·x-;,,, 1, and e'k2 ·x-iwzt do not merely add coherently, as expectedr with
linear superposition, but interact and (with small probability) transform into two
,t ~1 ;t\- 1. --1 1. 'T'L' .1.'
, co ~3 UH~ ~4· > »>0 •~ HUlC Ul
*M. Born and L. lnfeld, Proc. R. Soc. London A144, 425 (1934). See M. Born, Atomic Physics,
Blackie, London (1949), Appendix VI, for an elementary discussion.
1
An investigation of the effect of a Born Infeld tvue of nonlinearity on the atomic enPr~v levels in
sunerheavv elements h"s heen >rlP hu 1 RofPidri " ' n · '" 1 D "'--'~' , ·.,.
no , to< /00'70.,.
'When two of the photons in Fig. 1.3 are virtual photons representing interaction to second order
with a static nuclear Coulomb field, the process is known as Delbriick scattering. See Section 15.8 of
1 M Jm1ch "ncl F D · J'he_ ThonYu nf A ""' m, _12_ • " ' u
11 " " ' , ~ , ,
Sect. 1.3 Linear Superposition 11
f"k4
.Ita t
' .., ~
l
(
l.. e"':,./
\;7 \.e-
\
e _},_
e-
Figure 1.3 The scattering of light by light.

k2
Schematic diagram of the process by which
kl photon-photon scattering occurs.
quantum electrodynamics can be expressed, at least for slowly varying fields, in

terms of electric and magnetic permeability tensors of the vacuum:
D, = Eo 2:
k
E;kEk, B; = /-Lo 2:k /-LikHk
where
e~fi ' _,
Eik O;k -t- 4 7 LL~~-
-'
Clr)Oik -t-
'
I CD;DkJ -t ...
·~
~
{T <;'\
e~-fi
/-Lik O;k "'f "'" 4 71Z(CI5 1':JO;k ""T I ~i~kJ + ···
T>
CQ
J
•• '
'IS' \·
resurrs werenrsroorainecroynarer am:rKN.Xerin"T'Jj:J. we ooserve mat in me
/.' "· . -' . J -~ .. _,
-· 'T'L
classical limit (fi ~ 0), these nonlinear effects go to zero. Comparison with the
classical Born-Infeld expression (I.4) shows that for small nonlinearities, the
quantum-mechanical field strength
bq -_ y'45; ~ ea2 -~ 0.51 ea2

2 fie r 0 r0
plays a role analogous to the Born-Infeld parameter b. Here r 0 = eblmc2 =

2.8 X 10- 15 m is the classical electron radius and ealr5 = 1.8 X 1020 V/m is the
electric field at the surface of such a classical electron. Two comments in passing:
(a) the E;k and /-Lik in (I.S) are approximations that fail for field strengths ap-
nroachin~>: b or when the fields varv too raoidlv in soace or time ( fi!mc settin!!
~ ··~1 ·' ~n.--1 Mnu•2 nf t;m~l· fh'\ thP rh~nrP n11nwr;r~l rn;nr;_
' '' -~ 2.
'
-urvq~GT"-' o ,, VC
'UUlplUU<lUlJ HUl
·~
ll, >lll'-C '
Uq lll
'
YtancK s constant n.
In analai!V with the oolarization P -- D - EnE we sneak of the field-
' ,rlon+ ' ;M fT <;\ ~0 • I
"" Tn ~.irl;t;,-,n +n tho o~~•-
--:r: ~ ,. '-
'" "'15 u . lll5lll
'
u
-::
,.
y "IS"' U<
•.. 1 .
lU'-"-
~
·o• puuu
_,
''"''Y
smarr sm:rrs m atomic energy levels. The dommant contnDutJon mvolves a virtual
electron-positron pair, just as in Fig. 1.3, but with only two photon lines instead
*H. Euler and B. Kockel, Naturwissenschaften 23, 246 (1935).

12 Introduction and Survev
-~ ~- H<L -L .1 <L ., .' r .'

. cuv <' mv <vu<, '"'-', vVW lU lllC llld>> Ul LHC
and is decreed to vanish. If the photons are virtual, however, as in the electro-
L ,1, .. L ' • ' ~ ~ ~-
_..1
'5 ,v • •v•
any externally applied field, the creation and annihilation of a virtual electron-
positron pair from time to time causes observable effects.
Vacuum polarization is manifest by a modification of the electrostatic inter-
action between two charges at short distances, described as a screening of the
"bare" charges with distance, or in more modem terms as a "running" coupling
constant. Since the charge of a particle is defined as the strength of its electro-
magnetic coupling observed at large distances (equivalent to negligible momen-
tum transfers), the presence of a screening action by electron-positron pairs
closer to the charge implies that the "bare" charge observed at short distances
IS larger than the charge dehned at large distances. -uuantltatlvely, the lowest
' nn~ntn~ .1 ~ " ,]. ;-, tlo r' . .1 .L • .1 • L
·-. - - J · r • 5J •""
two charges Zte and Z 2e, corrected for vacuum polarization, is
V(r) = he
Z1Z2 a
r
2a
1 +-
31T
r 2m
dK
v KL -
K
2
4m
2
2m 2 )e-Kr
I1+ -
K2
(1.6)
where a is the fine structure constant ( = 1/137), m is the inverse Compton wave-
'.,,_,
1 maoo,
,.,._,
1 uy u 11 J_· u1c
.
_ 0 Ial, a >upct r
.. ..l Ul I UKawa
ootentials (e "'lr) is the one looo contribution nf ,]] thP · '' n:1ir' Tt inrrP"""'
the magnitude of the potential energy at distances of separation inside the elec-
tron Compton wavelength (hi me = aa 0 = 3.86 X 10- 13 m).
Because of its short range, the added vacuum polarization energy is unim-
portant in light atoms, except for very precise measurements. It is however
important in high Z atoms and in muonic atoms, where the heavier mass of the
muon (m" = 207 me) means that, even in the lightest muonic atoms, the Bohr
radius is well inside the range of the modified potential. X-ray measurements in
~· • "_L • ·~
-uwoo " • p• -a "'-''-'Ul dli:O V• Ul Llle V
polarization effect in (I.6).

'T'L "rl. -~ " "
v~ ,
_, . ~ vvuo.uu•,
.,_
lHaL 1~, au
rr
V '-' ~ll
.' Ul
r
interaction that changes with momentum transfer, is illustrated in electromag-

h" ovh;J..;t;n~ tho c. •~t." ,] p, -~
-J ·-o L
+L
'5J \ .v,.
IT ,-\
T-Tl n2\ 47TZ1 Z 2 a(Q 2 ) /T ~'

\~ J ~L I)
Q2
The 1/Q 2 dependence is characteristic of the Coulomb potential (familiar in

Rutherford scatterinp-) but now thP •• , ;, ~ h,; tho cn_r~llPrl rnnn;n/1
2
coupling constant a(Q ), the reciprocal of which is
W1 = ~"' - "
1
[a(Q
2
~\VJ J
lnl Tf5/3l
\ " ' t:
(I.8)
Here a(O)
l
0
10
= 11137. 036 ... is the fine structure constant, e is the base of natural
~nrl n2 ;<the> cnn~rp nf tho
~ -.
.L I .. .\
•
'T'L
expression (I.8) is an approximation for large Q 2/m 2. The running coupling a(Q 2 )
increases slowly with increasing QL (shorter distances); the particles are pene-
trating inside the cloud of screening electron-positron pairs and experiencing a
larger etfectJve product of charges.
Sect. 1.4 Maxwell E nations in Macrosco ic Media 13
Since the lowest order vacuum polarization energy is proportional to a times

the external char es, we describe it as a linear effect, even thou h it involves in
a) the square of the internal charge of the electron and positron. Small higher
order effects, such as in Fig. I.3 with three of the photons corresponding to the
. . .
the classical domain of sizes and attainable field strengths there is abundant ev-
idence for the validity of linear superposition and no evidence against it. In the
a om1c an su a om1c omam ere are sma quan um-mec amca non mear
effects whose origins are in the coupling between charged particles and the elec-
tromagnetic field. They modify the interactions between charged particles and
. . . .
absent.
quatwns m acroscop1c
thought of as equations giving the fields everywhere in space, provided all the
sources p and J are specified. For a small number of definite sources, determi-
. .
matter, the solution of the equations is almost impossible. There are two aspects
here. One is that the number of individual sources, the charged particles in every
atom and nucleus is rohibitivel lar e. The other as ect is that for macrosco ic
observations the detailed behavior of the fields, with their drastic variations in
space over atomic distances, is not relevant. What is relevant is the average of a
and macroscopic sources. It is shown in detail in Section 6.6 that the macroscopic
Maxwell equations are of the form (I.la) withE and B the averaged E and B of
simply multiples of E and B, respectively. The macroscopic field quantities D

and H, called the electric displacement and magnetic field (with B called the
. .
- c E
D a-~oa + P,.~ - L...
" aQ~f3 + ...
roscopically averaged electric dipole, magnetic dipole, and electric quadrupole,

and higher moment densities of the material medium in the presence of applied
and currents appear in the equations via P, M, and Q;,13 •

The macroscopic Maxwell equations (I.la) are a set of eight equations in-
~
.L'+
. mtroaucnon aua ;,urvey
equations can be solved formally by expressing E and B in terms of the scalar

potential and the vector potential A, but the inhomogeneous equations cannot
hP ~nhrPriuntil thP ' . "f;,Jrl. n ~nri H ~rp ''"~"'" ;n ~· ~ ~nrl " · 'rhaea
connections, which are implicit in (1.9), are known as constitutive relations,
J..l 1..1[1'.., DJ
H = H[E, B]
In addition, for conducting media there is the generalized Ohm's law,
J = J[E, B]
The square brackets signify that the connections are not necessarily simple and
m~u ~ ,r1 nn n~ot 1· · {~ 'o\ "'~" 1- -~-1· ''
.. , ·r r -, , .. , _, -J ' '
In most materials the electric quadrupole and higher terms in (1.9) are com-
nlPtP.lv nPuliuihlP. Onlv thP n' · nnd .,,;~ ' · · '" P ~nrl M ~ra e;n_
~ r -o
. > 111> UV<:O~ llVl are men , llUnvvv,, Ulill lllt:: ITilliVe"T
simple. There is tremendous diversity in the electric and magnetic properties of
0
.J'~
0
',ll -"' Ll. £ l ' '
' ~!' J "H ~"J~CUHOUv ' VVHH "'-H rn:;-m:rm ;u\.0 L!lit-
terials having nonzero P or M in the absence of applied fields, as well as more
.;,· . rl.'
;,·~1 .rl ' .,, ~·
, , ·~ ~ =.--.-m:o owuy VL " ' " ' "
properties is one of the provinces of solid-state physics. In this book we touch
nn lu uaru ' ' n ~nrl on -<: ; ~ 1l ~1 C' , lC rl
'J 'J 'J -.- 'J 'J T
vvv"~ >UI.011 it> ,.,uutu u<:: 1.0onsuueu tor a more sysremanc ana extensive
treatment of the electromagnetic properties of bulk matter.
T. _,_ r. '
.,_ .1.
LH VCH'-1 Vl lt::l. UWll , 1ur W<::itK <::nougn
L'-L.
fields the presence of an applied electric or magnetic field induces an electric or

' ' ' ,,· ' ,1 <L ' ' r •' '· ' P ' ' n •
!' !' 't' .v •av a>< ~ Ul lH<:O •U ll<:OlU, YY <:0
then say that the response of the medium is linear and write the Cartesian com-
~+ T\ ~nrl 1-1 ;n tha +,
' '*
D"' = L Eaf3Ef3
(I.lO)
Ha = L /L~f3Bf3
The tensors "'"' 13 and JL~ 13

are called the electric permittivity or dielectric tensor
,,, . .. .. .
auu 1111:0
P"'
111V<:OI~<:O . 1ney nze we nnear re-
sponse of the medium and are dependent on the molecular and perhaps crystal-
line structure of the material, as well as bulk properties like density and temper-
ature. For simple matenals the linear response is often isotropic in space. Then
Eaf3 and /L~f3 are diagonal with all three elements equal, and D = EE, H = JL'B
H/ IL·
. .
LV U<:O ,.,uu•u uc unaerswoa as nowmg ror me l:'ouner
. LOtp. ~"'v'
transforms m space and tlme of the field quantities. This is because the basic linear con-
nPPt;rm n onrll? fnrU onrl R\ POn hP _, -1. 'T'l
l.
Da(x, t) = ~ I I d 3 x' dt' Eu13 (x', t')E13 (x - x', t - t')
*Precedent would require writing B.= 2: 13 ~'-•I'H13 , but this reverses the natural roles of Bas the basic
magnetic held and H as the denved quantity. In Chapter S we revert to the traditional usage.
v . . . 11\
"· TA
where E"~(x', t') may be localized around x' = 0, t' = 0, but is non vanishing for some
range away from the origin. If we introduce the Fourier transforms D"(k, w), E13(k, w),
and Ea13 (k, w) through
r r
f(k w) = I d 3x I dt f(x t)e ik·x+iwt
J J
P~(T1!l\ '" h~ th · Pm>T;~r tro oc

-,· '- . t~ Af
Da(k, w) = 2: E. 13 (k, w)E13(k, w) (I.ll)

J3
A similar equation can be written H"(k, w) in terms of B13(k, w). The permeability tensors
are therefore functions of frequency and wave vector in general. For visible light or elec-
tromagnetic radiation of longer wavelength it is often permissible to neglect the non-
locality m space. 1 hen EoJ3 and ftoJ3 are tunctwns only ot trequency. 1 his IS the situatwn
discussed in Chapter 7, which gives a simplified treatment of the high frequency properties
of matter and explores the consequences of causality. For conductors and superconductors
long-range effects can be important. For example, when the electronic collisional mean
free path in a conductor becomes large compared to the skin depth, a spatially local form
of Ohm's law is no longer adequate. Then the dependence on wave vector also enters. In
the understanding of a number of properties of solids the concept of a dielectric constant
as a function of wave vector and frequency is fruitful. Some exemplary references are
given in the suggested reading at the end of this introduction.
n. L 1. 1 c_ I 1 A6 T L \ -'- .11
• v• vu~w "~ '"~""vu cum uc . v n "~'1' \" •v LLLJ '"~'~ uu
charges, regardless of their inertia, respond to applied fields, solids have dielectric
constants typically in the range of E•• IE0 ~ 2-20 with larger values not uncom-
mon. Systems with permanent molecular dipole moments can have much larger
and temperature-sensitive dielectric constants. Distilled water, for example, has
a static dielectric constant of EfEo = 88 at ooc and E!E0 = 56 at 100°C. At optical
frequencies onlv the electrons can respond significantlv. The dielectric constants
orP ;n thP ron<TP ~ 1~. 1 7 __ 1() , m;th ~ 1~. ?._':l fnr mnd cnl;.-lc · "'- hoc
c•~o
.I
.. ,
1 "7"7 ,-~"·
•-vv v
""
~·
-v
•'-
uw
. :t.. 1.
•w•s~,
"" ·v
. 11.
~
" ... ~~. "' .c

v• . •u•u•~
from 0 to 100°C.
The type of response of materials to an applied magnetic field depends on
the properties of the individual atoms or molecules and also on their interactions.
Diamagnetic substances consist of atoms or molecules with no net angular mo-
mentum. The response to an applied magnetic field is the creation of circulating
atomic currents that oroduce a verv small bulk ma!!netization oooosin!! the ao-
n];p.-1 fip].-1 w;th thP --' "'- • • nf ' ;n rT1 ()\ on .-I thP fnrm nf (l-Q\ th;<
/,
•
f-"0/-"aa
I
~
1
L
•'-
LHNHULU 1 LUv <HU~L
• ' "1-'
I
' 'uao \/-"0/-"aa " -'-
~UU~laH'-'v
/
.,
l.O X. lU . 1 nus mamagnensm 1s a very smau enecL u me oas1c arom1c unn or
the material has a net angular momentum from unpaired electrons, the substance
is paramagnetic. The magnetic moment of the odd electron is aligned parallel to
the applied field. Hence J.LoJ.L~a < 1. Typical values are in the range (1 - J.LoJ.L~.)
= 10- 2-10- 5 at room temperature, but decreasing at higher temperatures be-
cause of the randomizin!! effect of thermal excitations.
loon<>hr- m .t · ,], orP noromo1 · h11t hPr-"'"" nf · wt;nn< hi>-
11J "'£~.
_1_ . -"- 1. ": '- n .1 •'-
,.,_
ULUW0 1 OHV" "
Ul• U1H~1~1H U~HU 'IV,, ~~•vn U>~
-r ·~
(.lU'tu .r.. wr re, O-'U .r.. wr I'll), 1erromagneuc suoswnces snow spontaneous mag-
netization; that is, all the magnetic moments in a microscopically large region
called a domain are aligned. The application of an external field tends to cause
the domains to change and the moments in different domains to line up together,
leading to the saturation of the bulk magnetization. Removal of the field leaves
a considerable fraction of the moments still aligned, giving a permanent mag-
netization that can be as large as Br = J.LoMr ~ 1 tesla.
For data on the dielectric and magnetic properties of materials, the reader
can consult some of the basic physics handbooks* from which he or she will be
led to more speCihc and detailed compilations.
Materials that show a linear response to weak fields eventually show non/in-
ear oenavwr at mgn enough held strengtns as tne electromc or tome oscillators
are driven to large amplitudes. The linear relations (1.10) are modified to, for
example,
D -> €(1)£ + > E(Z) E E + ... (T 1 ?\
{3 {3,-y
For static fields the consequences are not particularlv dramatic but for time-
varying fields it is another matter. A large amplitude wave of two frequencies w1
and w2 generates waves in the medium with frequencies 0, 2w,, 2w?. w, + w?
w 1 - w 2 , as well as the original w 1 and w2 • From cubic and higher nonlinear terms
an even richer spectrum of frequencies can be generated. With the development
of lasers, nonlinear behavior of this sort has become a research area of its own,
called nonlinear optics, and also a laboratory tool. At present, lasers are capable
of generatmg light pulses with peak electric fields approaching 1012 or even 1013
V /m. The static electric field experienced by the electron in its orbit in a hydrogen
atom is er,la~ = 5 X 1011 V/m. Such laser fields are thus seen to be canable of
driving atomic oscillators well into their nonlinear regime, capable indeed of
destroying the sample under study! References to some of the literature of this
specialized field are given in the suggested reading at the end of this introduction.
The reader of this book will have to be content with basically linear phenomena.
1.:-J JJounaary conaltwns at Lnte1Jaces JJetween Different Medta

The Maxwell equattons (1.1) are differential equations applying locally at each
point in space-time (x, t). By means of the divergence theorem and Stokes's
theorem, they can be cast m mtegral form. Let Voe alfmte vOfume in space, S
the closed surface (or surfaces) bounding it, da an element of area on the surface,
anu n a unit norma1 LO we sunace aL aa pomnng outward trom tne enclosed
volume. Then the divergence theorem applied to the first and last equations of
~LHl) ytt:lU, lllC llllC!;HU Mii
{ {
'f u · n aa - 1 p d"x IT.13)
JS JV
r
<DR·nda-0 (T
,. 1.1\
/
Js
'LJ(L Hana/JooK OJ cnemrstry ana rnysrcs, eo. u. K. LIOe, 78th ed., CRC Press, Boca Raton, FL
(1997-98).
A~,>r;Nm Jn,6tufpnfDl, lJ, "" ,z, p~ n H r,. ' .U':111 "" ,v, -'-, o." "''
(1972), Sections 5.d and S.f.
Sect. 1.5 Boundary Conditions at Interfaces Between Different Media 17
The first relation is ust Gauss's law that the total flux of D out
is equal to the charge contained inside. The is
no net flux of B through a closed surface because of the nonexistence of magnetic
Similarly, let C be a closed contour in space, S' an open surface spanning
n' a unit normal at da pointing in the direction given by the right-hand rule from
the middle two equations in (l.la) gives the integral statements
H · dl = · n' da (T.15)
c
E • dl = (1.16)
These familiar integral equivalents of the Maxwell equations can be used

to deduce the · of various normal and
of the fields on either side of a surface between a
surface charge or current density at the interface. An appropriate geometrical
surface between two media with different

Similarly, the infinitesimal contour C has its long arms on
boundary and is oriented so that the normal to its spanning surface is tangent to
ume of the pillbox. In the limit of a very shallow pillbox, the side surface does
media. The boundary region is assumed to carry idealized surface charge and current
. .
other, with the normal n to its top pointing from medium 1 into medium 2. The
. . .
,. • ,. •J.. ' .L <1. LC< . {T 1 ~\ ..l fT 1 A\ A .1. >1. '

CULO VH lH~ L~>' LH \'''~/ ~"~ \'•'.'J• ~'"J '"' lVjJ GHU
bottom contribute. If the top and the bottom are parallel, tangent to the surface,
and of area ila, then the left-hand integral in (1.13) is
..l • ;1 rl j", {T 1 A\ H <L -'· ~ .1 •L c

, · ·r -o .0 p •o u< '"~ ~v a~
to produce an idealized surface charge density a-, then the integral on the right
in (I.13) is
3
Jv p d x = a- ila
"r• •• •-•
,/ ·r .o~H'O
c ~
VL U' UHU
_, ~
~ VH
.,,_
OlUv VL lHv
are related according to
(1.17)
(1.18)
In words, we say that the normal component of B is continuous and the discon-
.
.m u u y
. ·'"•'-
v• uw
"" ..
. VI U' Gl GHJ !-'VHH I> <O'fUGl LV lHv ~UIIG'-.0 \.:IlUige
density at that pomt.
In an analogous manner the infinitesimal Stokesian loop can be used to de-
termine the discontinuities of the tangential components of E and H. If the short
·o ·o·o ·o ·o
parallel to the surface and has length ill, then the left-hand integral of (1.16) is
[
Jc 1!. • dl - (t X n) · (E2 E 1 ) tJ.l
and similarly for the left-hand side of (I.15). The right-hand side of (1.16) vanishP<
of the short sides goes to zero. The right-hand side of ({15) does not va;ish,
however, if there is an idealized surface current density K flowing exactly on the
' . . . .
~uaa ...... ' " ~u"" cu Lne llll on Lne ngrn 01 ~1.1)) IS
Js· l J + aD] · da = K ·
iJt
t t ill
The second term in the integral vanishes by the same argument that was just
given. The tangential components of E and H on either side of the boundary are
••. _,. _, l
n x l1!.2 1!.t) u (1.19)

n X (H2 - H 1) = K (I.20)
u1 ~LLv J it
i, unuerstoou Lllat tne surrace current K nas only components parallel
to the surface at every point. The tangential component of E across an interface
is continuous, while the tangential component ofH is discontinuous by an amount
whose magnitude is eaual to thP m~· lP nf thP <nrf~rP r.r '"";h. ~nrl
whose direction is parallel to K x n.
The discontinuity equations (I.17)-(I.20) are useful in solving the Maxwell
equations in different regions and then connecting the solutions to obtain the
fields throughout all space.
I.
relations of such idealizations to the real world is perhaps worthy of a little dis-
First we consider the question of maintaining some conducting object at a
idea that the means does not significantly disturb the desired configuration of
from time to time, to have a conducting path or its equivalent from the object to
a source of char e far awa "at infinit " so that as other char ed or unchar ed
objects are brought in the vicinity, charge can flow to or from the object, always
maintainin its otential at the desired value. Althou h more so histicated
means are possible, metallic wires are commonly used to make the conducting
path. Intuitively we expect small wires to be less perturbing than large ones. The
reason is as follows:
potential diminishes indefinitely when the diameter of the wire is indefi-
dimensions will not be sensibly affected by the introduction of very

h as are used to form electrical con-
nexions between these bodies and the earth, an electrical machine, or an
electrometer.*
The electric field in the immediate neighborhood of the thin wire is very large,
considerable dimensions" the effects can be made small. An important historical
Leyden jars as his sources of charge, thin wires as conductors, and suspending
the objects in the room, Cavendish measured the amounts of charge on cylinders,
(the same sphere shown in Fig. 1.1) at the same potential. His values of capaci-
tance so measured are accurate to a few er cent. For exam e he oun t e
ratio of the capacitance of a sphere to that of a thin circular disc of the same
ra ms was . e t eorettca va ue 1s 1r •
There is a practical limit to the use of finer and finer wires. The charge per
a
conauctmg surtacer. To minimize the nerturbation of the s o"m"'

level, it is necessary to resort to other means to maintain potentials, comparison
methods using beams of charged particles intermittently, for example.
When a conducting object is said to be grounded, it is assumed to be con-
nected by a very fine conducting filament to a remote reservoir of charge that
serves as the common zero of potential. Objects held at fixed potentials are sim-
ilarly connected to one side of a voltage source, such as a battery, the other side
or wnicn is conne~o we common ·ground. !hen, when Imtlillly electrified
{)hif'l"t< <ITf' mOVf>rl Tf'l<ttiVP tn AnP ~nnthPr ;n on~h ~ '~" th ,t th • rJ"
v• ~·~~u•~ny u•~UUL LllCJI r lliCUS 1Cllldill uXeu, we appropnare
un~•~u,
amounts of charge flow from or to the remote reservoir, assumed to have an

inexhaustible supply. The idea of grounding something is a well-defined concept
in electrostatics, where time is not a factor, but for oscillating fields the finite
speed of propagation blurs the concept. In other words, stray inductive and ca-
pacitive effects can enter significantly. Great care is then necessary to ensure a
"[}Ood 2:round."
charge aensny or a surtace current densitY. The physical reality is that the chan,.e
or current is confined to the immediate neighborhood of the surface. If this region
has thickness small compared to the length scale of interest, we may approximate
the reality by the idealization of a region of infinitesimal thickness and speak of
a surface distribution. Two different limits need to be distinguished. One is the
limit in which the "surface" distribution is confined to a region near the surface
that is macroscopically small, but microscopically large. An example is the pen-
,
- H -' <L _1 • -' <L -l •'
•u~ OMU llldl lUI 111/',U CHUU)';ll 1110'!1
, , uuu <lllU )';UUU -o·
conductivities o can be macroscopically very small. It is then appropriate to in-
tegrate the current density J over the direction perpendicular to the surface to
obtain an effective surface current density Kerr·
The other limit is truly microscopic and is set by quantum-mechanical effects
in the atomic structure of materials. Consider, for instance, the distribution of
excess charge ot a conductin2: bodv in electrostatics. It is well known th<~t thi<
cnarge uensi1y u. 1uere Is no eJectnc held mslde tne conductor, out tnere Is, m
accord with (I.17), a normal component of electric field just outside the surface.
At the microscopic level the charge is not exactly at the surface and the field
does not change discontinuously. The most elementary considerations would in-
dicate that the transition region is a few atomic diameters in extent. The ions in
a meLat can oe wougm 01 as reJauveJV Immoblle ana Jocanzea to 1 an2:strom
or . thf' r _cl <>rP lP<< C"Andr<>;n,rl. ThP ~_nf__m .rlol nl
a1c ;11 , ;10 . L.J. 1ney come rrom a somrion ~e quantum-

mechanical manv-electron in whirh thP inn< of th" r.onrl,r-tnr <>rP
approximated by a continuous constant charge density for x < 0. The electron
density (rs = 5) is roughly appropriate to copper and the heavier alkali metals.
*N.D. Lang and W. Kahn, Phys. Rev. Bl, 4555 (1970); B3, 1215 (1971); V. E. Kenner, R. E. Allen,
and W. M. Saslow Phvs. Lett. lRA ?'i'i 11 Q77\
Sect. 1.6 Some Remarks on Idealizations in Electromagnetism 21
1.0 1- I
./
I
/
0.8 1-
I
I
I
I
I <oEn(x)! a
0.6 1- :/
A
f-
0.4 1-
0.2 1-
01-"'
"'""j i~
/1 I , i I I
~ X
0
(10- 8cm)
2 4 6
-0.2 '-
Figure 1.5 Distribution of excess charge at the surface of a conductor and of the
normal component of the electric field. The ions of the solid are confined to x < 0 and
are approximated by a constant continuous charge distribution through which the
electrons move The bulk of the excess charge is confined to wjtbjn +2 A of the
"surface."
The excess electronic charge is seen to be confined to a region within ±2 A of

the "surface" of the ionic distribution. The electric field rises smoothly over this
region to its value of u "outside" the conductor. For macroscopic situations
where 10- 9 m is a negligible distance, we can idealize the charge density and
electric field behavior as p(x) = £n'i(x) and E,(x) = u8(x)IE0 , corresponding to a
truly surface density and a step-function jump of the field.
\¥e see that the theoretical treatment of classical electromagnetism involves
several idealizations, some of them technical and some physical. The subject of
electrostatics, discussed in the first chapters of the book, developed as an exper-
imental science of macroscopic electrical phenomena, as did virtually all other
aspeets of electromagnetism. The extension of these macroscopic laws. even for
charges and currents in vacuum, to the microscopic domain was for the most part
an unjustified extrapolation. Earlier in this introduction we discussed some of
the limits to this extrapolation. The point to be made here is the following. With
hindsight we know that many aspects of the laws of classieal eleetwmagnetism
apply well into the atomic domain provided the sources are treated quantum
mechanically, that the averaging of electromagnetic quantities over volumes con-
taining large numbers of molecules so smooths the rapid fluctuations that static
applied fields induce static average responses in matter, and that excess charge
is on the surface of a conductor in a macroscopic sense. Thus Coulomb's and
Ampere's macroscopic observations and our mathematical abstractions from
them have a wider applicability than might be supposed by a supercautious phys-
• • cy'), .1.. c . ·~
simplmes matters!
References and Suggested Reading

The history of electricity and magnetism is in large measure the history of science
it<~ If w~ hOVP olrPOCJ. •it cJ Uf> • 0 ~1 ,, h· • ,, • "
tlorn a~<Jl!.
Another readable account, with perceptive discussion of the original experiments, is
N. Feather, Electricity and Matter, University Press, Edinburgh (1968).
The experimental tests of the inverse square nature of Coulomb's law or, in modern
language, the mass of the photon, are reviewed by
I. Yu. Kobzarev and L. B. Okun', Usp. Fiz. Nauk 95, 131 (1968) [trans!., Sov.
Phu< TTm 11 11R_Ll_Ot;Q\ 1
" • ' •J
<lllU
A. :-:.. LroJanaoer an M. M. Nieto Rev. Mod. Phvs. 43 277 (1971 ).

An accessible treatment of the gauge principle in the construction of field theories,
building on classical electrodynamics and ordinary quantum mechanics, can be found in
I. J. R. Aitchison and A. J. G. Hey, Gauge Theories in Particle Physics, 2nd ed.,
Adam Hilger, Bristol (1989).
Suggested reading on the topic of the macroscopic Maxwell equations and their der-
ivation from the microscopic eouations can be foun~ at the enrl of rh,nter li Tr., ho<i~
. ' . ,
UUUM uu ounu-ocac~~ ":' "'~"' 101
Asncrort ana IVlermm

Beam
Kittel
Wert and Thomson
Wooten
The second of these is aimed at electrical engineers and stresses practical topics like
semiconductors. The last one is mainly on optical properties. The need for spatial non-
locality m treatmg the surface impedance of metals (the anomalous skin effect) is discussed
in 1l nlo~P< hv
.
D
.
n; '
't't'
·"
,
A "·
·~ "'
~·
LACCUVmCO UHU ~'""""' J 'J ~u, 'UJ. V 1, eU. L.
Marton, Academic Press. New York (1954), pp. 1-45; Reports on Progress in
Physics, Vol. XXTTT, pp.176-266 (1960); The Dynamics of Conduction Electrons,
Gordon and Breach, New York (1965).
The concept of a wave-vector and frequency-dependent dielectric constant E(k, w) is
vy
Kittel, A_11.vanced Tavie D.
n PinP< ~- . "~""· ' " A ll . "'· . 'T -'· IH',L0\
1c.:-:.rern, o'>oua o'>tate Phystcs, Vol. D, eds. F. Seitz and D. Turnbull, Academic
Press, New York, pp. 299-408.
The field of nonlinear optics is now nearly 40 years old. Beginnings and introductions
can be found in
J. A. Giordmaine, Phys. Today 22(1), 38 (1969).
N. Bloembergen Am. J. Phvs. 35 989 !19671.
References :lj
Nonlinear optical phenomena and applications are discussed in

R. L. Sutherland, Handbook of Nonlinear Optics, Marcel Dekker, New York
(1966).
<;:r.mP tPvt< onrl mon .L On thP onh;P~t '>TP
..
~ . "' • ~ ~
v~y~,'
o"r" ~
~¥"~·"·
~ . ~-
, "''
" __ ,_
'"""J,.
~-
'"· H <1HU U. VV , '1- •unu ~ , VV H<OJ,
New York (1986).

Y. R. Shen, The Principles of Nonlinear Optics, Wiley, New York (1984).
ru- A...__..._
' - ' _..._ .L.L
P'TPD 1
.L ..L..J _..._ ' - _._
TV~=~
--rTTTT~ I I!
+A
I! V
+
--,::;~/"'to Cl /Y +i /> Cl
LJ I! L l...- f, I V..J f,U t, t, l...-...J
We begin our discussion of electrodynamics with the subject of electrostatics-

phenomena involving time-independent distributions of charge and fields. For
. .' .
--n:r<::>:ST
.'. . -.::n:TI>
. .
~lt:Vlt:W.
.
111 llll' ClldfJlt:l
weao not efaoorate significantly. We mtroduce concepts and defimtlons that are
important for later discussion and present some essential mathematical appara-
tus. In subsequent chapters the mathematical techniques are developed and
' . ~·
One point of physics should be mentioned. Historically, electrostatics devel-
oped as a science of macroscopic phenomena. As indicated at the end of the
Introduction, such idealizations as point charges or electric fields at a point must
oe vieweLras matnematical constructs that permit a descnptlon ot the phenomena
at the macroscopic level, but that may fail to have meaning microscopically.
1.1 Coulomb's Law

All of electrostatics stems from the quantitative statement of Coulomb's law
concerningllfe torce acting between charged bodies at rest With respect to each
other. Coulomb", in an impressive series of experiments, showed experimentally
that the force between two small charged bodies separated in air a distance large
comuared to their dimensions
varies directly as the magnitude of each charge,
varies inversely as the square of the distance between them,
. ,. ~ ... ...
~'
'15 '
d~
~'
_,
,..., J
.. ~
- 0 "''-" '-'Ha< 15'-'• auu
is attractive if the bodies are oppositely charged and repulsive if the bodies have
the same type of charge.
rur rt: u wa' t:XT tv tnat we LOLat wrce prouuceu on one
small charged body by a number of the other small charged bodies placed around
it is the vector sum of the individual two-body forces of Coulomb. Strictly speak-
in!!, Coulomb's conclusions auulv to char!!es in vacuum or in media of ne!!li!!ible
''<.,TIT~ ~ofo~ ~~~0;-ri~~n<;~~ ~J' ::.J..n~~~o ;~ ~;n1 ' • r'l A
"' -~
• J. 'b ~·
~.-
. "'
.L ...
TH •
nu:a;HIL .L' lt::IU
...-,. ~ Ll
Although the thing that eventually gets measured is a force, it is useful to intro-
duce a con cent one steu removed from the forces the conceut of an electric field
~•n ·~ o~~o ~~~nu ~f ;:J.. ' J...~~;oo A <J..n ~- • • • L.
, ,1 • j J.. /;,,1~
'J 'b'
24
Sect. 1.2 Electric Field 25
defined as the force per unit charge acting at a given point. It is a vector function
of position, denoted by E. One must be careful in its definition, however. It is
not necessarily the force that one would observe by placing one unit of charge
on a pith ball and placing it in position. The reason is that one unit of charge
may be so large that its presence alters appreciably the field configuration of the
array. Consequently one must use a limiting process whereby the ratio of the
fmce on the small test body to the charge on it is measm ed fm smalle1 aud smaller
amounts of charge.* Experimentally, this ratio and the direction of the force will
become constant as the amount of test charge is made smaller and smaller These
limiting values of magnitude and direction define the magnitude and direction of
the electric field E at the point in question. In symbols we may write
F = qE (1.1)
where F is the force, E the electric field, and q the charge. In this equation it is
assumed that the charge q is located at a point, and the force and the electric
field are evaluated at that point.
Coulomb's law can be written down similarly. IfF is the force on a point
charge qlo located at x1o due to another point charge q 2 , located at x2 , then
Coulomb's law is
(1.2)
Note that q 1 and q 2 are algebraic quantities, which can be positive or negative.
Th@ constant of proportionality k depends on the system of units us@d.
The electric field at the point x due to a point charge q 1 at the point x 1 can
be obtained directly:
(1.3)
as indicated in Fig. 1.1. The constant k differs in different systems of units.t In

electrostatic units (esu), k 1 and unit charge is chos@n as that charge that exerts
a force of one dyne on an equal point charge located one centimeter away. The
esu unit of charge is called the statcoulomb, and the electric field is measured in
statvolts per centimeter. In the SI system, which we employ here, k = (41TEo)- 1 =
10- 7 c2 , where fo"" 8.854 )( 10'- 12 farad per meter (F/m) is ealled the permittivity
of free space. The SI unit of charge is the coulomb (C), and the electric field is
measured in volts per meter (V/m). One coulomb (1 C) produces an electric field
Figure 1.1
*The diw eteness of elect1 ie eha1 ge (see Section L1) means that this mathematieallimit is imrossible
to realize physically. This is an example of a mathematical idealization in macroscopic electrostatics.
'The questwn of umts IS discussed m detail m the Appendix.
26 Chapter 1 Introduction to Electrostatics-SI
of approximately 8 9874 x l 0 9 Vim (8 9874 GVim) at a distance of 1 meter. One

electron (q 1.602 X 10 19 C) produces a field of approximately 1.44 X 10 9
V/m (1.44 nV/m) at 1 meter.
The experimentally observed linear superposition of forces due to many
charges means that we may write the electric field at x due to a system of point
charges q,, located at x,, i = 1, 2, ... , n, as the vector sum:
E(x)
l L
n
. I.Ji
X - X;
3 (1.4)
If the charges are so small and so numerous that they can be described by a
charge density p(x 1 ) [if t::..q is the charge in a small volume t::..x t::..y t::..z at the point
1 1
X , thtm !1q f3(X ) !1x !1y L\.z], the sum is replaced by an integral:
E(x) = - -
1 J p(x ) I x-
1 X
1
l3 d
3
x1 (1.5)
4 7rE0 X - X 1
.
1 1
where tPX dx dy 1
tlzl is a three-dimensional volume element at X 1•
At this point it is worthwhile to introduce the Dirac delta function. In one dimension,
the delta function, written li(x a), 1s a mathematically 1mproper functwn havmg the
properties:
1. o(x - a) = 0 for x * a, and
2. j li(x a) dx llf the regwn of mtegratwn mcludes x - a, and IS zero otherw1se.
The delta function can be given an intuitive, but nonrigorous, meaning as the limit of a
peaked curve such as a Gaussian that becomes narrower and narrower, but higher and
higher, in such a way that the area under the curve is always constant. L. Schwartz's theory
of distributions IS a comprehensive ngorous mathematical approach to delta functwns and
their manipulations.*
From the definitions above it is evident that, for an arbitrary function f(x ),
The integral of f(x) times the derivative of a delta function is simply understood if the
delta function is thought of as a well-behaved, but sharply peaked, function. Thus the
definition i
4. I f(x) o'(x- a) dx = -f'(a)
where a prime denotes differentiation with respect to the argument.
If tile delta function has as argument a function f(x) of the independent vatiable x,
it can be transformed accordmg to the rule,
1
5. o(f(x)) = ~ ldf o(x- x,)
d (x,) 1
X
where f(x) is assumed to have only simple zeros, located at x = x,.

In more than one dimension, we merely take products of delta functions in each
dimension. In three d1menswns, for example, With Cartesian coordmates,
1
A useful, rigorous account of the Otrac delta functiOn ts gtven by Ltghthtll. See also Dennery and
Krzywicki (Section Ill.l3). (Full references for items cited in the text or footnotes by italicized author
only will be found in the Bibliography.)
Sect. 1.3 Gauss's Law 27
0 if Ll V does not contain x = X
delta functions. For example,

n
charge density (1.6) into (1.5) and integration, using the properties of the delta function,
1.3 Gauss's Law
electric fields. There is another integral result, called Gauss's law, which is some-
obtain Gauss's law we first consider a point charge q and a closed surface S, as
q.-
q insideS
Figure 1.2 Gauss's law. The normal component of electric field is integrated over the
the charge by the inner side of the surface is 47T (zero).

-nr C':hant.er I to FIM·· . ..: -SI
element of surface area. If the electric field E at the point on the surface due to
the charge q makes an angle ()with the unit normal, then the normal component
of E times the area element is:
F .. n tin -
q COS() An {1 7\
47T<'. rL ' .
i< rli ~lnna thP linE> from thE> <
F. c>Jc>mont tn thP rh~r<TP
·o
n
·1
2
cos () da = r df!, where dfl is the element of solid angle subtended by da at
the position of the charge. Therefore
E · n da = _q_ dfl (1.8)

47TE0
If we now integrate the normal component of E over the whole surface, it is easy
In <PP th~t
r ,, ;f ,. . ..1 <'
f 5
E·nda=
T
,0
-u -1
if q lies outside S
(1.9)
This result is Gauss's law for a single point charge. For a discrete set of charges,
it is immediately apparent that
{ E · n da = : 2_ q, (1.10)
•t.. tho ""~ ;c moor ~n lu th~c.

~
'~ ·o·
•_J ••
' (' 1:'. ....
um1gc ucu~iLy PV•J, vau~~ ~ 1aw uecomes:
fs
E · n da = ..!_
Eo
J,v p(x) d x 3
(1.11)
where Vis the volume enclosed by S.

Equation (1.11) is one of the basic equations of electrostatics. Note that it
.... A
·r 't'
.t..
'i
1 c.
1~1
•• U<v <~<~v Uv< ovvH ~·n:rrg=,
the central nature of the force, and

the linear superposition of the effects of different charges.
Clearly, then, Gauss's law holds for Newtonian gravitational force fields, with
matter density replacing charge density.
It is interesting to note that, even before the experiments of Cavendish and
( oulomb. Priestley. takin2. uo an observation of Franklin that chanre seemed to
:..> ., •• • .... 1 ,, :..>
,
.c -' \.oUJJ, 1 A '
uy -' -OJ
'·~ ~· ~
'
wnn J'lewwn s Jaw or umversa1 gravitatiOn tnat the electrostatic Ioree must obey
an inverse square law with distance. The present status of the inverse square law
is discussed in Section 1.2.
~
. ~ oPP
--cT~lJCTr::ntlUI
.. ~
I'.,, ... UJ
p~
~LUW
'
.
~
'
l<l W \.o<lll 'U\0 LC Ul " ' Uv111g <ll1 ral 10Hil 01 Lilt: I a w 01
electrostatics. We can obtain a differential form 7i~. a di~nt;~1 ~em•Mion) hv
Sect. 1.5 Another Equation of Electrostatics and the Scalar Potential 29
usillg tne atvergence theorem. !he atvergence theorem states that tor any well-
behaved vector field A(x) defined within a volume V surrounded by the closed
surface S the relation
1 I
j A•naa Jv v · A a~x
5
holds between the volume integral of the divergence of A and the surface integral
of the outwardly directed normal component of A. The equatwn ill tact can be
used as the definition of the divergence (see Stratton, p. 4).
To apply the divergence theorem we consider the integral relation expressed
;n r:~uoo'o tl-.
f s
E · n da = 2. J p(x) d 3x
En v
Now the divergence theorem allows us to write this as
( rv · E - n/e,) d 3x - 0 (1.12)
Jv
for an arbitrary volume V. We can, in the usual way, put the integrand equal to
7<>r~ t~ ~ht~ ;n
V • E = ple0 (1.13)
which is the differential form of Gauss's law of electrostatics. This equation can
ttselt be used to solve problems m electrostattcs. However 1t ts often stmpler to
deal with scalar rather then vector functions of position, and then to derive the
vector quantities at the end if necessary (see below).
1.5 Another Equation of Electrostatics and the Scalar Potential

1ue sing1e equation ~i.LJ) is not ~;;uuugu Lu -r LY cuutplt:Ltoly Lilt: wrtoto cum-
ponents of the electric field E(x). Perhaps some readers know that a vector field
can be specified almost* completely if its divergence and curl are given every-
where ill space. Thus we look tor an equatwn specifying curl ]!; as a function ot
position. Such an equation, namely,
VxE=O (1.14)
LU ~- d. L ,. ~ '"'- -' - '- , ' ,_ ~\.
"""'"~ uu~~UJ HVH> VU1 0 . «J ~ HlVV \'•--')•
1 ( .r. '\ x- x'

.,-,
vr"\
4'7T<'A J ,_,. I 1..-- ..-'1 3
.J3 '
The vector factor in the integrand, viewed as a function of x, is the negative

gradient of the scalar 1/lx- x' 1:
'
~ x'l)
X X
--
lx- x'l 3 -v(lx
. _,, ,f <- •h T ,J, ~' ~ 1 Q AM

*fT.
.r
"
uniqueness.
•h o·
"'' '" '''""' "''A
30 Chaoter I Introduction to Electrostatics-Sf
C'.' •'-
CUv 51
..J: 1..
·'
d.
"VjJv1UUVH 1H oVHvO A, VUC HUl U1v 1Hlv51UUUH •A
' , u.v U H
be taken outside the integral sign. Then the field can be written
-I r n(x')
E(x)
47TEo
V J lx - x' I
d"x' (1.15)
Since the curl of the gradient of any well-behaved scalar function of position
vanishes (V x V!f! = 0, for all 1/J), (1.14) follows immediately from (1.15).
Note that V x E - 0 depends on the central nature of the force between
charges, and on the fact that the force is a function of relative distances only, but
does not depend on the inverse square nature.
In (1.15) the electric field (a vector) is derived from a scalar by the gradient
operation. Since one function of position is easier to deal with than three, it is
worthwhile concentrating on the scalar function and giving it a name. Conse-
quently we define the scalar potential <P(x) by the equation:
E- -V<P (1.16)
ThPn (1 1 <;\ ohmn< thQt thP <<-QlQr ;~1 ;< ;.,,,., · tPrm< ~f thP ,-h~r .. P -'
' . ' r - o· ~
"
by
<P(x) = _1_ J p(x') d3x' (1.17)

47TE0 lx-x'l
where the mtegratwn 1s over all charges in the universe, and <P is arbitrary only
to the extent that a constant can be added to the right-hand side of (1.17).
1 ne scalar potential nasa physical mterpretatwn when we consider the work
done on a test charge q in transporting it from one point (A) to another point
~D J in me presence 01 an e1ecrric neJQ J<,~XJ, as snown in dg. 1..). 1ne rorce acnng
on the charge at any point is
u ,u
'1
so that the work done in moving the charge from A to B is
W = - r
JA
F · dl = -a (B E · dl
JA
(1.18)
The minus si{Jn , because we are calculatinJ> the work clone nn the charJ>e
d. £ ~ '_, ~ .. ~'
lHv UvUVH V> lH'-' H'-'1U. H HH Ll1VH \L1U) lH\0 VVVH\. \.,<lH U\0 VVHll\OH
(B (B
w q )A V'l' • al q t a<v- q~<vs <JJA) ~ l.l'J)
Bi ~
....---- / "'
~dl/ // /
/
~//~A
~/
Fieure 1.3
Sect. 1.6 Surface Distributions of Charges and Dipoles 31
which shows that q can be interpreted as the potential energy of the test charge
in the electrostatic field.
From (1.18) and (1.19) it can be seen that the line integral of the electric field
of Stokes's theorem [if A(x) is a well-behaved vector field, Sis an arbitrary open
surface, and C is the closed curve bounding S,
J',(.c· A · dl = fs
(V x A) · n da
where dl is a line element of C, n is the normal to S, and the path Cis traversed
in a right-hand screw sense relative to n]leads immediately back to V x E = 0.
1.6 Surface Distributions of Charges and Dipoles and

Discontinuities in the Electric Field and Potential
One of the common problems in electrostatics is the determination of electric
field or potential due to a given surface distribution of charges. Gauss's law (1.11)
allows us to write down a partial result directly. If a surfaceS, with a unit normal
n directed from side 1 to side 2 of the surface, has a surface-charge density of
IT(x) (measured in coulombs per square meter) and electric fields E 1 and E 2 on
either side of the surface, as shown in Fig. 1.4, then Gauss's law tells us imme-
diately that
(E 2 - E 1) • n = ITIE0 (1.22)
is a discontinuity of u/e0 in the normal component of electric field in crossing a

surface with a surface charge density ff, the crossing being made in the direction
on.
The tangential component of electric field can be shov;n to be continuous
across a boundary surface by usmg ( I .21) for the Ime mtegral of E around a
closed path. It is only necessary to take a rectangular path with negligible ends
and one side on either side of the boundary.
An expression for the potential (hence the field, by differentiation) at any
point in space (not just at the surface) can be obtained from (1.17) by replacing
3
p d x by u da:
1
<l>(x) = -4-
7TEo
f
S
u(x')
IX - X
'Ida
,
(1.23)
For volume or surface distributions of charge, the potential is everywhere con-

tinuous, even within the charge distribution. This can be shown from (1.23) or
from the fact that E IS bounded. even though discontinuous across a surface
distribution of charge. With point or line charges, or dipole layers, the potential
is no longer continuous, as will be seen immediately.
Another problem of interest is the potential due to a dipole-layer distribution
on a surface S. A dipole layer can be imagined as being formed by letting the
surfaceS have a surface-charge density u(x) on it, and another surfaceS', lying
close to S, have an equal and opposite surface-charge density on it at neighboring
points, as shown in Fig. 1.5. The dipole-layer distribution of strength D(x) is
formed by letting S' approach infinitesimally close to S while the surface-charge
density rr(x) becomes infinite in sncb a manner that the product of g(x) and the
local separation d(x) of S and S' approaches the limit D(x):
lim u(x) d(x) = D(x)

d(x)~o
The direction of the dipole moment of the layer is normal to the surface S and
in the direction going from negative to positive charge.
To find the potential due to a dipole layer we can consider a single dipole
and then superpose a surface density of them. or we can obtain the same result
by performing mathematically the limiting process described in words above on
the surface-density expression (1.23). The first way is perhaps simpler. but the
second gives useful practice in vector calculus. Consequently we proceed with
Ill
Sect. 1.6 Surface Distributions of Charges and Dipoles 33
------- t
~ '
~~~
~~ s
.-::: s'- P'
0 "'-s·
Filmre 1.6 Dioole-laver lieometrv.
the limiting process. With n, the unit normal to the surfaceS, directed away from
J , a~ uu~ Ill rIg. LO, lilt: pUI UUt: lU lilt: lWU ClUSt: Sl IS
1 ( u(x 1 ) I
1 ( u(x 1 ) , "
'¥\A)
41TEo Js [x - X 1 [ uu 41TE0 Js· [x - X 1 + nd[ uu
'[), ,11 A .-l I. I --11 1 ·~ +1-. ol
I 'I ' 0 -~
fx + a[-\ where fa[ << [x[. We write a Taylor series expansion m three
dimensions:
1
[x +a[
= ! + a·
x
v(~) x
+ ···
In this way we find that as d ----> 0 the potential becomes
<P(x) = ~
41TEo
J
S
D(x')n · V'll
\IX -
1
X
,
I/
1
1da' (1.24)
In passing we note that the integrand in (1.24) is the potential of a point dipole
with dipole moment p = n D da '. The potential at x caused by a dipole p at X 1 is
1 p·(x-x 1 )
ci>( x) = ::;-- I '13
(1.25)
-u I· I
Equation (1.24) has a simple geometrical interpretation. We note that
1( 1 ) 1 _ cos 11 aa _
n•V ,[x-xl[ da- -[x-xlf2- -dD
where dD is the element of solid angle subtended at the observation point by the
area element da 1 , as indicated in Fig. 1.7. Note that dD has a positive sign if 1:1 is
s---
~ ..,n
/ ~~ ~
/
.
p..- ""'
:t __-/---- - \
....... _,/ ~/
-
__.
/
I:x-x''I'
.-·-
~- ~da' 1_7 Thte
dipole layer D on the area element da' is just
the negative product of D and the solid angle
· ·' ot P rln" tn th"
\ e1ernem au oy aa at t.
an acute angle (i.e., when the observation point views the "inner" side of the
dipole layer). The potential can be written:
1 I
'!'~X)
47TEo Js u~x ) au ~l.LD)
For a constant surface-dipole-moment density D, the potential is just the product

nf thP. m _, . . ' --1 h; 4,-,o,_ ~nrJ thP onlirJ ·m~l~ cnht~n~~~ ~t tJ,~ ~he, ;~n
l-'Vllll UJ Lilt; >UlWLOt;. 1' = ' Vl ll> ·-r •
There is a discontinuity in potential in crossing a double layer. This can be

seen by letting the observation point come infinitesimally close to the double
laver. The double laver is now imaP"ined to con<id nf hxrn ~nrto nnP hPina ~ <m~ll
disc directly under the observation point. The disc is sufficiently small that it is
sensibly fiat and has constant surface-dipole-moment density D. Evidently the
total potential can be obtained by linear superposition of the potential of the disc
and that ot the remamder. Yrom (l.Lb) 1t 1s clear that the potent1al of the disc
alone has a discontinuity of D/£0 in crossing from the inner to the outer side,
being - D/2£0 on the inner side and + D/2£ 0 on the outer. The potential of the
;n~~r nl~n~ ;tl ;, I ,] .,] tl -'" "' • • • •
cuoo
' '
me plane or me no1e. Lonsequenuy me total pOfeliTlaT]ump m crossmg the sur-
face is:
ih
"L
ih
1 ~ '~0 " ' -'--""
~
~~-
}
This result is analogous to (1.22) for the discontinuity of electric field in crossing
a surface-charge density. Equation (1.27) can be interpreted "physically" as a
potential drop occurring "inside" the dipole layer; it can be calculated as the
product of the held between the two layers of surface cnarge times the separation
before the limit is taken .
.... ,
---;-;;;; n. •
.L u•.J.JUH UHU
J .- .1
--r'•u._.~ ~.,-~
D ....
.. v•M
In Sections 1.4 and 1.5 it was shown that the behavior of an electrostatic field
can be descrihed hv the two iill ecm;Jtinn<·
V • E = pl£0 (1.13)
and
....• ~
~ v ~Ll'-t)
the latter equation being equivalent to the statement that E is the gradient of a
<r~l~r c. ion. the scillar iill <IJ·
1<' -'rim /1 1 L\
\ • v)
Equations (1.13) and (1.16) can be combined into one partial differential
equauon ror me smg1e runctwn lJJ(X):
VLC!> - - p!Eo (1.28)
This equation is called the Poisson equation. In regions of space that lack a charge
' . , "we; ouaw1
-' . . . .
woo equarwn:
V' 2 C!> = 0 (1.29)
Sect. 1.8 Green's Theorem 35
We already have a solution for the scalar potential in expression (1.17):
- " f •• \
1 ( p(x') _.13 ' (, 1 '"7\
~,~,
4m'o J Ix - x' I ~ ~ \'"''!
'T'.
•~ ·~u•J
:c. -''
~a~~UJ
,, +L +'- .
•au• uao
-'
~~~"
. -' -' . -"- •'-
•a~
n ..
• ~'1'
( ">0\
\'·~UJ• "~
operate with the Laplacian on both sides. Because it turns out that the resulting
integrand is singular, we invoke a limiting procedure. Define the "a-potential"
'!'a~X) uy
1 ( p(x')
'¥a~ X) A I a·x
'" ~o V \A A J T U
The actual potential (1.17) is then the limit of the "a-potential" as a---->- 0. Taking
the Laplacian of the "a-potential" gives
1 1
vz<PJx) ( n(x')V'2 d 3 x'
41TE0 J ,VrL + aL (1.30)
1 ( 3a 2
p(X a·x
"t'lTEo J _\' ' u J
where r = lx - x' 1- The square-bracketed expression is the negative Laplacian

of 1/Yr2 + a2. It is well-behaved everywhere for nonvanishing a, but as a tends
to zero it becomes infinite at r = 0 and vanishes for r =/= 0. It has a volume integral
eaual to 41T for arbitrarv a. For the of.. ion. · · -' : into two
regions by a sphere of fixed radius R centered on x. Choose F such that p(x')
changes little over the interior of the sphere and imagine a much smaller than
R and tending toward zero. If p(x ') is such that (1.17) exists, the contribution to
the integral (1.30) from the exterior of the sphere will vanish like a2 as a ---->- 0.
WP. thu' nP.P.rl · -'· nnlv th.e ., ntinn frnm in,irl.e thP. <nhPrP With ~
Taylor series expansion of the well-behaved p(x') around x' = x:one finds
•R r -
.? ~ I .Ja r .? ? ?.
v '<'a\A)
(rz + I'V'-J 6'1'' ... f Uf T V\U )
Eo Jo az)s12
T
n.·
~ .. ~~· .,.,
. • LJ.
]'~'~"
1
V'2 <Pa(x) = - - p(x) (1 + O(a2 /R 2 )) + O(a2 , a 2 log a) V 2p + · · ·
·u
In the limit a ---->- 0, we obtain the Poisson equation (1.28).

1 ne smgmar nature or tne L apJactan or 11r can De exmmtea rormauv m terms
of a Dirac delta function. Since V2 (1/r) = 0 for r =/= 0 and its volume integral is
-417, we can write the formal equation, V2 (1/r) = -4m5(x) or, more generally,
vz( 1 = -41To(x - x') (1.31)

lx- x'l
-. n ,... ,.
.L.O u . ......... .)' .L /to-., .......
If electrostatic problems always involved localized discrete or continuous distri-

butions of charge with no boundarv surfaces the general solution (1.17) wnulrl
be the most convenient and straightforward solution to any problem. There

would be no need of the Poisson or Laplace equation. In actual fact, of course,
if not mo<t of thP nr ' ' :of p], 't:>fir< i ' •finitP rPaion< of •
with or without charge inside, and with prescribed boundary conditions on the
bounding surfaces. These boundary conditions may be simulated by an appro-
• o' '' • r _, • • " • .r • ' _,_ • r- · '
~-ur.YITIUUllVllVl '0
"'" ' 0 Vl lW U· at """'"Y )•
but (1.17) becomes inconvenient as a means of calculating the potential, except
in simple cases (e.g., method of images).
To handle the boundary conditions 1t IS necessary to develop some new math-
ematical tools, namely, the identities or theorems due to George Green (1824).
These follow as simple applications of the divergence theorem. The divergence
theorem:
{ .o I
Jv-v ~ -:r_,~ ' I I UU
applies to any well-behaved vector field A deiined in the volume V bounded by

the closed surface S. Let A = cp V 1/J, where cp and 41 are arbitrary scalar fields.
Now
V • ~<P Vif!)- <P V"lf! + V¢ • Vlf! ~UL)
and
iJij!
cp Vlj! · n = cp- (1.33)
iJn
where iJ/iJn is the normal derivative at the surfaceS (directed outward from inside
the volume V). When (1.32) and (1.33) are substituted into the divergence the-
~~~~ •h~~~ ' rc.nn~'" -,;.,, . ; . .
'J ''J •
L 2
(4> 'i7 1j! + Vcp • Vlj!) d 3 x = { cp ~~ da (1.34)
If we write down (1.34) again with cp and ljJ interchanged, and then subtract it
from (1.34) the Vcb • Vw terms cancel and we obtain Green's second identitv or
,.--._
' .'
( ~2 ~2 •3 I iJij! acf>
-L \o/--. 'I' 'I' ~ 1 '+'o 'I' o. uu ~L;JJ)
The Poisson differential equation for the potential can be converted into an
interrral enuation if WP <'hoo,
the scalar notential and make u'e of '17 2 <1> -niF. From (1.i1 I w<'>' th>tt
~z" ·~· <
nw\A A ")o ov "
M.
wat \ L - ' - ' )
"~n•
~
(
I
L
..
, ~- I'
'¥~A )
M
U\A A ")
1
--;:;- 1-'V' "J
0
,,
UA
1
r
~
.... " '

u
a
\"
I 1I
n I
1 aw
n
HUH
" 1
uu
If the point x lies within the volume V, we obtain:
<ll(x) = __!__
47TE0
J.v f:1_XJ
R
d3x' + l Ts
47T s
1
Ran'
a _i_ I
an' ~) da' (1.36)
Sect. 1.9 Uniqueness of the Solution with Dirichlet or Neumann Boundary Conditions 37
If x lies outside the surfaceS, the left-hand side of (1.36) is zero.* [Note that this
is consistent with the inter retation of the surface inte ral as bein the otential
free volume, the potential anywhere inside the volume (a solution of the Laplace
equation) is expressed in (1.36) in terms of the potential and its normal derivative
only on the surface of the volume. This rather surprising result is not a solution
cification of the problem. This is discussed in detail in the next sections, where
techniques yielding solutions for appropriate boundary conditions are developed
using Green's theorem (1.35).
1.9 Uniqueness of the Solution with Dirichlet

or Neumann Boundary Conditions
at oun ary con ttlons are appropnate or t e otsson or ap ace equation
to ensure that a unique and well-behaved (i.e., physically reasonable) solution
will exist inside the bounded region? Physical experience leads us to believe that
specification of the potential on a closed surface (e.g., a system of conductors
Green's first identity (1.34).

We want to show the uniqueness of the solution of the Poisson equation,
2
V <P = -p/E0, inside a volume V subject to either Dirichlet or Neumann boundary
(1.37)
With the specified properties of U, this reduces (for both types of boundary
which implies V U = 0. Consequently, inside V, U is constant. For Dirichlet

boundary conditions, U = 0 on S so that, inside V, <1>1 = <1> 2 and the solution is
unique Similarly, for Neumann boundary conditions, the solution is unique
apart from an unimportant mbillmy additive comtant.
From the nght-hand side of (1.38) it is evident that there IS also a unique
solution to a problem with mixed boundary conditions (i.e., Dirichlet over part
of the surfaceS, and Neumann over the remaining part).
It should be clear that a solution to the Poisson equation with both and
a<t>/an specified arbitrarily on a closed boundary (Cauchy boundary conditions)
does not exist, since there are unique solutions for Dirichlet and Neumann con-
ditions sepamtely and these will in general not be consistent. This can be verified
w1th (1.36). With arbitrary values of <t> and a<t>/an inserted on the nght-hand s1de,
it can be shown that the values of <t>(x) and V<t>(x) as x approaches the surface
are in general inconsistent with the assumed boundary values. The question of
whether Cauchy boundary conditions on an open surface define a unique elec-
trostatic problem requires more discussion than is warranted here. The reader
may refer to Mane and Fnhbach (Section 6 2, pp 692 706) or to Sommerfeld
(Pmtial D(ffoential Equations in Physics, Chapter II) for a detailed discussion
of these questions. The conclusion is that electrostatic problems are specified
only by Dirichlet or Neumann boundary conditions on a closed surface (part or
all of which may be at infinity, of course).
1.10 Formal Solution of Electrostatic Boundary- Value

Ptoblem with Green Function
The solution of the Poisson or Laplace equation in a finite volume V with either
Dirichlet or Neumann boundary conditions on the bounding surface S can be
obtained by means of Green's theorem (1.35) and so-called Green functions.
In obtammg result ( 1.36 )-not a solutiOn-we chose the function 1/J to be
1/fx x' 1. it b~ing th~ pot~ntial of a unit point source, satisfying the equation:
~" 2 ( lx-1 x'l ) = -4m5(x - x') (1.31)
The function 1/lx - x'l is only one of a class of functions depending on the
variables x and x', and called Gnen functions, which satisfy (1.31). In general,
4mS(x x') (1.39)
where
1 I
G(x, x') lx x'I+F(x,x) (1.40)
with the function F satisfying the Laplace equation inside the volume V:
Sect. 1.10 Formal Solution of Electrostatic Boundary-Value Problem with Green Function 39
In facing the problem of satisfying the prescribed boundary conditions on 

or iJ<PiiJn, we can find the key by considering result (1.36). As has been pointed
out already, this is not a solution satisfying the correct type of boundary condi-
tions because both C!J anti aC!Jfan appear in the surface integral. It is at best an
integral relation for . With the generalized concept of a Green function and its
additional freedom [via the function F(x, x')], there arises the possibility that we
can use Green's theorem with 1/J = G(x, x') and choose F(x, x') to eliminate one
or the other of the two surface integrals, obtaining a result that involves only
Dirichlet or Neumann boundary conditions. Of course, if the necessary G(x, x')
depended rn detarl on the exact form of the boundary condrt10ns, the method
would have little generality. As will be seen immediately, this is not required,
and G(x, x ") satisfies rather simple boundary conditions on S.
With Green's theorem (1.35), , 1/J = G(x, x'), and the specified prop-
erties of G (1.39), it is simple to obtain the generalization of (1.36):
1
(x) = - - ( p(x')G(x, x') d 3 x'
47TE0 Jv
+ ~1 1 G(x x') iJ<l> _ (x') aG(x, x') da'
4 'lT Js , an ' an'
The freedom available in the definition of G (1.40) means that we can make the
surface integral depend only on the chosen type of boundary conditions. Thus,
for Dirichlet boundary conditions we demand:
Gn(x, x') - 0 for x' on S (1.43)
Then the first term in the surface integral in (1.42) vanishes and the solution is
For Neumann boundary conditions we must be more careful. The obvious

choice of boundary condition on G(x, x') seems to be
aG,~ (x, x') = 0 for x' on S
since that makes the second term in the surface integral in (1.42) vanish, as de-
SJred. But an apphcatwn of Gauss's theorem to (1.39) shows that
fsan'
aG
- da' = - 4'lT
Consequently the simplest allowable boundary condifton on GN ts

4'lT
an' (x, x') - for x on S (1.45)
s
where S is the total area of the boundary surface. Then the solution is
1 1 a
(x) = (>s + - p(x')GN(x, x') d 3 x' + ~ -, GN da' (1.46)
47TE0 v 4'lT s an
where (>s is the average value of the potential over the whole surface. The
customary Neumann problem is the so-called exterior problem in which the val-
ume V is bounded by two surfaces, one closed and finite, the other at infinity.
Then the surface area S is infinite; the boundary condition (1.45) becomes ho-
mogeneous; the average value ()s vanishes.
We note that the Green functions satisfy simple boundary conditions (1.43)
or (1.45) which do not depend on the detailed form of the Dirichlet (or Neumann)
boundary values. Even so, it is often rather involved (if not impossible) to de-
termine G(x, x') because of its dependence on the shape of the surfaceS. We
will encounter such problems in Chapters 2 and 3.
The mathematical symmetry property G(x, x') = G(x', x) can be oroved for
tha l"'!raa~ f,~~,;~~c c• • · tJ. , TY · ~h 1 >+ --;- -::J· • (1 A'l\ 1.
or ureen s meorem wnn rl H · · D. ~'
.-~ . . r .
ou wt: 'Y ry ts nm amomanc,<mr can oe 1mposea as a
·r
· rPnn;rPmPnt
~
* . .
,_~ a una1, rLanL rt:marK wt: nuLe Lne pnysiCllfllTIOanmg or 1'\X, x · )/47TE0.
Tt ;c ~ cr.lnt;r.n r.f thP T ~.-.~~~P Pnn<>t;r.n · 'rl '17 n~-::1 "~ tha 'nl
r ~ ~ r
of a system of charges external to the volume V. It can be thought of as the
potential due to an external distribution of charges chosen to satisfy the homo-
geneous boundary conditions of zero potential (or zero normal derivative) on
the surface S when combined with the potential of a point charge at the source
point x'. Since the potential at a point x on the surface due to the point charge
depends on the position of the source point, the external distribution of charge
F(x, x') must also depend on the "parameter" x'. From this point of view,
we see that the method of images (to be discussed in Chapter 2) is a physical
equivalent of the determination of the appropriate F(x, x') to satisfy the bound-
ary conditions (1.43) or (1.45). For the Dirichlet problem with conductors,
F(x, x')/47TE0 can also be interpreted as the potential due to the surface-charge
distribution induced on the conductors by the presence of a point charge at the
source point x '.
1.11 Electrostatic Potential Energy

and Energy Density; Capacitance
In :secnon 1.) n was snown mat me proauct or the scalar notent1al and the charge
point cnarge q, IS orougm rrom mnnny to a pomt x m a regwn of locahzed electnc

halrlc I h th ol 'nl ffi (, ..L'~L · ' • -~· .. \ +L -1
- J ' . \ YJ•
aone on me cnarge (ana nence ns potennaJ energvl 1s giVen bv
(1 Ll71
-~ (.
~J \' 'I
' . . r JJ
q 1(] 1, L, ... , n l) ar posll1ons x 1• 1nen
1 n-, n.
) (1 LIRI
*See K.-J. Kim and J.D. Jackson, Am. J. Phys. 61, (12) 1144-1146 (1993).
Seet. 1.11 Eleetrostatie Potential Energy and Energy Density; Capaeitanee 41
so that the potential energy of the charge q, is

n-1
w, = _!j_i_ .L qj (1.49)
47TE0 j~l Ix, - xj I
The total potential energy of all the charges due to all the forces acting between
them is:
w (1.50)
4rrco , 1 ;<t 1Xz
as can be seen most easily by adding each charge in succession A more symmetric
form can be written by summing over i and j unrestricted, and then dividing by 2:
(1.51)
It is understood that i = j terms (infinite "self-energy" terms) are omitted in the

double sum.
For a continuous charge distribution [or, in general, using the Dirac delta
functions (1.6)] the potential ene1gy takes the form.
W = _!__ Jr Jr p(x)p(x') d x d x' 3 3

(1.52)
87TEo lx - x' I
Another expression, equivalent to (1.52), can be obtained by noting that one of
the integrals in (1.52) is just the scalar potential (1.17). Therefore
W = ~ I p(x)<P(x) d 3 x (1.53)
Equations (1 51), (1 52), and (1 53) express the electrostatic potential energy
in terms of the positions of the charges and so emphasize the interactions between
charges v1a Coulomb forces. An alternative, and very fruitful, approach is to
emphasize the electric field and to interpret the energy as being stored in the
electric field surrounding the charges. To obtain this latter form, we make use of
the Poisson equation to eliminate the charge density from (1.53):
W = ~Eo I <l> V2<l> d'x
Integratwn by parts leads to the result:

r r
w 2 . 2 . (1.54)
where the mtegrat10n 1s over all space. In (1.54) all exphcrt reference to charges
has gone, and the energy is expressed as an integral of the square of the electric
field over all space. This leads naturally to the identification of the integrand as
an energy density w:
2
2
I h1s expresswn for energy dens1ty IS mtmtlvely reasonable, smce regrons of hrgh
fields "must" contain considerable energy.
There is perhaps one puzzling thing about (1.55). The energy density is pes
itive definite. Conseouent!v its .1. _, i< ihr Thi<

seems to contradict our impression from (1.51) that the potential energy of two
chames of opposite si!!n is neQ:ative. The reason for this 1mnarent con ·" ·
is that (1.54) and (1.55) contain "self-energy" contributions to the energy density,
whereas the double sum m (1.51) does not. To iiTustrate this, consider two point
charges q 1 and q 2 located at x1 and x2, as in Fig. 1.8. The electric field at the point
J> With coordmate xIS
1 nJv v.) 1 nfv v \
E +
47TEo lx- x1l 3 47TE0 lx - x2 l3
so that the energy density (1.55) is
n2 n2 n fv v \ 0 fv v \
327T"E0 W + +2 (1.56)
lx- x1l 4 lx- Xzl 4 lx- x,l31x- x213
Clearly the first two terms are "self-energy" contributions. To show that the third
term gives the proper result for the interaction potential energy we integrate over
all space:
n.n. r (v v .. ) 0 fv v \
167T2 E0; J lx - x1 l3 lx - x2l3 d"x

Wint (1.57)
A change of integration variable top= (x- x1 )/lx1 - x2 1yields

W
mt
= _1_ q,qz X _!_
47TEn I X1 - X? I 47T
Jp • In(p ++ nn)3 d3
n3 1 P
(1.58)
where n is a unit vector in the direction (x1 - x2). Using the fact that (p + n)/
13 - • • . .
II' 111. • r\ ~~ II' 111 J, ""' . - 0 ral can t:asuy ut: snuwn LU
have the value 47T, so that the interaction energy reduces to the expected value.
' L .L ' • ' '
' vu,._, "'-'"'5. u'-"''-'-" '-""'5'-u , '-" UC uy 1.0<111.01 _0 Lilt::
change in the total electrostatic energy of the system under small virtual displace-
~ ' r .
1110011". Vl <HI' <1100 111 woo pi• . LaT<;; lllUSL LJ<;; LaKt:ll lU
exhibit the energy in a form showing clearly the factors that vary with a change
-~ ' ' <L d. 1.
_0 "'-!-''
"' <UC!VH UHU CHV•?'- CHUC Ul'- '-VUOC< •
As a simple illustration we calculate the force per unit area on the surface
r ' .• ,_ .r '
v1 a WHU a 0 - if~XJ. Hl lUC ·o -
hood of the surface the energy density is
w - Eo IF.I2 - 2J?c. r1 <;o\

:z
If we now ima11ine a small eli -' ;\y nf <ln • 1 hl ~TP~ An nf
the conducting surface, the electrostatic energy decreases by an amount that is
the nroduct of ener<rv rlPn<itv t ' ancl thP PYr.lwlPrl w-.lnmP Av An·
~W = - <J
2
~a ~x/2E0 (1.60)
•p
1
nr
x/
I
'\. I
'\, I
Xi¥< ""
u ..,. ..LoO
Sect. 1.12 Variational Approach to the Solntion of the Laplace and Poisson Eqnations 43
[ h1s means that there IS an outward force per umt area equal to 0"272E0 w at
the surface of the conductor This result is normally derived by taking the product
of the surface-charge density and the electric field, with care taken to eliminate
the electric field due to the element of surface-charge density itself.
For a system of n conductors, each with potential V; and total charge
Q; (i = 1, 2, ... , n) in otherwise empty space, the electrostatic potential energy
can be expressed in terms of the potentials alone and certain geometrical quan-
Lilies called coefficients of capacity. Fo1 a given configmation of the conductms,
the linear functional dependence of the potential on the charge density implies
that the potential of the ith conductor can be written as
n
(i = 1, 2, ... , n)
where the Pu depend on the geometry of the conductors. These n equations can
be mverted to y~eld the charge on the nh conductor in terms of all the potentials:
n
(1 - L 2, ... , n) (1.61)
The coefficients C;; are called capacities or capacitances while the C;i, i i= j, are
called coefficients of induction. The capacitance of a conductor is therefore the
total charge on the conductor when it is maintained at unit potential, all other
conductors being held at zero potential. Sometimes the capaCitance of a system
of conductors is also defined. For example, the capacitance of two conductors
carrymg equal and opposite charges m the presence of other grounded conduc-
tors is defined as the ratio of the charge on one conductor to the potential dif-
ference between them. The equations (1.61) can be used to express this capaci-
tance in terms of the coefficients cij·
The potential energy (1.53) for the system of conductors is
n n n
w SQY, (1.62)
The expression of the energy in terms of the potentials V; and the C;i, or in terms
of the charges Q; and the coefficients Pu• permits the application of variational
methods to obtain approximate values of capacitances. It can be shown, based
on the technique of the next section (see Problems 1.17 and 1.18), that there are
variational principles giving upper and lower bounds on C;;. The principles permit
estimation ·.vith known error of the capacitances of relatively involved configu
rations of conductors. H1gh-speed computatiOnal techmques permit the use of
elaborate trial functions involving several parameters. It must be remarked, how-
ever, that the need for a Green function satisfying Dirichlet boundary conditions
in the lower bound makes the error estimate nontrivial. Further consideration of
this technique for ealeulating capacitances is left to the problems at the end of
th1s and subsequent chapters.
1.12 Variational Approach to the Solution of the Laplace

and Poisson Equations
v anatwnal methods play promment roles in many areas of classical and quantum
physics They provide formal techniques for the derivation of "equations of mo-
tion ana afso practicaf meflloasTor ootammg approximate, but often accurate,
solutions to problems not amenable to other approaches. Estimates of resonant
frequencies of acoustic resonators and energy eigenvalues of atomic systems
re>mP rP~riilu te> minrl
.. ..
~ ·r UlaL r "J SYSlt:lllS Ill t:4UlllUHU!llllaVt:
energy content is generalized to the consideration of energy-like functionals. As
an example, consider the functional
,, '' ~ I' - , I' ' •3 '; /~·

- '' U•3A
'l'I'J 2Jv•'1' ·'f
Jv 15'1' u "" ~LU.J)
where the function !fi(x) is well-behaved inside the volume V and on its surface
""~ .. -~ ,;z ,,, . ·' ....

S (which mav consist of several senarate surfaces) and p(x) is a snecified
• . T T UT.
'" "' ...I
,.
change in the functional when we change !fi __, 1/J + 81/J, where the modification
81/J(x) is infinitesimal within V. The difference 8I = I[ 1/J + 81/J] - I[ !fi] is
OT
'
1- - " " •3
'
I 0
,, "'
-= 'I' ~ Jv /5U'f' U A ~LU"<)
Jv
The neglected term is semipositive definite and is second order in 81/J. Use of
Green's first identitv with i = 81./J and 1]1 = 1./J vields
8l = Jv [- V !fi - 2
g] 81/J d 3 x + f s
81/J CJ!fi da
an (1.65)
Provided 81./J = 0 on the boundarv surfaceS (so that the surface inte2:ral vanishes).
., r: ' -' rr ·'-'
._, :£ _f .•• ' . ~
'l'f'J '/'\·')
V2!fi = -g (1.66)
Recalling that the ne2:lected term in (1.64) is semioositive definite, we see that
rr;/,1 io ~ minimnm ~ e~t;;.{;oo •> Pr>ieonn_lil-P Pnn~tinn mithin thP
"1
volume V and the departures 81/J vanish on the boundary. With lj1 __, <P and g __,
piE0, the minimization of the functional yields the "equation of motion" of the
electrostatic potential in the presence of a charge density and Dirichlet boundary
coilifttlons((J) Q"JVen on S ancl so ?id> - 0 there).
The derivation of the Poisson equation from the variational functional is the
formal aspect. Equally important, the stationary nature of the extremum of I[ !fi]
permits a practical approach to an approximate solution for !fi(x). We choose a
Lnar I lfi~X) • fi'¥~x, a, p, ... ) waL , on a normanzanon
constant A and some number of other parameters, a, {3, ... , and is constructed
to satisfy the given boundary conditions on the surface S. The function \(r may
be a sum of terms with the narameters as coefficients or a sin<>le function of
::1' t-~-~ . ' .--~;:: ~ ::;-... 1
'
_, , ..
. ....
or tne solution. ~lntmnon plays a role here!) CalcUlation ot ll!f!J giVes the tunc-
"t
1 £.
~J
..
tion, I(A, a, {3, . .. ). We now vary the parameters to locate the extremum (actually
a minimum) of I(A, a, {3, .. .). With the optimum parameters, the trial solution
isthe besf1:V;-ssibl,
-' r_ -'-
•
~ d.
im~ticm to thP trnP onlntinn mith thP
.U'-'
T
-' T
_ ,
'-''f'
. r
..
l~r fnnr_
. '
, '"'-' UVL <l 1> U v -
termined by the Dirichlet boundary values of !fi. For the Poisson equation, it is
determined by the source strength g(x), as well as the boundary values on S.
A ainerent runcrional IS necessary tor Neumann oounaarv conaJtwns. ~uo-
Sect. l.ll Variational Approach to the Solution of the Laplace and Poisson Equations 45
pose that the boundary conditions on ljJ are specified by (Jiji/(Jn Is = f(s), where s
locates a point on the surface S. The appropriate functional is
(1.67)
The same steps as before with + ' +I l'iifi lead to the first order difference in
(1.68)
The requirement that 8! vanish independent of 81/J implies
g vcithin F and f(s) on S (1.69)
Again the functional is a stationary minimum for ljlsatisfying (1.69). Approximate

solutions can be found by the use of trial functions that satisfy the Neumann
boundary conditions, just as described above for Dirichlet boundary conditions.
As a simple application to the Poisson equation, consider the two-dimen-
sional problem of a hollow circular cylinder of unit radius centered on the z-axis,
with an interior source density g(x) = g(p) azimuthally symmetric and inde-
pendent of z. The potential vanishes at p 1. The "equation of motion" for
ljJ (a function of p alone) in polar coordinates is
- a
1 -
pap
a+)
( p- =
ap
-g(p) (1.70)
Fot ttial functions we considet finite polynomials in powets of (1 p) and p. A

three-parameter functwn of the first type 1s
(1.71)
This choice might seem natural because it automatically builds in the boundary
condition at p = 1, but it contains a flaw that makes it a less accurate represen-
tation of ljJ than the power series in p. The reason is that, if the source density g
is well behaved and finite at the origin Gauss's law shows that •f! has a maximum
or minimum there ·.vith vanishing slope. The requirements at both the origin and
p - I are met by a three-parameter tnal functwnm powers of p:
'¥2 = ap2 + f3p3 + yp• - (a + f3 + y) (1.72)

We expect this trial function in general to be a better approximation to ljJ than
'lt 1 for the same number of variational parameters. [We could, of course, impose
the constraint, a 1 + 2{3 1 + 3y1 = 0 on (1.71) to get the proper behavior at the
origin, but that would reduce the number of parameters from three to two.]
The functional integral (1.63) for '1'2 is easily shovm to be
(1.73)
where en J!, g(p)(pn 1) pdp.

The integral for 'IF 1 has the same form as (1.73), but different coefficients.
As described above, we seek an extremum of (1.73) by setting the partial deriv-
atives with respect to the parameters a, {3, and y equal to zero. The three coupled
:ll(>ehr:lic linear eauations vield the "hest" va h,,,
')') .:;0 Ll'Jfb
-·
, . , , () 7
- L ·o
2450
{3= -420e2 + - - e3 - 420e 4 (1.74)
3
441
y = 210e2 - 420e3 + 2 e.
u1eM:: nuue' ""u uc iu>cllcu in10 ~ L t.J J 10 ~ive 11 'I' 2 Im in as a nor very iuuminaring
inn nf thf'. P OnP wmtlrl thPn finrl th ot thP "lr;nPt;P" flint\ ~ • m~e
. ,- .
. -,_ 1u u<tu we puLcu'i"' ~>cc,unu, urctCI\.eL ana opposire m stgn, a cnaracLer-
Jsttc of the extremum.
To go further we must specify g(p). The results for the best trial functions
'IF 1 and 'IF 2 are shown in Fig. 1.9 for the source density,
g(p) = -5(1 - p) + wV(1 - p) 5 (1.75)
; Ul >UUl\Ct: dllHllctly dllU 1>
. .
111<:: I' ~ . mar ts nor qune
fp~tnrPlP"" ThP "hP't" : fnr \II •p ~Q
~
-'7 0':1.1 ~n,-1 "
r-r
.J.lJ'otL. 111<:: Vdl lllt:: ViilUC,J['¥2Jmin
0
T.;)L\l I, CLJtupctlt::U ro
11<1>
I[ !/!]exact = -1.6017. The fractional error is 1.3%.

Note that the trial function 'IF 1 fails rather badly for p < 0.3 because it does
0.7 I I I I I I I I I
~
.... ....... .
... -~
.,
0.6 f---
..... ":'-.~
.-· ~ ·--~
0.5 f--- ,:;.- -
..
~: ...
F""" .·•
0.4 1-;--" ···· -
tI
0.3 f- /
.
•
Charge density
•
seal~\'\.• \
I!\
" I!\ -
-
,_
. / (arbitrarv
V.L
I ·, '\:-
( o)
I
.'
"-.. X·.
"'\.·.
i' "·,
0.0
_,
·- ...... __ ......
__..
~ .I
-0.1 f-
.I -
_n?
0.3
0.0
-- 0.1
_,./
0.2 0.3 0.4

0
0.5 0.6 0.7 0.8 0.9 1.0
Figure 1.9 Comparison of the exact solution 1/J(p) (solid curve) with two variational
approximations for the potential, '¥ 1 (dotted curve) and '¥ 2 (dashed curve). The charge
density (1.75) is indicated by the dash-dot curve (arbitrary scale).
Sect. 1.13 Relaxation Method for Two-Dimensional Electrostatic Problems 47
not respect the vanishing slope at p = 0. Nonetheless, it gives I[']t dmin = -1.5136,
which is somewhat, but not greatly, worse than ']t 2 (5.5% error). The insensitivity
.C Tf•T•l
Vo o_L I j <V ~HVoO oH
oL
U<~
. ·'c.
'HUO
'11 .
UOUOUU,~O
L
VV'H U
oL
""'~"5-"'
oL
UHU U
~
H' ~ '·
of the variational method. If the principle is used to estimate eigenvalues (related
t. tJ., ,], ,f Tr'Ttl\ ;t rl, ,]] TT, r1 >th r1 ,f ot' ,t· ,], .
1/J = ']t, it can faii 'b~dly, at least in .parts of the configuration space.
The reader will recognize from (1.70) that a polynomial source density leads
to an exact polynomial solution for 1/J, but the idea here is to illustrate the vari~
atwnal method, not to demonstrate a class ot explicit solUtions. t<Urther Jllustra~
t;nn ;< ]pft tn thP >< ~t thP Pnrl nf th;< ~nrll~tPr ~h· ·•· .
'T
1.13 Relaxation Method for Two-Dimensional

Electrostatic Problems
The relaxation method is an iterative numerical scheme (sometimes called iter~
mLV<. HHUCO 'LU~ lHCO l U~ lH<. , , U~ ~ ,
in two dimensions. Here we present only its basic ideas and its connection with
+J.. .1 +t. ~ v: ·..1. +t. T .1 :.• n· ·~"'
"l"
boundary conditions within a two~dimensional region S with a boundary contour
C. We imagine the region S spanned by a square lattice with lattice spacing h
(and the boundary contour C approximated by a step~ like boundary linking lat-
tice sites along C). The independent variables are the integers (i, j) specifying
t. • • tJ.. ..1, '·'-' +1. ',] ,], ,J' tJ. 'nl ,(," ;\ ot
1
eac h site.
. ' Th e potentia
.- . l va lues on t h e b oun d ary sites
. ,.. d g1ven.
are assume . Y\'' I
To P<t~hli<h thP ion~l of thP ,ol ~no! to !C. thP itf'rMivP
scheme, we imagine the functional integral I[ 1/J] over S as a s~m dver small do-
mains of area h", as shown in Fig. 1.10a. We consider the neighboring trial values
of the potential as fixed, while the value at the center of the subarea is a varia-
twnal quantity to be optimized. The spacmg 1s small enough to perm1t us to
rthP~d "'' ,rtPr ,f t~
Tr
· thP ..1. ' ' '
, -n H -,·
"' h
'J
I a!fi\ - 1 {,/, I a!fi - 1 {,,,

,,, \· ,,, \
ax NE h "~ ·w•
ay ;NE h H>' V,
and similarly for the other three quarters. The functional integral over the north·
east quarter is
1
INF - -
·- 2 o
dx
o
dy -
1
[ ax r r 2 2
ci/Jr + ci-/JrJ ay
(1.76)
- S [(1/Jo !fiN Y + (1/Jo !fiEYJ
The complete integral over the whole (shaded) subarea is evidently
I = ~ r(1/Jn - 1/JN ) 2 + (Wn - 1/fp )2 + (Wn - I]J, )

2
+ (1/Jn - 1/Jw)21 (1.77)
'+
Minimizing this integral with respect to 1/in gives the optimum value
1
~ 1/lo)optimum ~!/IN -t- !/IE -t- !/Is -t- 1/lw) ~l./15)
4
The trial values of the potential at the neighboring sites are labeled </JN, </Is, </JE, and </Jw,
surrounding sites (x).
The integral is minimized if if;0 is equal to the average of the values at the "cross"
points.
Now consider the whole functional integral, that is, the sum of the integrals
over all the subareas. We guess a set of if;(i, j) initially and approximate the
functional integral I[ if;] by the sum of terms of the form of (1.77). Then we go
over the lattice and replace half the values, indicated by the circles in Fig. l.lOb,
by the average of the points (crosses) around them. The new set of trial values
if;(i, j) will evidently minimize I[ if;] more than the original set of values; the new
set will be closer to the correct solution. Actually, there is no need to do the
1.22) by Taylor series expansion of any well-behaved function
where the "cross" and "square" averages are
((F(x, y)» = F(x, y) + (1.81)

Sect. 1.13 Relaxation Method for Two-Dimensional Electrostatic Problems 49
In (1.81) the Laplacians ofF are evaluated at (x, y). If F(x, y) is a solution of the
Laplace equation, the weighted averaging over the eight adjacent lattice sites in
(1.79) gtves Fat the center wtth corrections only of order h 6 • Instead of (1.78),
which is the same as (1.80a), a better iteration scheme uses if!new(i, j) = ((if!(i, j)»
I O(h 6 ). With either the "cross" or "square" averaging separately, the error is
O(h 4 ). The increase in accuracy with ((if!)) is at the expense of twice as much
computation for each lattice site, but for the same accuracy, far fewer lattice sites
are needed: ((N)) = O((N?13 ), where ((N)) is the number of sites needed with
((if!)) and (N) is the corresponding number with the "cross" or "square" average.
Equatwn (1.81) has an added advantage m apphcatwn to the Potsson equa-
tion, 'iPij, = -g. The terms of order h 2 and h 4 can be expressed directly in terms
of the specified charge density and the simplest approximation fur its Laplacian.
It is easy to show that the new value of the trial function at (i, j) is generated by
k2 h?
if!ncw(i, j) = ((lj1(i, j))) + ~ g(i, j) + ~O (g(i, j))c + O(h 6 ) (1.82)
wheie (g/c is the "cross" avetage of g, accmding to (1.80a).

A basic procedure for the iterative numerical solution of the Laplace or
Poisson equation in two dimensions with Dirichlet boundary conditions is as
follows:
1 A square lattice spacing h is chosen and the lattice sites, including the sites
on the boundary, are labeled in some manner [which we denote here as (i,j)].
2. The values of the potential at the boundary sites are entered in a table of the
potential at all sites.
3. A guess is made for the values, called <P 01ct{i,j), at all interior sites. A constant
value everywhere is easiest. These are added to the table or array of "start-
ing" values.
4. The first iteration cycle begins by systematically going over the lattice sites,
one by one, and computing ((<P(i, m> with (1.79) or one of the averages in
(1.80). This quantity (or (1.82) for the Poisson equation) is entered as
<Pnew(i, j) in a table of "new" values of the potential at each site. Note that
the sites next to the boundary benefit from the known boundary values, and
so their ((<P)) values are likely initially to be closer to the ultimate values of
the potential than those for sites deep in the interior. With each iteration,
the accuracy works 1ts way from the boundanes mto the mtenor.
5. Once all interior sites have been processed, the set of <P 01J(i, j) is replaced
by the set of <Pnew(i, j), and the iteration cycle begins again.
6. Iterations continue until some desired level of accuracy is achieved. For ex-
ample, one might continue itei ations until the absolute value of the diffet-
ence of old and new values is less than some preassigned value at every
interior site.
The scheme just outlined is called Jacobian iteration. It requires two arrays
of values of the potential at the lattice sites during each iteration. A better
scheme, called Gauss-Seidel iteration, employs a trivial change: one replaces
<Potd(i, j) with <Pnew(i, j) as soon as the latter is determined. This means that during
an Jteratwn one benefits tmmedtately from the tmproved values. 'I yptcally, at any
given site, ((<P)) is made up half of old values and half of new ones, depending
on the path over the lattice. There are many other improvements possible-
consult Pre.v.v et al., l'htnu:1ical Recipes, or some of the references cited at the end
of the chapter. The relaxation method is also applicable to magnetic field prob-
lems, as described briefly in Section 5.14.
Reje1 e11ces u11d Suggested Reading

On the mathematical side, the subject of delta funclions rs treated srmply but ngor-
ously by
Dennery and Kryzwicki

For a drscusswn of different types of partial differential equatiOns and the appropriate
boundary conditions for each type, see
Morse and Feshbach, Chapter 6
Sommerfeld, Partial Differential Equations in Physics, Chapter II
Comant and Hilbert, Vol U, Chapters III VI
The general theory of Green functions is treated in detail by

The general theory of electrostatics is discussed extensively in many of the older
books. Notable, in spite of some old-fashioned notation, are
Maxwell, Vol. 1, Chapters II and Iv
Jeans, Chapters II, VI, VII
Of more recent books, mention may be made of the treatment of the general theory by
Stratton Chapter III and parts of Chapter II
Readers interested in variational methods applied to electromagnetic problems can
Cairo and Kahan
Sadiku, Chapter 4
and
P6lya and Szegi:i
for elegant and powerful mathematical techniques.
The classic references to relaxation methods are the two books by R. V. Southwell:
Relaxation Method' in Engineering Srimce, Oxford I Iniversity Press, Oxford
(1940).
Relaxation Methods in Theoretical Physics, Oxford University Press, Oxford
(1946).
Physicists will be more comfortable with the second volume, but much basic material is
m the first. More modern references on relaxatiOn and other numerical methods are
Sadiku
Problems
1.1 Use Gauss's theorem [and (1.21) If necessary] to prove the following:
(a) Any excess charge placed on a conductor must lie entirely on its surface. (A
conductor by definition contains charges capable of movmg freely under the
action of applied electric fields.)
Ch. 1 Problems 51
(b) A closed. hollow conductor shields its interior from fields due to charges out-
side, but does not shield its exterior from the fields due to charges placed
inside it.
(c) rhe electnc field at the surface of a conductor JS normal to the surface and
has a magnitude a'/<0 , where a is the charge density per unit area on the
surface.
1.2 The Dirac delta function in three dimensions can be taken as the improper limit as
a---> 0 of the Gaussian function
D(a; x, y, z) = (21T)- 312 a- 3 exp[-

2 ~2 (x 2
+ y 2 + z2 ) ]
Consider a general orthogonal coordinate system specified by the surfaces u

constant, v constant, w constant, with length elernents du!U, du/V, dw/W in
the three perpendicular d1rectwns. Show that
o(x - x') = o(u - u') o(v - v') o(w - w') . uvw
by considering the limit of the G-aussian above Note that as a~ 0 only the jnfin-
itesimal hmgth element need be used for the distance between the points in the
exponent.
1.3 Using Dirac delta functions in the appropriate coordinates, express the following
charge distributions as three-dimensional charge densities p(x)
(a) In sphelical com dinates, a cha1 ge Q unifmmly distJ ibuted ave! a spherical
shell of radius R.
(b) In cylindrical coordinates, a charge A per unit length uniformly distributed
over a cylindrical surface of radius b.
(c) In cylmdncal coordinates, a charge Q spread uniformly over a flat circular
disc of negligible thickness and radius R.
(d) The same as part (c), but using spherical coordinates.
1.4 Each of three charged spheres of radius a, one conducting, one having a uniform
charge density within its volume, and one having a spherically symmetric charge
density that varies radially as r" (n > - 3), has a total charge Q. Use Gauss's theorem
to obtain the electric fields both inside and outside each sphere. Sketch the behavior
of the fields as a function of radius for the first two spheres, and for the third with
n
1.5 The time-averaged potential of a neutral hydrogen atom is given by
417"&,
e'"( ar)
<_P -q- - _ 1 + - _
r 2
where q is the magnitude of the electronic charge. and a - a0 /2, a0 being the
Bohr radius. Find the distribution of charge (both continuous and discrete) that will
give this poteiJti~l ~Ild iiJterpret yo'lf re~ult physic3lly
1.6 A s1mple capacttor ts a devtce formed by two msulated conductors adJacent to each
other. If equal and opposite charges are placed on the conductors, there will be a
certain difference of potential between them. The ratio of the magnitude of the
charge on one conductor to the magnitude of the potential difference is called the
capacitance (in SI units it is measured in farads). Using Gauss's law, calculate the
(a) two large, flat, conducting sheets of area A, separated by a small distance d;
(b) two concentric conducting spheres with radii a, b (b >a);
(c) two concentric conducting cylinders of length L, large compared to their radii
a, b (b >a).
(d) What is the inner diameter of the outer conductor in an air-filled coaxial cable
whose center conductor is a cylindrical wire of diameter 1 mm and whose
capacitance is 3 X 10- 11 F/m? 3 X 10- 12 F/m?
1.7 Two long, cylindrical conductors of radii a1 and a 2 are parallel and separated by a
distance d, which is large compared with either radius. Show that the capacitance
per unit length is given approximately by
1
C = 1TE0(ln - ~)
where a is the geometrical mean of the two radii.
Approximately what gauge wire (state diameter in millimeters) would be nee-
essary to make a two-wire transmission line with a capacitance of 1.2 X 10- 11 F/m
if the separation of the wires was 0.5 em? 1.5 em? 5.0 em?
1.8 (a) For the three capacitor geometries in Problem 1.6 calculate the total electro-
static energy and express it alternatively in terms of the equal and opposite
charges Q and - Q placed on the conductors and the potential difference
between them.
\UJ l!IC cucq;y UCliSllY U1 lllt: t:1t:Ul · llClU 111 cdUl ~'J>c ,, d
of the aoorooriate linear coordinate.
..., ae Lllc allla~Live wrce ue1ween conuUCLorS in llle paraue1 p1a1e capachor
(Problem 1.6a) and the narallel cvlinder canacitor !Problem 1.7) for
\3J uxeu cuarges un eacn conuuc1or;
,... ' " . ..1
'I"
· ,.1 '· ee L -• '
1.10 Prove the mean value theorem: For chan•P--free sn2ce th" v2ln" nf th"
. _,
..., .. 1 ... "o'
_,· ... . ' ... _, .c
"any spnere cemereu on mat pomL

•J " r
.
11< "-
"
, ., . "1-' •' -· ... ,£
" .,. ,
tne normal aenvanve or tne eJectnc ne1a rs grven oy
1 dl'.. I 1 1
E an \R, R2
where R 1 and R 2 are the principal radii of curvature of the surface.
1.12 Prove Green's reciprocation theorem: If <P is the potential due to a volume-charge
density p within a volume V and a surface-charge density a- on the conducting
surface S bounding the volume V, while <P' is the potential due to another charge
distribution p' and a-', then
L p<P' d'x + L a-<P' da = L p'<P d 3 x + L a-'<P da

1.13 Two infinite grounded parallel conducting planes are separated by a distance d. A
point charge q is placed between the planes. Use the reciprocation theorem of
Green to prove that the total induced charge on one of the planes is equal to ( -q)
times the fractional perpendicular distance of the point charge from the other plane.
(Hint: As your comparison electrostatic problem with the same surfaces choose one
whose charge densities and potential are known and simple.)
1.14 Consider the electrostatic Green functions of Section 1.10 for Dirichlet and
Neumann boundary conditions on the surface S bounding the volume V. Apply
Green's theorem (1.35) with integration variable y and 1> = G(x, y), ifi = G(x', y),
with V2y G(z, y) = -41Til(y- z). Find an expression for the difference [G(x, x')-
G(x', x)] in terms of an integral over the boundary surfaceS.
(a) For Dirichlet boundary conditions on the potential and the associated bound-
ary condition on the Green function, show that Gv(x, x' must be symmetric
in x and x'.
Ch. 1 Problems 53
(b) For Neumann boundary conditions, use the boundary condition (1.45) for
GN(x, x') to show that GN(x, x') is not symmetric in general, but that GN(x,
x') - F(x) is symmetric in x and x', where
1 (
r(XT- S7s-u,:vrx, YJ day
(c) Show that the addition of F(x) to the Green function does not affect the po·
tential <l>(x). See problem 3.26 for an example of the Neumann Green function.
1.15 Prove Thomson's theorem: If a number of surfaces are fixed in position and a given
total charge is placed on each surface, then the electrostatic energy in the region
bounded bv the surfaces is an absolute minimum when the char.,es are nlaced so
that every surface is an equipotential, as happens when they are conductors.
1.16 Prove the following theorem: If a number of conducting surfaces are fixed in po·
sffion wffifa gJvenTotllifflarge on eaffl, me mtroauclion ot an uncharged, msulated
conductor into the region bounded by the surfaces lowers the electrostatic energy.
1.17 A volume V in vacuum is bounded by a surface S consisting of several separate
conducting surfaces S;. One conductor is held at unit potential and all the other
conaucwrs at zero pmemia1.
(a) Show that the capacitance of the one conductor is given by
c = Eo r IV<PI 2
d 3x
where <l>(x) is the solution for the potential.

fh\ <;;hmu thot thP trnP conoc;toncP r ;o ol.uom laeo thon m Pnnol tA thP nnont;tu
-,- .,
C['l'] = Eo r IV'l' 2
1 d 3X
where 'I' is anv trial function satisfvin!! the boundarv conditions on the con·
ductors. This is a variational principle for the capacitance that yields an upper
hound.
T.Tir -consillerllfe cofiTfguranon Ofcoi\Guctors Drl'r65Iem 1.1"1, wllh all conductors ex-
cept S 1 held at zero potential.
.. , . ,, .., ,, '"
fn\ IOhAno thnt tho
~
... , ihf~\
"'J
Al 17
·"
surfaces S; can be written
<l>(x) = - -
41TEo
1
t st
<T1 (x')G(x, x') da'
where <T1 (x') is the surface charge density on S 1 and G(x, x') is the Green
function potential for a point charge in the presence of all the surfaces that
' ' ' . . " ··• -c ' "' ' '' •
'F ~ "T TI)·" ·~ mov uw. "'~
"'"~
energy is
W = ~ 7TE0
t t
st
da
St
da' <T1 (x)G(x, x')<Tt(x')
where the integrals are only over the surface S 1 .

\UJ --,n:ow-.:mrr llle ao ~Al"
~ da ~ da'<T!x\G!x x'\<Tix'\
7s,
c-'[<T] = 7s,
r 2
47TEn CJl <T(x) da

L JS,
.. '-
'] ~.,,
- '-' -" _/
,~,
--' .,, .
~
']
,_
variations of <T away from <T,. Use Thomson's theorem to prove that the
54 Chapter 1 Introduction to Electrostatics-51
't'
-' .£ ~- " 1
LV J 5' ' .' ' ··- .£ •'-
-
ductor S,.
1.19 For the cylindrical capacitor of Problem 1.6c, evaluate the variational upper bound
of ProDremT.T7b with the naive trial function, 'l'1 (p) - (b - p)l(b -a). Compare
<L . . ,1 .1o '<L oL ,o .1< £- Ll. 1 C ·_., ~ D. ,1 • ;1.. ~ .£
.' '
your results in terms of the functional form of '¥ 1 . An improved trial function is
treated by Collin (pp. 275-277).
I.:w In estJmatmg the capacitance ot a gtven connguratwn or conauctors, companson
,.,,,1.. ~"-"~''"-~"" '" · " r'. .~ . ' ,,, ,, ~- -
-r "''
tors in which the (n - 1) conductors held at zero potential are the same, but the
one conductor whose capacitance we wish to know is different. In particular, let the
'
ration have surfaces;, with s; totally insides,.

'" • L ~1 ~ ..
0 rl ' .~ •'- '"
(a) Use the extremum nrincinle of Section 1.12 and the variational nrincinle of
Problem 1.17 to prove that the capacitance C' of the conductor with surface
Si is less than or equal to the capacitance C of the conductor with surfaceS,
that encloses s;.
(b) Set upper and lower limits for the capacitance of a conducting cube of side a.
Compare your limits and also their average with the numerical value,
C = 0.655(47TE0 a).
(c) By how much do you estimate the capacitance per unit length of the two-wire
system of Problem 1.7 will change (larger? smaller?) if one of the wires is
renlaced bv a wire of snuare cro" · ' . 'irlP ;, Pnn~l tn it' rli~mPtPr?
1.21 A two-dimensional potential problem consists of a unit square area (0 :;; x :;; 1,
0 :5 y :5 1) bounded by "surfaces" held at zero potential. Over the entire square
. . . .
~ c-or=,u ""5"' \Per Ullll -o I Ill ZJ.
,_, A ,, •'- .•'-
~·'Y •
·r
"
\ · J
£ ,,_ . n ·
" L~>
" 'YY'J
ational" trial function 'l'(x, y) =A· x(1 - x) · y(1 - y) to determine the best
J
value of the constant A. [I use quotation marks around variational because

_, .. -" . 1
'Y 'Y "I- --J
,,_' '1'1.. /.,H ;, ,, £. ol..' ~'-'
. r. n. ,,_, • " 1 c ~
"7 '1' l'
2.16]
~
16 ;, sin[(2m + 1)7Txl coshf(2m + 1)7T(y - ~)l
'<o"'\· 'Y J 2 L._ I'
7T m=O (2m + 1)" cosh[(2m + 1)7T/2]
For y = 0.25 andy = 0.5, plot and compare the simple variational solution of
part a wifllllie exact smutwn asrunctwns of x.
1.22 Two-dimensional relaxation calculations commonly use sites on a square lattice with
spacing ~x = ~y = h, and label the sites by (i, j), where i, j are integers and x, =
,;, "· 'T'L .-1 . £ <1. ',1 /' ;> L • ~ L
~"'JJ ; Jll· , \•JJ ~t-r JJ
the average of the values at neighboring sites. [Recall the relevant theorem about
harmonic functions.] But what average?
W !IF(x, YJ 1s a werr:Dellavea function in the neighborhood of the origin, but
,o' ' ' .
,:J., L
·~• .
'
L
"} , .,L .,. •'-
~,
~
~,
"cross" sum
S, = F(h, 0) + F(O, h) + F( -h, 0) + F(O, -h)

can be exnressed as
j,4
S, - 4F(O, 0) + hL'i!LF + (F.mx + Fyyyy) + O(h")

12
Ch. 1 Problems !l!l
(b) Similarly, show that the "square" sum,

Ss = F(h, h) + F( -h, h) + F( -h, -h) + F(h, -h)
can oe expressea as
zV V
4 4
Ss -- 4F(O, 0) + 2h 2 V2 F - 3h (Fxxxx + Fyyyy) + h 2( 2

F) + O(h )
6
Here Fxxxx is the fourth partial derivative ofF with respect to x, evaluated at
x = 0. v = 0 etc. If V2F = 0 the avera!!es S ./4 and S 14 each give the value of
F(O, 0), correct to order h 3 inclusive. Note that an improvement can be ob-
tained by forming the "improved" average,
1 1
((F(O, O))) - S Sc + 4 S,
""" ~
((F(O, 0))) = F(O, 0) + :o 2
h V F
2
+ :~ V (V F) + O(h
2 2 6
)
If V2F = 0, then S gives F(O, 0), correct to order h 5 inclusive. For Poisson's
equation, the charge density and its lowest order Laplacian can be inserted
forth" '"mP. 'r.v
1.23 A transmission line consists of a long straight conductor with a hollow square region
in its interior, with a square conductor of one-quarter the area of the hollow region
centered in the empty space, with edges parallel to the inner sides of outer con-
ductor. If the conductors are raised to different potentials, the potential and electric
field in the space between them exhibit an eightfold symmetry; the basic unit is
sKetcnea m me accompanymg ngure. 1 ne erncacy 01 tue re.axadon mewou in ue-
termining the properties of the transmission line can be illustrated by a simple
calculation.
~aJ usmg amy tne tour mtenor pomts maicateu in tue ugure, write uown tne
relaxation equation for each point for the "cross" and the "improved" aver-
"";na cPhPmPC I riPiinPrl ;n 1 ??\ if the inner conductor has <P 100
V and the outer has <P = 0. By performing either the relaxation iteration
process or solving the set of algebraic equations for each scheme, find esti-
mates for the potential at each of the four points for the two schemes.
(b) From the results of part a make the best estimate (or estimates) you can for
the capacitance per unit length of the transmission line.
{r\ (, ,I\ Tr,;na vnnr fovor;tP •tMional tools rene at the relaxation cal-
culation with half the lattice spacing (21 interior points) and compare.
Answer: <P 1 = 48.87 V, <P2 = 47.18 V, <P 3 = 38.34 V, <P 4 = 19.81 V and C = 10.23
Eo F/m [from an accurate numerical calculation] .
• 3'
T
"'
.1
I
"
~
• 2' • 2 •3
I
I
Problem 1.-lj
56 Chapter 1 Introduction to Electrostatics-51
~·~~ ~ roisson equation prou1em or t'TODJem

lllC lWU-
1.21, a unit square with zero potential on the boundary and a constant unit charge
density in the interior, by the technique of relaxation. Choose h = 0.25 so that there
are nine interior sites. Use symmetry to reduce the number of needed sites to three,
at (0.25, 0.25), (0.5, 0.25), and (0.5, 0.5). With so few sites, it is easy to do the
· • ''' . hl -' . e -' ·
"' ~ ~!-'~' uuu u !-'~~~~· 0 VU< OUH JVU<O'
(a) Use the "improved grid" averaging of Problem 1.22 and the simple (Jacobian)
iteration scheme, starting with 47Te0 <1> = 1.0 at all three interior sites. Do at
least six iterations, preferably eight or ten.
(b) Repeat the iteration procedure with the same starting values, but using Gauss-
Seidel iteration.
(c) Graph the two sets of results of each iteration versus iteration number and
compare with the exact values, 47Te0 <1>(0.25, 0.25) = 0.5691, 4?Te0 <1>(0.5, 0.25)
(\ '7"l(\<: A ,,A " A "' A A
• , 0 • , v.J J V.7~JU. vu W<~ v• _
0
: <iiiU llllal
accura<:y.
~
·~
CtlAr l_DK L
., ,.- T -, ~
., .,
liOUnaary- vatue rrootems
~
•
1n Electrostatics: I
Many problems in electrostatics involve boundary surfaces on which either the

potential or the surface-charge density is specified. The formal solution of such
. . .
pr' was r m LlV, usmg Lne 01 vreen wncuuns.
In practical situations (or even rather idealized approximations to practical sit-
. • \ •• .l! · ' •'- r'. '· . . . .l
' / cuv ~>O• •V 'V>J V JJ
times not. Consequently a number of approaches to electrostatic boundary-value
problems have been developed, some of which are only remotely connected to
the Green function method. In this chapter we will examine three of these special
techniques: (I) the method of images, which is closely related to the use of Green
functions; (2) expansion in orthogonal functions, an approach directly through
the drtterenhal equatwn and rather remote tram the dtrect constructiOn ot a
Green function; (3) an introduction to finite element analysis (FEA), a broad
c1ass 01 numerica1 meLuous. f i major omissiuu is we use u1 cu111p1t;x-v; ·
IPrhnimH" inr1nrlin<> .-,.-.nformo 1 m: · for thP. trP.otmP.nt of two-rlimPn,ion" 1
problems. The topic is important, but lack of space and the existence of self-
contained discussions elsewhere accounts for its absence. The interested reader
may consult the references cited at the end of the chapter.
2.1 Method oj Images

The method ot Images concerns ttselt wtth the problem ot one or more pomt
charges in the presence of boundary surfaces, for example, conductors either
grounaea or new ar nxea poremia1s. unuer ravora01e conairions iris possio1e w
infer from the geometry of the situation that a small number of suitably placed
charges of appropriate magnitudes, external to the region of interest, can simu-
late the reauired boundarv conditions. These char2:es are called imal!e charl!es
·' •'- ,1-.L ,., .... ". . ... ·'
" '"
,]. .1.
.. 'J 'b' " 'b'
with image charges but not boundaries is called the method of images. The image
charges must be external to the volume of interest, since their potentials must be
solutions of the Laplace equation inside the volume; the ''particular integral"
(i.e., solution of the Poisson equation) is provided by the sum of the potentials
.
V>
....
CUV
_... • .l. • '-
CHv
-L
A stmple example ts a pomt charge located m tront ot an mhmte plane con-

rlnrtnr ot 7Prn ;ol oo ohnmn ;n 1'/;cr ? 1 Tt ;o p)por thot th;o ;o · 1Pnt tn
the problem ofthe original charge and an equal and opposite charge located at
the · -im"""" noint hP-hinrl thP nlane defined bv the nosition of the conductor.
,
,.,,.
58 Chapter 2 Boundary-Value Problems in Electrostatics: l-SI
I
I
I'<;-<!>= 0
I
~ _____ _.
-q
._ _____ lI _____ __
I q
;;;; q
I
T Fi<mre 2.1 Solution bv method of
1 images. The original potential problem
! is on the left, the equivalent-image
problem on the right.
2 2 Point Charge in the Presence of a Grounded

Conducting Sphere
As an IllustratiOn of the method of images we consider the problem illustrated
in Fig. 2.2 of a point charge q located at y relative to the origin, around which is
centered a grounded conducting sphere of radius a. We seek the potential <l.>(x)
such that <1.>( IxI a) 0. By symmetry it is evident that the image eharge q'
(assuming that only one image is needed) will lie on the ray from the origin to
the charge q. If we consider the charge q outside the sphere, the image position
y · will lie inside the sphere. The potential due to the charges q and q' IS:
q/47TE0 q '/47TE0
lx- Yl lx- Y'l
We now must try to choose q' and Iy' Isuch thatthis potential vanishes at lx I - a
If n is a unit vector in the direction x, and n' a unit vector in the direction y, then
<1.>( X) = + --'----"-~ (2.2)

lxn- yn'l lxn- y'n'l
If x 1s factored out of the first term and y' out of the second, the potential at
x = a becomes:
q '/47TE0
(2.3)
/p
I
/ -.J
I ~a
t.. //q~
- a
.....
/ I
y
A / Figure 2.2 Conducting sphere of radius

a, With charge q anel Image charge q .
Sect. 2.2 Point Charge in the Presence of a Gronnded Condncting Sphere 59
From the form of (2.3) it will be seen that the choices:

q q' y a
a a
make <l>(x = a) = 0, for all possible values of n • n '. Hence the magnitude and
position of the image charge are
2
q' = --q, y' =- (2.4)
y y
We note that, as the charge q is brought closer to the sphere, the image charge
grows in magnitude and moves out from the center of the sphere. When q is just
outside the surface of the sphere, the image charge is equal and opposite in
magnitude and lies just beneath the surface.
Now that the image charge has been found, we can return to the original
problem of a charge q outside a grounded conducting sphere and consider various
effects. The actual charge density induced on the surface of the sphere can be
calcnlated from the uormal derivative of <ll at the smface·
(2.5)
2
47Tay ( 1 + -a 2
a
- 2 -y cos 'Y
)3/2
y2
where y is the angle between x andy. This charge density in units of -q/47Ta 2 is
shown plotted jn Fig 2 3 as a functjou of l' for two values of yla The con centra-
"
Figure 2.3 Surface-charge density a mduced on the grounded sphere of radms a as a
result of the presence of a point charge q located a distance y away from the center of
the sphere. a is plotted in units of -ql47ra2 as a function of the angular position y away
from th@ radius to th@ charge for y - 2a, 4a. Inset shows lines of force for y - 2a.
¥ dF ~ (cr /2•olda
2
I d
,._ ..... _l .q
~ Figure 2.4
tion of charge in the direction of the point charge q is evident, especially for
yla = 2. It is easy to show by direct integration that the total induced charge on
Lll(; >jJHClC lo . LV LllC ; Luuc v1 u1e image cnarge, as il must oe, accoroing
to G~n"'' hw
r
~·
"'" 'v'"" .. 0 v11 LHc '1 ""u uc ueo in oillerent ways. vne
(the easiest) way is to write dow; immediately the force between the charge q
and the image charge q'. The distance between them is y - y' = y(1 - a2 /y 2 ).
Hence the attractive force, according to Coulomb's law, is:
2 \ 3 o'
2 -2
1 q a ( a
IFI = Ll~c. ,2 ( ;- 1 - 2l v
(2.6)
7--;
VI
,., ' 1 ,, £ ' ' ' '

'5~ . -~ 1Hv <Vl~v 10 UH LV LllC >jJHClC l l
UHv• Ov craw,~
1s proport10nal to the mverse square of the d1stance awayTrom the surface of the
sphere.
The alternative method for obtaining the force is to calculate the total force
acting on the surface of the sphere. The force on each element of area da is
( a 2 !2E 0 ) da, where a is given by (2.5), as indicated in Fig. 2.4. But from symmetry
it is clear that only the component parallel to the radius vector from the center
ot the sphere to Q contnbutes to tne total torce. Hence the total force actmr> on
., · r~"
{.
\. "1'
,] _.,
'Yr +" '· '5
_, . .,_
'I! 10 5' •v vy cuv -'.
<at.
L
F -
I I-
q2 ( r( 2r I
q_
321T2Eoa2 y
1 - I!_
y2
cos y
( 1 + a2 - 2a cos y
l y
r d!l (2.7)
Integration immediately yields (2.6).

Th~e whnl<e _,. inn h~' h~e~en h~'Pcl n thP nnrlP '.
'0
th ot thP ~~;nt
I"'
l,;lldl ~" '1 I> Lll(; . '"'"LudHy, llle resuns apply equauy ror lne cnarge
f1 ;wc;Ao th<e 'nhPrP Th<e nnlv rh~nop >Vu ;c ;n thP .~hov"P rlPnc;tu
·o
(2.5), where the normal derivative out of the conductor is now radially inward,
implying a change in sign. The reader may transcribe all the formulas, remem-
bering that now y < a. The angular distributions of surface charge are similar to
those of Fig. 2.3, but the total induced surface charge is evidently equal to -q,
independent of y.
:l.i I rp_ in thP nt

v
n l'h
·o
ln-.,ulntorl
--,
conaucun!! ..,onere
In the preceding section we considered the problem of a point charge q near a
grounded sphere and saw that a surface-charge density was induced on the
Sect. 2.4 Point Charge Near a Conducting Sphere at Fixed Potential 61
sphere. This charge was of total amount q' = -aq/y, and was distributed over
the surface in such a way as to be in equilibrium under all forces acting.
Tf wP. wi'h to ron<iclP.r thP. nrohlP-m of ~n in"Jl~tP..-1 · '" 'nhP.rP. with
total charge Q in the presence of a point charge q, we can build up the solution
for thP. i~l hv line~r · · Tn ~n ll sense_ we c~n im~P"ine
d. "LL d. .> .L /. "d. • .1. . • .>" "- ~
wac "'-'ow>' " " " "''-'IS'' , opu'-'''-' \"""no '1 u>ocw
over its surface). We then disconnect the ground wire and add to the sphere an
U< l-H<H!;C ~\.t 'f )• HH> UHH!;> lHC lUldl l-H<1l !;C Ull lliC >J-'liCl C UJ-' lU \.t·
To find the potential we merely note that the added charge (Q - q') will dis-
tribute itself uniformly over the surface, since the electrostatic forces due to the
point charge q are already balanced by the charge q'. Hence the potential due
to the added charge (Q - q') will be the same as if a point charge of that mag-
nitude were at the oriQ"in. at least for noints ont<i.-1,. the snhere.
The potential is the superposition of (2.1) and the potential of a point charge
(Q - q') at the origin:
a
Q +- q
1 q aq y
<l>(x) + (2.8)
"11l'Eo IX Yl aL lXI
y X- -
y2
y
- -
The force acting on the charge q can be written down directly from Coulomb's
law. It is directed along the radius vector to q and has the magnitude:
F = ~ !1_ Q _ qa-~Ly a) ~
(2.9)
41l'Eo l y(y2 - a2)2 y
1n rne nmn or y .->.-> a, rne rorce reauces 10 Lne usua1 ~ouwmo s 1aw wr LWO smau
charged bodies. But close to the sphere the force is modified because of the
mduced cnarge drstnbutton on tne surrace or tne spnere. J:<rgure L.:J snows me
force as a function of distance for various ratios of Q!q. The force is expressed
in units of q 2141TEoV 2 ; positive (negative) values correspond to a repulsion (at-
,\ H <h ,J, -L --' t. ,], rl <h f, ;o
.·" II d~r E .". 'f h:rr h 'J Q ?' h . ~ , h
attractive at a rstances. ven 1 t e c arge rs t e same srgn as q, owever,
th~ fAr~~ '- ~ttr~~·; ·~ ~· u~r· ~lAo~ r!;ot~n~PO Tn th~ ];m;t Af n '>"> n thP
•M
-J • ~ .,,
point of zero force (unstable equilibrium point) is very close to the sphere,
namely, at y = a(1 + ~yq!Q). Note that the asymptotic value of the force is
auameu as soon as llle cuarge q is more Lnan a <ew rauii away uom llle spuere.
This example exhibits a general property that explains why an excess of
charge on the surface does not immediately leave the surface because of mutual
,]," ,ft~ · · ,]~\, Ll.oo< ~ ~n .L >tAf·h~rnP;o ..1
f;~m the surface, the image f~;ce tends to attract it back. If sufficient work is
riAnP Af ~h~r~~ ~~n he> ..1 frAm the> onrfo~~ tA ' "' ' The> mClrlc
: ' • -:=' . .
1Uncuon 01 a meLatiS m targe pan JU>L lilt: worK uout: agam>L Lilt:"' ve
force to remove an electron from the surface.
2.4 Point Charf(e Near a ConductinK Sphere at Fixed Potential

Another problem that can be discussed easily is that of a point charge near a
rl •t.' h ,]rJ >t h rl •t-' '1 l7 Th >t" '1 ;o t~ <~ •< fAr
62 Chapter 2 Boundary-Value Problems m Electrostatics: I SI
5
' I
I
I
4
:
I
I Qlq- 3
31-
I
"'' 2 ?
I
I
I
/
q"
lc-
I
I
I
I 1
I
I~
v;u
_Q
_[ I I I I I I I
1 I2 3 4 5
ya~
I
VI
-1 1- I
I
I
I
-2 1-
~
i I
-3 - I
I
I
I
-4 r-- I
i
-5 - I
Figure 2.5 The force on a point charge q due to an insulated, conducting sphere of
radius a carrying a total charge Q. Positive values mean a repulsion, negative an
attraction. The asymptotic dependence of the fmce has been divided ouL 47TE0 Fy'1q 2 is
plotted versus y/a for Q/q - 1, 0, 1, 3, Regardless of the value of Q, the force is
always attractive at close distances because of the induced surface charge.
the charged sphere, except that the charge (Q- q') at the center is replaced by
a charge ( Va ). Th1s can be seen from (2.8), smce at IxI a the first two terms
cancel and the last term will be equal to V as required. Thus the potential is
<l>(x)
47TEolx Yl ~ ]+ Va
lxl
(210)
The force on the charge q due to the sphere at fixed potential is
F :2 [va I
47TE0 (y 2
qaia f
-
2
l Y
y
(2.II)
For corresponding values of 47TEo Valq and Q/q this force is very similar to that
of the charged sphere, shown in Fig 2 5 although the approach to the asymptotic
value (Faqly 2 ) is more gradual. Fm Va >> q, the unstable equilibrium point has
the eqmvalent location y - a(l + i V q/47TEo Va).
2.5 Conducting Sphere in a llnijrirm Electric /i'ield

by ldethod of/mages
As a final example of the method of images we consider a conducting sphere of
radius a in a uniform electric field E 0 • A uniform field can be thought of as being
Sect. 2.5 Conductmg Sphere m a Umform Electric Field by Method of Images 63
produced by appropriate positive and negative charges at infinity. For example,

if there are two charges + Q, located at positions z = + R. as shown in Fig. 2.6a.
then in a region near the origin whose dimensions are very small compared toR
thet e is an approximately constant electric field £ 0 2Qf47TE 0 R 2 par aile! to the
z axis. In the limit as R, Q--'> co, with Q/R 2 constant, this approximation becomes
If now a conducting sphere of radius a is placed at the origin, the potential
will be that due to the charges +Q at +R and their images +Qa!R at z
Q141TE0
2 rR cos 11) 112 (2.12)
aQ141TE0
112
a4 2a 2 r
+----cos IJ )
2
R R
where has been expressed in terms of the spherical coordinates of the obser-
vation point. In the first two terms R is much larger than r by assumption. Hence
we ean expand the radicals after factoring out R 2 • 8irnilarly, in the third and
fourth terms, we can factor out r2 and then expand. The result is:
= _1_ 2Q 2Q a3
- - r cos IJ + - - cos IJ + .. · (2.13)
41TE R2 Rz r2
where the omitted terms vanish in the limit R --" oo. In that limit 2Q/41TE0 R 2
becomes the applied uniform field, so that the potential is
E0 (r ;:) cos 0 (2.14)
The first term (- E 0 z) is, of course, just the potential of a uniform field £ 0 which
could have been written down directly instead of the first two terms in (2.12).
p .... __ "-,::_
+Q --- /0
~
-- --- -Q
z= -R
/ z=R
(a)
/f"p
~ /
+Q /_~Q }/\ \9 -Q
' - -R /\"'- >- R
z=- ~ z- ~
~ av - R
Figure 2.6 Conducting sphere in a uniform electric field by the method of images.
The second is the potential due to the induced surface-charge density or, equiv-
alently, the image charges. Note that the image charges form a dipole of strength
D - Qa!R x 2a2 1R - 4n"e E0 a 3 The induced smface charge density is
u = -E0 ~;I
3 r-a
= 3E0 E 0 cos 0 (2.15)
We note that the surface mtegral of thts charge denstty vamshes, so that there IS
no difference between a grounded and an insulated sphere.
2.6 Green Function for the Sphere; General Solution

for the Potential
In preceding sections the problem of a conducting sphere in the presence of a
point charge was discussed by the method of images. As mentioned in Section
1.10, the potential due to a unit source and its image (or images), chosen to satisfy
homogeneous boundary conditions, is just the Green function (1.43 or 1.45) ap-
propriate for Dirichlet or Neumann boundary conditions. In G(x, x') the variable
x' refers to the location P' of the unit source, while the variable xis the point P
at which the potential is being evaluated. These coordinates and the sphere are
shown in Fig. 2.7. For Dirichlet boundary conditions on the sphere of radius a
the Green function defined via (1.39) for a unit source and its image is given by
(2.1) with q ~ 41TEo and relations (2.4). Transforming variables appropriately,
we obtain the Green function:
G(x, x') = lx x' I (2.16)
oP'
x·/! .liP
/ ~I /I
I ! ;.__ "( I'Y/ i
~~
I
a :
/\- -- I I
y
\ / q>
'
,i
1/---- I
X "' \/[ --..J
/ ,\I
/ ~
/.
Figure 2.7
Sect. 2. 7 Conducting Sphere with Hemispheres at Different Potentials 65
In terms of spherical coordinates this can be written·
G(x, x') (2.17)
where y is the angle between x and x'. The symmetry in the variables x and x'
IS obvwus m the form (2.17), as IS the conditiOn that G 0 1f either x or x IS on
the surface of the sphere.
For solution (1.44) of the Poisson equation we need not only G, but also
aG/an '. Remembering that n' is the unit normal outward from the volume of
interest (i.e., inward along x · toward the origin), we have
aG
(2.18)
an' x'=a a(x 2 + a2 - 2ax cos y) 312
[Note that this is essentially the induced surface-charge density (2.5).] Hence the
solution of the Laplace equation outside a sphere with the potential specified on
its surface is, according to (1.44),
1
<t>(x) = 4-
J(a, f)', cp ')
a(x - a )
(x 2 +a-
2 2
2 2
)3/2 dil' (2.19)

7T axcosy
where dil' is the element of solid angle at the point (a, f)', p ') and cosy =
cos f) cos f)' + sin f) sin f)' cos( <f> - <f> '). For the interior problem, the normal
denvatJve IS radially outward, so that the sign of aGJdn' IS opposite to (2.18).
This is equivalent to replacing the factor (x 2 a2 ) by (a 2 x 2 ) in (2.19). For a
problem with a charge distribution, we must add to (2.19) the appropriate integral
in (1.44), with the Green function (2.17).
2. 7 Conducting Sphere with Hemispheres at Different Potentials

As an example of the solution (2.19) for the potential outside a sphere with
prescribed values of potential on its surface, we consider the conducting sphere
of radius a made up of two hemispherical shells separated by a small insulating
ring. The hemispheres are kept at different potentials. It will suffice to consider
the potentials as + Y, sinee arbitrary potentials ean be handled by superposition
of the solution for a sphere at fixed potential over its whole surface. The insu-
lating ring lies in the z = 0 plane, as shown in Fig. 2.8, with the upper (lower)
hemisphere at potential + V ( V).
+V--. / ~
/
....... \
::----- - --~
/ ~ y
""""
__/( ~J
""-. _/'--V
~
X t'il!nre :Z.!S
From (2.19) the solution for <P(x, {}, <P) is given by the integral:
<P(x, {}, <P) = v7T

4
r~ d¢'
rO
0
ul 0
d( cos{}')
(2.20)
' 2
uv• u2')
J-I d( cos {}')
(a 2 + x 2 - 2ax cos y) 312
By a suitable change of variables in the second integral (e' ~ 7T - {}'' <P' ~

.L
'I'
I 1 _\
.,, <\, :n J...~ A~" in the> fr.rm•
.? ,,
Ya\X
<P(x, {}, <P) =
47T
aT Jo d¢' Jo d(cos {}')[(a 2 + x 2 - 2ax cos y)- 312
(2.21)
-(a 2 + x 2 + 2ax cosy) 3/2]
J:lecause onne com plicatea aepenaence or cos yon me '<b' '(!J ,<fJ 1 auu \ u, cp),
equation (2.21) cannot in general be integrated in closed form.
As a special case we consider the potential on the positive z axis. Then
A~n A~C D' c;nAO D n ThP i >tiC\n i< Pl >nr ~ncl thP ial can 0
be shown to be
(z' a')
<P(z) = V 1 - (2.22)
zVz + a2
2
fil .<. u, ~
' <v ""
~
TT
' ao '~'iuu~u,
~ .L"
U'H~ UO
1
'5'
'. ,. o·
asymptotically as L .1.. ~. ' L "
\U A J <HT
'
- - a2) r~
va( (x 2 ' '
2 ?12
--312
<P(x, {}, <P) - d¢
2 d( cos{} )[(1 - 2a cosy)
4 1rx+a o o (2.23)
- (1 + 2acos y)- 312 ]
where a ax/(aL + xL). We observe that in the expans10n ot the ramcats onty
odd powers of a cos y will appear:
f(1 - ?rv r.o< ool- 312 - (1 + 2a cos -v) 312 1 - 6a cos -v + 35a3 cos 3 -v + ··· (2.24)
It is now necessary to integrate odd powers of cos y over d¢' d( cos {}'):
(2~ {1
cl .L' A{, oC tJ'\ A~C ·A~C D
.\
JO "'l' )o I
{2~ {1
(2.25)
7T
d<fJ )o a~ cos cos· y 4 cos tl~" cos·' 1:1)
)o 1:1 )
If (2.24) and (2.25) are inserted into (2.23), the potential becomes
3Va 2 x 3(x 2 - a2) a2x2

I
35
<P(x, {}, <P) = ~ 2 , 2 2\512 ) cos (} 1 + '1A ( 2 2\2
(3 - cos2{}) + · · ·
'\· I \' I
(2.26)
Sect 2 8 Orthogonal Functions and Expansions 67
We note that only odd powers of cos fJ appear, as required by the symmetry of
the problem. If the expansion parameter is (a 2 /x 2 ), rather than cl, the series takes
on the form:
3Va2 7a 2 1 -5 cos3 (! 3 '
<P(x fJ ri.) = - -
2
cos fJ - - -2 1 - -cos fJ + · · · (2.27)
' ' '+' 2x l2x 2 2
For large values of ria this expansion converges rapidly and so is a useful rep-
resentation for the potential. Even fo1 x!a 5, the second term in the series is
only of the order of 2%. It is easily verified that, for cos fJ = 1, expression (2.27)
agrees with the expansion of (2.22) for the potential on the axis. [The particular
choice of angular factors in (2.27) is dictated by the definitions of the Legendre
polynomials. The two factors are, in fact, P 1 (cos fJ) and P 3 (cos fJ), and the expan-
sion of the potential is one in Legendre polynomials of odd order. We establish
this in a systematic fashion in Section 3.3.] Further consideration of both the
exterior and interior prohlem of the two hemispheres is found in Problem 2 22
2.8 Orthogonal Functions and Expansions

The representation of solutions of potential problems (or any mathematical phys-
ics problem) by expansions in orthogonal functions forms a powerful technique
that can be used in a large class of problems. The par tic alar orthogonal set chosen
depends on the symmetries or near symmetries involved To recall the general
properties of or tlrogonal functions and expansions in terms of them, we consider
an interval (a, b) in a variable g with a set of real or complex functions U,,(g),
n = 1, 2, ... , square integrable and orthogonal on the interval (a, h). The ortho-
gonality condition on the functions Un( g) is expressed by
(2.28)
If n m, the integral is nonzero. 'Ne assume that the functions are normalized
so that the mtegral IS umty. Then the functiOns are Said to be orthonormal, and
they satisfy
(2.29)
An arbitrary function f(g), square integrable on the interval (a, b), can be
expanded in a series of the orthonormal functions U,(g) If the namber of terms
in the series is finite (say 1Y),
!W .__. L a,U,(O (2.30)

n-1
then we can ask for the "best" choice of coefficients a, so that we get the "best"
representation of the function f(g). If "best" is defined as minimizing the mean
(2.31)
it is easy to show that the coefficients are given by
an = f U~(g)f(g) d{; (2.32)
where the orthonormality condition (2.29) has been used. This is the standard
result for the coefficients in an orthonormal function expansion.
If the number of terms N in series (2.30) is taken larger and larger, we in-
tuitively expect that our series representation of f({;) is "better" and "better."
Our intuition will be correct provided the set of orthonormal functions is com-
plete, completeness being defined by the requirement that there exist a finite
number N 0 such that for N > N 0 the mean square error M N can be made smaller
than any arbitrarily small positive quantity. Then the series representation
(2.33)
n=l
with a" given by (2.32) is said to converge in the mean to f({;). Physicists generally
leave the difficult job of proving completeness of a given set of functions to the
mathematicians. All orthonormal sets of functions normally occurring in math-
ematical physics have been proven to be complete.
Series (2.33) can be rewritten with the explicit form (2.32) for the coef-
ficients an:
(2.34)
sum ot b!lmear terms u ~( c' l u "( t l must exist onlv m the neighborhood ot
.. .~
Lllld11LY <.,U11UlllU11 \L.L7 ), CX<CC)Jl llldl LHC lUIC> Ul LHC <CUHI y; ~
~ncl th<e · inclteY n h~vP hPP.n ·

UH.o H<VO l HUHV UO V L L' ill 10 llliO 0111100 <111U , <111 -
sion in terms of them heinP" a Fourier series. If the interval in xis r -n/2 n/21 thte
. . ·o· -.o· . . . '·

1ne ~ene~ eqmva1ent 10 (L.-'-' J IS cuswmaruy wnuen m rne rorm:
, "' Lmnx . 1 L7rmx
zrt.o T L.. .t-l.m I.OU>I I ' Dm >1111 I
u u
where
12
Am = -2a J" -a/2
f(x) cos (2 7rmx
a
) dx (2.37)
Bm = -2fa/Z f(x) sin (2 7rmx ) dx
a -a/2 a
Sect. 2.8 Orthogonal Fnnctions and Expansions 69
If the interval spanned by the orthonormal set has more than one dimension,
formulas (2.28)-(2.33) have obvious generalizations. Suppose that the space is
s
two-dimensional, and the variable ranges over the interval (a, b) while the
variable n has the interval (c. d). The orthonormal functions in each dimension
are U,(S) and V,(7J). Then the expansion of an arbitrary function f(s, 77) is
f({, 71) - 2: 2: a,, [~,(g)V,(71) (2 38)
n m
where
(2.39)
Ja Jc
If the interval (a, b) becomes infinite, the set of orthogonal functions Un(s)
may become a continuum of functions, rather than a denumerable set. Then the
Kronecker delta symbol in (2.29) becomes a Dirac delta function. An important
example is the Fourier integral. Start with the orthonormal set of complex
exponentials,
(2.40)
m- 0, +1, +2, , on the interval ( a/2, a/2), wjtb the expansion·
1
L
00
f(x) = ---r= Amei(2mnx!a) (2.41)
where
A, (2.42)
Va J-a/2
Then let the interval become infinite (a~ oo), at the same time transforming
dk (2.43)
A, --" - A(k)
a
I he resultmg expansiOn, eqmvalent to (2.41), IS the fourter mtegral,
f(x) = , ;,__
V 27T
r -oo
A(k)e'kx dk (2.44)
where
A(k) = -
vz;: J -oo
e >kxf(x) dx (2.45)
The orthogonality condition is
J_ ~· e'(k-k')x dx = 8(k - k') (2.46)

b7T - 00
while the completeness relation is
__!__
271'
r
J -oo
e;k(x-x') dk - o(x - x') (2A7)
These last integrals serve as convenient representations of a delta function. We

note in (2A4)-(2A7) the complete equivalence of the two continuous variables
x and k.
2.9 Separation of Variables; Laplace Equation

in Rectangular Coordinates
The pat tial diffet entia! equations of mathematical physics are often solved con-
veniently by a method called separation of variables. In the process, one often
generates orthogonal sets of functions that are useful in their own right. Equa-
tions involving the three-dimensional Laplacian operator are known to be sep-
arable in eleven different coordinate systems (see Morse a lUi Feshhach, pp. 509,
655). We discuss only three of these in any detail-rectangular, spherical, and
cylindrical-beginning with the simplest, rectangular coordinates.
The Laplace equation in rectangular coordinates is
-+ -+ -= 0 (2,48)
ax
2 2 2
ay az
A solution of this partial differential equation can be found in terms of three
ordinary differential equatiOns. all of the same form, by the assumptiOn that the
potential can be represented by a product of three functions, one for each
coordinate:
<l>(x, y, z) = X(x)Y(y)Z(z) (2,49)
SubstitutiOn mto (2.48) and dtvtston of the result by (2.49) yields
1 d 2X 1 d2Y 1 d 2Z
- - - -2 + - - - - + - - - -2 = 0 (2.50)
X(x) dx Y(y) dy2 Z(z) dz
where total derivatives have replaced partial derivatives, since each term involves
a function of one variable only. If (2.50) is to hold for arbitrary values of the
independent coordinates, each of the three terms must be separately constant:
i Lt A'" .2
~
X dx 2
1 d2Y -{32
--- (2.51)
r ay
.0•
X U ~
2
I
Z dz 2
where
If we arbitrarily choose r2 and {3 2 to be positive, then the solutions of the three

ordinary differential equations (2.51) are e±;ax, e";f3y, e"~z. The potential
(2.49) can thus be built up ftotn the product solutions.
2 2-
Sect. 2.9 Separation of Variables; Laplace Equation in Rectangular Coordinates 71
At this stage a and f3 are completely arbitrary. Consequently (2.52), by linear

superposition, represents a very large class of solutions to the Laplaee equation.
To determine a and f3 it is necessary to impose specific boundary conditions
on the potential. As an example, consider a rectangular box, located as shown
in Fig. 2.9, with dimensions (a, b, c) in the (x, y, z) directions. All surfaces of the
box are kept at zero potential, except the surface z = c, which is at a potential
v(x, y). It IS reqmred to find the potentml everywhere ms1de the box. Startmg
with the requirement that <P = 0 for x = 0, y = 0, z = 0, it is easy to see that the
required forms of X, Y, Z are
X= sin ax }
Y sin{Jy (2.53)
Z = sinh( V,-a,- 2 z)
2 -:-+---c/3=
To have nm• satJsfymg all the boundary conditiOns
except one,
<Pnm = sin(anx) sin(f3mY) sinh( 'YnmZ) (2.55)
The potential can be expanded in terms of these <Pnm with initially arbitrary
coefficients (to be chosen to satisfy the final boundary condition):
<P(x, y, z) = L
n,m=l
Anm sin(anx) sin(f3mY) sinh(rnmZ) (2.56)
There remains only the boundary condition <P = V(x, y) at z = c:
(2.57)
z=c r
t
I = V(X,Y )
/ / p-1=0
1=0-
- - --
y.., IJ
y
/ /
X 'a I' v
/
/ t.n Figure 2.9 Hollow, rectangular box
with five sides at zero
t~ . "'~ (. r\
tiaL while
~ t~ ~
/
potential ct> = V(x, y ).
"
• T "T
1 ni~ i~ ju~t a uuume ruurier series wr tne iunction v~x, y ). Consequently the
coefficients Anm are given by:
nm 4
absinh(Ynmc)Jo
r rl (
Jo
,-!,
--~
HI.
'-' 1
,\ '.I
'''"' 1
,.\ ' .f 0
'"m-' 1
,\ I'! <:0\
, .. ''
T£d. .I L L '.1 ~0££. "

;um< U~A HUO < UU<~<' --rrrn::rrz;=u- aU MA >lUC>, lilt;
required solution for the potential inside the box can be obtained by a linear
o - " ' l ' ' ' ~f o;v o~J,t:~ ' f, e~ '~
' '1'
' ' <:L\
, . . '!
"
, . . J,.
<:0\ {'I {'I
The problem of the solution of the Poisson equation, that is, the potential inside
the box with a charge distribution inside, as well as prescribed boundary condi-
tions on the surface, requires the construction of the appropriate Green function,
according to (1.43) and (1.44). Discussion of this topic will be deferred until we
have treated the Laplace equation in spherical and cylindrical coordinates. For
the moment, we merely note that the solution given by (2.56) and (2.58) is equiv-
alent to the surface integral in the Green function solution (1.44).
2.10 A Two-Dimensional Potential Problem;

uiun OJ a rourier ~eries
vve now urieuy Lne sowLion oy separation ot vanatJies 01 tne two-

dimensional Laplace equation in Cartesian coordinates. By two-dimensional
problems we mean those in which the potential can be assumed to be indepen-
dent of one of the iinates sav. 7. This is n~n~lhr nnlv ~n ~- · •~tinn hnt
may hold true to high accuracy, as in a long uniform transmission line. If the
notential is indenendent of 7. the basic solutions of the nrevinm sertinn '
to the products
e e
where a 1s any real or complex constant. The imposition of boundary conditions
on the potential will determine what values of a are permitted and the form of
the linear superposition of different solutiOns required.
A simple problem that can be used to demonstrate the separation of variables
wcnnique anu a1so w eswonsn connecnon wnn the use ot complex vanables 1s
;n,1;~·>1M1 ;n r;;~ ') 1 n Th~ ·.,J . tl. . n ~ ~ ~ {) . ' . '
·o t' -o- ' ' J ' '
subject to the boundary conditions that lrCfP II · r ~f th~ h~o;~ o~Jnt;Ano
shows that a is real and that, to have the potential vanish at x = 0 and x = a
for all II and aS V ~ 00 thP ]ine<Jr ~nmhin~tirmo ~r~ o-"y oin( ~v\ mith
a= nTTia. The linear combination of solutions satisfying the boundary conditions
on three of the four boundarv surfaces is thus
\' • J!
The coefficients An are determined by the requirement that <P = V for y = 0,

0 < x < a. As discussed in Section 2.8, the Fourier coefficients are
2 {a
An 1 lJJ(x U) sm(n7Tx/al dx 12.om
U JO
Sect. 2.10 A Two-Dimensional Potential Problem; Summation of a Fourier Series 73
With <P(x, 0) = V, one finds
4V 1 .
<P(x, y) = - - exp( -n1ry!a) sm(n1rxla) (2.61)
7r n
For small values of y many terms in the series are necessary to give an accurate
appreciable. The potential rapidly approaches its asymptotic form given by the
first term,
a
constant, provided the first term in the series is nonvanishing. The coefficient A 1
. .
sets in for y 2 a, regardless of the complexities of <P(x, 0). This is shown quan-
11 ~0
potential, the dotted, the first term (2.62). Close to the boundary (yla = 0.1) the
curves differ but for = 0.5 the · form is an
74 Chapter 2 Boundary-Value Problems in Electrostatics: I -Sl
1.0 r-
I
/
/ --- ' '
\
I \
no / \
~-\\
y/a- 0.1
/(
0.6 f-
77
~ (%,
lF
y)
I/ \\
0.4 _I I \ \
I I I
I
f- I \
! - I
I
~
Yl" 'u.~ I
/
I I
0.21-- I I
I I
uo
VI
//
I I I
x/a
I I I I
""\ I"\
1
Figure 2.11 Potentials at yla 0.1, 0.5 (along the dashed lines of l:<tg. L.lU) as
functions of xi a. The solid curves are the exact solution; the dashed curves are the first
term m me senes somrion lLO 1)·
~ • • 'L 1. .l • :~
lllCl<O alC l'VUlJCl wuc ~uu U~ 0 b'
r-1"'"<1 form. The series in (2.61) is one of them. We proceed as follows. Observing
that sin(}= Im(e;"), where Im stands for the imaginary part. we see that (2.61)
can be written as
<t>(x, v) = 4V Im L _!_ eUnnla)(x+!y)

n odd H
"
With the definition,
z = eUnla)(x+ly) (2.63)
this can be put in the suggestive form,
<t>(x, y) =
4
V Im L z
7T n odd n
At this point we can perhaps recall the expansion,*
' _, 1 __,
...
mu 1 -"
-t----zj £ zLo -r j£ ;!L'T
*Alternatively, we observe that~, 1 L"in) 2-;, oL' 'TT(T --z: . .....5 w,;vu wtn giv=
'5'" Z"ln - -lnl1 - Zl.
Sect. 2.11 Fields and Charge Densities in Two-Dimensional Corners and Along Edges 75
Evidently,
L
noddn
Z" 1 (1 z)
-=-In
2
+
1-Z
and
2v Im [ In
<l>(x, y) = --;:- (1 + z)l
1
z _ (2.64)
Since the imaginary part of a logarithm is equal to the phase of its argument, we
consider
1 +Z (1 + Z)(1 - Z*) 1 - IZI 2 + 2i Im Z
= =
1- Z ll - Zl 2
11 - Zl 2
The phase of the argument of the logarithm is thus tan- 1 [2 Im Z/(1 - IZI 2 )].
With the explicit form (2.63) of Z substituted, it is found that the potential
becomes
I
. 1TX
sm-
2V a
<l>(x, y) = - tan- 1 (2.65)
1T . 1TY
smh-
a
The branch of the tangent curve corresponds to the angle lying between 0 and
1r!2. The infinite series (2.61) has been transformed into the explicit closed form
(2.65). The reader may verify that the boundary conditions are satisfied and that
the asymptotic form (2.62) emerges in a simple manner.
The potential (2.64) with Z given by (2.63) is obviously related to functions
of a complex variable. This connection is a direct consequence of the fact that
the real or the imaginary part of an analytic function satisfies the Laplace equa-
tion in two dimensions as a result of the Cauchy-Riemann equations. As men-
tioned at the beginning of the chapter, we omit discussion of the complex-variable
technique, not because it is unimportant but for lack of space and because
completely adequate discussions exist elsewhere. Some of these sources are listed
at the end of the chapter. The methods of summation of Fourier series, with
many examples, are described in Collin (Appendix A.6).
2.11 Fields and Charge Densities in Two-Dimensional

Corners and Along Edges
In many practical situations conducting surfaces come together in a way that can
be approximated, on the small scale at least, as the intersection of two planes.
The edges of the box shown in Fig. 2.9 are one example, the corners at x = 0,
y = 0 and x = a, y = 0 in Fig. 2.10 another. It is useful therefore to have an
understanding of how the potential fields, and the surface-charge densities be-
have in the neighborhood of such sharp "corners" or edges. To be able to look
at them closely enough to have the behavior of the fields determined in functional
form solely by the properties of the "corner" and not by the details of the overall
configuration, we assume that the "corners" are infinitely sharp.
planes intersect at an angle {3. The planes are assumed to be held at potential V.
conductors or
possibly configurations of charges that specify the potential problem uniquely.
we are m
origin, but not in the absolute magnitudes, we leave the "far away" behavior
dinates. In terms of the polar coordinates (p, 2 =
Since the two terms are separately functions of p and ) = Ao + Boup"~lthough. not central to our present purpose, we note the general solution
of the Laplace equation in two dimensions when the full azimuthal range is per-
llllllt:U ' " • lUI , lUI lllt: ]JULt:llll<1l Ut:LWt:t:ll LWU "J , f1 U
and p = b, on which the potential is given as a function of rf>. If there is no

An A.. ;t ;o ogTu th<>t 'hP <> nAo;t;uP AT nP<Nt;uP ;ntPnPT AT ?PTA tA
' ·o ·o· ·o
ensure that the potential is single-valued. Furthermore, for v = 0, the constant
B 0 in (2.70) must vanish for the same reason. The general solution is therefore
j',
,j' ' "
'
~ , I.
"''
~\1'• "
'!') uo '-
uo L
m I' £...J, Unf'
n=1
~n
OW\"'f' ~n)
\
00 (2.71)
+ ~ bnp -n sin(nr/> + f3n)
~1
If the origin is included in the volume in which there is no charge, all the bn are
. . ..
LCl U. Vlll J a diiU ]JU>l Ll VC pUWCI> Ul J-1 appeaL H LllC u I> ,
the bn can be different from zero. In particular, the logarithmic term is equivalent
tA ~ ];n., <>h<>T<TP An thP <>v;o .,,;th <>h<>T<TP <1Pno;tu nPT nn;t lPn<Tth 1 - ~ 7 ~c.h~ <>O
-o r ·o -v v'
I> VV<Oll "-llUVVll.
For the situation of Fig. 2.12 the azimuthal angle is restricted to the range
n a '1'1, j. rJ, . m l7 j', ., 11 ~= n
c- "" c- "" n "J. A..
and ; = ~-. This requires that b0 = B 0 = 0 in (2.70) and b ~ 0 and A :: 0 in

'"
(2.69). Furthermore it reauires that v be chosen to make sin( vB) = 0. Hence
m7T
v = B , m = 1, 2, ...
and the general solution becomes

00
<P(p, 1>) = v+ ~ ampmrr/{3 sin(m 7T1>/{3) (2.72)

m=l
The still undetermined coefficients am depend on the potential remote from the
corner at p = 0. Since the series involves positive powers of p"'1f3, for small enough
p only the first term in the series will be important.* Thus, near p = 0, the po-
tential is approximately
lJJ~p, 1>) v -t- a 1p sm~ 7Tlf>l p) ~L.U)
The electric field comoonents are
a/R)

F(n m ) - --~
' ap /3
~L.I'+)
J;' (. A..\
1a<P 7ra1 (rr/{3)-1 •ncf. A.. I P.\
Am R -,
'"""'
~,
n r
'
*Here we make a necessary assumption about the remote boundary conditions, namely, that they
are such that the coefficient a, is not zero. Ordinarily this is of no concern hut special symmetries
might make a" or even a2 , etc., vanish. These unusual examples must be treated separately.
The surface-charge densities at cp = 0 and cp = f3 are equal and are approximately

(2.75)
The components of the field and the surface-charge density near p = 0 all vary
with distance as pE "'f3l 1 • This dependence on p is shown for some special cases
in Fig. 2.13. For a very deep corner (small {3) the power of p becomes very large.
Essentially no charge accumulates in such a coriJei. Fm p rr (a fiat surface),
the field quantities become independent of p, as is intuitively obvious. When
f3 > 7T, the two-dimensional corner becomes an edge and the field and the surface-
charge density become singular asp --7> 0. For f3 = 27T (the edge of a thin sheet)
the singularity is as p 112 • This is still integrable so that the charge within a finite
distance from the edge is finite, but it implies that field strengths become very
large at the edges of conducting sheets (or, in fact, for any configuration where
f3 > 7T).
The preceding two-dimensional electrostatic consJdcratJons apply to many
three-dimensional situations, even with time-varying fields. If the edge is a sharp
edge of fimtc length, as the edge of a cube away from a corner, then sufficiently
close to the edge the variation of the potential along the edge can be ignored.
I he two-dJmcnsJonal considerations apply, although the coefficient a 1 in (2.75)
may vary with distance along the edge. Similarly, the electrostatic arguments arc
valid even for time-vmying fields. The point heiC is that with time dependence
another length enters, namely. the wavelength. Provided one is concerned with
distances away from the edge that arc small compared to a wavelength, as well
as other relevant distances the behavior of the fields reduces to electrostatic or
magnctostatic behavior. In the diffraction of microwaves by a hole in a thin
conducting sheet, for example, the fields are singular as p - 112 as p --7> 0. where p
is the distance from the boundary of the hole, and this fact must be taken into
account in any exact solution of the diffraction problem.
The singular behavior of the fields ncar sharp edges IS the reason for the
effectiveness of lightning rods. In the idealized situation discussed here the field
strength mcreases without hm1t as p --7> 0, but for a thm sheet of thickness d with
a smoothly rounded edge it can be inferred that the field strength at the surface
w1ll be proportional to d 112 • For small enough d this can be vel y I at ge. In ab-
solute vacuum such field strengths arc possible; in air, however, electrical break-
down and a dischat gc will occur if the field sh cngth exceeds a certain value
(depending on the exact shape of the electrode, its proximity to the other elec-
Figure 2.13 Variation of the surface-charge density (and the electric field) with
distance p from the "eorner" or edge for opening angles f3 ,.14, ,.rJ,, ,., 3,.12, and 2'lT
Sect. 2.12 Introduction to Finite Element Analysis for Electrostatics 79
trades, etc., but greater than about 2.5 X 106 V/m for air at normal temperature
and pressure (NTP), sometimes by a factor of 4). In thunderstorms, with large
potential differences between the ground and the thunderclouds, a grounded
sharn condt edP"e or anoint (see Section 3.4) will have breakdown
~~~u<
.
m~uuu
·~
~
"'
H H<O< u u u HH<
.,_ ~ . vHU VL <Hv <H~H
•'-
p<~HUv
~
~u~
~ .£ •'- ~ ~-
,pmu
throuQ:h the a1r alonQ: wh1ch the JiQ:htnmQ: dJscharQ:e travels.
2.12 Introduction to Finite Element Analysis for Electrostatics

Finite element analysis (FEA) encompasses a variety of numerical approaches
for the solution of boundary-value problems in physics and engineering. Here
we sketch only an introduction to the essential ideas, using Galerkin's method
for two-dimensional electrostatics as an illustration. The generalization to three
dimensions is mentioned briefly at the end. The reader who wishes a deeper
introduction may consult Binns, Lawrenson, and Trowbridge, Ida and Bastos,
Sadiku, Strang, or Zhou.
Consider the Poisson equation, "il 2 !f; = -gin a two-dimensional region R,
;n, .. ,], L ~- ~' • ·~ ~-A .~rln~H r UTa •n~otrn~t tho
'0 ·o
. . .
vamsnmg mregra1,
I
1 I<PV-IJI+g<PJaxay u (L./0)
JJ<
where <P (x, y) is a test function specified for the moment only as piecewise con-
tinuous in R and vanishing on C. Use of Green's first identity on the first term
above leads to
L [V<P · V!f;- g<!J] dx dy =0 (2.77)
The surface integral vanishes because <P vanishes on C. Galerkin's method con-
sists first of approximating the desired solution •/l(x, y) by a finite expansion in
terms of a set of localized, linearly independent functions, <P,1(x, y), with support
only in a finite neighborhood of x = x,, y = y1. For definiteness, we imagine the
rep-ion R r1 hv" smJ>~re 1Mti"e with lauice sn""in" h. Th~en a nossihle choice
£. f.
LVL 'f'ii\Ao )' J 10 0
f ;, I I IUf· I ;~ ~m
'f'ijV'• J'! ' \' lA Ail•"}\-'- IJ' J'jl"'!

I " ' \~• UJ
'
lVl JA -"il' ' Y!l rt, UlllClWI~c, 'l'ij\-",J) V. lllt:O Ul dll UH:: o/ij UVCl ~Ulll
"• IY
the square lattice is unity. Other choices of the localized functions are possible,
of course. Whatever the choice, if the number of lattice sites, including the bound-
ary, is N 0 , the expansion of !f;(x, y) takes the form
(No)
!f;(x, y) = L
k,l
'l'kz<Pkl(x, y) (2.79)
Apart from the known values at sites on the boundary, the constant coefficients
'Itk 1 may be thought of as the approximate values of !f;(xk, y 1). If the lattice spacing
I. ' -11 .L <C · { , '7n\ !11 C ~t..i ' < <L
•• N oumu ·o· , •· ·~ 'l' \' . '! "F!'
true 1/J. orov1ded the coethcJCnts are chosen orooerlv.
cr~ .-l "'--' -'·. , ·~ .-l • -~ ,, '· .J.'
-·
"l' 'I'
(LJ 1 J LO oe Lne \l, 11 mncrion on rne expansion seL, wirn i ana j running sue-
cessively over all N internal sites of the lattice. The typical equation derived from
(2.77) is
(Nu)
"''
L..
k,l
"'kl
(
I "'l'ij~-"• YJ' • -
"'l'kf~.A, YJ
'
U.A
'
uy
. \
5\Aj, )'j}
(
I
' f
'f'ij\A• )' J
.\ -1.
~A
-'
')'
f">
,~.uv;
0()\
While the integrals are indicated as being over the whole region R, </>;;has support
only in a small region around the site(~,, y,). In (2.80) it is assumed that g(x, y)
clmdu ~n +h~ o~nl~ ~f +ha ~all oi7P tn hP ' l~tP£1 in thP
integral on ;he righ7 by its value at the lattice site. Once the ·i~tegrals have been
performed, (2.80) becomes one of N coupled inhomogeneous linear algebraic
• .T. '' d. •T• ' "
JWH>, 't'kf· UlC \CUI<}" w 0 kf
"4 LUL lllC 1V <H~ I }0 <VU
small number of sites near (x,, Y;), as indicated in Fig. 2.14 for the localized
function (2. 78). It is left as a problem to show that the needed integrals for the
functions (2.78) are
L </>;;(x, y)dx dy = h
2
k = i, l = j
(2.81)
r Q/'l
I ....,.L f. .\ T'1.L f. ,\A. A. £, k i+lt-i
'f'q ,~ , .J J 'f'kiV , .! J' •)'
)R -1/3 k = i, l = j ± 1
k = i ± 1, l = j ± 1
When the site li il is adiacent to the houndarv tnere are tnree or more terms
on the left-hand side of (2.80) that are ( -1/3) times known boundary values of
..-1.1 ~
~
/Off~
.L7HTL/. '1TI
'\'-" '"'
'IIIII I
I I I
\\'\
~ITT
"-ZLJ 17 'f--1... I r--1. 7'-/. Y.. XX..X'\...X'-.. \.Y
~
(k, l)
I ,, "
.,. ...... ~· •• •t. .L f .. \

1]\- ' ) ' '·
• I"> ..,0\
\' ' u;•
crt. rl- .,'j . t].
·"'' . tJ.
~omromJ are mose coup1ea oy me mregra1s on me ten in ~L.OVJ tur tHe w~,;a,izcu
function (2.78).
Sect. 2 U ··~ion to Finite Element Analvsis for Electrostatics 81
1/J. These can be moved to the right-hand side as part of the inhomogeneity. If
' I'> 0()\ ' ' <- T.T•I• £' '+1. T.T U v U ' rl
<W \' • 'J HH,_ 0 ~. n• T
lJr and G N-column vectors, the matrix K is a "sparse" matrix, with only a few
nonvanishing elements in any row or column. The solution of the matrix operator
equation by inversion of such a sparse matrix can be accomplished rapidly by
special numerical techniques (see Press et al.). Concrete illustration of this ap-
proach is left to the problems at the end of the chapter.
A square lattice is not optimal in many problems because the solution may
change more rapidly in some parts of the domain of interest than in other parts.
T. -'- • • ~ -'- ~~ A • L " • " "
H l OU'-H l ' Vll'-' IV HU ''-' U HH'-'l lll'-'OH. ~~~A~~ nHH U Oo lU
generic shape, but permitting different sizes, will be more flexible and therefore
superior. We describe the popular triangle as the basic unit in two dimensions.
ThP · -l~r PlPmPr>t ic ,.,r] t,-, hP cm~ll Pnmwh th~t thP fip]r] ""ri"hlP
changes little over the element and may be approxim~ed by a linear form in
each direction. The basic triangular element e(1, 2, 3) is shown in Fig. 2.15. Within
this region, we approximate the field variable 1/J(x, y) = 11e(x, y) = A + Bx +
Cy. The three values (!/Jr. !fJl, I/J3 ) at the nodes or vertices determine the coeffi-
cients (A, B, C). It is useful, however, to systematize the procedure for numerical
computatiOn by detimng three shape tunctwns Nj"'(x, y), one tor each vertex,
such that Nj"> = 1 when x = xj, y = yj and Nj"> = 0 at the other vertices. The
eh j', • j', lh ,J. •< ·~· ·,.1. •• • ,J, ~ •
· r u
Consider N\e) = a1 + b1x + c1y. Demand that

llJ T OJXJ T <: 1Y, 1
a1 + b 1x 2 + c1y 2 = 0
UJ UJ.>] ' qYJ U
The determinant D of the coefficients on the left is

1 x, y,
D = 1 x2 h = (x2 - x1)(Y3 - y,) - (x, - x,)(h - Yt)
1 x, y,
The determinant D is invariant under rotations of the triangle; in fact, D = 2S" where Se
is the area of the triangle. The coefficients (a 1 , b 1 , c1 ) are
a1 =
2~ e
(x2Y 3 - x,y2)
1
b1 = (h - h)
25e
-1
Figure 2.15 Basic triangular element

e(l, 2, 3) with areaS" for FEA in two
dimensions.
0~ Lhapter L nounuary- v atut! r •m
The other NY> can be written down by cyclic permutation of indices. The Ni and their
coefficients obey the following relations:
3 3 3 3
2: NV>(x, y)- 1; }.; a,- 1; }.; b1 - 0; 2,; c,- 0;
1-l 1=l 1=\ t=l
, , "T
Here Xe = (x 1 + x 2 + x 3 )/3 and Ye = (y 1 + y 2 + y3 )/3 are the coordinates of the center of

gravity of the triangular element e.
T ·1 -1 tl- D ~on
LIH; ·r ~ -0 u
be used in the Galerkin method as the localized linearly independent expansion

set. The field variable 1/J(x, y) has the expansion,
' "V •T•({)u{f\1 .\ (') Q')\
the tw~ triangles from either representation is the same weighted average of the
values at each end, with no contributions from the shape functions for the vertices
not in common.
WP return to the Pmsson eouatwn With Uinchlet bounuary conuitions anu
the vanishing integral (2.77). With the expansion (2.82) for if!(x, y), we choose
the test function cp(x, y) = N~e>(x, y) for some particular element e and vertex i
(only avoiding vertices on the boundary because we require cp = 0 on C). The
chmce reouces tne mtegrm ]ana we sU111 111 ~L.OLJJ tu u""' v ,v, wv pm
. ·-" Lt. • ··'
element chosen, just as did the choice of the localized function in (2.t-;U). The
integral, with the inhomogeneity transferred to the right-hand side, is
" r I
If g(x, y) changes very little over the element e, it can be approximated by its
,]. afv V \ ~t thP ~PntPr nf • Of fhp fri,lnO"lf> and factored OUt of
oe o e' ./ e/
the ngm-nanu integraL 1ne r u ·o'a' ;,
(2.84)
Je
For the lett-hand mtegral m lLt";.:lJ, me 1mearity or me snape 1uuuiu11>

that the integrand is a constant. We note that aNf'>;ax = b,, FJNV>!ay = c,, and
define
~L.1"D)
The coefficients k~) form an array of dimensionless coupling coefficients for the
" " " ' "e ' -L <1. <L ' tJ- chonP Af thP _.1
lf c. H 1> >Cl 'dlU v ouv•• ~y "'" -'"
Sect. 2.12 Introduction to Finite Element Analysis for Electrostatics 83
1 l
? -13
0750 760\
/ ;\
l
-- l
--
1
z
2-/3 2-/3
,7.
7
goo
1
1
v:< -----'---
~ 1
v~
2 2-/3
Figure 2.16 Examples of the triangular coupling coefficients. The "diagonal"
coefficients are at thr' corners (vertices) and the "off-diagonal" coefficients along the
sides, between vertices.
triangle, but not its orientation or size. Two examples are shown in Fig. 2.16,
where the diagonal elements k~) are located at the corresponding vertices (i) and
the off-diagonal elements k~) along the line connecting vertex i with vertex j.
With the definition (2.1\:'5) of the coupling coefficients, (2.1\3) becomes
3 s
" k(e)-qr(e) = .....!: g (i = 1, 2, 3) (2.86)
? 11 1 3 e
For each element e there are three algebraic equations, except when the side(s)
of the triangle form part of the boundary. The three coupled equations can be
writt€m in matrix form, k(e)lJt(e) c<e).
The result for one element must now be generalized to include all the tri-
angular elements spanning R. Let the number of interior vertices or nodes beN
and the total number of vertices including the boundary he No I abel the inter-
nal nodes with j = 1, 2, 3, ... , N, and the boundary nodes by j = N + 1,
N + 2, ... , N 0 . Now enlarge and rearrange the matrix k(e) ~ K, where K is an
N X N matrix with rows and columns labeled by the node index. Similarly, define
the N-column vectors, 'IT and G. For each triangular element in turn, add the
elements of k~e) and Sege/3 to the appropriate rows and columns of K and G. The
end result is the matnx equatiOn
K'IT = G (2.87)
K = (k,) with k,, = 2: kYf>

T
and i *j
N (2.88)
G, = - LT Sege -
3 i=N+l
I he summation over 1 means over all the tnangles connected to the mternal
node i; the summation over E means a sum over all the triangles with a side from
internal node i to internal node j. The final sum in G, contains, for nodes con-
nee ted directly to the boundary nodes, the known boundary values of rfJ the1 e
and the correspondmg klJ' values (not present m the matnx K). The reader may
ponder Fig. 2.17 to be convinced of the correctness of (2.88). Just as for the square
lattice, the N X N matrix K is a symmetric sparse matrix, with positive diagonal
elements. As mentioned earlier, there are special efficient methods of inverting
such matrices, even if very large.
1'1'1 Chapter 'L Boundary-Value l'romems m "'ecuosraiics; •· -:u
\
-;<\:
I / ;K. I
~~/
101~-f? __:=.
>fo5
~
~ 7
~
/104
Figure 2.17
. ·'· ·'
A part of the array of
'T
.; .. ~ ,,.,,
u
R
Boundary 1m assumed to have 100 internal nodes.
ThP .. hvinm oPnPplization of the trian~>le to three-dimensional FEA is to

add another vertex out of the plane to make a tetrahedron the basic element of
volume. Now four shape lunctwns, N'j''(x, y, z), are used to give an approxima-
tion to the field variable within the tetrahedron. The algebra is more involved,
but the concepTislffe same.
r\nr rl;c~ncc;~n ;< ~ h~ • intrnfl , 1n finitP PlPmPnt ,.n,.hr<i<. Manvvariants
exist in every branch of physics and engineering. National laboratories and com-
;gJ rnmn,.niPS have "canned" FEA nacka~>es: POISSON is one such pack-
age, developed at the Lawrence Berkeley National Laboratory jointly with
Ltvermore Natwnal Laooratory; n and LAI l~ are two deve1opeu aL
the Rutherford-Appleton Laboratory in Britain.

The method of images and the related technique of inversion are treated in many
books; among the better or more exrens1ve aiscussions are rnose oy
Jeans, Chapter VIII
Maxwell, Vol. 1, Chapter XI
Smythe, Chapters IV and V
The classic use of inversion by Lord Kelvin in 1847 to obtain the charge distnbutwn on
the inside and outside surfaces of a thin, charged, conducting spherical bowl is discussed
Ill
Kelvin. n. 186
Jeans, pp. 250-251
A truly encyclopedic source of examples of all sorts with numerous diagrams is the book
by Durand, espec1iilfyLllapters 111 and 1V. Durand discusses mverswn on pp. 1u 1 -11 '+.
Complex vanables and contormal mappmg techmques tor the sotulion 01 rwo-
dimensional potential problems are discussed by
JJUUtlliT, ·r
.,
Jeans, Chapter VTTT, Sections 306-337
Maxwell, Vol. I, Chapter XII
Morse and Feshbach, pp. 443-453, 1215-1252
Smythe,Tl1apter1v, ;:,ec1ions 'I.U'i-'+.51
Thnmsnn Chanter 3
A useful little mathematics book on conformal mapping is
Bieberbach
Ch. 2 Problems 85
There are, in addition, many engineenng books devoted to the subject, e.g.,
Gibbs
Rothe, Ollendorff, and Polhausen
Elementary, but clear, discussions of the mathematical theory of Fourier series and
integrals, and orthogonal expansions, can be found in
Churchill
A somewhat old-fashioned treatment ol Fonner senes and mtegrals, but w1th many ex-
amples and problems, is given by
Byerly
The literature on numerical methods is vast and growing. A good guidepost to per-
Paul L. DeVries, Resource I .etter cP-1 · Computational Physics Am T Phys

64, 364-368 (1996)
In addition to the books cited at the beginning of Section 2.12, two others are
P. Hammond and J. K. Sykulski, Engineering Electromagnetism, Physical Pro-
cesses and Computation, Oxford University Press, New York (1994).
C. W. Steele, Numerical Computation of Electric and Magnetic Fields, Van
Nostrand. New York (1987)
The first of these has a brief but clear discussion of FEA in Chapter 7, the second ticals
FEA and related topics in greater depth.
Problems
2.1 A point charge q is brought to a position a distanced away ftorn an infinite plane
conductor held at zero potential. Using the method of images, find:
(a) the surface-charge density induced on the plane, and plot it;
(b) the force between the plane and the charge by using Coulomb's law for the
force between the charge and its image;
(c) the total fmce acting on the plane by integrating t?/2E0 over the whole plane;
(d) the work necessary to remove the charge q from its position to infinity;
(e) the potential energy between the charge q and its image [compare the answer
to part d and discuss].
(f) Find the answer to part din electron volts for an electron miginally one ang-
strom from the surface.
2.2 Using the method of images, discuss the problem of a point charge q inside a hollow,
grounded, conducting sphere of inner radius a. Find
(a) the potential inside the sphere;
(b) the induced smface-chaige density,
(c) the magnitude and direction of the force acting 011 q.
(d) Is there any change in the solution if the sphere is kept at a fixed potential V?
If the sphere has a total charge Q on its inner and outer surfaces?
2.3 A straight-line charge with constant linear charge density A is located perpendicular
to the x-y plane in the first quadrant at (x0 , y 0 ). The intersecting planes x = 0,
y > 0 and v 0. x > 0 are conducting boundary surfaces held at zero potential
Consider the potential, fields, and surface charges in the first quadrant.
(a) The well-known potential for an isolated line charge at (x0 , y 0 ) is <l:>(x, y) =
(A/47TEo)ln(R2 /r2 ), where r2 = (x - x0 ? + (y - Yo? and R is a constant. De-
termine the expression for the potential of the line charge m tne presence or
the intersecting planes. Verify explicitly that the potential and the tangential
etectric ne1a vanisn-urrwc uutmumy surt<tt.Oc>.
(b) Determine the surface charge density a on the plane y = 0, x ""= 0. Plot a/ A
• ' 1 "1\
versusxror~x 0 'L,y 0 'h\-'·O '•Y" 'hauuvo •JO - 1.
(c) Show that the total charge (per unit length in z) on the plane y = 0, x ""= 0 is
2 _,, Xo
Qx = --A tan 1
1T Yo
What is the total charge on the plane x = 0?
(d) Show that far from the origin [p>>Po, wnere p - V lX + y") aru.r
p11 = v' (xi\ + yi\)] the leading term in the potential is
,., <1:> = 4A (XoYo)(xy)

~__....;,. asym 4
7TEo p
Interpret.
Z.4 A poim cnarge is ' . a HU"~~ u ~. " u Ulll lllC l.OCIIlCI v~ UH '5' ,
isolated, conducting sphere of radius R.
~
a) .
1ns10e or wnat msrance- rrum Lnc sunace ur-ure- 't '" ., '"" pvuu ~uac~
.
attracted rather than repelled by the charged sphere?
(b) What is the limiting value of the force of attraction when the point charge is
located a distance a (- d R) from the surface of the sphere, 11 a << K!
(c) What are the results for parts a and b if the charge on the sphere is twice
(half) as large as the point charge, but still the same sign?
[Answers: (a) d!R- 1 = 0.6178, (b) F = -q2 /(167TEoa2 ), i.e., image force, (c) for
Q - 2q, d/R 1 - 0.4276; for Q - q/2, d!R 1 - 0.8823. The answer for part b
is the same. J
2.5 (a) Show that the work done to remove the charge q from a distance r > a to
infinity against the force, Eq. (2.6), of a grounded conducting sphere is
Relate this result to the electrostatic potential, Eq. (2.3), and the energy dis-
. .c 0 • 1 1 1
(b) Repeat the calculation of the work done to remove the charge q against the
1Vl'-'<'• '-''!• \L..~), u~ au· ' ~ ~ 5, ~ · -l . <:J, · +~ >1 th • mnrlr
done IS
1 q a qa
vv
r
'"~ vu ' '"1" \ ~ • '
cussion of Section 1.11.

..:.o ~ <OlC'-'ll . fll UUlClll Ul u pvuu '1 v , -
ducting sphere is equivalent to that of three charges, the original and two others,
1 ~t tha nf th -L · ~ncl onnthPr ("the ima2:e Chaf2:e") inside
the now imaginary sphere, on the line joining the center and the original charge.
Ch. 2 Problems 87
If the point charge and sphere are replaced by two conducting spheres of radii
r, and rh, carrying total charges Oa and Q,, respectively, with centers separated by
a distanced > r" + rb, there is an equivalence with an infinite set of charges within
each sphere, one at the center and a set of images along the line joining the eenters.
The charges and their locations can be determined iteratively, starting with a charge
qa(1) at the center of the first sphere and qh(1) correspondingly for the second
sphere. I'he charge qh(1) has its image qa(2) within the first sphere and vice versa.
Then the image charge within the first sphere induces another image within the
second sphere and so on The sum of all the charges within each sphere must be
scaled to be equal to Q, or Qb.
The electwstatic potential outside the spheres, the force between the spheres,
etc. can be found by summing the contributions from all the charges.
(a) Show that the charges and their positions are determined iteratively by the
relations,
qa(j) = -r,qb(j- 1)/dh(j- 1), xa(j) = r;ldb(j - 1), da(j) = d - Xa(j)
qb(j) = -rbqu(j - 1)/da(j - 1), x 6 (j) = r~/da(j - 1), db(j) = d - x,(j)
for j = 2, 3, 4, ... , with du(J) = dh(1) = d, and Xa(1) = x 1,(1) = 0.
(bl Find the image charges and their loca1ions as well as the potentials on the
spheres and force between them by means of a suitable computer program.
[In eomputing the potential on each sphel e, evaluate it in diffei ent places. e.g.,
in the equatorial plane and at the pole opposite the other sphere. This permits
a check on the equipotential of the conductor and on the accuracy of
computation.]
(c) As an example, show that for two equally charged spheres of the same radius
R, the force between them when almost m contact IS 0.6189 limes the value
that would be obtained if all the charge on each sphere were concentrated
at its center. Show numerically and by explicit summation of the series that
the capacitance of two identical conducting spheres in contact is C/47TE0 R =
Reference: J. A. Soules, Am. J. Phys. 58, 1195 (1990).

2.7 Consider a potential problem in the half-space defined by z " 0, with Diiichlet
boundary conditions on the plane z = 0 (and at infinity).
(a) Write down the appropriate Green function G(x, x').
(b) If the potential on the plane z = 0 is specified to be = V inside a circle of
radms a centered at the origin, and - 0 outside that circle, find an integral
expression for the potential at the point P specified in terms of cylindrical
coordinates (p, ¢, z ).
(c) Show that, along the axis of the circle (p = 0), the potential is given by
<1> = v(t - va z+ z ) 2 2
(d) Show that at large distances (p2 + z2 >> a 2 ) the potential can be expanded in
a power series in (p 2 + z2 ) ', and that the leading terms are
Verify that the results of parts c and d are consistent with each other in
their common range of validity.
2.8 A two-dimensional potential problem is defined by two straight parallel line charges
separated by a distance R with equal and opposite linear charge densities A
anu -A.
,_,
I~\ <'1.
"
..,. ., ., -'· ~f ' ' n~t~ntiol T7 io o rir.
cular cylinder (circle in the transverse dimensions) and ftnd the coordinates
of the axis of the cyl~nder and its radius in terms of R, A, and V.
(b) Use the results of part a to show that the capacitance per unit length C of two
right-circular cylindrical conductors, with radii a and b, separated by a distance
J ,_ •
u ,
,.-. 21TEo
Jd2- b2
cosh_,
a2-
l
\ L.au
{CJ v enry tnat tne result tor c agrees wan tne answer in rruutem 1. 1 in '""
appropriate limit and determine the next nonvanishing order correction in
£ J J , _ _,
p
{,f) D. thP r"lr.11htinn nf 1hP itance oer unit length for two cvlinders
inside each other (d < Ib - a I). Check the result for concentric cylinders
(d = 0).
2.9 An insulated, spherical, conducting shell of radius a is in a uniform electric field E0 •
If the sphere 1s cut mto twonem1spneresoy a plane perpen01cular to tne new, nna
th~ fr>rr~ - th~ ,- , from ·w
-, · -1 t,-, ·
(a) if the shell is uncharged;

(b) if the total charge on the shell is Q.
"."' " ' -'· ·~ ,,J •rt;nn -"
..l. ,f ,]. mith
A 1
'b' · < ~ h·:r II -.-. . b f 0
. . .
separatiOn D. one of w 1ch has a sma hem1sphenca oss o radius a on 1ts mner
l
surface (D >> a). The conductor with the boss is kept at zero potential, and the
other conductor is at a potential such that far from the boss the electric field between
the plates is E 0 •
' ~-· _, ... .£ .
' '
.. ... ,I, ..l
'"' on the bos~, and,sketch th'!:r behavior as a function of distance (a/angle).

(b) Show that the total charge on the boss has the magnitude 31TE0 E 0 a 2 •
(p\ Tf . ~f th~ ~th~r ' ehPPT "(" ..l:«. i" 1 " noint '
q is placed directly above the hemispherical boss at a distanced from its center,
show that the charge mduced on the boss IS
a a
q' q 1 2 2
uvu u -
2.11 A line charge with linear charge density T is placed parallel to, and a distance R
away from, the axis of a conductmg cylinder of radms b held at fixed voltage such
tJ. tJ. ' niohoo< nt 'n+:n:t, Din..l
r
(a I the ma2:nitude and oosition of the ima2:e charge( s):

{D J tne potential at any pomt ~ expressea in pomr coo ruinates witn tne uri gin at
the "xis of th.e cvlindPr and the direction from the origin to the line chame as
the x axis), including the asymptotic form far from the cylinder;
(c) the induced surface-charge density, and plot it as a function of angle for
Rib = 2, 4 in units of Ti21Tb;
(d) the force per unit length on the line charge.
""· ., . on
0.7
2.12 Starting with the series solution (2.71) for the two-dimensional potential problem
with the potential specified on the surface of a cylinder of radius b, evaluate the
coefficients formally, substitute them into the series, and sum it to obtain the po-
tential inside the cylinder in the form of Poisson's integral:
1 (lrr b2- p2
~'
~\P• 'I' I
21T Jo ~,u, 'I' 1 b +p 2 2 - 2bp cos(</>' - </>) u'i'
What modification is necessary if the potential is desired in the region of space

hv thP ·" ,1, oncl · · ?
0
2.13 (a) Two halves of a long hollow conducting cylinder of inner radius b are sepa-
rated by small lengthwise gaps on each side, and are kept at different paten-
tials V and Vo. Show that the ootential inside is 2:iven bv
v, + Vo
... V, - Vo .I 2bo \
'!'~p. </>)
2 1T
ran \i>2 - p2 cos q>)
where </>is measured from a plane perpendicular to the plane through the gap.
(b) Calculate the surface-charge density on each half of the cylinder.
2.14 A variant of the preceding two-dimensional problem is a long hollow conducting
cylinder of radius b that is divided into equal quarters, alternate segments being
held at potential + V and - V.
(a) Solve by means of the series solution (2.71) and show that the potential inside
t\, .. e "
-J
4
LIV ~ / n \ n+Z <inf{L!n + ?\A,]
<P(p, "')
1T .~o ~b} 2n + 1
(b) Sum the series and show that
2
2V 1{ 2lb sin 2</> l
<P(o. d>) tan
1T \ 0 p
\~I "''-' "'-'u ""'-'' auu '-''!' -r HW<O.
2.15 (a) Show that the Green function G(x, y; x', y') appropriate for Dirichlet bound-
ary conditions for a square two-dimensional region, 0 :S x :S 1, 0 :S y :S 1, has
an expansiOn
~
G(x, y; x', y') = 2 2:

ll=l
g,(y, y') sin(n1rx) sin(n1rx')
"'"'" 5n\Y• Y I
·2 \
( ay' 2 - n 2 1T2 } g,(y, y') = -41To(y' - y) and g,(y, 0) = g,(y, 1) = 0
(b) Taking for g,(y, y') appropriate linear combinations of sinh(n1ry') and
cosn~n1ry) m tne two reg10ns, y < y ana y > y, m accoro w1tn tne oounoary
conditions and the discontinuity in slope required by the source delta function,
show that the explicit form of G is
G(x, y; x', y')
~
1
IS 2-. . h( ) sm~n1rx) sm~n1rx· J smn~n1ry<l smhLn1TU Y>)J
n=l n Slll n1T
where Y<(Y>) is the smaller (larger) of y andy'.

2.16 A two-dimensional potential exists on a unit square area (0 o=: x o=: 1, 0 o=: y o=: 1)
bounded by "surfaces" held at zero potential. Over the entire square there is a
UHHUHH c ') Uo UHH "'""15'" \J-'"' UHH ''-"15"' H< ~~c ~""15 "''- U<'-'-"
function of Problem 2.15, show that the solution can be written as
4 ~ sinf(2m + 1)17xl coshf(2m + 1)17(V- ~)1
'l'~X, YJ
1
1T'E0 "f:o (2m+ 1) 3 cosh[(2m + 1)17/2]
2.17 (a) Construct the free-space Green function G(x, y; x', y') for two-dimensional
c .,,_ L oL " c
~·~~oowouoo~o ':Y ""~puuu5 ''" ""' ~ ov \~ ~J '"~ uouuo
±Z, where Z is taken to be very large. Show that apart from an inessential
constant, the Green function can be written alternately as
G(x, y; x', y') = -ln[(x - x') 2 + (y - y'jl]
= -lnfo2 + o' 2 - 2oo' cos(<f>- <b')l
lb) Show exolicitlv bv seoaration of variables in oolar coordinates that the Green
function can be expressed as a Fourier series in the azimuthal coordinate,
r; - J... ~ pim(</>-<P'), ..J n n'l
where the radial Green functions satisfy
2. ...!:_ (p' iJg"') - mz g = -417 i5(p - p')

p' i!p' i!p' p'2 m p
Note that gm(p, p') for fixed pis a different linear combination of the solutions
of the homogeneous radial equation (2.68) for p' p, with a
rE c ;., of o1. r ~. • rl h t> rlolr C. •
' ., ·r r r · J
(c) Comolete the and show that the free-so ace Green fnn<Ctinn has the
expansion
G(p, <f>; p', </>') = -ln(p;) + 2 ~

~~ 1 ( ~: )"' · cos[m(</>- </>')]
m
where PAP>) is the smaller (larger) of p and p'.

2.18 (a) By finding appropriate solutions of the radial equation in part b of Problem
2.17, find the Green function for the interior Dirichlet problem of a cylinder
cC ~; I. C. f r 1.\ f\ C'. f' tf\\1 IO: "· ~ oL .1.·
' '""'"'" ., . ,. ~JJ· -.
to the free-space Green function of Problem 2.17. Then show that it can be
written in closed form as
or
(b) Show that the solution of the Laplace equation with the potential given as
w~o,q> Jon we cynnaer canoe expressea as rmsson s mregra1 or rromem L.IL.
\~J H H<H dH;; ~ 1Ul UIC \JlCCll lUI lllC LA<e< w . r

(b co. For situations in which the potential falls off fast enough as
p---> co, no mistake is made in its use.]
2.19 Show that the two-dimensional Green function for Dirichlet boun
densities ±A, as shown in the left-hand figure (the right-hand figure shows the elec-
tric field lines).
The charge density in two dimensions can be expressed as
(a) Using the Green function expansion from Problem 2.17c, show that the elec-
trostatic potential is
(b) Relate the solution of part a to the real part of the complex function
where z = x + iy = pe;"'· Comment on the connection to Problem 2.3.

(c) Find expressions for the Cartesian components of the electric field near the
origin, expressed in terms of x andy. Keep the k = 0 and k = 1 terms in the
states that if F(z) is analytic in a region R bounded by a closed curve C, then
Hint: You may wish to add an integral that vanishes (associated with the image
point) to the integral for the point inside the circle.
2.22 (a) For the example of oppositely charged conducting hemispherical shells sepa-
r~te.rl hv ~ tinv "~n. ~< in · ?.X <hnw th~t the. · · ·
~
,.
f. >\
.'
+l,
2 2
<P;n(Z) = V-a [ 1 - (a
Y 2- z ) 2 ]
z a a + z
~'>HU
> " <H~
£
>HO<
£.
>~ <~HHC V>
.. <H~ >H p v o~>O V> ~ "
UHU OHV <HU<
...
<H~J
agree wun ~L.LJ J wnn me appropnare suosmuuons.

(b) From the result of part a and (2.22), show that the radial electric field on the
positive z axis is
Va 2 a2
1'-,(Z) (j+-:;
~~ T U) ~
for z >a, and

v ' + (n/71 2 {]2
n,(z) -:;
u . ~~ \ ~/U) ) ~
for Iz I < a. Show that the second form is well behaved at the origin, with the
value, E,(O) = -3V/2a. Show that at z =a (north pole inside) it has the value
- (v2 ~ 1) Via. Show that the radial field at the north pole outside has the
vame v L v JU.
(c) Make a sketch of the electric field lines, both inside and outside the conducting
hemispheres, with directions indicated. Make a plot of the radial electric field
along the z axis from z = -2a to z = +2a.
2.23 A hollow cube has conductinQ walls defined bv six olanes x = 0 v = 0 z = 0 and
, J , •
"rl. ell; () A .J,oJA l7
~ ~ ~
r
The other four sides are at zero potential.
(a) Find the potential <P(x, y, z) at any point inside the cube.
{h\ PvolHotP thP
. of fhp nf thP ruhP numPriro llv
0 ,
orr1 orotP tn thrPP
• u u w 111<1:1 Y_ 1<01111> 111 11110 ><OU<O> 1> 11 1u 1\.nop 111
order to attain this accuracy? Compare your numerical result with the average
value of the potential on the walls. See Problem 2.28.
(c) Find the surface-charge density on the surface z = a.
l.l'l 1n tne two-aJmenswnat regiOn snown m r1g. LlL, tne angular runctwns appropnate
for Dirichlet boundary conditions at tjJ = 0 and tjJ = f3 are <P( tjJ) =Am sin(m7r</J/{3).
Show that the completeness relation for these functions is
2 ~
8(</J- </>') = R 2: sin(m1T</Jif3) sin(m7r</J'/f3) for 0 < </J, tjJ ' < f3
~ ~~
··-~
"r.
"·
_,
" ' ~~•v r
.. ' >H~~
.L
< U<VH5
...
<H~ ~ U~<OI UH
_,
UHs<~
n
fJ
between them, as in Fig. 2.12. A unit line charge parallel to the z axis is located
between the planes at position (p', </J').
(a) ::.now that (47rE0 ) times the ootentlal m the soace between the olanes, that 1s
thP nirirhJPt (irppn .e. (";( n d-.· n 1
d-.'\ ;, uivPn hv thP inlinitP <PriP<
G(p, </>; p', tjJ ') = 4 2: -1 p':!_~1 ~ p~m~l~ sin(m7r</J/ {3) sin(m7r</J 'I {3)
m=l m
Ch. 2 Problems 93
(b) By means of complex-variable techniques or other means, show that the series
can be summed to giVe a closed form,
G( A..· I ·" ') - I I ~p) -t- ~p r . " L~pp) . coS[ 7T\(/J + </>.)I 131
p, Y'• P • '~' - n (p)hi/3 + (p')2rr 1f3- 2(pp')~ 1 f3 cos[1r(</>- ¢'/!3]
(c) Verify that you obtain the familiar results when f3 = 7T and f3 = 1r!2.
2.26 The two-dimensional region, p ~ a, 0 :5 ¢ :5 /3, is bounded by conducting surfaces
at ¢ = 0, p = a, and ¢ = f3 held at zero potential, as indicated in the sketch. At
large p the potential is determined by some configuration of charges and/or con-
ductors at fixed potentials.
~o
I
I
I
I
"'- ""-......_ ________ / /
Problem 2.26
(a) Write down a solution for the potential 1>(p, <P) that satisfies the boundary
conditions for finite p.
(b) Keepmg only the lowest nonvamshmg terms, calculate the electnc field com-
ponents Eo and E,, and also the surface-charge densities u(p, 0), u(p, B). and
u( a, ¢) on the three boundary surfaces.
(c) Consider {3 7T (a plane conductor wlfh a halt-cylmder of radms a on It).
Show that far from the half-cylinder the lowest order terms of part b give a
uniform electric field normal to the plane. Sketch the charge density on and
in the neighborhood of the half-cylinder For fixed electric field strength far
from the plane, show that the total charge on the half-cylinder (actually charge
pel unit length in the z diiection) is twice as lmge as would reside on a strip
of width 2a in its absence. Show that the extra portion is drawn from regions
of the plane nearby, so that the total charge on a strip of width large compared
to a is the same whether the half-cylinder is there or not
2.27 Consider the two-dimensional wedge-shaped region of Problem 2.26, with f3 = 27T.
I his corresponds to a semi-mfinite thin sheet of conductor on the positive x axis
from x = a to infinity with a conducting cylinder of radius a fastened to its edge.
(a) Sketch the smface-charge densities on the cylinder and on tile top and bottom
of the sheet, using the lowest order solution.
(b) Calculate the total charge on the cylinder and compare with the total defi-
ciency of charge on the sheet neal the cylinde1, that is. the total difference in
charge for a finite compared with a = 0, assuming that the charge density far
from the cylinder is the same.
i
2.28 A closed volume is bounded by conducting surfaces that are then sides of a regular
polyhedron (n = 4, 6, 8, 12, 20). The n surfaces are at different potentials V,,
i = 1, 2, ... , n. Prove in the simplest way you can that the potential at the center
of the polyhedron is the average of the potential on the n sides. This problem bears
on Problem 2.23b, and has an interesting similarity to the result of Problem 1.10.
2.29 For the Galerkin method on a two-dimensional square lattice with lattice spacing
h, verify the relations (2.81) for the localized "pyramid" basis functions, <f>,/x, y)
= (1- lxlih)(l- IYiih), lxl < h, IYI < h, where x andy are measured from the
site (i, j). In particular,
r r r r 0
I dx I dv ,/,., .{y vi - h2 • I A· I "' "((A. . . • "((A. . . -

J J J J "" 3
r r r r
I Av I A, "CCA. . • "((A,
.,, ':. I "· I "· ...., .1. T'f.J. L'
J J 3, J J '·1 '·1 3'
Jdx Jdy V </>,+l.J+l • V<f>,J = - ~

2.30 Using the results of Problem 2.29, apply the Galerkin method to the integral equiv-
alent of the Poisson equation with zero potential on the boundary,
L dx dy[V <f>,,j • Vrfi - 47Tp<f>i,j] = 0 with rfi(x, y) = ,.,~ t r/J;·,r </>;·,r (x, y)
for the lattice of Problem 1.24, with its three independent lattice sites. Show that
you get three coupled equations for the r/J;,j values ( rfi, rfi2 , rfi3 ) and solve to find the
"Galer kin" approximations for the potential at these sites. Compare with the exact
values and the results ot the vanous IteratiOns oTYroDTem T.Nc. Comment.
ll/1 47TE0<PI.
_......,..,..A I~ ,.,
L,UJ-\.r l_ DK .J
"Til 1 T T 1 ~ 1 1
nounuury- vutue r· ruuterns
• T""""1 '
zn ntectrostaacs: 11 TT
In this chapter the discussion of boundary-value problems is continued. Spherical

and cylindrical geometries are first considered, and solutions of the Laplace equa-
tion are represented by expansions in series of the appropriate orthonormal func-
tions. Only an outline is given of the solution of the various ordinary differential
equations obtained from the Laplace equation by separation of variables, but the
properties of the different functions are summarized.
1ne promem 01 consuucnon or vreen tuncuons m terms or onnonormat
functions arises naturally in the attempt to solve the Poisson equation in the
• ' T ' ' ,_-, • , ,
vauvu' 0 . '""'"''' . ~ _~ V> ' - ' ' ' - ' ' - ' ' >UH'-'l'VH' a>'-' auu a}'-
plied to specific problems, and the equivalence of the various approaches to
. '1 . '-' . r~;
rl
r y
3.1 Laplace Equation in Spherical Coordinates

In spherical coordinates (r, IJ, ¢),shown in Fig. 3.1, the Laplace equation can be
written in the form:
1 a-;;2 (r<P) + 1 a . a<t> 1 a2<t>
r ar r s1n u au \
sm IJ
au
l+ r sm u mp
0 (3.1)
If a product form for the potential is assumed then it can be written:
<t> = U(r) P(IJ)Q(¢) (3.2)

r
When this is substituted into (3.1). there results the equation:
d2 U UQ d dP' UP d2Q
PQ -;;- + sin IJ +r --;; 0
ur r Slu u uu uu Slu u u<p
If we multiply by r 2 sin2 IJIUPQ, we obtain:
1 dLU 1 d dP 1 d"Q
r 2 sin2 IJ --+
2 - sin IJ diJ + (3.3)
U dr Pr 2 sin IJ d IJ Q d¢2 = 0
The ¢ dependence of the equation has now been isolated in the last term. Con-
<PflHPntlu th<>t tPrm mHd hP" N>n<t"nt mh;r-h
T -J
""11 (. ~2). mP
' /
2
1 d Q _.,.,2
,. L1·;\
(?. -
Q d¢L
n~
Jo,!
7U Lnapter J nounoary- v ame rromems m J;,tecrrosrancs: n -M
.l
/I
/ I
/'
ty I
I
I
I
I
I y
/ ''
X/ '" ',-J
Figure 3.1
This has solutions
!! e (.:) ..) )
For Q to be single valued, m must be an integer if the full azimuthal range is
allowed. By similar considerations we find separate equations for P(e) and U(r):
1 d dP mz
I Sill tJ J+ LV + l) .-:;- lr - u l j,())
Mil 0 UO UO I Mll 0
d 2 U l(l + 1)
dl-
r r
2 U=O (3.7)
.L
'u~·~
If 1
'\'
1\'
<J <O UHV<H~<
... <~U<
.I
~VHO<UH<o
J:<rom the torm ot the radial equation It Is aooarent that a sml!'le oower ot r
(rather than a power series) will satisfy it. The solution is found to be:
U = Ar 1+ 1 + Br- 1 (3.8)
hut l is as vet u nd1 1ined.
3.2 Legendre Equation and Legendre Polynomials

t ne t1 equa1ion wr r~t1J is cuswmarity expresseu in rerms 01 x cos t1,
of ()itself. Then it takes the form:
d [ dPl [ m
2
dx (1 - x2) dx + l(l + 1) - 1 - x2 p l = 0 (3.9)
,. ' .
' .'
_,
" '
, auu HO OV,UllVllO a<v
.T T
'"'' '"" IS"""''" "'JI
the associated LePendre functions. Before constdennP 11.9\ we outlme the

solution by power series of the ordinary Legendre differential equation with
m 2 = O·
'
A AP
- {1 - x")- + [([ + 1)p = 0 {3.1 0)
ax ax
We assume that the whole range of cos e, including the north and south poles, is
in the region of interest. The desired solution should then be single valued, finite,
>~nil · 11()]1' on thP intPrv>~l -1 < y < 1 in oriiPr th>~t it · "nh.,..;~11
Sect. 3.2 Le.,endre Eonation and Le.,endre Polvnomials 97
notential. The solution will be assumed to be renresented hv ~ o~r;~o nf

the form:
P(x) = xa f
j=O
axi (3.11)
where a is a parameter to be determined. When this is substituted into (3.10),

Lnere resuns Tne senes:
~
~ {(a + j)(a + j - 1)a;xa+j-z - [(a + j)(a + j + 1) - l(l + 1)]a1xa+i) = 0

-~
(3.12)
Tn th;o Pvn~no;,.,n thP .CC ;Pnt Af P~<>h · ,.,f v m11ot v~nioh oPn~r~tPlv HAr
j = 0, 1 we find that
;c -+ (\ +J..~~ ..f 1\ (\
"TT
"" '7
(3.13)
tt a 1 * U, then a( a + 1) - u
while for a general j value
(a + j)(a + j + 1) - l(l + 1)
aJ+2 = ai (3.14)
(a+ i + 1)(a + i + 2)
. .
A. s mougm snows maT me TWO retanons ~-'·UJ are eqmvatem ana maT
it is sufficient to choose either a 0 or a 1 different from zero, but not both. Making
+h~ C. · nJ..~;n~ m~ h~u~ ~r 1 Pr~~ 1A\ m~ 0~~ thot tJ..~
n (\ Ao ('l
.,
series has only even powers of x (a= 0) or only odd '" powers of x (a= 1)."
For either of the series a = 0 or a = 1 it is possible to prove the following
nronerties:
the series converges for x 2 < 1, regardless of the value of l;
. ,-1;
th~
·o
• ot v +1 unl~oo ;t . '"'~"
Since we want a solution that is finite at x = :+: 1, as well as for x 2 < 1, we demand
~-·~ · · •• c: ~-
'J !!
=~ · ·o
a
'
renee relation (3.14) will terminate only if l is zero or a positive integer. Even
then only one of the two series converges at x = :+:1. If lis even (odd), then only
thP"' ·n (, 1\ <P.riP< · 1~tP< *-ThP nnhmnmi~l< in P~rh r~<P h~\/P r 1 ~<
their highest power of x, the next highest being x 1- 2 , and so on, down to x 0 (x)
for l even (odd). By convention these polynomials are normalized to have the
value unity at x = + 1 and are called the Legendre polynomials of order l, P1(x).
The first few Legendre polynomials are:
Pn(x) = 1 .
P 1 (x) = x
p_(y\ - clcf'h 2 - 1\ (1 1 'i\
f 3(X) 2PX jX)

P 4 (x) = h35x 4 - 30x 2 + 3)
*For example, if l = 0 the " = 1 series has a general coefficient a, = a0 /(j + 1) for j = 0, 2, 4, .... Thus
the series is au(x + ~x 3 + ~x 5 + · · ·). This is iust a0 times the power series expansion of a function
Qofx) = kln(l + x)/(1 - x), which clearlv divemes at x = + 1. For each I value there is a similar function
Q1(x) with logarithms in it as the partner to the well-behaved polynomial solution. See Magnus et al.
(pp. 151 ff). Whittaker and Watson (Chapter XV) give a treatment using analytic functions.
98 Chapter 3 Boundary-Value Problems in Electrostatics: 11-SI
By manipulation of the power senes solutiOns (3~ ana p.l4J it is possime

to obtain a compact representation of the Legendre polynomials, known as
Rodrigues' formula:
1 d'
P1(x) =1- -1 (x 2 - 1) 1 (3.16)
2 1! dx
£ ,] A v+frn~ l
'" •J
VI .'
L' ' ""T
' ' ~ ' ,£ L
1 ue
Leg enure puty lln 111 a ~-o
the interval -1 < x < 1. To prove the orthogonality we can appeal directly to
the differential equation (3.10). We write down the differential equation for P/x),
· · ,]v hv P.Jr) and then inte~>rate over the interval:
r { [ •.
-1
Pr(x) -
d
dx
(1 - x 2 )
. ,
dP1
dx + 1(1 + 1)P1(x) dx = 0 (3.17)
T
+1.. "· • L
y !-"'
'
·~·· "b
J
dP, dPr
fl
-1
[ (x 2 - 1) dx dx
- + 1(1 + 1)(Pr(x)PL(x) J dx = 0 (3.18)
1t we now wrue uown ~-'·HI) witnt a11u, i"'"'

. .
ldllU wuu~•nu~m\••
• £. I~ 1 0\
Jjo
the result is the orthogonality condition:
"" "i ) 1/ / 1 / 1\1 ( n t ..\ n 1.~\ .L (\ (?.

.,
10\
ll~l 'V I '!'\~}' ,,~,
T iJ J
'"
For l i= l', the integral must vanish. For l = l', the integral is finite. To determine
its value it is necessary to use an explicit representation of the Legendre poly-
• • • "''~ ··~ • • .1." ....
r r
, iU~H <U~ -J
, "·!';•, "'UUll!';U<O>
1
N1 =
_
_
1 2 d' 2 I d _ 2 I
[P1(x)] dx- 221 (1!? _1 dx 1 (x - 1) dx 1 (x - 1) dx
1
Integration by parts l times ytelds tne resUlt:

( -1)1
Nl = 2zl(z!)z
r 1
d21
(x2 - 1)' dxzl (xz - 1)1 dx
The differentiation 21 times of (x 2 - 1)1yields the constant (21)!, so that

1
(21)! ( 11 2\1 .J
I " I "J2I(J1\2 I \ J
\"/
The remaining integral can be done by brute force, but also by induction. We
write the integrand as
,/
(1 - x2)t = (1 - x2)(1 - xzy 1 = (1 - xz)' 1 + - - (1 - xz)'
21 dx
1 nus we; 11a vc;
(~2 '(/!~}'
21 1
N1 = (
21
)N1_ 1 + )_ x d[(l -
1
X
2
Yl
"· ..~'> T .... -' I 0 .. 1.
'J .~·· 00
••
UH"b'~U~ •J
~ y~o.o ou '"
u•~ .
•~o•
,I
~·~· J
',U,
N. (2!- 1'lN. 1 N.
\ 21 21
or
(2/ + 1)NI = (2/- 1)NI 1 (3.20)
This shows that (21 + l)N1 is independent of l. For l - 0, w1th P 0 (x) - 1, we have
N 0 = 2. Thus N 1 = 2/(2/ + 1) and the orthogonality condition can be written:
II _ Pr(x)P1(x) dx
1
=
21
2
+
1
81'1 (3.21)
and the orthonormal functions in the sense of Section 2.8 are
U1(x) =
'I
1
t -r
2
1
P1(x) (3.22)
Since the Legendre polynomials form a complete set of orthogonal functions,

any function f(x) on the interval -1 < x < 1 can be expanded in terms of them.
1 ne Legenare series representation is:
00
./~X) "
~ r1. 1r 1 ~XJ ~.l.L..l)
[-0
..1.
A,
21 + 1
L
r
J 1
f(x)P,(x) dx (3.24)
As an example, consider the function shown in Fig. 3.2:

f(x) = +1 for x > 0
.1 f. ~ {I
Then
AI=
21 + 1
2 r 0
P1(x) dx -
Since P1(x) is odd (even) about x = 0 if l is odd (even), only the odd l
r
_ P1(x) dx
1
.rc i"'nt~ :H"' "'a. 1t from 7P.rn Tho" for I nrlrl
rl
A = (2! + 1) I PJx) dx (3.25)
Jo
II +1 :
I
I I
-1 I 10 I 1
: X--+-
I
I
I
-1
I : Figure 3.2
By means of Rodrigues' formula the integral can be evaluated, yielding
- -l)(l-1)/2 (21 + 1)(1- 2)!!

A, - ( 2 l + 1, (3.26)
2 --1!
L
'
,,.., ,,..,
""
~
\• lJ·· \~" 1l " "

!\~" 1l " "
!\~" "J} .) / ".) "- L ~· "'
HIU> lllC: >C:l1C>
tor f(x) Is:

r
H-'J '2 1 '
1\-')
7
8' 3\-'J T
11
16rs~X)
... ~.J.LI)
Certain recurrence relations among Legendre polynomials of different order

are m ev imegra1s, generaringn.igner oroer polynomials trom
lower order ones, etc. From Rodrigues' formula it is a straightforward matter to
_ 1. L1.
"'"" Ul<11
dP/+ 1 dP1-1 f?/+1\P_-n n 'HI\

-,
~
dx dx
~·.
Hll> IC:>Ull, w~w u~Hc•c:u'~"' cy_ua,ion p.1v ), can oe maue LO yiem
various recurrence formulas, some of which are:
~~ + l).tl+l ~Ll + l)Xl'zTTPz-1- ""V

d:~~ - x ~~ - (l + 1)P1 = 0 (3.29)
(x 2 - 1) dPz - lxP, + lP , = 0
ax
As an illustration of the use of these recurrence formulas, consider the evaluation

or rne integral:
rl
J-l -'' [\-'')' 1'\X)' UX'
T .n / \ n
'1 ~.J . .JV)
From the first of the recurrence formulas (3.29) we obtain an expression for
xP1(x). Therefore (3.30) becomes
1 ('
11
21
+ 1 )_ 1 Pr(x)l(l + 1)P1+ 1(x) + lP1 1(x)] dx
The orthogonality integral (3.21) can now be employed to show that the integral
vanishes unless l' = l + 1 and that for those values
2(1 + 1) lI = l
r1 (?/ _J_ 1\/?1 "\ , +1
J_ 1 xP (x)Pr(x) dx =
1
\"
"""
2l
'!
(3.31)
l' I 1
(Ll I }(Ll + l)
Sect. 3.3 Boundary-Value Problems with Azimuthal Symmetry 101
These are really the same result with the roles of l and l' interchanged. In a
similar manner It IS easy to show that
(3.32)
l' = l
where it is assumed that l' > l.
3.3 Boundary- ·value Pmblems with Azimuthal Symmetry

From the form of the solution of the Laplace equation in spherical coordinates
(3.2), it will be seen that for a problem possessing azimuthal symmetry m = 0 in
(3.5). This means that the general solution for such a problem is:
(3.33)
The coefficients A 1 and B 1 can be determined from the boundary conditions.

Suppose that the potential is specified to be V(ll) on the surface of a sphere of
radius a, and it is requin:)d to find the potential inside the sphere. If there are no
charges at the origin, the potential must be finite there. Consequently B 1 = 0 for
all I The coefficients A 1 are found by evaluating (3 33) on the smface of the
sphere:
V(ll) = .2: A a P (cos II)

1~0
1
1
1 (3.34)
This is just a Legendre series of the form (3.23), so that the coefficients A 1 are:
A1 =
21
+,
2a
1
j'"o V(II)P (cos II) sin II dll
1 (3.35)
If, for example, V(ll) is that of Section 2.7, with two hemispheres at equal and
opposite potentials,
(0 < II < 7r/2)

-v, ( 7r/2 < II < 7r)
then the coefficients are proportional to those in (3.27). Thus the potential inside
the sphere is
~
~( r ) P 3 (cos II) + r '

Cl>(r, II) = V 3 ~r P 1 (cos II) - 8
2 a
~~( a) P 5 ( cos II) · · · (3.36)
To find the potential outside the sphere we merely replace (rla) 1 by (a/r) 1+ 1 . The
resulting potential can be seen to be the same as (2 27), obtained by another
means.
Series (3 33), with its coefficients determined by the boundary conditions, is
a unique expansion of the potential. This uniqueness provides a means of ob-
IIIZ r 1 8 - ' ;n
.
. n. -"'
. . ...
W<O OVlUUVH VL a AHV ' ·'
.c
Vl <11'-'
p~ wa1 w
.. '
P'' ·'
r
l l Vlll
a limited domain namelv on the svmmetrv axis. On the svmmetrv axi< n 331
,_ {, ;th__,._
,.
d·
'
'¥\.C Y) "L J.lizY -t DzY 'J ~j.jf)
l 0
v<uiu '"' l''"i,ivc .:;. rut · z eacn term must oe muniptieu oy \ ·1)'. :sup-
pose that, by some means, we can evaluate the potential cl>(z) on the symmetry
axis. If this potential function can be expanded in a power series in z = r of the
form (3.37), with known coefficients, then the solution for the potential at
any point in space is obtained by multiplying each power of r 1 and ,-(1+ 1) by
P1( cos e).
At the risk of boring the reader, we return to the problem of the hemispheres
at equal and opposite potentials. We have already obtained the series solution
in two different ways, (2.27) and (3.36). The method just stated gives a third way.
For a point on the axis we have found the closed form (2.22):
cl>(z = r) = V [1 - V,z 2 a2J

r r +a2
This can be expanded in powers of a2 /r 2 :
cl>(z = r) = ~
v:ii 1~1
i (-l)f-1 (2j -1)~(j- ~) (~r
J. r
Comparison with expansion (3.37) shows that only odd l values (/ = 2j - 1) enter.
r
The solution, valid for all points outside the sphere, is consequently:
cl>(r, e) = ___!:::__ i(
v:ii 1 ~1
-1 )i- 1 (
2
j - !)~(j -
J.
}) ( ~
r
P 2;- 1(cos e)
This is the same solution as already obtained, (2.27) and (3.36).

An important expansion is that of the potential at x due to a unit point charge
at x':
1 oc I
2: r~:> 1 P 1(cosy)
Ix _ x 'I= t~o (3.38)
where r < (r >) is the smaller (larger) of IxI and Ix' I, and y is the angle between x
and x', as shown in Fig. 3.3. This can be proved by rotating axes so that x' lies
x'
~
I~
_L_'
·I ~ X
"I _.......~
I ' ~ .....::"
/~
/
./
II/
-
........... --~
'-.. I - - J.
_i
I
y
X --J Fhmre 3.3

Sect. 3.3 Boundary-Value Problems with Azimuthal Symmetry 103
along the z axis. Then the potential satisfies the Laplace equation, possesses
azimuthal symmetry, and can be expanded according to (3.33), except at the point
x = x':
1
'V
L; f " l
v·•zr + B zr (l' ll)Pz(cos ")
'
l ~0
If the point xis on the z axis. the right-hand side reduces to (3.37), while the left-
hand side becomes:
1
lx x'l lr r' I
Expanding, we find, for x on axis,
1
lx x'l
For points off the axis it is only necessary, according to (3.33) and (3.37), to
multiply each term by Pt(cos y). This proves the general result (3.38).
Another example is the potential due to a total charge q uniformly distrib-
uted around a circular ring of radius a, located as shown in Fig. 3.4, with its axis
the z axis and its center at z = b. The potential at a point P on the axis of
symmetry with z = r is just q!4TrE0 divided by the distance AP:
1 q
<P(z = r) = - ------::---...,--_2_--~
4TrE0 (r + c - 2cr cos a) 112
2 2
where c2 a2 + b 2 and a tan 1

(alb). The mverse d1stance AP can be expanded
using (3.38). Thus, for r > c,
<P(z = r) = _q_
47rEo
For r < c, the correspondmg form 1s:

00 l
<P(z = r) = _q_
4TrE0
.2; c~+I P 1(cos a)
z~o
z
z=r P
/
/
/
/
/
/
\
\
C\f I
\
-V
/
/ Fivnre 3.4 Rin~> of charge of radius a and total
,/ charge q located on the z axis with center at z = b.
q r~
<P(r, II) = - -
4 7TEo
L 00
l=O
1+1 P 1(cos a)P1(cos II)
r>
where r< (r>) is the smaller (larger) of rand c.
3.4 Behavior of Fields in a Conical Hole or Near a Sharp Point

Before turning to more complicated boundary-value problems, we consider one
. .
surrounding a pointed conical conductor.

The treatment of Section 3.2 for the Legendre differential equation needs
modification. With the assumption of azimuth symmetry, (3.1 0) is still applicable,
but we now seek solutions finite and single-valued on the range of x = cos II of
cos f3 < x < 1. Furthermore, since the conducting surface II = f3 is at fixed po-
tential, which we can take to be zero, the solution in cos II must vanish at II = f3
to satisfy the boundary conditions. Since we demand regularity at x = 1 it is
convenient to make a series expansion around x = 1 instead of x = 0, as was
done with (3.11). With the introduction of the variable
g= W- x)
Figure 3.5
Sect. 3.4 Behavior of Fields in a Conical Hole or Near a Sharp Point 105
where l has been replaced by v to avoid confusion. The corresponding radial

solutions for U(r)lr in (3.2) are r" and r " '.With a power senes solutwn .
.;.,
q~) ~- L ai~
j -tl
.~ ;, ••. rl ' •• _,n..... ,, ' "

'10\ . "1- '
,j' ' "
.£"' ' ~+ ,j' +h I
' F ~· c~
•eyuut" a V. HlC ICC I ll . UCLWCCll ve ILS ll1 lllC SCllCS
is then
a +I ( i - v)( i + v + 1)
~j.4U)
2
UJ \1 ' -'}
r
r"L
0 uo
1
i.V ·" •'-
cu~
.L
ovw"vu .v uuuy U< ~
A
v
/
\~vo "
u "-')• I.
~"' ~
tne senes representation
~ V)~V -t- l) ,. ~ V)~ V -t- l)~V -t l)~V -t L) ,.2

n I r\ 1 /0 A1 \
' v\~J i
~ ~ \ .. '!
11 11 ?I ?I
' .
WP fir,t th~t if " i< 7PTn nr ~ intPC>PT thP 'IMP< ThP
,,_ £. ,,_ '"I .i: /.-, A>\ •
.J
~UU ' •~u'J WU< lVI <J V I
A
V 0
1
i 0 ~•••• , "'~ 0~"~0 \•. 10
Legendre polynomials (3.15). For v not equal to an integer, (3.41) represents a

generalization and is called a Legendre jimction of the first kind and order v. The
series (3.41) is an example of a hypergeometric function 2 F1(a, b; c; z) whose
' . '
senes expansiOn 1s
) = 1 + ab ~ + a( a + 1)b(b + 1) z + ...
2
F (a b· c
2 1
' ' ' z c 1! c(c + 1) 2!
Comparison with (3.41) shows that the Legendre function can be written
Pv(x) = 2F1( -v, v + 1; 1; 1 -X)

2
(3.42)
Here we have returned to our customary variable x = cos e. The properties of

the hypergeometric functions are well known (see Morse and Feshbach, Chapter
5, Dennery and Krzywicki, Sections IV.16-18, Whittaker and Watson, Chapter
XIV). The Legendre function Pv(x) is regular at x = 1 and for lxl < 1, but is
singular at x = -1 unless v is an integer. Depending on the value of v, it has a
certain number of zeros on the range lxl < 1. Since the polynomial P1(x) has l
zeros for lx I < 1, we anticipate that for real v more and more zeros occur as v
gets larger and larger. Furthermore, the zeros are distributed more or less uni-
formly on the interval. In particular, the first zero moves closer and closer to
x = 1 as v increases.
The basic solution to the Laplace boundary-value problem of Fig. 3.5 is
Ar"Pv(cos e)
where v > 0 is required for a finite potential at the origin. Since the potential
must vanish at e = f3 for all r, it is necessary that
Pv( COS {3) = 0 (3.43)
This is an eigenvalue condition on v. From what was just stated about the
zeros of Pv it is evident that (3.43) has an infinite number of solutions, v = vk
(k = 1, 2, ... ), which we arrange in order of increasing magnitude. For v = v"

x cos f3 is the first zero of P,, (x) For l' l':z, x cos {3 is the second zero
of Pv,(x), and so on. The complete solution for the azimuthally symmetric poten-
tial in the region 0 < fl < fl is*
'P(r, II) (3.44)
In the spirit of Section 2.11 we are interested in the general behavior of the
potential and fields in the neighborhood of r = 0 and not in the full solution with
specific boundary conditions imposed at large r. Thus we approximate the be-
havior of the potential near r 0 by the first tem1 in (3.44) and write
cl>(r. II) ~ Ar"P,.(cos II) (3.45)

where now vis the smallest root of (3.43). The components of electric field and
the surface-charge density on the conical conductor are
r
ar
sin liP~( cos II)
1
Ee = -; all = Ar" (3.46)
A
Sill
Here the pnme on Pv denotes drfferential wrth respect torts argument. The fields
and charge density all vary as r"- 1 as r ~ 0.
The order v for the first zero of Pv( cos {f) IS plotted as a function of f3 Ill hg.
3.6. Obviously, for {3 <<1, v >> 1. An approximate expression for vin this domain
can be obtained from the Bessel function approximation, t
Pv(cos 8) J0 ((2v + 1) sm ~) (3.4 /)
valid for large v and fl < 1. The first zero of fo(x) is at x = 2.405. This gives
2.405 1
Since IF I and cr vary as r"- 1 there are evidently very small fields and very little
charge deep in a conical hole as {3 ~ 0. For {3 = 7T/2, the conical conductor
becomes a plane. There v- 1 and cr cc 1, as expected For f3 > n/2, the geometry
is that of a conical point. Then r < 1 and the field is singulm at t 0. For
B ~ 7T, v ~ 0, but rather slowly. An approximation for ( 7T - fl) small is
(3.48b)
This shows that for ( 7T - {3) = 10°, v = 0.2 and even for ( 7T - {3) = 1°, v = 0.1.
In any event, fm a nan ow conical point the fields near the point vary as r ' '•
*T1Jc mthoguuality of the functions Pvk(cos 6} on the interval cos {3 < x < 1 can De shown in the
same way as for P 1(cos 8) sec (3.1 /) (3.19). Completeness can also be shown.
tBessel functions are discussed in Section 3 7
Sect. 3.5 Associated Legendre Functions and the Spherical Harmonics Y1m(9, <)I) 1U7
4-
3f-
t
I r-
\
• \
2f- \
\
~
\, I
I
\
1-
' , ,__ ' \
n
0
I I I I I I I I
90
I I I I I I~ 180
fj (degrees) -
Ngure j,(J lhe order parameter v tor the hrst zero ot l'"(cosf3) versus f3. I he range
ll < (.! < Qll" ~. tn o · >I hniP whiiP Qll" < (.! < 1!lll" •« o ronirol
point. Near r = 0 the fields and surface-charge density are proportional to rv-l The
dashed curves are the approximate expressions, (3.48a) and (3.48b ).
where E << 1. Verv hi!!h fields exist around the ooint. The efficacy of such ooints
in lightning rods is discussed in Section 2.11.
. .
All ex1enueu uJscuss1on 01 pOLenllat pro01ems 01 LUIS genera1 Klllu uy K. ~~.
Hall [J. Appl. Phys. 20, 925 (1949)] includes graphs for a number of the roots vk
Vl ~.J.'t.J) ,, . Vl P·
3.5 Associated Legendre Functions and the

Spherical Harmonics Y1m(O, l/>)
So far we have dealt with potential problems possessing azimuthal symmetry with
solutwns ot the torm (5.55). Unless the range m tJ JS restnctea, as m :sect10n 5.4,
these involve only ordinary Legendre polynomials. The general potential prob-
tern can, nowever, nave aztmutnal vananons so mar m '1' u m (5.)) ana (j.'ol ).
ThPn UJP nPPci thP >li7Minn nf P.( rn< fl) n,m,]v thP <nlntinn nf {i Q) UJith
.~
tJdllU ~rt ~Vlll dlUHldlJ. 1:1 . lllC MlllC "' lVl lllC
.<e<>.enClr.e n · it "~n n.e -'- thM to h~ve finite solutions on the intervH I
-1 < x < 1, the parameter I must be zero or a positive integer and the integer m
can take on onlv the values -1 -(1 - 1),... 0 . . . (I - 1). I. The solution
having these properties is called an associated Legendre function PZV(x). For

positive rn it is defined by the formula*·
( 3.49)
If Rodrigues' formula is used to represent P1(x), a definition valid for both pos-
itive and negative m is obtained:
( l)m dl+m
pm(x) = (1 - x2)m!2 (xz - 1)1 (3.50)
I • 211! dxl+m
P[m(x) and P[(x) are proportional, since the differential equation (3.9) depends
only on m 2 and m is an integer. It can be shown that
(l- m)!
P[m(x) = (-l)m (l + m)! P[(x) (3.51)
For fixed m the functwns P!"(x) form an orthogonal set m the mdex l on the
interval 1 < x < 1. By the same means as for the Legendre functions the
orthogonahty relatwn can be obtamed:
2 (l + m)!
J -I
f''P(x)Pr(x) dx 8 (3.52)
The solution of the Laplace equation was decomposed into a product of

factors for the three variables r, e, and <f>. It is convenient to combine the angular
factors and construct orthonormal functions over the unit sphere. We will call
these functions spherical harmonics, although this terminology is often reserved
for solutions of the generalized Legendre equation (3.9). Our spherical harmonics
are sometimes called "tesseral harmonics" in older books. The functions
Qm( 4>) = e;m < 2-rr. The functions P[(cos 8) form a similar set in the index l
for each m value on the interval -1 < cos e < 1. Therefore their product P[Qm
will form a compl@te orthogonal S@t on th@ surface of the unit sphere in the two
indices l, m. From the normalization condition (3.52) it is clear that the suitably
normalized functions, denoted by Ybit(A, rp), are
l2! + 1
I (3.53)
v 417
From (3.51) it can be seen that
Yz,-m(e, 4>) = (-1)"'Y'l'm(e, 4>) (3.54)
The normalization and orthogonality conditions are
(3.55)
The completeness relation, equivalent to (2.35), is
8(1> 1> ')B(cos e cos IJ') (3.56)

I 0 m
*The choice of ph"'e fur Pi(<) i• that of ~tlugnus et al. and E. Il. Condon and G. H. Shortley in
The my vfAtumic Spect1a, Cmt1h1 idge Uui~e1 sity Pt e.<>.o; (1953). Fm explicit exp1 es.<>ious aud 1eettJ.'iion
formulas, see 1'Vlagnus et at., Seen on 4.3.
Sect. 3.5 Associated Legendre Fnnctions and the Spherical Harmonics Y1m(6, <!>) 109
For a few small/ values and m 2: 0 the list below shows the explicit form of the
Y 1m(R, cf>). For negative m values (3.54) can be used.
<;:DTJDDT<' 'T J-:r 'D V {CI A.\
un...._,,./
1
l = 0 Yoo = ;,-
v "T "
yll = -vri- 0"

sin ee;q'
l = 1
v fi "
"' 'lj47T
vu 1@ . 2a. z;q,
4 '1/ 27T
l L I 21 · ~·
!15sm tJ cos tJe
87T
V. - [I r' 2(1 l\
"" y 47T'" u
~
y33 = -i ~;: 3
sin ee3i.T.
''VW
....
_,~.,
c•
HH " "v,
Y,.J R rh) /21 + 1 PJms R) (i 'i7)
v 47T
An arbitrary function g(e, 4>) can be expanded in spherical harmonics:
g~tJ, 'P)
,.; "'
L; L. Flzmi lm~tJ, 'P) ~.J • .JO)
l=O m=-l
th ,j"f;, '
A - Jf dO Y* (R rh)<Y(R rh)
A point of interest to us in the next section is the form of the expansion for
...
u. vv llll ~ ·'--'' ), we uuu:
"
- ~ /21 + 1
[g(e, 4> )]e~o - L ,/ 4
.,. Azo (3.59)
where
Am =
fi;if dD P1( cos 8)g( 8, 4>) (3.60)
All terms in the series with m 0 vanish at 8 = 0. *

The general solution for a boundary-value problem in spherical coordinates
can be written in terms of spherical harmonics and powers of r in a generalization
of (3.33):
~ I
<P(r, 8, 4>) = 2: 2:
l=O m=-1
[Aimr 1 + B1mr-(l+l)]Y1m(8, 4>) (3.61)
If the potential is specified on a spherical surface, the coefficients can be deter-

mined by evaluating (3.61) on the surface and using (3.58).
3.6 Addition Theorem for Spherical Harmonics

A mathematical result of considerable interest and use is called the addition
theorem for spherical harmonics. Two coordinate vectors x and x', with spherical
coordinates (r, 8, 4>) and (r', 8', 4> '),respectively, have an angle I' between them,
as shown in Fig. 3.7. The addition theorem expresses a Legendre polynomial of
order lin the angle I' in terms of products of the spherical harmonics of the angles
8, 4> and 8', ¢':
I
PI( cosy) =
41T
2! + 1 m~-1 Yim(8', ¢')Yim(8, ¢)
2: (3.62)
where cos I' = cos 8 cos 8' + sin 8 sin 8' cos(¢ - ¢ '). To prove this theorem we
consider the vector x' as fixed in space. Then P1(cos y) is a function of the angles
. .
<",cp;-wrnn:rre- aug1es t1 , q; as par . H may oe . m a senes p.::li'l J.
I'
~-
'1\'.Ah YJ' "'
L.J "'
L.J '"'
.nl'm\U, ""I'm\"•
'f' } ' ' 'f'}
.' '~
\·J.UJ}
r~'
I =U m -L
7 ionn ~ ~;t n~ with 11 I " Tn OPP mhu

..
llllS
..
" LOUUI
lS S<J, llULe Llldl 11 :axes are
-J
1 su Llla Lx " uu LHe ~ axi>, llleu

--;;7 becomes the usual nolar an crle and p.r ('flo " ' '~ti,fiP, thP Prm ~tinn ·
Ul+ll
v - f'l( cos )I) -+ 1'1( cos 'Y) u (j,tJ4)
r
11\01\011\0U' LV .111<0'
' \0 1110 <lA\0,, 11 LH\0 dA\00 d1\0 llVVV 1UldlC:U'
~12.
'
V1KHO , ,,ll1'- ' ' ~.
TP ' '
to the nosition shown m Fi~>. 1.7 v"L - V'" and r is unchan~>ed.* Conseouentlv
--;;-; , \ ~ . " .. ,_ . "-
'~ '"'·.1,._,, u 10 a -r
.£d. £ -
-n: --n cr 1lldl ~.v~h
-
ot order 1. 1 n1s means mat It IS a linear combmation ot r ~~ s ot tnat order ontv:

~
----.rzc= 1J 2
m
7 f,;;-c:u
I
;cp )1' lm (tJ, <P) p.b))
'T'< A {£JI J. 1\ <

m\' ' I . 6' uy
--;:
,j (QI A..'\ I v* (£J -1.\P(o~.- ..\ d() f'l t;t;\
tm\'-' 'I \' , "I
' I
I
·~L -< ,L " ' '1'7'2

...,, ~-
,_ oL n2.'- ~ ~-.
't' 'b
''.'
Upol dLUI >CdliU . a11u <HaL all >cawr ; arc UIIUCI
Sect. 3.7 Laplace Equation in Cylindrical Coordinates; Bessel Functions 111
z
x'
8' Li'Y ~X
I /
~I
y
X Figure 3.7
To evaluate this coefficient we note that it may be viewed, according to (3.60),

as them' = 0 coefficient in an expansion of the function V 47T/(2/ + 1) Yim(e, cp)
in a series of 1;,.·(1~ [3) referred to the primed axis of (3 64) From (3 59) it is
then found that, since only one l value is present, coefficient (3.66) is
(3.67)
In the limit y ~ 0, the angles ( e, cp ), as functions of ( y, {3), go over into (6', cp ').
Thus addition theorem (3.62) is proved. Sometimes the theorem is written in
terms of PI( cos e) rather than Yzm· Then it has the form:
P 1( cosy) = P1( cos 6)P1( cos 6') (3.68)
+2 ±~!
m-1 l +m .
m~; P'/'( cos II)P/"( cos II') cos[m( i/) i/)' )]
If the angle 'Y goes to zero, there results a "sum r ale'' for the squates of Y 1m's.
2:'-l IYlm( e, cp) I = 2/41T+ 1

I
2
(3.69)
The addition theorem can be used to put expansion (3.38) of the potential
at x due to a unit charge at x' into its most explicit form. Substituting (3.62) for
P1( cosy) into (3.38), we obtain
1
lx x'l 47T2:
oc
2:
1
1~0 m~-1 21 + I r
I
f
1 Y;',(Il',</I')Y,,(Il,</l) (370)
Equation (3.70) gives the potential in a completely factorized form in the coor-
dinates x and x'. This is useful in any integrations over charge densities, etc.,
where one variable is the variable of integration and the other is the coordinate
of the observation point. The price paid is that there is a double sum rather than
a single term.
3. 7 Laplace Equation in Cylindrical Coordinates;

Bessel Functions
In cylindrical coordinates (p, r:p, z), as shown in Fig. 3.8, the Laplace equation
takes the form:
(3.71)
7
:
_/ ' '1 "
The separation of variables is accomplished by the substitution:

<P(p, ¢, z) = R(p)Q(cp)Z(z) (3.72)
In the usual way this leads to the three ordinary differential equations:
d 2Z 2
- - + v2 Q = 0
d¢2
(3.74)
(3.75)
The solutions of the first two equations are elementary

Z(z) = e+"z (3.76)
Q( ¢) = e"'"'"'
For the potential to be single-valued when the full azimuth is allowed, v must be
an integer. But barring some boundary-condition requirement in the z direction,
the parameter k is arbitrary For the present we assume that k is real and positive
The radial equation can be put in a standard form by the change of variable
x = kp. Then It becomes
I hts ts the Bessel equatiOn, and the solutiOns are called Bessel junctions of order
F. If a pOW€)£ S()fi()s solution of th€) form
(3.78)
is assumed, then it is found that

a= :!::v (3.79)
and
1
2j 2j 2
4j(j + a)
Sect. 3.7 Laplace Equation in Cylindtical Cootdinates, Bessel Functions 113
for j = 1, 2, 3, .... All odd powers of xi have vanishing coefficients. The recursion
formula can be iterated to obtain
(-1)lf'(a + 1)
azJ = 22ij! f'(j + a+ 1) ao
(3.81)
It is conventional to choose the constant a0 - [2af(a + 1)]- 1• Then the twc

solutions are
(3.82)
J v(x) (3.83)
These solutions are called Bessel functions of the first kind of order + 1'. The
series converge for all finite values of x. If vis not an integer, these two solutions
J" v(x) form a patr of 1mearly mdependent solut10ns to the second-order Bessel
equation. However, if v is an integer, it is well known that the solutions are
linearly dependent. In fact, for v m, an integer, it can be seen from the series
representation that
(3.84)
Consequently it is necessary to find another linearly independent solution when
v is an integer. It is customary, even if v is not an integer, to replace the pair
J=v(x) by Jv(x) and Nv(x), the Neumann function (or Bessel function of the sec-
ond kind):
sm v1r
For v not an integer, N,(x) is clearly linearly independent of J,.(x). In the limit
v-> integer, it can be shown that N"(x) is still linearly independent of J,,(x). As
expected, it involves log x. Its series representation is given in the reference
books.
The Bessel functions of the third kind, called Hankel functions, are defined
as linear combinations of J,,(x) and Nv(x):
H~1 l(x) = J,.(x) + iNv(x)}
(3.86)
H~2 l(x) = J,.(x) - iNv(x)
The Hankel functions form a fundamental set of solutions to the Bessel equation,
just as do J,.(x) and N,.(x).
'fh C 11H:C
f f lOftS Tl'' .1Y
J
u(l)
" l'' .u v , u(Z)
.L.L v a II Saf IS f y
· thc fCCtlfSIOH:
· . f~fffiUIas
(3 87)
(3.88)
where fl,.(x) is any one of the cylinder functions of order v. These may be verified
directly from the series representation (3.82).
For reference purposes, the limiting forms of the various kinds of Bessel
function are given for small and large values of their argument. For simplicity,
we show only the leading terms:
v
1
1 ,x,
v <'<' T fv\ ~ n SIO\
f(v + 1) 2
2 X
-
In + 0.5772 ... v=O
1T ,L_
Nv(x)-+ (3.90)
_r~) (~r v=!=O
In these formulas v is assumed to be real and nonnegative.

r;J "~ ~
X>> 1, V fv(x) -+ 'J -cos( x - - - - j

1TX 2 4 n o1\
12 I
V1T 1T
JV,.~x)-+ ~ sm X
1TX 2 4
The transition from the small x behavior to the large x asymptotic form occurs
in the region of x ~ v.
From the asvmototic forms (3.91) it is clear that each Bessel function has an
infinite number of roots. We will be chiefly concerned with the roots of Jv(x):
fv(xvn) = 0 (n = 1, 2, 3, ... ) (3.92)
X is the nth root of J .. (x ). For the first few integer values of v the first three
roots are:
v = 0, Xon = 2.405, 5.520, 8.654, ...
v = 1, X1n = 3.832, 7.016, 10.173, ...
v = 2, Xzn = 5.136, 8.417, 11.620, ...
--rur ., "TOOLS, Lllt: ~J ·r
Xvn = n1T + (v- ~) 1T

L
gives aaequate accuracy"("to at least three figures). Tables of roots are given in
'~" <~n\
-;:_,
,•• J T ••
,-, J JI .• ·
""'} u•u.n.u•uu•u vu.c
J <'. I
~l'. •v>)· ~f'·
' '" ·a
tlavmg touna rne so muon or tne raa1al part or tne Laplace equatiOn m terms
~~ D~oo~l ~-. • '"n ~n~ - ~,., ol " .,l, ' ' •• D ' ' ,J ~' o<" <'.
orthogonal, complete set of functions. We consider only Bessel functions of the

first kind, and we show that V p Jv(x,,npla), for fixed v > 0, n = 1, 2, ... , form
illl
,,
'O
' ., .
I 'COl Vll 11100 llllCOl Vill V
. ~
fl U.
~·
UlCO
. .
Ml<lllVll Milll'
.
W llll Ult;;
differential eauation satisfied bv I.(x .o/a):

.- \ -
I
1 -1
dJ:( Xvn ~ I 2 '.2 \ \
vn X,,,!:: I
p +!
pdp dp az p2 J,, a
0 (3.93)
Sect. 3.7 Laplace Equation in Cylindrical Coordinates; Bessel Functions 115
If we multiply the equation by plv(xv,-Pia) and integrate from 0 to a, we obtain
dfv( Xvn ~)
f Jv( Xvn' ~) :p p
dp
dp +
2\ I \
p \ Ti
I
(" I 2 p
Xvn v I TI
vn I A. {\
Jo a2 p" I' v vn v ·r
\ a; a;
Integration by parts, combined with the vanishing of (plvf~) at p = 0 (for v > 0)
and p = a, leads to the result:
dJv Xvn' {J) dfv( Xvn {J \

{" , a1 , a1 {" Xvn
2 v2\ p T p
-1 T -1 r.
v
Jo I' dp dp
~I'
Jo a2 p 2 Jt-'"JJ
1
""'vn' l"lJ ""vn
a1 , a
~I'
If we now write down the same expression, with n and n' interchanged, and
subtract, we obtain the orthogonality condition:
a
p
(x~n - x~~)
JU
pl,.l Xvn' {J fv( Xvn -
u; u,
l dp = 0 (3.94)
Adroit use of the differential equation, and the recursion formulas (3.87) and
. . .
~ CJ.oo J teaus LO we normanzanon mLegra.:
ra I \ I ?
), pJvl Xvn 1:!. fv( Xvn l:!.j dp = u [J,,+l(Xvn)] 2 8n'n (3.95)

o a a, 2
Assuminr> th8t the set of u. ~I f11nctions is comnlete. we c8n exn8ncl 811 8

function of p on the interval 0 < p < a in a Fourier-Bessel series:
00
,..'
f(p) - 2.-_ Avnfv[ Xm .!..... I (3.96)
n=l a
where
2
Avn = 2j2 .(J ,)
f" pj(p)J, XvnP
- ) dp
\ (3.97)
a e< Xm 0 a
Our derivation of (3.96) involved the restriction v > 0. Actually it can be proved
to hold for all v > -1.
Expansion (3.96) and (3.97) is the conventional Fourier-Bessel series and is
P"' Ll\CI 'J uppn;pildLC LV l Uldl /UH>OH " ' I' u ~e.g., JU'
Dirichlet boundary conditions on a cylinder; see the following section). But it
will be noted that an alternative expansion is possible in a series of functions
vP J,.(Yvnpla)
. .
where Yvn is the nth root of the equation [dlv(x)]!dx = 0. The
. . . .
1' wat, 111 prOVIHg we onnogonanty 01 we tuncuons, au tnat 1s uemanueu
is that the quantity rpUkp)(dldp)J,(k'p)- pUk'p)(dldp)JJkp)l vanish at
the end points p = 0 and p = a. The requirement is met by A = x,.nla or A =
Yvnla, where lv(Xvn) = 0 and J~(Yvn) = 0, or, more generally, by p(dldp)J,.(kp)
-r JU,.V'PJ vat tne en~ J;;mms, wun A a constant muepenuem or K. tne expan-
sion in terms of the set 1
J,.( v ,pia) is especially useful for functions with van-
ishing slope at p =a. (See Problem 3.11.)
A Fourier-Bessel series is only one type of expansion involving Bessel func-

tions. Some of the other possibilities are:
Neumann series:
n=O
2: anlv+n(Z)
n=O
Schlomilch series: L a,J"(nx)

n=1
The reader may refer to Watson (Chapters XVI-XIX) for a detailed discussion
of the properties of these series. Kapteyn series occur in the discussion of the
Kepler motion of planets and of radiation by rapidly moving charges (see Prob-
!ems 14.14 and 14.15).
Before leaving the proper ties of Bessel functions, we note that if, in the
separation of the Laplace equation, the separation constant k 2 in (3.73) had been
taken as k 2 , then Z(z) would ha~e been sin kz or cos kz and the equation for
R(p) would have been:
(3.98)
dp 2 pdp
W1th kp - x, th1s becomes
d2R + ~ dR - (r + ;:)R -0 (3.99)
The solutions of this equation are called modified Bessel functions. It is evident
that they are just Be~~el functions of pure imaginary argument The usual choices
of linearly independent solutions are denoted by lv(x) and K,,(x). They are de-
fined by
(3.101)
and are real functions for real x and v. The1r hmlfmg forms for small and large
x are, assuming real I' > O·
(3 102)
'x
In + U.)I/L · · · , v- u
2
-=;;-exT=;> p.LU.))
r(v) I)' v
? . y
, {)
(3.104)
Sect. 3.8 Boundary· Value Problems in Cylindrical Coordinates 117
3.8 Boundary- Value Problems in Cylindrical Coordinates

The solution of the I aplace equation in cylindrical coordinates i~ <P
R(p )Q( cp )Z(z), where the separate factors are given in the previous section. Con-
sider now the specific boundary-value problem shown in Fig. 3.9. The cylinder
has a radius a and a height L, the top and bottom surfaces being at z = L and
z = 0. The potential on the side and the bottom of the cylinder is zero, while the
top has a potential <P = 1l(p, cp ). We want to find the potential at any point inside
the cylinder. In order that <P be single valued and vanish at z = 0,
Q(cf;) =A sinmcf; + B cosmcp

Z(z) = sinh kz
where v = m is an integer and k is a constant to be determined. The radial factor

IS
R(p) = Clm(kp) + DNm(kp)
If the potential is finite at p = 0, D = 0. The requirement that the potential vanish

at p a means that k can take on only those special values:
Xmn
k
a
where Xmn are the roots of lm(Xmn) = 0.

Combining all these conditions, we find that the general form of the solution
IS
~ ~
<:fl(p, ¢>, z) Z Z
m=O n=1
lm(kmnP) sinh(kmnZ)(Amn sin m¢
+ Bmn cos mcp)

At z= L, we are given the potential as 1/(p, cp ). Therefore we have
m.n
z
/ t = V(p,c/>)
Ll
R
/
'
t=O- ;...,...
a
~-
~,.~-- ---,
/ ' '-t-0
-r
/ .... c;y.,
~~0
~LLO \_,lldfJltOl CJ • 'ilJU" C Ill
. .....,,.
This is a Fourier series in cp and a Fourier-Bessel series in p. The coefficients

are, from (2.37) and (3.97),
- ' ""mnL) T r2~
L ra
Amn
L
2 r2
"11T'OO
n ,\ dcr I dp pV(p, cp )1m(kmnP) sin mcp
' ~~ "'""'-'
\·•'V-'U)
-
L
, '
'1\"mnL.
" J r2~
dcp
ra
Jo dp pV(p, cp)lm(kmnP) cosmcp

Bmn
Tra ]~+1(kmna)
2
o
with the proviso that, form = 0, we use ~Bon in the series.

The particular form of expansion (3.105a) is dictated by the requirement that
tha ,t;~1 · ~t ~ tl f n r ' • "J ~ ~nrl ~t r~ n fnr ' • ~ l:'nr
r- "J
different boundary conditions the expansion would take a different form. An

example where the potential is zero on the end faces and equal to V( cp, z) on the
.,
--s:rrn::;sl
- . - ~
C1K lCll d' rlUUlClll CJ.';/ lUl lllC .
The Fourier-Bessel series (3.1051 is annronriate for a finite interval in n.

0 < p < a. If a ---+ oo, the series goes over into an integral in a manner entirely
analogous to tne transltwn trom a tngonometnc .l:'ouner senes to a .l:'ouner m-
tegral. Thus, for example, if the potential in charge-free space is finite for z > 0
and vanishes for z ---+ oo, the general form of the solution for z > 0 must be
X
( n. m zI = f
m=oJo
dk e-"zJ .. (knlrA .Jkl sin mm + B .. (k) cos mm 1 (3.106)
lf111e poTenT1a11s spec1hed over the whole plane z - U to be V(p, cp) the coeth-
n;anh• ~ra rl • ' h,
"J
X r~
V(p, c/J) = m2;0 1a dklm(kp)[Am(k) sinmcp + Bm(k) cosmcp]
The variation in cp is just a Fourier series. Consequently the coefficients Am(k)

and Bm(k) are separately specified by the integral relations:
.!_ ,~ vro. sinmcp ~
<b) Am(k')
dm = J.Jk'o) dk' (3.1071
Tr Jo cosmcp_ - () lilm(k')
These radial integral equations of the first kind can be easily solved, since they
TT -'~' ;: ;., ,, • .< .< •
'"1 'u pu" r- """ -ow"
' '
00
1 ' " ' ' " '-')' UA

' "T ~" '
Jo AJm\"-')JmV' k U\" ")
' '"
\J.lUO)
'~n
can be exploited to invert equations (3.107). Multiplying both sides by plm(kp)

and mtegrating over p, we find with the help of (3.108) that the coefficients are
-' . -' '~
cry . .. ' • '' -~•~ -' r •'~
"'V"'- m '-" v" "''- r
-' . ._ "·
v .
'"
2~
I
~
flm\K) sm mcp
n 11.\
m\"1
= K dp p I dcp V(p, cp)Jm(kp)
-~
(3.109)
' . .
~ V,~~'CllC' [j.lVOJ.
While on the subject of expansions in terms of Bessel functions, we observe
•'- "- r_ • T /,~ ~' <~ "'- -' ~ / ' ' <~ ~' " '
v\' ') •U v, "-'-\V) i 0 "VHU U '-V>HP"'-''-• V"
Sect. 3.9 Expansion of Green Functions in Spherical Coordinates 119
(ink) set of functions on the interval, 0 < x < oo. For each m value (and fixed
cp and z), the expansion ink in (3.106) is a special case of the expansion,
A(x) = L A(k)lv(kx) dk, where A(k) = k Jo xA(x)lv(kx) dx (3.110)
An important example of these expansions occurs in spherical coordinates, with

spherical Bessel functions,j 1(kr), l = 0, 1, 2, .... For present purposes we merely
note the definition,
r;
}I~<.) y2z Jz+112~<.J ~.J.lll)
[Details of spherical Bessel functions may be found in Chapter 9.] The ortho-
O"Ofl:l 1itv H'>1Mion (:\1 mn t1v 1' ·~
r~ ~
rLh(kr).if(k'r) dr = k 15(k - k') (3.112)

Jo 2 2
The completeness relation has the same form, with r--" k, k--" r, k'--" r'. The
~ • . 1. • . 1 ~ • £. • 1 • cL
< VUU'-'1" OIJH'-'U'-'al 1..1'-''>''-'1 ·r lVI a IS'''-'" 1 10 " ' ' - ' "
(X • - 2k2 r~ ,.
.nv J Jo 2 '. " '
r1~1\c)}l\'"') ((f\c' \'VlHOICO r1~1\c)
7T Jo ' .n.v "'~"" J "' ~J.IIJ)
Such expansions are useful for current decay in conducting media or time-
denendent ma2:netic diffusion for which an2:ular svmmetrv reduces consideration
to one or a few l values. See Problems 5.35 and 5.36.
.J.Y nxp • ~
UJ u. ~~ .. r •
rtUiti:i
•
tn ·r
• .
, IC.:UI -
•
~~
..
To handle problems involving distributions of charge as well as boundary values
for th.e 1ti:1 1 (i " ,o111tion' of th.e Poi"on <en11Mion) it i' """"""~ru to rl.et.er-
mine the Green function G(x, x') that satisfies the appropriate boundary con-
d1t10ns. Otten these boundary condltwns are speclhed on surfaces ot some sep-
arable coordinate system (e.g., spherical or cylindrical boundaries). Then it is
. .
convement to express rne ureen runcnon as a senes or proaucts or me runcnons
Q· ;QtP tr> thP ~;n~tP • ;n """"';"" WP firot ;11"' thP tunP r>f Pvn~n.
-rr r Jr r
sion involved by considering spherical coordinates.
For thP. ""'" nf nn '- · "'rf"""' :1t · 1
w.e "1r.e:1rlv h:1v.e th.e
expansion of the Green function, namely (3.70):
I
r
.1
I
I
= 47T .2: .2:., ..,,
to•v m
.1
1
,:, Yim(O', cp')Yzm(O, ¢)
I~ ~ ~· >
LL ~ £.
"'-.n..- lHUI \'VCO \'VIMI IV a '-'"1 "" "''-' '-" '-''-'" up-
propriate for the "exterior" problem with a sphencal boundary at r - a. The
.1. • .1:1. £. ~C. <L • C, .C<L r<. C• • {.-,1,;:\ TT.'
I~OUH 10 J_ IIVIH IU~ IV .... V' UL~ ~·~~U LUU~UVU ,~.IV f• -o
expansion (3.70) for both terms in (2.16), we obtain:
C1
G(x x') 47T ).

1 r~ 1 r a" 1 Y*,n(O', a rr
To see clearly the structure of (3.114) and to verify that it satisfies the boundary
conditions, we exhibit the radial factors separately for r < r and for r > r 1 1
:
1 azl+l
I 0 \ 1+1 ;-;::c;- r -,:;:c;- r <-- r
r> 1 (
u- r I '
-- (3.115)
r 1>+1 a rr 1
[ a21+1] 1
r'' - ,,z+l ,z+l , r> r 1
,..,, " • 11 ..... c. . •'- I .I •1. ..L" • I C. .'

' '
as required. Similarly, as r or r 1 --7 oo, the radial factor vanishes. It is symmetric
in rand Y 1 • Viewed as a function of r, for fixed r 1 , the radial factor is just a linear
combination of the solutions r 1and r-(1+ 1 \ of the radial part (3.7) of the Laplace
~~ R linP.Rr comhinMion for r < r 1 Rnrl for
r > r The reason for this, which will become apparent below, is connected with
1
•
the fact that the Green function is a solution of the Poisson equation with a delta
function inhomogeneity.
NowTITa"T we nave seen the general structure or tne expanswn ot a ureen
function in separable coordinates we turn to the systematic construction of such
expansions from first principles. A Green function for a Dirichlet potential prob-
!em satisfies the equation
~2~•
xV~A, A ") t
M
/IV~ A A ") '~
~.J.IIV)
' '
subject to the boundary conditions G( x, x 1 ) = 0 for either x or x 1 on the boundary

surfaceS. For spherical boundary surfaces we desire an expansion of the general
rorm p.ll'l ). J-u:coruingty we exptoir me iacr L!lar me ueua runcrion can oe
written*
1
8(x- X
1
) = 2 8(r- Y
1
) 8(1>- cp 1
) 8(cos 8- cos 8 1 )
and that the completeness relation (3.56) can be used to represent the angular
delta functions:
'I = I
or. 1\ 0/. 1\ '\;' '\;' v* fill .LI\V ill .L\ I~
\ .. 11 "7\
~~/
7 tm\"
" r2 "" ' " I ' tm\ ' '!-'I 'I
Then the Green function considered as a function of x can be exnanded as

= I
G(x, x 1
) = L 2:
l=O m=-l
Azm(rlr 1
, 8', c/> 1 )Yzm(IJ, cp) (3.118)
'" .1. • • • C/~11"7\

\ .. 'I
. .l
\
(~ " ' " •
. Jl .
/0 " L \ '
\• VI
. .l •1. '
A 1,(rlr 8 1
,
1
, cp = 1
) g1(r, Y
1
)Yi,(IJ 1
, c/> 1 ) (3.119)
*To express Ol'x- x') - 8(x 1 - xj)Oix2 - x;\81x3 - x;\ in terms of the coordinates ( g, g,, g3 ), related
tolx. x. x:\ via the Jacobian J(x. -:2\ we~e ~- · · is iilx · x'\ ,Pr He.nre
1
B(x- x') = IJ(x, g;)l B(gl -go o(g,- m8(6- g))
See Problem 1.2.
Sect. 3.9 Expansion of Green Functions in Spherical Coordinates 121
with
1 ,-J2 1fT -+ n d.
-;; (rRz(r, r')) Rt(r r') 0 o(r r') (3.120)
r ur r r
The radial Green function is seen to satisfy the homogeneous radial equation
(3. 7) for r =!= r'. Thus it can be written as
Ar 1 + Br-<1+ 1 ) for r < r'
gt(r, r') = { A 'rt + B 'r-(1+1) for r > r'
The coefficients A, B, A', B' are functions ofr' to be determined by the boundary
conditions, the requirement implied by o(r - r') in (3.120), and the symmetry of
aJ r r'i in r "nri r' th"t thP '" "· <1 "rP
vamsumg 01 g 1 ~r, r J 10r r a anu r u. ~onsequenuy g1v, r J uecomes

1
A(r - : : :1} r < r'
g1(r, r') = (3.121)
B'(,L - b~~} r > r'
'
The symmetry in rand r' requires that the coefficients A(r') and B'(r') be such
that R1(r r') can be written
1
' \
t. '' rl t u II
b "' , ' -1 < rl+ 1 1 1 ,t+ 1
< >
where r < (r >) is the smaller (larger) of r and r'. To determine the constant C we
must consider the effect of the delta function in (3.120). If we multiply both sides
of (3.120) by r and integrate over the interval from r = r' - E to r = r' + E,
where E is very small, we obtain
r d ' d ' 47T
- r,,,r, r')l - r,,,(r r')l n 1?.'ii
~ar r'-E r
-' '
1:' or r r, r < r . ttence
{ :, [rgt(r, r')] L+. = c(r·~ - ;~;::) r:, (? - ~~;+11) l~,·

= - ~ r1- (:.r+1J r~ + (l + 1)(:r+1J
\ I
\ I /
t \ I /
. ->,111 n;.
r'
Similarly
Substituting these derivatives into (3.123), we find:
Combination of (3.124), (3.122), (3.119), and (3.118) yields the expansion of the
Green function for a spherical shell bounded by r = a and r = b:
= I I 21+1 \ I \
G(x, x') = 4'lT 2: 2: I *1,~,' <:fJ ''TTim\'T• '+') I u

I '
_l_
) I rl+1
'>
1 \ 21+ 1 r< - rl+1 - h21+1 I
u
Ih
~
(2! + 1) 1-
(3.125)
For the special cases a~ 0, b ~ oo, and b ~ oo, we recover the expansions (3.70)
and (3.114), respectively. For the "interior" problem with a sphere of radius b,
we merely let a ~ 0. Whereas the expansion for a single sphere is most easily
obtained from the image solution, the general result (3.125) for a spherical shell
is rather difficult to obtain by the method of images, since it involves an infinite
set of ima es.
3.10 Solution of Potential Problems with the

Spherical "Green Function Expansion
The general solution to the Poisson equation with specified values of the potential
on the boundary surface is (see Section 1.10):
1
d>(x) - - - { p(x')G(x, x') d 3 x' (3.126)
4'lTE0 Jv 4'lT s an'
For purposes of illustration let us consider the potential inside a sphere of radms
b. First we will establish the equivalence of the surface integral in (3.126) to the
method of Section 3.5, equations (3.61) and (3.58). With a = 0 in (3.125), the
normal derivative, evaluated at r' = b, is:
(3.127)
Consequently the solution of the Laplace equation inside r b with 

V(O', 4> ') on the surface is, according to (3.126):
For the case considered, this is the same form of solution as (3.61) with (3.58).
There is a third form of solution for the sphere, the so-called Poisson integral
(2.19). The equivalence of this solution to the Green function expansion solution
"'""" Jo~U u•'
.
• ......... ~
.
•
. . .. ,'1
~-v
. .
./' ........
/
~ /"~
\
........
c~ ./
y
/ a
X
X "-.... _./
J Figure 3.11
~h ~raP
of radius b.
()
Ring of charge of radius a and total
inside a rL ,.1 ,.1, •
is implied by the fact that both were derived from the general expression (3.126)
and the image Green function. The explicit demonstration of the equivalence of
(2.19) and the series solution (3.61) will be left to the problems.
We now turn to the solution of problems with charge distributed in the vol-
. . . . . .
ume, so mat me vomme mtegrm m ~5.1LO) IS mvo1vea. n IS surnc1ent to cons1aer
problems in which the potential vanishes on the boundary surfaces. By linear
..
OU}''-''r v• a V< "''-' , '-'~uauvu, "'" ~ •a• ouuauvu '-'aH V'-'
obtained. The first illustration is that of a hollow grounded sphere of radius b
with a concentric ring of charge of radius a and total charge Q. The ring of charge
. . . . . .
IS wcaLeu m me x-y ptane, as snown m rig. -'.11. 1ne cnarge uensny or me nng
can be written with the help of delta functions in angle and radius as
p(x') = ~ 8(r' - a) 8(cos I!') (3.129)

27ra
In the volume integral over the Green function only terms in (3.125) with m = 0
will survive hpr~ll r:l.'i7) anel 1 herin P"
that a --> 0 in (3.125), we find

r
<P(x) = ___:___
47TE0
J p(x')G(x, x') d x' 3
(3.130)
= l{ L P,(O)r~( r'+I
47rE0 t~o
l - ...!...::..._
b2l+J )P,(cos fi)
>
'· ' >)' >0 lU'-

HVVV '< \' va< be-<) Vl ' aHU U. '-""'!', "''- W'-l lH<H
Pzn+ 1 (0) = 0 and P2n(O) = [(-1t(2n- 1)!!]/2"n!, (3.130) can be written as
<P(x)
J{ L ( -1n:n 1)!! 2n
r<
1
2n+1
r;:·
'4n+1
p z,( cos e) (3.131)
v
""'"~On 0
"" '" ,' >
In the limit b--> oo, it will be seen that (3.130) or (3.131) reduces to the expression
at the end of Section 3.3 for a ring of charge in free space. The present result can
be obtained alternatively by using that result and the images for a sphere.
A second example of charge densities, illustrated in Fig. 3.12, is that of a
hollow grounded sphere with a uniform line charge of total charge Q located on
the z axis between the north and south poles of the sphere. Again with the help
ot delta tunctwns, the volume-charge density can be wntten:
Q 1
P\"" J ,z LV\ lOU> 0 ') T U\ICU> 0 T 'JJ \.J.LJL)
~·· "-
"-V ~ '"
Linear
density
Q
2b
:r
The two delta functions in cos fJ correspond to the two halves of the line charge,
above and below the x-y plane. The factor 27rr' 2 in the denominator assures that
the charge density has a constant linear density Q/2b. With this density in (3.126)
we obtain
<l>(x) = Qb L oo
[P1(1) + P 1(-1)]P1(cos ()) J" r~ (1+1-
1 b~;: 1 I )
dr' (3.133)
87rE0 1~0 o r>
The integral must be broken up into the intervals 0 < r' < r and r < r' < b.
Then we find
fob = ( rl+l
1
-
,1 ) J' r'l dr'
b21+1 0 + rl
fbr ( r'/+1
1
-
r'z ) dr'
b21-1
(3.134)
For l - 0 this result is indeterminate .•A.pplying L'Hospital's rule, we have, for
r ~lim~
o z~o
[
1
-
!!._ ( )
(,;)']~lim [-!!._elln(r/b)J ln(~)
z~o dl
=
r
(3.135)
dl l
This can be verified by direct integration in (3.133) for l = 0. Using the fact that
P 1( -1) = ( -1) 1, the potential (3.133) can be put in the form:
¢(x) 4 7r:ob {lnG) + ~ zJ;k/ 1f) [1

1
GfJPzj(cos 6)} (3.136)
The presence of the logarithm fur l 0 reminds us that the potential diverges
along the z axis. This is borne out by the series in (3.136), which diverges for
cos () * 1, except at r b exactly. The peculiarity that the logarithm has ar-
gument (b/r) mstead of (blr sm II) IS addressed m Problem 3.8.
The surface-charge density on the grounded sphere is readily obtained from
(3.136) by differentiation:
cr(()) =
a<P
Eo-
Q [
- 47rbz 1 + ~
00
(4} + 1)
(2i + 1) Pzj(cos ())
J (3.137)
ar r~b
Sect. 3.11 Expansion of Green Functions in Cylindrical Coordinates 125
The leading term shows that the total charge induced on the sphere is -Q, the
other terms integrating to zero over the surface of the sphere.
3.11 Expansion of Green Functions in Cylindrical Coordinates

ThP. P.Yn<~n<ion
dHVlU> dllVlllCl
of thP
Ul \JlCCll
.
·•· 1ti"l of" nnit noint rh"r<JP. in rvl' .-L · 1l
~ -.
CXjJdl,IUII>.
..-1: •gtps
VVC ]J''-'"'-''" LIIC
initi<~l stens in ~ener"l enomrh f:'lshion to nermit the nre to he re:'lclilv

.-l. .-l <0. .... ~- "· •• • .1 .1..1 .,,_ .1: .-1.' .11. .-1
T •~ ~ ~·~~u ><v• yv y•v ""u~yuuuu~u•
ary surfaces. The starting point is the equation for the Green function:
47T
V~G(x, x') = - - o(p- p') 15(4>- 4>') o(z- z') (3.138)
p
where the delta function has been expressed in cylindrical coordinates. The 4>
and z delta functions can be written in terms of orthonormal functions:
o(z - z ') = - 1 foo dk eik(z z') = -1 foo dk cos[k(z - z ')]

27T -oo 7T 0
(3.139)
1 00 ""'
Then substitution into (3.138) leads to an equation for the radial Green function
g,(k, p, p'):
1 d ( dg,) - (k +-;;:
pdp p dp
m
gm
2
2
) 47T
= - - ; o(p- p') (3.141)
For p *
p' this is just equation (3.98) for the modified Bessel functions, I,(kp)
and K,(kp). Suppose that r/J1 (kp) is some linear combination of!, and K, which
satisfies the correct boundary conditions for p < p', and that tf!2 (kp) is a linearly
independent combination that satisfies the proper boundary conditions for
p > p'. Then the symmetry of the Green function in p and p' requires that
~j.l4L)
The normalization of the or t/J, Wo is determined bv the clisron · in slooe

·" .-11. ,,_ .-1.1 • • •
O V J "''-' U'-'HU "
ag,
dp + dp - p'
where I± means evaluated at p = p' ± E. From (3.142) it is evident that
[d:mp - d:,p - ]+
= k(t/J1 t/J~ - t/J2 t/Ji) = kW[t/Jb t/lz] (3.144)
n. ·'-'· ; in F.IP<' ·,.., n. -ST
where primes mean differentiation with respect to the argument, and W[ if!~o if!2]
is the Wronskmn of if!1 and if!2 • EquatiOn (3.141) 1s of the Sturm-LIOUVIlle type
'
.!!:_ n{y) dy + .,{y)v = 0 n.14'i)
dx dx
rl i t ' ,111. fh >t tJ, o1' ,f t , Fn ·~rlu 'n.-1, .-1, >t cnlntinno ,f
. . . ~ ... . .
SUCH an eqUaL!On IS propOrLIOnai LO l-liP~XJJ. rrence llle puS>Iuuny Ut Mll>tylllg
(3.143) for all values of p' is assured. Clearly we must demand that the normal-
. . -'-' '- "'· -'-
(3.146)
Tf no hot --'·
th~r~ :o~r~ sn·c. <>.Jk n. n') mnst he finite at n 0 :'lnd
vanish at p--? oo. Consequently if!1 (kp) = Alm(kp) and if! 2 (kp) = Km(kp). The
constant A is to be determined from the Wronskian condition (3.146). Since the
nr. ,!,' ,] f {1/v\ f, ,11 ,],.~ ,f it rl .. M nnt m~tt~r ,.1-
we evaluate it.ru~ing the limiti~g f~rms (3.102) and (3~103) for small x [or (3.104)
fnr braP r 1 urP flncl
1
W[lm(x), Km(x)] = -- (3.147)
so that A = 47T. The expansion of lllx - x' I therefore becomes:
This can also be written entirely in terms of real functions as:
1 =-
4 J~ dk cos[k(z - z')]
I
x - x' I 7T o (3.149)
X ~10 (kpJK0 (kp>) + _2'; cos[m( 4> - 4> ')]Im(kp<)KmkP>)

m=!
A number of useful mathematical results can be obtained from this expan-

sion. If we let x'--? 0, only them = 0 term survives, and we obtain the integral
1 2 roo
1 cos KZ r... 0 ~ Kp) aK p.DUJ
YPT£ "'"
Tf "'''" ·'- n2 in f11 <;()) hv R 2 n2 + n 12 ?nn 1 rn<( rl-. rl-. ') thPn ""'" h:<VP
vu "''- ,._n-uuuu o<u'- "''- " ' ''-'"'- ''-''- jA A 1 "'"' <. v, '·'-'·• JUOC \J·'~' 1
with z' = 0. Then comparison of the right-hand sides of (3.149) and (3.150)
f ..... ,1- ........ c. .11 -· .c .\ 1. ..... ... . .... -~ .
\ " »o •u ~) •u
Kn(kYo 2 + o' 2 - 2oo' cos(cb- d>')) =

t'\.1 'ill
10 (kp<)K0 (kp>) + 2 _2'; cos[m( 4> - 4> ')]Im(kpJKm(kp>)
m=l
Sect. 3.ll Eigenfunction Expansions for Green Fnnctions 127
• ,_ . 1. ,,_ --' _,_

T.
•u UH~ .U~ < • ~~ ·'·
UH " H ~ ~UH
.1
lU~~ Ul~
1:
<HHH
• L
" V U>"l UH
T
£.
<Vl
d.
"'~
Green function for (two-dimensional) polar coordinates:

1 --
In! 2 2
p + p' - 2pp' cos(¢- ¢')
I 1 ' ~
1
1 \ m (3.152)
2ln • I+ 2 2: - ~ m~1 m P>
Thio 1htirm C'On hP uPrifiPrl hu o >ti,-. C'r>ndr,,-.tir>n r>f thP tmr>.
. r . J .J
dimensional Green function for the Poisson equation along the lines leading to
(3.148). See Problem 2.17.
'-l 1?
~
Pi·o "J
.L'.
1rtinn F:·
"C
•dnnfi! fnr
J
r;., Runrtinnfi!
Another -1.
iaue for obtaininll ,;""" of Green fun, is the use of
eigenfunctions for some related problem. This approach is intimately connected
with the methods of Sections 3.9 and 3.11 .
.
'T\
"t'
:+.
.'J
.,J., ~
"J ••• '
• rl. • 1"
'!-'"
.1.
"" ·~
ent1al equatiOn of the form
V2 1/f(x) + rf(x) + A]l/l(x) = 0 (3.153)
If the solutions if!(x) are required to satisfy homogeneous boundary conditions
<J., -+, <' ,+ ,J., ,1, .+. Tl <l- 1'1 1 <:'J\ :11 . .1 1..
\" • ~I
'
well-behaved (e.g., finite and continuous) solutions, except for certain values of
A. These values of A, denoted by A,, are called eigenvalues (or characteristic val-
...
Ut:O) <lllU lilt; MJlUlllJII~ lf'n~"-) <llt; ·o· 'J Ult; ·o !<HUt; -
entia! equation is written:
\P,/dx) + f(x) + A 1,/dx) = 0 n 1'\4)
By methods similar to those used to prove the orthogonality of the Legendre or

.Ut;~~Cl lUll\CllVll~, 1l \C<lll Ut; >HVWH lll<ll llli:O _0 <llt; Vl 0
r i/l!(x)if!n(x) d x = 3
omn (3.155)
where the eigenfunctions are assumed normalized. The spectrum of eigenvalues

An may be a discrete set, or a continuum, or both. It will be assumed that the
totality of eigenfunctions forms a complete set.
Suppose now that we wish to find the Green function for the equation:
V~Ci(x, x') + lf(x) + AJCi(x, x') - -4m'i(x x') (J.D6)
where Ais not equal to one of the eigenvalues A of (3.154). Furthermore suooose
that the Green function is to have the same boundary conditions as the eigen-
functions of (3.154 ). Then the Green function can be expanded in a series of the
PinPnf,n,-.f ;,.. 10 r>f fhP f,
G(x, x') = L an(x')i/ln(x) (3.157)
*The reader familiar with wave mechanics will recognize (3.153) as equivalent to the Schri:idinger
PnnMinn for " n"rticle in · . 'I
128 Chapter 3 Boundary-Value Problems iu Electrostatics: II-SI
Substitution into the differential equation for the Green function leads to the
result:
)' n
m
{y'){)t - ,\ ),/, {y) - -J,.,s;{y - Y ')
n 1 "~'
H ,,,; ,,, ]., ,,,. 'rl, ]., ,/,*f,e\ rl' ,,., ,, 17 ,,., ,,.,
·r-•, ' 'rn\"/
( ) 'J fi d
gonality conditiOn 3.155 reduces the left-hand Side to one term, an we n :
0
' d
a (x') = 47T l/l~(x') (3.159)
n An- A
Consequently the eigenfunction expansion of the Green function is:
G(
X, X
') = 4
1T
2: l/l~(x')l/ln(x)
l l (3.160)
n n
For a continuous spectrum the sum is replaced by an integraL
Specializing the foregoing considerations to the Poisson equatiOn, we place
f(x) = 0 and A = 0 in (3.156). As a first, essentially trivial, illustration we let
~j.D'I) oe tne wave equanon over au space:
rvz + f.2) ,,,, ('I[) - () (i1fi11
with the continuum of eigenvalues, e, and the eigenfunctions:

!/1kx-
( )_ 1 k•X
f"1 ••>312e
I
(3.162)
\' '/
. . . .
1nese eiv ,,Ions nave ueua 1uncuun normanzauun:
Jr l/Jk'(x)l/Jk(x) d"x - 8(k k') (3.163)

rr'L
~HvH'
,,.
~~~~H~H<t; W
f'J
\•
1 Lf\\
.~VVJ•
•L
U<v
' ."'- . ... r' ~. L
H~U
••
H<v
'
1 1 { ,, eik·(x-x')
lx- x'l 27T2 Ju 1\,
e ~.J.~U'+)
1nis is jusr me mree-aimensiona1 l:'ourier imegra1 represemarion or 11 IX x I·

As a second example, consider the Green function for a Dirichlet problem
'- -~~ '- , A
a UVA UJ W\.- O>A < ' A v, Y v, ,(, v, A u,
y = b, z= c. The expansion is to be made in terms of eigenfunctions of the wave
V'i'
(Y' 2 + kTmn)!/lzmn(X, y, z) = 0 (3.165)

where the eigenfunctions which vanish on all the boundary surfaces are
I 12 , I,
!/Jhnn(X, y, Z) = \J b
a c
sin(
a
'
Sill ~
b
Sill
~)
and (3.166)
[2 mz nz
k2lmn -- '7T 2( -+-+-)
az bz cz
Sect. 3.13 Mixed Boundary Conditions; Conducting Plane with a Circular Hole .L..::'J
The expansion of the Green function is therefore:
' 32 (3.167)
G(x, X ) = 'TTnhr
m7ry'
l.
X '\' u u u u I \ ~ \ ~ I
LJ
l,m,n=1
z2
-,- \'. 'I 'Jl"" -r

3.9 and 3.11, namely, (3.125) for spherical coordinates and (3.148) for cylindrical
~: lMf" WP. writP. rlnwn thP. :m,]o<Tml< P.Yn,n<ion for thP. rP.rt"n<Tnl"r hoY Tf
the x and y coordinates are treated in the manner of (0, 4>) or ( </>, z) in those
cases. while the z coordinate is · ·" " out for <nPri>~l treatment. we obtain the
Green function:
~ I \ I ,\ I \ I ,\
G(x, x') = m7r ~ sin( t7rx I sin t7rx ) sin( rm'_!_) sin( m7ry
ab l,m~l a 1 \ a 1 b 1 b
(7. 1hR\
X sinh(K1mzJ sinh[K1m(c - z>)]
K 1m sinh(K1mc)
where K 1m = 7r(P/a 2 + m 2 /b 2 ) 112 • If (3.167) and (3.168) are to be equal, it must
J, t\, t\, /7. 1 t;'7\ ' t\, D. ,,, ' t\,
interval (0, c) of the one-di~en.~on~l Green function in z i~C3.168):

n7rz'
Sill--)
sinh(K.. z~) sinhrK jc- z )l 2 ,;:.. c
L, \ 2 Sill c I
c n~l 2 ( n7r
Kim+ -
~
The verification that (3.169) is the correct Fourier representation is left as an

exercise for the reader.
Further illustrations of this technique will be found in the problems at the
end of the chapter.
3.13 Mixed Boundary Conditions; Conducting Plane

Wlth a Circular Hole
The potential problems discussed so far in this chapter have been of the orthodox
1.' ". ···~· •• t.
~~
"· • J "•u~
'. u ~
.c ~ • J !"" (,
\' .11
•J T'\' •• t.l •\
"J
the whole boundary surtace. ln the umqueness proot tor solutiOns ot the Laplace
· p, ' ,.,j: {<1,, •I; 1 0\ ;1 . ,j rl nl L lh >I . "J, rl
.. "1· •. , • • r. ' ' .
ary conamons, wnere me potenuaJ IS specillea over pan or me oounaary ana ns
norm"] rl,
UHHfU'-'
. VV'
,
,,· i<
ty-vaau\.o C'
. '" rl ouPr thP
, ~·
· -lPr
l\.oHU lV
"ko ]p,.r] to mPll.rlPfinPrl
. .,.
lH'-' IJVOOlUHHY Va
mixP.rl hnnnrl,rv rlitinn< whP.n m>~kin<T thP. nnim .£ "~rl tn i<T,.;nrP.

uu~uy•~ ua 1u~•n .au~•·
0 uun~uuuuu~~,auuauo
boundary conditions are much more difficult to handle than the normal type.
To illustrate the difficulties encountered with mixed boundary conditions, we
an a
circular hole of radius a cut in it, and with the electric field far from the hole
either side of the plane. The geometry is sketched in Fig. 3.13. The plane is at
z = 0; the hole is centered on the · · of coordinates; the
components are z> z
z < 0. The problem may seem contrived, but with £ 0 = 0 or E 1 0 it has
"small" is defined as small compared to a wavelength so that electrostatic con-
siderations can e Section 9.5
Since the electric field is specified far from the hole, we write the potential
as
= + (3.170)
E <J>(l)
If the hole were not there, <1>< 1 l would be zero. The top surface of the sheet would
have a uniform surface ch and the bottom surface a
density E 0 E 1 • The potential can thus be thought of as resultmg from a rear-
rangement of surface charge in the neighborhood of the hole. Since this charge
z
Sect. 3.13 Mixed Boundary Conditions; Conducting Plane with a Circular Hole 131
:-.mce rne rorat z component or electric ne10 must De continuous across z u in

the hole we must have (foro< a)
0
-E0 + £(1>1 z z=O+ -- -E 1 + £ z >1 z=O
Because £~1 ) is odd in z, this relation determines the normal component of the
electric field to be
£(1!1
z z:....:O +-- -£(1!1
1
z z=O- -- 2 (£0 -E) 1
provided (x, y) lie inside the opening (0 U1Ull;IO ~U f' .._ ~), lllt: " " llt:1U 1> llUl ""llUWll, UUl lllt: f1Ult:lllld1 1> Lt:!U uy
hypothesis. From (3.170) this means that cp(l) = 0 there. Note that in the opening
we do not know the potential. We therefore have an electrostatic boundary-value
;tl, t } J"AJJ, • • -1;1; ,.
y
,1-,1,
·o . rl h r/,
"J
i)cp(l)
- :Z(lco n1) tor u 0 l(p, z) = r dk A(k)e-klziJ0 (kp) (3.172)

n
' LJ '. . _, '- ., '
l'-' J-'~V'-''-''-''-'"'5 LV 01010 HVVV r1~1\.) >0 UIOLIOL , UJ " ' " LJVL
we relate A(k) and its derivatives at k = 0 to the asymptotic behavior of the

potential. For large p or Iz I the rapid oscillations of l 0 (kp) or the rapid decrease
Af -klzl ;mnlu th~t th~ ;n+~nr~ 1 :n 1?. 177\ r~. · , ;to ;m, ~nntr;h,t;nno frAm
·o- ,-. -,-
the region around k = 0. The asymptotic behavior of cp(l) is therefore related to
tne Denav10r or A~k) at small k. we assume tnat A~k) can De expanaea m a
T; ,1, s~ri~s llronnrl I< ()·
;, e d 1A
fl~K)
£:a l! dkf\Y J
With this series inserted into (3.172), the potential cp(l) becomes
(1) -
ci> (p, z) - l~o de (0) Bz(p, z)
i: dzA
(3.173)
where
Bz(p, z) = ~ t dk k 1e-kiziJ0 (kp) (3.174)
Th~ int~<Jrlll n.174) r:m Pvirl~ntlv h~ writt~n
1 I
d \I (oo
Dz "'-'Jo~~<f'J
l! dlzl} Jo '"""
Using a result from Problem 3.16c, we find that B 1 is
I
d 1
B, ;, [ r1 I :1 )(
I •I
2
-c
2) (3.175)
1--'
132 Chap tel 3 Boundaty-Value Pwblems in Electrostatics: II SI
The reader should not be surprised to find that explicit calculation yields
Pz(icos 1:11)
B
r
where cos 8 = zlr and r - V p 2 + z 2 . The asymptotic expansion (3.173) is thus
an expansion of the spherical harmonic form (3.33):
(3.177)
As is discussed in the next chapter, this expansion in powers of r- 1 is called a

multipole expansion. The l = 0 coefficient, A(O), is the total charge (divided by
4we5 ) The l 1 coefficient, dA(O)!dk, is the dipole moment in the z direction,
and so on. Once the function A(k) is known these quantities that describe the
asymptotic behavior of the potential can be evaluated without explicit construc-
tion of the potential itself.
We are now ready to discuss the mixed boundary value problem. With the
assumed form (3.172) for cf>(ll, the boundary conditions (3.171) become a pair of
mtegral equations of the first kmd for A(k):
f oo
()
dk kA(k)l0 (kp) = 1(£0 - E 1) for 0 < p <a
0 fma<p<oo
Such pairs of integral equations, with one of the pair holding over one part of
the range of the mdependent vanable and the other over the other part of the
range, are known as dual integral equations. The general theory of such integral
equations is complicated and not highly developed.* Just over a hundred years
ago H. Weber solved the closely related problem of the potential of a charged
circular disc by means of certain discontinuous integrals involving Bessel func-
tions. We appeal to a generalization of Weber's formulas. Consider the dual
integral equations,
r dy yg(y)Jn(yx) = x" for 0 < x < 1

(3.179)
J dy g(y)Jn(yx) - 0 for 1 < x < oo
()
Exammation of the formula of Sonine and Schafheitlin for the integral of

1,,/at)lv(ht)r" (see Watson, pp. 398 ff, or :Jagmts et al., p. 99) shows that the
solution for g( y) is
f(n + 1) f(n + 1) l,,, HY)

g(y) (3.180)
f(n + ~) (2y) 112
In this relation j,(y) is the spherical Bessel function of order n (see Section 9.6).
*One monograph, I. N. Sneddon, Mixed Boundary Value Problems in Potential Theory, North-
Holland, Amsterdam, and Wiley-Interscience, New York (1966), is devoted to our subject. See also
Sect. 3.13 Mixed Boundary Conditions, Conducting Plane with a Cir tolar Hole 133
For our pair of equations (3.178) we haven = 0, x = pia, y = ka. Therefore

A(k) is
A(k) a Clkls ka]~ (3.181)
The expansion of A(k) for small k takes the form,

2 3
(Eo - E 1 )a [ka (ka) + .. ·]
A(k) 37T 10
This means that total charge associated with <.[1< 1 l is zero and the leading term in
the asymptotic potential (3.177) is the l = 1 contribution,
(3.182)
falling off with distance as r~ 2 and having an effective electric dipole moment,
(z ~ 0) (3.183)
The reversal of the effective dipole moment depending on whether the obser-
vation point is above or below the plane is a consequence of the fact that a true
dipole potential is odd in z, whereas (3.182) is even. The idea that a small hole
in a plane conducting sheet is equivalent far from the opening to a dipole normal
to the surface is important in discussing the consequences of such openings in
the walls of waveguides and cavities. Figure 9.4 depicts the origin of the dipole-
like field as a consequence of the penetration of the field lines through the hole
to terminate on the side with the smaller constant field. The picture is given
quantitative meaning through (3.182) and (3.183).
The added potential <.[JUl in the neighborhood of the opening must be cal-
culated from the exact expression,
(3.184)
7T JO
The integtal,* after an integration by parts to replacej 1 withj0 , can be expressed

as a sum of the imaginary parts of the Laplace transforms (for complex p) of
.1, (kp)lk for v - 0, 1 The result, after some simplifications, is
(Eo - E,)a [ fR - A lzl tan (3.185)

a
Some special cases are of interest. The added potential on the axis (p 0) is
*For integrals of the kind encountered here, see Watson (Chapter 13), Gradshteyn and Ryzhik,
},1agnus, Ohel'hettinger, and Seni, or the Hatenzan jHanuscript Pre-ject.
For lzl >>a thus reduces to (3.182) with r = lzl, while for lzl---> 0 it is approx-
imated by the first term. In the plane of the opening (z = 0) the potential cp(l) is
(Eo EI) \la2 p2
for 0 a). The tangential electric field in the
opening is a radial field,
(3.186)
The normal component of electric field in the opening is, from the first equation
in (3.171 ), just the average of the uniform fields above and below the plane, that
IS,
(3.187)
We note that the magmtude of the electnc field has a square root singularity at
the edge of the opening, in agreement with the considerations of Section 2.11.
The surface-charge densities on the upper and lower sides of the conducting plane
in the neighborhood of the hole can be evaluated in a straightforward manner.
The explicit calculation is left to the problems.
Equipotential contours near the circular bole for the full potential (3 170)
are shmvn in Fig. 3.14 for the situation where £ 1 0. At distances mo1e than
2.0
•. u
. ;(·~
0.4
~
0.2
I : ~OJ 0.1
0
~- -/
I
I
I
I
!
Figure 3.14 Equipotential contours near a circular bole in a conducting plane with a
normal electric field E0 far from the bole on one side and no field asymptotically on the
other ( E 1 = 0). The numbers are the values of the potential <P in units of a E0 • The
distribution is rotationally symmetric about the vertical dashed line through the center
Ch. 3 Problems 135
two 01 three times the radius away from the hole, its presence is hardly
disce rni bIe.
The classic problem of a charged conducting disc is discussed in detail by
Sneddon (op. cit.). The mixed boundary conditions for the disc or hole can be
avoided by separating the Laplace equation in elliptic coordinates. The disc (or
hole) is then taken to be the limiting form of an oblate spheroidal surface. For
this approach, see, for example, Smythe (pp. 124, 171) or Jeans (p. 244).

The subjects of the special functions of mathematical physics, the solution of ordinary
differential equations, hypergeometric functions, and Sturm Liouville theory are covered
in many books. For the reader who does not already have his favorite, some of the pos-
sibilities are
Arfken
Dennery and Kryzwicki
Morse and Feshbach
Whittaker and Watson
A more elementary treatment, with well-chosen examples and problems, can be found in
Hildebrand, Chapters 4, 5, and 8
A somewhat old-fash10ned source of the theory and practice of Legendre polynomials
and spherical harmonics, with many examples and problems, is
Byerly
For purely mathematical properties of spherical functions one of the most useful one-
Magnus, Oberhettinger, and Soni

For more detailed mathematical properties, see
Watson, for Bessel functions
Bateman Manuscript Project books, for all types of special functions
Electrostatic problems in cylindrical, spherical, and other coordinates are discussed
extensively in
Durand, Chapter XI
Stratton, Chapter III
Problems
3.1 Two concentric spheres have radii a, b (b >a) and each is divided into two hemi-
spheres by the same horizontal plane. The upper hemisphere of the inner sphere
and the lower hemisphere of the outer sphere are maintained at potential V. The
other heulispheres are at zero potential
Determine the potential in the region a -s: r -s: b as a series in Legendre poly-
nomial~ Include term~ at least np to l - 4 Check yonr ~olution against known
results in the lirniting cases b ===;o; oo, and a ===;o; 0.
3.2 A spherical surface of radius R has charge uniformly distributed over its surface
with a density Q/47rR 2 , except for a spherical cap at the north pole, defined by the
cone () = a.
(a) Show that the potential inside the spherical surface can be expressed as
Q ~ 1
2
8 1TEo l=O 21 + 1
[P'*' (cos a)
R
where, for l = 0, P1-1(cos a) = -1. What is the potential outside?
(b) Find the magnitude and the direction of the electric field at the origin.
(c) Discuss the limiting forms of the potential (part a) and electric field (part b)
as the spherical cap becomes (1) very small. and (2) so large that the area with
charge on it becomes a very small cap at the south pole.
3.3 A thin, flat, condncting, circular disc of radius R is located in the x-y plane with its
center at the origin. and is maintained at a fixed potential V. With the information
that the charge density on a disc at fixed potential is proportional to (R 2 -l)- 112 ,
where pis the distance out from the center of the disc,
(a) show that for r > R the potential is
<P(r, 11, ¢)
2V R
=--
1T r l=o
2~ ( -1 )1
21 1
+
R
r
P 21 (cos 11)
(b) find the potential for r < R.

(c) What is the capacitance of the disc?
J.4 The surface of a hollow conducting sphere of inner radius a is divided into an e•en
number of equal segments by a set of planes; their common line of intersection is
the z axis and they are distributed uniformly in the angle </J. (The segments are like
the skin on wedges of an apple, or the earth's surface between successive meridians
of longitude.) The segments are kept at fixed potentials ±V, alternately.
(a) Set np ~ series representation for the potential inside the sphere for the gen
era! case of 2n segments, and carry the calculation of the coefficients in the
series far enough to determine exactly which coefficients are different from
zero. For the nonvanishing terms. exhibit the coefficients as an integral over
(b) For the special case of n 1 (two betnispheres) determine explicitly the po-
tential up to and including all terms with l = 3. By a coordinate transformation
verify that this reduces to result (3.36) of Section 3.3.
3.5 A hollow sphere of inner radius a has the potential specified on its surface to be
<f> V(e, if>). Prove the eqmvalence of the two forms of solution tor the potentml
inside the sphere:
(a) <P( ) = a(a2 - r2) I V(11', </J') dfl'

x47f .r (r 2
I tt" 2til cos r) 312
~here cosy cos 8 cos 8' +sin 8 sin 8' cos(¢ ¢').
(b) <P(x)
where Azm = J dfl' Y'im(O', </J')V(I1', ¢').

3.6 Two point charges q and -q are located on the z axis at z - +a and z - -a
respectively.
(a) Find the electrostatic potential as an expansion in spherical harmonics and
powers of r for both r > a and r < a.
£'L '> . 1-::t.,
(b) Keeping the product qa = p/2 constant, take the limit of a ---> 0 and find the
,:, . n . . -~ • • ~· ' ·' '' • -' •
1VJ T V. HHO <0 UJ U~HHOWH< U U<VH/5 W~ .C UMO auu JCO
< I .
potentm.
(c) Suppose now that the dipole of part b is surrounded by a grounded spherical
ohPll ,f rorl; ~ · ..,;th thP nr;n;n ]Cl, EnPor · ;nn finrl thP
--r ·r
potential everywhere inside the shell.

3.7 Three point charges (q, -2q, q) are located in a straight line with separation a and
with the middle chafQe (- 2a) at the origin of a grounded conducting soherical shell
nf rorl;no h oo · "' • ". ;n thP
/ q~~=0
~
ya q-----; 2n
~ /
Problem 3.7
(a) Write down the potential of the three charges in the absence of the grounded
sphere. Find the limiting form of the potential as a ---> 0, but the product
qa2 = Q remains finite. Write this latter answer in spherical coordinates.
(b) The presence of the grounded sphere of radius b alters the potential for r < b.
The added potent1al can be v1ewed as caused by the surtace-charge density
induced on the inner surface at r - b or by image charges located at r > b.
Use linear superposition to satisfy the boundary conditions and find the po-
tential everywhere inside the sphere for r < a and r > a. Show that in the
" ;, (\
'
(r, fJ, <P) ---> Q ( 1 - 5r') P 2 (cos fJ)

2 7rE0 r 3 b
"lll Th . 0 n ,J;n. ,f thP (:: n,;\ fnr thP . ol • • ' 0 nrnHnriPrl
sphere with a uniformly charged wire alo~g a diameter. Very close to the wire (i.e.,
for p = r sin fJ << b), the potential should be that of a uniformly charged wire,
namely, 'P- ((d/471"Eob) ln(blp) + 'P0 • The solutwn (j.lJ6) does not expliCitly have
this behavior.
(a) Show by use of the Legendre differential equation (3.10) and some integration
, LU>CL U) Wt> lllC: djJjJllJjJlldliO ' . 111 >jJll\011\.,dl
uy 1-'"' "•
to perm1t the solutwn (j.lJ6) to be wntten m the alternative torm,
I \ 2j
2b \ 4j + 1
Q
~
I
dJ(y) 1
ln 1 r I P . .( rn< R\
471"E0 b r sine -1 2j(2j + 1) b
in which the expected behavior near the wire is manifest. Give an interpre-
tation of the constant term <P0 = -Q!47TE0h. Note that in this form, for any
rib < 1 the Legendre polynomial series is rapidly convergent at all angles.
(b) Sho" by use of the expansion (3.38) that
and that therefore the charge density on the inner surface of the sphere, Eq
(3.137), can be exp~essed altematively as
0 ( li)
The (integrable) singular behavior at f) = 0 and f) = 7T is now exhibited ex-

plicitly. The series provides corrections in ln(l/fJ) as fJ--> 0.
3.9 A hollow right circular cylinder of radius.b has its axis coincident with the z axis
and its ends at z = 0 and z = L. The potential on the end faces is zero, while the
potential on the cylindrical surface is given as V( ¢, z). Using the appropriate sep-
aration of variables in cylindrical coordinates, find a series solution for the potential
anywhere inside the cylinder.
3.10 For the cylinder in Problem 3.9 the cylindrical surface is made of two equal half-
cylinders, one at potential V and the other at potential - V, so that
V(¢,z)={-~
for - 7T!2 < ¢ < 7T!2
for 7T/2 < ¢ < 37T/2
(a) Find the potential inside the cylinder.

(b) Assuming L >> h, consider the potential at z = L/2 as a function of p and¢
and compare it with two-dimensional Problem 2.13. ·
3.11 A modified Bessel-Fourier series on the interval 0 :5 p :5 a for an arbitrary function
f(p) can be based on the "homogeneous" boundary conditions:
dJ (k' )
At p = 0, p I v (k p) "dp P = 0
d A
At p =a, d ln[I,(kp)] = - ~ (A real)
The tint condition restricts l' The second condition yields eigemralues k - y,,Ja,
where Y~m is the nth positive root of x dlv(x)!dx I A lv(x) 0.
(a) ~how that the Bessel functions of diff€mnt eigenvalues an: orthogonal in the
be expanded on the interval in the modified Bessel Fourier series
f(p)
with the coefficients An given by

The dependence on A is implicit in this form, but the square bracket has al-
ternative forms:
For A --> oo we recover the result of (3.96) and (3.97). The choice A = 0 is
another simple alternative.
3.12 An infinite, thin, plane sheet of conducting material has a circular hole of radius a
. . . . . " . . . .
ptane, 11umg we uote, uut separateu .rom we suee, uy a very narrow msutatmg
ring. The disc is maintained at a fixed potential V, while the infinite sheet is kept
at zero potential.
(a) Using appropriate cylindrical coordinates, find an integral expression involv-
. . . .
mg oesset tunctwns wr tue potenllat "' ctny iJOITI< ctuuvc we 1-''"nc.
(b) Show that the potential a perpendicular distance z above the center of the disc
IS
z
2
v
(c) Show that the potential a perpendicular distance z above the edge of the disc
IS
1- K(k)
'/jU
where k = 2a/(z 2 + 4a 2 ) 112, and K(k) is the complete elliptic integral of the
first kind.
3.13 Solve for the potential in Problem 3.1 using the appropriate Green function ob-
in thf'. tf'Yt "nd :c. thM thf'. nht"inf'.d in thi< w"v with thf'.
direct solution from the differential equation.
3.14 A line charge of length 2d with a total charge Q has a linear charge density varying
as (d2 - z2 ), where z is the distance from the midpoint. A grounded, conducting,
-'- · ,] <hf'll nf innPr r"dim h '> rl i< "M thf'. mirlnnint nf thf'. lillf'. ,].,
\31 rinu we pOLemiai everywuere insiue we spuericat sueu as an expansion in

Legendre polynomials.
(b) Calculate the surface-charge density induced on the shell.
(c) Discuss your answers to parts a and bin the limit that d <<b.
·o . -r -'- • . •_, :c
. ~ op••~•~ v> wuwo u uuu my u >O w" . w~uww
of conductivity (]"'. Inside the sphere there is a uniform (chemical) force in the z
direction acting on the charge carriers; its strength as an effective electric field
entering Ohm's law is F. In the steady state, electric fields exist inside and outside
thP. <nhf'.rf' "nrl ._, nn it<
\it} I:']llU lllC ClCCllHC llClU ~lll Jll l U I ' } dllU '-UllCIIL CVCIJWIIClC Ill
space. Determine the surface-charge density and show that the electric dipole
moment of the sphere is p = 47rE0 ifa 3 FI(if + 2(]"').
146 Chaplet 3 Buundaty-Value Pwblems in Elechostatics. II Sl
(b) Show that the total current flowing out through the upper hemisphere of the
sphere is
2mJ'
I= · 1ra 2 F
(T + 2(r'
Calculate the total power dissipation outside the sphere. Using the lumped
circuit relations, P = PRe = IVe, find the effective external resistance R" and
c·
(c) Find the power dissipated within the sphere and deduce the effective internal
resistance R; and voltage V,.
(d) Define the total voltage through the relation, V, = (Rc + R,)I and show that
V, 4aP/3, as well as Vc I F; V,. Show that IV, is the power supplied by
the "chemical" force.
Reference: W. M. Saslow, Am. .!. Phys. 62, 495-501 (1994).
3.16 (a) Starting from the Bessel differential equation and appropriate limiting pro-
cedures, verify the generalization of (3.1 08),
1
k IJ(k
. k') r {}l,(kp)J,(k, p) rip
or equivalently that
-1 8(p - p') =
(" kl,lkp).T ,(kp') dk
p JQ
where Re( v) > -1.

(b) Obtain the following expansion:
1
lx- x'l
(c) By appropriate limiting procedures prove the following expansions:
1
.To(kY p> + p' 2 - 2pp' cos¢) = ~ e'"'cb.Jm(kp).T,(kp')

tn=-cc
ln=-cc
(d) From the last result obtain an integral representation of the Bessel function:
.T,(x)
Compare the standard integral representations.

3.17 The Dirichlet Green function for the unbounded space between the planes at
z = 0 and z = L allows discussion of a point charge or a distribution of charge
between parallel conductmg planes held at zero potential.
(a) Using cylindiical cooidinates show that one fu1111 of the G1een function is
4 ~
= e sm
L n=l !n=-cc
Ch. 3 Problems 141
(b) Show that an altemative form of the Green function is
G(x, x')
m c;o JU , I
3.18 The configuration of Problem 3.12 is modified by placing a conducting plane held
at zero potential parallel to and a distance L away from the plane with the disc
insert in it. For definiteness put the grounded plane at z = 0 and the other plane
with the center of the disc on the z axis at z = L.
(a) Show that the potential between the planes can be written in cylindrical co-
ordinates (z, p, 1>) as
<P(z, p) V r dA 1 1 (A)J0 (Ap/a) 11tfhft

(b) Show that in the limit a -> oo with z, p, L fixed the solution of part a reduces
to the expected result. Viewing your result as the lowest order answer in an
expansion in powers of a- 1 , consider the question of corrections to the lowest
order expression if a is large compared to p and L, but not infinite. Are there
difficulties? Can you obtain an explicit estimate of the corrections?
(c) Consider the limit of L-> co with (L - z). a and p fixed and show that the
results of Pwblem3.12 are recovered. 'vVhat about corrections fm L >>a, but
not -> oo.
3 19 Consider a point charge q between two infinite parallel conducting planes held at
zero potential. Let the planes be located at z = 0 and z = L in a cylindrical coor-
dinate system, with the charge on the z axis at z = z0 , 0 < z0 < L. Use Green's
reciprocation theorem of Problem 1.12 with problem 3.18 as the comparison
problem.
(a) Show that the amount of induced charge on the plate at z = L inside a circle
of radius a whose center is on the z axis is given by
where <t>(z 0 , 0) is the potential of Problem 3.18 evaluated at z = z0 , p = 0.

Find the total charge induced on the upper plate. Compare with the solution
(in method and answer) of Problem 1.13.
(b) Show that the induced charge density on the upper plate can be written as
u(p) = - .!L
2w
r
4l
dk sinh(kzo) kfo(kp)
smh(kL)
This integral can be expressed (see. e.g .. Gradshteyn and Rvzhik. p. 728, for-
mula 6 666) as an infinite series invohring the modified Bessel functions
(c) Show that the charge density at p = 0 can be written as the series
3.20 (a) From the results of Problem 3.17 or from first principles show that the paten-
tial at a point charge q between two infinite parallel conducting planes held
at zero potential can be wrjtten as
<P(z, p) SUI Sill o

71"EoL n-1 L L L
pmm z
•• r_,
"'
n,
" "' T -' .. -• '6'
•• '. ··-
IJ..\
Zo·
f""'oJo,.Jo+a +ha ' ,,4, ... ' .... , .. / o\
'V\1-'J
< ... 'L\~
' -\ +l. L
-a:mr U jJ jJCl -~Oi\fJJrr'
q ~I · . I nr.zo \ I nr.p \
u ;;:::1 \ .,
I -\ >\n T7
'L\PJ
'\ L ) '"\ L I
Discuss the connection of this expression with that of Problem 3.19b and 3.19c.
(c) From the answer in part b, calculate the total charge QL on the plate at z = L.
By summing the Fourier series or hy other means of comparison, check your
answer against the known expression of Problem 1.13 [C. Y. Fong and C.
Kittel, Am. I. Phys. 35, 1091 (1967).]
3.21 (a) By using the Green function of Problem 3.17b in the limit L -> co, show that
the capacitance of a fiat, thin, circular, conducting disc of radius R located
parallel to, and a distance d above, a grounded conducting plane is given by
r~ Ju
r nLIZ.n\~ln\ dn
2
<1~~-
I e -Lka)
dkf1
L Jo {
I pa(p) ap
•v
, ' 'J
'\!' . ..
'> cue. '-lieU!'>"
, , .,
- O Vll cue; UlO'-•
,.
(b) Use the expression in part a as a variational or stationary principle for C 1

with the approximation that a(p) = constant. Show explicitly that you obtain
the correct limiting value for c- 1 as d << R. Determine an approximate value
of c-' for an isolated disc (d >> R) and evaluate the ratio of it to the exact
result, 4r.E0 /C = ( r./2)R- 1 .
(c) As a better trial form for a(p) consider a linear combination of a constant and
(R 2 - ~)- 112 , the latter heing the correct form for an isolated disc.
For part b the following integrals may be of use:
r d{J,;t)r -- -
4
3r. '
r 0
dt ft(t) = !
t 2
3.22 The geometry of a two-dimensional potential problem is defined in polar coordi-
natesov ffie surfaces(/)= 0, </> = 7J. ando = a as indicated in the sketch.
/"'\
7 'A
g
/ / \
\
'! Problem 3.22
Using separation of variables in polar coordinates, show that the Green function
can be written as
~ 4 ( -1- p~rri~) sin ( -

mr.</J)
- sin ( -
mr.</J')
G(p, </J; p', </> ') --
~~ 1 -
m P<
m7r/{3
p';,mi/3
- -~-
a2mr.!f3 {3
-
f3
~ ,, e
'Y '
~ ~~
'""·
Ch. 3 Pwblems 143
3.23 A point charge q is located at the point (p', cp', z') inside a grounded cylindrical
box defined by the surfaces z = 0, z = L, p = a. Show that the potential inside the
box can be expressed in the following alternative forms:
~ ~ eim(<J>-<J>')Jm(X:nP)Jm(xm;P')
<P(x. x') q 2: 2:
<J>(x, x') = _q_

1TE L
i: i:
m=-~ n=l
e'm(<P-<1>') sin Sill
n7Tz'
L
e'm(<P-<P'l sin( k7rz I sin k-rrz' k, XmnP )Jml XmnP:)

Lq L L u u I
<P x x' --
Discuss the relation of the last expansion (with its extra smnmation) to the other
two.
3.24 The walls of the conducting cylindrical box of Probl<lm 3.2.3 ar€l all at z€lro pot€lntial,
except for a disc in the upper end, defined by p = b < a, at potential V.
(a) Using the various forms of the Green function obtained in Problem 3.23, find
three expansions for the potential inside the cylinder.
(b) For each series, calculate numerically the ratio of the potential at p = 0,
z = L/2 to the potential of the disc, assuming b = L/4 = a/2. Try to obtain at
least two-significant-figure accuracy. Is one series less rapidly convergent than
the others? Why?
(Abramowitz and Stegun have tables; Mathematica has Bessel functions, as
does the software of Press et al.)
3.25 Consider the surface-charge densities for the problem of Section 3.13 of the con-
ducting plane with a circular hole of radius a.
(a) Show that the surface-charge densities on the top and bottom of the plane for
lT -£ E + ~<T
<T-(P)
(Eo
~lT(p) Eo Sl11
How does 6.<T(p) behave tor large p! Is 6.<T(p), defined 111 terms of <IJE'l, zero
for < a? Ex lain.
. .,,_ . .
""!AA
~~~ ~·-
~
~
n
- n .L• .H N
(b) Show by direct integration that

rR rR
hm 27r I dp p(a+ + a_) + 27rEo I dp p(l::o b,) - u
R~~ a 'U
Interpret.
3.26 Consider the Green function appropriate for Neumann boundary conditions for the
<. ~
.. l TTL <L ' • ..£, , LC.. L
, ~J
-
a <- o. 1 o be able to use_\ 1.4D) ror tne pmemia1, impose tne simple constraint ( 1.45 J.
Use an expansion in spherical harmonics of the form.
G(x, x') = i: glr, r')Plcos y)

I 0
where g1(r, r') = r~lr~:;' + fz(r, r').

(a) Show that for I > 0, the radial Green function has the symmetric form
rl
Rz(r, r J ;:;-+
>
1
1 [I + 1 rr' l + I (ab fi-1 + a2!+1 ( r + r,z ) ]
(b21+1 _ a21+l) -~- ( ) I + 1 (rr')l+ 1 r' 1+1 rl+'
\U J 011UW llldl lUi l u

2
g0 (r, r') = ..!_ - ( a2 a+ b2 ) -;;
1 + .f(r)
r>
where Hr) is arbitrarv. Show exnlicitlv in 11.46) that answers for the notential
.nr. . ". ·' •., ~- .\
[Th-~ arbitra;i~ess in the 'N:~mann Green function can be removed by sym-

metrizing g0 in rand r' with a suitable choice of f(r).]
3.27 Aoolv the Neumann Green function of Problem 3.26 to the s1tuatwn m wh1ch the
normal electric field is E, = - E 0 cos f) at the outer surface (r = b) and is E, = 0
on the inner surface (r = a).
(a) Show that the electrostatic potential inside the volume Vis
r r"< R ,-,3
(x) Eo 1 - p' \1 + 2r3 I
where p = alb. Find the components of the electric field,

I \
cus u u \ f
"'"", '"
{.
"'
·u 1 _ p3
1
r3 ,
D f •
, D\ TO' Slll U
1 - p3
1 .L
2r3)
u \
(b) Calculate the Cartesian or cylindrical components of the field, E 2 and EP, and
,], ' ''" ,], ,],
l'
·"" " . ' . ,_ '· . :,
01 p li.J.
HAPTER4
spherical harmonics, but contact is established with the rectangular components

for the first few multipoles. The energy of a multipole in an external field is then
. . . . .
statics is sketched but a careful treatment is deferred to Cha ter 6. Dielectrics
and the appropriate boundary conditions are then described, and some typical
Finally the question of electrostatic energy and forces in the presence of dielec-
tries is discussed.
xpanswn
A localized distribution of char
is nonvanishing only inside a sphere of radius R around some origin.* The po-
armomcs:
(4.1)
opole term, l = 1 are the dipole terms, etc. The reason for these names becomes
obtained from the integral (1.17) for the potential:

1 p(x') 3 I
lx- x'l
with expansion (3.70) for 1/lx - x'l. Since we are interested at the moment in
(4.2)
*The sphere of radius R is an arbitrary conceptual device employed merely to divide space into
' ' . . . .
the expansion in multi poles is valid at large enough distances.
145
146 Chapter 4 Multipoles, Eleetrostaties of Maeroseopie Media, Dieleetries SI
Consequently the coefficients in (4.1) are:
(4.3)
These coefficients are called multipole moments. To see the physical interpreta-
tion of them we exhibit the first few explicitly in terms of Cartesian coordinates:
q00 ,~{p(x')d
v ~
3
x' ,~q
v
(4.4)
qll = - jl; I (x' - iy')p(x') d 3 x' - jl; (Px - ipy)

{Ll '(\
Ii Ii
qlo-~471"• Zp(x')dox'-~47T"Pz .
q22 = ~ jg I (x' - iy'?p(x') d'x' =

1~ jg (Q11 - 2iQ12 - Q22)
q21 = - ~~~ J z'(x' - iy')p(x') d x'

3
= -rJ~: (Ql3- iQz,) (4.6)
qzo =~,~I (3z' 2

-
2
r' )p(x') d 3 x' = ~, ~ Q 33
' '
Only the moments with m > 0 have been given, since (3.54) shows that for a real
charge density the moments with m < 0 are related through
ql,-m q lm
In equations (4.4)-(4.6), q is the total charge, or monopole moment, p is the

electric dipole moment:
p
J
f x'p(x') d x' 3
(4 8)
and Q;1 is the traceless quadrupole moment tensor:
(3x;xj t' 2 8;)p(x') d 3x'

Q;;
.{ (4.9)
We see that the lth multipole coefficients [(2/ + 1) in number] are linear com-
binations of the corresponding multipoles expressed in rectangular coordinates.
The expansion of (x) in rectangular coordinates
47rE0 r r 2 ;,1
by direct Taylor series expansion of lllx- x' I will be left as an exercise for the
r l (4.10)
reader It becomc;s increasingly cumbersome to continue the expansion in (4 10)

beyond the quadrupole terms.
The electric field components for a given multipole can be expressed most
Sect. 4.1 Multipole Expansion 147
easily in terms of spherical coordinates. The negative gradient of a term in (4.1)

••• "·" • 1 ].. .t. . ·'
• ~I-'
(l + 1) Ytm( 8, c/1)
E (IT,
~Ll -t- l)Eo r
1 . a,_
1 a
E ¥,.J(I cb) (4.11)
~Ll + l)Eo r ~ dtJ
1 1 zm Y. ( n rh)
F n.
(2l + l)E0 r Sill {I
aY1m/ae and Y1m/sin (I can be expressed as linear combinations of other Y 1m's, but
the expressions are not particularly illuminating and so will be omitted. The
proper way to describe a vector multipole field is by vector spherical harmonics,
discussed in Chapter 9.
J:<Or a aipo1e p a10ng me z axis, me news in V'l.ll) n ro me rammar
form:
Lp COS t1
Er = 3
47TE0 r
p Sill tJ (4.12)
Eo= 3
47TE 0 r
E"' = 0
These dipole fields can be written in vector form by recombining (4.12) or by
directly operating with the gradient on the dipole term in (4.10). The result for
the field at a point x due to a dipole p at the point x0 is:
3n(u · n) - D
1'-~X) ~'I.U)
47TE0 lx - x 0 1
3
n ;, ~ unit " ;, frAm ~ tA ~

·v
There are two important remarks to be made. The first concerns the rela-
tionship of the Cartesian multipole moments like (4.8) to the spherical multipole
moments (4.3). The former are (l + 1)(! + 2)/2 in number and for l > 1 are more
numerous than the (2! + 1) spherical components. There is no contradiction here.
The root of the differences lies in the different rotational transformation prop-
ertJes ot the two types of multtpole moments; the Cartestan tensors are reductblc,
the spherical, irreducible~see Problem 4.3. Note that for l = 2 we have recog-
• cl th . h
"J
"">:' . . ,, ,/, .. r'c . n m
,.
"~•mAl~ m. •• (A o\
\ . ,.
The second remark is that in general the multipole moment coefficients in

the expansion (4.1) depend on the choice of origin. As a blatant example, con-
oirlPr " nn;nt .t, o lnrotPrl ot v. (r IL A-..\ Tto otiol hoo " mnltinC\lP
r
expansi~n of the f~rm (4.1) with multipole moments,

tv* rn .1. \
'11m u tm\~,,. -ru1
ThPcP "rP nt iohina fnr "11 I W> in .,] nnlu thP I () mn lJ innlP
-J -r
~· .c •• • 0 .c •• n.
'10() v> V <~ " . r V< '"~ <V~U,<VH V ' <~ t-'V U '5' • I'
charges + e and -e at x0 and x1o respectively, the multipole moments are
qlm elro r zm~t1o, (/Jo) '

T1I tm~t1Jo CfJJ)J
148 Chapter 4 Multipoles, Electrostatics of Macroscopic Media, Dielectrics-SI
Now the 1 = 0 multipole moment of the system vanishes, and the 1 = 1 moments
qw =
i(Yo
These moments are independent of the location of the origin, depending only on
the relative position of the two charges, lmt all higher moments depend on the
location of the origin as well. These simple examples are special cases of general
theorem (see Problem 4.4). The values of q1m for the lowest nonvanishing mul-
tipole moment of any charge distribution are independent of the choice of origin
of the coordinates, but all higher multipole moments do in general depend on
the location of the origin.
Before leaving the general formulation of multipoles, we consider a result
that is useful in elucidating the basic difference between electric and magnetic
dipoles (see SectiOn 5.6) as well as in other contexts. Consider a localized charge
distribution p(x) that gives rise to an eleetrie field E(x) throughout space. VIe
wish to calculate the integral of E over the volume of a sphere of radius R. We
begin by examining the problem in general, but then specialize to the two ex-
tremes shown in Fig. 4.1, one in which the sphere contams all of the charge and
the other in which the charge lies external to the sphere. Choosing the origin of
coordinates at the center of the sphere, we have the volume integral of the electric
field,
{ {
L.R 3
E(x) d x - -] r<R
V<Pd 3x (4.14)
This can be converted to an integral over the surface of the sphere:

( {
LR E(x) d x - - LR R
3 2
df! <P(x)n ( 4.15)
(a) (b)
Figure 4.1 Two configurations of charge density and the spheres within which the
volume integral of electiic field is to be calculated.
s ....t. 4.1 . . "'· 149
where n is the outwardly directed normal (n = x/R). Substitution of (1.17) for

the potential leads to
r n2 r r
),<R E(x) d x 3
47T€0
Jd 3
x' p(x') ),
r~R
dil
lx - X
,
I
(4.16)
To perform the angular integration we first observe that n can be written in terms
of the snherical an2:les (8 m) as
n = i sin () cos cjJ + j sin 8 sin cjJ + k cos ()
Evidently the different components of n are linear combinations of Y1m for
I 1 ""''" WhPn n 15<1 r.r (':\ 7()\ ;, ; intr. (.:11{;\ nrthncrr.nolih1 nfthP V.
. ehmmate
will . : aII but the
' l = 1 term m. th e senes.
. Thus we h ave
{ dil
n r< { dil n cos 'V (4.16')
Jr-R IX XI r> J
where cos y = cos 8 cos 8' + sin () sin ()' cos( cjJ - cjJ '). The angular integral is
equal to 47Tn'/3, where n' = r'!r'. Thus the integral (4.16) is
r RL r r
Jr<R E~XJ ax 3eo J a x r>2 n p~x ) ~Lf.ll)
where (r <• r>) = (r', R) or (R, r') depending on which of r' and R is larger.
If the sphere of radius R completely encloses the charge density, as indicated
in Fig. 4.1a, then r< - r and r> - R m (4.1 I). The volume mtegral of the electnc
field over the sphere then becomes
{ E(x) d 3 x =
p
(4.1R)
J r<l< Jto
where p is the electric dipole moment (4.R) of the charge distribution with respect
to the center of the sphere. Note that this volume integral is independent of the
size of the svherical re2:ion of integration provided all the charge is inside.
Tf <h MhPr h~nrl , <hP e;<,~<iAn ie •>e rbni~tPrl in Rirr A 11-. mith thP
-r ·cr ·o
'
all exterior to the sphere of interest, r< = R and r> = r' in (4.17). Then we have
( l?fv\
., Fl3Y - R3 ( Fl3Y' ~ ,-,(v'\
Jr<R 3eo J r'"
From Coulomb's law (1.5) the integral can be recognized to be the negative of
47T€o times the electric field at the center of the sphere. Thus the volume integral
·" v .
f r<R
E(x) d 3 x = -47T R 3 E(O)
3
(4.19)
tn otner woras, tne average vame or tne electric neJO over a spnericat vomme
containing no charge is the value of the field at the center of the sphere.
The result (4.18) implies modification of (4.13) for the electric field of a
dipole. To be consistent with (4.18), the dipole field must be written as
-~
Ju~p
·' I:' -
A
'" po(x - x 0 )
E(x) = _:':_ .. , (4.20)
47TE0 lx - x0 l3 3
llH;; UClld aoes not comrioute ro me new away rrom me site of

the dipole. Its purpose is to yield the required volume integral (4.1R), with the
convention that the spherically symmetric (around x 0 ) volume integral of the first
term is zero (from angular integration). the singularitv at x - x, "~"~ina ~n
otherwise ambiguous result. Equation (4.20) and its magnetic dipole counterpart
(5.64), when handled carefully, can be employed as if the dipoles were idealized
point dipoles, the delta function terms carrying the essential information about
" . . . .
tll<- ·-o "'11''-' 1uuus m cnarge anu currenL
4.2 Multinole Exnnn<!inn nf thP V-nM~ .. nf n rhnroP ni<.:tril. .. ~:~-

~~ " ~
in an External Field
If a localized charge distribution described by p(x) is placed in an external
potential ct>(x), the electrostatic energy of the system is:
W = Jp(x)ct>(x) d 3
x (4.21)
Tf the 1t~ v~rvina C\VPT thP TPO"lC\Il whPrP~norrl;rr;hlo

then it can be expanded in a Taylor series aro;nd a suitably chosen origin: '
1 azct>
ct>(x) ct>(O) + x • Vct>(O) + ~ L Lj X;X; •
UAj UAj
(0) + ... (4.22)
"" i
<T •••
' ..
\.) lll<- v< we : uuu n · v ""• LHe 1aSL LWO Lerms can De
rewritten. Then (4.22) becomes:
l "'\' "'\' aEj rrn

d:J(y) cbfn) ~ INn\
., -/
' 1 ax. \-/
2...,., j
'
Since V · E = 0 for the external field, we can subtract
1
6
z..,
•
'"'""
~\VJ
HUH! LHt: li"l LC!Hl LU nnauy me expansion:
ll>(X) ll>(U) x · E(O)

6
1
+f (3x;xj r"8)
'1 iJ X;
aE
(0) + ... (4.23)
When this is inserted into (4.21) and the definitions of total charge, dipole mo-
ment (4.8), and quadrupole moment (4.9) are employed, the energy takes the
form:
1 iJE
w - q'P(O) p · E(O) : 2. 2.· Q;j " (0) + ... (4.24)
6 ' 1
ax
'
..•
This expansion shows the characteristic way in which the various multipoles in-
~
_,
-"w . . ·~· "'-''
' " . . .
-,w; '-'Hdlt;"' wu11 tll<O pvt<OHllctl, "•"' ··~
.
wnn tne
electric field, the quadrupole with the field gradient and so on.
T, •1 .
"'eL
r
,],
-o
th
"· -r "" v•
.~ .1
t'~'"''-'~·~· .,,.._,, vo •
nuclei can possess electric quadrupole moments, and their magnitudes and signs
reflect the nature of the forces between neutrons and protons, as well as the
Sect. 4.3 Elementary Treatment of Electrostatics with Ponderable Media 151
shapes of the nuclei themselves. The energy levels or states of a nucleus arc
described by the quantum numbers of total angular momentum J and its projec-
tion M along the z axis, as well as others, which we will denote by a general index
a. A given nuclear state has associated with it a quantum-mechanical charge
density* PJMa(x), which depends on the quantum numbers (J, M, a) but is cylin-
drically symmetric about the z axis Thus the only nonvanishing quadrupole mo-
mentis q 20 in (4.6), or Q 33 in (4.9). 1 The quadrupole moment of a nuclear state
is defined as the value of (1/e) 011 w1th the charge dens1ty PrMa(x), where e 1s the
(bMa (4 25)
The dimensions of QJMa are consequently (length?. Unless the circumstances

arc exceptional (e.g., nuclei in atoms with completely closed electronic shells),
nuclei are subjected to electric fields that possess field gradients in the neighbor-
hood of the nuclei. Consequently, according to (4.24), the energy of the nuclei
will have a contribution from the quadrupole interaction. The states of different
M value for the same J will have different quadrupole moments QJMm and so a
degeneracy in M value that may have existed will be removed by the quadrupole
coupling to the "external" (crystal lattice, or molecular) electric field. Detection
of these small energy differences by radiofrequency techniques allows the deter-
mination of the quadrupole moment of the nucleus.*
The interaction energy between two dipoles p 1 and p 2 can be obtained di-
rcctly from (4.24) by using the dipole tleld (4.20). Thus, the mutual potential
w,2 (4.26)
where n is a unit vector in the direction (x1 - x2 ) and it is assumed that x 1 Xz. *
The dipole-dipole interaction is attractive or repulsive, depending on the orien-
tation of the dipoles. For fixed orientation and separation of the dipoles, the
value of the interaction, averaged over the relative positions of the dipoles, is
zero. If the moments are generally parallel, attraction (repulsion) occurs when
the moments are oriented more or less parallel (perpendicular) to the line joining
their centers. For antiparallel moments the reverse is true. The extreme values
of the potential energy are equal in magnitude.
4.3 Elementary Treatment of Electrostatics with Ponderable Media

In Chapters 1 2, and 3 we considered electrostatic potentials and fields in the
presence of charges and conductors, but Ito other ponderable media. 'Ne there
"See Blatt and Weisskepf (pp. 23 ff.) for an elementary discussion of the <J:11antum "'peels of the
'"The quadrupole moment of a nucleus," denoted by Q, is defined as the value of QJMo in the state
M = J. See Blatt and Weisskopf, loc. cit.
152 Chapter 4 Multipoles, Electrostatics of Macroscopic Media, Dielectrics-Sf
rore no ms1 oetween microscopicneurs ana macroscopic news, a!-

though our treatment of conductors in an idealized fashion with surface charge
densities implied a macroscopic description. Air is sufficiently tenuous that the
ne!!lect of its dielectric orooerties causes no "Teat error our results so far 'u"
T r •• D •• ,j. ,; ,J .>£ '•L -L
<5"0 W
~
<U
"'.u
H~<'-»
in ponderable media whose respective electric responses must be taken into ac-
count. In the Introduction we indicated the need for averaging over macroscop-
ically small, but microscopically large, regions to obtain the Maxwell equations
" · 1tP. for m · nh lP.nll Thi< i< -::lnnP in ~ nl f~ehicm in
Chapter 6, after the Maxwell equations with time variation have been discussed.
For the present we merely remind the reader of the outlines of the elementary
discussion of polarization in a fashion that glosses over difficult and sometimes
suoue aspens 01 me averagmg proceaure ana tr e mtroduct10n ot tne macrosco01c
quantities.
The first observation is that when an averaging is made of the homogeneous
equation, V X Emicro = 0, the same equation, namely,
V X~ u l4.L I)
holds for the averaged, that is, the macroscopic, electric field E. This means that
the electric field is still derivable from a potential ct>(x) in electrostatics.
Tf lln P.], · fip](] i~ llnnliP.rl to " mP.rlin~ ~~r1, '"' nf ~ l~rc'" '"r .-,f
T ·o
atoms or molecules, the charges bound in each molecule will respond to the
applied field and will execute perturbed motions. The molecular charge density
,,,,_ ,. _, T'J. ••. _, .1' ' _, ' .,, • "
'l'~" ~• ~u~ VVHl Uv L'-'llL HVIII
what they were m the absence ot the held. ln s1mpTe substances, when there is
no applied field the multipole moments are all zero, at least when averaged over
many molecules. The dominant molecular multipole with the applied fields is the
rlinnlP ThPrP i~ th11< nrnrlllrPrl in thP mPrli11m ~n · n~lnr;
·r
· P { ,>: -L
,. · r
. '· <
' '
u l''-'I u u n 'V<U<H'-'; IS''"'" uy
P(x) = L N,(p,) (4.28)

i
_,_ '.,_ •,'

V>llviv Jli 10 "'-
Lllv
~·
-r -' VI Lll'- <Lll ' J l ' ' - Ul
" ' '
lll lllC , lllt.O
average is taken over a small volume centered at x and N, is the average number
per unit volume of the ith type of molecule at the point x. If the molecules have
a net charge e, and, in addition, there is macroscopic excess or free charge the
-o
rl ~· the- ~ 'J
. . ] , p] ;n ~
p(x) = L
i
N,(e,) + Pexcess (4.29)
thP
T Te11~1lu ,J ul~r ie TJ..~n
••
~· .....,.,.
<J..
'J --o ·o
7C>rro
" "J
excess or free charge (suitably averaged).
If we now look at the m "' from " nf viPur "'" r~n m~
•• .• 1 <: .• u 1- 1' .. " .LllC
'
J.. ' " ' 't'
r oul'v< r ~J "
HVlll VI
each macroscopically small volume element LlV at the variable point x'. Thus
the charge of Ll Vis p(x') Ll V and the dipole moment of Ll Vis P(x') Ll V. If there
are no higher macroscopic multipole moment densities, the ootential Lltl>(x, x')
Sect. 4.3 Elementary Treatment of Electrostatics with Ponderable Media 153
caused by the configuration of moments in

given without approximation by
il V can be seen from (4.10) to be
M>(x, x') = - 1-
47TE0
I p(x')
Llx - x' I
ilV + P(x') · (x -, x')
lx - x' lo
ilv] (4.30)
provided x is outside il V. We now treat il V as (macroscopically) infinitesimal,

put it equal to d 3 x ', and integrate over all space to obtain the potential
into
1
<l>(x) = - -
47TE0
J d x' lx -1 x'
3
I
[p(x') - V' • P(x')] (4.32)
This is just the customary expression for the potential caused by a charge distri-
bution (p - V · P). WithE = - V<l>, the first Maxwell equation therefore reads
1
V · E = - [p - V · P] (4.33)
Eo
The presence of the divergence of P in the effective charge density can be un-
( 4.33) becomes the familiar

V·D = p (4.35)
Equations (4.27) and (4.35) are the macroscopic counterparts of (1.13) and (1.14)
of Chapter 1.
As discussed in the Introduction, a constitutive relation connecting D and E
is necessary before a solution for the electrostatic potential or fields can be ob-
tained. In the subsequent sections of this chapter we assume that the response
of the field is linear. This excludes from
Figure 4.2 Origin of polarization-charge density.

Because of spatial variation of polarization, more
molecular charge may leave a given small volume
than enters it. Only molecules near the boundary are
shown.
154 Chapter 4 Multipoles, Electrostatics of Macroscopic Media, Dielectrics-51
L ,1, · ~~~~o Ac o f,~tho~ · -··~ , mo CH

-rr
, thot tho
·"
is isotropic. Then the induced polarization Pis parallel to E with a coefficient of
· 1~litv th~t i< in,-' ' Clf ,. · •
• ' • •
P = ~"oXeE (4.36)
The constant Xe is called the electric susceptibility of the medium. The displace-
ment D is therefore proportional to E,
D = EE (4.37)
where
E' = E'o(l + Xe) (4.38)
.. . c/c_ + " ;c roollPrl tho d; - 1 ' C\T TP b tivP
;otho-'
r 'J
1
' "'
electric permittivity.
If thP. rlif>lPrtrir i< nC~t "" lv · ir hut ~ k" uniform then "'is indenendent
of position. The divergence equations (4.35) can then be written
- ~
• • .0:.
1-"' ~
' ' ~mJ
~ "ToJJ
All problems in that medium are reduced to those of preceding chapters, except
that the electric fields produced by given charges are reduced by a factor E'0 /E'.
The reduction can be understood in terms of a polarization of the atoms that
produce fields in opposition to that of the giVen charge. One Immediate conse-
quence is that the capacitance of a capacitor is increased by a factor of E'l ~"o if the
empty space between the electrodes IS hlled with a dielectnc with dielectnc con-
stant E'/!'0 (true only to the extent that fringing fields can be neglected).
H we Uni10fm meuium uueS llUL Jill au Ul lilt: 'IJdlOt: WliCJC LllCIC dlC LHC
fields or, more generally, if there are different media juxtaposed, not necessarily
. . '' ~-' _, _,. .
HI lH'-'H 1' ·r , VVOO lllUOL Ul'-' '{I VI
on D and E at the interfaces between media. These boundary conditions are

· <-
~~>>0 ~~
<L £,.11
H~H>
.<
<W
11 ''i' •
>YH
• C
"~·
• ~K.
T C 'T'L
• •
.1<
~
<L
the normal components of D and the tangential components of E on either side

,j' ot-'oh th h -'- A;t; -nl;rl-+~~ t;~o- • •~ ~c moll ~c
·; 'J ·; ·o
'
static fields,
~u2 U1) • n21 IT
(4.40)
(E 2 - E 1) X n 21 =0
where n21 is a unit normal to the surface, directed from region 1 to region 2, and
rr io thP. m: · <11rf~rP-rh~r<w ' · " " thP hC111nrl~rv -' (nnt inrlurl-
ing the polarization charge).
4.4 Boundary- Value Problems with Dielectrics

The methods of earlier chapters for the solution of electrostatic boundary-value
problems can readily be extended to handle the presence of dielectrics. In this
section we treat a few examples of the various techniques applied to dielectric
media.
To illustrate the method of images for dielectrics we consider a point charge
q emoeaaea m a semi-innnite aietectric E:1 a aistance a away rrom a plane inter-
face that separates the first medium from another semi-infinite dielectric E2 • The
appropriate solution to the equations:
E2V. E = 0, z<O (4.41)

and
V X E = 0, everywhere
subject to the boundary conditions at z = 0:
Et
lim Ex = lim Ex (4.42)
Since V x E = 0 everywhere, E is derivable in the usual way from a potential

<1>. In attempting to use the image method it is natural to locate an image charge
at a point P described by cylindrical coordinates (p, cp, z) will be
completely analogous to the problem with a conducting material in place of the

dielectric E2 for z < 0. But we now must specify the potential for z < 0. Since
there are no ch in the z < 0, it must be a solution of the
equation without singularities in that regiOn. '"""'au
that for z < 0 the potential is equivalent to that of a charge q" at the position A
of the actual charge q:
1 q"
--- z< 0 (4.44)
q
Smce
a( 1 a ( 1 d
z:-::0 z=O
while
a:UJ. z~o =
a(
dp R2
1) z=O
.llHO
' 'UV ' LJ -·· . ·~-
q q q
~ (q + q') = ' q"

"· "-
~ ' '
E2 Et )
q' =- q
E2 + E1
(4.45)
2
q" = ( 2E )
E2 + Et q
For the two cases E2 > E1 and E2 < E1 the lines of force (actually lines of D) are
shown qualitatiVely m Fig. 4.5.
'T'h ,] ' ' '" -' L ....... T '' ,~,_ ~,,,
r 'b' b' •J • ""•
P = EoXeE, so that -V · P = -EoXeV • E = 0, except at the point charge Q. At
the surface, however, Xe takes a discontinuous jump, VXe = ( E1 - E2 )/ Eo as z passes
..
LHlvu~H <. v. uH>
.
,
. ~·. .
waL LlHOlv 1> a pvuu
- .
-s1
.
-cnargto uensny un
the nlane 7 - 0:
.In D I< '-''
p01 \ - L l/ 1 21 \~·~V}
\
~~l'\\ ~ !\\ I
---
I
~'\\
------ """ ~ .....
~
I
~ v
I/
...----- ~n~ ____.... ~ \\--
____..../ VI, \ ~ ·;; / \"'-
/ / VI '
7 r77 \
/j / I I
I
e2 >el e2 <el
Figure 4.5 Lines of electric displacement for a point charge embedded in a dielectric
<1 near a semi-infinite slab of dielectric <2 •
Sect. 4.4 Boundary-Value Problems with Dielectrics 157
where n21 is the unit normal from dielectric 1 to dielectric 2, and P, is the polar-
ization in the dielectric i at z = 0. Since
P, = ("' - <'o)E, = -( <'; - <'o)V<P(O")
It Is a simple matter to show that the polanzation-charge density is
(4.47)
27T <'r ( <'z + <'r) (pz + d2)3/2
In the limit Ez >> E 1 the dielectric Ez behaves much like a conductor in that the
electric field inside it becomes very small and the surface-charge density (4.47)
approaches the value appropriate to a conducting surface, apart from a factor of
Eo!Er.
The second illustration of electrostatic problems involving dielectrics is that
of a dielectric sphere of radius a with dielectric constant E/e0 placed in an initially
uniform electric field, which at large distances from the sphere is directed along
the z axis and has magnitude E8 , as indicated in Fig 4 6 Both inside and out-
side the sphere there are no free charges. Consequently the problem is one of
solving the Laplace equation with the proper boundary conditions at r = a. From
the axial symmetry of the geometry we can take the solution to be of the form:
INSIDE:
(4.48)
OuTSIDE:
<Pout = 2: [B,r 1 + C1r-(l+I)]P1(cos 8) (4.49)

!~a
From the boundary condition at infinity ( - E0 z = -Ear cos 8) we find that
the only nonvanishing B 1 is B 1 = -Ea. The other coefficients are determined
from the boundary condll1ons at r - a:
TANGENTIAL E:
-~ a<P,n I _ ~ iJtPoutl
a a8 . ' a a a8 . ' a
(4.50)
NORMAL D:
r-a
When the series (4.48) and (4.49) are substituted, there result two series of
Legendre functions equal to zero. Since these must vanish for all IJ, the coef-
Figure 4.6
-t:o
~,.,~ ~ua~n~> ~
. . m
.
-~·
'' -c
'
ficient of each Legendre function must vanish separately. For the first boundary
condition this leads (through orthogonality of P\ = aPtfa8) to the relations:
- '---'1
A1 - -E0 +-----:;-
(4.51)
r.
At - a2t+l for l =F 1
while the second gives (through orthogonality of P 1):
2-
c,
(<'i<'o)Al = -E0 -
a3
(4.52)
( '"--'I
<'i<'o)lAt = -(l + 1) 21+1 for l =F 1
~
1 liC CL[ UdllVH> 111 ~ 't.J 1) illlU v+.JL) !Cilll UC >ill IUMY UlllY
with A C, 0 for all l =F 1 The remainin!! coeffi, · are !!iven in terms of

the applied electric field E 0 :
A1 = - ( 3 ) E0
2 + <'i<'o
I -1
( 4.53)
1
cl I ,j,~ + ) )a3Eo
'T'~ . '1 . .].,
'
I \
J
<Pin= - clc.u + 7
E 0 r cos(}
(4.54)
<'~<'o
3
 out = -Ear COS 8 + ( - 1) Eo 2a COS 8

Ei<'o + 2 r
. .
lllC 1-'v' Lilt: SjJilt:ft: Ut:SLuucoS a (;UilSLillll uem paraue1 w
the annlied field with m'H>nitncie
'l
Ein I E 0 < E 0 if <' > <'o (4.55)
<' <'n + 2
n, •. · r1. •• ,].
-r ·~ . . '1 .
.. ..,_ ..
,]. •• •. .h
~.-.-
>I' rl h ,Jrl c;
-v r
,), th
11C1U 01 illl C1CICL111C •r aL LHe ungm WlLu u1p01e momenL:
<'i<'o 1
p 47T<'o )a Eo3
(4.56)
,EIEo -r- Lj
/--~ -7-"'.:t+
Eo p l Eo
--- +
...
\ /' _, 4-
/l
.,.
-
Figure 4.7 Dielectric sphere in a uniform field £ 0 , showing the polarization on the left
~ •L _, . . -'- .,,_ . . . . .
UH~ •a~ ~au• fi~ H H H HO
' "l-'1 Q' ~·~~" •~ u~•u " " "'~ "5'"·
Figure 4.8 Spherical cavity in a dielectric
with a uniform field applied.
oriented in the direction of the

preted as the volume mtegral of the
(4.57)
It is constant throughout the volume of the sphere and has a volume integral
(4.58)
This can be thought of as producing an internal field directed oppositely to the
in Fig. 4.7.
as shown in Fig. 4.8, can be handled in exactly the same way as the dielectric
· · shows that the results
for the cavity can be obtained from those of the sphere by the replacement
Eo ---+
to E 0 , and of magnitude:
Similarly, the field outside is the applied field plus that of a dipole at the origin
· · to the field and with moment:
p=
4.5 Molecular Polarizability and ectric Suscep

In this section and the next we consider the relation between molecular prop-
although a proper treatment necessarily would involve quantum-mechanical con-
classical analysis.
~ore exammmg how the detailed properties ot the molecules are related
+~ +h~ "'" ;h;];+u mn ~""' ~olr~ ~ '' ' ' +h h ,],-l, ,,; +h
•y
' 'b
molecules in the medium and the applied field. The susceptibility is defined
through the relation P = EoXeE, where E is the macroscopic electric field. In
-ra:reneu- -wrrere- -r are targe were IS utue muercu"e
between the macrosconic field and that actin" on anv molecule or "roun of mol-
ecules. But in dense media with closely packed molecules the polarization of
neighboring molecules gives rise to an internal field E; at any given molecule in
addition to the average macroscopic field E, so that the total field at the molecule
is E + E,. The internal field E; can be written as the difference of two terms,
E; = Enear - Ep (4.61)
where Enear is the actual contribution of the molecules close to the given molecule
ariifKp 1s ffle contribution from those molecules treated man average contmuum
• ,,· ,., ., • " .-... .1 · · n "'" · L • ''"
~n ~J r · '1'6
close to the molecule in question we must take care to recognize the specific
atomic configuration and locations of the nearby molecules. Inside some mac-
. .
lUSCOi' • y --srrmrr,-uuT -urrge; v we
the smoothed marro · · Pnniv:>IPnl nf thP nP:>rhv mnlPrnbr LUI~:~~~:~~~~~
and replace it with the correctly evaluated contribution (Enear)· This difference
is the extra internal field E;.
The result (4.18) for the integral of the electric field inside a spherical volume
of radius R containing a charge distribution can be used to calculate Ep. If the
volume V is chosen to be a sphere of radius R containing many molecules, the
total dipole moment inside is
4,.R 3
p--P
3
nrovirlP-rl V ;, so <m:> 11 thM P ;, i:> llv rnn<t:>nt thrnn"hnnt thP. vnlnmP. ThP.n
(4.18) shows that the average electric field inside the sphere (just what is desired
I Or-x;:;} IS
3
Ep = - - 3
47TR
f
r<R
E d 3x = --
p
3E0
(4.62)
The internal field can therefore be written
1
E; = - P + Enear (4.63)
--::rEo
liletleiO due to the molecules ncar by is more difficult to determine. Lorentz

I. 1'JO\_L "''" <. ' • • .L' 1 ' V '_1.
\~· Jf ·r ncar "Y
lattice site. The argument depends on the symmetry of the problem, as can be
seen as follows. Suppose that inside the sphere we have a cubic array of dipoles
~."-'7,-witrr-arr-.:neir m--umems constant in magnimae ana
ClnPntp.-J :>Jnncr thP <:>mP .-J' • (rpm, · th:>t thP .L i< m:>o . 1Jiu
small even though it contains very many molecules). The positions of the dipoles
arc given by the coordinates X;jk wlth the components along the coordinate axes
t;~ '~ 1-~\
'J' '
.,, • ...h~~~ ' +h l..u.'
~t"
'
'6•
.:i · ·
' J'
1. ,t. .L
r
' · "
Sect 4 5 Molecular Polarizahility and Electric Susceptibility 161
Figure 4.9 Calculation of the internal field:

contribution from nearby molecules in a
simple cubic lattice.
negative integer values. The field at the origin due to all the dipoles is, according
to 4.13 ,
E = " 3(p . X;jk)x;;k - xtkP
-6 4 1TEoX 5"k (4.64)
. "k
l,j, I]
The x component of the field can be written in the form:
(4.65)
Since the indices run equally over positive and negative values, the cross terms
involving (ijp 2 + ikp 3 ) vanish. By symmetry the sums
are all equal. Consequently

f3i 2 - cP + / + e)]Pl
El = ti 41TEoa3(i2 + ? + k2)5/2 = 0 (4.66)
Similar arguments show that the y and z components vanish also. Hence
Encar = 0 for a simple cubic lattice.
If Enem - 0 for a highly symmetric situation, it seems plausible that Enear - 0
also for completely random situations. Hence we expect amorphous substances
to have no internal field due to nearby molecules. For lattices other than simple
cubic. the components of E""" are related to the components of P through a
traceless tensor sat' that has the symmetry properties of the lattice Nevertheless,
it is a good working assumption that Encar = 0 for most materials.
The polarization vector P was defined in (4.28) as
where (Pmot) is the average dipole moment of the molecules. This dipole moment
is approximately proportional to the electric field acting on the molecule. To
exhibit this dependence on electnc field we define the molecular polanzabzhty
'Ymol as the ratio of the average molecular dipole moment to Eo times the applied
field at the molecule. Taking account of the internal field (4.63), this gives:
(Pmol) = Eo'Ymoi(E + E,) (4.67)
162 Chapter 4 Multipoles, Electrostaucs of Macroscopic Medm, Dielectncs SI
'Ymoiis, in principle, a function of the electric field, but for a wide range of field ·
strengths is a constant that characterizes the response of the molecules to an
applied field. Equation (4.67) can be combined with (4.28) and (4.63) to yield:
P- Nrma{ EoE + ! P) (4.68)
where we have assumed E""" - 0. Solving for P in terms of E and using the fact
that P = EoXeE defines the electric susceptibility of a substance, we find
e
1
1- 3 Nrmol
as the relation between susceptibility (the macroscopic parameter) and molecular
polarizability (the microscopic parameter). Since the dielectric constant is
E/E0 1 + Xe. If can be expressed m terms of 'Ymoh or alternahvely the molecular
polarizability can be expressed in terms of the dielectric constant:
'Ymol = ~ (:;: + ?) ( 4.70)
This is called the Clausius-Mossotti equation, since Mossotti (in 1850) and
Clausius independently (in 1879) established that for any given substance
( El Eo - 1)/( E! ~:0 + 2) should be proportional to the density of the substance.* The
relation holds best for dilute substances such as gases. For liquids and solids,
(4. 70) IS only approxnnately vahd, especmlly 1f the dielectric constant is large.
The interested Hlader can refer to the books by Bottcher, Dcbyc, and Frohlich
for further details.
4.6 Models for the Molecular Polarizability

The polarization of a collection of atoms or molecules can arise in two ways:
the applied field distorts the charge distributions and so produces an induced
dipole moment in each molecule;
the applied field tends to line up the initially randomly oriented permanent dipole
moments of the molecules.
To estimate the induced moments we consider a simple model of harmomcally
bound charges (electrons and ions) Each charge e is bound under the action of
a restoring force
where m is the mass of the charge, and w0 the frequency of osclllatwn about
equilibrium I lnder the action of an electric field E the charge is displaced from
its equilibrium by an amount x given by
mw?ix = eE
1
At optical frequencies, €1<0 n 2 , where n !S the mdex of refractwn. W1th n 2 replacmg <i<o m (4.70),
the equation is sometimes called the Lorenz Lorentz equation (1880).
Sect. 4.6 Models for the Molecular Polarizability 163
Consequently the induced dipole moment is
Pmot = ex - --2 E (4.72)

mw 0
This means that the polm izability is y e2 /mw5e 0 • If there are a set of charges e;
with masses mi and oscillation frequencies wi in each molecule then the molecular
polarizability is
(4.73)
To get a feeling for the order of magnitude of y we can make two different
estimates. Since y has the dimensions of a volume, its magnitude must be of the
order of molecular dimensions or less, namely Yet :S 10- 29 m 3 . Alternatively, we
note that the binding frequencies of electrons in atoms must be of the order
of light frequencies. Taking a typical wavelength of light as 3000 A, we find
w = 6 X 10 15 s- 1 • Then the electronic contribution to y is Yet ~ (e fmw E0 ) ~
2 2
0.88 X 10 29 m', consistent with the molecular volume estimate. For gases at
NTP the number of molecules per cubic meter is N = 2.7 X 1025 , so that their
susceptibilities should be of the order of Xe < 10 3 . This means dielectric con
slants differing from unity by a few parts in 103 , or less. Experimentally, typical
values of dielectric constant are 1.00054 for air, 1.0072 for ammonia vapor, 1.0057
for methyl alcohol, 1.000068 for helium. For solid or liquid dielectrics, N ~ 1028
- 1029 molccules/m 3 . Consequently, the susceptibility can be of the order of unity
(to w1thm a factor 10 1) as IS observed.*
The possibility that thermal agitation of the molecules could modify the re-
sult (4.73) for the induced dipole polarizability needs consideration. In statistical
mechanics the probability distribution of particles in phase space (p, q space) is
some function f(H) of the Hamiltonian. For classical systems,
f(H) = e-HikT (4.74)
is the Boltzmann factor. Fm the simple problem of the harmonicdlly bound
charge with an applied field in the z direction, the Hamiltonian is
m
H = - p2 +- w6x 2 - eEz (4.75)
2m 2
where here pis the momentum of the charged particle. The average value of the
dipole moment in the z direction is
mo
If we introduce a displaced coordinate x' = x - eEzfmw5 then

2 2
1 mw 2 e E
11 pz + g (x')z (4.71)
2m 2 2mw5
*See, e.g., CRC Handbook of Chemistry and Physics, 78th ed., ed. D. R. Lipe, CRC Press, Boca
Raton, FL (1997-9g).
and
'
I .~ I .~
e· Jc,
/UJlfUA e.,; 2 1111 J
\ "'"'O;
<Pmol) = ( 4.78)
Jd p Jd x' f(H)
3 3
Since His even in z' the first integral vanishes. Thus, independent of the form
of f(H), we obtain
e
'PmoV mw6--zs
just as was found in (4.72), ignoring thermal motion.
The second type of polarizability is that caused by the partial orientation of
otherwise random permanent dipole moments. This orientation polarization is
1moortant m "oolar substances such as HCl and HoO and was first discussed bv
.-,. .1. /1<>1 " ' -A 11 -1. .1 ~ --1: .1.
ry~ ~· 'l' 1' ·~y
p 0 , which can be oriented in any direction in space. In the absence of a field,

thermal agitation keeps the molecules randomly oriented so that there is no net
dipole moment. With an applied field there is a tendency to line up along the
field in the configuration of lowest energy. Consequently there will be an average
T' T ~ T T d ' d d T T 'T ' . 0' d T -,
Ull'UHO , lU llll~ Wv llUtv tHat llHO Ul llHO
Ismvenov---
H = H0 - Po· E (4.79)
where H 0 is a function of only the "internal" coordinates of the molecule. Using
the Boltzmann factor (4.74), we can write the average dipole moment as:
I 'noE cos 1:1
I d!! Po cos IJ exp
J K1
\Pmolf l4.1:SU)
Jd!l exp (p EkTcos 1:1) 0
where we have chosen E along the z axis, integrated out all the irrelevant vari-
~hlPo ~nrl --' th~t nnhr thP wnt nf /n-. \ " ' tn thP fiPlrl io ,. " 1t
-rrunrzeru.-nr rar,~ >areu LU unny, excepL m 1uw
temperatures. Hence we can expand the exponentials and obtain the result:
1 p6
<Pmol) = 3 kT E (4.81)
"Fl1P ~ nol>~ri7:c~tion :i -1 on thP. r>~tnrP. ,.~ ·hp

" ;"
' C'UT<IITCllCI.Otlll .. ,;
Wlll\.011 lllC appueu
~
liClU iliUM UVCli.OUlllC lilt:

..
I Ul
'l'
thermal ag1tat10n.
In general both types of polarization, induced (electronic and ionic) and
orientation, are present, and the general form of the molecular polarization is
1 n~
11. _, - '\1, + - 14XZl
3E0 Kl
Sect. 4.7 Electrostatic Ener!!V in Dielectric Media 1()5
7 Polar
t r r T T T. ~
'Ymol ~~~~Nonpolar
Fignre 4.10 Variation of molecular

polarizability 'Ymot with temperature
lUl !JUid1 d11U IIUII!JUictl Ml!J" '.
ItT- 'Ymol versus I
lhlS snows a temperature ctepenctence ot the torm (a + D/1) so that the two
t >o nf oAl• ,,.;7~t;nn ~~" h~ o~n"r"t~rl ;~ont" llu "" · -'' >tori ;" p;~ A 1f)
::or r ·r- ·r ' ·e
ror po,ar mo,ecu.es, sucu as n L I auu rr2 v , llle uuse1 veu !-'"' LL -.-
moments are of the order of an electronic charge times w-R em, in accordance
with molecular dimensions.
4.7 Electrostatic Energy in Dielectric Media

In Section 1.11 we discussed the energy of a system of charges in free space. The
result obtained there,
W = ~ Jp(x)<l.>(x) d x 3
(4.83)
for the enenrv clue to>~ ch~nre clensitv n(x) and,, notential <J.>(x) cannot in P"eneral
'-
uv
'-
v ''-'' a~ n
.
' Ul VUl ·r
_,
.,JllVll Vl
c '. '
llv
... • ~·-
1 Hv
reason becomes clear when we recall now (4.KI l was obtamed. We thought ot
tJ.. ho ,] ,>;, • t: ... 'J.. .. t .rl J.. ohl;n h;t hu h;t tJ..~
·o "b" 0 "J "b. 'J
elemental cnarges, bnngmg eacn one m trom innnitely tar away agamst the action
of the then existing electric field. The total work done was given by (4.83). With
dielectric media, work is done not only to bring real (macroscopic) charge into
position, but also to produce a certain state of polarization in the medium. If p
and <l.> in ( 4.83) represent macroscopic variables, it is certainly not evident that
(4.83) represents the total work, including that done on the dielectric.
To be general in our description of dielectrics, we will not initially make any
assumptions about linearity, uniformity, etc., of the response of a dielectric to an
applied field. Rather, let us consider a small change in the energy oW due to
SUll Ul op 111 lllt: llldClUM.OU!-'11.0 p m au syace.
The work clone to 1lish this ch:mo-e is
r
;;w- ,';n(v\d>(v\ ,-J3y (4 R41
where <l.>C x l is the notential due to the char!Te densitv n( x l alreadv nresent. Since
.,_,
'1"1 ...
''
~1.
~"
"·,, • '1- • • •J- _,,
·r
'1 ·• s::n.
0 ...., '0~\ / ' on

up \U>J j \ ~-"·' J
Then the energy change oW can be cast into the form:
ow = J E . oD d x 3
(4.86)
where we have used E = -V<P and have assumed that p(x) was a localized charge
distribution. The total electrostatic energy can now be written down formally,
at least by allowing D to be brought from an initial value D = 0 to its final
valueD:
If the medium is linear, then

W = rd x 3
r E · oD (4.87)
E · oD = ~o(E · D) (4.88)
and the total electrostatic energy is
W-~
- 1 ( E·Ddx
3
(4.89)
~
uJI> 13M lC>Ull <.;all uc u: HlLU V'>.O.J) uy UMH)!, n "'±' allu " • u
p, or by going back to (4.84) and assuming that p and <Pare connected linearly.
~L cL "1\ • ~··~· • H ' ··.L '- '~ • • ,.
~HUO H~ o~~ ""'" \ - .U.'J <J VUHU H<UL• r <y un<y <j <HC UCHUnu• <J H OCW.
Otherwise the energy of a final configuration must be calculated from (4.87) and
might conceivably depend on the past history of the system (hysteresis effects).
A problem of considerable interest is the change in energy when a dielectric
object with a linear response is placed in an electric field whose sources are fixed .
. ..
-n-- llldl lllC llt:lU -"'0 UUC LV a ICt:llalll Ul 1.011<11 )!,C>
p0(x) exists in a medium of electric susceptibility Eo, which may be a function of

"" /£. cL " cL ''-"'" £LL _\ ~L """
r \'V' lH~ 'HVOH~HL ~o •o HVL lH~
0 n<UHHJ VL LH~ C<1'-UU<HJ• ~H'-
electrostatic energy is
Wn = !L { En • Dn d
J
3
x
where D 0 - E0J<; 0. Then With the sources nxed m position a dielectnc obJect ot
volume V1 is introduced into the field, changing the field from E 0 to E. The
presence or me oojeci can oe aescnoea oy a suscepiioiJiiy E~XJ, wnicn nas me
v~liiP ~- · 'riA II. <~nrl ~- ._,_ II. Tn """irl m· ,,.tir,.] ilifnrultiPo \XJP r"n
imagine E(x) to be a smoothly varying function of position that falls rapidly but
continuouslv from "· to Fn M the ecl!>e of the volnme V. The ener(tv now has the
value
r
W1 = ~ / E ·D d x
3
n. _., 'T'l. .- re h
"'"~
.1.
~ '" '6J
llf
1 ( m n. v n. \ ~'3
" 2J ,~
~ ~o ~o, ~
(4.90)
1 I 1 I
2
J ~~ · Uo u · ~o) a x +
2
J ~~ + ~o) • ~u Uo) a-x
Sect. 4.7 Electrostatic Energy in Dielectric Media 167
The second integral can be shown to vanish by the following argument. Since
V X (E + E 0 ) = 0, we can write
E + E 0 = -Vet>
Then the second integral becomes:
I = - ~ I V<P • (D - D 0 ) d 3 x
Integration by parts transforms this into
I = 2
1 r<PV • (D - D 0 ) d-3 x = 0
since V · (D - D 0 ) = 0 because the source charge density p0 (x) is assumed

unaltered by the insertion of the dielectric object. Consequently the energy
change is
W = l1 I (E • D0 - D · E 0 ) d-3 x (4.91)
The integration appears to be over all space, but is actually only over the volume
V1 of the object, since, outside V1, D = E 0E. Therefore we can write
W - --
1
2 v,
J ( E1 - E0 )E • E 0 d 3 x (4.92)
If the medium surrounding the dielectric body is free space, then using the def-
mttJon of polanzat10n P, (4.92) can then be expressed m the form:
1
w 2
{ P E 0 ti 3 x
Jv,
(4.93)
where P is the polarization of the dielectric. This shows that the energy density
of a dielectric placed in a field E 0 whose sources are fixed is given by
(4.94)
This result is analogous to the dipole term in the energy ( 4.24) of a charge dis-
tribution in an external field. The factor~ is due to the fact that (4.94) represents
the energy density of a polarizable dielectric in an external field, rather than a
permanent dipole. It is the same factor ~ that appears in (4.88).
Equations (4.92) and (4.93) show that a dielectric body will tend to move
toward regions of increasing field E 0 provided E1 > E 0 • To calculate the force
acting we can imagine a small generalized displacement of the body of Then
there will be a change in the energy 8W. Since the charges are held fixed, there
is no external source of energy and the change in field energy can be interpreted
as a change in the potential energy of the body. This means that there is a force
actmg on the body:
(4.95)
where the subscript Q has been placed on the partial derivative to indicate that
the sources of the field are kept fixed.
168 Chapter 4 Multipoles, Electrostatics of Macroscopic Media, Dielectrics-Sf
In practical situations involving the motion of dielectrics the electric fields

are often produced by a configuratiOn of electrodes held at fzxed potentials by
connection to an external source such as a battery. To maintain the potentials
constant as the distribution of dielectric varies, charge will flow to or from the
battery to the electrodes. This means that energy is being supplied from the
external source, and it is of interest to compare the energy supplied in that way
with the energy change found above for fixed sources of the field. We will treat
only linear media so that (4.83) is valid. It is sufficient to consider small changes
in an existing configuration. From (4.83) it is evident that the change in energy
accompanying the changes 8p(x) and 8<1>(x) in charge density and potential is
8W = ~ J(p 8<1> + 8p) tPx (4.96)
Comparison with (4.84) shows that, if the dielectric properties are not changed,
the two terms in (4.96) are equal. If, however, the dielectric proper ties are dltei ed,
E(X) ~ E(X) + OE(X) (4.97)
the contributions in (4.96) are not necessarily the same. In fact, we have just
calculated the change in energy brought about by introducing a dielectric body
mto an electnc field whose sources were fixed (op - 0). Equal contnbubons m
(4.96) would imply oW = 0, but ( 4.91) or ( 4.92) are not zero in general. The
reason for this difference lies in the existence of the polarization charge. The
change in dielectric properties implied by (4.97) can be thought of as a change
in the polarization-charge density. If then (4.96) is interpreted as an integral over
both free and polarization-charge densities (i.e., a microscopic equation), the two
contributions are always equal. However, it is often convenient to deal with mac-
roscopic quantities. Then the equality holds only if the dielectric properties are
unchanged.
The process of altering the dielectric properties in some way (by moving the
dielectric bodies, by changing their susceptibilities, etc.) in the presence of elec-
trodes at fixed potentials can be viewed as taking pldce in two steps. In the first
step the electrodes are disconnected from the batteries and the charges on them
held fixed (Sp 0). 'Nith the change (4.97) in dielectric properties, the energy
c an e IS
1 r
oW1 - J p o 1 is the change in potential produced. This can be shown to yield
the result (4.92). In the second step the batteries are connected again to the
electrodes to restore their potentials to the original values. There will be a flow
of charge op2 from the batteries accompanying the change in potential*
8<1> 2 = -8<1> 1 . Therefore the energy change in the second step is
(4.99)
*Note that it is neeessary merely to know that il4> 2 §4> 1 on the eleetrodes, sinee that is the only
place where ftee charge resides.
Ch. 4 Problems 169
since the two contributions are equal. In the second step we find the external
sources changmg the energy m the opposite sense and by tw1ce the amount of
the initial step. Consequently the net change is
aw (4 .100)
Symbolically
8Wv = -8W0 (4.101)
where the subscript denotes the quantity held fixed. If a dielectric with £C)E0 > 1
moves into a region of greater field strength, the energy increases instead of
decreases. For a generalized displacement dg the mechanical force acting is now
F< = +Ca~t (4.102)

The derivation of the macroscopic equations of electrostatics bv averaging over ag-
Mason and Weaver, Chapter I, Part III

Van Vleck, Chapter 1
Rosenfeld also treats the classical electron theory of dielectrics. Van Vleck's book is
devoted to electric and magnetic susceptibilities. Specific works on electric polarization
phenomena are those of
Bottcher
De bye
Frohlich
Boundary-value problems with dielectrics are discussed in all the references on elec-
trostatics in Chapters 2 and 3.
Our treatment of forces and energy with dielectric media is brief More extensive
diset1ssions, inel uding forces on liquid and solid dielectrics, the electric stress tensor, elec
trostnctwn, and thermodynamic effects, may be found in
Abraham and Becker, Band 1, Chapter V
Durand, Chapters VI and VII
Landau and Lifshitz, Electrodynamics of Continuous Media
Maxwell, Vol. 1, Chapter V
Panofsky and Phillips, Chapter 6
Stratton, Chapter II
Problems
4.1 Calculate the multipole moments q1m of the charge distributions shown as parts a
and b. Try to obtain results for the nonvanishing moments valid for all/, but in each
case find the first two sets of non vanishing rnornents at the vel) least.
1 "71\ ~'-
-·
--r
A
-r
-. '
~·
. -~
,.
'
~·-· ~·
-t-q'
-q a ~ ~ q
,y
-2q r~ ·V
A" ?Tq
/ a
%/
(a) (b)
Problem 4.1
(c) For the charge distribution of the second set b write down the multipole ex-
pansion ror me porennat. J<..eepmg onq 1ne lowest-ureter term m the expan-
sion, plot the potential in the x-y plane as a function of distance from the
. . f, rl' ot .~
·o
(di Calculate directlv from Coulomb's bm thP Pvo~t iol fm h in thP

x-y plane. Plot it as a function of distance and compare with the result found
in part c.
Divide out the asymptotic form in parts c and d to see the behavior at large distances
more CJeany.
4.2 A point dipole with dipole moment p is located at the point x0 . From the properties
or me aerivarive or a uirac aeua runcrion, snowtnarTor carcliTIITion ot the potential
'P or the energy of a dipole in an external field the dinole can be described bv an
effective charge density
"C"\"'1
1-\
r
'r'IN.
,. \
"U)
·~
-~
~ . H
'" ov
"-
" ' <Hv
_,..
•v•v
..
\ " ' • ' / Ul lllC r- 1> -,. uy tuc
'
(21 + 1) multipole moments q 1m· On the other hand, the Cartesian multi pole
moments
Q~~Y = J p(x)x"y~zY d 3x
.... , , ,
n
r -o -o r ', a• v
~, ·
d_ n
,..
(I + 1)(1 + 2)/2 in number. Thus, for I > 1 there are more Cartesian multipole
moments than seem necessary to describe the term in the potential whose radial
dependence is r- 1- 1 •
. .
C>llU lll<ll W 1111C lllC 'Jim . UHUCl l ;-as-rrremrcr01e T . Ien-
sors of rank I, the Cartesian multipole moments correspond to reducible spherical

nf ron b· I I ') I 4 I . ..L I . 0 nr 1 fnr I P"Pn nr nrlrl
respectively. Check that the number of different tensorial components adds up to
Lne LOLat numuer 01 '-'anesian tensors. vv ny are oniy me q ''" neeaea m me expanswn
(4.1)?
'I.'I \OJ rrove we roaowmg meorem: r or an arnnrary charge Cl!stnnutwn p(x) the
values of the (21 + 1) moments of the fi"t 1ishino m11ltinnlP. orP. inrlP-
pendent of the origin of the coordinate axes, but the values of all higher mul-
tipole moments do in general depend on the choice of origin. (The different
moments q 1m for fixed I depend, of course, on the orientation of the axes.)
\DJ A uistriomion nas muaip01e moments q, p, !lij• ... w!lh respect to one
-o :
set of coordinate axes, and moments q', p', Qij, ... with respect to another
Ch. 4 Problems 171
set whose axes are parallel to the first, but whose origin is located at the point
R = (X, Y, Z) relative to the first. Determine explicitly the connections be-
tween the monopole dipole and quadmpole moments in the two coordinate
(c) If q* 0, can R be found so that p' - 0? If q * 0, p * 0, or at least p * 0, can

R be found so that Qi; = 0?
'
4.5 A localized charge density p(x, y, z) is placed in an external electrostatic field de-
scribed by a potential cf:l("l(x, y, z). The external potential varies slowly in space
over the region where the charge density is different from zero.
(a) From first principles calculate the total .force acting on the charge distribution
as an expansion in multipole moments times derivatives of the electric field,
up to and including the quadrupole moments. Show that the force is
F = qE( l(O) + {V[p · E(

0 0
1
l(x)]} 0 + { V[ - 2; Qjk -a£(
6 j,k
--
1 l
axk
0
(x) J} o
+ ···
Compare this to the expansion (4.24) of the energy W. Note that (4.24) is a
(b) Repeat the calculation of part a for the total torque. For SimpliCity, evaluate
only one Cartesian component of the torque, say N,. Show that this campo-
4.6 A nucleus with quadrupole moment Q finds itself in a cylindrically symmetric elec-
tric field with a gradient (aE,/az)o along the z axis at the position of the nucleus.
(a) Show that the energy of quadrupole interaction is
w = -4e Q (aE,)
az o
(b) If it is known that Q = 2 X w-zs m2 and that W/h is 10 MHz, where
il i~ Planck'~ comtant, calculate (il E,liJz) 0 in units of el41r£ea~, where
a0 4rr~Ji?lme2 0.529 >< 10- 10 m is the Bohr radius in hydrogen.
(c) Nuclear charge distributions can be approximated by a constant charge density
throughout a spheroidal volume of semimajor axis a and semiminor axis b
Calculate the quadmpole moment of such a nucleus, assuming that the total
charge is Ze. Given that Eu 153 (Z = 63) has a quadrupole moment Q =
2.5 X I() 28 m 2 and a mean radius
R = (a + b)/2 = 7 X 10- 15 m
determine the fractional difference in radius (a - b)! R.
4.7 A localized distribution of charge has a charge density
(a) Make a multipole expansion of the potential due to this charge density and
determine all the nonvanishing mnltipole moments Write down the potential
at large distances as a finite expansion in Legendre polynomials.
172 Chapter 4 Multipoles, Electrostatics of Macroscopic Media, Dielectrics-51
(b) Determine the potential explicitly at any point in space, and show that near
the origin, correct to r2 inclusive,
determine the magnitude of the interaction energy, assuming that the unit of
charge in p(r) above is the electronic charge and the unit of length is the
X 10
swer as a frequency by dividing by Planck's constant h.

The charge density in this problem is that for the m = :±: 1 states of the
m is of the same order
found in molecules.
4.8 A very long, right circular, cylindrical shell of dielectric constant •/•0 and inner and
has a dielectric constant of unity.
(b) Sketch the lines of force for a typical case of b = 2a.

your
cylinder in a uniform field, and a cylindrical cavity in a uniform dielectric.
4.9 A is located in free a distance d from the center of a dielectric
sphere of radius a (a < d) and dielectric constant EIE0 .
(a) Find the potential at all points in space as an expansion in spherical harmonics.
the sphere.
4.10 Two concentric conducting spheres of inner and outer radii a and b, respectively,
(b) Calculate the surface-charge distribution on the inner sphere.

Ch. 4 Problems 173
4.11 The following data on the variation of dielectric constant with pressure are taken
from the Smithsonian Physical Tables, 9th ed., p. 424:
Air at 292 K
Pressure (atm) ~€o
20 1.0108 Relative density of air as a function of

40 1.0218 ptessme is given in AlP Ilandbook,
60 1.0333 [3rd ed., McGraw-Hill, New l' ork
80 1.0439 (1972), p. 4-165].
100 1.0548
Pentane (C 5 Hu) at 303 K
Pressure (atm) Density (glcm3 )
1 1.82
103 1.96
4 X 103 0.796 2.12
Test the Clausius-Mossotti relation between dielectric constants and density

for air and pentane in the ranges tabulated. Does it hold exactly? Approximately?
If approximately, discuss fractional variations in density and ( Ei£0 1). For pentane,
compare the Clausius-Mossotti relation to the cruder relation, ( E!£0 - 1) cc density.
4.12 Water vapor is a polat gas whose dielectric constant exhibits an appreciable tern-
perature dependence. The following table gives experimental data on this effect.
Assuming that water vapor obeys the ideal gas law, calculate the molecular polar-
izability as a function of inverse temperature and plot it. From the slope of the
curve, deduce a value for the permanent dipole moment of the H 2 0 molecule (ex-
press the dipole moment in coulomb-meters).
T(K) Pressure (em Hg) (e/e0 - 1) X 105
393 56.49 400.2

423 60.93 371.7
453 65.34 348.8
483 69.75 328.7
4.13 Two long, coaxial, cylindrical conducting surfaces of radii a and b are lowered
vertically into a liquid dielectric. If the liquid rises an average height h between the
electrodes when a potential difference V is established between them, show that
the susceptibility of the liquid is
where p is the density of the liquid, g is the acceleration due to gravity, and the
susceptibility of air is neglected.
Quasi-Static Fields
In the preceding chapters we examined various aspects of electrostatics (i.e., the

fields and interactions of stationary charges and boundaries). We now turn to
s ea y-s a e magne 1c p enomena, ara ay s aw o m uctton, an quas1-stat1c
fields. From a historical point of view, magnetic phenomena have been known
in ancient times; the mariner's compass is a very old invention; Gilbert's re-
electrostatics the basic laws of rna netic fields did not follow strai htforwardl
from man's earliest contact with magnetic materials. The reasons are several, but
the all stem from the radical difference between rna netostatics and electro-
statics: there are no free magnetic charges (even though the idea of a magnetic
charge density may be a useful mathematical construct in some circumstances).
This means that magnetic phenomena are quite different from electric phenom-
ena and that for a long time no connection was established between them. The
basic entity in magnetic studies was what we now know as a magnetic dipole. In
direction. That direction is by definition the direction of the magnetic-flux den-

• 0
not perturb the existing field. The magnitude of the flux density can be defined
0 0 0
N=r..tXB (5.1)
where 1..1. is the magnetic moment of the di ole, defined in some suitable set of
units.
Already, in the definition of the magnetic-flux density B (sometimes called
the magnetic induction), we have a more complicated situation than for the elec-
tric field. Further quantitative elucidation of magnetic phenomena did not occur
untl t e connectiOn etween currents an magnetic
over the cross-sectional area and speak of a current of so many amperes flowing
alon the wire.
Sect. 5.2 Biot and Savart Law 175
Conservation of charge demands that the charge density at any point in space
be related to the current density m that neighborhood by a contmmty equatiOn:
ap
-+V·J=O (5.2)
(}l
This expresses the physical fact that a decrease in charge inside a small volume
with time must correspond to a flow of charge out through the surface of the
small volume, since the total amount of charge must be conserved. Steady-state
m~crnf'tir nh >~rf' C'h>~ · -' hv no C'h~n<>f' in thf' nf't C'h>~r<>f' .-IPn<itv
anywhere ~in space. Consequently in magnetostatics
v .... u ~-1 . .J)
We now proceed to discuss the experimental connection between current and

,,· t'l rl, '
'J
rl <. ,,,,J-.);,J, th~ 1-..ooi~ lnnw ~f'~n~~.
-., ;~.,
5.2 Biot and Savart Law

T.
LH
, 0 , 0 f'o
LUL/ ~~~O<~U
_, - ,_
VVO~~ '~U "· " "-"
.• LL
<HUC WH~O ~W'J'"/S
..
~·~~CH~ ~UH~•ao jJ'VUU~vU
tions of permanent magnetic dipoles placed in their neighborhood. Thus the

·" • _f'l, rl. i<· u· ,. ' rl "· -~· (1 Q'){)\ h~e· n~rl
Ampere (1820-1825), in much more elaborate and thorough experiments, estab-

li<hP.-1 thP h~<ir · >Pnt,.] bm< rPbtincr thP m· · in.-lnrtion R tn thf' rnr.
,.,-r,,,' ee '-
0
_ , d _ A JLL .L
''-'"'o auu UOOH.-U W~ •aw " ' HH~v V~<WvvH VHv '-UHvH< QUU OUVW'-'. ~H -o
not in the form in which Ampere deduced it, the basic relation is the following.
Tf' -11 ' .1. .+ .< 1. ''- f. . . ' <\, -'' . .f' t'l. ,\ ,f'
\l' 'b 'I
filamentary wire that carries a current I and x is the coordinate vector from the
PlPnwnt nf ]pncrth <A ~n Al- nnint P "' chnmn in 'Rin <; 1 thPn thP p]p_
·o· r ' ·o
mental flux density dB at the point P is given in magnitude and direction by
(dl x x)
UD Kl 3 ~J,'-t)
lxl
ll SUOUlU Uto LllaL ~-'·'-+)"an LIIVtol,to '4Uiilt: laW, JUM a , " LU, laW
of electrostatics. However, the vector character is very different.
A word of caution about (5.4). There is a temptation to think of (5.4) as the
magnetic equivalent of the electnc field (1.3) of a pomt charge and to identify
I dl as the analog of q. Strictly speaking this is incorrect. Equation (5.4) has
meanmg only as one element ot a sum over a contmuous set, tne sum representing
the magnetic induction of a current loop or circuit. Obviously the continuity
. .
equallvn ~J •.J J ~~ nm lUI Lilt: :m 1 u1 Manumg -Lilt:
d~
------ T
dB
~"'-.
/ ~
/ "'-. ' .
/ J<Igure :..J. tc!emema1 magnenc moucnon an
'-.,..p due to current element I dl.
176 Chapter 5 Maguetostatics, Faraday's Law, Quasi Static Fields SI
current comes from nowhere and disappears after traversing the length dl! One
apparent way out of this difficulty is to realize that current is actually charge in
motion and to replace I dl by q• where q is the chmge and • its velocity. The
flux density for such a charge in motion would be
V X X
x·
in close correspondence with (5.4). But this expression is time dependent and
furthermore is valid only for charges whose velocities are small compared to that
of light and whose accelerations can be neglected. Since we are considering
steady-state magnetic fields in this chapter, we stick with (5.4) and integrate over
circuits to obtain physical results.*
In (5.4) and (5.5) the constant k depends in magnitude and dimension on the
system of umts used. as dtscussed m detail in the Appendix. In Gaussian units,
in which current is measured in esu and magnetic induction in emu, the constant
is empirically found to be k = 1/c, where c is the speed of light in vacuo. The
presence of the speed of light in the equations of magnetostatics is an initial
puzzlement resolved within special relativity where vic has a natural appearance.
In Gaussian units, E and B have the same dimensions: charge divided by length
squared or force per unit charge.
In SI units, k = J.Loi47T' = 10- 7 newton per square ampere (N/A 2 ) or henry
per meter (H/m). Here B has the dimensions of newtons per ampere meter
(N/A · m) while E has dtmenswns of N/C. B times a speed has the same dimen-
sions as E. Since c is the natural speed in electromagnetism, it is no surprise that
in SI units E and cB form the field-strength tensor p•v in a relativistic description
lion to determine the magnetic-flux density due to various configurations of

current -carrying wires. For example, the magnetic induction B of the long straight
wire shown in Fig. 5.2 carrying a current I can be seen to be directed along the
normal to the plane containing the wire and the observation point, so that the
lines of magnetic induction are concentric circles around the wire. The magnitude
of B is given by
(5.6)
where R is the distance ftom the observation point to the wire. This is the ex-
penmental result first found by Biot and Savart and is known as the Biot-Savart
law. Note that the magnitude of the induction B varies with R in the same way
as the electric field due to a long line charge of uniform linear-charge density.
*There IS an apparent mconsistency here. Currents are, after all, charges m motion. How can (5.4),
integrated, yield exact results yet (5.5) be only approximate? The answer is that (5.5) applies to only
one charge. If a system of many charges moves in such a way that as the unit of charge goes to zero
and the number of charges goes to infinity it produces a steady current flow, then the sum of the
exact relati~istie fields, including accele;ation effects, gives a magnetostatic field equal to the field
obtained by integrating (:5.4) over the Circuit. I his rather subtle result Is discussed tor some specml
situatiOns m Problems 14.23 and 14.24.
Sect. 5.2 Biot and Savart Law 177
dl~
8 X
I
Fil!nre 5.2
This analogy shows that in some circnmstances there may be a correspondence

between electrostatic and magnetostatic problems, even though the vector char-
acter of the fields ts dtfferent. We see more of that mlater secttons.
Ampere's experiments did not deal directly with the determination of the
f€)iation b€)tW€)€)n curr€)nts and magn€)tic induction, but W€)re conc€)rn€)d rath€)r
with the force that one current-canying wire experiences in the presence of an-
other. Since we have already introduced the idea that a current element produces
a magnetic induction, we phrase the force law as the force experienced by a
current element / 1 dl 1 in the presence of a magnetic induction B. The elemental
force is
(5.7)
If the external field B is due to a closed current loop #2 with current 12 , then the
total force which a closed current loop #1 with current / 1 experiences is [from
(5.4) and (5.7)]:
(5.8)
The line integrals are taken around the two loops; x 12 is the vector distance from
line element dl 2 to dl 1 , as shown in Fig. 5.3. This is the mathematical statement
of Ampere's observations about forces between current-carrying loops. By ma-
nipulating the integrand it can be put in a form that is symmetric in dl 1 and dl 2
and that explicitly satisfies Newton's third law. Thus
dl 1 X (dl2 X x!2) _ -( . 12
) ~. (dl1 • x 12 ) (5.9)
3 - dl 1 dl 2 IX 13 + dlz IX 13
IX 12 1 12 12
178 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-Sf
ThP tPrm in· -' .£. ,-1;££ ',1 ,_ +1-.~. .-/1 £".
J ~
r ~
~
.,. -
quently it gives no contribution to the integral (5.8), provided the paths are closed
or extend to infinity. Then Ampere's law of force between current loops becomes
II. r r r,.~. • ,.~. •~
Fl2
41T M2 ff lxnl3 (5.10)
showing symmetry in the integration, apart from the necessary vectorial depen-
' ~n ~
"lL'
Each of two long, parallel, straight wires a distanced apart, carrying currents
11 and 12 , experiences a force per unit length directed perpendicularly toward the
other wire and of magnitude,
dF 1,12
f.-to
-~
~J.ll)
dl 21T d
.,_
'T'J, ·~ ·~ . {,
v -~
,1 • o\ ;<+>
-1 "' m ·~
I
\ V1'JJVOH'-')
. -'
directions. The forces that exist between current-carrying wires can be used to
define magnetic-flux density in a way that is independent of permanent magnetic
dipoles.* We will see later that the torque expression (5.1) and the force result
(5.7) are intimately related.
T£ .-1 ' T/ __ , ' _, ' ~- ' A
H J ~\A) 10 Hl Uo ' - A " ' - ' HUl "llUJI. U~AJ, lllt;O
elementary force law implies that the total force on the current distribution is
F = f J(x) X B(x) d x
J
3 {:'i 1?1
Similarly the total torque is

(
1"1 J x A~ .. A H) --
a·x ~:l.U)
'T'J, _, ~~· .-1 1 ... _,.

'"
~
'1 ;]1 ' '-' H UlOU
• 0 ·-rr Hl '-''-'"'"
bon 5.7.
--
;),.J
OPP
uwer • • p
nq ....... ions OJ JJt.lagnewscaucs ana Amperes Law

The basic law (5.4) for the magnetic induction can be written down in general
form for a current density J(x):
, r v ~·
B(x)
41T
.J J(x') xI x-x '13 d"x' (5.14)
!his expressiOn tor H(x) IS the magnetic analog of electnc field in terms of the
~h~rAP rl~ncitn•
1 x- x' _../3,'
l?.fv\ ( nfv '\ f<: 1 <:\
\' . "I
41TE0 J ' ' 'lx- x'l 3
Just as this result for E was not as convenient in some situations as differential
equations, so (5.14) is not the most useful form for magnetostatics, even though
It contams m pnnc1p1e a descnptlon of all the phenomena.
*In fact, (5.11) is the basis of the internationally accepted standard of current. See the Appendix.
Sect. 5.3 Differential Equations of Magnetostatics and Ampere's Law 179
'' ' ' ' I~ 1 A\ <L
' V UUldlll LIIC UlllClClllldl Cl.[l <v \• , , Tf 0 n~ ~o~ <H~ •
just above (1.15) to transform (5.14) into the form:
B(x) = A~ V X
!Lo r I~J(x')~'I d 3 ' X (5.16)
-,-
{<; 1 fi) it follow' · 'i<~tP.lv th~t the diver"ence of B vanishes:

V •B = 0 (5.17)
This is the first equation of magnetostat1cs and corresponds to v x ~ - u m
p), 1tir' Rv :m~l.-.av 'uith ,], ·•· · "'P now calculate the curl of B:
vXB = !Lo v Xv X r I J(x')·l d 3 x' (5.18)

.,. I"' ~ I
UT:<L ot. ' ' '1'7 /'1'7 .. \ .... /'1'7 .. \ t"l2 .. L rh;t, •~• fio]rl
'... 'J ' ., ' 'I 'J
A, expression (5.18) can be transformed into

r I ' r /1\
(5.19)
1 \ I 1 \
V(lx - x'l V' lx- x'll
ana
I
V2 ' = -47T<S(x - x')

lx- x'l
the integrals in (5.19) can be written:
1
V X B = - !Lo V
47T
J J(x') • V' ( I
X- X
, )
3
d x' + JLoJ(x) (5.20)
1
Integration by parts yields
V X B = JLoJ + !Lo V
47T
JV'lx-x'l
• J(x') d x' 3
(5.21)
But for steady-state magnetic phenomena V • J = 0, so that we obtain

V X B = JLoJ (5.22)
This is the second equation of magnetostatics, corresponding to V · E = p/E 0 in
electrostatics.
In electrostatics uauss s Jaw (1.11) 1s tne mtegrm wrm 01 Llle equaLion
V · E = olEn. The inte2:ral eauivalent of (5.22) is called Ampere's law. It is ob-
tained by applying Stokes's theorem to the integral of the normal component of
(5.22) over an open surfaceS bounded by a closed curve C, as shown in Fig. 5.4.
Thus
I VxB·nda=JLol J·nda (5.23)

...--- --------- -----... "~

f ~~\\\\\\)c
/':.
" \ \ \ \\ \ \.Y
\ '-...\_ \
dl Figure 5.4
is transformed into
1 {
Jc D • UJ p., 0 Js" · n uu ~ J.L'-t}
Since the surface integral of the current density is the total current I passing
thrmwh the c ( A ' •, lllw c:ln he in the form·
~ n rll
ru
1 (<:_ ')<:_\
, I
lr
Just as Gauss's law can be used for calculation of the electric field in highly
symmetric situations so Ampere's law can be employed m analogous
circumstances.
5.4 Vector Potential

Hlt: UdM\.- UIHC:l C:lllldl ld W> VI >Ldlll.-> dl c;
V X B = /LoJ
(5.26)
V • D u
The problem is how to solve them. If the current density is zero in the region of
'" ·~· ~"''
'1'7 D
~
C\
~'
+L
'"~ ~~P'' '" ...
-~ +L
~ '~~·~· "'~
. " D
as tne grament or a magnettc scatar potenttat, n v<vM. 1nen p.LOJ reauces

to the Laplace equation for <PM, and all our techniques for handling electrostatic
problems can be brought to bear. A large number of problems fall into this class,
but we will defer discussion of them until later in the chapter. The reason is that
the boundary conditions are different from those encountered in electrostatics,
>1nrl the nrnhlem' ""'"llv · m"' · merli" with mll<metic
.L
different from free space with charges and currents.

A !!eneral method of attack is to exuloit the second eauation in (5.26). If
... ... ~
C\
~ ~·~<JHH~L~
0
...
~ <HUOC
L
V~
•'-
CH~ ~UU
.. 1 -~
VL OV<H~
<: ·'-'
u~•u
& I .
~\~J•
.II ~ +L
'"~
vector potential,
R{xl = V X A{x) {'i.271
We have, in fact, already written Bin this form (5.16). Evidently, from (5.16),
the general form of A is
A(x) = /Lo
47T
Jlx-
J(x') d x' + V'[t(x)
x'l
3
(5.28)
Sect. 5.5 Vector Potential and Ma2netic Induction for a Circular Current Loop 181
"T'l- rlrl .rl
b' "' -~ ' .
magnetic induction B, the vector potential can be freely transformed accord-
'J
,1. £. •Tf n 1-
.. -· '· b'
;n·o tr.
<T
A --c> A + V'l_r (5.29)

This transformation is called a gauge transformation. Such transformations on A
:1re nossihle hec:111se ('i ?7) '"' on lv the c11rl of A The £. ~ of
transformations allows us to make V · A have any convenient functional form we
wish.
If (5.27) is substituted into the first equation in (5.26), we find
v ;<. ~ v ;<. 1-'l.) fLo"

or (5.30)
V(V ·A)- V2 A = 1-LoJ
If we now exploit the freedom implied by (5.29), we can make the convenient
,]- • z ...,.
.~ (\ "T'J- nl- ,J ~ , , +1-
b' ' b ' ' . .,.
potential satisfies the Poisson equation,
VLA = - J.LoJ (5.31)
v. ' ri' ... ·' .
. nl. . ...•
~. ,,, .. ~. ~ .
bounded space is (5.28) with']! = constant:
A(x) = fLo
4'lT
J J(x') dV
lx- x'l
(5.32)
The condition ']! = constant can be understood as follows. Our choice of gauge,
'[7 A () ~- +. '172\Tf () • +ha ~ro+ +ar"' ~~ f"'- ')Q\ hoo ~nr~ ,.
' ' \' • 'I 'b'
2
because of V' · J = 0. If V ']t = 0 holds in all space,']! must be at most a constant
nrnv;r1Pr1 thPTP QTP nr. <AHTC'P< Qt ;nfin;tv
'
..
5.5 Vector Potential and Magnetic Induction
for a Circular Current Loop
As an illustration of the calculation of magnetic fields from given current distri-
butions, we consider the problem of a circular loop of radius a, lying in the x-y
plane, centered at the origin, and carrying a current/, as shown in Fig. 5.5. The
current density J has only a component in the ¢ direction,
a)
J,r = I sin 1:1' 8( cos 1:1') b(r (5.33)
u
The delta functions restrict current flow to a ring of radius a. The vectorial current
rl. ·;., .I r:1n hP
J
;++a•
T r .J.. '.
'I'
r
'<P
.J.. ' .
.,.. J ,.. .,
f<: '2A\
Since the geometry is cylindrically symmetric, we may choose the observation

"'
noint in the Y-7 nl~nP ( ,/, rn for nll of r~lr11btion ''inrP thP ~7im11th~l
*The choice is called the Coulomb gauge, for a reason that will become apparent only in Section 6.3.
~~-
Jj)~ cnapter :J lVtagnetostancs, l'araday's Law, '-!nasi-Matic Fields SI
z
p
11\
1\.
\
~ ~ I
v I
I
r " I
'-... A /
I
I
X Figure 5.5
integration in (5.32) is symmetric about <P 1 = 0, the x component of the current

does not contribute. This leaves only they component, which is Aq,. Thus
A ( ) _ f.Lol
q, r, 8 - 4 ~a
Jr 12 d 1
r d
1
n~ sin 8 cos <P .S(cos 8 )B(r - a)
1
Ix - 1
1
I
1
(5.35)
X
where lx- X I = [r 2 + r 12 - 2rr 1(cos 8 cos 8 1 +sin 8 sin 8 1 cos <P 1)] 112 •
1
We first consider the straightforward evaluation of (5.35). Integration over

the delta functions leaves the result
A (r 8) = f.Lo l a r~
COS <PI dcp I
2 2 (5.36)
"' ' 4~ o (a + r - 2ar sin 8 cos <P 1) 112
This integral can be expressed in terms of the complete elliptic integrals K and E:
Aq, (r, 8) = f.Lo 4Ia

4~ Va 2 + r 2 + 2ar sin 8
[ (2 - k )K(~)
2
k
- 2E(k) l (5.37)
where the argument k of the elliptic integrals is defined through
f/2
'
"TIM om v
n2 + ,2 + 7. o;n Cl
.....
"T''
··r
·+
'" ·~~~··~ ..,
L a r .. n
D
r '
"''
• {) ;l{) \ o•u
1 iJ
H"- - - - (rA.1. l ~.J.:JO)
r ar
n .-.
v
~"'
can a1so oe expresseu in terms 01 empnc Imegrals. l:Sut the results are not oartJc-
nl,rhr il · '"tinu ln<Pfnl h
rut u ___./ r, u '--'-- r, Ul u <--<-- L, au
f.,r
.. l Vt:
· ;;\
··;
"1Jr1:l.50) m powers
of aLrL smL8/I aL + rL)L leads to the followinu ~ 1~tP f.-.r thP
• ,1 •
,
r
<P '
l 2 • 8 [
A (r 8)= f.Loarsm
4 (az + r2)312
1 +
15 ar . 28
2 2 sm
S(az + rz)z
+ ...
l (5.39)
Sect. 5.5 Vector Potential and Magnetic Induction for a Circular Current Loop 183
To the same accuracy, the corresponding field components are
f.}
thola2 cos 8 1 -'-
l5a 2 r2 sin 2 8 -'- ...
r
2( a" + r 2 )'12 4(a 2 + r2 )" - (5.40)
thola2 sin 8 l5a 2 r2 sin 2 8( 4a 2 - 3r 2 )
Be = - 2a 2 - r2 + or 2 + ···
Af 2
'\'
2\512
I
.,. 2\2
I
These can easily be specialized to the three regions, near the axis (8 << 1), near
the center of the loop (r << a), and far from the loop (r >>a).
Of particular interest are the fields far from the loop:
'
tho L ) cos 8
B = - I 1ra 2 -----;;-
L7r , f<:A1\
.,
\''
tho
sin 8 2
Be = - (l1ra) --3
471" r
companson w1th the electrostatlc d1po1e tields (4.1LJ snows tnat me magnenc
fields far away from a circular current loop are dipole in character. By analogy
with electrostatics we define the magnetic dipole moment of the loop to be
m = 1rla2 (5.42)
We see in the next section that this is a special case of a general result-localized
current distributions give dipole fields at large distances; the magnetic moment
of a plane current loop is the product of the area of the loop times the current.
Although we have obtained a complete solution to the problem in terms of
empnc mtegraJs, we now umstrate me use or a spnerica1 narmonic expansion Lu
point out similarities and differences between the magnetostatic and electrostatic
problems. Thus we return to (5.35) and substitute the spherical expansion (3.70)
r I r I~ 1
'w I"' "'I
A
q,
= thul
n
Re 2.; Y,m( 8, O)
')/ -'- 1
rr' 2 dr' d!1' .S(cos 8')8(r'- a)e'"'' r~.
r/+1
Y* (8' ¢')
lm '
. -
{<;
\- .,
.A'l\
The presence of eW means that only m = + 1 will contribute to the sum. Hence
I \ -
A
q, -
- 2
7rtho a
2.: "l ru~t1, UJ
2[ + 1 r >
r<
Y,,l 7r 01 (5.44)
1-1
1+1 2'
uLy
•1-.
1~ a
f'
< \' >/
\ • ••
_
,11,
Ull L
. fl'
\' 'E''7
-\ • r .-1 'rh~ cnn~r~-
'
-'- t~rl nn~n-
'1"
~r
0
I 21 + 1 nlm·
\I a....Jn + n ' '~ v J
'
r
0 for l even
- .I ,_n+lT'I 3\ (5.45)
~· '' \ ---T ~ 27
) L' for l = 2n + 1
47rl(l + 1) r(n + 1)1@
184 Chapter 5 Magnetostatics, Faraday's Law, Qnasi-Static Fields-SI
--nrerr7r1>-cmrue-w:
A
1> -
__ f.Lola
4 n-0
- 2"( n + 1)1. 2n+2
i(
-1)"(2n - 1)!! r~n+l pl (
2n+l COS e
)
(5.46)
r>
' -~
~u ""..
~)
-~
~L.ft
'--~
~ )~ L.ft
~"
..) ,~- .. ' , / ..)
~ ~
..J " '

~, aau
_, ..
LI1\0 n
n
v
"~ .
in the sum is unity by definition. To evaluate the radial component of B from

(5.38) we need
-=--r,
'
11
1
2 nil:\1
I\~ /J
Til-'-
"\'
1\Df.-\
•I 1\~' I
{<:: A'"7\
\" . 'I
dx
Then we find
r 1 \nf'"l 2n+l
~·u·
T. oo 1 \11
.. <
"
1'l p f ~no il\ (<; AQ\
"'--' 2nn! ,2n+2 m->\ "/ \" "I .
2r n=O r>
The e component of B is similarly
(,., _j_ ,, 1 2n
.I
Be = _ f.Lolii' i (-1)"(2n+ +1)!1)!! 2n +1 a· a
Pin+l(cos e) (5.49)
(~Yn
4 n~o 2n(n
r\
~·
111\:0 -r r
'.
11111. ' ' ' >
r
lVl I
' .. '
U, d11U ll1\0 1VW\01 1111\:0 lVl I
'. r
' U. ~VI I
~
~- U, VlllJ
' .. ll1\0
n = 0 term in the series is important. Then, since P}(cos e) = -sine, (5.48) and
(5.49) reduce to (5.41). For r << a, the leading term is again n = 0. The fields
are then eamvalent to a magnet1c mauctlon uol!La m the z d1rect1on, a result
that can be found by elementary means.
We note a characteristic difference between this problem and a correspond-
ing cylindrically symmetric electrostatic problem. Associated Legendre polyno-
• > " >' T • " •
.. , ~ywcu 'Ya~ a~• 1Cd11 L!C Ll dlCCU LU ~ 111~
the vector character of the current and vector potential, as opposed to the scalar
properties of charge and electrostatic potential.
Another mode of attack on the problem of the planar loop is to employ
0~ o:~~ :~ m,J;~rlT:ool , J n+ . .-1 ,;o {7. '"7{\\ ,+ ' ,;o
"1' --:r \-. ~I ·r
'
IX x we may use me cynnanca1 rorm ~.:l.l'fi\) or ~.:l.l'f~) or war or rro01em
1
3.16b. The application of this technique to the circular loop will be left to the
problems.
5.6 Magnetic Fields of a Localized Current Distribution,

Magnetic Moment
. _,
"'
~
~b
L' , r
'
~Ull\:0111 U1~Ll
.'
tHat
. '
•~
'. '
in a small region of space, "small" being relative to the scale of length of interest
+n +hn n< • A ln+n •o~+ .:.. +h:o
, :~ .~ ,1 ;+l- +l- ,1
~"'. :OJ
LLIC expansion, can oe maae usmg vecwr spnenca1 narmonics.
*This is not the only way. Scalar potentials can be used. See J. B. Bronzan, Am. 1. Phys. 39, 1357
(1971).
Sect. 5.6 Magnetic Fields of a Localized Current Distribution, Magnetic Moment 185
Figure 5.6 Localized current density

J(x') gives rise to a magnetic induction
at the point P with coordinate x.
These are presented in Chapter 9 in connection with multi pole radiation. We

will be content here with only the lowest order of approximation. Assuming
Ixi >> lx' I, we expand the denominator of (5.32) in powers of x' measured
relative to a suitable origin in the localized current distribution, shown schemat-
ically in Fig. 5.6:
(fJ · V'g + gJ · V'f + fgV' • J) d 3 x' = 0 (5.52)
This can be established by an integration by parts on the second term, followed

by expansion of fV' • (gJ). With f = 1 and g = x;, (5.52) with V' · J = 0 estab-
lishes that
I f;(x') d 3 x' = 0
The first term in (5.51), corresponding to the monopole term in the electrostatic
expansion, is therefore absent. With f = x ;, g = x j and V' · J = 0, (5.52) yields
I( x'J.1 + x'.J.)
l J d 3 x' = 0
l
The integral in the second term of (5.51) can therefore be written
x· I x'J.l d 3 x' = L.JJ

"" j
X· I x'.J.
3
j t d x'
-~ 2: xi }
I 3
(x;Ii - xjl;) d x'
186 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-SI
It is customary to define the magnetic moment density or magnetization as
.M(x) = 21 [x X J(x)] (5.53)
and its integral as the magnetic moment m:

1 r
m- 2 J x' X J(x') d 3 x' (5.54)
Then the vector potential from the second term in (5 51) is the magnetic dipole
vector potential,
A(x) = 1-Lo m X x (5.55)

47T lxl3
This is the lowest non vanishing term in the expansion of A for a localized steady-
state current distribution. The magnetic induction B outside the localized source
can be calculated directly by evaluating the curl of (5.55):
/.to [3n(n · m) - m]
B(x) 47T lxl3 (5.56)
Here n is a unit vector in the direction x. The magnetic induction (5.56) has
exactly the form (4.13) of the field of a d1pole. Th1s 1s the generahzat10n of the
result found for the circular loop in the last section. Far away from any localized
current distribution the magnetic induction is that of a magnetic dipole of dipole
moment given by (5.54).
If the current is confined to a plane, but otherwise arbitrary, loop, the mag-
neue moment can be expressed m a s1mple form. If the current I flows in a closed
circuit whose line element is dl, (5.54) becomes
X X dl
For a plane loop such as that in Fig. 5.7, the magnetic moment is perpendicular
to the plane of the loop. Since~ lx x dll = da, where dais the triangular element
of the area defined by the two ends of dl and the origin, the loop integral gives
the total area of the loop. Hence the magnetic moment has magnitude,
lml = I X (Area) (5.57)
regardless of the shape of the circuit.
,- ---..\
I
/
da>~dl ""
I ~
~
L_
I\ /
............ /
N g ure ':"!.!
Sect. 5.6 Magnetic Fields of a Localized Current Distribution, Magnetic Moment 187
me currem msuwunon rs provwea oy a numoer or cnargeo parncres wnn

H
charges a, and masses M-in motion with velocities v the magnetic moment can
be expressed in terms of the orbital angular momentum of the particles. The
current density is
J = 2: q;V;O(x - X;)
where X; is the position of the ith particle. Then the magnetic moment (5.54)
becomes
The vector product (x x v,) is proportional to the ith particle's orbital angular
momentum, L; = M;(x; x v;). Thus the moment becomes
(5.58)
; ,;..m;
Tf ,11 t~ · IM :n m. ,t; '" ~ th cgm~ -'- _t,-,_mgcc rgt;,-, { n./~4. of~;(\

r ·o '""' -, · /'
the magnetic moment can be written in terms of the total orbital angular mo-
mentum L:
e e
m=-2:L=-L (5.59)
2M;' 2M
This is the well-known classical connection between angular momentum and
m: 1t holrl< for orhit~l motion PvP.n on thP. ir ~r~IP Rnt
this classical connection fails for the intrinsic moment of electrons and other
e1emenLary panrcres. ror e1eetrons, Lne unnn~ll: 1~ ~nguuy mure lll<lll
twice as large as implied by (5.59) with the spin angular momentumS replacing
L. Thus we speak of the electron having a g factor of 2(1.00116). The departure
...
·" •'-
~· ~ ~ ...
. "·
~ ...
. ,1. . ,] ,]. 1- • • • • ,, • • • ....
quantum-mechanical effects which we cannot consider here.

Before leaving the topic of the fields of a localized current distribution, we
~r.nc;,-l~r th~ cnh~r;~g]" ;n 'r.f th~ m~ ' · ;n,-1 ,,; R Tuct ~<;nth~
electrostatic case discussed at the end of Section 4.1, there are two limits of
interest, one in which the sphere of radius R contains all of the current and the
other where the current is completely external to the spherical volume. The vol-
ume integral of B is
(5.60)
The volume integral of the curl of A can be integrated to give a surface integral.
lhUS
where n is the outwardlv directed normal. Substitution of (5.32) for A and an

interchange of the orders of integration permits this to be written as
"' I I
The angular integral is the same one as occurred in the electrostatic situation.
Making use of ( 4.16 1 ) , we therefore find for the integral of B over a spherical
volume,
r r I ,? \
I
Jr<R
~
uo u
•3
A
/Lo I
3 J
·n r<
1 2
r r>
,. I "
~
I •3 ,
~J.OlJ
where (r <• r>) are the smaller and larger of r 1 and R. If all the current density is
contained within the sphere, r< = r 1 and r> = R. Then
r 7 •.
B d· x - --"--- m ( 'i fi?i
Jr<R j
where m is the total magnetic moment (5.54). For the opposite extreme of the
current all external to the sphere, we have, by virtue of (5.14),
3
4
r B d 3x = :R B(O) (5.63)
~·
un.- '"~uu~
.I '~ f~\ ..
\-'·""'} auu \-'·"J} "'"" 'vc
'~ '~
• Wllll Lllt:ll uu; -
parts (4.18) and (4.19). The difference between (5.62) and (4.18) is attributable
to the difference in the origins of the fields, one from charges and the other from
circulating currents. If we wish to include the information of (5.62) in the mag-
netic dipole field (5.56), we must add a delta function contribution
">. "
R(v\ ,. !:'::._0 -'"\" "'} m ' o·
d.~ lvP "!
. Of"\
.\"/ \ ..
1<:: LA\
T/
-
The delta function term enters the expression for the hyperfine structure of
atomics states (see the next section).
'7 l<'nrPo nnrl 'T nn nnrl T<'nn~. nl" n T .J:-n.J r .. ~~n-t

'7. "0.." ~J
UlstrliJutwn m an b'Xternat Magnetic lnductwn

If a localized distribution of current is placed in an external magnetic induction
B(x), it experiences forces and torques according to Ampere's laws. The general
expressions for the total force and torque are given by .(5.12) and (5.13). If the
1~1 m~ · incl11<>tirm cl rmll u w +h" ~" :~- ~f "'
·o 'T'.~J"'
.. 'J
. . '
' " 1u1u '""' uum1nanL Lerms m Lne torce anu
'"' '"' "'"" uc
torque. A component of B can be expanded around a suitable origin,
Bk(x) = Bk(O) + x · VBk(O) + · · · (5.65)
Tl.
I H'-'H
...
H1'-' HH
.,,,
H VI
. £ ,, £
U1'-' IW'-''-'
/"' '~
\J,JL.} '
r; - 2..
jk
Eijk
L
Hk(U) Jf lj(x') d 0
X + J1 (x 1
)X
1
• VB;(O) iFx 1 + ... (5.66)
Here E;ik is the completely antisymmetric unit tensor ( E;Jk = 1 for i = 1, j = 2,

k = 3, and any cyclic permutation, E;ik = -1 for other permutations, and E;ik =
V lVl LWV Vl i11uices equatJ. 1ne vomme mtegraJ or J vamsnes ror steaav
Sect. 5.7 Force and Torque on and Energy of a Localized Cnrrent Distribution 189
currents; the lowest order contribution to the force comes from the second term
in (5.66). The result above (5.53) can be used [with x ~ VBk(O)] to yield
F; - 2: E;1k(m x V)1Bk(x) (5.67)
jk
After differentiation of Bk(x), xis to be put to zero. This can be written vecto-
rially as
F = (m x V) x B = V(m · B) - m(V · B) (5.68)
Smce V · B 0 generally, the lowest order force on a locahzed current d1stn
bution in an external magnetic field B is
F = V(m ·B) (5.69)
This force represents the rate of change of the total mechanical momentum,
including the "hidden mechanical momentum" associated with the presence of
electromagnetic momentum. (See Problems 6.5 and 12.8, and the references cited
at the end of Chapter 12.) The effective force in Newton's equation of motion
2
of mass times acceleration is (5 69) augmented by (1/c )(dldt)(E X rn), where E
is the extemal electric field at the position of the dipole. Apart from angular
factors, the relative size of the two contributions is (cEIL) versus (EtJr.), where L
is the length scale over which B changes significantly and A is the free-space
wavelength of radiation at the typical frequencies present in a Fourier decom-
position of the time-varying electric field.
A localized current distribution in a nonuniform magnetic induction expe-
riences a force proportional to its magnetic moment m and given by (5.69). One
simple application of this result is the time averaged force on a charged particle
spiraling in a nonunifmm magnetic field. As is well known, a charged particle ifl
a umjorm magnetlc mductwn moves m a circle at nght angles to the field and
with constant velocity parallel to the field, tracing out a helical path. The circular
motion is, on the time average, equivalent to a circular loop of current that will
have a magnetic moment given by (5.57). If the field is not uniform but has a
small gradient (so that in one turn around the helix the particle does not feel
significantly different field strengths), then the motion of the particle can be
discussed in terms of the force on the equivalent magnetic moment. Considera
tion of the signs of the moment and the force shows that charged particles tend
to be repelled by regions of high flux density, independent of the sign of the1r
charge. This is the basis of the "magnetic mirrors," important in the confinement
The total torque on the localized current distribution is found in a similar

way by inserting expansion (5.65) into (5.13). Here the zeroth-order term in the
expansion contributes. Keeping only this leading term, we have
r
N - J x' X [.I X B(O)J d 3 x:' (5 70)
Wntmg out the tnple vector product, we get
N = J[(x' · B)J - (x' · J)B] d 3 x'

11)0 Chanter 5 Mal!netostatics Faradav's Law Ouasi-Statk Fi<'l<l<- -ST
The first integral has the same form as the one considered in (5.66). Hence we
can write down its value immediately. The second integral vanishes for a localized
steady-state current distribution, as can be seen from (5.52) with f = g = r'. The
leading term in the torque is therefore
m X B<Ol l'i 711
'
This is the familiar expression for the torque on a dipole, discussed in Section
5.1 as one of the ways of defining the magnitude and direction of the magnetic
induction.
The potential energy of a permanent magnetic moment (or dipole) in an
eXLerna• magnenc new can De ootameo rrom either the force 15.691 or the tormJP.
('; 711 Tf Wf' • th<> fAr~<> ~o th<> ne> >t: "' •< •• ',] H
we find
U = -m · B (5.72)
For a magnetic moment in a uniform field, the torque (5.71) can be interpreted
<L • -' • • e n •
~" '"~~ ,..., "'"''""'"'"' v' v vvnu ''"''l:''"'"' tv we aug1e ueLween n anum. 1 ms
well-known result for the i"l e~ rlinnlP ~hnw~ th"t thP. rlinnle..t ,.J,
< • ;, Q]C .H.< <>. .C.U. •' • • . ~
tv "''"' "'"''""' '"'"' pvou,vu Ul 'v ,...,,, pvtCHlldl CHei~Y·
We remark in passing that (5.72) is not the total energy of the magnetic
moment in the external field. In bringing the dipole m into its final position in
the field, work must be done to keep the current J, which produces m, constant.
Even though the final situation is a steady state, there is a transient period initially
• ..L •
,I,+L ' ~ • • • •
"~ '"~ l " ' " ' d i l l m:;,uo dli:O lllllC·uc;pc;Hut:lll. 1 illS nes OUlSiue our presenL
consideratiOns. ( onseauentlv we leave the ··~"""ion of thP of ml'1anPtir-

fields to Section 5.16, following Faraday's law of induction. ~,
The energy expression (5.72) can be employed in the treatment of magnetic
effects on atomic energy levels, as in the Zeeman effect or for the fine and hy-
perfine structure. The fine structure can be viewed as coming from differences
inelleJ:lDLOf."n ·' ,•, · · 'm" ·•" in the-m~ · ·Hald ·
. • e
c. ~· . .
mHO~~,, HUH • llllC OllUI.LUHo, WlLH lllC !Canon or lllOmaS.Jl.reces-
s10n, IS discussed briefly in Chapter 11. The hyperfine interaction is that of the
magnetic moment 1-LN of the nucleus with the magnetic field produced by the
electron. The interaction Hamiltonian is (5.72) with m = 1-LN and B equal to the
magnetic field of the electron, evaluated at the position of the nucleus (x = 0).
This field has two parts; one is the dipole field (5.64) and the other is the magnetic
field DrOdllr-PrJ
_, . . . hv thP orhitl'11 motion nf thP p] 'o ~h·>r~o 'rho ].,H •
., . ,.
.. 0 uy ~ J.J) auu 1.au ue -r as Hmbitait U) J.LoeLI'+711Ylr ,
where L = x x mv is the orbital angular momentum of the electron about the
nucleus. The hyperfine Hamiltonian is therefore
87T
HFS
1 (.,.,.,\(~ \
,... ~,
+ u . UM 3 T • u
r L r m
The expectation values of this Hamiltonian in the various atomic (and nuclear
spin) states yield the hyperfine energy shifts. For spherically symmetric s states
Sect. 5.8 Macroscopic Equations, Boundary Conditions on B and H 191
the second term in (5.73) gives a zero expectation value. The hyperfine energy
comes solely from the first term:
(5.74)
For l * 0, the hyperfine energy comes entirely from the second term in (5.73)
*
because the wave functions for l 0 vanish at the origin. These expressions are
due to Fermi, who obtained them from the Dirac equation (1930). In applying
(5.73) and (5.74) it should be remembered that the chargee is negative and that
JJ.e points in the opposite direction to the electron's spin. The energy difference
(5.74) between the singlet and triplet states of the ls state of atomic hydrogen is
the source of the famous 21 em line in astrophysics.
The difference of the "contact" term in (5.73) from the electric dipole form
(4.20) allows us to draw a conclusion concerning the nature of intrinsic magnetic
moments. While orbital magnetic moments are obviously caused by circulating
currents, it is a priori possible that the intrinsic magnetic moment~ of elementary
partlcles such as the electron, positron, muon, proton, and neutron are caused
by magnetic charggs, arranged in magnetically neutral configurations (no net
magnetic charge). If the electron and proton magnetic moments were caused by
groups of magnetic charges, the coefficient 87T/3 in (5.74) would be replaced by
411/3! The astrophysical hyperfine line of atomic hydrogen would be at 42 em
wavelength, and the smglet and tnplet states would be reversed. I he expenmen-
tal results on positronium and muonium, as well as the magnetic scattering of
neutrons, giVe strong additiOnal support to the conclusion that intrinsic magnetic
moments of particles can be attributed to electric currents, not magnetic charges*
5.8 Macroscopic Equations, Boundary Conditions on B and H

So far we have dealt with the basic laws (5.26) of steady-state magnetic fields as
microscopic equations in the sense of the Introduction and Chapter 4. We have
assumed that the current density J was a completely known function of position.
In macroscopic problems this is often not true. The atoms in matter have elec-
trons that give rise to effective atomic currents, the current density of which is a
rapidly fluctuating quantity. Only its average over a macroscopic volume is
known or pertinent. Furthermore, the atomic electrons contribute intrinsic mag-
netic moments in addition to those from their orbital motion. All these moments
can give rise to dipole fields that vary appreciably on the atomic scale of
dimensions.
The process of averaging the microscopic equations to obtain a macroscopic
description of magnetic fields in ponderable media is discussed in detail in Chap-
ter 6. Here, just as in Chapter 4, we give only a sketch of the elementary
*There is a caveat that all particles must have the same origin for their moments. For a pedagogical
discussion of the experiments, sec J. D. Jackson, The nature of intrinsic magnetic dipole moments,
CERN Report No. 77-17, CERN, Geneva (1977), reprinted in The International Community of Phys-
icists: Essays on Physics and Society in Ilonor of Victor Frederick Weisskopf, ed. V. Stefan, AlP Press/
Springer-Verlag, New York (1997).
derivation. The first step is to observe that the averaging of the equation,
V • Bmicro = 0, leads to the same equation
V ·B = 0 (5.75
for the macroscopic magnetic induction. Thus we can still use the concept of a
vector potential A(x) whose curl gives B. The large number of molecules or atoms
per unit volume, each with its molecular magnetic moment rn;, gives rise to an
average macroscopic magnetization or magnetic moment density,
M(x) = L N;(m,> (5.76)
where M is the average number per unit volume of molecules of type i and (m;)
is the average molecular moment in a small volume at the point x. In addition
to the bulk magnetizatiOn, we suppose that there IS a macroscopic current density
J(x) from the flow of free charge in the medium. Then the vector potential from
a small volume~ Vat the point x' will be
~A(x) = JLo [J(x') ~ V

41T lx-x'l
+ M(x') x (x - x') ~
lx-x'l3
v]
This is the magnetic analog of (4.30). The second term is the dipole vector po-
tential (5 55) I etting ~ V become the macroscopically infinitesimal cPx ',the total
vector potential at x can be written as the integral over all space,
A(x) JLo ( [ J(x') + M(x') X (x - x')]d,x'

(5.77)
The magnetization term can be rewritten as follows:
(x- x') d3x' =I M(x') XV'( 1 ) d3x'

I M(x')lx-x'l
X
3
lx-x'l
Now an integration by parts casts the gradient operator over onto the magneti-
zation and also gives a surface integral. If M(x') is well behaved and localized,
the surface integral vanishes. The vector potential (5.77) then becomes
JLo ( [J(x') + V' X M(x')] d 3x'
A(x) (5.7B)
The magnetization is seen to contribute an effective current density,

(5.79)
The macwscopic equivalent of the microscopic equauon, V x Bmicro
,u 9 .Imiere' can be read off from (5 78) If the equations (5 26) have (5 32) as a
solution, then (5.78) implies that J + JM plays the role of the current in the
macroscopic eqmvalent, that is:
V X B = JLo[J + V X M] (5.80)
The V x M term can be combined with B to define a new macroscopic field H,
called the magnetic field,
1
H B M (5.81)
JLo
Sect. 5.8 Macroscopic Equations, Boundary Conditions on B and H 193
'T'~ .~ .1. {" '1t:\
' \' . ',,
V X H = J
(5.82)
V·B = 0
The introduction of H as a macroscopic field is completely analogous to the
--'nrtinn nf 0 fnr thP . l t"tir fif'lrl Thf' . f'(lll (<; RJ\
have their electrostatic counterparts,

V·D = p
(5.83)
VxE=O
We emphasize that the fundamental tields are 1<: and H. They satisfy the homo-
geneous equations in (5.82) and (5.83). The derived fields, D and H, are intro-
duced as a matter of convenience, to permit us to take into account in an average
way Llle COnLriUULIOUS lU p anu .J Ul lllC . c;uargc' auu c;uuciH,,
To complete the descnptwn of macroscopic magnetostatJcs, there must be a
constitutive relation between H and B. As discussed in the Introduction, for
isotropic diamagnetic and paramagnetic substances the simple linear relation
D .IJ f<: OA\
~ ,. • T I
holds, p., being a parameter characteristic of the medium and called the magnetic
permeability. Typically p.,/ p.,0 differs from unity by only a few parts in 105 (p., > p., 0
for paramagnetic substances and p., < p.,0 for diamagnetic). For the ferromagnetic
snhst:mces ('i.R4) m11st h<' n,n],.c<'rl hv" nnnline"r fnnctional relationshin
B = F(H) (5.85)
The phenomenon of hysteresis, shown schematically in Fig. 5.8, implies that B is
not a single-valued function of H. In fact, the function F(H) depends on the
• • ' ~· ' o .- ' ' 'TT' '
LJ v• }'HOj Vl UHO Lia1. H1'-' Hl'-1 " " }'\01111\0aUIIllJ ,_..~H) 1~
detined as the denvatJve ot H with respect to H, assummg that H and H are

parallel. For high-permeability substances, p.,(H)/ p., 0 can be as high as 106 . Most
untreated ferromagnetic materials have a linear relation (5.84) between B and
H for very small fields. Typical values of initial relative permeability range from
lUtOlU'.
,. _...- I / -
( I I I H
)__/v
I I
., ::
<::R . '~~~n R ;~ o
·o
. ' . ' ·:o .£0T
<~HV
Figure 5.9
The complicated relationship between B and H m ferromagnetic matenals

makes analysis of magnetic boundary-value problems inherently more difficult
than that of similar electrostatic problems. But the very large values of rela-
tive permeability sometimes allow simplifying assumptions on the boundary
conditions.
The boundary conditions for B and H at an interface between two media are
derived in Section I.5. There it is shown that the normal components of B and
the tangential components of H on either s1de of the boundary are related ac-
cording to
(B 2 - B 1) • n = 0 (5.86)
n X(~ H 1) K (5.87)
where n is a unit normal pointing from region 1 into region 2 and K is the ide-
alized surface current density. For media satisfying linear relations of the form
(5.84) the boundary conditions can be expressed alternatively as
/JQ.
B2 X n =- B1 X n (5.88)
f.LI
or
H2 · n- H1 • n, (5.89)
/JQ.
If f.Lr >> f.L 2, the normal component of H 2 is much larger than the normal com-
ponent of Hr. as shown in Fig. 5.9. In the limit (p;1 /p;2 ) - oo, the magnetic field
H 2 IS normal to the boundary surface, mdependent of the direction of H 1 (barring
the exceptional case of H 1 exactly parallel to the interface). The boundary con-
dition on H at the surface of a material of very high permeability is thus the same
as for the elec1ric field at the surface of a conductor. 'N€ may therefore us€
electwstatic potential theory for the magnetic field. The surfaces of the high-
permeability material are approximately "equipotentials," and the lines of Hare
normal to these equipotentials. This analogy is exploited in many magnet-design
problems. The type of field is decided upon, and the pole faces are shaped to be
equipotential surfaces. See Section 5.14 for further discussion.
5.9 ldetlwds u.fSulving Buundary- Value Problems

in Magnetostatics
The basic equations of magnetostatics are
V ·B 0, VxH J (5.90)
~mviul!i "'~""'
""CI. "·" i UI
-~· •Ill .LJ~
with some constitutive relation between B and H. The variety of situations that
can occur in practice is such that a survey of different techniques for solving
boundary-value problems in magnetostatics is worthwhile.
A. Generally Applicable Method of the Vector Potential

Because of the first equation in (5.90) we can always introduce a vector
potential A(x) such that
B=VxA
If we have an explicit constitutive relation, H = H[B], then the second equation
. {C()f\\ 1-. •
u ,~.
'!
V X H[V X A] = J
This is, in general, a very complicated differential equation, even if the current
,-1; ;J., ,,; ' ' .. L .L IJ rl D · ,1..+. rl H'nr ];n~~r m~rl;~ ..,;th
·r "Y'J
'
B = pH, the equation becomes
\
'
vx iVxA =J (5.91)
u
If M is constant over a finite region of space, then in that region (5.91) can be
V(V ·A) - V2 A = J.LJ (5.92)

With the choice of the Coulomb gauge (V ·A = 0), this becomes (5.31) with a
modified current densitv. ( ul uo )J. The situation closely parallels the treatment
of uniform isotropic dielectric media where the effective charge density in the
. . . . . . . .
. ClfUdllUlllS Eop!E. ;)UJUllUllS~c;Hl HlU" U\0
'
matched across the boundary surfaces using the boundary conditions (5.88)
or(<; RQ\
B. J = 0; Magnetic Scalar Potential

Tf thP. «1mP finite re!!ion of soace the second
lt ' · · ''"' in
equation in (5.90) becomes V x H = 0. This implies that we can introduce a
magne1ic scawr puteruiw 'l'M su<.:11 LnaL
H = -V<PM (5.93)
just as E = - V<P in electrostatics. With an explicit constitutive relation, this time
.c • ., . ' .
Ul D DLD.J, lllC ~. D V CljUdllUll l:dll UC
..., ~c -~ ' "

-q
• VL -v'PMT
Again, this is a very complicated differential equation unless the medium is linear,
Ill <.:ase llle
.-, I ..,,y., ' n ( <:: ()A\
\I-" M! \' • 'I
If M is at least piecewise constant, in each region the magnetic scalar potential

lll<O T "i
'
t"'2ffi ()
M
The solutions in the different regions are connected via the boundary conditions
(5.89). Note that in this last circumstance of piecewise constancy of Jh, we can
also write B - -V'JI" with V2']!, - 0. With thi' ~ltf'rn~tivf'. "·~l"r nntf'nti"l th~
boundary conditions (5.88) are appropriate.
The concept of a magnetic scalar potential can be used fruitfully for closed
~~~~" ~· ~· I+ h ·~ •+ ih . . _, • •• _,._., _,
·r ·m r -r '5'"
subtended by the boundary of the loop at the observation point. See Problem
5.1. Such a potential is evidently multiple-valued.
C. Hard Ferromagnets (M given and J = 0)

A common practical situation concerns "hard" ferromagnets, having a mag-
netizatwn that IS essentially independent of applied fields for moderate field
strengths. Such materials can be treated as if they had a fixed, specified magne-
tization M(x).
(a) Scalar Potential

Since J = 0, the magnetic scalar potential ct>M can be employed. The first
equatiOn m (5.lJO) IS wntten as
V·B = Jl-oV · (H + M) = 0
1 nen with ~:J.'olj) It oecomes a magnetostat1c l'msson equation,
V2cJ:> M = -pM (5.95)
wnh me errecuve maRnetzc-cnarRe aensztv,
PM= -v · M (5.96)
The solution for the potential ct> M if there are no boundarv surfaces is
<l>w(x) = _ __!_IV' . M(x') d3x' (5.97)
" 41T lx- x'l
H 1\lr . .. 11 ~ 0~ rl rl 1
·" rl
, L
JJ p«H~
-• -'
H>«J U~
~
p~HUHH~U <V
yield
1 I l
4 1T J
'~'M\X) 1H\X ) • V ax
lx- x'l
Then
V'(lx ~ x'l) --
-v(lx ~ x'l)
mav be used to ~ive
1 ( M(x')
ct>M(x) V· d 3 x' (5.98)
't1T J IX X"/
In passing we observe that far from the region of non vanishing magnetization
the notential mav hf' · l!'ltf'n hv
1 '1 {
'VM~X)-
41T VI r I· J M(x') d-x
... ~
3
41Tr
-'•>' m .... ........
~/·
.><OUo - • aou<O •
""
where m = fM d 3 x is the total magnetic moment. This is the scalar potential of

a dipole, as can be seen from the electrostatic ( 4.10). Thus an arbitrary localized
distribution of magnetization asymptotically has a dipole field with strength given
by the total magnetic moment of the distribution.
While physical distributions of magnetization are mathematically well be-
haved and w1thout d1scontmmt1es, 1t 1s sometimes convement to Idealize the re-
~ litv ~ncl treM M(x) ~s if it were cliscontinuous. Thus. if a "hard" wnet
• .L I/
" •<'c ('
' ·r
. .,_ 1\Jrf-\ .
'J '\"!
. ..], 1/
, " •• .t ' <'n 11.
suddenly to zero at the surface S. Application of the divergence theorem to PM

(5.96) in a Gaussian pillbox straddling the surface shows that there is an effective
magnetic surface-charge density,
aM = n·M (5.99)
where n is the outwardly directed normaL Then instead of (5.97) the potential is
P'iven hv
1 I V • M(x ) ,, 1 ,), n • M(x) da
"'M~"'J A. I, I If U A
A T I •I
~.J.LUU)
I I I I
An important special case is that of uniform magnetization throughout the vol-

ume V. Then the first term vanishes; only the surface integral over aM contributes.
It is important to note that (5.98) is generally applicable, even for the limit
of discontinuous distributions of M, because we can introduce a limiting proce-
dure after transforming (5.97) into (5.98) in order to discuss discontinuities in M.
Never combine the surface integral of aM with (5.98)1
(b) Vector Potential
11 we cnoose LO wrne n v x A 10 sansty v • n v auwmancauy, wen we

write the second equation of (5.90) as
V X H = V X (B/ f.Lo - M) = 0
This leads to the Poisson equation for A in the Coulomb gauge,
'112A = - f.LoJM (5.101)
mhPrP .J. i< thP m~ crnPtir ,-1, '•· ('; 7Q) ThP. <oh1tion for thP.
. -' . d . .L
~
c L _, .£.
,. " ' u<v V< 'J "' 00
r n• v 1\Jrf, '\
A(x) =j
41T lx- x'l
d 3 x' (5.102)
as was already shown in (5.78). An alternative form is given by the magnetization

term in (5.77).
If the distribution of magnetization is discontinuous, it is necessary to add a
surface integral to (5.102). Starting from (5.77) it can be shown that forM dis-
continuously falling to zero at the surface S bounding the volume V, the gener-
alization of (5.102) is
/ __ , f.Lo I V X !Vl~X ) J3 I f.Lo ,.j., tn~x ) x n J. I fC 1 fY!>
.•
~\AJ
.1~ /, fv v 11 ~A
.1~ Jc lv v11 ~u
,~ ~J!
' '
The effective surface current (M X n) can also be understood by expressing the
boundary condition (5.87) for tangential H in terms of B and M. Again, if M is
constant throughout the volume, only the surface integral survives.
5.10 Uniformlv Marmetiud Snhere

1o Illustrate tne amerent metnoas possible tor the solutiOn of a boundary value
problem in magnetostatics, we consider in Fig. 5.10 the simple problem of a
'1'11""' m rauiu~ a, wiw a uniwrm permanem magnetization !VI or magnituae M
0
and parallel to the z axis, embedded in a nonpermeable medium.
~.
HlOO
. -'
~r H
" . . .~
Vl MllU<lV11" 111Ul Vl pall L~U) Ul lilt: p!• ~~ ~t:lOllUn,
via the magnetic scalar potential in spherical coordinates and a surface magnetic~
-L " fa\ HT:<L " " u " •• >< n <' _' ·
'M\VJ• .. VM 'lH lHQ '-'"' u, LH'- 'VlUUVU
o~3 ~ ~ H
(5.100) for the potential is
<PM(r, O) =
M a
4~
2
J d!l' lxcos- 0'x' I
With the expansion (3.38) or (3.70) for the inverse distance, only the l = 1 term
'Ul VlVC,, lllC, llldl "
_ 1 r<
 M(r, 0) - - M 0a2 2 cos 0 (5.104)
3 r>
where (r<, r>) are smaller and larger of (r, a). Inside the sphere, r< -- rand
"T'L ,y, {1 rl\U D {1!'1\U ·CT',. . ~,_u "
> · M ,. 'J ·u• u ,_ ? ·u~·
induction inside the sphere are therefore

1 2J.Lo
H;n =-3M, B,n =3M (5.105)
We note that B;n is parallel to M, while H,n is antiparallel. Outside the sphere,
 , 111<0 jJULC111ldl " UlU'
1 cos 0
'l'M IVI 0 U ().lUbJ
3 r2
This is the potential of a dipole with dipole moment,
47Ta3
m=--M (5.107)
3
For the sphere with uniform magnetization, the fields are not only dipole in char~
. .
a.:Lel asympL(JLJCaHy, uUL a1so c1ose LO LIIe spnere. tor tms special geometry~ ana
nn]~,\ thPrc> ~rP nn hiahPr m11l·'
thi<
~· ,. .£ ~
u1'- 1111._, " ' ~ a11u • • at'-

_, ..: '
,1
.
m , •ts· J,, . '","' '"""~ 01 n ate JU'
closed naths but those of H terminate on the surf~"" • thPrP i< ~n .u. ivP
.~. ~l
'6' ~· 'J M·
M =MoE3 ~p
~8
/
/
"",
! J<;
0
" 1fi
Sect. 5.10 Uniformly Magnetized Sphere 199
/
I \ I \
B H
Figure 5.11 Lines of B and lines of H for a uniformly magnetized sphere. The lines of
B are closed curves, but the lines of H originate on the surface of the sphere where the
.cc, "'rf"rP " " ,., rP,iriP'
Brief mention should be made of employing (5.98) instead of (5.100). With

M = M0 e 3 inside the sphere, (5.98) gives
1 M0
M(r, 8) = - -
"
-:;-
v~
a c r 12 dr 1 r df! 1 1
, . - - - - ---:
I
1
I
(5.108)
Now only the l = 0 term in expansion of the inverse separation survives the
angular integration and the integral is a function only of r. With Ar!Az = cos 8,
the potential is
M 0 cos 8
a
Uf
r
•u
Y
12
I>
dr
1
1
Integration over r leads directly to the expression (5.104) for M.
An alternatiVe solutwn can be accomplished by means ot the vector potential
and (5.103). Because M is uniform inside the sphere the volume current density
JM vanishes, but there is a surface contribution. With M = M 0 e 3 , we have
M X R 1 = M 0 sin 8 1 Eq,
= M 0 sin 8 -sin </> 1
(
1
E1 + cos </> e2 ) 1
Because of the azimuthal symmetry of the problem we can choose the observa-
tion point in the x-z plane ( 4> = 0), just as in Section 5.5. Then only the y com-
• I ' ' ' ' ~. ' ' •
r u v• iH ~ u ~unuv~ """'' v 1"' "'" , !S""'!S au
component ot the vector potential,
/.Lo 7 { sin 8 1 cos 4> 1
A JYJ au I ua I. I I
I I
where X
1
has coordinates (a, 8 1 , 4> 1 ). The angular factor can be written
1
sin 8 cos </> = - 1
r;· Re[Y1,1 (8 1 , </> 1 ) ] (5.110)
Thus with expansion (3.70) for lx- x' I onlv the l = 1 m = 1 term will surviv"
Consequently
\
(5.111)
where r< (r>) is the smaller (larger) of rand a. With only a 4> component of A,
the components of the magnetic induction Bare given by (5.38). Equation (5.111)
evidently gives the uniform B insirle and the ~: ~• . fielrl outside. as~- '"'• ·~-
5.11 Magnetized Sphere in an External Field; Permanent Magnets
' magnetic in-

of the linearity of the field equations we can superpose a uniform -
duction Bn - unHn throu<Thout all snace. Then we h~vtc the nrnhlPm -.,f ~ ,n;_
formly magnetized sphere in an external field. From (5.105) we find that the
magneric inaucrion ana ueld mside the sphere are now
lfLo
Bin= Ba +3M
~J . .llL)
n. lR 1M
!Lo 3
We nn~thP 'i< nnt ~ •wrm~nPntlu~ ~h;o~+.
rather a paramagnetic or dia~agnetic substance of p;rmeability IL· Then' the
magnetization M is a result of the application of the external field. To find the
magnitude of M we use (5.84):
(5.113)
Thus
2/Lo 1 1
B0 + ~ M = !L - B0 - - M (5.114)
3 !L - /Lo '
1n ~:J.ll:l)
We note that this is completely analogous to the polarization P of a dielectric

so here m a umform electrir fif'lrl (4 'i71
For a ferromagnetic substance, the arguments of the preceding paragraph
rau. tcquanon 12_.1D) Implies that the magnetizatiOn vanishes when the external
fic>lrl u~niohPo . The> c>v' '" • rl' '' · .. 1 . •
""'
u
nonlinear relation (5.85) and the phenomenon of hysteresis allow the creation of
ntcrm~nent m Wtc r~n onluP
H;n and Bin by eliminating M:
.
inno (<; 11 ?I fnr nnP rc>lot'
'
h
.
(5.116)
1 ne nysLeresJs curve provides the other relatiOn between Bin and Him so that
'L'O ,~]HPO{'•,nJ-. ·1 >, ,1.0,1.1 I:', ' {"11t:\ ~
Sect. 5.12 Magnetic Shielding, Spherical Shell of Permeable Material in a Uniform Field 201
\ \B \ \
'\:X 1\ ·~ \
\ \ f'U"
\ \
\ \
\ \
Y\\\ Figure 5.12
to a !me w1tl1 slope -Lon the hysteresis 01agram with mtercept jJ5 0 on tne y axzs,
go ;n p;o <; 1 ') <;;:, fnr Pvgrr>niP thgt thP >1 fip]cJ ;o ;n, nnt;l thP
-o· · ··rr ' '
ferromagnetic sphere becomes saturated and then decreased to zero. The internal
Band H will then be given by the point marked Pin Fig. 5.12. The magnetization
can be found from (5.112) with B 0 = 0.
The relation (5.116) between B;n and Hm is specific to the sphere. For other
•+h ,I 'T'h ,hi. ,f +h , I"
._, ~ .1 • _,
0 1"'
r. '_
exacuy anu suows LHaL Llle S1ope 01 LllC nncs ~-'·''UJ ~<tnge tn•m zno tut a uaL
disc to - oo for a long needle-like object. Thus a larger internal magnetic induction
can be obtained with a rod geometry than with spherical or oblate spheroidal
shapes.
..........
: 1') "J-1.
H.L
.~·
"".s··~··~
(:'I. •. , J.•
·.s'
(:',.I.
".I:
•
.~..,. ~··~··
.1 l:'l. .11
of Permeable Material in a Uniform Field

Suppose that a certain magnetic induction B 0 = J.toHo exists in a region of empty
space initially. A permeable body is now placed in the region. The lines of mag-
netic induction are modified. From our remarks at the end of Section 5.8 con-
0
,-l; ,f
·o
]-,' -~
'b' r
'~ ;]; •. ' .
-'-'
'
•+ +h fi.,],-l ];
.
•• •• .-l
to oe normal to me surrace or me ooay. carrymg me analogy wnn conauctors

further, if the body is hollow, we would expect the field in the cavity to be smaller
than the external field, vanishing in the limit JL---+ oo. Such a reduction in field is
said to be due to the magnetic shielding provided by the permeable material. It
is of considerable practical importance, since essentially field-free regions are
often necessary or desirable for experimental purposes or for the reliable working
of p], · clPvirP<
au
1-\.' He Ul Lllt: Ul
'"" we "
spherical shell of inner (outer) radius a (b), made of material of permeability J.t,
and placed in a formerly uniform constant magnetic induction B 0 , as shown in
Fig. 5.13. We wish to find the fields B and H everywhere in space, but most
particularly in the cavity (r < a), as functions of J.t· Since there are no currents
. ' . .
prescnr, rnc magncLic ncJU n IS ucnvamc uorn a pOLenuai, n . V'l'M·
FnrthP.rmnrP. <inrP R 11H thP. eli v.R () hf'r.Omf'S
' " " p(ll
_,
...
..
TT A'
V H>
•'-
CH~ 'UHVUO •
'T'L
~HUO
<L
"'~
._, ffi
~M
•'-
'"~
T
Figure 5.13
equation everywhere. The problem reduces to finding the proper solutions in the
different regions to satisfy the boundary conditions (5.89) at r = a and r = b.
For r > b, the potential must be of the form,
~ a!
<PM= -Hor cos()+ L, l+i Pl(eos e) (5.117)
1~o r
to give the uniform field, H = H 0 , at large distances. For the inner regions, the
a<1<b
r <a
The boundary conditions at r =a and r =bare that He and B, be continuous.

In terms of the potential <PM these conditions become
(5.119)
The notation bee means the limit r ~ b approached from r "= b, and similarly for
a These fom conditions, wbicb bold for all angles e, are sufficient to determine
tl1e unknown constants in (5.117) and (5.118). All coefficients with l '/' 1 vanish.
The l - 1 coefficients satisfy the four simultaneous equations
3 3
".:,()', + /.L 'b 3 P.{oJJ - 2 f,L ' 'Y1 -b 3 HQ~

(5.120)
a3 f31 + 'Yl -a 3 (\ = 0
J.L'a 3 {3 1 - 2J.L'Y1 -a 3 81 = 0
Here we have used the notation J.l.' = J.LIJ.Lo to simplify the equations. The solu-
tions for a 1 and 81 arc
(5.121)
Sect. 5.13 Effect of a Circular Hole in a Perfectly Conducting Plaue 203
Figure 5.14 Shielding effect of a shell of highly permeable material.
a dipole field (5.41) with dipole moment a 1 oriented parallel to H 0 . Inside the
cavity, there is a uniform magnetic field parallel to H 0 and equal in magnitude
to - .. For >> the moment and the inner field - become
(5.122)
We see that the inner field is proportional to fL _,. Consequently a shield made
of high-permeability material with fL/ fLo ~ 103 to 106 causes a great reduction in
plane with an asymptotically uniform normal electric field. Its magnetic counter-
part has a uniform tangential magnetic field asymptotically. The two examples
. .
Before sketching the solution of the magnetostat1c boundary-

we must discuss what we mean by a perfect conductor. Static magnetic fields
204 Fields-SI
because of its magnetic properties, not its conductivity, unless of course there is
current flow inside. With fields it is often otherwise. It is shown in
nonconductor, fields
tivity. For any nonvanishing w, therefore, the skin depth 8 ~ 0 as u ~ oo. Os-
cally varying fields as w ~ 0, provided at the same time that wu-> oo. Then the
magnetic field can exist outside and up to the surface of the conductor, but not
inside. The boundary conditions (5.86) and (5.87) show that B · n = 0, n x H =
K at the surface. These boundary conditions are the magnetostatic counterparts
conductor, where in this last relation u is the surface-ch

conductivity!
medium surrounding the plane is uniform, isotropic, and linear and that there is
a m z > far
from the hole, and zero field asymptotically for z < 0. Other possibilities can be
obtained by linear superposition. Because there arc no currents present except
(5.123)
The reversal of sign for the added potential c[J(l) below the plane is a consequence
of the symmetry properties of the added fields-Hj.'l and JIS'l arc odd in z, while
H~ l and ¢<'l are even in z. This can be inferred from (5.14) with the realization
1
that the effective current is only on the surface z = 0, as is the effective
dk A(k)e-klziJ,(kp) sin 4>
X
Sect. 5.13 Effect of a Cit culm Hole in a Petfectly Conducting Plane 295
Only m = 1 enters because the hole is cylindrically symmetric and the asymptotic
field varies as y = p sin</>. From the boundary conditions on normal B and tan-
gential H we find that the boundary conditions on the full potential ct>M are
ct>M continuous across z = 0 for 0 M - 0 at z - 0 fora<p<oo
These requirements imply the dual integral equations,
j ()
dk A(k)J1 (kp) - H 0 p/2 for 0 < p <a
f oo
()
dk kA(k)JJ(kp) = 0 for a< p < oo
These are closely related to, but different from, the electrostatic set (3.178) or
(3.1 79). I he necessary pair here arc
{00
J dy g(y)J,(yx) x" for 0 < x < 1
r
0
(5 126)
dy yg(y)ln(yx) = 0 for 1 < x < oo
with solution,
(5.127)
In (5.125) we have g = 2A(k)/J-10 a2 , n = 1, x = pia, and y = ka. Hence
(5.128)
The added potential is therefore
2Hoa2
7r
r d'<' ; ('<a)e-kizl
Jo Jl •
T ('<p) sin"'
.rl '
(5.129)
By methods similar to those of Section 3.13 it can be shown that far from the
opening the added potential has the asymptotic form
<tPl(x) ~ 2Hoa3 . y (5 130)
This is the potential of a dipole aligned in the y direction, the direction of H 0 .

Because of the signs in (5.123), the circular hole is equivalent at large distances
to a magnetic dipole with moment
3
m = ± - H0 for z "= 0 (5.131)
3
where H 0 is the tangential magnetic field on the z = o+ side of the plane in
the absence of the hole. Later (Fig. 9.4) we show qualitatively how the magnetic
field lines distort to give rise to the dipole field In the opening itself (z 0,
0 < p < a) the tangential and normal components of the magnetic field arc
?flo a
Hz(p, 0)
Comparison with the corresponding electrostatic problem in Section 3.13

shmvs similarities and differences. Roughly speaking, the roles of tangential and
normal components of fields have been mterchanged. I he effectiVe dipoles pomt
in the directions of the asymptotic fields, but the magnetic moment (5.131) is a
factor of 2 larger than the electrostatic moment (3.183) for the same field
strengths For arbitrarily ,baped bole, the far field in the electrostatic case is still
that of a dipole normal to the plane, while the magnetic case has its effective
dipole in the plane, but now the direction of the magnetic dipole depends on
both the field direction and the orientation of the hole (the hole has an aniso-
tropic magnetic susceptibility).
5.14 Numerical Methods for Two-Dimensional Magnetic Fields

Magnetic fields in the presence of iron or other highly permeable materials can
be evaluated numerically in two dimensions by the relaxation method described
in Section 1.13 or, more generally, by the method of finite element analysis of
Section 2.12. The problems can be classed as "interior" or "exterior," depending
whether the current flow and/or magnetized material and desired field are within
the same region.
First consider the boundary conditions for the field components at the
smooth interface of a highly permeable medium and a nonpermeable one. Lo-
cally, the interface can be approximated by a plane. The boundary conditions
are that the tangential component of H and the normal component of B are
continuous across the interface, if there are no surface currents. Figure 5.16 is a
Figure 5.1ti Illustration of the effcet of large permeability on the components of the
magnetic induction and magnetic field on either side of an interface. The sketch has
J.L "" 5J.L 0 , not a very high permeability!
"<OU• ~d, <
. .. .. . . ~
- ~· .. "' _, .. ?1\'7
~ n '
.,
'
'J<.t:LC!l Ul lilt: Ut:lldVlUl Ul lllt: llt:lU T ITs; \V ~-
~ '5• '
UU\ OHV""'5 J,7
both B and H components. For a given "external" field B(o) in the nonpcrmeable
region, the components of B (and H) in the highly permeable medium are more
closely parallel to the interface. The magnitude of the magnetic induction just
inside the highly permeable medium is
2
IB 12 = B(0)2
j_
+ /!:_
.. ~
B(O)Z
II
whiJP thP nPr nnit volume (see Section 5.16) there is
__!__ IB 12 = __!__ B(0)2 + __!!:___ B(0)2

2JL 2JL j_ 2p}j II
These two relations are immediately useful in learning the appropriate boundary
conditions of "exterior" and "interior" problems in the limit JLI JLo ---> oo.
The most familiar static magnetic fields arc those around a permanent mag-
net of high permeability or an iron core excited by remote current-carrying wind-
. . . . . ' .
mgs. 111e m ;L Is Lilt: '"5'v" u y woo
permeable pole face or faces-the arCllefYi)al"extenor ·problem. 1t we suppose
thM thP. PnPrcrv within thP. hiahhr •hlP medium is finite the enemv
relation shows that, as JLI f.Lo ---> oo, the parallel component of the magnetic field
outside must vanish: the "external" magnetic field at the surface is perpendicular
to the interface. These are just the boundary conditions for the electrostatic field
at the surface of a conducting boundary, as mentioned at the end of Section
5.8. If there are no currents within the nonpermeable region of interest, then
V x H = 0 there and we can write H = -V<PM. The magnetic scalar potential
'.
,.
satisfies the Laolace eauation V2 <P" = 0 with the "pole pieces " surfaces of
..
. '1· .~,
, .,~.,1~~" ..,;+J.. a1 +n+;N ;e
.... "'
lnlPtP
- '-' • L '' ...
L'Ul we _J lJUl --.::Jr "Tinvl"H jJl \V \' 'V '-"'
menswns, with steady current now only m tnc tnira mrecLion in a uni1orm, 11igwy
permeable conducting medium. We are interested in the magnetic induction
within the medium-for example, a long iron third rail of a subway system. The
current flow produces a magnetic induction both inside and outside the medium.
Whatever the magnitudes of the parallel and perpendicular components just out-
side, the boundary conditions assure that B is parallel to the surface of the me-
dium just inside as JLI f.Lo ---> oo.
H we currenL uensiLy 11as OlllY a z
. '
ll, J z~-"• y !• "''-' Y'-'\AV< ,-
A has onlv a z comooncnt. A.!x, v), which satisnes the l:'oisson equatwn,
\;2 .1 T 'T'J..a fi;l.-1 ~~~ o~~ R :>.1 /;;, R -ilA lilr R 0.
·r
TC -~'
• r -,', " ' " ' - ' n , . '1 •'- L -'· r ,., +h D ,,-I
u cuv " " ' - ' ' •w •v '] ~
IllJ:<Ig. :l.l I, We nave n • ~ V X A) ~n X V j_) • A lJ lJll L. lilt: [Sl·

j_ . UjJt:lctLVl
in the x-y plane can be resolved into components parallel to and perpendicular
to n. The boundary condition thus becomes
aAz = 0
az
wncrc at IS an element or arc Jengtn awng L. 1 ne vec,or po,en,Ia, b cvuMctlll
alon~> the boundarv curve C. Furthermore we can infer that in the interior regwn
.. <;,1rl1.' ... ~ .. 11~1 •~ +\,~ "'~•~"~" of rondont A R~roll<(" R
D ••
•
• ,,
~ no., '""'
" ,.
Vl Hll'-'0 "
Vl" lVl'-''-' 10 5' ''-'11
•UJ d_
uv vv
-' ,< A
z r •r
n
"'"·"'' permeable
cylindrical conductor with current flow along its length.
of contours of constant
mcrements m of the field as well as its
direction.
numc11•c.m u1cunJu>
V 2Az = -p.Jv boundary conditions must be specified. That seems to mean the
1s ar
to within addition of the gradient of a scalar function X· With the choice,
. = pJz within R with the homogeneous boundary condition A; = 0 on the

C. The value of on C · ·
needed. With J,(x, y) specified, the solution by the relaxation technique proceeds
as in Section 1.13.
num codes exist to solve more realistic magnetic field prob-
lems where, for example, the different permeable materials have but not
arc giVen at
5.15 Faraday's Law of Induction

The first quantitative observations relating time-dependent electric and magnetic
in circuits
sient current is induced in a circuit if (a) the steady current flowing in an adjacent
circuit is turned on or the acent circuit with
is moved relative to the first circuit, (c) a permanent magnet is thrust into or out
of the circuit. No current flows unless either the accnt current or there
IS motiOn. the transient current flow to a changing
magnetic flux linked by the circuit. The changing flux induces an electric field
Seet. S.lS J<araday's Law of Induction 209
Figure 5.18
The electromotive force around the circuit is
(5.134)
where E' is the electric field at the element dl of the circuit C. Faraday's obser-
vations are summed up in the mathematical law,
dF
dt
The induced electromotive force around the circuit is proportional to the time
1 ate of change of magnetic flux linking the circuit. The sign is specified by Lenz's
law, which states that the induced current (and accompanying magnetic flux) is
in such a direction as to oppose the change of flux through the circuit.
The constant of proportionality k depends on the choice of units for the
electric and magnetic field quantities. It is not, as might at first be supposed, an
independent empirical constant to be determined from experiment. As we will
see immediately, once the units and dimensions in Ampere's law have been cho-
sen, the magnitude and dimensions of k follow from the assumption of Galilean
invariancc for Faraday's law. For SI units, k = 1; for Gaussian units, k = c-I,
where c IS the velocity of light.
Before the development of special relativity (and even afterward, when in-
vestigators were dealing with relative speeds that were small compared with the
velocity of light), it was unde1stood, although not often explicitly stated, by all
physicists that physical laws should be invariant under Galilean transformations.
That is, physical phenomena are the same when viewed by two observers moving
with a constant velocity v relative to one another, provided the coordinates in
space and time are related by the Galilean transformation, x' = x - vt, t' = t.
In particular, consider Faraday's observations. It is expected and experimentally
verified that the same current is induced in a secondary circuit whether it is moved
while the primary circuit through which current is flowing is stationary or it is
held hxed whtle the pnmary czrcuu ts moved in the same relative manuel.
Let us now consider Faraday's law for a moving circuit and see the conse-
quences of Galilean invariance. Expressing (5.135) in terms of the integrals over
E' and 8, we have
fc E' • dl = -k :t L B · n da (5.136)
Figure 5.19
The induced electromotive force is proportional to the total time derivatlve of

the flux the flux can be changed by changing the magnetic induction or by
changing the shape or orientation or position of the circuit. In form (5.136) we
have a far-reachmg generahzatlon of Faraday's law. The circuit C can be thought
of as any closed geometrical path in space, not necessarily coincident with an
electric circuit. Then (5.136) becomes a relatiOn between the fields themselves.
It is important to note, however, that the electric field, E' is the electric field at
dl in the coordinate system or medium in which dl is at rest, since it is that field
that causes current to flow if a circuit is actually present.
If the circuit Cis moving with a velocity v in some direction, as shown in Fig.
5.19, the total time derivative in (5.136) must take into account this motion. The
flux through the circuit may change because (a) the flux changes with time at a
point, or (b) the translation of the circuit changes the location of the boundary.
It is easy to show that the result for the total time derivative of flux through the
moving circuit is*
d f
d t .d
1 B·nda [ iJB • n da
J5 f!l
+ f. c
(B X v) · dl (5.137)
Equation (5.136) can now be wntten m the form,
f. [E'
c
k(v X B)] · d! k
r iJB
Js iit • n do (5 138)
This is an equivalent statement of Faraday's law applied to the moving circuit C.

But we can choose to interpret it differently. We can think of the circuit C and
surfaceS as instantaneously at a certain position in space in the laboratory. Ap-
plying Faraday's law (5.136) to that fixed circuit, we find
1Jc E • dl = ~k Js iJB
at
•n da (5.139)
*For a general vector field there i> "" odded term, f,(V B)•• n da, whieh gives the eontrilmtion of
th® sources of the veetor field swept over by tlte moving circuit. The gene1al Iesu1t follows most easily
ftom the use of the convective deiivative,
'.
at
- - -
~ ar + (v • V)B - -
ill
+ V x (B X v) + v(V ·B)
where vis treated as a fixed vector in the differentiation. Use of Stokes's theorem on the second term
yields (5.137).
Sect. 5.15 Faraday's Law of Induction 211
where E is now the electric field in the laboratory. The assumption of Galilean
invariance implies that the left-hand sides of (5.138) and (5.139) must be equal.
This means that the electric field E' in the moving coordinate system of the circuit
E' E + k(v x B) (5.140)
To determine the constant k we merely observe the significance of E'. A charged

particle (e.g., one of the conduction electrons) essentially at rest in a moving
circuit will experience a force qE'. When viewed from the laboratory, the charge
represents a current J = qv8(x- x0 ). From the magnetic force law (5.7) or (5.12)
it is evident that this current experiences a force in agreement with (5 140) pro-
vided the constant k is equal to unity (SI) or 1/c (Gaussian).
lhus we see that, wtth our choice of units for charge and current, Galilean
covariance requires that the present constant k be equal to the constant appearing
in the definition of the magnetic field (5.4). Faraday's law (5.136) therefore reads
1Jc E ' · dl = - !!_

dt
Js B · n da (5.141)
whcr c E' is the electric field at til in its rest frame of coordinates. The time
denvattvc on the right is a total time derivative (5.137). As a by-p10duct we have
found that the electric field E' in a coordinate frame movmg wtth a veloctty v
relative to the laboratory is
E' = E + v X B (5.142)
Because we considered a Galilean transformation, the result (5.142) is an ap-

proximation valid only for speeds small compared to the speed of light. (The
relativistic expressions are derived in Section 11.10.) Faraday's law is no approx
tmatton, however. The Galilean transformation was used merely to evaluate the
constant kin (5.135), a task for which it was completely adequate.
Faraday's law (5.141) can be put in differential form by usc of Stokes's the-
orem, provided the circuit is held fixed in the chosen reference frame (to have E
and B defined in the same frame). The transformation of the electromotive force
integral into a surface integral leads to
Since the circuit C and bounding surface S' arc arbitrary the integrand must
vanish at all points in space.
Thus the differential form of Faraday's law is
aB
V X E + - = 0 (5.143)
We note that this is the time-dependent genera!tzatton of the statement,

V x E - 0 for electrostatic fields
212 Chapter .:; Magnetostatics, Faraday's Law, Quasr-Stanc Frelds SI
5.16 Energy in the Magnetic Field

In discussing steady-state magnetic fields in the first 14 sections of this chapter
we avoided the question of field energy and energy density. The reason was that
the creation of a steady-state configuration of currents and associated magnetic
fields involves an initial transient period during which the currents and fields are
brought from zero to the final values. For such time-varying fields there are
induced electromotive forces that cause the sources of current to do work. Since
the energy in the field is by definition the total work done to establish it, we must
consider these contributions.
Suppose for a moment that we have only a single circuit with a constant
current I flowing in it If the flux through the circuit changes, an electromotive
force cg is induced around it. To keep the current constant, the sources of current
must do work. To determme the rate, we note that the time rate of change of
energy of a particle with velocity v acted on by a force F is dE!dt • ·F. With
a changing flux, the added field E' on each conduction electron of charge q and
mean velocity v gives rise to a change in energy per unit time of qv · E' per
electron. Summing over all the electrons in the circuit, we find that the sources
do work to maintain the current at the rate
t
the negative sign followmg from Lenz's law. This is in addition to ohmic losses
in the circuit, which are not to be included in the magnetic-energy content. Thus,
if the flux change through a circuit carrying a current I is 8F, the work done by
the sources is
8W =I 8F
Now we consider the problem of the work done in establishing a general
steady-state distribution of currents and fields. We may imagine that the buildup
process occurs at an infinitesimal rate so that V · J = 0 holds to any desired degree
of accuracy. Then the current distribution can be broken up mto a network of
elementary current loops the typical one of which is an elemental tube of current
of cross-sectional area ,lO' following a closed path C and spanned by a surface S
with normal n, as shown in Fig. 5.20.
VIe can express the increment of work done against the induced emf in tenns
of the change in magnetic induction through the loop:
,l(8W) = J ,lO' ( n · 8B da
Js
/__..-- 1\1.</,\ \\\C~~

I
r-; "\~qj \ - -~~ \
\. \ l---~---\1 \
Figure 5.20 Distribution of current
~ / density broken up into elemental current
,oops.
Sect. 5.16 Energy in the Magnetic Field 213
wuctc Lllt: c11.ua u. ~y~ <> we; ate u vmy vu~

.. . .1 -1 .1
~"~""·
If we express B in terms of the vector potential A, then we have
ll(SW) = 1 Ll£T { (V x SA)· n da

co
UT:>L
,. .C n '
, •'- •'-. L.
"'"
.,
Af>;U!\- T A~~ >;A. r/1
/
Jc
but 1 Ll£T dl is equal to J d· x, by detimt10n, smce dl1s parallel to J. tvldently tne

sum over all such elemental loops will be the volume integral. Hence the total
.
u•~•
c
V>
.1.
vu• uv
~ 1.
VJ «»
•'- .1 --1. + "~
~
.,..,
>;,t.{,.\. +J..
vector potential is
SW = r SA· J d x 3
(5.144)
An ion · ·'- ino- thP

" m~anPtj,.. fields rather than J and SA can be
obtained by using Ampere's law:
n '· TT T
.,.,,
;;;w - ( >;A • ru "' J.n fi 3 r ( 'i 14-'i)

J
1ne vector taemny,

V · (P X Q) = Q · (V X P) - P · (V X Q)
can be used to transform (5.145):
SW = I [H · (V x SA) + V · (H x SA)] d x
3
(5.146)
· utirm i<
Tf thP fip],-1 ,-1." __, ..... h, 1.-....~1;7,r1 the second intc~>ral vanishes.
With the definition of B in terms of A, the energy increment can be written:
SW = I H · SB d x
3
(5.147)
This relation is the magnetic equivalent of the electrostatic equation ( 4.86). In

its present form it is applicable to all magnetic media, including ferromagnetic
substances. If we assume that the medium is para- or diamagnetic, so that a linear
. .
remuon exlSLS ucLwccn n anu n, Tntm
I I • UD
1 "
zU~ll. •
-
D)
If we now bring the fields up from zero to their final values, the total magnetic
energy will be
w=11H·Bd 3 x (5.148)
This is the magnetic analog of (4.89).

The magnetic equivalent of (4.83) where the electrostatic energy is expressed
in terms of charge density and potential, can be obtained from (5.144) by assum-
ing a linear relation between J and A. Then we find the magnetic energy to be
(5.149)
The magnetic problem of the change in energy when an object of perme-

ability p; 1 is placed in a magnetic field whose cur rcnt sources arc fixed can be
treated in close analogy with the electrostatic discussion of Section 4 7 The role
of E is taken by B, that of D by H. The original medium has permeability J.Lo and
existing magnetic induction B0 . After the object is in place the fields are B and
H. It is left as an exercise for the reader to verify that for fixed sources of the
field the change in energy is
W = -1 r (B · H0 - H • B 0 ) d 3x
2 J V1
where the integration is over the volume of the object. I his can be wntten m the
alternative form ·
w
Both g1 and u 0 can be functions of position, but they are assumed independent
of field strength.
If the object is in otherwise free space, the change in energy can be expressed
in terms of the magnetization as
(5.150)
It should be noted that (5.150) is equivalent to the electrostatic result (4.93),

except for sign. This sign change anses because the energy W consists of the total
energy change occurring when the permeable body is introduced in the field,
including the work done by the sources against the induced electromotive forces.
In this respect the magnetic problem with fixed currents is analogous to the elec-
trostatic problem with fixed potentials on the surfaces that determine the lields.
By an analysis equivalent to that at the end of Section 4 7 we can show that for
a small displacement the work done against the induced emf's is twice as large
as, and of the opposite sign to, the potential-energy change of the body. Thus,
to lind the force acting on the body, we consider a generalized displacement t;
and calculate the positive derivative of W with respect to the displacement:
(5.151)
The subscript 1 implies fixed source currents.

The difference between (5 150) and the potential energy (5.72) for a per
manent magnetic moment in an external field (apart from the factor~- which is
traced to the hnear relatiOn assumed between M and B) comes from the fact that
(5150) is the total energy required to produce the eonfigmation, whereas (5.72)
Sect. 5.17 Ener2v and Self- and Mutual Inductances 215
includes only the work done in establishing the permanent magnetic moment in
the field, not the work done in creating the magnetic moment and keeping it
permanent.
-
;:J. _l I
....
L!Jitt:rgy
-• .., •r • ll ·" . r T. •
UltU Jf:lj- UUU H.L.HUU • .LUUU'
A. Coefficients of Self- and Mutual Inductance

e e ~-
J USL as LflC ~pl 0< lLO UL T LUL <1 >J>LCOHI UL co
h"lcl l'lt
eli· lt p]. ic notential is useful (Section 1.11 ). the concept of
self- and mutual inductances are useful for systems of current-carrying circuits.
lmaQ:ine a svstem of N distinct current-carrving circuits, the ith one with total
current I,, in otherwise empty space. The circuits are not necessarily thin wires
~ tney can oe nus oars, etc. J om are assumeu tor Lnc p Lu ue
The total energy (5.149) in terms of an integral of J · A/2 can be expressed as
N N N
U!
1 )' T.T 2 + )' -.\;' MIT (5.152)
.,
2t=i_ i=l J>i
.u. ~- nf thP ith .--ir.--nit ~n.-1 ~If .. i< thP mntn~l in,~
"' '
T • ••
between the ith and jth circuits. To establish this result, we first use (5.32) for the
vector potential to convert (5.149) to
r r Tf~\ Tf~'\
w ~ J d x J d x' 3 3
lx- x'l
(5.153)
The integrals can now be broken up into sums of separate integrals over each
circuit:
"
W = f.Lo L J d 3 x, L J d 3 xj J(X;) ~ J~~i)
87T ,~1 1~ 1 Ix, Xj I
In the sums there are terms with i = j and terms with i =!= j. The former define
the first sum in (5.152), the latter, the second. Evidently, the coefficients L, and
M,, arc given by
f.Lo (
·'
( ,, J(x,) • J(x;)
/" "
L; JZ ), u X; ), u -'; ~J.LJ ' }
4 1T i c, c, lx;- x;l
and
Un I I Jlxl•J(x-l
M,i ), a-x), a·x- ~J.U.J)
47Tf;fj C; ' C; I lx;- xjl
Note that the coefficients of mutual inductance M,, are symmetric in i and j.
These general expressions for self- and mutual inductance are the rigorous
versions of the more elementary definitions in terms of flux linkage. To establish
the connection, consider the expression for mutual inductance (for which the
. . . . . . · r u · ,_ + \ rr• ~-
amU!guiLles 111 Lne '"" u< uuA 1u1 ,...-u-muu"'"""'"' m'"' '"~"'J·
integral over d 3x' times f.Loi47T is just the expression (5.32) for the vector potential
• -, • • · •' "+' " " -1 ' +L T ~ • • t~ ith
-"'~-"i} <ll 1-'VOlllUll Ai 111 nu ~H~~H JJ j 'D ·'
circuit. If the ith circuit is imagined to be negligible in cross section compared to

21ll _Cb ,,, c_5_ M< v. .... "'-L .., n. · "' · o_Ei,...... Jrr_
the overall scale of both circuits, we can write the integrand J(x 1) d 3 x for the
integration over th" volnme of th"' ith ,..;r,..nit ~o T ,J 3 T. A ,.~, .. ~ A ·
differential in the sense of current flow. With the vector potential sensibly con-
stant in the cross-sectional integral at a fixed position along the circuit, the mutual
inductance becomes
1
M I} = -] . ] . · I '
, 1
fc cl.
A I] · d! = ~
].
1
f sl.
(V x A)
'J · n da
where Ar is the vector potential caused bv the ith circuiLat the inteP'rlltion noint
theorem has been used to obtain the second form. Since the curl of A is the
magnetic induction B, the area integral is just the magnetic-flux linkage (5.133).
--'I'], .il ~ 1 • _, •
~ ••-v " ' U >HU • 1~
(5.156)
" " " ' " , ij i~

magnetic uux 1rom ctrcun 7nnKea wttmn cucmt 1. !'or selt-tnduc
Ule
tance, the physical argument is the same, but the ambiguity in the meaning of
the self-flux linkage F11 requires a return to the rigorous expression (5.154) based
on the magnetic energy.
1:: • . '
,,_
.,
"''-'
ol ,..]
.---
,Jr.
= ,..]
u eilller me conuucwrs carrymg the current are permeable or the medium be-
...
'-'"'-'~
. . '"'" lltal.
tween the conductors is (J.L *

J.Lo), (5.152) is valid, but (5.153) is not. It is then
best to use the expression (5.148) for the magnetic energy in terms of the fields
on the left-h:mrl ~irlf' of(<; 1 'i?) in """' '"" nf th. . ·'. ,. ] .
~ .. _ 1'" v• ~u"" ,., L u11ut u1 1v1 12 u1 2 1(u In tue vonage oatancc m
lumped circuit equations follows immediately from relating the time derivative
of the linked flux (dF/dt) to the induced emf jg through (5.135).
B. Estimation of Self-Inductance for Simple Circuits

The self-inductance of simple current-carrying elements can be estimated by
ration 01 Lne magn~c energy. :suppose a cuc01ar wire of cross-sectional
radius a carrying a steady current I forms a loop of circumference C and "area"
A (the quotation marks remind us that, since the loop may not be planar, A may
stand for a oroiected area). We imll<>inP th"t thP.lnnn thnJHrh_r,..JgtiuP!~· ·-'-'
size. An example IS sKetcnea m 1<1g. 5.21. There are three length scales here
the wire radius, the dimensions of the loop, represented by C/2w or A l/2, and the
outside region, r >> C/2w. From (5.152), the relation between the self-inductance
(5.157)
e, "''-' ,.::Hgtu ~ca,e 01 tHe wire raums, we may tgnore tne curvature and consider
Sect. 5.17 Energy and Self- and Mutual Inductances 217
Diameter= 2a
"Area"= A
Circumference= C
Figure 5.21 Closed current-carrying

circuit made of a wire of radius a, length
C, and (projected) area A.
the field inside and outside the wire as if it were straight and infinitely long. If
the current density is uniform throughout the interior, from symmetry and
Ampere's law (5.25) the magnetic induction is azimuthal and equal to
where P< (p>) 1s the smaller (larger) of a and p. We have assumed that the wire
and the medium surrounding it are nonpermeable. The contributions to the in-
ductance per unit length from inside the wire and outside the wire, out to a radius
Pmax, a e
f.Lo
81T'
The radial integral outside the wire is limited top < Pmax because the expression
fur Bq, fails to represent the magnetic induction at distances of the order of the
mtddlc length scale. If we look to the interior of the loop, it is clem that for
p = 0( C/27r) - O(A 112 ) the isolated straight wire is a very poor representatiOn
of the current pattern. Thus we expect* Pmax = O(A 112 ). There is, of course, a
contribution to the inductance from the outside region at distances beyond Pmax·
There, at distances large compared to A 112 , the slow falloff of the magnetic in-
3
duction as 11 pis replaced by a dipole field pattern with IB I = O(f.L 0 m/47Tr ), where
m = O(IA) is the magnetic moment of the loop of wire. Because of the rapid
decrease of the field beyond Pmaxo the contribution per unit length to the induc-
tance from large distances (i.e., p 2: A 112 ) can be estimated to be
dLct;pole
dl =0 ( 471'
C J~ r 2 (f.L 0 /A/47Tr 3 ) 2 dr ) = O(f.LoA 2 /47rp;,axC)
3
2
JLof Pmax
If we set Pmax = (t''A) 112 , where t'' is a number of order unity (containing our
ignorance),
dLct;poie = O(f.LoA 112/47rC)

dl
*If the circuit shape is such that A << C 2 , as for an elongated loop, a different estimate of Pmox may
be appropriate [e.g., Pmox = O(A!C)J.
zu; Chapter 5 Ma!!netostatics FaradaY's Law. o .... d.Stati<- · ,sr_
a contribution of order unity compared to the logarithm above. Upon combining

the different contributions, the inductance of the loop is estimated to be
I gA_ \ 1
L = f.Lo C lnl-1 +- (5.1'iRI
~7T a _io '
nere we nave cxmottea tne mtcnor contnbution explicitly and indicated the
uncertainty in the proper value of Pmax and the size of the exterior contribution
through the number g, of order unity.
Four comments: First, if the wire has a magnetic permeability f.L, the interior
contribution becomes~~ JLI2f.Lo· Second, for a thin wire bent in a circle of radius
large com oared to the wire radius " nrPri<P. "" k11lMion ( <PP 1-'1 ) p , 'i ":\7) ohn
th<Li ~ I'. A I 4 (l "1"7"2 _'I'l. d c. · L • -L
0 • •u ., ~~~ - ,
m5u - lU lll<tl llle ~.
skin depth of the wire is small compared to its radius, the interior contribution
is absent because the current is confined to near the surface of the wire (see next
section). Fourth, if the single turn of wire is replaced by a tight coil of N turns,
with the effective cross-sectional radius of the bundle being a, the self-inductance
.• Jll_2 • •L . -1.
u~v ~~~ •v.
Exercise
Consider a circuit made up of two long, parallel, nonpermeable, circular wires of
radii a 1 and a2 , separated by a distance d large compared to the largest radius.
'--'uucom 11vw~ up one wire anu uacK along tne otner. 1gnore the ends. Use the
__mefuruj l'lhovr to <how thl'lL1he. ,.]L;nrlnrt~nrP nPr" ciLl ·•1 · · ~1
ll (va;a;
r ~ ~ ~
dL
dl
f.Lo
= -:;; n
gd )
+ 41J
where g is of order unity. Can you find a reliable value of g, within the approxi-
>UUUVH~
5.18 Quasi-Static Magnetic Fields in Conductors;

Eddy Currents; Magnetic Diffusion
_':_l:l_e_111_agnewsrarics or me I1fS!__!_'I_ sections ot thts chapter are based on Amoere 's
],.w "nrl thP <>h<PnrP of "'"''"·'t. ,J. 'b' • A · Q • "1" H:c •'- _
~---..
<Hv >UU5 Jo~J,
netic induction varies in time, an electric field is created, according to Faraday's

law; the situation is no longer purely magnetic in character. Nevertheless, if the
time variation is not too rapid, the magnetic fields dominate and the behavior
can be called quasi-static. '_'Quasi-static" rcfers.lo.lhe__re_cim for whirh the..:finite_
. rl ~f J;nht
~o
}
v
,J •• rl __, "'-'--'
·~
__, :c ,,
u~ u ""·_J_ l'
ms.y. c.aiu in Olller, e4mvarem woras 1t ts tne regtme where the svstem is
__Sll1a_]}_ rom with __the_ • 1. • ~, · "'~ ,..,L -n.n1h r1 · • 1- • 1-, d.
nant time scale of the problem. As we learn in subs;quent chapters, such a regime
permits neglect of the contribution of the Maxwell displacement current to
Ampere's law. We consider such fields in conducting media, where Ohm's law
Sect. 5.18 Quasi-Static Magnetic Fields in Conductors; Eddy Currents; Magnetic Diffusion 219
relates the electric field to the current density and so back to the magnetic field
via the Ampere equation. The relevant equations are
V X H- J, V · B- 0 ' V X E + - - 0, J = lTE (5.159)

at
With B = V x A, Faraday's law shows that the curl of E + aA/at vanishes. As a
result, we can write E aAfat V $. With the assumption of negligible free
charge and the time-varying B as the sole source of the electric field, we may set
the scalar potential = 0 and have E = -aA/at. Note that we have the subsidiary
conditions, V · E = 0 and V ·A = 0. For media of uniform, frcquency-indepen-
dent permeability J.t, Ampere's law can be written V X B = J.tJ = J.tlTE. Elimi-
nation of B and E in favor of A and use of the vector identity, V x V X A =
V(V ·A) - V2 A, yields the diffusion equation for the vector potential,
aA
V2 A = J.tlT- (5.160)
at
This equation, which obviously also holds for the electric field E, is valid for
spatially varying, but frequency-independent IT. If the conductivity is constant in
space, it follows that the magnetic induction B and the current density J also
satisfy the same diffusion equation.
The structure of (5.160) allows us to estimate the time rfor decay of an initial
configuration of fields with typical spatial variation defined by the length L. We
put V2 A O(i'•JL 2 ) and f!Aif!t O(AIY). Then
Alternatively, (5.161) can be used to estimate the distance L over which fields
exist in a conductor subjected externally to fields with harmonic variation at
L o(vk) (5.162)
For a copper sphere of radius 1 em, the decay time of some initial B field inside
is of the order of 5-10 milliseconds; for the molten iron core of the earth it is of
the order of 105 years. This last number is consistent with paleomagnetic
data the last polanty reversal of the earth s held occurred about 106 years ago;
there is some evidence for a decline to near zero about 5 X 104 years ago and a
rise back to its present value.
A. Skin Depth, Eddy Currents, Induction Heating

A simple quantitative illustration of the fields described by (5.160) is afforded
by the situatiOn shown m Fig. 5.22: A semi mhmte conductor of umform con-
ductivity IT and permeability J.t occupies the space z > 0, with empty space for
z < 0. The surface at z 0 is subjected to a spatially constant, but time-vat ying,
magnetic field in the x direction, Hx(t) = H 0 cos wt. We seek a steady-state so-
lution of (5.160) for z > 0, subject to appropriate boundary conditions at z 0
and finiteness at z ~ + oo. Continuity of the tangential component of H and the
z
Figure 5.22 At the surface of a semi-infinite conducting permeable medium, a spatially
constant magnetic field, H.(t) = H 0 cos wt, is applied parallel to the surface at z = o-.
A localized magnetic field and current flow exists within the medium in the region
z < 0(8).
an x component,
function of z and
Because the diffusion equation is second order in the spatial derivatives and
Thus, the boundary value on H, is Hx = H0 e-""', where taking the real part is
understood. The steady-state solution for Hx(Z, t) can be written
Hx(Z, t) = h(z)e-;wt
where, from (5.160), h(z) satisfies
A trial solution of the = eikz leads to the
i5 = I2
VJ.UTW
(5.165)
For copper at room temperature (cr- 1 = 1.68 X 10-~ il · m), i5 = 6.52 X

w-z;~ m, where v = w/27r. For seawater, i5 = 240/~ m (see Fig. 7.9
Sect. 5.18 Quasi·Static Magnetic Fields in Conductors; Eddy Currents; Magnetic Diffusion 221
Since the field varies in time, there is an accompanying small electric field.
From Ampere's and Ohm's laws, together with the existence of only Hx(z, t), we
find that there is only a y component of E, given by
1 dHx -1 + i H e-z /8 elZ. 18 -tw!
.
E =--=
Y (J dz ifo o
Taking the real part and writing 1/ifo = p.,8w/2, we have

p.,w8
Ey = \12 H 0e-z1" cos(z/8 - wt + 37T/4) (5.167)
To compare the magnitude of the electric field and the magnetic induction, we
form the dimensionless ratio,
E ./cuH - ()( wS!c\ << l
hu
·o
th~ nn•xoi.ot ~•i~
-.-
lllill lliC llCIU.
Th~ Ji~lrlo
-· ~
UlC llClU "

.' .
~r~
.. ~ '•L
', ...
lv m>~unPtir with "
u
1.
-" ~
-~~ ~UH~H<
rlPMitv (for z > 01.

,""1)
T - ~T'<
·y
- 1-l.o- z/8 c-r><f7lf> ,,,, + ,,./4\ !'i.168\
8
mhr><P · -' in 7 i< "n o •<l~C'l
u
Ky(t) = r ly(Z, t) dz = -H0 cos wt
For very small skin depth, the volume current flow in the region within 0(8) of
the surface acts as a surface current whose magnitude and direction is such as to
reduce the magnetic field to zero for z >> 8. See Section 8.1 for more discussion
relevant to waveguides and cavities.
There is resistive heating in the conductor. The time-averaged power input
per unit volume is Prcsislivc = (J • E) (recall P = IV = V 21R in a simple lumped
resistor circuit). With (5.167) and (5.168), we find
1
presistive
= - p.,wH6e-2z/8 (5.169)
2
The heating of the conducting medium to a depth of the order of the skin depth
is the basis of induction furnaces in steel mills and of microwave cookers in
kitchens (where the conductivity of water, or more correctly, the dissipative part
of its dielectric susceptibility, causes the losses-see Fig. 7.9). References to more
• • e • • ' ' ' · L ' " >L .--1
ldlC Vl cuuy \-UllCH" auu wu• M~H H~UUH6 ~
ot the chanter.
~.
.
JJ';JJ' UJ
r.,
11
.
<fl' r~·
<ICIU» ' ' <n
,n .J. .
......... ...• , • .J:
~~~
unrusion or magneric neJUs in conuucting utcu;a ""u uc ldlCU WHll lllC

c.
,J., "f tw" infinitP 1 I <'llTTPnf . oarallel to each other and
d : L.U d}'<tl l, dl Z -u a11u' L. u,
'•L •
UH
. -"
medium of nermeabilitv u and conductJvttv if. Lne currents are sucn tnat m tne
region, 0 < Iz I < a, there is a constant magnetic field H 0 in the x direction and
7Prt> fiPlcl ""'";r1., ~ ,. ;,~, '" ,, r1 · • · · •• _,. · "
' -- y .~ .. y
T TT I 0
Jy '"OL U\.(. "! U\'G uI J
At time t = 0, the current is suddenly turned off. The vector potential and mag-
netic field decay according to (5.160), with variation only in z and t. We use a
Laplace transform technique: Separate the space and time dependences by
writilli!..
roo
TT / ,, I -vt"i:r
"X\'-• 'J Jo ~ "\F' '-I up
Substitution into the diffusion equation (5.160) for Hx leads to the wave equation,
2 2 2
(d /dz + k )h(p, z) = 0, where k 2 = 1urp. Since the situation is symmetric about
z = 0,
. the appropriate. solution ish cc cos(kz). With a change of variable fromp
w K 1u~orm mregraJ, 11x~Z, t)_tJ_ecomes
H,(z, t) = r 2
e-k ' 1"'"h(k) cos(kz) dk (5.170)
The coefficient function h(k) is determined by the initial conditions. At t = o+,

the magnetic field is \
r
11x~Z, U ) J 0
fl(k) cos(kz) dk Hal (")(z + a) ®(z a)] (5.171)
where ®(x) is the unit step function, t)(x) = 0 for x < 0 and ®(x) = 1 for x > 0.
Exploiting the symmetry in z, we can express the cosine in terms of exponentials
. and write
1 I
I fl!kle •+ dk Hcllt'!!z + a) ®17 nlI l:'i.17?)
L J oo
where h( -k) = h(k). Inversion of the Fourier integral yields h(k),
h(k) = --;;
H0 f" a e ''Z
.
1 2H0
dz = 'TI"k sin(ka) (5.173)
ThP. fr-.r the> morrne>t;~ /ie>lrl ot •>IL±i . ' {'\ . .. £.
r ,
·o
' '
2H0
Hx(Z, t) = ---;;- 0 sin K cos
e-vtK -K- r(~Z) KJdK (5.174)
2> 1 • . .
"' ·~·~ •
I
..":':'_ "' 1 10 raLe [See ~.J.JOLIJ. 1ue mregntLln
a 1-uar
]J.li'+J canoe expressed as the sum of two term' "~"h irlP-ntifiNl with~ rf>nrf>-
onn ,f ... ~-
<Peg) = - 2
y:;;: r 0
'
e-x' dx = -2
71'
r 0
e x'l4g2 Sin X
X
dx (5.175)
1 mo IvouH 10
Ho '1 + lzlLa_ '1 - 171/n\

11x(Z, t) ljJ I + <PI I 15.17())
L LV vt 2Vvt /
,.,,.,,.,
\..>11. " . . . .J
1.2 I I I I I I I I I I I
1 n
/ \.
0.8 -
I t = 0.05
\
-
<.;
3 ~ 0.6 - -
~ •
/
0.4 -
v " '\
-
...- ~ ~
t=4
-- --
0.2 t- -
...........
...- ./
I
/
L./
I
I
I
\
'
\
'-
.......... ....,_
0.0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
zla
Figure 5.23 Magnetic field distributions given by (5.172) for vt < 0, and (5.176) for
vt = 0.05, 1, 4 as a function of z/a. The outward diffusion with time for t > 0 is
manifest; the rise and fall of the field in time at a fixed position can be noted for
1<17l/a<7.
To understand qualitatively the meaning of the solution we note first that

<P(-g) =-<Peg), second that <Peg)~ 1- (1/y:;;:)[1- 1!2g" + · · ·Jexp(-g") for
g---> oo, and third that ( g) = (2g/y:;;:)(1 - g"/3 + · · ·) for Igl << 1. For vt ~ 0,
thP. lP.nt< in (<; 171'\) ~rf'. hraP in m~anitnrlp· thP snlntinn nhvinnslv -'
" ' I ~I ~·
tv tw:; ut;ut-uauu_~uc ut ~J.ll"-j, u~ ICI.[UHCU. rut tuut; tuuc~ U£"1 ~--- i),
Hx(z, t) ~ H 0 1VTrvt, independent of jzj/a to leading order in an expansion m
11-v;:;f. This result is misleading, however, because the coefficients of the higher
terms in llvt are z-dependent. A more revealing result is obtained by expanding
the error functions in Taylor series in 112-v;:;f to the third order. The result is
TT
T-1(7 t\ "O -lzl'14••taL 1 _,_ '

i rl7 2 n,tn 2 1\ _,_ . " (<; 177\
X , '
0
,-
,; ~"' 1?vt
Note that the approximate expression vanishes as vt ---> 0, as it should for any
lz I >a, and goes to Hx = H 0 /YTWt for vt >> lz l12a. For lz Ita< 5, it is within
a few percent for any vt > 1. At a given position, the field as a function of time
• I? . ? ' • • ' ~~'' - " '
ua~ a '" Vt I"' I I "-U l CAU\..l lVl lllC "!-'!-'" LllUll ~.l,l/1 )j, lUIIUWCU
hv thte vterv <low >P ~S r · l / L "i23 shows the snatial rlistrihutions of

..,
M\.. llClU dl llACU lllll\..,,
'""'

rro01ems or sreaay-srare current now In an . 1 CMMI V<O 1 are
to electrostatic notential nroblems with the current densitv rcnlacin£ the disnlacement
224 Chapter 5 Ma~netostatics. Faradav's Law Onasi-Static Fi.,Jd•- -ST
and the conductivity replacing the dielectric constant. But the boundary conditions are
generally different. Steady-state current flow is treated in
Jeans, Chapters IX and X
Smythe, Chapter VI
MaQnetic iPirl< rlnP. to rli<trihpt;Ano onrl h mrl· ., - ol ~ 1-1
'
in magnetostatics are discussed, with numerous examples, by
Durand, Chapters XIV and XV
Smythe, Chapters VII and XII
The atomic theorv of ma<>netic · ohtlu folk in thP rl nf
,], c. .. _, ·''
'
'~1.
' 6~ ~J
Abraham and Becker, Band II, Sections 29-34

Durand, pp. 551-573, and Chapter XVII
Landau and Lifshitz, Electrodynamics of Continuous Media
Rosenfeld, Chapter IV
more ue,ai,eu rrearmems or tne suoject, mcmumg crystallographic aspects and the topic
of ferromagnetic domains can h" fonnrl in
Chakravarty
Craik
Kittel
Ziman, Chapter 10
• c . '
~]
" . cu "'~ ona;;ucou~__l:"_':'':":""' Ol auoy' auu lJllll;'l pr L llllOT·
ma1ion as weu as me meory IS
R. M. Bozarth, Ferromagnetism, Van Nostrand, New York (1978).
Numerical methods in magnetism are discussed in
Ida and Bastos
~uuu
and m the other references cited in the References and Suooested Readin<> at thP. enrl of
Chapter 2.
Faraday's law of induction is somewhat tricky to apply to moving circuits such as
Faraday discs or homopolar generators. There are numerous discussions of these ques-
tions. A sampling from one journal comprises:
D. R ror<nn Am 1 Phys 24 1?1i (19~1i)
D. L. Webster, Am. J. Phys. 31, 590 (1963)
E. M. Pugh, Am. J. Phys. 32, 879 (1964)
P. J. Scanlon, R.N. Hendriksen, and J. R. Allen, Am. J. Phys. 37, 698 (1969)
R. D. Eagleton, Am. J. Phys. 55, 621 (1987)
A o-onrl Af tJ. ,f ~ <.
'££
:oo '] 'b
~
•J 'D HUH> <10 0 u:~~v~vUl UU~o, ~0 0 nvU LJY

Stratton, Chapter IT
Inductance calculations and forces are handled lucidly by
At ,]. rl o. ~·· :, .~.
, D. ,,-IT , r-1 , TTT
auu w <Uauy lt;X".
-=
, ... ... - ..~
cup;~ LH couuy ~utremS anu inuucrion neanng IS UlSCUSSeu Wl(ll
many examples in
Smythe, Chapter XX
l\!1 ' flp]rl,· in ~Anrln,
flnirl< nr~ nlou • e •n ol nl ., , ~J.
,_ .... ,-1, •'-
r ·:o
.£
- r -,
~" «vHU< pwuvco "' cuv u~

. r wu '"'"' euces are
"'· "· rarKer, Losmzcat!Vlagnetzc l'retas, uxrord Umvers1ty Press, Uxford(1979).
Ya. B. Zel'dovich, A. A Ruzmaikin, and D. D. Sokoloff, Magnetic Fields in
Astrophysics, Gordon & Breach, New York (1983).
Ch. 5 Problems 225
Problems
5.1 Startmg wiflfT!feC!llTerentJal expression
X X
JLol "'
u~
41T u• lx- x'l3
for the magnetic induction at the point P with coordinate x produced by an incre-
ment of current I dl' at x', show explicitly that for a closed loop carrying a current
I the magnetic induction at P is
B = JLol vn
41T
W<Ol~
~·
H
..
10 UO<O ,
, .. •s•
~ '-
"]
•'- 1.
"Y
,,_
. ,. n T'-'. ~-
to a magnetiC scalar potential, JLol!L/'+1T. 1ne s1gn convennon ror LUe sotiu
'l'M
onoiP ie thot () ic• nm-;t'vp views the "inner" side of the surface soan-
if thP noint p
ning the loop, that is, if a unit normal n to the surface is defined by the direction
of current flow via the right-hand rule, n is positive if n points away from the point
P, and negative otherwise. This is the same convention as in Section 1.6 for the
electric dipole layer.
5.2 A long, right cylindrical, ideal solenoid of arbitrary cross section is created by stack-
ing a large number of identical current-carrying loops one above the other. with N
coils per unit length and each loop carrying a current/. [In practice such a solenoid
could be wound on a mandrel machined to the arb1trarv cross sect1on. Arter tne
:1 , X ' -~I. ;,~ "\ th .,hPI ounoolcl hP ' 1
'6' ·~·· -r -u•
(a) In the approximation that the solenoidal coil 1s an wear current sheet and
infinitely long, use Problem 5.1 to establish that at any point inside the coil
the magnetic field is axial and equal to
H= NI
and that H = 0 for any point outside the coil.
(b) For a realistic solenoid of circular cross section of radius a (Na >> 1), but still
munne m 1englll, suow t11a1 u1e sm uwgu~uc u~IU JU" uov
solenoid faveraoed axiallv over several turns) is not zero but IS the same m
" " ••• '" _·-~·
,]. tl·
" 0 "
1, even u Jvu --7 . LU"1f'U1<0 H<OWO H101Uv UUU UUL
5.3 A right-circular solenoid of finite length Land radius a has N turns per unit length
and carries a current /. Show that the magnetic induction on the cylinder axis in
the limit NL--> oo is
JLuNl
B z = --(cos
2 (J
1
+ cos (J,)
-
c
W ll~l t; LlJl;
' '"~ H1 UOv "C,Ulvo
:
8.1
! •••••••••••••.•••• ~
I
I --- ./1-82 ;
Problem 5.3
5.4 A magnetic induction Bin a current-free region in a uniform medium is cylindrically

symmetric with components B,(p, z) and BP(p, z) and with a known B,(O, z) on the
. . • . c . c • • . • • .
ax" ut "'Y· t
HUU~ Ul lUv UAlUl HvlU ··y '-•
az z)J
[a B,(O,
2 2
p )
B,(p, z) = B,(O, z) - ( + ···
4 2
s p(p, z) - rI ~p) aB,(O,

.
z) + 1 p 3 \ o3 B,(O, z)
, ;1 + ...
(b) What are the magnitudes of the neglected terms, or equivalently what is the
criterion defining "near" the axis?
5.5 (a) Use the results of Problems 5.4 and 5.3 to find the axial and radial components
of magnetic induction in the rontrnl rua;,., ( 171 << L/21 of ,_.J, 1o " ·
lllC ~Ull \f-' · U }·
(b) Use Ampere's law to show that the longitudinal magnetic induction just out-
side the coil is approximately
~,,p u , <.J
For L >> a, the field outside is negligible compared to inside. How does this
axial component compare in size to the azimuthal component of Problem
5.2b?
(C) ~now tnat at tne end of the solenmd the magnetic induction near the axis has
components
f.LolVl
B, = -2-, B = ±f.LoNI
P 4
(£!.)
a
5.1! A cylindrical conductor of radius a has a hole of radius b bored oarallPl tn onrl
-" "' · ,J •< • "'" "' ·• u ' ,\ •r, .
uniform throughout the re~aini;~ metal of th~· cylinder ~~d is parallel to the axis.
Use Ampere's law and principle of linear superposition to find the magnitude and
the direction of the magnetic-flux density in the hole.
5.7 A compact circular coil of radius a carrvin" a current I(, rh N tmn•· Porh ..,;,"
r -o· ·
(a) By elementary means [Eq. (5.4)] find the magnetic induction at any point on
the z axis.
(b) An identical coil with the same magnitude and sense of the current is located
H ·. -"-' " "' ' · ·
cooruina,e origin relocated at tne pomt m!Ciway between tne centers of the
two coils determine the ma<>netic ;nrlnrt;nn on the axis....neaL.1he_m;o;n "' ""
expansion in powers of z, up to z4 inclusive:
2 2
-if.Lula \ 3(b - a2 )z 2 15(b4 - 6b 2 a2 + 2a 4 )z 4 •••
B' - d3 -)
1+ 2d 4 + 1<id 8 +
(c) Show that, ott-axis ncar the origin, the axial and radial components, correct
to second order in the coordinates, take the form
I
.J. r 2 1!. J•
p
2/'
Ch. 5 Problems 227
(d) For the two coils in part b show that the magnetic induction on the z axis for
large Iz I is given by the expansion in inverse odd powers of Iz I obtained from
oHv OHWH < . .
vo p a H v v y uov
• .
,u
I
I~ I·
(e) If b = a, the two coils are known as a pair of Helmholtz coils. For this choice
of geometry the second terms in the expansions of parts b and d are absent
t n · crL "'_.., .,_ · · · . .,_ "- ""
\ vz J• • • v J• • uv ·~ ~u• '"' '5' • •J ~ • "'· • HUo oo <Hv
maximum permitted value of Iz IIa if the axial field is to be uniform to one
part in 104 , one part in 102 ?
5.8 A localized cylindrically symmetric current distribution is such that the current flows
only in the azin;_uthal direction; the current density is a function only of rand IJ (or
p and z): J = <)>J(r, e). The distribution is "hollow" in the sense that there is a
current-tree regwn near the ongm, as well as outside.
(a) Show that the magnetic field can be derived from the azimuthal component
" ' lli<:O .vovo jJUl<:0HlH11 0 Willi d ..,.
A .(r A)- - !1-o )' m. rLPUro' Al

47T /.
in the interior and
A<t(r, IJ) = - !1-o

4 7T L ILL
L ,-t.-1p1 (cos
L
IJ)
outside the current dJstnbutwn.

(b) Show that the internal and external multipole moments are
m
L
= -
1
L(L + 1)
Jd x 3 ,-L-tpt (cos
L
IJ) J(r I!)
,
dHU
1
u,
L~L -r I)
r d 3X YL p) (COS IJ) ]( T
J
IJ)
5.9 The two circular coils of radius a and separation b of Problem 5.7 can be described
" ' v -" v v v ""'
J''"~"vuo <Hv v~ ~vHOHJ
J = <f>i8(p - a)[8(z - b/2) + 8(z + b/2)]

\ilJ U>'H!; lli<:O Ul r J.O, " " lli<:O dllU llllu-
tipole moments for L = I, ... , 5.

ih \ T T.-' ,,_ of U, ohl. ~Q rl. -" ; ,, '
"0 •': .,: . ,
an expression for Bz on the z axis and relate it to the answer of Proble; 5.7b.
5.10 A circular current loop of radius a carrving a current I lies in the x-v plane with its
"' thP. oriain
(a) Show that the only nonvanishing component of the vector potential is
A.p(p, z) =tJ-ulil
- 1c dk cos kzl1 (kp<)K 1 (kp>)
7T 0
where P< (p>) is the smaller (larger) of a and p.

(b) Show that an alternative expression for A<t is
r~
Aq,(p, z) = t::',~'u Jo dk e-klziJ1 (ka)J1 (kp)

228 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields SJ
(c) Write down integral expressions for the components of magnetic induction,
using the expressions of parts a and b, Evaluate explicitly the components of
8 on the z axis by performing the necessary integrations.
5.11 A circular loop of wire carrying a current I is located with its center at the origin
of coordinates and the normal to its plane having spherical angles 80 , ¢ 0 There i~
an appli@d rnagn@tic fi@ld, Bx B 0 (l + f3y) and By B 0 (1 + [jx).
(a) Calculate the force acting on the loop without making any approximations.
Compare your result with the approximate result (5.69). Comment.
(b) Calculate the torque in lowest order. Can you deduce anything about the
higher order contributions? Do they vanish for the circular loop? What about
5.12 Two conc@ntric circular loops of radii a, b and curr@nts I, I', msp@ctiv@ly (b < a),
have an angle a between their planes. Show that the torque on one of the loops is
about the line of intersection of the two planes containing the loops and has the
magnitude.
N
flo7Tll' b
2
f (n + 1) r(n + 1) 2;L.a\2n
I~ l P~ ,,(cos a)
La n-0 ~Ln "T 1) _1 ~n "T L)l \2) a;
where Pi( cos a) is an associated Legendre polynomial. Determme the sense ol the
torque for a an acute angle and the currents in the same (opposite) directions.
5.13 A sphere of radius a carries a uniform surface-charge distribution a. The sphere is
rotated about a diameter with constant angular velocity w. Find the vector potential
and magnetic-flux density both inside and outside the sphere.
5.14 A long, hollow, right circular cylinder of inner (outer) radius a (b). and of relative
permeability ll, is placed in a region of initially uniform rnagn@tic flu)[ density 8 0
at light angles to the field. Find the nux density at all points in space, and sketch
the logarithm of the ratio of the magnitudes of 8 on the cylinder axis to 8 0 as a
function of log 10 fl, for a2 /b 2 = 0.5, 0.1. Neglect end effects.
5.15 Consider two long, straight wires, parallel to the z axis, spaced a distance d apart
and carrying currents I in opposite directions. Describe the magnetic field H in
terms of a magHetie seahn potential $ M, with II V$ M·
(a) If the wires are parallel to the <: axis with positions, x - + d12, )' - 0, show
that in the limit of small spacing, the potential is approximately that of a two-
dimensional dipole,
<f>M =
where p and rf> are the usual polar coordinat@s.

(b) The closely spaced wires are now centered in a hollow right circular cylinder
of steel, of inner (outer) radius a (b) and magnetic permeability f1 = fl,flo·
Determine the magnetic scalar potential in the three regions, 0 b. Show that the field outside the steel cylinder is a two-
dimensional dipole Geld, as ill part a, but with a strength reduced by the factor
F
a
Relate your result to Problem 5.14.
(c) Assuming that f1, >> 1, and b =a+ t, where the thickness t << b, write down
an approximate expression for F and determine its numerical value for
f.L, = 100.
Problem 5.15
5.16 A circular loop of wire of radius a and negligible thickness carries a current l. The
loop is centered in a spherical cavity of radius b > a in a large block of soft iron.
Assume that the relative permeability of the iron is effectively infinite and that of
the medium in the cavity, unity.
(b) What is the radius of the "image" current loop (carrying the same current)
to a semi-infinite slab of material having relative permeability f.L, and filling the half-
· induction can be calculated the

medium of permeability f.L, by an image current distribution, J*, with
f.L, -1)
( f.L, + 1 fx(x, y, - z),
f.L, -1)
( f.L, + 1 fy(x, y, -z),
1)
- ( f.L ' - J (x v -z)
f.L, + 1 z ' . '
(b) Show that for z < 0 the magnetic induction appears to be due to a current
5.18 A circular loop of wire having a radius a and carrying a current I is located in
f.L· Find the force acting on the loop when
Determine the · form of answer to . a and b when d >> a.

Can you obtain these limiting values in some simple and direct way?
5.19 A magnetically "hard" material is in the shape of a right circular cylinder of length
L and radius a. The cylinder has a permanent magnetization M 0 , uniform through-
out its volume and parallel to its axis.
(a) Determine the magnetic field Hand magnetic induction Bat all points on the
axis of the cylinder, both inside and outside.
(b) Plot the ratios BIJL0 M 0 and H/M0 on the axis as functions of z for L/a = 5.
5.20 (a) Starting from the force equation (5.12) and the fact that a m<tgnetization M
inside a volume V bounded by a surface S is equivalent to a volume current
density JM = (V X M) and a surface current density (M x n), show that in
the absence of macroscopic conduction eurrents the total magnetic ftJI ce on
the body can be written
F [.rv (V · M)B, d'x +[ JS

(M · n)B, da
where B, is the applied magnetic mductwn (not mcludmg that of the body in
question). The force is now expressed in terms of the effective charge densities
PM and uM. If the distribution of magnetization is not discontinuous, the sur-
face can be at infinity and the force given by just the volume integral.
(b) A sphere of radius R with uniform magnetization has its center at the origin
of coordinates and its direction of magnetization making spherical angles 110 ,
</> 0 • If the external magnetic field is the same as in Problem 5.11, use the
expression of part a to evaluate the components of the force acting on the
sphere.
5.21 A magnetostatic field is due entirely to a localized distribution of permanent
magnetization.
(a) Show that
provided the integral is taken over all space.

(b) From the potential energy (5.72) of a dipole in an external field, show that for
a continuous distribution of permanent magnetization the magnetostatic en-
ergy can be written
W = ~0 I 3
H ·H d x = -~0 I 3
M ·H d x
apart from an additive constant, which is independent of the orientation or

position of the various constituent magnetized bodies.
5.22 Show that in general a long, straight bar of uniform cross-sectional area A with
uniform lengthwise magnetization M, when placed with its fiat end against an inti-
nitely permeable Rat surface, adheres with a force given appwxiliJately by
Relate your discussion to the electrostatic considerations in Section 1.11.

5,23 A right circular cylinder of length l and radius a has a uniform lengthwise mag
~~~
Lb. !:l l'robtems ~J~
(a) Show that, when it is placed with its flat end against an infinitely permeable
plane surface, it adheres with a force
K(k) E(k) K(k1) E(k 1 )

F = 2J.L0 aLM2 I<
-
I<"
'
?n
k k, "
\14a + U'2
Ya + 2
e
(b) Find the limiting form for the force if L >> a.
5.24 (a) For the perfectly conducting plane of Section 5.13 with the circular hole in it
and the asymptotically uniform tangential magnetic field H 0 on one side, cal-
culate the added tangential magnetic field H<'l on the side of the plane with
H 0 . Show that its components for p > a are
n l I\ 2H0 a 3 xy
X
7f p'\ll- a2
I
IJ<'l = LH 0 a y + no a I a· . -I
·~
~ ~1
---Sill
Y 7f o4 \l o 2 - a7 7f {J2 ,{J
(b) Sketch the lines of surface current flow in the neighborhood of the hole on
both sides of the plane.
5.25 A flat nght rectanglliarToop carrymg a constant current 11 IS pTacca near a lOng
strairrht wire carrvin<r a current 17 • The loop is oriented so that its center is a per-
pendicular distance d from the wire; the sides of length a are parallel to the wire
and the sides of length b make an angle a with the plane containing the wire and
the loop's center. The direction of the current I 1 is the same as that of 12 in the side
of the rectangle nearest the wire.
(a) Show that the interaction magnetic energy
w - ( J ·A d 3 x - 1 F
J
(,], co 10' '" '" 0 >tir flnv frnm L lin lcinrr thP xonhr rirrnit . ·w
I 1), is
J.Loi1Iza In 4d
2
+ b2 + 4d b cos a
w, '+U u cos a
'+7f L't(/"TU
~b) CalculatetlfeTorceoetween me wop ana me wire ror nxerrcuncm>.

,_,
,_, n. .1 .I ,,: f, . 1. xf ro.1;no mhA«o ~1o~o ;o ~· JlJp]
"1" '"
.
'1.
-r r
to the w1re and makes an angle a with respect to the plane containing the
center or me wop ana ll!e wire. o.uow l!laL '"" inL~•· ent:rgy "
?
W12 /Lo'J'z u · "'" 1~ v~ u /U I
J:<ina tneTorce.
ld\ D, h.x<h 1~~ xe mhPn .J >> n h thP i
eh~"' <lxot inn rPdn<'PO (()
X" ·r
W 12 = m · B, where m is the magnetic moment of the loop. Explain the sign.
'i.21i A twn-wire transmission line consists of a oair of nonoermeable narallel wires of
radii a and b separated by a distance d > a + b. A current flows down one wire
Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-SI
and back the other. It is uniformly distributed over the cross section of each wire.
Show that the self-inductance per unit length is
L = ~0 1 + 2 In
41T
5.27 A circuit consists of a long thin conducting shell of radius a and a parallel return
wire of radius b on axis inside. If the current is assumed distributed uniformly
the ·
where
and E are the complete elliptic integrals.
5.29 The figure represents a transmission line consisting of two, parallel perfect conduc-
tors of arbitrary, but constant, cross section. Current flows down one conductor and
per unit length Cis

LC =
where~ and e are the permeability and the permittivity of the medium surrounding
the
5.30 (a) Show that a surface current density K( q,) = I cos ¢!2R flowing in the axial
ular to the cylinder axis. Show that the field outside is that of a two-dimen-
(b) Calculate the total magnetostatic field energy per unit length. How is it divided
1s per the system, viewed as a long circuit

with current flowing up one side of the cylinder and back the other?
Ch. 5 Problems 233
5.31 An accelerator bending magnet consists of N turns of superconducting cable whose

con ent configutation can be described appr oxirnately by the axial cuuent density
J,(p. ck) = (~~)cos ck 8(p - R)
The right circular current cylinder is centered on the axis of a hollow iron cylinder
of mner radms R (R > R). The relative dimensions (R, R a few centimeters and
a magnet length of several meters) permit the use of a two-dimensional approxi-
mation, at least away from the ends of the magnet. Assume that the relative per-
meability of the iron can be taken as infinite. [Then the outer radius of the iron is
irrelevant.]
(a) Show that the magnetic field mstde the current sheath IS perpendicular to the
axts of the cylinder in the direction defined by ¢ = ± 1r12 and has the
(b) Show that the magnetic energy insider = R is augmented (and that outside
diminished) relative to the valnes in the absence of the iron (Compare part h
of Problem 5.30.)
(c) Show that the inductance per nnit length is
5.32 A circular loop of mean radius a is made of wire having a circular cross section of
radius b, with b << a. The sketch shows the relevant dimensions and coordinates
for this problem.
Problem 5.32
(a) Using (5.37), the expression for the vector potential of a filamentary circular
loop, and appropriate approximations for the elliptic integrals, show that the
vector potential at the point P near the wire is appt oximately
A, = (~t 0 I/27T)[ln(8a/p) - 2]
where p is the transverse coordinate shown in the figure and corrections are
of order (p/a)cos ¢and (p/a) 2 •
(b) Since the vector potential of part a is, apart from a constant, just that outside
a straight circular wire carrying a current I, determine the vector potential
inside the wire (p <b) in the same approximation by requiring continuity of
~-
"'0 mm
aensuy mswe tne w1re:
Aq, = (p,0 J/47T)(1 - p% 2 ) + (!-'1,J/27T)[ln(8a/b) - 2], p m" 1 h ·~- thl'~ind
-
L = p,0 a[ln(8a/b) - 7/4] /
Are the corrections of order b/a or (bla?? What is the change in L if the
currem IS assumea to 11ow only on the sur ace ot the Wire (as occurs at hiQh
frequencies when the skin depth is small compared to b)?
5.33 Consider two current loops (as in Fig. 5.3) whose orientation in space is fixed, but
whose relative separation can be changed. Let 0 1 and 0 2 be origins in the two
1 <; cd al ' --< ·'- < ~ ~ L ..<.' .L
-y o JjJo ·~ ~~ UHU ~2 V~ ~VV<" oO V< lllO Ulj a!lU
m 2 , respecuve1y, or rne loops retcrred to the respective ongms. Let K be the relative
coordinate of the origins, directed from loop 2 to loop 1.
(a) Starting from (5.10), the expression for the force between the loops, show that
it can he written
F12 = l 1l 2 VRM12(R)
where M 12 is the mutual inductance of the loops,
u In\ 1-Lu ,1, ,1, m1 · a1 2

12\"/
47T 1 1 I x 1 - x2 + RI
and it is assumed that the orientation of the loops does not change with R.
(b) Show thM the mntn"l · et. "P "< " fnn~t;nn nf R '• 0~'"'.
,,
the Laplace equation,
V;',M12 (R) = 0
1 ne impor:r_ance or this result IS that the umqueness ot so utwns o the Laplace
equation allows the exploitation of the properties of such solutions, provided
a solution can be found for a particular value of R.
5.34 Two identical circular loops of radius a are initially located a distance R apart on
u u~•o p~• l'~ LU lll<Oll
(a) From the expression W 12 = J d 3x J 1 • A 2 and the result for Aq, from Problem
5.10b, show that the mutual inductance of the loops is
r~
1VJ12 p,0 1ra 1 uK e JJ\KaJ
•u
(b) Show that for R > 2a, M 12 has the expansion,

3 5 7
/.Lo1Ta fa ,1a \ /) a
"
'"12 .., lo -'I ol
...
Q D I
\ 'I -
(c) Use the techniques of Section 3.3 for solutions of the Laplace equation to
snow _lllliL_Lile muLUal Inductance tor two coplanar JdenlicaJ_ Circular loops of
radius a whose rpn(f>TS "ri' hv" rl' OP"R > ?o i··
Ml2 =
/.Lo7Ta
---
4
(a' R
3
+ ~ a)
4R
1 5
+ 375 ( !!:__
64R
7
+ ...
Ch. 5 Problems 235
(d) Calculate the forces between the loops in the common axis and coplanar con-
figurations. Relate the answers to those of Problem 5.18.
5.35 An insulated coil is wound on the surface of a sphere of radius a in such a way as
to produce a uniform magnetic induction B 0 in the z direction inside the sphere
and dipole field outside the sphere. The medium inside and outside the sphere has
a uniform conductivity IJ and permeability IL·
(a) Find the necessary surface current density K and show that the vector poten-
tial describing the magnetic field has only an azimuthal component, given by
B 0 a 2 r< .
A.p = - - 2 sill e
2 r>
where r < (r >) is the smaller (larger) of rand a.
H UJll JlUW UJl, j VV lUI LllC "b' UL VCII > U<OJ" ~uu ~m, cu~ u~~uy
or rne magneuc neta 1s aescrioea oy me anruswn equation, ~J.tUUJ. u>iug a

Laplace transform and a sphencal Bessel functwn expanswn (J.J U ), show
that the vector notential at times t > 0 is Qiven by
·'·'· . I kr
~ Sill tJ In e . ]t~tc)}l\ n ux.
• ·' r r~t" and i,(x) is the · -1 Bessel

•• •• . "~lrl . th
Bz(O, t) = B0 <P(, ;-;;---':) - , ;-----. expl - - )

L v 4vtj v rrvt 4 vt
(c) Show that the total magnetic energy at time t > 0 can be written
Wm = 6B~a3JOO e 2vtk'[j,(k}f dk
!L {)
Show that at long times (vt >> 1) the magnetic energy decays asymptotically
as
(d) Find a corresponding expression for the asymptotic form of the vector poten-
tial (at fixed r, e and vt ~ oo) and show that it decays as (vt)- 312 as well. Since
the energy is quadratic in the field strength, there seems to be a puzzle here.
Show by numerical or analytic means that the behavior of the magnetic field
at time tis such that, for distances small compared to R = a( vt) 112 >> a, the
field is uniform with strength (B0 /6rr 112 ) (vt)- 112 , and for distances large com-
pared toR, the field is essentially the original dipole field. Explain physically.
5.36 The time-varying magnetic field for t > 0 in Problem 5.35 induces an electric field
and causes current to flow.
(a) What components of electric field exist? Determine integral expressions for
the components of the electric field and find a simple explicit form of the
current density J = ~JE at t = o+. Compare your result with the current density
of Problem 5.35a. Find the asymptotic behavior of the electric fields in time.
(b) With Ohm's law and the electric fields found in part a, show that the total
power dissipated in the resistive medium can be written
p = 12B~a'v (~ e-zvtk'[kj,(k)f dk
fL Jo
Note that the power is the negative time derivative of the rhagnetzc energy,
wm.
(c) Because of Ohm's law, the total electric energy is We = E 0P/2a. The total
energy is the sum of We and W m; its time derivative should be the negative of
the power dissipation Show that the neglect of the energy in the electric field
is the same order of approxiination as neglect of the displacernent current m
the equations governing the magnetic field.
PTER6
• '
onserva zan aws
In the preceding chapters we dealt mostly with steady-state problems in electric-

ity and in magnetism. Similar mathematical techniques were employed, but elec-
tric and magnetic phenomena were treated as independent. The only link
between them was that the currents that pro uce magnettc e s are astca y
electrical in character, being charges in motion. The almost independent nature
0
the independence. Time-varying magnetic fields give rise to electric fields and
magnetic fields. The full import of the interconnection between electric and mag-
netic fields and their essential sameness becomes clear on! within the framework
of special relativity (Chapter 11). For the present we content ourselves with ex-
aminin the basic henomena and deducin the set of e nations known as the
Maxwell equations, which describe the behavior of electromagnetic fields. Vector
tromagnetism. Conservation laws for energy and momentum and transformation

. .
topic of magnetic monopoles.
.. .
6.1 Maxwell's Displacement Current; Maxwell Equations

The basic laws of electricity and magnetism we have discussed so far can be
summarized in differential form by these four (not yet Maxwell) equations:
COULOMB'S LAW V •D = p
RE'S LAW
ABSENCE OF FREE MAGNETIC POLES V·B = 0

...
tions. Consequently, from a logical point of view there is no a priori reason to
237
238 Chaoter 6 Maxwell Eouations Macroscooic F.J, ...
d. d_ '" ,_ - u ' ' ' '
'1" ,, UW< <H'-' '-''-jUUUVHO n u HUH.< lUl lllllC-UC\- ;Ill llClUS,
ln tact tne euuatJons m set (o.l) are mcons1stent as thev stand.

, . . . . ,. (6 1)' -r d d~Jf h ·: . '
to see th e mcons1stency m equations . an to mo 1 y t em mto a consistent
set that implied new physical phenomena, at the time unknown but subsequently
verified in all details by experiment. For this brilliant stroke in 1865, the modified
set of equations is justly known as the Maxwell equations.
The faulty equation is Ampere's law. It was derived for steady-state current
phenomena with V · J = 0. This requirement on the divergence of J is contained
right in Ampere's law, as can be seen by taking the divergence of both sides:
V · J = V · (V x H) =0 (6.2)
While V · J = 0 is valid for steady-state problems, the general relation is given
by the continuity equation for charge and current:
v ... ' u (tU)

Ut
VVlliil lYliiXWCll ~iiW Wii' llliil lllC C4 ~\CUllVCllCU !ll{O a

vanishino diveroence bv usino Ccmlomh's l:c~w lA 1\ Thn<
V·J+-=V·
ap I
J+-j=O
AI>
(6.4)
at at
Then Maxwell replaced J in Ampere's law by its generalization
ai>
J~J +-
at
for time-dependent fields. Thus Ampere's law became
V X H = J + -
aI> (6.5)
At
et;ll th~ e~~~ <~lh, •~r;fi~rl 1· f, -• rl· •+ -• 1- ·'
' -r ' -.~ r '
ni<eaHy <eonsis,en, wiLu llle conLinuiLy equation ( O.j) wr time-uepcnacm
current the converse of Faradav's law. This ncccssarv ,..-J.-Jiti.-m to A mni'.rP.'s hw

"r •r "J "b , • •v . " '"'-'~'-' ""
no electromagnetic radiation, and the greatest part of the remainder of this book
would have to be omitted. It was Maxwell's prediction that light was an electro-
magnetic wave phenomenon, and that electromagnetic waves of all frequencies
could be produced, that drew the attention of all physicists and stimulated so
much theoretical and experimental research into electromagnetism during the
last part of the nineteenth century.
The set of four equations,
:Sect. fi.~ vector and :Scalar l'otentmls ~-'':1
known as the Maxwell equations, forms the basis of all classical electromagnetic
phenomena. When combined with the Lorentz force equation and Newton's sec-
ond law of motion, these equations provide a complete description of the classical
dynamics of interacting charged particles and electromagnetic fields (see Section
6.7 and Chapters 12 and 16). The range of validity of the Maxwell equations is
discussed in the Introduction, as are questions of boundary conditions for the
normal and tangential components of fields at interfaces between different me-
dia. Constitutive relations connecting E and B with D and H were touched on
in the Introduction and treated for static phenomena in Chapters 4 and 5. More
is said later in this chapter and in Chapter 7.
The units employed in writing the Maxwell equations (6.6) are those of the
preceding chapters, namely, SI. For the reader more at home in other units, such
as Gaussian, Table 2 of the Appendix summarizes essential equations in the
commoner systems. Table 3 of the Appendix allows the conversion of any equa-
tion from Gaussian to SI units or vice versa, while Table 4 gives the corresponding
conversions for given amounts of any variable.
6.2 Vector and Scalar Potentials

The Maxwell equatiOns consist of a set ot coupled first-order partial differential
.... ,1. +I. .•. .• '1 rl · r..,u. rr1-
·~]
can oe s01veu as mey swnu m simple snuanons. nm n IS onen convemem to

·lnrP 1ti ~ 1< , r.ht~inin a " <m<> IIPr .L r.f ..J_r.rr1Pr ir.n<. ,
.~ . . . . . ~
'"uMymg , Ul Llle IVliiXWell eqUiiLIUIIS •u...,dlii.OiiiiY. VV e <tre a1reauy 1a-

miliar with this in Pl ic" :mel m!'l 1tirs WP llSP.cJ thP
• I. · .1 rh ..1 >1. . .1 •
"'u' = uu~ '"~ •~~•u• pu' ~.
Smce V • B - 0 still holds we can define B in terms of a vector potential:

R-VxA {(, 71
Th +h ,.,_ ~
.... (?; ?;\
, .. ·n
n. ..1.
'J
,
.
1.
, ~ ; ..
vx (E +iJA)
- =0
iJt
(6.8)
This means that the quantity with vanishing curl in (6.8) can be written as the
gradient of some scalar function, namely, a scalar potential <1>:
iJA
E+-= -Vel>
at
or (6.9)
iJA
E = -Vel>--
at
The definition of B and E in terms of the potentials A and ci> according to (6.7)
and (6.9) satisfies identically the two homogeneous Maxwell equations. The dy-
namic behavior of A and ci> will be determined by the two inhomogeneous equa-
tions in (6.6).
At this stage it is convenient to restrict our considerations to the vacuum
240 Chapter 6 Maxwell Equations, Macroscopic Electromagnetism, Conservation Laws-SI
form of the Maxwell eguat10ns. I hen the mhomogeneous eguat10ns m (6.6) can
be '.vritten in terms of the potentials as
V2 <P + at (V ·A) = -p!E0 (6.10)

2
V2 A - -12 -a A - V( V · A + -12 -a<P) = - f.LoJ (6.11)
c 2
at c iJt
We have now reduced the set of four Maxwell equations to two equations. But
they are still coupled equations. The uncoupling can be accomplished by exploit-
ing the arbitrariness involved in the definition of the potentials. Since B is defined
through (6.7) in terms of A, the vector potential is arbitrary to the extent that
the gradient of some scalar function A can be added. Thus B is left unchanged
by the transformatiOn,
A~ A'- A+ VA (6.12)
For the electnc field (6.9) to be unchanged as welL the scalar potential must be
simultaneously transformed,
iJA
<P ~ <P' = <P - - (6.13)
at
The freedom implied by (6.12) and (6.13) means that we can choose a set of
potentials (A, <P) to satisfy the Lorenz condition (1867),*
1 iJ<P
V ·A +-
2 - = 0 (6.14)
c iJt
This will uncouple the pair of equatiom; (6.10) and (6.11) and leave tvw inho
mogeneous wave eguat10ns, one for <P and one lor A:
piE0 (6.15)
(6.16)
Equations (6.15) and (6.16), plus (6.14), form a set of equations equivalent in all
respects to the Maxwell equations in vacuum, as observed by Lorenz and others.
6.3 Gauge Transformations, Lorenz Gauge, Coulomb Gauge

The tran~fonnation (6.12) and (6.13) is called a gauge It ansfm mat ion, and the
mvanance of the fields under such transformations is called gauge invariance. To
see that potentials can always be found to satisfy the Lorenz eondition, suppose
that the potentials A, 1 a2 A
V ·A +2- - = 0 = V · A +2- - + V
2
A- -- (6.17)
c at c at c2 iJt 2
*L. V. Lorenz, Phil. Mag. Ser. 3, 34, 287 (1867). See also p. 294.
Sect. 6.3 Gauge Transformations, Lorentz Gauge, Coulomb Gauge 241
r
A r~n hP fn11nrl to ~atisfv
(6.18)
the new potentials A , cp will satisfy the Lorenz condition ano tne wave equatwns
(6.15) and (6.16).
Even for potentials that satisfy the Lorenz condition (6.14) there is arbitrar-
iness. Evidently the restricted gauge transformation,
A -c> A-t VA
();\ (6.19)
<l>-c><P--
at
where
vzi\ - .!.. azi\ = o

2 ' 2
(6.20)
preserves the Lorenz condition, provided A, <P satisfy it initially. All potentials
in this restricted class are said to belong to the Lorenz gauge. The Lorenz gauge
is •onlv 11sP.ci first hP.rmJsP it 1P~c1~ t(l the wave eauations (6.15) and r6.16).
U~U. = UHY ''l' ""5'' .,.
independent of the coordinate system chosen and so fits naturally inta the con-
siderations of special relativity (see Section 11.9).
Another useful gauge for the potentials is the so-called Coulomb, radiation,
~L • • >L · ..1. . • L
V ·A= 0 (6.21)
From (6.10) we see that the scalar potential satisfies the Poisson equation,
p!Eo ~U.LL)
with solution,
1 r o(x', t)
<v~x, t) 47TEo J lx - x' I a x ~ U.L.J)
The scalar potential is just the instantaneous Coulomb potential due to the charge
density p(x, t). This Is the ongm of file name "CouTomD gauge.
The vector potential satisfies the inhomogeneous wave equation,
1 il<P
-uoJ + -:; V - (6.24)
c dt c '"
The term involving the scalar potential can, in principle, be calculated from
(6.23). Since it involves the gradient operator, it is a term that is irrotational, that
• J.. ._,_. -1 'T'J.." <h ·• ;, ,] ~ -"nn n;P~P nf
Ul<O '-'UlliOlll UIOII'ILJ· I 1110 '-'UlliOlll U\Cll'ILY ~UJ <IllY •<V< u~<YJ ~uu v~ ""'~" u,
the sum of two terms,

J = J, + J, (6.25)
.. , ·~ tne wngiruainat or irromcional current a na--n as v x J 1 u,

while J, is called the transverse or solenoidal current and has V • J, = 0. Starting
from the vector identity,
VX(VXJl VIV•J) V'"J (I)?())
together with Y'2 (11 Ix - x' I) = -4mS(x - x'), it can be shown that J 1 and J, can
be constructed explicitly from J as follows:
_ --
J,- 1 V J
V' • J d 3 X , (6.27)
41T lx- x' I
• 1 ...., .., r J _73 I
I I 'I u A
I I
With the help of the continuity equation and (6.23) it is seen that
1 ii<l>
c 01
Therefore the source for the wave equation for A can be expressed entirely in
terms of the transverse current (6.28):
(6.30)
This. of course is the ori!!in of the na~ ''tran . """""" Th" nmne_ '-'radia_-
• " £_ •'-
H• uvetuwuuu <llCO IOIVCOII uy LIICO
vector potential alone, the instantaneous Coulomb potential contributing only to

the ncar fields. This gauge is particularly useful in quantum electrodynamics. A
y_ -mecuanical aescription or pnotons necessitates quantizatiOn ot only the
vector potential.
The Coulomb or transverse gauge is often used when no sources are present.
Then = 0, and A satisfies the homogeneous wave equation. The fields are
cJ,
iJA
E = --
iJt (6.31)
In passing we note a peculiarity of the Coulomb gauge. It is well known that

electromagnetic disturbances propagate with finite speed. Yet (6.23) indicates
<L <L ,1_ ._," • • •
• ·~ ... ~ aw• P' "~' "" JUMJ eve1ywuere 111 space. 1ue
vector_Q_otentlal on the other hand ~Mi~fies_ th" w"v" (!l '()) _wi_tb__its_
implied finite speed of propagation c. At first glance it is puzzling to see how
obviously unphysical behavior is avoided. A preliminary remark is that it is the
fip]rk not thP. ;~lo th~t ~n ·~~ Ll. ' 1 ~• ,,· • • <L <L
v~:rs~: """ ~"' ~ n.Li"\) exrenas over au space, even It J IS localized."'
*See 0. L. Brill and B. Goodman, Am. J. Phys. 35. 832 (1967) for a detailed discussion of causality
m tne ~ouwmu gauge. :-.ee also l'roblem o.20.
Sect. 6.4 Green Fnnctions for the Wave Equation 243
6.4 Green Functions for the Wave Equation

The wave equations (6.15), (6.16), and (6.30) all have the basic structure
uz,y, 1 cP'¥ A. +f, t\ ({:. C!')\

cz i:Jt2 • J \" ' 'I \' .. -,
where f(x, t) is a known source distribution. The factor cis the velocity of prop-
~umtion in the medium here to he ut ··on.
To solve (6.32) it is useful to find a Green function, just as in electrostatics.
We consider the simnlc situation of no boundarv surfaces and nroceed to remove
el, ,J; . rL -1. ~ -'· 1:', t ,f,
••
;.~
frequ~~cy. We su;pose that-;'l'(x, t) and f(x, t) have the Fourier i~tcgral
r .
representations,
'l'(x, t) = - 1 'l'(x, w)e-'w' dw
r .
27T -oo
rr:. 'n\
. I
'
f(x, t) = -1 f(x, w)e-'w' dw
27T 00
with the inverse transformations,
roo
'l'lx w) - I 'l'lx t)e'w' dt
J-oo
(6.34)
£1.
J \~, ~,
.\ r
I.
£1 .
J\~
.,\ iwt ,.../,
• • ,~ ~·
When the representations (6.33) arc inserted into (6.32) it is found that the
FnnriPr £
\j!(y ro~\ '"tj,fip, thP ii"J.,, w.< l-TPimhnlt7 wnvP Pnuntinn
(V2 + k 2 )'1'(x, w) = -47Tf(x, w) (6.35)
for each value of w. Here k = wlc is the wave number associated with frequency
w. In this form, the restriction of no dispersion is unnecessary. A priori, any
connection between k and w is allowed althou2:h causalitv imooses some restric-
tions (see Section 7.10).
Equation (6.35) is an elliptic partial differential equation similar to the
Pni"nn Pfl11<>tinn tn urhirh it rPr1HrP' fnr I. () ThP ~ c. inn (';(y , v 1 \
• 1
. .
appl OpllaLt;: LU ~ O . .J.J) SallSHeS Llle I ey1
(\7 2 + 1.2 \(';(y v'\ = -4'7TI)(Y - y'\ ((,1(,\
If there are no boundary surfaces, the Green function can depend only on R =
x - x' and must in fact be snhcricallv svmmetric that is denend onlv on
In I c. .C <J.. T .1. ,].. •1
"" r,
·~
D
·r -,
(3.d, itis evident th~t G"(R) sati;fies
l'
1 A2
R dR 2 (RGk) + k 2 G" = -4m5(R) (6.37)
Everywhere except R = 0, RGk(R) satisfies the homogeneous equation

2
with solution,
Furthermore, the delta function in (6.37) has influence only at R -> 0. In that
limit the equation reduces to the Poisson equation, since kR << 1. We themforc
know from electrostatics that the correct normalization is
. 1
hm Gk(R) =- ( 6.38)
kR~o R
The general solutwn for the Green function IS thus
(6 39)
e::!::.ikR
with A + B = 1. With the convention of (6.33) for the time dependence, the first
term in (6.39) represents a diverging spherical wave propagating from the origin
while the second represents a converging spherical wave.
The choice of A and B m (6.39) depends on the boundary conduwns zn tzme
that specify the physical problem. It is intuitively obvious that, if a source is
quiescent until some time z 0 and then begins to function, the appropriate
Green function is the first term in (6 39) corresponding to waves radiated out-
ward from the source after it begins to work. Such a description is certainly
correct and also convenient, but is not unique or necessary. By suitable specifi-
cation of the wave amplitude at boundary times, it is possible to employ the
second term in (6.39), not the first, to descnbe the actwn of the source.
To understand the different time behaviors associated with Gi+l and Gk-l
we need to const1 act the concsponding time-dependent Green functions that
satisf
4wll(x x')ll(t t') (6.41)
Using (6 34) we see that the wurce tenn for (6.35) is
The solutions are therefore Grl(R)e'wt'. From (6.33) the time-dependent Green
functions are
(6.42)
2wJ-oo R
where T t t' is the mlative time appearing in (6.41 ). The infinite-space Green
function is thus a function of only the relative distance R and the relative time
Sect. 6.4 Green Functions for the Wave Equation 245
'T" ~onrcP. ~nrl

' •tion noint Fnr ~ nnn, ,. ivP mPrlinm ~ ~'
k = wlc, the integral in (6.42) is a delta function. The Green functions are
I
R'
Gl"l(R, r) =~a T
-
+-
c
(6.43)
or, more explicitly,

I
r Ix-x 'I I\
a( t' t+
c
l i -'~X, t; X , I) ~ 0.4'1)
lx- x'l
The Green function a<+ lis called the retarded Green function because it exhibits
the causal behavwr assocmtcd w1th a wave dtsturbance. The argument ot the
delta function shows that an effect observed at the point x at time tis caused by
the action of a source a distance R away at an earlier or retarded time, t' =
t - Ric. The time difference Ric is just the time of propagation of the disturbance
from one point to the other. Similarly, G' -'IS called ttl e advanced <ireen tunctwn.
Particular integrals of the inhomogeneous wave equation (6.32) are
'lf(")(v t) - ( ( r;(ec)(v t· ""' t')f(y' t') rl'r' rlt'

J J
To specify a definite physical problem, solutions of the homogeneous equation

may be added to either of these. We consider a source distribution f(x', t') that
is localized in time and space. It is different from zero only for a finite interval
. . ..
U1 lllllt: il1UU11U l U. lWU ~ill il1C t:ll' • 111 Lilt: 111M ll "
assumed that at time t ---> - oo there exists a wave '1'1n(x, t) that satisfies the
homogeneous wave equation. This wave propagates in time and space; the source
turns on and generates waves of its own. The complete solution for this situation
. . .
at au nmes 1s evlGenuy
'l'(x, t) = 'l';n(X, t) + II a<+l(x, t; x', t')f(x', t') d 3 x' dt' (6.45)
The presence of G' ~~ J !!Uarantees that at remotely earlv times t before the source
has been activated, there is no contribution from the integral. Only the specified
wave '1'10 exists. The second situation is that at remotely late times (t---> + oo) the
wave is given as 'If out(x, t), a known solution of the homogeneous wave equation.
Then the complete solution for all times is
'l'(x. t) = wout(x. t) + II a<-l(x. t; x·. t')f(x'. t') d 3 x' dt' (6.46)
1\.T r r\, rh rl r:. ,., .. ; • rJ.,~r nA c;~n~l i'rAm t\,~

·o·
ch~ 11
exist explicitly after the source shuts off (all such signals are by assumption in-
eluded in 'If out)·
Th" 1onP.~t nhv~ic~l ~itnntion is described bv (6.451 with 'I'~
;.~, ·~
,., ...
0. It is
AA\
' . .,
{I'. . .l
..,..
-1: ; .....
'J.
'If( ) _
X, t -
I [J(x', t')]ret d3 ,
Ix-x 'I X (6.47)
i
246 Chapter 6 ''

Maxwell Equations, Macroscopic Electromagnetism, Conservation Laws-SI
The square bracket [ ],e, means that the time 1' is to be evaluated at the retarded
time, 1' = t- lx- x' I!c.
The initial or final value problem at finite times has been extensively studied
Ill one two and three dimensions The TP.~nP.r m~v rd"r tn ~J~n·o nnd Po<ld. ol.
(,
\l"t''
QA'O QA7\ " • 1 <L <L • ll
'"~ mv•~ VL" 1 mu.u~•
·~·
0 . .) .neraraea ;:}Otutwnsjor the J<tetas: Je}lmenko's &eneraltzations

of the Coulomb and Biot-Savart Laws; Heaviside-Feynman
Expressions for Fields of Point Charge
Usc of the retarded solution (6.47) for the wave equations (6.15) and (6.16) yields
<P(x, t) =
4
Jd x' R1 [p(x', l')],et
1
1TEo
3
(6.48)
A(x, t) = :; Jd x' ~ [J(x', l')],et
3
whterP we h~vP R v v' "'ith R I 'I

I
,rl fl ol. ,.\ o
'J
D/D
'T'L ,1 ~ ' 1. ' '
' \
""''-'woe 5"'-'" uy LML'-'UL \vp. lell.). 111 jJlllllCljJlC, lfOm tueSe tWO
u~•~-.~ '
equatiOns tne electnc and magnetic fields can he .. to.l hnt it i~ nftPn .,f.. l
+. ]., _, -' .1 .1. £ ' <1. ~ "
w• uo~ "~'u' Hl
•~sw•
'
Vl LHC 'lJUll-C>.
.caner U!recuy rrom me Maxwell equatwns or by use of the wave equations
for <P and A, (6.15) and (6.16), and the definitions of the fields in terms of the
potentials, (6.7) and (6.9), we can arrive at wave equations for the fields in free
space with given charge and current densities,
1 iiE
2 --
VE - = _ _!_ ( -Vp _ _!..2 iJJ) (6.49)
cz iJt2 Eo c iJt
and
1 a2 B
'1728---= -p.. 0 V X J (6.50)
c2 at 2
~L , • • ,
"''-' nu•~ lUi CdLll Ul Lilt: \...dlLCS!an ueiQ COmponen(S IS Ill me IOfffi
~yuctlllJll
(6.32_1. The retarded so]ntinn< (6.47) for..1heJif'lc1~ ""n immeciiMPlv hP · _in_

+1. ·" 'J
£.
'
1
E(x, t) = __!:__ J d 3 x' -V'p- 1 iJJ (6.51)
47TE0 R c? at' ;et
and
B(x, t) = /l.oo
41T
Jd x' lR [V' X net
3
(6.52)
These preliminary expressions can be cast into forms showing explicitly the static
limit. .d.±L ' +!.. L ' Ll. '-' ' -' • •
· ~J _ _._. "''-' 'I pmuu• U'-''"u'"'-''
1rom llle re1araea mtegranas. 1nere 1s a subtlctv here because V'l fl.. *I V'fl. .
ThP mP~nina ,f '<7' rl. +1. rl .-1 L ,], ' • 1
"' '" ~ ' ' '""' '
'+L
-o u "r 0
llA'-'u, "'"' mcaning outsiuc Lne re1arueu macKeLIS a spatial gradient w1tn respect
: for thP FiPIII• 247
to x', with x and t fixed. Since [f(x', t')].et = f(x', t - Ric), it is necessary to
r "'~ _, " ~' · ' i _•~ n •• ;~·
' - V H ' - ' - ' <V< "''-'A -r , <HUVUU'-'-U UHVUt;H VJ-
is taken outside. Explicitly, we have

R 'a
V'(l - Ric) = V'[p]Let - r ri;, (6.53)
Let ret
and
fl:'' v Tl
' "
- u· "m
'- '
-1-
-.::r.J'
at' Let
')( V'(t R/r)
-
(6.54)
1 aJ
V' X [J]Let + - -::--; xR =
"' = Let
If these expressions arc substituted into the preliminary forms of the solutions
and an integration by parts is performed on the first (gradient or curl) term in
"""h ""~P. wf'. "ffiV<' at
_
E(x, I ) -
1
41TEo
I d
3
X
, { R
R2
, ,]
[p(x ' t ) cet + cR
R [iJp(x',
at'
t')J Let
(fi 'i'i)
__1 [aJ(x',
c2 R iJt'
t')J }
Let
and
B(x, t) =:;I dV {[J(x', t')]ret X : 2 + [iJJ(~;: t')tt c~} X (6.56)
If the charge and current densities are time independent, the expressions reduce
to the familiar static expressions (1.5) and (5.14). The terms involving the time
,J, . •1fi"PS nnrf fhP btinn . ' thP ~ JiocatiOnS tO time-d icnt
OVU''-'-0• •uvov C •v ' ' - O U H 0 0 OV '"" b'
the Coulomb and Biot-Savart laws, were popularized in this author's text,
(Iefimenko).
In n~eeino ""' nnt0 thgf '- •oP thP iniPnrgn<f•· ~rP. tn hP ;lS fnnrtinns
r '
of x, x', and t, with t' = t - lx - x' Ilc, the time derivatives in the integrands have
the property
iJf(x', t') iJ • "
' LJ~X' t }JLet
This relation facilitates the specialization of the 1efimenko formulas to the Heav-
ISide -i"eynman expressiOns tor the ncrrrs or a poim cnarge. wim p~x , r J
qo[x' r 0 (t')] and J(x', t') - pv(t'), (6.55) and (6.56) specialize to
~ -
R a R iJ v '
E =-q- -----:;; +- - --, - (6.58)
KIC ~ret Let
Let ~ ""
q1TEo <:lJr L UL
and
' '
1-'41 vXK a VXK 'f ~n
D \ u.J7)
2
417 KR Let cat KR ret
:l41'S Chapter 6 Maxwell Equations, Macroscopic Electromagnetism, Conservation Laws SI
Here R is the distance from the position of the charge to the observation point·
H. ts a umt vector tram the charge toward the observation point; vis the charge's
velocity; K = 1 - v · R/c is a retardation factor. [Sec Problem 6.2.] It is important
cv uvc.,; waL now were is a uiuerencc Detween dL· · ·J,e1/iJI ana LiJ · · · /iltJcct because
x' --? r 0 (t'), where r 0 is the position of the charge. The fields are functions of x
~ .. ~ ., wuu ,' , I"' •o~' J l'c., "Y"H"'n s expression ror Lnc cJectnc neld Is
E -_ _ q
47TEo
{fR.J
R cet
+ [R]ret.!!_ rR.J + _.!.._ [R'] cet }
2
c at R cet c at
2 2 2 (6.60)
while Heaviside 's expression for the magnetic field is

_u_ ,a_ -->L-X v 1 _;)_
B (6.61)
47T K
2R2
ret
+ c [R Jret at K
1ne eqmvalcnce ot tne two sets or expressions for the fields follows from some
>4'n1 ~l ~L
6.6 J.Jerivation of the Equations of Macroscopic Electromagnetism

The discussion of electromagnetism in the preceding chapters has been based on
the macroscopic Maxwell equations,
.., n
- {)
~
-
T
~
~
iJB
at
v
_an_ (6.62)
v. u p VXH J
at
' " " ' " .... dllU . . " " ' lllC HWv' vscopic eiecuic ana magneuc ncm quantities, u and
Hare corr<"nondin<> "· · "fi,]rl< ·' -' •~ 1i' ~ cl R ,~, ,~, ,~, ·" · ·
P and the magnetization M of the material medium by
D = E0 E + P, H='B-M (6.63)
f.Lo
Similarly, p and J are the macroscopic (free) charge density and current density,
' - t. , l<L ...L • • •
'"''J· ""'""usu '""''"' "''iuctuu11S are ,aml!Iar anu .o,any acccpm01e, we
ha~e yet to present a serious derivation of them from a microscopic starting point.
T} . "' ,+: . ,.,- " . +t . . • . .
-J ...... v.~ , '" . " ' " uvuvdUUil remalllS WlLulll
a classical framework even though atoms must be described quantum mechani-
.calhL ThP ~ .t . ~ .. jJ, .±J, - -'- . J_
. .
d1scusswn I I 11 I ,,h I . I • '). h ,- 'h"'u"~- .. ,1
c ose y para c s t e c ass1ca one, wit quantum-mec amca expecta-
tion ·' rP.nhrin<> thP. d~ssir.al C:"~nt;t;M ;n t} f, .I " ,1. • T\.
at 1J!e end of the ch<ill.ter.

We consider a microscopic world made up of electrons and nuclei. For di-
mensions 1arge comparea to llJ m, the nuclei can be treated as point systems,
as can the electrons. We assume that the equations governing electromagnetic
L wr mesc poim cnarges are the nucroscoptc Maxwell equations,
... L rJIJ
,,
V •e = ry/E0,
Sect. 6.6 Derivation of the Equations of Macroscopic Electromagnetism 249
where e and b arc the microscopic electric and magnetic fields and TJ and j are
the microscopic charge and current densities. There are no corresponding fields
d and h because all the charges are included in TJ and j. A macroscopic amount
of matter at rest contains of the order of 1023 " 5 electrons and nuclei, all in in-
cessant motion because of thermal agitation, zero point vibration, or orbital mo-
LiOn. 1 IIC miCrOSCOpiC ereCLrOmagneLi<.: HelUS r -~ u y '"'-'".. '-"a' 1!,00> v '" y
extremely rapidly in space and in time. The spatial variations occur over distances
of the order of 10 10 m or less, and the temporal fluctuations occur with periods
ranging from w-n s for nuclear vibrations to 10- 17 s for electronic orbital motion.
JY!acroscopic measuring aevices generally average over imervars iu space auu
time much larger than these. All the microscopic fluctuations are therefore av-
~ or · · -~1.,r; .~1., '' n~,l ·' ' uo~u;nn m~, ;r rn · · •
'b' ' b' 'b •J '7 '7 ~
' '
such as appear in the macroscopic Maxwell equations.
tne question or wnat type or averagmg is appropriaLe mus' ue examineu
with some care. At first glance one might think that averages over both space
.1 . ·~· .~. Unr tJ..;o ;o ~nt t~ .. ~ r\~1., o o~.- ;~] ~vrr~a;na ;, '~rv
• 'J -r
(Parenthetically, we note that a time averaging alone would ~ertainly not be
sufficient, as can be seen by considering an ionic crystal whose ions have small
zero point vibrations around well-defined and separated lattice sites.) To delimit
•'- _, . ··'- • . ' . . ,f .1. . '"
'""" '""' ·r 'P ·r -
nomena to work, we observe that the reflection and refraction of visible light are
adequately described by the Maxwell equations with a continuous dielectric con-
stant, whereas x-ray diffraction clearly exposes the atomistic nature of matter. It
is plausible therefore to take the length Lo = 10- 8 m = 102 A as the absolute
tower nmn w me macroscopic uomain. 1 ne periou 01 Wllll
light of this wavelength is Lclc = 3 x 1o- 17 s. In a volume of L5 = 10- 24 m 3 there

are, in ordinary matter, still of the order of 106 nuclei and electrons. Thus in any
region of macroscopic interest with L >> L 0 there are so many nuclei and elec-
trans that the nuctuattons Will be comprerery wasnco om oy a spatiar averaging.
On the other hand, because the time scale associated with L is actually in the
r~n~~
'b
nf , ''nrl rnnJ~~nJor rnnl;nno
'
~ t;rnP.~uPrq(J;n(J W011lrl 1101 h.,; -
priate. There is, nevertheless, no evidence after the spatial averaging of the mi-
croscop1c t1me nuctuatwns or the medmm. This 1s so oecause, 111 me aoscnce or
special preparation and the establishment of ordering over macroscopic dis-
+.
'
~t_ ' '.+: ,C +L • . T
• g_,lcl, ~~tPrl nUM rl;o_-
tances of order L. All that survive arc the frequency components corresponding
to oscillators driven at the external, applied frequencies.
The spatial average of a function F(x, t) with respect to a test function f(x)
~ ·" _,·
r
(F(x t)) - I djx' Hx')Hx - x' t) (6.65)
J
,]., U-\ . ,] n. ;, onrn~ ~~:~J..)- -'- -' nf v n ~nd nnrm:>h7Pd In

. , . . IJ , ••,h .,.h11 . . f( ) b
umty over a space. It IS s1mp est, t oug not necessary, to 1magme x to e
nonnegative. To preserve without bias directional characteristics of averaged
physical properties, we make f(x) isotropic in space. Two examples are
j
r < R
f(x) = 47TR 3 '
n
\. u, (6.66)
3/2 -r 2!R 2
f(x) = (1rR 2 ) e
' "~M. » a..;.J

I
I
I
L»
I
I : :
: l .... :
:·~~
I
I
I I I
I I I
I I I
i : i r
0
Figure 6.1 Schematic diagram of test function .f(x) used in the spatial averaging
procedure. The extent L of the plateau region, and also the extent !J..L of the region
where .f falls to zero, are both large compared to the molecular dimension a.
The first example, a spherical averaging volume of radius R, is a common one in

the literature. It has the advantage of conceptual simplicity, but the disadvantage
of an abrupt discontinuity at r = R. This leads to a fine-scale jitter on the averaged
quantities as a single molecule or group of molecules moves in or out of the
averaging volume. A smooth test function, exemplified by the Gaussian, climi-
n~t.,, <nrh ri;J'fc:. ult;,, '-' -' ;,, ,I, '·I. -' -''
_' _r 'b' ' . • .
> v• cuua""''Y, """' '"''' '' J~X) UOeS nOL neeu LO ue SpCCiuCu Ill ueLaH; all tnat
arC needed are lJ COni ;nnitv ~n.-J lllf'." iP< th:lt , ~ r~ _
..Jl . CT' .1 ..• • -,
'"'Y ~ ' " . ' ' v ' ovllC~ ·~ , II Ul j ~X) UVCl Ul> , Ul ; U!Inen-
s1ons as m01catea scnemahcallv in Fi!!. 6.1. This is a !!rcat virtue.*
siaer mese opera_tious wnn respect to avcragmg accordmg to_1(J._(J:)J_. tovidentlv.

WP Ji:lVP
;, r ov I ov\
v' _tl - _j_ "' _l
ax J iJx- \axJ
and {(:. ?.7\

• J
( Ti'(y t\\ I \
iJt
lhe operatwns of space and time differcntiMinn thus lUte WifJi thP :lVPr-
aging operation.
We can now consider the averaging of the microscopic Maxwell equations
(6.64). The macroscopic electric and magnetic field quantities E and Bare defined
as the averages of the microscopic fields e and b:
t<.(X, t) \e(x 1!J (f. f.R\
n~x, t) \U~x, t )I
~ d . . .
uumogeneous equaL!ons m ~ o.u'+ J uecome rne cor-

"'"'" "'"' a v"''a1S"'' v• "'"' " " '
responding macroscopic equations,
(V · b) = 0 ~ V•B = 0 (6.69)
iJb)
Vxe+- iJB
=o~vxE+-=0
\ at at
*We arc here following the development of G. Russakoff, Am. f. Physics 38, 1188 (1970).
The averaged inhomogeneous equations from (6.64) become
e0 V • E = (17(x, t)) (6.70)
1-'-o V X B - Eo at = (j(x, t))
Comparison with the inhomogeneous pair of macroscopic equations in (6.62)

indicates the already known fact that the derived fields D and H are introduced
by the extraction from (17) and (j) of certain contributions that can be identified
with the bulk properties of the medium. The examination of ( 17) and (j) is there-
We consider a medium made up of molecules composed of nuclei and elec-

trons and in addition, "free" charges that are not localized around any particular
molecule. The microscopic charge density can be written as
(6.71)
where x1 (1) is the position of the point charge q1. To distinguish the bound charges
from the free ones, we decompose 17 as
YJ = 1Jfree + 1Jhound (6.72)

an
'l7bound = 17,(x, t)
n
(molecules)
where 17, IS the charge dens1ty of the nth molecule,
n.(x, t) - 2: IJ• .S(x - x,) (6.73)
In these and subsequent equations we suppress the explicit time dependence,

since the averaging is done at one instant of time. We proceed by averaging the
charge density of the nth molecule and then summing up the contributions of all
molecules. It is appropriate to express the coordinates of the cha1 ges in the nth
molecule with respect to an origin at rest in the molecule. Let the coordinate of
that fixed point in the molecule (us!Jally chosen as the center of mass) be x,(t),
and the coordinate of the jth charge in the molecule be x1,(t) relative to that
ongm, as md1cated m Fig. 6.2. The average of the charge density of the nth
molecule is
r
(17,(x, t)) = J d 3 x' f(x') 17,(x- x', t)
zq
j(n)
1 J 3
d .t' f(x') 8(x x' x1, x,) (6.74)
Smce x1, IS of order atom1c d1menswns, the terms 111 the sum have arguments
differing only slightly from (x - x,) on the scale over which f(x) changes appre-
Figure 6.2 Coordinates for the nth molecule. The origin 0' is fixed in the molecule
(usually it is chosen at the center of mass). The jth charge has coordinate X;n relative to
0', while the molecule is located relative to the fixed (laboratory) axes by the
coordinate Xn.
ciably. It is therefore appropriate to make a Taylor series expansion around

(x - xn) for each term. This gives
(Q~)af3 =3 (6.77)
In terms of these multipole moments the averaged charge density of the nth
(71,(x, t)> = q,f(x - x,) - p, • Vf(x - x,) (6.78)

1 , iPJ(x - x,)
± ~{Q) ±···
If we attempt to VIew this equatiOn as the direct result of the defimtwn (6.65) of
the spatial averaging, we sec that the first term can be thought of as the averaging
of a point charge density at x xno the second as the divergence of the average
of a point dipole density at x = x,, and so on. Explicitly,
(71,(x, t)> = (q,8(x - x,)) - V · (p,8(x - x,)) (6.79)
1 a
2
We thus find that, as far as the result of the averaging process is concerned, we
can view the molecule as a collection of point multipoles located at one fixed point
in the molecule. The detailed extent of the molecular charge distribution is im-
portant at the microscopic level, of course, but is replaced in its effect by a sum
of multipoles for macroscopic phenomena.
An alternative approach to the spatial averaging of (6.65) via Fourier trans-
forms gives a valuable different perspective. With the spatial Fourier transforms
1
g(x t) - - - { d 3k g(k t)eik·x and
straightforward substitution into (6.65) leads to the expression for the average of
' '
1 r 3
(F(x, t)) - (Z1r) 3 J d k j(k, t)F(k, t)e'k·x (6.81)
an illustration of the "faltung theorem" of Fourier transforms. The convolution

of (6.65) has a Fourier transform that is the product of the transforms of the
separate functions in the convolution. Thus
FT (F(x, t)) = }(k)F(k, t) (6.82)

The notation FT is introduced to stand for the kernel multiplying the exponential
=
in the first integral above [FT g(x, t) g(k, t)] to avoid a clumsy and confusing
use of the tilde.
A crucial aspect of ](k) is that ](0) = 1, as can be seen from its definition
and from the fact that f(x) is normalized to unity. For the Gaussian test function,
the Fourier transform is
FT f(x) = }(k) = e-k'R't4 (6.83)
Evidently the Founcr transform (6.82) of the averaged quantity contams only
low wave numbers, up to but not significantly beyond km ... = 0(1/R). the inverse
of the length scale of the averaging volume. But because ](k) __, 1 for wave
254 Chanter 6 Maxwell Eon . Ma,· . _c, . '1. ..li.I_
.t. ,11 " il. "" " .

' ~~ , •~ ,,.~ LlUt;ttopresenLa{JOll , UH.d L \1 :"""• '}/ b~."'"' a
of the long-wavelength aspects of F(x, t). Only the small-scale (large wave num.
hPrl ~ ~rP
. ... .
·"
. '
'Consider the averaging of the charge density of the nth molecule shown in
" <
·~·
•'-
~ ~ ~·
Fi!!.h.? ThPFn11riPr ·' "fth ~ rl '_ ·•

0
FT ( 1J,(x, t)> = f(k)ijn(k, I2_ 16.841_

where
ij,(k, t) = J d3x• 1'/n(x', t)e-ik·(x'-xn)

Here we have taken the spatial Fourier transform relative to x,. The qualitative
behaviors of the two factors in (6.84) are sketched in Fig. 6.3. Since the suooort
for the product is confined to comparatively small wave numbers, it is appropriate
to make a Taylor series expansion of the Fourier transform ij,(k, t} for
smau iKI,
fUk t) _ii_ffi tl + lc . V. n ..
(0 tl ±
'
...
Explicitly, we have
ij,(k, t) = I d'x' 1'/n(x', t)[l - ik · (x - x,) + · · ·]

or
1'/n~K, I) qn ZK • Pn + quadrupole and higher (6.85)

in terms of the molecule's multinole . The .rl _, .. 1 .1.
density can therefore be written as

r
(1'/n(x, t)) = (?:131 d3k e'k·(• x,.lf(k)[q, - ik . Pn + .. ·]
~U.OV)
'
= qnf(x - Xn) - Pn • Vj(x - Xn) + · · ·
We have arrived at (6.78) by a different and perhaps longer route, but one with
the advantage of giving a complementary view of the averaging as a cutoff in
wave number space, a point of view stressed by Robinson.
The total microscopic charge density (6.72) consists of the free and bound
11-
'\(k)
\ iilk "
--.....J ..........
\
\
\ ....... \\ I~ _L"'...
\/
I \
\ L
' I
k
Figure 6.3
-
Qualitative behavior of the Fourier transforms in (6.84)_for the transform
or the averaged molecular charge density ('Tin(x, t)).
charges. Summing up over all the molecules (which may be of different species)
,mel ~eomhinino with thP. fr"" rh~r2:es we find the averaf!ed microscooic charge
density to be
a2
V • P(x t) + ) 0 '.n(X t) + ··· (6.87)
wncrc pis me macroscopic cnarge aen,ity,

'
p(x, t) = ( 2: q/i(x - xJ + (6.88)
n I
(molecules)
P is the macroscopic polarization,
P(x, t) = \ ~ Pn8(x - Xn)) (6.89)

(molecules)
ana 1.! af3 IS tne macroscoptc quaurupate uensity,

I \
(6.90)
(molecules)
When (6.87) is inserted in the first equation of (6.70), it gives
=p (6.91)
a -uo< a f5 'I'
/C. L'"1\ ct-.· +t.. .+ +1-. • • ;~ A;enl· flMf

\' • •J
have components,
~f3 --+···
aQ~(3
D a = EE
0 a
+P-
a axf3 (6.92)
The first two terms are the familiar result (6.63). The third and higher terms are
present in principle, but are almost invariably negligible.
· · · .l / !\ n .c ·
~U -~- 'lll<; nvu•uo• ' \J!• •vv •
and the presence of velocities the derivation is considerably more complicated

than the earlier treatment of (TJ), even though no new principles are mvolved.
UJ, ~~1., th~ ~~enlte Jo,, • 10' thP OIIT" rlPhik tn 0 -'-
1
'fnr lhn'P rPOcl-
.t'h o enJOY
ers w . 'J
sue h cha ll enges.
' W eub egm
. u wit
: h t h c microscopic
. • current d ens1ty,
•
(6.93)
wh<'re v. = dx/dt is the velocitv of the ith char2:e. A2:ain the sum is divided into
one over the free charges and one over the molecules. The current density of the
fllll lit: l:dll Ut: dVt:ldgt:u
..Jl<~• a~
.
w
-~"
\v• "T} •v i',"'v
.
1.• I ·" "" {, \£1, \ ft: OLl\
\Jn\"• '!! ~ '1j\ jn nJJ \ n 1n1 \ · '/
j(n)
Here we have assumed nonrelativistic motion by writing the velocity of the jth
charge as the sum of an internal relative velocity vjn and the velocity v, = dx,ldt
.. . . . ,,
or rne ongm u 111lll(; r IUHI " ' " pu•11c u11 tHe u =·
~,.. rr ·
·lms, ,.. hcroscopic ~Ie<uomagnensm, Lonservauon Laws -Sl
Taylor series expansions and vector manipulations. A portion of the current in-
volves the molecular magnetic moment,
(6.9<;1
j(n) L
1 ne nnat resun ror a component ot the averaged m1croscopic current density is
\
\molecules)
(L nL\
\' . J!
_'i~~dUUvO W WOO OU<HvO >UOv V' a~~ 111(; mac-

roscopic current density
' ' '

1"'
\ j
~(. .\ 'ln n '\'
<;:(,
'1] J '\
"'I
M_(_x t) = ( ) m 8(x - x.) \ (fi QRI
Tf the free "charges" also possess intrinsic magnetic moments, these can be in-
cluded in the definition of Min an obvious way. The last terms in (6.96) involve
the electric molecular moments and molecular velocities and hP aiuPn ~n
easy interpretation, except in special cases (see below).
vv nen IJ 1 is insenea m the second equation ot ~b. /U ), there results the mac-
. ir Ll. ' -M~ •11 M"'~t;An Af ((C, (C,')\ ;,)-, H .1. ' .1 ' <: .1.1
quantity H given in terms of Band the p;o~erties of the medium as·o·

n n) lVl" -t- \ 2 (p, X v,)" b(X Xn))
,r-<o " \ n I
(molecules)
fL nn\
\ v. J!
l'V a I rn•\ r. \
__ ,"""'
n ___ ,
\~n/of3\ n/y
>:f.
'\~ ~n; I
•o o~uv v< \V.JJj ' " '"'"' w"uum <voun, \v.U.J). UlLO
other terms aregenerally_ extremely small· first because the molecular velocities
v, are small, typically thermal velocities in a gas or lattice vibrational velocities
in a ,u,iu auu, '"''-'onu, oecause tne veLOCities nuctuate and tend to average to
zero macroscopically. An exception occurs when the medium undergoes bulk
• D. • ~· ,, . .
velocity v. Neglecting any other motion of the molecules, we put v" = v for all
. . .
n. 1 ncn ~ o.':l':l J oecomes, arrer a mue mampUJauon,
1 ... ,..
•
TT /T>.
v "U' .,
D\ v
\ .. .,
f"' 1 (\(\\
f.-to
.. l. n i~ aivP.n hv {F. Q?) Thi~ 'hnw' thM fnr" mPdium in motion the electric
polarization P (and quadrupole density Q"13 ) enter the effective magnetization.
Equation (6.100) IS the nonrelativistic limit of one of the equations of
Minkowski's electrodynamics of moving media (see Pauli, p. lU.) ).
The reader may consult the book by de Groot for a discussion of the rclativ-
• -" £ • • ' -'-· • •• -+•l. .
'""'" '"v"'"'""vuo, ao VV'-'H ao •v• a o<w
From the standpoint of logic and consistency there remains one loose end. In
definin!! the molecular auadruoole moment (Q;.L" by (6.77) we departed from
our convention of Chapter 4, Eq. (4.9), and left (Q ~)" 13 with a nonvanishing trace.
Since we made a point in Chapter 4 of relating the five independent components
ot the traceless quadrupole moment tensor to the tL:t -r l) spncrica1 narmonics
for l = 2, we need to explain why six components enter the macroscopic Maxwell
.. ,; H ' ~ -~ ' ,1:.. "''"~ "1'
~, .. ,1, m~Jp >I { () \ h'l
'1' • T .•
means of ( 4.9), then we have
U.! n}af3 ~'d:n}af3 ""

L, 'lj~Xjn!
2~
Uaf3
/r ~•
\V.-'.V-'-J
j(n)
T .-lnf'inn" mc>"n ., "h"r"" r~d;," ,2 ~f thP ~ ,1 1l:lr c'· , di,trihution

0 c
by
c
2
n
"'
L.J_ '11\"
j(n)
/. \2
Jn!
where" is some convenient unit of char<'e for examnle. that of a oroton we can
write (6.101) as
rn•\ rn 1 .L nv2 J<
'~"'"!'
\ 'l;;,rl/Of{j
n ""'
The macroscopic quadrupole density (6.90) thus becomes
'
Q af3 = Qaf3 + 61~ 2:n er~8" 13 8(x - x,))
(molecules)
where Q" 13 is defined in terms of (Qn)af3 just as in (6.90). The net result is that in
the averaged microscopic charge density (6.87) the traceless quadrupole density
n ,]. ~<>0 I hP dPn,;h, {)' "n.-1 thP ,:h"raP ..1. - " n is :lll! _ , hv :ln :ld-
_, .' . ',
'-"' ''"""•
I
'
1-' I-'free ""
L. <JnU~A
. An)'
1~2
6 "'
LJ vo
2N.
,~,~
~,,I
\ \ f L
\ ~-
1 W>\
~ .,
·,, ,,,
n
,,,, n
·, __ ,_. Je,)
The trace of the tensor Q ~ 13 is exhibited with the charge density because it is an
1 n .:t · · • .+ T~ ,J ~h,rno
<1. .1 · '
-,. ·
ana mean square ramus terms wgemer acmauy '"l- """'-''ll lllto ll"L LWU Ltollll~ Ill
an expansion of the l = 0 molecular multioole as we go beyond the static limit.
In the> ~ .tr"n UNVP nnmhPr - r thPv 1P,j 1.-l tn thP fir,t twn
~''"'
..........,. vH<lf-'"'' v , ,nacroscoptc JC,tecuomagneusm, '-'onservauon Laws -Sf
terms in the expansion of the charge form factor in powers of k 2 • This can be
seen from the definition of the form factor F(k 2 ) for a charge density p(x):
F(k 2 ) = I d x o(x)(eik•x),ooO
3
J(
sin kr
J a X p~X)
kr
= Jpd x - 3
~k2 Jr p d x 2 3
+ ...
With the correspondence k - -iV, the general equivalence of the form factor
expansion and (6.102) is established.
~n an in,eres,ing monograpn auuaea to aoovc, rr.oomson gives a mscussion
nf thP inn'" t-hn · · .t; -1 th • · ·
, • . "1" ..... , • • 'J- . 'iua-
hons Similar to ours. However, he makes a dtstmctwn between the spatial av-
eraging (6.65) with the test function f(x), called "truncation" (of the wave
-'- - , \ J. ' _,.. ' • _, ' ' '
'r '! -'.!._ ...... , auu '"'- ocw - an.da~~"5 v<Cl VdliUUS
sorts of ensembles. Robinson holds that each macroscooic_12roblem hl'ls its nwn
appropriate lower limit of relevant lengths and that this sets the size of the test
function to be used, before any considerations of statistical averaging are made.
p..,.,.,t;.,,.,., Tl.
,;7
-J -o -·
,,~ rr~-
'-' .
---· ,.. "J_
.......
,/! v,
·..s_~
nnd 71.4 ...... n.,tum fnr n \'1,
-_,
·<- nfrh. ,] ~
~ -J -o
and Electromaf!netic Fields

1ne wrms m Lnc mws 01 conservatwn or energy ana momentum arc important
results to establish for the electromagnetic field. We begin by considering con-
servation of energy, often called Poynting's theorem (1884). For a single charge
q the rate of doing work by external electromagnetic fields E and B is qv · E,
where v is the velocity of the charge. The magnetic field docs no work, since the
magnetic force iS perpendicular to the velocity. If there exists a continuous dis-
;1- ,t' ,l' ,J- • A •'- _' _c .l • _,_ '- ,,_ ~ •
'b' l CH'- 'V'U< l<lC'- Vl UV>Hb '"'" VJ Ul'- 1'-lUO 111 d
finite volume Vis

r
3
J ·E d x (6.103)
v
This power represents a conversion of electromagnetic energy into mechanical

w "'""' 5Y. " HluS• ue uamnceu uy a corresponumg raLe 01 uecrease or
P.nP.T!JV in thP. p]P.rlrnm~ · • finl-1 m;th;n thn unl11mn V T, · ;t <l' _
vation law explicitly, we use the Maxwell equations to express (6.103) in other
terms. Thus we use the Ampere-Maxwell law to eliminate J:
1
Jv
_j_ • __E___d_3 y - r
Jv
F. . ru ')( H\ F ..
an
at _
_d3_y_ f{:.
\- .1 fiA\,I
If we now employ the vector identity,

_..,_
• ~~ " ll) n·~v"~)
"' • ~" " ll)
Sect. 6.7 Poynting's Theorem and Conservation of Energy and Momentum 259
and use Faraday's law, the right-hand side of (6.104) becomes

r r ;;n ;JR
Jv J · E d x =
3
-Jv V · (E x H) + E · -
iJt
+ H·-
CJI
d·x ( 6.105)
1o proceed runner we maKe two assumpTiOITS:\Tj llll: u.a-.,,

"'
linear in its electric and magnetic properties, with negligible dispersion or losses,
and (2) the sum of (4.89) and (5.148) represents the total electromagnetic energy
.-1 +].,
" '
, v ' v H 1V1
C '
U W V W1JW/S H v
'
..
. 0 - Ll. '"'•L oL o
·.- ' 0]
energy density denoted by
1 '"'
2 \'
..,. n TT\
., {{:. 1 {\{:.\
\ .. 'I
(6.105) can be written
-t J. E d3x = t iJu
- + V ·(EX H) d 3 x
iJt
(6.107)
. dlUlUaty, Llllo~-ctrimv
10 liJCO
' ' . . , . ....
'"~ •v••
c• v
.c ~
, . Pr
',.1
continuity equation or conservation law,

dU
- + V · S = -J · E (6.108)
iJt
The s o+ino flow is called the Povntim? vector. It is given

by
S=EXH (6.109)
and has the dimensions of (energy/area X time). Since only its divergence appears
in the conservation law, the Poynting vector seems arbitrary to the extent that
the curl of any vector field can be added to it. Such an added term can, however,
have no physical consequences. Relativistic considerations (Section 12.10) show
that (6.109) is unique.
1 ne ,. .1 u1 uJe 1111 Ul Ulllel
c
\U,iU<)
., "'""
\U.<UU) HW11VUH
f ,,...~,
V1
is that the time rate of change of electromagnetic energy wnnm a certam volume,
plus the energy flowing out through the boundary surfaces of the volume per
' ' ' .1 •'- ' .coL -1 .1. .-1. ~ +h hol.-1. ,n ~h
UHH cuu~, <O ~'-[ UU< \V \Hv '"IS' 'J
sources within the volume. This is the statement of conservation of energy. The
assumptions that follow (6.105) really restrict the applicability of the simple vcr-
,inn nf n. ina'< .... tn V}lrllllm m"' . or . . fields. Even
for linear media, there is always dispersion (wt!h accompanying losses). Then the
right-hand side of (6.105) docs not have the simple interpretation exhibited in
(6.107). The more realistic situation of linear disoersive media is dtscussed m the
next section.
The emphasis so far has been on the energy of the electromagnetic fields.
lhe worK done per unit time per umt vomme ny tne 11etus ~·' · ~J is a conversion
of electromagnetic energy into mechanical or heat energy. Since matter is ulti-
mately composed of charged particles (electrons and atomic nuclei), we can think
• • ' • • Co\. " '.\
u1 liltS raLe 01 cunverSton as a 1a1e u1 Iuueasl: u1"' "'bY . j.IU< ""'"'" V1 "'"'
per unit volume. Then we can mtcrpreCFoyntmg s theorem tor tne mzcroscopzc
fields (E, B) as a statement of conservation of energy of the combined system of
260 Chapter 6 Maxwell Equations, Macroscopic Electromagnetism, Conservation Laws-Sf
particles and fields. If we denote the total energy of the particles within the
volume Vas Emech and assume that no particles move out of the volume, we have
dEmcch =
dt
Jv J . E d3x ( 6.110)
Then Poynting's theorem expresses the conservation of energy for the combined
dE
-
dt
=-
d (Emech + Efield)
dt
= -
fs
n · S da (6.111)
where the total field energy within Vis

r r
E J rP ~o iv (E 2
1 c2 B 2 ) tPx (6.112)
The conservation of linear momentum can be similarly considered. The total

electromagnetic force on a charged particle is
F = q(E + v x B) ( 6.113)
If the sum of all the momenta of all the parl!cles m the volume Vis denoted by
pflieeh: we can wrjte, from Neurton's second la\V,
dPmech =
dt
f v
(pE +J X B) d3x (6.114)
where we have converted the sum over particles to an integral over charge and
current densities for convenience in manipulation. In the same manner as for
Poynting s theorem, we use the Maxwell equations to eliminate p and J from
1 iJE
p = E0 V · E, J=-VxB-E
0 - (6.115)
/Lo iJt
With (6.115) substituted into (6.114) the integrand becomes
t riE
2
l
Then writing
B
iJE
X -. = - - (E
a X B) + E X -.
iJB
dt at dt
2
and adding c B(V · B) 0 to the square bracket. we obtain
pE + J X B E0 [E(V · E) + c2 B(V · B)
iJ
- E X (V X E) - c2 B X (V X B)] - Eo iJt (E X B)
The rate of change of mechanical momentum (6.114) can now be written

dP cl r
(6.116)
= Eo t [E(V · E) - E x (V x E) + c2 B(V · B) - ClB x (V x B)] d 3 x

Sect. 6.7 Poynting's Theorem and Conservation of Energy and Momentum 261
We may tentatively identify the volume integral on the left as the total electro-
magnetic momentum Pfieid in the volume V:
(6.117)
The integrand can be interpreted as a density of electromagnetic momentum.

\Ve note that this momentum density is proportional to the energy-flux density
S, with proportionality constant c- 2 •
To complete the identification of the volume integral of
(6.118)
as electromagnetic momentum and to establish (6 116) as the conservation law

for momentum, ·.ve must convert the volume integral on the right into a surface
integral of the normal component of something that can be identified as mo-
mentum flow. Let the Cartesian coordinates be denoted by x,., a = 1, 2, 3. The
a - 1 component of the electric pnt of the integrand in (6 116) is given explicitly
by
[E(V · E) - E X (V X E))t
[E(V · E) E x (V X E)], - k a (EJ 13 (6 119)
T"~ = E0 [EaE~ + c2 B"B~ - ~(E · E + c B • B)oa~]

2
(6.120)
we can therefore write (6.116) in component form as
- a T"~
3
-d (Pmcch + Pfield)a -
- "L,
") d X (6.121)
dt ~ v ax~
Application of the divergence theorem to the volume integral gives
d
dt
(Pmech + Pneld)a = J: Lfl T"~n~ da
fs
(6.122)
where n is the outwmd nonnal to the closed smface S. Evidently, if (6.122) rep
resents a statement of conservatiOn of momentum, 2:flTa 13 n 13 Js the ath component
of the flow per unit area of momentum across the surface S into the volume V.
In other ·.vords, it is the force per unit area transmitted across the surface S and
acting on the combined system of panicles and fields inside V. Equation (6.122)
can therefore be used to calculate the forces acting on material objects in elec-
tromagnetic fields by enclosing the objects with a boundary surfaceS and adding
up the total electromagnetic force according to the right hand side of (6.122).
262 Chapter 6 Maxwell Equations, Macroscopic Electromagnetism, Conservation Laws-SJ
The conservation of angular momentum of the combined system of particles

and fields can be treated in the same way as we have handled energy and linear
momentum. This is left as a problem for the student (see Problem 6.10).
The discussion of electromagnetic momentum and the stress tensor in fluids
and solids entails analysis of interplay of mechanical, thermodynamic, and elec-
tromagnetic properties (e.g., Oe/OT and IJe/IJp). 'Ne refer the reader to Landau
and Lifshitz, Electrodynamics of Continuous Media (Sections 10, 15, 16, 31, 35),
Stratton (Chapter 2), and, for a statistical mechanical approach, to tic Groot (Sec-
tion 13). We note only that, although a treatment using the macroscopic Maxwell
equations leads to an apparent electromagnetic momentum, g - D x B (Min-
kowski, 1908), the generally accepted expression for a medium at rest is
1 1
g ? E X H JLoEoE X H
c
2 s (6.123)
'Nc note that g is the electromagnetic momentum associated with the fields.
There is an additional co-traveling momentum within the medium from the me
chanica! momentum of the electrons in the molecular dipoles in response to the
incident traveling wave.* The Minkowski momentum of a plane wave is the
"pseudomomentum" of the wave vector (k = nw!c or hk = n(hw )!c for a photon).
6.8 Poynting's Theorem in Linear Dispersive Media with Losses

In the preceding section Poynting's theorem (6.108) was derived with the restric-
tion to linear media with no dispersion or losses (i c D - FE and B - ,uH), with
E and JL real and frequency independent. Actual materials exhibit dispersion and
losses. I o dtscuss dtsperston tl ts necessary to make a Fourier decomposition in
time of both E and D (and Band H). Thus, with
E(x, t)
D(x, t)
the assumption of linearity (and, for simplicity, isotropy) implies that D(x, w) =
E(w)E(x, w), where E(w) is the complex and frequency-dependent susceptibility.
Similarly, B(x, w) = JL(w)H(x, w). The reality of the fields implies that E(x, -w)
E*(x, w), D(x, w) D':'(x, w), and E( w) E*(w). The presence ofdisperston
carries with it a temporally nonlocal connection between D(x, t) and E(x, t),
discussed in detail in Section 7.10. As a consequence, the term E · (nD/tJt) in
(6.105) is not simply the time derivative of (E · D/2).
We write out E · (OD/Ot) in terms of the Fourier integrals, with the spatial
dependence implicit,
aD
E· J dw J dw'E¥(w')[ twE(w)] . E(w)e 'w wt
at
*Sec R. E. Peierls, Proc. R. Soc. London 347. 475 (1976) for a very accessible discussion, of which
Problem 6.25 is a Simplmed versiOn. See also R. Loudon, L. Allen, and D. F. Nelson, Phys. Rev. R
55, 1071 (1997).
Sect. 6.8 Poyntmg's Theorem m Linear Dispersive Media with Losses 263
Split the integrand into two equal parts and in one make the substitutions,
w ___,. - w', w' ___,. - w, and use the reality constraints to obtain
E ·aD
-
at
1
=-
2
Jdw Jdw'E*(w')[-iwE(w) + tw'e"'(w')]·
. . E(w)e-'·( w-w
'JI
(6.124)
We now suppose that the electric field is dominated by frequency components

in a relatively narrow range compared to the characteristic frequency interval
over which E(w) changes appreciably. We may then expand the factor iw' E*( w')
in the square brackets around w' - w to get
[· · ·] = 2w Im E(w) - i(w- w')- (wE*(w)) + .. ·
Insertion of this approximation into (6.124) leads to
E·
aD
at
f f dw dw' E*(w') . E(w)w Im E(w)e i(o o')l
(6.125)
a1
+--
at 2
Jdw Jdw' E*(w'). E(w)-
dw
[wE*(w)]e-'d .< w-,,, ') I
There is a corresponding expression for H ·a Blat withE ___,. H and E ___,. J.L on the
right-hand side.
First of all note that if Eand J.L are real and frequency independent we recover
the simple connection between the time derivative terms in (6.105) and au/at,
with u given by (6.106). Second, the first term in (6 125) evidently represents the
conversion of electrical energy into heat (or more generally into different forms
of radtatwn"'), whtle the second term must be an effective energy density. A more
transparent expression, consistent with our assumption of the dominance of E
and H by a relatively narrow range of frequencies can be obtained by supposing
that E E(1) cos(w0 1 + a), H H(t) cos(w 0 t + (3), wheiC E(t) and II(t)
are slowly varying relative to both l!w0 and the inverse of the frequency range
over which E( w) changes appreciably. If we substitute for the Fourier transforms
E(w) and H(w) and average both sides of the sum of (6.125) and its magnetic
counterpart over a period of the "carrier" frequency w0 , we find (after some
straightforward manipulation),
j
\E • at
aD aB) = w
+ H ·at 0 Im E(w0 )(E(x, t) · E(x, t))
(6.126a)
u [(
+ w0 Im J.L( Wo)(H(x, t) • H(x, t)) + _e
where the effective electromagnetic energy density is
1 [d(wE) ]
(6.126b)
+ ~ Re[ d(wJ.L) (Wo)]<H(x, t) · H(x, t))

dw
*For example. if the dominant frequencies are near an atomic resonance of the medium where ab-
*
sotptiou is imput taut (Im e 0), the te-emissiem of the radiatiofl ahsorhed at w may be at (X)', v.fiere
w' ::5 w.
264 Chapter 6 Maxwell Equations, Mauoscopic Elcchomagnelism, Consenalion Laws SI
The presence of the factors d(wt:)!dw and d(wu.)/dw was first noted by Brillouin
(see Brillouin, pp. 88-93). Our treatment is similar to Landau and Lifshitz, Elec-
trodynamics of Continuous Media (Section 80).
Poynting's theorem in these circnms~anccs reads
aueff + v . s = -J . E - Wo Im E( Wo)(E(x, t) . E(x, t))
Wo Im fL( Wo)<H(x, t) • H(x, t))

The first term on the right describes the explicit ohmic losses, if any, while the
next te1ms Iepiesent the absorptive dissipation in the medium, not counting con-
duction loss. If the conduction current contribution is viewed as part of the di-
electric response (see Section 7.5), the -J · E term is absent. Equation (6.127)
exhibits the local conservation of electromagnetic energy in realistic situations
where, as well as energy flow out of the locality (V · S i= 0), there may be losses
from heating of the medium (1m E i= 0, Im g i= 0), leading to a (presumed) slow
decay of the energy in the fields.
Lumped circuit concepts such as the resistance and reactance of a two-terminal

linear network occur in many applications, even in circumstances where the size
of the system is comparable to the free space wavelength, for example, for a
resonant antenna. It is useful therefore to have a general definition based on field
concepts. This follows from consideration of Poynting's theorem for harmonic
time variation of the fields. We assume that all fields and somccs have a time
dependence e-;w,, so that we write
F(x, t)
The field E(x) is in general complex, with a magnitude and phase that change
with position. For product forms, such as J(x, t) • E(x, t), we have
J(x, t) • E(x, t) = HJ(x)e + J*(x)eiwt] • [E(x)e iwt + E*(x)eiwr]
iwt (6.129)
- ~ Re[J*(x) • E(x) + J(x) . E(x)e- 2 iwt]
For time averages of products, the convention is therefore to take one-half of

the real part of the product of one complex quantity with the complex conjugate
For harmonic fields the Maxwell equations become

V ·B 0, V X E twB 0
(6.130)
V · D = p, V X H + iwD = J
*The tteatment of this section pmallels that of Fano, Chu, anti Adler (Sections 8.2 and 8.3). The
reader can find in this book considerable further discussion of the connection between lumped circuit
and field concepts, examples of stray capacitances in inductors, etc. See also the lirst two chapters of
Adler Chu and Fano.
Sect. 6.9 Field Definitions of Impedance and Admittance 265
where all the quantities arc complex functions of x, according to the right-hand
side of (6.128). Instead of (6.103) we consider the volume integral
-1
2
f v
J*. Ed 3 X
whose real part gives the time-averaged rate of work done by the fields in the
,] l7. T, -' .1 oan+ dr;~+lu ~nrnll~l;nn tho e1 'frr.m {(; ]()1\ tr. {(; Hl7\
·r - y 1
' '
wto na vto
1 I .0 1 I "' .o
• I'- U X ,., I ., • Lv " n' CWU'J U X
'"' I " ~
"''"'"
\V.iJi)
= -1
2
J [-V · (E
v
X .
H*) - tw(E · D* - B · H*)] d 3 x
We now define the complex Poynting vector
S = ~(E X H*) (6.132)
and the harmonic electric magnetic energy densities,
We= ~(E • D*), Wm = hB •H*) (6.133)
Then (6.131) can be written as

, r r r
-'- I I* , I< ,-J3v + ');,,, I fw w \ ,/ 3 v + <t> s .n rln - n (() 114)
2 Jv Jv Js
Thi• ;, th~ "n"loo- of(() 107) fnr hl-lrmonir fi~lcl• Tt is a comolex eauation whose
real part gives the conservation of energy for the time-averaged quantities and
whose imaginary part relates to the reactive or stored energy and its alternating
flow. If the energy densities we and w m have real volume integrals, as occurs
for systems with lossless dielectrics and perfect conductors, the real part of
(6.134) is
t~ Re(J* · E) d 3 x + { Re(S · n) da = 0
showing that the steady-state, time-averaged rate of doing work on the sources
. .
m v oy me neJas IS equal ro me av liOWUT }'U w c 1 imu w to v uuuugu
the boundarv surface S as calculated from the normal component ot Kc ~- ThiS
io inot urh~t urr.nlcl hP f'~lf'nbtPcl frnm ~Pr form of Povntinp's tr m
'' " •r .'
\V.iVl) 11 W\0 a>OUlll\0 lllctl LillO COIICOlt;J
.' ' -o~OWUUJ
' ' ' • ·'- -·
tJUH UHU
.-1 1-•w
cauy nuctuaung pan. wnn losses m me componcnLs 01 we sys,cm, "'"'

term in ( 6.134) has a real oart that accounts for this dissipation.
The complex Poynting theorem (6.134) can be used to define the input im-
pedance of a general, two-terminal, linear, passive electromagnetic system. We
imagine the system in the volume V surrounded by the boundary surfaceS, with
only its input terminals protruding, as shown in Fig. 6.4. If the complex harmonic
input current and voltage arc / 1 and V,, the complex power input is ~J';'V,. This
Figure 6 4 Schematic diagrams of arbitrary, two terminal, linear, passive

electromagnetic systems. The surface S completely surrounds the system; only the input
terminals protrude. At these terminals, the harmonic input current and voltage are I,
and V,, with the input impedance Z defined by V, = ZI,. The upper diagram applies at
lon frequencies \lvhcte tadiationlosses me negligible, while the lower one wilh its
coaxial-line input permits discussion of radiation resistance.
can be written in terms of the Poynting vector by using (6.134) applied to all of
space on the outside of S as
!I;V, = -£ S·n da (6.135)
where the unit normal n is outwardly directed, as shown in Fig. 6.4, and we have
assumed that the input power flow is confined to the areaS, (the cross section of
the coaxial line in the lower diagram of Fig 6 4) By now considering (6 134) for
the volume V surrounded by the closed surfaceS, the right-hand side of (6.135)
can be written in terms of integrals over the fields inside the volume V:
1 ( {
(6.136)
s-si
The surface integral here represents power flow out of the volume V through the
surfaceS, except for the input surfaceS .. If the surface (S - S,) is taken to infinity,
this integral is real and represents escaping radiation (see Chapter 9). At low
frequencies it is generally negligible. Then no distinction need be made between
S, and S; the upper diagram in Fig. 6.4 applies.
The input impedance Z R zX (clectncat engmeers please read as Z
R + jX') follows from (6 I 36) with its definition, V, ZI;. Its real and imaginary
1 r _/'_ r
(6.137)
Sect. 6.10 Transformation Pro ertics of Electroma netic Fields and Sources 267
X= - 2
4w Re (w, - we) d 3 x - Im J* · E d x
I v v
(6.138)
at high frequencies. At low frequencies, in systems where ohmic losses are the
on! a reciable source of dissi ation, these ex ressions sim li ' to
(6.140)
Here O" is the real conductivity, and the energy densities w, and We (6.133) are
..
actance has a plausible form: if magnetic stored energy dominates, as for a
o. itive etc. The different fre uenc de-
(6.139) and (6.140) for the conductance and susceptance of the complex
6.10 Transformation Properties of Electromagnetic Fields
an
that the significance of such requirements and the limitations that can be thereby
placed on the form of the relations is sometimes overlooked. It is useful therefore
to discuss explicitly the relatively obvious properties of electromagnetic quanti-
ties under rotations, spatial inversions, and time reversal. The notions ave 1rect
application for limiting phenomenological constitutive relations, and are applied
m t e next sect10n w ere t e quest10n o magnet1c monopo es 1s 1scusse .
. . . . .
their relation to the general conservation laws is familiar to the reader from
given here.
of a point such that the sum of the squares of the coordinates remains invariant.
268 Chapter 6 Maxwell Equations Macrosconic Elcctroma!metism Conser ' Law•- __o;:r
"· ,].,
.11 ~ <L _, .r. "T'L
UH V> _O UUHJHH , > H~ U
unea
coordinates x ~ are given in terms of the original coordinates x f3 by
X~ - ~ aaf3Xf3 (6.141)
{3
1 ne requirement tO nave ~x·) (X) restncts the real transrormatton coetficients

aa{3 to be orthogonal,
2: a"f3aay = a{3y (6.142)

"
The inverse transformation has (a 1 )all = af3a and the square of the determinant
v< '"" '""";" ~uJ ;,_ "'-lua1 ''-'unity. 1ne vame uer~aJ -r 1 corresponus w a proper
rotation, obtainable from the original configuration by a sequence of infinitesimal
.. L ~,{_\ • •
'"~I'"• .. ~~<\<'J .l ~~ ''~ OCJ <1
<111 HHJ-'>VJN> pmsa
>VCUCaH>, 0
rotation.
Dh
'J
,1
'"1'
:.;
-
. ~
-'
UO >VCUO
£
10 V l >UUVUO >UHft r -
ing on how they transform under rotations. Coordinates x,, velocities v,, momenta
n. h:lVP. .~.
tts_that_ •· o, -''
-, t,-, the> hoo·~-£ ,,· 1,
(6.141) and are tensors of rank one, or vectors. Scalar products of vectors, such
as X • X Or V • D· are inyari:lnt 11n.-if'.r rnt,.tinn< :1n.-i <O :lrP nf ron 1- 7Pr~ ,
or scalars. Groups of quantities that transform according to
D a{3 "
L_ aaya{31'o D y8 (tl.l4:l)
y,l'o
-11 " - ,_
~
U>\, lo Ul, I.:Ul
->UHft ,_ , • .lllC lV<dA WCU rensor oc>
IS one such group of quantities. Higher rank tensor transformations follow

du
-' .
In considering electromagnetic fields and other physical quantities, we deal
with OnP nr mnrP fn ',-,f •;note>o onrl
- ·-r
,t] · 1 .
~
',1-.l
There then arises the choice of an "active" or a "passive" view of the rotation.
We adont the active ViPW- -lhP ,. l:liP OVP' :lrF' rrm• 'rl, rl h· ·rl onrl tho
physical system is imagined to undergo a rotation. Thus, for example, two
charged particles with initial coordinates x and x form a . th"t 11n.-iPr "
rotation is transformed so that the coordinates of the particles are now x; and
x~, as shown in Fig. 6.5. The components of each coordinate vector transform
according to (6.141), but electrostatic potential is unchanged because it is a func-
tion only of the distance between the two points, R = fx 1 - x2 f, and R 2 is a sum
of scalar products of vectors and so is invariant under the rotation. The electro-
static potential is one example of a scalar under rotations. In general, if a physical
quantity rp, WhiCh 1s a tunct1on of various coordinates denoted collectively by x1
(possibly including coordinates such as velocities and momenta), is such that
wnen me pnysica1 sysrem IS rotated with x, __, x,, the quantity remams unchanged,
d>'(x:)- d>(x,) (() 144)
then </> is a scalar function under rotations. Similarly, if a set of three physical
" T T /., \ I 1 -"- ~\ r o • •
'1' '-'\~iJ\~ ~,-,--'J' Lvtauu"'""'"'Y .accurumg
to
V~(x;) - L a" 13 V13 (x,) (6.145)

{3
then the V" torm the components of a vector, and so on for higher rank tensors.
Sect. 6.10 Transformation Properties of Electromagnetic Fields and Sources 269
Ql
Q2 _1!---;lxl'
!/.
~--· ~· I
Ql
1/L/---- Figure 6.5 Active rotatiOn ot a system

X
0 of two charges.
Differential vector operations have definite transformation properties under

rotations. For example, the gradient of a scalar, V </>, transforms as a vector, the
divergence of a vector, V · V, is a scalar, and the Laplacian operator V2 is a scalar
operator in the sense that its application to a function or set of functions does
not alter their rotational transformation properties.
Special mention must be made of the cross product of two vectors:
A-BxC (6.146)
In component form this vector shorthand reads

Aa = Lf3y Eapy Bp Cy
where Eaf3y = + 1 for a= 1, f3 = 2, y = 3 and cyclic permutations, Eaf3y = -1 for

· · · _,. ·' n .c •'-
Ulllt:J jJt::llllU , aHU 'Ul LWU Ul lllUIL' I ~'ium. ~~~uuo~ v1 cu~
presence of two vectors on the right-hand side, the cross product has some at-
tributes of a traceless antisymmetric second-rank tensor. Since such a tensor has
only three independent components, we treat it as a vector. This practice is jus-
. . .
nneu, 01 course, UlllY I
.
as llUau"urms lv
. " ""
_,. ·o tV ~U.i'H)·
In actual fact, the transformation law for the cross product (6.146) is
A~ = det(a) 2:, aa 13 A 13 (6.147)
{3
For proper rotations, the only kind we have considered so far, det(a) = + 1; thus
• • • • • • ' 0' To o
~0 . .1 '+I) IS Ill -o- Wllll we DaSTC' ltlVH ~U.i'H)• U"Uvt
proper rotations, the cross product transforms as a vector.
B. Snatial Reflection or Inversion

Spatial reflection in a plane corresponds to changing the signs of the normal
--r
' ,-,f tho> .,., ,-,f oll ,.,,.,;nt~ :1nrl to le:1vin<I the cnmnonents
~·
parallel to the plane unchanged. Thus, for reflection in the x-y plane, xi=
270 Chapter 6 Maxwell Equations, Macroscopic Electroma<metism Conservation Laws -~I
f.
\' l' Jl' '
'
,,
.\
{,
\"' l' J D
\ c.
'-'VU'-'0!'
~,,.
1V ''
.il
_,
v1 au
three components of every coordinate vector through the origin, x, ~ x; =-xi·
<;;:n~t;~ l ;n ' nr ·"· '•nn ;e ~ ,-l;e~ ·~t ... : <~ ·• , >.
-r "·
""111
two coordinates, cannot in general be accomplished by proper rotations. It cor-
resnonds to det( a) ·1. and for the strai!'h m-1 i ion An~rgtion is oiun
by (6.141) with a" 13 = -[ja!l· It follows that vectors change sign under spatial
inversion but cross products which behave accordinP to (6.14Tl do not. WP ~r~
thus forced to distinguish two kinds of vectors (under general rotations):
1
~ '
UtUf VtLI-UIO V,i\,1Vl0)' 111<11 li<11101Uilll
~V< JU>< ., LU ~ o.1 '+.J J ana ror
x ~ x' - -x behave as
v ~ V' = -v
Axial or pseudovectors that transform according to (6.147) and for x, ~ x; = -x,
behave as
A~A' =A
Similar distinctions must be made for scalars under rotations. We speak of scalars
or pseudoscalars, depending on whether the quantities do not or do change sign
under spatial inversion. The triple scalar product a· (b x c) is an example of a
pseudoscalar quantity, provided a, b, c are all polar vectors. (We see here in
passmg a dangerous aspect ol our usual notatiOn. The wntmg of a vector as a
does not tell us whether it is a polar or an axial vector.) The transformation
[JIV[JtoiLies or nigner ranK tensors unaer spanat mvcrs1on canoe deduced directly
if they arc built up by taking products of components of polar or axial vectors.
r . '- " .c ' • • ' • "
.'11<111<11 lllVClMVll Wllll a ~· "1) , we
T£
H a Ul ' U l l " ' ' '
call it a true tensor or just a tensor, while if the factor is ( -1 )N + 1 we call it a

' ' .< • -1. AT
c. Time Reversal
'T'ho h~,.;~ 1~ <0 ,.- '~
r -o
..
-' '" ''! '
-" cv d.
{,
,.
·• •L
'"'-'
_,
sense of direction of time. This does not mean that the equations arc even in t,
hnt that llnrler the time ff'UPr<g) 1tion f -> t' - f thP r~l~tp,.j ,<
r ·o
• oJ
quantities transform in a consistent fashion so that the form of the equation is
the same as before. Thus for a narticlc of momPnt11m n ~net nosition x mnvin"
in an external potential U(x), Newton's equation of motion,
u~ = -VU(x)
dt
is invariant under time reversal provided x ~ x' = x and p ~ p' = -p. The sign
change for the momentum is of course intuitivelv obvious from its relation to
the velocity, v = dxldt. The consequence of the in variance of Newton's laws under
time reversal is that, if a certain initial configuration of a svstem of particles
evolves under the action of various forces into some final configuration, a possible
state of motion of the system is that the time-reversed final configuration (all
positions the same, but all velocities reversed) will evolve over the reversed path
to the time-reversed initial configuration.
The transformatiOn properttes of various mechanical quantities under rota-
Sect. 6.10 Transformation Properties of Electromagnetic Fields and Sources 271
Table 6.1 Transformation Properties of Various Physical Quantities under Rotations,

Spatiallmersion and 'fime Re•ersal"
Rotation
(rank of Space Inversion Time
Physical Quantity tensor) (name) Reversal
I. Mechanical
Coordinate X I Odd (vector) Even
Velocity v 1 Odd (vector) Odd
MomentJJm p 1 Odd (vector) Odd
Angular momentum L=xxp 1 Even (pseudovector) Odd
Fotce F 1 Odd (vector) Even
Torque N=xXF 1 Even (pseudovector) Even
Kinetic energy p 2 /2m 0 Even (scalar) Even
Potential energy U(x) () Even (scalar) Even
<, ectromagnetzc
Charge density p 0 Even (scalar) Even
Current density J 1 Odd (vector) Odd
~}
Electric field
Polarization 1 Odd ("ector) Even
Displacement
Magnetic induction B
MagnetizatiOn
Ma netic field H
Poynting vector S=ExH 1 Odd (vector) Odd
Maxwell stress tensor T"~ 2 Even (tensor) Even
aFar quantities that are functions of x and t, it is necessary to be vcrv clear what is meant by
evenness or oddness under space inversion or time reversal. For example, the magnetic induction is
such that under space inversion, B(x, l)-> B 1(x, t) ~ + B(-x, t), while under time reversal,
tions, spatial inversion, and time reversal arc summmized in the fit st part of Table
6.L
D. Electromagnetic Quantities
Just as with the laws of mechanics, it is true (i.e., consistent with all known
experimental facts) that the forms of the equations governing electromagnetic
phenomena are invariant under rotations, space inversion, and time reversal. This
tmphes that the different electromagnetic quantities have well-defined transfot-
mation properties under these operations. It is an experimental fact that electric
charge is invariant under Galilean and Lorentz transformations and is a scalar
under rotations. It is natural, convenient, and permissible to assume that charge
ts also a scalar under spahal mverston and even under time reversal. The point
here is that physically measurable quantities like force involve the product of
charge and field. The transformation properties attributed to fields like E and B
thus depend on the convention chosen for the charge.
Wtth charge a true scalar under all three transformahons, charge density pis
also a true vcalar From the fact that the electric field is force per unit charge, we
see that E is a polar vector, even under time reversal. This also follows from the
1v1axweu equa,wn, v ·" p!E0 , smce Dotn S!Qes must transrorm m the sam"
m · thf'
>He u•>< '"u11u• tHe~· . equation representing rarall_ay s _raw,
transforms as a pseudovector under rotations and spatial inversion, and is even

under time reversal. To preserve the invariance of form it is therefore necessary
that the magnetic induction B be a pseudovector, odd under time reversal. The
left-hand side of the Ampere-Maxwell equation,
1 aE
- V X B -E0 - = J
J.Lo at
can be seen to transform as a polar vector, odd under time reversal. This implies
th~t thP ,/o · T · ·'- . .U .J. • ,/
nom us uenm11on m _L_erms ot charge ttmcs vcloctty.

Wre h~VP inst '""n th:>t thP m· · · fi0lrl< ~nrl <nurrP< h~uP ,11 ~ .~ '
·r . . . .
" . ·" , ""' , spaum mverswn, anu umc reversal.
From the derivation_0_f_the m M~xwell ecm:ltions_in__S_ectirm li li ~nrl
the definitions of P, M, etc., it can be seen that E, P, D all transform in the same
way, as do B, M, H. The various transformation properties for electromagnetic
quantities are summarized in the second part of Table 6.1.
To illustrate the usefulness of arguments on the symmetry properties listed
in Table 6.1, we consider the phenomenological structure of a spatially local
constitutive relation specifying the polarization P for an isotropic, linear, non-
dissipative medium in a uniform, constant, external magnetic induction B0 . The
relation is first order in the electric field E, by assumption, but we require an
expansion in powers of B0 up to second order. Since Pis a polar vector, and even
under time reversal, the various terms to be multiplied by scalar coefficients must
transform in the same way. To zeroth order in B0 , only E is available. To first
orrlrer in R. nossihlf> ' ;,unlu;nrr I< ];n0•>rlu ~rp
F.xR "R " R ...
All these arc permitted by rotational and spatial inversion grounds, but only
those involving odd time derivatives transform properly under time reversal. For
the second order in B0 , the possibilities are
aE
(B 0 • B0 )E, (E · B0 )B 0 , (Bo ·B)-
a at , · · ·
Here only the terms with zero or even time derivatives of E satisfy all the re-
quirements. The most general spatially local expression for the polarization, cor-
recr to secona oruer m tQC constant magncttc field Bo. is thus
1 iJR
I" - Xu!!-.+ _X:L X Bo + v,!R ·BolE + h!E • B~Jk + · · · (fi.14R)
Eo U<
where tne x, arc real scalar coeftictents and hi!! her time derivatives of E r~n orrnr
1-\.L 10w Hequencies, llle response or essenuauy au matenal systems ts vta electnc
Sect. 6.11 On the Question of Magnetic Monopoles 273
forces. This means that at zero frequency there should be no dependence of P

on B0 , and a more realistic form is
-1 P = xoE + X1 -aE X Bo + X~(Bo • Bo) -~E2 + x3 (~E

- 2 • Bo ) Bo (6.149)
Eo at at at
where we have exhibited only the lowest order time derivatives for each power
of B 0 . At optical frequencies this equation permits an understanding of the gy-
rotropic behavior of waves in an isotropic medium in a constant magnetic field*
Another cxamnlc the Hall effect 1s left to tne problems. lt. as well as tner-
d!scussea m Lanaau ana Ltrstutz ( op. cu.).

T,
. . .
remxea m consnmnve re1anorrs.--rur
.
, we
1hn ~~no· • ~f
w cm1"'
.ul'u'"'''u'"'u'y l'uwc1
-timP ovm m11ot hP
molecules is described phenomenologically by the constitutive relations, P =

EoXoE + gaB/iJt and /LoM = xbB + f aE/iJt. The added terms involve pseudoscalar
quantities g and f that reflect the underlying lack of parity symmetry for chiral
substances. (Quantum mechanically, nonvanishing g or f requires both electric
and magnetic dipole operators to have nonvanishing matrix elements between
the same pair of states, something that cannot occur for states of definite parity.)
6.11 On the Question of Magnetic Monopoles

At the nresent time ( l'JIJ~) there IS no cxncnmentaJ ev10ence ror rne exisTence
,f ~• ' D -• · ,fl •} ·~ ocn Af ~n n~rlu hr;u;.,nf th0_
Jl-' -o'
oretlcal argument ol U1rac, Life searCffTor monopmes is reneweu w vco1 a
sucn as "'""' lnHn we , avauau1co. '-''"'" o m 0 ,

below, is that the mere existence of one magnetic monopole in the universe would
offer an explanation of the discrete nature of electric charge. Since the quanti-
zation of charge is one of the most profound mysteries of the physical world,
Dirac's idea has great appeal. The history of the theoretical ideas and experi-
mental searches up to 1990 are described in the resource letter of Goldhaber and
Trower." Some other references appear at the end of the chapter.
There are some necessary preliminaries before examining Dirac's argument.
One question that arises is whether it is possible to tell that particles have mag-
netic as well as electric charge. Let us suppose that there exist magnetic charge
and current densities o. and J in addition to the electric densities, Pe and J"
• ·~ oUo
au
V • U Peo v "
(F. l'iOI
ilR
V · B- pm• -V X E-- + J
at m
*See Landau and Lifshitz, Electrodynamics of Continuous Media, p. 334, Problem 3, p. 337.
'P. A.M. Dirac, Proc. R. Soc. London Al33, 60 (1931); Phys. Rev. 74,817 (1948).
'A. S. Goldhaber and W. P. Trower, Resource Letter MM-1: Magnetic Monopoles, Am. J. Phys. 58,
429-439 (1990).
~I'+ Chapter o tVIaxweu ~quanons, tnacroscoptc t<:lectromagnettsm, l:onservation Laws SI
The magnetic densities are assumed to satisfy the same form of the continuity
"4' · a> un.. ,~... · · . n appears 1rom Lnese equmions Lnat the ex1s-
tence of magnetic charge and current would have observable electromagnetic
0. ·~
, , " wo:ovo:ot, lllO:O tuuuwtug .. uu"·'.l"r
I<; - E cos{; + Z 0 H' sin{;, Z 0 D - Z 0 D' cos{;+ B' sing

(6.151)
Z 0 H = - E' sin {; + Z 0 H' cos {;, B = - ZnD' sin I; + B' cos 1:
'r " 'b' '" • lVllllS

such as E X H, (E · D + B • H), and the components of the Maxwell stress tensor
T . invllri;mt. Tf th<e "" eP< orP .rl ;,., thP <ornP mou
L.oOfJe L.oQfJ~ l..V> S ' I'm Sill i_;, Lode Lode COS i_; -;- dm Sill i_;
(6.152)
Pm = - Zop; sin f; + p~, cos{;, J m = - Z 0 J; sin t; + J ~~ cos t;
then it is straightforward al2:ebra to show that the 2:eneralized Maxwell iom
(6.150) are invariant, that is, the equations for the primed fields (E', D', B', H')
are the same as (6.150) with the primed sources present.
The invariance of the equations of electrodynamics under duality transfor-
, o11vwo ""'' ~. ~, a ma11er 01 convenlion 10 speaK 01 a panic1e posscssmg
an electric charge, but not magnetic charge. The only meaningful question is
.L <L -11 ·~1 L " ' ' '
·~u•~• ~" t'~ . .w • w u~•~ UH:O oalllO:O tallU Ul tV lll.. l..llaJbo;o. 11 lllCY
do, then we can make a dualitv transformation choosin2: the an!':!" f. '" thot
Pm = 0, Jm = 0. We then have the Maxwell equations as they are usually known.
If, by convention, we choose the electric and magnetic charges of an electron
to be qe = -e, qm = 0, then it is known that for a proton, qe = +e (with the
r " Vl l..llVl U\.otllg llfe\ :I) -r qe~proLOn) l'e lU -) ana
lqm(nuclcon)l < 2 X 10- 24 Z 0 e.
This extremely small limit on the magnetic charge of a proton or neutron
follows directly from knowin2: that the avcra2:e ma2:netic field at the "'rf""" of
the earth is not more than 10- 4 T. The conclusion, to a very high degree of
precision, is that the particles of ordinary matter possess only electric charge or,
equivalently, they all have the same ratio of magnetic to electric charge. For
v<ll'-'', uHotaL"'-'• ym ""''-', llle 4' 01 magnetic cnarge is more open, but no
positive evidence exists.
The transformation properties of Pm and Jm under rotations, spatial inver-
sion, and time reversal are imnortant. From the known behavior of E "nrl R in
the usual formulation we deduce from the second line in (6.150) that
Pm is a pseuaoscatar aenstty, odd under tzme reversal, and

T A A '
Since the svmmetries of n under both moti"l i

opposite to those of Peo it is a necessary consequence of the existence of a particle
wim oom c1ecmc ana magnetic charges that space mverswn and time reversal
arc no longer valid symmetries of the laws of physics. Tt is a fact, of course, that
*'T'h . th ". , ~ .
• ' '" ' •0 • f"Qo<Q,HHHv C>UC>MUHHUC>UH<>a•
of the presence of the dimensionful parameters Eo and JLo in the SI system. Magnetic charge density
differs in dimensions from electric charge density in SI units. For users of Gaussian units. out Zo-> 1.
Sect. 6.12 Discussion of the Dirac QuautJzatwu Couddwn 27!!
these symmetry principles are not exactly valid in the realm of elementary par-
uclc phystcs, but present evidence is that their violation is extremely small and
associated somehow with the weak interactions. Future developments linking
electwmagnetic, weak, and peilraps stt ong, interactions may utilize pmticles cm-
rying magnetic charge as the vehicle for violation of space inversion and time
reversal symmetries. With no evidence for monopoles, this remains speculation.
In spite of the negative evidence for the existence of magnetic monopoles,
let us turn to Dirac's ingenious proposal. By considering the quantum mechanics
of an electron m the presence of a magnetic monopole, he showed that consis-
tency required the quantization condition,
eg ag n
(n 0, +1, +2, ... ) (6.153)
where e is the electronic charge, a e2 f4 rre0 nc is the fine structure constant

(a = 1/137). and g is the magnetic charge of the monopole. The discrete nature
of electric charge thus follows from the existence of a monopole. The magnitude
of e is not determined, except in terms of the magnetic charge g. The argument
can be reversed. With the known value of the fine structure constant, we infer
the existence of magnetic monopoles with charges g wlmse magnetic "fine >l!ac-
ture" constant is
137 2
4 l'l
Such monopoles are knovm as Dirac monopoles. Their coupling strength is cnor
mons, making their extraction from matter with de magnetic fields and their
subsequent detection very simple in principle. For instance, the energy loss in
matter by a relativistic Dirac monopole is approximately the same as that of a
relativistic heavy nucleus with Z = 137n/2. It can presumably be distinguished
fwm suclt a 11 uclcus if it is bwuglrt to rest because it will not show an inct case
in ionization at the end of its range (see Problem 13.11).
6.12 Discussion of the Dirac Quantization Condition

Semiclassical considerations can illuminate the Dirac quantization condition
(6.153). First, we consider the deflection at large impact parameters of a parttcle
of charge e and mass m by the field of a stationary magnetic monopole of mag-
nettc charge g. At suffictently large tmpact parameter, the change m the state of
motion of th@ charg@d particle can b@ d@termined by computing the impulse of
the force, assuming the particle is undcflccted. The geometry is shown in Fig. 6.6.
The particle is incident parallel to the z axis with an impact parameter b and a
speed v and is acted on by the radially directed magnetic field of the monopole,
B - gr/41Tr 3 , according to the Lorentz force (6.113). In the approxtmatwn that
the particle is undellectcd, the only force acting throughout the collision is a y
component,
(6.154)
276 Chapter 6 Maxwell Equations, Macroscopic Electromagnetism, Conservation Laws-Sf
Jl L ~H
I ?:
----- p----£1 v e, m Bz
I
: v
b r
/
/ J<'igure 6.6 ~haJXed_IJarticle
g v " m"unPtic•
_,
0 large impact parameter.
The impulse transmitted by this force is

~vbfoo & ~
ll.py = 4.. 00 fh2 -+ ,2,2)312 = ').,.h (6.155)
<;:inr~e thP imnnl"' i< in thP ' thP ~·>rti0l~ io ,.jpflp~f, ,,.j Jt .• +" _,
~+ ,-,, L L :._ ' ' <~~ -' ' • ·' · • •
V< < 'b' v.v, <Hm >o, " ' <H~ angutar lllldl Ulll;:;\,l!UII. ' - " _, liiCO jJdll]l,ll;:; S
momentum is changed by the collision, a result that is not surprising in the light
of the noncentral nature of the force. The magnitude of the change in angular
momentum is somewhat surprising, however. There is no z component of L ini-
tially, but there is finally. The change in Lz is
eg
AT -hAn - {F. 1 <;F.\
- 27T
The change m the z comr anent of anrl'\ll"r 1t11m of the ]p io incJp_
pendent of the impact parameter b and the speed v of the charged particle. It
depends only on the product eg and is a universal value for a charged particle
passing a stationary monopole, no matter how far away. If we assume that any
change of angular momentum must occur in integral multiples of h, we are led
immediately to the Dirac quantization condition (6.153). *
The peculiarly universal character of the change in the angular momentum
to. DO J or a cnargea particle m oassmg a ma2:net1c monooole can be understood
h" l'ClnoiclPrinu fhr ·~nuu] •r m~mP~h•m .1 ' ·~ h ,],.j, ,f
'
J ~ ~ r
iu Lm; p•~o~ .. --~ ut a point magnetic monopole. H rne monopole g 1s at
x = R and the charge e is at x = 0, as indicated in Fig. 6.7, the magnetic and
electric fields in all of space are
1 1 1'
g 1) g n' e e n
H = - 4.,., V . r' = <f.,... ,•2' E= -d- v -
·~ r
= ---
d 2 (6.157)
-v \. I -v
mhPrP r'
I
~A
~·
UJ anu "'•
lv R I
'
~~opv'-'llvvl.
r
,
[~I
• n "'
I .-1
'
' lite angutar momenLUm Lem lS giVen uy ue VOlUme

,
.~
-·
integral ofx X_!!. ~ (F, X Hl!r2 i< thP p], ll1P'nPt~m <1Pnoihr
*This argument is essentially due to AS. Goldhaber, Phys. Rev. 140, B1407 (1965).
Sect. 6.12 Discussion of the Dirac Quantization Condition 277
e Vn
l/ ·Y
x-""
/ Figure 6.7
The total momentum of the fields Pem (volume integral of g) vanishes. This fol-
lows from the fact that Pem is a vector and the only vector available is R. Thus
Pem = (R/R)P, where P is the volume integral of g • (R/R). But g · R oc
R · (n x n'). Since R lies in the plane defined by the vectors nand n', the triple
.. ].. . .n ••
,].
~
rl. ,. ' rl '
"· e1n •
CJ'J.,' . - J., .
-o
•• •J.. ••• _,
means that the angular momentum
Lem -- c?
1 J X X (E X H) d 3 X (6.158)
is independent of choice of origin. To evaluate Lem one can first substitute from
(6.157) for the electric field:
41T {1 {1m
T
em ) r
~ /,
\.
~ IJ\
., ,13
J r'
.,.
.(, IJ\1
'JJ
.13
J..l..o
Using a vector identity from the front flyleaf, this can be expressed as
4'lTLem = -e J (B · V)n d x 3
where B = J..1.. 0 H. Integration by parts gives
A _T ( f'" D\ J3 ( In \ J
-. 1 .L.Iem ~ J .., • ~~ U A ~ Js .. ,~ USJ •u
where the second integral is over a surface S at infinity and n5 is the outward
normal to that surface. With B from (6.157) this surface integral reduces to
~gi41T)J n au u, smce n ts raamuy atrectea ana nas zero angular average. :since
B is caused bv a ooint monooole at x = R its diver2.ence is V • B = !! Mx - R).
The field angular momentum is therefore*
eg R
L -- (6.159)
4'lT J(
*This result was first stated by J. J. Thomson, Elements of the Mathematical Theory of Electricity and
Magnetism, Cambridge University Press, Section 284 of the third (1904) and subsequent editions.
... "'
~ vo w " . .v "' v v ..
momentum, Thomson deduces the magnetic part e(v X B) of the Lorentz force.
278 Chapter 6 Maxwell Equations, Macroscopic Elcctroma211etism Conservation Laws -"T
Tt ;o -•· .1 ol, th 1' c. +1. ,J +1. •

~ ~b' 'b' nas
auu
magnitude equal to the product of the charges (in SI units) divided by 47T. If we
nnw think nf the rnlli<inn • ~f J;";~ h h on rl thP tr.to l on m ] .• r m,
·o " " Ul
the system, that is, the sum of the angular momenta of the particle and the
electromagnetic field we see that the total an g:uJar m lliDis_con · " 'rJ.
change (6.156) in the angular momentum of the particle is just balanced by the
change in the electromagnetic angular momentum (6.159) caused by the rever<ol
of the direction R. A systematic discussion of the classical and quantum-me-
chanica! scattering problem, including the electromagnetic angular momentum
is given by Goldhaber (Zoe. cit.).
'T'1. 'T'L < • •
i Uv i ivOUH \ U.LJ7 J wa' u,c;u uy Jdlld. auu l cnuy by
Wilsont to derive the Dirac condition (6.153) by semiclassical means. To get n/2
. "' -1 ,f' ,I ,J "'-"' -' . '-' " ..
JJ " J ,,,_ HviU iO ' H <O , y
to postulate half-integral quantization of L"'"' a somewhat undesirable hypothesis
fr.rth00l. '"'''".
Finally, we pr~sent a simplified discussion of Dirac's original (1931) argument
" ' 10 to (6.1_53_l_ln_ c~; 0,.,.,.;ncr thP merh~nir<_of_an_ ·" ;., th
1
presence of a magnetic monopole it is desirable to change as little as possible of

the formalism of electromag:netic inteprtions and to keen, for __examnlP thA
interaction Hamiltonian in the standard form,
e e
H mt = e<l> - - p · A + -Z A · A
m m
where and A are the scalar and vector potentials of the external sources. To
do this with a ma2,netic chame it is w•r to
-·
• ll n_m:tifi_ce___Ihp m
00
charge g is imagined to be the end of a line of dipoles or a tightly wound solenoid

.;0
that stretches off to infinity as shown in Fig. 6.8. The monooole and its attorh0rl
string, as the line of dipoles or solenoid is called, can then be treated more or
less normally within the framework of conventional electromagnetic interactions
where B = V x A, etc. From (5.55) we see that the elemental vector potential
dA for a magnetic dipole element dm at x' is
I \
0 1
dA(x)
..,~ dm x V \lA AI
I (6.160)
Thus for a string of dipoles or solenoid whose location is given by the string L
the vector potential is
r I \
AL(x)=-Ji.:j dlxV!I ~ , 1 (6.161)

47T L \ X - X 111
For all points except on the string, this vector potential has a curl that is directed
radially outward from the end of the string, varies inverselY with distance
squared, with total outward flux g, as expected for the B field of monopole g. On
the string itself the vector potential is singular. This singular behavior is equiv-
alent to an intense field B' inside the solenoid and bringing a return contribution
of flux (-g) in along the string to cancel the pole's outward flow. So far we have
*M. N. Saha, Indian J. Phys. 10, 141 (1936); Phys. Rev. 75, 1968 (1949).
'H. A Wilson. Phys. Rev. 75, 309 (1949).
Sect. 6.1~ Uiscussion of the unac (.luanuzauon ~onomon
--fl
~17
/B B
"A 1/ "~ v /
/ ·~ / I~
\
l
d1
!
I
I
f
I
0
L ' X
__.p
I
I
/ _;ff
r;r
(
!
/
(
f X/
~
~
\
\
Figure 6.8 Two representations of a magnetic monopole g, one as the termination of a

. .
tine o1 aipotes ana me otner as me ena ora WUUllU HU, UUlll SLilllgS
stretching off to infinity.
just described a long thin solenoid. To exhibit the field of the monopole alone
we write
Bmonopolc = V X A - B'
where B' exists only on the string (inside the solenoid). Dirac now argued that
·• ,, • • rd _, .,, • ,/ ,-,, +1-
<V u"'"" uv tuv """' v. "'" "' "~
' VTTIT"U" ·r
with a long thin solenoid, it is mandatory that the electron never "see" the sin-
gular field B'. He thus required the electronic wave function to vanish along the
· Th;, ~r~ ,,_ • • 1~•, h ~o J--,,,n · · ;7,.-1 1--.nt rli<rn"inn nf <nrh
·o· · r J
leads us too far afield and is not central to our limited purpose. Dirac's later work
U ';!41S) treats tne questton or me unoDservaoitity or tne strings in ue1ai1.
If (6.161 1 for A. (x) is accented as the annronriate vector ootential for a
monopole and its string L, there remains the problem of the arbitrariness of the
. . -'- -'-' _, _,
Ul Lilt; • 'b' _J, Lllt; PHYMCiil UUSt;l V w,-.,, '"vuou ""' .r vu
where the string is. We now show that a choice of different string positions is
equivalent to different choices of gauge for the vector potential. Indeed, the
requirements of gauge invariance of the Schrodinger equation and single-val-
uedness of the wave function lead to the Dirac quantization condition (6.153).
Consider two different strings L anCfT', as snown m .btg. b.'J. The difference ot
the two vector potentials is given by (6.161) with the integral taken along the
0 o L ~ T ' T 0 d. ~ ~ ~ - L C 1 +I. ' I. •
paw v L L aovu"u '"'-'~---=y-rn::nn~•u,. ' ,.
Ac(x) = AL(x) + ..£. Vilc(x) (6.162)

47T
where ilc is the solid angle subtended by the contour Cat the observation point
x. Comparison with the gauge transformation equations, A--+ A' - A + vx.
m m' m f1t~\t~ ,,;:;,\ ~h~"'" thnt .. nh .. ~ .... '~ ct · " ' frr.m l tr. T' ;, ,,m;v.
\ '/\~"- .,, 0
•
alent to a gauge transformation, x= gilci47T.
' ·~
Figure 6.9 Two different strings L

and L' give monopole vector
-- r1 porenua1s amenng oy a gauge
tr "tinrti • tht> ""
OJ)'I/"--------y of the solid angle flc(x) subtcnded at

the observation point P by the
surfaceS spanning the contour C =
X
L'- L.
cou:o"' """" ' tm: 1orm m we ""curoumger equauon mvanant,

provided the wa~e function is transformed according to
1/J ~ if/ = lj!e'exlr.
where e is the charge of the particle and xis the gauge function. A change in the
location of the string from L to L' must therefore be accompanied by a modifi-
cation of the nh"'"c...Of the w"v" fn of thf'. cl
(6.163)
Since De changes suddenly by 47T as the electron crosses the surface S, the wave
function will be multiple-valued unless we require
eg
'"
This is the Dirac quantization condition (6.153). It follows from the general re-
quirements of gauge invariance and single-valucdncss of the wave function, in-
dependent of the location of the monopole's string.
The preceding discussion of magnetic monopoles presents only the most ba-
oi~ 11. " '" • __,;+; ,;." ,j" ;\, ·•" "
.. -r . . ,. cu•••
"vnu•uvu, '1'" a• a 'iuanmm erectrouynamtcs wtw magneuc monopoles ana
· : C llnd nthPr o ThP · I rP"r1Pr ron nnronP •J.~ c,J...;~~•
' "J
through the article by Goldhaber and Trower (op. cit.) and the references at the
end of the chapter.
h./.1 ,..., l7 1-"'0fPntin/..: { l :I
Tt 1s sometimes useful to utilize potentials other than the standard scalar and
vector potentials as auxiliary fields from which to determine the fundamental
electromagnetic fields. The most important of these are the polarization potentials
"Inc acrnonstnl[10ll Is very easy. See. tor example. H. A. Kramers, Quantum Mechanics, North-
Holland, Amsterdam, (1957): Dover reorint (1964). Section 62.
~
"ru. UoJ·~ - - "'

'l!i!1
Ul nertz vcuv•o, lllll uy ~

. •_o • " " A o \
-o· V-'V')· r>.o tH-.o
suggests, these potentials put the electric and magnetic polarization densities to
the fore. We consider linear, isotropic media with sources of external polarization
densities, Pext and Mext• but no separate macroscopic charge or current. The me-
dia are described by electric and magnetic susceptibilities, E and J.L. [The realistic
situation of frequency-dependent quantities can be abstracted by choosing a
unique sinusoidal time dependence and then using Fourier superposition.]
~·-
l
. "-'-'
" " <HU"1 1
. · ll"1UO U 1 "
T\ -IC -1- D n .u -1- u .. M lf.1f..11

t:XL'
'
Then with the standard definitions (6.7) and (6.9) of the fields in terms of the
scalar and vector potentials, the macroscopic Maxwell equations yield the wave
equations,
2
J.LE -
il A
-
V' 2 A = J.L -
ilPext
- + J.Lo'V X Mext (6.165a)
at 2
at
az<P 1 ......
/"' ;u2
'<'72 ....
c
ext
fr:.1r:.<:h\
\"' .,
with
a<P
v .A + J.LE - =0
at
as the Lorenz condition. Two vector polarization potentials, Tie and Tim, are
introduced by writing A and <P in a form paralleling the structures of the right-
hand sides of the wave equations (6.165), namely,
• dlle ..., . l'l":' TT

,fcc.. ccr:.\
~
/" ;;, 1"0
~
urn, ·" e
1
'!
"
When we substitute these definitions into (6.165), we find that the Lorenz con-
dition is automatically satisfied. The wave equations become the following equa-
tions for Tie and Tim:
V • [ 2
V' Tic - J.LE
2
a lle +
----af Pext J= 0
(6.167a)
2
rl :~2JT il TI
J.L V'Llle J.LE -;;- + Pcxt + J.LoV X VLllm J.LE .? + lVlext -U
(}[ (}[ U<
(6.167h)
From (6.167a) we find that the square-bracketed quantity can at most be equal
to the curl of some vector function, call it (J.Lol J.L)V. When this form is inserted
into (6.167b), we have a vanishing curl of a vector quantity that must therefore
be equal at most to the gradient of some scalar field, call it at;/at. The result is
that the Hertz vectors satisfy the wave equations,
a2TI. Uc
J.LE --:;- V~lle ~'ext v " v ~ o.10caJ
Ut I"
ri2Jl av aE
J.LE 7 VLllm - "Mext + + v ( b.lbi'Sb)
at Ut vo
282 Chapter 6 Ma:nlell Equations, Maeroseopie Eleetromagnetism, Consenation Laws SI
It is left to the problems (Problem 6.23) to show that the arbitrary functions v
and gmay be removed by a gauge transformation on the polarization potentials
We may thus set V and g equal to zero with no loss of generality.
The electric and magnetic fields are given in terms of the Hertz vectors by
E e m
V(V ·He) p.. p.. 0 V X (6.169a)
E at 2
at
B= ,.~.;v
an c +
x __ MoV X V X Ilm (6.169b)
at
Ontside the source Pext the wave equation (6.168a) can be used to express E in
a form analogous to (6.169b) for B wnh the roles of the electnc and magnetic
Hertz vectors interchanged.
The wave equations for Ile and Ilm have solutions that are particularly simple
if the external polarization densities arc simple. For example, a time-dependent
magnetic dipole at the point x0 has a magnetization density,
m(t)lllx
From the form of the wave equation (6.41) and its solution (6.47), we deduce
that the magnetic Hertz vector is
m(t- v;;:ER)
where R = lx- xol·

Tllustrations of the usc of polarization potentials can be found in Born and
Wolf, in Stratton, and in Panofsky and Phillips, who discuss elementary multipole
radiation in terms of a Hertz vector. We find it adequate to work with the usual
potentials i\. and or the fields themselves.

The conservation laws for the energy and momentum of electromagnetic fields are
discussed in almost all text books. For example,
I'anofsky and Phillips, Chapter 10
Stratton, Chapter II
Landau and Lifshitz, Electrodynamics of Continuous Media (Sections 15, 16, 34)
discuss the Maxwell stress tensor in some detail in considering forces in fluids and solids.
as has already been mentiOned. I he descnptwn of resonant cavit!Cs as cucuit elements is

treated in a classic paper by
W. W. Hansen, J. Appl. Phys. 9, 654 (1938).
A thougflt provoking discussion of the deiivation of the 111ac1 oscopic equations of
Ch. 6 Problems 283
The derivation of the macroscopic Maxwell equations from a statistical-mechanical point

of view has long been the subject of research for a school of Dutch physicists. Their
conclusions are contained in two comprehensive books,
de Groot
de Groot and Suttorp
A treatment of the energy, momentum, and Maxwell stress tensor of electromagnetic
fields somewhat at variance with these authors is given by
Haus and Melcher

For the reader who wishes to explore the detailed quantum-mechanical treatment of
d1electnc constants and macroscopic field equatwns m matter, the tollowmg are suggested:
S I Adler, Phy< Rev 126 413 (1962)
B. D. Josephson, Phys. Rev. 152, 21 (1966).
G. D. Mahan, Phys. Rev. 153, 983 (1967).
Symmetry properties of electromagnetic fields under reflection and rotation are dis-
cussed by
Argence and Kahan
The subject of magnetic monopoles has an extensive literature. We have already cited
the paper by Goldhaber and his review with Trower, as well as the fundamental papers
of Dirac. The relevance of monopoles to particle physics is discussed by
J. Schwinger, Science ](iS, 757 (1969).
The interest in and status of searches for magnetic monopoles up to the 1980s can be
found in
R. A. Carrigan and W. P. Trower, Magnetic Monopoles, NATO Adv. Sci. Inst.
Series B, Physics, Vol. 102, Plenum Press, New York (1983).
The mathematical topics in this chapter center around the wave equation. The initial-
value problem in one, two, three, and more dimensions is discussed by
Morse and Feshbach (pp. 843-847)
and, in more tnathernatical detail, by
Hadamard
Problems
6.1 In three dimensions the solution to the wave equation (6.32) for a point source in
space and time (a light flash at t' = 0, x' = 0) is a spherical shell disturbance of
radius R = ct, namely the Green function G(c) (6.44). It may be initially surprising
that in one or two dimensions, the disturbance possesses a "wake," even though
the source is a "pmnt" m space and !Jme. I he solutwns for fewer d1menswns than
three CllO be found by svperposjtjon jo the superfiJJOJJS dimcnsion(s) to eliminate
dependence on such variablc(s). For example, a flashing line source of uniform
amphtude IS eqmvalent to a pomt source m two dimensions.
(a) Starting with the retarded solution to the three dimensional wave equation
(6.47), show that the source f(x', t') = ll(x')ll(y')ll(t'), equivalent to a t = 0
point source at the origin in two spatial dimensions, produces a two-dimen-
sional wave,
2c8(ct - p)
']! X, y, t)
p
whet e p2 x 2 I y2 aHd 8(0 is the unit step function [®(g) 0 (I) if

g < (>) 0.]
(b) Show that a "sheet" source, equivalent to a point pulsed source at the origin
in one space dimension, produces a one-dimensional wave proportional to
'lf(x f) - 27rcEJ( ct lxl)
o.~ 1 ne cnarge anu current densities tor a smgJe pomt cnarge q can De wnttcn formo 11.,
u ~
p(x', t') = qil[x' - r(t')]; J(x', t') = qv(t')il[x' - r(t'))
where r(t') is the charge's position at time t' and v(t') is its velocity. In evaluating
expressions involving the retarded time, one must putt' = t,·et = t- R(t')lc, wher"
R X rit'Ubut R X x' (t') inside the delta fu J5l
taJ As a preummary to denvmg the Heav1s!dc Fcynman expressions for the elcc-
tric and magnetic fields of a point charge, show that
J d'x' il[x' - r(t,et)l = ~

where K 1 v ·Ric. Note that K is evaluated at the ret~rc:lec:l ;m.,
\DJ ;:,ranmg wun me Jenmenko general!zatwns or tne couJomtJ an<.l Hiot Savart
laws, use the expressions for the charge and current densities for a point charge
and the result of part a to obtain the Heaviside-Feynman expressions for the
electric and magnetic fields of a point charge,
r A
q R i) R i) v
...., n2 -r
.n 2
' co ~I
ret cv _nn ret c '" '"' ret
auu
B =
~47T0 q{[vxR]
KR2 cet + cat
[vxR]}
KR ret
iJ
(c) In our notation Feynman's expression for the electric field is
q
A
K ]](Jeer d K
- d' ,.,; 1
"' 4m"o R2 ret c at R2 ret
c2iJ 2 , . .
1
Jrct
while Heaviside's expression for the magnetic field is
B = ~oq vxR 1 i) vxR

47T K2R2 + dRl. at _K_
S~lPn~P ~f th" ""~ cPtc ~f ...£=..±1 c.£.. I rl .

•
References: 0. Hcaviside, Electromagnetic Theory, Vol. 3 (1912), p. 464, Eq. (214).
R. P. Feynman, The Feynman Lectures in Physics, Vol. 1 (1963), Chapter 28, Eq.
(28.3).
6.3 The homogeneous diffusion eauation ( 5.160) for the vector potential for ouasi-static
fiPlrlc ;n ,rJ " . rl' 1 ,, .. ,, .,. - '-'
.£
~·
V< <HC W<w,
A(x, t) = J d x' G(x- x', t)A(x', 0)

3
where A(x', 0) describes the initial field configuration and G is an appropriate

1\,ernet.
Ch. 6 Problems 285
(a) Solve the initial value problem by usc of a three-dimensional Fourier trans-
form in space for A(x, t). With the usual assumptions on interchange of orders
of integration, shov: that the Green function Ius the Fourier representation,
X - X , t - -
21T(
)-_, . e
and it is assumed that t > 0.

(b) By introducing a Fourier decomposition in both space and time, and perform-
ing the frequency integral in the complex w plane to recover the result of part
a, show that G(x x', t) is the diffusiOn Green functiOn that sat1s1Ies the ill-
homogeneous equation.
ac;
/Pl(x x')i5(t)
at
and vanishes for t < 0.
(c) Show that if (Tis uniform throughout all space, the Green function is
1 "'·1~1
2
G(x, t; x', 0) 0(t)(::r exp( xT)

(d) Suppose that at timet' = 0, the initial vector potential A(x', 0) is non vanishing
only in a localized region of linear extent d around the origin. The time de-
pendence of the fields is observed at a point P far from the origin. i.e., IxI =
r >> d. Show that there are three regimes of time, 0 < t ""' T 1 , T 1 ""' t ""' T 2 ,
and t >> T2 • Give plausible definitions of T1 and T2 , and describe qualitatively
the time dependence at P. Show that in the last 1egime, the vecto1 potential
is proportional to the volume integral of A(x', 0) times t 112 , assuming that
integral exists. Relate your discussion to those of Section 5.18.B and Problems
6.4 A uniformly magnetized and conducting sphere of radius R and total magnetic
moment m = 47TMR 3 /3 rotates about its magnetization axis with angular speed w
In the steady state no current flows in the conductor. The motion is nonrelativistic;
the sphere has no excess charge on it.
(al By considering Ohm's law in the moving conductor. show that the motion
induces an electric field and a uniform volume charge density in the conductor.
? 3
(b) Because the sphere is electrically neutral. there is no monopole electric field
outside. Use symmetry arguments to show that the lowest possible electric
multipolarity is quadrupole. Show that only a quadrupole field exists outside
and that the quadrupole rnontent tenso1 has non vanishing co111ponents,
Q 11 = -4mwR 2 !3r?, Q,, = Q22 = -Q"/2.
(c) By considering the radial electric fields inside and outside the sphere, show
that the necessary surface-charge density (T(IJ) is
4
rr( li) =
4
";;R 2 • ~w · [ 1 - ~ P 2( cos li)]
(d) The rotating sphere serves as a unipolar induction device if a stationary circuit
is attached by a slip ring to the pole and a sliding contact to the equator. Show
286 Chapter 6 Maxwell Equations, Macroscopic Electromagnetism, Conservation I.aws-SI
+l. <L ,. • •
'"''- . 0 v• '"'- . uv<u U Ulll lilt: t:lj UdlUl w we pole
contact (by any path) is jg = JLomw/47rR.
[See Landau and Lifshitz, Electrodynamics of Continuous Media, p. 221, for au
alternative discussion of this electromotive forcc.l
u... ,. . vn.ovu;..; cnarge uistriomion proouces an electrostattc field,
V. 1!:
Into this field is placed a small localized time-independent current density J(x),
which generates a magnetic field H.
(a) Show that the momentum of these el
formed to
Pfieid = 21
c .
r<PJ d'x-
provided the product <PH falls off rapidly enough at large distances. How
rapidly is "rapidly enough"?
(b) Assuming that the current distribution is localized to a region small compared
~ . . . . .
w cH<O ' ' " ' " " Ul vauauuu U> " ' " <Ol<Ovlu..: uem, ' tne eieCtrostanc PD_lentull_
m a 1 ay1or senes and snow tnat
1
Pficlu = 2 E(O) X m
c
w " ' ' ' " 1'.\ V) is Llle electric uem at tne current dtstnoutwn ana m ts tne magnetic
moment, (5.54), caused by the current.
(c) Suppose the current distribution is placed instead in a uniform electric field
E...o_(_fillin>' all soace ). Show that...Jlil..1llil1r hnw lirM<erl i< th" I~
'J: D ~J ~ dHl<O!Olal '-VHlU' 'HV<H
equal to minus one-third of the result of part b, yielding
Compare this result with that obtained by working directly with (6.117) and the
considerations at the end of Section 5.6.
a w1oiua, coi, o, mean ramus a ana IV rurns, w1tn a smau
umiorm cross section of area A (both height and width small compared to a).
The toroid has a current I flowing in it and there is a point charge Q located
at its center. Calculate all the components of field momentum of the system;
show that the component along the axis of the toroid is
w..e,e Llle sign uepenas on me sense or tne current now m tne coil. Assume
that the electric field of the cha.J:2:e. c nnimn"rl"rl intn thP ,f
nonvanishing magnetic field, as would happen for a toroid that is actually a

set of N small nonconducting tubes inside which ionized gas moves to provide
the current flow.
Vl l l' v.Jv.
(b) If Q - 10 " C ( ~ 6 X 10 12
electronic charges), I - l.O A, N - 2000, A -
4 2
10- m , a = O.l m, find the electric field at the toroid in volts per meter, the
inuuccion in tesm, anu Llle eLectromagnetic momentum m newton-
Ch. 6 Problems 287
seconds. Compare with the momentum of a 10 p.g insect flying at a speed of
[Note that the system of charge and toroid is at rest. Its total momentum must
vanish. There must therefore be a cancelmg "hidden ' mechamcai momentum-see
Problem 12.8.
6.7 The microscopic current j(x, t) can be written as
j(x, t) = 2.: q;v;8(x - xj(t))
j
where the point charge qj is located at the point xj(t) and has velocity vj = dxj(t)/dt.
Just as for the charge density, this current can be broken up into a "free" ( conduc-
tion) electron contribution and a bound (molecular) current contnbutwn.
Following the averaging procedures of Section 6.6 and assuming nonrelativistic
addition of velocities, consider the averaged current, ij(x, t)).
(a) Show that the averaged current can be wntten m the form of (6.96) wnh the
definitions (6.92), (6.97), and (6.98).
(h) Show that for a medium whose internal molecular velocities can be neglected,
but which is in bulk motion (i.e., v,. = v for all n ),
1
- B - H = M + (D - EoE) X V
!Lo
This sho\\S that a moving polarization (P) produces an effective magnetization
density.
Hints for part a: Consider quantities like (dp,.ldt), (dQ~<;>idt) and see what they
look like. Also note that
df (x - x,.(t)) = -v,. · Vf(x - xn(t))
68 A dielectric sphere of dielectric constant E and radius a is located at the origin.

There is a uniform applied electric field £ 0 in the x direction. The sphere rotates
with an angula1 velocity w about the z axis. Show that there is a magnetic field
5
3 E-E0 a
<PM = - EoEo
where r > is the larger of r and a. The motion is nonrelativistic.

You may use the results of Section 4.4 for the dielectric sphere in an applied
field.
6.9 Discuss the conservation of energy and linear momentum for a macroscopic system
of sources and electromagnetic fields in a uniform, isottopie medium described by
a permittivity E and a permeability p.. Show that in a straightforward calculation
the energy density, Poynting vector, field-momentum density, and Maxwell stress
tensor are given by the Minkowski expressions,
S=ExH
g = p.EE X H
r
T;jEE,Ej + p.H,Hj ' c E" +
2(jij E p.li~Jl
What modifications arise if E and p. are functions of position?
2~~ rh~ntPr li Maxwell Eouations. M r.
6.10 With the same assumptions as in Problem 6.9 discuss the conservation of angular
momentum. Show that the differential and integral forms of the conservation law
are
rl "
iJ( ( :J:mcch + ;£field) + V •M 0
and
d
d
11 un
V'·mcch
c.
ohfieJd) "
'
"'-
(
I .. · ,., uu v
where the field angular-momentum density is
.'fr.c~d = X X g = f.LE X X (E X H)
<>Hu 111c 11ux vi <>Hguwr momentum is aescrioea oy me tensor
Note: Here we have used the dyadic notation for M,1 and T,r A double-headed
__a_fairlj,c_ -'- · mP~nino Pm Pv~mnlP n .~ ·'- ''"
' '\~ ~A ~• .,.__,. . . , J
•u ~,. . , • • • ij· ~ "" -"""' m '-"11 u<O "'"'"11 "' a " ' " u-nmK Lensor,
M,ik = T,1xk ~ T,kxr But in the indices j and k it is antisymmetric and so has only
three independent elements. Including the index i, M,1" therefore has nine compo-
nents and can be written as a pseudotcnsor of the second rank, as above.
6.11 A transverse nlane wav" i< inrirlPnt nmm"ll" in ""r mm rm " -'- ·-'-"
flot ' " "o
(a) From the law of conservation of linear momentum, show that the pressure
(called radiation pressure) exerted on the screen is equal to the field energy
per unit volume in the wave.
II.\ T •'- • ~> • c
,~, <11<0 <Oaau CH<O HUA U'-'-" U _'_"'1S"'""'-
"'" v~ I llUIII lflC u~
sun is approxJma,e!y 1.4 KW/m. u an mterp1anetary saup1ane" nad a sail of

mass 1 g/m 2 of area and negligible other weight, what would be its maximum
acceleration in meters per second squared due to the solar radiation pressure?
How does this compare with the acceleration due to the solar "wind" (cor-
puscular radiation)?
u.>~ u1e uc,iui1iun or tne aominance r u LJJ or a two-termmal linear
passive network in terms of field quantities by means of the complex Poynting
theorem of Section 6.9.
(a) Bv considering the comnlex coniwmte of (6.134\ obtain <>eneral
o·
radiation loss.
(b) Show that at low frequencies the expressions equivalent to (6.139) and (6.140)
are
1 I '"
I II
sneers or mmenswns a and b separated by a distance d small compared to a or b.

Current is fed in and taken out uniformly along the adjacent edges of length b.
Ch. 6 Problems 289
the input impedance or admittance using the field concepts of Section 6.9.
(a) Calculate the electric and magnetic fields in the capacitor correct to second
order in powers of the frequency, hut neglecting fringing fields.
(b) Show that the expansion of the reactance (6.140) in powers of the frequency
consisting of a capacitance C = E0 ab/d in series with an inductance L =
connected to a current source hy axial leads, as shown in the sketch. The current
in the wire is l(t) = J0 cos wt.
Problem 6.14
in powers of the frequency (or wavenumber), neglecting the effects of fringing

fields.
(b) Calculate the volume integrals of w, and wm that enter the definition of the
these energies are
w d3x = _1_ II,I2 d

e '
(c) Show that the equivalent series circuit has C = 7TE0a2/d, L = f1-odi87T, and that
an estimate for the resonant frequency of the system is w,e, = 2v'2 cia. Com-
pare with the first root of J0 (x).
6.15 If a conductor or semiconductor has current flowing in it because of an applied
of electric field in the direction orthogonal to hoth the applied electric field (direc-
reflections and the assumption of Taylor series expansions around zero mag-
have the form
+ /Mf2J + f3z(H • J)H

where p0 is the n~sistivity in th8 absencg of the magnetic field and R is called
(h) What about the requirements of time tevctsal invatiance?

6 16 (a) Calculate the forcg in ngwtons acting on a Dirac monopole of tile minimum
111agnetic clratge located a distance 0.5 A from and in the median plane of a
magnetic dipole with dipole moment equal to one nuclear magneton ( efi/2mp).
(b) Compare the force in part a with atomic forces such as the direct electrostatic
force between charges (at the same separation), the spin-orbit force, the hy-
pcrfine interaction. Comment on the question of binding of magnetic mono-
poles to nuclei with magnetic moments. Assume that the monopole mass is at
least that of a proton.
Reference: D. Sivers, Phys. Rev. D2, 2048 (1970).
6.17 (a) For a particle possessing both electric and magnetic charges, show that the
generalization of the Lorentz force in vacuum is
(b) Show that this expression for the force is invariant under a duality tramfor
mation of both fields and eharges, (6.151) and (6.152).
(c) Show that the Dirac quantization condition, (6.153), is ggneralized for hw
pat tides possessing elecLI ic and tnagnetic charges eb g 1 and e2• g 2 , respccnvely,
to
and that the relation is invariant under a duality transformation of the charges.
6.18 Consider the Dirac expression
- L
A(x) - 47T.
r dl' lx -(xx'l'- x')
L
X
for the vector potential of a magnetic monopole and its associated string L. Suppose
for definiteness that the monopole is located at the origin and the string along the
negative z axis.
(a) Calculate A explieitly and show that in sphe1 ical comdinates it has
componcn s
A, 0, 0,
7TT Sill 1)
(b) Verify that 8 = V X A is the Coulomb-like field of a point charge, except

perhaps at e = 7T.
(c) With the 8 determined in part b, evaluate the total magnetic flux passing
through the circular loop of radius R sin e shown in the figure. Consider
e < ?T/2 and e > 7T/2 separately, but always calculate the upward flux.
(d) From# A· dl around the loop, determine the total magnetic flux through the
loop. Compare the result with that found in part c. Show that they are equal
for 0 < (J < 7112, bnt have a wnstmlt difference for 7#2 < !) < 'IT. Interpret this
Ch. 6 Problems 291
./ !
I
I
I
I
I
L Problem 6.18
6.19 (a) Apply space inversion to the monopole vcctm potential of l'wblent 6.18 and
show that the vector potential becomes
A~ = -g (1 + cos fJ) = -_f._ cot(!!.)

47Tr sin fJ 47Tr 2
with the other components vanishing. Show explicitly that its curl gives the
magnetic field of a magnetic monopole, except perhaps at (;! = 0. [Remember
the space-inversion properties of the magnetic charge']
(b) Show that the difference, 8A = A' - A, can be expressed as the gradient of
a scalar function, indicating that the original and space-inverted vector poten-
tials differ by a gauge transformation.
(c) Interpret the gauge function in terms of Fig. 6.9. [Hint: Choose the contour C
to be a rectangle lying in a plane containing the z axis, with three sides at
infinity.)
6.20 An exalttple o1 the pt cscr valion of causality and finite speed of pt opagation in spite
of the use of the Coulomb gauge is afforded by a dipole source that is flashed on
and off at t - 0. The effective charge and current densities are
p(x, t) 8(x)8(y)b'(z)b(t)
!Jx, t) - 8(x)8(y)8(.:)8'(t)
where a prime means differentiation with respect to the argument. This dipole is
of unit stnmgth and it points in the nsgati"s z direction
(a) Show that the instantaneous Coulomb potential (6.23) is
1 z
(b) Show that the transverse current J, is
J,(x, t)
where the factor of 2/3 multiplying the delta function comes from treating the
gradient of z lr 3 according to (4 20)
(c) Show that the electric and magnetic fields arc causal and that the electric field
components are
1 r ~ 0
t-Ax. t) - 4 - -
7TE0 r
-o"(r- ct) +-
r
o'(r- ct) - -2 o(r- ct)
r
sin e cos e cos rb
E, is the same as Ex, with cos replaced by sin </J, and
1 c o'(r- ct) O(r - ct
~ '
'-~z\A• ·'
')
. ?
ow vu \' L<) \·' ~v.
?
I) . I
47TE0 r
fJ
r r2 I
Hint: While the answer in part b displays the transverse current explicitly, the less
expncn wrm
J,(x, t) = -o'(t)[ E3 o(x) +

4~ V ~ G) J
can be used with (6.47) to calculate the vector potential and the fields for part c.
An alternative method is to usc the Fourier transforms in time of J, and A, the
Green function (6.40) and its soherical wave exnansion from Chanter 9.
6.21 An electric dipole of dipole moment p, fixed in direction, is located at a position
r 0 (t) with respect to the origin. Its velocity v = dr 0 /dt is nonrelativistic.
\aJ ;:,now Lnat llle uipo1e s cnargc anu currem denstttes can tle expressed tormally
as
n(x tl - -(n • V\fi(y r.Jtl\· lfv t\ - _,,(~ , U\,l;{v .it\\
"
{h\ <;;hm,v tho< thP ofLrrntPr movinn c1inniP n;.,pc •'c· t~ o
"' -'
oonn .• t '
"~'"
and an electric quadrupole field in 'addition to an electri~"'dipole fleld, with
moments
m = ~p x v
and
(), - 1(r.,.TJ. + .n.\ ?._ • n,l;
rTh<"ff> "r" nf rC>IIf"' <till hi<>hPr .... 1
(c) Show that the quasi-static electric quadruoole field is

1 1
"\") 47TEo r4 lun\n. foJ\" • PJ .Jro[n • p) 5p[n • roJ 5n[r0 • p JJ
where n is a unit vector in the radial direction.

6.22 (a)
..
For the off-center, slowly moving, electric dipole of Problem 6.21, show that
IH~
. .
l.jUUO>-OiaHL HoLIVI r
.
' l"
. .
. uy we currem HOW a"oc1ateu WJLH
the dipole motion is
uov(n. n) u. I (u X vi X X J rn( X • Vl + y( X • n l1
A[X, t)
47Tr2 47T 2 ,.3 ---=F2 ,.3
w11c;u.o "'" w>e '"'11'u' "'" , wu11 :iiTv ana p) is me vecwr
potential of the magnetic dipole whose moment is !'iven in Problem 6.21. The
added term is symmetric in v and p.
(b) ~how that the magnetic field of the symmetric term is
Bsym = 3p.o n
------., X [p(v • n) + v(p • n1 )1
om
Ch. 6 Problems 293
(c) By calculating its curl, show that B,,m is consistent with being the quasi-static
magnetic field associated with the electric quadrupole field of Problem 6.21 c.
(d) Show that the total magnetic field (computed from the first form of the vector
potential, i.e., the sum of B,ym and the magnetic dipole field) is
B vx
4'1T
Comment.
6.23 The wave equations (6.168) for the Hertz vectors contain arbitrary source terms
involving the functions V and f Consider the gauge transformations
II~ - lie + 11-oV X G - Vg; 11'-ll

m m
-n-
r at
where G and g are well-behaved functions of space and time.
(a) Show that, if G and g satisfy the wave equations
then the new polarization potentials II~ and 11,;, satisfy (6.168) with vanishing
V and g.
(b) Show that the gauge transformation on the Hertz vectors is equivalent to a
gauge transformation on A and <1>. What is the gauge function A of ( 6.19) in
terms of G and g?
6.24 A current distribution J(x, t) localized near the origin varies slowly in time.
(a) Use the Jefimenko expressions (6.55) and (6.56) for the retarded fields to
evaluate the quasi-static fields far from the current distribution. Assuming that
there are no electric multipole moments and retaining only the magnetic di-
pole contributions, show that the magnetic and electric fields at the point
(x = rr, t) to first order in an expansion in successive time derivatives are
ric) · r)r m(t ric)]
/.La 1 • am(t - ric)

E ? r X
4'1T r at
(b) The construction and excitation of an infinite, straight, right circular solenoid
of radius a, with N tums per unit length, are such that its current 1{t) is the
same everywhere along its length and is changed very slowly in time. Show
that the fields far from the solenoid are approximately
where pis the perpendicular distance from the axis, provided max( Idlldt/ll)
<< clp A long solenoid with a time-varying current bas an electric field outside
it, in contrast to the static situation. Verify that Faraday's law is satisfied.
2Q4 rhanter 6 Maxwell Enu..tiono. M, . ~
T ~
F
.
1~ F.
'"'
~-
'IS .au
'•'- _,_ T
<u~ ~v•~u.- .
.v<w ~'J,--:>IIOW' lllat 111 lilt: aipo]e
approximation the force acting on a neutral atom at rest can be expressed as
ar "'""' = (d · V) E + dx B
dt
where d is the atomic dipole moment and E and B are the electric and mag.
' LLL _,_ ' .£
<UO U< <H~ OH~ W <H~ U<VU<,
(b) For a uniform plane wave of frequency win a nonmagnetic tenuous dielectric
~· '<L ' " .r L ' '
v<, o< lll<ll lllv lllllv l<llv Ul '-'llallge Q[
"\~h
mechanical momentum per unit volume gmech accompanying the electromag-

netic momentum " . (6.1! R) nf th" "'" · ;,
dgmech = ! (n2 _ I) dgem

U< L lU
[see Peierls (Joe. cit.) for corrections for dense media and non-uniform waves.]
Note of explanation:
"rL
u]- '
~
" " . ~ \ ~" ' •u ... ,.
v<l<lv'V'
_,
UK
of Danish physicist Ludvig V. Lorenz's name instead of Dutch physicist Hcndrik A. Lo-
rentz's with the relation (6.14) between the scalar onrl · YPt it;, of.
that in 1867 Lorenz, in a paper entitled "On the identity of the vibrations of light with
· . c;u11c:ue>, \UJJ. "i'-J exptuiteu ll1e teLarueu somrions ror rnc pmcmiaJs, aenved
(6.14) and equations equivalent to wave equations for the electric field, and discussed the
-' · · ~f 1;nht ~. ~ •' -~-
ously with Maxwell. H~ A. Lorentz has ample recognition j,~ physics ter~inology without
the mis-attribution of (6.14) to him (by others. beginnin!!: around 1900). As Van Blade!*
observes, it is up to textbook authors to accord Lorenz his due.'
*J. Van Blade!, IEEE Antennas and Propagation Magazine 33, No.2, 69 (April1991).
tAn earlier author who deplored the lack of recognition of Lorenz's contributions is A. O'Rahillv.
1\lectromagnetic Theory, Dover Publications, New York (1965) [originally published as Electromag-
netics, Longman Green and Cork University Press (1938)], footnote, p. 184.
This chapter on plane waves in unbounded, or perhaps semi-infinite, media treats
. . . . .
dia-their transverse nature, linear and circular polarization states. Then the
. . . .
coefficient of liquid water over 20 decades of frequency. Then comes a simplified

. . . . . .
in a conducting fluid. The ideas of phase and group velocities and the spreading
. . . . .
erties of a medium are discussed in some detail, including the Kramers-Kronig

. . . . .
eludes with a treatment of the classic problem of the arrival of a signal in a
. . . .
recently subjected to experimental test.
7.1 Plane Waves in a Nonconducting Medium

A basic feature of the Maxwell equations for the electromagnetic field is the
existence of traveling wave solutions which represent the transport of energy
from one point to another. The simplest and most fundamental electromagnetic
waves are transverse, plane waves. We proceed to see how such solutions can be
· · ·m 1 onconductin media described b s atiall constant er-
meability and susceptibility. In the absence of sources, the Maxwell equations in
an infinite medium are
V · B = 0' VxE+-=0
VxH--=0
an
V·D = 0
' at
Assuming solutions with harmonic time dependence e-iwr, from which we can
· · · ations for the am-
V·D = 0
'
V X H + iwD = 0
296 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-51
For uniform isotropic linear media we have D = EE B = ttH Where E ancl

may in general ~e. complex functions of w. We assume for the present that they
au:; 1"m <111u t: • '" ~11v 1osses J· 1 nen rne equanons ror .., anu n are
V X E- iwB = 0 V XB + iwttEE = 0 (7 ')\
The zero-diverg_ence ffiUations are not indenendent but are_ohtainf'el hv t~J.c;"

divergences in (7.2). By combining the two equations we get the Helmholtz wa;~
equation
E
(V + f-tEul") u (7.3)
B
Consider as a possible solution a plane wave traveling in the x direction, eikx-'"''.
(7 'I\ "'"' fine! thP
-, · •~ ,, <l -' 1~ ,,-1 •l c.
"Y
' ' T
w are related by
k=~w (7.4)
1 he pnase vetoczty ot the wave is
w 1 c
u=-=--=-
k ~ n' n= J:fL E
fLoEo (7.5)
The ouantitv n is called the index of refractinn :mel is nsuallv a '' inn r.f fr~_
quency. The primordial solution in one dimension is

u(x, t) = aeik._-.:-iwt + be-ikx-it,Jt
(7.6)
Usin!!: w ku from (7.5) this ""n hP

uk(x, t) = aeik(x-ut) + be-ik(x+vt)
If the medium is nondispersive (~tE independent of frequency), the Fourier su-
..
l'"l" L11t:01em ~L.'+'+ J ana ~L.'IJ) can oe useu w consrruct a general solutwn
of the form
u(x, t) = f(x - ut) + g(x + ut) (7.7)
where f(z) and g(z) are arbitrary functions. Equation (7.7) represents waves
rravenng m me posmve and negative x drrectwns wrth speeds equal to the phase
velocity u.
u LHt: mc:uium is uispersive, me oasrc sotunon U .D) Sllll nows, out when we
hnild liD a Wave aS an ' • fl • nf Y "n.-J f thP .-lionPr<ir.n "r~cln~M
modifications. Equation (7.7) n~ longer holds. The ~ave ch~nges sh~pe as it

orooaQ:ates (see Sections 7.8 7_9 ;mel 7_11)
We now consider an electromagnetic plane wave of frequency wand wave
vecwr K KD ana reqmre that rt satlsty not only the Helmholtz wave equation
{7 'I\ hnt ~l·r. ~n t~ J\A", ,11 · Tt · · .-1' '~ 1' ·
\ -; ''i o •o U ~HH ' ~ J \ o.J} 10 '--0
sentially kinematic; those imposed by the Maxwell equations, dynamic. With the
1tion thM thP nh ·~ol -' -•. on.-1 ,f· f>,],-1, -< . ,,-ll ,,-
co~pl~x quantities, we writ~ the

'J
the real parts of plane wave fields as

E(x, t) = ~eikn·x-iwt
r7.8)
D~.t, L)
""'"'
u
.
"'' m. '"·
.., . .U.J' '10'7
"'
'"~'~ ""v, "" ~-. u
..1
·~
~·
rc .L .. ..,. ,,-1 D •op
(7.3) provided
2
k2 n · n = JLEw (7.9)
To recover (7.4) it is necessary that n be a unit vector such that n · n = 1. With
the wave equation satisfied, there only remains the fixing of the vectorial prop~
erties so that the Maxwell cauations (7.1) are valid. The divergence equations in
(7.1) demand that
n ·~ = 0 and n ·~ = 0 (7.10)
This means that E and B are both perpendicular to the direction of propagation
n. Such a wave is called a transverse wave. The curl equations provide a further
restncuon,
~ = ~n x~ (7.11)
.
~'-
Ul\0
;-
V Jh~ l-all ' " '"
•'-
'LJIO ~
• ..1. .
llHl\011 V
;-
f-C" '
'TTTCC, ... ~
~'
.~
~
defined in (7.5). We thus see that cB and E, which have the same dimensions,
have the same magnitude for plane electromagnetic waves in free space and differ
hu thP ' ..1. nf '· ;,..n ;n nnnrlprohl, .... In pnf';nPPrin<> litPr"tllrP thP.
J r
magnetic field H is often displayed in parallel to E instead of B. The analog of
(7.11) for His
<It n X 70/L ~I. I 1 )
where Z = v;fi is an impedance. In vacuum, Z = Z 0 = ~ = 376.7 ohms,

the imoedance of free soace.
If n is real, (7.11) implies that~ and~ have the same phase. It is then useful
to introduce a set of real mutually orthogonal unit vectors (E 1, E2 , n), as shown
m t<1g. 1.1. 1n terms or tnese unit vectors tne neJO SLrengws 0 auu ~ are
~ = E1 E 0 , ~ = Ez~Eo (7.12)
UI
~ = E2Eb, ~=-E,~Eb (7.12 1 )

rl, ;} .rl } 17 Q\
·~
T7
0
..1 Dl
0 ''-'·'],,
.1 'T'L
'~.
".l_\.'1
and (7.12) or (7.12 1

) is a transverse wave propagating in the direction n. Tt rep-
z
tk=kn
~
~:~~
-., "-.... /
~
y
X/
/ Filmre 7.1 Propagation vector k and two
nrthn<'nnal notarization vectors e and e~.
298 Chapter 7 Plane Electtomagnetic Wa>es and Wave Propagation SI
resents a time-averaged flux of energy given by the real part of the complex
Poynting vector:
S =-EX H*
2
The energy flow (energy per unit area per unit time) is
S = - /- I Eo 12 n (7.13)
2'Jf.L
The time-averaged energy density u is correspondingly
This gives
u 0
2
The ratio of the magnitude of (7.13) to (7.14) shows that the speed of energy
flow is v - 1/v/iE, as expected from (7.5).
In the discussion that follows (7.11) we assumed that n \\as a real unit vectOI.
This does not yield the most general possible solution for a plane wave. Suppose
that n is complex, and written as n = nR + in1 . Then the exponential in (7.8)
becomes
e e
The wave possesses exponential growth or decay in some directions. It is then
called an inhomogeneous plane wave. The surfaces of constant amplitude and
constant phase are still planes, but they are no longer parallel. The relanons ( 7.10)
and (7.11) still hold. The requirement n · n - 1 has real and imaginary parts,*
n~ - ny = 1 (7.15)
'I he second of these conditions shows that nR and n, are orthogonal. The coor-
dinate axes can be oriented so that nR is in the x direction and n1 in they direction.
The first equation in (7.15) can be satisfied generally by writing
n - e, cosh !J + ie2 sinh !J (7 16)
where !J is a real constant and e 1 and e 2 are real unit vectors in the x and y
directions (not to be confused with e 1 and e 2 !). The most general vector~ sat-
isfying n · ~ 0 is then
(7.17)
where A and A' are complex constants. For !J =F 0, ~in general has components
in the direction(s) of n. It is easily verified that for !J 0, the solutions (7.12) and
(7.12") are recovered.
We encounter simple examples of inhomogeneous plane waves in the dis-
cussion of total internal reflection and refraction in a conducting medium later
in the chapter, although in the latter case the inhomogeneity arises from a com
*Note that if n is complex it does not have unit magnitude, that is, n · n - 1 does not imply In 12 - 1.
Sect. 7.2 Linear and Circular Polarization; Stokes Parameters 299
plex wave number, not a complex unit vector n. Inhomogeneous plane waves
form a general basis for the treatment of boundary-value problems for waves
and arc especially useful in the solution of diffraction in two dimensions. The
interested reader can refer to the book by C/emmow for an extensive treatment
7.2 Linear and Circular Polarization; Stokes Parameters

I fie plane wave ( 7.8) and ( 7.12) a wave with its electric field vector always in
IS
th~ direction € 1 . Such a wave is said to be linearly polarized with polarization
vector c 1 . Evidently the wave described in (7.12') is linearly polarized with po-
larization vector c 2 and is linearly independent of the first. Thus the two waves,
El = EtEleik·x-iMt
E2 = e2E2eik·x-itt~t
(7.18)
1'1 ith
can be combined to give the most general homogeneous plane wave propagating
in the direction k = kn.
E(x, t) (c I E I + E
2
E 2 )eik·x-iwt (7.19)
The amplitudes £ 1 and £ 2 are complex numbers, to allow the possibility of a
phase difference between waves of different linear polarization.
If E 1 and E 2 have the same phase, (7.19) represents a linearly polarized wave,
with its polarization vector making an angle () = tan_, ( E 2 / E 1) with c 1 and a
magnitude E = Y E~ + E~, as shown in Fig. 7.2.
If E, and E 2 have dzjferem phases, the wave (7.19) is elliptically put'wized.
To understand what this means let us consider the simplest case, circular polar-
ization. Then E 1 and E 2 have the same magnitude, but differ in phase by 90°.
The wave (7.19) becomes:
E(x, t) = Eo( c, ± iEz)eik·x-iwt (7.20)
with E 0 the common real amplitude. We imagine axes chosen so that the wave
1s propagatmg m the positive z direction, while c 1 and c 2 are in the x and y
directions, respectively Then the components oftbe actual electric field, obtained
by taking the real part of (7.20), are
Ex(x, t) = E 0 cos(kz - wt) }
(7.21)
Ey(x, t) = +E0 sin(kz - wt)
Eor·
/ '\8 I -'·
wave.
300 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-SI
At a fixed point in space, the fields (7.21) are such that the electric vector is
constant in magnitude, but sweeps around in a circle at a frequency w, as shown
in Fig. 7.3. For the upper sign (E 1 + iE 2 ), the rotation is counterclockwise when
the observer IS facmg mto the oncoming wave. This wave is called left circularly
polarized in optics. In the terminology of modern physics, however, such a wave
is said to have positive helicity. The latter description seems more appropriate
because such a wave has a positive projection of angular momentum on the z
axis (see Problem 7.29). For the lower sign (E 1 - iE 2), the rotation of E is clock-
wise when looking into the wave; the wave is right circularly polarized (optics);
it has negative helicity.
The two Circularly polanzed waves ( 1.20) form an equally acceptable set of
basic fields for description of a general state of polarization. V·k introduce the
complex orthogonal unit vectm s.
(7.22)
with properties
E.~ • E~ = 1
Then a general representation, equivalent to (7.19), is
E(x, t) = (E+E+ + E_E_)e;k·x-;,,, (7.24)
where E, and E are complex amplitudes If E.- and E have different magni
tudes, but the same phase, (7.24) Iepiesents an elliptically polarized wave with
prmcipal axes of the elhpse m the directions of E 1 and E 2 . The ratio of semimajor
to semiminor axis is f (1 + r)/(1 - r) [,where E_f E+ = r. If the amplitudes have
a phase difference between them, E_/E+ = re 1", then it is easy to show that the
ellipse traced out by the E vector has its axes rotated by an angle ( a/2). Figure
7.4 shows the general case of elliptical polarization and the ellipses traced out by
both E and B at a given point in space.
For r + 1 we get baek a linearly polarized wave.
The polanzatwn content of a plane electromagnetic wave IS known If It can
be written in the form of either (7.19) or (7.24) with known coefficients (£ 1 , £ 2 )
or (E+, E_). In practice, the converse problem arises. Given that the wave is of
the form (7.8), how can we determine from observations on the beam the state
of polarization in all its particulars? A useful vehicle for this are the four Stokes
E
"' ;> X
J
\ L
"....... /
/
Figure 7.3 Electric field of a circular!yp olarized

E(x, t) =Eo (<1 + i<:~)eik· x- iwt wave.
Sect. 7.2 Linear and Circular Polarization; Stokes Parameters 301
y y
n----
~
\ ~)
/~
E
cB'\ 1\
/ X
~ ~f X
I ~ / \~
\( v \ ~\/
Figure 7.4 Electric field and magnetic induction for an elliptically polarized wave.
parameters, proposed by G. G. Stokes in 1852. These parameters are quadratic

· ol. <; ,],.1 11. ,.1 1-. ,.1 ' ,,.1 11..~,~1-. ;.,tonoitu moo on 'rm lu
in conjunction ;ith a linear polarizer and a qu;;ter-wave,plate or equivalents:

Their measurement determines completely the state of polarization of the wave.
The Stokes parameters can be motivated by observing that for a wave prop-
agating in the z direction, the scalar products,
v
'l
v
' -2
v
, * ' * I<' (7 ')<;\
' . .'
are the amplitudes of radiation, respectively, with linear polarization in the x
• • •.• -• • • • ... .•• • • • 1-. ,,. ;, ,.1 •
UVH 0 f'VWHLUUVH <H w~--_y o t> ., o "b'
helicity. Note that for circular polarization the complex conjugate of the appro-
priate polarization vector must be used. in accord with (7.23). The squares of
these amplitudes give a measure of the intensity of each type of polarization.
Phase information is also needed; this is obtained from cross products. We give
aenmnons or me :->toKes parameters wirn respecr ro uow we tiuee Af tho n>~~n;t,rloe onrl r>JotiVP nh>I<P< (lf thP r(lm.
ponents. For the latter purpose we define each of the scalar coefficients in (7.19)
and (7.24) as a magnitude times a phase factor:
.c, u 1e
;~
, -=z ~
-
(7.26)
E_ = a_e;;;_ E+ =a
+eiDI '
In terms of the linear polarization basis (e 1 , e 2 ), tne :->toKes parameters are~
e, • E + E2 • E = a~ + a~
2 2
So = I 1 I 1
s, = Ie 1 • E 2 - e 2 • E 2 = a~ - a~
1 I 1
(7.27)
s2 = 2 Rer(e 1 • E)*(e 2 ·E)] = 2a,az cos(~ - 8,)
s 3 = 2 Im[(e 1 • E)*(e 2 ·E)] = 2a 1a2 sin(~ - 81 )
Tf 1h · ·~nlor nAior;?~t;~n h~eie fc c \ '" neorl in<to 1<1 thP <1Pfinitirm< rP:.oi
r "T
le>l< · El =a~+ a2_

2
s0 = le! • El 2 +
s, L. KeJ~E+ Cu>~ u_ • r,r~e- · J'.)J
U+) L.a+a- 1"7 ')Q\
\"' 'I
s 2 = 2 Im[(e! · E)*(e>l< ·E)] = 2a+a- sin(8 - 8+)
s, le·: ·ElL IE .. El -a+ a_
*The notation for the Stokes parameters is unfortunately not uniform. Stokes himself used (A, B, C,
U); Ullln WUCHilp d'C \1, \L, U, •) dHU \'• m, L, •>)· UUO "]·
'lll? r'ho~<~r
--r
7 Dlo~. .. Ulr
-.,
. <'T
1 ne expressions ~ /.L 1) ana t !.LIS) snow an mtereffing rearrangement of roles of

the Stokes parameters with respect to the two bases. The parameter s0 measures
. . . . . .
Luc '"'.a"""' tuLtoH>~Ly ut Luc w"v" Iu eiLller case. 1ne parameter s 1 gives tne pre-
ponderance of x-linear polarization over y-linear polarization, while s2 and s3 in
. . . . .
Luc """" ~·vc 1m'"" . vvc >toc~s 3 nas tne inter-
pretation of the difference in relative intensity of positive and negative helicity,
while in this basis s 1 and s2 concern the phases. The four Stokes parameters are
not maependent, smce they depend on only three quantities, a1 , a2 , and 0_ - 01 .
They satisfy the relation
s2 _2+2+2
0 - s1 s2 s 3 (7.29)
ut><; . u1 Luc _r 1 ''"PS neeueu 10 measure me :-.toKes parameters
and so determine the state of polarization of a plane wave would take us too far
-" . u " ' ~
· lUl Ut:LilllS. r>.l>U llto~leCteQ,
.£. "'- " " . , .
au~•u. "~ '~'~' wv •vau~• w ~ . v< .LJ • .LJ
except for the barest mention, is the important problem of quasi-monochromatic

radiation. Beams of radiation, even if monochromatic enough for the purposes
at hand, actually consist of a superpositiOn otnmte wave trams. By Fourier's
theorem they thus contain a range of frequencies and are not completely mono-
cnromatic. vne way or v1ewmg tms Is to say that the magmtuaes and phases
(a 1 81) in (7.26) vary slowly in time, slowly, that is, when compared to the fre-
l w. 1nt: """"' vame ;-,toKes parameterslffenoecome averages over a rei-
atively long time interval, and are written as
s2 = 2(a 1 a2 cos( 0_ - 81 )>
_,
£.
.LV' , .. L <Hv
-r u"
-'-
"'~'~ -.:rn::c
-I
uu5•~
L
Lllllt: a·· ~w~c;.
vne consequence or tne averagmg process Is that the Stokes parameters for a
quasi-monochromatic beam satisfy an inequality,
s 02 > s2
1 + s2
2
+ s2
3
..
"'"'"' uw11 we;
.
"LJuaHLJ, ~ I.L"1 J· nawrattiguL, even u monocnromanc to a mgn
. . .
degree, has s 1 = s2 = s3 = 0. Further discussion of quasi-monochromatic light

'
<l11U l'<llll<ll 1 1;;111.-c; 1.-ctll uc;
.
. 111 vurn u.nu VVUlJ, ~,. lU.
An astrophysical example of the use of Stokes parameters to describe the
-~ ,] · ,,: · .U. ·" "]. +1. "- e · ' ' _, •. e
l' " JJ ·~ o•~u-:_ v ~!- d11U 1 -, "'] !a'
diation from the pulsar in the Crab nebula. The optical light shows some small
amount of linear polarization, while the radio emission at w = 2.5 X 109 s- 1 has
a nign aegree or linear polanzatwn. · At netther1requency IS lliere evidence for
circular polarization. Information of this type obviously helps to elucidate the
lllt:UL Ull uum Ulese 1ascinating oojects.
I . .:J HeJlecuon ana HeJracuon OJ ntecrromagnetlc Waves

~• r I' ~ • • •
Ul U L · Lrut:rtu<-c net LJtc•curlCS
The reflection and refraction of light at a plane surface between two media of
different dielectric properties are familiar phenomena. The various aspects of the
phenomena divide themselves into two classes.
*See The Crab Nebula and Related Supernova Remnants, eds. M. C. Kafatos and R. n. C. Henry,
Lamonage umversny rress, J~ew rorK ll~~J).
Sect. 7.3 Reflection and Refraction of Electromagnetic Waves at an Interface 303
1. Kinematic properties:
(a) Angle of reflection equals angle of incidence.
(b) Snell's law: (sin i)/(sin r) n'ln, v;bere i, rare the angles of incidence
and refraction, while n, n' are the corresponding indices of refraction.
2. Dynamic properties:
(a) Intenstttes of reflected and refracted radtatwn.
(b) Phase changes and polarization.
The kinematic properties follow immediately from the wave nature of the
phenomena and from the fact that there are boundary conditions to be satisfied.
But they do not depend on the detailed nature of the waves or the boundary
conditions. On the other hand, the dynamic properties depend entirely on the
spectfic nature of electromagnetic fields and thetr boundary condmons.
The coordinate system and symbols appropriate to the problem are shown
in Fig. 7.5. The media below and above the plane z = 0 have permeabilities and
permittivities f-L, E and 1-L', <', respectively. The indices of refraction, defined
through (I .5), are n - V f-LE/ JLoEo and n' - v'f-L' <'I JLoE0• A plane wave wtth wave
vector k and frequency w is incident from medium f-L, E. The refracted and re-
flected waves have wave vectors k' and k", respectively, and n is a unit normal
directed from medium f-L, e into medium p!, e'.
Accordmg to (7.18), the three waves are:
INCIDENT
kXE (7.30)
B=v'/U k
REFRACTED
E' = EOeik'·x iwt
k' X E' (7.31)

B'
k'
n
/k'
;J.'t' v /
" J
j.IE
/ ~
y.- ~k" '

/
/
"'
Figure 7.5 Incident wave k strikes plane interface between different media, giving rise
to a r€ll€lct€ld wav€l k" and a r€fract€d wa"€ k'.
304 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-S!
The wave numbers have the magnitudes
lkl - lk"l - k - ui\/p:f

(7 33)
lk' I- k' - wv/]7
The existence of boundary conditions at z = 0, which must be satisfied at all
points on the plane at all times, implies that the spatial (and time) variation of
all fields must be the same at z = 0. Consequently, we must have the phase factors
all equal at z = 0,
(7.34)
independent of the nature of the boundary conditions. Equation (7.34) contains
the kinematic aspects of reflection and refraction. We see immediately that all
three wave vectors must lie in a plane. Furthermore, in the notation of Fig. 7.5,
k sin i = k' sin r = k" sin r' (7.35)
Since k" = k, we find i = r'; the angle of incidence equals the angle of reflection.
Snell's law is
sini = k' = /J.L'E' = n' (7.36)

sinr k 'J.LE n
The dynamic properties are contained in the boundary conditions-normal
components of D and B are continuous; tangential components of E and H are
continuous. In terms of fields (7.30)-(7.32) these boundary conditions at z = 0
are:
[e(Eo + Eo) t 'E'j

o • II 0
[k X E0 + k" X E3 k' X Eb] · n 0
(E 9 + E~ - E9) X n 0 (7.37)
[ ; (k X E 0 + k" X E 0) ;, (k' X E~)J X n 0
In applying these boundary condttions n IS convement to consider two sep-

arate situations, one in which the incident plane wave is linearly polarized with
its polarization veetot peipendiculaJ to the plane of incidence (the plane defined
by k and n). and the other in which the polarization vector is parallel to the plane
of incidence. The general case of arbitrary elliptic polarization can be obtaitJCd
by appropriate linear combinations of the two results, following the methods of
Section 7.2.
We first consider the electric field perpendicular to the plane of incidence,
as shown in Fig. 7.6a. All the electric fields are shown directed away from the
viewer. The orientations of the B vectors are chosen to give a positive flow of
energy in the direction of the wave vectors. Since the electric fields are all parallel
Sect. 7.3 Reftection and Refraction of Electromagnetic Waves at an Interface 305
B'
k'
•
B
(a)
k'
Figure 7.6 Reflection an

refraction with polarization (a)
perpendicular and (b) parallel to
(b) the plane of incidence.
to the surface, the first boundary condition in (7.37) yields nothing. The third and
fourth equatiOns m (7.37) give
(7.38)
,~ (Eo - E3) cos i - ~ E 0 cos r = 0
~ J.L 'V J.L
while the second, using Snell's law, duplicates the third. The relative amplitudes
of the refracted and reflected waves can be found from (7.38). These are:
E PERPENDICULAR TO PLANE OF INCIDENCE
Eb 2n cos i
0
(7.39)
E"
The square root in these expressions is n' cos r. but Snell's law has been used to
express it in terms of the angle of incidence. For optical frequencies it is usually
permitted to put J.LI J.L' = 1. Equations (7.39), and (7.41) and (7.42) below. are
most often employed in optical contexts with real n and n ', but ~hey are also
valid for complex dielectric constants.
If the electric field is parallel to the plane of incidence, as shown in Fig. 7.6b
the boundary conditions involved are normal D, tangential E, and tangential H'
[the first, thtrd, and fourth equatwns m (7.37)]. The tangential E and H contin
uous demand that
cosi(£0 - E 0)- cosr E 0 = 0
(; 77 (7.40)
Normal D continuous, plus Snell's law, merely duplicates the second of these
equations The relative amplitudes of refracted and reflected fields are therefore
E PARALLEL TO PLANE OF INCIDENCE
Eo 2nn' cos i
u
' n' 2 COS!+ nVnd n 2 sm21
J.L'
(7.41)
2 2 2 2
I!_ n' cos i - nVn' - n sin i
E" '
For normal incidence (i = 0), both (7.39) and (7.41) reduce to

2n
\ !.'IL)
J.LE - 1
E'Q _ 'JJ.L'E n'- n
~
E0 J.LE ' -I.
n' + n
1
\ u' t=
where the results on the right hold for J.L' = J.L. For the reflected wave the sign
convention is that for polarization parallel to the plane of incidence. This means
that if n' > n there is a phase reversal for the reflected wave.
7.4 Polarization by Reflection and Total Internal Reflection;

Goos JJiinclzen Effect
Two aspects of the dynamical relations on reflection and refraction are worthy
of mentiOn. The first is that for polarization parallel to the plane of incidence
there is an angle of incidence, called B1ew!Jte1 '.1 angle, for which there is no re
fleeted wave. With J.L' = J.L for simplicity, we find that the amplitude of the re-
.,.,.......
~A ~ -• , ), , D . a
... '~'~·-·
~-n
.'""'"· .... -•- ~M•
~07
--
wave m ~ t.'ll) var --ure- :or Ol~<OL[UdllUUlv '"''-'1 ' ~
:m~>-Ie
.
ls = tan
-1
(n')
-;; (7.43)
For a typical ratio n' In = 1.5, i8 = 56°. If a plane wave of mixed polarization is
incident on a plane interface at the Brewster angle, the reflected radiation is
completely plane-polarized with polarization vector perpendicular to the plane of
incidence. This behavior can be utilized to produce beams of plane-polarized
light but is not as efficient as other means employing anisotropic properties of
some dielectric media. Even if the unpolarized wave is reflected at angles other
than the Brewster angle, there is a tendency for the reflected wave to be pre-
dominantly polarized perpendicular to the plane of incidence. The success of dark
glasses that selectively transmit only one direction of polarization depends on
this fact. In the domain of radiofrequencies, receiving antennas can be oriented
to discriminate against surface-reflected waves (and also waves reflected from
the ionosphere) in favor of the directly transmitted wave.
The second phenomenon IS called total mternat rertectwn. 1ne wora mter-
n~l" ;mnJ;.,< th~t the> ;n, · -' ~n.-1 ,..,fl.,~t<>.-1 m~u<>e !-lTP. in " mP.riium nf hraP..r inriP.Y
or re1racnon man me rerr ~-nj. ~llCll > ldW ~/.JU) MIUW> llldl, 11
n > n' then r > i. Conseouentlv r - 7T/2 when i - in. where

I
; - c;n fJ -T (7.44)
n
For waves incident at i = i0 , the refracted wave is propagated parallel to the
surface. There can be no energy flow across the surface. Hence at that angle of
incidence there must be total reflection. What happens if i > i0 ? To answer this
we first note that, for i > i0 , sin r > 1. This means that r is a complex angle with
a purely imaginary cosine.
cos r = i (. ·r sm t
sin i0
- 1 (7.45)
The meaning of these complex quantities becomes clear when we consider the
propagation factor for the refracted wave:
eik'·x = eik'(xsinr+zcosr) = 1
e-k'f(sini/sinio)?-1] /?_zeik'(sini/siniu)x
(7.46)
This shows that, for i > i0 , the refracted wave is propagated only parallel to the
Sll .c.
VvvU>O
...
:md is attenuated exnonentiallv bevond the interface. The attenuation
>W
.
'J
£. _, . . .J: •L L ..1.
. ' •• ·u·
"
tven mougn news exist on tne otner swe 01 LI!e sunace Lllere is no energy
flmn thrmwh the> eurf~~" l-l'e>n~e> ~ ;n-;:;::-;:;;-;1 r<>fiP.rtinn nrrurs fnr i > i. ThP.
o·
ldlOl\. Ul llUW lOdll 'UC VCllllCU
A . ~ ' '
uy ' ' ••
U<v .uuv·u v>u 0 vu
..1
"v""u'
.1
vvou·
nonent of the PovntinP" vector iust mside the surtace:

1
S. n - - Rern · (E' x H'*ll (7.47)
2
-'"
__ T ____ 17\______ _
f 1/ \ '
Figure 7.7 Geometrical interpretation of the Goos-

Hiinchen effect, the lateral displacement of a totally
/ \ imernauy-renecrco oeam or raman on oecause of the
/ \ penetration of the evanescent wave into the region
of smaller index refraction.
mith H 1 (1. 1 X 1<' 1 ) / . , 1 mo fin..l
S·n = -
2wp.!
Re[(n · k') IEbl 2
]
But n · k' = k' cos r is purely imaginary, so that S · n = 0.

'f'~ ,], . . :1 ;.:; AC\ .< . I " •t.
" - 'J \ • ·'I , UUW0 . . , "0 ... ~ ~YY"~Y" ll\0 l.J.UdU-
tity to replace me square root appeanng m the Fresnel tormula,_ (/.:JY)and (7.41).
Inspection shows that the ratios E3/E0 are now of modulus unity, as is expected
physically for total internal reflection. The reflected wave does, however, suffer
a phase change that is different for the two kinds of incidence and depends on
t~ ,]. ,f " ".d '"' 1 .. 1 1\ 'T'L -l -l __L .,.
~ \' 1· u~0~ ~~u u~ LU
convert one Kma or po1anzatwn mto another. t<resnel's rhombus is one such
device, whereby linearly polarized light with equal amplitudes in the plane of
incidence and perpendicular to it is converted by two successive internal reflec-
tions, each involving a relative phase change of 45°, into circularly polarized light
I D -lHT.IC
evanescent wave penetratmg mto the region z > 0 has an exponential

1 ne
decay in the perpendicular direction, e-z/fi, where s-l = kYsin2 i - sin 2 io. The
penetration of the wave into the "forbidden" region is the physical origin of the
Goos Hiinchen....effe£1.; _jf_~ h""m nf r"rli~tinn h"vinP" " finite._
wrtn respect to tne preGictwn of a geometrical ray reflected at the boundary.* If

we imagine that the beam is reflected from the plane a distance 8 beyond the
boundary, as indicated in Fig. 7.7, the beam should emerge with a transverse
di<n)~CP.ffiP.nt nf n ?.fl in; 1\lfrwo ,] ~oJ~nlotir>n (e,.p p, ~],J.
,.
'1 '1\ nL
•. I
•'- •L" .1. " •·p
uw• uuo HUH~ "~""" •o llle srare OI ~Vlll<OWlldL, Wllll L/ • Vll
polarization of the radiation. The first -order expressions for D for the two states
of linear polarization are
....1.
n .,, 1'"1 AO\
\ . Jl
~ -...sitti-1
where ..\ is the wavelength in the medium of higher index of refraction.
The phenomenon of internal reflection is exploited in many applications
where It IS desired to transmit lioht 11f In<< in intPn<itv Tn npPio·•r on..l
*F. Goos and H. Hanchen, Ann. Phys. (Leipzig) (6) 1, 333-346 (1947). For an extensive discussion
of the effect, with many references, see the four-part article, H. K. V. Lotsch, Optik, 32, 116-137,
189-204, 299-319, 553-568 (1970).
Sect. 7.5 Frequency Dispersion Characteristics of Dielectrics, Conductors, and Plasmas jU'J
particle physics, plastic "light pipes" are used to carry light from scintillators
(excited bv the oassag:e of a char2:ed oarticle or energetic photon) to photomul-
tipliers, where the light is converted into useful electrical signals. If the light pipe
is large in cross-sectional dimension compared to the wavelength of the light
involved, the considerations presented here for a plane interface have approxi-
mate validity. In telecommunications, optical fibers exploit total internal refiec-
tion for transmission of modulated light signals over long distances. The vanous
transverse dimensions of a multilayered fiber are not always very large compared
to a wavelength. Then the prectse geometry must be taken mto account; the
language of modes in a waveguide may be more appropriate-see Chapter 8.
7.:> Prequency uiur~· .. :v ..

..
Conductors, and Plasmas
In Section 7.1 we saw that in the absence of dispersion an arbitrary wave train
(7.7) travels wtthout distortton.Tn reality alfmed1a snow some dtsperston. unty
1: · cl .++. · · ·~ thP ' · ~f '"~t;on
~ ~ u '5' 'i' ' ' r
be treated as constant in frequency. Of course, all the results of the preceding
. th~t . .1. ~ o;n"]p ~~mn~ IPnt ~TP V~]jr] in thP
-r lrP Of
""T
dispersion. The values of fJ- and E need only be interpreted as those appropriate
to the frem•Pnru hein<> 'iered. Where a suneroosition of a rang:e of freauen-
cies occurs, however, new effects arise as a result of the frequency dependence
of E and u. To examine some of these conseauences we need to develop at least
a simple model of dispersion.
A. Simple Model for e(w)

Almost all of the physics of dispersion is illustrated by an extension to
time-varying fields of the classical model described in Section 4.6. For simplicity
we neglect the difference between the applied electric field and the local field.
The model is therefore appropriate only for substances of relatively low density.
[This deficiency can be removed by use of (4.69 ), tt desired. J The relative per-
meability will be taken equal to unity. The equation of motion for an electron of
charge e bound by a harmomc force (4.11) ana actea on oy an etecuic ne1a
E(x, t) is
m[x + yx + w~x] = -eE(x, t) (7.49)
wnere y measures me pnenomenotogicat aamping 10rce. 1v1a w1<ec 0
·
are neglected in (7.49). We make the additional approximation that the ampli-
. . _, . u "- ~
LUue Ul IS SllHill •0 LU pt:nllll t:V v< "'" . ""'u ac "'"

average position of the electron. If the field varies harmonically in time with
iwt " '· ' · cl 1. •L ·
HI, W<l~t: ,LH\0 ·r ~J
1~
p -ex- V·JVj
_, . .,_~_,
we suppose l!laL l!lelt: aJt: Jy

11 i pt:J t'~' UHH V VVHH L, lVHO
molecule, and that, instead of a single binding frequency for all, there are f,
electrons per molecule with binding frequency w; and damping constant y;, then
the dielectric constant, EIE 0 = 1 + Xeo is given by
E(w) Ne 2
-E- = 1 + - .2;. f( w2
1 1
- w2 - iwy)
1
1
(7.51)
where the oscillator strengths f, satisfy the sum rule,
With suitable quantum-mechanical definitions off, y, and w, (7.51) is an ac-

curate description of the atomic contribution to the dielectric constant.
B. Anomalous Dispersion and Resonant Absorption

The damping constants y1 are generally small compared with the binding or
resonant frequencies w1• This means that E( w) is approximately real for most
frequencies. The factor (wT - w2 ) - 1 is positive for w < w1 and negative for,,,>
wi. Thus, at low frequencies, below the smallest w1, all the terms in the sum in
(7.51) contribute with the same positive stgn and E(w) IS greater than umty. As
successive w1 values are passed, more and more negative terms occur in the sum,
until finally the whole sum is negative and e (w) is less than one. In the neigh-
borhood of any w;, '
of course, there is rather violent behavior. The real part of
the denominator in (7.51) vanishes for that term at w = w1 and the term is large
and purely imaginary. The general features of the real and imaginary parts of
E( w) around two successive resonant frequencies are shown in Fig. 7.8. Normal
dic~po.~ion is associated with an increase in Re e(w) with w, anomalous disperswn
with the reverse. Normal dispersion is seen to occur everywhere except in the
neighborhood of a resonant frequency. And only ·.vhere there is anomalous dis-
persian is the imaginary part of E appreciable. Since a positive imaginary part to
E represents dissipation of energy from the electromagnetic wave into the me-
dium, the regions where Im E is large are called regions of resonant absorption.*
'I he attenuation of a plane wave is most directly expressed in terms of the
real and imaginary parts of the wave number k. If the wave number is written as
a
2
then the parameter a is known as the attenuation constant or absorptiOn coef-

ficient. The intensity of the wave falls off as e-az Equation (7 5) yield~ the con-
nection between (a, (3) and (Re E, Im E):
2 w
(3 - - = 2 Re E! Eo
*If Im E < 0, energy is given to the wave by the medium; amplification occurs, as in a maser or laser.
See M. Borenstein and W. E. Lamb, Phys. Rev. AS, 1298 (1972).
Sect. 7.5 Frequency Dispersion Characteristics of Dielectrics, Conductors, and Plasmas 311
I
_/ __..,
Re • ~
/
v
lm E
J \. A
w
0
Figure 7.1S Keal and 1magmary parts ot the dielectnc cons rant E\ W)l Eo m rne
neighborhood of two resonances. The region of anomalous dispersion is also the
c.
'i . ,.J •• L •· L •
'!'
If a << {3, as occurs unless the absorption is very strong or Re E is negative, the
attenuation constant a can be written approximately as
Im E(w)
a \I .J.J)
Re E(w)JY
'
Wll<:Ol<:O fJ V 1'-<:0\tltO) WI'".
, ~· r
111'- 11· ' ' ...... J p
,1
••
divided by 2'lT is thus given by the ratio, Im EIRe E.
r. T v" .. ~ ~- ~ .. u~t.
. ., v,. . r. J. ;.,;t" ..
-. ~.r .
In the limit w --? 0 there is a qualitative difference in the response of the

medium depending on whether the lowest resonant frequency is zero or nonzero.
Prw inonl· tho>~ ·~nt frP/lllPnrlT io different from zero. Then at w
0 the molecular polarizability is given by (4.73), corresponding to the limit w =
0 m (7.51). The elementary aspects of d1electncs m me sranc nmn nave oeen
discussed in Section 4.6. ,. •'-
,, . ' uo ~ "C"~~ .c
H >Ul11C 11 i J 0 Ul 111<:0 ' LUll> jJCOl ' •
" ' '"~ o~ '"'
havini> w" = 0 the dielectric constant is simmlar at w = 0. If the contribution of

the free electrons is exhibited separately, (7.51) times Eo becomes
u2+
E(w) (7.56)
Eb( w) + i mw,;{''Yo - zw
. u)
where Eb( w) is the contribution of all the other dipoles. The singular behavior
can be understood if we examine the Maxwell-Ampere equation
and assume that the medium obeys Ohm's law, J = uE and has a "normal"
dielectric constant Eb. With harmonic time dependence the equation becomes
V H= -iw(
X Eb +i :)E (7.57)
lf, on the other hand, we did not insert Ohm's law explicitly but attributed instead
all the pwpe1 ties of the medium to the dielectric constant, we would identify the
quantity in brackets on the right-hand side of (7 57) with E(u•) Comparison with
(7 .56) yields an expression for the conductivity:
.f0 Ne 2
(T = ( ) (7.58)
m Yo - iw
This is essentially the model of Drude (1900) for the electrical conductivity, with
f 0 N 6emg the number of free electrons per umt volume m the medmm. The
damping constant 1'o lj0 can be determined empirically from experimental data
on the conductivity. For copper, N 3 X 102 ~ atoms7m' and at normal temper-
atures the low-frequency conductivity is u = 5.9 X 107 (!1 · m)- 1 • This gives y0 /.f 0
= 4 X 10 s • Assuming that .fo ~ 1, this shows that up to frequencies well
13 1
beyond the microwave region (w :S 1011 s- 1) conductivities of metals are essen-

tially real (i.e., current in phase with the field) and independent of frequency. At
higher frequencies (in the infrared and beyond) the conductivity is complex and
varies with frequency in a way described qualitatively by the simple result (7 58)
The problem of eleet1 ieal conductivity is 1eally a quantum mechanical one in
which the Pauli pnnc1ple plays an 1mportant role. The free electrons are actually
valence electrons of the isolated atoms that become quasi-free and move rela-
tively unimpeded through the lattice (provided their energies lie in certain inter-
vals or bands) when the atoms are brought together to form a solid. The damping
effects come from collisions involving appreciable momentum transfer between
the electrons and lattice vibrations, lattice imperfections, and impurities.*
The foregoing considerations show that the distinction between dielectrics
and conductors is an art1fic1al one, at least away from w 0. If the medmm
possesses free electrons it is a conductor at low frequencies; otherwise, an insu-
Iat01. t B at at nonzeto 1f equencies I " con d uct1vrty
. lie . . " contn"b utron
. to E\f w) J "1')
("' .5
merely appears as a resonant amplitude like the rest. The dispersive properties
of the medium can be attributed as well to a complex dielectric constant as to a
frequency-dependent conductivity and a dielectric constant.
·'·See R. G. Chambers, Electrons m Metals and Semiconductors, Chapman & Hall, New York (1990),
or G. Lehmann and P. Z1esche, Electronic Properties of Metals, Elsevier, New York (1990).
1
Tn terms of the quantnm-mech:mical band stmctnr• of th• solid, the conductor has some eleetrons
ift a partially filled band, hhile the insttlatm has its bands filled to the ful1 extent pennitted by the
Pauli principle. A fiee · electron must have nearby energy-conservmg quantum states to whtch H
can move. In a partially filled band there are such states, but a filled band has, by definition, no such
states available.
D. High Frequency Limit, c>rJfflsma Frequency

At frequencies far above the highest resonant frequency the dielectric con-
stant (7.51) takes on the simple form
E(w) =
1
w ere
(7.60)
The frequency w,, which depends only on the total number 1'v'Z of electrons per
unit volume, ts called the plasma frequency of the medmm. The wave number ts
given in the limit by
(7.61)
Sometimes (7.61) is exptessed as ul ui), I c2 k 2 , and is called a dispersion
relatwn or equatiOn for w - w(k). In dtelectnc medta, (7.59) applies only for
w w;.
2
>> The dielectric constant is then close to unity, although slightly less,
and increases with frequency somewhat as the highest frequency part of the curve
shown in Fig. 7.8. The wave number is real and varies with frequency as for a
mode in a waveguide with cutoff frequency wP. (See Fig. 8.4.)
In certain situations, such as in the wnosphere or in a tenuous electronic
plasma in the laboratory, the electrons are free and the damping is negligible.
Then (7.59) holds over a wide range of frequencies, including w < wP. For fre-
quencies lower than the plasma frequency, the wave number (7.61) is purely
imaginary. Such waves incident on a plasma are reflected and the fields inside
fall off exponentially with distance from the surface. At w 0 the attenuation
constant is
(7.62)
On the laboratory scale. plasma densities are of the order of 1018 - 1022 electrons/
m~. This means w, 6 X 1010 6 X 1012 s-r, so that t~·pically attenuation lengths
(a- ) are of the order of 0.2 em to 2 X 10- 3 em for static or low-frequency fields.
1
The expulsion of fields from within a plasma is a well-known effect in controlled

thermonuclear processes and is exploited in attempts at confinement of hot
lasma.
The reflectivity of metals at optical and higher frequencies is caused by es-
sentially the same behavim as fm the tenuous plasma. The dielectric constant of
a metal ts gtven by (7.56). At htgh frequenctes (w >> y0 ) thiS takes the approx-
imate form,
where w; - 2
1Ui !m*E0 is the plasma frequency ~o;quared of the conduction elec
trans, given an effective mass m* to include partially the effects of binding. For
w << wP the behavior of light incident on the metal is approximately the same
as for the plasma described by (7 59) The light penetrate~ only a very short
distance into the metal and is almost entirely reflected. But when the frequency
is increased into the domain where E(w) > 0, the metal suddenly can transmit
light and its reflectivity changes drastically. This oeeurs typically in the ultraviolet
and leads to the terminology "ultraviolet transparency of metals." Determination
of the critical frequency gives information on the density or the effective mass
of the conduction electrons.*
E. Index o.fRefractwn andAbsorptwn Loefficwnt

of Liquid Water as a Function of Frequency
As an example of the overall frequency behavior of the real part of the index
of refractiOn and the absorptiOn coefficient of a real medium, we take the ubiq-
uitous substance, water. Our intent is to give a broad view and to indicate the
tremendous variations that are possible, rather than to discuss specific details
Accordingly, we show in Fig 7 9, on a log log plot with 20 decades in frequeftey
and 11 decades in absorption, a compilation of the gross features of n(w) =
ReV !1-EI/1-oEo and a(w) = 2 Im v;E w for liquid water at NTP. The upper part
of the graph shmvs the intctesting, but not spectaculat, behaviot of n(w). At very
low frequencies, n(w) = 9, a value arising from the partial orientation of the
permanent dipole moments of the water molecules. Above 1010 Hz the curve falls
relatively smoothly to the structure m the mfrared. In the v1s1ble regwn, shown
by the vertical dashed lines, n(w) = 1.34, with little variation. Then in the ultra-
violet there is more structure. Above 6 X 1015 Hz (hv = 25 eV) there are no data
on the real part of the index of refraction The asymptotic approach to unity
shown in the figure assumes (7.59).
Much more dramatic is the behavior of the absorption coefficient a. At fre-
quencies below 10 8 Hz the absorption coefficient is extremely small. The data
seem unreliable (two different sets are shown), probably because of variations
in sample purity. As the frequency increases toward 1011 Hz, the absorption
coefficient inct eases rapidly to a ·- 104 m \ corresponding to an attenuation
length of 100 11-m in liquid water. This is the well-known microwave absorption
by water. It is the phenomenon (in moist air) that terminated the trend during
World War II toward better and better resolution in radar by going to shorter
and shorter wavelengths.
In the infrared region absorption bands associated with vibrational modes of
the molecule and possibly oscillations of a molecule in the field of its neighbors
cause the absorption to reach peak values of a = 106 m - 1 . Then the absorption
coefficient falls precipitously over 7~ decades to a value of a < 3 x 10- 1 m - 1 in
a narrow frequency range between 4 X 1014 Ih and 8 x 1014 Hz. It then rises
agam by more than 8 decades by 2 X 10 15 Hz. This is a dramatic absorption
window in what we call the visible region. The extreme transparency of water
here has its origins in the basic energy level structure of the atoms and molecules.
The reader may meditate on the fundamental question of biological evolution
on this water-soaked planet, of why animal eyes see the spectrum from red to
*See Chapter 4 of D. Pines, Etementwy Exchadum; in Solids, W. A. fienjamin, New York (1963), for
a discussion of these and other dielectric properties of metals in the optical and ultraviolet region.
More generally. see F. Wooten, Optical Properties of Solids, Academic Press, New York (1972) and
Handbook of Optical Constanty of 5oUds ed F D Palik Academjc press, Boston (1991)
10 2 104 10 6 108 1010 1012 1014 1016 1018 1020 1022

t 10 I I I II I I I
~
~ 51-
r- "\ :: -
~'1'\
u
~ II
'§ II
0
.A
""'
"0 1 'I' II
-
-=
,.
Visible ::
,..; '-
'
II
102 104 10 6 108 1010 1012 1014! 1 1016 1018 1020 1022
1n6
I
::
10 5 t- I I -
I I
I I
I I I
104 t- I I I -
I I I
)\ :: \
'"
.v
)v II
II
II 1\
w·
t / I
I
I
I
I
I
K
edge
in
I
~
' E 10 - I I oxygen -
~ - I I
~
"
~
"' 1
I
I
I
I
I
I \
I
/
~0 Sea
water /
/ I:
\
Ij ~-
~ 10-I - / I
Q / I
o_ /
0 / I
-" /
<( 10-2 - / I I -
10--3 ,!
/
/
/
-- -
I
I
I
I
I
I
I
I
I
-
I I
I I I
1 km 1 1m 1 em 11'ml I 1A
10--4 - _I __
- /
1 eV
I I
1 meV
t I
I! I
1 eV 1I 1 keY
I
1 MeV
-
10-s I I I It I 1t! :. I t I It
102 104 106 108 1010 1012 1014 1016 1018 1020 1022
Frequency (Hz)
J<Jgure t.'l !he mdex ot retractwn (top) and absorption coethC!ent (bottom) tor liqUid
water as a function of linear frequencv. Also shown as abscissas are an energy scale
(arrows) and a wavelength scale (vertical lines). The visible region of the frequency
spectrum is indicated by the vertical dashed lines. The absorption coefficient for
seawater IS mdicated by the dashed diagonal lme at the lett. Note that the scales arc
logarithmic in both directions.
vwter ana or wnv tne grass IS green. lV!Otner l'lature nas certainlY expwnea ner
window! In the very far ultraviolet the absorption has a peak value of a = 1.1 X
,,...g
iV '"
-J
U< V
~
,
'""
<V
TT. '~'
< "" \~'
-"'
'-' . }•
~L'
UH'
•
<O
d.
-J
d.
a< <H'-' r
_,
·oJ
¥
<OUJp'
corresponding to a collective excitation of all the electrons in the molecule. The
attenuation is given in order of magnitude by (7.62). At higher frequencies data
are absent until the photoelectric effect, and then Compton scattering and other
high-energy processes take over. There the nuclear physicists have studied the
absorption in detail The behavior is basically governed by the atomic properties
and the density, not by the fact that the substance is water.
At the low frequency end of the graph in Fig 7 9 we have indicated the
absorption coefficient of seawater. At low frequencies, seawater has an electrical
conductivity u = 4.4 n- 1 m- 1 . From (7.57) we find that below about 108 Hz a""
(2J-L 0 wu)112 . The absorption coefficient is thus proportional to Vw and becomes
very small at low frequencies. The line shown is a (m- 1 ) = 8.4 X 10- 3 Vv(HZ).
At 102 Hz, the attenuation length in seawater is a 1 ~ 10 meters. This means
that 1% of the intensity at the surface will survive at 50 meters below the surface.
If one had a large fleet of submarines scattered throughout the oceans of the
world and wished to be able to send messages from a land base to the submerged
vessels, one would be led to constder extremely low-frequency (ELF) commu-
nications. The existence of prominent resonances of the earth-ionosphere cavity
in the range from S Hz to a few hundred hertz (see Section 8.9) makes that regton
of the frequency spectrum specially attractive, as does the reduced attenuation.
Vv'ith wavelengths of the otder of 5 X 103 km, very large antennas are needed
(still small compared to a wavelength!).*
7.6 Simplified Model of Propagation in the Ionosphere

and Magnetosphere
The propagation of electromagnetic waves in the ionosphere is described in
zeroth approx1matJon by the d1electnc constant (7.59), but the presence of the
earth's magnetic field modifies the behavior significantly. The influence of a static
external magnetic field is also present for many laboratory plasmas. To illustrate
the influence of an external magnetic field, we consider the simple problem of a
tenuous electronic plasma of uniform density with a strong, static, uniform, mag-
netic induction BG and transverse waves propagating parallel to the direction of
B0 . (The more general problem of an arbitrary direction of propagation is con-
tained in Problem 7.17.) If the amplitude of electronic motion is small and col-
lisions are neglected, the equation of motion is approximately
mx eB 0 X x eEe iwt (7 .63)
where the influence of the B field of the transverse wave has been neglected
compared to the static induction B0 and the electronic charge has been written
as -e. It is convenient to consider the transverse waves as circularly polarized.
and a similar expression for x Since the direction of Be is taken orthogonal toe,
and e 2 , the cross product in (7 .63) has components only in the direction e 1 and
·:·pm detailed discussion of ELF communications, see the conference pioceedings, ELF/VLF/LFRa-
dio Propagation and Systems Aspects, (AGARD-CP-529), Brussels, 28 September-2 October, 1992,
AGARD, Neuilly sur Seine, France (1993).
Sect. 7.6 Simplified Model of Propagation in the Ionosphere and Magnetosphere 317
E2 and the transverse components decouple. The steady-state solution of (7.63)

lS
e
X= ----~----E (7.65)
where w8 is the frequency of precession of a charged particle in a magnetic field,
ws=- (7.66)
m
The frequency dependence of (7.65) can be understood by the transformation of
(7.63) to a coordinate system precessing with frequency w 8 about the direction
of B 0 . The static magnetic field is eliminated; the rate of change of momentum
there is caused by a rotating electric field of effective frequency ( w :±: w8 ), de-
pending on the sign of the circular polarization.
The amplitude of oscillation (7.65) gives a dipole moment for each electron
and yields, for a bulk sample, the dielectric constant
(7.67)
The upper sign corresponds to a positive helicity wave (left-handed circular po-
larization in the optics terminology), while the lower is for negative helicity. For
p1 opagation antiparallel to the magnetic field 8 0 , the signs are reveiSed. This is
the extension of (7.59) to include a static magnetic induction. It is not completely
general, since it applies only to waves propagating along the static field direction.
But even in this simple example we see the essential characteristic that waves of
right-handed and left-handed circular polarizations propagate differently. The
ionosphere is birefringent. For propagation in directions other than parallel to
the static field B0 it is straightforward to show that, if terms of the order of w1
are neglected compared to w2 and ww 8 , the dielectric constant is still given by
(7.67). But the precession frequency (7.66) is now to be interpreted as that due
to only the component of B 0 parallel to the direction of propagation. This means
that '"R in (7 67) is a function of angle the medium is not only birefringent, but
also anisotropic (see Problem 7 .17).
For the ionosphere a typical maXImum density of free electrons 1s 1018 1012
electrons/m 3 , corresponding to a plasma frequency of the order of w~. = 6 X 106
-6 X 107 s- 1 . If we take a value of 30 fLT as representative of the earth's magnetic
field, tl1e p1 ecession frequency is w 8 6 X 106 s 1 •
Figure 7.10 shows E+!E0 as a function of frequency for two values of the ratio
of ( wP/w 8 ). In both examples there are wide intervals of frequency where one of
E"+ or E' is positive Vlhile the other is negative. At such frequencies one state of
circular polarization cannot propagate in the plasma. Consequently a wave of
that polanzatwn mc1dent on the plasma wlll be totally reflected. I he other state
of polarization will be partially transmitted Thus, when a linearly polarized wave
is incident on a plasma, the reflected wave will be elliptically polarized, with its
major axis generally rotated away from the direction of the polarization of the
mc1dent wave.
The behavior of radio u1aves reflected from the ionosphere is explicable in
terms of these ideas, but the presence of several layers of plasma with densities
20
-\ ;:
~
Wp
-· =
roB
2.0 4
I
I
Wp
wa = 0.5
I I
15 I- I < I
10 I-
1
2 I-
\,/
..........
I
I
I I
I I
I
51- I
2
1 I-
'~+ :
~
t 0
11
I
t I
I I
---
0
I
/'E_~
11
I 2
'+/f
c-::!::./c 0 €-:!:.leo I (JJ
rua I
_, ' OJ
7 I
I 7- I
I
-10 -I I
I I -2 I
I
I
15 - ! _.,
I I
I I
-20 I-
i -4 -
I
Figure 7.10 Dielectric constants as functions of frequency for model of the ionosphere
(tenuous electronic plasma in a static, uniform magnetic induction). <±(w) apply to the
right and left circularly polarized waves propagating parallel to the magnetic field. w 8 is
the gyration frequency; w, is the plasma frequency. The two sets of curves correspond
and relative positions varying with height and time makes the problem consid-
erably more complicated than our simple example. The electron densities at var-
ious heights can be inferred by studying the reflection of pulses of radiation
transmitted vertically upwards. The number n0 of free electrons per unit volume
increases slowly with height in a given layer of the ionosphere, as shown in Fig.
7.11, reaches a maximum, and then falls with further increase in height. A pulse
of a given frequency w 1 enters the layer without reflection because of the slow
change in n 0 . When the density n 0 is large enough, however, wP(h 1 ) = w1• Then
the d1electnc constants (I .6!) vamsh and the pulse 1s reflected. The actual dens1ty
lle where the rellection occurs is given by the roots of the right hand side of (7.67).
By observing the time interval between the initial transmission and reception of
the reflected signal the height h 1 corresponding to that density can be found. By
varying the frequency w1 and studying the change in time intervals, the electron
"o max I
1-------------T-~-----
Figure 7.11 Electron density as a

01 m a rayer 01 Lm;
ionosphere (schematic).
Sect. 7.7 Magnetohydrodynamic Waves 319
density as a function of height can be determined. If the frequency w1 is too high,

the index of refraction does not vanish and very little reflection occurs. The
frequency above which reflections disappear determines the maximum electron
density in a given layer. A somewhat more quantitative treatment using the
Wentzel-Kramers-Brillouin (WKB) approximation is sketched in Problem 7.14.
. . . . .
k_ ~ wP
c
I
y w8
w
This corresponds to a highly dispersive medium. Energy transport is governed

by the group velocity (7.86)-see Section 7.8-which is
~
vg(w) = 2vP(w) ~ 2c - -
wP
range. With the estimates given above for wP and w8 and distances of the order
of 104 km, the reader can verify that the time scale for the whistlers is measured
in seconds. Further discussion on whistlers can be found in the reading sugges-
tions at the end of the chapter and in the problems.
7. 7 Magnetohydrodynamic Waves
In the preceding section we discussed in terms of a dielectric constant the prop-
agation of waves in a dilute plasma in an external magnetic field with negligible
. . . . .
usually neglected. The nonrelativistic mechanical motion is described in terms of

a single conducting fluid with the usual hydrodynamic variables of density, ve-
locity, and pressure, with electromagnetic and gravitational forces. The combined
system of equations describes magnetohydrodynamics (MHD).
The electromagnetic equations are those of Section 5.18, with the Ohm's law
320 Chapter 7 Plane Electromagnetic Waves and Wave Propagation SJ
in (5.159) generalized for a fluid in motion to J = u(E + v x B), in accord with

the discussion of Section 5.15. The generalization of (5.160), but for the magnetic
induction, is
aB 1
- = V x (v x B) + - V2 B (7.68)
at f.LU
whei e fur simplicity we have assumed that the conduct1v1ty and permeability are
independent of position
Consider the idealization of a compressible, nonviscous, "perfectly conduct-
ing" fluid in the absence of gravity, but in an external magnetic field. By perfectly
conducting we mean that the conductivity is so large that the second term on the
right-hand side of (7.68) can be neglected-the diffusion time (5.161) is very long
compared to the time scale of interest. The hydrodynamic equations are
av 1
p-
at + p(v • V)v = -Vp - - B
f.L
X (V X B)
The first equation is conservation of matter; the second is the Newton equation
of motion with the mechanical pressure force density and the magnetic force
density, J X B, in which J has been replaced by V X H. The magnetic force can
l
--B x (V x B)= -V - B 2
f.L
c)
2f.L
+ -(B·V)B
1
2
The first term represents the gradient of a magnetic pressure; the second is an
additional tension. Equation (7.69) must be supplemented by an equation of
state.
In the absence of a magnetic field, the mechanical equations can describe
small amplitude, longitudinal, eompressional (sound) waves with a speed s, the
square of which IS equal to the denvat1ve of the pressure D with respect to the
density p at constant entropy. With the adiabatic gas law, p - KpY, where y is
the ratio of specific heats, s 2 = yp 0 ! p0 . By analogy, we anticipate longitudinal
MHD waves in a conducting fluid in an external field B 0 , with a speed squared
of the order of the magnetic pressure divided by the equilibrium density,
To exhibit these 'Naves we consider the combined equations of motion (7.68) and
(7.69), with the neglect of the V2 Bigu term m ( 7.68), with an unperturbed con-
figuration consisting of a spatially uniform, time-independent magnetic induction
B 0 throughout a stationary fluid of constant equilibrium density p0 . We then allow
for small-amplitude departures from equilibrium,
B = B0 + B 1 (x, t)
V V X t
Sect. 7.7 Magnetohydrodynamic Waves 321
11 equations ( l.b'ol) and ( /.M) are JinearizeainLITe smau quamiiies, mey oecome:
dp]
- + poV • v1 = 0
at
av 1 B0
Po-+ s 2 Vp1 +-X (V x B 1) = 0 (7.71)
at f.L
iJB1
at
Y , ~' •1 " -
~ '
~o, "v
where s 2 is the square of the sound velocity. These equations can be combined
lU yitau au !Uf v1 illUlle:
.0
0 1 2
iJt: - s V(V · VJ) + VA X V X [V X (v 1 X vA)] = 0 (7.72)
·" wP h"vf' intrnclnrPcl " i"l Alfven velocit":

B
VA (1./j)
~
The wave equation (7.72) for v1 is somewhat involved, but it allows simple
solutions for waves propagating parallel or perpendicular to the magnetic field
direction.* With v1(x, t) a plane wave with wave vector k and frequency w:
v1 (x, t) = vleik·x-iwf (7.74)
equation (7.72) becomes:
-w2 v1 + (s 2 + u::,)(k • v1)k + vA · k[(vA • k)v1 (7.75)
(vA • v1)k (k • VJ}VAJ u
If k is perpendicular to vA the last term vamshes. Tnen tne somt10n tor v1 1s a
1- · r~; ,/ m::l i,- m::lUP mith " -h~o~ ·1 itv·
0
"
H. - ,; <2 _j. ,2 (7 71i\
""
Note that this wave propagates with a velocity that depends on the sum of hy-
drostatic and magnetic pressures, apart from factors of the order of unity. If k is
-
parallel to vA, (7.75) reduces to
Wv~ - 2
w )vl + c~ 1)k (VA • v.)vA = 0 2
(7.77)
There are two types of wave motion possible in this case. There is an ordinary
. . .. , , -1 . -1 d. -'
WdV<;;~YJr lUK<illUYA)Wllll <J cV<H'-'
velocity s. But there is also a transverse wave (v 1 • vA = 0) with a phase velocity

equal to the Alfven velocity vA· This Alfven wave is a purely magnetohydrody-
.~ .rlo ~~1" ~~ <ho m~m ;,_ fip]cJ (tpnoinn\ ::lncl th0
• .1..
, "1- :nh -'· 'J
densit~ (inertia).
For mercury at room temperature tneAifven velocity 1s /.bl H 0 (tesla) m/s,
*The determination of the characteristics of the waves for arbitrary direction of propagation is left
<V ' 'VU<~H< .W.
~
1'\
\ ~ ~ \\ ~ ~ u ~
1'\ 1'\ 1'\ 1'\ 1'\ I'\ I'\
T
k
I II I I I II
( ( I ( ( I k
) l )\) l) l))\)
n·n 7'7' 1'1' r 7' 7'
I I I I I I I I I
Figure 7.12 Magnetohydrodynamic waves.
compared with the sound speed of 1.45 X 103 m/s. At all laboratory field strengths
the Alfven velocity is much less than the speed of sound. In astrophysical prob-
!ems, on the other hand, the Alfven velocity can become very large because of
the milch smaller densities. In the sun's photosphere, for example, the density is
of the order of 10- 4 kg/m3 ( ~6 X 1022 hydrogen atoms/m 3 ) so that v A = 105 B(T)
m/s. Solar magnetic fields appear to be of the order of 1 or 2 X 10- 4 T at the
surface, with much larger values around sunspots. For comparison, the velocity
of sound is of the order of 104 m/sin both the photosphere and the chromosphere.
The magnetic fields of these diffe~ent waves can be found nom the thi1d
e ua 1on ill
fork .l B0
0 for the longitudinal k II B0 (7.78)
The magnetosonic wave moving perpendicular to B 0 causes compressions and

rarefactions in the lines of force without changing their direction, as indicated in
Fig. 7.12a. The Alfven wave parallel to B0 causes the lines of force to oscillate
back and forth laterally (Fig. 7.12b ). In either case the lines of force are "frozen
in" and move with the fluid
Inclusion of the effects of fluid viscosity, finite, not infinite, conductivity, and
the displacement current add complexity to the analysis. Some of these elabo-
rations are treated in the problems.
7.8 Superposition of Waves m One Dimension; Group Velocity

In the precedillg sectwns plane wave soluhons to the Maxwell equations were
found and their properties discussed. Only monochromatic waves, those with a
definite frequency and wave number, were treated. In actual circumstances such
idealized solutions do not arise. Even in the most monochromatic light source or
the most sharply tuned radio transmitter m receiver, one deals with a finite (a!-
though perhaps small) spread of frequencies or wavelengths This spread may
originate in the finite duration of a pulse, in inhe1 ent broadening in the somce,
or ill many other ways. Sillce the bas1c equatwns are lillear, 1t IS ill principle an
Sect. 7.8 Superposition of Waves in One Dimension; Group Velocity 323
elementary matter to make the appropriate linear superposition of solutions with

different frequencies. In general, however, several new features arise.
1. If the medium is dispersive (i.e., the dielectric constant is a function of the
frequency of the fields), the phase velocity is not the same for each frequency
comoonent of the wave. Conseauentlv dtfferent components ot the wave
travel with different speeds and tend to change phase with respect to one
another.
2. In a dispersive medium the velocity of energy flow may differ greatly from
'" ,] ' '''J , ') 1. "'' T -~ ;,.,n
j, In a dtsstpattve medmm, a pulse ot radtatton w111 oe attenuatea as n travels
with or without distortion, depending on whether the dissipative effects are
or are not sensitive functions of frequency.
The essentials of these dispersive and dissipative effects are implicit in the
.,.], ,j' n. ' ' n~rl ' ''" ,.
{<:', ' ,.;
.Q\',. 10', o;m--" ' mP '·'
·:o· T J,
,,
Wi1Vt:~ 111 UlllY Ullt: "V"• 'J """ u.o
. lilt:• 0 '• dlllr
of as one of the components of the electromagnetic field. The basic solution to
the wave equation has been exhibited in (7.6). The relationship between fre-
quency w and wave number k is given by (7.4) for the electromagnetic field.
Either w or k can be viewed as the independent variable when one considers
making a linear superposition. InitiaiiV we wililfriCf tt most convement to use k
as an independent variable. To allow for the possibility of dispersion we will
consider w as a general function of k:
w = w(k) (7.79)
;-,mce tne msperstve properties cannot depend on wneLIIer we wave 1rave1s w
the left or to the right, w must be an even function of k, w( -k) - w(k). For most
wavelengths w is a smoothly varying function of k. But, as we have seen in Section
7.5, at certain frequencies there are regions of "anomalous dispersion" where w
varies rapidly over a narrow interval of wavelengths. With the general form
(7 7QI n11r "''- '""t _,. ion r~n ~nnlv enuallv well to electromagnetic
_, _, n. _,. ,..; .~ ,, ,
HU •~o, '"·~o,
·~ '5'
that k and w(k) are real, and so exclude dissipative effects. "
From the basic solutions (7.6) we can build up a general solution of the form
.{,
'\' ' ·'
•J
1
,f)'17'J-
r= A fl.\ ,ikx-i<v(k)t ATE
~"T
('7 5111\
'I
00 '
The factor 1/y'2; has been inserted to conform with the Fourier integral notation
of (2.44) and (2.45). The amplitude A(k) describes the properties of the linear
superposition of the different waves. It is given by the transform of the spatial
amplitude u(x, t), evaluated at t = 0*:
1 roo
A{k)- - - I u(x O)e ' " dx (7 .81)
V27T7 oo
If u(x, 0) represents a harmonic wave e'kox for all x, the orthogonality relation
(2.46) shows that A(k) = y'2; o(k - k 0 ), corresponding to a monochromatic
·~· ,_, '.L

"6'
..L .. ' ._, -"
'' This omission is of no consequence for
.Ll p, 'ol
equation we must specify not only u(x, 0) but also au(x, 0)/at.
the rest of the material in this section. It is remedied in the following section.
travelmg wave u(x, t) - eiko-' iw(kul', as reqmred. If, however. at t - 0. u(x, D)

r@pr@S@nts a finit@ wav@ train with a length of order Ax, as shown in Figure 7.13,
then the amplitude A(k) 1s not a delta functwn. Rather, 1t 1s a peaked functwn
with a breadth of the order of !::..k, centered around a wave number k 0 , which is
the dominant wave number in the modulated wave u(x, 0). If !::..x and !::..k are
defined as the rms deviations from the average values of x and k [defined in terms
of the intensities Iu(x, 0) 12 and IA(kWJ, it is possible to draw the general
cone uswn:
(7.82)
The reader may readily verify that, for most reasonable pulses or wave packets
that do not cut off too violently, !::..x times !::..k lies near the lower limiting value
in (7.82). This means that short wave trains with only a few wavelengths present
have a very wide distribution of wave numbers of monochromatic \'Va~cs, and
conversely that long smusmdal wave trams are almost monochromatic. Relation
(7.82) applies equally well to distributions in time and in frequency.
The next question is the behavior of a pulse or finite wave train in time. The
pulse shown at t = 0 in Fig. 7.13 begins to move as time goes on. The different
frequency or wave-number components in it move at different phase velocities.
Consequently there is a tendency for the original coherence to be lost and for
the pulse to become distorted in shape. At the very least, we might expect it to
propagate With a rather d1fferent veloc1ty from, say, the average phase velocity
of its component waves. The general case of a highly dispersive medium or a
very sharp pulse with a great spread of wave numbers present is difficult to treat.
But the propagation of a pulse which is not too broad in its wave-number spec-
trum, or a pulse in a medium for which the frequency depends weakly on wave
number, can be handled m the following approximate way. The wave at time t
is given by (7.80). If the distribution A(k) is fairly sharply peaked around some
value k 0 , then the frequency w(k) can be expanded around that value of k:
dw
w(k) = w0 +- (k - k 0 ) + (7.83)
dk 0
' llx
r
u (x, 0) I
I
-- -- - - ~
\ I
!"'--Ilk~
.L
I
1/
m
I I
I
I
\
\
\ n•
.
-~-
"'n A L
·'
KO n· ~ umte extem ffiiTI'ITs-rourier spectrum in
wave number.
Sect. 7.8 Superposition of Waves in One Dimension; Group Velocity 325
and the integral performed. Thus

ei[ko(dw!dk)lo-wo]t Joo
u(x, t) = A(k)e'[x-(dw!dkllotlk dk (7.84)
y'2; -x
From (7.81) and its inverse it is apparent that the integral in (7.84) is just u(x', 0),
where x' x (dwldk)la t:
(7.85)
This shows that, apmt fiom an overall phase factor, the pulse travels along an
distorted in shape with a velocity, called the group velocity:
dw
Vg = dk (7.86)
0
If an energy density is associated with the magnitude of the wave (or its absolute
square), it is clear that in this approximation the transport of energy occurs with
the group velocity, since that is the rate at which the pulse travels along.
For light waves the relation between wand k is given by
where c is the velocity of light in vacuum, and n(k) is the index of refraction
expressed as a function of k. The phase velocity is
w(k) c
v =--=-- (7.88)
p k n(k)
and is greater or smaller than c depending on whether n(k) is smaller or larger
than unity. For most optical wavelengths n(k) is greater than unity in almost all
substances. The group velocity (7.86) is
c
v = ------~------ (7.89)
g n(w) + w(dnldw)
In this equation it is more convenient to think of n as a function of w than of k.
For normal dispersion (dnldw) > 0, and also n > 1; then the velocity of energy
flow is less than the phase velocity and also less than c. In regions of anomalous
dispersion, however, dnldw can become large and negative as can be inferred
fiom Fig. 7.8. Then the group velocity differs greatly fiom the phase velocity,
often becommg larger than c or even negative. I he behav10r of group and phase
velocities as a function of frequency in the neighborhood of a region of anoma-
lous dispersion is shown in Fig. 7.14. There is no cause for alarm that our ideas
of special relativity are violated; group velocity is generally not a useful concept
in regions of anomalous dispersion. In addition to the existence of significant
absorption (see Fig. 7.8), a large dnldw is equivalent to a rapid variation of w
with k. Consequently the approximations made in (7.83) and following equations
are no longer valid. Usually a pulse with its dominant frequency components in
the neighborhood of a strong absorption line is absorbed and distorted as it
travels As shown by Garret and McCumber,* however there are cjrcumstances
•C. G. B. Garrett and D. E. McCumber, Phys. Rev. A I, 305 (1970).

':1')1::
'"" ,- ~ "'- m .... ion-
-
'""" >LliU """ I -"·
I I :
~-~~I
n(w)
1
-r- ~---t---
I j I
-
1 : t
w--
I "gil\ I I
I I I \r 1
c H- "p-:~J -Y-~ . TnclPY nf ,f, ,~,;A~ ~r _\

· 7_14.
~,
~ I I as a function of frequency w at a
region or anomalous d1spcrswn; phase
I i T velocitv" •mel <JTC>lln _, ..--,- ""
w--- functions of w.
in which "group velocity" can still have meaning, even with anomalous disper-
sion. Other authors* subsequently verified experimentally what Garrett and
McCumber showed theoretically: namely, if absorbers are not too thick, a
Gaussian pulse with a central frequency near an absorption line and with support
narrow compared to the wlcllli Of the line (pulse wide in time compared to 1/y)
otoo ,~,;tl, n '.1..1. ' ' 1.. 0 • '
r 7 -,-,- '!-' , « rm::>TV U~ ~1000 ILO -,- , llle
peak of which moves at the group velocity (7.89). even when that nuantitv i<
negative. Physically, what occurs is pulse reshaping-the leading edge of the
;- . ' . .. . .
1-' •o ~'-'M a""'" " u~ Lue Lrauwg euge. L.onmuons can oe such that the
veak of the o-reatlv attenuatPcl nll loP f'~A~ thp n ~ 0 0 <L .1
u 'T
of the incident pulse has entered it! (That is the meaning of negative group
verocrry.) Smce a Uauss1an pulse does not have a sharply defined front edge,
there is no question of violation of causality.
Some experiments are described as showing that photons travel faster than
the speed of light through optical "band-gap" devices that reflect almost all of
the incident flux over a restricted range of frequencies. While it is true that the
centroid of the very small transmitted Gaussian pulse appears slightly in advance
.£~ • • • • •
"" .~'"u"' v1 u<tvco~> 1asrer rnan c. Tile

main results are explicable in conventional classic"! " " orp =v.
amined in Problems 7.9-7.11. A review of these and other experiments has been
"iven oy Lniao anu C)reinoerg .
"'"' TlJ. . p •• ~ ..
lo7 ~~~u~ UJ Ute Jprr:-, OJ a r utse as--rrrropagates
.. -
:~ ~ .--.~~---~.
, ~1:
... .. ,._. .
~
'HAl .
...
.:1.·
•0
To illustrate the ideas of the preceding section and to show the validity of the
"oncep1 oi group ve10cny, we now cons1der a specific model for the dependence
*S. Chu and S. Wong, Phys. Rev. Letters 48, 738 (1982); A. Katz and R R Alfano, Phys. Rev. Letters,
49, 1292 (1982); s_ Chu, and s_ Wono ibid.1293. R Seoard and., "· _,_ Phuo T ett •no. "' 0 lo ()0<\
'R Y. Chiao and A.M. Steinberg, in Progress in Optics, VoL 37, ed. E. Wolf, Elsevier, Amsterdam
" '"-'
\ 7],-p- tuu,
Sect. 7.9 Illustration of the Spreading of a Pulse as It Propagates in a Dispersive Medium 327
of frequency on wave number and calculate without approximations the propa-

gabon of a pulse m this model medium. Before specifying the particular model
it is necessary to state the initial-value problem in more detail than was done in
(7.80) and (7.81). As noted there, the proper specification of an initial-value
problem for the wave equatiOn demands the mitial values of both functiOn u(x. 0)
and time derivative au(x, 0)/at. If we agree to take the real part of (7.80) to obtain
u(x, t),
u(x, t) = -1 -1-
2 Vb
r
J m 00
A(k)eikx-i<v(k)t dk + c.c. (7.90)
then it is easy to show that A(k) is given in terms of the initial values by:
1
A(k) = - -
Joooo
. i au
e-•kx u(x, 0) + - - (x, 0) dx (7.91)
Vv'e take a Gaussian 111odulated oscillation

u(x, 0) - e x'i 1 L' cos k 0 x (7.92)
as the initial shape of the pulse. For simplicity, we will assume that
u
- (x, 0) = 0 (7.93)
This means that at times immediately before z 0 the wave consisted of two
pulses, both moving toward the origin, such that at t = 0 they coalesced into the
shape given by (7.92). Clearly at later times we expect each pulse to reemerge
on the other side of the origin. Consequently the initial distribution (7.92) may
be expected to split into two identical packets, one moving to the left and one
to the right. The Fourie1 amplitude A(k) for the pulse desciibed by (7.92) and
3 i
(7.94)
The symmetry A(- k) - A(k) 1s a reflectiOn of the presence of two pulses trav-
eling away from the origin, as is seen below.
To calculate the waveform at later times, we must specify w = w(k). As a
model allowing exact calculation and showing the essential dispersive effects, we
assume
w(k) (7.95)
where vis a constant frequency, and a is a constant length that is a typical wave-
length where dispersive effects become important Equation (7 95) is an approx-
imation to the dispersion equation of the tenuous plasma, (7.59) or (7.61 ). Since
the pulse (7.92) is a modulated wave of wave number k - k 0 , the approximate
arguments of the preceding section imply that the two pulses will travel with the
group velocity
(7.96)
and will be essentially unaltered in shape provided the pulse is not too narrovc
(7.97)
The integrals can be performed by appropriately completing the squares in the

exponents. The result is
u(x, t) =
.
r
(x - va k 0 t? 2
-
exp
?'
la llt
LL i
T 2 I 2 2
L.
,~J I'
~
lpp >v• i/, 1 ' {/, 1. \
L
. 2 HL -r ·u·
7 \' •u •UJ
1 + 1
LL
(7.98)
Equation (7.98) represents two pulses traveling in opposite directions. The peak
amplitude of each pulse travels with the group velocity (7.96), while the modu-
lation envelop remains Gaussian in shape. The width of the Gaussian is not
constant, however, but increases with time. The width of the envelope is
L(t) =
rLLz + (at
2 )21,,2
J (7.99)
Thus the dispersive effects on the pulse are greater (for a given elapsed time),
the sharper the envelope. The criterion for a small change in shape is that
L >> a. Of course, at long times the w1dth of the Gauss1an mcreases linearly w1th
but the time of attainment of this asymptotic form depends on the rano (Lla).
A measure of how rapidly the pulse spreads is provided by a comparison of r (t)
2
given by (7.99), with Vgt va k 0t. Figme 7.15 shows two examples of curves of
the position of peak amplitude (vgt) and the positions vgt::!:: L(t), which indicate
the spread of the pulse, as functions of time. On the left the pulse is not too
narrow compared to the wavelength k 0 1 and so does not spread too rapidly. The
Sect. 7.9 Illustration of the Spreadmg of a l'ulse as U Propagates m a utspersiVe tvtemum J~7
koL>>l
II i ""'
Fi ..ure 7.15 Chan~>e in shane of a wave nacket as it travels along. The broad packet
containing many wavelengths (k 0 L >> 1), is distorted comparatively little, while the
narrow packet (k 0 L :s 1) broadens rapidly.
pulse on rne rignt, nowever, is so narrow i · · _J LwH IL " n:ry ""' _J 'l-
out and scarcely represents a pulse after a short time.
'''- -'- "' '- ' ' ' -' ' ..1 ,-, ',1 .l. 1"7 WI\ .f
•u0 u "''"' ''"'"uuo auv '-' HU"''-' U'-''-'" "'-'"' -r \ . -J
initial pulse shape and dispersion relation (7.95), their implications are of a more
general nature. We saw in Section 7.8 that the average velocity of a pulse is the
group velocity vg = dwldk = w'. The spreading of the pulse can be accounted
for by noting that a pulse with an initial spatial width dx 0 must have inherent in
it a spread of wave numbers dk ~ (lldx 0 ). This means that the group velocity,
wnen evamatea tor vanous k values wnnm me pUlse, nas a spreau in iL 01 Llle
_,
(7.101)
LlXo
At a time t this implies a spread in position of the order of dugt. If we combine

the uncertainties in position by taking the square root of the sum of squares, we
ooram me w1arn Llx~r) at nme r:
I , 2
(7.1 02)
We note that (7.102) agrees exactly with (7.99) if we put dx 0 = L. The expression
~~-~UL) !Ul '-l.A~!) M!UWS LHC ~ u ~
. . " T v", uuuuv" pu>o'-'
0
-' O[" _,_ . .
rapidly because of its broad spectrum of wave num hers, and vice versa. All these
• ..1, _,. • •• ' " " " ~- ;~c ThPu form thP hoc;c of thP
'J -J J
Heisenberg uncertainty principle. In wave mechanics, the frequency is identified

with energy divided by Planck's constant, while wave number is momentum di-
vided by Planck's constant.
The problem of wave packets m a dtsstpattve, as weu as msperstve, memum
<1- -" •n<- ...1 f"', ' ""~ j:,= ..1;, •ccPrl onoh1t;rollu hnt thP on-
. '<' . : . . -" J -~ • -' •
aiyLtCai exprt"'Ions are nuL r · nnerpreLeu 1'" y • .m T ""-
"" '""~ wllve nackets are attenuated and distorted appreciablv as thev prop-
agate. The reader may refer to Stratton (pp. 301-309) for a discussion of the
problem, mcludmg numenca1 exampres.
7.10 Causality in the Connection Between D and E;

Kramers-Kronig Relations
A. Nonlocality in Time
Another consequence of the frequency dependence of E(w) is a temporally
-1 • L • .
u~n· LHC , u V'~A, 1) auu llH:O vlv..:UlC HeJQ
E(x t). If the monochromatic comnonents of freauencv ware related hv
O(x ,..) - F( ,.,)F,{y ,,,) (7 1 ncn
thP -'
_, nn timP -~~ h~ . ~et ~• rl ~ · P . it. 'T'.
r • I
the spatia ~ ~ mtegra
coor d'mate as a parameter, t h e Founer . : an d requeue~~
Is-.'m time f
can be written
D(x, t) = -1-
v2; -oo
r D(x, w)e-""' dw .
and (7.104)
1 roo
D(x, w) = - -
'\~
J ···oo
D(x, t')eiwt' dt'
urith
'
-'ina Pnn~ti~~e •~~ ~<', 'T'h~ onhotit ,,·
'
,.- t7 1 n'l\
' . .. '
.-. -nr.,. ' -,s.\
1 roo
D(x, t) = ,~ J 00
E(w)E(x, w)e iwt dw
-·· . ,, _,
'xr~ n~ in o~rt t~ .D
-r
-~ 07/.
' -,
.\
. ·-
r r
~v ~·-·
D(x, t) = ]__ dw E(w)e-iwt dt' eiwt'E(x, t')

21T = -oo
With the assumption that the orders of integration can be interchanged, the last
expression can be written as
D(x t) F. lUx t\ + {oo (';(. ' t ,-) r/,- t7 1 n'>>

c J-oo
where G( T) is the Fourier transform of Xe - E(w)IE0 - 1:
00
1
G(T)- f.,(,,,)/.,, Jlp-iWT r/r, t7 1 nr;>
2TT ~-oo
~
'
Equations (7.105) and (7.106) give a nonlocal connection between D and E, in
""'-J.. n "' · _, .d <~. _, • "'-'-' • •L ,, '" '·
· r <H~ • ll~<U Uo ollll~O VH<~< ollUU to U t ~ W) 10
*Equations (7.103) and (7.105) are recognizable as an example of the fa/tung theorem of Fourier
;"'v!5'"',, ;, ~\'h ~\'h .:,,/ aml U\WJ, O\W), L\WJ are LWO '~" u1 wnccions re1a1eu in pairs Dy tne
Fourier inversion formulas (7.104), and
C\W) U\W)O\W)
th
"'
''
"6 ,
1 r~
C(t) yih L A(l' )B(t t') dt'

Sect. 7.10 Causalitv iu the Connection Between D and E; Kramers-Kronig Relations 331
independent of w for all w, (7.106) yields G(7) oc 8(7) and the instantaneous
connection is obtained, but if E( w) varies with w, G( 7) is nonvanishing for some
values of 7 differentTrom zero.
B. Simole Model for G(r), Limitations

To illustrate the character of the connection implied by T7.f05J ana (7.106)
we consider a one-resonance version of the index of refraction (7.51):
.{ .. \!~ 1 2 f, .•2 2 ;",.•)-! (7107)
P•
The susceptibility kernel G(7) for this model of E(w) is
-iwT
(7.108)
27r J = w0 w ryw
The integral can be evaluated by contour integration. The integrand has poles in
+ho 1 · J..~lf" ~~~~o ~+
ly y2
w 12 = --
2 +- v0 , where ~ = w5 - 4 (7.109)
!'or 7 <- u tne contour canoe croseu 1u Lllt: · uau-r w1uwuL -o '""
value of the integral. Since the integrand is regular inside the closed contour,
the integral vanishes. For 7 > 0, the contour is closed in the lower half-plane and
the mtegral 1s g1ven by 7m times me resiaues m me two p01es. 1ue Kerucr
(7.108) is therefore
Sin Vn7
tl(7) ~/.llUJ
vo
where 0(7) is the step function [0(7) = 0 for 7 < 0; 0(7) = 1 for 7 > 0]. For the
dielectric constant 17.'l1 l the kernel 01 7) 1s 1ust a Jmear supemosnwn 01 Lerms
like (7.110). The kernel G( 7) is oscillatory with the characteristic frequency of
the medium and damped in time with the damping constant of the electronic
oscillators. The nonlocalitv in time of the connection between D and E is thus
confined to times of the order of y 1 . Since y is the width in frequency of spectral
lines and these are typically 107 -10 9 s-\ the departure from simultaneity is of
the order of 10- 7 -10- 9 s. For frequencies above the microwave region many
vJ -o ·o · J
the displacement D at a given instant of time.

Equation (7.105) is nonlocal in time, but not in space. This approximation is
.1"-' "-' -'d. • .1 • • A"+• ·" -' f:.,J;I, J.. onln +hn+ • ]Q~<>P
.UHU i'~VHUvU >!' ·o
compared with the dimensions involved in the creation of the atomic or molecular
polarization. For bound charges the latter scale is of the order of atomic dimen-
. ' ' ·• £ -'.".1. ' +L •• c. . +." ,j, ,f"
w can be expected to hold ror rrequencws weu oeyona me visible range. ror
conductors, however, the presence of free charges with macroscopic mean free
• . • 1 ' ' '' .• .• 1.
pau" lllt: vr a ~ \ w} v~ u \ w} '""a" UCT ~ _., +v
frequencies. For a good conductor llke copper we have seen that the dampmg
constant (corresponding to a collision frequency) is of the order of y 0 ~ 3 X 10
13
s- 1 at room temperature. At liquid helium temperatures, the damping constant

mav be 10 -.; times the room temperature value. T akmg the ""ffcilir verocny m
hydrogen (c/137) as typical of electron velocities in metals, we find mean free

paths of the order L ~ c/(137y0 ) ~ 10-4 m at liquid helium temperatures. On
the other hand, the conventional skin depth 8 (5.165) can be much smaller. of
the order of 10 7 or 10 8 m at microwave frequencies. In such circumstances,
Ohm's law must be replaced by a nonlocal expression. The conductivity becomes
a tensorial quantity depending on wave nnmher k and frequency,,, The associ
ated departures from the standard behavior are known collectively as the anom-
alous skm effect. I hey can be ut1hzed to map out the Fermi surfaces in metals.*
Similar nonlocal effects occur in superconductors where the electromagnetic
properties involve a coherence length of the order of 10- 6 m.t With this brief
mention of the limitations of (7.105) and the areas where generalizations have
been fruitful we return to the discussion of the physical content of (7.105).
C. Causality and Analyticity Domain of c=(w)

The most obvious and fundamental feature of the kernel (7.110) is that it
vanishes for r < 0. Th1s means that at bme t only values of the electnc field prior
to that time enter in determining the displacement, in accord with our funda
mental ideas of causality in physical phenomena. Equation (7.105) can thus be
written
D(x, t) { r
c0 E(x, t) I Jo G(i)E(x, t (7.111)
This is, in fact, the most general spatially local, linear, and causal relation that
can be written between D and E in a uniform isotropic medium. Its validity
transcends any specific model of c(w). From (7.106) the dielectric constant can
be expressed in terms of G( r) as
c(w)!co = 1 + r G(r)e;WT dr (7.112)
This 1 elation has several interesting consequences. From the reahty of D, E, and
therefore G(i) in (7 111) we can deduce from (7 112) that for complex w,
E*(w*)lc0 (7.113)
Furthermore, if (7.112) is viewed as a representation of e(w)/e0 in the complex
w plane, it shows that c(w)/c 0 is an analytic (unction o(w in the upper half-plane,
provided G( r) is finite for all r. On the real axis it is necessary to invoke
the "physically 1 easonable'' requirement that G(r) --> 0 as r--> co to assure that
c(w)/cg is also analytic there This is twe for dielectrics, but not for condnctors,
where G( r) --> ulc0 as r--> co and c( w)!c0 has a simple pole at w = 0 ( c--> iu!w
as w--> 0). Apart, then, from a possible pole at w - 0, the dielectric constant
c(w)lc0 is analytic in w for Im w > 0 as a direct result of the causal relation (7.111)
*A I!. Pippanl, in Reports en !'Fegress in Physics 23, 176 (1960), and the arhele entitled "The D)
namics of Conduction Electrons," by the same author in Low- Temperature Physics, Les Houches
Summer School (1961), eds. C. de Witt, B. Dreyfus, and P. G. de Gennes, Gordon and Breach, New
York (1962). The latter article has been issued separately by the same publisher.
1
See, for example. the article "Superconductivity" by M. Tinkham in Low Temperature Physics, op.
Sect. 7.10 Causality in the Connection Between D and E; Kramers-Kronig Relations 333
between D and E. These properties can be verified, of course, for the models
111 I • .J.11 <111U / • .J.L.
The behavior of E( w )!Eo - 1 for large w can be related to the behavior of

G( T) at small times. Integration by parts in (7.112) leads to the asymptotic series,
iG(O) G'(O)
E(w)/E0 - 1 = - - - - - + ···
w w2
where the argument of G and its derivatives is T = o+. It is unphysical to have
G(O-) = 0, but G(O+) of. 0. Thus the first term in the series is absent, and
• . 0 • •
E~ w )I Eo l 1a11s ou aL mgu trt:4uem.atos as w , J uSL as was tOuHu m ~ 1 •.JY) 1u1 LHto
oscillator model. The asymptotic series shows, in fact, that the real and imaginary
etc ,f ,(. ,\!. 1 J. ,j- f, ]. ·~ ,] >O
r '"'/ -v -o·
1\ I I 1\
Re[E(w)!E0 - 1] = 01--zl, Im E(w)/E0 = 01 3 (7.114)
w w·
These asymptotic forms depend only upon the existence of the derivatives of
f"'f _\ _, f\+
~\ '} U>VUHU < v •
u. Aramers-Aroml( xetatwns
The analyticity of E( w)!Eo in the upper half-w-planc permits the usc of Cau-
,j- ' tl- t, >lot th~ ·~ol onrl · ,n:noru no ct ~f d.,\/. th~ r~ol ovio
-J r ·' • C ·u ·
For any point z inside a closed contour C in the upper half-w-plane, Cauchy's
theorem gives
E(z)IE0 = 1 +~f. [E(w'!!Eo- 1 ] dw'

2m c w - z
The contour C is now chosen to consist of the real w axis and a great semicircle
M infinitv in the nnner h"lf-n],ne From the "svmntotic exnansion iust discussed
or the specific results of Section 7.5.D, we see that EIE0 - 1 vanishes sufficiently
rapidly at infinity so that there is no contribution to the integral from the great
semicircle. Thns the ::mchv inte!'rl'll c>1n he
1 (~ [E(w')/E0 - 1] dw'
dzVEo 1 + Lm
(7.115)
J oo w z
where z is now any point in the upper half-plane and the integral is taken along
th~ rPol ovio Tolrinn th~ ]; it oo th~ ~~mnl~v frc>nnc>n~u >o~hc>< th~ rc>ol ovic
-rr
from above, we write z = w + i8 in (7.115):
roo
LE~wpE0 lJ dw'
E( w)!E0 = 1 + l
2m
. J-~ w' - w- i8
(7.116)
p, ,] tJ. ,f tl-
'" . tJ. -'· f, th
distortion of the"contour along the real axis by giving it an infinitesimal semicir-

cular detour below the ooint w' - w. The denominator can be written formally
as
i
pf '
i
.j. _,.np,(,,,' ,,, (7 11 71
w' - w- i8 w' - w
..1..14 Chapter 7 Plane Electromagnetic Waves and Wave PropagatiOn-S!
where P means principal part. The delta function serves to pick up the contri-
omion rrom me smau semicircle going in a positive sense nanway arouna the
pole at w 1 = w. Use of (7.117) and a simple rearrangement turns (7.116) into
E(w)/Eo = 1 + ~P r= [E(~:~~o: lj dw 1
(7.118)
r
The real and imaginary parts of this equation are
1
1 Im E( w )1Eo
Re E(w)/E0 = 1 + '"P dwl
= w' - w 1'7 11m
\ '!
1
1 P
Im E(w)IE0 = -- r==[Re E( W )I Eo I
- 1] d
W
1
These relations, or the ones recorded immediately below, are called Kramers-
Kronig relations or dispersion relations. They were first derived by H. A. Kramers
(1927) and R. deL. Kronig (1926) independently. The symmetry property (7.113)
shows that Re E(w) 1s even m w, while Im E(w) IS odd. The mtegrals m (7.119)
c
can thus be transformed to span only positive frequencies:
1 1
2 W Im E(w )/E0
,,,~z
1
Re E(w)/E0 = 1 + "'"P ,,,2 dw {'7 1 '1(\\
.
Im E( w)IE0
2w p
71'
r
JO
[Re E(w )1Eo - 1] dwl
w w
1
\ 'J
In writing (7.119) and (7.120) we have tacitlv assumed that E(w)IE0 was regular
at w = 0. For conductors the simple pole at w = 0 can be exhibited separately
with little further complication.
The Kramers-Kronig relations are of very general validity, following from
little more than the assumption of the causal connection (7.111) between the
polarization and the electric field. Empirical knowledge of Im E( w) from absorp-
tion studies allows the calculation of Re E(w) from the first equation in (7.120).
1 ne connectwn between absorptton ana anomalous d1sperswn, snown m t'lg. 1.l'l,
is contained in the relations. The presence of a very narrow absorption line or
. .
oanu at w w0 can oe approx1mareu oy raKing
7rK
Jill E~W) o~w Wo} + · · ·
2Wo
wnere n.. is a constam anu me ums inuicate me omer ~smommy varying) comri
butions to Im E. The first equation in (7.120) then yields
RP "'(,,,,
K (7 1 ?1)
" + w~- w"
fnr thP hPh~v;e>r ,.,f RP d ,,,\ nPQr hnt ne>t PvQc>thl ~t
-o
'·'- ThP tPrm if' n.,f,
. '/
' . . ·r.
lilt: Stuwty varymg pan 01 Kt: En trum rne more remote Lu
Tm F. Th" · 1rttinn (7 1 ?1) PYhihit, thP. rr~nirl vr~rir~tinn nf RP F( ,.,) in the
·-LL -L ..1 .£ .L 1.' -L . ..,, .c . ,.~ 0 . .c "' • • ..1•L A
~ VL UU UVOVLjJUVU u u v , O U V n U UL ~ Lb' loU LVL UUvO VL U U H v •
more realistic descrintion for Im E would lead to an exnression for Re E in com-

y
,~,. ...1 '+l- +l- L ,J., ,J.,
• . "''
"b'
'7 0
.. 'T'L ..1. ,.· , f +J.,;c ;0
left to the problems at the end of the chapter.

Sect. 7.11 Arrival of a Signal After Propagation Through a Dispersive Medinm 335
Relations of the general type (7.119) or (7.120) connecting the dispersive and
absorptive aspects of a process are extremely useful in all areas of physics. Their
widespread application stems from the very small number of physically well-
founded assumptions necessa1:y for their derivation. References to their appli
cation in particle physics, as well as solid-state physics, arc given at the end of
the chapter. We end with mention of two sum rules obtainable from (7.120). It
was shown in Section 7.5.D, within the context of a specific model, that the di-
electric constant is given at high frequencies by (7.59). The form of (7.59) is, in
fact, quite general, as shown above (Section 7.10.C). The plasma frequency can
therefore be defined by means of (7.59) as
Provided the falloff of Im E(w) at high frequencies is given by (7.114), the first
Kramers Kronig relation yields a sum rule for w~:
w~ =
2
1T
r
Jo w Im E( w)IEo dw (7.122)
This relation is sometimes known as the sum rule for oscillator strengths. It can
be shown to be equivalent to (7.52) fv1 the dielectiic constant (7.51 ), but is ob-
viously more general.
The second sum rule concerns the integral over the real part of E(w)
and follows from the second relation (7.120). With the assumption that
[Re E(w')IE0 - 1] = -w~lw' 2 + 0(1/w' 4 ) for all w' > N, it is straightforward to
show that for w > N
Im E(w)/E0 + j [Re E(w')IE0

1TW N 0
It was shown in Section 7.10.C that, excluding conductors and barring the un-
physical happening that G(O+) -F D, Im E('•') behaves at large frequencies as u1- 3
It therefore follows that the expression in curly brackets must vanish. We are
thus led to a second sum rule,
p
N Re E(w)/E0 dw - 1 + N 2 (7.123)
J0
which, for N ~ oo, states that the average value of Re E( w)!E 0 over all frequencies
is equal to unity. For conductors, the plasma frequency sum rule (7.122) still
holds, but the second sum rule (sometimes called a superconvergence relation)
has an added term '1Tui2E0 N, on the right hand side (see Problem 7 23) These
optical sum rules and several others are discussed by Altarelli et a!.*
7.11 Arrival of a Signal After Propagation Through

a Dispel'sive Medium
Some of the effects of dispersion have been considered in the preceding sections.
There remains one important aspect, the actual arrival at a remote point of a
*M. Allarelli, D. L. Dexter, H. M. Nussenzveig, and D. Y. Smith, Phys. Rev. B6, 4502 (1972).
336 Chapter 7 Plane Electromagnetic Waves and Wave Propagation -SI
wave train that initially has a well-defined beginning. How does the signal build
up? If the phase velocity or group velocity is greater than the velocity of light in
~. '
-r
"'-" "~ '~' ~n~
-, - -r
ote rlnPe th~ eian~l HJl'lttC f'"tP
J
than allowed by causality and relativity? Can the arrival time of the disturbance
hP ;:;!u~n ~n Hn~~h!;;unn -1 .C:. itirm? ion' were examined authori-
tatively by Sommerfeld and Brillouin in papers published in Annalen der Physik
in 1914.* The original papers, plus subsequent work by Brillouin, are contained
. ~ _,._,_ _, . . .. ·'· ur. n . "I'C'v~"~ l7nl. . h
'" '" • -r ·o ••,.. ·J • • J
Brillouin. A briefer account is given in Sommerfeld's Optics, Chapter III. A com-
.L.• -1: • 1 .,_ rl t, •hn' -~11, ~n~nli~~t~rl t \X/p h~ot nnh. thP nn~l.
y . 'J • • ,, . ' ,. ~ -
nanve remures. 1 ne reaaer c<111 lllVlC UC:ldll Ill llli:O ULCU v< lllCOimu~~ ctHO
second edition of this book, from which the present account is abbreviated.
. . ' . _,_ . . _, ~.
rv1 we a ' " " " uaw .. J ·u~-
uum on a semi-infinite uniform medium of index of refraction n( w) filling the

region x > 0. From the Fresnel equations (7.42) and Problem 7.20, the amplitude
ot the electnc held ot the wave ror x .> u IS given oy
roo r ')
u( Y t) I A( w)ei<\wJx -;," dw ( /.lL4)
J 00 1 + n(w)_
wh"r"
A(w) = - 1 r U;(O, t)eiwt dt (7.125)

27T -oo
is the Fourier transform of the real incident electric field u;(x, t) evaluated just
. .
OULsiuemememum,aLx v . '""wavcou1
' " ' .
"~wJ"
k(w) = w n(w) (7.126)

L
""rl ;, o-"n"r" lhr 'l"Y

·
with nositive ima!!inarv nart corresnondin!! to absorp
~, _,, ,~ · :, •• te ,.
UUU VI --uurr:rrg-pr~l' __ 0
• _J ou~u•a m~ cry u 1.
the wave number can be treated as real for most purposes, but there is always
• -1, '" ' " " t fTI, ,tl • olJu ,l 'th~t in {7 J ')L[\.fre.an"n"V- not mP
-r ·o ~ · L· •
wave number, is used as the independent variable. The change from the practice
of Sections 7.8 and 7.9 is dictated by the present emphasis on the time develop-
. . '
men t or rne wave<rr<rTIXerr IJV uJL 111 s1-'"c"'. J
We suppose that the incident wave has a well-defined front edge that reaches
A A ~ .{{\ ,\ {\ ~- ~ (\ U"th ' ' .nnl - '- ' lll\1
A V HV< VvlVlv I V. • .u '\' , <J r -J
reasonable mathematJcllireqmrements, thiS conditiOn on u(U, t) assures LnaL.!i~WJ

is analytic in the upper half-w-plane [just as condition (7.112) assured the anal-
..
yticity ot E( wTTifereTvenerauy, A( w) will nave sing ~ ' v i lldll-w·
plane determined by the exact form of u(x, t). We assume that A( w) is bounded
for Iwl ~ oo.
The index of refraction n(w) is crucial in determmmg the detailed nature or
the propagation of the wave in the medium. Some general features follow, how-
*A Sommerfeld, Ann. Phys (Leipzig) 44, 177 (1914). L. Brillouin, Ann. Phys. (Leipzig) 44,203 (1914).
An exhaustive treatment is given in K. E. Oughslun and G. C. Sherman, Electromagnetic Pulse
1
Prova•ation in Causal Dielectrics. Springer-Verlag, Berlin (1994).

Sect. 7.11 Arrival of a Signal After Propagation Through a Dispersive Medium 337
ever, from the global properties of n( w). Just as E( w) is analytic in the upper half-
w plane, so IS n(w). Furthermore, (7.59) shows that for lwl --'> oo, n(w) --'>
1- w~/2w2 • A simple one-resonance model ofn(w) based on (7.51), with resonant
frequency w0 and damping constant ')', leads to the singularity structure shown in
Fig. 7.16. The poles of E(w) become branch cuts in n(w). A multiresonance ex-
pression for E leads to a much more complex cut structure, but the upper plane
analyticity and the asymptotic behavior for large Iwl remain.
The proof that no signal can propagate faster than the speed of light in
vacuum, whatever the detailed properties of the medium, is now str aightfor wm d.
We consider evaluating the amplitude (7.124) by contour integration in the com-
plex £!!plane Since n('") ==> 1 for wl --'> oo the argument of the exponential in
I
(7.124) becomes
i<f>(w) i[k(w)x wt] --'>

c
for large Iwl. EVIdently, we obtain a vanishing contribution to the integral by
closing the contour with a great semicircle at infinity in the upper half-plane for
x > ct and in the lo\ver half-plane for x < ct With n(£<>) and A (u•) both analytic
in the upper half-w-plane, the whole integrand is analytic there. Cauchy's theo-
rem tells us that if the contour is closed in the upper half-plane (x > ct), the
integral vanishes. We have therefore established that
u(x, t) = 0 for (x - ct) > 0 (7.127)
provided only that A(w) and n(w) are analytic for Im w > 0 and n(w)--'> 1 for
Iu>l--'> oo Since the specific form of n(w) does not enter, we have a general proof
that no signal propagates with a velocity greater than c, whatever the medium.
For ct > x, the contour is to be closed in the lower half-plane, enveloping
the singularities. The integral is dommated by different smgulanties at different
mw
Re w
X X
Wh Wd
Figure 7.16 Branch cuts defining the singularities of a simple one-resonance model for
the index of refraction n(w). For transparent media the branch cuts lie much closer to
(but still below) the real axis than shown here. More realistic models for n( w) have
more complicated cut structures. all in the lower half-w-planc. The crosses mark the
possible locations of singHlarities of A(w).
338 Chapter 7 Plane Electromagnenc Waves and Wave Propagafton SI
times Brillouin and Sommerfeld nsed the method of steepest descent* to eva!-
uate (7.124) in various regimes. We sketch the chief aspects using the concepts
ol the less ngorous method of stationary phase. The method of stationary phase
is based on the idea that the phase </J(w) in an integral such as (7.124) is generally
large and rapidly varying. The rapid oscillations of e'"' over most of the range of
integration mean that the integrand averages almost to zero. Exceptions to the
cancellation occur only when </J(w) is "stationary," that is, when p(w) has an
extremum. The integral can therefore be estimated by approximating the integral
at each of the points of stationary phase by a Taylor series expansion of if5(w)
and summing these contributions.
We use the idea of stationary phase to discuss the qualitative aspects of
the arrival of the signal without explicit use of the integration formulas. With
</J(w) = k(w)x - wt and k(w) given by (7.126), the stationary phase condition
dk dn t
c - - n(w) + w - - - for t > t0 - xlc (7 128)
dw dw t0
The earliest part of the wave occurs when t/t0 is infinitesimally larger than unity.
From the global properties of n(w) we see that the point of stationary phase is
at Iwl ~ 00 , where n ~ 1. Explicitly, we have
dk
0
dw
showing that the frequency of stationary phase w., = wPt\12(tlt0 - 1) depends

only on tlt0 and w~, a global property of the index of refraction. The incident
wave's A( ws) is presumably very small. The earliest part of the signal is therefore
extremely small and of very high frequency, bearing no resemblance to the in-
cident ¥/aYe. This part of the signal is called the first (}I" S(}mmet feEd pt ecmsm.
At somewhat later times, the frequency w, slowly decreases; the signal grows
very slowly in amplitude, and its structure is complex.
Only when tlt 0 in (7.128) reaches n(O) is there a qualitative change m the
amplitude. Because w - 0 is now a point of stationary phase, the high frequency
of oscillation is replaced by much lower frequencies. More important is the fact
that d 2 k( w )!dw2 = 0 at w = 0. In such circumstances the stationary phase
approximation fails, giving an infinite result. One must improve the approxi-
mation to include cubic terms in the Taylor series expansion of <P ( w) around
"' "', The ampljtnde is expressible in terms of Airy integrals (of rainbow
fame). The wave becomes relatively large in amplitude and of long period for
times t :2': n(O)t0 • This phase of development is called the second or Brillouin
At still later times, there are several points of stationary phase. The wave
depends in detail on the exact form of n(w). Eventually, the behavior of A(w)
begins to dominate the integral. By then the mam part ol the wave has arnved
at the point x. The amplitude behaves in time as if it were the initial wave prop-
agating with the apptopriate phase velocity and attenuation.
The sequence of arrival of the tiny high-frequency Sommerfeld precursor
the larger and slower oscillating Brillouin precursor, and then the main signal,
*~@@J-effreys and Jeffreys (Seetion 17.04) or Botti turd Hlvlf (Appendix III) for a discussion of this
method, originally developed by P. Dehye.
Ch. 7 References 339
and indeed their detailed appearance, can differ greatly depending upon the
specifics of n(w), A(w), and the position x of observation A textbook example
can be found in Oughstun and Sherman (op. cit., Fig. 9.10, p. 383).

The whole subject of optics as an electromagnetic phenomenon is treated authori-
Born and Wolf.

Their first chapter covers plane waves, polarization. and reflection and refraction, among
other topics. Stokes parameters are covered there and also by
Stone.
A very complete discussion of plane waves incident on boundaries of dielectrics and
conductors is given by
Stratton Cha tcr IX.
Another good treatment of electromagnetic waves in both isotropic and anisotropic media
is that of
Landau and Lifshitz, Electrodynamics of Continuous Media, Chapters X and XI.
A more elementary, but clear and thorough, approach to plane waves and their properties
Adler, Chu, and Fano, Chapters 7 and 8.

I he umque opt1cal properties of water (Fig. 7.9) and the usc of Kramers-Kronig
relations arc discussed by
~~. \V. \Villimns and E. T. At akawa, Optical a1Jd diclccllic pi ope1 ties ufrnater ials
relevant for biological research, in Handbook on Synchrotron Radiation, Vol. 4,
eds. S. Ebashi, M. Koch, and E. Rubenstein, North-Holland, New York (1991),
See also Vol. 1 (1972) of

Franks. Felix, ed., Water: A Comprehenr;dve Treatise in 7 volumes Plenum Press
New York (1972-82).
Propagation of waves in the ionosphere has, because of its praetieal impo1 ta11ee, an
extensive literature. The physical and mathematical aspects are covered in
Budden
Wait
The special topic of whistlers is discussed in detail in
R. A. Helliwell, Whistlers and Related Ionospheric Phenomena, Stanford Uni-
versity Press, Stanford, CA (1965).
The topic of magnetohydrod~·namic and pla,·ma waves has a very large literatl!re.
The American .T ournal of Physics Resource Letters,
G. Bekefi and S. C. Brown, Plasma Physics: Waves and radiation processes in
plasma, Am. J. Phys. 34, 1001 (1966)
C. L. Grabbe, Plasma waves and instabilities, Am. J. Phys. 52, 970 (1984)
will lead the reader to appropnate references for further study. For dJScussJon of MHD
in the context of astrophysics, sec
Oxford I Iniversity Press_
A survey of the basic physics involved in the interaction of light with matter can be
found in the semipopular artJcle by
V. F. Weisskopf, Sci. Am. 219, 3, 60 (September 1968).
The propagation of waves in dispersive media is discussed in detail in the book by

Brillouin.
The distortion and attenuation of pulses in dissipative materials are covered by
Stratton, pp. 301-309.
The generalization of dispersion to nonlocality in space as well as time via E( w, k) is
an important concept for inelastic electromagnetic interactions. Problem 7.26 introduces
some of the ideas. The uses of are discussed in
Interactions in Solid State Plasmas,
in Solid State
I, Academic Press, New
F. Stern, in Solid State Physics, Vol. 15, eds. F. Seitz and D. Turnbull, Academic
7.1 For each set of Stokes parameters given below deduce the amplitude of the electric
field, up to an overall phase, in both linear polarization and circular polarization
bases and make an accurate drawing similar to Fig. 7.4 showing the lengths of the
axes of one of the ellipses and its orientation.
(a) s 0 = 3, s 2 = 2, S3 = -2;
(b) s0 = 25, s2 = 24, s3 = 7.

7.2 A plane wave is incident on a layered interface as shown in the figure. The indices
of refraction of the three non permeable media are n, nz, n 3 . The thickness of the
Problem 7.2
Ch. 7 Problems 341
that there is no reflected wave for a frequency «Jo. What thickness d and index
of refraction n 2 are necessary?
7.3 Two plane scmt-mfimtc slabs of the same umform, tsotroptc, nonpermeable, lossless
dielectric with index of refraction n are parallel and separated by an air gap (n = 1)
of width d. A plane electromagnetic wave of frequency w is incident on the gap
from one of the slabs with angle of incidence i. For linear polarization both parallel
to and perpendiculm to the plane of incidence,
(a) calculate the ratio of power transmitted into the second slab to the incident
power and the ratio of reflected to incident power;
(b) fori greater than the critical angle for total internal reflection, sketch the ratio
of transmitted power to incident power as a function of d measured in units
7.4 A plane-polarized electromagnetic wave of frequency w in free space is incident

normally on the flat surface of a non permeable medium of conductivity u and di-
electric constant E.
(a) Calculate tfie amphtude and phase of the reflected wave relative to the inci-
dent wave for arbitrary u and E.
(b) Discuss the limiting cases of a very poor and a very good conductor, and show
that for a good conductor the reflection coefficient (ratio of reflected to inci-
dent intensity) is approximately
w
R=1-2-8
c
where 8 is the skin depth.
7.5 A plane polatized elechornagnetic wave E Eie*·"-i<'!)l is incident notlllally on a
flat uniform sheet of an excellent conductor (<T >> wE0 ) having a thickness D. As-
suming that in space and in the conducting sheet p.,/ !Lo = EiE0 = 1, discuss the
reflection and transmission of the incident wave.
(a) Show that the amplitudes of the reflected and transmitted waves, correct to
the first order in ( E 0 w/<T) 112 , are:
e
'
E, (1 - e 2A) + y(1 + e 2A)
E, 2ye-A
E, 2
(1 - e A) + y(1 + e 2A)
w ere
and IJ \/2J(;;r;;;. is the penetration depth.

(b) Verify that for zero thickness and infinite thickness you obtain the proper
limiting results.
(c) Show that, except for sheets of very small thtckness, the transmtsston coeffi-
cient is
8(Rc y)ze 2D!B
T = ------;;-~---"-':-----,--=c;
1 - 2e cos(2D/ 8) +
ZD!a e
4 18
D
2
Sketch log T as a function of (D/8), assuming Rc y- 10
Define "very small thickness."
1.6 A plane wave of frequency w rs mcident normally from vacuum on a semi-infinite

slab of material with a complex index of refraction n(w) [t~ 2 (w) - €(w)/E0].
(a) Show that the ratio of reflected power to incident power is
2
1 - n(w)
R=
1 + n(w)
while the ratio of power tmnsmitted into the medium to the mcrdent power
IS
(b) Evaluate Refiw(E • D* - B · H*)/2] as a function of (x, y, z). Show that this
rate of change of energy per unit volume accounts for the relative transmitted
power T.
(c) For a conductor, with n2 = 1 + i(aiwE0), areal, write out the results of parts
a and h in the limit o,o• <"< "'· Express your answer in terms of S as mueh as
possible. Calculate~ Re(J* ·E) and compare with the result of part b. Do both
enter the complex form of Poyntmg's theorem?
7.7 A ribbon beam of plane-polmized radiation of wavelength A is totally reflected
internally at a plane boundary between two nonpermeable media with indices of
refraction nand n' (n' < n). As discussed in Section 7.4, the ratio of the reflected
to incident amplitudes is a complex number of modulus unity, E 0/E0 = exp[i<f>(i, i 0)]
for the angle of incidence i > i 0 , where sin i0 = n' ln.
(a) Show that for a "monochromatic" ribbon beam of radiation in the z direction
with an electric field amplitude, E(x)e(ikz iw<), where E(x) is smooth and finite
in transverse extent (but many wavelengths broad), the lowest order approx-
imation in terms of plane waves is
E(x, z, t) = £ JdK A(K)e'K<+ikz •wt
where e is a polarization vector, and A(K) is the Fourier transform of E(x),

with support in K around K = 0 small compared to k. The finite beam consists
of plane waves with a small range of angles of incidence, centered around the
geometrical optics value.
(b) Consider the reflected beam and show that for z > t0 the electnc field can be
expressed approximately as
E"(x, 7, t) - e" E(x" ox) exp[ik" • X io•l + i<J>(i, io)]
where e" is a polarization vector, x" is the x coordinate perpendicular to k",

the reflected wave vector, and i5x = -(1/k)[d<P(i, i0 )/di].
(c) With the Fresnel expressions of Section 7.3 for the phases <f>(i, i 0 ) for the two
states of plane polmization, show that the lateral displacements of reflected
beams with respect to the geometric optics position are
A Sill i
D~ and D,
Sill 1 cos i · sin i0
The displacement is known as the Goos Hanchen effect (op. ctt.).
78 A monochromatic plane wave of frequency "' is incident nonnally on a stack of
layers of various thicknesses t, and lossless indices of refraction ni. Inside the stack,
the wave has both forward and backward moving components. The change in the
Ch. 7 Problems 343
wave through any interface and also from one side of a layer to the other can be
described by means of 2 X 2 transfer matrices. If the electric field is written as
E = E 1-eikx + E_e-ikx
in each layer, the transfer matrix equation E' = TE is explicitly
(a) Show that the transfer matrix for propagation inside, but across, a layer of
index of refraction ni and thickness ti is
where ki = niwlc, lis the unit matrix, and ak are the Pauli spin matrices of
quantum mechanics. Show that the inverse matrix is T*.
(h) Show that the transfer matrix to cross an interface from n1 (x < x 0 ) to
Tintcrtacc(2, 1) =
1 (
l
n +1
-(n _ 1)
-(n- 1)) =
1
(n + 1) _ a (n - 1)
1
n + 1 2 2
where n = n 1/n 2 .
(e) Show that for a complete stack, the incident, reflected, and transmitted wa>•es
where t are the elements of T the product of the forward-going transfer

0
matrices, including from the material tilling space on the incident side into the
first layer and from the last layer into the medium filling the space on the
transmitted side.
7.9 A stack of optical clements consists of N layers with index of refraction n and
thickness f~o separated by air gaps (n 2 = 1) of thickness t2 . A monochromatic plane
.. a.e is i11cident normally. 'Nith appropriate thicknesses, a modest number of layers
can cause alrnost totalte:flection of a given range of freqttclleies, even for Hormal
n values (e.g., 1.3 < n < 1.8).
(a) Show the transfer matrix for the stack is T,ack = TN (I cos a 2 - ia, sin a 2 ),
where a2 = wt2 /c, and the single air gap plus foil transfer matrix is
T- (114n)/[(n + 1? cos(a1 + a 2) (n lY cos(a 1 a 2 ))1
+ ia3 [(n + 1) sin(a1 + a 2 ) + (n - 1) sin(a 1
2 2
- a 2 )]
+ 2a1 (n2 - 1) sin a 1 sin a2
2
- 2a2 (n - 1) sin a 1 cos a2 }
with a = nwt /c.
(b) If all the layers (both air gaps and foils) have optical thicknesses of a quarter-
wavelength of the incident wave, show that
T = -exp(-Aa1), where A = !n(n)
is (roughly) the amplitude "decay constant" per layer. Show that the fractional
transmitted intensity is
4 ?N
N ln(n2 )]
The asymptotic form holds for n 2 N >> 1.

''""~ .. an:. anu wave rropaganon -Sl
7.10 An arbitrary optical element of length L is placed in a uniform nonabsorbing rne.

dium with index of refraction n( w) with its front face at x = 0 and its back face at
x = L. If a monochromatic plane wave of frequency w with amplitude •/';n,(w, x, t)
= exp[ik(w)x- iwt] is incident on the front face of the element, the transmitted
wave amolitude is '''~. J w x t) Tl ,.,\ Pvnr;u ,,,\I r l \ ' ·:]. "" '" . ,.._ ive
' \ w) auu 1'""0<0'1'\W )·

A plane wave of radiation 1/Jinc(x, t), consisting of a coherent superposition of
different frequencies centered around w = w0 , with support A(w) narrow on the
scale of variation of r(w), </>(w) and k(w), is incident on the optical element. Show
that the transmitted wave for x > L is approximately
where {3 is a constant phase and x' = x - L, t' - t - T. The transit or group

delay time (sometimes attributed in another context to E. P. Wigner) is T ~
[d<f>(w)/dw]w~w,· If cT < L, some authors speak of superluminal propagation
through the element. Discuss.
7.11 A simple example of the transit time of the preceding problem is afforded by a slab
of lossless di"l' ·~'"'· ~f · " ·~-' :~-'~ -•· · · .
(a) ror a plane wave incident normallv show thM thf' .noanitnriP ~f thP
mitted amplitude is
4n
Ir( w) I = --;;oo;===~cr=;=====o~~~~=:=::==~=:=::=~2
Y[(n + 1)Z - (n - l? cos(2zW + [(n - 1? sin(2z)]
mhilP it< '"""~ :e
\ft L)
</>( w) z + arctan L(n + 1)2 (
- n - 1)2 cos(2z)
where z = nwd/c.
(b) Neglecting dispersion, show that for z--> 0 and z = rr, Irl = 1.0 and cT/d =
2
(n + 1)/2, while for z = rr/2 (quarter-wave plate), lrl = 2nl(n 2 + 1) and
-~U 2 2 l5i
'\IT .,. 'J· 0uuw atw--ura:I'Cl!a, averageu over any mteger num er
oi quaner-wavelengtn opl!cal aths, is (cT/d) n. Does this result 1P11 "~"
anything about what you might expect for the observed transit time of a long
wave train (!1w/w << 1) through a piece of window glass? Explain.
(c) Calculate numerically and plot the results as functions of z for the magnitude
of the transmission amplitude, its phase, and the transit time in units of die
for n 1.5 and n 2.0.
~. H ~·- · " ' <H • uismroances in goou conductors 1s governed bv
Tffe rrequency-aepenaent conductivitv77.58T. Consider lonaitudinal clPcctrir fi~
in a conductor •><ina nh.n'< I om th. ,,· 'J ' ." .,_ • . ' ..
of Coulomb's law.
(a) Show that the time-Fourier-transformed charge density satisfies the equation
[IT(w) - iwE0 ]p(x, w) = 0
\UJ '-""'t; <H<O repreSelllaLJOn IT\ W) 1T0 /(l lWT), Where ITo EnW Tan TIS a
will oscillate with the plasma frequency and decay in amplitude with a decay
constant A= 1/27. Note that if you use IT(w) = IT(O) = ITo in part a, you will
Ch. 7 Problems 345
hna no oscmanons ana extremety lafJiU umufJiug willl Llle \''lVH!'J uc-.ay -.vu-
stant Aw u,/ Eo.
Reference: W. M. Saslow and G. Wilkinson, Am. f. Phys. 39, 1244 (1971).

7.13 A stylized model of the ionosphere is a medium described by the dielectric constant
(7.59). Consider the earth with such a medium beginning suddenly at a height h
and extending to infinity. For waves with polarization both perpendicular to the
plane of incidence (from a horizontal antenna) and in the plane of incidence (from
a vertical antenna),
.... ,-.,
\")
£.
'"" """'
•
.
n
"
there IS a range or . or 0
• •
:
•
• w1 w w-.u
•
,fl
•
' "
'" f<>r
. '
'", '"'u.,
L £
larger angles there is total reflection back toward the earth.

(b) A radio amateur operating at a wavelength of 21 meters in the early evening
finds that she can receive distant stations located more than 1000 km away,
but none closer. Assuming that the signals are being reflected from the F layer
of the ionosphere at an effective height of 300 km, calculate the electron
~ · C' • with the known maximum and minimum F layer densities
·' ry v 1nl2 3 · tha rl:!~,,;..,P onfl · 1?.4) Y tn" m 3 ot niPht.
7.14 A simple model of propagatiOn Of radiO waves m the earth s armospnere or iono-
sphere consists of a flat earth at z 0 and a nonuniform medium withE- e(z) tor
z > 0. Consider the Maxwell equations under the assumption that the fields are
independent of y and can be written as functions of z times e•<kx-•")
(a) Show that the wave equation governing the propagation for z > 0 is
d2 F 2
" 2 + q (z)F = 0
'
q 2 (z) = alJ.LoE(z) - k 2
and F = Ey for horizontal polarization, and
2
q (z)
2
= w J.LoE(Z) + - -
2 - -
2
1 d E
2E dz
3
4e2
(
dE
-
dz
y-e
with F = V EIE0 E, for vertical polarization.
lb\ Use the WKB anoroximation to treat the propagation of waves directed ver-
t;r"lh intn thP inn ' .(k 0) assumin!! that the dielectric constant is
given by (7.59) with a plasma frequency w"(z) governed by an electron density
like that shown in Fig. 7.11. Verify that the qualitative arguments in Section
7.6 hold, with departures in detail only for w ~ wp,m>X·
(c) Using the WKB results of part b and the concepts of the propagation of a
pulse from Section 7.8, define an effective height of the ionosphere h'(w) by
calculating the time T for a pulse of dominant frequency w to travel up and
be reflected back (h' = cT/2). [The WKB approximation is discussed in most
books on auantum mechanics.l
' "6' l) ~-
. ,.,
t.l!l 1 ne paniauy ionizeu \ .. , 'J
frequencies as an electromc prasma m a weaK magnetic nem.--nre- uruau·'l'""" uou
nulses from a pulsar allow determination of some average properties or tne mrer-
stellar medium (e.g., mean electron density and mean magnetic field). The treat-
ment of an electronic plasma in a magnetic field of Section 7.6 is pertinent.
..i4b Chapter 7 Plane Electromagnetic Waves and Wave Propagation-S!
In\
,-, -,·
T,
-c: 'h -'- . C .. U
, MIUW that
"
<oo~ '"S
<L \
•uu• •"'::"'\ ~p!
c times the transit time of a pulse of mean frequency w from a pulsar a distano.
R away is
2
ct(w) = R + e 2
2 Eon1eW
Jne(z) dz
where ne(z) is the electron density along the path of the light.
(b) The presence of the magnetic field causes a rotation of the plane of linear
polarization (Faraday effect). Show that to lowest order in the magnetic field
the oolarized li!!ht from th<e nnls"r h"' its . . Jrot"t"rl tl uwh on ..5"'
88(w):
e3
j",(J( •• ,\ ( " e 17\R l-z'i1L
2e0 cm e2w2 J II • '
where B 1,(z) is the component of B parallel to the path of the light.

(c) Assummg you nad au mdependent measure of the pulsar distance R, what
-1.. • ,].-1 "A"
0
tions, if any, about the polarization are necessary?

~oJ..p ;n n ·rl. · ,f, /, ·c \ orl / R \? \X/1
, .. '"'I''
7.16 Plane waves propagate in a homogeneous, non permeable, but anisotropic dielectric.
The dielectric is characterized by a tensor e,1, but if coordinate axes are chosen as
the principle axes, the components of displacement along these axes are related to
the electric-ficld wnt, hv n F.P· (i 1 )_ " where..E..are...t.he p; "'·
of the matrix e,1.
(a) Show that plane waves with frequency wand wave vector k must satisfy
k X (k x E) + f.LoW2D = 0
(b) Show that for a given wave vector k = kn there are two distinct modes of
propagation wltu anrerent pnase veJocilies v wl K lnat satlsry the Fresnel
equation
3
n2
2:2'2=0
i=l v -vi
where v, = 1/~ is called a principal velocity, and n, is the component of

n along the ith principal axis.
(c) Show that Da ·Db - 0 where D D, are the disolacemen ts associated with
the two modes of propagation.
7.17 Lonsu1er the problem of dispersion and waves in an electronic plasma when a
11. "J m" >Pt;~. ~ ;nn 1> . . " . 7 {,
- -v · r • · ·
(a) Show that in general the suscentibilitv tensor v" ( rv) defined thrmwh n -
2. 1J=jkEk, and Ejk = E0 ( 8jk + Xj~c), is
w2p
r 2~ ...1..1.
,,, w2( w2 - w~) ,. jk 2l..
B j"k R jkl •/J
where b is a unit vector in the direction of B0 .

(b) By straightforward diagonalization of the dielectric tensor Efk or by an airtight
].., rl •L .1.. " .1. .£ C' " L C " " .\
•pp <U •~•uno ~• .v, u u u " <:;l~<:;HVW
U'-'' Oj' j I , L, J ,
Ch. 7 Problems ~4/
(c) A plane wave ( w, k = kn) must satisfy the vector equation of Problem 7.16a.
Show that in terms of X;k the electric field and wave number must satisfy the
three homogeneous equations,
(1
o, -, +
1'\P. i'M.fn .1<'\
-, + ~ . H". - fl ; - 1 ? '
'. 0
k "'' '
where g- (cklw)L. Keeping only first-order terms m an expansion of X;k m

powers of w8 /w, show that the effective dielectric constant for propagation of
the plane wave is
wP2 - wP2 wB b • n
E+/E 1 +
w w
ror positive and negative ne11cny waves.

7.18 Magnetohydrodynamic waves can occur in a compressible, nonviscous, perfectly
conducting fluid in a uniform static magnetic induction B0 . If the propagation di-
rection is not parallel or perpendicular to B0 , the waves are not separated into
purely longitudinal (magnetosonic) or transverse (Aifven) waves. Let the angle
between the propagation direction k and the field Bo be e.
. . . . .
\RJ ->uOW ,ua, were are ,uree utueren, waves Wlw puase ve,OCh!es g1ven uy
ui = (VA COS e?
u~.3 = ~(s 2
+ u::,) ± H(s 2 + u::,f - 4s 2u::, cos 2 8]" 2
whPrP ' i< thP vPlnritv in thP flnirl "nrl " (R~I"n") 112 i< thP Alfv6n
.. ]
'J.
(b) Find the velocity eigenvectors for the three different waves, and prove that
the first (Alfven) wave is always transverse, while the other two are neither
longitudinal nor transverse.
(c) Evaluate the phase velocities and eigenvectors of the mixed waves in the ap-
nrnximMinn thet 11 >> s. Show thM for nne w"ve the nnlv "nnreci"hle com-
nnnPnt nfvPlnr;tv ;, noroiiPl tn thP "'"' · · fip]fl , wh;]p for thP nthf'r thf' nnlv
' _, ,_,'.; _, •-
'
u<op~•,
oL
'"' oL
'V oU~ H~<~ UH~ "'<H oU~ puu~
0 ~
""
UHU ~(}•
7.19 An approximately monochromatic plane wave packet in one dimension has the
instantaneous form, u(x, 0) = f(x) e 1k0", with f(x) the modulation envelope. For
each of the forms f(x) below, calculate the wave-number spectrum IA(k) 2 of the 1
packet, sketch Iu(x, 0) 12 and IA(k) 12 , evaluate explicitly the rms deviations from the
means Ll.x and Ll.k ~dcnned m terms or tne mtenstttes ju~x, un and jA~k)l'), and
test inequality (7.82).
(a) f(x) = Ne-alx!/2
(b) f(x) = Ne-" 2x) ;4
Nil - n; lrl\ fnr "' IY I < 1

(c) HX) (\ fr i -I --, 1
1\T f, I -1 <'
(d) f(x) - !'
0 for lxl >a
7.20 A homogeneous, isotropic, nonpermeable dielectric is characterized by an index of
refraction n( w), which is in general complex in order to describe absorptive
P"
348 Chapter 7 Plane Electromagnetic Waves and Wave Propagation SI
(a) Show that the general solution for plane waves in one dimension can be
written
u(x, t) = _l_ J~ dw e-''"'[A(w)e'<wlcJn(c,Jx + B(w)e-i(w!cJn<wJx]

Vh -~
whcr@ u(x, t) is a component of E or B.

(b) If u(x. t) is real. show that n( "') n*(u1)
(c) Show that, if u(O, t) and au(O, t)/ax are the boundary values of u and its deriv-
ative at x = 0, the coefficients A ( w) and B( w) arc
A(w)} _ 1 1 Joo iwt [ _ ic au ]

{ B(w) - 2 Yh -x dt e u(O, t) + wn(w) ax (0, t)
7.21 Consider the non local (in time) connection between D and E,
D(x, t) e0 {E(x, t) + Jdr G( r)E(x, t r)}
with the G( r) appropriate for the single-resonance model,

e(w)le0 = 1 + w~(w5- w2 - iyw) 1
(a) Convert the nonlocal connection between D and E into an instantaneous re-
lation involving derivatives of E with respect to time by expanding the electric
field in the integral in a Taylor series in '· Evalttate the iuteg1 als over G(r)
explicitly up to at least d2 E!dt 2 •
(b) Show that the series obtained in part a can be obtained fm mally by eonver ling
the frequency-representation relation, D(x, w) = e(w)E(x, w) into a space-
time relation,
D(x, t) = { : )E(x, t)
1
v.here the valiable w iu e(w) is replaced by w---. i(B/Bt).

7.22 I Isc the Kramers Kronig relation (7 120) to calculate th@ real part of e(w), giY@ll
the imaginary part of e(w) for positive w as
(a) Im ele0 = A[O(w- w1 ) - O(w- W;,)], W;, > w1 > 0
Ayw
(b) Im ele0 = ( )2 _
w02 -w2 2 2
+yw
In each case sketch the behavior of Im e(w) and the result for Re e(w) as
functions of w. Comment on the reasons for similarities or differences of your
7.23 Discuss the extension of the Kramers-Kronig relations (7.120) for a medium with
a static electrical conductivity u. Show that the first equation in (7.120) is unchanged,
but that the second is changed into
Im E ( w)
u
= - - -
2w
p
Joo [Re e( w') -
,z z
Eo J d ,
.w
[Hint: Consider e(u>) iulto as analytic for Im U! > 0J

7.24 (a) Use the relatwn (7.113) and the analytiCity of e(w)/e0 for Im w > 0 to prove
that on the positive imaginary axis E(ltl)/Eu is real and monotonicf:!Uy decreasing
Lh. 'I Problems J .. 7
away from the ongm towara umty as w---> zoo, prov10ea 1m E ~ u tor rea1
positive frequencies. Assume that (7.114) holds for real w.
(b) With the assumption that lm E vanishes for finite real w only at w = 0, show
that E( w) has no zeros in the upper half-w-plane.
(c) Write down a Kramers-Kronig relation for E0 /E(w) and deduce a sum rule
similar to (7.122), but as an integral over Im[E 0 /E(w)].
(d) With the one-resonance model (7.107) for E(w) determine lm E(w) and
Im[l!E(w)] and verify explicitly that the sum rules (7.122) and part c are
satisfied.
7.25 Equation (7.67) is an expression for the square of the index of refraction for waves
propagating along field lines through a plasma in a uniform external magnetic field.
Using this as a model for propagation in the magnetosphere, consider the arrival
of a whistler signal (actually the Brillouin precursor and subsequently of Section
7.11).
(a) Make a reasonably careful sketch of c dkldw, where k = wn(w)/c, for
the positive helicity wave, assuming wP/w8 <: 1. Indicate the interval where
c dk/dw is imaginary, but do not try to sketch it there!
(b) Show that on the interval, 0 < w < w8 , the minimum of c dkldw occurs at
wlw 8 =~'provided wP/wR :o- 1. Find approximate expressions for c dk/dw for
w near zero and for w near w8 .
(c) By means of the method of stationary phase and the general structure of the
solution to Problem 7.20a, show that the arrival of a whistler is signaled by a
rising and falling frequency as a function of time, the falling frequency com-
ponent being the source of the name.
(d) (Optional) Consider the form of the signal in the Brillouin precursor. Show
that it consists of a modulated waveform of frequency w0 = wJJ/4 whose en-
velope is the Airy integral. This then evolves into a signal beating with the
two frequencies of part c.
7.26 A charged particle (charge Ze) moves at constant veloc1ty v through a meamm
· ·" hy a dielectric function E( o. w )/Eo or eauivalcntlv. bv a conductivity func-
. "J . '5 ~ ... , . -' '"

prox1mauon Lnar Lne ""'u " <w.; ""IS""'" IS' u1 '"" " uu~
current flow obeys Ohm sTaw, J(q, w) u(q, wJE(q, w).
p(q, w) ,~ ;-1 ll(w q · v)
,~, r
o(o. w)
1.-.<s ner unit time can
,mr 7 2o 2 ( rf'n r~ I
I 1 aw w lm 1o( w q. v)
-= f7T
3
'J
2
'V . C\ 'f' ~,
• 'h"' r~r -'- ,,\1 1 i< -' ~ t.-. '1.-.« "nrl

~ '
"'IS ~J ,
7.27 The angular momentum of a distribution of electromagnetic fields in varnn.

given by
.
1 r
2 J d x X X (E X B)
3
L -
Jl.oC
wnere tne integration is over all space.

(a) For fields produced a finite time in the past (and so localized to a finite region
u1 >pace J snow tuat, proviaea me magneuc neJa 1s eummatea m tavor of the
vector potential A, the angular momentum can be written in the form
~
_:'_
L =
J.l.oC
Jd x 3
E x A + L
j=1
Ei(x X V)Ai
The first term is sometimes identified with the "spin" of the photon and the
second with its "orbital'' angular momentum because of the presence of th"
angular momentum operator Lop = -i(x X V).
(b) Consider an expansion of the vector potential in the radiation gauge in terms
of olane waves:
A(x, t) = ). ( {Pk r<,(k)aJk)e'k·x-iwt + c.c.l

A J ~L1T)
The polarization vectors EA(k) are conveniently chosen as the positive and
negative helicity vectors E± - (11V2)( E1 ± iE 2) where E1 and E2 are real
orthogonal vectors in the plane whose positive normal is in the direction of k.
<'L •L • · • • •
""~" cauc en• llUJI; dVCJ<l]';C lJi lilt; 111M ICUU Ui L ~<til UC Wflllen as
3
T 2 { dk Ml {1,\ 12 I {J,\ 121
spin -v·JJ
Jl.oC J (27T)' 'Ll +v·JJ I J
Can the term "snin" "nanhr hP • from thi' ?

I -•.
Calcu ate the energy of the field in terms of the plane wave expanswn of A
and compare.
7.28 A circularly polarized plane wave moving in the z direction has a finite extent in
the x and y directions. Assuming that the amplitude modulation is slowly varying
l the wave IS many wavelengths broad), show that the electric and magnetic fields
are given approximately by
. a£0 ,
a£0 ± 1
E(x, y, z, t) = [ E 0 (x, y)(e 1 ± .
ze 2) + ki 1- - e,
1
eikz-iwt
ax ay
B = +i"V;E E
d .
" " "L' -j

.. , V~~JU" '" <H~ A 0 Y, ~ UH
7.29 For the circularly polarized wave of Problem 7.28 with E 0 (x, y) a real function of x
,,-l nl ,], '' ~
• JO ·c '~u" r 'VJ _, l lU UJC
direction of propagation. Show that the ratio of this component of angular mo-
mentum to the energy of the wave in vacuum is
L,
- = +w-1
u
Interpret this result in terms of quanta of radiation (photons). Show that for a
cylmancauy symmetnc, finite plane wave, the transverse components of angular
momentum vanish.
Ch. 7 Problems 351
7.30 Starting with the expression for the total energy of an arbitrary superposition of
plane electromagnetic waves (7.8, 7.11) in otherwise empty space, show that the
total number of photons (defined for each plane wave of wave vector k and polar-
ization e as its energy divided by lick) is given by the double integral
_ <o
N - 4.,?/ic
J J
3
d x
3
d x
,
2
[E(x, t) • E(x', t) + c B(x, t). B(x',
lx - x' 12
t)]
~ ~
LtlAY 1 bK <'S
Waveguides, Resonant Cavities,

and Optical Fibers
Electromagnetic fields in the presence of metallic boundaries form a practical

. . . . ., . .
"'1'""'' OI tne SUD]eCt OI ,dUll' -r tldlllOC. r u 111~11 uc• '"v'v "'"
. ltho or~ Af th~ A ·-ipr nf mP1 P.T< ()T lr'<S the 0nlv nractical WaV Of !!ener-
·o
ating and transmitting electromagnetic radiation involves metallic structures with
dimensions comparable to the wavelengths mvolved. At much higher ~tmrareaJ
c. . _,. .• . . "'1. ' . " . • ~. ,1 ·~~~" . . '".
" ., ' "'- -r ·r
dustrv. In this chanter we consider first the fields in the neighborhood of a con-
ductor and discuss their penetration into the surface and the accompanying re-
ststlve losses. Tlfen Tfle problems or waves guiaeo in -rn:mow wiX,'1''" "."u
nf rMAn•mJ ~o,;,;o<• "r~ -" frnm " hirlv 11 vi,
illustrations included along the way. Attenuation Ill waveguides and Q
values of cavities arc discussed from two different points of view. The earth-
. .,. s; as a 11vvco• ,, " -.avny .
. ., _, 11'-Al. ~· "' .
mnlti;,nnP "n.-1 <inalP-mnrl" nrnna<>ation in ontical fibers. The normal mode
expansion for an arbitrary field in a waveguide is presented and applied to the
helds generated by a locattzed source, wtth oner mcnuon or tne use 01 we nor-
mal mode expansion in the treatment of obstacles in waveguides by variational
methods.
8.1 Fields at the Suiface of and Within a Conductor

In Section 5.18 the concept of skin depth and effective surface current was intra-
duced by a simple example of a planar interface between conductor and vacuum,
with a spatially uniform, time-varying magnetic field at the interface. Here we
generalize the circumstances, at least conceptually, even though the mathematics
is much the same.
.
riTSl
. . .
:r a SUl'Ilil.:e' Wllll Ullll
. ' ' "
11 Ull C\-lvU VUl' V<llU w ¥ 'J
"
u ''
...
,-nnn""'"r nn nn~ oin~ ;.,tn " nnnr.nnrluctin<> medium on the other side. Then,
just as in the static case, there is no electric field inside the conductors. The
charges mstde a perfect conductor are assumed to be so moouc tnat Liley mov~
· · -'- • •1- "',1rl. ..H. hmu ""~;,1 " ,cl olw"VS
~J ' ~y,
produce the correct surface-charge density~ (capital~ is used to avoid confusion

with the conductivity u):
n.n-2: (8.1)
.. ,.....
J J. .
Sect. 8.1 Fields at the Surface of and Within a Conductor 353
to give zero electric field inside the perfect conductor. Similarly. for time-varying
magnetic fields, the surface charges move in response to the tangential magnetic
field to produce always the correct surface current K:
n x H = K (8.2)
to have zero magnetic field inside the perfect conductor. The other two boundary
conditions are on normal Band tangential E:
n · (B -Be) = 0 (8.3)
n x (E- Ec) = 0
where the subscript c refers to the conductor. From these boundary conditions
we see that just outside the surface of a perfect conductor only normal E and
tangential H fields can exist, and that the fields drop abruptly to zero inside the
perfect conductor. This behavior is indicated schematically in Fig. 8.1.
The fields in the neighborhood of the surface of a good, but not perfect,
conductor must behave approximately the same as for a perfect conductor. In
Section 5.18 we saw that inside a conductor the fields are attenuated cxponcn-
(a) (b)
Figure 8.1 Fields near the surface of a perfect conductor.

354 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
is much more rapid than the variations parallel to the surface. This means that
we can sately neglect all denvatJves with respect to coordmatcs parallel to the
surface compared to the normal derivative.
n mere exists a tangcmiat n omslUe me surtace, boundary conditiOn (8.4)
11
implies the same H inside the surface. With the neglect of the displacement
. 11
.
current ut Llle , tue veu cun eyuauu11'
1
l'.,e V X tle
(f
(8.5)
He= - - V X Ee
J-LeW
where a harmonic variation e-iwr has been assumed. If n is the unit normal out-
ward from the conductor and I; is the normal coordinate inward into the con-
ductor, then the gradient operator can be written
V = -n u
ag
neglecting the other derivatives when operating on the fields within the conduc-
tor. With this annroximation the Maxwell curl ion< {R 5\ ' w.
1?-,, 1" v iJHc

IT ag (8.6)
i iJEc
He
J-LeW
II A ag
These can be combined to yield
a2 2i
ae (n X He) + i? (n X He) = 0
(R 7)
n•Hc=O
where i5 IS the skm depth dehned preVIously:
'") \ 112
8= -- (8.8)
J-LeWIT
The second equation in (8.7) shows that inside the conductor His parallel to the
surface, consistent with our boundary conditions. The solution for He is
H, = Hlle -<'"e'<'" (8.9)
--' u · .~
"II
. '1
. h ''" .·' " .•L" ~
10 c\
\' ·V)
•L
electric field in the conductor is approximately
E - 'J J-LcW (1 -
LIT
i)(n X H,,),-iCI8,ii;/8 (R 10)
ThPcr than the el~~tfi~ fi~kl f"llTfh~fffi()rP thPv <hnm th~t fnr ~ annrJ r'nn-
Sect. 8.1 Fields at the Surface of and Within a Conductor 355
t E
H
Figure 8.2 Fields near the surface of a good, but not perfect, conductor. For ~ > 0, the
dashed curves show the envelope of the damped oscillations of He (8.9).
ductor, the fields in the conductor are parallel to the surface* and propagate
normal to it, with magnitudes that depend only on the tangential magnetic field
H that extsts JUSt outstde the surface.
1
From the boundary condition on tangential E (8.3) we find that just outside
the surface there exists a small tangential electric field given by (8.10), evaluated
at g = 0:
(8.11)
In this approximation there is also a small normal component of B just outside

the surface. This can be obtained from Faraday's law of induction and gives Bj_
of the same order of magmtudc as E I'he amphtudes of the helds both mstde
1
••
and outside the conductor arc indicated schematically in Fig. 8.2.

The existence of a small tangential component of E outside the s1ufacc, in
addition to the normal E and tangential H, means that there is a power flow into
the conductor. The time-averaged power absorbed per unit area is
(8.12)
*From the continuity of the tangential component of Hand the equation connecting E to V X H on
either side of the surface, one can show that there exists in the conductor a small normal component
of electric field, E,. · n = (iwE!a)E", but this is of the next order in small quantities compared with
(X.lO). Nole that Ol:lf discussion here presnppo~es ~ Jangential component of H In situations in which
the lowest order approximation is essentially electrostatic, the present treatment is inapplicable. Dif-
ferent approximations must be employed. See T. H. Boyer, Phys. Rev. A9, 68 (1974).
~·· _, . . . . . .
' " ' ' ''-''U" llllt:aprnauuH a, OHmiC lOSses Ill we ooay of
'-'«H "'"' 5""'" u -r
.•.'" ..
the conductor. According to Ohm's law, there exists a current density J ncar the
..... -VH ___ W••
J = uEc = ~ (1 - i)(n x H 11
)e-W-i)!a (8.1:ii
u
The time-avera2.cd rate of dissi ,,· of . · ner nnit volume in · L . 1.

1 T '17• /1/"1_\ITI2 ,,
>0 zu ~ \'' ~v J luI ' C<U VVHll<:Oll Jll v-~U7 )· " ' " ' lll ~ v• \J-~u7 1 m z 1eaas
directly to (8.12).
ThP C'llr~ T ;o n~~h ,..I '' ,11 •• 'd •• ' o± L ,1 d
' . J "'"
surface of the conduct;r that it is equivalent to an effective surface current Keff:
roo
Keff = JJ [)
d{; = n X H 11 (8.14)
Comparison with (8.2) shows that a good conductor behaves effectively like a
perfect conductor, with the idealized surface current replaced by an equivalent
,f, >t ..V,t. ', .11 • ...1: .:t. rl +L ,L .. 11 '- ,.. '
0 -~- ,_c ~ -•J """~"' UUC HWC<O,
thickness at the surface. The power loss can be written in terms of the effective
surface current:
dP]o" __1_ K 12
.J - ,., _
0 1
eff (8.15)
This shows that 11 uo plays the role of a surface resistance of the conductor.*
Fm1~tion {R 1'i) with K . 0..,;,,.." hu
.0
(S/.1il\ Ar (SI. 1")\ .,;11 ~llm,; "' t
' ' •
,1 "'''
approximately the resistive losses for practical cavities, transmission lines, and
WaVe!?.UideS. nrovided We h~VP ' 'for thP fiPkk in thr idP•"Ji7Pd ~rnhlPm ~f
. ,<; :. rl •
"J.
8.2 Cylindrical Cavities and Waveguides

A practical situation of great importance is the propagation or excitation of elec-
tromagnetic waves in hollow metallic cylinders. If the cylinder has end surfaces,
it is called a cavity; otherwise, a waveguide. In our discussion of this problem the
boundary surfaces are assumed to be perfect conductors. The losses occurring in
practice can be accounted for adeouatclv by the methods of<: · 8.1 A rvlin-
..... . '1 .f. (' -·
b' · - · - • ~l
. .'
>Ul
. - '· . '"''
'5•
0
l'Vl " l J l '- >0 o>HV 'H m ' U,J,
plicity, the cross-sectional size and shape arc assumed constant along the cylinder
!lYi< Wi1h ~ cin.us.oidaLti~P ~
-'
.,
orlPnC'P n i<vt fA• thP hPJ.d.o.' '.d _tl
-J
-''
,
the Maxwell equations take the form
'{i'vl< ; ,,R U.R (\
(8.16)
v X B = -ifLEWE V· E = 0
"lhe coethcient ot proportiOnality linking E and Kerr is called the surface impedance Z,. For a good
1
conductor (8.11) vields Z (1 iVa-8 but the con cent of '"'to. · mce or · d" h"'' ,;,;rler
"nnl;r" hiliho
Sect. 8.2 Cylindrical Cavities and Waveguides 357
Figure 8.3 Hollow, cylindrical waveguide of arbitrary cross-sectional shape.
where it is assumed that the cylinder is filled
Because of the cylindrical geometry it is useful to single out the spatial vari-
Appropriate linear combinations can be formed to give traveling or standing

waves in the z direction. The wave number k is, at present, an unknown param-
eter that may be real or complex. With this assumed z dependence of the fields
the wave equation reduces to the two-dimensional form
[V't + (f.LEW
2
- e)]{:} = 0 (8.19)
where v; is the transverse part of the Laplacian operator:

(8.20)
to the z axis:
E = Ez + E, (8.21)
where
Ez = zEZ (8.22)
E, = (z X E) X z
and zis a unit vector in the z direction. Similar definitions hold for the magnetic
field B. The Maxwell equations (8.16) can be written out in terms of transverse
(8.24)
It is evident from the first equations in (8.23) and (8.24) that if Ez and Bz are
dependence is given by (8.18). Explicitly, assuming propagation in the positive z

direction and the nonvanishing of at least one of Ez and Bz, the transverse fields
are
(8.26a)
and
For waves in the opposite direction, change the sign of k.

Before considerin the kinds of field that can exist inside a hollow c
we take note of a degenerate or special type of solution, called the transverse
There are
given by the infinite-medium value,
k = k0 = w f.LE (8.27)
as can be seen from (8.19). The second consequence is that the magnetic field,
deduced from the first e nation in 8.24 IS
TEM ~ TEM
are structures for which this is the dominant mode. (Sec Problems 8.1 and 8.2.)
An im ortant ro ert of the TEM m 1
The wave number (8.27) is real for all w. This is not true for the modes occurring
e
Sect. 8.3 Waveguides 359
continuous across the boundary. Thus, for a perfectly conducting cylinder the
boundary conditions are
n X E = 0, n·B = 0
where n is a unit normal at the surfaceS. It is evident that the boundary condition
~n P io
z
r:;l -n (Q 10\
From the component of the first equation in (8.24) parallel ton it can be inferred
tnat tne corrcsponamg Dounaary conamon on li, IS
aliz =0 (8.30)
an 5
where a/an is the normal derivative at a point on the surface. The two-dimensional
wave equations (8.19) for Ez and Bn together with the boundary conditions (8.29)
and (8.30), specify eigenvalue problems of the usual sort. For a given frequency
w, only ccrtam values ot wave number k can occur ltyptcal wavegmde sttuatwnJ,
f, · /, , ], .t · , ],M • " (~_,~;~"I ~ouitu oitH_
' o· ' _ \·or ·o
ation). Since the boundary conditions onE, and Bz are different, the eigenvalues
will in -1 ho ;/it"t"o •' ThP fif'ld< 1hll< n•<111r~llu .-liui.-1~ thpmeoluoe intn t\xrn
. . u . ''
dtstmct categones:
TRANSVERSE MAGNETIC (TM) WA YES
n r. _,_ . L ~,. ,-,r"
'J ' z IS
P n ' · hnnn.-brv rnn.-litinn aBz 11

- an s
The designations "electric (or E) waves" and "magnetic (or H) waves" are some-
times used instead of TM and TE waves, respectively, corresponding to a spec-
thcatlon ot the axtal component ot the news. The vanous TM and Tb waves,
,], tJ. 'rPl\A" :f ;, · •' .•1i1 .. 10 o loto eot ~f holrle t~ ' .,
:n arbitrary electromagnetic dist~rbance in a waveguide or cavity.
o.J n·aveguwes
ror me propaganon or waves mswe a noHow wavegmoe 01 unuorm cross secuon,
it i< +. -' frnm (R ?(';~ h \ th~t thP tr: m~"n..tir ~nd P1Pr1rir fiPl.-1< for hoth
UV< dllU u::. WdVt" dlC 1' uy
(8.31)
where Z is called the wave impedance and is given by

k- k & ('rl\1[\
EW k0 -y E
(8.32)
z = '"' k-. r;;
k- k l;; (TE)
where k 0 is given by (8.27). The plus (minus) sign in (8.31) goes with z depen-
~ a . . .
, e ~e )· 1ue uausverse ue10s are ueter. uy tue -o a1news,
according to (8.26):
TMWAVES
ik
E,= :+: y V,I/J
TE WAVES
..
H=
t
+ fl<
- 2
vt 1/1 (8.33)
I'
where 1/fe"''kz is Ez(Hz) for TM (TE) waves* and y 2 is defined below. The scalar
function r/J satisfies the two-dimensional wave equation (8.19),
(V; + y 2 ) 1/1 = 0 (8.34)
where
/'2 = J-tEW2 - k2 (8.35)
, ..
.L •
Jj' -· <L L -'
'J
o.l iJI/f /o ~"

'I' IS " " \U,JV)
iJn s
£ . ~-· ~~~\
\ ry
Equation (8.34) for 1/J, together with boundary condition (8.36), specifies an
p; >lnP nrnhlPm Tt ;< P~n~l;uP
Roughly speaking, it is because 1/J must be oscillatory to satisfy boundary condi-

tion T8.36J on opposite sides of the cylinder. There will be a spectrum of eigen-
values y~ and corresponding solutions 1/JA, A = 1, 2, 3, ... , which form an or-
tnogonal set. 1 nese amerent solutions are called tne moaes of tne gwae. to or a
given frequency w, the wave number k is determined for each value of A:
k~ = J-tE w2 - I'A
2
(8.37)
'"
--ncW'\0'
. ~
cauuUJJ"' "'1 'ICY WA,
/'A
~0
t:JJA
v;;, . .)0)
then the wave number can be written:

kA- V/IEVw - (J)A ( I-\.3LJ)
We note that for w > w the wave number k is real· of the A " can
<L ,_, .,.,; . £ ,,
t" ~!" :.b • ''-''i '
""-'"
d.
UWH <Hv
.££ £.
"'"-'"!' ' <0 ~b
'"-A
inarv; such modes cannot propagate and are called cutoff modes or evanescent
modes. The behavior of the axial wave number as a function of frequency is
shown qualitatively in Fig. 8.4. We see that at any given frequency only a finite
number of modes can propagate. It is often convenient to choose the dimensions
*We have changed from E and B toE and Has our basic fields to eliminate factors of p, when using
the wave imnedances. ILike ordinarv im ·-'· wove. . 'vnltooe onn ·..rrenl ond .L
··~""""~
Sect. 8.4 Modes in a Rectangular Waveguide 361
---------------~-~-~-~~
Figure 8.4 Wave number k, versus
of the guide so that at the operating frequency only the lowest mode can occur.
wave JJ-EW,
wavelength in the guide is always greater than the free-space wavelength. In turn,
the phase velocity uP is larger than the infinite space value:
v
p k,
=-=--
~ Jl-c~r ~
>--
8.4 Modes in a Rectangular Waveguide

As an important illustration of the we
consider the propagation of TE waves in a rectangular waveguide with inner
dimensions a, b, as shown in Fig. 8.5. The wave equation for 1/J = Hz is
with boundary conditions ol/f/i:Jn = 0 at x = 0, a andy = 0, b. The solution for 1/J
1/J~n(x,
m7TX) cos
y) = H 0 cos ( -a- (n7ry)
b (8.42)
2
"Ymn =
2 m 2 nb22)
7T ("""";t + (8.43)
X
"T'L _, • .L .,_
•'- '"'
.. "
_,
~ L ,._ .. ~ ~
~Uv H <HUC
U'v -r vU>Hv' '~ ' v y VJ U'v <HV Jc" '' e
integers m, n. For there to be nontrivial solutions, m and n cannot both be zero.

cr" ,,_,H4',
mn b'
),
'J
2 2 1/2
7T m n)
Wtnn = '\~ ,2 + h2 (8.44)
' \
If a > b, the lowest cutoff frequency, that of the dominant TE mode, occurs for
m- 1 n - 0:
' 7T
(8.4 'i\
''" V J.L€ a
This corresponds to half of a free-space wavelength across the guide. The explicit
neJUs tor tms moue, ueno1eu oy 1.c1 _0 , are:
\
llz n 0 cos
1TX
a
,e -
I
ika T-T .;n I
1TX ikz-;wt
T-l
'A
-
·u
\
0 (51. ,1{,\
\ .,
7T a
I
. waJ.L
. 1TX eikz-iwt
Ey = z - - H 0 sm(
. ~
where k = k 1 ,0 is given by (8.39) with w, = w 1 _0 . The presence of a factor i in fix

(and Ey) means that there is a spatial (or temporal) phase difference of 90° be-
tween H, (and Ey) and Hz in the propagation direction. It happens that the TE 1•0
mode has the lowest cutoff freauencv of both TE and TM modes * and so is the
one used in most practical situations. For a typical choice a = 2b the ratio of
cutoff frequencies w for the next few modes to w, n are as follows:
~n 0 1 2 J
m~
0 2.00 4.00 6.00
1 J.UU L.L'+ '+. l.J
2 2.00 2.84 4.48

~
-.
~ {l{l
-. ~ {,1 ' {l{l
4 4.00 4.48 5.66

5 5.00 5.39
F. "'nn
There is a frequency range from cutoff to twice cutoff or to (alb) times cutoff,
. . . . . _._..... -
Vvl 1~ , Wll\CUO lllC U_Cl,() lll(JUC 1~ 111\C UlllJ jJl' ·r--
yond that frequency other modes rapidly begin to enter. The field configurations
*This is evident if we note that for the TM modes E, is of the form

r _( m-rrx \ . _( nny \
r
., ·o '\ h I
'\ a I
while y 2 is still given by (8.43). The lowest mode has m ~ n ~ 1. Its cutoff frequency is greater than
th"t of the TE mncle hv I he f"rtnr 11 + a2fh2) liZ
Sect. 8.5 Energy Flow and Attenuation in Waveguides 363
of the TE 1,0 mode and other modes are shown in many books, for example,
Amerzcan mstztute or rnyszcs 11anaoooK Lea. u. to. uray, -'ra emu on, !Vtcuraw-
Hill, New York (1972), p. 5-54].
I).;J
.... '(;Y
......
riUW UIIU
• . .
!lUll 111
TH
rr u" "'
...
._, .£ ..L '
'T'L
~Hv ICU
U '1 " ' V< Uvv<>VH 0,., £<V<
.£ " .
vJHHUHvCU
.1: "- ' .1
U,J V< U<VHlO<J U HU' U
cross-sectional shape can be extended to include the flow of energy along the
'.-L rl tl, t+. .+ tJ.. rl 1. . . tJ.. ,11, J., <.: •
'b uucv
conductivity. The treatment is restricted to one mode at a time; degenerate modes

orP mc>ntir.nc>rl r.nlu The> flr.m r.f PnPrnu io rlc>oC'rihPrl hu the> C'r.m·>l.
. fl.
Poynting vector: --
-- ·oo
' J
S - ~(1!: X H") ( ~.47)

· . real nart !lives the time- nux of ener!'v. For the two tvnes of fielrl
""
we find, using (8.31) and (8.33):
-
2
€ z 1Vri/JI + i : 1/J V,I/J*
"
wk
2·/
JL z IVri/JI 2
- i r 1/J* Vri/J
(Q AO\
\' . J!
where the upper (lower) line is for TM (TE) modes. Since 1/J is generally real,*
~ .
~~" ~"" lll<ll tll~h::all~V~'~" <.Vll'l l l :~; n-~~· ''-~v~,t~
lor>s not · 1te to lP. time-~ven"'"' lnx · r>nr>r<>v 1 vllt';;;; h':u
"'otl 1~ ~~u
1~n, 1e
axial component of S gives the time-averaged flow of energy along the guide. To
evaluate the total power flow P we integrate the axial component of S over the
cross-sectional area A:
r
P = J S • z da = 2·l
r
A
wne
' €
JL
L (V,I/J)* • (Vri/J) da (8.49)
By means of Green's first identity (1.34) applied to two dimensions, (8.49) can
be written:
wk
P=-
€
-~ If;* a!J! dl - ( If;* 'i/~1./J da (8.50)
L')l JL JC un JA
where the first inte2.ral is around the curve C which defines the boundarv surface
'T't.. . • • I.. rl
V<. £ ...
cHv VJ.1: rl • ~<HJ H>O .1
U O<V< • ... .'JlNJ L'.£<:
L'VH<
£. '... .• 1rll-
Hv<U .£ 1.
'J V
conditions (8.36). By means of the wave equation (8.34) the second integral may
hP rC'rlllC'C'rl tn thP nnrmoli7otinn intC'{JroJ v
fr.r ,J,, (', 1110nthr 1hP 1r~n·
' J
power IS
7 112
l w
p
LV JL€ ,wA
I 1 w2
WA
J
I
€
-~
1/J* 1/J da (8.51) I
. . . . . . . .
'H lS f'U>SWlC lU CAClCC d gU1ue 1H >UCH d < lHdl d !;HCH lHUUC Ul HHCdl LUlHUUWUom oH <HUULO
has a complex 1/J. Then a time-averaged transverse energy flow can occur. Since it is a circulatory
flow however it really represents only stored energy and is not of great practical importance.
where the upper (lower) line is for TM (TE) modes, and we have exhibited all
the frequency dependence explicitly.
It is straightforward to calculate the field energy per unit length of the guide
in the same~ flow Th~ r~sult is
U=!(-~J{E}I
2 w" 1-L A
1;*1Jda (8.52)
r'Amnor;oAn it} H ll D •. 1- <1- D rl TT .. ' ·>

~·
, r - •·~y~• llt
constant of proportionality has the dimensions of velocity (velocity of energy
flow) and is just the group velocity:
_n
'- ' '
:..._ = '"' ' = ----=----
~,
1 -2 = Vg (8.53)
U W /-LE y;;k 'V w
as can be verified by a direct calculation of vK = dw!dk from (8.39), assuming

that the dielectric filling the ~>nicle is ·ive. We note thllt v . is~· lc>oo
than the velocity of waves in an infinite medium and falls to zero oat cutoff. The
product of phase velocity (8.40) and group velocity is constant:
1
(R <;A\
""r o
/-LE
an immediate consequence of the fact that w ilw oc k ilk.
Our considerations so far have applied to waveguides with perfectly con-
ductimr walls. The axial wavP. numhP.r k-. "'"~ PithPr rP!ll or nurP.lv · '"";nor)' Tf
the walls have a finite conductivity, there will be ohmic lossc~ and the power flow
along the guide will be attenuated. For walls with large conductivity the wave
number will have small additional real and imaginary parts:
'(n\
K,\ KA -r- a" -r- zp, ~ 1\ ••)))
where k\0 ) is the value for perfectly conducting walls. The change a, in the real
"t ,f H •L . •" • ,., • '
r 5' 'J •u•n v~vvp• avu• v• lLVH VVIHOH
kJ,. u. 1ne anenuanon constant {3" can be found etther by solvmg the bound-
--c>
ary-value problem over again with boundary conditions appropriate for finite
conductivity, or by calculating the ohmic losses by the methods of Section 8.1
~nd usinP . of PnPr<nr WP mill fir~t 11~(' thP httPr tPrhn;n .. ~ Th~
p~ •v• "
"~
.1
~
±1..
•uv ou•~
."
VVHl LJCO
'
to' VCOH UJ
'
P(z) = Pae-2f3Az (8.56)

Thu~ thP ,.tt ·<1inn ;<
· er
..,;.,~n h.,
·.;
1 dP
{3,\ = --- (8.57)
2P dz
_.1 DU . . c±J, . ,,
"'' ·~ •• Jc' ~ n-
~·
H<
• '1
1''-'1 UHH
"_i
"
Vl ll1CO
Accorctmg to tne results ot Section 8.1, this power loss is

1
dP
dz
J In "
2aB Jc
Hl 2 rll (\1.
\'
<:11.\
Sect. 8.5 Energy Flow and Attenuation in Waveguides 365
where the inte~>ral is around the boundarv of the ~>uidc. With fields (8.31) and
{Q 'J'J\
\ • J J
.
"J
• ,1. ... +.
b'
_, .
2
--
1 al/f
-
~: = 2~8 (:J fc
2
p}w~ an
dl (8.59)
- 1 wz' ' wz
2
- -
1 - --1 In x V,I/JI 2 + --11 1/11
UEN~ w w
where agam the upper (lower) Jme appltes to TM (lt) moues.
Since the transverse derivatives of 1/J are determined entirely by the size and
shape of the waveguide, the frequency dependence of the power loss is explicitly
exhibited in (8.59). In fact, the integrals in (8.59) may be simply estimated from
the fact that for each mode:
( V, + JLEWATiJl -u C1S.OU)
This means that, in some average sense, and barring exceptional circumstances,
the transverse derivatives of 1/J must be of the order of magnitude of vr;;kw"l/f:
( al/fan ) ~ (In x V,l/!1 ~ JLEW~( II/JI

2
/
2
/ (8.61)
r, HPntlu
-o' T
thP l;nP ; -o >le ;,-, (Q
\
<;Q\ ~~n hP rPhtPn tn thP. ,]i7~tion
2
integral of I1/11 over the area. For example,
2
fc ~~
al/f
an
dl = g"JLE C
A
J I1/11
A
2
da (8.62)
where C ts the ctrcumterence andA tstffe area or cross section, wniJe ~" is a
ilim ,],~~ nnmbPr of the order of unitv. Without further knowled~>e of the
shape of the guide we can obtain the order of magnitude of the attenuation
constant {3" and exhibit completely its frequency dependence. Thus, using (8.59)
with (8.62) and (8.51), plus the frequency dependence of the skin depth (8.8), we
find
"
w
2
'E 1 ' L ."'A;
I
WA
w~rz
{3- -- g)o, + 1JA( (8.63)
" - \j JL a-8" 2A I ( w
1--
wz
where u 1s the conductlvtty ( assumea u1cfepellifent ot trequency ), i'>" 1s the sJon
~
...
"t thP rntnff frP()llPnr\1 "n£1 J:. 'YL "rP dimensionleSS numberS Of the order
'J.
1:), 'T'l\"
" • , 'lA
()
For a given cross-sectional geometry it is a straightforward matter to calcu-

late the dimensionless parameters g" and 'Y/A in (8.63). For the TE modes with
n = 0 in a rectangular guide, the values are gm,o = a/(a + b) and YJm,o =
'11.1/ 1.\ r:'. .l..l. .1 ' c~; .·
, ,
ol. ·n ~f ~r.-lPr
~V'\U v ,. , ~,
. '
umty, as cxpccteo.
The ~>eneral behavior of B, as a function of frcouency is shown in Fig. 8.o.
Minimum attenuation occurs at a frequency well above cutoff. For TE modes
the relative magnitudes of g" and YJA depend on the shape of the guide and on A.
1--
t
f3, 1-- :\ -------:--
i' 1
~ -:;:::::::=
TE
.
. .,. '"'. "tt, or' ~~netont a•A
1- i I
I
as a funct1on of frequency for
I typical TE and TM modes. For TM
I
I
v~
' I I I modes the minimum attenuation
0 I 2 3 ~ § Ul,;(.;Ul~ at WIWJ.. -vi3, IC::~tudkM uf
w/w~~ cross-sectional shape.
Comeqncntly no general statement can he made about the exact frequency for
minimum attenuation. But for TM modes the minimum always occurs at Wmin =
\13w,. At high frequencies the attenuation increases as w112 • In the microwave
r cgion typical attenuation con~tants for copper guides are of the order
{3, ~ 10- 4 w,lc, giving lie distances of 200-400 meters.
The approximations employed in obtaining (8.63) break down close to cutoff.
Evidence for this is the physically impossible, infinite value of (8.63) at w = w"
8.6 Perturbation of Boundary Conditions

The usc of energy causer vation to determine the attenuation constant {3, is dir eet
and has intuitive appeal, but gives physically meaningless results at cutoff and
fails to yield a value for a," the change in the real part of the wave number. Both
these defects can be remedied by use of the technique called perturbation of
boundary conditions. This method is capable, at least in principle, of obtaining
answers to any deslfed degree of accuracy, although we shall apply n only to the
lowest order. It also permits the treatment of attenuation for degenerate modes,
mentioned briefly at the end of this section and in Problem 8.13. The effect of
small distortions of cross section can also be treated See Problem 8 12
For definiteness we consider a single TM mode with no other mode (TE or
TM) degenerate or nearly degenerate with it. The argument for an isolated TE
mode is similar. To reduce the number of sub- and superscripts, we denote the
(unperturbed) solution for perfectly conducting walls by a subscnpt zero and the
(per tm bed) solution for ., ails of finite conductivity by no sub or superscript.
Thus the unperturbed problem has a longitudinal electric field E, = 1/10 , where
0, 0 (8 64)
and y6 is real. For finite, but large, conductivity, Ez = 1/J is not zero on the walls,
but is given by (8.11). To lowest order, the right-hand side of (8.11) is approxi-
mated by the unperturbed fields. By use of the first equation in (8.23) and (8.33),
the perturbed boundary condi1ion on •/' can be expressed as
!{!Is (8.65)
Sect. 8.6 Perturbation of Boundary Conditions 367
where the small complex parameter f is*
(8.66)
Here fL, and fL arc the magnetic permeabilities of the conducting walls and the
medium in the guide, respectively, ois the skin depth (8.8), and w 0 is the cutoff
frequency of the unperturbed mode. The perturbed problem, equivalent to
(V? + --/) 1/1 = 0, (8.67)
If only the eigenvalue ·lis desired, Green's theorem (1.35) in two dimensions
can be employed:
( [cr
lA
v;,,u- ,p1~cf!l rta - ~c l\u a<t>
an
- ¢ iii/J dt
an
J
where the right-hand side has an inwardly directed normal [out of the conductor,
in conformity with (8.11) and (8.65)]. With the identifications, •/! = 1/J and 4> =
¢'/';, and use of the wave eqnations (8.64) and (8.67), and their boundary eondi-
lions, the statement of Green's theorem becomes
(8.68)
Since f is assumed to he a small parameter, it is normally consistent to approx-

imate 1/J in the integral on the left by its unperturbed value I/J0 . This leads to the
final result,
(8.69)
From (8.51) and (8.59) of the preceding section one finds that the ratio of integrals
on the right-hand side of (8.69) enters a previous result, namely,
(8 70)
t 1/10 da
where [3< 0 ! is defined by (8.57) and (8.63). This means that (8.69) can be written
as
e = k<OJZ + 2(1 + i)k(o) f3co) (8.71)
a result that holds for both TM and TE modes, with the appropriate f3co) from
Section 8.5. For pol >> {3( 0 ), (8.71) reduces to the former expressiOn (8.)5) w1th
*~lmc gcnerall)', f ean be exp1essed in terms of the stufaee imJ3edanec Zs as f (itx:J/J:Lw6)h:..

Jm'f -chapter tl Wavel!;uides, Resonant Lav1ties, and Optical Fibers-SI
sensiote resuns oecausc me commnanon K .. , Jj "'IS nmte and well behaved m thf'.
-o r
IS LV nm a snarp um; ir we waus areTesslllan perrect conductors
" " ,. " J 0 '7 ..
HCOL[' lll<H lllllCO COli V' CUlUll. " 111 a SllaiV
The discussion of attenuation here and intl1P · '" ~Prtinn i~ rP~1rirtPr1
" ' VH'-'
_,
a ~ m ~ gu;al
UH1'-'·
~
v1
~
. llllS ""'-'
.
aoorox1mat1on IS adeauatc. lt however It har ens that a TM and a TE mnriP
.-1 •+ (. +• ,] •~ £ C\ ~ A\ ''
-o \' 'b ' v h "''-'11
any perturbatiOn, no matter now small, can cause sizaoTe mixmg of the two
Ul
we pr . memuu occurs in me penurueu uounoary condition l1i.D:J ), wncre

~ .~. ~ .
we •1 llUTLH.:u nz, as weu as 111e ue11vauve~z· Ana were IS, ot
COUrSC a COrrcsnonnino nPrtllrhf'.n hounnl'lrV itinn fnr T-T inunh;inn hnth
--L ~' ~· -' -"-'~ 'T'L -'-' c, -o
uul''-'' w• " v u ' " ' . ~ 11v l'' UH<O Ul I dl<O-Mdl<O jJL:llUI-
bat10n thcorv. most familiar m the context of auantum mechanics. The nertnrhP.-1
~ -11' -L' "" '•~•• '
a v 0 lllVll~ v --,;m;:; ··r- ITIIITJCOU 1 !Vi dllU u::,
modes, ana me attenuatiOn constants tor tne two modes have the charactenstic
n lrn n \ 1
1-' 2 \1-''IM 1-'TE/ 2 v \1-'TM PTE) "+ I"- I ~ 0./ L)
wuere PT.\1 auu PTE arc tnc va1ues tounu auove, ana A is a COUPling parameter.
ThP. . nf l'lttPnlll'ltion l'lnrl ion~ ;-;:;:;- "· ... cinn nPr_
·o -or
lUI . VI uy dl c; dUUl ~ :-.ee mso couin.
LllUll~
o.7 Kesonant Lavltles

Although an electromagnetic cavitY resonator can he of anv shane whatsoever
cytinorical waveguide. we assume mat me eno suffaces are01ane and perpen-

.. -, . . .' . ~. . .
uave uuuute Ivlly, wu11e Lue cavHy IS uuea wnn a 10ss1ess OleJectnc wnn
Lll<O ll<OIUS IS LllaL d]JlJI LU
A sm KZ + jj COSKZ
If the olane boundarv surfaces are at z 0 anrl7 d~rl>1rv
I. 7T f A ~
" l' A \1' v, , L, .•• ) ~ o.1 J 1
l'Vl llVl 111.01U~ lll<:O Vd111Mll11!j Ul L t dl .( U dllU--z ~
~ ' p7rZ
\P u, 1, L, . . . ) \<">.!'+)
10. ~•. 0..'7 D r 0 0
'\t>Q
<'' "
~"""'" 'Y ""
--' £ ""'""
·~
"' - ' ~ , "''"'
d_ .£ TT
--ur-rT <X ~ z
r. .-1
~
--1
"1'
p7TZ)'
I
Hz = 1/J(x, y) sin( (p = 1, 2, 3, ... ) (8.75)

d
Then from (8.31) and (8.33) we find the transverse fields:
TM FTFTDS
I
p7T . t( p7TZ
E, = -d'V2 sm d V,if;
(R 7f.\
H =iEw (p7TZ)
-cos - - Z• X V,if;
t ·l d
TE FIELDS
I
~
lWJ.L p7TZ I • ... '
"-t
I
2 ""I 7 I L "' ' t'l' {0 '"7'"7\
\c. 'J
I \
p7T p7TZ
H, = --'· 2 cos ------:;-- V,if;
l \ I
The boundary conditions at the ends of the cavity are now explicitly satisfied.
There remains the eigenvalue problem (8.34)-(8.36), as before. But now the
constant y is:
;n7T'2
r J.LEW I -;-I ~ 11./11)
-=
For each value of p the eigenvalue -0, determines an eigenfrequcncy wAp:
2
WAp
1
= J..l.€
and the corresponding fields of that resonant mode. The resonance frequencies
[ ( rl /'A
2
+
p7T
d (8.79)
form a discrete set that can be determined graphically on the figure of axial wave
number k versus trequency m a wavegmde (sec !'I g. 11.'1) oy aemanaing mat K
p7T!d. It is usually expedient to choose the various dimensions of the cavity so
that the resonant frequency of operation lies well separated from other resonant
. . . . . . . ..
ueL[' . ~ueu Lne cavny wu1 ucTCiaTIVeJ:Y="' upcolalluu auu "'v"
to perturbing effects associated with frequency drifts, changes in loading, etc.
An i
• tJ..
y
...
1t nr,rtir" 1 rP.~nn,nt
.11.
'• i~ the>. rio-ht circular cvlindcr nerhans
1-.
JJ 'J '6
<-~ 1-.~'~1-.<- ;,h~ c>ul;n.-lc>r ;c chnmn ;., H;n
·o -o ·o·
8.7, with inner radius Rand length d. For a TM mode the transverse wave equa-
tion for w = K subiect to the boundarv condition K = 0 at p = R, has the
c()lllt;,-,.,.
.1.1" -1-.\ - T:;" T i. "\ 0 ±im<P (R Rll\

'r "'', 'I ·m 1
where
,, - --;vntn
R
~i711 Chanter 8 Wave.,uides Resonant Cavitie•· anrl . _.:J
!'-...
R----,
/
-r d
- f.-
- '
'-. / /
X Figure 8.7
x,.. is the nth root of the equation, J m(x) = 0. These roots were given earlier,
following Eq. (3.92). The integers m and n take on the values m = 0, 1, 2, ... ,
oncl n 1 'J "l Th~ ~MAnon~~ f, •n · · •
'1 b JJ
1 J...Lm.n p7r /n n
mnp ~U.<H)
~~Rz d2
The lowest TM mode has m = 0, n = 1, p = 0, and so is designated TM0 , 1_0 .

Its resonance frequency is
2.405
'"'" ,; ,. ,T;J
~·
HlC ll IOAiJl' 1Vl 11110 IIClU' dl <:;
EJz = E 0 J0 e.405p
R )e-;wt
(8.82)
_ .
Hq, - -z ,, ,, £ 0 1 1
I
E ( 2.4Q5p
R
le-;,,,,
r
~·
UlC 1 '-''JUCll\,} 1Vl 111" . 01 u. ~UilSt:l[Ueuuy s1mp1e
tuning is impossible.
. H
"' '
For TE modes, the basic solution (8.80) still applies, but the boundary con-
dition on Hz[(aqdap)IR =OJ makes
-~".
'
'Y.
J(
mhP~P v' ;o thP nth rAAt Af T' (v\

'"""'
n T~
. .
,.. 0
~. ~. nl -~ rl
n, dlC lJClUW '

~lUi fft T .1, A u " a ulVIal fOOl)·
Roots of J ,;.(x) = 0
m = 0: xb.. = 3.832, 7.016, 10.173, ...
m = 1: x;., = 1.841, 5.331, 8.536, 0 ••
'l· v' "l n<;A , n n-,n

~
-"
f. '?()f.
0 ' ...
m 3: x' 4.201 8.015 11 ,,,.,
Sect. 8.8 Power Losses in a Cavity; Q of a Cavity 371
The resonance frequencies are given by

1 x'z 2 2 112
mn P TT )
(8.83)
Wmnp = '\ C R2 + ,12
r '
where m = 0, 1, 2, ... , but n, p = 1, 2, 3, .... The lowest TE mode has m =

. " .
ft . j/ i ' auu 1> iDl.l,l• H> llCOljUCOIIC,J I>
1 ~o\ l/2
LO'H I\ )
(J)lll = < ;-: 1 + 2.912 2 (8.84)
v JLER \ d I
while the fields are derivable from

i 1 QA 1 \
•1•
'f'
Hz H oq
J I R
I cos "P
"- sin d
e i<vt (8.85)
\
by means of (R.77). For d large enough (d > 2.03R), the resonance frequency
w 111 is smaller than that for the lowest TM mode. Then the TE 1 , 1,1 mode is the
tun dam ental oscJllatwn ot the cav1ty. 11ecause the trequency depends on the ratiO
d/R it is possible to provide easy tuning by making the separation of the end
.aces auJUSLaUle.
Variational methods can be exploited to estimate the lowest resonant fre-
• .£ •• • •. , , • ,.
Y' v1 ~av"''-''· " •auauvum P' -r auu ovmv vAampl'-'O G>'-' P" 111
the problems (Problems 8.9-8.11).
8.8 Power Losses in a Cavitv: 0 of a Cavitv

In the preceding section it was found that resonant cavities have discrete fre-
quencies of oscillation with a definite field configuration for each resonance fre-
quency. This implies that, if one were attempting to excite a particular mode of
oscillation in a cavity by some means, no fields of the right sort could be built up
unless the exciting frequency were exactly equal to the chosen resonance fre-
qucncy. ln actual tact there will not be a delta functiOn smgulanty, but rather a
narrow band of frequencies around the eigenfrequency over which appreciable
excitation can occur. An 1mportant source or this smearing out ot tne snarp fre-
quency of oscillation is the dissipation of energy in the cavity walls and perhaps
. . . . . ' " ,,;
111 lllCO LtiC lllCO '-'d V1l J. n. lll<Oa>Ul CO Vl lllCO M!al[- Ul 1' Ul UlC
cavity to external excitation is the Q of the cavity, defined as 2Tr times the ratio
£ ,,_ , , • ,,_ "- ' •••
v1 utv wuv-u··~•usvu ""~'5Y ocv1vu ' " " ' " ~a"'Y LV "'" 00 1voo p~• ~Y~'"·
n Stored energy 111 Rt:;\

~ ·u Power loss '· ·
HPrP ,,, io thP rPorm~n,..P frPmlPn,..u ~oo11mincr nn lnooPo Ru ... nnoPruMinn nf Pn-
ergy the power dissipated in ohm"ic losses is the negative of the time rate of
change of stored energy U. Thus from (8.86) we can write an equation for the
behavior of U as a function of time:
-1TT
- = _ _.2' u
dt Q
• "· , 1. (8.87)
If an m!Ual amount of energy 0 0 1s stored m the cavity, It decays away exponen-

tially with a decay constant inversely proportional to Q. The time dependence
in (8.87) implies that the oscillations of the fields in the cavity are damped as
follows:
E(t) (8.88)
where we have allowed for a shift Llw of the resonant frequency as well as the
damping. A damped oscillation such as this has not a pure frequency, but a
superposition of frequencies around w w0 + ilw. Thus,
E(t) = -1- f~ E(w)e-twt dw

V27T -~
.
where
E(w)
Tht\ integral in (K89) is elementary and leads to a frequency distribution for the
energy in the cavity havmg a resonant hne shape:
1
[E(w)[ 2 oc (8 90)
(w Wo
The resonance shape (8.90), shown in Fig. 8.8, has a full width rat half-maximum
(confusingly called the half-width) equal to w0 /Q. For a constant input voltage,
the energy of oscillation in the cavity as a function of frequency will follow the
resonance curve in the neighborhood of a particular resonant frequency. Thus,
the frequency separation i5w between half-power points determines the width r
and the Q of cavity is
Wo Wo
i5w r
Q values of several hundreds or thousands arc common for microwave cavities.
To determine the Q of a cavity we can calculate the time-averaged energy
stored in it and then determine the power loss in the walls. The computations
arc very similar to those done in Section 8.5 for attenuation in waveguides. We
consider here only the cylindrical cavities of Section 8.7, assuming no dcgcner-
ff\
I
I
1/ r-wo/Q =r
I i\ Figure 8.8 Resonance line shape. The
/I _!_ ~ full width I at ha.l.flliaxirnum {of thP
/ I ~ I _, .
" ~ '/ '1'
-1 •1.
-uy-•••
frequency w0 divided by the Q of the
' '
'
w~
w0 +Awj wo
cavity.
Sert. 8.8 Power Losses in a Cavitv: 0 of a Cavitv 373
acies. The energy stored in the cavity for the mode A, pis, according to (8.74)-
(R.77):
u= ~ {;}l1 + (~A:rJ L~~w da (8.92)
Wuere Llle Upper ~lOWer) dnt.: applit.:~ lU llVl ~ LC) ITIUUC,. >VI LHCO u n HIUUCO' wiw
p - 0 the result must be multiplied by 2.
The power loss can be calculated by a modification of (8.58):
r..
Pro» = 2~8 [fc dl
..•
r dz In X Hl~ides + 2 Lda In X Hl~nds J
.'·
(8.93)
~Vi
CT'~
HH
K
HiVU~O
..]
HW f' T "'l"O
U ~UOJ LV 0
d ..•
Pross= u~M [ 1 + ( ~A:r](1 + gA ~~) L1~1 2 da (8.94)
where the dimensionless number gA

is the same one that appears in (8.62), C is
the circumference of the cavitv. and A is its cross-sectional area. For p = 0, gA
mwf hP ro>nl•>~~rl hn ") /:.
_, c -~. "';,-', r'~A·,,
r'n~h;n;n~
.. L '
_,·o.;·
(Q ()")\
0
onrl
c
{Q
...·
Qd) ~·
c '- • ....-. .c '1-.
· 1a to (R Rl>)
v ' '
-. ~nrl
UolHIS ClVH \ lJ,lJ) LVL LH'-' '"-"' UCOfJLU U 0 '-' UUU H~ "' Vi CH~ ~· '"J•
Q=!!_c!_ 1
(8.95)
gA ~~)
Me 0
2( 1 +
• .,,.,. r.. n
Ul" 'LHCO~
' ' "· r.> ..]
WUCOLCO fJ'-'lHH.O<llJHHJ
f'-'c " lll'-' .u~ ~u ."•J· ~vi ~ u
'
~ o.'J)) must oe mumpnea oy L ana ~A replacea oy L~A· 1 IUS cxpress10n ror td' nas
an intuitive physical interpretation when written in the form:
Q = ;c (~) X (Geometrical factor) (8.96)
T7 ;< thP ""lnrn~ nf th~ <nrhrP ~r"~- Thte n of~

..
.,]., ,. onrl ";t< tnhl
"CVlUCHlly, <l}'<lll 110111 <1 I n•v v•

0
-' ' .
;Uti.Oal T , lllC .£ ••.
UL'-'
' ..]
by the fields to the volume of the conductor into which the fields penetrate be-
cause of the finite conductivity. For the TEr, 1 ,1 mode in the right circular cylinder
cavity, calculation yields a geometrical factor
I .]2
11 + 0.343-=
/'(
d
l + R ~ O.Y I)
3
(1 d + 0.244 R
+ 0.209 R d 3)
that varies from unitv for d/R = 0 to a maximum of 2.13 at d/R = 1.91 and then
~
..]
•ooM tn 1
.A"! oo .1/l) ""
'
.. _,
'
J::,J<(.Jll lUl \.t ap}'lll'" llUL UlllY LU '-'Y
~O.':IUJ ' .._,a lll'-'0 '-UUL UlOV LV ..]_'
V
cavttles ot arbitrary shape, w1th an appropnate geomerncat racwr or we oraer

of unity.
The usc of conservation of energy to discuss losses in a cavity has the same
advantages and disadvantages as for waveguides. The Q values can be calculated,
but possible shifts in frequency lie outside the scope of the method. The technique
of perturbation of boundary conditions, described in Section 8.6, again removes
the>oP ,.jpfi~;, ·~;, T. f, ,, tl,' ,, . ,, t tL 'rl. . •
' "bJ ~ "'" IUVll tlJ<I{
the answers can be deduced without performing the calculation explicitly. The
unperturbed problem of the resonant frequencies of a cavity with perfectly con-
u · o wau' ;, .- · uy ~ 15.04) or irs equivaJCm ror 1 t moaes. ;»Jmllarly, the
perturbed problem involves solution of (8.67) or equivalent. A result equivalent
to (8.69) evidently emerges. The difference (Yo - ·l) is proportional to (w5 - w2 )
where now w 0 1s the unperturbed resonant frequency rather than the cutoff fre-
quency of the waveguide and w is the perturbed resonant frequency. Thus the
analog of (8.69) takes the form
w6 - w2 = (1 + i)I (8.98)
,], T. tJ, ,.; ,f. .L T. +L 1: .c T " ··-
·rr JfJ HvO<U<Oo H< U<v UH<H ~· < v, U<v ~y
pan or w IS l1! Lw0 . J:<rom ( 8.88) this 1s to be identlhed With iw0 !2{2, and there-
fore I= w~/Q. Equation (8.98) can thus be written
(8.99)
""- "1' 'J uy \'-'·""} auu \U·Y~J• \U·7Jj· , -~ 1' MOt:ll LO

0
2
cause equal modifications to the real and imaginary parts of w • For large Q
values, the change in the resonant frequency, rather than its square, is
wo
D.w=Imw=--
2Q
'':'' '" ""'"Y' •v 'vu..u uy ""'_I-'' . v< , ' ' " lU"t:S. Hle
near equality of the real and imaginary parts of the change in w2 is a consequence
of the boundary condition (8.11) appropriate for relatively good conductors. For
very wssy sysrems or ooundanes w1th dttferent surtace impedances, the relative
magnitude of the real and imaginary parts of the change in w 2 can be different
from that given by (8.99).
In this section, as in Section 8.6, the discussion has been confined to non-
degenerate modes. Generalization to degenerate modes is treated in Problem
8.13.
0 0 D ... 1. .1 T. .L ~
u.~ ~ .... ••• ........ ~ •.., .,.,. U ~\.<:~
Schumann Resonances
··r " vu " • y '" t'' v n~v~ u y cuv , l l l l ll"Oll
as one boundary surface and the ionosphere as the other. The lowest resonant
mO\nP< "f "wh " <v<tPm ~" ,.,;rln~fl ,J' ·1 · J', tJ.
lMt\C wa v or tne oraer or magmtuae or the eann s raOius. ln such

. muM uo::
circumstances the ionosphere and the earth both appear as conductors with real
, ;,;'-''· vvu•nH\01 lld' a . , OI U 0.1 H m , W~e tne iOllO-
sphere has u 10 -10 n 'm -•. The walls of the cavity are thus far from
perfectly conducting, especially the outer one. Nevertheless, we idealize the phys-
ICal reauty ana conswer as a model two nerfectlv conductin!:!. · · <nhr·rP.,
Sect. 8.9 Earth and Ionosphere as a Resonant Cavity: Schumann Resonances 375
with radii a and b = a + h, where a is the radius of the earth (a = 6400 km) and
his the height of the ionosphere above the earth (h ~ 100 km). Furthermore, if
we are concerned with only the lowest frequencies, we can focus our attention
on the TM modes, with only tangential magnetic fields.* The reason for this is
that the TM modes, with a radially directed elcctnc held, can sat1sty the bounaary
,,-1;,· ·~ ~• '"~· ,.h;~~ t ,.;.,] .1 · fip]rJ ~t r n ~nrl r h mithcmt
ciable radial variation of the fields. On the other hand, the TE modes, with only
tangential electnc helds, muST nave a raaia1 variation or appr ITIIIr"
wavclcm•th between r = a and r = b. The lowest frequencies for the TE modes,
are therefore of the order of WrE ~ 7Tc/h, whereas for the lowest TM modes
"'TM L/U.
The general problem of modes in a spherical geometry is involved enough

that we leave it to Chapter 9. Here we consider only TM modes and assume that
+L r. ,U.,. · ' .-l -~+\, · ,t} ,] ·m~lo ..h .crho loct ;c ~~ roolrodr;~,· ·
it is kno:.Vn from -~onsideration of spherical h;rmo~ics that the relevant quantity

is l, not m. If the radial component of B vanishes and the other components do
not depend on cp, the vanishing of the divergence of B requires that only B<P is
nonvanishing if the fields are finite at (} = 0. Faraday's law then requires E 1, =
0. Thus the homogeneous Maxwell equations specify that TM modes with no cp
dependence involve only E,. E 8 , and Bq,. The two curl equat10ns ot Maxwell can
} ' ;~orl oftor occn~;~~ o t' _.-l. 1rlPnrP of P iw£ into
"' '
,,2
c2 H VXVXH u (ts.lUUJ
where the relative permeabilities of the medium between the spheres are taken
as umty. The cp component ot (ts.lUUJ 1s
oi~ R :R (sin(} r84,)

W [f I d
-;2(rB"') + ~. 2 (rB<P) + , 2 :JR = 0 (8.101)
The angular part of (8.101) can be transformed into
-
a [ -1- - a Sill
ae sin(} ae (
. (} rB
1')
J = -sin(}
1- -a ( .
ae
Sill (}
iJ(rB 4,))
ae
---
rBq,
sin2 (}
showing, upon comparison with (3.6) or (3.9), that the(} dependence is given by
the associated Legendre polynomials PI"( cos (}) with m = :+:: 1. It is natural there-
fore to write a product solution,
Ut(r)
B"'(r, e)=-- P)(cos (}) (8.102)
'
0.
ID'"'' • • 1.-l. th A: .• 1 ' .t' f... f "\
O<UUUH UHV CJ
\V.iV J "1' '1\' , .
•2 I > 2 If I 1 \
-=-=7\' '\'
+ --
rz u1(r) = 0 (8.103)
dr 2 cz
with l = 1, 2, ... defining the angular dependence of the modes.

The characteristic frequencies emerge from (8.103) when the boundary con-
*For a spherical geometry the notation TE (TM) indicates the absence of radial electric (magnetic)
field components.
ditions appropriate for perfectly conducting walls at r = a and r = b are imposed.

The radial and tangential electric fields arc
. 2 " . 2
IC " n~ , IC ",
r
wr sine ae \""' '"J<J•} wr '\' LJ r < f\'-VO VJ
2
ic a ic2 aul(r) 1
Ee = - - - (rBq,) = ---- P 1 (cos e)
wr ar wr ar
The vanishing of Ee at r = a and r = b imnlies that the boundary condition frw
u 1(r) is
0 for r a and r b (8.104)

dr
The solutions of (8.103) arc r times the spherical Bessel functions (see Section
9.6). The boundary conditions (8.1 04) lead to transcendental equations for the
characteristic frequencies. An example is left as a problem· for our present nur
poses a limiting case suffices. The height h of the ionosphere is sufficiently small
'"u <v '"" .au;u, u waL Lilt: 1imi1 ma <...<... 1 canoe assumeu. 1ne 1~1 + 1)/r
tt:rm in (8.103) can be approximated by its value at r =a. Tht: solutions of (8.103)
•• . .f ,\ " .f. ' ' 2. . .
u1(r) =A cos [q(r - a)]

where qh = wrr, n = 0, 1, 2, .... For n = 1, 2, ... the frequencies of the modes
',J .tl 1 tL ""· .J +L .J .cc.
'J '0 UH~ m e ' " rue ~• " ' UL Lllt
TE modes. Only for n = 0 are there very-low-frequency modes. The condition

q = 0 is equivalent to u1(r) = constant and
Wf- Vl(l + 1)- (8.105)

a
where the equality is exact in the limit hla ~ 0. The exact solution shows that to
first order in hla the correct result has a replaced by (a + ~h). The fields are
Ee = 0, r 2 E, rx P1(cos e), rB, rx Py(cos e).
The resonant frequencies (8.105) are called Schumann resonances.* They are
.1.1 • . ~ •
cAu • •v .. "'iu"""'""· wiLu u u"tuu 1\.lu, '"" urs, uve resonan, .requencies
are w 1127r = 10.6, 18.3, 25.8, 33.4, 40.9 Hz. Schumann resonances mamfest them-
nl. ,], ' tJ., .C .1 1 C.
, • 1- ·r ruur VL cAu J rv" «'-''-[' 'jJLUjJ-
agating around the earth. Lightning bolts containing a wide §J2ectrum of fre-
quencics, act as sources of radial electric fields. The frequency components near
the Schumann resonances are propagated preferentially because they arc normal
modes of the earth-ionosphere cavity. The first definitive observations of these
,L · tJ., _, ""'" t "h1.
t 'Jc' ~~ .um '"''- " " uvv, u• 0 """'"'"
that Nikola Tesla mav have observed them before 1900.* A tvnical noise nower
*W. 0. Schumann, Z. Naturfor;·ch. 72, 149, 250 (1952).

'M. Balser and C. A. Wagner, Nature 188, 638 (1960).
" ' u. u. ~"'""' • v , "~ .':_~l'•;uo, •~uo ), · _ in JViKota 1 esta, Lecrures anu rmems and Art1cles,
Nikola Tcsla Museum, Beograd, Yugoslavia ( 1956), this remarkable genius clearly outlines the idea
of the earth as a rcsonatill_g_circuit (he did not know of the ionosphere), estimates the lowest resonant
frequency as 6Hz (close to the 6.6 Hz for a perfectly conducting sphere), and describes generation
o ,,1 del• .. ,; ,fth 1, _ '•'- - ' · " ' nc --'- <- ,,_. · · · ·
• J.
·~
Sect. 8.9 Earth and Ionosphere as a Resonant Cavity: Schumann Resonances 377
Sf-
6!--
....
.t:
0:::
::>
:::-
g
-
:0
~
.e4 -
~
0:::
'
"0
"'
,,!\ ~
"' 2 f-
I
I
I II
I
II
I
v
I
I
I
I \X
I
Y.5
- I I
v
\~ .A.~o.
..Jio
0 I I I I I I '""' I
0 10 20 30 40 50
Frequency (Hz)
J J"'-'"
c
. "" .,\
L'lf;UHO Oo7 1 HU»C 1 >pC'-H UlU <H WW
'"'!' Oj>
ooservea m Lavangsaa1en, r'iorway 0~,--r:m::r. --rrrc- 1." 0~11UU\iliiH
resonances at ~;, 14, LU, and Lb tl.z, plus peaks at JL, J I, anu 'l.:l nz as weu as smauer
structure are visible. [After A. Egeland and T. R. Larsen, Phys. Norv. 2, 85 (1967).]
spectrum is shown in Fig. 8.9. The resonances are clearly visible. They shift
slightly and change shape from day to day, but have average linear frequencies
of R, 14, 20, 26, 32, 37, and 43 Hz for the first seven peaks. These frequencies
are {Jiven auite closelv bv 5.8\/Ul + 1) Hz the coefficient being 0.78 times
.n ,(
'\
n.;auy IIULeu, LHe
'I At; U" \
01 penecu y
"I'
'T'\.,
w au' "
1. ~l
HVIU "'-
Ulv
,f
~
•n :o~ ~
~b'
;,. nnt <
... <;"""
or.
W<
r.
o< oL
truth. The Q values are estimated to oe mflle oroer ot 4 to lU tor the nrst tcw
resonances, corresponding to rather heavy damping. The effect of the damping
on a resonant frequency is in the right direction to account for the differences
between the observed values and (8.1 05), but the simple shift implied by (8.99)
is only about half of what is observed. The Vl(l + 1) variation of the resonant
frequencies is, however, quite striking.
The simple picture of a resonant cavity with well-defined, but lossy, walls
accounts tor the mam teatures ot the"SC!lumann resonances, annuugn raiting in
some auantitative asoects. More realistic and detailed models and discusswn of
th .t:nnc """ -ho fmmrl '" o ~n .. :nm hu
-~
r:.o lo:c * o< wPlJ o< h;< mnn,
Vutep. UlC U>C UL

... ... w· 0
propagatiOn or clcctromagneuc waves arouna the eann u1scusseu in LI!e

dliU 1' , ~crry
IS
.,_
'"~
oOOKS
C1
.r
Ul
by Budden and Wait listed at the end of this chapter. Two curiosities may be
*J. Gakjs, J. Res. Nat. Bur. Stand. (U.S.) 69D, 1043 (1965). See also T. Madden and W. Thompson,
Rev. Geophys. 3, 211 (1965) and F. W. Chapman, D. L. Jones, J.D. W. Todd, and R. A. Challinor,
Radio Sci. l, 1273 (1966).
permitted here. On July 9, 1962, a nuclear explosion was detonated at high alti-
tude over Johnston Island in the Pacific. One consequence of this test was to
create observable alterations in the ionosphere and radiation belts on a world-
wide scale. Sudden decreases of 3 5% in Schumann 1 esonant fr equcncies were
observed in France and at other stations immediately after the explosion, tbe
changes decaying away over a period of several hours. This is documented in
Fig. 17 of the paper by Galejs.
The second cmiosity is the proposal* that Schumann resonances can sene
as "a global tropical thermometer." The average magnetic field intensity of the
fundamental Schumann resonance is expected to be strongly dependent on the
frequency of lightning strikes around the world (which are seen from satellite
observations to peak strongly in the tropics, ±23° latitude). The frequency of
lightning strikes at a number of sites in the tropics is known to be dramatically
correlated to the average temperature. This lightning-temperature relation pro-
vides the physical understanding of the remarkably close correlation of
Schumann resonance monthly 111ean magnetic field strength and monthly mean
surface temperature observed at Kingston Rhode Island, over a 55 year period
and suggests that the Schumann resonances can serve as a global thermometer!
8.10 Multimode Propagation in Optical Fibers

Optical fibers lie at the heart of high-speed, high-capacity telecommunications.
Y1stble or mfrared hght, modulated with the signal, is transmitted with little loss
through small silica fibers. The , cry great ft eqnency of the can ier light means
that very large bandwidths are available for the signals. The technology has ad-
vanced rapidly in the past 25 years; a voluminous technical literature continues
to gww. We can discuss only some of the basic pnnctplcs. !'he reader wishing
more can consult the references given at the end of the chapter.
Transmission via optical fibers falls approximately into two classes-multi-
mode or single-mode propagation. "Cores" (the region where most of the energy
flow is located) are typically 50 J.Lm in diameter for multimode propagation, com-
pared to a wavelength of the order of 1 J.Lm, while 5 J.Lm diameters are typical of
single-mode fibers We first consider multimode transmission for which the se-
migeometrical eikonal or WKB approximation is appropriate. Single-mode prop-
agation. is best. described in waveguide terms. These concepts arc treated in the
Optical fiber cables. of the order of 2 em in diameter, are actually nests of

smaller cables each containing six or eight optical fibets protected by secondary
coatings and buffer layers. The operative fiber consists of a cylindrical core of
radius a [2a = 0(50 J.Lm)] and index of refraction n~o surrounded by a cladding
of oulet tadius b [2b 0(1:50 J.Lm)] and mdex of refractiOn n0 < n 1 , as shown in
Fig. 8.10a. Since the wavelength of the light is 0(1 Mill), the ideas of geometrical
optics apply; the interface between core and cladding can be treated as locally
fiat. If the angle of mcidence i of a ray originating within the core is greater than
1
i0 , where i 0 sin (n 0 /n 1 ) is the critical angle for total intcmal Iefiection, the ray
will continue to be confined-it will propagate-as shown in Fig 8 1Ob and 8 1 Oc
*E. R. Williams, Science 256. 1184-1187 (1992).

Sect. 8.10 Multimode Propagation in Optical Fibers 379
Figure 8.10 Optical fiber core and cladding, with inner cylinder of index of refraction
section, showing a total internally reflected ray; (c) longitudinal section of the core,
() < 11max = cos - 1 (n0 /n 1 ), the critical angle for total internal reflection.
It is convenient to use the

the cylinder axis. Propagation occurs lS
also convenient to use the parameter
ll=
nf- = 1 (8.1 06)
1
us to estimate the number of

propagating modes. The transverse wave number kl_ = k() is limited because
() < ()max· Two-dimensional phase-space number density dN is
where the first factor is the spatial area, the second the wave-number volume
2
element, and the factor 2 is for two states of polarization. With d k = 27Tkl_ dk1_ =
8 dO =
to
N = 335. In contrast, single-mode propagation has a = 0(2.7 /-Lm) and
0 . Then N = 0(2), one for each state of polarization. Such a phase-space
geometries are possible. Consideration of Snell's law at successive interfaces

shows that if the indices of refraction decrease from layer to layer out from the
center, a ray leaving the axis at some angle is bent successively more toward the
axis until it is totally reflected. In fact, for an arbitrary number of layers outside
the core, the critical angle Omax COS 1(noute,htinnec), j ast as fOI the simple two-
index fiber. The limit of many layers is a "graded" index fiber in which the index
of refraction varies continuously with radius from the axis. Grading addresses
the problem of distortion caused by different optical path lengths for different
angles of launch, as we discuss below.
For multimode propagation, especially in graded fibers, the quasi-geomet-
rical description called the eikonal approximation is appropriate. We assume that
the medium of propagation is a linear, nonconducting, nonmagnetic material with
an index of refraction n(x) = V E(x)/E0 that varies in space slowly on the scale of
the local wavelength of the wave. 'Nith fields varying in time as e iwt, the Maxwell
equations for E and H can be combined to give Helmholtz wave equations of
the form
(8.108)
2 2
VH + J.LoW E(x)H zw(V E) X E ()
The assumption that E(x) changes little over a wavelength allows us to drop the
terms involving the gradient of E as the next order of smallness. Then the com-
ponents of the electric and magnettc fields sahsfy
(8.109)
Locally, the basic solutions ate "plane" waves; that is, there is a local wave num-
her Ik(x) I - wn(x)lc [t is suggestive to write without approximation to (8.109)
as yet,
lfr = eiwS(x)!c
(8.110)
where the function S(x) is called the eikonal. Insertion of (8.110) into (8.109)
leads to an equation for S'
2
2 [n (x) VS · VS] + i V2S - 0

c c
Consistent with the hypothesis of slow variation of n(x) on the scale of a wave-
length (and thus small change inS on the same scale), we neglect the last term
as h1gher order. We then have the ezkonal approximation of quasi-geometrical
optics,
(8.111)
To interpret the eikonal S and connect it to geometric ray tracing, we first
consider the expansiOn of S(x) in a Taylor series around some point x 0 :
S(x) S(x0 ) + (x x0 ) • VS(x0 ) + · · ·
The wave amplitude ifJ is then
ifJ (x) = exp [iwS(Xo) lexp l

.(x - x0 ) • ----''-""-
wVS(x0 )
c
r •' r • '<L 1~/. \ .'r'rC'f. \

UHO <V~~ll V~ '/'" lll<l~ V~ <1 nu•v nnu •\' ()J '\'"U/'
n(x 0 )wk(x~yc,
whereK(xOJls a una vecror in me airecrion or VJI.'XOJ.--riT o 1a1
we define k(x) by
VS = n(x)k(x) (8.112)
The amnlitude Mx) describe~ a wave front that is locallv plane and is propagatmg
in the direction defined by k(x). If we imagine advancing incrementally in the
direction of k, we trace out a path that is the geometrical ray associated with the
wave. Figure 8.11a sketches such a path. If the distance along the path is labeled
bv the variables the incremental chane.e Llr has associated with it an incremental
~o , r
becomes dr!ds =
k. We therefore can write a result equivalent to (8.112) to
describe the optical ray path r(s),
., .. , ds
Consider now the change in the left~hand side with s along the path,
r
d dr _ dS
But d!ds = k · V, so that, from (8.112), dS/ds = k · kn(r) = n(r). We thus obtain
an equation relating the coordinate r(s) along the ray to the gradient of the index
of refraction, a generalization of Snell's law,
' -'· ~
u n(r) d.. = Vn(r) (8.114)
ds s_
Rays in a circular fiber fall into two classes:
l. M erzazonat rays: rays that pass mrougn me cy1inuer axis; mey ~.v~
to modes with vanishing azimuthal index m and nonvanishing mtens1ty at
p = 0.
2. Skew rays: rays that originate off-axis and whose path is a spiral in space with
inner and outer t nrninP' noints in radius· thcv correspond to modes with non-
vanishing azimuthal index m and vanishing intensity at p = 0.
X X
~8
/"
nlxJ /
__ :
/ /r+~r \ ;/
1/ \
IL_I/~~~-z
tal (b)
"' .•: fh\
z u11ecuun WILli muex-urrerr

382 Chapter 8 wa,,eguides, Resonant Cm'ities, and Opti£al Fillers 81
For simplicity, we apply (8.114) only to the transmission of meridional rays in an

optical fiber, or equivalently to rays in a "slab" geometry. Let the propagation
of radiation be in the x-z plane, generally in the z direction with an index of
refraction that is "graded" in the x direction, i.e., n = n(x), as indicated in Fig.
8.11b. Suppose that a ray leaves the origin at an angle 0(0) with tespect }o the z
axis, as shown. A distances along the ray (at the pomt Pl the unit vector k makes
an angle e(x) with the z axis. Note that we write e(x), not O(x, z), because the
coordinate x of the point P on the ray determines the value of z, modulo multiple
values if the ray bends back toward the z axis. In terms of O(x), the derivatives
in (8.114) arc dxlds = sin e(x) and dz!ds = cos e(x). Then the vector equation
(8.114) has as its two components,
[n(x) sin fl(x)] and ds [n(x) cos O(x)] 0

ds dx
The second equation has as its integral, n(x) cos O(x) = n(O) cos 0(0). If n(x) is
a monotonically decreasing function of lx 1. for any given 0(0) there is a maximum
(and a minimum) value of x attained by the ray, namely, when cos O(xmax) = 1.
The index of refraction at IxI X max IS
n n(Xmax) n(O) cos[&(O)] (8.115)

The parametern is a characteristic of a given ray or trajectory [specified by 0(0)].
From n(x) we can deduce Xmax and so delimit the lateral extent of that trajectory.
To find the actual path x(z) or z(x) of the ray we must return to the equations
for x and z in terms of s. The first integral of the z component of (8.114) is, as
we have just seen, n(dzids) n. Ih1s means that we can replace dlds m the x
component of (8.114) by d/ds = (n!n)d/dz. The equation then reads
dn
-n- - ndx)
d (- -
n(x) dz dz dx
2
2 dx [n (x)] (8.116)
Equation (8.116) has the structure of Newton's equation of motion of a particle

of mass min a potential V(x), with t ~ z, m ~ n2 , and V(x) ~ -n 2 (x)/2. Just as
in mechanics, use of the "velocity" x' = dxldz allows one to write d 2 x/dz 2 =
d(x'"/2)/dx and find a first integral (conservation of energy in mechanics),
(8.117)
the constant of integration being determined by the condition that x' = 0 when
n(x) = n. The trajectory z(x) is found from the integral,
Here it is assumed that the ray began at the origin with angle 0(0). For x < Xmax•
the path represents one-quarter of a cycle of oscillation back and forth across
the x = 0 line, as shown in Fig. 8.12a. The half-period of the ray (from one
crossing of the z axis to the next) is
z = 2n fm"
a Yn 2
dx
(x) ~ n 2
(8.118)
Tn cli,rll« thP • timP nf ~ mmrP. ~]nnO <>n ODtical fiber We n~~cl to

... ,).
t> • J
'1
"' -r
'1
~
•• ). 1
'6'
_,, n1 ,n thP r~u
-J.
ThP'P n~th
r
lPnnth
-o
rrom A to 15 are
I"
Lohv as ana Lapt n~XJUS
JA
wnn as ~nm) az ~nmJ~aztax) ax Jn~x)l v" ~-'J It J u.<, Wt: LULU Lilt:
ohvsical and ootical oath len!'.ths for half a oenod to be
(Xmax nfr1 ;/y X rna-::
n 2 (x) dx
Lnhv 2 I and Lant 2 ( 15.lllJ)
JO VW~X) n v V It ~A)
"
The transit time of a ray of a given launch angle 8(0) 1s g1ven by tne opt1ca1 path
divided by c. For a length of fiber z >> Z, the transit time T(z) is
L
T(z) = ~~ (8.120)
z c
Different rays, defined by different 8(0) orn, have different transit times, a form
of dispersion that is geometrical. (Note that cZ/Lopl is the ray equivalent of the
group velocity within the fiber.) A signal launched with a nonvanishing angular
spread will be distorted unless n(x) is chosen to make the transit time largely
independent of n. With a graded profile that decreases monotonically with jxj,
rays with larger initial angles and so larger Xmax will have longer physical paths,
but will have larger speeds (phase velocities) c/n(x) in those longer arcs. There
is thus an inherent tendency toward equalization of transit times. The grading
can in fact be chosen to make all transit times equal (see Problem 8.14). A simple
example is shown in Fig. 8.12b. The fractional increase in optical path length Lop•
[divided by n(O)] relative to Z as a function of 8 is shown for a simple two-index
fiber and a Gaussian-graded fiber with the same values of n 1 = n(O) and n 0
(ll = 0.01). For 0 < 8 < 8max = 0.1414, the graded fiber has a fractional change
of less than 10- 5 ; for the simple fiber, the spread is 1%.
The geometrical dispersion resulting from different launch angles 8(0) has
its counterpart as intermodal and intramodal dispersion when the propagation is
described by discrete modes, as in the next section (see Problem 8.16). There is
also material dispersion from the optical properties of the dielectrics. The optical
path length Lapt (8.119) is then modified by having one of the factors of n(x) in
•'- • _..1 :~/0 11(\\ ,I _..1 1- .Jr. ,/ -\11.1 C. ;].·~~ thP nrnnnuPln~;h;
~ .. ~ .. ·-:&·_ . \' .
'! ·r •J 'T " ' , 'TJ' ' 1. O' .l '':1.,_ -~
~ 1
m llle lHlllHLt: 1~ ~La ty fLlll, V'-'Y ""'IS'- aL ll l..J HHO UHU
very high information transm1sston rates are possilife there. Absorptton ts a mm-
1 <;<; ,.m· I. . ~rp nf thP nrrlPr nf () ') clBikm (see Section 9.7 for
imnm ~t A
• ' n .1 • - '
UO~ HUJ'~'&" ~~ o« '6
1.' .. \
384 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibcrs-SI
0.2
I I I I _I I I
0.1 r-- ?\ ·",'

,, -~ ''
',, /
1\
\ -
........ '\'•,
,,' ~
2 L -~ ·, /
/ \
\
~ ~ ------,
c I
\
I \\
"
0
0
u
~ u.u
\
\ / \
XZ>---.A
\
"''c"
> \
\
i" \
f-
\
-0.1 -
\
\
\
\
\/
I
/
-;x- ~ ' ' ,,
,
-0.2 I I I I I I I I
0 1 2 3 4 5 6
z
(a)
10 2
I ' I I I I
I
~""'''""
w-3
r '
1-
1-
4
, """'
~
~ w-5
"'-<
~
~
~
~
0
w-6 ~
= _/Graded fiber
w-7 §"-
E / ~
1 n 81-
0 0.02
I / I
0.04 0.06 0.08 0.10 0.12 0.14

v
(b)
' .!11? fn\"R· · ·· .1 .. 1 n .In 1'1" "" AA
·o ,. .., -c max max "~·- ·~-· .... ,, '-' v.v1
(dashed curves) and a graded fiber of the same radius a, critical angle, and central
inaex, our wnn a vauss1an profile, n(x) n 1 e ' "' tor U < x < a (b a/V Ll) and n(x)
- n 0 for x > a. Note the difference in scales. Units are such that a - emax - VZ/10.
I h \ DiffPrPnrP•·..in. ·•h 1. ••• "' ." " " •• .·' . ". ·' " ,£_,_c.
actual length along; fiber: [Lopt/n(O)Z- 1], fo;~ simple two-index fiber with Ll = 0.01
ana tne uauss1an-graded tiber of (a); limax - V2/10. The compensation from faster
phase velocity at larger excursions away from the axis in the graded fiber is striking.
s.ll Modes in Dielectric Waveguides
While the geometrical ray description of propagation in optical fibers is appro-
priate when the wavelength is very short compared to the transverse dimensions
gation at a given frequency can occur only via certain discrete modes, each with
unique transverse field configurations and axial wave numbers. The bound rays
(IJ < !Jm"x) in the geometric description have their counterparts as bound modes,
with fields outside the core that decrease exponentially in the radial direction.
Unbound rays (IJ > (Jmax) correspond to the radiating modes, with oscillatory
guides. We now discuss modes in a planar guide and then introduce the circular
fiber.
the simple situation of a "step-index" planar fiber consisting of a dielectric slab

of thickness 2a in the x direction and infinite in the other two directions. We look
for waves that are traveling in the z direction and are independent of y. The
indices of refraction are n 1 and n 2 for the slab and its surrounding medium (clad-
(Jm"xis totally internally reflected; the light is confined and propagates in the z
direction, as discussed in the preceding section. The discrete mode structure oc-
curs when we consider the wave nature of the light. Instead of solving the bound-
ary-value problem, as for metallic waveguides, we keep to the optical description
as the
k sin IJ, and the other with k, = -k sin IJ. To have a stable transverse field con-
figuration and coherent propagation in the z direction, the accumulation of trans-
verse phase on the path from A to just beyond B (with two internal reflections)
must be an integer multiple of 27T:
Figure 8.13 Slab dielectric waveguide. Ray or normal to wave front at angle t! shown.
where pis a nonnegative integer, k = n 1 wlc, and cf> is the phase associated with
the total internal reflection, according to the Fresnel formulas (7.39) and (7.42).
These phases are easily found to be
(8.122)
where Ll = (nf - nD!2nf_ The subscripts TE and TM in waveguide language

correspond to the electric field being perpendiculm and parallel to the plane of
incidence in the Fresnel equations. Introducing the fiber parameter (frequency
variable) V = ka\12S.. and transverse variable {; = sin e;VZS.., (8.121) can be
written
(8.123)
where .f I forTE modes and f - 11(1 211) for TM modes.

The two sides of (8.123) are plotted in Fig. 8.14 fot V 1 and V 10. There
are seven TE and seven TM modes for V = 10. For small Ll the TE and TM
modes are almost degenerate. The left hand side of (8.123) shows that there ate
roughly N = 4 V/7T modes in all, a number that follows from the one-dimensional
phase-space estimate,
r"m"' d' 2k rYE 2V
An appropriate expression for the roots of (8.123) for TE modes is given

in Problem 8.15. The lowest approxtmatJon, vahd for V >> I and small p, ts
~(IE) (p + 1)1ri2(V + 1 ), showing eqnal ~pacing in p, as implied by the
phase-space argument.
Although our phase coherence argument relied only on the wave in the
interior of the slab, fields exist outside, too. Their influence is expressed in
the phases ¢. From (7 .46) we find that the fields outside the slab vary in x as
e<-lllxll, where
v'v/l (8.124)
a
For a fixed V, as the mode numbc1 p increases (~----.1), f3 gets smaller and smaller,
the fields extended farther and farther into the cladding For angles A > fl,"*
a> 1), {3 becomes imaginary, corresponding to unconfined transverse fields. The
slab radiates rather than confines the fields. In the waveguide regime, part of the
power propagates within the core (slab) and part outside (see Problem 8.15). For
V 1, 1oughly two thirds of the power of the TE0 mode IS earned w1thm the
core. When V >> 1, the lower modes are almost totally confined. Only for p
fJmax does any appreciable power ttavcl outside the core.
Note that if ll is very smalL em .. = ~ is small The longitudinal propa-
gation constant kz k cos II k for all the waveguide modes, as we saw in the
geometrical optics approach. For the TM modes, in which there is a longitudinal
Sect. 8.11 Modes in Dielectric Waveguides 387
\
\,
41-- "
3-
-
2-
1-
-
0
I I I I I I '-1 I
X
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
(a)
1.1
(b)
Fignre 8.14 Graphic determination of eigenvalues for planar slab optical fiber:
tan(Vx - prr/2) = fY1/x 2 - 1; x = sin e/\125. = e/emw V = ka\125. (f = 1 for
TE modes, f = 1 + 211 for TM modes); dashed curves have f = 1.04. (a) V = 1,
x = 0.739 (TE), 0.747 (TM). (b) V = 10, seven roots forTE and for TM (p = 0 .... , 6).
component of electric field En we have IEzlEx I tan IJ < Hmax v'ZK for small
Ll. Thus, to zeroth order in 11, the TM modes have transverse electric fields and
are degenerate with the TE modes. Appropriate linear superposition of two such
degenerate modes gives a mode with arbitrary direction of polarization in the
x-y plane. Such modes are labeled LP (for linearly polarized, although they can
be circularly or elliptically polarized as well). LP modes are approximate descrip-
tions in other geometries, such as circular, provided .il << 1, as is mentioned at
the end of this section.
B. Modes in Circular Fibers

Optical fibers come in a wide variety of cross-sectional shapes, many analyz-
able only by numerical methods. The circular fiber with an index of refraction
that is azimuthally symmetric is one of the simplest to discuss, but even it is more
complicated than the one-dimensional slab geometty of the pteccding section.
We give only an introduction here. The reader wishing to go into mote detalls
may consult the references cited at the end of the chapter.
We consider a fiber of uniform cross section with unit relative magnetic per-
meability and an index of refraction that docs not vary along the cylinder axis
but may vary in the transverse directions. For the present we do not restrict the
problem to a circular cylinde1. The Maxwell equations can be combined, as m
Section 8.2, with assumed propagation as e'k,z-iwr, to yield the Helmholtz wave
equations for H and E,
2 2
(8.125)
where we have written e = n 2 e0 • Just as in Section 8.2, the transverse components

of E and H can be expressed in terms of the longitudinal fields Ez and Hz.
Explicitly, the connections are
E, WJLoZ X V,h'z]
y
and (8.126)
where y = n w /c - k; is the radial propagation constant, as for metallic wave-

2 2 2 2
guides. If we take the z component of the equatwns (8.125) and use (8.126) to
eliminate the transverse field components (and assume that &n 2 /(}z 0), we find
generalizations of the two-dimensional scalar wave equation (8.34),
2
w
and (8.127
Our first observation is that, in contrast to (8.34) for ideal metallic guides,
the equations forE, and H, arc coupled. In general there is no separation into
purely TE or TM modes. We restrict further comments to the simple situation
of a core that is a circular cylinder of radius a with an azimuthally symmetric
index n(p ). The cross products on the right-hand sides in (8.127) arc proportional
7
to (Bn /Bp)(Bipocp)[E" Hz]. Only if the fields have no azimuthal vanatJon are these
right-hand sides zero; only in such circumstances are there separate TE and TM
modes. One might think that for a step-index fiber the transverse gradient of n7
would vanish, at least for p < a and for p > a; but there's the rub The change
from n = n 1 inside to n = n2 outside implies a transverse gradient,
V,n 2 2nyt:.a(p a)P
The equations are coupled, unless the fields are independent of az1mutfi. The
modes with both Ez and Hz nonzero arc known as HE or EH hybrid modes. In
Sect. 8.12 Expansion in Normal Modes; Fields Generated by a Localized Source 389
tangential E and H across p = a. Separation of variables in cylindrical coordi-
p<a
an
with the z and t dependences understood. Here ·l = niw2 /c 2 - k; and {3 2 =

2 · · ns t = a with the transverse com-
poncnts computed from (8.126), leads to a determinantal eigenvalue equation
for the various modes sec Pro em . ne n s t at t e an mo es
have nonvanishing "cutoff" frequencies, with the lowest corresponding to V =
n 1 waVU!c = 2.405, the first root of J0 (x). In contrast, the lowest HE mode
propagates in the fiber.

The azimuthally symmetric TE or TM modes correspond to meridional rays;
the HE or EH modes, which have azimuthal variation, say, as sin(m4>) or
cos(m4> ), correspond to skew rays. at s ew ray mo es ave ongt u ma
components of both E and H can be understood physically by considering the
total internal reflection of such a ray at p = a. Since the plane containing such a
. .
vector after reflection will have a different projection on the z axis than before,
as will the magnetic field vector. Successive reflections therefore mix TE and TM
waves; the eigenmodes have both Ez and Hz nonvanishing.
In fibers with very small , ca e wea y gm mg wavegm es m e 1 -
crature, the fields are found to have very small longitudinal components and are
closely transverse. The language of plane light waves can be employed. For ex-
are approximately linearly polarized and vary as 10 ( yp) in the radial direction.
In the "weakly guided" approximation, this mode is labeled LP01 .
The discussion so far (and some further aspects addressed in the problems)
provide a brief introduction to t e su Ject o optlca wavegut cs. e 1 era ure
is extensive and growing. The interested reader may gain entree by consulting
one of the references at the end of the chapter.
in a hollow uide arc described by an infinite set of characteristic or normal

modes consisting of TE and TM waves, each with its characteristic cutoff fre-
390 Chapter il Waveguides, Resonant Cavities, and Optical Fibers-SI
ubgtacle, u1 apertme in the guide, the fields are relatively simple, with only the
propagating modes (often just one) present with appreciable amplitude. Near a
source or obstacle, however, many modes, both propagating and evanescent,
must be superposed in order to descnbe the fields correctly. I he cutoff modes
have sizable amplitudes only in the neighborhood of the soarce or obstacle; their
effects decay away over distances measured by the reciprocal of the imaginary
part of their wave number. A typical practical problem concerning a source,
obstacle, or aperture in a waveguide thus involves as accurate a solution as is
possible for the fields m the vtcm1ty of the source, etc., the expansiOn of those
fields in terms of all normal modes of the guide, and a determination of the
amplitudes for the one or more propagating modes that will describe the fields
far away.
A. Orthonormal Modes
To facilitate the handling of the expansion of fields in the normal modes, it
is useful to standardize the notation for the fields of a given mode, treating TE
and TM modes on an equal footing and introducing a convenient normalizatiOn.
Let the subscript A or ,u denote a given mode One may think of A - 1 2, 3, ..
as indicating the modes arranged in some arbitrary order, of increasing cutoff
frequency, for example. I he subscnpt ,\ also conveys whether the mode is aTE
or TM wave. The fields for the A mode propagating in the posith·e z direction
are written
E\+ 1(x, y, z) = [EA(x, y) + EzA(x, y)]e'kAz (8.129)
H\+l(x, y, z) = [HA(x, y) + HzA(x, y)]eikAz
where EA, HA are the transverse fields given by (8.31) and (8.33) and EzA• HzA
arc the longitudinal fields. The wave number kA is given by (8.37) and is taken
to be real and positive for propagating modes in lossless guides (and purely imag-
inary, kA = iKA, for cutoff modes). A time dependence e iwt is, of course, under-
stood. Fm a wave propagating in the negative z direction the fields are
E\ ) [EA EzAJe ikAz (8.130)
H\-) = [- HA + HzA]e-ik,z
The pattern of signs in (8.130) compared to (8.129) can be understood from the
need to satisfy V · E = V · H = 0 for each direction of propagation and the
1eqni1 ei11ent of positive power flow in the dil ection of ptopagation. The overall
phase of the fields in (8.130) relative to (8.129) is arbitrary The choice taken
here makes the transverse electric field at z = 0 the same for both directions of
propagatiOn, JUSt as 1s done for the voltage waves on transmiSSion hncs.
A convenient normalization for the fields in (8.1:69) and (8.130) is afforded
by taking the transverse electric fields EA(x, y) to be real, and requiring that
( E, · E, da = /5,,. (8.131)
J '
where the integral is over the cross sectional area of the guide. [The orthogonality
of the different modes is taken for granted here. The proof is left as a problem
Sect. 8.12 Expansion in Normat Modes; Fields Generated by a Localized Somce 391
(Problem 8.18), as is the derivation of the other normalization integt als listed
below.] From the relation (8.31) between electric and magnetic fields it is evident
that (8.131) implies
r 1
(8.132)
and that the time-averaged power flow in the Ath mode is
(8.133)
It can also be shown that if (8.131) holds, the longitudinal components are nor-
malized according to
AVES
r
T AVES
(8.134)
As an explicit example of these normalized fields we !tst the transverse elec-

tric fields and also T-1,; and E, of 1he TF and TM modes in a rectangular guide.
The mode index A is actually two indices (m, n). The normalized fields are
Exmn 2nm
,~ COS
(m1rx) (n1ry)
Sill
'Ymnav ab a b
Eymn =
2
7r~
'Ymnb ab
sin(m7rx) cos(n1ry)
a b
(8.135)
Eymn
27Tm
,~
. (m1rx) cos (n1ry)
sm (8.136)
'Ymnavab a b
with 'Ymn given by (8.43). The transverse magnetic field components can be ob-
tained by means of (8.31). For TM modes, the lowest value of m and n ts umty,
htt1 for TF modes m - 0 or n - 0 is allowed. If m - 0 or n - 0, the normalization
must be amended by multiplication of the right-hand sides of (8.136) by 1/VZ.
:)";}~ Chapter ~ Waveguides, Kesonant Cavities, and Optical Fibers-Sl
B. Expansion of Arbitrary Fields

An arbitrary electromagnetic field with time dependence e-iwr can be ex-
panded m terms of the normal mode fields (8.129) and (8.130).* It is useful to
keep track explicitly of the total fields propagating in the two directions. Thus
tne aronrary news are wntten m the form
E- E\"J + E\->, H- H 1+> + Ht-> (8.137)
whP-rP.
J?(~) 'Y A (+)L'(+) u(+) 'Y A ( + )U( ce) '0 '
~A
-A ' "A "A (U·LJO)
~A "'-;:
Snecification of thP. .inn ,,-,,ffi · A ( +) onrl A ( ) rl inc>o +h <:~u
everywhere in the guide. These may be found from b~undary or source con~~:
bons m a vanety ot ways. Here is a useful theorem:
The fields evervwhP.rP. in thc> ""irlP orp -'· . IPrlnnitliiPlv hv onp, -~
0 • ., •• ,
of the transverse components of E and H m a plane, z = constant.

Proof' There is no loss in generality in choosing the plane at z = 0. Then from
(8.137). (8.138) and (f\_1)Q) (R nn) thP tron fic>lrlo "rP
F 'Y (A(+) -'- A (-)\F

~
'A
A ' 'A
(8.139)
H - ) (AC}J- AC-;->)H,
A'
1i 111e scatar prouuCL 01 ootn sides or tne nrst equatiOn 1s tormed w1th -"A and an
integration over the cross section of the guide is performed, the orthogonality
..
(0-~J.l.) tllll-'1110~
A~+) +A\.. l = J EA • E, da
Similarly the second equation, with (8.132), yields
A\+> - AJ,-> = z~ JHA • H, da

The coefficients A),'l are therefore given by
A\") = ~ { (EA • E, ::':: Z~HA • H,) da (8.140)

'T~. 0~ .~ ;+"' rl TT n oL " .
t t m ~ IS"~· «< .c v, cu~ "o m cu~ '"!'
(8.137) and (8.138) are determined. The completeness of the normal mode ex-
' o 'the> nn;t1 .++L ,,· +. oll
r "t' ~·
c. Fields Generated by a Localized Source

The news m a waveguide may be generated by a localized source, as shown
o"hc>moti"o Jl ;rl J:l' Q 1 <; rr'L -l ,· T/, ,\ ' -l ' '
. , ·o· · • , u\~, 'J w •~ •~• y HI Lilli'-'
as e· •wr. Because of the oscillating current, fields propagate to the left and to the
*We pass over the mathematical problem of the completeness of the set of normal modes, and also
only remark that more general time dependences can be handled by Fourier superposition.
Sect. S.U Expansion in Normal Modes; Fields Generated by a Localized Source 393
n n
Figure 8.15 Schematic representation of a

localized source in a waveguide. The walls
of the guide, together with the planes S 1
and S' define the volume cont(-Jjning the

s_ source.
right Outside the source, at and to the right of the surface S say, there will be1
,
only fields varying as e'kAz and the electric field can be expressed as
E = E(+) = L A\;lE\;> (8.141)
A'
wtth a correspondmg expresswn for H. On and to the left of the surface S _ the
fields all vary as e -ikAz and the electric field is
E = E(-) = LA\. lE~-) (8.142)
A'
To determine the eoeffieients A\+J in terms of J, ·.ve consider a form of the

energy flow equation of Poynting's theorem. The identity
V · (E x H\"l - E~-'-l x H) = J · E\"l (8.143)
follows from the souree-&cc Maxwell equa~ions for E\+l, II~+l, and the Maxwell
equations with source satisfied byE and H. Integration of (8.143) over a volume
V bounded by a closed surface S leads, via the divergence theorem, to the result,
(E X H\±l - E\±l X H) · n da = (8.144)
where n is an outwardly directly normal. The volume Vis now chosen to be the
volume bounded by the inner walls of the guide and two surfaces S+ and S_ of
Fig. 8.15. With the assumption of perfectly conducting walls containing no sources
or apertures, the part of the surface integral over the walls vanishes. Only the
integrals over S+ and S contribute. For definiteness, we choose the lower sign
in (8.144) and substitute from (8.141) for the integral overS+:
(
/, I
With the fields (8.129) and (8.130) and the normalization (8.133), this becomes
(
(8.145)
Js,
The part of the surface integral in (8.144) from S_ is
(
Js_ A' Js_
which can easily be shown to vanish. For the chmce of the lower s1gn m (8.144),
therefore, only the surfaceS, gives a contribution to 1he left-hand side Similarly,
for the upper sign, only the integral over S_ contributes. It yields (8.145), but
with A\-! instead of A~+!. With (8.145) for the left-hand side of (8.144), the
coefficients A~.c) are determined to be
(8.146)
2 Jv
where the field E~-J of the normal mode A is normalized according to (8.131).
Note that the amplitude for propagation in the posithe z direction comes from
integration of the scalar product of the current with the mode field describing
propagation in the negative z direction, and vice versa.
It is a simple matter to allow for the presence of apertures (acting as sources
or sinks) in the walls of the guide between the two planes S+ and S_. Inspection
of (8.144) shows that in such circumstances (8.146) is modified to read
(8.147)
where E is the exact tangential electric field in the apertures and n is out·:;ardly
directed.
I he apphcatwn of (8.146) to examples of the excitation of waves in guides
is left to the problems at the end of the chapter. In the next chapter (Section 9.5)
we consider the question of a source that is small compared to a wavelength and
derive an approximation to (8.146): the coupling of the electric and magnetic
dipole moments of the source to the electric and magnetic fields of the Ath mode.
I'he couplmg of waveguides by small apertures is also discussed in Section 9.5.
The subject of sources and excitation of oscillations in wa~eguides and cavities
is of considerable practical importance in microwave engineering. There is avo-
luminous literature on the topic. One of the best references is the book by Collin
D. Obstacles in Waveguides
Discontinuities in the form of obstacles, dielectric slabs, diaphragms, and
apertures in walls occur in the p1 actical usc of waveguides as carriers of electro
magnetic energy and phase information in microwave systems The expansion of
the fields in normal modes is an essential aspect of the analysis. In the second
(1975) edition of this book we analyzed the effects of transverse planar obstacles
with variational methods (Sections 8 12 and 8 13). Lack of space prevents inclu
sian of the material here. The reader interested in pursuing these questions can
refer to the second edition or the references mentioned below and in the Ref-
erences and Suggested Reading.
Theoretical and expernnental study of obstacles, etc. loomed large in the
immense radar research effort daring the Second World War. The contributions
of the United States during 1940-45 are documented in the Massachusetts Insti-
lute of Technology Radiation Laboratory Series, published by the McGraw-Hill
Book Company, Inc., New York. The general physical principles of microwave
circuits are covered in the book by Montgomery, Dicke. and Purcell. while a
Cb 8 References 395
compendium of results on discontinuities in waveguides is provided in the volume

by Marcuvitz. Collin, already cited, is a textbook source.

Waveguides and resonant cavities are discussed in numerous electrical and commu-
nications engineering books, for example,
Ramo, Whinnery, and VanDuzer, Chapters 7, 8, 10, and 11
The two books by Schelkunoff deserve mention for their clarity and physical insight,
Schelkunoff, Electromagnetic Fields
Schelkunoff, Applied Mathematics for Engineers and Scientists
Among the physics textbooks that treat waveguides transmission lines and cavities are
Slater
Smythe, Chapter XIII
Sommerfeld, Electrodynamics, Sections 22-25
Stratton, Sections 9.18-9.22
An authoritative discussion appears in
F. E. Borgnis and C. H. Papas, Electmmagnetic Wawguides and Reumater.~, Vol.
XVI of the Encyclopaedia of f'hy.liC.I, ed. S. Flugge, Spt in get Verlag, Berlin
(1958).
The books by
Collin
Harrington
are mtended for graduate engmeers and phys1c1sts and are devoted almost completely to
guided waves and cavities. The standard theory, plus many specialized topics like discon-
tinuities, are covered in detail. The original work on variational methods for discontinu-
ities is summarized in
J. Schwinger and D. S. Saxon, Discontinuities in Waveguides, Notes on Lectures
by Julian Schwinger, Gordon & Bn~ach, New York (196il).
Variational principles for eigenfrequencies, etc., as well as discontinuities, are sur
Cairo and Kahan

and also discussed by
Harrington, Chapter 7
Van Blade!, Chapter 13
The definitive compendium of formulas and numerical results on discontinuities,

junctions, etc., in waveguides is
Marcuvitz
The mathematical tools for the treatment of these boundary-value problems are pre-
Morse and Feshbaeh, especially Chaptet 13

PerturbatiOn of boundary condttwns ts dtscussed by Morse and Feshbach (pp. I 038 ft).
Information on special functions may be found in the ever-reliable
Magnus, Oberhettinger, and Soni, and in encyclopedic detail in
Bateman Manuscript Project, Higher Transcendental Functions.
Numencal values ot special functiOns, as well as formulas, are given by

Abramowitz and Stegun
Jahnke, Emde, and Losch
Two books dealing with propagation of electromagnetic waves around the earth and
in the ionosphere from the point of view of waveguides and normal modes are
n
'"-.
<
See also
Galeis
ol.
_, rl' rl. :J'
P. V. Bliokh, A. P. Nicholaenko, and Yu. F. Filtippov, Schumann Resonances in

the Earth-Ionosphere Cavity, trans!. S. Chouet, ed. D. L. Jones, IEE Electro-
~ •Nn>t. \X/. ~ ·
, VAl Q , D, . n -b· · , T ' 11noC\\
\ v ,.
There is a huge literature of the theory and practice of optical fibers for communi-
cations. Our discussion in Sections 8.10 and 8.11 has benefited from the comprehensive
book
A. W. Snyder and J.D. Love, Optical Waveguide Theory, Chapman & Hall, New
York (1983).
tsooKs with discussions ot the wavegmoe aspects, as well as much practical detail, are
J. M. Senior, Optical Fibre Communications, 2nd ed., Prentice-Hall, New York
(1992).
L. vassauo, uptzcat wavegwae concepts, blsev1er, New YorK (IYYI ).
Numerical methods are often required for optical waveguide geometries. A useful refer-
ence IS
r. A. ~ernanaez ana r. LU, lVltcrowave and Upttcat Wavegu1ae Analysts by the
Finite Element Method, Research Studies Press & Wiley, New York (1996).
Problems
8.1 Consider the electric and magnetic fields in the surface region of an excellent con-
ductor in the approximation of Section 8.1, where the skin depth is very small
• -1 In thP rorlii nf nf thP" -fo0a A~ tha e0nla ,f ·' ;,c; '' •
-, ._,
variation of the fields just outside.
(a) For a single-frequency component, show that the magnetic field H and the
currem aensny .1 are sucn tnat 1, tne time-averaged torce per umt area at the
surface from the conduction current, is given by
/Lc
• .. 41"TT'"
where tt IS the peaK parallel component of magnetiC field at the surface, /L,
11
is the magnetic permeability of the conductor, and n is the outward normal at

the surface.
,._, • 1..' •
H•L
'" e • • • ee e
f'n"
'"' ''J !--" " ' ' ' ' - ' " " ' - " "UlH_,l_,HC HUH<
an additional magnetic force per unit area? What about electric forces?
Cln-'''-
tCJ Assume mar me ne1as are a superposttlon ot dttterent frequencies (all high
enouah that the" · 1Mions still holrli '>hnw thot thP timP-'"'"rcoaPrl fnr~<"-
Ch. 8 Problems 397
takes the same form as in part a with IHIII 2 replaced·by 2(IH, n where the
angle brackets (- · -) mean time average.
8.2 A transmission line consisting of two concentric circulm cylinde1 s of metal with
conductivity u and skm depth 15, as shown, IS filled With a umform Iossless dielectnc
(JL, E). A TEM mode is propagated along this line. Section 8.1 applies.
(a) Show that the time-averaged power flow along the line is
P = R 7Ta
2
1Hol 2
In(~)
where H 0 is the peak value of the azimuthal magnetic field at the surface of
the inner conductor.
1....--- b ---;;.j
I
Problem 8.2
(b) Show that the transmitted power is attenuated along the line as
P(z) = P 0 e- 2 yz
where
,(,u! + ~)
u;
1 E
' 2uo \I u (h
In!-)
\".
(c) The charactei is tic impedance Z 0 of the line is defined as the 1atiu of the 'oltage
between the cylmders to the axial current flowing in one of them at any
position z. Show that for this line
Zo = 2 7T
1
RIn(~)
(d) Show that the series resistance and inductance per unit length of the line are
R--1_(!+!)
2'1T<M \a e
where ,u, is the permeability of the conductor The correction to the indue-
tance comes from the penetration of the flux into the conductor~ by a di~tance
8.3 (a) A transmiSSion !me conststs of two identical thin strips of metal, shown in
cross section in the sketch. Assuming that b >> a, discuss the propagation of
a TEM mode on this line, repeating the derivations of Problem 8.2. Show that
!J.L a\
'Y E I b)
R = _3_
a Jl, E
Problem 8.3
where the symbols on the left have the same meanings as in Problem 8.2.
(b) The lower half of the figure shows the cross section of a microstrip line with
a stnp of Width b mounted on a dielectric substrate of thickness hand dielectric
constant E, all on a ground plane. What differences occur here compared to
part a if b >> h? If b << h?
84 Transverse electric and magnetic "'<l"es are propagated along a hollow, right cir-
cular cylinder with inner radius R and conductivity t:r.
(a) Find the cutoff frequencies of the various TE and TM modes. Determine
numerically the lowest cutoff frequency (the dominant mode) in tenus of the
tube radius and the ratio of cutoff frequencies of the next four higher modes
to that of the dominant mode For this part assume that the conducthrity of
(b) Calculate the attenuation constants of the waveguide as a function of fre-

quency for the lowest two distinct modes and plot them as a function of
requency.
8.5 A waveguide is constructed so that the cross section of the guide forms a right
triangle with sides of length a, a, V'Za, as shown. I he medmm inside has t.L,
E - 1.
Ch. 8 Problems 3t)t)
(a) Assummg mflmte conduchv1ty for the walLs, determine the possible modes of
propagation and their cutoff frequencies.
r
a
t Problem 8.5
(b) For the lowest modes of each type calculate the attenuation constant, assuming
that the walls have la1ge, but finite, conductivity. CompaiC the nsult 111ith that
for a square gmde of side a made from the same material.
86 A reson:mt cavity of copper consists of a hollow, right circular cylinder of inner
radius R and length L, v'lith fiat end faces
(a) Determine the resonant frequencies of the cavity for all types of waves. With
(1/yr;;;. R) as a unit of frequency, plot the lowest four resonant frequencies
of each type as a function of RIL for 0 < RIL < 2. Does the same mode have
the lowest frequency for all Rl L?
(b) If R = 2 em, L = 3 em, and the cavity is made of pure copper, what is the
numerical value of Q for the lowest resonant mode?
8.7 A resonant cavity consists of the empty space between two perfectly conducting,
concentric spherical shells, the smaller having an outer radius a and the larger an
inner radius b. As shown in Section 8.9, the azimuthal magnetic field has a radial
dependence given by spherical Bessel functions, j,(kr) and nt(kr), where k = w/c.
(a)
(b) For l 1 use the explicit forms of the spheiical Bessel functions to show that
the charactenstic frequencies are given by
where h = b - a.
(c) For h/a << 1, verify that the result of part b yields the frequency found in
Section 8.9, and find the first order correction in h/a. [The result of part b
seems to have been derived first by J. J. Thomson and published in his book
Recent Researches in Electricity a11d Maglleti~m, Oxford Clarendon Press,
tion that the earth has a conductivity cre and the ionosphere has a conductivity cr;,
ljlirith corresponding skin depths f( and o,
(a) Show that to lowest mde1 in h!a the Q value is given by Q lWzi(Bc I 8;) and
determme the numencal factor N for alL 1.
(b) For the lowest Schumann resonance evaluate the Q value assuming cr, = 0.1
-:roo --;:::;J:; R "'~
., ~ 0 0
nil flnf;p,.l 0 ,,.
•
(c) Discuss the validity of the approximations used in part a for the range of
parameters used in part b.
un
0
A L • " H
cu
..
'5 o\Oof.<)"
u eo ·r
. ,.
. umo,u
..
oy a peneclly conuucLing cwseu sunace " \ wmcn may nave more man one Otscon-
nected part). A harmonic electric field inside the cavity satisfies the vector
Helmholtz equation,
V X (V X E) = k 2 E with P = w 2 !LE
ThP hnnnrbn rnnditinn i" n X F.- n ("nd n 0 R - mAn ~
(a) Show that
k~
r E* · [V X (V X E)] d 3x
I ""* ~ •3....
. "'u
Jv ~
is a variational principle for the eigenvalue k2 in the sense that a change of

E ~ E + 8E, where both E and 8~ satisfy the boundary conditions on S, leads
lU u111y •·UlUCl b ' Ill " •
(b) Apply the variational principle to the TM010 mode of a right cylindrical cavity
of radius R and length d, using the trial longitudinal electric field Ez =
E 0 cos( 7rpi2R) [no variational parameters]. Show that the estimate of the ei-
genvalue is
/ . ..2 . '
kR
"" - 4 2
Compare numerically with the known eigenvalue, the root x 01 of l 0 (x).
(c) Repeat the calculation of part b with Ez = E 0 [1 + a(p!R) 2 - (1 + a)(p!R) 4 ],
. .
"'"·"v a" a var pa1ameLer. .:~nuwlTillTTilrrnistriar runcrion me best
estimate 1s
TJ, ol +l.
kR = [ c7 _2v34)
v34
80 ·
68 +
~ ' -'
34
r 2
_, '-"
"'
L
.
'J I'
8.10 Use the variational principle of Problem 8.9 in terms of the electric field E to find
an estimate of the eigenvalue for k 2 for the TEm mode in a right circular cylinder
cavity of radius Rand length d with perfectly conducting walls. Use as a trial tunc-
tion Bz = B 0 (p/R) (1 - p/2R) cos 4> sin( 7rZid). fThis function satisfies the boundarv
~Anclitinns nf R - n M 7 - 0 "nd 7 - rl "nd ~n~~n R l -not n -
(a) First show that the variational principle can be reexpressed as
k
r (V X E*) · (V X E) d 3x
lv·~•3
Jv
(b) Show that the (transverse) components of the (trial) electric field are
E En(l - o/2R) sin <b sin( 71'Zid): E E.(J n/R\ cos rh sin( 71'7/d\
(C) LalcUJate tne curl ot .~<; and snow tnat tne aooroximation for kL is
e=~+""
2 2
5R d
Ch. 8 Problems 401
Compare with the exact result. For small enough d/R, this mode has a larger
· ,J . •" •• • cru
uw
"· UTI,
"J
,.I,'"'" •~ t voriotin ,J ' ,,
'"
be at all reliable?
'
(d) The original variational expression in Problem 8.9 has an equivalent integrand
in thco nn ' I?*. rvrv. 1?\ V"El. Discuss the relative mPrit< nf thio
integrand compared with the square of the curl of E in part a for the present
problem.
8.11 Apply the variational method of Problem R.9 to estimate the resonant frequency of
the lowest TM mode in a "breadbox" cavity with perfectly conducting walls, of
length d in the z direction, radius R for the curved quarter-circle "front" of the
_,,_ _, .1 "L n _, ", __ ,_, r ,_ ....:1.~+:.- .rl 1.... +l. ... 1.
'" ~uuuvA, uuu cu~ UHU ·' J 'C
~y u, u <.. x <.. 1{) ana ~x u, u <.. y <- KJ, respec,iv"'Y· u>c <uc '""' wncLion,
Ez = £ 0 (p/R)"(l - p/R)sin 2</J
for the only component of electric field present. This function gives vanishing tan-
gential component of E on the boundary surfaces; the index v is a variational pa-
rameter. Show that
(v + 2)(2v + 3)(.? + v + 4)
k 2R2 =
v(2v+ 1)
,'oL c ' ..
, .< /, D <. •h ' .1
""' " cv y cv 6'
function. Compare with the exact answer, kR = 5.13562, the first root of 12 (x).
8.U A waveguide with lossless dielectric inside and perfectly conducting walls has a
• ••tinnoJ .nr r thot rJ 'oJirrhth• frnrn >U io I rnnlniiT r, vh '
fields are known. The difference in boundaries is described by ll(x, y), the length
measured from C0 to C along the normal to C0 at the boundary point (x, y ). The
denvatlve aDtas along the boundary IS higher order m smau quanuues.
(a) If the eigenvalue parameters and solutions for C and C0 are ( 'l, 1/1) and ( y~, J/!0 ),
respectively, without degeneracy, show that to first order in 8
{ 1~.,. •
Teo ll(x, y) an V'o ;Pan"'·2 dl
?
y ril r
J,o I
2
J/!o 1 dx dy
where only the first (second) term in the numerator occurs for TM (TE)
modes. [Hint: Follow the same general approach as used m Sectwn ~.6 lor the
effects of finite conductivity.]
(b) Determine the perturbed value of y for the lowest TE and TM modes (TE,,o
and lNlu)In a rectangUlar gmde II me cnange m suape is as suown iu <uc
accompanying sketch.
8 8
b
k-
c ~rlo -
a Problem IS •.u
4UZ Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
.
"'" cr,,,
'd I . d" .
,,,
d
·-'-
u -J
" ..
b . b' h C
1 ea con !liOns, one must use egenerate-state pertur atwn t eory. ons1 er the
"d • u q
,, _, ",.], ..
two-dimensional (wave!!uide) situation in which there is an N-fold de2:eneracv in

the ideal circumstances (of perfect conductivity or chosen shape of cross section),
with no other nearby modes. There are Nlinearly independent solutions if!kl, chosen
to be orthogonal to the transverse wave equation ('i7 2 + Yrilif!kl = 0 i = 1 2 .
N. In response to the perturbation, the degeneracy is in general lifted. There is a
set of perturbed eigenvalues, y];, with associated eigenmodes, if!k, which can be
expanded (m lowest order) m terms or the N unperturbed e1genmodes: 1/J,
""i:,a,if!g> .
(,.) Shnw thM thP '" nf IR flRI fnr finitP <Cnnr' .. ( "nrl thP rrwrc
sponding expression in Problem 8.12 on distortion of the shape of a wave-
guide) is the set of algebraic equations,
L [( '}'
i=l
2
- Yfs)N/Ji, + Ll1,]a, = 0 (j = 1, 2, ... , N)
where
~ i!if!(i)* i!if!(i)
N. - ( l,!tUJi2 dn "nrl ;\ - f -'-'- --'-' d/
}A Jc an iin
for finite conductivity, and
Llji = fc 8(x, y) oifii/' iJif!f:' - 1/!u>* o"ifJi\' dl

iin an ° 2
iJn
for distortion of the boundary shape.
(b) The lowest mode in a circular guide of radius R is the twofold degenerate TE 11
A. ,.; ,J,.J, . 1-
, ;,~,
. 0' 'J
1/!(±J = B = Bc/1 (y0 p) exp(±i</>) exp(ikz- iwt)

7
ThP PietPnv"hw 1PIPr ;, " I R41/R ~; "' tn thP fir,t rnnt nf
dJ,(x)!dx. Suppose that the circular waveguide is distorted along its length
into an elliptical shape with semimajor and semiminor axes, a = R + LlR,
b = R - LlR, respectively. To first order in LlR/R, the area and circumfer-
ence of the guide remain unchanged. Show that the degeneracy is lifted by
the distortion and that to first order in LlR/R, Y, = i,(1 + ALlR!R) and y:i =
-lo(l - ALlR/R). Determine the numerical value of A and find the eigenmodes
as linear combinations of ljf±) Explain physically why the eigenmodes tum
VUL a' Liley UV.
8.14 Consider an optical fiber with a graded index of refraction for rays confined to the
x-z plane, n(x) = n(O) sech(ox). The fiber has large enough transverse dimensions
(x J to contain all ravs of interest which are evidentlv svmmetric about x - 0. The
invariant n = n(xmax) = n(O) cos IJ(O).
(a) Solve the eikonal equations for the transverse coordinate x( z) of the ray and
show that
n<Y- ,;nh- 1 r,;nh(,y 1 ,;nln<zll
where the origin in z is chosen when the ray has x = 0. Sketch rays over one
half-period for "launch angles" IJ(O) = 7rl6, 7rl4, and 7r/3.
(h\ Finrl the h"lf- 7 nf thP r"v. DnP' it dPnPnd nn n?
(c) Show that the optical path length for a half period, Lopt = Jn(x) ds is Lopt =
n!Q)Z, Comment on the effectiveness of this narticular "r"dinP nf th" index.
Ch. 8 Problems 403
Hmt: In the compmmion of Lapt• a useful change of variables is sinh( ax)

sinh( =max) sin t. The resulting integral can be done by contour integration:
7T
8.15 Discuss the TE and TM modes in the dielectric slab waveguide of Section 8.1l.A
as a boundary-value problem.
(a) Show that (8 123) emerges as the determining relation for both even and odd
modes (in x) and that even or odd p goes with the evenness or oddness of the
mode, defined by the symmetry in x of the transverse fields.
(b) Show that the eigenvalues oft; for the TE modes are given approximately by
2
(p + 1)7T [ (p + 1?7T ]
l; = 2(V + 1) l - 24(V + 1)3
The lowest order result is accurate for V >> 1 and smallp. Check the accuracy
of the full expression against solution of (8.123) by Newton's method for
2 3
(c) Calculate the power now in the z direction (per unit length in they direction)
Wlthm the core ( lxl <a) and m the claddmg (lxl >a) tor the even IE modes
and show that the fractions are
2
F = l [ 1 + sin(2Vt;)] and F _ l [ cos (Vt;) ]
cme S 2Vl; dad - S V\IJ _ l;2
where
where Cis the root of (8 123) for thepth mode Find corresponding expressions
for the odd TE modes.
8.16 The longitudinal phase velocity in the dielectric slab waveguide of the preceding
problem is uP = wlkz = c/(n 1 cos ()P). Intermodal dispersion occurs because the
dielectric media have dispersion and also because the group velocity differs intrin-
sically for different modes.
(a) Making the approximation that the dielectrics' dispersion can be neglected,
show that the group velocity vg = dw/dk, for the TEn mode is
c cos ()" [ 1 + {3"a ]
where ()P is the eigenangle of the pth mode (cos ()P - V1 - 2Lll;~) and {3P is
given by (8.124). Interpret the departure from vpvg = c21ni (as in metallic
R ......:;. oo) jn terms of the Goos H;incben effect and ray-like prop-
waveguides·, rp
agation at the simple phase speed c/n 1 . [Hint: Write the eigenvalue relation
(8.121) in teuns of the independent variable wand the dependent variable kz
and differentiate with respect to w.]
(b) Write a program to evaluate vg versus VIV,, where V, = p7T/2 is the threshold
frequency variable for the p'h mode. Make a plot of v glc for n 1 - 1.5, n2 - 1.0
for p = 0, 1, 2, ... , 6 as a function of V!V,(p = 1) on the range (0, 10).
(c) Relate the results of part b to the optical path length difference for the step-
index fiber shown in Fig. 8.12b. Can you generate a plot from the results for
vg(p) at fixed F for n 1 - 1.01, n 2 - 1.0 to compare with the "classical" ray
result?
r. '..1.
" -" _,'
" 1''7
'"
~-
' ." • u '

-o"•u• w •amus
," ~JHHUU~"' < HU- a). Assume that the fields vary as e'"'<l> 'tk,z ''"'. For bound modes, the fields in
the core (cladding) are proportional to ordinary (modified) Bessel functionslv(Kv)
with appropriate values of v and argument, as in (8.128).
*
(a) Show that form 0 the eigenvalue relation for the transverse parameter Ymn
(and f3mn) is
1
' + lK~\ m ( ni -+ n~l(l + 1_\
2
I n, 1'm + nz2 K'm \ 1 [ ] m "'
2
,_]l_j_m /3 Km \ Yfm {3 K ./ a" \ y' ~ sz;

wnere_r n 1 ur_!c · k; and W k; n2wL!c" while_Qrimes indicate deriv-
atives with respect to the argument, and the argument of J, (Km) is ya (f3a).
The first subscript on y is the azimuthal index m; the second designates the
nth root of the eigenvalue equation for fixed m.
(b) Determine the eigenvalue equation for the m = 0 modes (TE and TM) and
show that the lowest "cutoff" frequency corresponds to V = 2.405, the first
root of J0 (x ), where "cutoff" is the frequency below which the guide radiates
•'- thon
(C) :-.now tnat tne lowest HE mode (HE , l has no cutoff freauencv and thot
for V << 1 the ..,, f<n A P BIV' Pinel A oncl R in fa ,,
'1
and n 2 •
8.18 (a) From the use of Green's theorem in two dimensions show that the TM and
TE modes in a waveguide defined by the boundary-value problems (IU4) and
(8.36) are orthogonal in the sense that
r Ez,Ez""da = 0 for A * JL
fnr_IM • ..1 nr! o
z tnr_= -r
lin<T
0
raloti~n fm U . ..1 .
(b) Prove that the relatiOns (8.131) (K 134) form a consistent set of normalization
conditions for the fields, including the circumstances when A is a TM mode
and JL is a TE mode.
8.19 The figure shows a cross-sectional view of an infinitely long rectangular waveguide
with the center conductor of a coaxial line extending vertically a distance h into its
interior at z = 0. The current along the probe oscillates sinusoidally in time
with frequency w and its variation in so ace can be aooroximated "' Tl "\
T_ ',f(, .1~\1~ u\1 ·n ··-' '
1n
" ,, ., J-,
·o
.~.
. ..1 'T'J.
-'- " '

OU~H <H"< U<HJ > <H~ ~
10 llllJU<;; Ciolll fC m 1111; gutu~;, '
= I
l 0
:X h
J
I Problem 8.19
(a) Calculate the amplitudes for excitation of both TE and TM modes for all
(m, n) and show how the amplitudes depend on m and n form, n>> 1 for a
fixed frequency w.
Ch. 8 Problems 405
(b) For the propagating mode show that the power radiated in the positive z
direction is
~c I6 . -
P=--sm
wkab a
2
2crx) sm. 4(-wh)
2c
with an equal amount in the opposite direction. Here k is the wave number
for the TE 10 mode.
(c) Discuss the modifications that occur if the guide, instead of running off to
infinity in both directions, is terminated with a perfectly conducting surface at
" T HTJ., , L .C I "' ' ' ol., fl. fr fi •rl f ')
c r
What is the radiation resistance of the probe (defined as the ratio of power
flow to one-half the square of the current at the base of the probe) at
max1mum?
8.20 An infinitely long rectangular waveguide has a coaxial line terminating in the short
.
O>U<O W <H'- !S'.UU<O
. .
HIH 'H'-
,.
UHH UOUU<U
' ' .
"6
. ..
"t'
..
radius R whose center is a height h above the floor of the guide, as shown in the
accompanying cross-sectional view. The half-loop is in the plane z = 0 and its radius
R is sufficiently small that the current can be taken as having a constant value / 11
everywhere on the loop.
I. a
I ----;r
_'I'.
[:)· 2R
"'I f l
J Problem 8.20
(a) Prove that to the extent that the current is constant around the half-loop, the
TM modes are not excited. Give a ohvsical explanation of this lack of
excitation.
(b) Determine the amplitude for the lowest TE mode in the guide and show that
1ts value 1s mdepenaent ot the height h.
(c) Show that the power radiated in either direction in the lowest TE mode is
p = 16 z :!. ( 7T~)
16 b a
where Z is the wave impedance of the TE 10 mode. Here assume R << a, b.
8.21 A hollow metallic waveguide with a distortion in the form of a localized bend or
mcrease m cross sectwn can support nonpropagating \ oounu SLaLe ) conugura-
tions of fields in the vicinity of the distortion. Consider a rectangular guide that has
its distortion confined to a nlane as shown in the figure and TE '" as its lowest
propagating mode, with perpendicular electric field E" = .p. On either side of
' ' ,1., '-1..•
'"'··' ·• ' · "''"
·•· \~fith~nt "'' · · , ,f,
• •2 ~·~
.6'
E-o >lll\'IIYIUJ <OXiJ\·-l"-i.), Wll<Oli;"
.o
\WIC)
•
\
0
"'
/U) , o " ' - ~'"'"'
.
""" w
"
~'-0'-U J} "- " 1.,
a curvature K(s) = l!R(s) and a width w(s). Locally the element of area in the plane
is dA h(s, t) ds dt, where sIS the length along the gmae wau ana t tne transverse
coordinate, as shown in the figure, and h(s, t) - 1 - K(s)t. In terms of s and t the
T onlorion ;, I
2
'\' "' = -;: "'
1 a(
h "'
a.p
)+i:o(i:~~
406 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-Sf
If the distortions are very small and change slowly ins on the scale of the width
a, an ansatz for the solution is
1/J(s, t) =
u(s)
"\~ • smw(')
. [ rrt J
[The factor in the denominator is equivalent to the factor p 112 familiar from Bessel
functions that converts the radial part of the Laplacian in polar coordinates to a
simple second partial derivative (plus an additional term without derivatives).]
(a) Show that substitution of the ansatz into the two-dimensional wave equation
('\7 2 + w 2/c2 ).p = 0, leads to the equation for u(s), '
if small terms are neglected. Interpret v(s) in analogy with the Schri:idinger
. . .
(b) If the distortion is in the form of a bend through an angle with constant
(I
radius of curvature R >> a, show that for (}a/ R << 1 there is a "bound state"
at frequency Wo where
References: J. Goldstone and R. L. Jaffe, Phys. Rev. B 45, 14100 (1992); J.P.
Carini, J. T. Londergan, K. Mullen, and D P Mmdock, Phy' Rev 8 48, 4503
(1993).
I
I K(s) = l!R(s)
I
I
I
I
I
r
Problem 8.21
s
In Chapters 7 and 8 we discussed the properties of electromagnetic waves and
was said about the generation of sue waves. In t e present c apter we turn to
this question and discuss the emission of radiation by localized systems of oscil-
latin char e and current densities. The initial treatment is straightforward, with-
out elaborate formalism. It addresses simple systems in which electric dipole,
magnetic dipole, or electric quadrupole radiation dominates, or the sources are
I . ·. ca. . Th . ·
multipole expansion of a source in a waveguide is also treated, and the effective
multipole moments of apertures. These "elementary" discussions arc followed
. .
For a system of charges and currents varying in time we can make a Fourier
f h time de endence and handle each Fourier com anent sc aratel .
We therefore lose no generality by considering the potentials, fields, and radia-
tion from a localized system of charges and currents that vary sinusoidally in
p(x, t) = p(x)e-iwt
-iwt
(9.1)
*See Problem 9.1 for some of the subtleties that can arise over factors of 2. There are also factors of
electron atom our classical dipole moment p is replaced by 2e(f r i) for a transition from state i to
stater
408 Chapter 9 Radiating Systems, Multipole Fields and Radiation-51
It was shown in Chapter 6 that the solution for the vector potential A(x, t)
in the Lorenz gauge is
r r I \
T/, I
, 1\ I •I
A(x t) ~
d3r 1 dt' t' + t I {0 ')\
'-,
47T lx x'l
"'' c
provided no boundary surfaces arc present. The Dirac delta function assures the
causal behavior of the fields. With the sinusoidal time dependence (9.1), the
solution for A becomes
A_lx)
.
!:::.<: I
r
.H x')
0 ik1x-x'1
· d'x' (Q ':\\
47T J IX x· I - '
where k = wlc is the wave number, and a sinusoidal time dependence is under-
stood. The magnetic field is given by
1
H=-VxA (9.4)
I"" II
"
"uuv, .~ ··-
LUv OUU> '-''-'o "''-' . 11CIU I~
iZ
E = ko V x H (9.5)
where Z 0 = ~ is the impedance of free space.

(iiven " 1t cli~trihntir.n l(v'\ thP fip]ck rem in nrinrinlP ' t l~n<•t , >
-1 -1 L ., .
' ll.l ' . ' 'A
VJ --'---'- vu<:o UI LWU CX-
'"" )• H '- YHH ~7.J
amples ot direct mtegration of the source integral in Section 9.4. But at present
we wish to establish certain simple, but general, properties of the fields in the
limit that the source of current is confined to a small region, very small in fact
compared to a wavelength. If the source dimensions arc of order d and the wave-
length is A= 27Tc/w, and if d <<A, then there are three spatial regions of interest:
The near (static) zone: d << r <<A
The intermediate (induction) zone: d << r - A
The far (radiation) zone: d <<A<< r
We will see that the fields have very different properties in the different zones.
In the near~ li >lrlc h>lvP the'. rl nf . fie_L:k--"lith_
, ·~rli~l r-r.m_
~
r """ y
Wll11 lllitl 111 u 11 --"""-1:"_u fJ ern cs 01
the source. ln the far zone, on the other hand, the fields are transverse to the
radius vector and fall off as r-\ typical of radiation fields.
For the near zone where r <<A (or kr << 1) the exponential in (9.3) can be
replaced by unity. Then the vector potential is of the form already considered in
J. u1c: ~11V"'"' uis,ance can oe expanacu using p.tu J, wnn Lne result,
ILII ~ 47T Y... re ml 1

nm
kr--'*"0
&~XJ
47T 1.2 21 + 1 rt+1 J J(X' )r 'Y tm(tl, ¢')a- X (Y.b)
This shows that the near fields are quasi-stationary, oscillating harmonically as
e-'"'', but otherwise static in character.
• ~ £ • • •
' " '"" w• "'""" ~"" ~~ IJ 111e exponenna1 m ~~ ..J J OSCI11aLeS rafJillY ana ae-
Sect. 9.1 Fields and Radiation of a Localized Oscillating Source 409
termincs the behavior of the vector potential In this region it is sufficient to
lx - x' I = r - n · x' (9.7)

where n is a unit vector in the direction of x. Furthermore, If only the leading
term in kr is desired, the inverse distance in (9.3) can be replaced by r. Then the
vector potential is
. A(x) = -
hm
kr~oo
fLo -
eikr
47T r
I · · d3x'
J(x')e-,kn·x (9.8)
I hrs demonstrates that in the far zone the vector potential behaves as an outgoing
spherical wave with an angular dependent coefficient. It is easy to show that the
fields calculated from (9.4) and (9.5) are transverse to the radius vector and fall
1
off as r . They thus correspond to radratron fields. If the source dimensions are
small compared to a wavelength it is appropriate to expand the integral in (9.8)
in powers of k:
J(x')(n · x')" d 3 x' (9.9)
The magnitude of the nth term is given by
..!_
n!
I J(x')(kn · x')" d 3 x' (9.10)
Smce the order of magmtude of x' is d and kd is small compared to unity by

assumption, the successive terms in the expansion of A evidently fall off rapidly
v.'ith n. Consequently the radiation emitted from the source will come mainly
from the first nonvanishing term m the expansron (9.9). We wrll examine the first
few of these in the following sections.
In the intermediate or induction zone the two alternative approximations
leading to (9.6) and (9.8) cannot be made; all powers of kr must be retamed.
Without marshalling the full apparatus of vector multipole fields, described in
Sections 9.6 and beyond, we can abstract enough for our immediate purpose. The
key result is the exact expansion (9.98) fur the Green function appearing in (9.3).
For points outside the source (9.3) then becomes
A(x) fLolk ;§;, h)Il(kr)Y 1m(IJ, ¢){ J(x')Jt(kr')Y'fm((l', </>') d 3 x' (9.11)
,ffl ,J
If the source dimensions are small compared to a wavelength, Mkr') can be

approximated by (9.88). Then the result for the vector potential is of the form
of (9.6), but with the replacement,
(9 12)
r r
*Actually (9.7) is valid for r >> d, independent of the value of kr. It is therefore an adequate ap·
proximation ewn in tile near zone.
410 Chapter 9 Radiating Systems, Multipole Fields and Radiation SI
The numerical coefficients a; come from the explicit expressions for the spherical
Hankel functions. The right-hand side of (9.12) shows the transition from the
static-zone result (Q 1'>\ fr.r Ia .--.-- 1 tA th ,1." -·. ~- -?~ fv, :.... ~~ L
~ £ ~- 0 .:;! ........ ,".... -.
- ctuu Olller rypcs or ramatwn, we examine
the question of electric monopole fields when the sources vary in time. The analog
of (9.2) for the scalar potential is
<l>(x, t) = -=- J
41TE0
r
d 3 x' J dt' p~x, tJ_
Ix ~ x' I
oj t' + IX
c
X' I
~ t
The electric monopole contribution is obtained by replacing Ix ~ x' I -;. IxI =r

unuer me m egral. The result 1s
q(t' = t ~ ric)
<l>monopoie(X, t) =
41TE0
where q(t) 1s llie total charl>e Ol th~ ~nrP rhctraP io ~Artco~u~-1 rl

localized source is by definition one that does not have ~harge flowing into or
away from it, the total charge q is independent of time. Thus the electric monopole
nnrt of thP • c ;~1 f~n-1 h ,]..1~\ .~ 1 - 1." . r 'cy SCllllC HlC
*
,. . ~VUL'-'<0 <J VL •
uc1u' willl ic rime aepcndence e w•, w 0, have no monopole terms.

We now turn to the lowest order multipole fields for w 0. Because these *
fields can be calculated from the vector potential alone via (9.4) and (9.5), we
omit explicit reference to the '('e>hr 11 ;rt mh~t J'AliA 0
9.2 Electric Dipole Fields and Radiation

Tf onhr thP h~ot to~~ ;rt {0
.
0\ '
/
1
'l" '
•L
~~'"· l"
. ' .,.
A(x)
1-t eikr
= ....Q -
JJ(x')d x' 3
(9.13)
41T r
Examination of (9.11) and (9.12) shows that (9.13) is the l = 0 part of the series
and that it is valid everywhere outside the source, not just in the far zone. The
integral can be put in more familiar terms by an integration by parts:
r r
JJd 3
x' = ~ J x' (V' · J)d 3 x' = ~ iw J x' p(x')d 3 x' (9.14)
since from the continuity equation,
<wp -.... ~~-D)
Thus the vector potential is

lUAW elKr
l'l.(X) p (Q 111\
'+1T r
where
f
p J x' p(x')d· x' (9.17)
is the electric dipole moment, as defined in electrostatics bv (4.8\.

~
~~~ . ~~
---~
. ~· -• . "''~'"'. .... -·· A11
UlC toltoClllC ··r

-
0 H'-'UO HCHH
.. ' <n "
\7•">} aHU
-• •n ~'
\7,J} m~
.. TT
ck2 ,
47T \"
.\
l'!
eikr
r
1
i
1
ikr (9.18)
1 e.ikr 1 ik \
E k"lD X p) X D + ljn(n · p) PJI r3 rz }e
47TEo
1
We note that the magnetic ne1a IS transverse to tne rauius vecLor a, au •

but that the electric field has components parallel and perpendicular to n.
T. •t • A.',.•' •A . th~ +>~l,.lo t~t--~ An th~ l;m;t' 10 frmn<
~
ck2 eikr
u (n '>{ nl
47T r (9.19)
E = Z0H X n
showing the typical behavior of radiation fields.

In thte nte~r zone. on the other hand the fields approach
H =iw >) 1
- (In X p) ~
'17T r ~'J.LU)
1
E - - - r3n(n · n) - ul _!,
'17TEo r
The electric field anart from its oscillations in time is iust the static electric dipole
field (4.13). The magnetic field times Z 0 is a factor (kr) smaller than the electric
field in the region where kr << 1. Thus the fields in the near zone are dominantly
electnc m nature. 1ne magnenc new vamsnes, O• course, in llle lllllll
k __, 0. Then the near zone extends to infinity.

'T'l- • A r .. <1;~t~<1 n~r ,n;t oAE<1 :molP hv thP. o<rill~tincr
-cr r r
dipole moment p is
ar = ~ Re[r 2 n · E x H*] (9.21)

d[! 2
where E and Hare given by (9.19). Thus we find

0
~~ = ;2:~ 4
k j(n X p) X nl
2
(9.22)
The state of polarization of the radiation is given by the vector inside the absolute
value signs.* If the components of p all have the same phase, the angular distri-
bution is a typical dipole pattern,
2
dP c Z"
S!ll"tl ~'J.L:>)
d[! 32 7Tz k IPI
*In writing angular distributions of radiation we will always exhibit the polarization explicitly by
writing the absolute square of a vector that is proportional to the electric field. If the angular distri-
' .. ;· ~ <. ••· ..-,, · nnlo · · · · -' · it con then he"' · ·-1 bv takinQ the scalar nroduct
·r
of the vector with the appropriate polarization vector before squaring.
A1?
"Y
()
--
, ~· . " .."
. -., .
where the angle (}is measured from the direction of p. The total power radiated,
independent of the relative phases of the components of p, is
2_7_1,4
p 0
IPI 2
(9.24)
127T
A simnlP PV~mnlP nf ~n ol. ·i~ rlinnlP ' ~: .• ;0 n f, r1 1'
d . r r h ' . ·-
whose length rs small compared to a wavelengt . The antenna IS assumed to
be onentea along the z axis, extending from z (d/2) to z (d/2) with a
narrow gap at the center for purposes of excitation, as shown in Fig. 9.1. The
current is in the same direction in each half of the antenna, having a value / 0 at
th.e~ f~llino im~tPlv linP~ PrA nt th, Pnrlf:
"
I '"> I I\
I(z)e '"" = lo( 1 ~ ~¥ }e iwt

(9.25)
From the continnihr Pnn~tinn (Q 1 <;) th" linP~r- '

'0
'
'J '
' ~ . r ..
length) is constant along each arm of the antenna, with the value,
p '(z ) = ±2il-0 (9.26)

wd
..
"'~ ul'l'~' \'v
.
VCl) M);HUClll); app1u,
.
tuVeJ ValUeS
. .
; lUl !'"'
..
lVC ( ~
.
OI Z. 1ne
dipole moment (9.17) is oarallel to the z axis :mel h~s th,.- m~onit11rl"
r<d/2) iTA
p J-(d/2)
zp'(z) dz
2w
(9.27)
The an11,ular distribution of radiated nower is
dP Z 0 I6 ( kd) 2 • 2
dfl = 1287T2 Sill (} (9.28)
while the total power radrated rs
p Z 0 I6(kd) 2
__L<l_2<ll
'Hl7T
we see that tor a fixed mput current the power radiated increases as the square
of the frequency, at least in the long-wavelength domain where kd << 1.
The coefficient of 1612 in (9.29) has the dimensions of a resistance and is
called the radiation resistance Rmd of the antenna. It corresponds to the second
term in (6.137) and is the total resistance of the antenna if the conductivity is
••
%~~~~ ;r,n
~e/ I
v : -v
_.c::_
~ ...... I
L "' ...... ...... I
"J
./ d~~~
2 .t'Igure 'J.l Short, center-fed, linear antenna.
Sect. 9.3 Magnetic Dipole and Electric Qnadrnpole Fields 413
perfect For this short center-fed antenna R,ct 5(kd) 2 ohms. In pnnciple the
input reactance for the antenna can be calculated by applying (6.138) or (6.140)
of Section 6.9. Unfortunately the calculation depends crucially on the strong
fields near the gap and thus is sensitive to the exact shape and method of exci-
tanon. Since the system is an electric dipole and the electrostatic dipole field
dominates near the antenna, we can nevertheless say with certainty that the re-
actance is negative (capacitive) for small kd.
9.3 Magnetic Dipole and Electric Quadrupole Fields

The next term in expansion (9.9) leads to a vector potential,
A(x) :; e:' C ik) J J(x')(n · x') d x' 3

(9.30)
where we have included the correct terms from (9.12) to make (9.30) valid ev-
erywhere outside the somce This vector potential can be written as the sum of
two terms. One gi\ies a transverse magnetic induction and the other gives a trans-
verse electric field. These physically distinct contributions can be separated by
writing the integrand in (9.30) as the sum of a part symmetric in J and x' and a
part that is antisymmetric. Thus
(n · x')J ~[(n • x')J + (n · J)x'] + Hx' X J) X n (9.31)
The second, antisymmetric part is recognizable as the magnetization due to the
current J:
The first, symmetric term will be shown to be related to the electric quadrupole
moment density.
Considering only the magnetization tenn, we h"ve the vector potential,
A(x) = -
1)
ik!-Lo (n x m) -e"" ( 1 - -. (9.33)
47T r tkr
whet e 111 is the rnagnelic dipole nzunzent,
(9.34)
The fields can he determined by noting that the vector potential (9 33) is pro-
pot tiona! to the magnetic field (9.18) for an electric dipole. This means that the
magnetic field for the present magnetic dipole source will be equal to l!Z0 times
the electric field for the electric dipole, with the substitution p --+ mlc. Thus we
m
1 e"" l ik
H = - e(n X m) X n - + (3n(n • m) - m]l- - - \eik> (9.35)
47T r l\r3 r21'
StmJ!arly, the electnc field for a magnetic dipole source is the negative of Z 0
time~ the magnetic field for an electric dipole (with p --4 mic)·
Z eikr ( 1 )
E = - _Q P(n x m) - 1 - -. (9.36)
47T r tkr
414 Chapter 9 Radiating Systems, Multipole Fields and Radiation-SI
All the arguments concerninl> the ior of the fielrls in the nP"r ~nrl fo
zones are the same as for the electric dipole source, with the interchange
E -'> Z 0 H, Z 0 H -'> ~ E, p -'> mlc. Similarly the radiation pattern and total power
radiated are the same for the two kinds of dipole. The only difference in the
r uem~ ~e polarization. l:" or an electric mp()J_e _!ll_e ~ectnc vector
lies in the plane defined by n and p, while for a magnetic dipole it is perpendicular
to the plane defined by n and m.
The integral of the symmetric term in (9.31) can be transformed by an in-
-o·
1-
-J r
rl
c~m ...
21J [(n · x')J ~~ x'(n · x')p(x') d 3x'
+ (n • J)x'] d 3 x' = ~ "2 (9.37)
'T'L
~ .. ~
-J
<n
\
'~'
• iJ} llle
I " " UIOI:Oil U'IOU lV y • .J uy tWp.
. .
~mce
mtegral mvolves second moments of the charge density, this symmetric part cor-
responds to an electric quadrupole source. The vector potential is
~
,,
.. ..
JLoCk2 eikr /'
Q \
1 1
I I
r ...,..
~ ~
" /_ u -Il...L
!1-'\~ I ~ ~ ~
•n
:1 ••70)
"'
The complete fields are somewhat complicated to write down. We content our-
selves with the fields in the radiation zone. Then it is easy to see that
H ;/.rn '>( ,\/.,
(9.39)
E = ikZ0 (n X A) X niJLo
Consequently the magnetic field is
·d ikr r
H = ~-
., :.__ J (n X x')(n • x')p(x') d 3x' (9.40)
81r r
With definition (4.9) for the quadrupole moment tensor,
Q~ 13 = J (3x,x 13 ~ r2 8~ 13 )p(x) d 3x (9.41)

the integral in (9.40) can be written
r
nX J x'(n • x')p(x') d 3 x' = ~n X Q(n) (9.42)
The vector Q(n) is defined as having components,
n ) n nn (Q <!'\
(3
,.·
We note that it depends in magnitude and direction on the direction of obser-
'11 - H
t ~r •~o v•
• CoL
'"~ ovu•~~.
""•L
"au
oL
'"~o~
-' .r. ..
we uavc
me magnenc maucnon,
ick3 eikr
H = ~--n x Q(n) (9.44)
247T r
and the time-averaged oow<>r r>~rli"t"rl n"r 1mit solirl >lno-1"
dP c2 Z 0
-= k 6 l[nXQ(n)]xnl 2 (9.45)
dO 11527T2
Seet. 9.3 Magnetic Dipole and Electric Quadrupole Fields 415
where again the direction of the radiated electric field is given by the vector inside
the absolute value signs.
The general angular distribution is complicated. But the total power radiated
can be calcalated in a straightforward way With the definition of Q(n) we can
write the angulm dependence as
l[n X Q(n)] X nl 2 = Q* · Q -In· Ql 2 (9.46)
= 2: Q: 13 Qaynpny - 2: Q~~Qy8 nanf3nyn 8
a, , a,f3, y,O
The necessary angular integrals over products of the rectangulat components of

n are readily found to be
Then we find
Jl[n X Q(n)J X nl 2 d!l- 47TH 2:

a,/3
IQaal 1
Since Qas is a tensor whose main diagonal sum is zero, the first term in the square
brackets vanishes identically Thus we obtain the final result for the total power
radiated by a quadrupole source:
(9.49)
I he radmted power varies as the sixth power of the frequency fm fixed quad-
rupole moments, compared to the fourth power for dipole radiatiOn.
A simple example of a radiating quadrupole source is an oscillating sphe-
roidal distribution of charge. The off-diagonal elements of Qafl vanish. The di-
agonal elements may be written
(9.50)
Then the angular distribution of radiated power is

dP
- =
c Zok 6
.
Q 2 sm2 e cos
2
e (9.51)
n 5127T 2 a
This is a four-lobed pattern, as shown in Fig. 9.2, with maxima at (} - 7T/4 and
37T/4. The total power radiated by this quadrupole is
rZ0 k 6 Q'f,
p (9.52)
9607T
The labor involved in manipulating higlter terms in expansion (9.9) of the
vector potential (9.8) becomes mcreasmgly prohibitive as the expansiOn IS ex-
tended beyond the electric quadrupole terms. Another disadvantage of the pres-
ent approach is that physically distinct fields such as those of the magnetic dipole
416 Chapter 9 Radiating Systems, Multipole Fields and Radiation-Sf
,
I
\.
~ L~) \. / I
~\ I I /
/ _.....
I
/ J1\" \ /
\.. /
-
Figure 9.2 A quadrupole radiation pattern.
ana tne eJectnc quaarupole must be disentangled from the separate terms in
(9.9). Finally, the present technique is useful only in the long-wavelength limit.
A systematic development of multi pole radiation begins in Section 9.6. It involves
:'1 f"irlv .el:.'- m: 1~tir~l 1< h11t thP nrirP n••irl io ' ' ..L;ln 'T'J..
11 ·r '. .' .
H '"au HWL_, UlUt:lS lU Ut: . 111 Lilt: "1111e Way; we reSUJ[S
are valid for all wavelengths; the physically different electric and magnetic mul-
tipoles are clearly separated from the beginning.
9.4 Center-Fed Linear Antenna

A. Approximation of Sinusoidal Current
r·v1 1..1;;1 w;u 1auia,ing systems tne geometry 01 current now IS suwc1ent1y s1m-
ple that integral (9.3) for the vector potential can be found in relatively simple,
closed form if the form of the current is assumed known. As an example of such
a svstem we consider a thin linear ant.enn~ of l.enoth d _,_. -'- is · .A ~
,11 . • • •
~· . •
. Eo~l" ~· "'" -r . u1'- . 10 aoou1u<=u tv u'- v"'"'"'"u a1v1~1; tllt: z axrs
with its gap at the origin, as indicated in Fig. 9.3. If damping due to the emission
of radiation is neglected and the antenna is thin enough, the current along the
antenna can be taken as sinusoidal in time and space with wave number k = w/c
:.ncl is svmm<etri" on thP tmr. ~rmo r.f thP ~ntPnn~ ThP r11rrPnt .", ••
ends of the antenna. Hence the current density can be written
kd
~
1 Mll
'1 KIZI' 0 ~-'-J 0 V)t:3 ~ •,!.:)_;)
for Iz I < (d/2). The delta functions assure that the current flows only along the
z axis. I is the peak value of the current if kd > 7T. The current at the gap is
T T . ,{ lull'l\
·u "\' -r
wirn me curren.!_Qenslty (LJ.)3) the vector potential 1s m !he z...IDrect10n and
in the radiation zone has the form [from (9.8)]:
JL Ieikr (d/2) kd
AJx) 7 °
4,. .I
o;n { &I I\
I I
-ikzcmO A· (() <:A\
r -Ui?\ ? ' ·- 'J
'
Sect. 9.4 Center-Fed Linear Antenna 417
2
~ ~~~-
Coaxial [',_9 ........,..,

~
feed_
y
~~
/- m '-._
X~ "
-2~~~~
"' ~
-~-
"~ c. ~
., "-
The result of straightforward integration is
(9.55)
Since the magnetic field in the radiation zone is given by H = ikn X A/ /-Lo, its
magnitude is /H/ = k sin(} /A 3 /Ii-Lo· Thus the time-averaged power radiated per
unit solid angle is
I \ I
2
cos( Ku cos fJ) - cos! Kf!:)
2 \2 \ 2
dP - Z 0 l I
- - -- (9.56)
d!l 87T2 sin fJ
The electric vector is in the direction of the component of A perpendicular to n.

Consequently the polarization of the radiation lies in the plane containing the
antenna and the radms vector to the observatiOn pomt.
The angular distribution (9.56) depends on the value of kd. In the long-
wavelength limit (kd << l) it is easy to show that it reduces to the dipole result
(9.28). For the special values kd = 7T(27T), corresponding to a half (two halves)
of a wavelength of current oscJilatwn along the antenna, the angular d1stnbutwns
are
r I \
ol 7T
ICU> '1 ICV> U
\' I
2 2
kd = 7T
dP Z l sin fJ '
- -
~ -0 -
d!l 87T2 I
7T
\
(9.57)
4 n
vVO ? vv
, kd = 27T
' sin2 fJ
Ihese angular dJstnbutwns are shown in Fig. 9.7, where they are compared to
multipole expansions. The half-wave antenna distribution is seen to be quite
similar to a simple dipole pattern, but the full-wave antenna has a considerably
sharper distribution.
I he full-wave antenna d1stnbut10n can be thought of as due to the coherent
superposition of the fields of two half-wave antennas, one above the other, ex-
cited in phase. The intensity at (} = 7T/2, where the waves add algebraically, is
418 Chapter 9 Radiating Systems, Mnltipole Fields and Radiation-Sf
four times that of a half-wave antenna. At angles away from () - w/2 the ampli-
tudes tend to interfere, giving the narrower pattem. By suital>le attangement of
a set of basic antennas, such as the half-wave antenna, with the phasing of the
currents appropriately chosen, arbitrary radiation patterns can be formed by co-
herem superposition. The mterested reader should refer to the electncal engi-
neering literature for detailed treatments of antenna arrays.
B. The Antenna as a Boundary- Value Problem

Only fur infinitely thin conductors are we justilied in assuming that the cur-
rent along the antenna is sinusoidal, or indeed has any other known form. f.
finite-sized antenna with a given type of excitation is actually a complicated
boundary-value problem. Wtthout attempting solution of such problems, we give
some preliminary considerations on setting up the boundary value ptoblem for
a straight antenna with circular cross section of radius a and length d, of which
the center-fed antenna of Fig. 9.3 is one example. We assume that the conductor
is perfectly conducting and has a small enough radius compared to both a wave-
length A and the length d that current flow on the surface has only a longitudinal
(z) component, and that the fields have azimuthal symmetry. Then the vector
potential A will have only a z component. With harmonic time dependence of
frequency w and in the Lorentz gauge, the scalar potential and the electric field
are given in terms of A by
lC
<l>(x) = - V ·A
E(x) = ~ [V(V ·A) + k 2 A]
Since A = zAz(x), the z component of the electric field is
Bat on the surface of the perfectly conducting antenna the tangential component
of E vanishes We thus establish the important fact that the !'ector potential Az
(and also the scalar potential) on the surface of the antenna are strictly sinusoidal:
(a~2 + e )Az(P = a, z) = 0 (9.59)
This is an exact statement, in contrast to the much rougher assumption that the
current is sinusoidal.
An integral equation for the current can be found from (9.3). If the total
current flow in the z ditection is I(z), then (9.3) gives for A, on the surface of
the antenna
z') dz'
K(z z') (9.60)

Sect. 9.5 Multipole Expansion for Localized Source or Aperture in Waveguide 419
is the azimuthal average of the Green function eikRIR. The condition (9.59) leads
to the integra-differential equation
d2 0\ rzo+d
u z ) uz
dzz -r K } Jzo l~Z )1\.~Z ~~-01)
This can be regarded as a differential equation for the integral, or equivalently

one can integrate (9.59) and equate it to Az(p = a, z). The result is the integral
~'1'
a 1 cos kz + a2 sin kz =
Jz 0
+d
I(z')K(z ~ z') dz'
zo
The constants a, and ao are determined bv the method of excitation and bv the
boundary conditions that the current vanishes at the ends of the antenna.
1ne somrion or me imegrat equarion is nor easy. !:'rom me rorm m ~~.OUJ i1
i< rlPllr thllt whPn 7 1 7 rllrP m11<t hP tllkPn !lnrl thP finitP r!lrlill< i< im
For a ---+ 0, the current can be shown to be sinusoidal, but the expansion param-
eter for corrections turns out to be the reciprocal of ln(d/a ). Thts means that even
£. J/. IA3 oL '- ' .£ oL J C '" H'OI '"'- oL '
LW U/U LV <H'-1 ' - '-UH Uv vVH' VL "''- VLUvL VL •u. '" " ' - " "''-' '- •o u
current node near the place of excitation, such corrections can change the an-
tenna's input impedance drastically. Various approximate methods of solution
of (9.61) are described by Jones. A detailed discussion of his version of the theory
and the results of numerical calculations for the current, resistance, and reactance
' '" ~ r '0
Ul d -H:;U dHI dH:; !;lVCH UJ HWU:;n. d l C I,.,UCU Hl
the su2:2:ested readin2: at the end of the chauter.
at:
/o~
itA' • • I~:
.. ~-···r~·~
.In 17,
·r ~: ~~ -~'~-
··~·~·· J ~·
,- ......
,l~~nA ~.
or Aperture in Waveguide
If a source in the form of a probe or loop or aperture in a waveguide is sufficiently
small in dimensions compared to the distances over which the fields vary appre-
ctably, tt can be usetully approx1mated by Its lowest order multlpole moments,
,JJ, ,J. •+•' rl , nno>t-'~ ' ' ~·r~ cAn~~oc ;rtn th~ c~m~ lmu_
· r ·o
est order multipole moments will produce sensibly the same excitations in the
waveguide. Often the electric dipole or magnetic dipole moments can be calcu-
lillt:U l!Ulll SlilllC llt:lU,, Ul t:Vt:ll 0
_
0 • LVt:llll Lllt: SUUlCt: lS UUL
... :... ,.
trulv smalL thte multinolte texnansion P'ives a oualitative and often semiauantita-
., J
"' '5 'I'
'
A r, " '· . -'' r;, . -'

In <;:, Q 1") ;t HI~C th~t th~ ~mnl;tnrl~c Ll (+) f~~ ~f thP
Ath mode are proportional to the integral

(
J J • E\+' d"x
where the integral is extended over the region where J is different from zero. If
the mode fields E~+l do not vary appreciably over the source, they can be ex-
420 Chapter 9 Radtatmg Systems, Multipole Fields and Radiation SI
panded in Taylor series around some suitably chosen origin. The integral is thus
written, dropping the sub- and superscripts on E~"l:
I J·E d x = ~
3
3
1I l
J~(x) E"(O) + t:
3
1
ax; (0)
x 13 aE + · · ·Jd 3x (9.62)
From (9.14) and (9.17) we see that the first term is
E(O) ·I J(x) d 3 x = ~iwp • E(O) (9.63)
where p is the electric dipole moment of the source:
p = !....
WJ
r J(x) d x 3
I hts can be transformed mto the more faml11ar form (9.17) by the means of the
steps in (9.14), provided the surface integral at the walls of the waveguide can
be dropped. This necessitates choosing the origin fur the multipole expansion
such that J"x 13 vanishes at the walls. This remark applies to all the multipole
moments. The use of the forms involving the electric and magnetic charge den-
sities p and PM requires that (x,J13 ::':: x 13 J")x., · · · xv vanish at the walls of the
guide. The above-mentioned form for the electric dipole and the usual expression
(9.34) for the magnetic dipole are correct as they stand, without concern about
choice of origin.
The second term in (9.62) is of the same general form as (9.30) and is handled
the same way. The product J"x 13 is written as the sum of symmetric and antisym-
metric terms, just as in (9.31):
" laXfl-
L... aE" (0) = -1 "L... (Jaxfl ~ lflxa) laE"
-.- (0) ~ a£13 (0)
-.- J
a,f3 axp 4 a,{3 axfl axa (9.64)
The first (antisymmetric) part has been written so that the magnetic moment
density and the eurl of the eleetrie field are clearly visible. With the help of
Faraday's law V X E zwB, the antJsymmetnc contnbut10n to the nght s1de ol
(9.62) can be written
I l2:(}1,{3
laXp aE"
axf3
(o)J
antisym
d 3 x = iwm • B(O) (9.65)
where m is the magnetic dipole moment (9.34) of the source. Equations (9.63)
and (9.65) give the leading order multipole moment contributions to the source
integral (9.62).
Other terms in the expansion in (9.62) give rise to higher order multipoles.
The symmetric part of (9 64) can be shown, just as in Section 9 3 to involve the
flaceiess e{ecflic quadJUpo{e moment (9.41). The first step is to note that if the
surface mtegrals vanish (see above).
r r
Sect. 9.5 Multipole Expansion for Localized Source or Aperture iu Waveguide 421
Then the second double sum in (9.64), integrated over the volume of the current
distribution, takes the form
iw
~2 L -aE" (0) Jp(x) x~xfl d x 3
a,{l ax{3
The value of the double sum is unchanged by the replacement x"'xil ~

(xaxfl ~ ~r o"' 13 ) because V · E = 0. Thus the symmetric part of the second term
2
in (9.62) is
J[La,{3 laX{3-
aE" ( 0) J
ax{3 sym
d·x
3
= iw
~6 L
a,{3
aE" ( 0)
Qafl-.-
ax{3
(9.66)
Similarly an antisymmetric part of the next terms in (9.62), involving xflxy, gives
a contribution
J[21 L{3~' 'y

J,x 13xy rPE~
ax{3 axy
(0) J d x ~
3
~ iw
6
L
~ ' {3
Q,M13 -aB" (0)
axil
(9.67)
antisyTil
in(a,{l)
where Q;;"13 is the magnetic quadrupole moment of the source, given by (9.41) with
the electric charge density p(x) replaced by the magnetic charge density,
1
nM(..-1 ~ ~V. AA ~ ~-- V. (..- X .Jl (Q t1R1
2
If the various contributions are combined the expression (8.146) for the
• 1 itnrlc> A(+) h~o ~o ito ·' c>vn~noirm
-r -r
"
.7
A("') = i ---"" D • E\+l(O) ~ m • BH(O)
L
;JJ;'(+) \7.V7)
1 riRH
+- > 0 .. ,--" (0) ~ 0':.',--" (0) + ...
0 a,{l dX13 dXf3
lt should be remembered that here the mode helds b).., are normalized accordmg
to (R_nn. Thf> f>xn;msion is most llSdul if the source is such th:1t thf>
'rll rl' ,-1, ,.,], ..l ~ ;t h• ot t. 'T'h
· · .,. d · · f , b 0
• 'J • f · ·u ' · 0
-,...,...
twnmg an onentat10n o pro es or antennas to excite pre erentla y certam
modes can be accomplished simply by considering the directions of the electric
:1ncl m:lP"nf>tic clinol<'> (or hiP"herl m 1ts of the sourc<'> :1ncl thf> ~1 mocl<'>
<".1..1.
,
'C.
~ v~ ~ ,
''-
•u~ ~"'~" • •
-~
v~
'T''C
~~
..l.
'
'•'- ''-
o u u •u:-u
, ,1
~~m~ •
. ".1..1.
'
can ue prouuceu oy a magnellc uipO•e an,enna wuose uipO•e momenL IS paraue1
to the axis of the guide. TM modes cannot be excited by such an antenna, except
via higher multipole moments.
B. Aperture in Side Walls of Guide

Apertures in the walls of a waveguide can be considered as sources (or sinks)
of energy. In Section 8.12 it was noted that if the guide walls have openings in
the volume considered to contain the sources, the amplitudes A~"') are given by
422 Chapter 9 Radiating Systems, Multipole Fields and Radiation-51
(8.147) instead of (8.146). 'Nith the r,ssumption that there is only one apeiture
and no actual current densrty, the amphtude for excitatiOn of the Ath mode is '
A\") = - Z,
2
faperture
n • (E X H\+l) da (9.70)
where n is an inwardly directed normal and the integral is over the aperture in
the walls of the guide. If the aperture is small compared to a wavelength or other
seale of Vdtidtion of the fields, the mode field Hi l can be expanded just as before.
The lowest order term, with H\+l treated as constant over the aperture, evidently
leads to a coupling of the magnetic dipole type. The next terms, with linear
variation of the mode field, give rise to electric dipole and magnetic quadrupole
couplings, exactly as for (9.64)-(9.66), but with the roles of electric and magnetic
interactions interchanged. The result is an expansion of (9.70) like (9.69):
A A(ceJ = i wZ)\ [p e f·f E<"l(O)

A - me f·fB<"'l(O)
A + · · ·] (9.71)
4
where tire effective electric and magnetic dipole moments are
Pen = f
Ell (x • Etan) da
(9.72)
meff = ~f
lf.LW
(n X Etan) da
In these expressions the integration is over the aperture, the electric field E," is
the exact tangential field in the opening, and in (9 71) the mode fields are eva!
uatcd at (the center of) the aperture. Tire effeeti"e moments (9.72) are the equiv-
alent dipoles whose fields (9.18) and (9.35) (9.36) represent the radiation fields
of a small aperture in a flat, perfectly conducting screen (see Problem 10.10).
Comparison of (9.71) and (9.69) shows that the dipole moments (9.72) arc only
half as effective in producing a given amplitude as are the real dipole moments
of a source located inside the guide. The effective dipoles of an aperture are in
some sense half in and half out of the guide.
C. Effectif,e Dipole Moments ofApe~twes
On first encounter the effective dipole moments (9.72) arc somewhat mys-
terious. As already mentioned, they have a precise meaning in terms of the elec-
tric and magnetic dipole parts of the multipolc expansion of the fields radiated
through an aperture in a flat perfectly conducting screen (considered later: Prob-
lem 10 10) For small apertures they can also be related to the solutions of ap
propriate static or quasi-static iJoundary-valuc problems. Such problems have
already been discussed (Sections 3 13 and 5 13), and the resalts are appropriated
If an aperture 1s very small compared to the distance over which the fields
change appreciably, the boundary-value problem can be approximated by one in
which the fields "far from the aperture" (measured in units of the aperture di-
mension) are those that would exist if the aperture were absent. Except for very
elongated apertures, it will be sufficiently accurate to take the surface to be flat
and the "asymptotic" fields to be the same in all directions away from the ap
erture. For an opening in a perfectly conducting surface, then, the boundary-
value problem is specified by the normal electric field E 0 and the tangential
magnetic field H 0 that would exist in the absence of the opening. The fields E 0
0
tion in a wave uide or reflection of a lane wave from a screen, for exam !e. But
for the purpose at hand, they are treated as given. To lowest order their time
related to E 0 and the magnetic moment to H 0 • (See, however, Problem 9.20.)

The exact form of the fields around the opening depends on its shape, but
some qualitative observations can be made by merely examining the general
behavior of the lines of force. Outside a sphere enclosing the aperture the fields
may be represented by a multipole expansion. The leading terms will be dipole
fields. Figure 9.4 shows the qualitative behavior. The loop of magnetic field pro-
magnetic dipole whose moment is directed oppositely to H 0 , as indicated by the

1rec IOn o e momen m s own e ow. e magne JC e e ow e p ane
can be viewed as the unperturbed H 0 , plus an opposing dipole field (dashed lines
in Fig. 9.4) whose moment is oriented parallel to H 0 (denoted by mc-J below).
Similarly, the electric field lines above the plane appear to originate from aver-
tical dipole moment p(+J directed along E 0 , while below the plane the field has
1 h lfr · 1
p < l, directed oppositely to E 0 . The use of effective dipole fields is of course
tion is useful qualitatively everywhere, and the effective moments are all that are
needed to evaluate the couplings of small apertures.
The preceding qualitative discussion has one serious deficiency. While it is
correct to state that the electric dipole moment is always directed parallel or
antiparallel to E 0 and so is normal to the aperture, the magnetic dipole moment
0·
tangent plane, and the relative orientation of the aperture and the direction of
H 0 are relevant in determining the direction of ffictt· Since the effective moments
. . .
electric and magnetic polarizabilities of the aperture. The dipole moments can be
written
Petf = EoYEEo (9.73)
(me[[)~ = 2: y~(Ho)r;
larizability tensor. The magnetic tensor can be diagonalized by choosing principal
the dipole is viewed, as shown in Fig. 9.4. If there are fields on both sides of the
Other possibilities can be

424 Chapter 9 Radiating Systems, Mnltipole Fields and Radiation-S!
1_ ) ~I ~
'\{
-J
'/ "- '\.. !"
- ll<'
' "'U
m(+)
p(+)t
Ho_____,_
--
~
m(-l
l•r
I l
E0 1
(a) I h)
Figure 9.4 Distortion of (a) the tangential magnetic field and (b) the normal electric
field by a small aperture in a perfectly conducting surface. The effective diQ_ole
moments, as viewed from above and below the surface, are indicated beneath.
The polanzabilities yr. and 'Y~~ have the dimensions of length cubed. If a
typical dimension of the aperture is d, then it can be expected that the polariz.
aoiiiLies wiu oe a" umes numencal coetnc1ents ot the order 01 umty, or smaller.
The expression (9.72) for Perf can be seen to be of the form to yield such a result,
since Etan is proportional to E 0 , and the two-dimensional integral will give £ 0
times the cube of a len2:th that;, r' ;,tir of th.e :.n " thn
vectorial properties of Petf in (9.72) correspond to (9.73). On the other hand, the
expression in (9.72) for mctr is less transoarentlv of the orooer form even thouoh
dimensionally correct. Some integrations by parts and use of the Maxwell equa-
tions puts it into the equivalent and more satisfying form:
men = 2 { x(n · H) da (9.74)

J
where n • His the exact normal component of H in the aperture and the integra-
' ' tJ-. ,], •L L ' '_, •
r ~· cu~ at'~au~~. H ~o uv" ~v.u~uc wac cu~ ;uvu
between H 0 and mcff is of the general form shown in (9.73). For a circular oneninP
of radius R the effective dipole moments can be taken from the static solutions
of Sections 3.13 and 5.13. The results are
4E0 R 3 8R 3
Perl = -----;:;-- Eo, met£ = --::;- Ho (9.75)
J J
where the signs are appropriate for the apertures viewed from the side of the
surrace wnere ~ ana n are nonvamshing, as can be checked from Fig. 9.4. The
electric and magnetic polarizabilities are thus
4R 3 8R 3
'}'E = - ' ' '}'~ = ' 8aj3 (9.76)
The use of iv.e_din.-.1"

• "£
v~
.u"
LO LOfU -~"aY~"'I'S"· I
HV~~o \.-aU u~ UU\.-\OU IJaCK
• ' " rl
.!L thP
•
pj
".
'nn nt'
ne general tneory
WAA_
~~
was develooed by H. A. Be the t and has h.e.en :'lnnliecl fmitfullv to •'J"niriP :mrl
*Lord Rayleigh, Phil. Mag. XLIV, 28 (1897), reprinted in his Scientific Papers Vol. IV n. 305.
1
H. A. Bethe, Phys" Rev. 66, 163 (1944).

1-iect. !l.b 1-iphencaJ wave 1-iolutwns or the 1-icaJar Wave ~quation 't~::l
":u. ion nrohlPm< Tt i< <i in . ' '. th!lf thP rr ivP
dipole moments of arbitrary apertures can be determined experimentally by elec-
trolytic tank measurements.*
•c •1.. . ,, . _, "
'" <H~ Uo'~ '"
wavegmdes and diffractiOn are lett to several problems at the end ot the chapter.
Other material can be found in the list of suggested reading.
9.6 Spherical Wave Solutions of the Scalar Wave Equation

1n ~napters .) anu '+ spnerica1 narmonic expansions 1or tne sowtions 01 we
Laplace or Poisson equations were used in potential problems with spherical
boundaries or to develop multipole expansions of charge densities and their
fields. Our approach so far for radiating sources has been "brute force," with
creation of the lowest order multi poles more or less by hand. Clearly, treatment
of higher multivoles demands a more systematic avvroach. We therefore turn to
the development of vector spherical waves and their relation to time-varying
sources.
As ;; nrelnrl<e to th~e _1..
11 w:.vte nrohlPm WP ronsirlPr thP sr:1l:.r
·' '"'IS ';Lllv 'VU<'-v-ucoco ;a vco "'i uauvu,
/
wav" "' .r-.. ' !HOLU 'f' <} ~A,
1 D2 1/J
'iPrf! ~ -2 - 2 = o (9.77)
L '"
can be Fourier-analyzed in time as

rx
1/J(x, t) ~ J-x rf!(x, w)e u11t dw (9.78)
with ""r.h FonriPr rom1 '"tisfvina thP HP.lmholt7 w"'"' Prm,.tion
(\7 2 + k 2 )rf!(x, w) = 0 (9.79)

0 0 0 0 0
Wllll K w 1C • ror proo1ems possessmg 'J 'y p< vp<::iLies . some ongm,
it is convenient to have fundamental solutions appropriate to spherical coordi-
nates. The representation of the Laplacian operator in spherical coordinates is
given in equation (3.1 ). The seoaratwn of the an2.ular and radial vanables follows
the well-known expansion
1/J(x, w) = L
l,m
flrn(r)Ylm(O, cf>) (9.80)
where the spherical harmonics Y 1m are defined by (3.53). The radial functions
f. (r) '"tisfv tho> r,.rJ;,.] P(111"tion · of m
d2 2 d 2 l(l + 1)
-+
2
--+k ~
2 fl(r) = 0 (9.81)
rlr r rlr r
With the substitution,

~
1' 1 .. \ / .. \ In o"l\
J [\' J rl/2 [\' J \· 0
"!
~~ R rnhn Prn. 11? 1< ~0 1 <1 1,; /1 0" \· dll 1f){;Q /1 O'i?\
equatiOn (IJ.X 1) IS transformed into

2
[ -d + - d + k2
1 -
dr 2 r dr
- (l +2~?] u1 r - 0
r
()- (9.83)
This equation is just the Bessel equation (3.75) with v = l + 1. Thus the solutions
c. c .. \
W< J I""!_!_ U< v
ftm(r) = A/1'; ft+l!2(kr) + n;;_ Nl+ 112 (kr) (9.84)

r r
It is customary to define spherical Bessel and Hankel functions, denoted by
Mx), n 1(x), hj1· 2 l(x), as follows:
I \ 1/2
Uv\ I I. _{ v\
",--' 2x -' Ci'"'-- /
nt(x) = (2: r2Nt+1!2(x) (9.85)

\ 112
h)12l(x) = ( "'Tr I [hc1!2(x) ± iNt+ln(x)]
.,
<v' n.-mA, "t ~-<)IS Lilt:
' 12\
.lAHIJugaLe 01n 1" X). rrom me senes expansiOns
(3.82) and (3.83) one can show that
il(x) = (-x)ZG !Y(si:x)

(9.86)
1 1
1 d ' 'rr" r'
n 1(x) ·( X_Il II I
X uX X
For the first few values of l the explicit forms are:

cosx e;x
Jo'()
x = sinx
--, n 0 (x) = --- hC 1 l(x) = -.
X X ' o !X
' fv\ "'-'- ICVSA

1 •• \
1.-V>-< smx
J' •/ v2 " [\A}
' 2
h(11l(x) = -e'x
- (1 +-i)
X X
I 3 1 3 COSX I 3 1' smx
h(x) = 13 - -) sinx - .2 n 2(x) = - --- COS X - 3 -----:,--
' 3
A
_e:' -'I
· ix ,. , ~0'.01)
j,(l)fv\ I 1 -'- ~· ~ I
"L
X X x2
I 15 6\ 15 1
h(x) = I- 4 - --o; sinx - I 3 ---
I cosx
x x- x x
' 1" h ' 1" 1\
n,(x) I cosx ) ,;n r
X X x" X_}_
h(ll(x) = e'x ( l + 6i _ 15 _ lSi)

o x x x2 x3
Sect. 9.6 Spherical Wa>e Solutions of the Scala1 Wa~e Equation 427
From the series (3.82), (3.83), and the definition (3.85) it is possible to calculate
the small argument limits (x << 1, l) to be
1 2
. x 1 ~ 2(21x + 3) + ... )
]L(X) ~ (2/ + 1)!!
(
(9.88)
x 1+ 1 2(1 ~ 21)
where (21 + 1)! ! = (21 + 1)(2/ ~ 1 )(21 ~ 3) · · · (5) · (3) · (1 ). Similarly the large
argument limits (x >> 1) are
(9.89)
X 2
ix
The sphet ical Bessel functions satisfy the recursion formulas,
zL(.cr)
X
z;(x) + [1z, 1(x) (1 + l)z 1+ 1(x)] (9.90)

21 1
d
dx [xz 1(x)] = xz1-1(x) ~ 1z!(x)
where z1(x) is any one of the functionsMx), n1(x), hfl(x),h?l(x). The Wronskians
of the vanous pmrs are
1
W(j 1, n 1) (9.91)
The general solution of (9.79) in sphencal coordmates can be wntten

!f!(x) = L [Ai.:lh)il(kr) + A);;{hi2 )(kr)]Y1m(e, 4>) (9.92)
l.m
where the coefficients A b~! and A i~! will be determined by the boundary
conditions.
For reference purposes we present the spherical wave expansion for the out-
going wave Green function G(x, x'), which is appropnate to the equatiOn,
li(x x") (9.93)
in the infinite domain. This Green function, as was shown in Chapter 6, is
G(x, x') (9.94)
The spherical wave expansion for G(x, x') can be o[Jtained in exactly the same
way as was done in Section 3.9 for the POisson equatiOn [see especmlly (3.111)
and text following] An expansion of the form
(9.95)
l.m
. ...
A')!l "'-
-. " -. '
'"
n~mo auu
-~·
substituted into (9.93) leads to an equation for g1(r, r'):

d2 2 d l(l + 1)
~
2
- +--+k-
A, 2 ? gL = - 8(r - r'l (9 CJ(;\
-·
L
~
..
UH:O . we rv . ut uniLeuess at we ongm and
'""'
outgoing waves at infinitv is
".{y r') - AU kr )h(l)( kr_ ) {() ""''
' . 'I
The correct discontinuity in slope is assured if A = ik. Thus the expansion of the
Green function is
_ikJx=.' 00 I
1
,1,.,.
~
Iv v' I = ik 2: jz(krJh) l(kr,)

m -,
2: Y!m(8', cf>')YLm(8, cf>) (9.98)
'
nnr ,J,~<i< <r> f,r h ,J. <I- d.
·.r . 'l"l" d . · ''-'dliif h "'"' h f
wave equatiOn. We now reexamine t e angular unctiOns m or er to mtroduce
some concepts of use in considering the vector wave equation. The basic angular
<, ' _ili '.IL ' U 0> -'1.-,.-.-,, .1.'-L '
w• 'lm\U•'f'l\~-~~1, "'~ UlLne
equation
-
r 1 a1 1
- - - sin 8 -;- + ----;-;;-- a2 - Y,. = l(l + 1)Y,
-----, a=)
(9.99\
-'"' u [}[} ou I Mu- u acp
"'"' " VV 1011 l'dlU VV 11 111 , ics, wis equa1ion can oe wriLLen m we torm:
0
1.,-I lm l~l -t l)Izm (lJ.lUU)

. . 0
.11110 UIHeielll!al upera10r 1., 1.,~ -t Ly -t 1.,;, wnere
L = ~ (r x V) (9.101)
l
is li- 1 times the orbital angular-momentum operator of wave mechanics.

1 ne components ot L can be wntten conveniently in the combinations,
L+ = Lx + iLv = e'"'( _i!__ + i cot 8 __i!__)

\"" "'+'!
I "
L_ = Lx - iLy = e '"'I - - + 1 cot 8 - u . u )
(9.102)
a8 acf>
. a
Lz = -~-
iJcf>
•• .
"'· T
~l"
.I
'J JH
•I
au5um• ' " '
- '-'
, uuu ' '
~- ~
-.-
~
".f •
~
.l'lUUl
definition (9.1011 it is evident that
r · L- 0 (0.103)
.,_
holds as an operator equation. From the exnlicit forms (9.102) it is e~sv t"
that U is equal to the operator on the left side of (9.99).
rwm we exp.iciLtorms ~'f.lUL) ana recursiOn relatiOns ror Izm tne rollowmg
useful re 1 • c..Call..be___e~ t..l:- L ~ •
L+Ylm = Y(l- m)(l + m + 1) Yl,m+l

LYzm - V(l + m)(l m + 1) YL,m-1 (9.104)
Lzizm mYzm
Seet. 9.7 Multipole Expansion of the Eleetromagnetie Fields 429
Fmally we note the followmg operator equatwns concernmg the commutation
(9.105)
where
(9.106)
9: 7 Multlpole Expansion of the Electromagnetic Fields

With the assumption of a time dependence e '"' the Maxwell equations in a
source-free region of empty space may be written
V X E = ikZ0 H, V x H = -ikEIZ0
(9.107)
V·E = 0 V·H = 0
where k = wlc. If E is eliminated by combining the two curl equations, we obtain
forH,
V·H = 0
with E given by (9.108)
Altematively, II can be eliminated to yield

V·E 0
with H given by (9.109)
l
H=-··-VXE
kZ0
Either (9.108) or (9.109) is a set of three equations that is equivalent to the

Maxwell equations (9.107).
We wish to find multipole solutions forE and H. From (9.108) and (9.109)
it is evident that each Cartesian component of H and E satisfies the Helmholtz
wave equation (9.79). Hence each such component can be written as an expan-
sion of the general form (9.92). There remains, however, the problem of orches
tratmg the drfferent components in order to satisfy V · H 0 and V · E 0 and
to give a pure rnnltipole field of order (!, m) We follow a different and somewhat
easier path suggested by Bouwkamp and Casimit. * Consider the scalar quantity
r · A, where A is a well-behaved vector field. It is straightforward to venfy that
the Laplacian operator acting on this scalar gives
(9.110)
*C. J. Bouwkamp and H. B. G. Casimir, Physica 20, 539 (1954). This paper discusses the relationship
among a number of different, but equivalent, approaches to multipole radiation.
430 Chapter 9 Radiating Systems, Mnltipole Fields and Radiation-SI
From (lJ.lOS) "nrl (Q 1()Q) it

' ' 'VLU
\ " ,, •• "") v, \V K)~C'CI) U
The general solution for r • E is given by (9.92), and similarly for r ·H.
We now define a magnetic multipole field of order (l, m) by the conditions,
(M) _ l(l + 1) ) )
r • H 1m - k g1(kr Y 1m(O, </>
(9.112)
r • E);;;l = 0
where
(9.113)
The presence of the factor of l(l + 1)/k is for later convenience. Using the curl
equation in (9.109) we can relater· H to the electric field:
1 1
£ 0K r •H
l
. r · ( V X E) . (r X Vl • E L·E (9.114\
•.., ., 5 ,. oy (':I.JUJ). wim r•H given oy (Y.llL), the electnc held ot the
mag:netic multino]p mn~t ~"tidv
(9.115)
and r · E):;;J = 0. To determine the purely transverse electric field from (9.115),
we first observe that the operator L acts only on the angular variables (0, <f; ).
This means that the radial dependence of El;;;l must be given by g1(kr). Second,
the operator L acting on Y 1m transforms the m value according to (9.104), but
does not change the l value. Thus the components of El;;;l can be at most linear
combinations of Y1m's with different m values and a common l, equal to the l
value on the right-hand side of (9.115). A moment's thought shows that for
L · E);;;l to yield a single Y 1m, the components of El;;;l must be prepared before-
hand to compensate for whatever raising or lowering of m values is done by L.
Thus, in the term L_E+, for example, it must be that E+ is proportional to L, Y 1m.
Wh"t thi~ :.~m · tA ;,. th··• th~ ~~ · . fi ,Jrl ~• .u 1
JC(M) 7 (/rr\T V ffl ~ \

·rm -uc"· · f . <m\'' 'I'!
~
.... ;th r\
_\_7oU.~
u(M) ' ~ v(M)

-,m m
i<T
Rouation(lJ.111)) thPP ;,..finlrloAr. ,,; , , . ~1 ,f
order (l, m). Because the electric field (9.116) is transverse ;;;'·the radius ~ector,
these multipole fields are sometimes called transverse electric (TE) rather than
magnetic.
The fields of an electric or transverse magnetic (TM) multipole of order (l, m)
are specified similarly by the conditions,
(E) _ l(l + 1)
r·E 1m - - Z0 k ft(kr)Y1m(O,cp)
(9.117)
Sect. 9.7 Multipole Expansion of the Electromagnetic Fields 431
Then the electric multipole fields are

H~,~> = ft(kr)LYtm(I:J, c/>) (9.118)
E("l = iZo V X H(E)
lm k lm
The radial function f 1(kr) is given by an expression like (9.113).

Tl1e fields (9.116) and (9.118) are tl1e spherical wave analogs of the TE and
TM cylindrical modes of Chapter 8. Just as in the cylindrical waveguide, the two
sets of multipole fields (9 116) and (9 118) can be shown to form a complete set
of vector solutions to the Maxwell equations in a source-free region. The ter-
rninology electric and magnetic rnultipole fields will be used, rather than TM and
TE, since the sources of each type of field will be seen to be the electric-charge
density and the magnetic-moment density, respectively Since the vector spherical
harmonic, LY1m, plays an important role, it is convenient to introduce the nor-
malized form,*
(9.119)
with the orthogonality properties,
(9.120)
an
(9.121)
or a , , m, m.
By combining the two types of fields we can write the general solution to the
Maxwell equations (9 107)·
l,m (9.122)
E = Zo ~ U aE(l, m)V X j 1(kr)X1m + aM(l, m)g1(kr)X1m J

where the coefficients a"(l, m) and aM(l, m) specify the amounts of electric (l, m)
multipole and magnetic (l, m) multipole fields. The radial functions f 1(kr) and
gt(kr) are of the form (9.113). The coefficients aE(l, m) and aM(l, m), as well as
the relative proportions in (9.113), are determined by the sources and boundary
conditions. To make this explicit, we note that the scalars r · H and r · E are
sufficient to determine the unknowns in (9.122) according to
k (y* Hdft
Vf(l + 1) J
tml • (9.123)
ZoaE(l, m)f1(kr) = - V k
l(l + 1)
I Y *1mr • E dO
*X.;" is defined to be identically zero for I - 0 Spherically symmetric solutions to the source-free
Maxwell's equations exist only in the static limit k--> 0. See Section 9.1.
432 Chapter 9 Radiating Systems, Multipole Fields and Radiation-S!
" ' '

a -raurr; 7
''"llU' -o· U> > ' H dllU > 'I'.. -tree at lVVU 1 aiKf72. 1n
re1110n will therefore nermit a comnlete snecification of the fields inclurlino .-L
termination of the relative proportions of hfll and h}2 l in f 1and g 1. The use of the
scalars r · H and r · E permits the connection between the sources p, J and the
multipole coefficients aE(l, m) and aM(l, m) to be established with relative ease
(see Section 9.10).
Y.CJ roperues OJ Mumpote .Netas; ~nergy

I"
and Angular Momentum of Multipole Radiation
Before considering the connection between the general solution (9.122) and a
localized source distribution we examine the ~•if'., of thf' · ~· ·~ 1~1 m111ti
pun.-
_, £.U /oHL\
V·LLVJ auu
~ m
\J,UUJ· LH
~ o o\ L ...
<Hv uvaL "
LV'"\"'
.'
=-If~
ftlkr) IS proportiOnal to n 1, given by unless Its coefficient vamshes iden- l':Uil'S),

tically. Excluding this possibility, the limiting behavior of the magnetic field for
an electric (l, m) multi pole is
(E) k Ylm
H1m ~ --; L---,:;:-:;- (9.124)
' '
w11t:1t: Lilt: prupurLiuHaiity coeurcien, is CHosen ror laTer convenience. 1 o nna the
electric field we must take the curl of the right-hand side. A useful operator
identity is
iV X L = rV
2
- v( 1 + r :r) (9.125)
~· ' ~' ' <n ' o\ •

UI<O "'" uo:au \ 7 . l l O J I>
(E) -i (ylm)
E 1m ~ -l- Z 0 V X L r 1+ 1 (9.126)
Since (Y1m/r 1+ 1) is a solution of the Laplace equation, the first term in (9.125)
vanishes. Consequently the electric field at close distances for an electric ([ m)
,]<.' ,1
-,-
EIEJ
lm ~ -zv(Yim)
0 l+I (9.127)
r
This is exactly the electrostatic multipole field of S_ection 4. L We note that the
magnenc netu n 1m IS smauer m magnnuae man ~ 1m' IL 0 bV a tact or kr. Hence,
0 _, ""' 0 0
in thP nP.~r '7Clnf'. thf' m~crnPtir fif'lrl of ~n p] · • ~·~ · --; • ie ,;c '~,~h e~~ll»r
·r -o
LilaH wt: '"" 11t:1u. rur wt: • m -r ~O) eviaenuy Ine
roles of E and H are interchanged ;ccording to the transformation,
E(lo:) ~ - z ()
H(M) , HIE) ~ EIMl;zo (9.128)
Tn th» nT
·•inn 7on» (,.Z,. -..;-.., 1-,\ <1.
f~r rl.' ', .1 <:.U. _, T ' •'- ..
oounaary conaiiions imposea. t<or aennneness we consider the example ot out-
nnin n m~uP< i~tP tn r~rli~tinn hu ~ t\, ;~; '1 '""~- 0 ~,~~~ 'T'h~n
-: -or'' " ' .- -

" {0\
uuu j IV" J I> P" upu1 lU lll<O 01 uanKer lUnCLIOn nt -~KYJ. rrom wt:
Sect. 9.8 Properties of Multipole Fields; Energy and Angular Momentum 433
asymptotic form (9.89) we see that in the radiation zone the magnetic induction
for an electric (l, m) multipole goes as
ikr
H(E) __,. (~i)t+l ~ LY (9.129)
lm kr lm
Then the electric field can be written
(9.130)
Since we have already used the asymptotic form of the spherical Hankel function,
we are not justified in keeping powers higher than the first in (1/r). With this
restriction and use of the identity (9.125) we find
E/~l = ~Z0 (~i)l+ 1 ikr [

:, n X LY1m ~k
1 (rV' 2 ~ V)Yim J (9.131)
where n = (r/r) is a unit vector in the radial direction. The second term is evi-
dently llkr times some dimensionless function of angles and can be omitted in
the limit kr >> 1. Then we find that the electric field in the radiation zone is
ECEJ
lm
= Z UHCEJ
/m
x n (9.132)
v-;here Hf,!l is given by (9.129). These fields are typical radiation fields, transverse
to the radius vector and falling off as r 1 • For magnetic multipoles the same
relation holds because the Poynting vector is directed radially outward for both
types of multipole.
The m ultipole fields of a radiating wur ce can be used to calculate the energy
and angular momentum carried off by the radiation. For definiteness we consider
a linear superposition of electric (l, m) multipoles with different m values, but all
having the same l, and, following (9.122), write the fields as
Hl "' a (l
= L..JF:, h( 1 )(kr)e- 1'"'
m)Xlml
m (9.133)
For harmonically varying fields the time-averaged energy density is
0
u ~ (E • E* + Z 2 H • H*) (9.134)
In the radiation zone the two terms are equal. Consequently the energy in a
spherical shell between r and (r + fir) (for kr >> 1) is
f.Lodr
dU = -
2k
- 'V * ') ( )
2 L... aE(l, m aE l, m
m,m'
JX * 1 m· • X 1m dD (9.135)
where the asymptotic form (9.89) of the spherical Hankel function has been used.
With the orthogonality integral (9.120) this becomes
dU = fLo
-
-v IaE(l, m) l2
k? L... (9.136)
dr 2 - m
434 Chapter 9 Radiating Systems, Mnltipole Fields and Radiation-SI
munipo1es rne sum over m becomes a sum over l and m and [aE[~ becomes
la£1 2 + laMI 2 . The total energy in a spherical shell in the radiation zone is thus
an incoherent sum over all multipoles.
The time-averaged angular-momentum density is
1
m = - Re[r X (E X H*)] (9.137)
2c 2
1 ne Ulple cross proauct can be exoanded and the electnc field ( 9.133 l sub,titn+a~
m = f.Lo Re[H*(L · H)] (9.138)

2w
Then the angular momentum in a spherical shell between r and (r + dr) in the
radiation zone is
f:!:_our u "' *" " n ·' I r• ..,. •*..,. Jn.

m,m
With the explicit form (9.119) for X 1m, (9.139) can be written
dM~~
-
dr
-f.Lo-2
2wk
Re
m,m'
"'*aE(l, m ,)aE(l, m)I*
L.J Y lm'LY1m d[! (9.140)
From the properties of L Y 1m listed in (9.104) and the orthogonality of the spher-
._, . . . . .
,..,m ,.a, vv" Lilt: wuuwmg ''"'l'' . ,, tur Llle':::::afl:_eSian components
ot dlVl!dr:
dMx = 1-Lko 2 Re
dr 4W
2
m
[Y(l ~ m)(l + m + 1) a';;(l, m + 1)
(9.141)
+ Y(l + m)(l ~ m + 1) a';;(!, m ~ 1)]aE(l, m)
dMy = f.Lko 2 Im
dr 4w
L m
[Y(I ~ m)(l + m + 1) aW, m + 1)
fO 1 A')\
\ . ~,
(9.143)
These equations show that for a general/th-order electric multipole that consists
of a superposition of different m values only the z component of angular mo-
mentum is relativelY simule
or \~·1'DJ ana \~.l.:lOJ snows that

dM m dTT
(9.144)
dr w dr
independent of r. This has the obvious quantum interpretation that the radiation
from a multipole of order(/, m) carries off mh units of z component of angular
momentum per photon of energy hw. Even with a superposition of different m
values tne same mteroretatwn of 19.143\ holds with each mul · of .-lPfinite
Sect. 9.8 Properties of Multipole Fields; Energy and Angular Momentum 435
m contributing incoherently its share of the z component of angular momentum.

Now, however, the x andy components are m general nonvamsnmg, wnn mm-
tipoles of adjacent m values contributing in a weighted coherent sum. The be-
• • • • • -; o• • • /C. A· /C. ·~ ' ~- ,-
lldVlUl l.OUHI 111 \OI,.L"+U) d U U : - e = = uy w \7·~•~)-\J·~·v;"
in the quantum mechanics of a vector operator and its representation with respect
tn h~c;c nf r2 ~nrl T *. 'Th<> ~nm•l~r 1tnm nf mnl ' ,1 , fip]ck "" -'
c ·o
a classical example of this behavior, with the z component being diagonal in the
(l m) multioole basis and the x and v comoonents not.
The characteristics of the angular momentum just presented hold true gen-
erally, even though our example (9.133) was somewhat specialized. For a super-
position of both electric and magnetic multipoles of vanous lf, m) values, the
angular momentum expression (9.139) is generalized to
dl\'! - J.Lo RP ) ' rn*O' rn'\n.Jl rn\ -1- n*.O' rn'\n .([ m)l ( (1 . X .. )*X d[!
dr 2wkL t;;, ' J
l',m'
./' -lc *
-"-1m ''""
(0
\- . 1 !1<:1
The first term in (9.145) is of the same form as (9.139) and represents the sum
01 me ewctric ana magneiic munip01es separately. "'" '"'m " au u1tc1-
terence between electric and magnetic multipoles. Examination of the structure
-~' .1 ' .1 .L d. d. ' " ' L .1 ' .-1
V~ HO D HH~5~ U~ OHV '0 """ ' •o
netic multipoles whose l values differ by unity. This is a necessary consequence
of the parity properties of the multipole fields (see below). Apart from this com-
plication of interference, the properties of dM!dr are as before.
The quantum mechamcal mTerpreTation Of1'1.l '1-'1-) concernea me z compo-
nent of angular momentum carried off by each photon. In further analogy with
_, . .. " . .,,_ -~·'- .i
"' "''-'"' , we wvutu ·r- '"" tauv "'~ "'1~~ ~ ~
mentum to the square of the energy to have value

MW - (M~ + ~ +~q - l(l + 1) (9.146)
7- U2 - ul
But from (9.136) and (9.141)-(9.143) the classical result for a pure (1, m) multi-
pole IS
\"7-.l"t/)
"~~ " <v> uuo ~- 'l' ·o

fields for a single photon. If the z component of angular momentum of a single
r- r ·J "J r
ponents be uncertain, with mean square values such that (9.146) holds. On the
other hand, tor a state ot the radiatiOn tield contammg many pnotons l me c1as-
sica! limit), the mean square values of the transverse components of angular
•tum canoe negng= ··r- ""UJ" ' " " oyua1c v• '"" "· "·
*See for example, E. U. Condon and G. H. Shortley, The Theory of Atomic Spectra, Cambridge
University Press, Cambridge (1953), p. 63.
436 Chapter 9 Radiating Systems, Multipole Fields and Radiatiou-SI
Then the classical limit (9.147) applies. For a (1, m) multipole field containing N
photons it can be shown* that
[MCql(N)f N 2m 2 + Nl(l + I) - m2
(9.148)
Ihrs contams (9.146) and (9.147) as hmrting cases.

The quantum-mechanical interpretation of the radiated angular momenturn
per photon for multipole fields contains the selection rules lor multJpolc transi-
tions between quantum states. A multipole transition of order (l, m) will connect
an initial quantum state speciJied by total angular momentum J and z component
M to a final quantum state with .T' in the range II - ll < .T' < .T + I and M'
M - m. Or, alternatively, with two states (J, M) and(.!', M'), possible multipole
transitions have (I, m) such that /.! - I' I < l < J + J' and m = M - M'.
To complete the quantum-mechanical specification of a multipole transition
it is necessary to state whether the parities of the initial and final states are the
same or drlferent. The parity of the initial state is equal to the product of the
parities of the final state and the multipole field. To determine the parity of a
multipole field we merely examme the behaviOr of the magnetic field H 1m under
the parity transformation of inversion through the origin (r--> -r). One way of
seeing that 111m specifies the paiity of a ttl ultipole field is to recall that the mter-
action of a charged particle and the electromagnetic field is proportional to
(v ·A). If H 1m has a certain parity (even or odd) for a multipole transition, then
the corresponding A 1m will have the opposite parity, since the curl operation
changes parity. Then, because v is a polar vector with odd parity, the states
connected by the interaction operator (v ·A) will differ in parity by the parity of
the magnetic field H,m·
For electric multipoles the magnetic field is given by (9.133). The parity trans-
formatiOn (r--> r) IS eqUivalent to (r--> r, (j--> 1T 0, tj>--> tj> + 1r) in spherical
coordinates. The operator Lis invariant under inversion. Consequently the parity
pwper ties of H 1m for electric multipoles are specified by the transformation of
Y 1m(O, 4> ). From (3.53) and (3.50) it is evident that the parity of Y 1m is ( -1 t Thus
we see that the parity of fields of an electric multipole of order (l, m) is (-1) 1•
Specifically, the magnetic induction H 1 has parity ( -1l, while the electric field
m
E 1m has parity ( -1r+ 1, since E 1m = iZ0 V X H 1mlk.

For a magnetic multipole of order (l, m) the parity is ( -1) 1+1. In this case the
electric field E 1m is of the same form as H 1m for electric multipoles. Hence the
parities of the fields are just opposite to those of an electric multipole of the same
or er.
Correlating the parity changes and angular-momentum changes in quantum
tmnsitions, we see that only certam combmatwns ol multrpole transrtwns can
occur. For example, if the states have J = ~ and J' = ~. the allowed multipole
m ders arc l l, 2. If the pa1ities of the two states are the same, we see that
parity conservation restricts the possibilities so that only magnetic dipole and
electric quadmple transitions occur. If the states differ in parity, then electric
dipole and magnetic quadrupole radiation can be emitted or absorbed.
*C. Marette De Witt, and J. H. D. Jensen, Z. Naturforsch. 8a, 267 (1953). Their treatment parallels
ours closely, With our classiCal multipole coefficients aE(l. m) and aM(/. m) becoming quantum-me·
chanica! photon annihilation operators (the complex conjugates, a~ and a,~. become Hermitian con·
Sect. 9.9 Angular Distribution of Multipole Radiation 437
9.9 Angular Distribution of Multipole Radiation

For a general localized source distribution, the fields in the radiation zone are
given by the superposition
kr t,m (9.149)
E~Z0 HXn
The coefficients aE (l, m) and aM(l, m) will be related to the prop~Jrti~Js of the
source in the next section. The time-averaged power radiated per unit solid angle
lS
dP ~ Z 0 'V . t+ 1 )
dO ~ k 2 L.. ( ~t) [a..,(l, m)X1m X n + aM(l, m X 1m] (9.150)
2 /,m
Within the absolute value s1gns the dimensions are those of magnetic field, but
the polarization of the radiation is specified by the directions of the vectors. We
note that electric and magnetic multipoles of a given (l, m) have the same angular
dependence but have polarizations at right angles to one another Thus the mul-
tipole order may be determined by measurement of the angular distribution of
radiated power, but the character of the radiation (electric or magnetic) can be
detcunincd only by a polatization measurement.
For a pure multipole of order (l, m) the angular distribution (9.150) reduces
to a single term,
dO 2k2 a 'm tm
(9.151)
From definition (9 119) of X 1,. and properties (9 1 04), this can be transformed
into the explicit form:
dP(l,m) Zo la(l, m)l2 W ~ m)(l + m + 1) 1Yt,m+ll 2
_, lfl _, -\(I _j. 1 \ IV. . 12 -1. ~2 Iv 12
2\' "/\' -/ - ,,m _.11 m
(9.152)
Table 9 1 lists some of the simpler angular distributiOns.
The dipole distributions are seen to be those of a dipole oscillating parallel
to the z axis (m = 0) and of two dipoles, one along the x axis and one along the
y axis, 90° out of phas~J (m + 1). The dipole and quadrupole angular di~tribu-
tions are plotted as polar intensity diagrams in Fig. 9.5. These are representative
of I = 1 and l = 2 multipole angular distributions, although a general multipole
Table 9.1 Some Angular Distributions: IX,m( e, </>)I'
I 0 ±1 ±2
1 3
(1 + cos'll)
2 5
l 67T (1 - 3 cos2 0 + 4 cos4 0) - 5 (1- cos4 e)
Quadrupole 16 /l
438 Chapter 9 Radiating Systems, Multipole Fields and Radiation-Sf
I
- \
\ I
I \- 1/ " j l
I-
\. - 1--
)__
_/
)I\
"'-... __) -; \
l = 2, m = !2
l = 2,
m =!I
-
"" /
(
\ \\ I I
"-.. 1/ /
/ -.......
/ J1\
(
.__....... /
l = 2, m = 0
'\. "-.,...
""'
v
c '\f-
~
L_ 1\.. ./
) - 1--
l = 1, m = 0
\ /
1 1 +t
'
Figure 9.5 Dipole and quadrupole radiation patterns for purejl, m}_ multipoles.
distribution of order l will involve a coherent superposition of the (2! + 1) am·

pmuoes ror onrerent m, as shown m (Y.150).
It can be shown by means of (3.69) that the absolute squares of the vector
OjJH~llCi11
H<H , uucy we sum rme,
1
2: IXzm(&, W = Ll + (9.153)
m "'
LJ. .,'~
A' '.L
·~~-~--~" ~•ou.
.,
V>H
'
V,_ l>VLlV]J''- HV<U
. ~ .
a OVU''-'- <HaL
c
V1
a set of multipo]es of order l with coefficients a([ m) in-' -' of.., o11nPr·
Sect. 9.10 Sources of Multipole Radiation; Multipole Moments 439
posed incoherently. This situation usually prevails in atomic and nuclear radiative
transitions unless the initial state has been prepared in a special way.
The total power radiated by a pme multipole of order (l, m) is given by the
integral of (9.151) over all angles. Since the X 1m are normalized to unity, the
power radiated is
2
e
P(l, m) = Zo Ia(l, m) l
2
(9.154)
For a general somce the angular distribution is given by the coherent sum (9.150).
On integration over angles it is easy to show that the interference terms do not
contribute. Hence the total power radiated is just an incoherent sum of contri-
butions from the different multipoles:
(9.155)
9.10 Sources of Multipole Radiation; Multipole Moments

Having discussed the properties of multipole fields, the radiation patterns, and
the angular momentum and energy carried off, we now turn to the connection
of the fields with the sources that generate them. We assume that there exist
localized well-behaved distributions of charge p(x, t), current J(x, t), and intrinsic
magnetization .M,(x, t). Furthermore, we assarne that the time dependence can
be analyzed into its Fourier components, and we consider only hannonically
varying sources,
.M,(x)e-iwt (9156)
where it is understood that we take the real part of such complex quantities. A
more general time dependence can be obtained by linear superposition (see also
The Maxwell equations for E and H' = B/ /Lo are

V. H' 0, V X E - ikZ0 H' 0
(9.15 7)
V · E = p!E0, V x H' + ikE/Z0 = J + V X .M,
with the continuity equation

iwp = V · J (9.158)
It is convenient to deal with divergeneeless fields. Accordingly, we use as field
variables, H' and
E' E+ J (9.159)
WEo
In the region outside the somces, E' reduces to E and H' to H. In terms of these
fields the Maxwell equations read
i
V · H' = 0, V X E' - ikZ0 H' = - V X J
V · E' = 0, V X H' + ikE'/Z0 = V X .M.

440 Chapter 9 Radiating Systems, Mnllipole Fields and Radiation SI
The curl equations can be combined to give the inhomogeneous Helmholtz wave
equations
and (9.161)
These wave equations, together with V · H' 0, V · E' 0, and the curl equatiOns
giving E' in terms of H' or vice versa, are the counterparts of (9 108) and (9 109)
when sources are present.
Since the multipole coefficrents m (9.122) are determmed according to
(9 123) from the scalars r · H' and r · E', it is sufficient to consider wave equations
for them, rather than the vector fields E' and H'. From (9.110), (9.161) and the
vector relatiOn, r · (V X A) - (r x V) ·A iL ·A for any vector field A, we find
the inhomogeneous wave equations
2
(V' + k")r · H' iL • (J + V X :At) (9.162)
The solutions of these scalar wave equations follow directly from the develop-
ment in Section 6.4. With the boundary condition of outgoing waves at infinity,
we have
iklx-x'l
r · H'(x L' · [J(x') + V' x .M.(x')) d 3 x'
47T lx - x' I (9.163)
r • E'(x)
z0 k { e;klx-x'l
J
[
L ·:M:(x') +
1
V' x J(x')
J d x'
3
To evaluate the multipole coefficients by means of (9.123), we first observe that

the requirement of outgoing waves at infinity makesA)2 l = 0 in (9.113). Thus we
choose fz(kl) g1(k1) ftjll(kf) in (9.122) as the representation of E and H
outside the sources. Next we consider the spherical wave representation (9 98)
for the Green function in (9.163) and assume that the point xis outside a spherical
surface completely enclosmg the sources. '!hen m the mtegratwns m (9.163),
r< = r', r> = r. The spherical wave projection needed for (9.123) is
1 { * e;klx -x'l
47T f dQ Y . ,(ll, </>) lx x' I (9.164)
By means of this projection we see that aM(l, m) and aE(l, m) are given in terms
of the integrands in (9.163) by
The expressions in (9.165) give the strengths of the various multipole fields
outside the source in terms of integrals over the source densities .I and .M. They
Sect. 9.10 Sources of Multipole Radiation; Multipole Moments 441
can be transformed into more useful forms by means of the following identities:
Let A(x) be any well-behaved vector field. Then
L · A = iV · (r x A) (9.166)
L · (V x A) = zV (r · A ·2 ) ~ -iiJ
- (r V · A) 2
r iJr
These follow from the definition (9.101) of Land simple vector identities. With
A= .M, in the first equation and A= J in the second, the integral for aE(l. m) in
(9.165) becomes
k I . '~
U£~t, ITt) I ltV'-' ) 1 tm • • ~· "' Jill,)
v .,. lJ
1 ic 1 iJ
-r- ,-., v-lr • J) r p) a·x
" " ' u•
.L • mP hovP HOPrJ fQ,1 <;R\ tr>
•? -r •?
•U T ;n tPrmo ,-,f
...-. TToP ,-,f r< 'f thPC>rPm
vu "'"'. C'-'UU 1 <OfHOO,<OO V UJ "- 0 W UU'-' a HHC

0 tal1Vll UJ !-'01LO Vll CH\:0
third term casts the radial denvatJVe over onto the sphencal Hessel tunctwn. The
..
resun for the ·" · mul• · ,/, zt is
i) . . .
k2 ( cp , L'l tV") J T uc~ 1 • d)} t~ K.f)
a£U, m) j Yim a·x ('J.lbl)
' v '\' 1) ikV • (r X .M,)jJ(kr)
The analogous manipulation with the second equation in (9.165) leads to the
.1 •• . 1. .r.rc .
r i)
"2 r V • lf X J)]tlKT) -t- V • .M.
v•
LT}tl Kr )J
d·x
aM(l,m)
iV/(l + 1) J Yim ~ k2
(r · .M)Mkr)
~
(9.168)
These results are exact expressions, valid for arbitrary frequency and source size.
For many applications in atomic and nuclear physics the source dimensions
are very small compared to a wavelength (krmax << 1). Then the multipole co-
efficients can be simplified considerably. The small argument limit (9.88) can be
used for the spherical Bessel functions. Keeping only the lowest powers in kr for
terms involving p or J and .M,, we find the approximate electric multipole
coefficient,
ce+ 2 l + 1
aE(l, m) = i(2l + 1)!! (-~- r2 (Qtm + Q/m) (9.169)
where the multipole moments are
Qlm = I rtY'fmP d 3X
and (9.170)
Q' =
tm
~ik
(l + 1)c
I rtY* V · (r x .M,) d 3 x
tm
The moment Q 1m is seen to be the same in form as the electrostatic multipole

moment qtm (4.3). The moment Q!m is an induced electric multipole moment due
to the magnetization. It is generally at least a factor kr smaller than the normal

moment Q1m. For the magnetic multipole coefficient aM(l, m) the corresponding
lonP-w~v<>1Pnctth · 1~ti"n i<
aM(l, m) = (
ik/+2
2! + l)!!
c 1r2
-
+
1
- (Mtm + M/m) (9.171)
where the magnetic multipole moments are
Mlm = -
1
1
_,_
1
(I*
rY Lm V · (r X J)d3X,
and (9.172)
M'l m--
-I r Yim V · .M. d x
1 3
In contrast to the electric multipole moments Q1m and Q/m, for a system with
intrinsic magnetization tne magnetic moments M 1m and M tm are generally of the
same order of magnitude.
<H lH\0 lUllg· vva ··o . IillliL we 'e"' Cltoany tnat electric munipo!e news are
related to the elerJrir-rh~r<TP. dP.n<itV n Whilf> thP m~<TnPtir mn1tin{)1P fip1..J,
' ·~
determined by the magnetic-moment densities, (r x J)/2 and .M..
Y.ll mutupme l(aazatwn m Atoms ana Nuclei
Attnougn a rull discussiOn ot radiative transitions m atoms and nuclei requires a

-m,•0h~~;~nl + +~ ,];+ ' t. ,1 ..1 c.
T ' '1 0}' VV ~ -Y H V H > VUl
classical formulas by means of semiclassical arguments and simple estimates of

thf' .u. ''ivP. mnltin"1P 1t< Pird nf ~11 •a .~+ +t +t • ;,· .L
-,- .'
bility r (reciprocal mean life) for emission of a photon of energy hw is given by
the radiated power divided by hw. From (9.154) for the power and (9.169) and
(9.171) for the amplitudes aE and aM in terms of the long-wavelength multipoles,
we nna me transitiOn probability tor an electnc mult!pole (l, m),
r m - wZokzt l +1
E(l, ) - 2hf (21 + 1)! !F )jQim + Q/ml 2 (9.173)
l
For a magnetic multinolP. n, + n• ~ (1/r)fM. + M' 1
The effective multi pole moments can be e~timated as -~o order of magnitude
_L
as follows. Suooose that for the svstem under consideration thP .u.
is e, the effective mass of the radiating constituents is m, and the effective size is
n. l_ll_en-':".e enecuve magnenzanon Is _l.Jll. U(efiJmK'), where entm IS !fl_e ~-
fective magnetic moment of the constituents. The most naive estimates of the
HI
'T . are wen
_, hw
I \Lim I u~en J; IlL Lm I u 2 eK J
me
and (9.174)
~ IM,. + M: .. l- 0 eh Rt-I
e me
Sect. 9.11 Multipole Radiation in Atoms and Nuclei 443
With these order-ot-magmtude estimates some qualnanve reamres or atomic ana

_,
· r:'lclilltive tr:'lnsitions C:'ln be abstracted. In atoms and in nuclei the tran-
sition energies hw are invariably small compared to the rest energy mc2 of the
• I~~·, . • •1 • TC1 •
\CUll> • VV\0 UlU> M;c Llldll'.!iml ,...__ l'.!lml 1> a uv~•ou• r •
multioole transitions of order l( denoted bv El) are dominated by the transitional

charge density, with negligible contribution from the "magnetization charge."
.-.. <L _,,_ -, ..1 • .1<.' .1 . ' fAd[\ ,J1,-].,
~u "'~ ~'"~' ·~, ·r· \' • / ."'' 'J
'
rable contributions from the orbital and intrinsic magnetizations.
In atoms the electrons arc the radiating constituents. The size of the system
is R = O(a 0 /Zen). where a 0 is the Bohr radius and Ze11 is of order unity for valence
electron transitions and of order Z for K- or L-shell x-ray transitions. From
(9.174) the relative size of the magnetic multipole moments with respect to the
electric of the same order lis [M[ /c[ Q I - O(h/mcR) - O(Zerr/137). For the same
;+· +J..n + ;+:~~ ,. . • • ··;+;n" ,.,;11 hn in thP r"tin
'bJ.
\
,... {1\ f 72
~ e (9.175)
0
r E(l) (137?
Onlv for x-rav transitions in heavy elements are magnetic multipoles even re-
motely competitive with electric multipoles of the same order. [Note, however,
.,- ' ' ·' ' ' . ,,__ D1 f. +l,
ll!al LIIC JYI< Lie lldVIO Ulv lJP}- ~ t"~v~· "~" ·~ ... ~ ~· ~·
same/.]
Of interest is the relative size of transition probabilities for multipoles dif-
fering by one unit in order. Ignoring factors of order unity, we see from (9.173)
and (9.174) that
l E,M(l + 1)
= O(k2 R 2 ) (9.176)
l E,M(f)
r, •• . th +r~ w; t; ~~ n~ urhilP thP <i7P i<
~rn ~· ' 72. ~ ,2/(117)2
R = 0(137 h/mcZc~1 ). We thus find kR = O(Zcn·/137) and the ratio for successive
bl multipoles IS ot the same order as ('1.1 /) ). ror atomic LransiLions in wnicn we
an!!ular-momentum selection rules nermit several multiooles the lowest multi-
pole generally dominates. For example, if the initial and final angular momenta
1 n 1 ',, • ,, ' · ''- .11. ..1 •• • •
aLC J i a11u J i a11u Ulv HaVv Lllv "VV '"" 1. u•:•J, '"~ uu~ ~
are £1 and M2. The £1 transition will dominate by a factor of order (Zett/137) 4 .
If the parities are the same, the allowed transitions are Ml and E2. Now the two
,;+;~ nrnho'-''' '>e ~n~J..
. . ·'- · ·
']
'- -'-'
, ;,t, •· 'L' L'
smaller than for opposite parities. In atoms the dominant transitions are £1; high
angular momentum states de-exCite by a cascade or bl transmons, II at au
no<sihlP
In nuclei the situation is somewhat different. Successive multipoles of the
same type still ooey me estimate ~'i:TlUJ,-uutllie"u;,.,, · · ~udgic> vary · ·-
icantly. With the nuclear radius R = 1.4 A 113 X 10- 15 m as the effective size,
' " mn ].,~,n frO< rJ;_ Jl\lfnV\1 Ltll3f111fl p, . v"rv from" fpw kPV
l' 7T ' ~
to several MeV. In heavy nuclei, this corresponds to a range, kR < 10- 4 -10- 1 .
bvidently, tor energetic nuclear transitiOns successive muuip01es 01we same Lype
are not as suonressed as in atoms. For low ener!!ies however the suppression of
rate with multipole order is dramatic. M4 isomeric transitions with energies of
c-cn;;,_" • ' 1' .ro. TL ~1 '
LLO<O UlUvl Ul lVV 1\.v V Ul 110" ICdlJ lld Vv ll -~- Vl L~ULOo L
101 :uc '""'uvc tu transition rates oi rne same oruer, and for ~n
electric multipole of one higher order relative to a magnetic transition, are
r M(/) nm ') A -213\. r £(/ + 1) n, (fiw[MeV)) A

2 413
\ /A
In these estimates we have taken the effective magnetization to be roughly

3 efi/mNR 3 , with a g factor of 3 to account for the magnetic moments of nucleons.
nnr >t ,f ·~ ~1. • . . • •.
ctiuautc tu ,pecia, ptopenies or me nuc1ear srmes anu mteracnons. ln light to

medium mass nuclei, £1 transitions are strongly suppressed by the isospin sym-
metry of nuclear forces, at least at low energies. Ml transitions are far commoner
than El transitions and_iust as intense. In rare earth and__transur>~ni~ nnrJ,; ""'
. . ,J"+ 1 (\(\ ' ~L L P ' •
.~ OU _0 CHUU VU< H~ •o~ V< >ll'\lllllC3ll{
static and transitional quadrupole moments in these nonspherical nuclei. If al-
lowed by spin-parity, £2 transitions then compete favorably with M1 transitions.
A proper quantum-mechanical treatment of multi pole radiation can be found
in Rlott and ('h~ntPr YTT
r
A tn nnrlP~r ; •.· •no n :~ "
~u
in the References and Suggested Reading at the end of the chapter.
(} 12 /V/11 1/o Undjatinn_ on
As an illustration of the use of a multipole expansion for a source whose dimen-

. -Ll ' ,; ' • • •
O<Vuo U1V 1UV1V •v VVa ' VVIO lHIO ' llUJIJ a lflJll,
lmear. center-ted antenna, as shown in Fig. 9.6. We have already given in Section
9.4 a direct solution for the fields when the current distribution is taken to be
sinusoidal. This will serve as a basis of comparison to test the convergence of
thP ,]+" ~1. . Ulo tho t F -1 H . C.
''" • • • u
~UIL) . <. \UIL)o auu lV uavc a smau gap aLILS center so waLIL can De SU!laDJy
z
Figure 9.6 Linear, center-fed
tt antenna.
Sect. 9.12 Multipole Radiation from a Linear, Center-Fed Antenna 445
excited. The current along the antenna vanishes at the end points and is an even
function of z. For the moment we will not specify it more than to write
I(z, t) - I( lz I)e <w', (9.178)
Since the current flows radially, (r X J) 0. Furthermore there is no intrinsic

magnetization. Consequently all magnetic multipole coefficients aM(l, m) vanish.
To calculate the electric multipole coefficient a£(!, m) (9.167) we need expres-
sions for the charge and current densities. The current density J is a radial current,
confined to the z axis. In spherical coordinates this can be written for r < (d/2)
J(x) = r l(r)2 [o(cos (!- 1)- o(cos (! + 1)] (9.179)

27rr
where the delta functions cause the current to flow only upward (or downward)
along the z axis. From the continuity equation (9.158) we find the charge density
p(x) = ~ dl(r) o(cos (!- 1) - ;ccos (! + 1) (9.180)

r
ae(l, m) , r
These expressions for J and p can be inserted into (9 167) to give
Jrd/2 [
drj kriz( kr )J(r)
(9.181)
X I d[! Yim[o(cos (!- 1) - o(cos (! + 1)]
The integral over angles is
I d[! = 27rom.o[Y10 (0) - Y 10 ( 7r)]
showing that only m = 0 multipoles occur. This is obvious from the cylindrical
symmetry of the antenna. The Legendre polynomials are even (odd) about(!=
7r/2 for l even (odd). Hence, the only non vanishing multipoles have l odd. The
the angular integral has the value,
f df! ~/41f(2l I 1), l odd, m 0
With slight manipulation (9.181) can be written
aE(l, 0) = !5_ [47r(2l + 1)]1'2 {d/2 {- !!__ [rh(kr) diJ

27r l(l + 1) Jo dr dr
(9.182)
+ riz(kr)( ~:; + k 2
I)} dr
To evaluate (9.182) we must specify the current I(z) along the antenna. If
no radiation occurred, the sinusoidal variation in time at frequency w would imply
a sinusoidal variation in space with wave number k = wlc. But as discussed in
Section 9.4 .B, the emission of radiation modifies the current distribution unless
.J.Jtl l'honiPr Q R, .
. "' "· ..... . ..I ,
_..
the antenna is infinitely thin. The correct current J(z) can be found only by solving
a complicated boundary-value problem. Since our purpose here is to compare a
multipole expansion with a closed form of solution for a known current distri-
omion, we maKe llle same assumption aoout l(Z J as m ;)ectwn ':I.'+.A namelv
I(z) =Ism 2- klzl )

. (kd (9.183)
TisthP~
"" "nrl thP nhooP ;,. rhn,Pn In~ tha
r .
"'"
v · <ll lllC: <OHU> V<
. vv nn a sinusoiua1 curren L llle secona part of
lllC aT
the integrand in (9.182) vanishes. The first part is a perfect differentiaL Conse.
quently we immediately obtain, with I(z) from (9.183),
II? ?
1 41TUL + 1l kd kd
U£~l, V)
"'- u. l]z • l OQQ (9.184)
"'' ·~'
"
'J
~
"- "-
Since we wish to test the multipole expansion when the source dimensions
are comparable to a wavelength, we consider the special cases of a half-wave
<UHC:lllld ~KU 1TJ anu a <UH-wave anrenna (Ka L1TJ. 1 ao1e ':J.L. snows the 1 1
coefficient for these two values of kd, along with the relative values for l = 3, 5.
From the table it is evident that (a) the coefficients decrease rapidly in magnitude
as l increases, and (b) higher l coefficients are more important the larger the
source dimensions. But even for the full-wave antenna it is probably adequate
w Keep omy t 1 ana t j in the angular Oistnouuon ana certamly adequate
for the total power (which involves the squares of the coefficients).
With only dipole and octupole terms in the angular distribution we find that
the power radiated per unit solid angle (9.150) is
2
~p 71 11 ml2 1'1 "'
• vI
d[! 4kL
T.V
U£\~,
V6aE(1,0)
__1_X'.
o,u \ . ..,
(0 1Q<:\
The various factors in the absolute square are
3
ILY 10 1 smell
'+1T
63
I LY3,0 12 = 161T sin 2 11(5 cos2 11 - 1) 2 (9.186)
jVLl
~._,, 1,0). ' (LI 3,0) sm-up cos-u 1)
81T
Table 9.2 M"' ' ,, n '"' . ''" fm T ;nPor AntPnno
'·-' -"'" ' m~, -.-n-•

I'> n"
VJIUE\~,
n
UJ UE\~, VJ'UE\~, VJ
'6! x w- 2
1T
\I·;;t_
4.95 1.02 x w- 3
r:- I
L1T v 071' J U.jL4L Lj'J X 1U
u
Sect. 9.12 Multipole Radiation from a Linear, Center-Fed Antenna 447
With tnese angular tactors l'J.li:S)) becomes

' ~ T? I ~ \ r;:; 2
-~ A>
cu = A -'Loo' -'?T sin 02 1- I!.. "E\-'• UJ (5 ze- 1) (9.187)
df! 7T3 8 \/8 a,.(l 0) cos
where the factor A is equal to 1 for the half-wave antenna and ( 7T 2/4) for the full
wave. The coefficient of (5 cos 2 0- 1) in (9.187) is 0.0463 and 0.3033 for the half-
,-1 " " 11
.•.
A numerical compariso~ of-[he exa~~ .and approximate angular distributions,
(9.57) and (9.187), is shown in Fig. 9.7. The solid curves are the exact results, the
dashed curves the two-term multipole expansions. For the half-wave case (Fig.
9.7a) the simple dipole result [first term in (9.187)) is also shown as a dotted
curve. The two-term multipole expansion is almost indistinguishable from the
exact result for kd - ?T. Even the lowest order approximation IS not very tar ott
;n th;c <'QOP Pr>r thP fnll.mQup (p;<T 0 7h\ thP rJ;nr>lP ;mQt;r>n ;o
. . ·o. ~,- . -r r
evidently qmte poor. But the two-term multlpole expansiOn 1s reasonably good,
. . .
amenng oy less man )/o m me regiOn or apprecm01e ramanon.
The total oower radiated is according to (9.155).
n Zo "' 1 /1 "'12 {() 1 0 0 \

\ . Jj
1
2k2 -~ I"E\'• VJI
l odd
/(
..,-
.,
......
...._
// '\ ~
(
'
I I
~
\\
, ,_//
\\..
'• ~/.
J/
\
\. /
I
I
.·/" •, / '\
,/I \-.. ,,I \ .
/I \\_
I
'
.,, __
'
\\ j/ J
.... /'' ~
u_ "
,u,
_, ' ~-
Fi~ure 9.7 Comparison of exact radiation patterns (solid curves) for half-wave
(kd = 1r) and full-wave (kd = 27T) center-fed antennas with two-term multipole
;nn< (rl"<hPrl ~llrVP<l For thP h"lf-w"vP. th~e rJ;nniP. · · >~1;nn
(dotted curve) is also shown. The agreement between the exact and two-term multipole
_, · .n '-"- <
<~OU«O 00 , ovo
LJ
448 Chapter 9 Radiating Systems, Multipole Fields and Radiation-S!
51-
dipole approx.
/
I
I -
\
\
\ A multipoles
l(\1
-
\
\
\
(-
\
I
41- I I -
I
t.r-.
I
I I I
\I
\ \)' /.\
_!_fL_',\ I I I
31-- I
f-
f-
f-
21-
f \\i \ \ I I \
I./--\
I
/i
I .\
1
I
\
I
I
I .
•
1
.II -.:
Lxac! •, ~, : ; : • ~ ; I
1 u1poe!erm1 ·1 1 1 ; I 1 •, I
kd
Figure 9.8 Total power radiated by center-fed antenna with sinusoidal current
distribution (9.183) versus kd The ordinate is 47TPIZ.,fl with l the peak CJJrreut in
(9.183). T!J.g curvg labgJed "Long wavelength dipole approx." employs the long-
wavelength dipole moment (9.170) rather than the exact (9.167) used for the curve
labeled "Exact dipole term." The curve labeled "All multipoles" is the sum (9.188)
[actually up to E9 J.
For the half wave antenna the coefficients in Table 9 2 show that the pmver
radiated is a factor 1.00244 times latger than the simple dipole result, (3Z 0 i"l7r').
For the full-wave antenna, the power is a factor 1.10565 times larger than the
dipole form (3Z0 J2/47r).
A comparison of the total power (9.188) for the center-fed linear antenna
with the lowest multipole power, for both the exact lowest multipole and its long-
wavelength approximation, is shown in Fig. 9.8 versus kd. For kd < 27r, the power
is dominated by the El mnltipole, as we have jnst seen, but for larger kd the
higher multipoles contribute more and mme. It is notewm thy that the long-
wavelength dipole approximation departs significantly from the exact dipole re-
sult (and the total power) for kd > 1r. The departure, which becomes gross for
larger kd, is a consequence of differences between exact multi pole moments and
the long-wavelength approximation~ to them when the 'Navelength becomes com
parable to or smaller than source size.

The simple theory of radiation from a localized source drstnbuhon rs drscussed mall
modern textbooks. Treatments analogous to that given here may be found in
Stratton, Chapter 8
~~"'
~II. "> • ...... 7
More complete discussions of antennas and antenna arrays are given in applied
' , ""
Joruan anu
Kraus
Schelkunoff and Friis
Silver
ll> Ul -V<UUv J-H'
,_ .
_, .
' ""
J'~"'o ~
ll .. ... 'IN '
be found in
Hallen
Jones
Schelkunott, AdvanceaAntennalTfeory
The subject of excitation of waveguides by localized sources and the use of multipole
moments is discussed by
Collin
L ,L ~\ '
'
l lC lllCl<llUlC Ull tHC~' uu• " "!-'~'
of effective dipole moments has been cited in Section 95. The basic theory and some
applications appear in
Collin
-
Nwmgomery, UICKe, -
ana rurceu tPP· 110 ''- "''u l'l'· L.7v "-J
Van Blade!
The theory of vector spherical harmonics and multipole vector fields is discussed
thoroughly by
Dlall anu VY eiMKUpt, AJ-'. CD
Morse and Feshbach, Section 13.3
Applications to nuclear multipole radiation are given in
Blatt and Weisskopf, Chapter XII
, OE , Alll -uy 0. ""- auu ' -"- ' ' V'J uy m.
Goldhaber and A. W. Sunyar
"- _L,nnu•
9.1 A common textbook example of a radiating system (see Problem 9.2) is a config-
uration of charges fixed relative to each other but in rotation. The charge density
is obviously a function of time, but it is not in the form of (9.1).
Ia\ <::hom thot fnr rntot;na rhornP< nnP ive is to calculate real time-denen-
dent multi pole moments using p(x, t) directly and then compute the multipole
moments for a given harmonic frequency with the convention of (9.1) by in-
nr FnnriPr dPcomnosition of the time-denendent moments. Note that
_._.
' L ",_ •1. , L .1 .
'1 ,m
( t\ tn fnrrr l;nPM
~b
;nn< thot
are real oerore maKing tne connection.
(b) Consider a charge density p(x, t) that is periodic in time with period T = 21r/w0 .
By making a Fourier series expansion, show that it can be written as
.t. ,,
, ., r~\-'- ~ RPr7, n (v\, -inw,,t]
" " "' n"-;;l ' "
where
.
l'nV'J
' 1
-r1
I ,.\
, •J
inld r ..J.
This shows explicitly how to establish connection with (9.1).

450 Chapter 9 Radiatinl! Svstems, Multipole Fields and Radiation-S!
. ·o· ·o ~ ·o '5' J l' '~"; 0[

radms R at constant angular speed Wo, calculate the l = 0 and l = 1 multipo\e
moments by the methods of parts a and b and comnare. In method b eYnrA,
the charge density Pn(x) in cylindrical coordinates. Are there higher multi-
poles, for example, quadrupole? At what frequencies?
9.2 A radiating quadrupole consists of a square of side a with charges ±q al alternate
corners. The square rotates with angular velocity OJ about an axis normal to the
plane of the square and through its center. Calculate the quadrupole moments, the
radiation fields, the angular distribution of radiation, and the total radiated power
" . .' ' . . . . ,
" " " ' W<O CVHIS" va . r . " u . IS lllC . Ul lilt: rauranon.
9.3 Two halves of a spherical metallic shell of radius R and infinite conductivity are
'"parateu uy a very MitaH · · ;gap. f"\.n aaernating potential is app1iea between
the two halves of the sphere so that the potentials are ± V cos OJ!. In the long-
•h ,. ;, ",1 t>
., ,r,,· "'''"'· ' '" ·' x .c _, .
~
power, and the total radiated power from the sphere.

0 A A ~~1., thA A ,f 0 11 . t> ~ " "
· n ., ·rr > v•
the particle of charge q rotating about the origin in the x-y plane in a circle of radius
R at constant angular speed OJn. The motion is such that OJ,R << c.
(a) Find (Jx), (Iy), and (J,), in terms of cylindrical coordinates for all n. Also
determine the comnonenJS Of the • "moan • • " {y >( J )/? Onrl ;t.
divergence [which plays the role of a magnetic charge density for magnetic
multipoles, as in M 1, (9.172)].
(b) What long-wavelength magnetic multipoles (I, m) occur and al what frequen-
cies? [Remember that the multipole order l does not necessarily equal the
harmonic number n.]
(c) Use linear superposition lo generalize your amument to the four charges ro-
tating in Problem 9.2 at radius R = a/Vl. What harmonics occur, and what
magnetic multipoles at each harmonic? Is there a magnetic multipole contri-
outwn at tne ~L trequency ot J:'roblem Y.2'! Is It stgnificant relative to the £2
r"diation?
9.5 (a) Show that for harmonic time vaTiation at frequency OJ the electric dipole scalar
oncl · -•, in thAT nrPn? ao•
0 . 0
·-'"' tJ..- 1
"b
• ·' •~" ;, ,
eikr
(h{v\ · n • nf1
47TE0r" •
. fLoW eikr
Afv\ ,,,. • • /() 1<0\l
47T r r L ' • ~n
,, .. ,, _,. -'
,..., ("',]
wc;y
"" ..
\">·!I), ctllU UlC lllllC
ctlco );lVCll uy \"'·'"J·

e 1' UllUCl >LUUU.
'J
r ·'
•J-
. ' _, _, "
fiE ~ r
~.u
'"' 0C«H~HIS HVW '"~ IS~H<OW> ~AfH
pressiOn ror 'l', expana oom K lx x' 1 and t t · Ric to first order m
fx• 1/r to obtain the ·' · ~- ·' iok fr.r orhitronr timP vo ·
<Nx t) - 1 1 n •" + 1 " • ap,e,

47TEo r CT at
47Tr iit
Ch. 9 Problems 451
E(x t)
1
r1 + r a:) 3n(n • Peer) - Pcet
+
1
n x( n X cFp~et j
41TEo c dtj r cr (j[
(c) Show explicitly how you can go back and forth between these results and the
harmonic fields of (9.18) by the substitutions - iw <->a/at and peik,-iwr"'""' Pcer(t').
9.7 (a) By means of Fourier superposition of different frequencies or equivalent
meafls, show for a real electric dipole p(t) that the instantaneous radiated
power per unit solid angle at a distance r from the dipole in a direction n is
2
[n x dt~
d2
(t')
]
x n
where t' = t - ric is the retarded time. For a magnetic dipole m(t), substitute
(1/c)m x n for (n x p) x n.
(b) Show similarly for a real quadrupole tensor Qa~(t) given by (9.41) with a real
charge density p(x, t) that the instantaneous radiated power per unit solid
ang e ts
dP(t) Zu I f0 d Q
3
i
dll = 5767T'c4 L x dt' 3 (
0
' t') J
where Q(n, t) is defined by (9.43).
9.8 (a) Show that a classical oscillating electric dipole p with fields given by (9.18)
radiates electromagnetic angular momentum to infiHity at the rate
3
- - - - Im[p* X p]
dt 121TE0
(b) What is the ratio of angular momentum radiated to energy radiated? Interpret.
(c) For a chargee rotating in the x-y plane at radius a and angular speed w, show
that there is only a z component of radiated angular momentum with mag-
nitude dLzldt = e2 k 3a2!61TE0 . What about a charge oscillating along the z axis?
(d) What are the results corresponding to parts a and b for magnetic dipole
radiation?
Htnt: The electromagnetic angular nrotnentmn density conres ftorn nrote than the
transverse (radiation zone) components of the fields.
9.9 (a) From the electric dipole fields with general time dependence of Problem 9.6,
show that the total powe1 and the total rate of radiation of angular momentum
through a sphere at large radius r and time t are
P(t)
1
61TEoC
1 ("Poet
3
c r pm)
a Pcet
af
X a
2
dt
where the dipole moment p is evaluated at the retarded time t' = t - ric.
452 Chapter 9 Radiating Systems, Mnltipole Fields and Radiation Sf
(b) The dipole moment is caused by a particle of mass m and charge e moving
nonrelativistically in a fixed central potential V(r). Show that the radiated
power and angular momentum for such a particle can be written as
P(t) = .!...
m
(dV)z
dr
dL m T ~dV~ L
where .. e2 /61rc 0 mc3 ( 2e2 /3mc3 in Gaussian units) is a eharaeteristic time

Lis the particle's angular momentum, and the right-hand sides are evaluated
at the retarded time. Relate these results to those from the Abraham-Lorentz
equation for radiation damping [Section 16.2].
(c) Suppose the charged particle is an electron in a hydrogen atom. Show that the
inverse time defined by the ratio of the rate of angular momentum radiated
to the particle's angular momentum is of the order of a 4 c/a 0 , where a =
e2!41Tc0 hc 17137 ts the fine structure constant and a 0 ts the Bohr radms. How
does this inverse time compare to the observed rate of radiation in hydrogeil
(d) Relate the expressions in parts a and b to those for harmonic time dependence
in Problem 9.8.
9.10 The transitional charge and current densities for the radiative transition from the
m = 0, 2p state in hydrogen to the ls ground state are, in the notation of (9.1) and
with the neglect of spin,
p(r A ¢_ t) - . re-3r!2aov V e ivJof

\16 a 400 ~ 0
J(r, e. </>, t) =
-iu 0
-2- (r2 + ---; z a0 -)
p(r, e, </>, t)
where a0 = 4mE0 h 2 /me 2 = 0.529 X w-w m is the Bohr radius, Wo = 3e2 /327Tc,jw 0 is
the frequency difference of the levels, and u0 = e2 /47Tc 0 h = ac = c/137 is the Bohr
orbit speed.
(a) Show that the effective transitional (orbital) "magnetization" is
".M."(r, e. p, t) - -i 4 tan e(x sin <k - y cos <k) . p(r, e. ¢, t)

Calculate V · ".M.'' and evaluate all the nonvanishing radiation multipoles in
the long-wavelength limit.
(b) In the electric dipole approximation calculate the total time-averaged power
radiated. Express your answer in units of (hWo) · (a4 c/a 0 ), where a= e2 /47Tc0 hc
is the fine structure constant
(c) Interpretmg the classtcally calculated power as the photon energy (hw 0 ) ttmes
the transition probability, evaluate numerically the transition probability in
units of mciprocal seconds.
(d) If, instead of the semiclassical charge density used above, the electron in the
2p state was described by a circular Bohr orbit of radius 2a 0 , rotating with the
transitional frequency w0 , what would the radiated power be? Express your
answer in the same units as in part b and evaluate the ratio of the two powers
9.11 Three charges are located along the z axts, a charge + 2q at the ongm, and charges
q at z +a cos wt. Determine the lowest nonvanishing multipole moments,
Ch. 9 Problems 453
me angmar oistriourion or rauia,iou. a11u we <u<at ~ wutaLeu.

ka << 1.
1\ >.., A "1 "' - '- · -' , -··
•
R(O) = R0 [1 + f3P2 (cos IJ)]
has inside of it a uniform volume distribution of charge totaling Q. The small pa-
rameter f3 varies harmonically in time at frequency w. This corresponds to surface
waves on a sphere. Keeping only lowest order terms in f3 and making the long-
wavelength approximation, calculate the nonvanishing multipole moments, the an-
gular distribution of radiation, and the total power radiated.
0 1'l ThP ,nifmm ' riPn<itu nf n . . 1..1. <) 17 ;, rP.nlorPrl hv o
'b
With the same approximations as above calculate the nonvanishing radiation mul-
tipole moments, the angular distribution of radiation, and the total power radiated.
9.14 An antenna consists of a circular loop of wire of radius a located in the x-y plane
with its center at the origin. The current in the wire is
(,.\ Pinrl thP fm E H in the radiation zone without annroximations

r r
(b) What 1s the lowest nonvamsmng mumpme moment U!t,m or Mt,m). tVaiUa<e
this moment in the limit ka << 1.
9.15 Two fixed electric dipoles of dipole moment pare located in the x-y plane a distance
2a apart, their axes parallel and perpendicular to the plane, but their moments
directed oppositely. The dipoles rotate with constant angular speed w about a z axis
located halfway between them. The motion is nonrelativistic (wale << 1).
(a) Find the lowest nonvanishing multipole moments.
(h) Show thot thP mounPtic field in the radiation zone is aoart from an overall
ikr
H =
cpa k '[('x
- + 'Y . IJ e'.<~>] cos IJ -e
.,) cos (} - z, sm
2~ r
(c) Show that the angular distribution of the radiation is proportional to
(cos2 1J + cos4 1J) and the total time-averaged power radiated is
LJ7Tto
~ ~· . _,
9.16 A thin linear antenna of lenQth dis excited in such a wav that the sinusoidal current
' 0 full UoOUPlPnrrth nf A " ' 'hown in the fiuure.
..,....... --- - ' .....

/
.... ...... _____ ..,.. /
Problem 9.16
\"! . CAa >J .""! 'l' "b' '> ·o
JUUUllUITl!illdUUll.
(h) th<> tntol nnwPr radiated and find a numerical value for the radi-
· ,,
ation resistance.
917 Treat the linear ant€mna of Problem 9.16 by the multipole expansion method.
(a) Calculate the multipole moments (electric dipole, magnetic dipole, and elec-
tric quadrupole) exactly and in the long-wavelength approximation.
(b) Compare the shape of the angular distribution of radiated power for the lowest
nonvanishing multipole with the exact distribution of Problem 9.16.
(c) Determine the total power radiated for the lowest multipole and the corre-
sponding radiation resistance using both multipole moments from part a.
Compare with Problem 9.16b. Is theiC a pa1adox he1e?
9.18 A qualitative understanding of the result for the reactance of a short antenna 'Hhose
radiation fields are described by the electric dipole fields of Section 9.2 can be
achieved by considering the idealized dipole fields (9.18).
(a) Show that the integral over all angles at fixed distance r of <oiEI 2 - JLo IH12
IS
(b) I lsing (6 140) for the reactance, show that the contribution ¥"to the reactance
from fields at distances 1 > a is
X=
"
where I, is the input current.

(c) For the short center-fed antenna of Section 9.2 show that X a= - d 2 /247TEowa3 ,
corresponding to an effective capacitance 241T<oa 3 /d 2 . With a = d/2. Xa gives
only a small fraction of the total negative reactance of a short antenna. The
fields close to the antenna, obviously not dipole in character, contribute heav-
1ly. For calculatwns of reactances ol short antennas, see the book by
Schelkunoff and Friis.
9.19 Consider the excitation of a waveguide in Problem 8.19 from the point of view of
multipole moments of the source.
(a) For the linear probe antenna calculate the multipole moment components of
p, m, Q"~' Q:;j; that enter (9.69).
(b) Calculate the amplttudes for exclfabon of the I E 1 ,0 mode and evaluate the
power flow. Compare the multipole expansion result with the answer given in
Problem ~.19b. Discuss the reasons for agreement or disagreement. What
about the comparison for excitation of other modes?
9.20 (a) Verify by direct calculation that the static tangential electric field (3.186) in a
circular opening in a flat conducting plane, when inserted into the defining
equation (9.72) for the electric dipole moment Pcrr. leads to the expression
(c) Use the statrc normal magnetic field (5.132) for the corresponding magnetrc
boundary problem with a circular opening to compute via (9.74) the magnetic
dipole moment meff and compare with (9.75).
(d) Comment on the differences between the results of parts b and c and the use
of the definitions (9.72) in a consistent fashion. [See Section 9 of the article,
Diffraction Theory, by C. J. B. Bouwkamp in Reports on Progress in Physics,
Vol. 17, ed. A. C. Strickland. The Physical Society, London (1954).]
Ch. 9 Problems 455
9.21 The fields representing a transverse magnetic wave propagating in a cylindrical

waveguide or radius R are:
Ez = I m ( yr)eimrf.>eif3z-t(t)l , H, = 0
-mf3 E, k
E.p = 7 7' H
'
= --E
Zo/3 = k?- - f3 2 (k = w!c), where y is such that lm( yR) = 0. Calculate the ratio
of the z component of the electromagnetic angular momentum to the energy in the
field. It mav be advantageous to perform some integrations bv parts and to use the
~"'- . -' · .,;, rl h" J;' tn <i '], '],
" ""'"
-,.~~ spuerica1u01e 01 raums am a conuuccing meuium can serve as an e1eccromagnecic
J-\.
resonant cavity.
(a) Assuming infinite conductivity, determine the transcendental equations for
the characteristic frequencies Wzm of the cavity for TE and TM modes.
(b) Calculate numerical values for the wavelength A1m in units of the radius a for
the fo11r lowe't "· :for TF "nc:l TM w"ves.
. . . . . .
~CJ La!CUlate expncllly L!Le e1eetnc anu magneuc ue1us mstue LHe cavny wr Llle
lowest TE and lowest TM mode.
9.23 The spherical resonant cavity of Problem 9.22 has nonpermeable walls of large, but
finite, conductivity. In the approximation that the skin depth 15 is small compared
to the cavity radius a, show that the Q of the cavity, defined by equation (8.86), is
given by
n a f, rl.
'11 '1'<0
"" 15
Q = '!. (
8
1- l(l :
Xtm
1)) for TM modes
where x 1, = (a/c)w 1m for TM modes.

9.24 Disccuss the normal moc:l'" of oscillation of a nerfectlv conductin11 solid snhere of
n in frPP snorP (This nmhiPm wos -'- h"TTTh , in thP 1RRt1s I
(a) Determine the characteristic equations for the eigenfrequencies for TE and
TM modes of oscillation. Show that the roots for w always have a negative
imaginary part, assuming a time dependence of e-'"''.
(b) Calculate the eigenfrequencies for the l = 1 and I = 2 TE and TM modes.
Tabulate the wavelength (defined m terms ot tne real part ot tne trequency)
in units of the radius a and the decay time (defined as the time taken for the
energy to fall to e 1 of its initial value) in units of the transit time (ale) for
each of the modes.
CHAPTER 10
Scattering and Diffraction
L. -'- .C .L. A' -'- -'"'"· ~ .< .L' •<- .< ~ < d. "J
~•" 'J uu•~• ""5 vu '"~ •~w" ·~ o<.mc;S
involved-the wavelen2.th of the waves on the one hand and the size of thP.
-~ \' .. w•~•v•~ .... ~~'"'J""'"~""'~'- '"~"'"~ ·~ -~ "''"~·~~ouuuu
is large compared to the dimensions of the target, a simple description in terms

of lowest order induced multipoles is appropriate. When the wavelength and size
are comparable, a more systematic treatment with multipole fields is required.
In the limit of very small wavelength compared to the size of the target, semi-
geometric methods can be utilized to obtain the departures from geometrical
optics. We begin with the long-wavelength limit of electromagnetic scattering,
with some simple examples. Then we develop a perturbation approach to scat-
tering by a medium with small variations in its dielectric properties in order to
discuss Rayleigh scattering, the blue sky, and critical opalescence. To introduce
the more systematic approach with multipole fields, we first present the multipole
,~ion of an wnetic nlane wave and then annlv it to the · "'
Uittractwn IS treated next. first the scalar Huv2.ens-Kirchhoft theorv. then a
or comp1ememary screens. 1 nese TOols are appnea TO anrracnon oy a circUlar

aperture, with connection to the low-order effective multipoles of Section 9.5 in
the long-wavelength limit. Scattering at very short wavelengths and the important
optical theorem complete the chapter.
10.1 Scattering at Long Wavelengths

A. Scattering by Dipoles Induced in Small Scatterers
The scattering of electromagnetic waves by systems whose individual dimen-
sions are small compared with a wavelength is a common and important occur-
renee. m sucn Imeracnons IT IS convement to tnmK or me InCident (radlanon)
,,
. -~ . .
'l' wHH Lm; IuuucnL wave anu rauiaLe
. .
energy m uirecL!ons ower
. . . w<tu
thP. inn nf inriciP.nrP ThP PY~rt fnrm nf thP ~no-nbr d' nf r~cli~tPrl
.l L •L .1. ' · <-'c -L ' • < ' -'· -' L •'-
'"5" '~"•vu VJ u•v ~U"~'V'H OUJ'~'J'VOoHV" ~L '"' '1-' VJ ou~
mcident helds and m general depends on the state of polarization of the incident
wave. If the wavelength of the radiation is long compared to the size of the
scatterer, only the lowest multipoles, usually electric and magnetic dipoles, are
important. Furthermore, in these circumstances the induced dipoles can be cal-
culated from static or quasi-static boundary-value problems, just as for the small
apertures of the preceding chapter (Section 9.5).
456
Sect. 10.1 Scattering at Long Wavelengths 457
The customary basic situation is for a plane monochromatic wave to be in-

• ..1
em o Pnr o;mn];,.;tv thP · ""' mP.clinm is t~ k-Pn to h~v<e
01
' T r .1
;o • 1 1~ .1 '"' .,
1
fL, e, L H we, -uy -= '~ ~
""
mcwem po1anzanon vecwr is E0 , me inciuem ueiUs are
~inc t:oLoe "' (lU.l)
Hinc = no X Einc/Zo
where k = w!c and a time-dependence e-i<vr is understood. These fields induce
dipole moments p and m in the small scatterer and these dipoles radiate energy
in all directions, as described earlier (Sections 9.2, 9.3). Far away from the scat-
terer, the scattered (radiated) fields are found from (9.19) and (9.36) to be
1 eikr
Esc = - - k 2 - [(n X p) X n - n X m/c]
47TE0 r (1 0.2)
H,c = n X E,cfZo
WllC1C 11 d Ullll VCICtV1 111 tllC Ulll
1~ lH' ' y, m uuu'"' >o +L .J: VU~C1 •uy '"~
from scatterer. The power radrated m the d1rectwn n wnn pomnzanon E, per unir
~
.I. ; ..
-u, '"1'
; '"'" .. •.
solid ;mp-Je. ner unit inr;.-Jpnt flux (nower ner unit area) in the direction fin with
~• ~~~
.J
,....
nPr nn;t on];.-] onnlP
-o
oIt ;o
r
called the differential scattering cross section*:
1
rz -IE*. E 12
du 2Zo sc
-- (10.3)
dH (n, E; n 0 , Eo)
1 IEo* • Einc 12
220
The complex conjugation of the polarization vectors in (10.3) is important for
the correct handling of circular polarization, as mentioned in Section 7.2. With
(10.2) and (10.1), the differential cross section can be written
rfrr k-4
(n, E; n 0 , E0 ) IE · p + (n x e) · mrq ( IU.4)
UH \ '+7T<ooLo)
H1C > lH tllC IC1VM ~0
'
""-['"~"'] '"'" .
_, ..., +L
. . .
mp01e moments p ana m. 1ne vananon or rne unrerenllal 1anu tOta,, s~eaw:11u~
cross section with wave number as k 4 (or in wavelength as A ") is an almost
universal characteristic of the scattering of long-wavelength radiation by any fi-
nite system. This dependence on frequency is known as Rayleigh's law. Only if
both static dipole moments vanish does the scattering fail to obey Rayleigh's law;
the scattering is then via quadrupole or higher multi poles (or frequency-
dependent dipole moments) and varies as w6 or higher. Sometimes the dipole
scattering is known as Rayleigh scattering, but this term is usually reserved for
the incoherent scattering by a collection of dipole scatterers.
B. Scatterinr! bv a Small D1electnc Svhere

As a first, very simple example ot dipole scattenng we conswer a smau ui-
_, -L nf '"'" n with ' 1 on.-1 " nnifnrm isotronic dielectric constant
""ln tne engmccrmg uterature tne term mstatlc cross sectwn IS uscu tor 1T \U(hUHJ.
458 Chapter 10 Scattering and Diffraction-SI
E,(w). From Section 4.4, in particular (4.56), the electric dipole moment is found
to De
1:' 1
(10.5)
There is no magnetic dipole moment. The differential scattering cross section is

2
du _- k•a6 E, - 1 I
e~
.
· e 0 12
dfl E, +2 (10.6)
T'hco n~lori7oti~ cl ,cl ' ,1 -~ ~1 -' rl.' ~1

r :r. •J.r . ·- 0 • Ute
» ·; .
•cou 1ctuwuvu" p01anzeu mme p1ane uenneu uy LIIe utpo e moment
uy
direction (en) and the unit ·n
Typically the incident radiation is unpolarized. It is then of interest to ask
for the angular distribution of scattered radiation of a definite state of linear
polarization. The cross section (10.6) is averaged over initial polarization e 0 for
a fixed choice of e. Figure 10.1 shows a possible set of polarization vectors. The
scattering plane is defined by the vectors n 0 and n. The polarization vectors E&1>
and e(t) are in this plane, while e&2 > = e(2 ) is perpendicular to it. The differential
cross sectwns ror scattenn~> with notarizations e''' and e'"'. avera~>ed over initio!
k•a6 IL
dull - Er- 1
- -- cos2 8
dfl 2 E, +2 (10.7)
2
duJ. k•a6 E,- 1
-- --
d[l 2 E, +2
w II COl" lHC . II ctHU _[_ e po1arizmion paraue1 w ana perpendicular
to the scatterin~> nlane
.r · r1 · " .rl l
")
duJ. _ du,1
IT(e) = dfl dfl (10.8)
duJ. du11
~+
d[l dfl
IJ
\
\
\
Figure 10.1 Polarization and

propagation vectors for the
Y incident and scattered radiation.
Sect. 10.1 Scattering at Long Wavelengths 459
.-- ....
1~
'\. /
/
/ ' IT(B) ' ,
' ', /
/
\
1- ")(_ I 1
2\T """"T
2 d.,.
\
\
\
\
\
-
I dfl
\ -
I \
I \
I \
1- I I \ -
/ \
I \
01 0 -1
-cose
Figure 10.2 Differential scattering cross section (10.10) and the polarization of
scattered radiation (10.9) for a small dielectric snhere (dioole approximation).
From (10. 7) we find for the (electric dipole) scattering by a small dielectric sphere,
sin 2 11
Tl(ll) = < ,2n
(10.9)
i '-'VO
The differential cross section, summed over scattered polarization, is

2
Er - 1
da - k4 6
dD- a E, + 2
W + cos 11) 2
(10.10)
" fl. ~I . ~••

•o '" 2
a = Jd[!
da d[! = 87T k4a6
3
E,-
E,
1
+ 2
(10.11)
TJ, ,For :~J ~r~oo Oo {1111 11\ o~rl tho ~~lor;• nf thP -' r~rli-
\ '/ r
ation (10.9) are shown as functions of cos II in Fig. 10.2. The polarization IT( II)
has its maximum at II= 7T/2. At this angle the scattered radiation is 100% linearly
polarized perpendicular to the scattering plane, and for an appreciable range of
·o
ol • ;,J, "rl. ,j' Q .!') :o ...,.
~,.:too:. ·~ tlu nnbri·,pd ThP
., : nnbri7~tinn
'
;u<." u1 llle otue ~K y are au rmroiilJrTiliO f' II , auu' <U " • IH
tact, the mot1vatwn that led Kayleigh hrst to consH1er me pro01em. 1ne reaaer
can verify the general behavior on a sunny day with a sheet of linear polarizer
or suitable sunglasses.
r fiO:.rm • to hu n fiO:.nJnll ~ P .. r. -'- • !17 <'- .L

·o -' "J
An example with interesting aspects involving coherence between different

multipoles is the scattering by a small perfectly conducting sphere of radius a.
1ne e1ecrnc mp01e momem or sucn~ I 111 I L.J lU UC
3n (1{1 1'1\
l' '..-rrq)' lllC \ -;
0/~
~ Lnapter 1U :o~cattenng ana Utttracnon -:'11
The sphere also possesses a magnetic dipole moment. For a perfectly conducting
sphere the boundary condition on the magnetic field is that the normal compo-
nent of B vanishes at r = a. Either by analogy with the dielectric sphere in a
uniform electric field (Section 4.4) withe= 0, or from the magnetically permeable
spnere ~:sect10n ).11) wttn J1. - u, or by a simple direct calculation, it is found
thot tho> monno>t;,- mr>mPnt nf the> omoll onhc>ro> :c
T
... "
~ Hu
3TT
.a..a.inc ~1V.UJ
~
'. ' ' . are at ngnt
< u~ a "Y r . H~'-'1U<Oll1 vvavco 1111;; 1vvv -r · w each
other and to the incident direction.
The differential cross section (10.4) is
du (n,
dfl E,. n 0 , Eo ) -- k4 a6 IE * • Eo - 21(n X E *) • (n 0 X Eo )12 (10.14)
Tho> nr>lor;7ot;r>n ;PO onrl the> o ml rl" -~

_, _,.
are ~ore complic;ted than for the dielectric sphere. The cross sections analo~ous
to (10.7), for polarization of the scattered radiation parallel to and perpendicular
lU lH<O ' u~ >l-<llt<Ollll~, w llll ,- uLvu rnurnnon , are
au,, = K a Ieos ()- !12

dfl 2 2
(10.15)
k4a6 I
-duj_ = - - 1 - -21 cos () 12
d!l 2
The ditterential cross sectiOn summed over both states of scattered polarization
1- •
'
~ = k4 a 6 [~(1 + cos 2 ()) - cos f)] (10.16)
mh;]., the> nr>lor;.

r
(1 fl
..,
>2\ ;0
'
2
IT(e) = 3 sin f)
(10.17)
5(1 + cos 2
()) - 8 cos ()
The cross section and polarization are plotted versus cos() in Fig. 10.3. The cross
· . nas a ><r u"/5 uuuuvara peaKing causea oy erecrricciipore magnetic dipole
interferenre ThP nol,ri7,tion n'""hP.~ n - + 1 ot ll - hfl 0 onrl ;, "r " ' ; , ; , , thrmwh
--o
the whole angular range. The polarization thus tends to be similar to that for a
small dtelectnc sphere, as shown in Fig. 10.2, even though the angular distribu-
t:~no orn ~ .. :t · T' + 0 ] • • • - , {) 4 6v> .<
'1 • "b '"~ ,~,
the same order of magnitude as for the dielectric sphere (10.11) if (e,- 1) is not
omoll
Dipole scattering with its w4 dependence on frequency can be viewed as the

"
111co
,.
111co v~
1 co1.
..
lowest order approximation in an expansion in kd, where dis a length typical of
. 111<0
.
--.ar --r,--more- man me 1 v vv ..,,.
u~
order multiooles must be considered. Then the discussion is he~t 'i~hed

by use of a systematic expansion in spherical multipole fields. In Section 10.4 the
scattering oy a conaucnng spnere IS exammea from this pomt of view. When
/uf >> 1 • IOf;r>n mPfhe>rlo r>f 0 -''"· o~r+ Aon h ~~-1. rl · :llnc_
trated later in this chapter (Section 10.10). Whole books are d;v;(ed ~o the scat-
Sect. 111.1 Scattering al Long '.Yavelenglhs 461
T
I I I
' I I
2- )
1-
-
/ ,"!~, ....' ', /

k'
7 -
-
/
/
I '-..
I
I .... ....
I
__./ ........ ....
o,
1
I
' I I
' n
I
....
---- -- - I
--cos9
Figure 10.3 Differential scattering cross section (10.16) and polarization of scattered
radiation (10.17) for a small perfectly conducting sphere (electric and magnetic dipole
approximation).
tering of light by spherical particles possessing arbitrary fdo, €, fl'. Some references
to this literature are given at the end of the chapter.
D. Collection of Scatterers
As a final remark we note that if the scattering system consists of a number
of small scatters with fixed spatial separations, each scatterer generates an am-
plitude of the form (10.2). The scattering cross section results from a coherent
superposition of the individual amplitudes. Because the induced dipoles are pro
portional to the incident fields, evaluated at the position xj of the jth scatterer,
its moments will possess a phase factor, e'kno·xi. Furthermore, if the observation
point is far from the whole scattering system, (9.7) shows that the fields (10.2)
fot the jth scatterei will have a phase factor euikn·x .. The generalization of (10.4)
for such a system is
(10.18)
where g = kno - kn is the vectorial change in wave vector during the scattering.
The presence of the phase factors e•q·x/ in (1 0 18) means that, apart from the
forward direction where q = 0, the scattering depends sensitively on the exact
d1stnbutwn of the scatterers m space. The general behavior can be illustrated by
assuming that all the scatterers are identical. Then the cross section is the product
of the cross section for one scatterer times a structure factor,*
(10.19)
*We do not consider here the effects of multiple scattering; that is, we assume that the mean free
path fm sealtei ing i.< large wmparecl to the dimensions ef the seatteriRg array.
462 Chapter 10 Scattering and Diffraction-51
wntten out as a ractor times Its complex conjugate, ~(q) IS

2F(q) = 2: 2: e'q·(xl-xi)
; j"
If the scatterers are randomly distributed, the terms with j j' can be shown to *
give a negligible contribution. Only the terms with j = f' are significant. Then
~~ o\ N the total number of scatterers anrl the c~ott~rino ;, "'irl In h~
H~V
. '·•v ~ O~J'L'
..
yuo<HVH W
.c . ..1. - '
mu' nuum
•..
Lr,
.' Vlll<Cl
~.- .' llilllU, rne
~vmuv..rm:JITI>". ~
scatterers are very numerous and have a regular distribution in space, the struc-
ture factor effectively vanishes everywhere except in the forward direction. There
is therefore no scattering by a very large regular array of scatterers, of which
single crystals of transparent solids like rock salt or quartz are examples. What
~lllilll illllUUlll lJl ~l.-dll<Cllll!;
UU<C> Vl.-l.-Ul 1> :-uy -vr ; awayrrom
the oerfect Ja ttlce or bv imourities etc. An exnlicit illn•tpt;"" "],o
evidence for a restriction of the foregoing remarks to the long-wavelength ;~~
gime, is that of a simple cubic array of scattering centers. The structure factor is
well known to be
I I
2 Nrqta ' 2 N 2 q2a \ 2
I
N 3q3al \ -
') ') I ') I

Gl:( , \ M2 \ \ I
. \'1/
,\ \ I \
""
~lU."-V)
1
N~ sin 1 '1~ sinT ~
2 2 2 1
._
N 1
sin 1 ::! 2 ) ! Nj J
where a is the lattice spacing, Nt, N 2 , N 3 are the numbers of lattice sites along
the three axes of the array, N = N 1 N 2 N 3 is the total number of scatterers and
n. n. n. llTP thP 1Pnl' nf n lllnno th~ ovoc. At ohnrt '
;, . ' .. ' • u
VW ~ 7T), ~1U."-V) llil> w ""11 111t: nragg scanering conaiTion, q ,a U, L7T,

41T, ... , is obeyed. This is the situation familiar in x-ray diffraction. But at long
wavelengths only the peak at q,a = 0 is relevant because (q,a)max = 2ka << 1.
In this limit 2F(q) is the product of three factors of the form [(sinx,)/x,] 2 with
x, = N,q,a/2. The scattering is thus confined to the region q, :S 27TIN,a, corre-
sponding to angles smaller than AIL, where A is the wavelength and La tvoical
""~ro 11 .-1" nJ' tho o~oH. ·o -r
10.2 Perturbation Theory of Scattering, Rayleigh's Explanation

of the Blue Sky,* Scattering by Gases and Liquids,
Attenuation in Optical Fibers
A I' .1 Tl.on~•
._,
If the medium through which an electromagnetic wave is passing is uniform
in its properties, the wave propagates undisturbed and undefiected. If, however,
•·Annougn KaytcJgn s name snoma unaouotemy oe assocmted w1th the quantJtatJve explanatJon ot
the blue sky, it is of some historical interest that Leonardo da Vinci understood the basic uhenomenon
around 1500. In n"rticul"r his · 1Pnts with the scM+erino nf h" mnnil 'rnnh •'- ·
against a dark background (quoted as items 300-302, pp. 237 ff, in Vol. I of Jean Paul Richter, The
"'
Literary Works of Leonardo da Vinci, 3rd edition, Phaidon, London 1970) (also a Dover reprint
entitled 1he Notebooks of Leonardo da Vinci, Vol. 1, pp. 161 ff.) anticipate by 350 years Tyndall's
remarkably similar observations [J. Tyndall, Phi los. Trans. R. Soc. London 160, 333 (1R70)J.
Sect. lU.Z Blue Sky, Scattcrmg by Gases and Liquids, Attenuation m Optical Fibers '10-'
there are snatial (or temnoran variations in the electroma~>netic nr th.e

wave is scattered. Some of the energy is deviated from its original course. If the
variations in the properties are small in magnitude, the scattering is slight and
nerturbative methods can be emnloved. We ima~>ine a comnarison situation ror-
responding to a uniform isotropic medium with electric permittivity Eo and mag-
netic permeability f.Lo· For the present Eo and f.Lo are assumed independent of
freouencv althou~>h when harmonic time denendence is assumed this restriction
can be removed in the obvious way. Note that in this section Eo and f.Lo are not
the free-space values! Through the action of some perturbing agent, the medium
is suooosed to have small chan!!es in its resnonse to anolied fields so that
D * E0E, B *
f.LoH, over certain regions of space. These departures may be
functions of time and space variables. Beginning with the Maxwell equations in
the absence of sources
--
aB
V • B = 0, VxE=
at
(lU.Ll)
AD
V· D = 0 V X H=-
' at
it is a straightforward matter to arrive at a wave equation forD,
t"72n.
a2 D -...,. ' ...,. ~ in. r.\ ' a . .,. ~ in II\ i 1 " ,.,,.,,
~'"'"'{) atz \ C()UJ
"U At \' t 0 ., \ ~,
This equation is without approximation as yet, although later the right-hand side
will be treated as small m some sense.·
If the right-hand side of (10.22) is taken as known, the equation is of the
form of (6.32) with the retarded solution (6.47). In general, of course, the right-
narur sTITe IS urfKnown and l b.4/) must be regarded as an mtegral relatwn, rather
than a solution. Nevertheless, such an integral formulation of the problem forms
a fruitful starting point for approximations. It is convenient to specialize to har-
monictime vammon willfTrequency w1ortlfe unperturbed helds and to assume
that the departures (D - E0 E) and (B - f.LoH) also have this time variation. This
puts certain limitations on the kind of perturbed problem that can be described
by tne rormansm, our prevents tne d1scusswn rrom oecommg too mvo1ved. W1tn
a time dependence e-;w, understood, (10.22) becomes
ir72 7.2\n. fn. V\ ...,. " in
\ ) '"'"'"' \ cu ~, cu \
TT\
I 0 •J
(1" '>0\
\ . -'I
where k' = f.LoE 0w2, and f.Lo and Eo can be values specific to the frequency w. The
solution of the unnerturbedr!roblem with the rwht-hand side of (10.23) set eoual
to zero, will be denoted by D( 0 l(x). A formal solution of (10.23) can be obtained
from (6.45), if the right-hand side is taken as known. Thus
D = D(o) + -1
4'lT
l d 3 x' e""I' I
lx - x' I
X
+iE0w
V' X V'
V' X
X (D - E0 E)
(B - f.LoH)
(10.24)
*If prescribed sources p(x, 1), J(x, t) are present, (1 0.22) is modified by the addition to the left-hand
side of
;ul
lI Vp + JLo<o at(j
464 Chapter 10 Scattering and DitTraction !II
If the physical situation is one of scattering, with the mtegrand m (I 0.24) confined
to some finite region of space and n<o) describing a wave incident in somll direc-
lion, the field far away from the scattering region can be written as
D --'> D(o) + A -e (10.25)

'c r
where the scattering amplitude A,c is
1 V' X V' X (D - E0E)
Asc =- d3x' e--ikn-x'
(10.26
()
The steps from (10.24) to (10.26) are the same as from (9.3) to (9.8) for the
rad1atwn fields. Some mtegratwns by parts m (10.26) allow the scattering ampli-
tude to be expressed as
e r a·x e [n X (D - E0E)j X n
(10.27)
41T J EoW
- - n X (B - J.LoH)
k
The vectorial structure of the integrand can be compared with the scattered
dipole field (10.2). The polarization dependence of the contribution from
(D - E0E) is that of an electric dipole, from (B - t.toH) a magnetic dipole. In
correspondence with (10.4) the differential scattering cross section is
2
sc
(10.28)
dfl
where E is the polarization vector of the scattered radiation.
Equations (10.24), (10.27), and (10.28) provide a formal solution to the scat-
tering problem posed at the beginning of the section. The scattering amplitude
Asc is not known, of course, until the fields are known at least approximately.
But from (10.24) a systematic scheme of successive approximations can be
developed in the same wa~' as the Born approximation series of quantum
mechanical scattering. If the integrand in (10.24) can be approximated to first
order, then (1 0 24) provides a first approximation forD, beyond D(o). This ap
proximation to D can be used to give a second approximation for the integrand,
and an Improved D can be determined, and so on. Questions of convergence of
the series, etc. have been much studied in the quantum-mechanical context. The
senes IS not very useful unless the first few 1teratwns converge rapidly.
B. Born Approximation
We will be content with the lowest order approximation for the scattering
amplitude. This is called the first Born approximation or just the Born approxi-
mation in quantum theory and was actually developed in the present context by
Lord Rayleigh in 1881. Furthermore, we shall restrict our discussion to the simple
example of spatial variations in the linear response of the medium. Thus we
assume that the connections between D and E and B and H are
D(x) = [Eo + DE(x)]E(x) (10.29)
B(x) [Mo + 8/X(x)]H(x)
where 8e(x) and l'iJ.L(X) are small in magnitude compared with Eo and J.Lo· Ihe
differences appearing in (10.24) and (10.27) are proportional to DE and OJ.L. To
Sect. 10.2 Blue Sky, Scattering by Gases and Liquids, Attenuation in Optical Fibers 465
lowest order then, the fields in these differences can be approximated by the
unnerturbed fields:
D - EoE = 8E(x) n<Ol(x)

<Q (lU ...iU)
B - ttoH = <'>J.L(X) B(0 l(x)
f.Lo
If the unperturbed fields are those of a plane wave propagating in a direction Do,
so that D'"' and B'"' are
D(O)(X) = EoDoeikno·x
B(O)(x) = J~ Do X n<Ol(x)
I -u
the scalar product of the scattering amplitude (10.27) and E*, divided by D 0 , is
liF(x)
E* • Eo
E* • A(l)
sc _
k2 r d3 . 1q·x Eo
(10.31)
n - :;- xe >; • .{~\
v
+ (n X E*) · (n0 X E0 ) -,-~
f.Lo
wnere q Klllo n) IS tne diiierence or tne mcwent and scattered wave vectors.
The absolute square of (10.31) gives the differential scattering cross section
" '>0'
\ ~v.~v;.
If the wavelength is large compared with the spatial extent of 8E and 8J.L, the
IPnli~l in (10-11\ r~n hP 'P.t f'.nn~l tn nnitv T-hP ~mnlitnclP. ;, thP.n ~ dinnlP
approximation analogous to the preceding section, with the dipole frequency
dependence and angular distribution. To establish contact with the results already
obtained, suppose that the scattering region is a uniform dielectric sphere of
radius a in vacuum. Then 8E is constant inside a spherical volume of radius a and
vamsnes outsiOe. lhe mtegral m llU ...il) canoe perrormed tor armtrary IQI, wnn
the result,
e* . Asc 8E smqa- qa cosqa
= k'- (E* • Eo) q3
Do Eo
In the limit q ~ 0 the square bracket approaches a3/3. Thus, at very low fre-
quencies or in the forward direction at all frequencies the Born approximation
to the differential cross section for scattering by a dielectric sphere of radius a is
I fliT\ ?iF 2
lim I - ka IE~ • Eol' (lU ...iL l
q-o ,a!.L;Born 5Eo
WlLll \.lU.U) 'llUW' llldl lllC LHJHl <lj.Jj.JU
.. .
<lllU lllC CA<ll.-1 lUW
frequency result have the expected relationship.
C. Blue Sky: Elementary Argument

ThP' nf li(}ht hv 1}~,,, fin:t trPMPn nn~n1 ·
• I(} ]v hv Lnrcl Ravlei(}h
in his celebrated work on the sunset and blue sky,* can be discussed in the present
*Lord Rayleigh, Philos. Mag. XLI, 107, 274, (1871); ibid. XLVD, 375 (1899); reprinted in his Scientific
Papers, Vol. I, p. 87, and Vol. 4, p. 397. Rayleigh's papers are well worth reading as examples of a
..
l'"J'
rrameworK. ~nee _IQ_e magnetic moments ot most gas molecujes are ne!!li~>ihlP
compared to the electric dipole moments, the scattering is purely electric dipole
in character. In the preceding section we discussed the angular distribution and
polarization of the individual scatterings (see Fig. 10.2). We therefore confine
our attention to the total scattering cross section and the attenuation of the in-
cident beam. The treatment is in two parts. The first, elementary argument is
adequate for a dilute ideal gas, where the molecules are truly randomly distrib-
uted in space relative to each other. The second, based on density fluctuations
in the gas, is of more general validity. We now identify Eo with the electric per-
mittivity of free space.
If the individual molecules, located at xj, are assumed to possess dipole mo-
ments pj = EoYma 1E(xJ, the effective variation in dielectric constant .SE(x) in
( lU.jl) can De wntten as
\ . '~}
~ \ ·- • I \ I} 00
d[!
k• IYmol 12 Ie* .
dcr = 161T2 Eo
12u.;C
';!je q)
where 2/i( q) is given by (10.19). For a random distribution of scattering centers

the structure factor reduces to an incoherent sum, and the cross section is just
that for one molecule, times the number of molecules. For a dilute gas the mo-
lecular polarizability is related to the dielectric constant by Er = 1 + Nymob where
N is the number of molecules per unit volume. The total scattering cross section
per molecule of the gas is thus
?.k•
11 In 11 110.34\
u1TIV j'lTJV
where the last form is written in of the · rl. of__refi:action__ 1ino
ecme ror a unn mciOent energy !lux. ln traversmg a thickness dx of the gas, the
fractional loss of flux is Ncr dx. The incident beam thus has an intensity I(x) =
/0 e-~\ where a is the absorption or attenuation coefficient (also called the extinc-
tion coefficient) of (7.53) and is given by
2k4
a= Ncr= -In-
31TN
11 2 (10.35)
These results, (10.34) and (10.35), describe what is known as Rayleigh scattering,
the mconerell!. scattenflg DV !!as molecules or other randomlY_ distributed dioole
CPO PO Ph <Poft >T; . D ,]. . ,<' 4 1
Kay1eign s uerivation or ( lU.J)) was m me context or scattermg or l!!!ht bV

the a = E.cidenthz:_ fhP 1<4 nPnPnnPnPP mf'on< thot jn__the. • .;hJ •m
d " • ••· • . •
'"" '"u •o o"a""'"u '"ao• ctuu we VHHt:L IliUM. Llgm IeceJveu away 1rom me m·
mmea Deam oecomes mcreasmgly red m 1ts spectral composition, as well as di-
minishing in overall intensity. The blueness of the sky, the redness of the sunset,
the waneness of the winter sun, and the ease of sunburning at midday in summer
are all consequences ot Raylergh scattenng m the atmosphere. lhe mdex of

refraction of air in the visible region (4100-6500 A) and at NTP is (n - 1) =
2.78 X 10- 4 . With N = 2.69 X 1019 molecules/cm 3 , typical values of the attenu-
ation length A = a- 1 are A = 30, 77, 188 km for violet (4100 A), green (5200
A), and red (6500 A) light, respectively With an isothermal model of the at-
mospltere in wlticlt tire density varies exponentially with height, the following
intensities at the earth's surface relative to those incident on the top of the at-
mosphere at each wavelength can be estimated for the sun at zenith and sunrise-
Color Zenith Sunrise-Sunset

Red (6500 A) 0.96 0.21
Green (5200 A) 0.90 0.024
Violet (4100 A) 0.76 0.000065
These numbers show strikingly the shift to the red of the surviving sunlight at
The actual situation is illustrated in Fig. 10.4. The em ve A shows the power
spectrum of solar radiation incident on the earth from outside as a function of
photon energy. Curve B is a typical spectrum at sea level with the sun directly
103
1- -1. ' l ,', lo.6 ' ' 1
0.35
1- T f 2j0, \4 1j2 1j0 ~·~ Oj8 ~. Oj7 I
'ois' lo{
I
1- +-----Wavelength (microns) -
\J\ 1\ ,,_,..:>-:::::-....
v -
~,,
1-
1-
If
;;::
....E I
I \
\
'\
-!""
~
\\
"'
E 102 1--
2 1- I \
\
\
\
\
-:
~ \I
\\
1-
.
~
1-
1- \ Visible
~ 1- I region \ ~
\ \
1- I '\\ -
.
\~
"~" "·~·
I
I
1- I
I
I
\ \
10
0 1 2 3 4
Photon energy (eV)
Figure 10.4 Power spectrum of solar radiation (in watts per square meter per electron
volt) as a function of photon energy (in electron volts) Curve A is the incident
speetrm above the atmosphere. Curve B is a typical sea level spectrum 'Nith the snn at
the zenith. The absorption bands below 2 e V are chiefly from water vapor and vary
from site to site and day to day. The dashed curves give the expected sea-level
spectrum at zenith and at sunrise-sunset if the only attenuation is from Rayleigh
scattering by a dry, clean atmosphere
overhead.* The upper dashed curve is the result expected from curve A if the
only attenuation is Rayleigh scattering by a dry, clean, isothermal, exponential
atmos here. In re li ·
of water vapor, which has strong absorption bands in the infrared, and ozone,
which causes absorption of the ultraviolet, as well as other molecular species and
dust. The lower dashed curve indicates roughly the sunrise-sunset spectrum at
see even re
The formula (10.35) for the extinction coefficient is remarkable in its pos-
session of the factor N- 1 as well · ·
refraction. If there were no atomicity (N ~ oo ), there would be no attenuation.
onverse y, t e o serve attenuatiOn can e use to etermine N. This point
was urged particularly on Rayleigh by Maxwell in private correspondence. If the
determine N. Early estimates were made in this way and agree with the results
of more conventional meth ds.
D. Density Fluctuations; Critical Opalescence

An alternative and more general approach to the scattering and attenuation
o 1g t m gases an 1qm s is to consider fluctuations in the density and so the
index of refraction. The volume V of fluid is imagined to be divided into cells
actual number of molecules fluctuates around Nv in a manner that depends on

the r · ·
From the Clausius-Mossotti relation this can be written
Kondratyev, Radiation in the Atmosphere, Academic Press, New York (1969), Chapter 5.
With thi< fnr fil' fnr thP ith C'P1J thP. in1Pl1T~ 1 (1 0 11) nnw ~ <llm CIVPT
_1f L
~~u~,
e* ·All)
<C _ *
- e · e0
e(Er- 1)(E, + 2) 2: D.N ·e 1
iq·x
1 (10.37)
Do 12nNE, ;
. . .
m rormmg me aosomte square or ~ !V.-' 1) a suucmre racwr s1m11ar w ~1u.r ':1) Will
occur. If it is assumed that the correlation of fluctuations in different cells (caused
indirectly by the intermolecular forces) only extends over a distance small com-
pareu LO a waverengm, me exponemiar in ~ 1 u.-' 1 J can oe put eq uar w uniL y. 1 nen
the extinction coefficient a, given by
1
a="
r e* ·
n
A~~>
2
dD
·u
IS (10.38)
2
(w/c)4 (E,- 1)(Er + 2) D.N'f,
a
tnrN j NV
where D.Nt is the mean square number fluctuation in the volume V, defined by
D.N'f, = 2: D.NjD.Nr
II
the sum being over all the cells in the volume V. With the use of statistical
mechanics* the quantity D.N'f, can be expressed in terms of the isothermal com-
pressibility f3r of the medium:
< H2 ' I ,TT\
u.n v = NkT{3 f3r = - L Iu (10.39)
NV T• v aP )T
The attenuation coefficient (10.38) then becomes
4 2
w ( E, lHE, -t L)
a = (c 6~N
3
· NkTf3r (10.40)
This l~r first obtained bv Einstein in 1910 is called the

Einstein-Smoluchowski formula. For a dilute ideal gas, with IE - 11 << 1 and
NkTf3r = 1, it reduces to the Rayleigh result (10.35). As the critical point is
approached, f3r becomes very large (infinite exactly at the critical point). The
' rl ,,; <h '- ]. <h 'rh'e :c <ha ~i;,
known ~~critical opalescence. The large scatt~~ing is directly related to the large
fluctuations in density near the critical point, as stressed originally by
on< c~ f'~r fa:lc ha_
"· ,], ·h · f10IIA\
'". ./ ' u<
)
<h ~· ;,.· •al ~.
_r
:ni ~nr
Lilt: t,;Ull t:l lt:Hglll lUI lllt: grc::a tt:r waH a
wavelength, as first pointed out by Ornstein and Zernicke (1914).

For large correlation length A we must retain the exponential phase factors
lll ~ lU ..71 )· 1 HC MfUatt: Vl we, lllt:lllll VOl VC~ a UUUUlC
sum of 11N;I1Nje;q-\'' xi!, which can be expressed as a Founer transform of the

density correlation function. Because there is now additional angular dependence
from q, the angular distribution is no longer the simple dipole form. If a corre-
*See F. Reif, Fundamentals of Statistical and Thermal Physics, McGraw-Hill, New York (1965),
pp. 300-1, or L. D. Landau and E. M. Lifshitz, Statistical Physics, 3rd edition, Pergamon Press, New
York (1980), Chapter XII.
470 Chapter 10 Scattering and Diffraction-Sf
1
•• , .. '
•+ v .. 1• +•
IUe
-riAl~
_, . L - ,_
'"'"" u-., ouv v" '"'"
diiierential attenuatiOn coetnc1ent tor unpolanzea mclcfent nR!iatwn takes the
form
da(O) = ~ (1
dD 167T
+
cos
2 )
e a
[ 1 + AV!NkTf3T
1 + Azqz
1 (10.41)
where q 2 = 2( w/c) 2 (1 - cos 0) and a is given by (10.40). For Aq << 1, integration

<1. • 1• ~ •1 _,_. !L • • ' ' ,: ~ '~ ' c
"''"' ~,,, IS"'"'' ~iV."+V), UUL1U1 H "", tne
angular mtegratwn vtelds attenuatiOn proportional to lc/Awl lnl Aw/cl timAo
{1 () Ll()\
' •
ThA frPmtPn<>u rl
-, -J ·r
~
"'"''"U
QO 4 Q
·-:
·~~- •J..~
..
'~
~ ., .. _,
to roughly w 2 ; the scattered light appears "whiter" close to the critical point.
We note that, while our expressions diverge exactly at the critical point and
therefore are unphysical, a better treatment yields large but finite attenuation.
One consideration is that the correlation length A cannot become larger than the
dimensions of the fluid container.
LO LIIe eany JiLerawre can oe round in 1 aDellnsku, wno discusses
thP. ~nnlirMinn nf lio-ht · -tn rritir~ I n ,;nt .:;h,n Qn,1 ,r1_~¥r1.
'
.. r
r Ll
. 1'V1 " " " UL lllC -carr n, see
Rosenfeld (Chapter V, Section 6), or Landau and Lifshitz (op. cit.).
E. Attenuation in Optical Fibers

It is of interest that the ultimate limiting factor setting the maximum distance
between reneater units in ontical fiber transmissinn is the 11n ·-',hiP -
100
10
"' --,
c (\ -
·-
I I
""
'0
-~ I \ '
c ; -........ I I '
-ro Rayleigh·-~ "'\.J '
-·-·-·- _____\ ·-
./
" '
''
c scattenng ..............
!i'"' . , ''
'
''
0.1
~ ...
"' ...... ... , ......
......
Ultr.violet
.. --,!__
'
'
, --- .
---:
'
~ ~· I I
......
I
--- -- -- ''
'
'
0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8
Wavelength (I'm)
Fignre 10.5 Attenuation versus wavelength for a typical low-loss, single-mode silica
optical fiber (schematic). Rayleigh scattering sets the lower limit until infrared
absorption rises above 1.6 J-tm. The peaks in the observed attenuation are caused by
water (OH ions) dissolved in the Qlass.
Sect. 10.3 Spherical Wave ExpansiOn of a Vector Plane Wave "'H
tion caused by Rayleigh scattering, and by infrared absorption at longer wave-

. ... .. . " .·
1engms. 1 ne Isotnermai compresswuny or smca gmss 1s p, I X lU m-11~,
while the relevant temperature T = 1400 K (called the fictive temperature) is

Lllc HULL i <11 C 11 UZcll 111 ~ itf- r _0 Lllc allllcalllll,; -r LdLU1t: )·
The u. . V<'lliH" of (F. 1 )( E + 2)/3 1.30 in (10.40) is somewhat smaller

than the 1.51 inferred from an index of refraction of n = 1.45 at A = 1.0 J.Lm. The
net result is that a (km- 1 ) = 0.2/rA ( J.Lm)l 4 . The conversion to decibels per kilo-
meter (a factor of 4.343) gives a (dB/km) = 0.85/[A (J.LmW, shown as the dash-
aonea curve in J:<ig. lU.), wnicn aisp1ays a scnemaric rcprcscmadon or ,1 l';~u
data for a low-loss, single-mode optical fiber. For wavelengths less than 1.5 J.Lm,
Lllc all 1' uy Ray •0 • >CaLLClllll,;, pru' Lllc au>Ul}'LlUH uy 1111-
nurities such as the hvdroxvl ions from verv small amounts of water dissolved in
the glass. At wavelengths longer than 1.6 J.Lm, infrared absorption sets in strongly.
The minimum attenuation of about 0.2 dB/km occurs at A = 1.55 J.Lm. The ab-
sorption mean free path at the minimum is 22 km.
10.3 Sphencal Wave Expanswn of a Vector Plane Wave

In discussing the scattering or absorption of electromagnetic radiation by spher-
. -' .L" ' • ~ • ·' . . .£.. , L • .£
~~m ., 0 w OJ• ~u
0
em, u D uo~~u <V ·r
a plane electromagnetic wave in spherical waves.
For a scalar field !f;(x) satisfying the wave equation, the necessary expansion
can be obtained by using the orthogonality properties of the basic spherical so-
lutions h(kr) Y 1m(11, 1> ). An alternative derivation makes usc of the spherical wave
• • ibR. I 1 I ' ,, • '
, \7.70) v1 Lllc Ulccll ~" J·'lf")· " " 1eL I" 1 v11 uvLH 01ueo
of (9.98). Then we can put jx- x' I = r' - n · x on the left-hand side, where n is
a unit vector in the direction of x'. On the right side r> = r' and r< = r. Fur-
thermore we can use the asymptotic form (9.89) for h\ 1>(kr'). Then we find
eikr' eikr'
-4 ' e-ikn-x = i kk -r 2: (-iY+ 1j f (kr)Y*lm (II' ' d>')Y
, lm
(II , </>) (10.42)
7Tf r l,m
Canceling the factor e'k''!r' on either side and taking the complex conjugate, we
have the expansion of a plane wave
00 I
eik-x = 47T ~- i h(kr) 1
L Yim(ll.<f>)Y,m(ll', </>') (10.43)
/--=-0 m=-l
where k is the wave vector with spherical coordinates k, II', </>'. The addition
meorem ~.:l.OL) can oe usea to put mis in a more compact iorm
6, i (2l + 1)h(kr)P (cos y)

~
1
e'k·x = 1 (10.44)
where y is the angle between k and x. With (3.57) for P1 cos( y). this can also be
wntten as
~
eik-x = ~ {\147r(2l + 1) h(kr)YI_o('Y) (10.45)

We now '.vish to make an equivalent expansion for a eireularly polarized

plane wave with helicity ± incident along the z axis,
(1 0A6)
cB(x) e3 X E +iE
Since the plane wave is finite everywhere, we can write its multipole expansion
(9.122) involving only the regular radial functions h(kr):
(10.47)
. .
To determine the coefficients a,_(l, m) and b+(l, m) we utilize the orthogonality

properttes of the vector sphencal harmonics X 1m. For reference purposes we
summarize the basic relation (9 .120), as well as some other useful relations.
(
J l!1~r)A1·m·f • Lgt~T)AimJ au J lgl bu·bmm'
J[ft(r)XI'm']* • [V X gt(r)Xtm] dO = 0
~1U.48)
1 { rn ~ J:t.\v '* __rn_ /_\~
t:oiV
1 Jr.
k2 ) L Jl\' J· l'm'J L !"~lmJ u•"
1 i) i)
Ou·Omm' J Igl -t rJ 1 ~rg 1 )
er 2 iJr iJr
In these relations f 1(r) and g1(r) arc linear combinations of spherical Bessel func-
tions, satisfying (9.81). The second and third relations ean be proved using the
operator identity (9.125), the representation
r iJ i
v-----
r iJr r
for the gradient operator, and the radial differential equation (9.81 ).
To determine the coefficients a"'(l, m) and bo. (l, m) we take the scalar product
of both sides of (10.47) with Xim and integrate over angles. Then with the first
and second or tlrogonality relations in (10.48) we obtam
r
a±(l, m)h(kr) - J Xim • E(x) dil (1 0.49)
and
b. (l m)i.(kr) - c { X~"" • B(x) dQ (10 50)

'" J
With (10.46) for the electric field, (10.49) becomes
r (T l'; )*
(10.51)
Sect. 10.4 Scattering of Electromagnetic Waves by a Sphere 473
where the operators L"' are defined by (9.102), and the results of their operating
by (9.104). Thus we obtain
(1 0.52)
If expansion (10.45) for e'kz is inserted, the orthogonality of the Y1m's evidently
leads to the result,
a±(l, m) = iV 47T(2l + 1) lim,±t (1 0.53)
From (10.50) and (10.46) it is clear that

(10.54)
Then the multipole expansion of the plane wave (10.46) is

00
E(x)- ~ i 1V47T(2l + l)ft(kr)X1 .~ 1 ± k1 V

[ X ]!(kr)X1 ±1 ]
(10.55)
00 •
cB(x) = 2: iV47T(2l + 1) 1~ V X h(kr)X1,±1 + ih(kr)X1 ~ 1
For such a circularly polarized wave them values of m = ±1 have the obvious
mterpretation of ±1 unit of angular momentum per photon parallel to the prop-
agation direction This was established in Problems 7 28 and 7 29
10.4 Scattering of Electromagnetic Waves by a Sphere

If a plane wave of electromagnetic radiation is incident on a spherical obstacle,
as indicated schematically in Fig. 10.6, it is scattered, so that far away ftom the
scatterer the fields are represented by a plane wave plus outgoing spherical
waves. There may be absorption by the obstacle as well as scattering. Then the
total energy flow away from the obstacle will be less than the total energy flow
towards it, the difference being absorbed. We will ultimately consider the simple
example of scattenng by a sphere of radius a and infinite conductivity, but will
for a time keep the problem more general.
Tire fields outside tire sphere can be written as a sum of incident and scattered
(10.56)
wave
F'tgure 10.6 Scattenng of radtatwn By a locahzed ObJect.

where Einc and Binc are given by (10.55). Since the scattered fields are outgoing
waves at infinity, their expansions must be of the form,
Esc=~~ iV47T(2/ + 1) [ a~(l)h)i\kr)X1.~, 2: {3~/) V x h)ll(kr)X1.~ 1 ]

cB,. ~ ~ i'V47T(2l + 1) [ -i;~(/) V X hfll(kr)X1• 1 + ip (l)bfll(kr)X, +JJ
The coefficients a"- (l) and 73+(/) will be determmed by the boundary condrtwns
on the surface of the scatterer A priori, it is necessary to keep a full sum over
m as well as l in (10.57), but for the restricted class of spherically symmetric
problems considered here, only m = 2:1 occurs.
Formal expressions for the total scattered and absorbed power in terms of
the coefficients of a(/) and [3(!) can be derived from the scattered and total fields
on the surface of a sphere of radius a surrounding the scatterer, with the scattered
power being the outward component of the Poynting vector formed from the
scattered fields, integrated over the spherical surface, and the absorbed power
being the conesponding inwwd component formed fwm the total fields. With
slight rearrangement of the triple scalar products, these can be written
(10.58)
Pabs = _!{__ Re
2/Lo
I E • (n X B*) dfl (10.59)
Here n is a radial!} directed outward normal, Esc and B,c are given by (10.57),
while E and B are the sum of the plane wave fields (10.55) and the scattered
fields (10.57). Only the transverse parts of the fields enter these equations. We
already know that X 1m Is transverse. I he other type of term m (10.55) and (10.57)
. inVl(l + 1) 1 a
V X fz(r)Xzm = fz(r)Yzm + - - [r.fz(r)]n X X 1m (10.60)
r r ar
where fz is any spherical Bessel function of order l satisfying (9.81). When the
mulupole expanswns of the fields are mserted m (10.58) and (10.59), there results
a double sum over l and l' of various scalar products of the form XTm · Xrm,
Xim • (n X Xf'm•) and (n X X'im) • (n X Xl'm)· On integration over angles, the
orthogonality relations (10.48) reduce the double sum to a single sum. Each term
in the sum involves pwducts of spherical Bessel functions and derivatives of
spherical Bessel functions. Use of the Wronskians (9 91) permits the elimination
of all the Bessel functions and yields the follov.;:ing expressions for the total scat-
tering and absorption cross sections (the power scattered or absorbed divided by
the incident flux, 11f.Loc):
(10.61)
Sect. 10.4 Scattering of Electromagnetic Waves by a Spbet e 475
The total or extinction cross section is the sum of O"sc and O" abs:
O", = - ~ 2: (21 + 1) Re[ a (I) + f3(1)] (10.62)
Not surprisingly, these expressions for the cross sections resemble closely the
partial wave expansions of quantum-mechanical scattering.*
The differential scattering cross section is obtained by calculating the scat-
tered power radtated mto a gtven sohd angle element d!! and dlVldmg by the
incident flux. Using the result of Problem l0.6a, we find the scattering cross
section for incident polarization (E 1 :+: iE 2) to be
2
sc "' (10.63)
d!l = 2F f
The scattered radiation is in general elliptically polarized. Only if a±( I) = /3±(1)
for all I would it he circularly polarized This means that if the incident radiation
is linearly polarized, the scattered radiation will be elliptically polarized; if the
incident radiation is unpolarized, the scattered radiation will exhibit partial po-
larization depending on the angle of observation. Examples of this in the long-
wavelength limit were described in Section 10 I (see Figs 10 2 and 10 3)
The coefficients a±(/) and /3±(1) in (10.57) are determined by the boundary
cond!Uons on the fields at r a. Normally this would involve the solution of the
Maxwell equations inside the sphere and appropriate matching of solutions
across r a. If, however, the scattcrer is a sphere of radius a whose electromag-
netic properties can be described by a surface impedance Zs independent of po-
sttlon (for thts the radtal vanatwn of the fields JUSt inside the sphere must be
rapid compared to the radius), then the boundary conditions take the relatively
(10 64)
where E and 8 are evaluated just outside the sphere. From (10.55), (10.57), and
en X 8 = 2: iV411"(21 + 1) 1-' a
I
l
-~
VA
X
.
]I
a±(l)
+ ----;:;-
h(l)
l XVI
*Our results are not completely general. If the sum over m had been included in (10.57), the scattering
cross S€ction 'j!,'OlllG have a snm guer I and m with the absolute squares of a(! m) and (1(1 m) The
total cross section would stay as it is, with a(l)--> a(!, m ~ ± 1) and {3(1)--> {3(1, m ~ ±1), depending
on the state of polarization of the incident wave (10.46). The absorption cross section can be deduced
from taking the diffeiCilee of o 1 and Usc·
where ;r ka and all the spherical Bessel functions have argument x. The bound-
ary condition (10.64) requires that, for each I value and for each term X,., anct
n x X 1m separately, the coefficients of Etan and n X B be proportional, according
to
1 1
]t 2 Z0 x dx x ]! 2 1
jl + ~+il) hji> = i(i) ~ ~ l x(jl + ~;l) hi'))]

(10.65)
By means of the relation 2j, = h)ll + h)2 l, the coefficients a"(/) and f3=Jl) can be
h) 2 ) - i Zs I1- d (2)
- (xh 1 )
Zo x dx
a=~t) + I (10.66)
I Z \ 1 d
ni l d ~xn 1 ·)
L Zo X X _
w1th {3'"'(1) havmg the same form, but with ZsiZ0 replaced by its reciprocal. We
note that with the surface impedance boundary condition the coefficients arc the
same for both states of circular polarization.
For a given Z, all the multipole coefficients are determined and the scatter-
ing is known in principle. All that remains is to put in numbers. Before proceeding
to a specific limit, we make some observations. First, if Z, is purely imaginary
(no dissipation) or if Zs = 0 or Z,--+ oo, [a+(/)+ 1] and [{3"(1) + 1] are numbers
of modulus umty. 'Ih1s means that a"'(l) and f3±(l) can be written as
(10.67)
where the phase angles 81 and 8/ are called scattermg phase shtjts. Specifically
tan 81 = jz(ka)/n 1(ka)
.}
;Y (xj1(x)) r1 n (;R\
tan oI
d
dx (xn 1(x))
.... x=ka
if Z, 0 (perfectly conducting sphere) and 81 - 8/ for Zs--+ oo.

The second observation is that (10.66) can be simplified in the low- and high-
frequency limits. For ka << l, the spherical Bessel functions can be approximated
according to (9.88). Then we obtain the long-wavelength approximation,
-2i(ka) 21 + 1 x - i(l + l)Z)Z0
(21 + 1)[(21 1)!!] X+ ilZ,IZ0
and the same form for {3"(1), with (Z/Z0 ) replaced by its inverse. For ka >> I,
we use (9.89) and obtain
1 (10.70)
Sect. 10.4 Scattering of Electromagnetic Waves by a Sphere 477
' "' ' o _I ' T "' I I d.

Wllll Pee~!) lllc UOU<U, -u-vJ • ~" '""' CVH!S- va
Vld -c uuu•,
independent of the actual value of Z" the scattering coefficients a-(l), f3-(l)
become small very rapidly as l increases. Usually, only the lowest term (l = 1)
need be retained for each multipole series. In the opposite limit of ka >> 1,
~lU./U) snows tnatTor~a, llle 'vv cue1 · .uavc -r raure-uca5-
nitudes but phases that fluctuate widely. For l ~ lmax = ka, there is a transition
region and for l >> lmmo (10.69) holds. The usc of a partial wave or multipole
expansion for such a large number of terms is a delicate matter, necessitating the
caretul use or mgital computers or approximatiun u1 me type msct"'"u
in Section 10.10.
We specialize now to the long-wavelength limit (ka << 1) for a perfectly
conducting sphere (Zs = 0), and leave examples of slightly more complexity to
tne problems. un1y the l 1 terms m ~lU.O-'J are imponam. rrom ~1u.v"J we
find
-1 2i 3
a-(1) = " f3-(1) = - 'l (ka)
(10.71)
rrom 1 ao1e 'l.t we ootam tnc ausuJuLc >l[UaJcu ,
I v V 12 IV 12
3 {1 -1- ~Ao2tl)
The cross terms can be easily worked out:
Thus the long-wavelength limit of the differential scattering cross section is

du
dfl ~ a2 (ka) 4 [~(1 + cos 2 ll) - cos OJ (10.72)
Equation (10.72) is the same as (10.16), found by other means and is valid for
either state of circular polarization incident, or for an unpolarized incident beam.
The generalizations to arbitrary incident polarization and to different surface
I. .J. _,, ' L£• <J, ,J--1. •< '" .-1 ,4' <J. ,J. •<
uv 'J w t-' "t-' .
The general problem of the scattering of electromagnetic waves by spheres
of arbitrary electric and magnetic properties when ka 1s not smal11s comphcatea.
It was first systematically attacked by Mie and Debye in 1908-1909. By now,
I_ -'- .I £ I. L .J--1;,J. .-l <1-. .1-.' ,. T'\. :J, ,£ <1,
w vc t-' "t-' uavv t-' ., • o
aspects of this important problem can be found in the books by Kerker, King
and Wu, Bowman, Senior, and Uslenghi and other sources cited at the end ot tne
chapter. The book by Bowman, Senior, and Uslenghi discusses scattering by other
<
' 0>
' .; _,
LH<O >jJI
' oc~.
For scatterers other than spheres, cylinders, etc., there is very little in the
way of formal theory. The perturbation theory of Section 10.2 may be used in
478 Chapter 10 Scattering and Diffraction-S I
10.5 Scalar Diffraction Theory

Although scattering and diffraction are not logically separate, the treatments tend
to be separated, w1th d1ffract10n bemg associated w1th departures from geomet-
rical optics caused by the finite wavelength of the waves. Thus diffraction tradi-
tionally involves apertures or obstacles whose d1mens10ns are large compared to
a wavelength To lowest approximation the interaction of electromagnetic wa>es
is described by ray tracing (geometrical optics). The next approximation involves
the diffraction of the waves around the obstacles or through the apertmes with
a consequent spreading of the waves. Simple arguments based on Fourier trans-
forms show that the angles of deflectiOn of the waves are confined to the region
(} :-s Aid, where A is the wavelength and d is a linear dimension of the aperture
01 obstacle. Tire various approximations to be discussed below all work best for
>. /d << 1, and fail badly for ,\ d or A > d
The earliest work on diffraction is associated with the names of Huygens,
Young, and Fresnel. The first systematic attempt to derive the Fresnel theory
from first principles was made by G. Kirchhoff (1882). Kirchhoff's theory, despite
1ts mathematical mcons1stency and 1ts phys1cal defiCienCies, works remarkably
well in the optical domain and has been the basis of most of the work on dif-
fraction. Vv'e first derive the basic Kirchhoff integral and ns operative approxi-
mations, then comment on its mathematical difficulties, and finally describe the
modifications of Rayleigh and Sommerfeld that remove the mathematical
inconsistencies.
The customary geometry in diffraction involves two spatial regions I and II,
separated by a boundary surface S" as shown m F'lg. 10.7. The surface S2 is
generally taken to be "at infinity," that is, remote from the region of interest.
Sources in region I generate fields that propagate outward. I he surface 5 1 1s
supposed to be made np of "opaque" portions (the boundary conditions are
discussed below) and apertures. The surface S1 interacts with the fields generated
in region I. reflecting some of the energy absorbing some of it, and allowing
some of the fields, modified by their interaction, to pass into region II. The an-
gular d1stnbut1on of the fields m reg10n II, the d1ffract10n reg10n, IS called the
diffraction pattern. It is the diffracted fields in region II that we wish to express
in terms of the fields of the sources and their interaction with the screen and
apcrtmes on .S\ or more precisely, in terms of the fields on the surface S1 It
r---...., / -..,
1 ( 11
"\\ I
/ ~
Sources j_ s2 ( J~l
\Sourcej II
s J \ _i
__/ ~ /
Figure 10.7 Possible diffraction geometries. Region J contains the sources of radiation.
Regwn II IS the dtffractlon regwn, where the fields satisfy the radiation condition. The
right-hand figure is also indicative of scattering, with a finite scatterer in region I instead
of an active soufce, and the surface S1 an atbihaty tnathernatical sat face enclosing the
scatterer rather than a material screen with apertures.
Sect. 10.5 Scalar Diffraction Theory 479
should be obvious that the geometry and mode of description is equally appli-
cable to scattering, with the sources in region I replaced by a scatterer (thought
of as a source being driven by the incident wave).
Kirchhoff's method uses Green's theorem (1.35) to express a scalar field (a
component of E 01 B) inside a closed volume V in te1ms of the values of the field
and its normal derivative on the boundary surface S. Let the scalar field be
•lr(x, t), and let it have harmonic time dependence, e-;wr. The field 1!J is assumed
to satisfy the scalar Helmholtz wave equation,
(10 73)
inside V. We introduce a Green function for the Helmholtz wave equation
G(x, x'), defined by
(V2 + k 2 )G(x, x') = - 8(x - x') (10.74)
In Green's theorem (1.35), we put cp = G, 1/J = 1/J, make use of the wave equations
(10.73) and (10.74), and obtain, in analogy to (1.36),
1'(x) = £ [1/l(x')n' • V'G(x, x') - G(x, x')n' · V'!/J(x')] da' (10.75)
where n' is an inwardly directed normal to the surfaceS. Equation (10.75) holds
if x is inside V' if it is not the left-hand side vanishes
eikR
G(x, x') = R (10.76)
4
where R = x - x'. With this Green function, (10.75) becomes
da' (10.77)
This is almost the Kirchhoff integral. To adapt the mathematics to the diffraction
context we consider the volume V to be that of region II in Fig. 10.7 and the
surfaceS to cons1st of S 1 + S2 • 'Ihe mtegral overS 1s thus dJV1ded mto two parts,
one over the screen and its apertures (5 1 ), the other over a surface "at infinity"
(S 2 ). Sinee the fields in region II are assumed to be transmitted through 5 1, they
are outgoing waves in the neighborhood of 5 2 • The fields, hence 1/J(x), will satisfy
a radiation condition
eikr
"'---'> f(IJ, 1>) - (10.78)

r'
With this condition on 1/J at 5 2 it is easily seen that the contribution from 5 2 in
(10.77) vanishes at least as the inverse of the radius of the hemisphere or sphere
as the radius goes to infinity. There remains the integral over 5 1. The Kirchhoff
integral formula reads
1 ( e;kR ,.- '
1/J(x) = --::;- I n n' · V' 1/J + ikll da' (10.79)
L \ 'I
with the integration only over the surface 5 1 of the diffracting "screen."
.:IRO rhoniPT 1()
. •<~nil ;on -~l
To apply (10.79) it is necessary to know the values of 1/1 and a!/J/an on the
·"· S . I Tnless the nroblem has been solved bv other means these values arP
HV<~HVHHo
1 TT' -'-L L'L''
OUJc'y
-'- . ~t'Jc' ·~
,j' ,/, ~ rl • ,1./•
~"HI
on S1 in order to calculate an approximation to the diffracted wave. The

Kirchhoff approximation consists of the assumptions:
'
L
. '"
~· auu
'-1.
~ .
u•;nun
, . tl.
• m u o u '-' ''-'~J """~"' vu '-'1 T -r
2. The values of 1{1 and a!/J/an in the openings are equal to the values of the
incident wave in the absence of any screen or obstacles.
The standard diffraction calculations of classical optics are all based on the
.u T< • .L •'- •'- ~ ~n lu ' ~
;;ct:~y. The;: is, in fact,

a .serious mathematical inco~sistency in the.Jassumptions
of Kirchhoff. It can be shown for the Helmholtz wave equation (10.73), as well
•'- £0 • • .£
as tut we "'4' 'llJdl ll ljJ dllU Oljl/Ofl UV<H VH UHJ HUH'-' ouaac,c;, a~'-' L'-'~V
then 1{1 - 0 everywhere. Thus the only mathematically correct consequence of

the first Kirchhoff assumption is that the diffracted field vanishes everywhere.
This is, of course, inconsistent with the second assumption. Furthermore, (10.79)
rl. ',lrl (' t'- rl .1. ~• ,/, ~nrl :J,J,/:Jn
J '1
The mathematical mconsJsrencies mLTie"Kircnnon approximmion canoe re-

moved by the choice of a proper Green function in (10.75). Just as in Section
1.10, a Green function appropriate to Dirichlet or Neumann boundary conditions
r<~n hP ·• .rl Tf ,,, ;, known or annroximated on the surface S, a Dirichlet
r< r< I. • _£_ .
' - " ' - ' ' - ' " LUH'-'UVH '-'D\A• A"J• ow IOIJ "IS
£.
GD(x, x') = 0 for x' on S (10.80)

is reouired. Then a rreneralized Kirchhoff interrral, equivalent to (1 0.79), is
!/J(x) = J 1{1(x') -aGo

sl an
, (x, x') da' (10.81)
and a consistent annroximation is that !11 = 0 on S1 except in the openings and 1/1
to
.1
'-''i uao
...
<v " ' " '
• •1
'a''-' m
...
"'"' -r •0
•
•
T£ • ' -
•
'1 ~ • t. ,j' ,/, .
'I'
t
be approximated, a Neumann Green function GN(x, x'), satisfying

r~r:.
(x xT- u Tor x on :-, ( lU.ISLJ
dn
is employed. Then the generalized Kirchhoff integral for Neumann boundary
.
-"IUVHO ~.
'"'
!/J(x) = - J- ~ (x')GN(x, x')

s, n
da' (10.83)
, '- £. .L ~
'b' a "IJIJl' -.:<liT U'-'
For the imnortant snecJal circumstance m Which the surtace .'1, IS an mnnire
plane screen at z = 0, as shown in Fig. 10.8, the method of images can be used
to give the Dirichlet and Neumann Green functions explicit form:
1 e' e'~"
Cin "'(x. x') + (lU.i'l'+)
'I 'IT n H
·~
11ect. ~11.:. l!caiar untracnon ~ neory '+0-L
oP
P'o. /
'foi' ' Ht.O 1J llll LUl it
pomt source at r , a plane screen with

apertures, and an observation point at
P. The distances from the clement of
area da' in the aperture to the points P
and P' are r and r', respectively. The
angles 8 and 8' are those between r and
n, and r' and -n, respectively.
where R = x - x', and R' = x - x", x" being the mirror image of x'. Explicitly
we have
R = [(x - x') 2 + (y - y') 2 + (z - z'?F 12
R' = [(x - x'? + (y - y') 2 + (z + z'?t 2
The generalized Kirchhoff integral (10.81) ( 1/J approximated on S1 ) then takes
the form,
ifi(x) = }5__ J
2m s 1 R
eikR (1 + _i ) n' . R 1/l(x') da'
kR R
(10.85)
An analogous expression can be written for (10.83), both results attributable to

the ubiquitous Rayleigh.*
Comparison of (10.85) with (10.79) shows that (10.85) can be obtained from
(10.79) by omitting the first term in the square brackets and doubling the second
term. The Neumann result (10.83) specialized to a plane screen is equivalent, on
the other hand, to doubling the first term and omitting the second. It might thus
appear that the three approximate formulas for the diffracted field are quite
different and will lead to very different results. In the domain where they have
anv 1ahle valirlitv thev vielrl in fact. verv similar resnlts This ~"n he nn-
on one siue 01 a p•ane screen anu an ouservatiou puint r un Lilt: uwt:r siut:, as
shown in Fig. 10.8. The amplitude of the point source is taken to be spherically
oPtrir ~nrl Prm~l tr. ,ikr' lr 1 Rr.th P ~nrl P 1 ~rP <ooonmPrl tr. hP "'"""-
equivalent assumptions in ~ lU.IS)) ana irs Neumann oounaary conairion counter-

oart the diffracted fields for all three aonroximations can be written in the com-
mr.n fr.rm
" r ,ikr ,ikr'

1/J(Pl- - . I - 0(8, 8') da ( 1U.Ii6 l
Lm J apecture< r r
·-'" ;n h;< eorlv ~; ·

* ~ n ·. 10., u
482 Chapter 10 Scattering and Diffraction-S!
where the obliquity factor 0( 8, 8') is the only point of difference. These factors
are
cos 8 1)
( 1/J approximated on S
0(8, 8') = cos 8' ( :~ approximated on s1)
i(cos 8 + cos 8') (Kirchhoff approximation)

--~ •• - '"''IS· '"0
,1'
0
~ ~
are
• T'"
•v.u.
y-,
'w ut'vaw~~
•• L -1."
'b'
,f ~1
0
.. . _, . ' ....
large compared to a wavelength the diffracted mtens1tv IS confined to a n:Jrrm..
•
rl'
'b' •
" -1
''] ~;
~
~
_r
~ '< VV vCOll Llle
LWO exponen,Ia, taCLors m (lU.~OJ. n me source pmm t ana me observation
noint P :1r.e f:1r from th.e o ' ;n tPrmo nf thP •'•Jro rJ;mPno;r.no <ho ~h1'
'J ' ' T
factor in (10.86) can be treated as a constant. Then the relative amplitudes of the
different diffracted fields will be the same. For normal incidence all obliquity
factors are approximately unity where there is appreciable diffracted intensity.
In this case even the absolute magnitudes are the same.
The discussion above explains to some extent why the mathematically in-
consistent Kirchhoff approximation has any success at all. The use of Dirichlet
or Neumann Green functions gives a better logical structure, but provides little
practical improvement without further elaboration of the physics. An important
deficiency of the discussion so far is its scalar nature. Electromagnetic fields have
vector character. This must be incorporated into any realistic treatment, even if
approximate. In the next section we oroceed with the task of obtainin~> the vectnr
Prn •
-, r.f~> ,ff~r ! v: eh ~ ,ff ' 't ,_1__£ ,.]_
vu,
0 't'
JFI £ T.T D . .L .r ~1. T7° .1.1. . rr T. ~ .1

~u.v
• '7 ••'-' "J un;; .._..._., LIIIIU.J.J .I.IHIOJ;;I Ul
The Kirchhoff integral formula (1 0.79) is an exact formal relation expressing the
scattered or diffracted scalar field 1/f(x) in region II of Fig. 10.7 in terms of an
integral of 1/1 and iJI/f/iJn over the finite surface S1. Corresponding vectorial rela-
tions, expressing E and B in terms of surface integrals, are useful as a basis for
a vectorial Kirchhoff approximation for diffraction (Section 10.7) and scattering
(Section 10.10), and also for formal developments such as the proof of the optical
theorem (Section 10.11).
To derive a Kirchhoff integral for the electric field, we begin with (10.75) for
each rectangular component of E and write the obvious vectorial eauivalent
r
E(x) - <P fF.(n' • V'!;) !;(n' · V')F:l da' fHl R7)
Js
nrovided the ooint x is insirle thP. vnl11mP V hv thP ~ I-f PTP ~<
11 f'l "7<:\ •• ;. . '1. r• 1 , • _1 • ~
-1. "11
. -" " TT
\ 10 UH v'--\vU HHU \Hv
, . "~""- '. WC WlH
H ~
specny lito be tne mhmte-soace Green functiOn. ( 10.76) but for the nresent we
1o~ oo ;t ,] of." .C /1A "7A\ D. ',L .C
.• .J \ • T J• ~ HUH •v ~U~ ~~HmH '" \ '--lUO
vector calculus that apply to well-behaved functions, while G is singular at

x' = x, we must exercise some care. We imagine that the surfaceS consists of an
outer surface S' and an infinitesimally smaller inner surface S" surrounding the
liect. lU.b vector ~qmvalents ot tne Kircnnon Integral 'li"iJ
pomtx x. 1 nen, rrom Lrreen s tneorem, tne 1ert-nana swe or~ 1U.li 1) vamsnes.
Of course, evaluation of the integral over the inner surface S", in the limit as it
shrinks to zero around X 1 = x, gives -E(x). Thus (10.87) is restored in practice,
but by excluding the point X 1 = x from the volume V the necessary good math-
ematical behavior is assured. With this understanding concerning the surface S,
we rewrite (10.87) in the form
0 = £ [2E(n 1 • V1 G) - R
1
• V1 (GE)] da 1
The divergence theorem can be used to convert the second term into a volume
integral, thus yielding
r r
0 = Cf! 2E{nl · V 1 G) da' + I V'' 2 (GE) d 3 x'
Js Jv
With the use ot V"A - Vl v ·A) V X lV X A) tor any vector held A, and the
vector calculus theorems,
Jv V<P d x = fs n,P da 3
(10.88)
Jv V x A d x = fs (n x A) da 3
where <PandA are any well-behaved scalar and vector functions (and n is the
outward normal), we can express the volume integral again as a surface integral.
We thus obtain
I
u '~'~ [L~~n • v Lr) n ~v · ~Lr~)) -r n x ~ v x ~Lri'.))J aa
~~"Y"''b ~~•
•'-
HW
• A: A
~· .. ~· ~·
.c •'-
... ~ ':''~~~~· ' ""' ~~, ""~
A • 1. .
uo~ v•
.c
me JVtaxweu equanons, v · ~ U, V X 1'. zwn, we nna
0 = *~ [iw(n' x B)G + 2E(n' · V1 G) - n'(E · V'G) + n' x (V'G x E)] da'
Expansion of the triple cross product and a rearrangement of terms yields the
final result,
E(x) = £ [iw(n 1 X B)G + (n 1 X E) X V'G + (n 1 • E)V 1 GJ da' (10.89)

_,_ +L .I. T/ 1. A A J.. .c. (' ,,_ I A
cv '"~ ·~·"~ v Y_ '"':' v ~v oo~'"" "~ yv"'' ~ ~. ' "'
ana10gous expressiOn ror n can De omamea rrom ~ 1 u.liY) Dy means or me suD-
stitutions E ---'> cB and cB ---'> -E.
Pm {1 () RQI i< thP Pnniv<>lPnt ,-,f thP <r<>br c nb {1 () 7<;1 T,-,
,_ . '
tuv
: '
_ Hll~e•m \ <V. I 7
/•
j, WOO UHC v
'
ouy v<
r
. 0 u< " ' " ''""
l:'Ig. lU. 1 ana Jet me surrace .) De maae up or a nmte surrace .) 1 surrounamg tne
sources or scatterer and a surface So "at infinitv." There is no loss of 2:eneralitv
in taking 5 2 to be a spherical shell of radius r0 ---'> oo. The integral in (10.89) can
be written as the sum of two integrals, one over S 1 and one over S2 . On the
surface 5 2 the Green function (10.76) ts gtven, for large enough r0 , by
eikr'
V'G ~ -ikn'G
Then the contribution from S2 to (10.89) is
,), = ik,), [c(n' X B) - (n' X E) X n' - n'(n' · E)]G da'

J.~.2 Js 2
or
f = ikf (c(n' x B) - EJG da'
The fields in region II are diffracted or scattered fields and so satisfy the condition
of outgoing waves in the neighborhood of S2 . In particular, the fields E and B
are mutually perpendicular and transverse to the radius vector. Thus, on S2 ,
E = en' x B + 0(1/rA). This shows that
and the contribution from the integral over S2 vanishes as r0 ~ cc. For the ge-
ometry of Fig. 10.7, then, with S2 at infinity, the electric field in region II satisfies
the vector Kirchhoff integral relation,
E(x) = £ [iw(n' x B)G + (n' x E) x V'G + (n' · E)V'GJ da' (10.90)
where G is given by (10.76) and the integral is only over the finite surface S1 •
It is useful to specialize (10.90) to a scattering situation and to exhibit a
formal expression for the scattering amplitude as an integral of the scattered
fields o~er sl. The geometty is shown in Fig. 10.9. On both sides of (10.90) the
Figure 10.9 Scattering geometry. An incident plane wave with wave vector Ji., and
fi€lds (E, B,) is scatt€n:~d by an obstacle (the scatterer), giving rise to scattered fields
(E." B,) that p10pagate as sphetically dive1ging waves atla1ge distances. The smface S1
completely encloses the scatterer.
Sect. 10.7 Vectorial Diffraction Theory 485
fields are taken to be the scattered fields (E, B,), that is, the total fields (E, B)
minus the incident wave (E;, B;). If the observation ooint Pis far from the scat-
terer, then the Green function and the scattered electric field take on their as-
ymptotic forms,
l e'~' . ,
G(x, x') ~-- e-lk-x
47T r
ikr
E,(x) ---+ ::____ F(k, k 0 )
r
where k is the wave vector in the direction of observation, k 0 is the incident wave
vector, and F(k, k 0 ) is the (unnormalized) vectorial scattering amplitude. In this
];m;t ~·r; -il<r. Thllo (1() om"~" hP . ~·· ~~ . ~. . /", rl.
- r "75 ''JC • J
scattering amplitude F(k, k 0 ):
F(k, ko) = - _
47T s,
1
r e-ik·x'[ w(n' X B,) + k X (n' X E,) - k(n' · E,)] da' (10.91)
Note caretuuy now J<~J;:, R 0 ) depends explicitly on file outgomg d1rect10n of k.

.
The dependence on the incident direction specified by k 0 is implicit in the scat-
.
tc;1c;U Hc;1U> £. 5 <lllU D 5 • 01111.;c; We; Uldt .... .,
. .
• v, ll musr oe rrue matm t 1 v.'il 1
the component parallel to k of the first integral cancels the third integraL It is
therefore convenient to resolve the integrand in (10.91) into components parallel
and perpendicular to k, and to exh1b1t the transversality of F explicitly:
1
k ~ k , 'ck X (n' X B,)
F(k kn) X ,-' •X n' X 1?. nn' (1() 0?\
47TI J s, L
k
A hPrn~t;uPlu U!P r-~n ~d..- frw thP ~ rriP ~• ~ ~n.-1;~•;~~ n ;,,
•A :
<lllU ' lllUll 0:, 1 II" >0 101 V <011 U Y
.,. • I'tK, Ko)

i I
e [WE • tD »,) + E • ~R x ~n x ~<-,))J aa
477 fs,
l'.
(10.93)
The terms in square brackets can be interpreted as effective electric and magnetic
surface currents on S1 acting as sources for the scattered fields. The various equiv-
alent forms (10.91)-(10.93) are valuable as starting points for the discussion of
the scattering of short-wavelength radiation (Section 10.10) and in the derivation
of the ootical theorem (Section 10.11 ).
.
1n 7 TT ·'" .p:~1 n;J.'.t'. TJ.n~P.,
"JJ '··~··~·· ~ ··~~· J
·'".
'1'1-.a ;n1 V' ~1-.' ,H · , , ,1 r1n
•
omJ 1 .-1 <1 1
·"'
\ ~
~t't
• • • 0
mate tneory or amracuon m exacuy me same manner asaescrilled below (lU.7lJ)

for the scalar theory. Unfortunately, the inconsistencies of the scalar Kirchhoff
"1-'1-- ll1Vll f'<'l MM.
For the special case of a thin, perfectly conducting, plane screen with aper-
' ' 'L 1 •L ' ' ' ' ' ' •' '
d,uw IV Vv~IVUU" 1' , dl\.111 tV tile;
'
Ktrchhoff mtegral ( 10.1-H) or (10.~5 ), m wh1ch tile l:JouriOary cond1t1ons are sat-
isfied; these relations, moreover, are amenable to consistent approximations. The

plane screen is taken at z = 0, with the sources supposed to be in the region
z < 0, and the diffracted fields to be observed in the region z > 0. It is convenient
to divide the fields into two parts,
E = E(o) + E' B = B<OJ + B' (1 0.94)
'
where E<"J, n<oJ are the fields produced by the sources in the absence of any
screen or obstacle (defined for both z < 0 and z > 0), and E', B' are the fields
~~noM.J hu t!-,p nrPoPn~P n f thP nl~nP · J:;, ~ n V I Dl <C --''~~- •
-J r r ' '
uew~, wnite ior z <--- o, Lney are LIIe reueCLea nelC s. we will call E , H' tne scattered
fi~e]cl, wh~en · · hnth 7 < 0 >lncl 7 > 0 Th., · ~rk ~~n hP ~"
• • • d •.
NU'-'''-'U ao uavm15 "'"'" "' '"'-' o ·cOUllt'lll Ut:IlMlY aUU > ·cnarge
densttv that are necessarilv nroduced on the scr""n tn · · th" hm · Mn.
ditions. Certain reflection properties in z of the scattered fields follow from the
fact that the surface-current and -charge densities are confined to the z = 0 plane.
A vector potential A' and a scalar potential <J:>' can be used to construct E' and
B'. Since the surface current flow has no z component, A~ = 0. Furthermore,
A~, A;, and <J:>' are evidently even functions of z. The relation of the fields to the
potentials shows that the scattered fields have the reflection symmetries,
E~, E;, B~ are even m z
(10.95)
E~, B~, s; are odd in z
The fields that are odd in z are not necessarily zero over the whole olane z = 0.
Wh<erP. thP rnnclJw.tino- F' =I= 0 · ~;M ~n ;~t,-;1 onrf~~,-~1-,~r~.
• • - ~- ';=''
'J• v11 '"" <vvv >Iuvo w ' " " " . ~ ny, nom tangenuat
comnonents of B imnlv a surf>lrP.-rnrr<ent 't" Pnn~l ;n m~o-n;hulP ~nrl rl;rp~-
], ·•'" .--'· :r ~'" r. •'" •
that c tf , tf vamsh. Thls leacts to the statement that m the anertures of a
perfectly conducting plane screen the normal component of E and the tangential
components of B are the same as in the absence of the screen.
The generalized Kirchhoff integral (10.83) for Neumann boundary condi-
tions can be applied to the components of the vector potential A'. The normal
derivatives on the right can be expressed in terms of components of B'. The
result, written vectorially is
A'(x) = -
1 J e'kR
(n x B') - da' (10.96)
2 '1T screen R
In view of the preceding remarks about the surface current and the tangential
comnonents of B 110.96\ could nerhans have been writt"n rlmxm The
~~ __ , 1 - I
.. \"'J ") ' , I.
In (10.96) and (10.97) the integrand can be evaluated on either side of the screen
with n being normal to the surface. For definiteness, we specify that n is a unit
normal in the positive z direction and the integrand is to be evaluated at z = o+.
The integration extends over the metallic part of the screen; B;an = 0 in the
apertures. The electric field E' can be calculated from E' = (ilwf.LE)V x B'.
Equation (10.97) can be used for approximations in a consistent way. It is
Sect. 10.7 Vectorial Diffraction Theory 487
most useful when the diffracting obstacles consist of one or more finite fiat seg-
ments at z - 0, for example, a c1rcular d1sc. I hen the surface current on the
obstacles can be approximated in some way for instance, by using the incident
field B(o) in the integrand. We then have a vectorial version of the generalized
Kiichhoff's appwximation of the preceding section.
It 1s useful to construct an expresswn eqmvalent to (10.97) for the electric
field. From the symmetry of the somce-free Maxwell equations with respect to
E and B it is evident that the electric field E' can be expressed by analogy with
E'(x) (10.98)
2w s R
where 1t 1s assumed that E' 1s known on the whole surfaceS, at z - o+. The
upper (lower) sign applies for z > 0 (z < 0) It can he verified that (1 0 98) satisfies
the Maxwell equations and yields consistent boundary values at z 0. The reason
fm the difference in sign for z ""= 0, as compared to (10.97) forB", is the opposite
refiectwn properties of E' compared to B' Isee (10.95)].
There is a practical difficulty with (1 0 98) as it stands The integration in
(10.98) is over the whale plane at z 0. We cannot exploit the 'anishing of the
tangential components of the electric field on the metallic portions of the screen
because it is the total electric field whose tangential components vanish, not those
of E'. The difficulty can be removed by use of linear superposition. We add E(o)
to the integrand in (10.98) to obtain the full electric field, and subtract the cor-
responding integral. We thus have, for the diffracted electric field,
eikR
E'(x) = :!::...1:._ V
2w
X fs, (n X E) Rda' (10.99)
where
(10.100)
The integrand in (10.99) now has support only in the apertures of the screen, as
desired. But what is the extra electric field E('l? Just as (10.98) gives the extra
(diffracted) field for z > 0 in terms of a surface integral of itself over the whole
screen, so (10.100) is equal to the "source" field E(OJ in the region z > 0. But
because E(1 ) is defined by an integral over the surface at z = 0, it respects the
symmetries of (10.95). A moment's thought will show that this behavior means
that for z < 0 the sum E(o) + E(') describes the fields of the sources in the presence
of a perfectly conducting plane (with no apertures) at z = 0: E(') (and its partner
B(ll) are the reflected fields!
If in (10.99) we transfer E(l) to the left-hand side, we find for z > 0 the total
electric field, now called the diffracted field, given by
1
Ect;n(x) = - V X
J (n X
e;kR
E) - da' (10.101)
2 7T apertures R
where the integration is only over the apertures in the screen and E is total
tangential electric field in the apertures. In the illuminated region ( z < 0) the
total electric field is
(10.102)
where for both regions Edifr(x) is given by (10.101 ). This solution for the diffracted
electric field in terms of the tangential electric field in the apertures of a perfectly
conducting plane screen was I1rst obtained by Smythe.* It can serve as the basis
of a consistent scheme of approximation, with the approximate solutions for E..utff
satisfying the required boundary conditions at z = 0 and at infinity. Some ex-
amples are discussed in a later section and in the problems.
10.8 Babinet's Principle qf Complementary Screens

Before discussing examples of diffraction we wish to establish a useful relation
called Babinet's principle. Babinet's principle relates the diffraction fields of one
diffracting screen to those of the complementary screen. We first discuss the
principle in the scalar Kirchhoff approximation. The diffracting screen is assumed
to lie in some surface S, which divides space into regions I and II in the sense of
Section 10.5. The screen occup1es all of the surtace S except for certam apertures
The complementary screen is that diffracting screen which is obtained by replac-
ing the apertures by screen and the screen by apertures. If the surface of the
original screen is Sa and that of the complementary screen is Sb, then Sa + Sb =
S, as shown schematically in Fig. 10.10.
If there are sources inside S (in region I) that give rise to a field 11i(x), then
in the absence of either screen the field !fo(x) in region II is given by the Kirchhoff
integral (10. 79) where the surface mtegral 1s over the entire surface S. W1th the
screen Sa in position, the field 1/Ja(x) in region II is given in the Kirchhoff ap-
proximation by (10.79) with the source field 1/J in the integrand and the surface
integral only over Sb (the apertures). Similarly, for the complementary screen Sb,
the field 1/!h(x) is given in the same approximation by a surface integral over Sa.
Evidently, then, we have the following relation between the diffraction fields '""
(10.103)
This is Babinet's principle as usually formulated in optics. If ifJ represents an

incident plane wave, for example, Babinet's principle says that the diffraction
*W. R. Smythe, Phys. Rev. 72, 1066 (1947). See also Smythe, Section 12.18.
" "'0 ~. 0 ' ~ 0
0 •
-~ .. --
.c LI.SlQ
patterns away from the incident direction are the same for the original screen
and its complement.
The result (10.103) also follows from the generalized Kirchhoff integrals
(1 n Sl1 \ Ar (1 n Sl':t\ ;f th~ ~mnl;t .. CI~ nr ;h. nnrm~l riPr;, .tivP i~ t>ikten Pflll>il tn-th>~t
'
U< lllC ll
-/
WdVC 111 lllC ctpCllU~

. ' .. L
,~,..,~,~, '"
<L
.u~ opua v•
.. _£ ,,_
w~
Kirchhoff approximation. All these formulations of Babinet's principle are un-

satisfactory in two respects: They are statements about scalar fields, and they are
based on a Kirchhoff approximation.
' . ~ '
J-1. '-o ll 01 ~ , """ "<.-1U> <.-au u<.-
made tor a thm, pertectty conouctmg plane screen ana us comp1emem. 1 ne resuJL
follows from the two alternative formulations of this diffraction problem given
in the preceding section. The original diffraction problem and its complementary
problem are defined by the source fields and screens as follows:
~
'-./ K " " ' '-''"L

E(O) 8(0). Sa (10.104)
' '
COMPLEMENT
E(o) = c8<"l 8(o) = - E<"lfc· sb
' c '
'
.. d .+ •L. .1
UHO '
ana nas source ue10s wim opposi1e p01arizaLion c "' oll<.>.
'uao a -.-rrcrr=
·. v.
>'U1 lllC
screen Sa the electric field for z > 0 is, according to (10.101 ),

1 {., e'kR
E(x) = ;:;- V X _ (n x E) 0 da' (10.105)
For the complementary screen Sb we choose to use (10.97) instead of (10.101) to

express the complementary scattered magnetic field 8~ for z > 0 as
1
8:.(x) = - V X
J (n X 8:) -
e'kR
da' (10.106)
27T sb R
1n both (lU.lu::>) and (lU.lUb) the imegralion is over me screen ,)z, oecause or me
boundary conditions onE and 8; in the two cases. Mathematically, (10.105) and
(10.106) are of the same form. From the linearity of the Maxwell equations and
the relation between the original and complementary source fields, it follows that
m the reg10n z > 0 the total electncneiOTortfle screen .:,a 1s numencauy equal
to c times the scattered magnetic field for the complementary screen Sb:
E(x) = c8~(x)
The other fields are related by

H~X) -E~~X)/C
. . . . . . .
wnere me s1gn IS a ''l' -unrreT~ "1' uc111 v1 uu'~ _ 1au1uuv11 0
flux at infinity, just as for the source fields. If use is made of (10.94) for the
complementary problem to obtain relations between the total fields in the region
A n ,_- • - - ' '- _, ' • ~- ' <1-o' .-1 ;,,
"- v, ~ J P' Hc<p<c •v• _a .r , I:-'" •y -o
complement states that the ongmarTieiiTS\E,-nJ auu Lne ·r 1 y ue1u~
(E. R) are related according to
E - c8 c = E(o) (10.107)
8 + E)c = 8(o)
approximate, statements for the idealized problem of a perfectly conducting

screen. For · situations but
screens whose radii of curvature are large compared to aperture dimensions,
. .
says that the diffracted intensity in directions other than that of the incident field
is the same for a screen and its complement. The polarization characteristics are
but this · ·
source fields (10.104).
The rigorous vector formulation of Babinet's principle is very useful in mi-
crowave problems. For example, consider a narrow slot cut in an infinite, plane,
along the slot and the electric field perpendicular to it, as shown in Fig. 10.11.
The radiation pattern from the slot will be the same as that of a thin linear
antenna with its ·
9.2 and 9.4. The polarization of the radiation will be opposite for the two systems.
Elaboration of these ideas makes it possible to design antenna arrays by cutting
suitable slots in the sides of waveguides.*
10.9 Diffraction by a Circular Aperture;
The subject of diffraction has been extensively studied since Kirchhoff's original
and in microwave generation and transmission, where more accurate solutions

are needed. Specialized treatises are devoted entirely to the subject of diffraction
and scattering. We will content ourselves with a few examples to illustrate the
use to com-
pare the accuracy of the approximation schemes.
as constants.
Sect. 16.1} Diffraction by a Circular Apertme, Remarks on Small Apertmes 491
Fignre 10.12
factor kR in e;kR needs to be handled with some care. With r >> d, it can be
expanded as
2r
where n = x!r is a unit vector in the direction of observation. The successive
terms are of order (kr), (kd), (kd)(d!r), . ... The term Fraunllofer diffraction
applies if the third and higher terms are negligible compared to unity. For small
diffracting systems this always holds, since kd << 1, and we have supposed
d!r << 1. But for systems that are large compared to a wavelength, (kd 2 /r) may
be of order umty or larger even though dlr << 1. I hen the term fresneldzffractwn
applies In most practical applications the simpler Fraunhofer limit is appropriate.
Far enough from any diffracting system it always holds. We consider only the
Fraunhofer limit here (except for Problem 10.11).
If the observation point IS far from the dlffractmg system, expansion (9. 7)
can he used for R - lx - x' I Keeping only lowest order terms in (1/kr), the
scalar Kirchhoff expression (10.79) becomes
where x' is the coordinate of the element of surface area da', r is the length of
the vector x from the origin 0 to the observation point P, and k = k(x!r) is the
wave vector in the direction of observation, as indicated in Fig. 10.12. For a plane
surface we can use the vector expression (10.~0~), which reduces in this limit to
ie;h ( . ,
E(x) = - k X n X E(x')e-'k·x da' (10.109)
As an example of diffl action we consider a plane ·.vave incident at an angle

a on a thin, perfectly conducting screen with a circular hole of radius a in it. The
polarization vector of the incident wave lies in the plane of incidence. Figure
10.13 shows an appropriate system of coordinates. The screen lies in thex-y plane
with the opening centered at the origin. The wave is incident f1 om below, so that
the domain z > 0 is the region of diffraction fields. The plane of incidence is
taken to be the x-z plane. The incident wave's electric field, written out explicitly
in rectangular components, is
Ei = Eo(El cos a - E3 sin a)eik(zcosa+xsina) (10.110)
In calculating the diffraction field with (10.108) or (10.109) we will make the
customary approximation that the exact field in the surface integral may be re-
placed by the incident field. For the vect01 relation (10.109) we need
~
E 'z
tn X
0 uE- E
2
cos a eiksinux' (10.111)
Figure 10.13 Diffraction by a circular
Then, introducing plane polar coordinates for the integration over the opening,
we have
2
. ""E
E(x) te 0 COS a (k X Ez) JI" p dp Jf "
d{3eikp[siHuc<>Sfl sin8cos(¢ fl)] (lO.l1 2)
27Tr o o
where fJ, <fJ are the spherical angles of k. If we define the angular function,
t (sm 2 fJ + sin 2 a 2 sin f) sin a cos ¢>) 112
the angular integral can be transformed into
fe(kpi;)
27r Jo '
Then the radial integral in (10.112) can be done directly. The resulting electric
field in the vector Smythe-Kirchhoff approximation is
(10.113)
The time-averaged diffracted power per unit solid angle is
(10.114)
where
. D
' (10.115)
is the total power normally incident on the aperture. If the opening is large com-
pared to a wavelength (ka >> 1), the factor [2J 1 (ka~)lkagF peaks sharply to a
value of unity at i; 0 and falls rapidly to zero ('Nith small secondary maxima)
within a region t:J.g ~ (l!ka) away from g = 0. This means that the main part of
the wave passes through the opening in the manner of geometrical optics; only
slight diffraction effects oeeur. * For ka 1 the Bessel function varies compat a-
*To see this explicitly we expand garound the geometrical optics direction 8 ~ "· q, ~ 0:
g~ v (e - a) 2 cos2 a + </> 2 sin2 a
For ka >> 1 it is evident that kal; >> 1 as soon as 8 departs appreciably from "· or </> from zero. or
,..,.,
"ecr. J.u., uinracriun oy a . "'I'"' ouuo, , un '""""
"r/o'
tively slowly in angle; the transmitted wave is distributed in directions very dif-
ferent from the incident direction. For ka << 1, the angular distribution is entirely
determined by the factor (k X E 2 ) in (10.113). But in this limit the assumption of
an unperturbed field in the aperture breaks down badly.
The total transmitted power can be obtained by integrating (10.114) over all
angles in the forward hemisphere. The ratio of transmitted power to incident
io r~]]prJ thP frnn ' ;,., ·'"" :n~' T·
"
""< ~ rh rrr/2 T. ( /<tJJ:.)
T I dch I (cos"& + cosLch sin"&) Sill 8 de (10.116)
7T Jo Jo ~
In the two extreme limits ka >> 1 and ka << 1, the transmission coefficient
approaches the values,
Hit: lU11!;-WdVt;;lt;;11!;l11 1111111 VW '---'-- ~ ' £

e
~ ~(ka? cos a,
V' VU> Ul'r
,
e<vuo,
ka >> 1
ka << 1
IUO~
out It snows mat tne transmiSSIOn IS small ror very sma11 uo1eS. ror norma1 inci
den<Ce (a - 0) the transmission coefficient (10.116) can be written
T = r o
2
ff(ka sin e) ( -.-
2 - sin 8) de
Sill 8
With the help of the integral relations,
r/2 J~(z
o
sin e) ~e =
Sill 8
fzo
lzn(t) dt
t (10.117)
1"7T/2 r2z
I ,znr.. c;n tnc;n
., Jl ,.~ Jl - -r T l. (t\ tit
Jo 2z Jo
and the recurrence formulas (3.87) and (3.88), we can put the transmission co-
efficient in the alternative forms
1 =
1 --
ka m~o
2:
l><n+l(2ka)
T=
1 {lka T.Jt\ dt
1 -
2ka Jo
The transmission coefficient increases more orTess monotomcally as ka mcreases,
with small oscillations superposed. For ka >> 1, the second form can be used to
.L • . • .
vu<am au aoy "Y '"~vAl'''
1 1
T = 1 - - - - sin( 2ka - 7T) + · · · (10.118)
2ka 2y-:;:,(ka) 312 4
which exhibits the small oscillations explicitly. These approximate expressions
T , ' ·~ th~ o~n~r~l hah~";"" ~c ~ J'unr-t;,-..n ,-..f /no , hut thPV ~rp nnt VPrV ~l'l'llr~tP
f,
b'
':''
HlU11>, ,, Wt:ll '" lllUl t; • "PP1
'
uu•v '-''-''-'" vu~o,
.. L ~.
for the circular opemng. These are compared witt eacn otner m rne oooK oy
Kin<' and Wu (FiP'. 41 n 17.til. The correct asvmntotic expression does not contain
•• 11'11. (HI110\ cl ••
J ,,
·• ,+ •J..~ t~rrn ;n fZ..n\
, .. 3/2 ;c ,.,;,-.p oc /
\
large.
We now wish to compare our results of the vector Smythe-Kirchhoff ap-

proximation with the usual scalar theory based on (10.79). For a wave not nor-
mally incident, the question immediately arises as to what to choose for the scalar
function if!(x). Perhaps the most consistent assumption is to take the magnitude
of the electric or magnetic field. Then the diffracted intensity is treated consis-
tently as proportional to the absolute square of (10.79). If a component of E or
B is chosen for ijJ, we must then decide whether to keep or throw away radial
components of the diffracted field in calculating the diffracted power. Choosing
the magnitude of E for 1/J, we have, by straightforward calculation with (10.108),
as the scalar equivalent of (10.113). The power radiated per unit solid angle in
the scalar Kirchhoff approximation is
dP
- =
(ka) 2
p.--cosa
COS a + COS e) L 2J,(kat;)
L
df! (10.119)
l 41T
7 '"" rv kof:.
where P; is given by (10 !15). If th~ alt~rnative scalar formula (~0.85) is used,
the obliquity factor (cos a + cos e)/2 in (10.119) is replaced by cos e.
If we compare the vector Smythe Kirchhoff result (1 0 114) with (10 119), W€l
see similarities and differences. Both formulas contain the same "diffraction"
distnbutJon factor [JI(kat;)/ka/;] 2 and the same dependence on wave number. But
the scalar result has no azimuthal dependence (apart from that contained in/;),
whereas the vector expressiOn does. I he azimuthal vanatwn comes from the
polarization properties of the field, and must b~ abs~nt in a scalar approximation.
For normal incidence (a = 0) and ka >> 1 the polarization dependence is un-
important. The diffraction is confined to very small angles in the forward direc-
tion. Then all scalar and vector approximations reduce to the common
expressiOn,
dP
df! (10.120)
1T ka Sill 0
The vector and scalar approximations are compared in Fig. 10.14 for the
angle of incidence equal to 45° and f01 an aperture one wavelengtl1 in diamelet
(ka - 1r). The angular distribution is shown in the plane of incidence (containing
the electric field vector of the incident wave) and a plane perpendicular to it. The
solid (dashed) curve gives the vector (scalar) approximation in each case. We see
that for ka = 1T there is a considerable disagreement between the two approxi-
mations. There is reason to believe that the Smythe Kirchhoff result is close
to the correct one, even though the approximation breaks down seriously for
ka < ~.The vector approximation and exact calculations for a rectangular open-
ing yield results in surprisingly good agreement. even down to ka ~ 1.*
*See J. A. Stratton and L. J. Chu, Phys. Rev., 56, 99 (1939), for a series of figures comparing the
vector Smythe-Kirchhoff approximation with exact calculations by P.M. Morse and P. J. Rubenstein,
Phy< Rev 54, 895 (1938) The alert reader may be puzzled by the apparent disGrepancy in the dates
of ~mythe 's publication (lt~c. eit.) and of Stratl<'m aHd Cht!. The h• o ealetiiations ~ ield the smuc r csttlt,
though quite different in appearance and detail of derivation, the earlier one involving a line integral
around lhe boundary of the aperture as well as a surface mtegral over it.
Sect. 10.10 Scattermg m the Short· Wavelength Limit 495
I \
;// 1\I \ I \
// ; Ir \ !
lj I I I
i' 7 \ /
//
/
/ \\ //
/ '\V
B-+----
(a) (b)
Figure 10.14 Fraunhofer diffraction pattern for a circular opening one wavelength in
dmmeter m a thm, plane, conductmg sheet. The plane wave is incident on the screen at
45° The solid cmves are the vector Smythe Kirchhoff approximation. while the dashed
curves are the scalar approximation. (a) The intensity distribution in the plane of
mcidence (E plane). (b) The intensity distribution (enlarged 2.5 times) petpendiculm to
the plane of incidence (H plane).
The diffraction by apertures or obstacles whose dimensions are small com-

pared to a wavelength requires methods different from the Kirchhoff or
Kirchhoff-like approximation. The exact formula (10.101) for a plane screen can
be used as a starting point. If the radiation fields of (10.101) are expanded in
mnltipoles, as in Sections 9 2 9 3, effective multipole moments (9 72) and (9 74)
can be identified in terms of integrals of the exact electric field in the aperture.
The derivation of these effective moments is left as Problem 10.10. Once the
dipole moments of an aperture are known, the diffraction can be calculated
merely by using the dipole fields of Sections 9.2 and 9.3. The example of a circular
aperture with effective moments (9.75) is left to the problems. The whole dis-
cussion of the physical picture parallels that of Section 9.5.C and is not repeated
here.
10.10 Scattering in the Short-Wavelength Limit

Scattering in the long-wavelength limit was discussed in Sections ~0.1 and 10.2.
The opposite limit, similar to the Kirchhoff domain of diffraction, is a scattering
by obstacles large compared to a wavelength. Just as for diffraction by a screen,
the zeroth approximation is given by classical ray theory. The wave aspects of
the fields give corrections to this, with the scattering confined to angular regions
only slightly away from the paths of geometrical optics. For a thin, flat obstacle,
the techniques of Section 10.7, perhaps with Babinet's principle, can be used.
But for other obstacles we base the calculation on the integral expression (10.93)
for the scattenng amplitude m terms of the scattered fields E, B., on a surface
S, just outside the scatterer.
In the absence of knowledge about the correct fields E, and B, on the surface,
496 Chapter 10 Scattermg and Ddlrachon SI
we must make some approximations. If the wavelength is short compared to the

dimensions of the obstacle, the surface can be divided approximately into an
illuminated region and a shadow region.* The boundary between these regions
is sharp only in the limit of geometrical optics. The transition region can be shown
to have a w1dth of the order of (2/kR)wR, where R 1s a typical radius of curvature
of the mrface. Since R is of the order of magnitude of the dimension5 of the
obstacle, the short-wavelength limit will approximately satisfy the geometrical
condition. In the shadow region the scattered fields on the surface must be very
nearly equal and opposite to the incident fields, regardless of the nature of the
scatterer, provided it is "opaque." In the ill11minated region on the other hand
sponding incident fields, while the tangential B., and normal E, will be approxi-
mately equal to the mcident values [see (10.95)].
Because of the generality of the eontribution ftom tl1e shadow region, it is
desirable to consider it separately. We write
E* • F = E* • F,h + E* • Fill (10.121)

If the incident wave is a plane wave with wave vector k 0 and polarization Eo,
- E 0 E0 e'k,,·x
Ei -
B, = k 0 X E;fkc
the shadow contribution, from (10.93) withE, = - E;, B, = - B,, is
(10.123)
where the integration is only over the part of S1 in shadow. A rearrangement of

the vector products allows (10.123) to be written
E* • [(k + k 0) X (n' X Eo) + (n' • E 0)k0]e'iko-k)·x· da'
In the short wavelength limit the magnif!Jdes of ke • x' and k • x' are large com-
pared to unity. The exponential factor in (10.124) will oscillate rapidly and cause
the mtegrand to have a very small average value except in the forward direction
where k = k 0 . In that forward region, () :'S 1/kR, the second term in the square
bracket is negligible compared to the first because (E* • k 0 )/k 1s of the order of
sin A<< I (remember e* · k- 0 and k 0 k). Thus (10.124) can be approximated
by
27r
E¥ • Eo J 'h
e' "o·• (k0 • n') da'
*Fm a very sitnila1 tJ eatment of the scattering of a scalar wave by a sphere, see lVlorse and Feshbach
(pp. 1551-1555).
Sect. 10.10 Scattering in the Short-Wavelength Limit 497
The integral over the shadowed side of the obstacle has, in this approximation,
the remarkable property of depending only on the projected area normal to the
incident direction and not at all on the detailed shape of the obstacle. This can
be seen limn the fact that (k0 - n') da' k dx' dy' k d 2 xj_ is just k times the
projected element of area and (ko- k) · x' = k(l -cos 8)z' - kj_ · X1 = -k · xj_.
Here we have chosen ko along the z axis, introduced two-dimensional vectors,
xj_ = x'e 1 + y'e 2 , kj_ = kxe 1 + kye 2 in the plane perpendicular to k 0 , and approx-
imated to small angles. The final fonn of the shadow contribution to the scattering
when kR >> 1 and 8 << 1 is therefore
e -ik j_
•X
1
d2 X1. (10.125)
In this limit all scatterers of the same projected area give the same shadow-
scattering contribution. The polarization character of the scattered radiation is
given by the factor E* ·Eo. Since the scattering is at small angles, the dominant
contribution has the same polarization as the incident wave. In quantum-
mechanical language we say that the shadow scattering involves no 5pin flip
For example, consider a scatterer whose projected area is a circular disc of
radius a. Then
2
11 (ka sin 8)
2 (10.126)
ffll ka sin 8
and the shadow-scattering amplitude is
k 2 ( * ) l1(ka sin 8) (10.127)
The scattering from the illuminated side of the obstacle cannot be calculated
without specifying the shape and nature of the surface. We assume, for purposes
of illustration, that the illuminated surface is perfectly conductmg. In ut!lizmg
(10.93) we must know the tangential components of E, and B, on .S\ As men-
tioned in the introductory paragraphs of this section, in the short-wavelength
limit these are approximately opposite and equal, respectively, to the correspond-
ing components of the incident fields. Thus the contribution from the illuminated
(10.128)
Comparison with the shadow contribution (10.123) at the same stage shows a
sign difference in the first term. This is crucial in giving very different angular
behaviors of the two amplitudes The co11nterpart of (1 0.124) is
e. [k k0 ) X (n' X Eo) (n' • Eo)ko]ei(ko k) x' da' (10.129)

41Ti Jill
Fm kR » 1, the exponential oscillates rapidly as before, but now, in the forward
direction, where we anticipate the major contribution to the integral, the other
factor in the integrand goes to zero. This can be traced to the presence of (k - k0 )
in the first term, rather than the (k ± k0 ) of the shadow amplitude (1 0 124) The
illuminated side of the scattet er thus gh es only a modest contribution to the
498 Chaplet 10 Scatleting and Diffraction SI
scattering at small angles. This makes perfect sense if we think of the limit of
geometrical optics. The illuminated side must give the reflected wave, and the
reflection is mainly at angles other than forward.
To proceed further we must specify the shape of the illuminated portion of
the scatterer, as well as its electromagnetic properties. We assume that the surface
is spherical of radius a. Since the contribution is not dominantly forward, we must
consider arbitrary scattering angles. The integrand in (10.129 consists of a rei-
atlve y s owly varying vector function of angles times a rapidly varying exponen-
tial. As discussed in Section 7.11, tlie dominant contribution to such an integral
comes from the region of integration where the phase of the exponential is sta
tionary. If (8, ¢)are the angular coordinates ofk and (a, {3) those ofn', relative
to k0 , the phase factor is
j( a, {3) (k0 k) · x' ka[(1 cos 8) cos a sin 8 sin a cos({3 ¢)]
The stationary point is easily shown to be at angles a 0 , {3 0 , where
7T 8
a0 = - +-
2 2
f3o = ¢
These angles are evidently j11st those appropriate for reflection from the sphere
according to geometrical optics. At this point the unit vector n' points in the
dtrechon of (k k 0 ). If we expand the phase factor around a a 0 , {3 {3 0 , we
f(a, {3) = -2ka sin~ [1- ~(x 2 + cos ~y 2 ) 2

+ ·· ·1 (10.130)
where x a a0 , y f3 {3 0 . Then integral (10.129) can be approximated by

evaluating the square bracket there at a = a 0 , {3 = {3 0 :
E*. Fill = k:;o sin 8e-2ika,in(B/2)E*. E, Jdx e'lkasin(B/2)]x

2
Jdy e'[kusin(B/2)cos 2 (e/2]y 2
(10.131)
where e, is a unit polmization vectm defined by
n, bemg a unit vector in the direction of (k ko). The vector E, is just the
polarization expected for reflection, having a component perpendicular to the
surface equal to the corresponding component of Eu and a component parallel to
the surface opposite in sign, as shown in Fig. 10.15. The x andy integrals in
e'= dx = ~ rovided 2ka sin 8/2

2
(10.131) can be approximated usin
>> 1
E * •Fi u - o
E - e 2ikasin(e/2)E*
•E
r (10.132)
2
For 2ka sin(8/2) large, the reflected contribution is constant in magnitude as a
function of angle, but it has a rapidly varying phase; as () ~ 0, If vamshes as II.
Sect. 10.10 Scattering in the Short-Wavelength Limit 499
Figure 10.15 Polarization of the reflected wave relative to the incident polarization: n,
is normal to the surfacg at thg point appropriate for reflection according to geometrical
optics. To avoid complexity in the figure, the wave vectors k0 and k are not shown, but
they are perpendicular to Eo and En respectively, and so oriented as to make their
difference parallel to n,.
Comparison of the shadow amplitude (10.12 7) wlfh the reflected amplitude

(10.132) shows that in the very forward direction the shadow contribution dom-
inates in magnitude over the reflected amplitude by a factor of ka 2> 1 , while at
angles where ka sin () >> L the ratio of the magnitudes is of the order of
ll(ka sin3 ())112 • Thus, the differential scattering cross section (10.3), summed over
the outgoing and averaged over the initial polarization states, is given in the two
regions by
2
lt(ka sin e) 10
2
a (kaf () :-s -
du ka sin () ' ka
~LV. D:""fJ
dfl az 1
- () >>-
4' ka
The scattering in the forward direction is a typical diffraction pattern with a
central maximum and smaller secondary maxima, while at larger angles it is iso-
tropic. At intermediate angles there is some interference between the two am-
plitudes (10.127) and (10.132), causing the cross section to deviate from the sum
of the two terms shown in (10.133). Actually, in the present approximation thts
interference is very small for ka >> 1. There is more interference in the exact
solution, as shown in Fig. 10.16, where the dips below unity are indicative of
destructive interference.*
The total scattering cross section is obtained by integrating over all angles.
Neglecting the interference terms, we find from (10.133) that the shadow dif-
fraction peak gives a contribution of TTa 2 , and so does the isotropic part. The total
scattering cross section is thus 2TTa 2 , one factor of the geometncal proJected area
coming from direct reflection and the other from the diffraction scattering that
must accompany the formation of a shadow behind the obstacle. The latter part
of the total cross section can be shown to be independent of the detailed shape
of the scatterer in the short-wavelength limit (Problem 10.16). Similarly, for a
general scatterer that is "opaque," the reflected or absorbed part of the total
cross section will also be equal to the projected area, although without specifying
*For a linearly polarized wave incident, the amount of interference depends on the orientation of the
incident polarization vector relative to the plane of observation containing k and k0 . For Eo in this
plane the interference is much greater than for e0 perpendicular to it. See King and Wu (Appendix)
or Bowman, Senior, and Uslenghi (pp. 402-405) for numerous graphs with different values of ka.
"~
I ' I ' I I I I I I
-
-
1- 4 du
\ ' d{l
I
10
~ I
I
-
r- -
I
r- I
I
1- I
-
- ~~ I
\
I
I
\
\
i ! If \
Ii II \ /"\.
1-
r-
1-
v Y_l_
v
_l
v
I \
\J
/'-.....
- ~
- -
1- -
1- -
1- -
N~nre 10.16 Semilogarithmic plot of the >COttP.rin<> ern" fnr 0 nPrrP~tl·

,cl ~t' "'_a_funct; = ll_ . ~1. '±b ~ . "r -' -o
''" , " ' l' ·~ .• u. ~
incident and ka = 10. The solid curve is the exact result (King and Wu). The dashed
curve is the approximation based on the sum of the amplitudes (10.127) and (10.132).
the properties of the illuminated surface, we cannot say how it is divided between
scattering and absorption.
10.11 Optical Theorem and Related Matters

A fundamental relation, called the optical theorem, connects the total cross sec-
,;v .. vr a 'l:aLLt:rcr LO we imaginary part or tne rorwara scattenng amnii tude. The
•follow~ -' •~ >t~ · ,~ ·
A_ ~ •
_ .. __JC " ' " ' uvn, auu ' " " HO •pan m we quanLUm-mecuamcaJ scattenng OI
particles through the conservation of probability.
To establish the theorem, we consider the scattering geometry shown in Fig.
10.9. A plane wave with wave vector k0 and fields (E,, B,) is incident in vacuum
Sect. 10.11 Optical Theorem and Related Matters 501
r
vu a uuit<.-
• •'-
W~·
"
H~O
'_.I •'- ..c. p
']•
"]"~ f;,]rlo(IO'
\" ~-,
,,
U\
propagate out from the scatterer and are observed far away in the direction of
k. The total fields at all points in space are, by definition,
E-E+E, B- B, + H,
The scatterer is, in general, dissipative and absorbs energy from the incident
wave. The absorbed power can be calculated by integrating the inward-going
e"' Tl. · .C •'- .1 <:~-1..1. tJ, " C ·
-r v• "'" • ~J ... u, 6 ~ .,.
Pab' = - -
1
2J.Lo
f
s1
Re(E X B*) · n' da' (10.134)
The scattered power IS normally calculated by cons1aenng me asymptotic

form of the Poynting vector for the scattered fields in the region where these are
simple transverse fields falling off as 1/r. But since there are no sources between
_, '
.)1 ana mnnny, tne scauereu power Call t:l[Uauy U<.- <.-V<U - ao au -o'"'
overS, of the outwardly directed component of the scatterea J-'oyntmg vector:
P,catt = ,. ,
1
-,-u
f. 1
Re(E, X B~) · n' da' (10.135)
The total power P taken from the incident wave, either by scattering or
absorption, is the sum of (10.134) and (10.135). With some obvious substitutions
and rearralli!ements the total power can be wntten
P = -2-
1 f Re[E s X B * + E * x B,] · n' da' l l
J.Lo s,
With the incident wave written explicitly as

E, = EoEoe'ko·x (10.136)
1
en, kKo~;
the total power takes the form,

1
P = - - Re{ Ei;
2J.Lo
f s1
e-'ku·x'[ E~ • (n' X B,) + Ei; • ko X (~~ X E,) J da'}
Comoarison with 110.93) for the scattermJ), amplitude snows tnat me total power
is related to the forward (k = k 0 , E = Eo) scattering amplitude according to
2
P = - 7T Im1r Ec,* Eo* • F<ok = ko)l (10.137)
KLo
This is the basic result of the optical theorem, although it is customary to express
it in a form that is independent of the magnitude of the incident flux. The total
rrnoe em•finn ~ f • r~]]prJ thP.
<A> rmss section in ootics) is defined
" .; .. 1 n "- · · , I r. 12 n 7
<lO LHC l<lllU Vl lllC <V<UO jJV 1~> l " · !-' r ' I CuI •u•
Similarly, the normalized scattering amplitude f is defined relative to the ampli-

tude of the incident wave at the origin as
F(k, k 0 )
f(k, k 0 ) = (1 0.138)
Eo
0
0
Figure 10.17 A plane wave incident normally on a slab of dielectric of thickness d.
radiation (photons) this is indicated by the presence of the amplitude e6 · f for

scattered radiation with the same polarization finally as it was initially.
The optical theorem relates different aspects of the scattering and absorption
of electromagnetic waves for a single scatterer. It is also possible to connect the
forward scattering amplitude for a single scatterer to the macroscopic electro-
magnetic properties, namely the dielectric constant, of a medium composed of a
large number of scatterers. We will content ourselves with a brief elementary
discussion and refer the reader to the literature for more detailed and rigorous
treatments.* Consider a plane wave (10.136) incident normally from the left on
a thin slab of uniform material composed of N identical scattering centers per
unit volume, as shown in Fig. 10.17). The incident wave impinges on the scattering
centers, causing each to generate a scattered wave. The coherent sum of the
incident wave and of all the scattered waves gives a modified wave to the right
of the slab. Comparison of this modified wave at the observation point 0 with
that expected for a wave transmitted through a slab described by a macroscopic,
electric susceptibility E( w) then leads to a relation between E and the scattering
amplitude f.
Sect. 10.11 Optical Theorem and Related Matters 503
The thickness and the density of the slab are assumed to be so small that
only single scatterings in the slab need be considered and, as a consequence, the
effective exciting field at each scatterer is jm;t the incident field itself The scat-
tered field produced at the observation point 0 with cylindrical coordinates
(0, 0, z 0 ) by the N d 3x scatterers m the mfimtestmal volume element (z3x at the
point x(p, q,, z) in the slab is, in this approximation,
dE.,
where we have written the scattering amplitude in terms of the scattering angles
8 and ¢, with sin 8 = p/R, and have assumed that the observation point is many
wavelengths from the slab. The distance from the volume element to 0 is
rp
R = 2 + (zo - z?f 12 • The presence of the phase factor of the incident wave is
necessary to account for the location of the scatterers at x, rather than at the
migin of comdinates. The total scattered field is obtained by iuteg• at ion ovel the
volume of the slab:
Es (10.140)
Since pdp = R dR, this expression can be written
Es = NEo dR e'kR f(k, 8, </J) (10.141)
'<Vhere eos 8 (z 0 z)IR. We now treat e1kR tlR as a differential and integrate
by parts to obtain for the R integration,
1
f=
lz o-zl
dR eikR f(k, 8, </J) = .k
l
eikR f(k, 8, </J) =
R=lz 0 -zl
1
+ -:--k J= dR
(Zo R- Z) e
2
ikR
d(
d
) f(k, 8, </J)
l lzo-zl . COS 8
Provided the indicated derivative off is well behaved, the remaining integral is
of the mde1 of lf(k [z 0 z [) ti111es the original. Since we ha1e assumed that the
observation point is many wavelengths from the slab, this integral can be ne-
glected. Neglecting the oscillating contribution at the upper limit R ----> cc (this
can be made to vanish somewhat more plausibly by assuming that the number
N of scattering centers per unit volume falls to zero at very large p), we have the
resu t
r: -I dR eikR f(k, 8, </J) = f 1

e klzo-zl f(k, 0)
The scattered field at 0 is therefore
k lv'E0 f(k, 0) J., dz eik[z+ 'e '11
Since Ze > z by assumption, we have finally
Es - k NE0 f(k ' O)e'x'o d (10.142)

504 Chapter 10 Scattering and Diffraction SI
The total electric field at the observation point 0 is
(10.143)
correct to first order m the slab thickness d. I he amplitude at 0 for a wave with
the same polarization state as the incident wave is
(10.144)
Suppose that we now consider the slab macroscopically, with its eleetromag-
netic properties specified by a dielectric constant E( w )/Eo appropriate to describe
the propagation of the wave of frequency w = ck and polarization E 0 . A simple
calculation using the fotmulas of Chapter 7 shows that the transmitted wave at
z = Zo is given by
Eo · E macroscopic) = E 0 e' zo 1 + ik(Eho - 1) 2 (10.145)
correct to first order in d, but with no approximation concerning the smallnev~ of

IdEo ~ 1. Comparison of (10.144) and (10.145) shmrvs that the dielectric constant
can be written in terms of the forward scattering amplitude as
e
E(w)!Eo - I + 4nN Eo"' . f(k, 0) (10.146)
A number of observations are in order. It is obvious that our derivation has been
merely indicative, with a number of simplifying assumptions and the notion of a
macroscopic description assumed rather than derived. More careful consider-
ations show that the scattering amplitude in (1 0 146) should be evaluated at the
wave number k' in the medium, not at the free-space wave number k, and that
there is a multiplier to the second term that gives a measure of the effective
exciting field at a scatterer relative to the total coherent field in the medium. The
reader can consult the literature cited above for these and other details. Suffice
it to say that (10.146) is a reasonable approximation for not too dense substances
and p10vided correlations among neighboring scatterers are not 1mportant. It 1s
worthwhile to illustrate (10.146) with the simple electronic oscillator model used
in Chapter 7 to describe the dielectric constant. The dipole moment of the atom
IS given by (7.50), summed over the various oscillators·
p _ -r I f ; cW;2 - W2 - . )-1£gEg
lWYJ
From (10.2) we infer that the atomic scattering amplitude is
f(k) - - - - 2: f/wf - w2 - iwyJ 1

(k X Eo) X k
4 7TE0 m 1
The scalar product of E;'; with the forward scattering amplitude is then
Sect. 10.11 Ontical Theorem and Related Matters 505
Substitution into (10.146) vields the dielectric constant

NeL .., 0 0 -1
-r w ·~-
E~W)!Eo 1 L Jj~Wj tW'YJJ
A
\iV.i"T/}
co, 1
in agreement with (7.51).

Contact can be established between (10.146) and the optical theorem
(1 0.139) by recalling that the attenuation coefficient a is related to the total cross
section of a single scatterer through a = N(J", and to the imaginary part of the
wave number in the medium through a = 2 Im(k'). From (10.146) and there-
lations (7 .54) for the real and imaginary parts of k' in terms of e( w) we find
4TTN . * ( (10.148)
a = Nu, = Re(k') Im[Eo • f Re k', 0)]
where I have immoved (10.146) bv evaluating fat the wave number in the me-
Pnn~tion ~ iniiTrMP< that. if we
rlinm ~< rl. .cMih~rl
vy a
L
•
""
.• .
...-rrn-
-" ..
'. uv -r
. .
• 1 ,,_
•v
rl
r scat-
• ,_ ...
retanons wtH appear as oerore, provwea we aescnoe we CUL
bv using the local wave number k' in the medium. The same sttuation holds m
the scattering of electrons in a solid, for example, where the effective mass or
other approximation is used to take into account propagation through the lattice.
As a final comment on the optical theorem we note the problem of approx-
imations for f. The optical theorem is an exact relation. If an approximate ex-
pression for f is employed, a manifestly wrong result for the total cross section
may be obtained. For example, in the long-wavelength limit we find from (10.2)
and (10.5) that the scattering amplitude for a dielectric sphere of radius a is
f = ( E' - 1) a3 (k X Eo) X k
E1 + 2
ThP 1
-• "mnlitnrl<> i<
E 1
Eo · t(R Ro) k'a· ~lU.l'+~J
'1
r . I
D. 1 .].
" ' '1
'
.... -r
,1" rl' . ,n1o <1-n A-·;~~~
'
tl- {1 () 1 '1.0\ thPn
·:
ytews (]"' u. vn rne ower nana, we ,, ,v rnar rne lUla1 u u>> , I> 111 UH>
case equal to the scattering cross section (10.11 ):

2
87T . Er - 1
(J"sc
= - k4a6 (10.150)
3 E1 + 2
Even with a lossy dielectric (Im e * 0), the optical theorem yields a total cross
section,
~., -'- 3 T.
r
fT. (10.151)
IE, + L I"
wnile tne scattenng cross secnon remains ( LU. DU ). 1 uese seeming comrauic,iuu>
are reflections of the necessitv of different orders of approximation required to
,,_,, . L ,,_ < oA o:rlno n+ <1-~ A~ ' ,] ·~ Tn thP lon<T-
-: ' .' r _,. rl 0
wav 11 lllllll Jl , , 1 J LV <0 LUt;; lVl '"'~ '"'5 uwpu•u~v •v

"illh . .. .
rhontPT J () • <mrl
-"'
IJll;llt:l Ul W LU lUlU lllt:
111 r 11111; <Cl U"
" <CUlllllUU(]On Jn
the total cross section by means of the optical theorem. For lossless or nearly
lossless scatterers it is therefore simplest to determine the total cross section
directly by mtegrat10n ot tne dJtterent1a1 scattenng cross section over angles. For
dissipative scatterers, on the other hand, the optical theorem yields a nonzero
answer mar nas a airrerem ~ usuauy a wwer power) aepenaence on w ana other
parameters from that of the scattering cross section. This contribution is, of
course, the absorption cross section to lowest explicit order in w. It can be cal-
culated from first principles with (10.134), but the optical theorem provides an
elegant and convenient method. Examples of these considerations are given in
the problems. An analogous situation occurs in quantum-mechanical scatterinP"
by a real potential where the first Born approximation yields a real scattering
amplitude. The second Born approximation has an imaginary part in the forward
direction that gives, via the optical theorem, a total cross section in agreement
with the integrated scattering cross section of the first Born approximation.
H, v '""',. and ,,; ~~

d ~
Scatterin£ and diffraction are ~ in manv antics texts. References con

critical opalescence are cited in Section 10.2. The subject of losses in optical fibers and
Rayleigh scattering there can be pursued in
J. M. Senior Optical Fiber Communications Prentice Hall New York (1992).
Chapter 3.
H. Murata, Handbook of Optical Fibers and Cables, Marcel Dekker, New York
(1\11;1;),
The scattering of radiation by a perfectly conducting sphere is treated briefly in

Mm•P "nrl FP•hhorh Inn IRR? .. IRRI'>\
Panofsky and Phillips, Section 13.9
Much more elaborate discussions, with arbitrary dielectric and conductive properties for
thP mhPrP IMiP'' ,hl. 1\ orP oivPn hu
Born and Wolf, Sectio~ 13.5
Stratton, Section 9.25
Specialized monographs on the scattering of electromagnetic waves by spheres and
other obstacles are
n ~ ',_,
·~
' " , uu~ ~
H. C. van de Hulst, Light Scattering by Small Particles, Wiley, New York (1957)
Kerker
v;
'0 rl " ' · .
Kerker has a nice historical introduction, including quotations from Leonardo da Vinc1
and Maxwell in a letter to Ravleif!h. See also
Von RloclPI
The subject of diffraction has a very extensive literature. A comprehensive treatment

VL VVLU O~dhU L'-W <lllU VC\.OlUl ~, w llll 11wuy <lllU ugutcoo, •o
given by
Rnrn ond \X! A if r~ ,,,
-r
\/TTl TY rl YT
~·-
'"~ '~"~"
C. J. Bouwkamp, Diffraction theory, in Reports on Progress in Physics, ed. A. C.
1rl Vnl XVTT nn '1~-1()() ThP Phu<ir"l "'- T ~- 11Q~4\
Ch. 10 Problems 507
has an extensive list of references and treats a number of difficult points. More elementary
discussions of the scalar theory are found in
Slater and Frank, Chapters XIII and XIV
Stone, Chapters 8, 9, and 10

Sommerfeld also discusses his rigorous calculation of diffraction by a straightedge, one of
the few exact solutions in diffraction theory.
The hmgnage of diffraction is used increasingly in optical applications such as Abbe's
theory of the resolution of optical instruments and the phase-contrast microscope. An
introduction to this area from an electrical engineering viewpoint is provided by
J. W. Goodman,Introduction to Fourier Optics, McGraw-Hill, New York (1968).
Born and Wolf and also Stone discuss some aspects of these applications.
Mathematical techniques for diffraction problems are discussed by
Baker and Copson
Jones, Chapters 8 and 9
A. Rubmow1cz, Dze Beugungswelle in der Kirchhoftichen Theorie det Beugwtg,
2nd edition, PWN-Springer, Warsaw-Berlin (1966)
L. A. Weinstein (Vaynshteyn), The!H)' of Dif,hactio'l ar1d tile Factorization
Method, Golem Press, Boulder, CO (1969)
The proof that for short-wavelength scatteiing 01 diffraction the li ansition between
the illuminated and shadow regions is of the order of Rl(kR) 111 can be found in
V. A. Fock, Electromag11etiG Difjra('(ion and Propagation Problems Pergamon
Press, Oxford (1965)
Entirely omitted from this chapter is the use of variational principles for diffraction
and scattering. The reader may fill this gap by consulting
Cairo and Kahan

Harrington, Chapter 7
H. Levine and J. Schwinger, Commun. Pure Appl. Math. 3, 355 (1950)
Morse and Feshbach, Section 11.4
Problems
111.1 (a) Show that for arbitra1y initial polarization, the scattering cross section of a
perfectly conducting sphere of radius a, summed over outgoing polarizations,
is given in the long wavelength limit by
d!l 4 4
where n0 and n are the d1recuons of the incident and scattered radiations,
respectively, while e 0 is the (perhaps complex) unit polarization vector of the
incident tadiation (eli· e 0 1, n0 • e 0 0).
(b) If the incident radiation is linearly polarized, show that the cross section is
dO (e0 , no, n) - k a 8 (1 + cos IJ) cos (} sm"ti cos L</J

8
where n · n 0 cos eand the azimuthal angle if> is measured from the direction
of the linear polarization.
.. u auu v
\q vv naL " Lilt: rauo 01 reiTTIT ... ~, 'I' 7TIL,
'" v
</> 7T/2? txplam physically m terms or tne maucea mump01es and their
radiation patterns.
111.2 Electroma<'netic radiation with ellintic notarization described (in the notation nf
~ ,.,·\ L _,
" . ~,
"
•L
I" '
1
:,.'2 ( E+ + reiaE )
€ = ""
is scattered by a perfectly conducting sphere of radius a. Generalize the amplitude

in the scattering cross section (10.71), which applies for r = 0 or r = oo, and
ol. ,] ••· <l. • <. · · '" ],.n, _ '" ,pl. .<l. limit <;:J,, <l.
-o L'
da 5
~n = k 4 a 6 ;:; (1 + cos 2 11) - cos
3
0 - :;: (-1 +r r') sm. 11 cos(2</J -
2
2
a) -
Compare with Problem 10.1.

10.3 A solid uniform sphere of radius R and conductivity a acts as a scatterer of a
plane-wave beam ot unpolanzed radmtwn or rrequency w, wirn IJJJ{!C <--<.. J. 1 ne
conductivity is large enough that the skin depth /5 is small compared to R.
(a) Justify and use a magnetostatic scalar potential to determine the magnetic
<: .• 1~ ..1 <l.
-,• • · <l. • • i< · · f~pm,
-o
that w * 0.)
(b) Use the technique of Section 8.1 to determine the absorption cross section
ot the sphere. ~now that 1t vanes as ~W) proviaea a is inaepenaem or
frequency.
10.4 An unpolarized wave of frequency w = ck is scattered by a slightly lossy uniform
· · ~·~1 • ·-• - ' • "' n -• ,11, ·th >n th Thacnh.
T -, -o_' r
..
1> CHdldClt:llLt:U uy dll lY lCdl "~ ~VU•'<«<H c, auu a>~«< my
a. The parameters are such that the skin depth 8 is very large compared to the
radius R.
(a) Calculate the differential and total scattering cross sections.
(b) Show that the absorption cross section is
(07 ~·
aabs 127TR2
(E, + 2? + (Z0 alk?
{~\ H'rnm· nor1
r
o ..,.i,P ..1nnon 1ha o-· 1tarino omnlitnrlP onci n<P thP nnfiral
·'-
LO evawaLe we total cro" • I~ JUUl <UDWCl UU! IUv
sum of the scattering and absorption cross sections from parts a and b.
Comment.
10" 'T'hP <ro11Prino hu lhP Af 1n 4 wa' tr,Mer1 as nurelv elcc-
tric dipole scattering. This is adequate unless it happens that the real dielectric
constant E! Eo is very large. In these circumstances a magnetic dipole contribution,
"ven t hnuuh hiaher order in kR mav be imnortant.
(a) Show that the changing magnetic flux of the incident wave mduces an azl·
muthal current flow in the sphere and produces a magnetic dipole moment,
N'lTOZo ' . o\2 "·_ n

'I me
kJLo 30
Ch. 10 Problems 509
(b) Show that application of the optical theorem to the coherent sum of the
electric and magnetic dipole contributions leads to a total cross section,
2
~21rR (RZ11 e-) t ( E,
1 1
+ 2 )2 + ( Z 0 ulk ) 2 + 90 (kR)
2j
(Compare Landau and Lifshitz, Electrodynarrucs of Contmuous Medza,
. 323 .
10.6 (a) Show that for the scattered wave (10.57) the normalized scattering amplitude
IS
where the polarization vector of the incident wave is (e 1 :+: ie 2 )/v'2.

(b) Deduce an expression for the total cross section of u, from the optical the-
orem (10.139) and the above expression for f.
10.7 Discuss the scattering of a plane wave of electromagnetic radiation by a nonper-
meable, dielectric sphere of radius a and dielectric constant E,.
(a) By finding the fields inside the sphere and matching to the incident plus
scaltel ed wave outside the sphere, determine <'ithout any restriction on kti
the multipole coefficients in the scattered wave. Define suitable phase shifts
for t e pro em.
(b) Consid@r th@ long wav@l@ngth limit (ka << 1) and determine explicitly the
differential and total scattering cross sections. Compare your results with
those of Section tO.l.B.
(c) In the limit E, --> co compare yonr results to those for the perfectly conducting
sphere.
10.8 Consider the scattenng of a plane wave by a nonpermeahle sphere of radms a and
very good, but not perfect, conductivity following the spherical multipole field
approach of Section 10.4. Assume that ka << l and that the skin depth 8 < a.
(a) Show from the analysis of Section 8.1 that
Zs ko .
- = - (1 ~ l)
Zu 2
(b) In the long-wavelength limit, show that for l 1 the coefficients a:±(l) and
/3+(1) in ( 10.65) are
-
I
0\ 0 -
1 ~
i-
\ a; a
a+(l) 2i (ka)3
" ( 1 + ,.,() +'t,.,,-
()
- 'I -
(c) Write out explicitly the differential scattering cross section, correct to first
ord@r in !iia and lowest order in ka
(d) Using (10.61), evaluate the absorption cross section. Show that to first order
in 8 it is uah' ~ 37r(k8)a 2 • How different is the value if 8 ~ a?
10.9 In the scattering of light by a gas very near the critical point the scattered light is
observed to be whiter' (i.e., its spectrum is less predominantly peaked towmd
the blue) than far from the critical point. Show that this can be understood by the
fact that the volumes of the density fluctuations become large enough that
Rayleigh's law fails to hold. In particular, consider the lowest order approximation
to the scattering by a uniform dielectric sphere of radius a whose dielectric constant
E, differs only slightly from unity.
Ia) Show thM for ka >> 1 th~e rliff~er~enti"l cross section is sharnlv ne"k"rl ;n '"
c~; rl th , ,1 -~~ -rr
,J.
~"
1T
a= 2 (ka?l <, - ll a 2 2
with a k 2, rather thank", dependence on frequency.

(b) Show that for arbitrary ka the total cross section to lowest order in ( E, - 1)
is tne expression given in pan a, muuiptieu oy we wnction
F(z) = 1 + sz- 2 - ~z- 4 (1 - cos 2z) - z- 3 sin 2z
4(z 2 z 4) rz 1 -
JO
cost dt
t
where z = 2ka. [This result is due to Lord Rayleigh, 1914.]

.
.... .... 'T'J,
h-,·1 · 1
.ff .· ~ "J f d "· -~r~· (h d"ff d ·~u
,J. 1-.
as t e ocat10n o e ect1ve sources that pro uce ra 1at10n t e 1 racte fields).

An aperture whose dimensions are small compared with a wavelength acts as a
,f cl" ,, "' • "th th ' . ,f th ,,,. ,, J, .
.. "b
(a) Beginning with (10.101) show that the effective electric and magnetic dipole
moments can be expressed in terms of integrals of the tangential electric field
m the aperture as follows:
p = Ell ( (x • E,,n) da 19.7?)

~ r
m = -.-
lW~
J (n X E.,'") da
wnere E,an 1s tne exact tangential eJectnc neld m tne aperture, n IS tne normal
to the plane screen, directed into the region of interest, and the integration
is over the area of the openings.
lbl Show that the for the can be ~into
m = ~ Jx(n · B) da (9.74)
n ..L
~~ ' . uuvu•, ~uuu • uvu• u•~ ~u5~ v• '"~ up~au•~
some components of the fields are singular if the screen is infinitesimally

thick
.LU•.Ll A perrecuy conouctmg nat screen occupieS nan or tne x~y plane (1.e., x <. u )· 1\.
plane wave of intensity / 0 and wave number k is incident along the z axis from the
· n n·. .. •L .< "'""· ""-1. . · '" .t. ,11 ,t th
'b' ' . y
. .
x~y ptaue ueuneu uy z ""' ~ u. Let we cooruma,es Ot tue ouserva,Jou potu, u~
(X, 0, Z).
(a) Show that, for the usual scalar Kirchhoff approximation and in the limit
Z >> X and V kZ >> 1 the diffracted field is
!/I(X 0 Z t)
' ' '
= J'l2eikz-;,,{ 1
()
+ i)
'1,.
13- r~ 2
e1r dt
\
'
where g = (k/2Z) 112
X.
Ch.10 Problems 511
V'l c>uu• lll<tl lllt; lllliCll>LlJ <.;all ue NllllCll
1- 1.w = ~ [<cw + D + <sw + m 2
where C(l;) and S(l;) are the so-called Fresnel integrals. Determine the as-
ymptotic behavior of I for l; large and positive (illuminated region) and g
large and negative (shadow region). What is the value of I at X = 0? Make
a sketch of I as a function of X for fixed Z.
(c) Use the vector formula (10.101) to obtain a result equivalent to that of part a.
Compare the two expresswns.
"'..... "
_, ·' -'. .c ~. "' rl ·'- /, . . rl • • j
. "UI 1 autu'~ 1''-"'-'-" y '-""uu'-~"'5 uuc

'J r "U
uu a " o,_, '-'-"·
The incident wave vector makes an angle a with the normal to the screen. The
polarization vector is perpendicular to the plane of incidence.
(a) Calculate the diffracted fields and the power per unit solid angle transmitted
through the opening, using the vector Smythe-Kirchhoff formula (10.1 01)
with the assumption that the tangential electric field in the opening is the
unperturbed incident field.
... ·' . TT! -'-'- .££
V'l YUUl lOC>Ull Ul ~ ~
matJOn anu Wlut tuc tesuuut U.7

"
u
"''- r
0'-'
''-'-'"'
't'~
"'
...
"''-
plane ot mc1dence.
10.13 Discuss the diffraction of a plane wave by a circular hole of radius a, following
Section 10.9, but using a vector Kirchhoff approximation based on (10.90) instead
of the Smythe formula (10.101).
(a) Show that the diffracted electric field in this approximation differs from
(10.112) in two ways, first, that cos a is replaced by (cos I)+ cos a)/2, and
second, by the addition of a term proportional to (k x E 3 ). Compare with
the obliquity factors C of the scalar theory .
'" ·' '" ' ..... ....,

.rl . ·~· th;< ·~ff -
/h\ "'· ·'·
.c /1{\ 1 1 " c. ... .. oh . p; 1 n 1A th
lWU '5' '

.' '/
H>.
·o·
10.14 A rectangular opening with sides of length a and b ::>- a defined by x = ±(a/2),
y = ±(b/2) exists in a flat, perfectly conducting plane sheet filling the x-y plane.
A plane wave is normally incident with its polarization vector making an angle f3
with the long edges of the opening.
(a) Calculate the diffracted fields and power per unit solid angle with the vector
Smythe-Kirchhoff relation (10.109), assuming that the tangential electric
.
moou "' tHe . ,.
• • ·' . . uupe• w ' "-U
·o I> ' " ' -
{h) Calculate the corresnondin<> result of the scalar Kirchhoff approximation.

\CJ ror u a,p '+J , IW '+7T, ICUllljJUlto lllto •'-CVI UUU oo'-UWI
'
to the dJttracted power per umt solid angle as a runcuon or me ang1e t1 ror
<P - 0. Plot a graph showing a comparison between the two results.
10.15 A cylindrical coaxial transmission line of inner radius a and outer radius b has its
axis along the negative z axis. Both inner and outer conductors end at z = 0, and
the outer one is connected to an infinite plane copper flange occupying the whole
x-y plane (except for the annulus of inner radius a and outer radius b around the
nri<Jin 1 ThP t · · linP i< · bt " 1rv win its dominant TEM mode
·,!, •• nl ~It
,_ th """~-' '"" ha;~rr V T To • thP
r ,. ~~.
'"'"~"~'u"
." r >UUH CU . ·'cue ' e
''
•'- .1.
·'
ramauon, ana tne total power T'lflJiaT=.
10 16 (a) Show from (10 125) that the integral of the shadow scattering differential
and therefore is equal to the projected area of the scatterer, independent of

its detailed shape.
(b) Apply the optical theorem to the "shadow" amplitude (1 0. 125) to obtain the
total cross section under the assumption that in the forward direction the
contribution from the illuminated side of the scatterer is negligible in
companson.
10.17 (a) Using the approximate amplitudes of Section 10.10, show that, for a linearly
polarized plane wave of wave number k incident on a perfectly conducting
sphere of radius a in the limit of large ka, the differential scattering cross
section in the E plane (£ 0 , k0 , and k coplanar) is
2
da a
dD (E plane) = 4 [
4 cot>() Jf(ka sin IJ) + 1
- 4 cot IJ .!1 ( ka sin IJ) sin ( 2 ka sin ~) J

and in the H plane (Eo perpendicular to k 0 and k) 1s
2
da (H plane) = a 4 cosec 2 () .Ti(ka sin 0) + 1
+ 4 cosec IJ .T1 ( ka sin 0) sin ( 2 ka sin ~) J

(The dashed curve in Fig. 10.16 is the average of these two expressions.)
(b) Look up the exact calculations in King and Wu (Appendix) or Bowman,
Senior and Uslenghi (pp. 402-405). Are the qualitative aspects of the inter-
ference between the diffractive and reflective amplitudes exhibited in part a
in agreement with the exact results? What about quantJtatJve agreement?
10.18 Discuss the diffraction due to a small, circular hole of radius a in a fiat, perfectly
conductmg sheet, assummg that ka << I.
(a) If the fields near the screen on the incident side are normal E 0 e-iwt and
tangential B 0 e-'w', show that the diffracted electric field in the Fraunhofer
zone IS
wlHml lc is the wave vector in the direction of observation

(b) Determine the angular distribution of the diffracted rad1at10n and show that
the total power transmitted through the hole is
10.19 Specialize the disetission of Problem 10.18 to tile diffraction of a plane "ave by
the small, circular hole. Treat the general case of oblique incidence at an angle a
to the normal with polarization in and perpendicular to the plane of incidence.
(a) Calculate the angular distributions of the diffracted radiation and compare
them to the vector Smythe-Kirchhoff approximation results of Section 10.9
and Problem 10.12 in the limit ka << 1.
,..,~ 11\ 1::1't
.. ., "'u.'
" '
"
00
"
V'! C '" WL • Vii <t;.-nT.r ~ '= "/ "l'
intensity in the plane of incidence and compare the relative values with the
solid curve in Fig. 10.14. (Use a protractor and a ruler to read off the values
from Fig. 10.14 at several angles.)
(c) Show that the transmission coefficient [defined above (10.116)] for the two
states of polarization are
T, 64
7:T'7r
(ka)•l 4 + sin 00
"'~U> <X
2
l
I
64
I~ ;;;:;-? ( ka r cos a
Note that these transmission coefficients are a factor (ka) 2 smaller than those
given by the vector Smythe-Kirchhoff approximation in the same limit.
10.20 A suspension of transparent fibers in a clear liquid is modeled as a collection of
scatterers, each being a right circular cylinder of radius a and length L of uniform
dielectric material whose electric susceptibility differs from the surrounding me-
dium by a small fractional amount 15<1<.
(a) :Show that to lirst order m i5EIE the scattermg cross secrion per scauerer ror
unpolarized radiation of wave number k is
2 2
au = OE
K a·L· (1 + cos 2 1J) 1"-Jl\lhUJ. >CU\'JII'-"L.J
dfi E 32 q~a qiiL/2
where 11 (z) is the Bessel function of order unity and q (q~) is the component 11
of the wave number transfer parallel (perpendicular) to the cylinder axis.

(b) In the limit of very slender cylinders (ka << 1), show that the scattering cross
section, averaged over the orientation of the cylinder (appropriate for an
ensemble of randomly oriented fibers), is
I , I 2 r4 4T2
0 I. · .r,, ,..,, 2 ~
{ uu \
11 + cos 1J
2 Si( aL) I I
\d!lj E 32 ijL \ quL
where Si(x) = J: [(sin x)!x] dx is the sine integral (Abramowitz and Stegun,
p. 231) and q 2 = 2k 2 (1 - cos IJ).
(c) Plot the square-bracketed quantity in part b as a function of q 2 L2 on the
range (0, 100). Verify that the cross section is the expected one when
kL << 1 and show that when kL >> 1 (but ka << 1) the total scattering cross
section is annroximatelv
oO
1177" liE > 3 4 T

1 m~K~
Uscatt f.{)
fteUL
' kL' kL
'
Comment on the frequency dependence.
CHAPTER 11
Special Theory of Relativity
Beginning with Chapter 11 we employ Gaussian units instead of SI units for electro-
magnetic quantities. Explicit factors of c appear in a natural manner in these units,
making them more appropriate than SI units for relativistic phenomena. The Issue of
" . ,r ,,: , · ,;- · · c. ,;-A . •J. 1\Jf. ,JJ •t" .\.
another matter. Som~ 'workers, esp;cially quantum field theorists, prefe;Heaviside-

Lorentz units-see the Appendix.
The special theory of relativity has, since its publication by Einstein in 1905,
<1 lOU! Ill }lllYM'-'"• <1" lUI !;ldiiLCU d, l>CWLUll, l<tW'
of classical mechanics the Maxwell equations of electromagnetism or the

Schrodinger equation of quantum mechanics. Daily it is employed by scientists
m tne1r consiOeratJon or precise atomic pnenomena, m nuclear pnys1cs, ana a nove
all in high-energy physics.
. . . . . . . .
1 ne ongms 01 llle special llleory 01 retanvuy ne m eteCLromagnensm. Ill taCL,
nnP r<ln O<l\1 th<lt thP cfpvp], 1t nf thP M! •.11 · with thP. unifir<ltion
of electricity and magnetis~ and optics forced special relativity on us. Lorentz
above all laid the groundwork with his studies of electrodynamics from 1890
onwards. Poincare made important contributions, but it fell to Einstein to make
the cruCial generalization to all physical phenomena, not JUSt electrodynamics,
<>nrl tn drP« thP f<>r-rP<>l'hino ~ 1l<ltP ThP ~nPri<ll
-, HPni'P< n f thP <· ' . '
theory of relativity is now believed to apply to all forms of interaction except
large-scale gravitational phenomena. It serves as a touchstone in modern physics
for the possible forms of interaction between fundamental particles. Only theo-
ries consistent with special relativity need to be considered. This often severely
1: •
...
•1.
~
. ., • •
.
The e~perimental basis and the historical development of the special theory
of relativity, as well as many of its elementary consequences, are discussed in
many places. A list of books and articles is given at the end of the chapter. We
content ourselves with a summary of the key points and some examples of recent
~ "' ..
''-' -' ""'- . ..
£"..
"'"""mo. • "'-'" "''-' uao.v
_,, < ,_
<.Uv ouu,-
•- . ''-'OUHO
marized including coordinate transformations orooer time and time dilatation

the relativistic Doppler shift, and the addition of velocities. The relativistic energy
and momentum of a particle are derived from general principles, independent of
the force equation. Then the idea of the Lorentz group and its mathematical
. . . . . .
uescnpuon IS presen1eu anu a specmc represemanon m 1erms 01 '+ -" '+ m
514
Sect. 11.1 The Situation Before 1900, Einstein's Two Postulates 515
~u~ ~hf'~' vuo~o ~v~ <H~ w v~ ~~~~, . .~ ~uu~s~, <Hv ~v •uuuu~~ v~
electrodynamics, and the explicit transformation properties of electric and mag·

netic fields follow. The chapter concludes with a treatment of the relativistic
equations of motion for spin and a remark on the notation and conventions of
11. 1 The Situatinn B -~- 1900. E' in's Two Postulates

In the 40 years before 1900 electromagnetism and optics were correlated and
explained in triumphal fashion by the wave theory based on the Maxwell equa-
tions. Since previous experience with wave motion had always involved a medium
fnr thP inn nf it ""~' n~tllr~l fnr nh •<;~;de tn ~"llmP. thl'lt liaht
needed~ m~edium through which to propagate. In view of the known facts about
light, it was necessary to assume that this medium, called the ether, permeated
all space, was of negligible density, and had negligible interaction with matter. It
ex1sted solely as a vehicle tor the propagatiOn ot electromagnetic waves.
The hypothesis of an ether set electromagnetic phenomena apart from the
rest of physics. For a long time it had been known that the laws of mechanics
were the same in different coordinate systems moving uniformly relative to one
. VV C MJ llldl lllC laW' Ul dlC UlV '-'' lldll,-
formations. To emphasize the distinction between classical mechanics and elec-

tromagnetism let us consider explicitly the question of Galilean relativity for
each. For two reference frames K and K' with coordinates (x, y, z, t) and
{ ' ' ' '\ · .1. c1 ;th r~l~t;,.~ .1. ;, v th~ cn~N· ~nc1
': ' J ' ' • /' ~ -.1 ' -o . -.; ' . ~r . .
nme cooramates m me two rrames are re1mea accormng w uamean reJauvny
by
x' = x- vt 111 1 \
\ . 'I
provided the origins in space and time are chosen suitably. As an example of a
mechanical system, consider a group of particles interacting via two-body central
pULClllldl>. Hl dll lJlJVlUU> llUI lllC Clf Ul Ul lllC llll jJdlll\.-lC 111
the reference frame K' is
mi dv; "" V ij (I xi'

dt' = - V'i L.J - xj ' I) (11.2)
From the connections (11.1) between the coordinates inK and K' it is evident
that v; = v; - v, v; = V;, dvjdt' = dv;ldt, and x; - x; = X;- xi. Thus (11.2) can
~ t ·'· " •
m -av;
dtI
= -V l
L .
V(lx·- x-1)
lj l }
(11.3)
1
namely Newtons equatiOn ot monon m tne reterence rrame K.

Th~ nf th~ fnr~ nf th~ ~,-,, nf ,1, ,1 mPC'h~niC'c
the transformation (11.1) is in contrast to~the change in form of the equations

~ --cllapterTI --s-pecml Theory of Relativity-G
governing wave phenomena. Suppose that a field !f!(x', t') satisfies the wave
equaTion
,z
I
y -u '
--r <T
'2 \
lf!=O (11.4)
. rir' 2 - r 2 rit' 2
in the reference frame K'. By straightforward use of (11.1) it is found that in

terms of the coordinates in the reference frame K the wave equation (11.4)
1 1 ,., \
"
VT-----v·V---v·Vv·V
0
)!f!=O (11.5)
c2 at 2 c2 at c2
The form of the wave equation is not invariant under Galilean transformations.
Furthermore, no kinematic transformation of 1J1 can restore to (11.5) the appear-
~ '" " * ~ -' •a '"' ·'UHO 1avl\.
' ' Ule 111Va11a11vv
. .....,..
T
• • -' ~ _,., .
mations is quite acceptable. The wind throws our voices. Sound waves are com-
• ~ " -~ . . •• • • ·' ',1. "+1. e '
-p r ll_,U
'
reference frame K' in which (11.4) is valid is obviously the frame in which the
transmitting medium is at rest.
So it also appeared for electromagnetism. The vital difference is this. Sound
waves ana similar wave phenomena are consequences of Galilean classical me-
L • ~• • e . ~
e _~
' •' '- L
u1 1-""'"u"u '"'"'"""" uamv> vvuvlv "''-' r . an;

simple is well understood in terms of the bulk motions of the media of propa-
~~··
--s'
-.:::. ' · "'~· '
'
..• . ' " .,
'
"' ~ ... ....
truly ethereal with no manifestation or purpose other than to support the
propagation.
When Einstein began to think about these matters there existed several
possibilities:
1. The Maxwell equations were incorrect. The proper theory of electromag-

netism was invariant under Galilean transformations.
2. Galilean relativitv aoolied to classical mechanics but electromatmetism had
a preferred reference frame, the frame in which the luminiferous ether was
at rest.
3. There existed a relativity principle for both classical mechanics and electro-
magnetism, but It was not Galilean relativity. This would imply that the laws
of mechanics were in need of modification.
The first nossibilitv """ h:<rclhr vi"hlP ThP "m:<7ino- <ll~~~cc~s nfthP M:<YWP.ll
theory at the hands of Hertz, Lorentz, and others made it doubtful that the
*The reader might wish to ponder the differences between the wave equation and the Schri:idinger
rurraer .n . II in!<.. me ,_-,uuuuinger equation reaas
li 2 3</J'
'V' 2 "/ + V•'l - iii
L:m at
~ L Hd' LUO Mill" 1Ullll lUf lfle
2
VaVe
' 'I'" v "a
and </J = </!' exp[i(m/h)v · x - i(mu /21i)t]. The Schri:idinger equation is invariant under Galilean
transformations.
Sect. 11.1 The Situation Before 1900, Einstein's Two Postulates 517
equations of electromagnetism were m serious error. The second alternative was

accepted by most physicists of the time. Efforts to observe motion of the earth
and its laboratories relative to the rest frame of the ether, for example, the
Michelson-Morley experiment, had failed. But for this important experiment at
least, the null result could be explained by the FitzGerald-Lorentz contraction
hypothesis (1892) whereby objects moving at a velocity v through the ether are
contracted in the direction of motion according to the formula
L(v) (11 6)
This rather unusual hypothesis apparently lies outside electromagnetism, since it

applies to bulk matter, but Lorentz late1 argued that it was rooted in eleetrody-
namics. He and Poincare showed that the Maxwell equations arc invariant in
form under what are known as Lorentz transformations (see Section 11.9) and
that the contraction (11.6) held for moving charge densities, etc , in electrody-
namics. With the idea that matter is electromagnetic in nature (the discovery of
the electron encouraged this hypothesis), It IS plausible to assume that (11.6)
holds for macroscopic aggregates of electrons and atoms. Lorentz thus saved the
ether hypothesis from contradiction with the Michelson-Morley experiment.
Other experiments caused embarrassment to the ether idea. Fizeau's famous
experiments (1851, 1853) and later similar experiments by Michelson and Morley
(1886) on the velocity of light in moving fluids could be understood only if one
supposed that the ether was dragged along partially by the moving fluid, with
the effectiveness of the medium in dragging the ether related to its index of
refraction!
Apparently it was the implausibility of the explanation of the Fizeau obser-
vations, more than anything else, that convinced Einstein of the unaeceptability
of the hypothesis of an ether. He chose the third alternative above and sought
principles of relativity that would encompass classical mechanics, electrodynam-
ics, and indeed all natural phenomena. Einstein's special theory of relativity is
based on two postulates:
1. POSTULATE OF RELATIVITY
The laws of nature and the results of all experiments performed in a given
frame of reference are independent of the translational motion of the system
as a whole. More precisely. there exists a triply infinite set of equivalent
Euclidean reference frames moving with constant velocities in rectilinear
paths relative to one another in which all physical phenomena occur in an
identical manner.
For brevity these equivalent coordmate systems are called mertzal reference
frames The postulate of relativity, phrased here more or less as by Poincare is
consistent with all our experience in mechanics where only relative motion be-
tween bodies is relevant, and has been an explicit hypothesis in mechanics since
the days of Copernicus, if not before. It is also consistent with the Michelson-
Morely experiment and makes meaningless the question of detecting motion
518 Chapter 11 Special Theory of Relativity-G
2. POSTULATE OF THE CONSTANCY OF THE SPEED OF LIGHT

The speed of light is finite and independent of the motion of its source.
Thi< nnduiMP , c+ rl mhPn J:;j c+ • ~ ;+fn~A "~~~rl rl . . -'

r . r. - r ..·~y unJy
lll
Y""".
,) TPth'
-sec C>ectiou ' t.L.n ), IS s1mpncny itsen. ret it rorces on us snPh
~ nf nur iclP~< ~hnut <n~rP n~rl ti~o thn+ ;+ mne
~ -u' lUf
uwuy yvmo.
Because special relativity applies to everything, not just light, it is desirable

to express the second postulate in terms that convey its generality:
2'. POSTULATE OF A UNIVERSAL LIMITING SPEED

In every inertial frame, there is a finite universal limiting speed C for physical
entities.
Experimentally, the limiting speed Cis equal to the speed c of light in vacuum.
Postulate 2' (with the first postulate) can be used equally to derive the Lorentz
transformation of coordinates (see Problem 11.1 ). Our own derivation in Section
11.3 is the traditional one, based on Postulates 1 and 2, but, as Mermin has
, . Lilt: rm structure 01 Lne Lorentz transrormation can De deduced
from the first nostnlMe >~lone_ nlus some o : " withn11t .£ ~<Sll
",_ _, . £ ,._,_ ..
LV CH'- • r v~ "IS~"' uo "'
LHv vLU]JULvU }'<ll<llllt:lt:l lll<ll Ul' me
transformatiOn from the Galilean !see Problem 11.2\.
l, • c+ ,f +1- ',1+1- ..
~ '"
cy'l, ,f '1. .1 ~-
t h roug h expenments ·.) · J

1 · h · ·) · ''r IS's
IS eat wit man extensive Jterature. orne re erences are
f 1l
given at the end of the chapter. Of particular note is the "Resource letter on
relativity" published in the American Journal of Physics [30, 462 (1962)]. This
article contains references to books and journal articles on the history, experi-
mental verification, and laboratory demonstrations on all aspects of special
relativity.
In passing we remark that Einstein's postulates require modification of the
laws of mechanics for high-speed motions. There was no evidence at the time
indicating a failure of Galilean relativity for mechanics. This is basically because
relativistic particles and their dynamics were unknown until the discovery of beta
r>~vs >~rmmrl 19()()_ Poinc:1rP. h>~rl snoecul,toerl th~t thP ' .-.f lio-ht mi;,.ht hP ~
,. . . _, £
. ' -
. .
"'"""'IS LUl WUCvHU< f-'W Uvl'-O, UUL LW>lt:lll S Ul lt:ldllVllY
originated from his desire to treat all phvsical phenomena in the same wav rather
+1- f. .rl ""- " n1. 1
~·
• n
v
.£ <L
') -r
'"
special tneorv ror mecnamcal concepts liKe momentum and enemv are discussed
in Section 11.5.
11.2 Some Recent Experiments

Although we omit discussion of the standard material, appealing to the reader's
prior knowledge and the existence of many books on the special theory of rela-
tivity, there are two experiments worthy of note. One concerns the first postulate,
.1. +L • C. "+L ~"'"f. ·~ r · r •• ,-,
'J •V W
"'"'-' ~"" \ v U l lHUUUll Ul LHt: la'
relative to the ether l and the other the second nostulate.
•J><. lJ. Nlermm, KeJatlVlty wtthout hght, Am. J. Phys. 52, 119 124 (1984).
Sect. 11.2 Some Recent Experiments 519
~. ~· ..,.
The null result of the Michelson-Morley experiment (IR87) established that
the velocity of the earth through the presumed ether was less than one-third of
its orbital speed of approximately 3 X 104 m/s. The experiment was repeated
timP< with v~rinll< mnclifir~tinn< ~~"'~"< mith nn firm PviciPnrP .-.f mntinn
•~ouu '~<v "'~ ~"'~'. cy v• au a

~ oo 5"'"" uy ~
eta!. [Rev. Mod. Phys. 27, 167 (1955)].

As already noted, these null results can be explained without abandoning
the concept of an ether by the hypothesis of the FitzGerald-Lorentz contraction.
Ulv UOO .... UVCOIJ ~J
.. IUvl ~"'JO) Ul
'
Ul dU>lHpLIUll Ul
gamma rays ~cauea me IVIOssoauer enect) auows companson or rrequenCJes to

astounding precision and gives the possibility of very accurate ether drift exper-
iments based on the Doppler shift. In the Mossbauer effect the recoil momentum
from the emission or absorption of a gamma ray is taken up by the whole solid
rather than by the emitting or absorbing nucleus. This means that the energy of
· is tntallv nteP"liP"ihlte. A '" r"v is temitt<erl with th<e f11ll F,. of thP.
2
nuclear transition, not the reduced energy E = £ 0 - E612Mc , where M is the
mass of the recoiling nucleus, resulting from the recoil. Furthermore, with such
recoilless transitions there are no thermal Doppler shifts. The gamma-ray line
thus approaches its natural shape with no broadening or shift in frequency. By
.1 " .L -L " " oL ".o oo. "
nuc1ear resonance ausorpuon or use lL as an msLrumenLIOr Lne sLUuy 01 ex1reme1y

small changes of frequency.
To understand the principle of an ether drift experiment based on the
Mossbauer effect, we need to recall the classic results of the Doppler shift. The
pnase 01 a p1ane wave ts an mvanam quanmy, me same m au cooramate rrames.
This is because the elapsed phase of a wave is proportional to the number of
wave crests that have passed the observer. Since this is merely a counting oper-
ation, it must be independent of coordinate frame. If there is a plane electro-
magnetic wave in vacuum its phase as observed in the inertial frames K and K',
-tPcl hv th<e Galilte"n iinM<e tr" .£ ion (11 1) is
_~ n·x n' · x'

<P = w1 t - - - = w' t'- (11.7)
c c'
If t and x are expressed in terms oft' and x' from (11.1), we obtain
I \ I _,
f' 1 H
0
H
0
A 0 f' U
0
A l
L C C _ C
Since this equality must hold for all t' and x', it is necessary that the coefficients
oft', x{, x~. x~ on both sides be separately equal. We therefore find
n - n'
(11.8)
c' = c- n · v
These are the standard Doppler shift formulas of Galilean relativity.

Thte nnit w"v" n, n i< ""'n from (11 R) to hP an invariant thP. samP in
all inertial frames. The direction of energy flow changes, however, from frame
520 Cha ptcr 11 Special Theory of Relativity-G
to frame. To see this, consider the segments of a plane wave sketched in Fig. 11.1.
The segments can be thought of as schematic representations of wave packets.
At t t' 0 the center of the segment is at the point A in both K and K' If
inertial frame K is the preferred reference frame (ether at rest) the wave packet
moves in the direction n, arriving after one unit of time at the point B in frame
K. The distance AB is equal to c. In frame K' the center of the wave packet
arrives at the point B' after one unit of time. Because of the Galilean transfor-
mation of coordinates (11.1) the point B' differs from B by a vectorial amount
1', as indicated in the bottom half of Fig. 11.1. The direction of motion of the
wave packet, assumed to be the direction of energy flow, IS thus not parallel to
n in K', but along a unit vector m shown in Fig. 11.1 and specified by
en - v
m= (11.9)
len- vi
Since the experiments iw. oh e photon propagation in the labor atm y, it is
convement to have the Doppler formulas (11.8) expressed m terms of the m
appropriate to the laboratory rather than n. It is sufficient to have n in terms of
m correct to first order in vic. From (11.9) we find
n =( l - m em+-;_:-
· v0 ) Vo
(11.10)
where v0 is the velocity of the laboratory relative to the ether rest frame
K'
Figure 11.1
Sect. 11.2 Some Recent Experiments 521
r u
in the laboratory, and w1 in an inertial frame K 1 moving with a velocity v 1 =
u 1 + v0 relative to the ether rest frame. From (11.8) the observed frequencies
ar"
n • Vn
c
If w 1 is expressed in terms of the laboratory frequency w0 and the wave normal
• • • ? ? • •
n IS enmmateu oy means or ~ 11.1 u J, rne , ,~~· 1v utuct u ~~ , 1' coa"'Y
shown to be
I
"' ' '\
where u1 is the velocity of the frame K 1 relative to the laboratory, m is the

direction of energy propagation in the laboratory, w0 is the frequency of the wave
m tne 1aoorawry, ana v0 IS me ve1ocny or me Jaoorarory willl respect 10 we elller.
Equation (11.11) forms the basis of the analysis of the Mi:issbauer ether drift
experiments. It is a consequence of the validity of both the wave equation in the
. .. . .,
COlllC01 1 COM 11 d111CO dllU V lt:ldLrvrry-u::rTI LV VLH'-'1 "''-'' U<ll u auH;oo.
:smce It mvolves v0 , tt obviously predicts an ether ann euecL. ~onswer rwo

Mi:issbauer systems, one an emitter and the other an absorber, moving with ve-
locities u1 and u2 in the laboratory. From (11.11) the difference in frequency
uc, .. ~~ . emuter anu · IS
If the emitter and absorber are locatea on Tile opposite ends ot a rod of length
2R that is rotated about its center with angular velocity !l, as indicated in Fig.
11.2, then (u2 - u1 ) • m = 0 and the fractional frequency difference is
sin!lt l(voLI (11.12)

c2
where (v0 )-' is the component of v0 perpendicular to the axis of rotation.
A resonant absorption experiment of this type was performed in 1963 in
Birmingham.* The Moss bauer line was the 14.4 keV gamma ray in 57 Fe, following
the {3+ decay of 57 Co. The isotope 57Fe is stable and occurs with a natural a bun-
dance ot l.L'i'o; tne absorber was made with 1ron enncnea to ::JL7o m · 'J:<e. 1ne
cobalt source was em planted in 56Fe. The emitter and absorber foils were located
as in Fig. 11.2 with R = 4 em. The observed fractional width of the Mi:issbauer
line was 11wlw = 2 X 10- 12 • Counters fixed in the laboratory and located sym-
metrically along a diameter of the circle m the plane o tne source and absorber
recorded the gamma rays transmitted through the absorber. Two rotational
speeds, !lL = 1257 s- 1 and !lH = 7728 s-\ were alternated during each 4-hour
cyCJe wat uata were raKen anu a murna1 euect co 1u 1uco coa• w ' 1 u1auvH
*D. C. Champeney, G. R. Isaak, and A. M. Khan, Phys. Lett. 7, 241 (1963). See also G. R. Isaak,
Phys. Bull. 21, 255 (1970).
UJ
\
/ ~
I ~1
/
''"
2v
(vo)l
1?/
J
~ ./
\ U2 Fi ure 11.2
was sought. From (11.12) it can be seen that with fl ~ 6000 s- 1 and R = 4 em,
an ether drift velocity of 200 m/s would produce a total change of frequency of
the magnitude of the Mbssbauer line width. The data showed no diurnal change
in transmission to an accuracy of 1 or 2%. The authors conclude that the mag-
nitude of the component of v0 past the earth in a plane perpendicular to the
earth's axis of wtation is I(v 0 ).LI 1.6 -'- 2.8 m/s, a null result. An improved
experiment along the same lines in 1970 gave a limit of 5 em is (see lsaak op cit)
A conceptually similar experiment was performed in 1958 using ammonia
masers.* The ammonia molecules have a well-defined direction and nonzero
speeds when they enter the maser cavity. According to (11.11) there is therefore
a shift in the frequency. If the frequenctes of two masers whose ammoma mol
ecules travel in opposite directions are compared, there should be an observable
beat Lequency. Fm thermore, if the two masers are rotated together through 130",
the beat frequency should change by /!;w/u>g- 4 ln"'"1 • v0 l/c2 The null result of
this experiment set the component of ether drift velocity at less than 30 m/s.
These two Doppler shift experiments set observable ether drift speed limits
6000 and 1000 times smaller than the speed of the earth in its orbit and make
the idea that we can ever detect any motwn relal!ve to some "absolute 'reference
frame quite implausible.
B. Speed ofLighlfrom a Movin~ Source

The second postulate of Einstein. that the speed of light is independent of
the motion of the source, destroys the concept of time as a universal variable
independent of the spatial coordmates. Because thrs was a revolutwnary and
unpalatable idea, many attempts were made to invent theories that would explain
all the observed facts without this assumption. The most notable and resilient
scheme was Ritz's version of electrodynamics (1908 1911) Ritz kept the two
homogeneous Maxwell equations intact, but modified the equations involving
the sources in such a way that the speed of light was equal to c only when mea-
sured relative to the source. The Ritz theory is in accord with observation for
the aberration of star posil!ons, the Frzeau expenments, and the ongmal
'C. J. Cedmholm, G.!'. Bland, B. L. Havens, and C. H. Iownes, Phys. Rev. Lett. I, 342 (1958). See
also T. S. Jaseja, A. Javan, J. Mnrray, and C. H. Townes, Phys. Rev. 133, A1221 (1964).
~~orne
~"'"
;;J . .J
Michelson-Morley experiment. It is customary, however, to cite Michelson-

Morley experiments performed with extraterrestrial light sources (the sun or
other stars) and light from binary stars as establishing the second postulate and
ruling out Ritz's theory. Unfortunately, it seems clear that most of the early
evidence is invalid because of the interaction of the radiation with the matter
through which it passes before detection.*
oto th~t r1~ n~t onff~r fr~m
"1"'1.
,• , 'F
the criticism of Fox. The most definitive is a beautiful experiment performed at
CERN, Geneva, Switzerland, m llJ64. The speed oro uev pnotons proaucea
in th" rl""11v of v"rv ,n, i" n"ntr11l nions was measured bv time of fti2:ht over
paths up to 80 meters. The pions were produced by bombardment of a beryllium
• 0 _, £ £_ _, _,
LargtOL uy D'.L VC V plULUll' i111U 11i1U "!' '-'U <H'-'UOU~'-'U -r V~ HV"'
char2:ed oions oroduced in the same bombardment) of 0.99975c. The timing was
done by utilizing the rf structure of the beam. Within experimental error it was
£. ~ •L •L ~ .£d. .t. ' ~ L •L ,1. ',-11. '
~vuuu cuuc cuv -,- v~ '"" !-'"'"' ') -. ~ -_, 'b
source was equal to c. If the observed speed is written as c' = c + kv, where v
is the speed of the source, the experiment showed k = (0 ± 1.3) X 10- 4 .
The CERN experiment established conclusively and on a laboratory scale
the validity of the second postulate (2) of the special theory of relativity. Other
experiments* on charged particles and neutrinos independently establish the
validity of postulate 2'. See also Section 11.5.
,.. -,-, ~
_, I' ·'- ..._ -' .L' .- • .L 'T7
'--'• ~-, . .._. ~~- VJ uu;; Jl' VJ ~•15••• ••• '
"T'L rl -~1·-· . ' -~ f, rt · 1 ()9 fr~m m~~
suremen;; at infra;:d frequencies and lesser ;~curacy at higher frequencies (or

equivalently, the meter is defined to this preCisiOn). Une can ask wnetner tnere
;0 ~nu
... • _t f~r ~ fr~~n~n~U rlP~Pnrl~n~~ ~f
"1 - . -y
;hP rl Of pj, •
waves in vacuum. One possible source of variation is attributable to a photon

mass. 1 ne group vewcny m tnis case is
c(w) ( 2r2
= c 1 - :~ (11.13)
where the photon rest energy is liw0 . As discussed in the Introduction, the
mere existence of normal modes in the earth-ionosphere resonant cavity sets a
limit of w0 < 10c/R where R is the radius of the earth. From radiofrequencies
( w ~ 108 s 1 ) to w ----> oo, the change in velocity of propagation from a photon
1-A- 10
u•uoo
•
~o
o
'"'-'' v .v v
1.
ovOO
•L A
,,,
,
Another source of frequency variation in the speed of light is dispersion of

the vacuum, a concept lying outside special relativity but occurnng m models
with " ,. ·•~ .t;mP ThP eli of nnk:>r~ make it no~~ihlP to test this
*See the papers of criticism by J. G. Fox, Am.J.Pnj;s. "3lJ,297lT%2J; .i.i, 1 ( 1~n~ ); .1. Upt. :>oc. ':if,
967 (1967). The second paper cited is a detailed discussion of Ritz's emission theory and a critique
of the various arguments against it. See also T. Alvager, A. Nilsson, and J. Kjel!man, Ark. Fys. 26,
209 (1963).
'T. Alvager, J. M. Bailey, F. J. M. Farley, J. Kjellman, and I. Wallin, Phys. Lett. 12,260 (1964); Ark.
"' . ~1, 1Ll' 11 Qt;<;\
'G. R. Kalbfleisch, N. Baggett, E. C. Fowler, and J. Alspector, Phys. Rev. Lett. 43, 1361 (1979).
idea with high precision. Pulsar observations cover at least 13 decades of fre-
quency, with any one observing apparatus having a certain "window" in.. the
frequency spectmm The quite small time duration of the pulse from some pulsars
permits a simple estimate for the upper limit of variation on the speed of light
for two frequencies w 1 and uJ 2 inside the frequency window of each apparatus.
c D
where !::.tis the pulse duration and Dis the distance from the somce to observer.
For the Crab pulsar Np 0532, !::.t = 3 X 10- 3 sand D = 6 X 10" light-years so
that (c tit/D) - I. I X 10 14 . Vanous overlappmg observatiOns from 4 X 108 Hz
through the optical region and up to photon energies of 1 MeV indicate constancy
of the speed at the level of ticJc < 10 11 by this simple esnmatwn.'-· For higher
energies, an experiment at the Stanford Linear Accelerator;- compared the speed
of 7 GeV photons with that of visible light and found tide < 10 '. Up to very
high energies, then, there is no evidence for dispersion of the vacuum. The speed
of light is a universal constant, independent of frequency.
11.3 Lorentz Transformations and Basic Kinematic

Results ~.f." Special Relativity
J'.s is well known, the constancy of the velocity of light, independent of the mo-
lion of the source, gives rise to the relations between space and time coordinates
in different inertial reference frames known as Lorentz transformations. 'Ne de
rive these results in a more formal manner in Section 11.7, but for the present
summarize the elementary derivation and important consequences, omitting the
details that can be found in the many textbooks on relativity. The reader who
wishes more than a reminder can consult the books listed at the end of the
chapter.
A. Simple Lorentz Transformation of Coordinates

Consider two inertial reference frames K and K' with a relative velocity v
between them. The time and space coordinates of a point are (t, x, y, z) and
(t', x', y', z') in the frames K and K', respectively. The coordinate axes in the
two frames are parallel and oriented so that the frame K' is moving in the positive
z direction With speed v, as VIewed from K. For SimpliCity, let the ongms of the
coordinates in K and K' be coincident at t = t' = 0. If a light source at rest at
the origin inK (and so moving with a speed v in the negative z direction, as seen
from K') is flashed on and off rapidly at t = t' = 0, Einstein's second postulate
implies that observers in both K and K' will see a sphelical shell of 1adiation
expanding outward from the respective origins with speed c. The wave front
reaches a point (x, y, z) in the frame Kat a timet gi-ven by the equation,
(11.14)
*.T. M. Rawls. Phys. Rev. DS, 487 (1972).

B. C. Brown et al.. Phys. Rev. Lett. 30, 763 (1973).
-. ..... _ ll_l r ' ""cl R".;"
. · · RP•nlt• of Snecial Relativitv 525
Similarly, in the frame K' the wave front is specified by

,2 !2 f •2 •2 ' •2\ () f11 !A'\
\• J
' \
'
With the assumption that space-time is homogeneous and isotropic, as im-
ptieu oy Lne 11rst postutate, tne conneCTion oet' '""" '"" Lwv ' " ' ' vt
is linear. The quadratic forms (11.14) and (11.14') are then related by
cLt•2 _ (x'z + y'z + z'z) _ Az[cztz _ (xL + YL + zL)] (1U5)
where A = A(v) is a nossible chanPe of scale between frames. With the choice of
orientation of axes and considerations of the inverse transformation from K' to
Kit is straightforward to show that A(v) = 1 for all v and that the time and space
coordinates in K' are related to those in K by the Lorentz transformation
x /1 - y(xo - f3x t)
x; = y(x1 - {3x0 )
\ILlV) "
x~ = x2
r! - Y-
-- ct, x 1 - X,
where we have introduced the suggestive notation x 0 -- Z, Xz
-
y ''
auu aiOV LHC 1.-Vll'
- ~ '
AJ ~
v I rtl
R-
r , r:! -
IC
c (11.17)
_o "o
'Y ~1 VJ
The inverse Lorentz transformation is
Xo = y(xi1 + {3x 1)
'1
.{
\"I
' 0, ' \
~ 'U}
(11.18)
x2 = x2
x3 x3
It can be found from (11.16) by direct calculation, but we know from the first
postulate that it must result from (11.16) by interchange of primed and unprimed
• '' ' "•' ' • :, • • ~ ' --l" f11 1C\ f11 10\
vat WILH a 111~ ·~~~":""5 ·~ _\''·'~! ~ \· · ~Jo
the coordinates perpendicular to the direction of relative motion are unchanged

"'hilP thP n~r,..llP-1 'in,..tP nml thP timP "rP " This Ciln he con "
with the Galilean transformation (11.1).
Equations (11.16) and (11. 17) describe the soecial circumstance of a Lorentz
transformation from one frame to another moving with velocity v parallel to the
0 0 0
0
' H O L 1_ • ' d. £.
A 1 dXt>. H LHC dXJ;;' l l l 1"1.. dllU A
u"' '"" l'"' """'• .0 T V I ' " " uwu-.
K' in frame K is in an arbitrary directiOn, the generalizatiOn ot pl.lb) IS
Xi. 'VIXn H · xl
(y-1){0 (11.19)
' {32 \I" ., ...
.. \ 0
I
,R
r "U
~
the first
ThP fir<t Pnn,..tinn hPrP fnllnw< ,..]mn<t trivi,..llv_from inn in (11.16)
The second appears somewhat complicated, but is really only the sorting out of
comoonents of x and x' oarallel and oeroendicular to v for separate treatment
in accord with (11.16).
ThP rnnnPrtinn • R ::~nri 'V aivP.n in (11 171 Qnrl thP rQn<TP<
-o
n < D
t 1,
1 < 'Y < oo allow the alternative para~etrization, .
B- tanht
and so
'Y = cosh ( (11.20)
'Yf3 = sinh?
where ? is known as the boost parameter or rapidity, In terms of? the first two
equations of (11.16) become
x0 - x 0 cosh? x 1 smh?
(11.21)
xi = - x0 sinh ? + x 1 cosh ?
The structure of these equations is reminiscent of a rotation of coordinates, but
"•< <
uc •ypv
-L .1"
uvu~
.-l
C. "
v•
•uu~uvuo
.1
" ' . " ' .C
Ulc ·lll<O
• '
~u~u•m,
•
l<Oial!Ve
negative sign between the space and time terms in (11.14) [see Section 11.7 and
111 nc\1
\ Ojj•
D A TT.
~. ~
. ~~·~·~
The Lorentz transformation (11.16), or more generally (11.19), describes the

transformation of the coordinates of a point from one inertial frame to another.
Just as for rotations in three dimensions, the basic transformation law is defined
m terms of the coordinates of a point. In three dimensions we call x a vector and
speak of x1o x 2 , x 3 as the components of a vector. We designate by the same name
any three physical quantities that transform under rotations in the same wav as
the components of x. It is natural therefore to anticipate that there are numerous
physical quantities that transform under Lorentz transformations in the same
manner as the time and space coordinates of a point. By analogy we speak of
4-vectors. The coordinate 4-vector is (x 0 , x 1 , x 2 , x3 ); we designate the components
of an arbitrary 4-vector Similarly as ~A 0 , A~o A 2 , A 3 ), * where A~o A2o A 3 are the
components of a 3-vector A. The Lorentz transformation law equivalent to
\11.10) wr an arourary '+·Yenor IS
Ai1 = 'Y(Ao - 13 • A)
Ail - 'Y(AII - f3Ao) (11.22)
Al =A,
where the parallel and perpendicular signs indicate components relative to the
veracity v cp. 1 ne in variance irom one inerriarrrameco another embodied
through the second postulate in (11.15) has its counterpart for any 4-vector in
.lH\0
' lH ,
A~ 2 - IA'I 2 - A~- IAI 2 (11.23)
*Because we are deferring the explicit algebraic treatment of the Lorentz group to Section 11.7. we
do not write a sin ole svmbol for this 4-vector As written thP~I orP thP ot< ~fthP c~ntrouorinnt
4-vector A".
(11.24)
This result can be verified by explicit construction of the left-hand side, using
(11.22) for the primed components, or using (11.23) for the sum of two 4-vectors.
It is the Lorentz transformation analog of the in variance of A · B under rotations
in three dimensions.
C. Light Cone, Proper Time, and Time Dilatation

A fruitful concept in special relativity is the idea of the light cone and "space-
like" and "timelike" separations between two events. Consider Fig. 11.3, in which
the time axis (actually ct) is vertical and the space axes are perpendicular to it.
For simplicity only one space dimension is shown. At t = 0 a physical system, say
cone for times t > 0, that region is called the future. Similarly the lower half-cone
is called the past. The system may have reached 0 by a path such as AO lying
inside the lower half-cone. The shaded region outside the light cone is called
elsewhere. A system at 0 can never reach or come from a point in space-time in
elsewhere.
The division of space-time into the past-future region (inside the light cone)
and elsewhere (outside the light cone) can be emphasized by considering the
invariant separation or interval s 12 between two events P 1 (th x1 ) and P 2 (t2 , x2 ) in
space-time (we are reverting to t and x temporarily to avoid proliferation of
subscripts). The square of the invariant interval is
sf2 = c2 (t 1 - t2 ) 2 - IX1 - x2 12 (11.25)
For any two events P 1 and P 2 there are three possibilities: (1) si 2 > 0, (2) si2 < 0,
(3) si 2 = 0. If si 2 > 0, the events are said to have a time/ike separation. It is always
possible to find a Lorentz transformation* to a new coordinate frame K' such

that x; = x~. Then
s'f 2 = c 2 (t{ - t~) 2 > 0
In the frame K' the two events occur at the same space point, but are separated
;n t;~o U · ' •rr t~ ~;rr 11 "! ',t ~ 1~ ·_, •'- ' ' •
' ··o 'D' ' ' r '5' UUU tile
other lies in the past or future. If s'f 2 < 0, the events are said to have a svacelib
separation. Now it is possible to find an inertial frame K" where t~ = t~. Then
"2
. '"
I v"
'i '£
"12 __.. {'\
I
Tn V" tl-.o +
fF 11 3 .
,, rW'I'
h.,. . ..r h'l h h
,t
~' uute.
,,_
-·
I' .
. r ,
n terms o 1g. . , one event IS at t e ongm, w 1 e t e ot er ws m t e else-

uohPrP rPrrinn ThP fin~] nnooihilih• <2 () ·~n1;oo n l:~l.tl;/rn "" 'T'L
c r "' "" ' '<'>' -r · uu;;
events lie on the light cone with respect to each other and can be connected onlv
by light signals.
The division of the separation of two events in space-time into two
classes-spacelike separattons or t1melike separations with the light cone as the
1-.~.,nrl wo o r . ~ ot ' . T ' , ' 'T'. '+c
like separation in one coordinate system have a spac~like separation in ~II co:
nn1in~tP Thio mP~no th~t tuon on.-h onto · l-.o n•ononll, n~ -'
'J •
Since physical interactions propagate from one point to another with velocities
no greater than that of light, only events with timelike separations can be causally
related. An event at the origin in Fig. 11.3 can be influenced causally only by the
events that occur m the past regton of the light cone.
A n~+t-.o~ ,oof, 1 n, ,, ' · r. '-' ~-~
I .. ·
' ' ' L
.
mteness we WI'll t h'm k o f as, a partie
'" .. e, movmg w1t h an mstantaneous
· · u~(')
ve 1oCity u t
·o '
' . tn <nmP inPrti~l K Tn ~ timP . ,J At ;to . . nh ...
' mtes1ma J mvariant
d x = u d t. From ( 11.25) the square of the correspon d'mg m J ' VJ fi ' ' C
interval ds is
de 2 - r 2 dt 2 I dv 2 -
1 r 2 At2(1 Q2\
f
... horo Q _,. •gfo v•

ulr Tn tho
'J
"" +'-
"J
neously at rest the space-time increments are dt' =dT, dx' = 0. Thus the invariant
,] ;, fi< r fiT ThP. ;n.-r<>mont nf timP -~~ ;n tho inofgn+ ,no rod r.
of the system is thus a Lorentz invariant quantity that takes the form,
dT = dtv1 - {3 2 (t) dt
= '7:\ (11.26)
f\' I
'rt ,11 ",,_ . c "··~ . '
·r "r ·~ ~} y~• "~ u o Y''""'· T
H
• ''
<> "''-' ""''-' ao ''-'~"
m the rest trame ot the system. !'rom (ll.Lb) It tallows that a certam proper time
;nto~un 1 mill l-.o H f V ,. '•• .1
'£
OPP '
,
(T2 rl, rrz
t2 - t1 = JT, Y1 - {32(7) = L, y(T)dT (11.27)
Equation (11.27) or (11.26) expresses the phenomenon known as time dila-

tation. 1-\. moving ClOCK runs more swwty wan a stanonary clocK. !"or equal nme
'DY · · . c<.[Uutiun> ]lLWJ, t11e reauer can veruy tnat tnere ex1sts a Lorentz transrormanon
with f3 0. Explicitly, 1131 ~ lx, - x2l /c I1, - t2l·
Sect. 11.3 Lorentz Transfonnations and Basic Kinematic Results of Special Relativity 529
intervals in the clock's rest frame, the time intervals observed in the frame K are
greater by a factor of r > I. I h1s paradoxical result IS venfied dmly m h1gh-
energy physics laboratories urbere beams of unstable partjc]es of known Jjfetjmes
,0are tr anspm ted befm e decay ov et distances many many times the uppet limit
on the Galilean decay distance of cT0 • For example, at the Fermi Natwnal Ac-
celerator Laboratory charged pions with energies of 200 Ge V are produced and
transported 300 meters with less than 3% loss because of decay. With a lifetime
of To = 2.56 X w-s s, the Galilean decay distance is CTo = 7.7 meters. Without
time dilatation, only e- 30017 ·7 = 10- 17 of the pions would survive. But at 200 GeV,
r = 1400 and the mean free path for pion decay is actually ycT0 = 11 km!
A careful test of time dilatation under controlled laboratory conditions is
afforded by the study of the decay of mu-mesons orbiting at nearly constant speed
m a magnetic field. Such a test, mCidental to another expenment, confirms fully
the formnla (11 27) [See the paper by Bailey et al cited at the end of Section
A totally different and entertaining experiment on time dilatation has been

performed with macroscopic clocks of the type used as official time standards.*
The motion of the clocks was relative to the earth in commercial aircraft, the
very high precision of the cesium beam atomic clocks compensating for the rel-
atively small speeds of the jet aircraft. The four clocks were flown around the
world twice, once in an eastward and once in a westward sense. During the
journeys logs were kept of the aircrafts' location and ground speed so that
the integral in (11.27) could be calculated. Before and after each journey the
clocks were compared with Identical clocks at the U.S. Naval Observatory. With
allo'Nance for the earth's rotation and the gravitational "red shift" of general
relativity, the avetage observed and calculated time differences in nanoseconds
are -59 ± 10 and -40 ± 23 for the eastward trip and 273 ± 7 and 275 ± 21 for
the westward. The kinematic effect of special relativity is comparable to the gen-
eral relativistic effect. The agreement between observation and calculation es-
tablishes that people who continually fly eastward on jet aircraft age less rapidly
than those of us who stay home, but not by much!
D. Relativistic Doppler Shift

As remarked in Section 11.2.A, the phase of a wave is an invariant quantity
because the phase can be identified with the mere counting of wave crests in a
wave train, an operation that must be the same in all inertial frames. In Section
11 2 the Galilean transformation of coordinates (11 1) was used to obtain the
Galilean (nonrelativistie) Doppler shift formulas (11.8). Here v;e use the Lorentz
transformation of coordinates (11.16) to obtain the relativistic Doppler shift. Con-
sider a plane wave of frequency w and wave vector k in the inertial frame K. In
the moving frame K' this wave will have, in general, a different frequency w' and
wave vector k', but the phase of the wave is an invariant:
cp = wt - k · x = w't' - k' · x' (11.28)
[Parenthetically we remark that because the equations of (11 16) are linear the
plane wave in K with phase rf> indeed remains a plane wave in frame K' .] Using
*I c Hafele and R F Keatjpg, Science 177, 166, 168 (1972)

(11.16) and the same arguments as we did in going from (11.7) to (11.8), we.find
that the frequency w' = ck(, and wave vector k' are given in terms of w "" rk
0
and k by
(11.29)
k~ = kj_
The I orentz transformation of (k 0 k) has exactly the same form as for (x x)
4-vectors (11.24). This correspondence is, in fact, an alternate path from (11.28)
to the transformation law (11.29).
For light waves, lkl = k 0 , lk' I ""kf,. Then the results (11.29) can be expressed
in the more familiar form of the Doppler shift formulas
w' yw(1 - {3 cos 11) (11.30)
tan 11'
where 11 and 11' are the angles of k and k' relative to the direction of v. The
mverse equatiOns are obtamed by interchanging primed and unprimed quantities
and reversing the sign of {3.
The first equatiOn m (1 I .30) IS the customary Doppler shift, modified by the
factor of y. Its presence shows that there is a transverse Doppler shift, even when
IJ w/2. This relativistic transverse Doppler shift has been observed spectro-
scopically with atoms in motion (Ives-Stilwell experiment, 1938). It also has been
observed using a precise resonance-absmption Mossbauei experiment, with a
nuclear gamma-ray source on the axis of a rapidly rotating cylinder and the ab-
sorber attached to the circumference of the cylinder.*
11.4 Addition of Velocities, 4- Velocity

The Lorentz transformation (11.16) or (11.18) for coordinates can be used to
obtain the law for addition of velocities. Suppose that there is a moving point P
whose velocity vector u' has spherical coordinates (u', 11', cf>') in the inertial frame
K'. as shown in Fig. 11.4. The frame K' is moving with velocity v "" cB in the
positive x 1 direction with respect to the inertial frame K. We wish to know the
components of the velocity u of the point P as seen from K. From (11.18)
the differential expressions for dx 0 , dx 1 , dx 2 , dx 3 are
dx 1 Yv(dxj + {3 dxi,)
*H. J. Hay, J. P. Schiffer. T. E. Cranshaw, and P. A. Egelstaff, Phys. Rev. Leu. 4, 165 (1960). See
also I. E. Cranshaw m Proceedzngs of the international School of Physics, Yarenna, Course XX, 1961,
Academic Press, New York (1962), p. 208.
Sect. 11.4 Addition of Velocities, 4-Velocity 531
.
%1
p
~
-A
v
. i' --"""-.....
I
i
I
%3
rp' '-.1
v
X2'
0 -Xl!
/
~/ Figure 11.4 Addition of velocities.
where we have put a subscript on y to distinguish it below from Yu =

(1 - Lac 2) - 112 and Yu· = (1 - u' 2/c 2) - 112 . The velocity components in each frame
~r' ' ~ Av' /Av' ~nrl" ~A, !Av_ 'T'hio nw~n•• th~t thP •+• .-.fup].-,,-.itu
' I ·u• -r 'J
' <H 111 ' ~· '
ll dll ~I _ lU
0
--
u; 1
+v
u,
v · u'
1 + -2-
/" ""
\~~.~~,
u~
ut --
Yu( 1 +v-cz· u')

-
Th<e notl'ltion u, ancl nents of velocitv narallel and nernendic-

n to
ular, respectively, to v. The magnitude of u and its polar angles e, </>in the frame
K are easily found. Since u 2 /u~ = u 3 /u~, the azimuthal angles in the two frames
are equal. Furthermore,
,.. u sm tl
<UU U
" (,' ,..,-" il' + n\
and (11.32)
I ,.., 2
'
I ,•2 + , 2 + :),•, ....0 R'
y c I
u=
u'u
1 + r2 COS I)'
The inverse results for u' in terms of u can be found as usual from (11.31) and
(11.32), by interchanging primed and unprimed quantities and changing the sign
of v.
532 Chanter 11 Soecial Theorv of Relativitv-G
For speeds u' and v both small compared to c, the velocity addition law
( 11.31) reduces to the lialllean result, u - u + v, but 11 e1ther speed 1s compa-
rable to c modifications appear. It is impossible to obtain a speed greater than
. . . . . .
utat 01 ngm uy auumg tWO vetOCiues, even u eacu IS very ctose 10 c. ror the
simple case of parallel velocities the addition law is
' v
u = ., (11.33)
vu'
1 +-
c
If u' = c, then u = c also. This is an explicit example of Einstein's second pos-
.1 'T'L ~- .L .L £. •'- ~ " " f 1 '>"\ Ll.
LU.UCv. ~ Hv < ~UH H VH< CHv V':[ UUUVH >H ~ \ .~~ j <HCH U c
implies u = c for nonparallel velocities as well.
The formula for the addition of velocities is in accord with such observational
rests as Lne rizeau experimems on me speea or tigm in moving tiquias ana the
aberration of star positions from the motion of the earth in orbit.
UlC MlUUUlC Ul ~~L.:H) • 1l UUVIUU:> lll<ll lllC litW Ul ll, n OI
velocities is not that of 4-vectors, as given by (11.22) and of which (11.16) and
(11.29) are examples. There is however, a 4-vector closely related to ordinary
ve1oc1ty. To exh1b1t th1s 4-vector we rewnte lll.31 ). !:'rom the second equation
in (11.32) it can be shown directly that the factor (1 + v · u'/c 2 ) can be expressed
alternatively through
I ,\
V•U
Yu = y., Yu• I 1 + ----:;-
c:
l (11.34)
where y,, 'Yu, 'Yu· are the gammas defined by (11.17) for v, u, and u', respectively.
un ,.r1 1 .'>.A\., · · r1 • r 1 1 .'>.1.,\
\
,h
"'1'
· '-
'YuUII = 'Yv('Yu·U\ + uy,.) (• <>n

~~LJJ J
/U 1.. {U '
J.
Comparison of (11.34) and (11.35) with the inverse of (11.22) implies that the
four quantities ( 'YuC, 'Yuu) transform in the same way as (x 0, x) and so form a
4-vector under Lorentz transformations. These four quantities are called the
time and space components of the 4-velocitv ( Un. U).
An alternative approach to the 4-velocity is through the concept of proper
time r. Ordinary velocity u is defined as the time derivative of the coordinate
x(t). The addition law (11.31) for velocities is not a 4-vector transformation law
because time is not invariant under Lorentz transformations. But we have seen
that the time r is a Lorentz i · mt. We can thus 't· ,, a 4-vector
" .1
.• " '-"J ~'''·
.1. £ "LI.
" ' " ' " ~-,..,~'"' ~.< 0 , AJ " " " '"ov~~· '"
.£
'"
.\ ~
..
Using (11.26) we have
= dx 0 -_ dx 0 dt -_
U(_)-
'YuC
dr dt dr
\ll..JU)
ux ux Ul
U - - = - - - - = 'YuU
dr dt dr
VI~- • Ul a valll'-1'- U< ~
n" " ' " " " ' "'" """' uw< '"" ·r
orooortional to its total ener!!:v and momentum.
~ect. 1L!5 Ketativishc lvtomentum amt J!.uergy ot a Parncte ::1-'-'
11.5 Relativistic Momentum and Energy of a Particle

We next consider the relativistic generalizations of the momentum and kinetic
energy ot a particle. These can be obtamed tor charged particles tram the Lorentz
j',
rl ·~
• j', ,j' ,]. ,,· h.>lrl, ,,
ready e~~ablished by Lorentz before 1900, b~t it is useful to give a more general
riPriv~tinn t. --' nnhr on thP bur< of inn of ~nrl m, 1m onrl
on the kinematics of Lorentz transformations. This appr~'ach shows clearly the
universality of the relationships, independent of the existence of electromagnetic
interactions for the particle in question.
!-<or a particle With speed small compared to the speed ot light Its momentum
rl .,_, ,r, t. ~
p = mu
\H.J I)
E = E(O) + ~mu 2
where m is the mass of the particle, u is its velocity, and E(O) is a constant
.__,_
ifiPrl ~< thP rP<t PnPrrnr nf thP · ·1 Tn > 1'
.. .__,
inn< th<>
rest energies can be ignor~d; they c'antribute the same additive constant to both
sides of an energy balance equation. In special relativity, however, the rest energy
cannot be ignored. We will see below that it is the total energy (the sum of rest
energy plus kmetlc energy) ot a particle that ts s1gmhcant.
ur, · .•h JC; r1 • • • f +h ,, • r1 ,f .,],
consistent with the Lor~~tz transformation law (11.31) of velociti~s and r;ducing
to (11 '\7\ for nnnrPl · ' ' , mntinn ThP nnhr nn«ihlP -~li7otinn< rnn• '
• < ~
with the first postulate are

p = M(u)u
(11.38)
E = ~(u)
where M(u) and ~(u) are functions of the magnitude of the velocity u. Compar-
ison with (11.37) yields the limiting values,
M(O) = m
( 11.39)
a~ (O) = m
au 2 2
VV t; LJ1t: 1' lU1t: '"~Ull1jJL1Ull Llldl JVL\ U) dllU

-16\U) d1 t; Wt:ll-Ut:Jld Vt:U -
tonic functions of their arguments.
~- --1 • Ll. £. £ _, co. ' 'J .,_ .l . ll."
" ' uvcvuwuv wv v< J''\"J auu "\"} " v wv
of two identical particles and require that conservation of momentum and energy
hold in all equivalent inertial frames, as implied by the first postulate. In partie-
ular, we consider the collision in two frames K and K' connected by a Lorentz
transformation parallel to the z axis. A certain amount of algebra is unavoidable .
..
' U !'.Ct:jJ lL LU d lWll, LWU d} r 'ell<:: UjJC11. Vllt: 1~ LU ~t;L UjJ LllC VC1Ul.lllt:>
and directions of the particles in such a clever way that the algebra shakes down
".LL ' l ~L ".1 ".L
HHV au "auu LvHL lvOUH. Ulv VUlvl 10 w p•v"- a .>ua•;e,•u-
forward kinematic situation and proceed iudiciouslv. The first approach lacks
motivation. We adopt the second.
Let the inertial frame K' be the "center of mass" frame with the two identical
~ rhent<er 11 ofR,
0 0 0
-G
Ua'=v
I
7
ub'- -v
•
I
/ll,;t - v H"' L'''
lU Ihe
' c '
o~ auu uua• ''-'V'-HJ
I
• .,.
frame K' for the collision of two identical particles.
particles having initial velocities u~ = v, ub = -v along the z axis. The particles

collide and scatter, emerging with final velocities, u~ = v', ud = v". The various
velocities are indicated in FiP". 11 'i Tn K' thP >ti r.n '" <.
,., ,., • "
u ·~ u• •~u~
Pa, + Pb, = Pc' + Pd,

E'a + E'b = E'c + E'd
or, with the forms (11.38),
M(u)v .JIA,( V )V - .JIA,( V )V + .M.(v')v
!11.4())
0~V) -r 0~V) 0~V ) 'iDWJ
Because the particles are identical it is necessary that 'jg(u') = 'jg(u") and, with
the hypothesis of monotonic behavior of 'jg(u ), that v' = v". The second equation
in (11.40) then demands v' = v" = v. The first equation requires v'' = -v'. All
f011T ,1, iti<e~ h:oJVfe th<e ~:oJmf>. m:oJP"niflldfe With thP finc.J .,~l~~itiP< Pnll~J ~nrJ ~~
.. .. . . -. T
I" , J UM uu:;
d~ UHO v ,, • < Ill, < au 1e!UUVIOTis-sTaU.OU 1 ;1s snown
m the nght-hand d1agram of Fig. 11.6 where the scattering angle inK' is denoted
bye·.
We now consider the collision in another inertial frame K moving with a
velocity -v in the z direction with respect to K'. From the transformation equa-
"'
z z'
u.' ~
-"
--
Ua
* \Ud
b
•
--
-~ ..... ~-
% lib'
Tl' Tl''
Ngure J.J.,o miLiat anu una• vcwcny vecLors m 1rames K and K' tor tne colllsJOn or two
identical particles. The lengths and angles of the solid lines representing the velocities
correspond to (}' = 30° and {3 2 = ~. The dashed lines in K are the results of a Galilean
transformation from K' to K.
Sect. 11.5 Relativistic Momentum and Energy of a Particle 535
tions (11.31) for velocity it can be seen that particle b is at rest inK while particle a
is incident along the z axis with a velocity
?.v ::>cR
Ua (ll.41)
1
u I -t p
cz
wh<er<e R vic Th<e '1. itv rnmnnnPnt' nf th<e. fin:1l itiP., u. :mrl11 , in K l'ITP.
similarly
cf3 sin f)' ( ) = cf3(1 + cos 8')
(uc)x =
1 + {3 2 cos 8'
llc z
y(l + {3 2 cos 8') ,
(11.42)
cf3 sin 8' ( ) _ cf3(1 - cos 8')
(ud)x = - ud z-
y(1 - {3 2 cos 8') , 1 - {3 2 cos 8'
withy= (1 - {3 2)- 112 .
The equations of conservation of momentum and energy in the inertial frame
Kread
JUC~Ua)Ua T JUl~Uf>)Uh '
JUC~Uc)Uc ' '"
JUl~Ud)Ud
{11 A'!\
'IJ.(u ) + 'IJ.(u ) - 'IJ.{u _) + 'fj(u ,) \ '/
Tt ;, ".
•1. :1.
y
. ,1,
frnm
.... .. ...
(11 d.1) "nrl (11 d.?.)
~'-
'"
nr thP.lPft-h"nrl rli:wr"m nfFia
,1, ;,; .,11 ->:rr_ .' .
11 h thM
D
'1
1nus me aeLermination or Jllt~U) ana ~~u) Irom ~11.45) '"·we can,

however, consider the limiting situation of a glancing collision in which 8' is very
small. Then in the frame K, ud will be nonrelativistic and uc will differ only slightly
from ua. We can therefore make appropriate Taylor series expansions around
8' = 0 and obtain equations involving M(u), 't;(u), and perhaps their first deriva-
tives. Explicitly, the x component of the momentum conservation equation in
(11.43) is
M c f3 sin 8' M c f3 sin 8'
0 = (uc) y(1 + {3 cos 8') -
2 (ud) y(1 - {3 2 COS 8').
canceung common !actors and rearrangmg terms, we nave

l + {3" cos (}'
'"
JVL~ Uc)
1 Q2 ~~e il'
JVL~'" Ud)'
r- I
T\,. ,] ... ; ,];,! •• .11 lJ'

_,.
r
,]. fr il' n+
r
·~~ ~>f11
\'
A'l\
'I
shows that in that limit = ua, ud = 0. Thus we obtain
c {32)
Uc
M(ua) = +
_ {3 2 M(O) (11.44)
1
From (11.41) it is easy to demonstrate that

1 + {32 1
1 - {32
--
= Ya (11.45)
,/ ~
I Ua
2
2
~
nuu Lu~:o vcuu~:o JvL~vl '" uum ~ll.Y:I) we tnus nave
M(ua) Yam
or equivalently that the momentum of a particle of mass m and velocity u is
mu
p = ymu = ---,=====;; (11.46)
2
/1 -
\I c
Determining the functional form of )?;(u) requires more than the straightfor-
ward evaluation of the conservation of energy equation at e' = o+. We must
examine the equation for small e'. From ( 11.43) we have
'£(,. \ .J. '£((\\ gz(, \ .J. '£(, \
"C ~u~ ~d ~" ~ v1 u . 1 1vu1 ~ 11. "'"J '" ~' '·-''" J ' " ' uuu, correct to
order e' 2
inclusive.
wnere Ya IS given oy ( 1 !.'+.> l and 71 c"/3 e "/( 1 /3 'l IS a convenient exmm~icm

parameter. We now expand both sides of (11.47) in Taylor series and equate
coefficients of different powers of 7):
)?;(ua) + )?;(0) = )?;(ua) + 7) . (d)?;(~c) . au~'

due a'l') '-.,~o
'I
The zeroth-order terms give an identity, but the first-order terms yield
1
__ _!_ d)?;(ua) d)?;(ud) J
0 - 3 -'· 2 + -1. 2
ra a \ d ud 0
" " " wv .. uv"" uvu1'-'la"VLMLc vawe u1 llle seconu term 1rom ~1L.J":1J, we nna
d)?;(ua) m m
= - Ya•0 = -------,-;o
du2a 2 2\ 'l/2
lla
21-::2
111 yic1us tue expression,
(11.48)
ror me energy or_ a _IJ_ar_!!C'l_e ot mass m and ve oc1ty u, up to an ar_llitrarv constant

nf in ' P~rPnth, '·~11 .. mn ~n~ rl, th , t ' ,], • ' ' ·tho
vauvu VL ~" CUHUlllun can ue expresseu Ill terms Ot K.!lleL!C energle~

Sect. 11.5 Relativistic Momentum and Energy of a Particle
alone. Thus the undetermined constant in (11.48) is necessary and is not, as the
reader might have conjectured, the result of our Taylor series expansions. Note
that the kinetic energy T(u) is given unambiguously by
T(u) = c'(u) - '(O) ~ me'[ (l _[~) m - l] (11.49)
Equations (11.46) and (11.48) are the necessary relativistic generalizations

for the momentum and ener of a article consistent with the conservation laws
Decay processes are particularly transparent. Consider, for example, the decay
of a neutral K-meson into two photons, K0 --> yy. In the rest frame of the
K-meson, conservation of energy requires that the sum of the energies of the two
photons be equal to '{gK(O). For another decay mode of a neutral K-meson, into
two pions, the kinetic energy of each pion in the K-meson's rest frame must be
T, = ~'{gK(O) - '{g,(O)
Measurement of the pion kinetic energy (11.49) and knowledge of '€K(O) allows
determination of '{g,(O). In these examples and every other case it is found that
. . . .
mc 2
E = ymc 2 = ----r==';O (11.51)
/ z/
Y1 - c2
A second path to the results (11.50) and (11.51) is theoretical. Although the
expressions (11.46) and (11.48) for the momentum and energy of a particle were
found by applying the principles of special relativity to the conservation of energy
(11.36), that is p = mU. The time component of this 4-vector is p 0 = mU0 =

myuc. Comparison with (11.48) shows that the energy of a particle differs from
cp 0 by an additive constant ['{g(O) - mc2 ]. This means that the four equations of
.-;;'lSI r~ ,,, 11 .~n _rc_
--
energy and momentum conservation for an arbitrary collision process can be
written as
2: a
(Po)a - L b
(Po)b = flo
· · "ol fi, o I
(11 ""'
2.. Pa ~ Pb- a \ ->~)
a b
initial final
where 1Ll.n. a 11s a 4-vector w1tn a - ll and

cflo =
b
L [11:b(O) - mbc 2 ] -
2:a
[11:a(O) - mac 2 ]
final initial
From the first postulate, (11.52) must be valid in all equivalent inertial frames.
But if a - 0 in all inertial frames it can be seen from (11.22) that it is necessary
=
that Lln o· the 4-vector (!ln. a) is a null vector. If different tvoes or numbers nf
particles can occur in the initial and final states of some process, the condition
Ll 0 = 0 can only be met by requiring that (11.50) hold for each particle separately.
WP ~rP thn< ]p.-J tn (11 51)~' thP fnrm nf thP tntol
' u•.
Jlto VI
l ty Ul lllto [JalllCIC Call Uto ~
-r Ill oi its
momentum and energy from (11.46) and (11.51) as
r 2n
u = _.£ (11.5~)
1'.-
The invariant "length" of the energy-momentum 4-vector (p 0 = E!c, p) is

Po-P·
2 p-
- ( me )2 (11.54)
We see that the invariant property that characterizes a particle's momentum and
energy is its mass, m, sometimes caueu its rest mass. equation~ lL.Y+ ), comoinea
with the conservation equations, forms a powerful and elegant means of treating
relativistic kinematics in collision and decay processes (see the problems at the
end of the chanter). Note that (11.54) nermits the ener>'v E to be exnressed in
••••
E = Yc2p 2 + m 2 c4 (11.55)
UlCO J1 • " " ~lL'"tU), ~lL-11), dllU ~:ll •.JJ) lUl LLUlll, COllCl~y, dllU vc;-
locity of a particle are so universally accepted that it seems superfluous to speak
of experimental tests. It is perhaps worthwhile, nevertheless, to cite some !abo-
ratorv demonstratiOns. One is the connection between the kinetic energy ( 11.4'1)
•• t'
. ,1 cl •
• ''_!'
cl
•
t 'T'.•~ '-
--'·
~•
.c .l.
~uo
.c
~• nw • u
1.'
<HH~H~ ~ 'b'
from 0.5 to 15 MeV (accelerated through a known voltage in a Van de Graaff
;JO;, ,-1 ~t thP hP~m -L hv r~ lnrimPtnr) orP JrP.-1 hv h~vinu
u
' .n
J J J
UUISLS Ul ~Ll.l .J A 1\J SJ traVel a lllgllt paw lJI 0.'+ meters. 1-\.S UlC:
energy increases the transit time falls toward a limiting value of 2.8 X 10- 8 s, m
good agreement with (11.49). Verification of cas a limiting speed for material
*Some authors define the mass of a particle to be E/c 2 , designating it as m or m(u) and reserving the
symbol m 0 for the rest mass. We always use the word "mass" for the Lorentz invariant quantity whose
square appears in ( 11.54) .
'W ,, . Am 1 Ph'" "\2 'i'il !IQI>4\
Se..t 11 6 Mathematkal Properties of the Space-Time of Special Relativity 539
particles has been carried out for 11 GeV electrons ( y ~ 2 X 107) in the Stanford
experiment cited at the end of Section 11.2, where it was found that the electrons'
speed differed fractionally from e by less than 5 X 10- 6 . An undergraduate ex-
penment to venfy the relatiOn (I 1.55) between momentum and energy employs
a simple magnet with roughly 10 em radius of curvature for the momentum mea-
surement and aNal crystal for the energy measurement on beta rays.*
The specification of the kinematic properties of a particle (velocity, momen-
tum, energy) in any inertial frame can be accomplished by giving its mass and
either its velocity u or its momentum p in that frame. A Lorentz transformatiOn
(11.22) of (p 0 , p) gives the results in any other frame. It is sometimes convenient
to use the two components of p perpendicular to the z axis and a rapidity {;
(11.20) as kinematic variables. Suppose that a particle has momentum pin frame
K, with transverse momentum pj_ and a z component p 11 • There is a unique Lorentz
transformation in the z direction to a frame K' where the particle has no z com-
ponent of momentum. In K' the particle has momentum and energy,
(11.56)
Let the rapidity parameter associated with the Lmentz transformation fiom K
to K' be { Then from the inverse of (11.21) the momentum components and
energy of the particle in the original frame K can be written
E
p 11 = !l sinh {;, - = n cosh{; (11.57)
c.
with n = V p~ + m 2 The quantity !l/e is sometimes called the transverse mass
(because it depends on pJJ or the longitudinal mass (because it is involved in a
longitudinal boost). If the particle is at rest in K', that is, p, = 0, then the ex-
pressions (11.57) become
p = me sinh {;, E- me cosh? (11.58)
alternatives to (11.46) and (11.51).
One convenience of pyJ and ?(i) as kinematic variables is that a Lorentz
transformation in the z direction shifts all rapidities by a constant amount, ?(i) ~
~(i) Z, where Z is the rapidity parameter of the transformation. \Vith these
variables, the configuration of particles in a collision process viewed in the lab-
oratory frame differs only by a trivial shift of the origin of rapidity from the same
process viewed in the center of mass frame.
11.6 Mathematical Properties of the Space-Time

of Special Relativity
The kinematics of special relativity presented in the preceding sections can be
discussed in a more profound and elegant manner that simultaneously simplifies
and illuminates the theory. Three-dimensional rotations in classical and quantum
mechanics can be discussed in terms of the gronp of transformations of the co
*S Parker Am I Phys 40 241 (1972)

540 Chapter 11 Special Theory of Relativity G
ordinates that leave the norm of the vector x invariant. In the special theory of
relativity, Lorentz transformations of the four-dimensional coordinates (.Jio, x)
follow from the invariance of
We can therefore rephrase the kmemat1cs of spec1al relatiVIty as the considera-

tion of the group of all transformations that leave s 2 invariant. Technically, this
group is called the homogeneous Lorentz group. It contains ordinary rotations
as well as the Lorentz transformations of Section 11.3. The group of transfor-
mations that leave invariant
c f
is called the inhomogeneous Lorentz group or the Poincare group. It contains
translations and reflections in both space and time, as well as the transformations
of the homogeneous Lorentz group. We shall restrict our discussion to the ho-
mogeneous transformations and subsequently omit "homogeneous" when refer-
ring to the Lorentz group.
From the first postulate 1t follows that the mathematical equations expressing
the laws of nature must be covariant, that is, invariant in form, under the tt ans-
formations of the Lorentz group. They must therefore be relations among
Lorentz scalars, 4-vectors, 4-tensors, etc., defined by their transformation prop-
erties under the Lorentz group in ways analogous to the familiar specification of
tensors of a given rank under three-dimensional rotations. We are thus led to
consider briefly the mathematical stmctm e of a space-time whose norm is defined
We begin by summarizing the elements of tensor analysis in a non-Euclidean

vector space. The space-time continuum is defined in terms of a four-dimensional
space with coordinates x'\ x \ x 2 , x 3 . We suppose that there is a well-defined
transformation that yields new coordinates x' 0 , x'\ x' 2 , x' 3 , according to some
rule
via ID'( 0 1 2 3\
"' x \x,x,x,.x1 (a 0, 1, 2, 3) (11.60)
For the moment the transformation law is not specified.
Tensors of rank k associated with the space-time point x are defined by their
transformation properties under the transformation x -->- x'. A scalar (tensor of
rank zero) is a single quantity whose value is not changed by the transformation.
The interval s (11 59) is obviously a I orentz scalar For tensors of rank one,
called -vectms, tl'vo kinds must be distinguished. The fhst is called a contravariant
vector A" with four components A 0 , A\ A 2 , A 3 that are transformed according
to the rule
'"'
In this equation the derivative is computed from (11.60) and the repeated index
f3 1mphes a summatiOn over f3 0, 1, 2, 3. Ihus exphc1tly we have
Sect. 11.6 Mathematical Properties of the Space-Time of Special Relativity 541
We will henceforth employ this summation convention for repeated indices. A

covariant vector or tensor of rank one B, is defined by the mle
{3
B' - - B (11.62)
ax'a {3
a_
or, explicitly, by
The partial derivative in (11.62) is to be calculated from the inverse of (11.60)

with x 13 expressed as a function of x' 0 , xd, x' 2, x.J.
Note that contravariant vectors have superscripts and covariant vectors have
subscripts, corresponding to the presence of ax'"/ax 13 and its inverse in the rule
of transformation It can he verified from (11 61) that if the law of transformation
(11.60) is linear then the coordinates x 0 , x\ x 2 , x 3 form the components of a
contravariant vector.
A contravariant tensor of rank two F" 13 consists of 16 quantities that trans-
form according to
(11.63)
A covariant tensor of rank two, G "13 , transforms as

8
- axY ax G
G~f3 -
ax'" ax' f3
y
" (11.64)
and the mixed second-rank tensor H" 13 transforms as
(11.65)
The generalization to contravariant, covariant, or mixed tensors of arbitrary rank

shoUld be obv10us from these examples.
The inner or scalar product of two vectors is defined as the product of the
components of a covariant and a contravariant vector,
B·A=BA" (11.66)
"'
With this definition the scalar product is an invariant or scalar under the trans-
formation (11.60). This is established by considering the scalar product B' ·A'
and employing (11.61) and (11.62):
ax 13 ax"' ax 13
B'. A ' = - - BAY=-
{3
10'y
BAY=
{3
)'
o13 y BAY=
{3
B. A
The inner product or contraction with respect to any pair of indices, either on
the same tensor or one on one tensor and the other on another, is defined in
analogy with ( 11.66). One index is contravariant and the other covariant always.
The results or definitions above are general The specific geometry of the
space-time of special relativity is defined by the invariant interval s 2 , (11.59). In
~ Chapter 11 Special Theory of Relativity-G
differential form, the infinitesimal interval ds that defines the norm of our space
IS
•
2
(ds) - (dx 0 ) 2 - (dx 1 ) 2 - (dx- 2 12 - (rlr 3 1' { 11 o>'}
""''
'
Here we have used superscripts on the coordinates because of our present con-
v · . 1ui~ nuffn w metric is a speciat case or me general ainerennat length
element,
(ds) 2 = gap dxa dx 13 (11.68)
'
'Jl"'" /5af3 ·~ "~
/5f3a ,
" ..
' u•co ,
• '=~-ume or specral
relativity (in distinction to the curved space-time of general relativity) the metric
eom'<w ·, rl;n~~nnl ..,;.~ ol
'
goo = 1' g,, = g22 = g33 = -1 (11.69)
The contravariant metric tensor gaf3 is defined as the normalized cofactor of gaf3·
Prw flot -, .t;n, ;t ;< tlw oomp·
g"f3 = gu{3 (11.70)

Note that the contraction of the contravariant and covariant metric tensors gives
the Kronecker delta in four dimensions·
g ay gyf3 = 8a{3 (11.71)
where 8"13 = 0 for a* {3 and 8"a = 1 for a= 0, 1, 2, 3.
Comoarison of the invariant len2:th element ( ds1 2 in (11.681 with thP ~imibrlu
invariant scalar product (11.66) suggests that the covariant coordinate 4-vector
xa can be obtained from the contravariant xf3 bv contraction with f! .. o. that is
Xa = gapX(3 (11.72)
and its inverse,
a a{3 111 ..,.,,
·p \ . .>I
"
In fact, contraction with g" 13 or gaf3 is the procedure for changing an index on any
tensor from bemg contravariant to covariant, and vice versa. Thus
TO'· .a .. a(3 p ... .
0 ..p
and (11.74)
13
.. . a. =g af3 G ....
G··
vv iLu ute meuic Lensor _~ tt.OYJ iL wuows tnat it a contravanant 4-vector

h;c,~ 1ents A 0 A 1 A 2 A 3 its .;ont nortnPr hoc
·r
1Pnt< A
0 1 '7 • 1
r1 'r1] - .n '.nz -.t-1 '£'13 - .n . vv co vv lllC LillS as
A (A , A), A" lA, -Al (11./))
where the 3-vector A has comnonents A 1 A 2 A 3 ThP '""lor nr.,rhwt (11 flfll of
A
n·A DaA n·A -n-·"A

Sect. 11.7 Matrix Representation of Lorentz Transformations, Infinitesimal Generators 543
Consider now the partial derivative operators •Nith respect to x" and x, The
transformation properties of these operators can be established directly by using
the rules of implicit differentiation. For example, we have
f3
Comparison with (11.62) shows that differentiation with respect to a contravariant

component of the coordinate vector transforms as the component of a covariant
vectm opeuum. From (11.72) it follows that differentiation with respeet to a
covariant component gives a contravariant vector operator. We therefore employ
iJ"
a = ( axa
= ax" 0 • - v) (11.76)
The 4 divergence of a 4 -vector A is the invariant,
(11.77)
an equation familiar in form from continuity of charge and current density, the
Lorentz condition on the scalar and vector potentials, etc. These examples give
a first inkling of how the covariance of a physical law emerges provided suitable
Lorentz transformation properties are attributed to the quantities entering the
The four-dimensional Laplacian operator is defined to be the invariant
(11.78)
This is, of course, just the operator of the wave equation in vacuum.
11.7 Matrix Representation o.[Lorentz Transformations,

Infinitesimal Generators
We now turn to the consideration of the Lorentz group of transformations. To
make the mampulat10ns exphc1t and less abstract, If Is convement to use a matnx
representation with the components of a contravariant 4-vector forming the el-
1 2 3
f I 11 d. a I d fi d'
(11.79)
544 Chanter 11 Suecial Theorv of Relativitv- G
_,
' ' a''-' ' ~
-1.
U, , ,L -~ '
"" "c"'~' r ' UJ ~• W "'-

"''-' UOULU nay U,j summmg
"VvvCV<O ~U,
over the products of the elements of a and b, or equivalently by matrix multip!i-

cation of the transpose of a on b:
(a, b) = ab (11.80)
'T'~
oap
_1
"!·
•<
'-[ -~
_A _A
... .. .,
~
It 0 0 o\
0 -1 0 0
g UL81)
0 0 -1 0
0 0 0 -1;
with rf = I, the 4 X 4 unit matrix. The covariant coordinate vector is
x"\
-x'
gx =
-xz (1 1.82)
-x3
obtained by matrix multiplication of g (1 Ull) onx (11.79). Note that in the present
nuuuiun uu: 'cutar proaucr ~-ll.DD) ojrwo LJ-vecwrs reaas
a·b = (a, Rb) = ( Ra, b) = fiRb (11.83\
On the basis of arguments already presented in Section 11.3 we seek a group
of linear transformations on the coordinates,
x' =Ax (11.84)
.d ; o ~ --"-"-" '-" _L[____:z_ A ,' . ~1+~-+•< {. ,'\. ' _j_,£, '
HUHC,
' V 'IY'/
.'i'gx' = igx (1L85)

" -~ {1
~•
0 A; ' •'
\ «.<n J ' " " ' UHO HOH·uauu MUv J
' c. ' ' '
• " ' " v'{Uaucy,
iAgAx = igx
Since this must hold for all coordinate vectors x, A must satisfy the matrix
eauation
AgA = g ( 11.86)
r. _,.
Clf tho tr~~e<" .d ~ ,], ,],
"'
ately from (i Ul6 ). The first concerns the determinant of A. Taking the deter-
minant of both sides of (11 R(';\ aiw•' n·
det (AgA) = det g (det A) 2 = det g
<;:;n,--p clPt a -1 =1=0
0
detA = ~1
There are two classes of transformations: proper Lorentz transformations, con-

tinuous with the identity transformation and so necessarily having det A = + 1,
>~nn " _L 1f moFnr~nfinno H'nr i,_,.,nrnnor tron-C. · • ;, '• enffi.
-" · ·r - r • •
cient, but not necessary, to have det A = -1. The fact that det A = ~ 1 does not
unambi!lUOUslY__sm:t_o_ut_thte twn "'"'""' ;, " rnnstemltenrte nf~. J -" :.
of space-time. Two examples of improper transformations, A = g (space inver-

sian) with det A = -1 and A = -I.( soace and time inversion}_ with det A = + 1
;nnot· ,, th ·
r
. ,,
Sect 11 7 Matrix Representation of I.orentz Transformations. Infinitesimal Generators 545
The second property of A is the number of parameters needed to specify

completely a transformation of the group. Since A and g are 4 X 4 matrices,
(11 86) represents 16 equations for the 42 - 16 elements of A. But they are not
all independent because of symmetry under transposition. There are thus
16 (I+ 2 + 3) 10 hnearly mdependent equations for the 16 elements of A.
This means that there are six free parameters-the Lorentz group is a six-param-
eter group. The six parameters can be conveniently thought of as (a) three pa-
rameters (e.g., Euler angles) to specify the relative orientation of the coordinate
axes and (b) three parameters (e.g., components of ~) to specify the relative
velocity of the two inertial frames. Parenthetically we remark that for every
six-parameter A giving a proper Lorentz transformation, there is an improper
one represented by -A. From now on we consider only proper Lorentz
transformatwns.
The explicit construction of A can proceed as follows We make the ansatz
L
where L is a 4 X 4 matrix The determinant of A is*
If L is a real matrix, det A 1 is excluded. Furthermore, if L is traceless, then

det A = + 1. Thus, for proper Lorentz transformations, L is a real, traceless
4 X 4 matrix. Equation (11.86) can be written
gAg- A (11.88)
From the definition ( 11.87) and the fact that i = I we have
A-t = e-L
Therefore (11.88) is equivalent to

gLg = -L
gL gL
The matrix gL is thus antisymmetric. From the properties of g (11.81) it is evident
that the general form of L is
. I
\T : L:;O 1 --r:oz L03
-,----.--- -~-------;----- i--
L --
01 .' -,_
(11.90)
n
T
.''' -T. !. .
Lm ' -Lu -Lz3 0
The dashed lines are inserted to set off the 3 x 3 antisymmetric spatial matrix
corresponding to the familiar rotations in a fixed inertial frame from the sym-
metric space-time part of the matrix correspondmg to Lorentz transformations
or beests from one inertial frame to another.
*To prove this, note first that the value of the determinant or the trace of a matrix is unchanged by
a similarity transformation. Then make such a transformation to put L in diagonal form. The matrix
A wiH then be diag0nal with elements that afe the eXJlonentials of the eorresponding elements of L.
The result follows immediately.
1 ne matrix ~ 1 J.'JU ), w1tn Its s1x parameters 1s an explicit constructiOn [through

(11.87)] of the transformation matrix A. It is customary, however, to systema.tize
L and its six parameters by introducing a set of six fundamental matrices defined
by
lo
..• 0 0 0 o: 0 0 o'
---~----------------
• ---~-----------------
n n n n n:.
.• . " " 1
" 0 •
•
•
0 0 -1 "" o: •
0 0 0
,o •
• 0 1 0 o: -1 0 0
'o •
• 0 0 0
n .•
•
---~-----------------
n -1 n
s,- • (11 01\
u • I u () "/
,o• .. 0 0 0
o :..
1___ __________
1 0 0 0: 0
•
---~-----------
1 o\ 0:0 0 1
---1-·---------
1:
.• o:. o:•
KI= ,..., K7 = • ,..., K, = • ,...
v ' v u
U•
• ..
1 • U•
•
\v'• I \v' \' . I
The matrices S, evidently generate rotations in three dimensions, while the rna-
trices K, produce boosts. For reference, we note that the squares of these six
matrices are all diagonal and of the form,
lo 0 0 0 \
n 1
~i ~?
"i \}
0 -1 0 -1
lo 0 I
52- --1
3-
-1
n
\ v "I 111.92)
Kj{ 0
I
0
0
J ~~(' 0
0
1
0
J.
/1 (\ \
TT2 u
'"3 0
0 1
Furthermore, it can be shown that (E · S) 3 = -E ·Sand (E' · K) 3 = E' · K, where
F :m.-1 F' :1rP. :~nv rP.:11 nnit i . Thn~ :1nv · .-.f nn<> .-.r th<> · : r:1n
,
Uc; CA}'l CO>CU dO d · Vl lHC HldlliA IJJ llO oquaTe".
The l!:eneral result 111.901 for L can now be written alternativelv as
and
L ~ -~ · S- C• K} (11.93)
A = e-w-s-<;-K
Sect. 11.7 Matrix Representation of l~orentz Transformations, Infinitesimal Generators 547
wh.er<e w ~ncl r ~r<e ccon•t:mt 1- · Th.e thr.e<e 11, """h of ,,, 'mel r
correspond to the six parameters of the transformation. To establish contact with
earlier results such as (11.16) or (11.21), we consider first a simple situation in
which w = 0 and ~ = {;E 1 . Then L = -!;K1 and with the help of (11.92) and
_,
rq rq W<O llllU
A- e -v K1) K1 smn 1, + K 1 cosn!; (11.Y4)

]::;
-n
I ,\
<-V"l b """ b v v
-- -sinh!; cosh!; 0 0
A (11.95)
0 0 1 0
\ 0 0 0 1;
This mRtrix 1ds .e· •'· to th.e lRtion (11_? 1\* Tf [ 0 "ncl
w = wE 3 , the transformation is similarly found to be
I u u u
n N'l< ''' <in n
A- (11.Yb)
0 -smw cos w 0
0 0 0 1
,,j.' ,r '~ ,,, ,]. ·L
~
t.
"'' "
the 3-axis.
For a boost (without rotation) in an arbitrary direction,
A -(·K
1 ne noost vector ~ can De wntten m terms or tne relative ve1oc1ty li as
~ = ptanh- 1
f3
where f3 is_,_ a unit vector in. the direction of the relative velocity of the two inertial
H UHlvO. U l v }'UH.- UVVO< 10 lll<Oll
Ah lB) = e-il·K tanh-

1
f3 (11.97)
Tt . J.,ft t ;f, th ,, th' H
-J ·o· '
y -yf3, -yf32 -y/3:;

1 + (y-1)f3i ( y - 1)/3113> (y-1)/3,/33
-"R
A (R\-
w w .o
w
-yf3z
_\-_!' 'JPl/J:z
1 + ~ !' ~JP2 _\-_!' 1 JPzP3
[J.2 [J.2 [J.2
r r
\' i'/33
( y - 1)f3tf33
,z
( y - 1)13>/3:; 1 + ( y - 1)/3~ J
1-' 1-' 1-' '
(11 QR\
*The reader is reminded that in Sections 11.3, 11.4. and 11.5 no distinction is made between subscripts
and superscripts. All components of vectors there are to be interpreted as contravariant components,
;n •orrnrchnre ..,;,, Ill 7<;\
The equation x' = Aboost(l3)x is a matrix statement of the four equations of

(11.19).
The six matrices (11.91) are a representation of the infinitesimal generators
of the Lorentz group. Straightfor wa1 d calculation shows that they satisfy the
following commutation relations,
[5,, K 1] EijkKk ( 11.99)

[K,, KJl = -Eijl/)k
=
where the commutator notation is [A, B] AB - BA. The first relation corre-
sponds to the commutation relations for angular momentum, the second relation
merely shows that K transforms as a vector under rotations, and the final relation
shows that boosts do not in general commute. The commutation relations (11.99),
with the characteristic minus sign in the last commutator, specify the algebraic
structure of the Lorentz group to be SL(2, C) or 0(3, 1).
11.8 Thomas Precesswn

The description of Lorentz transformations in terms of noncommuting matrices
demonstrates that in general the result of successive Lorentz transformations
depends on the order in which they are performed. The commutation relations
(11.99) imply that two successive Lorentz transformations are equivalent to a
single Lorentz transformation plus a three-dimensional rotation. An example of
the kmematic consequences of the noncommutativity of Lorentz transformations
is the phenomenon known as Thoma8 ptecession.* To motivate the discussion
we first describe the physical context.
In 1926 Uhlenbeck and Goudsmit introduced the idea of electron spin and
showed that, if the electron had a g factor of 2, the anomalous Zeeman effect
could be explained, as well as the existence of multiplet splittings. There was a
difficulty, however, in that the observed fine structure intervals were only half
the theoretically expected values. If a g factor of unity were chosen, the fine
structure intervals were given correctly, but the Zeeman effect was then the
normal one. The complete explanation of spin, including correctly the g factor
and the proper tine structure interaction, came only with the relativistic electron
theory of Dirac. But within the framework of an empirical spin angular momen-
tum and a g factor of 2, Thomas showed in 1927 that the origin of the discrepancy
was a relativistic kinematic effect which, when included properly, gave both the
anomalous Zeeman effect and the correct fine structure splittings. The Thomas
precession, as it is called, also gives a qualitative explanation for a spin-orbit
mteractJon m atomic nuclei and shows why the doublets are "mverted" m nuclei.
The Uhlenbeck Goudsmit hypothesis was that an electron possesses a spin
angular momentum s (which can take on quantized values of ±h/2 along any
axis) and a magnetic moment I' related to s by
fL=--s (11.100)
2mc
*L. H. Thomas, Phil. Mag. 3, 1 (1927).

Seet. 11.8 Thomas Preeession 549
where the g factor has the value g 2. Suppose that an electron moves with a
velocity v in external fields E and B. Then the equation of motion for its angular
momentum in its rest frame is
f.t X B' (11.101)
where B' is the magnetic induction in that frame. We will show in Section 11.10
that in a coordinate system moving with the electron the magnetic induction is
B' (11.102)
where we have neglected terms of the order of (z?/c 2 ) Then (11.101) becomes
(-ds)
direst frame
=r..tX (B--xE v
C
) (11.103)
Equation (11.103) is equivalent to an energy of interaction of the electron spin:
U' = -1-l · ( B- ~ x E) (11.104)
In an atom the electric force eE can be approximated as the negative gradient

of a spherically symmetric average potential energy V(r). For one-electron atoms
this is, of course, exact. Thus
eE = - - - ( 11.105)
r dr
Then the spin-interaction energy can be written
~ g 1dV
0' s·B + (s · L) (11.106)
where L rn(r x v) is the electron's orbital angular momentum. This interaction

energy gives the anomalous Zeeman effect correctly, but has a spin-orbit inter-
action that is twice too large
The error in (11.106) can be traced to the incorrectness of (11.101) as an
equation of motion for the electron spin. The left-hand side of (11.101) gives the
rate of change of spin m the rest frame of the electron. If, as I homas first pomted
out, that coordinate system rotates, then the total time rate of change of the spin,
or more generally, any vector G is given by the well known result,*
(dG) ± l!lT X G (11107)
where wr is the angular velocity of rotation found by Thomas. When applied to

the electron spin, (11.107) gives an equation of motion:
~ds) (geB' (11.108)
*Sec, for example, Goldstein (pp. 174-177).

The corresponding energy of interaction is

u= U' + s. WT (11.109)
where U' is the electromagnetic spin interaction (11.104) or (11.106).
The origin of the Thomas precessional frequency wr is the acceleration ex-
perienced by the electron as it moves under the action of external forces. Con-
sider an electron moving with velocity v(t) with respect to a laboratory inertial
frame. The electron's rest frame of coordinates is defined as a co-moving se-
quence of inertial frames whose successive origins move at each instant with the
velocity of the electron. Let the velocity of the rest frame with respect to the
laboratory at laboratory time 1 be v(z) cf3, and at laboratory time t + 8t be
v(t + &) = c(J3 + 8J3). The connection between the coordinates in the electron's
rest frame at time t and the coordinates in the laboratory frame is
X (11.110)
At time t + i5t the connection is
x" (11.111)
It is important to note that these transformations of co or din ate from the !abo-
ratory to the rest frame are defined here in terms of pure Lorentz boosts without
rotations. 'Ne are interest<!d in the behavior of the coordinate axes of the elec-
tron's rest frame as a function of time. Thus we want the connection between
the two sets of rest-frame coordinates, x' at time t and x" at time t + 8t. This
relation is
x" =Ax'
where
For purposes of calculating Ar a suitable choice of axes in the laboratory frame

is shown in Fig. 11.7. The velocity vector 13 at time t is parallel to the 1 axis and
the increment of velocity 813 lies in the 1-2 plane. From (11.98) it follows that
(t 0 0 1
(I 1.113)
p+O(J 7'
/op
/ ,.
Figure 1L7
Sect. 11.8 Thomas Precession 551
Similarly we obtain from (11.98), keeping only first-order terms in 8jl,

I 3o s:.o _{, ,Q -'- 3 >:.o \ >:.o n\
I I ~ -r-i \1~ -,-'I I -:L
y- 1
- ( yf3 + y 3 8{31) y + l/3 8{31 ( 0 8{32 0
~ I
Aboo,,(jl + 8jl) = ,\
( y- 1
-y 8{32 D ) 8{32 1 0
/'-'
0 0 0 1
"
\ll.U_.,,
Straightforward matrix multiplication according to (11.112) yields

I 1 -y2 8{31 -y 8{32 Ol
0
I y - 1\
Y P1 0 I
{3
upz \1
AT= I )I - 1
(11.115)
YOPz I op2 ~ u
{3
0 0 0 1
This represents an infinitesimal Lorentz transformation that can be written in
tPrms nf thP. mMrirP.s S ~n.-1 K ~'
I 1\
AT I ~ y- )(jl X 8jl) • S ( y 2 813 11 + y 813 _~_) • K (11.116)
fJ
where 813 and 813-'- are the components of 813 parallel and perpendicular to jl,
11
respecuvety. 10 nrsr oruer mop, ~~LllOJ 1s eqmvatemw

Ar = Aboost(~jl)R(Ml) = R(~!l)Ahoo,(8jl) (11.117)
where
1 n\ 1 A 0 Tf
''hoost\~P/ ' ~p ·~
R(~!l) =I - ~!l ·S
are commuting infinitesimal boosts and rotations, with velocity,
~13 = y 2 813,, + y 813_~_
and angle of rotation,
y ~
~n = {32
Ill X 813 = y 13 X 813
y +1
Thus the pure Lorentz boost (11.111) to the frame with velocity c(jl + 8jl) is
equivalent to a boost (11.11 0) to a frame moving with velocity cjl, followed by
au
- LUlt:Hl£ ll >Llll!', uta Wllll V' c up _J
and a rotation ~n.

Tn nf thP · 1tinn nf thP ·w fr~mPs ~s : rPst fr~mPs
of the electron we do not want rotations as well as boosts. Nonrelativistic equa-

tions of motion like (11.101) can be expected to hold provided the evolution of
the rest frame is described by infinitesimal boosts without rotations. We are thus
led to consider the rest-frame coordinates at timet + 8t that are given from those
at time t by the boost A 6 aost(~l3) instead of A 1 . Denoting these coordinates by

x"', we have
x"' = Aboost(~(3)x'
Osmg (11.117), (11.112), and (11.110) we can express x"' in terms of the labora-
x"' = R( -Ml)Aboos1((3 + 8(3)x (11.118)

The rest system of coordinates defined by x"' is rotated by -~n relative to the
boosted laboratory axes (x"). If a physical vector G has a (proper) time rate of
change (dGidT) in the rest frame, the precession of the rest-frame axes with
respect to the laboratory makes the vector have a total time rate of change with
respect to the laboratory axes of (11.107), with
~n
lim (11.119)
Bt~o 8t 'Y + 1 c2
where a is the acceleration in the laboratory frame and, to be precise,
(dG/dt)restframe = 'Y- 1 (dG/dT)restframe•
The Thomas precession is purely kinematical in origin. If a component of
accelei ation exists peipendicula• to v, for whatever reason, then there Is a
Thomas precession, independent of other effects such as precession of the mag-
netic moment in a magnetic field.
For electrons in atoms the acceleration is caused by the screened Coulomb
field (11.1 05). Thus the Thomas angular velocity is
(11.120)
It is evident from (11.109) and (11.106) that the extra contribution to the energy
fnnn the Thomas precession reduces the spin orbit coupling, yielding
(11.121)
With g = 2 the spin-orbit interaction of (11.106) is reduced by~ (sometimes called

the 1homas factor), as required for the correct spin-orbit interaction energy of
an atomic electron.
In atomic nuclei the nucleons experience strong accelerations because of the
specifically nuclear forces. The electromagnetic forces are comparatively weak.
In an approximate way one can treat the nucleons as moving separately in a
short range, spherically symmetric, attractive, polential well, V:v(r) Then each
nucleon will experience in addition a spin-orbit interaction given by (11.109) with
the neghg1ble electromagnetic contribution U' omitted:
where the acceleration in wr is determined by V,(r). The form of w 1 is the same

as (11.120) with V replaced by VN. Thus the nuclear spin-orbit interaction is
a roximatel
(11.123)
Sect. 11.9 Invariance of Electric Charl!e' Covariance of Electrodvnamics 553
L 111 1'">0\ '+L . r .1 {111"11\ •t. -L +L T/ .-1 T/

'"~~"-'" ·o\_' · ·') \ · 'J N
are attractive (although V:"' is much larger), so that the signs of the spin-orbit
' ' "J'J-,' ~ ·~ •o <1-.~< ;n nu~1~; <J-.~ o;n~l~ ' ' louole f~rm
"in~:rted" d~~blets. With a reasonable form for V:"'' (i'~j .12.3) is in qualitative
mith thP r.l- ~ min.nrhit <nlittinr« in nnr-lPi *
0 0
T h e p h enomenon of Thomas
r
precessiOn
'.
1s
'
presented rom a more sop h'1st!-
' [
cated ooint of view in Section 11.11 where the BMT eauation is discussed.
11.9 lnvariance of Electric Charge;

r. . nf r.>f · J_
1irfii
-.,
--
Thte inv,ri:lnr.te in form of the enuations of electrodvnamics under Lorentz trans-
formations was shown by Lorentz and Poincare before the formulation of the
special theory of relativity. This in variance of form or covariance of the Maxwell
and Lorentz force equations implies that the various quantities p, J, E, B that
enter these equations transform in well-defined ways under Lorentz transfor-
mations. Then the terms of the equations can have consistent behavior under
Lorentz transformations.
Consider first the Lorentz force equation for a particle of charge q,
I \
rl~
::!:. = q( E + -" X B (11.124)
dt \
c
We know that p transforms as the space part of the 4-vector of energy and
momentum,
pa = (p 0 , p) = m(Uo, U)
where Po - E!c and ua 1s the 4-veloclty (11.36). lt we use the proper time T
,.
/11 "){;\ .
'J
.•. .-1 ,4' 4', ~•u. ,,; '
. /11
\.
1 "lA\
., 1-. '
dp
dT
q
C \' 'U'
ITT V T T " D\
., {11
\
1 "'"'
')
~ ~- . . •'- £ ~'- ~:
~'- 1 • • '-
L Hv LvH'HUHU OLUv LO W v 0!-'Uvv l-'UH \H U ' vv<VL. L Hv v~U T 0
ponent equation is the rate of change of energy of the particle (6.110):
dpo = q_ U. E (11.126)
dT C
If the force and energy change equa!Jons are to be Lorentz covanant, the
, .1-.
·o·
1-. .-1 , .-1, >- >1-.
-.-
.4' A •'- 'T'l.
'J
, • 1.
!'
--'· ...
or tnree racwrs, me cnarge q, me '1-vewcny, ana me eiecuomagnetic neJUs. u
tJ-. <• ,<-' ' • ,f j ·n ~f tJ-.~ <1-.roo >- ~ro lrnmun ~nrl T
t' 't'
covar • llltoll lilt: LllldllUll lHUptolllto' Ul lllt: lllllU ldUUI

" is heeL
c:'ln he est8 hi
_, r. '- ~- •'- . '-
LH.- ~ . 10 ' ' 'vu, ao lW ao Wv • • UL , Wv
ma2.nitudes of the chames of elementarv oarticles (and therefore of anv svstem
*See, for example, Section. 2.4c of A. Bohr and B. R. Mottclson, Nuclear Structure, Vol. 1, W. A.
Benjamin, New York (1969).
of charges) are mtegral multiples of the charge of the proton. In the published
literature,* it is experimentally established that the fractional difference betwe~n
the magnitude of the electron's charge and the proton's charge is less than 10--19
and unpublished results of King push this limit almost two orders of magnitude'
further.t The results of these experiments can be used to suppmt the invariance
of electnc charge under Lorentz transformatiOns or, more concretely, the inde-
pendence of the observed charge of a particle on its speed. In his experiments
King searched for a residual charge remaining in a container as hydrogen or
helium gas is allowed to escape No effect was obsen•ed and a limit of less than
19
10 e was established fm the net charge per molecule for both H 2 and He. Smce
the electrons in He move at speeds twice as fast as in H 2 , the charge of the
electron cannot depend significantly on its speed, at least for speeds of the order
of (O.Ol-0.02)c. In the experiment of Fraser, Carlson, and Hughes an atomic
beam apparatus was used in an attempt to observe electrostatic deflection of
beams of "neutral" cesium and potassium atoms. Again, no effect was observed,
and a limit of less than 3.5 X 10~ 19 was set on the fractional difference between
the charges of the proton and electron. Cesium and potassium have Z = 55 and
19, respectively Thus the K-shell electrons in cesium at least move with speeds
of order 0.4c. The observed neutrality of the cesium atom at the level of
10~ 18 -10~ 19 is strong evidence for the invariance of electric charge.'
The experimental invariance of electric charge and the requirement of
Lorentz covariance of the Lorentz force equation (11.125) and ( 11.126) deter-
mines the Lorentz transformation properties of the electromagnetic field. For
example, the requirement from (11.126) that U • E be the time component of a
4-vector establishes that the components of E are the time-space parts of a second
rank tensor F" 6 , that is, E · U F 06 U 13 • Although the explicit fmm of the field
strength tensor F" 13 can be found along these lines, we now proceed to examine
the Maxwell equations themselves.
For simplicity, we consider the microscopic Maxwell equations, without D
and H We begin with the charge density p(x, t) and current density J(x, t) and
the continuity equation
ap
-+V·J=O (11.127)
az
From the discussion at the end of Section 11.6 and especially (11.77) it is natural
to postulate that p and J together form a 4-vector !":
- c
*J. G. King, Phys. Rev. Lett. 5, 562 (1960); V. W. Hughes, L. J. Fraser, and E. R. Carlson, Z Phys.
D-Atoms, Molecules and Clusters 10, 145 (1988). The latter tabulates many of the different methods
1
The limits on the measured charge per molecule in units of the electronic charge for H 2, He, and
SF6 were determined as 1.8 :+: 5.4, -0.7 :+: 4.7, 0 :+: 4.3, respectively, all times 10- 21 • Private com-
munication from J. G. King (1975).
*Mentioning only the eleetrons is somewhat misleading. The protons and neutrons inside nnelei move
with speeds of the 01 de1 (0.2 0.3)c. Thus the helium Iesults of King already test the in variance of
charge at appreciable speeds. Of course, if one is content with invariance at the level of 10- 10 for
vic ~ 10- 3 the observed electrical neutrality of bulk matter when heated or cooled will suffice.
Sect. 11.9 Invariance of Electric Charge; Covariance of Electrodynamics 555
Then the continuity equation (11.127) takes the obviously covariant form,
a
"
r = o (11.129)
where the covariant differential operator a" is given by (11.76). That]" is a le-
gitimate 4-vector follows from the invariance of electric charge: Consider a large
number of elementary charges totaling oq at rest* in a small-volume element d 3 x
in frame K. They are idealized by a charge density p. The total charge 8q =
p d 3x within the small-volume element is an experimental invariant; it is thus
true that p' d 3 x' = p d 3 x. But the four-dimensional volume element d 4 x = dx 0 d 3 x
is a Lorentz invariant:
a( !0 d r2 r3)
d4 X ' -- X ' X ' X ' X d4 X -- d et A d4 X -- d4 X
0 1 2 3)
a(x , X , X , X
The equality p' d 3 x' = p d 3 x then implies that cp transforms like x 0 , namely, the
time component of the 4-vector (11.128).
In the Lorenz family of gauges the wave equations for the vector potential
A and the scalar potential are
1 aLA
2 ~ 2
~?·
v ,....
4'7T
. T
(11.130)
1 a"
-t-V•A u ~ll.l.:ll)
Tl- ~''<£~. . '1

'1'
+,
" u'
11 1 1 '11\\
,~
'I
. . ~,
•• - ,]
II ~ J l. I 0), W 1111 C L11C 11!!,11L -11d11U MUC> dl C LllC Vl a - vc-.Lvl.
Obviously, Lorentz covanance reqmres that the potentials , A) (11.132)
Then the wave equations and the Lorenz condition take on the manifestly co-
variant forms,
4
OA" = '7T l"
c
and (11.133)
aaA" = 0
The fields E and B are expressed in terms of the potentials as
1aA
E = - - - - V
c i:Jt (11.134)
B=VxA
*If there is a conduction current J as well as the charge density pin K, the total charge within d3 x is
not an invariant. See M¢/ler, Section 7.5. (His argument assumes the 4-vector character of cp and J,
however.)
The x components of E and B are explicitly

~
1 riA rit th ~1. ' ,rl ' fi,]rl, ' ~- ••• ' ,]] •'-
·r-J ' •
' -~ ""'
elements of a second-rank, antisymmetric field-strength tensor,
r;afJ - ri"' A {3 - ;JfJ A "' (11 l':lt;\
.. ,
Explicitly, the field-strength tensor is, in matrix form,
0 -E -E -E
pafJ = Ex 0 -B z By
(11.137)
Ev Bz 0 -Bx
v TJ TJ (1
\ z y X I
For reference, we record the field-strength tensor with two covariant indices,
I 0 Ex Ey Ez
-E 0 -B z By
F,.,/3 = g'".,Fyaga/3 = X
(11.138)
-E B 0 -B
-Ez -By B, 0
The elements of Faa are obtained from F"'fJ by putting E ~ -E. Another useful
quantity is the dual field-strength tensor '?fi"'P. We first define the totally antisym-
metric fourth-rank tensor E"'f3Y8:
+1 for a = 0. B = 1 'V = 2. o = 3. and

eafly8 = any even permutation
(11.139)
-1 for anv odd oermutation
(1 ;., t . ,-l; ,]
'J "1'
Note that the non vanishing elements all have one time and three (different) space
inrlirP~ ~nrl th~t " ..,afly8 Th P t, .,afly8 io ~ n • J •<nr nnrlPr on~ti~l
inversions. This c;;~ be seen by contracting it with f~ur different 4-vectors and
examinin2. the snace in of the mt rotationallv invariant
""''-
,_
d.
'iuuuucy.
J. - ' " - u
uuu.
"'~ u~•u-ow " uy
"
lo -B -B. -B
TJ 0 vz .[;'
y
'?Ji"'P - "Eapyop
2
8- X
(11.140)
y
By -E z 0 Ex
\BZ Ey -Ex 0 I
UlC t;It; ~
Ul LliC UUai Lt:hov• ;y "" ale UUUl r
.. Q
uy ,.. v n '7 v
and B ~ - E in (11.137). This is a special case of the duality transformation
fL 1 <01\
\ Voio>' J•
10 complete tne demonstratiOn ot tne covanance ot electrodynamics we
Sect. 11.9 lnvanance of Electnc Charge; Covariance of Electrodynamics 557
must write the Maxwell equations themselves in an explicitly covariant form. The
inhomogeneous equations are
V • E = 41Tp
In terms of F'"'3 and the 4-current J" these take on the covariant form
41T
a paf3
~
= - Jf3 (11.141)
c
Similarly, the homogeneous Maxwell equations
V. R- 0 , VxE+---0
c at
can be written in terms of the dual field-strength tensor as

a"?1'"13 = o (11.142)
In terms of F"13 , rather than ?Ji"f3, these homogeneous equations are the four
equations
(11.143)
where a, {3, yare any three of the integers 0, 1, 2, 3.
With the definitions of]" (11.128), A" (11.132), and F"13 (11.136), together
with the wave equations (11.133) or the Maxwell equations (11.141) and (11.142),
the covariance of the equations of electromagnetism is established. To complete
the discussion, we put the Lorentz force and rate of change of energy equations
(11.125) and (11.126) in manifestly covariant form,
- = m - - - - F"f3U (11.144)
dT dT C I'
The covariant description of the conservation laws of a combined system of elec-

tromagnetic fields and charged particles and a covariant solution for the fields of
a moving charge are deferred to Chapter 12, where a Lagrangian formulation is
presented.
For the macroscopic Maxwell equations it is necessary to distinguish two
field-strength tensors, E"f3 - (F, B) and G"f3 - (D H), where F" 13 is given by
(11.137) and G" 8 is obtained from (11.137) by substituting E > D and B ~ H.
I he co van ant form of the Maxwell equations is then
1T 13
a" G"13 =- 1 , (11.145)
c
It is clear that with the fields (E, B) and (D, H) transforming as antisymmetric
second-rank tensors the polarization P and the negative magnetization - M form
a similar second-rank tensor. With these quantities given meaning as macroscopic
averages of atomic properties in the rest frame of the medium, the electrodynam-
ics of macroscopic matter in motion is specified This is the basis of the electro
dynamics of Minkowski and othe1s. Fm fm the1 infonnation on this I ather large

and Important subJect, the reader can consult the literature cited at the end of
the chapter. "
11.10 Transformation of Electromagnetic Fields

Since the fields E and B me the elements of a second rank tens01 F'fl, their values
in one inertial frame K' can be expressed in terms of the values in another inertial
frame K according to
(11.146)
In the matrix notation of Section 11.7 this can be written

F' = AFA (11.147)
where F and F' are 4 X 4 matrices (11.137) and A is the Lorentz transformation
matrix of (11.93). For the specific Lorentz transformation (11.95), corresponding
to a boost along the x 1 axis with speed cf3 from the unprimed frame to the primed
frame, the explicit equations of transformation are
' -- y(E2 - {3B3)

E2 B~ = y(B2 + {3£3) (11.148)
' -- y(E3 + f3Bz)
E3 B~ = y(B3 - {3E 2 )
Here and below, the subscripts 1, 2, 3 indicate ordinary Cartesian spatial com-
ponents and are not covariant indices. The inverse of (11.148) is found, as usual,
by mterchangmg pnmed and unpnmed quantities and puthng B ~ -B. For a
general Lorentz transformation from K to a system K' moving with velocity v
relative to K, the transformation of the fields can be written
2
E' = y(E + l3 X B) - y:
1
13(13 • E)
2
B' gt(B ll X E) y + 1 ll(ll . B)
These are the analogs for the fields of (11.19) for the coordinates. Transformation
(11.149) shows that E and B have no independent existence. A purely electric or
magnetic field in one coordinate system will appear as a mixtme of elec1ric and
magnetic fields in another co01dinate f1ame. Of comse ce1tain 1estl ictions apply
(see Problem 11.14) so that, for example, a purely electrostatic field in one co-
ordinate system cannot be transformed into a purely magnetostatic field in an-
other. But the fields are completely interrelated, and one should properly speak
of the electromagnetic field pafl, rather than E or B separately.
If no magnetic field exists in a certain frame K', as for example with one or
more point charges at rest in K', the inverse of ( 11 149) shows that in the frame
K the magnetic field B and electric field E me linked by the simple I elation
8 f3XE (11.150)
Sect. 11.10 Transformation of Electromagnetic Fields 559
Note that E is not the electrostatic field in K', but that field transformed from
K' to K.
As an important and illuminating example of the transformation of fields,
we consider the fields seen by an observer in the system K when a point charge
q moves by in a straight-line path with a velocity v. The charge is at rest in the
system K', and the transformation of the fields is given by the inverse of (11.148)
/11 1 1C\\ HI <L <L -L ' <L ;,· '' ' -'
~ ~"l-'l-'~0~ " ' " ' '"~ ~""'5~ w
~0 'l
~' \ i ioiT/}•
0
·"' •
lllaL ILS CiosesL ulSLance 01 approacu 10 we ouserver IS u. r 1gure 1 1.o ~••uw~ a

suitably chosen set of axes. The observer is at the point P. At t = t' - 0 the
rwi oin < nf t h P hun -'' l~tP ·• · .J. ~nrl thP ,-h~roP n io ~t ito nl. .o. ••
_,. T. •L ·oc_ , , ·on
"'-'-'
.V
<L
<H~
•L
~~0~, ~,, '" <U~ HU<H~ H
T/1 d ..
"·~ ~
.L
'l-' , ' <L
are to De evamat~a, nas cooramates x 1 vt ,x 2 o, x 3 - u, ana is a aistance

2
r' = Vb + (ut')L awav from a. We will need to exnress r' in terms of the co-
nrrlin~tP< in 1<, ThP nnlu -'' l~tP -'in o • ,f. •~tinn io thP timP t'
-o o.
c 7> ' '' r TT T '' r
Yl' -~UI'-' )-'I] yt, Mlll.C A( V lVI LllC jJUliiL 1 111 Lll<:O HdlllCO 1'L Hl LllCO lCO" U<llllCO
K or tne cnarge tne eJectnc ana magnenc ne1as at me ooservanon pomt are
r.r qvt r.r qu )71 {)

~1 ~:2 OJ
r'3 ' r!3 '
B; = 0, B~ = 0, B"J = 0
In terms of the coordinates of K the nonvanishing field components are
E' = _ q-yvt E' qb (11.151)

• ( bz + -yzvztz)312 • 2 - ( bz + iv2t2)3!2
Then, using the inverse of (11.148), we find the transformed fields in the sys-
tern K:
q-yvt
EJ=Ei=-(bz zzz)"12
+ -yvt J
' -yqb (11.152)

Ez = -y£2 = (bz + iv2tz?;2
B3 = -y{3E~ = {3E2
with the other components vanishing.
X2 X2'
/
/
b
~~ vt
I
/q
v
X!
X!'
_/
., I
X3
J?;nnrP 11 !I D. · .. 1. nf ,-horoP n mnu;no ot >nt -' v on
': D .t . I.
" ' '

!'\bU Chapter 11 Special Theory of Relativity-G
Fields (11.152) exhibit interesting behavior when the velocity of the charge
approaches that of light. First of all there is observed a mag_netic in duel ion in th
x 3 direction already displayed in (11.150). This magnetic field becomes alm~;t
equal to the transverse electric field E 2 as {3--+ 1. Even at nonrelativistic veloriti"'
where y = 1, this magnetic induction is equivalent to
n" V r
B-2--~-
c r'
whtch ts JUSt tne approximate Ampere-Bwt-Savart expresswn for the magnetic
field of a moving charge. At high speeds when y >> 1 we see that the peak
" " " ' •v• ov ctcCtric uelU 1.:- 2 ~l u) oecomes equa1 to y nmes ns non relativistic
value. In the same limit, however, the duration of appreciable field strengths at
•• n• ~ . . .
'"'"' p v u a ' lO . J-\c "''-'UOUl'-' Vl LllC UlllC HH'-'l •m VVCI Wu!Cu me nelds
are appreciable is evidently

b
11t = - (11.153)
'}'V
As y increases, the peak fields increase in proportion, but their duration e.oes in
inverse proportion. The time integral of the fields times v is independent of ve-
locity. Figure 11.9a shows this behavior of the transverse electric and magnetic
fields and the longitudinal electric field. For {3--+ 1 the observer at P sees nearly
equal transverse and mutually perpendicular electric and magnetic fields. These
are mmsnngmsnao1e uom the helds ot a pulse ot plane polarized radiation prop-
agating in the x 1 direction. The extra longitudinal electric field varies rapidly from
pvsiti vc cv ucgative anu uas zero ume mtegraJ. n me ooserver s aetectmg ap-
paratus has any significant inertia, it will not respond to this longitudinal field.
~ .
-fU'-''"'Y •v• 1-'~ . put, • uc wu1 see umy we cransverse uews. lllis
equivalence of the fields of a relativistic charged particle and those of a pulse of
_, ~
. ..: . ., '
VVHl UC T
' . .
111 -r LJ. >H l ll..l(") llle
fields for {3 = 1 are given an explicit realization.
ThP liPid··~ .rl th~ ,j' p; 11 0. Lb ... -
deuce of the field~ ;t a fixed observation p~int. An alt~rnative descriptio~'"can

hP PiVen in ofthP :n~ti~J u~ri~tirm ,fthc> fip]rJo rc>J~tiuc> t" thP indnn '"
present position of the charge in the laboratory. From (11.152) we see that
E.!Eo vtlb. Reference to FiP 11 S -•- : thM thte teltertrir fitelrl i' th '" ~
along n, a unit radial vector from the charge's present position to the observation
point, iust as for a static Coulomb field. Bv exoressine. the denominator in
(11.152) in terms of r, the radial distance from the present position to the ob-
server, and the angle IjJ = cos- 1 (n · v) shown in Fig. 11.8, we obtain the electric
field in terms of the charge's present position:
nr
E (11.154)
r3y2(l _ {32 sinzl/J)3/2
1 ne magneuc mauctwn 1s g1ven by (11.150). The electnc field is radial, but the
lines of force are isotropically distributed only for {3 = 0. Along the direction of
· ~ l.fl • u, 7rJ, Lne 1ie1U strength is down by a tactor ot y relative to Isotropy,
while in the transverse directions ( IjJ = 1r/2) it is larger by a factor of y. This
~· ~ . . . .
'""""v•vvuo p~«vHl Vl " " lH IVII...C, >IlVWll llll'l~. I L7U, IS llle spatial snap
shot" equivalent of the temporal behavior sketched in Fig. 11.9a. The compres-
Sect. 11.11 Relativistic Equation of Motion for Spin in External Fields
-rr !.-----
'J27 b2 ___ _ vt ______,._
---
Figure 11.9 Fields of a uniformly moving charged particle. (a) Fields at the
of the charge.
sion of the lines of force in the transverse direction can be viewed as a conse-
quence of the FitzGerald-Lorentz contraction.
or Slowly Varying External Fields
discussed in Section 11.8 on Thomas precession. Here we exploit the ideas of
what is known as the BMT equation of motion for the spin.* With the magnetic
V. L. Telegdi, Phys. Rev. Lett. 2, 435 (1959). The equation actually has much earlier origins; Thomas
. . .
by (11.100), the rest frame equation of motion for the spin,
ds ge
- = - - s x B' ( 11.155)
dt' 2mc
where primes denote quantities defined in the rest frame and s is the spin in that
frame This equation applies to a particle of mass m, charge e, spin s and a
magnetic dipole moment with Lande g factm of g. It is a classical equation, but
is the same as the quantum-mechanical Heisenberg equation of motion for the
spin operator or, equivalently, the equation of motiofl for the polarization ~ector
of the system.
A. Covariant Equation of Motion

To obtain a relativistic generalizatiofl of (11.155) it is first t1eeessary to gen-
eralize the spin s from a 3-vector in the particle's rest frame. There are two
avenues open. One is to recall from the end of Section 11.9 that P and - M form
an antisymmetric second rank tensor. This suggests that JL, hence s, may be gen-
eralized to a second-rank tensor S"f3. A simpler alternative is to define an axial
4 vectm S" in such a manner that it has only three independent components and
reduces to the spm s m the particle's rest frame.* If S" denotes the components
of the spin 4-vector in the inertial frame K, the time-component in the rest frame
K' ts, accordmg to ( I 1.22),
where U" is the particle's 4-velocity. We see that the vanishing of the time-
component in the rest frame is imposed by the covariant constraint,
(11.156)
In an inertial frame where the particle's velocity is cl3 the time component of
spin is therefore not independent, but is
It is useful to display the explicit connection between S" and the rest-frame spin
s. Use of (11.19) or (11.22) and (11.157) yields
s s y (13 . S)l3 (11 158)

+ 1
and the inverse expressions
S = s + _Y_ (13 · s)l3

y + 1
Specification of the rest-frame 3-vector spins determines the components of the

4-vector spm S"' m any inertial frame.
*The spm 4-vector S" is the dual of the tensor S"/3 in the sense that S"- (112c)•"f3Y8 U 8 S,8 , where U"
is the particle's 4-velocity.
Sect. 11.11 Relativistic Equation of Motion for Spin in External Fields 563
The obvious generalization of the left-hand side of (11.155) is dS"/dr, where

Tis the narticle's moner time. The ri2.ht-hand side must therefore be exmessible
HClUS r
.
as a 4-vector. We assume that the equation is linear in the spinS" and the external
.
. l l lOdH dlSU IIIVUIVC U
. . . .
dHU UU IUT, lllt: ldlLCl uemg lll£
.. llSCll.
Higher time derivatives are assumed absent. And of course the equation must
reduce to (11.155) in the rest frame. With the building blocks S", F" 13 , U", dU"/dr
;I .~, .< 1: ('a ;I vaf3 , A
,
... · ·
'J
• +I,
I ~ 1 Tf3'
F" 13 S f3' (S AFA'"U1L )U" ' s
13
- )U"
dr
Other possibilities, such as F" 13 U,lSAUA), (UJA'"UJS", and (SAFA'"U) dU"!dr,
either vanish, are higher order in F" 13 , or reduce to multiples of the three above.
The equation of motion must therefore be of the form
13
A (SAFA'"U )U" + ___]_
-dS" = A 1F" 13 S 13 + __2 A ( S -dU ) U" (11.160)
2 2 13
dr c '" c dr
where A1o A 2 , A 3 are constants. The constramt equation (ll.Db) must hold at
~ 11 t;m,., Th;, rPmJ;r,.,
•
d dTI d.'\"
- (UaS")- S " - + U"-- 0
dr dr dr
hence
13
I
V'l
A \ TT
"'2/~a•
naBP
Uf3 ,.
t• A
'"'u 13
\(' dU
dr
" (
\·" .,
"'"'
If nonelectromagnetic or field gradient forces are allowed, at least in principle,

it is necessary that A 1 = A 2 and A 3 = -1. Reduction to the rest frame and
comparison with (11.155) gives A 1 = ge/2mc. Thus (11.160) becomes
-dS" = - ge [ F" 13S

13
+ -12 U"(SAF A'"U ) - -12 U" ( sA-
d UA) J (11.162)
dr 2mc c '" c dr
Tf thP p]po orrn,ot;~ fiPlrlc ~rP un;frorm ;n cn~~P ror ;f rrr~rl;Pnt fnr~P ];],.,
·r cr
V(r-t ·B), (5.69), can be neglected, and there are no other appreciable forces on
. .
me panicle, ns transJanonaJ monon IS aescnoea oy UL144):
au - _!___ F"~'U (11.163)

;/,.,. - mr· {3
'ThP (11 1{;')\ '- thP RMT omonr;nM'

-, T
/a \
AC::" a 1
I£
0
-=- '?. p•f3S 13 + -2 - 1 I U"(S,FA'"U,.) (11.164)

dr me 2 c 2
B. Connection to the Thomas Precession

The covariant equation (11.162), or its special case (11.164), contain the
Thomas precession of the spin. It occurs in the final term in (11.162), the term
LH<tl WitS Sr
.. uy LUe lt:l[Ull• ~LL.LJU) llldl Lilt: sp111 '+ ue Ullllug-
~£,4 ~'-
" • ~L
•J
.£ .. ~
·~
"'J
onal to the 4-velocity. To exhibit the Thomas precession explicitly, we consider

the equation of motion for the rest-frame spins. Using the result
•
dU' dv
S , - = -yS · - (11.165)
dr dr
and (11.158) for sin terms of S, we find that the equations,
dS 2 (
-=F+yp S ·dp)
-
dr dr
~nd
""::.'!
Jn
- K -+ -v 2 f
"' . '!
dr dT
can be combined to give, after some simplification,

2
ds = F - yp F0 + ')' [s X ( [3 X dp)] (11.166)
dr y + 1 y + 1 dr
In these equations (F0 , F) stand for the time and space components of the terms
with coefficient (ge/2mc) in (11.162). Since (F0 , F) form a 4-vector, with F;, =
p • F, the first two terms in (11.166) can be recognized as the torque F' evaluated
in the rest frame. Dividing both sides by y and using the definition (11.119) for
the Thomas precession frequency, we find that (11.166) becomes
ds 1
- =- F' + Wr X S 111.1671
uc ')'
:-.mce 1' 1s g1ven ov me ngnt-nana s1ae or ( 11. )) \. th1s 1s mst I 11.10/l ot Sect10n
11 Q
. .
rUI " 111 eleCLromagueLIC uelUS w11ere ( 1 1.1 O.:l J nows,
dp e
- = - [E + p X B - PCP . E)] (11.168)
dt ymc
We also have, from the transformation properties (11.149) of B,
-1 F , = - ge s x [ B - y (p · B)p - p x E ] (11.169)
')' 2mc y + 1
When these expressions are inserted into (11.167), it becomes
ds e {g 1 lg \
1'
-:; 0 A
', ~
r
/u
\~
, I
y
~
~
\P • DJP
H '~A\
~~~.~ 'V)
I ~ Tliv~
2 " + 1 t•
Th;o fnrm nf thP Pm,~t;nn
-, nf mnt;nn nf t~ ,, . 'T'l,
' , , · on
-, "1.
•~" ·~~~
'
~··.~L~)
r
V' ~7L' ~uy. Ul.J,
C. Rate of Change of Longitudinal Polarization

As an example of the use of (11.170) we consider the rate of change of the
component of spin s parallel to the velocity. This is the longitudinal polarization
Sect. 11.12 Note on Notation and Units in Relativistic Kinematics 565
or net helicity of the particle. If is a unit vector in the direction of 13, the p
longitudinal polarization is p·
s. It changes in time because s changes and also 13
chan~>:es. bxolicJtiV. we have
d • • ds 1 • • dl3
- (13 • s) = 13 • - +
dt dt
B[s - (13 • s)l3] · -
dt
Using (11.168) and (11.170), this can be written, after some algebra, as
I \ I f3 \ -
d ' e g ' g 1
rlt (13 • s) = - mr
- s . I. ? - 1 113 X B + ? - R E
.L
(11.171)
where s.L is the component of s perpendicular to the velocity.

Equation (11.171) demonstrates a remarkable property of a particle with
ff ? - Tn o moffnot;~ holrl tho c~;~ ;n CH~h o thot tho
r ·r r
-~
. .
-.oT pOtaJ raT , wnarever me monon or me parnc1e.
If the particle is relativistic (/3 ~ 1), even the presence of an electric field causes
the longitudinal polarization to change only very slowly, at a rate proportional
to o;-L times the electric field comnonent nernendicular to v.
The electron and the muon have g factors differing from the Dirac value of
2 by radiative corrections of order a/7T = 0.00232. Because (g - 2) is so small,
'1. .1 • ..]." '1 . • _, 1. _, -1 - " ,. .
·o t' ·o •s
nericnercrmanges relatively slowly. ThiS phenomenon permits very prec1se mea-
surements of the quantity a (g - 2)/2, called the anomaly or the anomalous =
magnetic moment. The values of a provide accurate tests of the validity of quan-
tHm • ' ·~· ' 0 H'nr mHnnc 1ooo;_ lnn>T;hul;nollu · ·' ot h;rth- tho
·o r
change in polarization is detected by means of the characteristically asymmetric
angular distribution of the decay electron from the muon relative to the direction
of muon polarization. For electrons from beta decay the initial longitudinal po-
lanzatwn IS + R. Its chan<'e w1th time IS detected bv chan<'es in the asvmmetrv
of Mott scattering (e-) or the angular distribution of the annihilation photons
from positronium formed in an intense magnetic field (e+). The precision attain-
. .1..1 1. +L • + -L . • . ..1" -• ..1 L
J]
+L ' ' ..1. ·*
"7 ""1
a( e-) = 1 159 652 188.4 (4.3) X 10- 12

a(e+) = 1159 652187.9 (4.3) X 10 -12
nf Cc\ 1 1.-;;<: 0"1A {0\ v 10 9
"" '
These results are in good agreement with the predictions of quantum electro-
dynamics, as discussed in detail in the review by Kinoshita.
,-,, ... . . _,
"P
.
t'
. . ,_,,
= = u <v <Hv
... t''
-'-' ••
· U< <Hv vHU V0
.•. £
"
Lnapter lL.
-,-,--,-, 7\J,.-;;; "" .. 7\J"" • , .. n ..~ ; .. D - l .-.• : ....:ro v: .. o ... nt-:,. ..
In dealing with Lorentz transformations and relativistic kinematics, it is conve-

-n±e-nTTo1IOopTa consistenT, simpre nmaLion ana set or units. we nave seen that
*e-, e+: VanDyck, Schwinberg, and Dehmelt, Phys. Rev. Lett. 59, 26 (1987); p,±: J. Bailey et al.,
Nucl. Phys. B 150, 1 (1979). See also the review, T. Kinoshita, ed., Quantum Electrodvnamics. World
Scientific Sinoanore7i9<)()\.
various nowers of the velocitv of li~ht c annear in the formulas of sneci~l rPlot;
ity. These tend. to make the formulas cumbersome, although their presence {a-
c1htates extractmg nonrelatlv1stlc hm1ts (by lettmg c ---> oo ). In domg relativistic
· >otir< it io r>
' . 'J
tA <
Tr
•c• ,]] ,f · h
. 'J
cuitohl. •h · ,,.
...,.,,
vve auopt lile convenllon tnat au momenta, energies, anu masses are measured
in energy units, while velocities are measured in units of the velocity of light. In
particle kinematics the symbols,
p cp
E E
m stand for mc2
v -
c
'T't.. ,,_ _, •1 -?
• "~0 ... ~ 'HVH V~' o ~~H >HUH1 UHU <V•U>
'" '""~" UO L
p + m , a particle s veloCity IS v - pll:, and so on. As energy umts, the electron

volt (eV), the megaelectron volt (1 MeV = 106 eV), and the gigaelectron volt
(1 GeV = 109 eV) are convenient. One electron volt is the energy gained by a
nortirlP mith P · rhor<TP whPn it folic thrnn<Th o iol 'r" Af
~olt (1 eV = 1.602 X 10- erg= 1.602 X 10- roule).
12 19
In addition to eliminating powers of c, it is customary to denote scalar prod-
v
-~ 1.
"v•o vy u
.l _,
uv•
1. . ." ' .l • t.. .1. '•'- -'-
l-'' vu-
u~•• ·~ ~u v~• ·~~u
"J ' '""
ucts ot J-vectors denoted by a dot between boldtace symbols, as usual. Thus we
have
a · b - aaba = a0 b 0 - a•b
Four-vectors may be written with or without an index. Thus conservation of

energy and momentum may appear as
P-p+q
or
pa a a
'1
'
n .r. rrPro: nnrl c-. -' flpnrlino

-., 00 ·o
The theory of relativity has an extensive literature all its own. Perhaps the most lucid,
though concise, presentation of special and general relativity is the famous 1921 encyclo-
peuia anic1e 1orougm up w ume in 1~:10 J oy
Pauli
Another authoritative book is
MIZ)ller
Among the older textbooks on special relativity at the graduate level are
Ahomni
Bergmann, Chapters I-IX
and, more recently,
A C:hent~" 11 enrl 7
D.
.
Schwartz
,..,_
. .,. 1 _, ~
Tonnelat, Part II
Ch.ll References 567
"'" ., ,L
'] ... ~
"
~'
']• .,, . ' ·•vu~~~ ·~ oo ~v ·~u
c ..
•ov•u u
physicist's point of vie;;.:"in

Anderson, Chapters l· -)
Bergmann, Chapter V
Schwartz, Chapter 5
The flavor of the original theoretical developments can be obtained by consulting the
iointlv collected papers of
Einstein, Lorentz, Minkowski, and Weyl
and in the collection of reprints of
Kilmister
A very individualistic discussion of special relativity, with extensive references and
perceptive commentary, can be found in the two books by
Arzelies
The main experiments are summarized in
Ml')ller, Chapters 1 and 12
Condon and Odishaw, Part 6, Chapter 8 by E. L. Hill
The barely mentioned topic of the electrodynamics of moving media is treated by
Cull wick
Ml')ller, pp. 209 f£.
Pauli, pp. 99 ff.
Penfield and Haus
,--, •• 1' . ~1- " ' D c• " , h , ,].. . 1-.' r'l-. •y
5 by reading his prefac~ to the second editi~n.
Relativistic kinematics has been relegated to the problems. Useful references in this
Baldin, Gol'danskii, and Rozenthal

ttageaorn
Sard, Chapter 4
Another neglected subject is the appearance of rapidly moving objects. This fasci~
nating topic illustrates how careful one must be with concepts such as the FitzGerald-
Lorentz contraction. Some oertinent oaoers are
R. Penrose, Proc. Cambridge Philos. Soc. 55, 137 (1959).
1 • "'· 11 DL D "" 1fH1 t1om
: 'J . • , \ '}'
V. F. We1sskopf, Phys. Today 13: 9, 24 (1960).

!Yl. L. twas, Am. J. rnys. ""· L15J (l'Jtll}.
G. D. Scott and H. J. VanDriel, Am. J. Phys. 38, 971 (1970).
For the reader who wishes a more elementary or leisurely introduction to special
relattvtty, there are a host ol undergraduate texts, among them
Bohm
Fevnman, Vol. 1, Chapters 15-17
n. •
Mermin
"'mU1er
Sard
Smith
Tevlor enrl IPr
Sard's book is specially recommended. It is at an intermediate level and has a thorough
discussion of Thomas precession and the motion of spin. Hagedorn also discusses the
~uu ~'iuauuu w u~cau.
"'· • '· ,. ., A • ncr n .T
Special Relativity SRT-1, A;;,. J. Phys. 30, 462 (1962).

Problems
•.
11.1 Two equivalent inertial frames K and K' are such that K' moves in the positive x
direction with speed v as seen from K The spatial coordinate axes in K' are
parallel to those in K and the two origins are coincident at times t = t' = 0 .
.'
\dJ
no
",L
J]!Y
'' .£
'J! - ~'i u" atcuce 01
_,
rntterent mert1a1 trames (hrst postulate ot relativity) reqmre that the m0 , 1

general transformation between the space-time coordinates (x, y, z, t) and
(x', y', z', t') is the linear transformation,
x' = f(v 2)x - vf(v 2)t; t' = g(v 2 )t - vh(v 2 )x; y' = y; z' =z
and the inverse,
x - f(u')x + uf(u ')I ; t - l!(U ')t + uh(v')x · v- v'· z- 7'
where f, g, and h are functions of v2, the structures of the x' and x equations
are determined by the definition of the inertial frames in relative motion,
and the signs in the inverse equation are a reflection of the reversal of roles
of the two frames.
(b) :-.now tnat consistency ot tne JhltJal transrormauon ana ns mverse reqmre
t=l! and {
2
- v2 fh - 1
(c) If a physical entity has speed u' parallel to the x' axis in K', show that its
speed u parallel to the x axis in K is
u' + v
u=
1 + nu'(h/f)
T To;n~ tho
2
~notnlot
r
"Jo {,
\' ·-' F ·". ··~ >arl r \
-r
0 h th "' ~
f/C is required and that the Lorentz transformation of the coordinates re-
suits. The universal limiting speed C is to be determined from experiment.
11.2 Consider three inertial frames and coordinates K(x, t), K'(x', t'), and K"(x", t").
Frame K' moves in the x direction with speed u 1 relative to K; frame K" moves
,,;th "~QQrl " ralot;.,Q tn V' n ,,.l ·~ ,,.l ' V ll ' ' • tl
-r·
"T
:' , " -r
~ H.l \: Lilt; I t;SUllS Ul paiLS a
and b), (x , t") -> (x , r) ___,. (x, t) and (x , t") ___,. (x, t) directly, show that
I h(u 2 )/f(u 2) I is a universal constant with the dimensions of an inverse speed
squared.
This approach obtains the Lorentz transformation without reference to elec-
tromagnetism or the second postulate, but requires experiment to show that
hlf > 0.
T • V P 'T'Prlah-Ir;; wno ;~ tlon Thonr> nF D ' . PrP«
--r<C ~-"TT==:T.
11.3 Show explicitly that two successive Lorentz transformations in the same direction
are equivalent to a single Lorentz transformation with a velocity
v1 + u2
u 1 2\
"""L I
_, _,
~·· -" •y
oL
·]!
" - -' ' '
'
11.4 A possible clock is shown in the figure. It consists of a flash tube F and a photocell
P shielded so that each views only the mirror M, located a distance d away, and
mounted rigidly with respect to the ftashtube-photocell assembly. The electronic
Ch. 11 Problems 569
innards of the box are such that when the photocell responds to a light flash from
the mirror, tne nasntube 1s with a ne»i' ble delay and emits a short flash
toward the mirror. The clock thus "ticks" once every (2dlc) seconds when at rest.
lu
p li'
-'-VI~
Problem 11.4
(a) Suppose that the clock moves with a uniform velocity v, perpendicular to the
,. c. n,-, u -' '' • • ·
•, " ·~ cv . . . . ~vo~• •~• ••s LHC '"'"UJIU pu"u'"'" ut
relativity, show by explicit geometrical or algebraic construction that the ob-
server sees the relativistic time dilatation as the clock moves by.
(b) Suppose that the clock moves with a velocity v parallel to the line from PF
LV JYL t;ULJ LLLdL u n c , LUU, LLLt; '"'U'"L\." UU>CL •cu I0 IICK more ' <J DY Ine
same time dilatation factor.
11.5 A coordinate system K' moves with a velocity v relative to another system K. In
K' a particle has a velocity u' and an acceleration a'. Find the Lorentz transfor-
mation law for accelerations, and show that in the system K the components of
acceleration -" and orli~nlor to v are
2 3/2
( v
\' -;:>! '
3 11
V•U a
1 + - -' )
(
2-
a,= 3 c!l~ + ·2 x (a' X u') )

v · u' c
1 + -- )
( c2
'f ·~ ~ua" m '"~ J "-LVV. \JUt; UL d ot;L Ul L'¥LH>
born in 2080 remains on earth; the other rides in the rocket. The rocket ship is so
that It has an acceleratiOn g m its own rest frame (this ,-,ok-P, the
'"uuScLu~Lt:u
~~~nnantS feel at h< Tie), Jt ateS in a -JinP noth fnr, <; Vf'or' (hv it,
----,
erates for 5 years, decelerates for 5 years, and lands on earth. The twin in the
.0 ' <A
0 V~ -~ LO ~V J~a<O VLU,
(n) WhM '"'"r i' it nn Porth?
(b) How far away from the earth did the rocket ship travel?
•~ ·~•] ~ •~my u.uo~u~u "!'~up d
d apart on they axis for a race parallel to the x axis. Two starters, one beside each
mon mill hrP thPir ctortinn ot clinhth, 1t timM niuinn o ho ~:

"!" LO
the better of the two runners. The time difference in K is T.
(a) For what range of time differences will there be a reference frame in K'
which there is no handicap, and for what range of time differences is there
a name K" m wmcn tnere 1s a true (not apparent) nanOJcap·!
'L' r-. " -" "<1. ot. T .£. " ot. £. V'
'I " '1-'1'' vpnate
for each of the two possibilities in part a, finding the velocity of K' relative
to K and the snace-time nositions of each snrinter inK'.
11o8 (a) Use the relativistic velocity addition law and the invariance of phase to dis-
cuss the Fizeau experiments on the velocity of propagation of light in moving
liquids. Show that for liquid flow at a speed v parallel or antiparallel to the
J-"'Lll Ul LllC tigm LllC 'JlCCU Ul LllC tigm, a' UUMC:l ;eu iu llle lillJULiiLUTy, iS giVen
to first order in v by
c ,( 1
u = :t-:\ ::!: v 1 - 2 + -wdn(wl)
-----;;-:-
VJ \ I
. ;c tho f<onnon~"
"1" "o
nf tho J;nht ;n tho lo
"0 .,
{;n tho J;
"""!."
:~ ..
IU
outside it) and n(w) is the index of refraction of the liquid. Because of the
extinction theorem, it is assumed that the light travels with soeed u' -
c!n( w') relative to the moving liquid.
(b) Consult the paper of W. M. Macek, J. K :Schneider, and R. M. Salamon
[I. Appl. Phys. 35, 2556 (1964)] and discuss the status of the Fizeau
At-
111) Ani " ' ·· '"I T • tr"mfnrm"ti,-,n "nrl it<· · ""n hP "'
x'a = (g(J'fl + Eufl)xf3
X (g + E ")X~
wnere E " anu E " are inunitesimal.

(a) Show from the definition of the inverse that •"'~ = - E"~.
(b) Show from the preservation of the norm that E"~ = - E~".
.. . . .
~CJ ny Wllllll!', LllC l1 wr 11 terms u1 <.:unu avanant components on uout
sides of the equation, show that E"~ is equivalent to the matrix L (11.93).
. .
'"'"o-LII ~a} I'Ul Llle LULClllZ UUUM iillU lUliillUJ ffiiilll<.:C' n. iillU" 'JJUW Llliil
\" . ")
no'
" 0
n
"
(E' ° K) 3 = E' ° K
where E and E' are any real unit 3-vectors.
(b) use tne results or part a to snow tnat
exn(- lB o K) = I - Ro K sinh l + (B o K) 2 fcosh l - 11

11.11 Two Lorentz transformations differ bv an infinitesimal amount. In the notation of
Section 11.7 they are represented by A 1 = eL, A 2 = eL ' 8L. Without using explicit
matrix representations show that to first order in 8!" the Lorentz transformation
A~ A 2A, can be wntten as
A =I+ 8L + ;! [L, 81"] + ;! [L, [L, 8L]] + :! [!", [L, [L, 8L]]] + · · ·
Hint: The early terms can be found by brute force, but alternativelv consider the
~ o • • • >U+M) e," '
1J ''-' '-- 'Y m " Vl lllC VjJCl<llVl -'"'\") c c <lllU ULCH }'
A= 1.
·~
Ln . .L.L rroDJems ;J LL
11.12 Annlu thP rPsnlt of ProhlPm 11 11 to a nurelv aloebraic deviation of (11.116) on

'T'' ·r
(a) With
p . D.\ lilllll p}
T
f3
w
tn 00
j_\ )l'{, ,h -la•\
L + 8L
{3'
where {3' = V (13 + 813 .f + (813 j_) 2 , show that

1
813 j_ • K( tanh - I {3)

8L = -·l 813 11
• K -
f3
(b) Using the commutation relations forK and S, show that
C1
C2
= [I", 8L] = -
= [I", C 1]
( r
tanh- 1{3
= (tanh 1{3) 2 8Lj_

f3 (13 X 8l3j_) • S
C3 = [L, C2 ] = (tanh- 1{3) 2 C 1

C4 = [I", C3 ] = (tanh- 1{3) 4 8Lj_
where 8L j_ is the term in 8L involving 813 j_·
(c) Sum the series of terms for AT= A 2A] 1 to obtain
'Y2
Ar =I- (y2 813 11 + y 8l3j_) • K- 'Y + (13 X ll13J • S
1
correct to first order in 813. [See D. Shelupsky, Am..!. Phys. 35, 650 (1967).]
11.13 An infinitely long straight wire of negligible cross-sectional area is at rest and has
a uniform linear charge density q 0 in the inertial frame K'. The frame K' (and the
wire) move with a velocity v parallel to the direction of the wire with respect to
the laboratory frame K.
(a) Write down the electric and magnetic fields in cylindrical coordinates in the
rest frame of the wire. Using the Lorentz transformation properties of
the fields, find the components of the electric and magnetic fields in the
laboratory .
..
ih\ Ulh ,,
.,
'b'
.. ••
rl
•'-
.t rl.
'
""' iotPrl mith thP mirP in ito rPd
'].
(c) From the laboratorv charge and current densities, calculate directly the elec-
lrir ond moanPtir liPide in- thP '- '- r,.,mnare with the results of nart a.
.n
.L.L,J'I \a} eXpreSS llle LOrelllz SCalarS T - T "~' CJ' -I"~' auu J' '"'J a~ w' l<OUHo VL <0 auu
IS. Are there any other mvanants quadratic m tne ne10 strengtns E ana H .
._. ,], .•..
,..,
/L
c',':"
. ·' '- . c .. u
. c .. u .
·r ,,_·o.
u~•u w vu~ '"~' "'" u "'"~
'
auu ao a
'
rramet wnat are me crneria imposeo onE anu o sucuLUaLlllere is an inenia1
frame in which there is no electric field?
rl n , 1-1 th. tPnOAr
,_,
! ..\ 1:;',
.a
"' , V , D +. th c .. LI
-r "'"~
v . VVllill lUllllel ue tuuueu. "uac "' ~ '"~" ll> \Cali ~ApL ~o-
s1ons m terms ot the J-vector helds'l

11 1.:: Tn o rPrtoin n fromP o dotic . _,
ir fiPirl F .. is n"r"llP.l In lhP. Y '
• . rl n ")c;' ,. th •
, rl
, '" , ·u -u J r
,],
,
572 Chapter II Special Theory of Relativity-G
making an angle IJ with the axis. Determine the relative velocity of a reference
frame in which the electric and magnetic fields are parallel. What are the fields in
that frame for () << 1 and ()--> ( 7T/2)? #
11.16 In the rest frame of a conducting medium the current density satisfies Ohm's law,
J' uE', where CT is the conductivity and primes denote quantities in the rest
frame.
(a) Taking into account the possibility of convection current as well as conduc-
lion cunent, show that the covariant generalization of Ohm's law rs
where U" is the 4 velocity of the medium.

(b) Show that if the medium has a velocity v = cl3 with respect to some inertial
frame that the 3-vector current in that frame is
J = /'if[E + 13 X B - 13(13 • E)] + pv

where pis the charge density observed in that frame.
(c) If the medium is uncharged in its rest frame (p' = 0), what is the charge
density and the expression for J in the frame of part b? This is the relativistic
generalization of the equation J - a{E + v X B (see p. 320).
11.17 The electnc and magnetrc fields (I 1.1 52) of a charge m umtorm motron can be
obtained from Coulomb's law in the charge's rest frame and the fact that the field
strength F"~ is an antisymmetric tensor of rank 2 without considering explicitly the
Lorentz transformation. The idea is the following. For a charge in uniform motion
the only relevant variables are the charge's 4-velocity U" and the relative coordi-
nate X"= x; - x~, where x;andx~ are the 4-vector coordinates of the observation
point and the charge, respectively. The only antisymmetric tensor that can be
formed is (X" U~ - X~ U"). Thus the electromagnetic field F"~ must be this tensor
multiplied by some scalar function of the possible scalar products, X.X", X.U",
(a) For the geometry of Fig. 11.8 the coordinates of P and q at a common time
inK can be written x; = (ct, b), x; = (ct, vt), with b · v = 0. By considering
13
the general form of F" in the rest frame of the charge, show that
Verify that this yields the expressions (11.152) in the inertial frame K.
(b) Repeat the calculation, using as the starting point the common-time coor-
dinates in the rest frame, x;" (ct', b ~t') and x~" (ct', 0). Show that
where Y'" X:" x~" Verify that the fields are the same as in part a Note
that to obtain the rcstJ!ts of (11.152) it is necessary to use the time z of the
observatiOn pomt P m K as the t1me parameter.
(c) Finally, consider the coordinate x; = (ct, b) and the "retarded-time" coor-
dmate x~- let R, l3(ct R)l where R is the distance between P and qat
Ch. 11 Problems 573
the retarded time. Define the difference as Z" = [R, b- J3(ct- R)]. Show
thM in IPrm< nf 'l" onc1 na lhP fip]n io
q (Z"U~ - Z~U")
F"~ = -
\ 3
( ~ U"Z")
11, Hl TJ..
_, rl . <;,],.j, ,, .c . _, ,, .L
., ·o
"oh< 1"
·o·
with speed v = {3c, given by (11.152), become more and more concentrated as
p-> 1, as IS mmcmeu m rig. ''·"'· Lnoose axes so tnm we cnarge moves awng Lne
z axis in the positive direction, passing the origin at t = 0. Let the spatial coordi-
nates of the observation noint be (x v. z) and define the transverse vector r with
components x andy. Consider the fields and the sonrce in the limit of {3 = I.
(a) Show that the fields can be written as
F. 7n r, i'ifrt 7 )· R ?n v X r, i'ifrt 7\
r~ r~
where vis a unit vector in the direction of the particle's velocity.

(b) Show by substitution into the Maxwell equations that these fields are consis-
tent with a 4-vector sonrce density,
J" = qcv"I'Pl(r,)8(ct - z)
where the 4-vector v" = (1, v).

(c) Show that the fields of part a are derivable from either of the following
'+-vecLOr pmenna1s,
0 z ~ . ,, ' •
"' "' ~'1 v\ co 'J " ' \ ' " l h "1 "
v
or
A 0 = 0 = A'·, A 1 = -2q8(ct - z) V 1 ln(Ar1 )
where A is an irrelevant parameter setting the scale of the logarithm.

Show that the two potentials differ by a gauge transformation and find
thP """"" fnnrticm v
Reference: R. Jackiw, D. Kabat, and M. Ortiz, Phys. Lett. B 277, 148 (1992).
11.19 A n~rtide. of m~« M ~nn 4- '"'"~ p into two n~rtirl<e< of m~""' m.
and m 2 .
(a) Use the conservation of energy and momentum in the form, p 2 = P- P~o
and the in variance of scalar products of 4-vectors to show that the total en-
Pro" of thP fi"t norliC'lP in thP rPd fromP of thP dPC'm<ino n• . ,J. i'
' '
M 2 + mf- m~
Rl = ,.,.,
and that E 2 is obtained by interchanging m 1 and m 2 .
\D J ;:,now mar me Kmetzc energy 1, or me nn panicle m me same rrame IS
T, = t:J.M I - ~- ;;;; )
wnere LlJVI lV1 m, m2 IS me mass excess or \! vame or me process.

(,•\ ThP. rh"r<rPd ni-mP<nn ( M HQA MPVl ·1 intn " mn-mP<nn lm.
.,
InC '"I ' L H \ -' I.
\' 'L .,.
C\\ ,.-.,,_ .L •< ... . .c ...
--yrzr -ci!apter lT --s-pectal Theory of Relatmty-G
mu-meson and the neutrino in the pi-meson's rest frame. The unique kinetic
. . .
~L!;Y Ul LHC 111UU11 1> lllC lC Ul <.l L'lu-uuuy U<O~<.ty. H CHLCLCU lmpor-
tantly in the discovery of the pi-meson in photographic emulsions by Powell

on.-1 in 1 Od.7
11.20 The lambda particle (i\.) is a neutral baryon of mass M = 1115 MeV that dec""o
with a lifetime ofT= 2.9 X w-w s into a nucleon of mass m 1 = 939 MeV and a
pi-meson of mass m 2 = 140 MeV. It was first observed in flight by its charged
aecay moae~---> p -,= 7T in ITol.ITrcnamoers. The charged tracks ongmate from a
sin ale noint and have the annearance of an inverted vee or lambda. The ,
identities and momenta can be inferred from their ranges and curvature in the
magnetic field of the chamber.
(a) Using conservation of momentum and energy and the invariance of scalar
--pr ; or 'f-vecwrs snow tnat, ir tne opening ang1e tt oetween me two tracks
is measured, the mass of the decaying particle can be found from the formula
M-r = m"""'f + m:f + 2E,E2 - 2p 1p 2 cos li
where here p 1 and p 2 are the magnitudes of the 3-momenta.
(b) A lambda particle is created with a total energy of 10 Ge V in a collision in
the top plate of a cloud chamber. How far on the average will it travel in the
'- - ' cl. . -0 U1< ,< . -1 . :11 ·
~) 'b. 'b T 'b .. b
10 Ge V lambda if the decay is more or less isotropic in the lambda's rest

frame?
11.21 If a system of mass M decays or transforms at rest into a number of particles, the
sum Of whose masses IS less than M by an amount t..M,
(a) show that the maximum kinetic energy of the ith particle (mass m,) is
I 'T' \ Au/. ' m, t..M\
\ I}max
M 2M)
(b) determine the maximum kinetic energies in MeV and also the ratios to t..M
for each of the particles in the following decays or transformations of particles
arrest:
JL-->e+v+v
K+--> 7T' + 7T- + 7T
K ~ --> e± + 7T0 + v
K± ---> JL ± + + v 1r
0
p + j5 ___,. 27T+ + 27T- + 7To

p + p--> K+ + K + 37T(]
11.22 The nresence in the universe of on onnorPnthr nnifnrm "oPo" nf "· radi-
ation at a temperature of roughly 3K gives one mechanism for an upper limit on
""""!lTe" energies or pnorons mat nave traveled an appreciaDJe aisrance since tneir
creation. Photon-photon collisions can result in the creation of a charged particle
" · · • '" • -•· "\ " - ' • .. <<' · • ol ~<
" 7 '6J
"mass" of the two photons. The lowest threshold and also the largest cross section
occurs for an electron-positron pair.
(a) Taking the energy of a typical3K photon to beE= 2.5 10- 4 eV, calculate X
Tile energy for an incident photon such that there is energy just sufficient to
~olra "" al. " " "C", ";l, " • 1. ;h thi<
-,- ,. . ,. "b "b"
threshold value, the cross section increases to a maximum of the order of

(e2 /mc2 ) 2 and then decreases slowlv at hi2:her ener2:ies. This interaction is one
Ch. 11 Problems 575
mecnanism ior tne aisappearance or sucn pnOluus as mey rrave1

distances.
(b) There is some evidence for a diffuse x-ray background with photons having
energies of several hundred electron volts or more. Beyond 1 keY the spec-
trum falls as E-" with n = 1.5. Repeat the calculation of the threshold inci-
dent energy, assuming that the energy of the photon in the "sea" is 500 eV.
11.23 In a collision process a particle of mass m 2 , at rest in the laboratory, is struck by
• .< -' D ' •L .n.· . •L
u ~ 1• J'I.AB 'OJ LAB·
two initial particles are transformed into two others of mass m 3 and m 4 • The con-
liguratwns ot the momentum vectors m the center of momentum (em) frame (tra-
ditionally called the center-of-mass frame) and the laboratory frame are shown in
the figure.
_,..q'
PL.Ae
-..·
r•
Laboratory frame em frame
Problem 11.23
(a) Use invariant scalar products to show that the total energy Win the em frame
has its square given by
w2 2 -'- 2 ..,, c-
nnrl thnt thte ccms 1-
l' w
(b) Show that the Lorentz transformation parameters l3cm and 'Ycm describing the
, .L ,f th fr · th~ lohnrotnrH orP
-' -'
m. + E
'Ycm
w
(c) Show that the results of parts a and b reduce in the nonrelativistic limit to
the familiar expressions,
p'
11.24 The threshold kinetic energy Tth in the laboratory for a given reaction is the kinetic
. • . • • co • ' "
~H~'oJ V' CU~ !-'""''-'''-' VH a -, <a'o'-'' JUOC V "'"~'-' CH'-'
center of mass energy W equal to the sum of the rest energies of the particles in
the final state. Calculate the thresholc! lunellc energies tor the tollowmg processes.
Exoress vour answers in MeV or Ge V and also in units of the rest ener2:v of the
incident particle (unless it is a massless particle).
\3) n-meson pnotoproaucnon, yp ---> 7T p
\'"p 7.>U.. "''-' , H<rr LJJ.V H>'-' V)

(b) Nucleon-antinucleon pair production in nucleon-nucleon collisions, for ex-

am e
(c) Nucleon-antinucleon pair production in electron-electron collisions, e- e-·_,
e- e-pp and e+ e---> pp (me = 0.511 MeV).
11.25 In colliding beam machines such as the Tevatron at Fermilab or the numerous
e ' e storage nngs, counterrotatmg relativistic beams of particles are stored and
made to collide more or less head-on in one or more interaction regions. Let the
particles in the two beams have masses m 1 and m 2 and momenta p 1 and p 2 , re-
spectively, and let them intersect with an angle (J between the two beams.
(a) ~how that, to order (m!pf inclusive, the square of the total energy in the em
W
2 _
- 4p 1p 2 cos 2 -
2
e + (p, + Pz) (mi +-
-tn~)
p, Pz
(b) Show that the em merbal frame has a velocrty 111 the laboratory grven by
(p 1 + p 2 ) sin 0/2
f3cm = (£ I + £) .
z sm "'
where
p, + Pz)
tan a = tan-0
( p, - P2 2
The angle a is defined in the figme
(c) Check that the results of part b agree wrth those of Problem 11.23b.
(d) If the crossing angle is 0 = zoo and the colliding protons have p 1 = p 2 = 100
GeV/c, is the laboratory frame a reasonable approximation to the em frame?
Consider, for example, a proton-proton inelastic collision involving pion pro-
duction and examine the collinearity of two pions produced with equal and
opposite momenta of 10 GeV/c in the em frame.
l3cm
l
0/2\. ~ jo12
;_:s.---p; P2~----J
rromem ~~-~;:,
11.26 In an elastic scattering process the incident particle imparts energy to the station-
ary target. The energy !:J.E lost by the incident particle appears as recoil kinetic
energy of the target. In the notation of Problem 11.23, m 3 = m 1 and m 4 = m 2 ,
while !:J.E = T4 = £ 4 - m4.
(a) Show that AE can be expressed in the following different ways,
tiE cos 0')
PLAB Sill 4
2
!:J.E=-
2m2
Ch.ll Problems 577
mhPr~ n2 -f n n_ \2 fn n_ )2 [" "- \2 ;< th~ T · ;m,or;ont

{, --'\
(b) Show that for charged particles other than electrons incident on stationary
electrons (m 1 >> m 2 ) the maximum energy loss is approximately
f1f:max = 2·lf3 2 m,
where y and {3 are characteristic of the incident particle and y << (m 1 /m,).
r.;,., th.;c rP";lt o · ; tot;nn h,. Pnn• · · · thP rPl· ,-.,,n;o;nn
in the rest frame of the incident particle and then transforming back to the
laboratory.
(c) For electron-electron collisions show that the maximnm e.n.erov tr~n<fPr is
max = ( y - 1 )m e
f1£(e)
11.27 (a) A charge density p' of zero total charge, but with a dipole moment p, exists
m reference trame K · .. ,.here 1s no current dens1ty m K'. lhe trame K' moves
with a velocity v = l3c in the frame K. Find the charge and current densities
p and J in the frame K and show that there is a magnetic dipole moment,
m = (p X 13)/2, correct to first order in {3. What is the electric dipole moment
. " ..
lV lllC MWC V<UC< W
' . '"
{J.
(b) Instead of the charge density, but no current density, in K', consider no
charge density, but a current density J' that has a magnetic dipole moment
~:. ~ d. -~ ~ ' ' ' T/ -' .. L <L ~- -~ •
'5'
{3 there is an electric dipole moment p = 13 x m in addition to the magnetic
dipole moment.
.. .,_
..
.
~0 ~
'~· -' r ~~'
.-p
..
-"
>« •o " ' .n.u
'
C!C\Clll\C ··~ ; 111> JUMY <H I ' " ' aL LLLC U 1g111 lll Llle "- 11<1> ,_ ,
<II' = p · r'!r'', and A' = 0 (and thus only an electric field). The frame K' moves
with uniform velocity v = Be in the frame K.
,. ,
' ' n• .. e TT
'
·H. cv mot v<u<e< m
c n
p
•'
tu<e t'
..
"" '""
p. R A = 13 (p. R)
=-- ,
R3 R3
where R = x - x0 (t), with v = dXoldt at time t.

(b) Show explicitly that the potentials in K satisfy the Lorentz condition.
(c) Show that to first order in {3 the electric field E inK is just the electric dipole
field (centered at x0 ), or a dipole field plus time-dependent higher multipoles,
if viewed from a fixed origin, and the magnetic field is B = 13 X E. Where is
d. · ,-~ • '~'_,. "~~- n _, _,_,
as "~ "· v u.~w.
11.29 Instead of the electric dipole potential of Problem 11.28, consider a point magnetic
moment m in the moving frame K', with its potentials, ' = 0, A' = m x r' !r' 3
(and so only a magnetic field).
(a) Show that to first order in B the potentials in K are
(13 X m) · R (m X R)
<II= ~3 , A= ~3
Note that the scalar potential is the same as the static potential of the electric
dipole moment of Problem 11.27b. [But this gives only the irrotational part
'
Ul LllC
. '
llClU.J
(b) Calculate tne eJectnc ana magnetic news 111 K rrom me potentials and show
that the electric field can be expressed alternatively as
[3n(n · 13) - 13]
E = Ectipo~c(Perr = 13 X m) - m X R3
1 13 ) 3 [m(n · 13) + l3(n · m)]

E E"' "d n.. H x m + n x
L L lC
-.,
where 8 is the magnetic dipole field. In light of Problem 6.22, comment on
the interpretation of the different forms.
11.30 An isotropic linear material medium, characterized by the constitutive relations
(in its rest frame K'), D' = <E' and JLH' = 8', is in uniform translation with
vefodtv v 111 Tile mertJaT frame K. BY exploiting the fact that F , = (E 8) ~nrl
r' fn U\ tronofm~ oe · -' '"~1, A_ · ,~rio• T
show that the macroscopic fields D and H are given in terms of E and 8 by
D = EE + ·l( E- ~)[f3 2E~ + 13 X 8]
TT 1~ J 1\, n2~ 0 ~~
" r \ JL ' ~ t
R ,,.;;:JR orP romnonP,:;-t;;' hr to v

11.31 Consider a hollow right-circular cylinder of magnetic insulator (relative perme-
abilities E and fL and inner and outer radii a and b) set in rotation about its axis
at angular speed win a uniform axial magnetic field B0 . In 1913 the Wilsons mea-
sured the voltage difference between its inner and onter surfaces caused by a radial
internal electric field. Assuming- that locallv the relations of Problem 11.30 holrl
. . "t- ' .. ) . p
---a:nu--z>, wmcn are muepenuem or z ana </J, so1ve we equanons v • u u ana
V X H = 0 within the cylinder and show that the internal fields are
EP = -
fLWPBo
c(1 - w2p2/~)
(
1 - fLE ;
1)
_, d. £ -' . . ' '
\'
I 1
V = £-B0 (b 2 - a 2 )11 - - )
2c \ fLE
This experiment was an early validation of special relativity and Minkowski's elec-
trodynamics of material media in motion. If you are curious about how the Wilsons
made a magnetic insulator, look up the paper.
RPfPrPnrp· M w;J,.-.n ""rl H A w;knn Prnr Rm ~nr ' J AliQ QQ_1()/i
CHAPTER 12
Dynamzcs o7Re1atzvzstzc Partzcles

and Electromaf!netic
-=-
Fields
The kinematics of the special theory of relativity was developed in Chapter 11.
n' ., • ' • · T ·• P r ·• •
<VC llU" LULU LV LllC ljUC~llUll Ul >J • 111 LllC 111M jJ<lll Ul lllC l.JldjJlCI WC
discuss the dvnamics of charQ"ed narticle motion in external electromaP"netic

fields. The Lagrangian approach to the equations of motion is presented mainly
To mTroaucetlfe concept ot a Lorentz mvanant act10n trom wh1ch covanant dy-
namical equations can be derived. The transition to a Hamiltonian, with the
aennition or me canonical momentum, is men aiscussea. :-.everat sections are
_] .L -' tr. thP mr.t;r.n r.f ~ _] • .I ;n Pl • ~n.-1 m~~ ,.· hPl.-lo ()llr
o r
treatment of motion in a uniform static magnetic field is followed by considcra-
tlon of motion in a combination of electric and magnetic fields. Then the secular
changes (drifts) of a particle's orbit caused by nonuniform magnetic fields and
the adiabatic invariance of the linked flux are discussed. The problem of a rela-
•" ·~,:~T .. ~-~ . f. -• ,;· . ,. -' -•·,1 . rlrl _, r1
'5 ·o cry -e • 6' . r '
it is shown that to order u2 /c 2 it is possible to eliminate retardation effects and
write a Lagrangian (the Darwin Lagrangian) in terms of the instantaneous po-
sitions and velocities of the particles.
In the last five sections of the chapter the emphasis is on fields. First, the
~K. .11 . -' · ' C. • .L..l T " 'Tl. -'"' _,
-:1 ~5' ~u,
Lagrangian describing a "photon" with mass is presented and its consequences

in resonant circuits, transmission lines, and cavities described, as well as its man-
tresrarion in superconaucrors. A covariant generalization or me tlamiuonian tor
fields is next discussed, along with the conservation laws of energy, momentum,
' ' r ~• ' •• • •
dliU 1U1 , UUlll ~UUl l.t; 11 <:;<:; dliU 111 111 Jll Wllll
charP"ed narticles. The chanter ends with a derivation of the inv,.ri,.nt GrPP.n
functions that form the basis of the solution of the wave equation with a given
4-vector currentaens1ty as source.
., 'I .,
<-.iOOZ
T
~~
·-!--~ ---1 LT- -:1-.- ~~
:~- C- ~ n~J-,.•
·---· F"l.
.......... ~ .........
J
....,, .... ,15'"' ...
~ ·· ·fp in R Jnl 11 1 ... · - - -... ot-j,-. Ri~l-1~
·o
The equations of motion
"- ..
:::!:: = e E+-xB (12.1)
dt c
dE
eu • n ~lL.LJ
dt
~/'J
580 Chapter 12 Dynamics of Relativistic Particles and EleL1romagnetic Fields-G
for a particle of chargee in external fields E and B can be written in the covariant
form (11.144):
"
d U'" = _!___ F'"~ U (12.3)
dT me ~
where m is the mass, T is the proper time, and ua = ( yc, yu) =palm is the
4 ~' · nf thP. n~rtidte
Although the equations of motion (12.1) and (12.2) are sufficient to describe
the general motion of a charP"ed narticle in external electromagnetic fieirl<
(neglecting the emission of radiation), it is useful to consider the formulation of
the dynamics from the viewpoint of Lagrangian and Hamiltonian mechanics. The
Lagrangian treatment of mechanics is based on the principle at least action or
Hamilton's principle. In nonrelativistic mechanics the system is described by gen-
.er" lized · 1Mes a.(t) "nd ~es--;.i,7tT. The Lagrangian Lis a functinn~l
of q, and q, and perhaps the time explicitly and the action A is defined as the
time inte2:ral of L along a oossible oath of the svstem. The principle of'least action
states that the motion of a mechanical system is such that in going from a con-
figuration a at time t1 to a configuration b at time t2 , the action
A =
('2 L[q,(t), q;(t), t] dt (12.4)
J ,,
is an extremum. By considering small variations of the coordinates and velocities

away from the actual path and requiring 8A = 0, one obtains (see Goldstein,
Chapter 2) the Euler-Lagrange equations of motion,
I \
(T,~
-~=0 (12.5)
dt , aq,, aq,
We wish to extend the formalism to relativistic particle motion in a manner
consistent with the special theory of relativity and leading for charged particles
in external fields to (12.1) and (12.2) or (12.3). There are several levels of so-
phistication possible. The least sophisticated and most familiar treatment contin-
ues with ordinary coordmates, velocities, and ttme and generauzes trom the non-
relativistic domain in a straightforward way. More sophisticated is a manifestly
covanant discussion. We hrst presenTTife erementary approacn ana men inaicme
the manifestly covariant treatment.
A. Elementary Approach to a Relativistic Lagrangian

.. . _, - ' ~ ~ U~ ~~ _,
<U a lCldll v"ll'-' Ld!;"'"!;'"" lUI " }'Ul U~l~ "' ~A.vUWl HvlUO ·~ H
consider the question of the Lorentz transformation properties of the Lagrangian.

" ~ ' r · o ' • ' •' • • .1 ~L T
llv Vl lCldllVHJ lll<O uvu~oo cuo ~
> 1 vm '"'-' " " ' '"'

scalar because the equatiOns oT motiOn areaetermmed by the extremum conai-
· S: A "· Tf ·~•"' ,-1,~~ +h~ · '~ t;nw ~ ;nto (1? .1\
t'
'
~,
at /'aT, me actiOn mregra1 oecomes

I .,
7t -1, yr;7TT ~ <L.UJ
;-.ince proper rime is invariant rne conairion rnar x ""u uc ;" var;um '"4 u;rc~ uw•
yL be Lorentz invariant.
. . r oL
1111;; L<i!;I<LIIgiall 101 U llt:t:~ a Wll<,UVIl Vl <HC Vl ""'
particle and its mass, but cannot depend on its position. The only Lorentz invar-
Sect. 12.1 Lagrangian and Hamiltonian for a Relativistic Charged Particle 581
iant function of the velocity available is UaUa = c2 • Thus we conclude that the
Lagrangian for a free particle is proportional to y- 1 = Y1 - {3 2 • It is easily seen
that
1
ISthe proper multiple of y to yield, through (I 2.5), the free-particle equatiOn
of motion,
d
dt (ymu) 0 (12.8)
The action (12.6) is proportional to the integral of the proper time over the
path ftom the initial propel time r1 to the final proper time -r2 . This integral is
Lorentz invariant, but it depends on the path taken. For purposes of calculation,
consider a reference frame in which the particle is initially at rest. From definition
(11.26) of proper time it is clear that, if the particle stays at rest in that frame, the
integral over proper time will be larger than if it moves with a nonzero velocity
along its path. Consequently we see that a straight world line joining the initial
and final points of the path gives the maximum integral over proper time or. with
the negative sign in (12.7), a minimum for the action integral. This motion at
constant velocity is, of course, the solution of the free-particle equation of motion.
The general requirement that yL be Lorentz invariant allows us to determine
the I agrangian for a relativistic charged particle in external electromagnetic
fields, provided we know something about the Lagrangian (or equations of mo-
tion) for nonrelativistic motion in static fields. A slowly moving charged particle
is influenced predominantly by the electric field that IS denvable from the scalar
potential <1>. The potential energy of interaction is V = e<l>. Since the nonrelativ-
istic Lagrangian is (T- V), the interaction part Lint of the relativistic Lagrangian
must reduce m the nonrelatJvistJc limit to
(12.9)
Our problem thus becomes that of finding a Lorentz invariant expression for )'L,nt
that reduces to (12.9) for nonrelativistic velocities. Since ¢1 is the time component
of the 4-vector potential A a, we anticipate that yL;nt will involve the scalar prod-
uct of A a with some 4-vector. The only other 4-vectors available are the momen-
tum and position ~ectors of the particle. Since gamma times the Lagrangian must
be translationally invariant as well as Lorentz invariant, it cannot involve the co-
ordinates explicitly. Hence the interaction Lagrangian must be*
(12.10)
e
Lint = -e<l> +- U • A (12.11)
*Without appealing to the nonrelativistic limit, this form of L,n, can be written down by demanding
that 1" int be a I orentz invariant that is (1) ljnear jn the charge of the particle (2) linear jn
the electromagnetic potentials, (3) tran~lationally invariant, and (4) a functjon of no hjgher than the
first time derivative of the particle coordinates. The reader may consider the possibility of an inter-
action Lagrangian satisfying these conditions, but linear in the field strengths p·~. rather than the
::m~ cnapter 1L uynamics 01 Ketativistic--r~ -u
The mmhin>~tion of (1? 7) ~nrl (1? 11) • •1 "c, the comolete relativistic L>~ i~
for a charged particle: ,

u2 e
T 2 /. ~ ,ffi /o~
"" '"L ,/~ 2 \~"'·'<-)

~
Verification that (12.12) does indeed lead to the Lorentz force equation will be
left as an exercise for the reader. Use must be made of the convective derivatiw
[dldt = (a/at) + u · V] and the standard definitions of the fields in terms of the
potentials.
ThP ,J mnmPntnm P rnnina"t" tn the nosition lte x is nh_
..
ldl!it:U uy LUt: '
aL e
P, = -:;-- = ymu, + -:A, (12.13)
UUj
Thus the conjugate momentum is

e
P=p+-A (12.14)
c
•
WUIOHO J'
,,
ymu , uno
,. . 'T'l. u.. ., . TT '
<O U
function of the coordinate x and its conjugate momentum t' and ts a constant of
the motion if the Lagrangian is not an explicit function of time. The Hamiltonian
is defined in terms of the Lagrangian as
H=P·u-L (12.15)
Tho uolnritu" ~net ho
-J
Pli~in~tPrl frn~ (1 "! 1•:;) in f:wor of P ~nrl y From (1? 11)
'
VI \ H d <) W\0 UHU Llldl
cP- eA
u 2
(12.16)
eA
~ P-~ + m2c2
c
. . '
VVnen llllS IS SUI • lillO ~!L.l.J) aiillllllU L \~"-.H•j, UHC la""'
on the form:
<I ~If~» nA\2 ::;: ~ 2,-.4 -!- ,CD (1? 17)
'7
Again the reader may verify that Hamilton's equations of motion can be com-
bined to yield the Lorentz force equation. Equation (12.17) is an expression for
th tnt~ I U/ nf tho 'nlo Tt .-l;fforc f;:n thP frPP.- ' on >r~u hv the
addition of th~ potential ;nergy eel> and by the replacem~nt p ---> [P - (e/c)A].
Th<>.<P. two ,.-,.. inns :.n•. >Jrtn"llv onlv one 4- r change, This can be seen
~-. ,, '_, ~
' '
,, "T'L. ·~
uy u< 0:'¥ Ill \ l L . . l ) dHU >ljUdllll!; u o.u~o. >u~u
(W- ecl>) 2 - (cP - eA? = (mc) 2 (12.18)

This ts 1ust the 4-vector scalarproauct,
PaP"'= (mc) 2 (12.19)
mhorP
1
e ,)
p"' = E
r , p) = ~1 (W - e(I>), P - ~A (12.20)
Sect. 12.1 Lag1 angian and Hamiltonian for a Relativistic Charged Particle 583
We see that the total energy Wk acts as the time component of a canonically
conjugate 4-momentum pa of which P given by (12.14) is the space part. A man-
ifestly covariant approach, discussed in the following paragraphs and also in
Problem 12.1 leads naturally to this 4-momentum.
In passing we remark on the question of gauge transformations. Obviously
the equations of motion (12.1) and (12.2) are invariant under a gauge transfor-
mation of the potentials. Since the Lagrangian (12.10) involves the potentials
explicitly, it is not invariant. In spite of this lack of invariance of L under gauge
transformations it can be shown (Problem12.2) that the change in the Lagrangian
is of such a form (a total time derivative) that tt does not alter the action mtegral
or the equations of motion.
B. Manifestly Covariant Treatment of the Relativistic Lagrangian

To make a manifestly covariant description, the customary variables x and
u are replaced by the 4-vectors xa and ua. The free-particle Lagrangian (12.7)
can be written in terms of ua as
me-vlu ua
(f (12.21)
Then the action integral (12.6) would be

T2
A = -me
J VYJ;Jr
71
dr ( 12.22)
This manifestly invariant form might be thought to provide the starting point for
a variational calculation leading to the equation of motion, dUaldT 0. There
is, however, the equation of constraint,
or the equivalent constraint,
U dUry = 0 (12.24)
u dr
on the equations of motion. This can be incorporated by the Lagrange multiplier

technique, but we pursue a different, equivalent procednre The integrand in
• a
\ /d T d T ' u {3
that is, the infinitesimal length element in 4-space. This suggests that the action
integral (12.22) be replaced by
f' J
2
A = -mcguf3 dXa dXf3 ds (12.25)
,, ds ds
where the 4-vector coordmate of the particle is x"(s), with s a parameter that is
a monotonically increasing function of r, but otherwise arbitrary. The action
integral is an integral along the world line of the particle, and the principle of
least action is the statement that the actual path is the longest path, namely the
·1?
goeaeSJC.·· !fie V dl<O llU ' - ' dllU Ulv mo, UUl S ·J
is considered _as arbitrary. Only after the calculus of variations has been COJII-
pltOLCU UU WC
u
'-J 5 ds ds uo
\~~·"-V)
J <L (1') ')'1\ A • ' n~~;~t;nn<>l N,],--,].

auu '1' • \ · "/'
with (12.25) yields the Euler-Lagrange equations,
me d =0 (12.27)
ds
I
~ ds ds _
or
v
"' d?-
. -'
a~ ·~ lUI ucc-pa• ""''-' .
For a charged particle in an external field the form of the Lagrangian (12.11)
·' ,; • r · C. .C d . · ' ~ ·
llldllll<O 'J'-'V>UHUUC<V' V 'O
("' I aB dxa dx~ e dxa A a{ ..\ .J.
--rr J,, ·y 6
ds ds c ds \-'<7
\ •I
lT. :1• ' · .1 ·.1..1 .. t h , P ,], T <>rrronrro orm<>t;Ano

r 'l' J ·o· ·o -.- '
d ' aL - :c.ar n (1? 1()\

ds ;.I dx a l
L ds
'l:Y ..,.
r _ r mr /~'a~ dxa dx~ + e dxa Aa(x) 02 _31)

V ds ds c ds _
Pvnl;,--;th, (1? 'Ul\ """" rlivision hv the sauare root and use of (12.26) becomes
dT C dT C dT
";nrP dA aldT - ( dx !dT\ a~ A a this eauation can be written as
(12.32)
UT C UT
which is the covariant eauation of motion (12.3) in different notation.

The transition to the conjugate momenta and a Hamiltonian is simple
enough, but has problems of interpretation. The conjugate momentum 4-vector
is defined by
., f
(12.3J)
*The geodesic is the longest path or longest proper lime tor limcl!ke separatmn at evems . .>cc
Rohrlich, pp. 277-278.

Sect. 12.2 Motion m a Uniionn, Static Magnetic t<Ield ::lO::l
_, _, ..
cr'l-.
'"b' ' •• (1,., '2'2\
_'- ' 'I
••• (!')
\ '
1~\
'/'
..
be traced to the properties of the Lorentz space-time. A Hamiltonian can be

defined by
fl = P aua + L (12.34)
Elimination of ua by means of (12.33) leads to the expression,
n
- 1
ra
eA.\
1
eA"' I e \ e
. /1
m\ c) r -c~ \ra ~ Iia) 1 J ~ LC.-'-')
c c
Hamilton's equations are
~: = :~ = ! (pa _ ~ A a)
and (12.36)
dr = _ afl =~(PI'_ eAI') il"A~'
dT ax,. me c
. e a 22 I e
where we have made use of the constramt pa - -A •) = m c Pa - - A a ) (
1
after differentiation. These two equations can be immediately shown to be equiv-
alent to the Euler-Langrange equation (12.32).
. . . . .
vv nue llle namuwman auove Is 1urmauy Ly, u 11as severaqw .
The> fir<t i< th~t it i< hv ~ ·" · inn ~ T nrPnt'l <rohr not "n PnPrrnrlik-P- rm,.ntitv
Second, use of (12.23) and (12.33) shows that fl = 0. Clearly, such a Hamiltonian
formulation differs considerably from the familiar nonrelatiVIStic versiOn. The
.-1. .<. D. (, ?.0 U \ <. .-1; ' ,< +!-.' .-1 ••• •
\lT' ·;
Hamiltonians.
12.2 Motion in a Uniform Static Marmetic Field

AS a llfSL Imporcam examp1e 01 Lne uynamics 01 cnargeu paniCles in e1eCLromag
nPtir r.Pl.-1< "'"' rc>n<ir1Pr thP · in " nnifnrm d"tir m• · in.-lnrtinn R
The equations of motion (12.1) and (12.2) are
dp e ~
dE A
"~ ~~·
v
~.
'"~·-)
dt ( T
dt
where here the particle's velocity is denoted by v. Since the energy is constant
in time, the magnitude of the velocity is constant and so is y. Then the first
equation can be written
dv
-=VXI•h (12.38)
at
where
eB ecB
w =- -
ymc
= -
~
(12.39)
IS tne gyration or precessiOn rrequency. 1 ne monon aescrmea oy ~ lL...Jil) IS a

· ,], · · ,], · tn U on.-1-" nnifn~ tronclotinn· r~orollPI tn R ThP
solution for the velocity is easily shown to be
v(t) = u,e 3 + wsa(e 1 - ie 2)e-h'n' ( 12.40)
586 Chapter 12 Dynamics of Relativistic Particles and Electromagnetic Fields-G
where e 3 is a unit vector parallel to the field, e 1 and e 2 are the other orthogonal
unit vectors, u11 is the velocity component along the field, and a is the gyration
'" cy'j, ,,; •• ' " •• ,, ,f.;: . . t. .1. •
·.- 'I' 111en -~u.
uue 1.0a11 "'e lllaL \ Lu-+u J represen,s a countercwcKwise rorarwn \ ror posrtiw·
,) whP.n ;n thP nf R A •• t. . ;, ,;p]rJ,. '" ~-
J
u"-
placement of the particle,
x(t) = X 0 + u11 te 3 + ia( e 1 - ie 2 )e-iw"1 (12.41)
The path is a helix of radius a and pitch angle a= tan- 1 (u 1 1w 8 a). The magnitude
of the gyration radius a depends on the magnetic induction B and the transverse
momentum p.L of the particle. From (12.39) and (12.40) it is evident that
.D
Cf> _l c~~
ThiS torm IS convement tor the Cletermmatwn of nart1cle momenta. The rad;",
of curvature of the path of a charged particle in a known B allows the determi-
nation of its momentum. For particles with charge the same in magnitude as the
electronic charge, the momentum can be written numerically as
p.l(MeV/c) = 3.00 X 10- 4 Ba (gauss-em) = 300 Ba (tesla-m) (12.42)
17.1 M. in ( .'nmhinPrl I lnit"nrm .\'tntir 11

.
7 ~ 7
ana Mavneuc l'leta.<>
We now consider a charged particle moving in a combination of electric and

magnetic fields E and B, both uniform and static, but in general not parallel. As
an important special case, perpendicular fields will be treated first. The energy
equation (12.2) shows that the particle's energy is not constant in time. Consc-
quently we cannot obtain a simple equation for the velocity, as was done for a
T
.. ·
static ma!!netic field. But an appropriate Lorentz tr,.nsform,.tion simnljfip, th.e
,, r-. T
" ..... <V a ~VV<'
_,. £
U <U.H~
K movmQ: With a velocitv u wrtn respect to the ori!Tmal frame. Then the lorentz
fnrp,> Ptm~tjnn fnr thP rn~rt;p]p ;, T<' ;,
T
-dp' = e ( E' + v' X B')

dt' c
where the primed variables are referred to the system K'. The fields E' and B'
are given by relations (11.149) with v replaced by u. Let us first suppose that
IE I < IB 1- If u is now chosen perpendicular to the orthogonal vectors E and B,
EXH
u c n2 (1L.4J)
~
we hnd the helds m K to be
E' = 0,
I E~ = r( z E + X B) = 0
(12.44)
1 I B2 - E2 1/2
Bll = 0, B~ =- B = n2 ) B
[ ~
;o,ecr. H.-' JVtonon m \.-omomeo, unnorm, ;o,rauc J<,tecrrtc and Magnetic l'telds ::'!l'H
wnere 11 ana _1_ rerer w me airec1ion or u. m me rrame K the only he!a actmg 1s
a static magnetic field B' which points in the same direction as B, but is weaker
than B by a factor y- 1. Thus the motion inK' is the same as that considered in
Section 12.2, namely a spiraling around the lines of force. As viewed from the
original coordinate system, this gyration is accompanied by a uniform "drift" u
perpendicular toE and B given by (12.43). This drift is sometimes called the
E X B drift. The drift can be understood qualitatively by noting that a particle
that starts gyrating around B is accelerated by the electric field, gains energy, and
so moves in a path with a larger radius for roughly half of its cycle. On the other
half, the electric field decelerates it, causing it to lose energy and so move in a
tighter arc. The combination of arcs produces a translation perpendicular to E
'~ ' . . . . . . . .
u11u u a> >HuwH 111 '''10· "'"'· 111c u1 u1u< ,, em or we stgn 01
the charge ot the particle.
ThP rlrift vPlf'\ritv n (1? <11\ h~' '~ ' ~~ mP~ . >~ ~nlu ;f ;t ;o loco th~n tho
,1, , f ,. ~· .1. :< lvi ~ 1u'1 ;, l v l - In I .,: . ,""" . "'.u .
'':' ' .. , :J I 'I
'J I • I' I ~I I '1'. V"O
waL Llle pan•cte is conunuauy acce1ermeu m we UJrection or ~ ana its average
energy continues to increase with time. To see this we consider a Lorentz trans-
formation from the original frame to a system K" moving with a velocity
ExB
u' = c £2 (12.45)
relative to the first. In this frame the electric and magnetic fields are
E"I' = 0' E" =

j_
_!_ E =
i
( 2 2r2
E - B
E2 E
(12.46)
u' X E\
ll'l u, nJ. 'YI n I u
v
I
-rnus in Llle sysLem 1\. we pa"~'-":' is acLeu on uy a purely-ereciTu:rratic"ueJa wnicn
causes hyperboliC motion Wlth ever-mcreasmg veloCity (see Problem 12.3 ).
The fact that a particle can move through crossed E and B fields with the
uniform velocity u = eElB provides the possibility of selecting charged particles
according to velocity. If a beam of particles having a spread in velocities is nor-
mally incident on a region containing uniform crossed electric and magnetic
fields, only those particles with velocities equal to eEIB will travel without de-
flection. Suitable entrance and exit slits will then allow only a very narrow band
of velocities around eElB to be transmitted, the resolution depending on the
geometry, the velocities desired, and the field strengths. When combined with
mf'\mPntum ,1 ·• · "''"h ~' ~ rlPfiPrtina m~anPt >1 >co 17 v D ' ' · ,. '
0
' -J
1..
+e 1~11flll~fll
vvvvv
-e~
u
Fignre 12.1 E X B drift of charged
B
particles in crossed fields.
!'ll'Sl'S Chapter 12 Dynamics OfRefativtstic Yarttcles and ~tectromagneuc-r-iems: OG
1'~' w "~15"-""'"'oY .. ~'"·

If E has a component parallel to B, the behavior of the particle cannot be
' ' " -L " .1 '~ T'ho e~~l~r ~rArlu~o-1<' • 1l ;, coT
·r ' · r
invariant quantity (see Problem 11.14), as is (B 2 - E 2 ). When the fields were
·:~.~·~· (1<'-. n ()\ it \Xl~< nr><<ihiP tr. finrl ~ T r.rpnt7 fr~mp whterte F. {I
it IBI >lEI, o~·B = 0 if lEI> IBI. In those coordinate frames the motion was
r<>l · __ <imnl<> Tf F. R. =I= 0 ·" ·•· ic and mao-netic fields will exist simnlb-
neously in all Lorentz frames, the angle between the fields remaining acute or
obtuse deoendin!! on its value in the original coordinate frame. Consequently
motion in combined fields must be considered. When the fields are static and
uniform, it is a straightforward matter to obtain a solution for the motion in
Cartesian components. This will be left for Problem 12.6.
12.4 Particle Drifts in Nonuniform. Static Mal!netic Fields

In astrophysical and thermonuclear applications it is of considerable interest to
know how particles behave in magnetic fields that vary in space. Often the vari-
ations are gentle enough that a perturbation solution to the motion, first given
by Alfven, is an adequate approximation. "Gentle enough generally means that
the distance over which B changes appreciably in magnitude or direction is large
compared to the gyration raoms a ot the particle. Then me tow est oruer approx
imation to the motion is a spiraling around the lines of force at a frequency given
Dy the local varue ortiTe magnt:tic inauction. 1n me nexL appn , Luc uw"
undergoes slow changes that can be described as a drifting of the guiding center.
'. ' -1 · · " •'- £! - U L '..J -__, ' _.1; ,;
UlC 111M LJjJC Ul 'T VJ "'~ "~~u <v v~
perpendicular to the direction of B. Let the gradient at the point of interest be

' •' T " . r -'- ' ,; ;;1-, D {I 'J'> · - >~ hre> ._-1 thP
H< "''-' VJ U~'-' ' ' '
gyration frequency can be written

e 1 aB
ws(x) = - - B(x) = w0 1 +- - n ·x (12.47)
ymc Bo ag o
In (12.47) g is the coordinate in the direction n, and the expansion is about the
. .
Oflglll 01 !;OUr
.
WlltHO WB
~· "'
WQ• 0111'-'<0 tll\0
,. . ,-~.
UJ U " U"
_,_
,
..
motion parallel to B remains a uniform translation. Consequently we consider
-,-~ " " .o . Hl."' ..1. "·thP
VlllJ < 111 tlJ\0 U UHO o V h ' V H l V U V H o >l lf, _]_ '0 'b 'U '
uniform-field transverse velocity and v, is a small correction term, we can sub-

""'" \- 'I -,
uv_j_ (! '1 AQI

dt • _!_ ~ "-'BV"!
ana, Keeping omy nrst- oruer terms, OuLaiu we "1-'l"

aB
X Wo
dt ag o
Sect. U.4 Particle Drifts in Nonuniform, Static Magnetic Fields 589
From (12 40) and (12 41) it is easy to see that for a uniform field the trans-
verse velocity v0 and coordinate Xo are related by
v0 = - w 0 X (x0 - X)l
- (12.50)
1
(x0 X) 2 (w 0 X v0 )
where X is the center of gyration of the unperturbed circular motion (X = 0

here). If (w 0 x v0 ) is eliminated in (12.49) in favor of x0 , we obtain
ti;l = [ v1 - _!_ ("B) w0 X x0(n • x0) l X W0 (12.51)
vc = (v,> = -
B0
-
ago
w0 X ((x0 )j_(n · x0 )> (12.52)
To determine the average value of (x0 ), (n · x 0 ), it is necessary only to observe

that the rectangular components of (xe)~ oscillate sinusoidally with peak ampli
tude a and a phase difference of 90°. Hence only the component of (x 0 )j_ parallel
to n contributes to the average, and
((xa)jn x0 )) H (12.53)
2
(12.54)
2 B0 ag o
An alternative form, independent of coordinates, is
(12.55)
From (12.55) it is evident that, if the gradient of the field is such that a IVBIB I
<< 1, the drift velocity is small compared to the orbital velocity (wBa). The par-
ticle spirals rapidly while its center of rotation moves slowly perpendicular to
both Band VB. The sense of the drift for positive particles is given by (12.55).
For negatively charged particles the sign of the drift velocity is opposite; the sign
change comes from the definition of wB. The gradient drift can be understood
qualitatively from consideration of the variation of gyration radius as the particle
moves in and out of regions of larger than average and smaller than average field
strength. Figure 12.2 shows this qualitative behavior for both signs of charge.
Another type of field variation that causes a drifting of the particle's guiding
center is curvature of the lines of force. Consider the two-dimensional field shown
in Fig. 12.3. It is locally independent of z. Figure 12.3a shows a constant, uniform
magnetic induction B0 , parallel to the x axis. A particle spirals around the field
lines with a gyration radius a and a velocity wBa, while moving with a uniform
velocity v 11 along the lines of force We wish to treat that motion as a zeroth-order
approximation to the motion of the particle in the field shown in Fig. 12.3b, where
the lines of force are curved with a local radius of curvature R that is large
com ared to a.
./
VB
Fignre 12.2 Drift of charged particles
due to transverse gradient of magnetic
field.
~ H'-' HhWUH~'-'' • !"' •v
spiral around a field line, but the field line curves off to the side. As far as the
.C <'- .__,. ;c ~nn .-1 th;c ;c Mm;uqlnnt tn q · · ,,
acceleration of ~agnitude vW!R. This acceleration can .be viewed as arising f;~~
an effective electric field
(12.56)
in addition to the magnetic induction B 0 . From (12.43) we see that the combined
rr • ' · ~ ' ' ' <'- , · · .-1 • , A. :.1'. ,/, .;,,
(12.57)
With the definition of ws = eB 0 /ymc, the curvature drift can be written
(12.58)
in which R is the radius

vector from the effective center of curvature to the position of the charge. The
<ian in (12.581 is annronriate for nositive chanres and is indenendent of the sign
of v11 • For negative particles the opposite sign arises from ws.
A more straightforward, annougn peaestnan, aerivation oi ~lL..JO) cau uc
given by solving the Lorentz force equation directly. If we use cylindrical coor-
(a) (b)
Filmre 12.3 (a) Particle moving in helical path along lines of uniform, constant
magnetic induction. (b) Curvature of lines of magnetic induction will cause drift
perpenaicuJan:u-rrre-x~.
Sect. 12.4 Particle Drifts in Nonuniform, Static Magnetic Fields 591
dinates (p, cp, z) appropriate to Fig. 12.3b with origin at the center of curvature,
t~ • • rl >+• ~ ,] A. •+ Q Q (Of~\ Th~n +h~ f~•~A
t:L C:Clll ue "'""'Y LV t;lVt; Ult: llllt:to t:ljl •.
.. A-2
P - P'f' = -wn -R Z.
p
p;jJ + 2pep = 0 ( 12.59)
n
7- ";.
p
The second equation has a first integral, p 2 ep = Rv 11 , a constant. The third equation
has a first integral, z = w s ln(p/R) + u0 , where u0 is a constant of integration.
With the zeroth-order trajectory a helix with radius small compared to R, it is
natural to write o = R + x and exoand ( o!R)n and In( o!R) in vowers of x/R. Then
7 .r + 11. ~nrl thP r~rli~ 1 Pnn~tinn nf mntinn r~n hP ~- · l~tPrl hv
·· 2 2) 2 VI V:l
x + (Wn + 3 R x = R - 2 wnVo
which describes simple harmonic oscillations in x around a displaced equilibrium

?.
where we have assumed v1 << wsR. The mean value of z is then

v~
(z> = v0 + wn(x> = _LR. (12.60)
Wn
CT'l-" .• ,,_ ..!.:c. '- {10<:0\
~HW WJ~Jt • ~ • 5'. ") \ • "J•
.t'Or regwns 01 space m wmcn mere are no curreH<> rne 11 unn Vc
(12.55) and the curvature drift vc (12.58) can be combined into one simple form.
This follows from the fact that for a two-dimensional field such as shown in
Fig. 12.3b V x B = 0 implies
j_~ -
B W
Evidently then, for a two-dimensional field, the sum of vc and vc is a total drift
velocity,
(lL.bl)
Wni\ \ I\D I
WllCOlCO U j_ WBU " lllCO lldl"VCO'"t; LJ

~
Vl !;Jl<lllVll. 1 Vl
'
·eo
' ' llVW
relativistic particles in thermal equilibrium, the magnitude of the drift velocity is

172 T(K)
(12.62)
VD(cm/s) = Ofm\
,. -, Q(~oucc\
r -r -J
tV
'
. '
. HVt •
.
"~
"' in
., '
in hrnP< nf
~
LallVH
is a toroidal tube with a strong field supplied by solenoidal windings around the
592 Chaptet 12 n,namies of Relativistic PMtieles and Eleetromagnetie Fields G
torus. With typical parameters of R = 1 meter, B = 103 gauss, particles in a 1 eV

plasma (T = 104 K) will have drift velocities uD ~ 1.8 X 103 cm/s. This me'YJs
that they will drift out to the walls in a small fraction of a second. For hotter
plasmas the drift rate is correspondingly greater. One way to prevent this first-
order drift in tm oidal geometries is to twist the torus into a figure eight. Since
the particles generally make many circuits around the closed path before drifting
across the tube, they feel no net curvature or gradient of the field. Consequently
they expenence no net dnft, at least to first order in l!R. This method of ehmt-
nating drifts due to spatial variations of the magnetic field is used in the
Stellarator type of thermonuclear machine, in which containment is attempted
with a strong, externally produced, axial magnetic field.
12.5 Adiabatic bzvariance of Flux Through Orbit of Particle

The various motions discussed in the preceding sections have been perpendicular
to the lines of magnetic force. I hese motwns, caused by electnc nelds or by the
gradient or curvature of the magnetic fi€ld, arise became of the peculiarities of
the magnetic-force term in the Lorentz force equation. To complete our general
survey of particle motion in magnetic fields, we must consider motion parallel to
the lines of force. It turns out that for slowly varying fields a powerful tool is the
concept of adiabatic invariants. In celestial mechanics and m the old quantum
theory, adiabatic invariants were useful in discussing perturbations on the one
hand, and in deciding what quantities were to be quantized on the other. Our
discussion will resemble most closely the celestial-mechanical problem, since we
are mterested m the behavior of a charged particle in slowly varying fields, which
can be viewed as small departures from the simple, uniform static field consid-
ered in Section 12.2.
The concept of adiabatic invariance is introduced by considering the action
integrals of a mechanical system. If q; and p; are the generalized canonical co-
ordinates and momenta, then, for each coordinate which is periodic, the action
integral J, is defined by
(12.63)
The integration is over a complete cycle of the coordinate q;. For a given me-
chanica! system with specified initial conditions the action integrals l; are con-
stants. If now the properttes of the system are changed m some way (e.g., a change
in spring constant or mass of some particle), the question arises as to how the
actton mtegrals change. It can be proved* that, if the change in property is slow
compared to the relevant periods of motion and is not related to the periods
(such a change is called an adiabatic change), the action integrals are invariant.
This means that, if we have a certain mechanical system in some state of motton
and we ntake an adiabatic change in some property so that after a long time we
end up with a different mechanical system, the final motion of that different
system will be such that the action integrals have the same values as in the initial
*See, for example, M. Born, The Mechanics of the Atom, Bell, London (1927), or I. Percival and
D. Riehmds, Jntwdttction ltJ Dynamics, Cambridge URP.ersity Press, Cambridge (1932), ~octiGn 9 4
Sect. 12.5 Adiabatic Invariance of Flux Through Orbit of Particle 593
system. Clearly this provides a powerful tool in examining the effects of slow
changes in properties.
For a charged particle in a uniform, static, magnetic induction B, the trans-
verse motion is periodic. The action integral for this transverse motion is
(12.64)
where P .L is the transverse component of the canonical momentum (12.14) and

dl is a directed line element along the circular path of a part1clc. From (12.14)
we lind that
1 = f -ymv.L • dl + ~ f A · dl (12.65)
Since v.L is parallel to dl, we find
T t 2
ym£·> 8 a de + : t~· dl (12.66)
Applying Stokes's theorem to the second integral and integrating over ()in the
first integral, we obtain
J = 27rymwsa2 + ~c f s
B · n da (12.67)
Since the line clement dl in (12.64) is in a counterclockwise sense relative to B,

the unit vector n is antiparallcl to B. Hence the integral over the circular orbit
subtracts from the first term This gives
(12 68)
c
making usc of we eB!ymc. I fie quant1ty H7ra 2 1s the flux through the part1cle's
orbit.
If the particle moves through regions where the magnetic field strength varies
slowly in space or time, the adiabatic invariancc of J means that the flux linked
by the part1cle's orb1t rcmams constant. If B increases, the radius a will decrease
so that B7ra 2 remains unchanged. This constancy of flux linked can be phrased
in several ways involving the particle's orbit radius, its transverse momentum, its
magnetic moment. These different statements take the forms:
p .L are a 1a atlc mvanants
where 1-t (ewsa 2!2c) is the magnetic moment of the current loop of the particle
in orbit. If there arc only static magnetic fields present, the speed of the particle
is constant and its total energy docs not change. Then the magnetic moment JL
is itself an adiabatic invariant. In time-varying fields or with static electric fields,
g is an adiabatic invariant only in the nonrclativistic limit.
Let us now consider a simple situation in which a static magnetic field B acts
/ r igure J.~.'l
mainly in the z direction, but has a small positive gradient in that direction. Figme
12.4 shows the general behavior of the lines of force. In addition to the z com-
ponent of field there is a small radial component due to the curvature of the lines
of force. For simplicity we assume cylindrical symmetry. Suppose that a particle
is spiraling around the z axis in an orbit of small radius with a transverse velocity
vi0 and a component of velocity t• 110 parallel to B at ;; - D, where the axial field
strength is B 0 • The speed of the particle is constant so that any position along the
Z aXlS
VII
z + vz.L -_ vz() (12.70)
where v5 = uf0 + VITo is the square of the speed at z = 0. If we assume that the
flux linked is a constant of motion, then (12.69) allows us to write
(12.71)
'Nhere B is the axial magnetic induction. Then we find the parallel velocity at any
position along the z axis given by
2 B(z)
v.LO Bo (12.72)
Equation (12. 72) for the velocity of the particle in the z direction is equivalent
to the first integral of Newton's equation of motion for a particle in a one-
dimensional potential*
V(z)
If B(z) increases enough, eventually the right hand side of (12 72) will vanish at
some point z = z0 . This means that the particle spirals in an ever-tighter orbit
along the hnes of force, convertmg more and more translatwnal energy mto
energy of rotation, until its axial velocity vanishes. Then it turns around, still
sp1ra!mg m the same sense, and moves back m the negative z direction. The
particle is reflected by the magnetic field, as is shown schematically in Fig 12 5
Equation (12.72) is a consequence of the assumption that pf!B is invariant.
To show that at least to first order this invariance follows directly from the
Lorentz force equation, we consider an explicit solution of the equations of mo
*Note, however, that our discussion is fully relativistic. The analogy with one-dimensional nonrela-
tivistic mechanics is only a furmal one.
Sect. 12.5 Adiabatic Invariance of Flux Through Orbit of Particle 595
....
-- ...........
--,...--.....__
7 "'\?' "
I
l
I
z
- \
e:::-----
~
_.. I Figure 12.5 Reflection of charged

I uarticle out of reoion of hioh field
•v
strength.
uon. If the magnetlc mductwn along the ax1s 1s B(z). there wlll be a radial com-
ponent of the field near the axis given by the divergence equation as
(12.73)
where pis the radius out from the axis. The z component of the force equation
lS
.. e . e . aB( z)
z =- (-pcpB) = - - p2 cp-- (12.74)
mc r 2 me iJz
where (f> is the angular velocity around the z axis. This can be written, correct to
first ordt~r in tht~ small variation of B (z), as
z.. ~ -v~o
-- iJB(z)
- (12.75)
2B 0 dZ
where we have used p 2 (f> = -(a 2 w8 ) 0 = -(v~ 0 /w 80 ). Equation (12.75) has as its
first integral (12.72), showing that to first order in small quantities the constancy
of flux linking the orbit follows directly from the equations of motion.
The adiabatic invariance of the flux linking an oroit is useful in discussmg
particle motions in all types of spatially varying magnetic fields. The simple ex-
ample described above illustrates the principle of the "magnetic mirror":
Charged particles are reflected by regions of strong magnetic field. This mirror
property formed the basis of a theory of Fermi for the acceleration of cosmic-
ray particles to very high energies in interstellar space by collisions with moving
magnetic clouds. The mirror principle can be applied to the containment of a hot
plasma for thermonuclear energy pwduction. A magnetic bottle can be con
structed with an axial field produced by solenoidal windings over some region of
space. and additional coils at each end to provide a much higher field toward the
ends. The lines of force might appear as shown in Fig. 12.6. Particles created or
injected into the field in the central region will spiral along the axis, but will be
reflected by the magnetic m1rrors at each end. 11 the ratio of maximum field Bm
in the mirror to the field B in the central region is very large, only particles with
a very large component of velocity parallel to the axis can penetrate through the
ends. From (12.72) it is evident that the criterion for trapping is
"mirror" machine for the

B containment of a hot plasma.
If the particles are injected into the apparatus, it is easy to satisfy requirement
Then the of 1s the rate at which are
scattered by residual gas atoms, etc., in such a way that their components
violate
Another area ese 1s to
magnetic fields. The motion of charged particles in the magnetic dipole fields of
sun or can
here and the drift velocities of Section 12.4. Some aspects of this topic are left to
belts).
12.6 Lowest Order Relativistic Corrections to the Lagrangian
we are
particle in external electromagnetic fields described by the vector and scalar po-
If we now consider the problem of a conventional Lagrangian description of the
possible only at nonrelativistic velocities. The Lagrangian is supposed to be a
the finite velocity of propagation of electromagnetic fields is taken into account,
the other particles depend on their state of motion at "retarded" times. Only
when the retardation effects can be neglected is a Lagrangian description in terms
of instantaneous and velocities In view of this one might think
that a Lagrangian static , 1.e., to
in (vic). We will now show, however, that lowest order relativistic corrections can
an
to the order of (u/c) 2 inclusive.
Seet. 12.6 Dar'\\ in Lagrangian 597
If attention is directed to the first particle, this can be viewed as the negative of
the product of q 1 and the scalar potential <1> 12 due to the second particle at the
position of the first. This is of the same form as (12.9). If we wish to generalize
beyond the static limit, we must, according to (12.11 ), determine both <1> 12 and
A 12 , at least to some degree of approximation. In general there will be relativistic
corrections to both <1> 12 and A 12 . But in the Coulomb gauge, the scalar potential
is given correctly to all orders in vic by the instantaneous Coulomb potential.
Thus, if we calculate in that gauge, the scalar potential contribYtion C!> 12 is already
known. All that needs to be conside1 ed is the vector potential A 12 .
If only the lowest order relativistic corrections are desired, retardation effects
can be neglected in computing A 12 • The reason is that the vector potential enters
the Lagrangian (12.11) in the combination q 1 (v1 /c) · A 12 . Since A 12 itself is of the
order of u27c, greater accuracy in calculating A 12 is unnecessary. Consequently,
we have the magnetostatic expression
~
A12-
!
C
IIX1 - X
3
Jlx') d x'
'I (12.78)
where J, is the transverse part of the current due to the second particle as dis-
cussed in Section 6.3. From equations (6.24) (6.28) it can be shown that the
transverse current is
(12.79)
When this is inserted in (12.78), the first term can be integrated immedmtely.
Thus
By changing variables to ! x' x2 and integrating by parts, the integral can

be put in the form,
A12 = q2v2 - !12_ V

cr 41Tc '
I v2 • y
l
1
IY - rl
d3y
The integral can now be done in a straightforward manner to yield
A12 = ~2 r~ - !v,( v2 r. r) l
The differentiation of the second term leads to the final result
(12.80)
With expression (12.80) for the vector potential due to the second particle
at the position of the first, the interaction Lagrangian for two charged particles,
including lowest order relativistic effects. is
L = q1qz (12.81)
ml
r
This interaction form was first obtained by Darwin in 1920 It is of importance
in a quantum-mecltanical discussion of relativistic corrections in two-electron
atoms. In the quantum-mechanical problem the velocity vectors are replaced by

their corresponding quantum-mechanical operators (Dirac a's). Then the inter-
action is known as the Breit interaction (1930).* '
For a system of interacting charged particles the complete Darwin
Lagrangian, correct to order l/c2 inclusive, can be written down by expanding
the free-particle I agrangian (12 7) for each particle and summing up all the in
teraetion terms of the £01111 (12.81). The result is
LDacwin -
- -21 L-
""' m,v,2 + 81 2 ""' 4 1 ""'' q,q,
L- m;V; - -2 L- - ..
, c , '" rlJ (12.82)
+ 12 L-
""' ' -q ,q; [V; • v + (V; • r,
1
' ) (v • r,
1 1
' )]·
1
4c i,J rii
where r,1 [x; x;[, r;; is a unit vector in the drrecnon x, X;. and the pnme on
the double summation indicates the omission of the (self-energy) terms, i = j.
Although the Darwin Lagrangian has had its most celebrated application in the
quantum-mechanical context of the Breit interaction, it has uses in the purely
classical domain. Two examples are cited in the suggested reading at the end of
the chapter. See also the problems.
12.7 Lagrangian for the Electromagnetic Field

In Section 12.1 we considered the Lagrangian formulation of the equations of
motion of a charged particle in an external electromagnetic field. We now ex-
amine a Lagrangian description of the electromagnetic field in inttlraction with
specified external sources of charge and current. The Lagrangian approach to
continuous fields closely parallels the techniques used for discrete point particles. t
The finite number of coordinates q;(t) and q,(t), i = 1, 2, ... , n, are replaced by
an infinite number of degrees of freedom. Each point in space-time xa corre-
sponds to a finite number of values of the discrete index i. The generalized co-
ordinate q, is replaced by a continuous field ch(x), with a discrete index (k = 1,
2, , rl) and a continuous index (x"). The generalized velocity(], is replaced by
the 4 vector gradient, ;J'"ifi". The Euler Lagrange equations follow from the sta-
tionary property of the action integral with respect to variations 8c!>k and 8(af3tj>k)
around the physical values. We thus have the following correspondences:
L = L L,(q,, q,) ~ J:£( 4>ko a"tj>") d x

3
'
£~( (:~)
aL
aq,
*See H A Betbe and F F Salpetcr_ Qua~:~ tum Mechanics of One- af-:ld Two !ileGtronAtems, Springer-
Verlag, Berlin; Academic Press, New York (1957), pp. 170 ff.
For more detail and or background than given in our abbreviated account, see Goldstein (Chapter
"
1
12) or other references cited at the end of the chapter.

~ ........
~ect. U.c/ Lagrangian lOr toe J;IeCirOmagneuc NeiO ;;)77
wnere :x ts a Lagrang1an aensuy, corresponaing to a ueunite point in space-Lime

and eouivalent to the individual terms in a discrete oarticle Lagran!':ian like
-;-.
{17 Q')\ J;'nr thP -' ·~~r>Pti~ fipjrJ thP " tP.'" ~nrl "VP.cloritiP.'" ~rP. Aa
' . ~,(
dllU U r1 ,
The actwn mtegral takes the torm,
A = JJ9:, d x dt = J9:, d x
3 4
(12.84)
The Lorentz-invariant nature of the action is preserved provided the Lagrangian

density 9:, is a Lorentz scalar (because the four-dimensional volume element is
invariant). In analogy with the situation with discrete particles, we expect the
free-field Lagrangian at least to be quadratic in the velocities, that is, af3A a or
F" 13 . The only Lorentz-invariant quadratic forms are Fa~" 13 and F" 13',!faf3 (see
Problem 11.14). The latter is a scalar under proper Lorentz transformations, but
a pseudoscalar under inversion. If we demand a scalar 9:, under inversions as well
as proper Lorentz transformations, we must have ;£fcee as some multiple of
F" 13paf3. The interaction term in 9:, involves the source densities. These are de-
scribed by the current density 4-vector, J"(x). From the form of the electrostatic
and magnetostatic energies, or from the charged-particle interaction Lagrangian
(12.10), we anticipate that 9:,; 01 is a multiple of J aA a. With this motivation we
postulate the electromagnetic Lagrangian density:
1 1
;£= - - F paf3 - - J A a (12.85)
167T af3 C a
The coefficient and sign of the interaction terms is chosen to agree with (12.1 0);
the sign and scale of the free Lagrangian is set by the definitions of the field
strengths and the Maxwell equations.
In order to use the Euler-Lagrange equation in the form given in (12.83),
we substitute the definition of the fields and write
7TgA 0 gv,.(a"'A"- a"A")(aAAv- iJ"AA) - ~ J A"

1
9:, = - (12.86)
16 c "
In calculating a;£!a(a 13Aa) care must be taken to pick up all the terms. There are
four different terms, as can be seen from the following explicit calculation:
iJ;£ 1 { 8 I' 8 upAv- 8 "8 "FA"}
a(a 13Aa) = - 167T gA"g"" + 8:' 8:"F""- 8:" 8:AF""'
Because of the symmetry of g" 13 and the antisymmetry of F" 13 , all four terms are
equal and the derivative becomes
iJ;£ 1 1
of oB a\ -- --;- Ff3a = -;- Faf3 (12.87)
'\. I
"r], o+h
r
·+ ,f +h "'-·' T
'b' 'b'
n~nn+;~n
'1'
;o
A::f 1
]lL.M)
art c J"
c ,. ,, <"~1..1
~·c oL
U<UO LH'-' '-''i' V< <HVUVH V< Cllv~ ' -~-~
1 1
- a13F"c - J (12.89)
47T c
ouu Chapter IL uynamtcs ot Kelativtsttc l'arttcles and Electromagnetic Nelds -lj
These are recognized as a covariant form of the inhomogeneous Maxwell equa-

LlOnS -~ LL ~ '+1).
The Lagrangian (12.85) yields the inhomogeneous Maxwell equations, but
llU l lll<O ~ Vll<OO, Hll> I>
. . lll<O
..
lUVH V~ lHv
~
llv~U >ll t:llg{Q
tensor F"' 13 in terms of the 4-vector ootential A A was chosen so that the homo-
geneous equations were satisfied automatically (see Section 6.2). To see this in
our present 4-tensor notation, consider the left-hand side of the homogeneous
equations (11.142):
aa ?Ji"'13 = !2 ao: Eaf3A!LF li.IL
_ :1 &af3A!L :1 A
-u ~- -,.
= E"13A'"a a A
But the differential operator iJ"'iJA is symmetric in a and A (assuming A,. is well
. .a; . . . - .
l), Ill U dllU Jl. UlU' lllC CUlllldCUUll Ull U llllQ
,\ ·-'les_ The • " 1eous l\A'. ·11 eouations are satisfied triviallv.
t
The conservation of the source current density can be obtained from (12.89)
bv taking the 4-divergence of both sides:
1 1
iJ"'iJPFf3a- a"'J"
47T c
The left-hand side has a differential operator that ts symmetnc m a and {3, while
l:' •• . • A • '1- ,,; • ,.1 I.
f3a • .b.
a"Ja = 0 (12.90)
12.8 Proca Lagrangian; Photon Mass Effects

The conventional Maxwell equations and the Lagrangian (12.85) are based on
the hypothesis that the photon has zero mass. As discussed in the Introduction,
it can always be asked what evidence there is for the masslessness of the photon
or equivalently for the inverse square law of the Coulomb force and what con-
0 • • • • '
O<Oljl \'VVU~U HO>Ull HV~ll a HUll' lHd>O. rlo o~ ~V~
such considerations is the Lagrangian formulation. We modify the Lagrangian

density (12.85) by adding a "mass" term. The resulting Lagrangian is known as
the Proca Lagrangian Proca having been the first to consider it (1930, 1936). The
Proca Lagrangian is
2
'
5£Proca = -~ F F''f3 +!::... A A"'-~ J A"'
11'1,. af3 )<,. "' r a
(12.91)
The parameter f.L has dimensions of inverse length and is the reciprocal Compton
waveJengtn or me pnoton UL myctn). msteaa ot ~lLIS'J), me Yroca equanons
nf ' ~rP
~{3 l:' 2,1 47T T (1 'J O'J\

I' -,
P"'
" c " '
with the same homogeneous equations, a"?Ji"'f3 = 0, as in the Maxwell theory. We
observe that in the Proca equations the potentials as well as the fields enter. In
Sect. 12.8 Proca Lagrangian; Photon Mass Effects 601
contrast to the Maxwell equations, the potentials acquire real physical ( observ-
able) significance through the mass term. In the Lorenz gauge, now required by
current conservation, (12.92) can be written
(12.93)
and in the static limit takes the form
If the source is a point charge q at rest at the origin, only the time component
. . .
flo <v ts nonvamsnmg. u raKes rne spn' LV sy: "'- , uAawa 1uHH
e-tu
<l>(x) = q ~ (12.94)
r
This shows the characteristic feature of the photon mass. There is an exponential
falloff of the static potentials and fields, with the lie distance equal to 1-t 1. As
->: _,. •'- T- ~- • ~ -'- • n . . 1-1 ''"''" ,1-, ·,1<-
alters the ch~~acter of the earth's magnetic field sufficie~tly to permit us to set
quite stringent limits on the photon mass from geomagnetic data. It was at one
timo 011 "* th•>t rolo<' -1-
"J
oimnlo lo
·-r -J
. ltO . '· Tr
circuits could improve on even these limits, but the idea was conceptually flawed.
There is enough subtlety involved that the subject is worth a brief discussion.t
The st~rtin" noint of the ar"ument is (12.9ill in the absence of sources. If we
assume harmonic time and space variation, the constraint equation on the fre-
quency and wave number is
2 2• 2 2 2
This is the standard expression for the square of the energy (divided by h) for a
panicle or momcumm nK ana mass ,:crm::~ 'umc H'Y""'H
(cavity or lumped circuit). Suppose that when f.-t = 0 its resonant frequency is w0 ,
*
while for 1-t 0 the resonant frequency is w. From the structure of (12.95) it is
. • •'- -1 •
(12.96)
' -,- -J ·o
wand w0 for a given photon mass. This suggests an experiment with lumped LC
circuits. The scheme would be to measure the resonant frequencies of a sequence
of circuits whose w 2 values are in known ratios. If the observed resonant fre-
quencies are not in the same proportion, evidence for 1-t 0 in (12.96) would be *
found. Franken and Ampulski compared two circuits, one with a certain indue-
tance L and a capacttance C, hence wtth w0 - ( LCT , ana anotner wttn tne same
inductor, but two capacitances C in parallel. The squares of the observed fre-
'P. A. rranken and li. w. AmpulskJ, f'hys."Kev~!lJ.

'Shortly after the idea was proposed, several analyses based on the Proca equations appeared. Some
of these arc A. S. Goldhaber and M. M. Nieto, Phvs. Rev. Lett. 26, 1390 (1971 ); D. Park and E. R.
Williams Phvs. Rev. Lett. 26. 1393 (1971): N. M. Kroll Phvs. Rev. Lett. 26, 1395 (1971); D. G.
Boulware, Phys. Rev. Lett. 27, 55 (1971): N. M. Kroll, Phys. Rev. Lett. 27, 340 (1971).
602 Chapter 12 Dynamics of Relativistic Particles and Electromagnetic F1elds-G
quencies, corrected for resistive effects, were in the ratio 2: 1 within errors. They
thus inferred an upper limit on the photon mass, pointing out that in principle
What is wrong with the idea? The first observation is that lumped circuits
are by definition incapable of setting any limit on the photon mass.* The lumped
ctrcmt concept of a capacttance ts a two-termmal box wtth the property that the
current flow I at one terminal and the voltage Vbetween the terminals are related
by 1 C dV!dt. Similarly a lumped inductance is a two terminal box with the
governing equation V = - L dl/dt. When two such boxes are connected, the
cunents and voltages aJ e necessarily equal, and the combined system is desc1ibed
by the equation, V + LC d 2 VIdt 2 = 0. The resonant frequency of a lumped LC
eiieUJt 1s w0 (LC)- 112 , period.
It is true, of course, that a given set of conducting surfaces or a given coil of
wire will have different static properties of capacitance or inductance depending
on whether JL = 0. The potentials and fields are all modified by exponential
factors of the general form of (12 94) The question then arises as to whether one
can set a meaningful limit on JL by means of a "tabletop" experiment, that is, an
experiment not with lumped-circuit elements but with ones whose stzes are mod-
est. The reader can verify, for example, that for a solid conducting sphere of
radms a at the center of a hollow conductmg shell of mner radms b held at zero
potential, the capitance is increased by an amount JL 2a 2 b/3, provided JLb << 1. It
then turns out that instead of the fractional difference,
that follows from (12.96) with w~ = (LC)- 1 , the actual effect of the finite photon
mass IS
(12.98)
where d is a dimension characteristic of the circuit and '''e is the resonant fre-
quency for JL = 0. This makes a "tabletop" experiment possible in principle, but
very insensitive in practice to a possible photon mass.
Although the estimate (12.98) says it all, it is of interest to consider the effect
of a finite photon mass for transmission lines, waveguides, or resonant cavities.
For transmission lines, the effect of the photon mass is the same as for static
lumped-circuit parameters. We recall from Chapter 8 that for JL = 0 the TEM
modes of a transmission line are degenerate modes, with propagation at a phase
velocity equal to the velocity of light. The situation does not alter if JL 0. The *
only difference is that the transverse behavior of the fields is governed by
(V; - JL2 )1/J = 0 instead of the Laplace equation. The capacitance and inductance
per umt length of the transmiSSion !me are altered by fractional amounts ol order
JL2 d 2 , but nothing else. (The result of Problem 5.29 still holds.)
For IE and TM modes m a wavegmde the situation is more complicated.
*I am mdebted to E. M. Purcell for emphastzmg that this is the point almost universally missed 01 at
least glossed over in discussions of the Franken-Ampulski proposal.
Sect. 12.9 Effective "Photon" Mass in Superconductivity; London Penetration Depth 603
The boundary conditions on fields and potentials must be considered with care.
Analysis shows (see Kroll, op. cit.) that TM modes have propagation govemed
by the naive equation (12.96), but that TE modes generally propagate differently.
In any event, since the cutoff frequency of a guide is determined by its lateral
dtmenswns, the generally mcorrect estJmate (12.97) becomes the same as the
proper estimate (12.98).
For resonant cavities a rigorous solution is complicated, but for small mass
some simple results emerge. For example, for a rectangular cavity, (12.96) holds
to a good approximation for modes with l, m, n all different from zero, but fails
If any mode number ts zero. Th1s 1s because the fields behave m the dtrectwn
associated with vanishing l, m, or n as static fields and the arg11ments already
made apply. The low-frequency modes (Schumann resonances) of the earth-
ionosphere cavity, discussed in Section 8.9, are of particular interest. These modes
have a 1adial electiic field and to the zeroth order in hlR, where It is the height
of the ionosphere and R the radius of the earth, are TEM modes in a parallel
plate geometry. Thus their propagation, hence resonant frequencies, arc unal-
tered from their p., = 0 values. To first order in h!R there is a mass-dependent
change in resonant frequency. The result (see Kroll's second paper cited above)
ts that (12.97) ts modtfied on its right-hand side by a multiplicative factor g
0 44 hiR for the lowest Schumann mode With h - 70 km, g - 5 x 10- 3 Ibis
means that the resonant frequency of Wo = 50 s- 1 is a factor of (1/g) 112 = 14less
effective in setting a limit on the photon mass than naive considerations imply.
12.9 Effective "Photon" Mass in Superconductivity;

London Penetration Depth
A counterpart of Proca electrodynamics is found in the London theory of the
eleetromagnetie behavior of superconductors, formulated to explain the Meissner
effect. The Meissner effect (1933) is the expulsion of a magnetic field from the
interior of a superconducting material as it makes a transition from the normal
state (T > Tc) to the superconducting state (T < Tc). If the field is applied after
the material is superconducting, it does not penetrate into the sample, or rather
it penetrates a very small distance called the London penetration depth AL (typ-
ically a few tens of nanometers). Rather than being a perfect conductor, a
superconductor is perfectly diamagnetic. It is this phenomenon, which is a con-
sequence of an effective "photon" mass for fields within a superconductor, that
We begin a simple phenomenological discussion by assuming that the current

flow within a superconductor is caused by the nonrelativistic motion of charge
carriers of charge Q, effective mass mQ, and density "Q· If the average local
velocity of these carriers is v, the current density is
In the presence of electromagnetic fields the current can be expressed in terms

oi the canomcal momentum P through (12.14), P mQv + QA/c:
A
664 Chapter 12 Dynamics of Relativistic Particles and Electromagnetic Fields G
The superconducting state is a coherent state of the charge carriers with vanishing
canonical momentum. (P = 0 was an assumption by the Londons, but now has
a firm quantum-mechanical foundation~see Kittel, Chapter 12.) The effective
current density within a superconductor is therefore
With this current density inserted in the Lorenz-gauge wave equation for A
[(6.16), but in Gaussian units], the wave equation takes the Proca form (12.93),
but with no source term:
where f.J.. 2 47rQ 2n 0 Jm 0 c 2 . It follows from (12.99) that the boundary condltwn
on A at an interface between normal and superconducting media across which
no current is flowing is that the normal component of A vanishes. In the static
limit and planar geometry, the solution of the London equation akin to (12.94)
is A oc e'-ILX, showing that the London penetration depth is AL = f.J..- 1 :
/ m Q cz
' -
"L - (12.1 00)
'1/ 47rQ 2 n 0
'fltee ff ecttve
. " p1:oton
I . t Jeft
" n1ass ISIIZy ,Fl/tLc.Ince
_ '7 S. t Ire cIratge earners
. are
surely related to electrons in the material, we express the charge Q in units of e,
the protonic charge, the mass m 0 in units of me, the electronic mass, and write
the density of carriers in units of the inverse Bohr radius cubed. Then the rest
energy of the "photon" can be written
2
(my)effC = r;
The dimensionless quantity in square brackets is presumably of order unity. The
rest energy of the "photon" is thus of the order of the Rydberg energy, that Is,
a few electron volts.
Experimentally and theoretically, it is known that the charge carriers in low-
temperature superconductors are pairs of electrons loosely bound by a second-
order interaction through lattice phonons. Thus Q = -2e, m 0 = 2me, and
n 0 = nen/2, where neu is the effective number of electrons participating in the
current flow. ,11., useful formula is tt 2 8wr0 nQ, where ru 2.818 X 10- 15 m is
the classical electron radms. With n 0 - 0(1022 em 3 ) we find
A6 = f.L = 0( 4 X 10 em)
The BCS quantum-mechanical theory* shows that at zero temperature, n 0
nen/2 = 2Er,N(0)/3, where EF is the Fermi energy of the valence band and N(O)
is the density of states (number of states per unit energy of one electronic spin
state) at the Fermi surface. For a degenerate free Fermi gas, neff is equal to the
total density of electrons, but in a superconductor the density of states is modified
by the interactions and resulting energy gap. Using half the total number of
*I Bardccn, L N Cooper. J R Schridfer, PIJys. r~ev. lOS, 1175 (1957).

Sect. 12.10 • • • !>ntl . ....
' .. T • -n T nn•o (i;(ll:;
wr nQ iu ~LL.~VVJ yi~ OI!'J oraer-or-

'vu' _!:""_ Ullll
magnitude estimates for A~.. In passing we note that in high-temperature (cupric
oxide) superconductors penetration depths are found to be an order of magni-
tuae smaller tnan m conventional sunerconductors.
Measurements of AL, especially its temperature dependence, can be accom-
plished by incol])_oratir~g_ the superconducting specimen into a resonant circuit
and studying the shift in resonant frequency with change in temperature. In cir-
cumstances in which AL is small compared to both the wavelength A associated
\XIiththP l!'lnt' ' nrl<ho -•.o;?o ~o;mnl. ,].,]' (p -' .1"1"1{)\
paralleling Section 8.1 leads to a purely re~~tive ·;urface imped~~ce, . "'

877 2 A ?. '7T.\
Ls l A· ~ uauss1an umts) or Zs l · Z 0 (SI units)
c A
With onr Ilion ahon1_ tim" " .l. (p·iwt\ thP' .-hnrP ;o · '
• . - r . ,- ,. ' .
'
correspondmg to an mductance per umt area, L = MoAL (SI umts).
Our sketch of the simple London theory addresses only the Meissner effect,
.l • II . --'I'le '~ ' .,_ _,
~H ~' . - .> <H~ 1-"Ul-''-'"''-''
.-U" UUU ._, · VL T •
tors, the physical size of the coherent state (coherence length ~).and many other
J' 1'.,11, .J.J_ .l .1 .L ol. • • • ,:,
, 'J 'JU]'"~""~' '~'-j - llllCOUlJ.
1 ne
reaaer w1snmg to learn more about superconductivity may consult Ashcroft
and Mermin or Kittel and the numerous references cited there. An alternative,
ver11aps more p11ysicar, avvroacn (also oy t. 1 onaon) to tne 1 onaon eauations
is addressed in Problem 12.21.
12.10 Canonical and Symmetric Stress Tensors;

~
11..-
.
· vuttun Laws
A. Generalization of the Hamiltonian: Canonical Stress Tensor

In particle mechanics the transition to the Hamiltonian formulation and con-
servation of energy is made by first defining the canonical momentum variables
aL
p, = aq,
and then introducing the Hamiltonian
TT
"'
£.J J>i'-li
l
T
~
"~ ~""
~lL,J.U.L)
'
H
.' 1
'-'UH <H'-'H UO. 'llVVVll lHUl UH{U! V iJlV ULf(lt v. ru1 we;
.. HC
navmg a Hamlltonian density whose volume integral over three-dimensional
space is the Hamiltonian. The Lorentz transformation properties of 'iJe can be
guessea as IO!!ows._::,_mce the energy ot a particle IS the time comnonent of a
4-vector, the Hamiltonian should transform in the same way. Since H = f 'iJe d 3 x,
and the invariant 4-volume element is d 4 x = d 3 x dx 0 , it is necessarv that the
l-l!>miltnni"n .-1. "'f n <h ,; .. ,f .l • 1.
tensor. If the Lagra~gian density for some fields is a function of the field variables
cPk(x), a"'cf>k(x), k = 1, 2, ... , n, the Hamiltonian density is defined in analogy

·.c f1'"1 1A1\
\ 'I
\
oW
" ':'<:
"'f "Y '1' (171fl'"l\
~ a(a!k) at "I
The first factor in the sum is the field momentum canonically conjugate to cPk(x)
and acf>k/at is equivalent to the velocity q,. The inferred Lorentz transformation
properties of ':1e suggest that the covariant generalization of the Hamiltonian den-
. . •'- . .i .
OHJ <0 <HV
.rn
raf3 '\' Uoh
;,{3 A-. aaf3cp (17.1fl':l\
~ a( aacf>k) " -,
For the free electromagnetic field Lagrangian
1
:£em = - - - F F"""
l61T !LV
thP <'~Mo ofrP« • ~o~~ io
--;;cp
T"'f3 aflAA g"'P :£em
a(aaAA)
where a summation over A is implied by the repeated index. With the help of
(12.87) (but notice the placing of the indices!) we find
T"'f3 = - -} g"'""F""Aa 13A A - g"f3 :£em ( 12.104)
CT'. _, .". •• .c •'- '~ ;]-. >ntc ur;th

·~
X ~)<; Jr )!IS 'IT ana ~ Jl.UCS) we nna
Too = __!_ (Ez + Bz) + _1 V. (<PE)

81T 41T
o: 1 I (12.105)
T ' = 1T (E x B), + 1T V · (A,E)
4 4
"'
I I a
1 ' ~~ X -.J)i T ' \V A 'I'D )i
0
\ '¥Dj)
"T ·u
"'"
In writing the second terms here we have made use of the free-field equations
V · E - 0 and V X B - aE/ax0 - 0. If we suppose that the ilelds are localized m
some finite region of space (and, because of the finite velocity of propagation,
they always are), the integrals over all3-space at fixed time in some inertial frame
of the components T 00 and T 0 ; can be interpreted, as in Chapter 6, as the total
energy and c times the total momentum of the electromagnetic fields in that
fr~mP"
r 1 r
2
I T 00 d'r - I (F, + RL) d0 X - £""''
J i51T J ( 12.1 U!J)
JT 0
' d 3 x = - 1T
4
1
J(E x B) d x = cP; t
3
field
-~-
Sect. U.lU t:anomcal and symmetnc Stress 1 ensors; \.-Onservauon Laws UU/
These are the usual (Gaussian units) expressions for the total energy and mo-
mentum of the fields, discussed in Section 6.7. We note that the components T""
,.1 r0i <t nl• . <'• +> •< ,,-ln~,-1 ,.lnh~:+;r.nc r.f PnPr<m ·1 • ~nrl
·oJ -J
momentum density by added divergences. Upon integration over all space, how-
ever, the added terms g1ve no contributwn, bemg transrormea mto surrace m-
tegrals at infinity where all the fields and potentials are identically zero.
The connection of the time-time and time-space components of Taf3 with the
field energy and momentum densities suggests that there is a covariant general-
ization of the differential conservation law (6.108) of Poynting's theorem. This
~,u_ · 1l · 't~tPmPnt ;,
a r 13 = o (12.107)
In proving (12.107) we treat the general situation described by the tensor (12.103)
and the Euler-Lagrange equations (12.83). Consider
iJ:£ f3
aaT'" 13 = 2: aa ofo -1. \ a cPk
- af35£
\ • a "KJ
-
~ a5£ a5£ o.
a· .L
2_. cl a iJ (a cP ) · O'<f5k 'T iJ ( iJ cP ) ilafJ' <pk
k _ ak ak
By means of the equations of motion (12.83) the first term can be transformed
so that
a T" 13 = 2:k [ -_dcPk

a5£ a5£ a13 (a cf> ) J -
-_ a13 cf>k + a(iJacPk) a135£
a a k
Since 5£ = 5£( cf>k, iJ"cf>k), the square bracket, summed, is the expression for an
implicit differentiation( chain rule). Hence
aaT" 13 = a135£( cPko aacf>k) - o135£ = 0
The conservation law or continuity equation (12.1 07) yields the conservation
or total energy ana momentum upon mtegrauon over au 01 .J-space atuxeu Lime.
Exolicitlv. we have
r r {
u J i!"J ~,. a· x ao)i 7r X T J d;l . a·x
n me ne1as are JOcanzea me secona integral~ a aivergence J gives no c lULIUll.
Then with the identifications (12.106) we find

d d
dt £field = 0, dt Pfield = 0 (12.108)
In this derivation of the conservation of energy (Poynting's theorem) and

momentum and in the definitions (12.106) we have not exhibited manifest co-
. . . .
vanance. 1 ne resuns are vauu ror-:rrruu~c1 vta ar 1cM 111 LHc 11amc 111 lllC
fields are specified. But the question of transforming from one frame to another
has not been addressed. With a covariant differential conservation law, aaT" 13 = 0,
.... .1 • .• .• • .1. !1..1. .,.., • ... ,], •
~uv "': u v T u r • -b'
(12.106) do not appear to have the transformation properties of the components

of a 4-vector. For source-free fields they do in fact transform properly lsee Prob-
l<>m 1 'J 1 R ~nrl l)n/..vJ;~/., A ,r1;v ll. 1 <;\ h,< ;, >I cln nnt Tn :wnicl hmrino-
-r <: 7 --o
electromagnetic energy and momentum defined separately in each inertial frame,
without the customary connection between frames, one may construct explicitly
covariant integral expressions for the electromagnetic energy and momentum, 0 \
which the forms (12.106) are special cases, valid in only one reference frame.
This is discussed further in Chapter 16 in the context of the classical electroma _
netlc se -energy pro em. ee o r tc , ectlon - , or an exp ICitly covariant
treatment of the conservation laws in integral form.)
B. Symmetric Stress Tensor

The canonical stress tensor T" 13 , while adequate so far, has a certain number
of deficiencies. We have already seen that T 00 and T 01 differ from the usual
expressions for energy and momentum densities. Another drawback is its lack
of symmetry-see T 01 and T;o in (12.105). The question of symmetry arises when
we consider the angular momentum of tlte field,
Lfield
1
41TC
J x >< (E x B) tPx
The angular momentum density has a covariant generalization in terms of the

third-rank tensor,
(12.109)
I hen, JUSt as (12.10 I) 1mphes (12.108), so the vamshmg ol the 4-divergence
a M"13 Y = o
implies conservation of the total angular momentum of the field. Direct calcu-
lation of (12.110) gives
0
With (12.107) eliminating the first and third terms, we see that conservation of
angular momentum reqmres that J'"f5 be symmetnc. Two final cntJCJsms of T"/3,
(12.104), are that it involves the potentials explicitly, and so is not gauge invariant,
and that its trace (T~) is not zero, as reqmred for zero-mass photons.
There is a general procedure for constructing a symmetric, traceless,
gauge-invariant stress tensor ®"f3 from the canonical stress tensor T"f3 (see the
references at the end of the chapter). For the electromagnetic T"f3 of (12 104) we
proceed directly. We substitute af3AA = - pA/3 + aAA/3 and obtain
1 1
af3 - __':__
T - 4'77' (12.111)
The first terms in (12.111) are symmetric in a and f3 and gauge invariant. With
the help of the source free Maxwell equations, the last term can be written
T'ff
(12.112)
......t 11.111 r, • -• """I . . . ' J.!>WO llOQ
The tensor Tif has the following easily verified properties:

fi\ ,1 T"f3 - (l
,-,
(ii) JTfjfa d x = 0 3
Thus the differential conservation law (12.107) will hold for the difference
eraf3
\_
raf3\ '~ ·
U I
• ,lrl ~. raf3 u. ·••
' • •• "b'
'1 '1 ' ' . {1 ')
\ '
1{\t:\
"I
~.
the total energy and momentum of the fields will also be valid in terms of the
difference tensor. We are therefore free to define the symmetric stress tensor 0"13 :
('<)ap - Tap - T{J'
or
I \
<>all 1 auT' ..,AB 1 aB T' rnA ""' 111\

OJ
/5 ' JLA' 4 15 ' ,.v \L'·L ~~
4,-
Explicit calculation gives the following components,

1
eoo = - ( E" + BL)
1'>7T'
0 1 (12.114)
0 ' = 7T' (E X B);
4
z8;JCE
1 1
(><)if = L
E;E; + B;B; - 2
+ B
2
)
. 47T'
The indices i and f refer to Cartesian components in 3-space. The tensor 0" 13 can
hP writtPn in ,]. ...,Mir m:>triv fnrm :>~
0"{3 =
____ ..'
cg ,
u •________
(12.115)
' ,(,U\
cg : . 1 ij
In (12.115) the time-time and time-space components are expressed as the energy
and momentum densities (6.106) and (6.118), now in Gaussian units, while the
space-space components (12.114) are seen to be JUSt the negative of the Maxwell
stress tensor (6.120) in Gaussian units, denoted here by T&M) to avoid confusion
with the canonical tensor T" 13 • The various other, covariant and mixed, forms of
the stress tensor are
u ..'' _______
______ -cg , u '' -cg
' ea{3-
- '
----.J--------
0af3 =
-c~> ! - T(M) CP" :
' T(M)
\ ' I \ ' I
I \
''
u cg
''
(<) a {3 = ----- -·---------
'
-cg: T~vr)
\ I
. . .
1ne wuerenua1 conservallouww
aa0"13 = o (12.116)
.C-tl\
..rru ~· ·~ - . ...,.-~ .. .
~
~ .. . -· • ~-LO
OUL •.
~
'J
embodies Poynting's theorem and conservation of momentum for free fields. Fnr
example, with {3 = 0 we have
•
I
r. ' r>aO
1 au ... <'
v UaV
"' "'
\ I
where S = c 2 g is the Poynting vector. This is the source-free form of (6.108).

~R-i
3
ag - ~ -a
o = a 0>"; = -' r<MJ
a- ,1t . ,1r. l]
J
a result equivalent to (6.121) in the absence of sources. The conservation of field

anPular momentum defined throw>h the tensor
M"f3Y = {><)"f3xY - 0"Yxf3 (12.117)
is assuTerruy-pz.TIOJ allllllTe symmecty <Jri>'T , as a1reaay aiscussea. 1 nere are
evidently other conserved quantities in addition to energy, momentum, and an-
gular momentum. The tensor M 013 Yhas three time-space components in addition
to the snace snace comnonents that "ive the an"ular momentum densitv ThP"'
three components are a necessary adjunct of the covariant generalization of an-
gular momentum. Their conservation is a statement on the center of mass motion
{ ·~ 1 '1 1(\\
,...,.
"
r r. .•. T nw~ 1'~. IN .•. .... .. ,_, • •n
'J ·o ·o
with Charged Particles
In the nresence of external sources the La"ranPian for theM l erm~tinn'
is (12.85). The symmetric stress tensor for the electromagnetic field retains its
form (12.113), but the coupling to the source current makes its divergence non-
"~ . ·'-. cr~~ ~nlm.ln+:~- ~· .t:.~ .-1' • o+•n' .-].
~ -o 'b
. f3
()a(")"
1
= -:;-- rJIL(FILJAf3) + ~ a13 (FILAF~'-A)
1
-- - (a~'-F
!LA
)F'f3 + F!LA u~'-Pf3 + -21 F!LA af3FILA
47T
The first term can be transformed by means of the inhomogeneous Maxwell

equations (12.89). Transferring this term to the left-hand side, we have
I 1
aa 0"13 +- F 13 AJA = -8 F /LA (a~'-FAf3 + iJ~'-F' 13 + a 13 F~'-')
c 7T
The reason for the peculiar grouping of terms is that the underlined sum can be
replaced, by virtue of the homogeneous Maxwell equation (a~'-pA/3 + af3ptLA +
d1 -ur.ov 'fFFP"' +~. Tllus we ootam
aa €-)"13 + -1 F 13AJA = -1 F!LA (a~'-FA(3 + aAF~'-13)
c 8 7T
Sect. 12.10 Canonical and Symmetric Stress Tensors; Conservation Laws 611
But the right-hand side is now the contraction (in f.L and A) of one symmetric and
one antisymmetric factor. The result is therefore zero. The divergence of the
stress tensor is thus
(12.118)
The time and space components of this equation are

1 au ' 1
c at ' v • "', c
.. ·~ ~ lL.ll'J)
and
1
pE; + - (J X B); (12.120)
c
'J ' U U V H V< W UHU H>VU ''1 'V< \..-llctplCJ tJ
for electromagnetic fields interacting with sources described by J" = (cp, J). The
negative of the 4 vector on the right-hand side of (12.118) is called the L~vn"'~
force density,
{ 13
1
=- pt3AJ, =
(1- J · E, pE + -1 J X B \ (12.121)
r.; \ r.; c
If the sources are a number of charged particles, the volume integral of f 13 leads
mrougn me Lorentz rorce equation ~lL.lJ to tne tlme rate of change of the sum
of the energies or the momenta of all particles:
J J ax dt
With the qualification expressed at the end of Section 12.10.A concerning co-
. .
Vd> , lllC 'b>dl UVCJ :J-_r illllXCU lllllC Ul lllC Jert-nanQ SlOe Of ~1L.II15)
is the time rate of change of the total energy or momentum of the field. We
therefore h>~vP thP con ion of 4-m- 1h1m for th" ""'m'· · -' r.f
(1? 1??)
J at
The discussion above focused on the electromagnetic field, with charged par-
ticles onlv mentioned as the sources of the 4-current densitv. A more enuitable
treatment of a combined system of particles and fields involves a Lagrangian
having three terms, a free-field Lagrangian, a free-particle Lagrangian, and an
interacTion Lagrangian mat invOlves oom new ana particle degrees ot lreedom.
Variation of the action integral with respect to the particle coordinates leads to
thP T fr.rr-P Pnn~tir.n ind ~<in I<Pr-tir.n 1? 1 . ..,]..;]~ »n-:n+' ·- ~< +~ f:nL-1
-~_ ' .J • '
" 1. "
'"{;U\.HUIItal<O' !;lVC' lll<O lVWAWCU <O'{UdllUll,, ' " Ill -rz:"l.--:rnJWCVCl,
self-energy and radiation reaction effects are included, the treatment is not quite
so strai!!htforward. References to these .• >lrP . M thP Pnr1 ni thP
--r
Mention should also be made of the action-at-a-distance approach associated
1'>1?. r'l . I~
. : of Relativistic Particles and Electromal!netic Fields G
· c" ' ' ., ' ~ ' ...1 D.,l,l. rrt _,.

Wllfl Ute V> ->l-UVV , ~ couvuco, "~ • · vu cue
rh ~
narticles and an invariant action orinciple is postulated with the inter_
action term involving integrals over the world lines of all the particles. The idea
of electroma2:netic fields and the Maxwell equations is secondary. This approach
is the basis of the Wheeler-Feynman absorber theory of radiation.*
12.11 Solution of the Wave Equation in Covariant Form;

Invariant Green Functions
The electromagnetic fields paf3 arising from an external source r(x) satisfy the
mnomogeneou, 1vtctxwe11 "Ll'
.
u,•
~~n
47T '"
c
.. c •' . -' ' . r "- •• 1. •• • '
VV lUI Llle V< Ul'-- H'-'><!0 W Vl <V >
47T 13
OA 13 - a13(aaA") =- 1
c
If the potentials satisfy the Lorenz condition, aaA" = 0, they are then solutions
nf the four-dimensional wave eouation
4
OA 13 - 7T J13(x) (12.123)
(;
The solution of (12.123) can be accomolished by finding a Green function D(x, x')
for the equation
DxD(x, x') = o< 4l(x - x') (12.124)
wHere o·'"-~x
~ "} 0 <' M
A
n·
} » u
~-
•v
.... , '1 ..l ,] . ~-
A "\--'0 AO) "\'"-
In the absence of boundary surfaces, the Green function can depend only on the
• •·cr a a >arr!. nf. ' T)f. >\ nt.\~nrlf1'11~A\
"t' I<OUVl ~ ~ A • '\· , 1 \' I \'•/ 'I
becomes
UzD(z) - B'''(z)
We use Fourier integrals to transform from coordinate to wave number space.
The Fourier transform D(k) of the Green function is defined by
,. ., 4
D(z) = t'1 ~\ rd k D(k)e-ik
4
z (12.125)
_,_
' I.
o~o
,, " " ' ' +h ~t-
i,-,n ,-,f th~ rl~lb fnnrtir>n h~ina
"-
_;:(4){.\
1 { A4b- 0 -ik·z (1? 171i\
,~,
(27T)4 J
"'· .L +1-.n th~ I. r.r~~n ~- i,-,n ic
T
I {1 ?..127)
nriE\
,.
k·k
*J. A. Wheeler and R. P. Feynman, Rev. Mod. Phys. 21,425 (1949).

Sect. 12.11 Solution of the Wave Equation in Covariant Form; Invariant Green Functions 613
The Green function D(z) is therefore
D(z) (12.128)
Because the integrand in (12.128) is singulm, the expression as it stands is

ambiguous and is given definite meaning only by the handlmg of the smgulantJes.
'Ne proceed by performing the integration over dk 0 first. Thus
D(z) = - (277)4 ] d3k eik·z ]_m dko k~ - K2 (12.12,9)
where we have introduced the notation, K = Ik 1. The k 0 integral is given meaning

by considering k 0 as a complex variable and treating the integral as a contour
integral in the k 0 plane. The integrand has two simple poles, at k 0 = ± K as shown
in Fig. 12.7. Green functions that differ in their behavior are obtained by choosing
different contours of integration relative to the poles. Two possible contours are
labeled rand a in Fig. 12.7. These open contours may be closed at infinity with
a semicircle in the upper or lower half-plane, depending on the sign of Zo in the
exponential. For z0 > 0, the exponential, e-ikozo, increases without limit in the
upper half-plane. To use the residue theorem, we must therefore close the con-
tour in the lower half-plane. The opposite holds for z0 < 0.
Consider now the contour r. For Zo < 0, the resulting integral vanishes be-
cause the contour is closed in the upper half-plane and encircles no singularities.
For Zo > 0, the integral over k 0 is
The Green function (12.129) is then
The integration over the angles of k leads to
!S"~(
.1. ., I
dK sin(«R) sin(«zo) (12.130)
IL
------+- --+------r
-K K
-----+----------a
Figure 12.7
614 Chapter 12 Dynamics of Relativistic Particles and Electromaenetic Fields-G
where R = lzl = lx- x' I is the spatial distance between xa and x"' Using some
simple trigonometry and a change of variable, we can write (12.130) as
8( Zo) (oo d.
"' m
u,~Z)
87T2R J-oo UKL e e "j
· · i11Legra1s are jusLuirac ueHa mnctions. l::Secause Zo :> u and R > 0

:~:e ,,., int~<>r~l is ~~w~vs z~ro ThP. GrPPn £. inn fnr Hr rio th~'"'·
.co
n rr r')
8(xo - xb) M v
v' '{)\ {1'"1 11,
47TR \ ")
Here we have reintroduced the original variables x and x'. This Green function
is called the retarded or causal Green function because the source-point timex~
IS always earlier than the observation-point time x 0 • Equation (12.131), or its
D · '>1. lA :n\ 1 iwR/c · " ~ .,. ~ ; .
-, ·• 0 ,,. '"! • ,>Olll'-' '-'''-'ClHUHCLIOn
of outgoing waves of Chapter 6.
U/;th th~ ~hni~n nf thn ' , Tl' 1 '"I "7 " 11 _, _, .. 1 ~
""'' . ' ·'J ' VII ~lUll
yields the advanced Green function,
8[ -(x 0 - x6)] .S(xo

Da(x x') xb + R) (12.132)
"tiiH
1 nese vreen mncLwns can oe pm m covanant torm by use ot the lollowmg

identity:
8[(x - x'?J = 8[(xo - x6? - lx - x' n
= 8[(x0 - xb - R)(x0 - xb + R)]
1
2R LO~Xo Xo K) + li(Xo X 0 + l<)J
Then, since the theta functions select one or the other of the two terms, we have
DrCx - x') = ~ 8(x0 - x~) 8f(x - x'fl

L7T (12.133)
n f. f\ 1 CJ{ , \ Nl r\21
a I 27T ,.. u '0/ Jl \- ! J
The theta functions, apparently noninvariant, are actually invariant under proper
Lorentz transformations when constrained by the delta functions. Thus (12.133)
gives t!Ie Green functions an explicitly invariant expression. The theta and delta
; .
.J.Ull'-'llVllO . ''"
111 \.l.L • .J.JJ) OllUW llldl ULC
. . .
eu ~auvanceuJ vreen runcnon IS uu-
ferent from zero only on the forward (backward) light cone of the source point.
~1
L HUV UC ''-'" UUW11111 LC11110
'T'L ' .C<L ' "" '""' · '
>v ~•u>vnurv'-''1' C\Hoo.l.oCJj
of the Green functions:
Aa(x)- A~(x) + 4 7T { d 4 x' D{x- x')F'(x') (12.134)

c J
or
4'1T r
A~(X) -A~UI(X) + c J d x' Du(x x')r(x') (12.135)
Ch. 12 References 615
where At:; and A~ut arc solutions of the homogeneous wave equation. In (12.134)
the retarded Green function is used. In the limit x 0 --+ - oo, the integral over the
sources vanishes, assuming the sources are localized in space and time, bccansc
of the retarded nature of the Green function. We sec that the free-field potential
A~Jx) has the interpretation of the "incident" or "incoming" potential specified
at x 0 --+ - oo. Similarly, in (12.135) with the advanced Green function, the ho-
mogeneous solution A~m(x) is the asymptotic "outgoing" potential, specified at
x 0 --+ + oo. The radiation fields are defined as the difference between the "out-
going" and the "incoming" fields. Their 4-vcctor potential is
47T r
A~ad(x) - A~ut x')J"(x') (12.136)
where
O(z) - D, (z) - Da(z) (12.137)
is the difference between the retarded and advanced Green functions.
The fields of a charged parncle moving in a prescribed path will be of interest
in Chapter 14. If the particle is a point charge e whose position in the inertial
frame K is r(1), its cltatge density and cunent density in that frame are
p(x, t) = e 8[x - r(t)]
(12.138)
J(x, t) = e v(t) 8[x - r(t)]
where v(t) - dr(t)!dt IS the charge's vcloctty m K. I he charge and current den-
sities can be written as a 4-vector current in manifestly covariant form by intro-
ducmg the charge's 4-vcctor coordmatc r"( T) as a function of the charge's proper
time T and integrating over the proper time with an appropriate additional delta
l"(x) - ec Jr dT 0"( T) sC4l[x r( T)] (12.139)
where 0" 1s the charge's 4-veloclty. In the merttal frame K, r" [ct, r(t)] and
U" = ( yc, yv). The usc of (12.139) in (12.134) to yield the potentials and fields
of a moving charge is presented in Section 14.1.

The Lagrangian and Hamiltonian formalism for relativistic charged particles is
li eated in e~ ery ad~ anced mechanics textbook, a.; well as in books on electrodynamics.
Some useful references are
Corben and Stehle, Chapter 16

Goldstein, Chapter 7
Landau and Lifshitz, Classical Theory of Fields, Chapters 2 and 3
The motion of charged particles in external electromagnetic fields, especially inho-
mogeneous magnetic fields, is an increasingly important topic in geophysics, solar physics,
and thermonuclear research. The classic reference for these problems is
Alfven and Fi:ilthammar
but the basic results are also presented by

Lhandrasekhar, Lhapters7 afill3
l'l.,mmnw "nd Dou<>hertv Chanter 4
Linhart, Chapter 1
Spitzer, Chapter 1
The book by
Rossi and Olbert
covers the motwn ot charged parttcJes m the hetus 01 me eann anu sun. 1 ne ropic of
adiabatic invariants is treated thoroughly by
~·
"V'" "· JtJ
Another important application of relativistic charged-particle dynamics is to high-

energy accelerators. An mtroductwn to the physics prootems or mis ueta wiu oe round m
. " .' ' .c '- c.
•u
.>uuno u1 " ' " p1 ·r ao '<CU ao lll<e ua"" 1-'"J"' 111
Livingston and Blewett

~ ' H'~rror; onrl '<~nrP
ThP ' T onronn;on (17 Q')\ hoo lwPn

·rl "'" '""fnl tool in thf>
of low-frequency oscillations in plasmas by
A. N. Kaufman and P. S. Rostler, Phvs. Fluids 14, 446 (1971).
It has also been used in an interesting paper by
S. Coleman and J. H. Van Vleck, Phys. Rev. 171, 1370 (1968)
to resolve a paradox m the conservatiOn of momentum and J'iewton s thud Jaw.
The issue of "hidden momentum" addressed bv Coleman and Van Vleck is treated
in a number of pedagogical papers. A valuable early one is
W. H. Furry, Am. J. Phys. 37, 621 (1969)
More recent works include:
L. Vaidman, Am. .J. Phys. 58, 978 (1990).
v. nmzuu, .lim. J. rnys. o:., "" tl""' )·
V. Hnizdo, Am. J. Phys. 65, 92 (1997).
The subject of a Lagrangian description of fields is treated in
D.
, r>L TH
T

GoJrl,tPin l'h"nt"r 11
Low
Schwartz, Chapter 6
Soper
ThP n~nprol ~ondn•~t;on of thP c"~~~tr;~ no> - · mt 'In"' Aa~ from
ya~ is des~ribed in
Landau and Lifshitz C/assicaiTneorv Of Fields, SectiOns 3:2 and '14
The narticular ouestions of the interaction of chare:ed oarticles and electromae:netic
f;,,t,1, rl th .·,,,,1. ~ ;nlo>He OrP
Barut, Chapters IV and V

Landau and Lifshitz Classical Theorv o7 Fields Chapter 4
Schwartz, Chapter 7
and m an especially thorough taSTiionoy
Rohrlich
. ~
" n
WUV ll'-'QlO 111..00<011" auu CT
of the energy, momentum and mass of electromagnetic fields, consult Rohrlich's book,
ilNI ritPrl "nrl "JSO
F. Rohrlich, Am. .J. Phys. 38, 1310 (1970)
Ch. 12 Problems 617
The invariant Green functions for the wave equation are derived in almost any book
on qvantum field theory One such book with a concise covariant treatment of classical
eleetwdynamies at its beginning, is
I htrnng, Secflon l.2 and AppendiX II
Two reviews on the subject of the photon mass, already cited in the Introduction, are
A. S. Goldhaber and M. M. Nieto, Rev. Mod. Phys. 43, 277 (1971 ).
I. Yu. Kobzarev and L. B. Okun', Uspek. Fiz. Nauk. 95, 131 (1968) [English
transl., Sov. Phys. Uspek. 11, 338 (1968)].
Problems
12.1 (a) Show that the Lorentz invariant Lagrangian (in the sense of Section 12.1 B)
L = - mU.U"- q_ UaA"
2 c
gives the correct relativistic equations of motwn for a parttcle of mass m and
charge q interacting with an external field described by the 4-vector potential
A"(x).
(b) Define the canonical momenta and write out the effective Hamiltonian in
both covariant and space-time form. The effective Hamiltonian is a Lorentz
invariant. What is its value?
12.2 (a) Show from Hamilton's principle that Lagrangians that differ only by a total
flme denvaflve of some funcuon of the coordinates and time are equivalent
in the sense that they yield the same Euler-Lagrange equations of motion.
(b) Show explicitly that the gauge transformation A" ~A" + a"A of the poten-
tials in the charged-particle Lagrangian (12.12) merely generates another
equivalent Lagrangian.
12.3 A particle with mass m and charge e moves in a uniform, static, electric field E 0 .
(a) Solve f01 the velocity and position of the particle as explicit functiolls of time,
assummg that the tmttal veloctty v0 was perpendtcular to the electnc field.
(b) Eliminate the time to obtain the trajectory of the particle in space. Discuss
the shape of the path for short and long times (define "short" and "long"
times).
12.4 It is desired to make an E X B velocity selector with uniform, static, crossed,
electric and magnetic fields over a length L. If the entrance and exit sht wtdths are
Ax, discuss thtl inttlrval Au of vtllociti€s, around thco m<:an vain<: u cEIB, !h~t is
transmitted by the device as a function of the mass, the momentum or energy of
the incident particles, the field strengths, the length of the selector, and any other
relevant variables. Neglect fringing effects at the ends. Base your discussion on
the practical facts that L ~ few meters, Emax ~ 3 X 106 V/m, Ax ~ 10- 3 -10- 4 m,
u 0.5-0.995c. (It ts mstruchve to constder the equatwn of motton m a frame
moving at the mean speed 11 along the beam direction as well as in the laboratory)
References: C. A. Coombes et al., Phys. Rev. Il2, 13W (1958); P. Eberhard, M. L.
Good, and H. K. Ticho, Rev. Sci. Instrum. 31, 1054 (1960).
12.5 A particle of mass m and charge e moves in the laboratory in crossed, static,
uniform, electric and magnetic fields. E is parallel to the x axis; B is parallel to the
(a) For IE I < IB I make the necessary Lorentz transformation described in Sec-
tion 12..3 to obtain tlxplicitly param€tric coquations for thco particle'> trajectory
(b) Repeat the calculation of part a for IE I > IB 1.
U 6 Static, uniform electric and magnetic fields, E and B, mah an angle of e with
respect to each other.
(a) By a suitable choice of axes, solve the force equation for the motion ol a
particle of charge e and mass m in rectangular coordinates.
(b) For E and B parallel, show that with appropriate constants of integration
etc., the parametric solution can be written '
x = AR sin </>, y = AR cos </>, z =- Vl + A 2 cosh(p<f>)

p
ct = R V1 + A 2 sinh 
p
where R = (mc 2 /eB), p = (£/B), A is an arbitrary constant, and p is the

parameter [actually c/R times the proper time].
12.7 A constant uniform magnetic induction B in the negative z direction exists in a
region limited by the planes x = 0 and x = a. For x < 0 and x > a, there is no
magnetic induction.
(a) Determine the total electromagnetic momentum G in magnitude and direc-
tion of the combination of a particle with point charge q at (x0 , y0 , z0 ) in the
presence of thts magnettc mductwn. Ftnd G for the charge located on either
side of and within the region occupied by the magnetic field. Assume the
particle is at rest or in nomelati.istie motion.
(b) The particle is normally incident on the field region from x < 0 with nome-
lativistic motncntump. Asstuning thatp > qBafc, detertnine the coinponents
of momentum after the particle has emerged into the field-free region, x >a.
Compare the components of the sum of mechanical (particle) and e]ectro-
magnetic momenta initially and finally. Why are some components of the
sum conserved and some not?
(c) Assume that p < qBa/2c and that the initial conditions are such that the
particle's motion is confined within the region of the magnetic induction at
fixed z. Discuss the conservation, or lack of it, of the components of the sum
of mechanical and electromagnetic momentum as the particle moves in its
pa .
12.8 In Problems 6.5 and 6.6 a non vanishing momentum of the electromagnetic fields
was found for a charge and a current-carrymg tormd at rest. I fits paradox ts among
situations involving "hidden momentum." Since the field momentum is propor-
tional to llc 2 , you may infer that relati.isrie effeets ma} enter the considerations.
(a) Consider the charge carriers in the toroid (or other current-carrying systems)
of mass nz, charge e, and indnidualtneehanieal rnomentum p ynn, and
the current density J = env, where n is the number density of carriers and 'Y
is the relativistic Lorentz factor I Ise conservation of energy for each charge
carrier to show that, for the "static" field situation of Problem 6.5, the total
mechanical momentum of the charge carriers,
3 3
Pmech = { d x ynmv = - \ { d x <PJ
~ c '
just opposite to the field momentum of Problem 6.5a.
(b) Consider the toroid of Problem 6.6 to be of rectangular cross section, with
width w and height h both small compared to a, and hollow tubes of uniform
cross sectton Aw. Show that the electrostatic potential energy difference be-
Ch. 12 Problems 619
rween me inner ana omer verrtc:IT a> u1 em;u 1uue y•e•u• 1m: ~nange
in ymc 2 necessary to generate a net vertical mechanical momentum equal
and opposite to the result of Problem 6.6a, with due regard to differences in
units.
Reference: Vaidman, op. cit.
12.9 The magnetic field of the earth can be represented approximately by a magnetic
dipole of magnetic moment M = 8.1 X 1025 gauss-cm3 • Consider the motion of
• _, . d. . " .£d. d. ,-l. tl. . .<tl.' ,.,, ·'
•<' <O '6'
field (Van Allen electron belts). [Note that M points south.]
(a) Show that the equation for a line of magnetic force is r = r0 sin 2 0, where e
• tl. nl ,]. ,]. • ,-1,\ ,-l fmm thP OV;< nf thP cJ;nnJp
,.. ·o· ,. -r
and find an expression for the magnitude of B along any line of force as a
mncuou 01 tJ.
(hI A · · ·Jv -' · ].., rird"' ormmcl a line of force in the eauatorial
plane with a gyration radius a and a mean radius R (a << R). Show that the
. ' .
jJ<UH'-1'-' <lLHHUliUU r
.. ~ . . ' . . '· " .
·'·'J
in time according to
where w8 is the frequency of gyration at radius R.

(c) If, in addition to its circular motion of part b, the particle has a small com-
ponent of velocity parallel to the lines of force, show that it undergoes small
oscillations in e around e = 7rl2 with a frequency n = (31V"i)(a!R)w 8 . Find
. .
me cnange m wngnuae percycre-or "'
(oil F<1r on P' <1f 10 MPV ~c;nPt;r PnPr<JV M 0 mean radius R 3 X 107 m
find w 8 and a, and so determine how long it takes to drift once around the
... ' ' ' • . ' •• ·'" • • 1. • ,.,
'-UHH UHU H V IVH5 " <• ~J
Calculate the same quantities for an electron of 10 ke V at the same radius.

12.10 A charged particle finds itself instantaneously in the equatorial plane of the earth's
• '.I~ I. ,-l C •· ' <:~1~\ ~:. 1) f thP >nt.>r nf
t he eart h . Its velocity

·o \' · vector at that mstant
• rnak es an angle a w1t• h t h e equatonal
•
p1ane \u 1 tu~ ran a). Assuming mat me !'"•';".".'!-';""' a1vu),; w" J;He> uttuJ""
with a gyration radius a << R, and that the flux linked by the orbit is a constant
of the motion, find an equation for the maximum magnetic latitude A reached by
the particle as a function of the angle a. Plot a graph (not a sketch) of A versus a.
Mark parametrically along the curve the values of a for which a particle at radius
R in the equatorial plane will hit the earth (radius R 0 ) for RIR 0 = 1.2, 1.5, 2.0, 2.5,
3, 4, 5.
1?.11 r, •~ thP nrPo' nf thP <n;n nf 0 ffiUOn initia!lV lon<>itudinallV notarized
as the muon moves in a circular orbit in a plane perpendicular to a uniform mag-
. r '~
"""" ""'u ~.
(a) Show that the difference f1 of the spin precession frequency and the orbital
gyration frequency is
eBa
!1=-
;nrl mhPrP n (" 2)/2 is the ma<1netic

moment anomaly. (Find equations of motion for the components of spin
••• . ,; "' ' ,. • " • .,-l" '" • ' ·' •
the radius , ector from the center of the circle to the particle, and the magnetic
field,)
(b) For the CERN Muon Stm age Ring, the orbit radius is R 2.5 mete1s and
B = 17 X 103 gauss. What is the momentum of the muon? What is the time
dilatation factor y? How many periods of precession T = 21TI!1 occur per
observed laboratory mean lifetime of the muons? [m,. 105.66 MeV
To = 2,2 X 10- s, a = a/21T].
6
'
(e) Express the difference frequency !1 in units of the orbital rotation frequency
and compute how many precessional periods (at the difference frequency)
occur per rotation for a 300 MeV muon, a 300 MeV electron, a 5 GeV elec-
tron (this last typical of the e + e- storage ring at Cornell).
12.12 In Section 11.11 the BMT equation of motion for the spin of a particle of charge
e and a magnetic moment with an arbitrary g factor was obtained.
(a) Verify that (11.171) is the correct equation for the time derivative of the
longitudinal component of the rest-frame spin vectors.
(b) Let n be a unit 3-vector perpendicular to ~ and coplanar, with ~ and s (n is
generally time dependent). Let e be the angle between ~ and s. Show that
the time rate of change of e can be written as
-de = - e
dt me
(g )
[ - - 1 n·
2
A
(~ X
A
B) + - (g/32 - -f31) n · E
A ]
where E and B are the fields in the laboratory and c~ = cf3~ is the particle's
instantaneous velocity in the laboratory.
(c) For a particle moving undeflected through an E X B velocity selector and
with (n x ~) · B = B, find d&/dt in terms of the gyration frequency eB/ymc.
(d) By defining the two 4 ,ectors, L" ( r/3. 1~) and N" (0, B.), show that d(}f
dr can be written in the quasi-covariant form
ae
dr
where U" is the particle's 4 velocity.
12.13 (a) Specalize the Darwin Lagrangian (12.82) to the interaction of two charged
particles (m~o q 1 ) and (m 2 , q 2 ). Introduce reduced particle coordinates, r =
x1 x2 , v v1 v2 and also center of mass coordinates. Write out the
Lagrangian in the reference frame in which the velocity of the center of mass
vanishes and evaluate the canonical momentum components, Px &L/&Idx,
etc.
(b) Calculate the Hamiltonian to first order in l/c 2 and show that it is
2 ' \ 4 , \ I. 2 f.
r
)
Compare with the various terms in ( 42.1) of Bethe and Salpeter [op. cit.
(Section 12.6), p. 193]. Discuss the agreements and disagreements.
12.14 An alternative Lagrangian density for the electromagnetic field is
(a) Derive the Euler-Lagrange equations of motion. Are they the Maxwell equa-
tions? Under what assumptions?
Ch.12 Problems 621
(b) Show explicitly. and with what assumptions that this Lagrangian density
... fmm {1?

+<
5<~\ h\i " L[_ •
_, ~
DnP< thi< orlrlnrl L!.olivPr<JPnrP "ffprt
12.15 Consider the Proca equations for a localized steady-state distribution of current
that has only a static magnetic moment. This model can be used to study the
observable effects of a finite photon mass on the earth's magnetic field. Note that
if the magnetization is .M.(x) the current density can be written as J = c(V x.M.).
(a) Show that If .M, - m.f(x), where m IS a fixed vector and f(x) IS a localized
scalar function the vector potential is
A(x) = -m x V J f(x') le:"_'~·.·: d x' 3
(b) lf the magnetic dipole is a point dipole at the origin [.f(x) = o(x)], show that
the magnetic field away from the origin is
A A ,I fLY\e"" ? ?
e ,.u
D\AJ L-'•1• • IIIJ HIJ\ i
1-"' 31-"
3) r'
Ill
r
(c) The result of part b shows that at fixed r = R (on the surface of the earth),
the earth's magnetic field will appear as a dipole angular distribution, plus
an added constant magnetic field (an apparently external field) anti parallel
tom. Satellite and surface observations lead to the conclusion that this "ex-
rerna1 new IS tess man 4 x w : rimes me oip01e neto ar me magne1ic
equator. Estimate a lower limit on fL _,in earth radii and an upper limit on
the photon mass in grams from this datum.
This method of estimating fL is due to E. Schrodinger, Proc. R. Irish
Acad. A49, 135 (1943). See A. S. Goldhaber and M. M. Nieto. Phys. Rev.
Lett. 21, 567 (1968).
12.11\ (a) 0.
· '" with th" Proc" Laaranaian densitv f12.91) and followinP the same
f, tl. ,J. . h ,Jr!, -h tl. ,, th ' " oh •oo_
r -
'b)- '· +l. n.
'J
, , ,/A, •
0"13=_..!:._
4 77 [g"YF yA
F"J3+lgaf3p
4 Av
F"v+ fL 2 (A"Af3_lg"J3AA")]
2 A
(b) For these fields in interaction with the external source Jf3, as in (12.91 ), show
lll<lL lllC Ulllt:lt:lllHtl V<lliUll JaW: l<ll\.C lilt: Millt: lUllll a> lUi lilt: Clt:~llU-
magneuc neJOs, name1y,

J pAJ3
(J @uf3 = _A_ _
a
c
(c) Show explicitly that the time-time and space-time components of 0"1' are
1
0° 0 =- f£ 2 + B 2 + u 2(A 0A 0 +A. A)l
1577
0'o = 4~ [(E X B), + fL2A'Ao]
12.17 Consider the "Thomson" scattering of Proca waves (photons with mass) by a free
electron.
(a) As a nreliminarv. show that for an incident nlane wave of unit amnlitude
.n • a . '"'~' ~r nn;t ;tnrlo ,,;~n
A
.-v. . ". "' ""
,. "' ;~,
r
, . '"
"b' "'
/t\ " " " ' ' +<
'6 . \:! \'! , .
energy fluxes (measured by 8 30 ) are F, = wk/877 and F1 = (fL/w) 2F,. Show
. .
dl'V 1U1 dWH>d1 J n•vu ""'' cu~ rauu u1 ume-ave1 ageu energy !lux
to energy density is
\0'0) k I v}
\0"0)
p
w y' w2
as expected for particles of mass p...

(b\ For nolariz"tion "'' initi"ll" "nd nolari~ "fi"~"'" th"t th" "T> -
son" cross section for scattering is
du 1"*•"·12Fout
1
(F Fo) - 2f"
d!l F
where r0 is the classical electron radius, E 0 is a factor for the efficiency of the
incident Proca field in exciting the electron, and the final factor is a ratio of
.' ._ LV
. . .
"'~ uu 0 UUA~O. , ""' "' cue vatue u1 Lo.
(c) For an unpolarized transverse wave incident, show that the scattering cross
section is
I \
" ' ' 2
\;~i), = t~) 1 + cos 2 0 + \~) sin20
where the first term is the f"mili"r tr" •to · sr • Ht "n.-1 tho
.• 1 •• '1 ~-
_,
·-o
(d) For a longitudinally polarized wave incident, show that the cross section,
summed over outgoing polarizations, is
2 2
·'· "\ I
I d ri) 1
= ! -'=-)
W
r6 sin 0 + \~
2
W
cos2 0
where the first term is the longitudinal to transverse scatterinP and the second
is longitudinal to longitudinal.
Note that in the limit p../w ~ 0, the longitudinal fields decouple and we
recover the standard Thomson cross section.
H • ., ~-
'·J)
,
-" • ' < '-
VL LU~ UU~'f>~U~~ LU~VL~m
l Vl llldl WUl ULH~lWlM:O,
for source-free electromagnetic fields confined to a finite region of space, the 3-

space integrals of 0° 0 and 0°' transform as the components of a constant 4-vector,
• ,. _,- ,. . n
ao UJ \L~.lVVJ·
12.19 Source-free electromagnetic fields exist in a localized region of space. Consider

the various conservation laws that are contained in the integral of auMu~y = 0 over
all space, where Ma~y is defined by (12.117).
(a) Show that when f3 and yare both space indices conservation of the total field
angular momentum follows.
. .
\UJ JHV W U " " W HCH p V LH~ •uuv~
':':"": = L~ em
dt Eem
where X is the coordinate of the of m"« of thP '""""tir fiPlcls
defined by
y ( A3 { .J3
J I
where u is the electromagnetic energy density and E,m and P ,m are the total
energy and momentum of the fields.
Ch. U Problems 623
12.20 A uniform superconductor with London penetration depth AL fills the half-space
x > 0. The vector potential is tangential and for x < 0 is given by
Ay = (ae'kx + be-ikx)e-iwr
Find the vector potential inside the superconductor. Determine expressions for
the electric and magnetic fields at the surface. Evaluate the surface impedance z.,
(in Gaussian units, 417/C times the ratio of tangential electric field to tangential
magnetic field). Show that in the appropriate limit your result for Z, reduces to
that given in Section 12 9
12.21 A two-fimd model for the electrodynamics of superconductors posits two types of
electron, normal and superconducting, with number densttles, charges, masses, and
collisional damping constants, nj, e}, mi, and 'Yi· respectively (i = N. S). The elec-
trical condlJCtivity consists of the SllW of two terms of the DnJde form (7 58) with
f 0 N ~ ni, e --> ei, m --> mi, y0 ~ Yi· The normal (superconducting) electrons are
distinguished by YN i= 0 ( Ys = 0).
(a) Show that the conductivity of the superconductor at very low frequencies is
largely imaginary (inductive) with a small resistive component from the nor-
mal electrons.
(b) Show that use of Ohm's law with the conductivity of part a in the Maxwell
equations results in the static London equation for the electric field in the
limit w ~ 0, with the penetration depth (12.100), provided the carriers are
identified with the superconducting component of the electric fluid.
CHAPTER 13
Collisions, EnerRY Loss, and

Scattering of Charged Particles;
r 1/1 n-rl" n"J/1 lr r11., ,
'--" I " {., I \...- I " I'-' l / V
n-r/1 rl Tv rt"J/1 C'l; +;......,. -./1 D
&A- I " '-"' _L I &A- I "/J " " "VI " _L \.. U
rw rl; rw +;......,. -.~
·HJIL
In this chapter we consider collisions between swiftly moving, charged particles,

.
dHU VH LIICO
.. .
ll y 1111;
.
with special emphasis on the exchange of energy between collision partners
.
llUlll ""Ule"
. .
,LlVll~ -urr
. .
. we also
treat Cherenkov radiation and transition radiation, phenomena associated with
charged particles in uniform motion through material media.
.,]
A fast charged particle incident on matter makes collisions with the atomir
'c ~nrl nnro].,; Tf thA n~rt;,..]., ;c l , . ' (, ...
K meson, proton, et~.), the :ollisions with electrons and with nuclei hav: differen~
consequences. The light electrons can take up appreciable amounts of energy
11 Vlll lllCO p<tHicu:: wiuruUL C<tUSi11g signincanr aenections, wnereas me
massive nuclei absorb very little energy but because of their greater charge cause
scattering of the incident particle. Thus loss of energy by the incident particle
occurs almost entirelv in collisions with ,], 1s Th-, rJ.,-fl.,,..tirm of thP · ~~"
c ..
HWJI
: ..
H~
'.J .J."
YU'
. 1•
VH VU
<' lldl1U,
"
llUlll
d .
<W•
... . .
"
0 _J
.
collisions with the atomic nuclei. The scattering is confined to rather small angles,
so that a heavy particle keeps a more or less straight-line path while losing energy
until it nears tne ena ot 1ts range. !'or mcwent e!ectronsooth energy loss and
scattering occur in collisions with the atomic electrons. Consequently the path is
much less straight. After a short distance, electrons tend to diffuse into the rna-
., . ,,. n<~th
~·
UlCO ~UUJCOv< Vl
r ..
tPri,.] r"thPr th>'ln ""in "
•lV~~ a u u JJ __; 1~ i111
.• 1
"r l li1lll Ullt: i111U lS

in several books (see references at the end of the chapter) where numerical tables
and graphs are presented. Consequently our discussion emphasizes the physical
Ideas mvolved rather than the exact numerical formulas. Indeed "f,,n ml<~ntum-
mechanical treatment is needed to obtain exact results, even though all the es-
sential features are classical or semiclassical in origin. All the orders of magnitude
nf th<> m•~nh•m
'1
.cc ·~r<> <>~cilu rl<>nu~hlA <". ·~ ;;,n .,_nn~<n' ••
-J -J r
' '-' • ,,;11
' T
oe seen.
We begin by considering the simple problem of energy transfer to a free
electron by a fast heavy particle. Then the effects of a binding force on the
el ~re ~ncl thP. chssic"l Rohr ~ <>nArm• lncc ic nl..<n;nAcl
-OJ
A description of quantum modifications and the effect of the polarization of the

medium is followed by a discussion of the closely related phenomenon of
I'"'L .. 1. ..1." • • • .1. ~• ·• • • • r •
m "' ·r '~'" um~. UlCOH LHCO Vl H<"
cldent particles oy nuclei ana multiple scattenng are presented. Fmally, we treat
624
"· ... " . • "J
T · u . ., ro.~"' · /,;.,.::;:
transition raaiarion oy a panicle passing rrom one meaium ro anomer or airrerem

. ,] .
13.1 Energy Transfer in a Coulomb Collision Between Heavy

7nclilenfPartlCfe anastatwnary ---pree Electron;
Energy Loss in Hard Collisions
~swtrrparrtcJe or cnarge ze ana mass M ~energy!'- yMc, momentum t -
n>~,\ ,.~lli<iPo mith ~n ;,. p] nf ,.·h~rnP ~P ~nd m~oo vn H'nr ;,.
·o
collisions the binding of the electron in the atom can be neglected; the electron
can be considered free and initiallv at rest in the lahoratorv. For >~ll in,.irl.,nt
particles except electrons and positrons, M >> m. Then the collision is best
viewed as elastic coulomb scattenng m the rest trame ot the mcident particle.
The well-known Rutherford scattering formula is
L
2
dcr r ze ~~"~~4 e /1 '2 1 \
,~ .,
dD 2pv 2
where p = yf3mc and v = {3c are the momentum and speed of the electron in
the rest frame of the heavy particle (exact in the limit Mlm ----> co). The cross
sectiOn canoe giVen a Lorentz-invariant form by relating the scattering angle to
••~ A .&. ' n2 ' 1\2 .,-,_ _, • •
q .. 1- I .
''1 -~
J~
2 2 2 '
Q = 4p sin ( 8/2). The result is
7
dcr r( ze-:::;:;
~ = 471'
l (13.2)
UI,L \PG\L
where {3c, the relative speed in each particle's rest frame, is found from {3 2 =
1 - (Mmc 2 /P · p )".
~- .
~ -rora 5" 1c.u ·oJ --ross-r uy uu; .pctlllClt:, Llldl "•
the kinetic energy imparted to the initially stationary electron, is proportional to
(13.2). If we evaluate the invariant Q 2 in the electron's rest frame, we find Q 2 =
2mT. With Q 2 reolaced bv 2mT, (13.2) becomes
dcr 27rz 2 e 4
~U.:l)
dT mc2{32Tz
. . . . .
1:..4 ~1-'·-' J 1~ LIIe cru~~ ;_per umL · Hlle1va1 1ur e11ergy IU~~ 1 uy
the massive incinent . :in :l ,....,]Jioinn with :l fr"'"' o+o+' 1:lTV .,].,c-
tron. Its range of validity for actual collisions in matter is
~ ~ ~
.1. min .1 '.lmax

. .
\'min-, n\WJ
-wrrrr-I min--seruy-uur ;-or
: n\Wi ~~~ ~~nesum~re
of th" m"an pff ir ' · ,. "' PnPr<m\ :lncl T
We can find T max by recognizing that the most energetic collision in the rest frame
of the incident narticle occurs when the electron reverses its direction. After such
a collision, the electron has energy E' = ymc 2 and momentum p' = yf3mc in the
direction ot the inciaent particles velocity m me Jaooratory. 1ne ooost to the
' ' >tn~• nivPo
Tmax = E - mc 2 = y(E' + {3cp') - mc 2 = 2y2 {3 2 mc 2 (13.4)

0~0 Lnapter u Energy Loss, l!lcanermg; LnerenKov ana 1ransmon Kamanon -I.J
We note in passing that (13.4) is not correct if the inc~dent particle ~as too high
dll vW.-LfSJ• UlC CXdlOL dl.,WCl lUl 1 max lld> d lll lllC Or, 1J
1 + 2mE/M 2 c 2 + m 2 /M 2 • For muons (M!m = 207), the denominator must be
_\_ • '£ cL • -U A r"' • H
uuv 11 " ' " '-11'-' 51 " -r , au<'- cv ~· ~" v< IS""uc""" ,-v, pro-
tons that energv is roughlv 340 GeV. For equal masses it is easv to see th"'
Tmax = ('y ~ 1)mc 2 .
When the spin of the electron is taken into account, there is a quantum-
mechanical correction to the energy loss cross section, namely, a factor of
1 p.? •
Mil \OIL)
?
\1
'"
P 11 1 max)·
1 I
du Z1rz 2 e4 T '
1 p ~U.J)
\dT Jqm mczf32T2 I Ttnax
The energy loss per unit distance in collisions with energy transfer greater
. -' tl-.rnno-1-. m"ttPr mit!-. N nPr 11nit vnh
tl-."n
r . bo
- frw " !-.""""
.
J
each with Z e ectrons, 1s g1ven y the mtegra ,
JJ ~- o "
,
dE (T > s)
---;- = NZ r"""x T du·~ dT
UL
UA
"" ~U.tl)
2 4 r ',, zaz 2'
Z1rNZ In {32
mczf3z s
In the result (13.6) we assumed s << T max and used (13.5) for the energy-transfer
' ,o . ' . . . .
\.OlU» • 111C LCllll, -p , lll lllC >LtUdlC UJ. . , LllC 1 ·>pill
contribution. Equation (13.6) represents the energy loss in close collisions. It is
vwy
_ \_ - L' ~
'u11u l-'"'
• ~ ~
~ -- " I
"\ = 1
.\ L
Uv'-a~ov
'- · T L 1..
, 11UO Uv'-11 _0
' ~
An alternative, classical or semiclassical approach throws a different light on the

physics of energy loss. In the rest frame of the heavy particle the incident electron ap-
proaches at impact parameter b. There is a one-to-one correspondence between b and
· _'- n _,_ .r. ~ t.. •
U UH HO
<H'- 0 Lv V \Ovv L LVUOvLH L,',Ljo LHv vHvLOJ HUOWOvO L vUH Uv
2z 2e4 1
T(b) = muz . bz + b(c)2
mm
with b~fn = ze 2/pu. For h >> h~fn the energy transfer varies as b 2 , implying that, if the
energy transfer is greater than e, the impact parameter must be less than the maximum,
2 4 1/2
h(cJ rc \ ~ f 2z e I
muLe
When the heavv oarticle oasses throue:h matter it "sees'' electrons at all oossible imoact
parameters, with weighting according to the area of an annulus, 27Tb db. The classical
..
IV" pea UJJH LUL W Hll ll dJIMC:l )';1 C:d LC:l Uldll c "
N
""
(b(c) (c) 2 4 2
dE~~ _\ ~ _ _,7 max ~/CoL "- ~ _.,7 z e '- !J(c)
n{ax ( S )
,__ ..'7\,
In
'} " \" }' <U
dx " Jo mc2{32 b(c)

mm
Substitution of bm"' and bmin leads directly to (13.6), apart from the relativistic spin cor-
rection. That we obtain the same result (for a spinless particle) quantum mechanically
and classically 1s a consequence ot the val1d1ty ot the Kuthertord cross sectwn m oom
If we wish to find a classical result for the total energy loss per unit distance, we must
address the influence of atomic binding. Electronic binding can be characterized by the
Sect. 13.2 Energy Loss from Soft Collisions; Total Energy Loss 627
frequency of motion (w) or its reciprocal, the period. The incident heavy particle produces
appreciable time-varying electromagnetic fields at the atom for a time t!.t ~ b!yv [see
(11.153)]. If the characteristic time tJ.t is long compared to the atomic period, the atom
responds adiabatically-it stretches slowly during the encounter and returns to normal,
without appreciable energy being transferred. On the other hand, if M is very short com-
pared to the characteristic period, the electron can be treated as almost free. The dividing
line is {w)l:!.t = 1, implying a maximum effective impact parameter
beyond which no significant energy transfer is possible. Explicit illustration of this cutoff
tor a charge bound harmomcalty ts found in Problems 13.2 and 13.3.
If (13.8) is used in (13.7) instead of b~~~x(e), the total classical energy loss per unit
distance is approximately
dE z2e4
= 27TNZ - - - ln(Bz)
2 2
(13.9)
dx classical me f3
where
(13.10)
In (13.10) we have inserted a numerical constant A of the order of unity to allow for our
uncertatnty 1n h~~~x· The parameter 1J ze 2 7fw is a characteristic of quantarn-mcchanical
Coulomb scattering: 71 << 1 is the strongly quantum limit; 71 >> 1 is the classical limit.
Equation (13.9) with (13.10) contains the essentials of the classical energy loss for-
mula derived by Niels Bohr (1915). With many different electronic frequencies, (w) is the
geometric mean of all the frequencies w1 , weighted with the oscillator strength .f1 :
Z ln(w) 2: h In w 1
(13.1!)
Equation (13.10) is valid for rt > 1 (relatively slow alpha particles, heavy nuclei) but
overestimates the energy loss when 71 < 1 (muons, protons, even fast alpha par!icles). We
see below that when 1'J < 1 the correct result sets 1'J 1 in (13 10)
13.2 Energy Loss from Soft Collisions; Total Energy Loss

The energy loss in collisions ·.vith energy transfers less than a, including those
small compared to electronic binding energies, really can be treated properly
only by quantum mechanics, although after the fact we can "explain" the result
m semtclasstcallanguage. The result, first obtained by Bethe (1930), is
E
dx (T< e) (13.12)
me f3
where
yv(2me) 112
(13.13)
h(w)
The effective excitation energy h{w> is given by (13.11), but now with the proper
quantum-mechanical oscillator strengths and frequency differences for the atom,
mcludmg the contribution from the continuum. The upper limits on the energy
t;;?!ol ~·
-r
n T oo .
-.,, , .. .. - .. . co
transfers is assumed to be beyond the limit of appreciable oscillator strength.

Such a limit is consonant with the lower limit s in Section 13.1, chosen to make
the electron essentially free.
The total energy loss per unit length is given by the sum of (13.6) and (13.12):
dF: 72('4
47rNZ Unl H.\ Wl rn 1 A\
ax mew 'I
where
2·'lf3 2 mc 2
Bq = (13.15)
fi(w)
The general Dehavwr ot Doth the classical ariCICluantum-mechamcal enf>rav
1~cc lnc ;c ;11 -' ;n R; 1 'l 1 'l'l, r _, .& . ,
' UL
-o· · · 'J 'J
the incident heavy particle, the mass and charge of the electron, and the mean
excitation energy fi(w). For low energies ( -yf3 < 1) the main dependence is as
11{3 2 , while at high energies the slow variation is proportional to ln( -y). The min-
imum value of dE!dx occurs at -yf3 = 3. The coefficient in (13.12) and (13.14) is
' ~11 ~1 {) 1<:{) 2('1'7/A \ H'~' ' ~' •' '
"1 '- u LO UHO
\~~'"'Jf'
HoV " v o o 0 uumuer
ana A me mass numoer or me matena1 wmle o r!!/em·\ 1s 1ts dens1tv. Sin""'
2Z/A = 1, the energy loss in MeV· (cm 2 /g) for a singly charged particle in alu-
minum is approximately what is shown in Fig. 13.1. For aluminum the minimum
energy loss is roughly 1.7 MeV·(cm2 /g); for lead, it is 1.2 MeV·(cm2 /g). At high
energies corrections to the behavior in Fig. 13.1 occur. The energy loss becomes
heavv-oarticle specific throu!!h the mass-deoendent denominator D in T. 'mt1
10-2
'I 'I " II I I 'I I I II I II
-
~ l
Vi -~
" \\
1:'
-~
D
10-1 1- -
~
.lJ
~
kl
"" \\,\
----
\\ ~
\""- -- -----""':."
10-o .,1
\
' -----,1 -- -- -- --- .,I
-
.
< 10 keV
.I
10-1 100 101 102 103 10 4
YB
tigure 13.1 bnergy loss as a functiOn of y/3 of the incident heavv varticle. The solid
curve is the total energy loss (13.14) with h(w) = 160 eV (aluminum). The dashed curve
is the energy loss in soft collisions (13.12) with e = 10 keV. The ordinate scale
corresponds to the curly-bracketed quantities in (13.12) and (13.14), multiplied by 0.15.
Sect. 13.2 Energy Loss from Soft Collisions; Total Energy Loss 629
has a different energy variation and dependence on the material, because of the
density effect discussed in Section 13.3.
The restricted energy loss shown in Fig. 13.1 is applicable to the energy loss
' f', .~ +-~~Jr. ;~ ~J.. -y
' · ~~"': · D' with · . nrP~tPr
y u.' u
r 'L L
mau ~VKt:V navt: u raugt: LV -r OH HV~H ''-~
The density of blackening along a track is therefore related to the restricted

energy loss. Note that it increases more slowly than the total for large yf3-as
ln( y) rather than ln( y 2 ). A semiclassical explanation is given below.
Comparison of Bq with the classical Be (13.10) shows that their ratio is 7J - ze"/hv.
To understand how this factor arises, we turn to semiclassical arguments. Be is the ratio
of b!i,~x (13.8) to b;;;/" = ze 2 /ymv 2 • The uncertainty principle dictates a different brnin for
7J < 1. In the rest frame of the heavy particle the electron has momentum p = ymv. If it
is oescrioeo oy a uansverse1y Iocaiizeo wave .- ~LU lLO
': ""~ a>
well as posstble), tne spread m transverse momenta a.p arounu zero mus1 sallSlY up<--<-- p;
otherwise. its longitudinal direction would be ill-defined. This limit on IJ.p translates into
an uncertainty IJ.b in impact parameter, IJ.b >> hlp, or in other words, an effective quan-
tum-mechanical lower limit,
h
lJ~ ~D-10)
ymv
Evidently, in calculating energy loss as an integral over impact parameters, the larger
of the two minimum impact parameters should be used. The ratio b;;;/nlb~~[, = 7J· When
~ ' ~\ rl f·n 1 ~\ ' ' tJ..~
'I , "L .1 ' .11 . 1: '
·rr
' • C.
, \I n ' 1 VJ -1'
' '
expreSSLUII LOT D.
The value of Bq(e) in (13.12) can also be understood in terms of impact parameters.
The soft collisions contributing to (13.12) come semiclassically from the more distant
collisions. The momentum transfer op to the struck electron in such collisions is related
tJ.. t ,f T ' w t~ "~ •~ ~• I/Z (')~ thP othPr h,n,-l thP 1 .1: ~
_, 'C>J
-'· '- '':, A "\

'r. " ~o>rto;n tho+ tJ..P
y "Y -¥ •
collision produces an energy transfer less thane, we must have IJ.p < DPmax = (2me) 112 ,
hence IJ.b > h/(2me ) 112 • The effective minimum impact parameter for soft collisions with
energy transfer less than e is therefore
T,(q)fo\
n (H17\
rnm (2me ) 11"
For collisions so limited in impact parameter between (13.17) and bmax = yvl(w). we find
yv(2me) 112
R (" \
n(w)
Ill a
0, WlLll ' lt"UlL.
The semiclassical discussion of the minimum and maximum impact param-

,], 'o. •J.. C, <t.. -'' • <t..ol. 'cL ,;~ thl.
·o: o·
tne resuictea ana total energy 1osses. At nign ._, 5 1"'' we t:Hcre;y
dependence is through dE!dx oc ln(B) ln(bmaxlbmin)· J:<or tne total energy loss,
the maximum impact parameter is proportional to y, while the quantum-
mechanical minimum impact parameter (13.16) is inversely proportional to y.
The rat1o vanes as y . !'or energy loss resrricreaTo energy uansters 1ess mane,
the minimum imnact narameter (13.17) is indenendent of y, leading to B, (e) oc y.
Despite its attractiveness in making clear the physics, the semiclassical de-
. . . . .
sc s y a war-
rants discussion. Classically, the energy transfer T in each collision is related
. . c 24 22
em
13.1 . With increasin b the ener
bmax = yv/(w) it becomes
(13.18)
smaller than the ionization potential, indeed, smaller than the minimum possible
. . .
We know, however, that energy must be transferred to the atom in discrete

quantum JUmps. tmy amount o energy sue as s1mp y cannot be ab-
sorbed by the atom. We might argue that the classical expression for T(b) should
parameters from bmax = yv/(w) and lead to wrong results. Could bmax nevertheless
? . . . . n
a ecte . o, tlme- epen ent pertur at10ns o a quantum system cause s1gm Icant
excitation only if they possess appreciable Fourier components with frequencies
in another direction. The classical expressions must be interpreted in a statistical

sense.
The classical concept of the transfer of a small amount of energy in every
co ISion IS mcorrect quantum-mec amca y. Instea , w 1 e on t e average over
many collisions, a small energy is transferred, the small average results from
appreciable amounts of energy transferred in a very small fraction of those col-
. . .
that the quantum-mechanical mechanism of discrete energy transfers and the
classical process with a continuum of possible energy transfers can be reconciled.
The detailed numerical agreement for the averages (but not for the individual
amounts stems rom t e quantum-mec amca e mt10ns o t e osc1 ator
strengths fj and resonant frequencies wj entering (w). A meaningful semiclassical
(13.19)
Sect. 13.3 Density Effect in CoUisional Energy Loss 631
ThPrP ~rP onin ~nrl exch~nPe

.. -'-
t:llt:l;L' 111 LV uno
argument of the logarithm; the other effects only contribute to the aaaea
constant.
The expressions for dE!dx represent the average collisional energy loss per
unit distance by a particle traversing matter. Because the number of collisions
oer unit distance is finite. even though large, and the spectrum of possible energy
r · · ,. · ' ,1 .11: · · ;r1~ th~r~ ~r~ flu~h•~•;~nc rl thP >!Vf'r-
'
age. These tlucTuauons proouce srragg1ing in o or mp<u 'i"'c ud-
versing a certain thickness of matter. If the number of collisions is large enough
and the mean energy loss not too great, the final energies of a beam of initially
monoenergetic particles of energy E 0 are distributed in Gaussian fashion about
the mean E. With Poisson statistics for the number of collisions producing a given
·transfer T it can be shown (see e.g. Bohr. Sectwn-z:3 or'Ross1, :sectwn
~. f j UHH <H'-' . O.j_UU "bJ
(13.20)
where tis the thickness traversed. This result holds provided 0 << E and 0 <<
(1<
,-
7:'\
-n
~nrl ~lc~ () .._,.._, T ?012 R 2 rnr 2 Fnr 1:. · · · · >the last
•• • • • . • . L ~
conmnon UJU y raus. >11c11 uno tuv11 111 c11c•g•c'" 11vL u"' , vu•
is described by the Landau curve. The interested reader may consult the refer-
ences at the end of the chapter for further details.
13.3 Densitv Effect in Collisional Enerf!.v Loss

ror particles that are nor roo reJallvlsnc, Lne ooserveu energy Joss is givc11 dv
curately by (13.14) [or by (13.9) if TJ > 1] for particles of all kinds in media of all
types. For ultrarelativistic particles, however, the observed energy loss is less than
predicted by (13.14), especially for dense substances. In terms of Fig. 13.1 of
(dE!dx), the observed energy loss increases beyond the minimum with a slope of
rmwhlv nnf'-h:.lf th:.t nf the theoretical curve corresoondml' to onlv one oower
v1 ·y 111 wc "' );' . v1 u"' 1v5uuuuu " ' \ ,~. • • J u ' . r "b' -y
emulsions the energy loss, as measured from grain densities, barely increases
above the minimum to a plateau extending to the highest known energies. This
again corresponds to a reduction of one power of y, this time in Bq(r=) (13.13).
This reduction in energy loss, known as the density effect, was first treated
tneorencallv bY t ermi 1.f9Zrn J. 1n our uiscussiun so wr wt: nave vwo
th"~t ;"---;:;;:;;:- ""'lirl in rlPnof' o11 Wf' h"v" ~«,;;:;:;-;:;-.-1 that it is
legitimate to calculate the effect of the incident particle's fields on one electron
in one atom at a time, and then sum up incoherently the energy transfers to all
the electrons in all the atoms with bmin < b < bmax· Now bmax is very large
tn atnmic dimensions esnec1allv for larl'e 'V. ( onseouentlv m aense
the tvnical atom m ouestwn It DIS comoarable to Dmax· lhese atoms, muuenceu
themselves by the fast particle's fields, will produce perturbing fields at the chosen
atom's position, modifying its response to the fields of the fast particle. Said in
.
paniCleS HelUS ilUm Lnelf !fcc-
. ..
another way, in dense media the dielectric polarization of the material alters the
.
VdiUC' LV L11V'C <.,Hdld'-LCHM!<., Vl """''
. .
632 Chapter 13 Energy Loss, Scattering; Cherenkov and Transition Radiation-G
fields in a dielectric. This modification of the fields due to polarization of the

meomm must oe taken mto account m calculatmg the energy transferred in dis-
t~nt ~nll" ' H'n~ ~lno~ ~nlJ;o;nno th~ ' 'A. •• ' ' :tJ.. -'
atom at a time. . Ie caIcuI atwn

. . Th en th e free-partlc . ...wit
. h out po 1anzat10n
· · e ffu'y
ects une
will
·~~J, n~r~
ThP iJiviiJina ; ~lno~ rl:ot~nt ~nll:o; L >n on;
h r J .u . ' f I .h
r . '
t e order of atomic dimensions. Smce the joimng o two ogant ms Is involved
. . <0 Vl
in calculating the sum, the dividing value of b need not be specified with great
preCISIOn.
We wlll determine the energy loss in distant collisions (b >a), assuming that
the fields in the medium can be calculated in the continuum approximation of a
macroscopic dielectnc constant Elw). lt a IS of the order of atomic dimensions
th:o n ' >nt:nn ,.,;n ,t h rl f', tJ, -> -~ tL .r •• .n· ' ' '
-.-~ 0' , UUL
will be valid for the great bulk of the collisions.

'T'hP n~nhlPm nf ~ "' " ' thP p] ' fi~lrl :n tJ..~ "" · - rl, • t\.
._,
r
fast particle moving with constant velocity can be solved most readily by Fourier
transforms. If the potentials A~-'(x) and source density lJx) are transformed in
space and time according to the general rule,
F(x, t) = ~ { d 3k { dw F(k, w)eik·x-iwt (13.21)

~L7TJ J J
then the transformed wave equations become

-
2
2 .
'T7T
k - : 2 E(w) <l>(k, w) = ~~ p(k, w)
-,
2 - 4 (13.22)
e w d w) A.(k ,,,) -
c
7T
c
I(lr ,,,\
L
'f'hP iJ;pl, c( ,,,\ ~nnpgro "h~ :~n llu :n ~Ao;t:Ano rl:~tnt ,;1 \.
• Tr 'J t" ')
the presence of D in the Maxwell equations. The Fourier transforms of

p(x, t) = ze B(x - vt)
auu" ~LJ.LCJ)
.T(x t) - vn(x t\
are readily found to be
ze
p(k, w) = 7T B(w - k · v)
2 (13.24)
J{Jr
,. ' ..-'
\ ,{1.
.' '
..\
·-,
From (13.22) we see that the Fourier transforms of the potentials are
?.7P !'i( ,,, lr • v\
<Pik wl
ElW) w
k 2 - 2 E(w)
L
(13.25)
and
v
A(k, w) = E(w)- <l>(k, w)
~
Sect. l5.3 Density Enect m t..;othsiOnal Energy Loss ().).)
From the definitions of the electromagnetic fields in terms of the potentials we

obtain their Fourier transforms:
r
E(k, w) = i WE(w) -V - k <f>(k, w)
c c
J
(13.26)
v
B(k, w) = iE( w)k X - <J:>(k, w)
c
In calculating the energy loss to an electron in an atom at impact parameter b,
we evaluate
I /
""''--' ~ I •
0
I:. UL "'~ L'\.10 LWA\WJ
0
I:. \WJ UW
\'·'·"'')
"~ • vf. ,,\ ;c th<> Rmn-i<>r .£

;n t;~, of th<> ,]. '· "' .~t' rl I<'{ ,\
,,_ ...,"'-'. .£_ r ,,_ _, .c -1 J. _,. '',''
iO LW.- i VUUvi U iH CHHV Vi u • v '-'''-''-'' ''"' <H U }-''-'' r
distance b from the path of the particle moving along the x axis. Thus the required
electric field is
E( w) =
1
(21T)3/2
J d 3k E(k ' w)eu'k 2
(13.28)
where the observation point has coordinates (0, b, 0). To illustrate the determi-
nation of E( w) we consider the calculation of E, (w), the component of E parallel
to v. Inserting the explicit forms from (13.25) and (13.26), we obtain
2ize { wE(w)v o(w- vk,)
£.1~W)
-f. ,\f,., .\3/2
\"/\' ., I a ''x 2 K! 2 ~ l.J.LY)
L
- k2 - - ,o(r.Jl
c"
The mtegral over dk 1 can be done Immediately. Then
E 1(w) = -
Ltzew
( 2 1r)312v2
___!:__ - {32
E(w)
J -oc
dk2 eihk, J -oo k~
ak3
+ k~ + ,\ 2
where
wz w2 wz
A2 = 2- - -
2
E(w) = 7 [1 - /3 2E(w)] (13.30)
v c v-
The integral over dk 3 has the value 1r/(A 2 + k~) 112 , so that E 1 (w) can be written
L.(.U.U ... roo
~
ihk
P,(,,,) R2 I tiL
-
(11111
V21T rl E( w) J-oo (,\" + k~)"£
The remaining_ inte2.ral is a reoresentation of a modified Bessel function.* The

rPclllt ;o
i7Pf•
'? 'l/£
1
E,(w)- --- - - - - BL K,.( Ab l (13.32)
v· 1T E~W)
*See, for example, Abramowitz and Stegun (p. 376, formula 9.6.25); Magnus, Oberhettinger, and Soni
(Chapter XI), or Bateman Manuscript Project, Table of Integral Transforms, VoL 1 (Chapters I-III).
634 Chanter 13 R Loss. • • Ro -'·' •
where the square root of (13.30) is chosen so that A lies in the fourth quadrant.
A similar calculation yields the other fields:
1 112 '
Eo( w) - ze f 2 A KJ>.h)
v 7T E(W)
In the limit E( w) ---'> 1 it is easily seen that fields (13.32) and (13.33) reduce to the
results of Problem 13.3.
To find the energy transferred to the atom at impact parameter b we merelv
write down the ~lizatir.n r.f f11 ?7\·
llE(b) = 2e 2: fJ Re
J
r iwxf(w) • E*(w) dw
where xi ( w) is the amplitude of the jth type of electron in the atom. Rather than
use (7.50) for xf(w) we express the sum of dipole moments in terms of the mo-
lecmar polanzaollltv ana so the dielectric constant. '1 hus
where N is the number of atoms per unit volume. Then the energy transfer can
be written
1\J::fh\ "n I J.\lv/.\12.1
The energy loss per unit distance in collisions with impact parameter b > a
is evidently
dE
= 2TTN roo llE(b)b db (13.35)
' b>a Ja
u anu ~LJ.J..l) are msenea mro ~Ll.-''1-J ana

~lJ.JLJ ~1-'.-':lJ, we nna, atter
some calculation, the expression due to Fermi,
~ (ze;z Re Jo(
00
1
( ddE) = iwA*aK1 (A*a)K0 (Aa)(-() - {3 2 ) dw (13.36)
X b>a 7r V E W
wv :v uuw wruu 5 u a ~,;~ u1 ramus a arouna

the path of the incident particle. By conservation of energy this is the energy lost
per unit time by the incident particle. Thus
dE 1 dE c r= ., .D D J._
A I
'T'l n 1 A .C ' - > '

a VL <1l1lv <~ -, <V UU d l VHv
pomL on Lne cynnuer over au time. usmg ax v at, we nave
dE) ca Joo
( -d
X b>a
= - - _ BJ(t)E 1(t) dt
2 oo
~ nrr
.. Lu•• .r-.1!!"
UJJ
'J
T ..
Hl llHO ~'
' LU
' 1<1
...
y tuiO vau 'uc; ' .
Ll<OU llllU d
r
H <0'-j U<Olll.- J
'"'•
'
(~~)b>a = -caRer Bj(w)E (w) dw 1 (13.37)
With fields (13.32) and (13.33) this gives the Fermi result (13.36).
The Fermi expression (13.36) bears little resemblance to our earlier results
tor energy loss. tlut under conC11tlons where polanzatlon el!ects are ummportant
it vields the same results as before. For examn!e for n Jt.ivistic n: • . I
({3 << 1) it is clear from (13.30) that A = w!v, independent of E(w). Then in
(13.36) the modified Bessel functions are real. Only the imaginary part of 1/E(w)
contributes to the integral. If we neglect the polarization correction of Section
4.5 to the internal field at an atom, the dielectric constant can be written
..~. 4nNe ' \ '

2
A,.,\ 1
f; (1 'l 'lR\
m '7 w7 - w--r - iwL
')
'
where we have used the dinole moment exnression (750). AssuminP that the
second term is small, the imaginary part of 11 E( w) can be readily calculated and
substituted into (13.36). Then the integral over dw can be performed in the
narrow-resonance approximation. If the small-argument limits of the Bessel
functions are used, the nonrelativistic form of (13.9) emerges, with B" =
vla(w). If the departure of A from wlyv in (13.30) is neglected, (13.9) emerges
with Be = yvla(w).
The densitv effect evidentlv comes from the oresence of comolex anmments
in~ ·' f11nrtinno ' · 1o- tn t~ kin a intn ~rrnunt ,d ,,,\ in
2
(13.30). Since E( w) there is multiplied by {3 , it is clear that the density effect can
be really important only at high energies. The detailed calculations for all ener-
gies with some explicit expression such as (13.38) forE( w) are quite complicated
and not particularly informative. We content ourselves with the extreme relativ-
is tic limit ({3 = 1). Furthermore, since the important frequencies in the integral
over dw are optical frequencies and the radius a is of the order of atomic dimen-
,_
ry
..
sions l Aal ~I wale\<< 1. Conseouentlv we can annroximate the Bessel functions
~ "-·
,. ;, /'1 1 ()'J\ T t
~ ~--"".
~1 ~·· ·~··
-.::c· ,.
:.
"'7'
.-. .. ~
r(
CXfll ~LJ,.JU) 1•
(dE)
- 2
=- -?- (ze? Re iw -1- - 1 ){ ln (1.123c) - -1 ln[l - E(w)] } dw
dx b>a 7T c o E( w) wa 2
(13.39)
It is worthwhile right here to point out that the argument of the second logarithm
io
~~-
~~h·~lhr
,_.
f1
·, ;
. ·c
-~ ..
R 2"(,,,\l Tn thP limit"
_, - ,
. ·r
... -o • ••
1 thio lna tPrm
.u
' ",. ,
aivP~ ~ f~rtnr 'II
" · ' · {1., ( l \

. ~ ~ .I .\
·\ 'I
in thP
-1- 1
~ ·as Jl E~WJJ, mereoy removmg one power or y rrom

the loganthm, m agreement w1th expenment.
The integral in (13.39) with E( w) given by (13.38) can be performed most
easily by using Cauchy's theorem to change the integral over positive real w to
one over positive imaginary w, minus one over a quarter-circle at infinity. The
integral along the imaginary axis gives no contribution. Provided the r 1 in (13.38)
are assumed constant, the result of the integration over the quarter -circle can be
written in the simple form:
al!, ~.Ct:)
'" 7
Wp ' l.lLjC
I c2 1111 ~Li.4UJ
dx h>a awP
where wP is the electronic plasma frequency
?
W2 = ... 'Ill V LC
P m (13.41)
. .
1ne corresponomg relanVIsnc expresswn wnnom me oensny enecr IS
(dE) = (ze):w~ ln(1.123yc) (13.42)

dx b>a c a(w)
. . . . .
vve see wat we uensuy euect prouuces a srmpnucauon m maL me asympronc
energy loss no longer depends on the details of atomic structure through (w)
(13.11), but only on the number of electrons per unit volume through wP" Two
substances having verv different atomic structures will produce the same enemv
1
>~~~
~-
>~>
,]. .1. . .
u~uv Y_~>UV>V~
• ~1.
p>~
"_.I." . . . .
<>~V~ U>VU
"·
~vu~nwo
..
mv o
• 1.. ...
- u • u m u>v
... "·
~VUOHJ
of electrons is the same in each.

Since there are numerous calculated curves of energy loss based on Bethe's
formula (13.14), it is often convenient to tabulate the decrease in energy loss due
tr. thP Thi~ ;~ ill~t thP {11 4nl :mrl {11 4?\·
• J
. (dE) __ (ze) w~ In()

cwP)
IIm~d- 2
(13.43)
f3~1 X C W
'
'-
Without
density
correction ...- ...- ...-
...-
T \
...-
/
-
~
I
...- ...-
/
~~~ / Total
t- .......
I I I I I I
~ 1 1 1n 1~2 103 104
h-11 ......
l''igure 13.2 bnergy loss, mcludmg the density ellect. The dashed curve Is the total
energy loss without density correction. The solid curves have the density effect
incorporated, the upper one being the total energy loss and the lower one the energy
loss Clue to mCIIv!Ciual energy transters ot less than lU ke v.
-sect. 13.4 Cherentwv KadJatiOn OJ/
For photographic emulsions, the relevant energy loss is given by (13.12) and
(U.UJ with 8 lU keY. WlthtlfeaensiTy correction apptiea, rnis oecomes con-
stant at high energies with the value,
dE( E) {zdw~ /2me 2 8 '

---+ !n h2w2 J (U.44J
dx 2c 2 p
For silver bromide, hwP = 48 eV. Then for singly charged particles (13.44), di-
vided by the density, has the value of approximately 1.02 MeV· (cm 2 /g). This
energy loss is in good agreement with experiment, and corresponds to an increase
dUUVC lllC
. . ·~~ ~·
VdlUC U i l C ' ' llldll .1U /0. r'/Su'v
.
'"",.., -'- .'
<V<a'
'
-o) lV'' iJ.~
and loss from transfers of less than 10 keV for a typical substance. The dashed
' LL ~ d ~- ' C<L ,<,' <'. -' ' ££ .
vU' V v " <Hv 'vUl Vv 'Vl <VLal .
00
lV" VVuu~~c c;
.,..
.L J."#
~ rn
\.-lttl
r
'II
T>
••
'~·
•
The density effect in energy loss is intimately connected to the coherent response
of "' ,.1; ~ to thP. "'"' of " relativistic narticle that causes the emission of
Cherenkov radiation. They are, in fact, the same phenomenon in different lim-
itin2: circumstances. The exmession (13.36) or better (13.37), represents the en-
ergy lost by the particle into regions a distance greater than b = a away from its
path. l:1y varymg a we can examme how the energy 1s ueposiLea mrougnom Lne
medium. In (13.39) we have considered a to be atomic dimensions and assumed
1 Aa <.<. 1. 1~ ow we raKe me opposite 1imir. 11 Aa .->/' '• LHc
1 · • 1 1
fnnrtinn~ r~n hP · ,~tPrl t,;, thPir ~~um · •- · Then thP. tiP.Irls (11.1?)
rr
and (13.33) become
zew e Ab
E 1(w, b)---+ i - 7 1 - 1 --
c {32€( w) V5J)
.,.., lA
Ez(W, b)---+ VE(w) v b e "v
(U.4J)
H 3( w, b) ---+ FfETw)EZ('w, o)
The integrand in (13.37) in this limit is
re- A* 1 e-(A+A')a
( -caBiE 1 ) ---+ - 2 -[ w 1 - 2
- (13.46)
c \)A {3 E(w)
crt. •1
t-'
...... 't-'
. '
... ·o
••. rl f~~nn~nriM
"l'
R;uM
o·
lh~
·oJ
deposited far from the path of the particle. If A has a positive real part, as is
•llv trnP thP IPnti~ 1 ~- in (11 41i) will cm1se the · to wm-
ish rapidly at large distances. All the energy is deposited near the path. This is
not truP. on lv ,\ i < ~ irnn · Then the exoonential is unitv: the
expression is independent of a; some of the energy escapes to infinity as radiation.
From (13.30) it can be seen that A can be purely 1magmary It E(wJ 1s real (no
-' ' -' -' n2 / ' ' • H " -' -' ' ' _ p -' > '
'l-'"V"J auu p c~WJ J. -~ ,"''IIIIU' 't-'H~H ~
•
in the interests of simplicity we will assume that E( w) is essentially real from now
on. The condition {3 2 E( w) > 1 can be written in the more transparent form,
c
v > --;== (13.47)
~
This shows that the speed of the particle must be larger than the phase velocity of
the electromagnetic fields at frequency win order to have emission of Cherenkov
radiation of that frequency.
Consideration of the phase of A as {3 2 Echanges from less than unity to greater
tance along the path of the particle:
(dE)
dx md
= (ze?
C
2
JE(w)>(l/{3 2)
w( 1 (13.48)
The integrand obviously gives the differential spectrum in frequency. This is the
Frank-Tamm result, first published in 1937 in an explanation of the radiation
observed by Cherenkov in 1934. The radiation is evidently not emitted uniformly
in frequency. It tends to be emitted in bands situated somewhat below regions
of anomalous dispersion, where E(w) > {3- 2 , as indicated in Fig. 13.3. Of course,
if f3 = 1 the regions where E(w) > {3- 2 may be quite extensive.
Another characteristic feature of Cherenkov radiation is its :m'"'"'
tan Be= (13.49)
From the far fields (13.45) we find

1
cos Be = , 17""\ (13.50)
{3v E(w)
Sect. 13.4 Cherenkov Radiation 639
y
2
z 3 Figure 13.4
The criterion {3 2 E > 1 can now be rephrased as the requirement that the emission
angle &c be a physical angle with cosine less than unity. In passing we note from
Fig. 13.4 that Cherenkov radiation is completely linearly polarized in the plane
containing the direction of observation and the path of the particle.
The emission angle &c can be interpreted qualitatively in terms of a "shock"
wavefront akin to the familiar shock wave (sonic boom) produced by an aircraft
in supersonic flight. In Figure 13.5 are sketched two sets of successive spherical
wavelets moving out with speed d'vf€ fiom successive instantaneous positions of
a partlcle movmg wrth constant vclocrty v. On the left v rs assumed to be less
than, and on the right greater than, ci\/E For u > cl\fE the wavelets interfere
so as to produce a "shock" front or wake behind the particle, the angle of which
is readily seen to be the complement of fJc. An observer at rest sees a wavefront
movmg m the drrectlon of &c.
The qualitative behavior shown in Fig 13 5 can be given quantitative treat-
I
I
I
I
vt~ I
I I
j-E---y'E
.E. ----.1
-I
V< C/-y'E
Figure 13.5 Cherenkov radiation. Spherical wavelets of fields of a particle traveling

less than and greater than the velocity of light in the medium. For v > c/Vc, an
electromagnetic "shock" wave appears, moving in the direction given by the Cherenkov
angle &c.
(13.21). For example, the vector potential takes the form,
.L
1 .L ~.
but tractable, approximation that E is a constant the integral can be done in closed
. 2
of inte ration. Choosin the contour for the k 1 inte ration so that the otential
vanishes for points ahead of the particle (x - vt > 0), the result is found to be
A(x, t) = 13 2
(13.51)
inside the Cherenkov cone and zero outside. Note that A is singular along the
. . .
functional dependence will remove the mathematical singularity in (13.51).

The ro erties of Cherenkov radiation can be utilized to measure velociti
ployed to select a small interval of frequency and so improve the precision of

velocit measurement. For ver fast articles :S 1 a as rna be used to ro-
vide a dielectric constant differing only slightly from unity and having ( E - 1)
vana e over w1 e lilll s y varymg e gas pressure. oun mg ev1ces usmg
Cherenkov radiation are employed extensively in high-energy physics, as instru-
. . .
13.5 Elastic Scattering of Fast Charged Particles by Atoms

In Section 13.1 we considered the scattering of electrons by an incident heavy
particle in that particle's rest frame in order to treat energy transfers to the
m ni a sa e of swift
particles, whether heavy or light, through matter because of interaction with the
atoms. Char ed articles are elasticall scattered b the time-avera ed otential
created by the atomic nucleus and its associated electrons. The potential is
tions (see above (13.5)]. At small angles, all particles, regardless of spin, scatter
accordin to the small-an le Rutherford ex ression
Sect. 13.5 Elastic Scattering of Fast Charged Particles by Atoms 641
Even at IJ 7rl2, the small-angle result is within 30% of the exact Ruthelford
formula. Such accuracy is sufficient for present purposes.
The singular nature of (13.52) as 8---> 0 is a consequence of the infinite range
of the Coulomb potential. Because of electronic screening, the differential scat-
tering cross section is finite at 8 = 0. A simple classical impact parameter cal-
culation (following Problem 13.1b) with a Coulomb force cutoff sharply at r =a
gives a small-angle cross section
du 1
(13.53)
d!l
where &min IS the classical cutoff angle,
(13.54)
A b!ltt!lr form of SCf!l!lll!ld Coulomb int!lraction is V(r) (zZe 2 /r)e-' 1a, with
113
a = 1.4 a0 Z- (from a rough fit to the Thomas-Fermi atomic potential). A
classical calculatiOn with such a potential giVes a small-angle cross sectiOn lor
8 ---> 0 that rises less rapidly than 8- 4 , but still is singular at 8 = 0. Quantum
mechanically. either the Born approximation or a WKB eikonal approach yields
a small-angle cross section of the form (13.53) with {)min the quantum-mechanical
cutoff angle
Z 113 me
{)(q)
111111
= -fi = - - .- (13.55)
where p i~ the incident momentum (p rMv), and m i~ the electron'~ rna~~- In

passing, we note that the ratio of classical to quantum-mechanical angles {)min is
T/ = zZe 2 /fiu, in agreement with the corresponding ratio of minimum impact
parameters [see below (13.16)]. For fast particl!ls in all but the highest Z
substances, T/ < 1; the quantum-mechanical expression (13.55) should be used
or min·
At comparatively large angles (but still small in actual magnitude) the scat-
tering cross section departs from (13.53) because of the finite size of the nucleus.
For charged leptons ( e, f.L, T) the influence of the finite size is a purely electro-
magnetic effect, but for hadrons ( 7r, K, p, a, etc.) specifically strong-interaction
effects also arise. Since the gross overall effect is to lower the cross section below
(13.53) at larger angles for whatever reason, we examine only the electromagnetic
aspect. The charge distribution of the atomic nucleus can be approximated
crudely by a uniform volume distribution inside a sphere of radius R, falling
sharply to zero outside. The electrostatic potential inside the nucleus is parabolic
in shape with a finite value at 1 0.
.? I .. ?. \
J"L-"' I 1 - -'--- ) r< R
2R 3R 2
2
ror ~U.JO)
zZe
- D
r
The classical scattering cross section from such a potential exhibits singular be-
havior at a maximum angle given approximately by the classical formula (13.54),
hHt mah n ---? R Thic nh. ic ~ ~An•OPrHoPn~P r.f thP " _,
"" '
!I\V) " '' V.
'->f'\V)If'
. ' . '
<tl V
. .
u, 11>111!; LV d
·o
' " ' LHau v
tum aL J u" "
A
'6''"
l(,
T
..
and falling again for larger b. The maximum translates into a vanishing derivative
dO/db and so an infinite differential cross section. The bizarre classical behavior
is the vestige of what occurs quantum mechanicallv. The wave nature of thP
T
- ...
.
-"'' .~. .
•
'
·o • •J ". "'·
•
•• '5 Vl
electromagnetiC waves by locahzed scatterers, dtscussed m Chapter 10. At short
wavelengths, the scattering is diffractive, confined to an angular range tlH = llkR
where k =pi h. Depending on the radial dependence of the localized interaction,
the scatterm<> cross sect10n mav exhihit wi<><>les or secondarv maxima anrl min.
ima, but it will fall rapidly below the point Coulomb result at larger angles. Said
another way, in perturbation theory the scattering amplitude is the product of
the Coulomb amplitude for a point charge and a form factor F(Q 2 ) that is the
ruuner u rm u1 me cnarge uisLriouLiuu. 11Ie tunu tacwr is uennea
2
to be unity at Q = 0, but becomes rapidly smaller for (QR) > 1. Whatever the
viewpoint, the finite nuclear size sets an effective upper limit of the scattering,
A
h 274 me "~
rnax nfi -:11!3 ,.n
\'J.Jf}
The final expression is based on the estimate, R = 1.4 A 113 X 10- 15 m. We note
that &max >> Om;n for all physical values of Z and A. If the incident momentum
ic cr. cm~ll th~t A . > 1 . th.P nnrlP~r ci?P h~c nr. i~hlP .u. r.n thP c~ot.
.--rurarraT «rrg=, <lmax --rwneiT jJ .JV !Vie VIC, CVii - r lU
50 MeV kinetic energy for electrons and 1.3 MeV for protons. Only at higher
energies are nuclear-sized effects important. At p = 50 MeV/c, Hmin = 10- 4 radian
in aluminum.
Tl-~ ,J h~hnui~r ~~ <hn • ch . v; •A >
·o· 1 "1 ."'.
The dot-dash curve is the small-angle Rutherford formula (13.52); the solid curve
shows the qualitative behavior of the cross section including screening and finite
nuclear size. The total scattering cross section can be obtained by integrating
rr
113.53) over the total solid angle
2
= J dfi
8 dH
U du .
Sill (J d() dc/J = 27T czZe · ( )" (13.58)
pu 0 (j
2
+ 2
(jmm
:
~-~ I
.
1 'I "'
I
I
~
I
--
~l'ii
o;;
1-l
I
I
-
1
"" I
I
-I
I "\ ~e•
I
T \ ·, -;:;; .1~/,; Atr.~i~

; '
\ " JIII,;IUUIH/; Ul <Ol<O'-llUlll'-
Bmin Bmax at small angles and finite nuclear size at

loge- large angles.
""ct. ~.>.o tnean ~ Ang•e or
'
l U< ......
.,LA'">
The result is
2' 2 I. 2' 2
_{ 2zZe \ Ll- 2 21 2zZe \ f11 <;o\
.I nn
~
fiv '
The final expression is obtained by use of (13.55) for l:lmin· It shows that at high
velocities the total scattering cross section can be far smaller than the classical
geometrical area 7ra 2 of the atom.
,,., A nuiP nt _,.,

: .
li.h ru· A n• r l.llstrwuuon
p .
OJ muulpte ~cuuering
~
~ ~
Rutherford scattering is confined to very small angles even for a point Coulomb
field, and for fast particles l:lmax is small compared to unity. Thus there is a very
large probability for small-angle scattering. A particle traversing a finite thickness
of matter will undenm verv manv small-an~>le oenectwns and Will generally
-" uu 0-'•~ oL d. . '1 .. •• 1. .L .
~w~·s~ u• u 3wuu u " ~ -1' - r • b'
number of deflections. Only rarely will the particle be deflected through a large
angle; since these events are infrequent, such a particle will have made only one
such collision. This circumstance allows us to divide the angular range into two
regions-one region at comparatively large angles, which contains only the single
scatterings, and one region at very small angles, which contains the multiple or
comoound scatterin!'s. The comnleteaistrioutwn m angle can be approximated
.-_,
L
VJ
+L
-~ -·
plural scattering must allow a smooth transition from small to large angles.
-'·
~·
'T'L -'' o+
-~
, 11 .A
The important quantity in the multiple-scattering region, where there is a

large succession of small-angle deflections symmetrically distributed about the
incident direction, is the mean square angle for a single scattering. This is defined
uy
J 1:12 aa dfl
dfl
(1:12) = (13.60)
Jdfl
dad
fl
ur;.~ ol-. .+ (' 1 ,., "- ,]..
~1'1'
(1:1 2) = 21:1~,;n lnl t:lmax) (13.61)

emin
If the quantum value (13.55) of l:lmin is used along with l:lmax (13.57), then with
A 'J7 (1'J.k1\h~e+ho"" ,] i'~r~·
' '\- ~ .
0 - H~·
\tl( "117min rm:zu'tL- ) \ l.::l.ULJ
If nuclear size is unimportant (generally only of interest for electrons, and per-
haps other particles at very low energies), l:lmax should be put equal to unity in
(13.61). Then inst~ad of (204Z- 113 ), the argument of the logarithm in (13.62)
I 1 fY'I 1/2
L
I ..!:. I
Z"" me I
644 Chapter 13 Energy Loss, Scatteling, Clterenko• and Transition Radiation 6
r-.:- ---------- -?1

'- _,-:/ I
/ '- I /
X
'--¥
Incident
direction
It is often desirable to use the projected angle of scattering ()',the projection

being made on some convenient plane such as the plane of a photographic emul-
sion or a bubble chamber, as shown in Fig. 13.7. For small angles it is easy to
show that
(13.63)
In each collision the angular deflections obey the Rutherford formula (13.52)
suitably cut off at &min and ()max• wlth average value zero (when viewed relative
to the for;.;ard direction, or as a projected angle) and mean square angle (fP)
given by (13.61). Since the successive collisions are independent events, the
central-limit theorem of statistics can be used to show that for a large number n
of such collisions the distribution in angle will be appwximately Gaussian around
the forward direction with a mean square angle (0 2 ) = n(8 2 ). The number of
collisions occurring as the particle traverses a thickness t of material containing
N atoms per unit volume is
n Ncrt (13.64)
This means that the mean square angle of the Gaussian is

2
2
(0") 21rNezze ) ln(llmax) t (13.65)
Or, using (13.62) for ( () 2 ),
(13 66)
The mean square angle increases linearly with the thickness t. But for reasonable
thicknesses such that the particle does not lose appreciable energy, the Gaussian
will still be peaked at very small forward angles Parenthetically, we remark that
the numerical coefficient in the logarithm can differ from author to author-for
Sect. 13.6 Mean Square Angle of Scattering; Angular Distribution of Multiple Scattering 645
example, Rossi has 175 instead of 204. We also note that in practice the Gaussian
'UHICO
.
approximation holds only for large n-see the last paragraph of this section for
. ·'. .
l <lllUH UH l l l " p v l l l l ,
The multiple-scattenng distnbutwn tor the proJecrea angJe or scanering is

1 (}'2
PM(e') de' expl IICA2\
d()' (13.67)
v 7T\'-"'! ,~ 1/
where both positive and negative values of ()' are considered. The small-angle
l:>n<J.. .~ . ..! r olo (1
,. "!<;')\~on
-, J..~
-r
;n Of thP nroiPI'fPn ~n<J]p ~- OOPrl
I \ 2
du = 1r I 2zZe 2 _..!.._ (13.68)
3
d()' 2 DU ()'
This gives a single-scattenng distributiOn for the proJectea ang1e:

1 2 2
PJ()') d()' Nt dud()' 7T Ntl 2z Z e ' I d()'

;,- (13.69)
£If] -y- flU u
The single-scattering distribution is valid only for angles large compared to (0 2 ) 112
oncl • otM to;] t~ tJ..~ nonce; On • ,; ot;~n
0
If we express angles in terms of the relative projected angle,

()'
/1'1'7(\\
u
/'"'2\112
\ -. 'I
the multiple- and single-scattering distributions can be written

'
PM( a) da = y;;;. e a' da
(13.71)
1 da
l'sta) aa
8 ln(204Z- ) a 3
113
where (13.66) has been used for (0 2 ). We note that the relative amounts of mul-
t;nJ, ~nrl ,;nul" · "'' ~r" · of thickness in these units and de
p~nd only on Z. Even this Z dependence is not marked. The factor 8ln(204Z- 113 )
has the value 36 for Z = 13 (aluminum) and the value 31 for Z = 82 (lead).
v: 1 '1 0 ,J.. •J.. d l- •'- · -~-.J..~ nnottPr;nn cJ;otr;hnt;~n oc o fo ·
'b' -. 0
of a. The transition from multiple to single scattering occurs in the neighborhood

of a = 2.5. At this point the Gaussian has a value of 1/600 times its peak value.
• • ,. .• ' • • .11 " • .: · ' · . _,
HlU> lHCO • UHb Ul'lHl . t;LVCOO UHlJ U o ~·) 01HUH \UH ~
-
scattering curve.
There are two things that cause departures from the simple behavior shown
m Fig. U.tl. The Uaussian shape IStnelimiiing rorm or me angmar aisuiomion
for very large n. If the thickness tis such that n (13.64) is not very large (i.e.,
n :S 200), the distribution follows the single-scattering curve to smaller angles
than rY = 2.5. and is more sharnlv neaked at zero angle than a Gaussian.* On
...
lH'-o
d •
VlH'-o1
L ~
HUH~ 0
'£
H
d.
lH'-o
d •• -'-
lO 151
-'-
'b' o
•J..
'1'
.1.
'b'
/(i\2\
o
becomes comparable with the angle ()max (13.57) which limits the angular width
-~ ' ' ' ,] · ..1;"' .;t · v ... ~·~n•~• •'-'-' '"''"'' thP mHlt;n]p. -
'b c-ce
*For numerical evaluation for very thin samples (e.g., gases), seeP. Sigmund and K. B. Winterbon.
"· _, ,_ "· ,,_ -'- "" , ."" ~.ry ""ry"
1
I I I I I
"'I
\
\PM(a)
'v
\
\ I
I
\:
slO 2f-
\:I
-
i:(
--- \
\
\ \}_
'v
_, 'R !\ "
""' ~
I\
I ,\
I
'u
I ~
0 1 2 3 4 5 6 7
a--
Figure 13.8 Multiple- and single-scattering distributions of projected angle. In the
· ~r nlnrol · (, ')_~\ t~ d· o~ d · ,1; • • th • ·o ;,;
fr;m the ;mall-angle mul~iple scattc'ring (approximately Gaussian in shape) to the wide-
angle single scattering (proportional to a 3 ).
ing curve extends in angle beyond the single-scattering region, so that there is
no single-scattering tail on the distribution (see Problem 13.8).
13.7 Transition Radiation

A charged particle in uniform motion in a straight line in free space does not
radiate. It was shown in Section 13.4, however, that a particle moving at constant
velocity can radiate if it is in a material medium and is moving with a speed
greater than the phase velocity of light in that medium. This radiation. with its
,. ,1 .< Ll lio.l/2\ · ~ -'' ·
'b' c \1'-' 1 , •v • -~~ ·~ ~ -~ •~•uoo~
IS another type ot radiation, translfzon radwtzon, hrst noted by limsburg and

;n 1 0.1.-; th >t · ~ ;tt, ,-1 .,h n} .-1 ' ,-1,-1, ,], f, m
one medium i~to another. Far from the bound~~y i;the first medium, the p"article
has certain fields characteristic of its motion and of that medium. Later, when it
is deep in the second medium, it has fields appropriate to its motion and that
medium. Even if the motion is uniform throughout, the initial and final fields will
• ~,££. ,... "' 1-. • _, • • • .• d.
ou ~ ~-.-~• ~"' ~-~~" . ~ "'~ 1-'' ~!-'~' oo~v. ~ "~~ •J
the fields must reorganize themselves as the particle approaches and passes
thr~nnh th, · .r T, t}' ,f ' ,,· ,f t} R >lrlo
sh~ken off as tra~sition r;diation.

'
are
Sect. 13.7 Transition Radiation 647
Ez
',,
' '......
• ------,;;,)1 ~·
P(x',
....,._E
k /: p
~ x/ :
i a i
:r / I z'
I
! \
I v
/ Eb
I
I
I
y
:/ :... i
8 "' ·------------- .__ ....._, I
I
'...j
Ea
" v
ze
Figure 13.9 A charged particle of charge ze and velocity v is normally incident along
the z axis on a anifunn serni-infinite dielectric rnediunr occupying the half-space z > 0.
The transition radiation is observed at angle ()with respect to the direction of motion
of the particle, as specified by the wave vector k and associated polarization vectors
E and E .
Important features of transition radiation can be understood without elab-

orate calculation.* We consider a relativistic particle with charge ze and speed
v = {3c normally incident along the z axis from vacuum (z < 0) on a uniform
semi-infinite medium (z > 0) with index of refraction n(w), as indicated in Fig.
13.9. The moving fields of the charged particle induce a time-dependent polar-
ization P(x', t) in the medium. The polarization emits radiation. The radiated
fields from different points in space combine coherently in the neighborhood of
the path and for a certain depth in the medium, giving rise to transition radiation
with a chmacteristic angular distribution and intensity.
The angular distribution and the formation length D are a direct consequence
of the requirement of coherence for appreciable radiated intensity. The exciting
fields of the incident particle are given by (11.152). The dependence at a point
x' = (z', p', 4>') on inverse powers of [p' 1 + ·l(z' - vt) 2 ] implies that a Fourier
component of frequency w (a) will move in the z direction with velocity v and so
have an amplitude proportional to e 1"'z'iv, and (b) will have significant magnitude
radially from the path only out to distances of the order of p/nax = yu!w. On the
*The """" fur a ~ualitative cliseussion has been impressed on me b) numerous questions from eo!
leagues near and far and by V. F. Weisskopf on the occasion of a seminar by him where he presented
.'
VlllCl 'HaHU,' lllC lllHC- pv•, aCT rcn:esa-wave wnose torm
m the radiatiOn zone IS
eikr
A =- exp[ -ik(z' cos fJ + p' sin fJ cos¢')]
r
whe:e_A is proportional to t~e ?ri~ng field of_the incident particl~, k = n(w)wlc
auu ILl~ lllal Llle laUlallVll" Uvovl leU 111 tne X-Z anU Ill Ille fOrWard
hemisnhere. Annreci~hl~e cnher~ent ·itinn frnm lt :in th~
dium will occur provided the product of the driving fields of the particle and the
generated wave does not change its phase significantly over the region. The rei-
evant factor in the amplitude is
I \
.I . w ' w .I .\ . w ' W \)f.' '
" <. '
A
"'t' '
"" "'t' "\~! vv~ v vAt' ' ll~ Mll 1/ '-'VO "
'f'
\ I -
w 1
= exp z- - - n(w) cos fJ z' exp -i w n(w)p' sinfJcos¢']
c f3 . c
In the radial direction coherence will be maintained only if the phase involving
p' is unity or less in the region 0 < p' :s p/nax where the exciting field is appreciable.
Thus radiation will not be annreci8hl~e 1m1~ess
- n(w) .!___sin fJ :s 1
c w
or
n(w)yfJ :S 1 (13.72)
for 11 >> 1. The angu ar distributiOn IS therefore confined to the forward cone
-yfJ :s 1, as in all relativistic emission processes.
The z' -dependent factor in the amplitude is
.w 1
exp z- ~- n(w) cos fJ z'
c
The deoth d( w) uo to whirh • is rna· --' is t
; [~ -n(w) cos fJ] d(w) = 1
ur~
-r~
;~~to ,( .•\
., -'. 1 (. 2 /'")
-' p
2\ f.
I
<.
"1
~t. ... .,
_,
-o
wnere LnerenKov ramation aoes not occur, f:J - "T"TTZY tor a relatlvistlc
~ncl "" • {I
1 ·~
<
2-yclwP
d(v) = 1
(13.73)
v+ v
. . .
wutac we ua vc tu•xu a uHHeu>IOmess 1requency vanao1e,
w
v- (B./4)
-ywP
Sect. 13.7 Transition Kadiation o..,
we aenne the tormatwn ten(?tfl u as the largest value or a( v) as a runctJon or v:
D = d(1) = ')'C (13.75)

wP
For substances with densities of order of unity, the plasma frequency is wP -

3 X 10 16 s-\ corresponding to an energy liwP- 20 eY. Thus cfwP = 10- 6 em and
even for 'Y :c 103 the formation length D is only tens of micrometers. In air at
NTP it is a factor of 30 larger because of the reduced density.
The coherence volume adjacent to the particle's path and the surface from
which transition radiation of frequency w comes is evidently
I
. \ 3
1
V(w) 'TTO~ :7wJ dJw) 27T1il
Wp ll(l -r v-)
This volume decreases in size raoidlv for v > 1. We can therefore exoect that in
~~,.,~ • HT -;:-
~ -r "~ rof tr~noitiron ro~iotiron ,;11
-o
extend up to, but not appreciably beyond, v - 1.
We have obtained some insight into the mechanism of transition radiation
and its main features. It is confmed to small angles in the forward direction
( ')'8 :'S 1). It is produced by coherent radiation of the time-varying polarization
in a small volume adjacent to the particle's path and at depths into the medium
up to the formation length D. Its spectrum extends up to frequencies of the order
of w- ')'Wp. It is possible to continue these qualitative arguments and obtain an
estimate of the total energy radiated, but the exercise begins to have the ap-
pearance of virtuosity based on hindsight. Instead, we turn to an actual calcula-
tion of the phenomenon.
An "X"rt rolrnbtiron of · ion rocli>~tion io '. ·' l<omP .<. lC'PO .
n-n ~·
6
on- n< <l ,.., ·' ., ~t-
UL . ' ,.,
. . ~t't'
LITIILIS<laeauare Ior mosT s ana IS pnys1cauy rransparem.

It IS based on the observatiOn that for freauencies above the optical resonance
rPo-ion ~ inrlPv~ " io not for fro-;, nnihr ThP inriciPnt "rlP'o fip]cJo
' e ·~
_,. ·' .
~ , aU..- HVL HlJ Ulll\01\0lll lH lllv a u u lH
L '·
This means that the Fourier component of the induced polarization P(x', w) can
be evaluated approximately by
1
P(x', w)- [ E(w17T- JE;(x', w) (13.76)
where E; is the Fourier transform of the electric field of the incident particle in
UOPnnm .Tho · Af tho mouo rorliotorl hu tho nAI · · mnot ho rio_
., -..-..,"'
~
r
.I \< ._,, d . -;]'
~ .vpv 'Y• uv ''"' ''"''• vuu '"'-' vav'"' r Ul ' \ Ul )IL "T r J}-'llaU.- LV
Lire meaium. TITis isoecause pnaseairrerences are important, as already seen m

the nualitative discussion.
~0 ' . rorl;ot;An fiolrl frAm tho nAlor;?Ot;An Pfvl
marx-~~.
,.
:}", ,.
,. .,\ A3 v ;n tho uAlnmo
• u "<
. -/
1
m e ,. • r1
R , ... , ..~·)
I
Ubrad A ... u A
where k is the wave vector in the direction of observation and R - r - k . x'

With the SUbstitutinn nf f11.7hl ~nrl ~n in inn Cl\/Pr thP h~lf-<n~,.., ~' ' ,;
the total radiated field at frequency w is
Emd = -
e
r c
E(w) - 1 kz
47T
J. (kz'.>O
X E;) X ke-ik·x' d 3 x'
With the approximation,

2
p
/
C~WJ
' '.J. ~.U./1)
w2
LITe rna1atecrneraTor w-::> w"Lfecomes
e'h 1 -w~ l (
v flr v 1<'.\
..,
y lr -ik·x' ,.J3v'
0 {1'1 '70\
r 47TC
2 Jz'>O ,. \ • Jj
~~~-UUJJ,lH" ' ~~' ' ' ' fM' .1•

"llldl lllC ClltOigy
radtated has the dltterentlal spectrum m an angle and energy,
I IJ [k X
>2 T \ 4 r 2
<T L L p
3
= E,(x, w)] X ke-ik·x d x (13.79)
dw d[! 327T3 c z>O
' . .
we uuvmg ue1us I'.; a1e
1-.uLe l!laL uy Lue r u uner u nu u1 Lne ue10s
of Section 11.10. In our approximation it is not necessary to use the more elab-
~" <n • ,,:,-~ ''~ • ,, • r~· ~n.l · .,
~' .J..J . .J dllU .J..J, . HI lllC VI .J."I!',• .J..J.7 lll<O
fields are (see Problems 13.2 and 13.3)
~ I \
E,(x, w) = ~- _ e''"zlvKI( 2)
' 7T yv 2 yv
f11R())
/2 7Pt.l ',,,n
.t.z(X, w) - zV·; ;. yzvz e "Ko
yv
The integral in (13.79) can be evaluated as follows. We first exploit the fact that
the z dependence of E, is only via the factor e 1wzlv, and write
F = J [k
z>O
X E,(x, w)] X ke-ik·x d 3 x
r ( r~
l
= diJ dylk X E;]FD X ke ""x")., dz exp l - - k cos
v
e )z
w
!'T exp !I k cos IJ jZ
,v I
( dx ( dvrk x El. " x ke ikxsinfJ
w J J
--kcosiJ
v
Tile upperilmlt Z on the z mtegrat10n 1s a formal dev1ce to show that the con-
tributions from different z values add constructively and cause the amplitude to
-g-row~-v:-ITeyonirTn"e-rreplllVTlle rapirny rotating pnase prevents runner
enhancement. For effectively semi-infinite media (slabs of thickness large com-
Sect. 13.7 Tiansition Radiation 651
pared with D) we drop the oscillating exponential in Z on physical grounds* and

obtain, for a single interface,
F= ( w i
- - k cos(}
) If dx dy[k
v
The electric field transverse to k can be expressed in terms of the components
Er, Ez and the polarization vectors Ea and Eb shown in Fig. 13.9 as
[k X EJ X k- (EP cos tJ cos </J
where (} is the polar angle of k and the prime has been dropped from the a:d-
muthal angle of integration. The component parallel to Eb integrates to zero
because it is odd in y. Thus, substituting from (13.80), we have
lEa rr k . 8 l X
lEa (2 zew ( r dx dy e ikx>ine
~-
2
( k cos (}) & }V J)
[ x ( w 12 2) sin (} ( w 1 2 2) J
K1 VX +y K0 VX +y
\lx 2 + y 2 yv w iJx yv
ze sin e( k cos (} u> )

vl I
F = Ea
2
- {( dx dye ikxsin fl Ko ( -w \lx2 + l
\
v w ' I
v --kcosel
1!
The remaining integral can be evaluated from the cosine transform,
The result for F is
2y'2; ze sin e(k cos (}

~
- w
try
2) ~
(13.82)
F = Ea ( &J )( w2 )
*A less cavalier treatment of the dependence on thickness is necessary for foils that are not thick
compared taD, ar when a staek af foils is emflloyed. See Problems 13.13 and 13.14.
In the approximation of relativistic motion ( y >> 1), small angles (IJ << 1), and
high frequencies (w >> wp), this becomes
F (13.83)
where v IS the d1menswnless frequency variable (13.74) and Y/ - ( y1J) 2 is an

appropriate angular variable. With dfl = d¢ d(cos 1!) = d¢ d"l)l2y 2 , the energy
distribution in v and Y/ is
(13.84)
Angular distributions for fixed v values are shown in Fig 13 10 At low frequen
cies the spectrum peaks at 17 1 and then falls relatively slowly as 11 1 until the
value Y/ v 2 IS reached. I hen It falls off as .,., 3 • For v > I, the spectrum peaks
at 71 > ~ and falls at 71 3 for 71 :» 1. At 71 0 the denominator in (13.84) is
I T I I
-
-
• = 0.1
10- 1 ,-
I ',
.....
P. \, -
It \
!
,:,01-..
""1-..
..... ,....
1--
""\\ '\
'' '· u'>'> I

' ,,
•v
-2
"•=!'
' ' ', ' '
f- ', -
f- ''
1-- '
' ...................
1--
..................
1-- ................ -
....
10-3 I I I I I
0 2 4 6
Figure 1310 Angular distributions of transition radiation at 1' () 1, 1' 1 and 1' >> l
The solid eur.es are the ltOJillalizcd angular distributions, that is, the ratio of (13.84) to
(13.81). I he dashed curve rs v4 trmes that rat10 m the lrmil v--"' oo.
:Sect. u. 7 Transition Radiation ()!)j
~1 -r v-r, snowmg Lnat wr v >> 1 there IS negligible mtens1ty at any angle [cf.
coherence volume V(w), above].
'T'h '"' •< .1 •• 1...1
"OJ ~r 'uuuv•~ 'I> '"
' ~
dl z2e2ywp .I
1\ ..,
{1 .., 2\ , 1 _l_ I _Li 0~\
dv 7rC
\ I
vz ,~~.uJ!
It has the small and large v limits,

r
dl z 2e2ywP 2 ln(1/ev), v << 1

dv = 'TTr 1 (13.86)
v ---~ J.
6v 4 '
'T'h . 0 h 1 1 .1 • ,..,, ' " ' ' 'T'l.
_w '. 'b. Jb p•v•"' ~ 'b· ~J-'·'·· ~"~ op~~· IUlll
diverges logarithmically at low frequencies, where our approximate treatment

ralls m any event, but It has a finite integral. The total energy emitted in transition
radiation per interface is
I= o dvr
dl dv =
z2e2ywP
3c =
zz
3(137) yhwP (13.87)
10
' I I
~ -
~
lt- \~ -
1-
\ -
\\
~\
~~~ 10-1 t-
_,...., -
~
r- -
I
I
I
•n-2 I
r
~\1 -
\
10-3
10-·
I
10-l
I
I
\ 10
p = w/'Yw,
1~ 11 '-1. '1; r1 f. 11/T\f.JTU •.\ .< .. . ~'
a functwn
. of v = w I ywP. 'h
f e h" '"' ., .
d as hed curves are t e two approximate expressiOns m . . ao
\U.UUJ•
From Fig. 13.11 we can estimate that about half the energy is emitted in the range
O.l < v < 1. In quantum language, we say that an appreciable fraction of the
energy appears as comparatively energetic photons. For example, with y = 1Q3
and hwP 20 e V, these quanta are in the soft x-ray regwn or L to LU Ke v.
The presence ofthe factor of yin (13.87) makes transition radiation attractive
~e ~ ,.,,"h"ni'm for the identification of narticles and nerhans even measurement
of their energies, at very high energies where other means are unavailable. The
nresence of the numerical factor 11(3 X 137) means that the probability of en
ergetic photon emission per transit of an interface is very small. It is necessary
to utilize a stack of manv foils with gaps between. The foils can be quite thin,
needing to be thick only compared to a formation length D (13.75). Then a
particle traversing each foil will emit twice (13.87) in transition radiation (see
Problem 13.13). A typical set-up might involve 200 Mylar fmls or thicKness 20
JLm, with spacings 150-300 JLm.* The coherent superposition of the fields from
the different interfaces, two for each foil, causes a moauTatwn or tne energy and
angular distributions (see Problem 13.14).
,,l'"' ··oo .. .... •• ·o

D I'
..... ._J ._.
··-
~·
.l-1 .C •L .-<' NiP I< Rrohr
,f · -' mottor :nt "
' Y ~ r ··o
all his life. A lovely presentation of the whole subject, with characteristic emphasis on the
' of classical and auantum-mechanical effects, appears in his comprehensive
review article of 1948:
Bohr
"Numef!CaJ tableS and grap!IS (IT energy·JOSS Uala, aS~ <UUUU""• aHO b;'""
by
· rot.
n
, j ~, ·"
Segre, article by H. A. Bcthe and J. Ashkin
~oo ole
U. Fano, Annu. Rev. Nucl. Sci. 13, 1 (1963).

Rossi gives a semiclassical treatment of energy loss and scattering similar to ours. He also
considers the question of fluctuations in energy loss, including the Landau -:>yrnon meory.
pffprt ron thr pnerav loss bv extremelv relativistic narticles is discussed,
'•L
""" ,.lt~ C. ·'- >C • ~r·onh:~ol~~
o..r
R. M. Sternheirner, in Methods of Experimental Physics, Vol. SA, Nuclear Phys-
ir-< Pnrt 11 Prl< L. C. L. Yuan and C. S. Wu Academic Press, New York (1961),
pp. 4-55.
-'· "·' . "' .. nl Tt o .n];not:An tro • rl
~·
- 'J r " ·rr
IS descnoed Ill tne oooK oy r uan anu vv u, jus1 me · , auu at>u ;.,
n M Rit<ron Prl TPrhninuP, in Hi<'h Ener<'V Phvsics Interscience New York
,.
\V'")•
L<
..
L JVL
"
tranK, Usp. r1z. Nauk IH, 1"" \1 '1m J Lrranst. ,)OV. rny>. v>y. o, , "-7 \ uuu;" J·
... ,...,. .. b • • ,.
• ... h
. .
F. G. Bass and V. M. Yakovcnko, Usp. Fiz. Nauk 86, 189 (1965) [trans!. Sov.
"' ~
'Hy>• Vofj., U 0 .LU \I7'<TJ7T•
_, -< . ' • f, .~·.u o:~'· " · -1 '· ·' · ' · A ~li7 ??n
'
(1995), W. Bruckner et al., Nucl. Instrum. Methods A 378,451 (1996), and J. Ruzicka, L. Krupa, and
V. A Fadeiev, Nucl. Instrum. Methods A 384, 387 (1997).
,,..,..
Ln. ~u l'roblems "--'--'
The calculation of transition radiation from the traversal of interstellar dust grains by
energetic particles, done in the same approximation as in Section 13.7, is given by
L. Durand, Astrophys. J. 182, 417 (1973).
A review of both Cherenkov radiation and transition radiation with much history, is
given by
A' ~..
V • L. -.:::rrmwurg; "Y· "~· Hc<Ur. ~""• w J J \P7UJ L'=·
oA"
"7·'·
' , "~"
Co o
"~.
(1996)1.
For current applications of both Cherenkov and transition radiation, however, the reader
must turn to specialized journals such as Nuclear Instruments and Methods A. Volume
367 of that journal (1995), a conference proceedings, contains descriptions of several par-
tide physics detectors based on these and other principles.
Problems
13.1 If the light particle (electron) in the Coulomb scattering of Section 13.1 is treated
classically, scattering through an angle II is correlated uniquely to an incident tra-
jectory of impact parameter b according to
b = zc ""
m• cot;;-
0 au
where p = ymv and the differential scattering cross section is au =
dH sine de·
(a) Express the invariant momentum transfer squared in terms of impact param-
e1er anu snow """ "'~ . o. <WHM~' ' \U) "
2z2e4 1
T(b)
mu~ 0~
I U~in
2 2 2
nun = ze /pv and T(O) = T max = 2y{3 mc ·
where b(c)
(b) Calculate the small transverse impulse Ap given to the (nearly stationary)
light particle by the transverse electric field (11.152) of the heavy particle
q ze as It passes by at large 1m pact parameter D m a (nearly) s1rmgm nne
''" ... , ·~~;, v;ncl thP ~ tron<fPr T IA.n\ 2 1?~ ;n tPrrn< nf h ~om-
r T •
pare with the exact classical result of part a. Comment.
13.2 Time-varying electromagnetic fields E(x, t) and B(x, t) of finite duration act on a
charged part1cle ot cnarge e ana mass m oouna narmomcauy w me origin wiu,
natural frequency w0 and small damping constant r. The fields may be caused by
a passing charged particle or some other external source. The charge's motion in
resoonse to the tieiOs 1s nonreJatlvJStJc ana small m amp11tuae compareu LO •ue
uau"erreuo:o--.:rre-
.
W
.
,,
" ' " HUH< UO o " ' , >HWH
...
scale of spatial variation of the fields (dipole approximation). Show that the energy
. . .
1TeL
t:.E = -IE(wu)l 2
'"
where E( w) is the symmetric Fourier transform of E(O, t):
E(O, t) = , ,r
!') ~ ~
~ dw;
E(w)e •w• E(w) = ,-k J ~ E(O, t)e'w' dt
U.J Thg gxt@rnal fi@ids of Probi@m 13.2 are caused by a charge ze passing tile origin
in a straight-line path at speed v and impact parameter b. The fields arc given by
11.152 .
(a) Evaluate the Fourier transforms for the perpendicular and parallel compo-
nents of the electric field at the origin and show that
where /; = wb/yv, and Kv(/;) is the modified Bessel function of the second
kind and order v. [See references to tables of Fourier transforms in Section
13.3.]
(b) Using the result of Problem 13.2, write down the energy transfer D.E to a
harmonically bound charged particle. From the limiting forms of the modified
Bessel functions for small and large argument, show that your result agrees
with the appropriate limit of T( b) in Problem 13.1 on the one hand and the
arguments at tile end of Section l3 .1 on the adiabatic behavior fat b » yvlwo
on the other.
13.4 (a) Taking /i(m) 12Z e'l in the quanturn-rnechanical eneigy-loss fonnula, cal-
culate the rate of energy loss (in MeV/cm) in air at NTP, aluminum, copper,
and lead for a proton and a mu meson each with kinetic energies of 1O, 100,
1000 MeV.
(b) Convert your results to energy loss in units of Mev · (cm21g) and compare
the values obtained in different materials. Explain why all the energy losses
in J'l4gV-(cm2 /g) arc within a factor of 2 of @aci! other, whereas the values in
MeV/em differ greatly.
U.S Consider the energy loss by close collisions of a fast, but nomelativistie, heavy
particle of charge ze passing through an electronic plasma. Assume that the
screened Coulomb interaction V(r) - ze 2 exp( -k,r)/r. where k 0 is the Debye
screening parameter, acts between the electrons and the incident particle.
(a) Show that the energy transfer in a collision at impact parameter b is given
approximately by
where m is the electron mass and v is the velocity of the incident particle.
(b) Determine the energy loss per unit distance traveled for collisions with im-
pact parameter greater than bmin· Assuming kubmin << 1, show that
where b'"'" is given by the larger of the classical and quantum minimum
impact parameters [(13.16) aHd abo, e].
13.6 The energy loss in a plasma from distant collisions can be found with Fermi's
method for the density effect. Consider the nom elativistie limit of (13.36) with the
relal!ve dielectnc constant of a plasma given by (7.59) augmented by some
w + iwl'
Ch. 13 Problems 657
Assume that the arguments of the Bessel functiOns arc small ( correspondmg to a
speed of the incident particle large compared to thermal speeds in the plasma).
(a) Show that the energy loss (13.36) for knb > 1 becomes
(dE)
dx koh>l
=
mJ
2 2
2z ~ r Re(..!.!!!_)
)o s(w)
ln(l.123knv) dw
w
(b) With the assumption that r << wP in s( w ), show that the formula of part a
yields
Combine with the close-collision result of Problem 13.5 to find the total en-
ergy loss of a nonrelativistic particle passing through a plasma,
where A is a number of order unity. The presence of wP in the logarithm

suggests that the energy loss may be quantized in units of hwr In fact, elec-
trons passing through thin metallic foils do show this discreteness in energy
loss, allowing determination of effective plasma frequencies in metals. [See
H. Raether, Springer Tracts in Modern Physics, Vol. 38, cd. G. Hohler,
Springer-Verlag, Berlin (1965), pp. S4-157.]
13.7 With the same approximations as were used to discuss multiple scattering, show
that the projected transverse displacement y (see Fig. 13.7) of an incident particle
is described approximately by a Gaussian distribution,
P(y) dy =A exp[ 2~::)] dy

where the mean square displacement IS (y') - (x 2/6)(0 2 ), x bemg the thickness of
the material traversed and (8 2 ) the mean square angle of scattering.
13.8 If the finite size of the nucleus is taken into account in the "single-scatlei ing" tail
of the multiple-scattering d!stn5ul!on, there 1s a cnttcal thickness x, beyond which
the single-scattering tail is absent.
(a) Define xc in a reasonable way and calculate its value (in em) for aluminum
and lead, assuming that the incident particle is relativistic.
(b) For these thicknesses calculate the number of collisions that occur and de-
termine whether the Gaussian approximation is valid.
13.9 Assuming that Plexiglas or Lucile has an index of retraction of 1.50 in the visible
region, compute the angle of emission of visible Chcrcnkov radiation for electrons
and protons as a function of their kinetic energies in MeV. Determine how many
quanta with wavelengths between 4000 and 6000 A are emitted per centimeter of
path in Lucile by a l MeV electwn, d 500 MeV p10tun, and a 5 GeV plUton.
13.10 A particle of charge ze moves along the z axis with constant speed n passing
frequency w is given as a functjon of spatial coordinate X by
ze
<P(w, x) = - (-)
VE W
h'iM 11 '( gnrJ
.. ion _(
I·'"'·
""."'" '· auu I'

VA
2
y
2 ,. . '
a•" ""'" '-Y """""a'
. ,.
V
£ "'- .L
1H'- VU0'-1
pomt.
(b) Assuming that E is independent of frequency and that {32E < 1, take the
. . . .
rouner transrorm wnn respect w w or me expressiOn m pan a ana ontam
ct>(x, t). Calculate the electric and magnetic fields and compare them to the
"-1..1 i l l 1"1\ CJ.. •J.. -•1. •<' •1. <.
\ "!' _, -o
' y
,, rt:piac:c:u uy r p E) '" '
~·
(c) Repeat the calculations of parts a and b with {32< > L Show that now
ct>(w,x) = ~ /;e;w,lv[ -No(lwl Py{3zE-

VE(w)'/2 V
l) :±: ilo(lwl PY{32E-1)]
V
10r w"" '" LarcmaLe llle remaining rourier rransrorrnw uoLain cv~x, '!· KeJate
your answer to the result given in Section 13.4 fur A( x, t) .
.tJ •.LL A magnetic monopo1e wiLn magneric cnarge g 1' • • . Lnruugu anu 1oses
energy by collisions with electrons, just as does a particle with electric charge ze.
(a) In the same approximation as presented in Section 13.1, show that the energy
loss per unit distance is given appruxnnately by (13.14), but With {3g, ze-->
yielding
'rl F' !72P2 I 1 ,;2~,2
47INZ In(
dx }magnetic mc2 h(w)
monopole
' '
VVIllllllt; lJlldt: t:UllUlllUU U. UJ) U\Clt;J JJJJJJIJJ!; lJJt; ~
\UJ
'""
z
charge, what value is necessary for an ordinary charged particle in order
that it lose energy at relativistic speeds at the same rate as a monopole?
~ e · ' e 'r". · _' ~· "
W1 <H'- 0 '"" "' 1 a ""' " V1 "'i' •w '"' 'IS· u.'
and comment on the differences.
.L , ' ' -' .£
' "'"" '"L
'
~vo~ ~ '"w'"'"'" I""""'" u< '-"«11S" '"
~v wu '"o '"""IS <. 'I'
z
f3c. lhe halt-space :s: U IS hlled w1th a umtorm ISotropic d1electnc medmm With
z
plasma frequency w1, and the space > 0 with a similar medium whose plasma
,_ 1
'
2
h
' .... ' ' ·'
h f ' .
.
'
f s '
,. '
. 13 7
' '" y ' : ~.. ' ' ..
verses t e mter ace, usmg t e approx1matwn o cctwn •..
'1'
(a) Show that the radiation intensity per unit circular frequency interval and per
unit so1ia angle is given approximately ny
2
d 2l z 2e2fP 1 1
dwdf1 ~c 1 WT 1 w~
y + -w2+ rP
- y + -w2+ rP
-
'
wuere u IS Lilt:_ ~ugre ur lC:HillVC lU
'
Lilt: Vt:IUt:lly Ul Lilt; f'W u dllU
y = (1 - {32)-112.
\UJ ~UV wa< "''-' 1u<a1 """'ISJ 10
I=~- (w, w2t

·y
3c (w1 + w2)
n. 1 'l r, ' ..
th ~, .;,; ,-1; ..1
' c
o trouoro;no • rJ;_
-o
h ,I. . ;0,;~
electric foil of thickness a perpendicular to its path. Assuming that reflections can
be ignored because
l[n(w)- 1]/[n(w) + 1]1
IS very small, show t11at tile dttterentJal angular and trequency spectrum 1s g1vcn
by the single-interface result (13.84) times the factor,
I 1 a
';!!• - 4 sm'I"J, With \") - vi I + 7 + "' I~
~~
\
Ch. 13 Problems 659
Here D = yc/wr is the formation length, v = wlyw"' and n = (y0) 2 . Provided

a >> D, the factor ?F oscillates extremely rapidly in angle or frequency, averaging
to ('Ji) = 2. For such foils the smoothed intensity distribution is just twice that for
a single interface. Frequency distributions for different values of r = 2D/a are
displayed in Fig. 1 of G. B. Yodh, X. Artru, and R. Ramaty, Astrophys. J. 181, 725
13.14 Transition radiation is emitted by a relativistic particle traversing normally a uni-

form array of N dielectric foils, each of thickness a, separated by air gaps (effec-
tively vacuum), each of length b. Assume that multiple reflections can be neglected
for the whole stack. This requires
2
n(w) - 1 wP 1
=-<<-
n(w) + l 2
4w N
(a) Show that if the dielectric constant of the medium varies in the z direction
as E( w, z) = l - ( w~l w2 )p(z), the differential spectrum of transition radiation
is given approximately by the single-interface result (13.84) times
'Ji = IlL I dz p(z)e'wzlv exp( -i cos e r k(z') dz') 1

2
where p(O) = l by convention, 1-' = wlv - k(O) cos II, and k(z)
(w/c) YE(w, z).
(b) Show that for the stack of N foils
"" _ . '"' sin [N(0 + 'It)]

;Je -
4 sm2 "' . 2 [""
Slll 6 + ''']
'J."
where ® is defined in Problem 13.13 and 'It = v(1 + rt)(b/4D). Compare

G M Garihyan Zb Ebp Teor Fiz 60, 39 (1970) [trans! Sov Phy;· IETP
33, 23 (1971 )].
The practical theory of multilayered transition radiation detectors is
treated in great detail by X. Artru, G. B. Yodh, and G. Mennessier, Phys.
Rev. D 12, 1289 (1975).
U.lS (a) Find the number NY of transition radiation quanta with frequencies greater
than wP emitted per interface, starting from the energy spectrum (13.85).
Show that for y >> 1,
I 1T C
where terms of order lly have been neglected.

(b) Using the result from part a for the number of photons and the value
fuoP 20 eV, find the mean energy of the photons (in ke,T) for 1' 103 ,
104 , 105 .
13.16 A highly relativistic neutral particle of mass m possessing a magnetic moment 1-'
parallel to its direction of motion emits transition radiation as it crosses at right
angles a plane interface from vacuum jnto a dielectric medium characterized at
high frequencies by a plasma frequency wP. The magnetic moment 1-' is defined in
the particle's rest frame. [The particle could be a neutron or, of more potential
interest, a neutrino with a small mass.]
(a) Show that the intensity of transition radiation is given by (13 79), provided
the electric field of the incident particle E,, is given by (f3JL!yze) times the
partial derivative in the z direction of EP in (13.80). Note that the electric
field actually points azimuthally, but this affects only the polarization of the
rad1atwn, not 1ts mtensity.
(b) Show that m the combmed hmit of y >> 1 and w >> w1, the intensity distn-
hutions in angle and frequency arc given by (13.84) and (13.R5), each mul-
(c) By expressing JL in units of the Bohr magneton J.Ln - en/2mec and the plasma
frequency in atomic units (hw0 = e 2 /a 0 = 27.2 eV), show that the ratio of
frequency distributions of transition radiation emitted by the magnetic mo-
ment to that emitted by an electron with the same speed is
2 2
4
d!,Jv) _ a (..f!:_) (hwp) . ,7
die( v) - 4 J.Ls hwo
where a = 11137 is the fine structure constant and v = w/ywP is the dimen-
sionless frequency vanablc.
(d) Calculate the total energy of transition radiation, imposing cons!lrvation of
energy, that is, v < Vmax = mc/hw". [This constraint will give only a crude
estimate of the energy in the quantum regime where v'""' < 1 because the
derivation is elassieal throttghout.] Show that th€l ratio of total energies for
the magnetic moment and an electron of the same speed can be written as
I = ;~ (:J G:Y · G(vmax)
where G = 1 for Vmax » 1 and G = (10 z?.axhr) . [ln(l/ Vmaxl -2/3] for
Vma> << 1. For fixed particle energy and magnetic moment, how does the
actual amount of radiated energy vary with the particle's mass for very small
Hint: the integrals of Section 2.7 of Gradshteyn and Ryzhik may be of use, a!-
though integration by parts is effective.
CIIAPTER 14
Radiation by lv!oving Charges
It is well known that accelerated charges emit electromagnetic radiation. In

Chapter 9 we discussed examples of radiation by macroscopic time-varying
charge and current densities, which are fundamentally charges in motion. But in
one class of radiation phenomena the source is a moving point charge or a small
number of such charges. In such problems it is useful to develop the formalism
in a way that relates the radiation intensity and polarization directly to properties
of the charge's trajectory and motion. Of particular interest are the total radiation
emitted, the angular distribution of radiation, and its frequency spectrum. For
nonrelativistic motion the radiation is described by the well-known Larmor result
(see Section 14.2). But for relativistic particles a number of unusual and inter-
estmg effects appear. It rs these relauvrstic aspects that we wish to emphasize. In
the present chapter a number of general results are derived and applied to ex-
amples of charges undergoing prescribed motions, especially in external force
Deflection of ultrarelativistic electrons in magnetic fields found in accelera-

tors, but also in plasmas and astrophysical contexts, leads to copious emission of
radiation called "synchrotron radiation." The basic properties of synchrotron
radiation are derived in Sections 14.5 and 14.6. The broad frequency spectrum,
often corresponding to millions of harmonics of the basic frequency of particle
motion, finds uses in solid-state physics and biology wherever intense beams of
x-rays are desirable. These apphcatrons have led to the creatiOn of dedrcated
"light sources" with special "insertion devices" called wigglers and undulators.
The physics of these magnetic structures, designed to produce spectral "lines"
(actually narrow peaks) of very high brightness and adjustable photon energy, is
discussed in Section 14 7
14.1 LienartJ. lViechert Potentials and Fields for a Point Charge

In Section 12.11 it was shown that if there are no incoming fields the 4-vector
potential caused by a charged particle in motion is
4 '
(14.1)
where D,(x x') is the retarded Green function (12 133) and
l'"(x') - ec { dTV'"(T) o< 4l[x' - r( T)j (14.2)

J
661
662 Chapter 14 Radiation by Moving Charges-G
is the charge's 4-vector current, V"( T) its 4-velocity and r"( 7) its position. Inser-
tion of the Green function and the current into (14.1) gives, upon integration
over d 4 x',
r
A"(x) = 2e J dTV"(T) 8(x 0 - r0(7)] o{[x- r(TW) (14.3)
The remaining integral over the charge's proper time gives a contribution only
at T- T0 , where To is defined by the light-cone conditiOn,
ry - r( 'To) 12 - () (1d. .1\
and the retardation requirement x 0 > r0 ( 70). The significance of these conditions
is shown diagrammatically in Fig. 14.1. The Green function is different from zero
only on the backward light cone of the observation point. The world line of the
particle r( T) intersects the light cone at only two points, one earlier and one later
than x 0 . The earlier pomt, r"( To), IS the only part of the path that contnbutes to
the fields at x". To evaluate (14.3) we use the rule,
5:{. \
o[f(x)] 2::
i I dj'
'
II~
dx ~
FX
'
where the noints x = ~L <~re the zeros of f(x)~ assumed to be linear. We need
d
~ [x- r(TW = -2[x- r(T)] 0 Vi!(T) (14.5)
dT
evaluated at the one point, T = T0 . The 4-vector potential is therefore
eV"(T)
K'(x) (14.())
V · [x- r(T)] FTo
wncre To 1s aennea oy (14.4) ana tne retaraauon requirement.

The potentials (14.6) are known as the Lil!nard-Wiechert potentials. They
..
an: uue11 lll 11a11L, UUL )JC111a[l' lllU1t: , 101111 ' " llJIIlJ" ~.
The light-cone constraint (14.4) implies x 0 - r0 ( To) = Ix - r( To) I - R. Then

T7. fv rl - v_r
,, V~ rJ~_\1 v. rv
c-
,.( ~- \1
''
,~
"
= -ycR - -yv • nR (14.7)
-ycK(l J.i • ll)
Time
~ ~ /
~
·'- I '- /
/
A
~
r(To)
/ ·"pace
/
\ " .
-~~
, 1A 1
Sect. 14.1 Lienard-Wiechert Potentials and Fields for a Point Charge 663
where n is a unit vector in the direction of x - r( T) and f3 = v( T)!c. The potentials

(14.6) can thus be written
<P(x, t) = lc1-; · n)RL,'

A(x, t) = l ef3
(1 - (3 • n)R
J cet
(14.8)
The subscript "ret" means that the quantity in the square brackets is to be eva!-
uated at the retarded time To, given by, r 0 ( To) = x 0 - R. It is evident that for
nonrelativistic motion the potentials reduce to the well-known results.
The electromagnetic fields F"fl(x) can be calculated directly from (14.6) or
(14.8), but it is simpler to return to the integral over dT, (14.3). In computing F"fl
the differentiation with respect to the observation point x will act on the theta
and delta functions. Differentiation of the theta function will give o[x0 - r 0 ( T)]
and so constrain the delta function to be 8(- R 2 ). There will be no contribution
from this differentiation except at R = 0. Excluding R = 0 from consideration,
the derivative iJ"A 13 is
iJ"Afl = 2e J dT Vfl(T) ll[x 0 - r0 (T)] iJ"o{[x- r(T)] 2 } (14.9)

The partial derivative can be written
d dT d
a"o[fl = a"f · - o[fl = i!"f · - ·- o[fl
df df dT
where f = [x - r( T)?. The indicated differentiation gives
a"o[f] = - (x - r)" !!:_ o[fl

V · (x- r) dT
When this is inserted into (14.9) and an integration by parts performed, the result
IS
iJ"Afl 2e
r
I dT
A
,.
{. 00 \aT7/3
I
orxo rc( T) 1 OHx r( TWl (14.10)
J aT V ·~X r)
Hl Lilt:
lallUH uy 1''"'' '"" lllUll Ul lllt: lllt:ld gtves uu cuu·
tribution, as already indicated. The form of (14.10) is the same as (14.3), with
V"( T) rPnl~rPcl hv thf' clPrivMiv~e tcerm ThP r~n tht" hce rce~cl nff hv <nh<ti-
-~U~H U~H+
(1 A
,.~.~,,
£:\ 'T't.
+HV
"'-l.l
HV+~
+1..
w
~
rl - ( y - r)"V/3- { y - r)i3V"
F"P " (14.11)
V · ~X T) aT V · ~X T)
a ...
uv~'-' ' " auu y <H'-' v~ •· .n.H'-'~ <HVH W'-'
"
w oe evamareu ar me remraea proper nme T0 .
The field-stren2'th tensor F"/3 (14.11) is manifestlv covariant but not overlv
,1· ;+ Tt ' ,,.- ,r,. 1 h th h ,],-1, .... .-Ill
.. .-1 ,]' ;+
_, ' . •'·
£. ' £
vc '""'
d. -1.
"""'IS" ~ V'
auu .
<UUVH,
~
vc '""' "'IS~' "'

needed to carry out tne dnterenttatton m (14.11) are
~X T) ~1\, 1\ll), v ~ yc, ycpJ
riV" . . .
- - ln'B ·B. cyLB + cy"B(B · Bll flA l"J\
aT \' -;
rl riV"
- IV· (x - rll - -c" + (x - r), --
aT aT
664 Chapter 14 Rad•anon by Movmg Charges G
where~ = dl3/dt is the ordinary acceleration, divided by c. When these and (14.7)
are employed the fields (14.11) can be written in the inelegant, but perhaps mor@
B = [n X E],e, (14.13)
n R p n "-yr;;- ~!11
J!;(x, t) e + (14.14)
L ·/(1 - 13. n)JR2 tel c (1 - 13 • n) 3 R ret
Fields (14.13) and (14. 14) divide themselves naturally into "velocity fields," which
are ~ndependent of accelerattOn, and "acceleratiOn fields," whrch depend linearly
on 13. The velocity fields are essentially static fields falling off as R- 2 , whereas
the acceleration fields are typical radiation fields, both E and B being transverse
to the radius vector and varying as R- 1.
For the special circumstance of a particle in uniform motion the second term
in (14.14) is absent. The first term, the velocity field, must be the same as that
obtained in Section 11.10 by means of a Lorentz transformation of the static
Coulomb field. One way to establish this is to note from (14.11) for F" 13 that if
V" is constant, the field is
2
F" 13 = ec 3
· [(x - r)"V13 - (x - r) 13 V"] (14.15)
[V · (x- r
in agreement with the third covariant form in Problem 11.17. It may be worth-
while, nevertheless, to make a transformation of the charge's coordinates from
its present position (used in Section 11.10) to the retarded position used here in
order to demonstrate explicitly how the different appearing expressions, ( 11.152)
I hen from the tnangle OMP' we haveR sine- b, so that
[(1 ~ n)Rf r (14.16)
~
/ \
/
/~
// /
/7
/
0
/~/
b
'. r \ /
\../"\~ v \/ Fl
' ' lp 1M
{JR:--..;...---vt----+<
Figure 14.2 Present and retarded pos1!Ions of a charge m umtorm mot10n.
Sect. 14.2 Larmor's Formula and Its Relativistic Generalization 665
The transverse component E 2 from (11.152),
eyb
E2 = (b2 + lv2t2)312 (14.17a)
can thus be written in terms of the retarded position as
D
E2 = e (l4.17b)
"12(1 - 13 • n)3 R3 ret
!hiS IS JUSt tne transverse component of the veloc1ty field in (14.14). The other
components of E and B come out similarly.
14.2 Total Power Radiated by an Accelerated Charge:

Larmor's Formula and Its Relativistic Generalization
Tf :1 ~hllnne ;, "' liter! hnt ;o · '· ~ ;n ~ .c.
•• mhPrP ;, ,, ;,
"J
is small compared to that of light, then in that coordinate frame the acceleration
field in (14.14) reduces to
e nx (n x ~)
Ea =;:: R
(14.18)
re<
The instantaneous energy flux is given by the Poynting vector,

c c
S = - E x B = - IE 12
n (14.19)
47T 47T a
This means that the power radiated per unit solid angle is*
AP 2
I RKI 2 - In X (n X rnJ2 ,.
(1,1")(1\
au 47T 47TC
If 0 is the amde " thf'. :<rrPlf'rMinn v :mrl n "' •hnwn in Fi<> 1L17. thPn ~
the power radiated can be written

dP 2
e 1 ·.12 2/C\ ~n
..
47TC3 I I o \.l" ·..f.>.L)
~
d[!
'J'h;c PV~ ;J. ;, tl, c~ 2/C\ .1 ,
" -,_uv
' "
vu " '
,.]."
HH~U ~0 H
resun. vve nme rrom ~ 14.1~) that the rad1at1on Is polanzed m the plane contammg
vand n. The total instantaneous power radiated is obtained by integrating (14.21)
." '. _,
V n.-.L UH OVHU • .LHUO
2 e2
P = ::;-
-'
31vl 2
~
(14.22)
This is the familiar Larmor result for a nonrelativistic, accelerated charge.
*As noted in Chapter 9, in writing angular distributions of radiation we always exhibit the polarization
explicitly by writing the absolute square of a vector that is proportional to the electric field. If the
angular C!JStnbutwn tor some parlicular polarization is desired, it can be obtained by taking the scalar
product of the vector with the appropriate polarization vector before squaring.
/v
I
v
/
/
/
/ Figure 14.3
Larmor's formula (14.22) can be generalized by arguments about covariance

under Lorentz transformations to vield a result that is valid for arbttrarv velociti'"
of the charge. Radiated electromagnetic energy behaves under Lorentz trans for-
mation like the zeroth component of a 4-vector (see Problem 12.18). This can be
used (see Rohrlich, p. 109ff.) to show that the power Pis a Lorentz invariant. If
we can unu a Lorentz mvanan, tuat reuuces tO tue LulluOt tvtuwtu \l'+.LLJ 10r
f3 << 1, then we have the desired generalization. There are, of course, many
Lorentz invariants that reduce to the desired form when f3 ---7 0. But from (14.14)
it is evident that the general result must involve only J3 and J3. With this restriction
<L ~- .c ~ <L ol.. • 1- • cr. ..: ~ <>.
UH U>~ V>~~> V> ~~u.uu•v~ •uu• ~UH uppvu> 0 oH~ >V~UH •~ U '>'1UV• V HH~ OHV
appropnate generanzatwn we wnte Larmor s rormma m me suggestive rorm:
2 -e -
p _ -
- 3 m 2 c3
2
( dpdt dp)
--.- dt
(14.23)
where m IS the mass ot the charged particle, and p Its momentum. 1 ne Lorentz
invariant generalization is
P-
2 ez I dpl-' dp" \ (1J. 7.1\
'I mLr' dT dT '
where dT = dtly is the proper time element, and p" is the charged particle's
momentum-energy 4-vector.* To check that (14.24) reduces properly to (14.23)
~s R ---7 0 we ev~luMe the 4-vectnr sc~l~r ~-
_ dp" dp" = (
dT dT dT
dp)" __!_ (dE)"
dT
= ( dp)" _
C
2
dT
{3 2 ( dp)
dT
(14.25)
Tf (1.1 7.1\ ;0 ~ ;n r1f thP

_, ~ncl
_, ;nn hv mP~ns nf
Lo yu<L
• ?.
auu
~ '
I' Y'" '
- ,_
•v uuw.u wv
d. T •'
'~
~- -'• ,_ '"''"
\iU/Uj•
-
2 e2 • •
p = :0- 1'6[(J3)2 - (J3 X J3?J (14.26)
..
Vllto dltod Vl djJ} Vl lllto ltoldl!Vl,lll,; vA~ LVL I jJVW\:01 "
that of charged-particle accelerators. Radiation losses are sometimes the limiting

factor in the maximum practical energy attainable. For a given applied force (i.e.,
" "" . '- . ,_

.. ·c-_ •'-· ~·-· ·'b "I·
·~'- '- • 1 -~ '-
" • '"'I"•• "J "J
products of 4-vectors or tensors of higher rank. The available 4-vectors are pv and dpvfdT. Only form
(14.24) reduces to the Larmor formula for f3 ---> 0. Contraction of higher rank tensors such as
vv(dv"ldT) can be shown to vanish or to give results proportional to (14.24) or m'-
Sect.14.2 Larmor's Formula and Its Relativistic Generalization 667
a given rate of change of momentum), the radiated power (14.24) depends in-
versely on the square of the mass of the particle involved. Consequently these
radiative effects are largest for electrons.
In a linear accelerator the motion is one-dimensiOnal. From ( 14.2)) 1t IS ev-
ident that in that case the radiated power is
2 ez dn
1' 2 3
~l'I.LI)
3m c dt
_, ._,
cr~
>U~>U<V~
.c .c
. 'I' ·~ -~
~ ~
,f tho
•
per unit distance. Consequently
2
P-
2 ez IdE (14 7RI
3 m--rc-., dx
showing that for linear motion the power radiated depends only on the external
t-hat determine the rate of change of narticle energy with distance not on
the actual energy or momentum of the particle. The ratio of power radiated to
power supplied by the external sources is
n , 2 ' Jr ,-, I 21. 2\ ATi'
\'
2 3
-7
2
(14.29)
(dE/dt) 3 m c v dx 3 mc dx
wnere me last rorm no10s ror relativistic particles ~p -7 .t). cqua,ion ~ t'+.L" J snows
that the radiation loss in an electron linear accelerator will be unimportant unless
the gain in energy is of the order of mc 2 = 0.511 MeV in a distance of e 2/mc 2 =
-n "·' • r~ ·~14•'-" .1cr. · .1 ·
"-·0"- A J.\J \.olll, Ul U1. lllv U1Uv1. U1. "- J.V IV1.\.o " ' " ' 'JY_•~m ~ '~'6J 6'
less than 50 Me V/m. Radiation losses are completely negligible in linear accel-
erators, whether for electrons or heavier particles.
' -ho~~orleoo' " ;~~;r~nlor
_, : JikP thP -1-
or betatron. In such machines the momentum p changes rapidly in direction as

the particle rotates, but the change in energy per revolution is small. This means
that
dp 1 dE
A. = ywlpl »-;::A~ (14.30)
Then the radiated power (14.24) can be written approximately

2 ez 2 2
e c
r ~ .l 't.CH)
3 mzc3 'Y w IPI' 3 p2 p '}'
where we have used w = ( c{31 p ), p being the orbit radius. This result was first
.'
0 uy ru m H>""'· .tue r LYe- .0 TC:JSS" )J\.o I IUVll 1.>
;:;p - 27rp P - 471' ez [-!3,,4

(141?)
cB 3 0
where 1/p is actually 1!27rtimes the path integral around the ring of [1/p(s)f For
hi<Th-enen>v electrons (R- 1) this has the numerical value
BE(MeV) = 8.85 X
w-2 ~vev )J (14.33)
p(meters)
Tn thP fird - 1 GeV ;:;p 1 p 0.1
""n"L n
1 ke V per revolution. This was less than, but not negligible compared to, the
energy gain of a few kilovolts per turn. At higher energies the limitation on
available radiofrequency power to overcome the radiation loss becomes a dom-
inant consideration. In the 10 GeV Cornell electron synchrotron, for example,
the orbit radius is p 100 meters, the maximum magnetic field is 3.3 kG, and
the rf voltage per turn IS 10.5 MV at 10 GeV. Accordmg to (14.33) the loss per
turn is 8.85 MeV. These same general considerations apply to electron-positron
storage rings, where rf power must be supplied just to maintain the beams at a
constant energy as they circulate. At the LEP ring in Geneva, Switzerland, for
beams at 60 Ge v the loss per turn is about 300 Mev per electron.
The power radiated in circular electron accelerators can be expressed nu-
merically as
P (watts) - 106 8E (MeV) J (amp) (14.34)
where J is the circulating beam current. This equation is valid if the emission of
radiation from the different electrons in the circulating beam is incoherent. In
the largest electron storage rings the radiated power amo11nts to tens of watts
per microampere of beam. While this power dissipation is a waste to high-energy
physicists, the radiation has unique properties that make it a valuable research
tool. These properties are discussed in Section 14.6, and in greater detail for
dedicated "hght sources" m SectiOn 14.7.
14.3 Angular Distribution of Radiation Emitted

hy an Accelerated Charge
For an accelerated charge in nonrelativistic motion the angular distribution shows
a simple sm 2 ® behavior, as given by (14.21), where® is measured relative to the
direction of acceleration. For relativistic motion the acceleration fields depend
on the velocity as well as the acceleration. Consequently the angular distribution
is more complicated. From (14.14) the radial component of Poynting's vector can
~ {_.!_ ~]
2
[S. n] = n x [(n - 13) x }
(14.35)
;et 47TC R2 (1 - 13 . n)3 cet
It is evident that there are two types of relativistic effect present Que is the effect
of the specific spatial relationship between 13 and ~' which will determine the
detailed angular distribution. The other is a general, relativistic effect arising from
the transformation from the rest frame of the particle to the observer's frame
and manifesting itself by the presence of the factors (1 - 13 · n) m the denomi-
nator of (14.35). For ultrarelativistic particles the latter effect dominates the
whole angular distribution.
In (11.35) S n is the energy per unit area per unit time detected at
an observation pomt at time t of radiation emitted by the charge at time t'
t - R(t')/c. If we wanted to calculate the energy radiated during a finite period
of acceleration, say from t' = T1 to t' = T 2 , we would write
(, r 2 1 [R(T2),c] j t' r, dt
(14.36)
<>no. , ..,_, Ul
..
• " • I« -\ I J J ' d.
lllU> WC >CC llldl lllC _ dlJU rur IT)' 'o \~ · UJ \~"u' J• cu~ 1-'V '~'
radiated per unit area in terms of the charge's own time. We therefore define
th" nnwer radiated ner unit solid an<'le to be
dP(t') R 2 (S • n) dt (14.37)
If we imagine the charge to be accelerated only for a short time during which ~
and ~ are essentially constant in direction and magnitude, and we observe the
radiation far enough away from the charge that nand R change negligibly during
• • ,_ . • ,,_ .1. J' ''- .
UlC ldllUll llHCl Vdl, lllCll ~ 1 '+ ..~ •v cu~ v
of the energy radiated. With (14.35) for the Poynting vector, the angular distri-
hlltinn i'
dP(t') e2 In X ((n - ~) X ~w (14.38)

dfl 41TC \T ll •lf)'
aile!. If e is the angle of obser~ation measured from the common direction ~f ~

;mrl R thPn (14 'R) reduces to
dP(t')
(14.39)
au '11TC \_T ~UJ
For f3 << l, this is the Larmor result (14.21). But as f3--;. 1, the angular distri-
bution is tipped forward more and more and increases in magnitude, as indicated
schematically in Fig. 14.4. The angle Omax for which the intensity is a maximum
-
1 1- cn-2 1\
1 {1 A Af'l\
7f '/ \ , I
3{3' -
2y
;" tho l;m;,;~n .,nlHn f~· Q ~ 1 Tn thi, '~mp limit thP nP~k-
intensity is proportional to y 8 • Even for f3 = 0.5, corresponding to electrons of

~SO ke V kinetic energy, Omax = 38.2°. For relativistic particles, Omax is very small,
being of the order of the ratio of the rest energy of the particle to its total energy.
Thus the angular distribution is confined to a very narrow cone in the direction
01
.
. ror SUCII •0 .
. .
lllcarrgumrrn:srrruuvu \ l."T ...n J ~au vv vvucu... u
. /• . .
approximately
dP(t') s ezvz: 8 \ yey (14.41)
,-,. = - - 3 - ')' 11 2,-,2\5
UH \ I I
Bmax Figure 14.4 Radiation pattern for

' "' '" U>
mol! on. Tne two patterns are not to
scale, the relativistic one (appropriate
--
--- J--
for "' 21 havinf! been reduced bv a
factor ~ 102 for the same acceleration.
dP
- -
d{l
/i\ /: '\
~
I
/ I
I
I ~ J
I
I
I
I
--r"' -1.0 l !\ ): ~
I I
I: : !I
0 1.0
-y8~
~igure L'l.::l -o uisrriomion or rauiarion wr relativistic panic1e.
The natural angular unit is evidently y- 1. The angular distribution is shown in

~· ' ~ '•' ' ' · •' · ~• _' n 1
> 1!;• l ' ·,J Wll11 -o '-U lU 111<:0~<:0 U1111~. H1'- )-''-0"- V'-'-U!O 01 yv z, 011U lJIC
half-oower ooints at "~() = 0.23 and "~() = 0.91. The root mean sauare anr>:le of
emission of radiation in the relativistic limit is
( 2)1/2 - ~ - me
(} - - D (14.42)
r
This is typical of the relativistic radiation patterns, regardless of the vectorial
relation between 13 and 13. The total power radiated can be obtained by inte-
. ..
graung ~ t'+.Y1) uve1 au -c . 1 nus
2 e2
f'(t') - 3 c 3 vy (14.43)
in agreement with (14.26) and (14.27) .

.• ,' r ' '. ., _,.
"
V1 Oll!;Ula1 '
a '-1la1!;<:0 111U1~1lllJUl1V11 V1 1
£
•
£
to lll01 1V1
instantaneously circular motion with its acceleration B oeroendicular to its ve-

locity 13. We choose a coordinate system such that instantaneously 13 is in the z
direction and ~ is in the x direction. With the customary polar angles e, q, defining
the direction of observation, as shown in Fig. 14.6, the general formula (14.38)
reuuces LU
dP(t') ez lvl 2 sin2 () cns 2 m
1 (14.44)
dD 41TC
3
(1 - f3 COS 1:1) 3 lCl - f3 cos ef
We note that, although the detailed angular distribution is different from the
', ' , 'I ' • • -I , , , -I. • £ . ~
'-0''-• 111'- ~o111<:o '-1101 0'-1<:011~11'- 1 '-'la 11 v " " ' - ' r , 01 1V1 VOIU
rz
r•"
' "
\M I
/ I
-~ "', I y
fJ;t' -;;: I
'
/ '.J Figure 14.6
Sect. 14.4 Radiation Emitted by a Charge in Arbitrary, Extremely Relativistic Motion 671
!'lncrlP< i< Tn thP rPl'•· · •+ir limit (" '>'> 1) thP ononbr rli<trihntirm ron
b e ~wntten
. ' approximate
. ly
dP(t') 2 e2
=- 3 'Y
6
/1
lvl 2 1 -
4·l82 cos 2 </>
(14.45)
"("\ 2n2\3 /1 2 n2\2
~·· " ~ ,~
f " I ,~
f " !
~L
~WC 'VVJ ~ -~ ~'-' ':'"e'~ W W "'". "J-'J-'' VAUU<lUVll" t;lVCOll UJ ~.1 ....."-j,
just as for one-dimensional motion. The total power radiated can be found by
/1 A A A\ .11 .1 +. (1 A ..;t:\.
T/~"~U ~< U~U< ,~~.~~,.
\ ~
? • I?
P(t') = :: ~ I• I 'Y4 (14.46)
3 c3
It is instructive to compare the power radiated for acceleration parallel to
tne velOcity ~l4.4j)Or ~l4.LI) w1tn tne power ramatea ror acceleratiOn perpen-
dicular to the velocity (14.46) for the same magnitude of applied force. For cir-
Clilar motion, we magnimue 01 we raw 01 cnange 01 momentum~ wnicn is equa1
to the "1 ' .] for<ee) ;, wnv n mentlv (14 4n) ""n he written
2 \ 2
P. (t') - ~ _e_ .,zl dp I (14 47)
3 m"c dt
When this is compared to the corresponding result (14.27) for rectilinear motion,
we find that for a given magnitude of applied force the radiation emitted with a
transverse acceleration is a factor of y" larger than with a parallel acceleration.
ld d U. ,I.' rtinn R ... ittPn h .. n rh. ~, 0

in A rhitrnr" ~,,
R: "'" n ·'·rtivistic Motion

For a charged particle undergoing arbitrarv. extremelv relativistic motion the
radiation emitted at any instant can be thought of as a coherent superposition of
comnouuons commg trom me componems 01 acce1erauon paraue1 w ana per-
pendicular to the velocity. But we have just seen that for comparable parallel
..
LHC 11 Ulll LHC pa! dlH:l -r 1' uc:gnglt11e
~~·~ !-'"~' 1I -v2 )
~
tred to that from the ne1 · ular ~nt l nse~
quently we may neglect the parallel component of acceleration and approximate

the radiation intensity by that from the perpendicular component alone. In other
words, the radiation emitted by a charged particle in arbitrary, extreme relativ-
1suc monon IS approx1mareq me same as mar emnrea oy a parncw movmg
instantaneously along the arc of a circular path whose radius of curvature p is
glVcou uy
u c
p=-. =-:- (14.48)
uj_ uj_
where u is the perpendicular component of acceleration. The form of the angular

j_
distribution of radiation is (14.44) or (14.45). It corresponds to a narrow cone or

searchlight beam of radiation directed along the instantaneous velocity vector of
tne charge.
~. ,]. . +]. f, ;,· .-1. +I- •• +I-
radiation to a narrow pencil par.allcl to the velocity has important consequences.

The radiation will be visible only when the particle's velocity is directed toward
thP ' Fnr ~ nortirlP in · · th~ ~1- ·mill ..1~•~~• ~ .1.
L .£ c ~ ~-
t-' ·~
~·
.L r
v v H V< V< ·~<J OUVH UW~ UU<UUV '
~Ul d u1 >U'-'11 uursts
If
.
the oart!cle
~<L .C <L L
IS m oenod1c motiOn). as sketched in Fi". 14.7. Since the
,£ <L -~ .£
f
1 <L
' "'~ l-'·~ V
.,
U~L
• uwy
' VL
'. ~
.. au~;
or tne oraer or
d=£
'Y
corresponding to a time,
M=_p_
yu
while illuminating the observer. To make the argument conceptually simple,
neglect the curvature of the path during this time and suppose that a sharp rec-
tangular pulse of radiation is emitted. In the time lit the front edge of the
pulse travels a distance,
n
1J-cl1t--
'Y
t ---l ~L
~t--Lo-
'x -A"
/
I
I
I
I I
/
/
A•
"
./
'
~ -~ "
P(t)
t T o--
_Lo
c
_, f<;-- T =bc
t__.._
'6'
.H~
. ..
,.
<.oumo a opu m 1,
._,
!''"'"'"
. ~·
mar an ooserver ar me pomt A aetects as snort oursts ot radiatiOn ot tlme duratiOn
"""·. '" "-7'u .,

T- Lie. occurrin!! at re!!ular intervals T - L ~lc. The n~·n-;;th ;, aivPn hv !14 4CJ\
.1. 1 D, _,. -, ' .. .. - -~ _,.
. ·~ ·:
parnctes, seeK. r. lSien, Am. J. rnys. 'IU, 'IO (lY/L).
Sect. 14.5 Distribution in Frequency and Angle of Energy: Basic Res nits 673
Since the particle is moving in the same direction with speed u and moves a
distance d in the time tlt, the rear edge of the pulse will be only a distance
T
T p p
L u u .L ~.L'-t.'t:1)
(3 '}' 2y3
behind the front edge as the pulse moves off. The pulse length is thus L in space,
~r T 1~ ;n t;~~ Hr~..,.; ~AnArol omH~Antc oh~nt thA "P~nr;Ar '~ ;t;~n r ~ · ~
·o
."' . ~r
.
-wave-rrams-TTT ·r rurrrurrrre- ~"r Wlll l-Ul!Ialll -yy IDle
frequency components up to a critical frequency,
\
c c 1"3 f1A <:rn
'· \ • -I
T n
For circular motion c/p is the angular frequency of rotation w 0 and even for
arbitrary motion it plays the role of a fundamental frequency. Equation (14.50)
eh~me thot o rPlot;,net;~ nort;P1~ p~;te o ' ' ~f frAm · nn t~ 3
r -. •-r
.' T
nmes me 11 !La! ,. y. tn a LVV JVte v syncnrotron, 'Ymax 'IVV, wnue

1r~ ~
w 0 ~ 3 X 10 s- . The frequency spectrum of emitted radiation extends up to

8 1
~2 X 10 16 s- 1 , or down to a wavelength of 1000 A, even though the fundamental

frenuencv is in the 100 MHz ran<Te. For the 10 GeV machine at Cornell 'V ~
,~-1()4 I "1"'1.. '.

" "U '1 ~ 1()6 oL
c
'"1A' 1Cl19 1
, -r
to 16 keY x-rays. In Section 14.6 we discuss in detail the angular distribution of
the different frequency components, as well as the total energy radiated as a
function of frequency. In Section 14.7 we show how to modify the spectrum with
~· rlPv;PP<
,... _,_. ,_ .
14.5 Distribution in Frequencv and AnJ![e of EnerJ!v
_r rn
--uy
~
""
. ,... r
~· Ulc:;U '-'UUI l')"'~•
r1<.-t..-c:;lc:;l LIU~lt..- ~UX~UH~
~ ~
The qualitative arguments of Section 14.4 show that for relativistic motion the
radiated energy is spread over a wide range of frequencies. The range of the
fr~~H~n~u ~k 0--AO"-~·~ ,fv.
"'
0 HoOO AO •
"JT "J ·r "r "b' ·

The argument can be made precise and quantitative by the use of Parseval's
theorem of Fourier analysis.
The general form of the power radiated per unit solid angle is
df'J!) = IA(t)ll (14.51)

dO
wnere
""
A(t) = 4c7T) [RE]cct (14.52)
E bemg the electnc held 1 14.14). ln 114.)1) the mstantaneous oower ts exoressed
. ., - '
-·
, ., "-" .. {
'J
._, • . C' 1A'>\ ,·
. J ,,
to consider a frequency spectrum in terms of the observer's frequencies. For
definiteness we think of the acceleration occurring for some finite interval of time,
-ora:r-reasr -urrTIJrTl ~mnes, so tnat me LOW! cHc1~;y
radiated is finite. Furth,rmnrP th" nl nnint is f~r Pnnno-h

away tram the charge that the spatial regwn spanned by the charge whlle accel-
' ' -'- ~ om~ll o~1;r1_~ ,J,_ ,1, .t +1- -1- • •
l' .
The total energy radiated per unit solid angle is the time integral of (14.51 ):
dW r~
dfl - J_x IAUJI at (14.53)
This can be expressed alternatively as an integral over a frequency spectrum by

use of Fourier transforms. We introduce the Fourier transform A( w) of A(t),
A(w) = - 1 f~ A(t)eiwt dt
(14.54)
v27T -w
anu ILS u• ''"I>'",
A(t) = ~
1 1- A(w)e-iwt dw (14.55)
v L7r J X
Then (14S-I) can be wntten

dW
-
1 {x dt
{x dr,
{x dr.i A*(,..'\. A( ..\oi(w'-•o)t (111 <;?;\
d!! "l7rJ-x J-x J-x ' ~J
Interchanging the orders of time and frequency integration, we see that the time
integral is just a Fourier representation of the delta function o( w' - w). Conse-
quently the energy radiated per unit solid angle becomes
dW
dfl
r,
J -oo
A(._,,,2 "·
' '
{1A C:"J\
\ I
The equality of (14.57) and (14.53), with suitable mathematical restrictions on

we wucLiuu .a~' J, is a specia1 case 01 rarseva1 s tneorem. lt IS customary to m
OnlV OVef • • - C. • 'inrp__the_ oion ~f fhP frPniiPnru h~o n~
physical meaning. Then the relation,
dD Jo dw dD aw ~14.:11'5)
defines a quantity that is the energy radiated per unit solid angle per unit fre-
quency interval:
(14.59)
(14.60)
1nis resun r - in a quamitative way me oenavior OIT!fe power radiated as a
By using (14.14) for the electri~ field of an acceler~~~d charge we can obtain
a general expression for the energy radiated per unit solid angle per unit fre-
• oi ,1 • ,f ' ,J •'- • -• .C •'- "_T "'·
Sect. 14.5 Distribution in Frequency and Angle of Energy: Basic Results 675
must calculate the Fourier transform (14.54) of A(t) given by (14.52). Using
gration from t to t', thereby obtaining the result:
A(w) =
Since the
R(t') = x - n · r(t') (14.63)

where x is the distance from an origin 0 to the observation point P, and r(t') is
the position of the particle relative to 0, as shown in Fig. 14.8. Then, apart from
an
diated per unit solid angle per unit frequency interval (14.60) is accordingly
(14.65)
For a specified motion r(t) is known, f3(t) and Jl(t) can be computed, and the
can be evaluated as a function of w and the direction of n. If accelerated
motion of more than one arge sum
one for each charge, must replace the single amplitude in (14.65) (see Problems
Even though (14.65) has the virtue of explicitly showing the time interval of
. .
zero, a simpler expression for some purposes can be obtained by an integration
by parts in (14.64). It is easy to demonstrate that the integrand in (14.64), ex-
IS a
0/0 Chapter 14 Kadiation by Moving Charges-G
Then an integration by parts leads to the intensity distribution:

2
d2J ezwz
-- r~ n X (n X J3)eiw(t-n·•(t)lcJ dt
..1. ..10 ::;---2 (14.671
u1e 1eauer may rignuy asK wnerner (14.0~) IS correct mall circumstances as it
stands. Suppose that the acceleration is different from zero only for T 1 < t < T2 .
HIL il ' " " ' . ' '
" "y '"~" '" 1 auuu 111 \ l'"t.U 1 J uVer au L!flle. 1 ne preCISe answer IS tnat
(14.67) can be shown~ by adding and subtracting the integrals over the times when
the velocity is constant, to follow from (14.65) provided ambiguities at t = =too
are resolved bv insertin~r a conver f"rtnr "_,.,, in the intes:rr:mcl and__taki on
the limit E --7 0 after evaluating the integral. In processes like beta decay, where
the classical description involves the almost instantaneous haltin~r or settinQ in
motion. of charges,.
extra care must be taken to specify each particle's velocity as
a puysicauy sensiOJe runcnon or nme.
We remind the reader that in (14.67) and (14.65) the polarization of the
· · ;, -r · uy llle uirec1ion or me vecror mtegrar m eacn. lhe
in · _af r~cli~tirm of H "ert:o~in fixpAi_nill · · r~n hP "hbinPcl~l ·
the scalar product of the appropriate unit polarization vector with th~ vecto~
inte~rral before forminQ the ahcnl •te souare
For a number of charges e1 in accelerated motion the integrand in (14.67)
involves the replacement
N
eJ3e i(w/c)n·r(t) ~ 2:, e. p·e i(wic)n·r/t)

(14.68)
1 J
j~l
In the limit of a continuous distribution of charge in motion the sum over j

au ;,uegrat over Lne currem uensJLy J(X, t):
1 (
e~e ;,, '" ,.• ,,, --7 c J dox J(x, t)e 1\wtcJn·x (14.69)
Then the intensity distribution becomes

d 21
dw d[! -
w2
417 2 r 3
I I dt 3
d x n x[n X J(x, t)]eiw[t-(n·x)lc]
2
(14.70)
,]+ th L .1.. .I C.. ct. p

'UVH< cu~ "'
~" ~~• oviUllUU Ul LillO
v
wave;
equation for the vector potential.
ILl;;
•
p, __£
··- .-1' D ,J," .... . JL ~ .1 • • .
rh;: ·o
-~-~ -r
• -'
-·· .....
;.,.. T,
~J ~
._,
..,,, fl'
Ho:;u u,r u ~._.,.UHO'•.>HL
•••
.J l l l .....
. . . v ..
Tn Sedion_14 4 UTP thP r~cli<t±inn

O~UT th~t .;t+n.-1 ~ nl ol . . .
J J J
particle subject to arbitrary accelerations is equivalent to that emitted bv a par-

ticle movinQ. '"nf'm"h' M rnnot"nt An ~n ·rr ·r i~tP ~;r. ,], .~-._
The radiation is beamed in a narrow cone.in the direction of the velocity_ v~ctor
and IS seen by the observer as a short pulse of radiation as the searchlight beam
• o • L • •
o ·~~["" 111'-' UU''-'' ValiUU
~~•voo
Sect. 14.6 Frequency Spectrum Emitted in Instantaneously Circular Motion 677
To find the distribution of energy in frequency and angle it is necessary to

calculate the integral in (14.67). Because the duration of the pulse is very short,
it is necessary to know the velocity J3 and position r(t) over only a small arc of
the trajectory whose tangent points in the general direction of the observation
pomt. Ftgure 14.9 shows an appropriate coordinate system. The segment of tra-
jectory lies in the x-y plane with instantaneous radius of curvature p. Since an
integral will be taken over the path, the unit vector n can be chosen without loss
of generality to lie in the x-z plane, making an angle {)(the latitude) with the x
axis. Only for very small {)will there be appreciable radiation intensity. The origin
of time is chosen so that at t = 0 the particle is at the origin of coordinates.
The vector part of the integrand in (14.67) can be written
n x (n x 13) - 1{ € 1
sin(;) + e~ cos(;) sine] (14 71)
where e = e 2 is a unit vector in the y direction, corresponding to polarization in

11
the plane of the orbit; e.L = n x e 2 is the orthogonal polarization vector corre-
sponding approximately to polarization perpendicular to the orbit plane (for {)
small). The argument of the exponential is
(14.72)
Smce we are concerned wtth small angles IJ and comparattvely short Urnes around
t = 0, we can expand both trigonometric functions in (14.72) to obtain
(14.73)
where {3 has been put equal to unity wherever possible. Using the time estimate
plcy for t and the estimate (A 2 ) 112 (14 42) for A, it is easy to see that neglected
terms in (14.73) are of the order of -y- 2 times those kept.
With the same type of approximations in (14.71) as led to (14.73), the radi-
ated energy distribution (14.67) can be written
2
(14.74)
'!.
1'\ E
i\r-;
1\ ~ /
I 8 y
I
~lf/-
/
~ '
v
T/ l.:I.Q
where the amplitudes are*

1 c2t3
A,(,,,) C roo f Pl<n· i W .. -1- fj2 It -1- ,/t
" pJ-oo 2 '11" 3oL
I w 1 CL(.
11 d. .,
'7<:.\
.ri.l~W)
Loo exp l 2 -r tl" It + at
fJ
2 ")I I 3p2
fJ
2
A change of variable to x --
rc0C 2
+
02 and introduction of the
parameter g,
I 312
~
wp .L 2\
~
":t 2 u I ~ l'l./0)
\
allows us to transform the integrals in A 1,(w) and A.l(w) into the form:
A 11 (w) = £.C ( ")I\ + 02) r-oo

x exp[i~g(x + ~x 3 )] dx
(14.77)
I \ 112 roo
A ( "'' c
p 111
1
\ "1/L
-1- 112
J 00
. ~~[;~ C{'
"L~ \"
..L L~3\ l A~
/J
ThP. intPor,l< in (14 77) "rP idPn · "< A irv; ·o

le Ar ~~t~rn~fiu~lu ~e ~.Arl
J "J
ified Bessel functions:
r sin[~g(x
X + ~x 3 )] dx = ~ K213W
r0
cos[~g(x + ~x 3 )] dx = ,"?i'
1 Kl!3W
(14.78)
l'rm<PnllPnthr thP PnProv r"di"tf'd nPr 11nit J', "~"~" . ::J t"n~r •mit ,A]irl nn~l~
"J ·o·
" '
"
d 2I
dw df! =
e2
37r2C (r(
wp
-;:
1
"Y2 + 02)
The first term in the square bracket corresponds to radiation polarized in the
TK~!3(g) + (1/-yz)
02
+ 02 Kfl3w J (14.79)
plane of the orbit, and the second to radiation polarized perpendicular to that
_,
r
We now proceed to examme th1s somewhat como lex result. First we inteQrate
'11 j', . rl h rl +< •< rl" _; _,
r
A
'1." '5Y ·o ro
dl
df! =
d 1
o dw df! dw
2
=
7 e 1
2
16 -;; (1/l + 02 ) 512

r1 + 75 (1/l)0 + 02J 2
(14.80)
*The fact that the limits of integration in (14.75) are t ~ :too may seem to contradict the approxi-
mot;Anc morln ;nITA. 'f, /lA 77\ IJA 7>\ n "nt. •L ' . . •L _;_ _,
r
-~
, •• •• J}
'v'Y .' ~· auu "TIJ:<O "b" •; zero wr ume>
JIIUCJJ lllan uJose necessary ro mamtam tile vanm y ol (14./J). lienee tne upper and lower
lilmts on the mtegrals can be taken as infinite without error. OnlY for freauencies of the order of
...
w lc/n) ·- w does the annroximation fail. But we h <PPn ;n 1A J. thot fm •'-
particles essentially all the frequency spectrum is at much higher frequencies.

"""'• '""·u
This shows the characteristic behavior seen in Section 14.3. Equation (14.80) can
be obtained directly, of course, by integrating a slight generalization of the
circlll>~r-motion nower formula {14.44) over all times. As in !14.79). the first term
in (14.80) corresponds to polarization parallel to the orbital plane, and the second
to perpendicular polarization. Integrating over all angles, we find that seven times
as much energy is radiated with parallel polarization as with perpendicular po-
bri7~tinn Th;_,~ r~ili~tinn frnm ~ ~Pb . . . ]]v mnvin~> char~>e is verv stron~>lv
'l' 'J•r r
The properties of the modified Bessel functions summarized in (3.103) and
(3.104) show that the intensity of radiation is negligible for g >> 1. From (14.76)
we see that this will occur at large angles; the greater the frequency, the smaller
d. '' _, ~' L -' ~'~'-'-"'- 'HL -'·' "-' .-1:' 'f'J.,' ,1, +J.,
"'~ ~uu-.,m _0 _ uc,yuuu "''-'' '-' "" U'-' 0 .0 ·~
Lne rauimion is 1arge1y conuneu LO me p1ane comaining tHe • , ,, uy

(14.80), being more confined the higher the frequency relative to c/p. If w gets
too large, however, we see that g will be large at all angles. Then there will be
negligible total energy emitted at that frequency. The critical frequency we be-
yond which there IS negligible radiatiOn at any angle can oe ueuneu oy ~ l!L
for () = 0. Then we find*
(14.81)
'L pj 'L me" j p
This critical frequency is seen to agree with our qualitative estimate (14.50) of
Section 14.4. If the motion of the charge is truly circular, then c/p is the funda-
mental frequency of rotation, w 0 • Then we can define a critical harmonic fre-
quencv w, ~ new 0 , with harmomc number,
(14.82)
Since the radiation is oredominantlv in the orbital plane tor 'Y >> 1, It IS
instructive to evaluate the angular distribution (14.79) at () = 0. For frequencies
well below the critical frequency (w << we), we find
e2 f(~) 2 3 \113; Wp \2/3
For the opposite limit of w >> w, the result is
e o c
These limiting forms show that the spectrum at () = 0 increases with frequency
roughly as w 213 well below the critical frequency, reaches a maximum in the neigh-
,_ nf bO "nil thPn ilrnn' f'Ynnnpnfi,Jhr tn 7P.Tn ~hmrP that
0
nCV.
~·~ " • _, £'. "£ ·~ .-1 1- .-1
*Our present definition of w,. differs from earlier editions. The present one, defined originally by
Schwinger (1949), is in general use.
{;;jill}
~,
·1d. ~
. ' h" . "" f:}. G
the angle fJ, at which {(fJe) - {(0) + 1. In the low-frequency range (w << we),
{(0) is very small, so that {( fJe) = 1. This gives
•'lr \ 113 1 \ 113
1 2w
fJ, ~1'1.1'1))
wp ')'
w}
We note that the low-frequency components are emitted at much wider angles
than the average, (0 2 ) 112 - y- 1 • In the high-frequency limit (w >we), {(0) is large
compared to unity. Then the intensity falls off in angle approximately as
d2J d21 3w-y 2 FP12w (1 A Qt:\
~ \ • J J
dw dfl dw dfl e~o
.. ' ' ' " .J L "- 1 I.

lHU> lllC 'b' ' vy '"~ ~• r '
I. 112
f)e = .:._
I
\7'
\
(14.87)
This shows that the high-frequency components are confined to an angular range
. . -' . .
much smaller than average. J:<Igure I4.1U snow~ ve1y we "'-o u"'"
L • j', j', ·~ •n~~;Qo 0 ~ o1f ,-.,-,m .J "'ith Of thP. OT<iPT Of and much laf!!ef
-, . -,- .
than we- The natural umt of angle yfJ IS used.
The frequency distribution of the total energy emitted as the particle passes
by can be found by integrating (14.79) over angles:
dl
dw
("12
---=<] -w/2
d2J
d W dfl wo--v-rru
n •n .,
~
r
1-~dwdfl
d21 -liJ (1 A RR\
\
~,
(remember that fJ is the latitude). We can estimate the integral for the low-
freauencv ramw bv usinl! the value of the angular distribution (14.83) at fJ - 0
and the critical angle fJ, (14.85). Then we obtain
dl d 2l ez wp
dw - Z7rfJe dw dfl
(14.89)
8=0 - -z' c
eh 'th~t thP < 1ses ~s w113 for w << w . This gives a very broad,
·urn·
fiat spe~trum at frequencies below we. For the high-frequency limit where w >>we
we can integrate (14.86) over angles to ootam the reasonably accurate resun,
1/2
,iJ e2 w
dw
VJ7r/2 c il w,
e c ~l't.)IV)
t-
'\
t I
.... I~ \: -
-
~~ _'\. Filmre 14.10 Differential frequency
~ F ,,; ~r ;nlP l'm
~W"-."-.Wc
r--. .......... -c frequencies comparable to the critical
frequency "'"' the radtatwn ts connnea
\ ~---- • nl •e , f thn Ar.-1ur , f '
much smaller (larger) frequencies, the
1 For
0 1 2
"Y9- angular spread is larger (smaller).
Sect. 14.6 Frequency Spectrum Emitted in Instantaneously Circular Motion 681
r
A proper integration of (14.79) over angles yields the expression,*
:~
2
1
V Je 1' w K, 1,(x) dx (14 91)
In the limit w << w, this reduces to the form (14.89) w1th a numencal coeffiCient
3.25, while for w >>we it is equal to (14.90). The behavior of di!dw as a function
of ft equency is shown in Fig. 14.11. The peak intensity is of the order of e2 1k.
and the total energy is of the order of e 2 yw)c = 3e 2 y 4 / p. This is in agreement
with the value of 41Te 2 y 4/3p for the radiative Joss per revolution (14.32) in circular
accelerators.
The radiation represented by (14.79) and (14.91) is called synchrotron radi-
ation because it was first observed in electron synchrotrons (1948). The theoret-
ical results are much older, however, having been obtained for circular motion
by Schott (1912) although their expression in the present amenable form is due
to Schwinger. For periodic circular motion the spectrum is actually discrete, being
composed of frequencies that are integral multiples of the fundamental frequency
w 0 = c/ p. Since the charged particle repeats its motion at a rate of c/21Tp revo-
lutions per second, it is convenient to talk about the angular distribution of power
radiated into the nth multiple of w0 instead of the energy radiated per unit fre-
quency interval per passage of the particle. To obtain the harmonic power ex-
pressions, we merely multiply di!dw (14.91) or d 2 1/dw d[! (14.79) by the repetition
rate c/27Tp to convert energy to power, and by w0 = c/p to convert per unit
frequency interval to per harmonic. Thus
\ 2 .o
ar n - .L ( c) a 1
An-?,.,.,. n 7i:::an
0
2 (14.92)
1 I C\ dl
Pn = ~
~
( -
\F'
l -:;-- w nw 0
These results have been compared with experiment at various energy synchro-
trans. t The angular, polarization, and frequency distributions are all in good
agreement with theory. Because of the broad frequency distribution shown in
Fig. 14.11, covering the visible, ultraviolet, and x-ray regions, synchrotron radi-
ation is a useful tool for studies in condensed matter and biology. We examine
synchrotron light sources and some of the insertion devices used to tailor the
spectrum for special purposes in the next section.
Synchrotron radiation has been observed in the astronomical realm associ-
ated with sunspots, the Crab nebula, and from the particle radiation belts of
Jupiter. For the Crab nebula the radiation spectrum extends over a frequency
range from radiofrequencies into the extreme ultraviolet, and shows very strong
polarization. From detailed observations it can be concluded that electrons with
*This result and the differential distribution (14.79) are derived in a somewhat different way by
J. Schwinger, Phys. Rev. 75, 1912 (1949). Schwinger later showed that the first-order quantum-
mechanical corrections to the classical results involve the replacement of w -> w(1 + hw!E) in
w- 1 d 2 !Jdw dD. or w- 1 dlldw [Proc. Nat/. Acad. Sci. 40, 132 (1954)] and are thus negligible provided
hw, << E, or equivalently, y << (pmc!h) 112
'F. R. Elder, R. V. Langmuir, and H. C. Pollock, P~ys Re1• , 74, 52 (1948)· D H Tomhon!ain and
P. L. HartmaA, Phys. Ret·., 102, 1423 (1956); G. I!atllow, E. Freytag, and R. HaoRsol, l. 4ppl P~ys,
37, 3449 (1966).
00~ Chapter 14 Kamauuu uy 1nuviug \...Har 6 ~, -'lJ
0.6 I ' , I ' I I '

I'..
I "\
0.5 "\
?;
"'
0.4
no
-
\ "\.
'\.
-
-
-
~
0.2 1- -
U.l
0.0
0
l I
1
I
2
,
y
.......
' I
3
--- , I
4 5
(a)
0.6 ~ l II I I I
f-
0.5 f--
f-
/
/
/ \ \
\
-
-
v
u.
~ 03
0.2
t:-
\ \
-
0.1
nn
.01
t-
l
.1
' '
1
\ ........
-
-
-
10
y
(b)
Fi!mre 14.11 Normalized synchrotron radiation spectrum (lll)(dlldy) =

r~
10\ I K. o(x) dx. where v w/w and I 41Te"y4 /3p: (a) linear abSCISSa sciife
Jy
and (b) logarithmic abscissa scale.
enerl!ies ran !ling up to 1013 e V are emitting synchrotron radiation while moving
.1 h ,j' ,j rl-;tc ; , n rnn~not;n ;,rJ, • r.f thP r.rnPr r.f Jn- 4 U~llSS
~-
(see"Pr65Iem 14.LOJ. 1 ne raaio emtsswn<rr "Tu 1v1nz 11om J - crrurrr
energetic electrons trapped in Van Allen belts at distances from a few to 30-lOU
radii (R 1 ) from Jupiter's surface. Data from a space vehicle (Pioneer 10, Decem-
ber 4, 1973, encounter with Jupiter) passing within 2.8R1 showed a roughly dipole
magnetic field with a dipole moment of 4R] gauss. Appreciable fluxes of trapped
electrons with energies greater than 3 MeV and a few percent with energies
cal, but the language of photons can be use , 1 es1red. The number o photons
per unit frequency interval is obtained by dividing the intensity distribution
(14.91) [or (14.79)] by hw. Then the photon frequency distribution is
57r
N=-ya (14.94)
V3
(hw) (14.95)
energies, fundamental wavelengths (27rp) of the order of hundreds of meters give

rise to synchrotron photons of wavelengths down to 10 10 meter (1 angstrom) or
less, corresponding to keV x-rays.
14.7 Undulators and Wigglers for Synchrotron Light Sources
photons for research and was utilized initially in a "parasitic" mode by biologists
and condensed matter physicists. Curved crystals or other devices were used to
select s ecific fre uencies from the continuum. As a lications rew the need
development of magnetic "insertion devices" called wigglers and undulators to

be placed in the synchrotron ring. The magnetic properties of these devices cause
peaks. The basic formula for the radiation is still (14.67), although here we use
invariance arguments and Lorentz transformations to make the results more
ally an electron and usually moving relativistically, ( y >> 1), is caused to move
transversely to its general forward motion by magnetic fields that alternate pe-
. . . .
~ ~ ~ ~ r--71
fS /jffJ@~J@ N S N S
~
/ ..-z
x/y~~/7/71
-r-V ~r~/ ~ L:Y
(a)
/
/
n A t
t ~
'-o
7
~ /
(b)
Figure 14.12 (a) Schematic diagram of alternating-polarity bending magnets for a

wiggler or undulator. (b) Sketch of approximately sinusoidal path of electron in the x-z
plane. The magnet period is A0 , the maximum transverse amplitude is a, and the
maximum angle is !fi0 •
or me panic1e, is SKercneu in rig. l'l.lLa. 1 ne palll u1 tHe paLLi"'"' ;, i11 we 11u1i-

zontal (x-z) plane.
A. Qualitative Features
T£ •'-
• -1.' • .£ •L . ,. -'-' • ' +J.. 'nl 0 .+J.. .;n
u <H~ p~~ :o v "'~ '" ''- •uo y •t
be approximately sinusoidal in the transverse direction with the same period, as
sketched in Fig. 14.12b. We have x =a sin(21Tz/.\ 0 ), with the maximum amplitude
a aepenaem on tne strength oTTffe wiggrer s magnetic Ht:JU anu Lnt: _pa,,i'-"' s
energy. The maximum angular deviation l/10 away from the forward direction is
proportional to a; it is an important parameter, which distinguishes undulators
from wigglers. We have
I \
,/, I dx l 21Ta ,_ " mh0r0 I< ?~ll (1.1 QA)

·u ·u· ' v ·v
rl7 l
is the fundamental wave number of the system. [Actually, the time taken for the
n"rtir.lte tn traverse one neriod of the maonet structure ts 1 ,\,/£3c and so the
real fundamental wave number of the radiation is {3k 0 • For y >> 1 the difference
is insi!!nificant.l
v. ____, 1 +> _,. ;++ ~ ;:;: -' n~rtiC"lP ie N-.nhnPrl tn "
• 'J ·o r
narrow angular region of angular width !18 = 0(1/y) about the actual path. As
thP ·,.1, • in ito r.oC"ill~tnrunoth eL-PtC"hcrl in Fio 141?h thP" ""lioht"
r -, r-
,. ~
oeam 01 rauiation wi11 HiCK oacK anu ronn auuut tHe 1u1 watu uit . ...., -
Sect. 14.7 Undulators and Wigglers for Synchrotron Light Sources 685
tatively different radiation spectra will result, depending on whether 1/Jo is larger
or smaller than !l 8.
(a) Wiggler (!fro » JiO)

For 1/Jo >> /l(), an observer detects a series of flicks of the searchlight beam,
with a repetition rate given by the relation, v0 = w 0 /2w = ck 0 /2w. With /\ 0 of the
UlUICl Ul d liCW , Vo V~lV UHL)· lllC, 1> VClJ lllUICll <t>
in an ordinary synchrotron with bunches spaced a few centimeters apart. The
of r>~rliMion f\Xtf\ncl' to'" if\' th>~t >~rP -v1 tim<'\' thP h'"i" frPmlf\n"v
f! = c/R, where R is the effective radius of curvature of the path. The minimum
value of R is generally the one of interest. It occurs at the maximum amplitude
of the transverse motion and is
1
ThP wicrcrlPr r,.rJi,.tirm 1m;," 'mnnth fp"h •1 - " 1m vP.rv m11rh lil<P
the syn~hrotron radiati~n spectrum of Fig. 14.11, with a fundamental frequency,
n = 2wcifJ0 //\ 0 , and a critical frequency ·l times this value. If the wiggler magnet
structure has N periods, the intensity of radiation will beN times that for a single
pass of a particle in the equivalent circular machine.
L ' £ •• ' • " < L T/ • ' " • ' L
H" cv mw "''-' !-"'~' n, a ""IS~'-', uy
K = YI/Jo
A wiggler is characterized by K >> L In terms of K, its critical frequency is
?,.r
(14.98)
U>Cl> Vl_>Jlll..lll ll!;lll >VUli..CO> liCllU lU . Ul W<lV ldlllCl llldllllV
quency. The critical wavelength is ~
(14.99)
(h) Undulators (!fro<< aO or K << 1)

u 'flo'"--- uu, llle >ecuuutgut uearu ur u1uVe> ~ :-: y - r a • e u tu
its own angular width. This means that the radiation detected by an observer is
"n ,.]mo<t ' " "itinn nf thP ih11tinn". from ,.11 thP n"·ilhtinn< nf
the trajectory. For perfect coherence and an infinite number of magnet periods
~ anu indnitesima• angumr resoLUtion or tne uetector ), Uie rauiation wou1u LJe
monochromatic. For finite N the spread in frequency is !lw/w = 0(1/N); finite
"nmlhr "' "kn rml<P<" <nTP"rl hPrml<P nf thP nnnnlPr <hift NPvPrthP-
H .. oo, WC, UC,'-[< •J ·r C U~H UV~H UH UHUC ~0 r a
series of peaks in practice, but with a most intense "fundamental").
The freauencv of the "line" from an undulator can be estimated bv consid-
ering the particle in its rest frame. The FitzGerald-Lorentz contraction means
. <'- £. • • "- ._, • • _,
Wa< H< lHG< UGOH'-' CH'-' OUU'-'Wl'-' ~> 1 UJ lH'-' 1'""~'-'~'-' VV1lH Q
penod A0 /y. lhe trequency ot stmple dtpole radtatton m that trame ts thus
w' = y(2wc/Ao). In the laboratorv frame the relativistic Donnler shift. w' =
yw(1 - f3 cos IJ) = w(1 + l1J2 )/2y, leads to a spectral line at an angle IJ with
frequency
w (14.100)
Note that at small angles ( yiJ << 1) this frequency has the same y-dependence
as the wrggler's cntlcal frequency, (14.98), for a fixed K.
B. Some Details of the Kinematics and Particle Dynamics

We wish to consider the particle in its average rest frame, in which it executes
oscillations both transversely and longitudinally. If its initial Lorentz parameters
are '}'and fJ, they remain unchanged because the magnetic field does no work on
the particle. But because of the transverse motion, the particle's average speed
in the z direction, f3c, and its associated y, are le~s than the instantaneous param-
eters. The average rest frallle moves with speed {:Jc with tespect to the laboratory.
One way to find j3 and y is to consider the path shown in Fig 14 12b and
compute its length for one cycle:
('o rAo
s= Jo V1 + (dxldz) 2 dz = Jo [1 + ~(dx!dz) 2 + · · ·1 dz (14.101)
or
(14.102)
Here we have assumed that !/to << 1, and we assume below that y >> 1 Since
the particle travels this path as speed f3c, we infer that
f3
,8(1 (14.103)
1 + rfi~/4
Even though f3 = 1 and % << 1, so that /3 = 1, the difference between f3 and f3
produces a finite (not infinitesimal) difference between y and y:
1 2)
-2 1 73" 1 {32(1 1
1o
2Q
1'
=y 2
+ 1 o/2
2 0 = 1' -2(1 + ~K2)
We therefore find
Since K >-> 1, is possible even if lf!a << 1, y can differ significantly from 'f', at least
for wigglers.
The transverse motion has been assumed to be sinusoidal. How is that con-
nected to the structure of the magnet that causes the motion? With {3 and y
constant, the x component of the Lorentz force equation can he written
x = -ef3zBylym, where /3yBz is assumed to be negligible or zero. Approximating
z - ct and /3, - 1, we have
ymc d x ymc k 2 • k
By(Z) = - - - -2 = -- oa sm oZ (14.105)
e dz e
The requisite magnetic structure is By= B 0 sin k 0 z, where B 0 = ynzc 2 k6ale. Since
K - yk 0 a, the important parameter R can be expressed m terms of the known
field of the magnet and its period,
(14.106)
l 1.n
actual magnet stmctme will be periodic, but not sinusoidal. V'/e can,
however, make a Fourier decomposition of the actual By in multiples of k 0 . Each
component will contribute to the motion. The fundamental will dominate. For
simplicity, we keep only that contribution.
The longitudinal oscillations can be found, at least approximately, from the
constancy of {3. We have 13; = {3 2 - {3;. Since lf3xl << {3, we can write
)
But x = a sin k 0 z "" a sin(k 0 ct). Thus f3x "" k 0 a cos(k 0 ct). We then have the
component of Jl in the ;; direction as
Rz\:ft\-! ~ 8 - lea
I 2 e
1
2 cos 2 fk ct\
\: 6 -j
= {3 - ik6a 2 [1 + cos(2k 0 ct)]
2
= {3 - - cos(2k 0ct)
4y2
Integrating cf3z (t) once with respect to t, we find the longitudinal and transverse
motions to be
AoK .
z(t) and x(t) sm(k 0 ct) (14.107)
27Ty
C. Particle Motion in the Average Rest Frame

It is informative to examine the particle's motion in the frame K', moving
with speed 73 in the positive z direction. The Lorentz transformations equations
are
x' = x, z' = y(z - {3ct), ct' = y(ct - {3z)
Substituting z(t) from (14.107) into the last equation, we have
ct' r[u(l ff) + !~;2 sin211l

where (} = k 0ct. Neglect of the last term gives the first approximation, t = yt'.
Then with this result inserted into e, we find a better approximation,
or
(} -k
y 0 ct '
Usually the first term is adequate, but in computing time derivatives in the mov-
· <- •'- · .1- .-rr · · "·"
·'5 0 •~•n• •o J >H.CH u LClll uv J·
The narticle's coordinates in the movinQ frame are
K
x'(t')- k sm&(t')- a smO(t')
')I 0
-K2
z'(t') = --;..-- sin 20(t') = Ka .
sm2& (t' )
8y k 0 8V1 + K 212
'1'1-
"'"6'
',].
·" • t
-o· . !-'~'"~"' ~· "'~ '~"''
' •?
A 1\.U
z' +2(z')max ·An j1 2 where (z')max n 2I k
l -= • 1~
~·
1
"
''·
'~ -'-
'5U"-• +'T.+J ouv
<Hv
-' .. , v•....
""-
' .
0 VlUlllll lllt: lllUV111 0 ll<llllt: lUI lfie
>O jJQHLvLv
regime K >> 1. ForK = 1, the z' amplitude is 0.576 times as large as is shown.
ForK<< 1, the z' oscillations are negligible; the motion is simple harmonic in
the transverse direction.
An important feature of the motion in the moving frame is the maximum
>pceu UL Lilt: patll'-'LC. >tl anon ytetas me square or me par-
~
tide's sneed in the movinQ "· • ~~~~u

- - 2
') J<2 J<4 v2
f3'" 2 + K2 cos"& + 4(2 + K2)2 cosL2& 1 2u:2 + K 2) cos 20
(14.109)
where it is now safe to put &(t') = yk 0 ct'. The last factor comes from the form
of d&(t')!dt'. The two limits of K are instructive. ForK << 1, the leading term
gtves
f3' = K cos e, K<< 1 (14.110a)
corresponding to nonrelativistic simple harmonic motion. This limit is for an un-
dulator. In the opposite limit, K--'> oo, the leading behavior is
{3' = 1 - (cos 2 &- D2 , K--'>oo (14.110b)
In this (strong wiggler) limit, the particle's speed varies between 3c/4 and c in
the course ofthe motion, quite relativistic. From Problems 14.12, 14.14, and 14.15,
one can infer that the radiation in the moving frame consists of many harmonics
of the basic frequency, with an angular distribution that is far from a simple dipole
nottPrn ThP ];1. r"cli"tirm nottPrn J', ~ n • -' >r ;o t ..1.
scribed by the contrib~tions from the successive segm~~ts -~f'the path whose
tangents point in the direction of observation.
\.
/
~
._,.
--.........
........ .......-
............ --
...............
' )' X
Figure 14.13 Orbit of the particle in
me movmg (average rest) trame tor K
>> 1 ThP. . ,. . thP
direction of motion in the laboratory
rrarne.
.roo
anu ; IU£ •
D. Radiation Spectrum from an Undulator

When K << 1, the motion in the average rest frame is very simple. The
particle moves in nonrelativistic simple harmonic motion along the x axis. It emits
monochromatic dipole radiation whose power differential distribution is
where k' - yk 0 ts the wave number m the movmg trame. lhe coordmates are
shown in Fi2:. 14.14. Now k' 2 sin2 (•) can be written as k' 2 sin 2 E> = k' 2 - k' 2 cos 2 1-.l
= k~ + k; With K = yk 0 a = yk 0 a for K << 1, the power angular distribution
2 2
(14.111)
To find the laboratory spectrum in angle and frequency (actually, either angle
or frequency), we exploit certain in variances. Since the phase-space density d 3 k/w
is a Lorentz invariant, it is useful to consider w' d 3 P' /d 3 k', rather than dP' !dfl'.
Inserting a delta function o(k' - yk 0 ) to assure the monochromatic nature of the
radiation in the moving frame, we have
2 2T.r2 N U -.1. \- -131.1
(14.112)
where d 3 k' = k' 2 dk' dfl'. Consider now d 3 P'. If we multiolv bv the time !::.t' it
takes for one period of the magnet structure to pass by the particle in the moving
C. lA 1 --,:; •-~\ .L ,,_ -1: -1 ' -1' oL
\~' "0 !f-'v "0 yvJ• '"' cwC · '5J" 1-'~' 1-'~uvv uuv "'~
invariant element of phase space. If we divide by nw' = lick', we obtain the

differential number d 3 N' of photons emitted into d 3 k' Iw' per passage of a magnet
· r1 R· th ,1, ,, ,h · · · •• ;. ur, th
~rite the connection betwee'n the laboratory differen~ial radi.ation spectrum and
the spectrum in the moving frame as
!::.t' w d 3 P'
!::.t w'
k'
;:
" ' R' I :
t 1/
Figure 14.14 Radiation direction and angles
in ovPrO<>P rP.,t fromP. PortirlP motion j, in thP.
x-z plane.
With !:it' Il1t = 1/y and d 3 kl w = k dk dille, we have

3
dk dfl
(14.113)
All that remains is to express the primed quantities in terms of the laboratory
·variables. The Lorentz ti ansfot mations are
k'y = ky = ..~,
k sin () sin 1.{/, cf>' = cf>
z
k' = yk(1 - 73 cos IJ)
I Ising the constraint of the delta function, W€ hav€
If we make the appropriate approximations for y >> 1 (i.e., () << 1,

73 = 1 - l/2y2 , etc.), (14.113) can be written
dYJ dk dcf>
8[k(1 + TJ) - 2y 2 ko] (14.114)
2
where 17 ( yiJ) is the natural angle vmiable to replace cos IJ. Note that, be-
cause of the delta functton, the frequency and angular distributions are not
independent.
(a) Angular Distribution
If we choose to integrate over the frequency spectrum dk, we find the angular
distribution of power to be
d2P e 2 cy2 K 2 k6 (1 - TJ? + 4TJ sin2 cf>
-'· -' , 5 (14.115)
•t ·'!' ~ \ 'IJ
(14.116)
where
(14.117)
is the total power radiated It is easy to verify that the average valu€ of '1'1 is
(b) Frequency Distribution

To obtain the frequency distribution emitted into an angular range, YJ 1 <
7J < Tl?, we integrate (14.114) over d<J> d"Q. The result is
dP
dv = 3P[v(1 - 2v + 2v)] for Vmin < V < Vmax (14.118)
where v - k72y 2 k 0 and Vmin - 17(1 + TJz), Vmax - 17(1 + TJ1 ). lhe complete
normalized frequency spectrum is plotted in Fig. 14.15a: the sharply peaked spec-
" I I I I I
31-- I -
II /:I
!/' I
I
I
2-
: :-
;: I
I
I I
I
I
I
I
I
1-
/ I
I
I
:-
I
/ :
I
:
i
o/
/ I I I I
I
I
i
I
I
i
0.0 0.2 0.4 0.6 0.8 1.0
v
(a)
10 I
I -
/
1- / -
~
-
/\
/
v., -
0.01
0.1 1
(b)
>AH:: 1~\
'· ,, v ~~ 1 " . '_.! n 1
The dash~d lin~~ indicate the freq:enc~ i~erval visible if the angular acceptance is
u < ytJ < 3 .. lD) Log-Jog plot ot mtensny or runaamema1 ana secona narmonic 10r 1'..
0.5 with a sinusoidal magnetic field. In real undulators, the spectrum shape depends on
details of the undulator structure.
()~~ Chapter 14 Radiation by Movmg Charges-G
trum between the dashed lines corresponds to an angular acceptance 0 < 7) < 1/9
~tl <- 1/YJI). l'Wte mat mis spectrum is ror perrecuy sinusoiaa1 mmion or me par-
tide at all times. If the number N of magnet periods is finite, the duration of the
. . . ~ . .,_ r ~ . ., r_ • -'
'Y '"'"""'" "'""""'• "''-' wan_, uam vnu uav'-' a H• :
in frequency of the order of liN. For large N this spread is generally small com-
-' +L -' <'. +: . .
c~ ., •~ ~Y ~u~ ~ .~ -~'"' ,
'
For small, but not negligible, K, there are higher harmonics. These can be
,~ ,f f, ~ 1-.'-L ,]t; ,], rll- t1- B.mr. ~· >ht ~.,t.'
·~
sho;~ in Fig. 14.13. The seco~d harmonic comes fro~"a cohe;ent su-;~rposition
nfthP fip]ck nf ~ rJinn]p in thP 7 _.I.' r7' CX oin ")I)( t'\] ~nrJ ~ nW -'· liP f'~ll<~rl
by the x' motion. See Problem 14.27. Th~ resulting fr~quency ~pectr~m is shown
in Fi2.. 14.15b forK= 0.5 with hi2.her harmonics decreasin2. in intensitv. at least
forK< 1.
(c) Energy of Photons and Number Emitted per Magnet Period

The radiated power 1s g1ven by (14.11/) and the max1mum energy ot photons
in the fundamental is hwmax = 2y2 k 0 hc (at 7) = 0). The amount of energy radiated
per passage or one magnet penoa 1s fj,f:- r fj,f, wnere fj,f A 0 tc. 1ne numoer
of photons N.y emitted per magnet period can thus be estimated to be NY ;c

'> r?.> •
' '-"" ""'max v \ u n _h w "'-'' '-' u .o "''-' ""'-' " "
on (14.118) divided by hw gives
"·
- ~
... l •
lUVH
277 2
Ny = "i aK (14.119)
E. Numerical Values and Representative Spectra and Facilities

ThP /( onrl/i,, ~rP aivPn fnr · 1 l< in ,] (, .1.
1tnr\
. b ,- max u ,-
umts y
K = eB 0 eBoAo
2
-- = 93.4 B 0 (T)A. 0 (m)
k 0 mc 27Tmc 2
and
9.496[£ (GeVW
hwmax(eV) = (l + K 2 /?\), Jm)
Typical undulators have B 0 = 0(0.5 T), A0 = 0(4 em), E = 0(1-7 GeV). Hence
K- 0(2) and fiwmax- 0(80 eV-4 keV). Wigglers have B 0 - 0(1 T) and A0 -
n f'>n ,\ rr1- v n f'>n\
\' 'I' \' T I'
1 nere are aozens or syncnrotron ngnt racllltles arouna tne wona. 1 yp1ca1 or
thP -' r].,,~: r1 f"rilitio>< ("< nf 100>1\ "rP
,- . '
Advanced Light Source (ALS), Lawrence Berkeley National Laboratory,

F, = 1.5 GeV
National Synchrotron Light Source (NSLS), Brookhaven National Laboratory,
F - 07"i ') <; GPV
European Synchrotron Radiation Facility (ESRF). Grenoble, France, E = 6 GeV

-- ..... _ ,
.....
.....
1.5 GeV undulator \ 7 GeV undulator \
'
\ \
I \
I
bending magnet
E (keV)
Figure 14.16 Representative photon spectra for actual light sources. The bending
magnet and wiggler spectra are continuous and are closely proportional to (14.79),
evaluated at @ 0. Th<l undulator cnPl€lS are th€l €lllY€liop€ls of a series of sharp peaks at
multiples of the fundamental. See text for definition of brightness.
Tristan Light Source, KEK National Laboratory, Tsukuba, Japan, E = 6.5 GeV
Advanced Photon Source (APS), Argonne National Laboratory, E 7 GeV
The lower energy facilities provide photons in the tens of eV to several keV
range; the high-energy facilities extend to 10-75 ke V, and even higher at reduced
flux. Figure 14.16 shows some representative spectra of actual light sources. The
spectral brightnesses indicate the typical capabilities available at relatively low-
energy rings such as the ALS and the higher energy rings such as the APS. For
undulators the smooth curves represent the envelope of the narrow "lines."
Brightness or brilliance is defined as the number of photons per second per mil-
liradian in the vertical and horizontal directions per 0.1% fractional bandwidth
in photon enog:y, divided by 2n times the effective source area in square milli-
meters. High brilliance rather than high tlux is generally desired.
F. Additional Comments
There is a vast amount of detail about synchrotron light sources, the design
of beams, the transport of photons to expenments, and so on. We make only a
few comments here.
I. An undulator's fundamental frequency Wmax can be tuned by varymg the
undulator parameter K by changing the gap in the magnet structme and so
changing B 0 [see (14.106)].
2. The simple undulator with beam oscillations in the horizontal plane provides
linearly polarized light. Circular polarization can be provided by use of a
helical undulator designed to make the transverse trajectory an ellipse. Al-

ternately, two undulators at right angles with an adjustable longitudinal spac-
ing between them can be used to produce circular polarization or any other
state because of the coherent superposition of the radiation from all the
magnet periods.
3. Free electron lasers are closely related to wigglers and undulators. An un
dulator can be thought of as radiating in the forward direction at frequency
Wmax by spontaneous emission. Addition of a co-traveling electromagnetic
wave of almost the same frequency provides the possibility of interaction and
stimulated emission and growth of the wave.
Further details about the sources and about their uses in research can be
found in the references cited at the end of the chapter
14.8 Thomson Scattering of Radiation

If a plane '>Nave of monochromatic electromagnetic radiation is incident on a ftee
particle of charge e and mass m, the particle will be accelerated and so emit
radiation. This radiation will b@ @mitt@d in dir@ctions other than that of the in
cident plane wave, but for nonrelativistic motion of the particle it will have the
same frequency as the incident radiation. The whole process may be described
as scattering of the incident radiation.
According to (14.20) the instantaneous power radiated into polarization state
€ by a particle of charge c in nonrelativistie motion is
dP
(14.120)
d!l
The acceleration is provided by the incident plane wave. If its propagation vector
is k 0 , and its polarization vector e 0 , the electric field can be written
ik ·x-hut
Then, from the force equation for nonrelativistic motion, we have the
acceleration,
(14.121)
If vve assiJme that the charge moves a negligibl@ part of a wavelength during one
cycle of oscillation, the time average of Ivl 2 is 2Re(v · v·'). Then the average power
per unit solid angle can he expressed as
(14.122)
Since the process is most simply viewed as a scattering, it is convenient to intro-

duce a scattering cross section, as m Chapter 10 defined by
du Energy radiated/unit time/unit solid angle
(14.123)
d!l Incident energy flux in energy/unit area/unit time
Sect. 14.8 Thomson Scattering of Radiation 695
~I
'"'
.. ·~
The incident energy flux is just the time-averaged Poynting vector for the plane
wave, namely, c I Eo I2 /8'J"f. Thus from (14.122) we obtain the differential scattering
cross section,
(14.124)
The scattering geometry with a choice of polarization vectors for the outgoing
wave is shown in Fig. 14.17. The polarization vector e 1 is in the plane containing
n and k 0 ; e 2 is perpendicular to it. In terms of unit vectors parallel to the coor-
dinate axes, e 1 and e 2 are
E1 = COS 8( ex COS </J + e sin </J) - ez sin 8
For an incident linearly polarized wave with polarization parallel to the x axis,
the angulm distribution sunnned over final polarizations is (cos 7 0cos 2 if> + sin\p),
while for polarization parallel to they axis it is (cos 2 8 sin2 ¢ + cos2 p). For un-
polarized incident radiation the scattering cross section is therefore
(14.125)
This is called the Thomson formula for scattering of radiation by a free charge,
and is appropriate for the scattering of x-•ays by electrons or gamma rays by
protons. The angular distribution is as shown in Fig. 14.18 by the solid curve. The
total scattering cross section, called the Thomson cross section, is
IJT -
~
3
(!t._)
87T ~mcz- (14.126)
The Thomson cross sectiOn rs equal to 0.665 X 10 24 cm 2 for electrons. I he umt

of length e 2 /mc 2 = 2.82 X 10- 13 em. is called the classical electron radius. since a
classical distribution of charge totaling the electronic charge must have a radius
of tlds 01 der if its electrostatic self energy is to equal the electron mass.
The classical Thomson formula is valid only at low frequencies where the
momentum of the incident photon can be ignored. When the photon's momen-
tum hw/c becomes comparable to or larger than me, modifications occur. These
,.~
-
,~
~~\·
,,,
i\\~
ol
!G-
~ 0.5 I-
I\\
I \ \
II \ \\
\\ \\ \ ~ _,/
/
-- --
>ei
b
"'
..
.!I.
\ • \ \ .....<:
~- ' ...
\\ ' ":.:....
0.25 ..... ~:·
..... ··~ ...........
-
'\·,..... ,,' ~:.:.:.················:.:.::::.::. ..........

'·.,·..·•. .....' -....... ----- -~
------------ 0.50 -
','·····..................................................
~--- _____ _.C.!~---
I
01 0 -I
....,, 0 L.AB
.. _,
~· . UHI · o1 c~ottPrin or. ~1
·" .-1' ' 1
ch~rged particle initially at rest in the laboratory. The solid curve is the clas.~ic;l ~
Thomson result. The dashed curves are the quantum-mechanical results for a sninless
particle, with the numbers giving the values of fiw!mc 2 • For fiwlmc 2 = 0.25, 1.0 the
-~ ~ -'- ... .•• £. • 1 . .
<Hv >vouuo <V> opm 2 pvuu pm m IUmJ.
can be called quantum-mechanical effects, since the concept of photons as mass-

1 • -' • oL .1 ' · '
>~oo )!«<""'~, '"" "uuu 'bJ >o " " " 'Y II. ~puce,
Newton!), but granting that, most of the modifications are purely kinematical.
"T'J. ' ' ,1, ' •• _, • " ' '"' ~'-
·'Jc ·v~v v~ "!:" nauy uy .,.nuu. ~"'-
energy or momentum of the scattered photon is less than the incident energy
because the charged particle recoils during the collision. Applying two-body rel-
ativistic kinematics to the process, we find that the ratio of the outgoing to the
incident wave number is given by the Compton formula,
k' 1
z,. = J;
1 + -2 (1 - cos 8)
me
where (I is the scattering angle in the laboratory (the rest frame of the target). A
quantum-mecnamcal calculatiOn ot tne scattenng of photons by spinless pomt
n~ ' nf ~h<>r<TP o <>nrl m<>oo ~ ;PTrlc th
r
'-'
·o
, '
, J
dcr e 2 \ 2;k''2 "

I mcz I k IE • Eo I ~l't.lLI)
dfl
to be compared With the classical expression (14.124). In the radiation gauge the
~· ' .. 1 . _, • •• d. _, ' _, _,. _, ~·-
'1' ~u•" ao "'" -,- ' • ""
2
factor (k' /k) comes entirely from the phase soace. Its oresence causes the dif-
' >1 ~rncc tn "· rPl"t; t. •• ·'' 1.
'b'
.1.
·o ,
as shown by the dashed curves in Fig. 14.18. Also shown in the figure by the
Ch.l4 References 697
dotted curves are the quantum-mechanical results for photon-electron scattering,

that is, the scattering by a point spin particle described by the Dirac equation. 1
The curves are generally similar to those for spinless particles, but are somewhat
1. I .I L -~ • 1. -•L , I ' • ~
•u•s~• u• •u•_o• u '5 'J ·
The integral over angles of (14.127) is elementary but slightly involved. We quote
only the limiting forms for nw << mc 2 and nw >> mc 2 :
y
(}"
1 - 2 ""'2 + ...
mc '
nw << mc 2
- -- (14.128)
O"r 3 mc 2
---
'+ fi{J)
, nw >> mc 2
For scattering by electrons the low-frequency limit is the same, but at high fre-
quencies there is an additional multiplicative factor, [~ + 1ln(2hw/mc 2 )].
ror proLOns me aepanures rrom me 1nomson rormUla occur m pnmon en-
ergies above about 100 MeV. This is far below the critical energy nw ~ Mc 2 ~
1 GeV, which would be expected in analogy with the electron Compton effect.
The r. is th~t ~ nroton is not a noint narticle like the electron with nothin!!
but electromagnetic interactions, but is a complex entity having a spread-out
cnarge mstnbutwn With a raams ot tne omer ot u.ts X lU em caused by tne
strong interactions. The departure (a rapid increase in cross section) from Thorn-
. . .
son scaLLenng occurs m pnmon energtes 01 LUe oraer or me rest energy 01 tne p1
(140 MPVl

The radiation by accelerated charges is at least touched on in all electrodynamics
textbooks, although the empnas1s vanes considerably. The relatJVJStlc aspects are treated
in morP. or lf>« nf>t"i] in
Iwanenko and Sokolow, Sections 3'!-43
Landau and Lifshitz, Classical Theory of Fields, Chapters 8 and 9
Panofsky and Phillips, Chapters 18 and 19
~- ;~ .. 1.-1 1:1. - .,
·'· --- "'- •• ; '")(l .-l ~"
' '
Extensive calculations of the radiation emitted by relativistic particles, anticipating
many resuns reoenveo m rne perioo t'oi'+U-l'oi:Ju, are presemea in rne imeresring mono-
Qranh bv
Schott
__,, ,__ _,
~-
"J -"
'1-"r T"J'' ·'· t"'J •
.-1
matter physics, material science, and biology. Synchrotrons and electron storage rings as
such are discussed in detail in a classic reference
M. Sands, "The physics of electron storage rings," in Proceedings of the Inter-
nattuflUl UJ I llJ"CO ~' 1TUnt, L.UUTOe l''U. 'fU, cu., U. 1'
'
Academic Press New York (1971), pp. 257-411.
*For electrons the cross section equivalent to (14.127) has Ie* ·Eo I' replaced by
I. * .12-+ (k- k'jl f1-+ I~* v ~\. 1~- v ~*\1

4kk'
It is known as the Klein-Nishina formula for Compton scattering.
0~0 Chapter 14 Kadianon by Movmg Charges -G
A
_, -" llions_"rP ~ in rlPt"il in
A. G. Pacholczyk, Radio Astrophysics, Freeman, San Francisco (1970); Radio
Galaxies, Pergamon Press, Oxford (1977).
Plasma ohvsics aoolications ""' · -' hv
n~t ,,;
The classic reference on the subject of wigglers and undulators is

H. Motz, J. Appl. Phys. 22, 527 (1951).
The production and characteristics of synchrotron radiation from bending magnets, un-
dulators, and wigglers and the many uses are covered exhaustively in the five-volume
work,
HanaoooK on Synchrotron Radiation eds. E. E. Koch and others Vol' 1 A 1 R
2, 3, 4, North-Holland, Amsterdam (1983-1991).
In Volume 1A, Chapter 2, S. Krinsky, M. L. Perlman, and R. E. Watson cover all of the
theory and comparison with exoeriment.
An unpublished 1972 treatment of undulators and wigglers by E. M. Purcell, very like
that of Section 14.7, appears in
Proceedings, Wiggler Workshop, SSRP Report 70/05, eds., H. Winick and T.
Knight, Stanford Linear Accelerator Center (1977), p. IV-18.
The scattering of radiation by charged particles is presented clearly by
T ,rl d_l_" .
, r1. . ' - t 'T't. '}' 'Jc.-,, .u. , <' ' .LLii -,
~
~- ~-
'~- -
uynamtcs OJ LOnanuous Meata, Lnapters At v ana A v
~-.
••
14.1 Verify by explicit calculation that the Lienard-Wiechert expressions for all com-
ponents of E and B for a · :le · 1P with ~ · _aarPP with thP
ones obtained in the text by means of a Lorentz transformation. Follow the general
method at the end of Section 14.1.
1d, A .-.f "' .I .. n
'b 'J •
ne~r t = t0 , its v;ctorial position can be expanded in a Taylor series with fixed
vector coefficients multiplying powers of (t - t0 ).
eOUSlY at reSt
'"' uuvn ""''• ' " m lllt:l Llill UilJHe Wllt;J t; tm; puilH.:Je 1>
at the origin but has a small acceleration a, the Lie nard-Wiechert electric
field, correct to order 1/c2 inclusive, at that instant is E = Eu + Ea. where
'"~ . _'_ uu~ H>UH llCJU> <U C
E" = e r2 + e2 [a - 3r(r ·a)]; Ea = _.!:__[a- r(r ·a)]

r 2c r c2 r
anu Lnar rne total electnc held to this order IS
n r e ' ' '

,2 '},2.1" .,. "
"JJ
The unit vector r points from the origin to the observation point and r is the
magnitude of the distance. Comment on the r dependences of the velocity
""u acce1er~ion helds. Where 1s tne expanswn likely to_t>_e Vllil(l_.
(b) What is the result for the instantaneous magnetic induction B to the same
{r\
order? Comment.
<:how th"t th.,~/,2 +. . ,, 'f;,Jrl. .-~;
·o·
~ .. ••
curl of the electric field is V x E = e(r x a)/c r • From Faraday's law, find
2 2
the magnetic induction B at times near t = 0. Compare with the familiar

_, lrV
Ch. 14 Problems 699
m arbltrary motwn, an alternativeTotiTeT!enam -vvi~~~~~" ~AP' · . \1'1.1" ;, "
E = e[;,] + e[R] :t [;,]

ret C rtt ret
+ ec'd; 1
, [n],e,
where the time derivatives are with respect to the time at the observation point.
The magnetic field is given by (14.13).
Using the fact that the retarded time is t' = t - R(t')!c and that, as a result,
dt
dt'
show that the form above yields (14.14 when tne lime amerentwtwns are
performed.
14.4 Using the Lienard-Wiechert fields, discuss the time-averaged power radiated per
unit solid angle in nonrelativistic motion of a particle with chargee, moving
(a) along the z axis with instantaneous position z(t) = a cos w 0 t,
(b) in a circle of radius R in the x-y plane with constant angular frequency w 0 .
Sketch the angular distribution of the radiation and determine the total power
raawtea in eacn case.
1.:1_<; A -'- · .;",;~ nf rh"raP 7P m"ss m and kinetic enemv E makes a head-
pulsive and described by a potential V(r), which becom~s greater thanE at close
distances.
(a) Show that the total energy radiated is given by
2
£> w =
4 z 2e 2
- - 2 3
'm foc -dV
1- dr
:-r.=:c===c=~
3 m c ~ 2 ''"'" dr Y V(rmin) - V(r)
r . io thP -' ..,f h in the collision.
~b J 11tne mteraction is a l.-OtiTomiY,. :rnr--vc-r) Z/-e If, >llUW llH1L lll~ LUldl
energy radiated IS
8 zmv6
t.W=---
3
45 Zc
where u0 is the velocity of the charge at infinity.
14.6 (a) Generalize the circumstances of the collision of Problem 14.5 to nonzero
angular momentum (impact parameter) and show that the total energy
radiated is given by
' 2 2
112
r~
t> w - ___:<:__ I- l 1-liE-V(r)---l dr
wu~•~ min , ut~ ~•u,~>e Ul app- \' !•

L mou 0 , wnere D is me impacr parametc1, ctuu u 0 i, '"c
(E = mv612).
(b) Specialize to a repulsive Coulomb potential V(r) = zZe2 /r. Show that £> W
can be written in terms of impact parameter as
2zmv 50 I t'\
£> W = -- -t- 4 + 1- 5 11 + --;-) tan- 1 t
LC \ ·'I
wu~• ~ • -u
2
'HU()I ~'-'t
2•
" ' " ' " auu u
.. .,. .,.
or cwsesr approacn in a nearr=urr
Show that in the limit oft going to zero the result of Problem 14.5b is
recovered, while for t >> 1 one obtains the approximate result of Problem
(e) Using the relation behveeH the scattering angle (} and t ( cot (}f2), show that
Ll W can be expressed as
tan ~]
(d) What changes occur for an attractive Coulomb potential?
14.7 A nonrelativistic particle of charge ze, mass m, and initial speed u0 is incident on
a fixed charge Ze at an impact parameter b that is large enough to ensure that the
particle's deflection in the course of the collision is very small.
(a) I Ising the Larmor power formula and Newton's seeond law, calculate the
total energy radiated, assuming (after you have computed the acceleration)
that the parllcle's traJectory is a straight line at constant speed:
(b) The expression fmmd in part a is an approximation that fails at small enough
impact parameter. For a repulsive potential the closest distance of approach
at zero impact parameter, r, = 2z7e 2 /mv5, serves as a length against which
to measure b. The approximation will be valid for b >> r'" Compare the
result of replacing b by r, in part a With the answer of Problem 14.5 for a
head-on collision.
(c) A radiation cross section x (with dimensions of energy times area) can be
defined classically by multiplying Li W(b) by 2r.b db and integrating over all
impact paramet<lrs. Beeause of the divergeHee of the expression at small b,
one must cut off the integration at some b = bmin· If, as in Chapter 13, the
uncertainty principle is used to specify the minimum impact parameter, one
may expect to obtain an approximation to the quantum-mechanical result.
Compute such a cross section with the expression from part a. Compare your
result with the Bethe-Heitler formula [N- 1 times (15.30)].
14.8 A swiftly moving particle of charge ze and mass m passes a fixed point charge Ze
man approximately strmght-hne path at impact parameter b and nearly constant
speed 11 Show that the total energy radiated in the encounter is
LiW
This is the relativistic generalization of the result of Problem 14.7.

14.9 A particle of mass tn, eha1 ge q, moves in a plane pe1pendiculat to a unifonn, static,
magnetic induction B.
(a) Calculate the total energy radiated per unit time, expressing it in terms of
the constants already defiried and the ralio y of the par!icles's total energy
(b) If at time t = 0 the particle has a total energy E 0 = Yomc 2 , show that it will
have energy E - ymc 2 < E 0 at a time t, where
2q'B 2 y Yo
provided y >> 1.
Ch. 14 Problems 701
(c) If the particle is initially nonrelativistic and has a kinetic energy T 0 at t = 0,

what is its kinetic energy at time t?
(d) If the particle is actually trapped in the magnetic dipole field of the earth
and rs sprrahng back and forth along a !me of force, does n radrate more
energy while near the equator, or while near its turning points? Why? Make
quantitative statements if you can.
14.10 A pailicle of charge e moves at constant velocity l3c for 1 < 0. During the shot t
trme mterval, 0 < 1 < 15.1, rts velocrty remams m the same drrectron, but rts speed
decreases linearly in time to zero. For I > !:;1, the particle remains at rest.
(a) Show that the radiant energy emitted per unit solid angle is
dE (2 - {3 cos 8) [1 + (1 - {3 cos 8) 2 ] sin2 8
dO (1 - {3 cos 8) 4
where 8 is the polar angle relative to the direction of the initial velocity.
(b) In the limit of y >> 1, show that the angular distribution can be expressed
as
\0J7· and that the expression for the

total energy radiated is in agteetnent wiih the result front (14.43) in tl1e saJJLe
lml.
1411 A particle of charge ~g and ma~~ m move~ in external electric and magnetic field~
E and B.
(a) Show that the classical relativistic result for the instantaneous energy radiated
per unit time can be written
2 z4 e4
P = - - ·2- 3
3m c
y'[(E + 13 X B? - (13 · E?J
where E and B are evaluated at the position of the particle and y is the
particle's instantaneous Lorentz factor.
(b) Show that the expression in part a can be put into the manifestly Lorentz-
invariant form.
where r0 - e 2 fmc 2 is the classical charged particle radius

14.12 As m Problem 14.4a a chargee moves m srmple harmomc matron along the z axrs,
z(t') = a cos( w 0 t').
(a) Show that the instantaneous power radiated per unit solid angle is
where {3 = a<v 0 /c.

(b) By performing a time averaging, show that the average power per unit solid
(c) Make rough sketches of the angular distribution for nonrelativistic and rel-
ativistic motion.
14.13 Show explicitly by use of the Poisson sum formula or other means that, if the
motion of a radiating particle repeats itself with periodicity T, the continuous
frequency spectrum becomes a discrete spectrum containing frequencies that are
integral multiples of the fundamental. Show that a general expression for the time-
averaged power radiated per unit solid angle in each multiple m of the fundamen-
tal frequency w 0 = 27r!T is:
2
~· = ~:~~: Lhlwa v(t) X n exp[imwo(r- n · :(t))] dt
14.14 (a) Show that for the simple harmonic motion of a charge discussed in Problem
14.12 the average power radiated per unit solid angle in the mth harmonic is
dPm ezcpz z 2 T2
n =--2m tan eJm(mf3 cos e)
du 27ra
{h\ "'- .. . " ldUldlC:U 1> dll lll llLC:
aJ anu nas Lne vame

L e
, ~ 3 woa
2 . •'-
14.15 A o: ._,_ nf P
~.- .. , -c -, ~a·
{,.) Sh~w thot th~ ~vo~t f, · t~ .• 1 ,. .< ~; ~

•'- •'- -'·. _, -r ·o ·t• ·--·-·--
'"u v' ~o "
dPm
dO =
2
e w'6R
27rc3
2
m
2{[dJm(mf3 sin
d(mf3 sin e)
e)] 2
+ {32
cot>e .P:. ( . )}
m m/3 sm e
where f3 = woR!c, and lm(x) is the Bessel function of order m.

(b) Assume nonrelativistic motion and obtain an approximate result for dPm!dO.
Show that the results of Problem 14.4b are obtained in this limit.
(c) Assume extreme relativistic motion and obtain the results found in the text
for a relativistic particle in instantaneously circular motion. [Watson (pp. 79,
249) may be of assistance to you.]
14.16 Exploiting the fact that k 0 d 3 N/d 3 k, the number of quanta per invariant phase-space
element d 3 klk 0 , is a Lorentz-invariant quantity, show that the energy radiated per
unit frequency interval per unit solid angle, (14.79), can be written in the invariant
and coordinate-free form
where dT is the proper time interval of the particle of mass m, p'" is the 4-momen-
tum of the particle, k~" is the 4-wave vector of the radiation, and E~o E2 are polar-
ization vectors parallel to the acceleration and in the direction E1 x k, respectively.
The parameter is
- 2v'2 (p . k)3/2
g - 3m . (I d 2 (p · k)ldT2 1) 112
Ch. 14 Problems 703
This exoression can be used to obtain the results of Problem 14.17 in an alternative
manner. Hint: In proceeding with a solution, it is useful to expand k · r(t) around
t = 0 in terms of the velocity, acceleration, etc. and compare with (14.72). One
finds. for example, that wc 2 /p 2 = - k · d 2 v(O)!dt 2 , and, because the energy is
constant,
n2
~-
m·c ldL(p · k)ldrl
14.17 A particle of charge e and mass m moves relativistically in a helical path in a
uniform magnetic field B. The pitch angle of the helix is a (a = 0 corresponds to
circular motion).
(a) By arguments similar to those of Section 14.4, show that an observer far from
the helix would detect radiation with a fundamental frequency
Wo = ------::;--
<.:U' U
and that the spectrum would extend up to frequencies in the order of
w, = 23 yw l
8 cos a
,,
'I-'
... ;
oR!"~'·
'·!
~ora ;n .-1. · · 1n tho ro.-l;m ~f ~f
(b) From part a and the results of Section 14.6, show that the power received by
d 2 p ~ 3e 2~
WR
dw d!J - 81r3 c cos 2 a
(
W
w,
r [
the observer per unit solid angle and per unit circular frequency interval is
(1 + y<f!)
2 2 2
K 213 (1;)
~
+ 1 + y<f! 2
"'
2
2
K 113 (1;)
J
where Wn and w. are defined above. f = ( w/2wJ(1 + V<h 2 ) 312 and ,r, is the
o.:::Ta :< ' ' ralot;"" tn ;ha 'o .1
., 0 •
Fig. 14.9. '

14.18 (a) By comparison of (14.91) with (14.79), show that the frequency spectrum of
the received power for the situation in Problem 14.17 is
dP \13e 2 ywB \ w
-:1 l" I LT[ I
'7 c
where G(x) = x L~ K 513 (t) dt and the other symbols are as in Problem 14.17.
This expression shows that the shape of the power spectrum in units of wl w,
is unchanged by the spiraling.
" . n• .. .. ' c
.. 1-y==
. ' ..
'"' ""' ~v=
') 0 2,,,2_, 4
p
.)C
Comparison with (14.31) shows that the total received power is independent
of the pitch angle of motion.
fin doin2: the inte!!ration over solid an!!les in nart a note that <iJ = 0 corresnonds
i~ "'- ~n '" 1
-:rzl;rT -constcrenne angular and trequency spectrum ot rac!Jatwn produced by a magneuc
moment in nonrelativistic motion, using (14.70) and the fact that a magnetization
density .M produces an effective current density Jeer = cV X .M.
/U'f Lnapter 14 Kamanon oy lVIovmg l:harges -G
(a) Show that a magnetic moment fL with magnetization, .M = JL(t) 8[x - r(t)],
in nonrelativistic motion gives a radiation intensity (energy radiated per unit
solid angle per unit frequency interval),
2
d2[ma' w4 r
dw dfl 4-rr"c 3 J J at n x fL\ l)e y
(b) The magnetic moment is located at the origin and is caused to precess by an
external torque such that !Lx JLo sm w0 t and JLz JLo cos w0 t for the time
interva 1 t - T/2 tn t Tn --'- •. -rn~ '>'> 1 .'<hnw thot th ·-~~-, ·o
distribution of the radiation is very strongly peaked at w = w 0 , that the an-
gu.ar uis,riuULIOn or ramauon IS proportiOnal to \1 + sm·11 sm-cp), and that
for T ~ oo, the total time-averaged power radiated is
(P) = 2w;\3 ;li

3c
vv"'pm~ cu~ •~ou willl Llle power CatCUtaLeu uy LUe melllOU 01 .:>eCttOn \U.
A ,,, ' ' 'n .f •L L
>A ""
t
tr
hrr·J_ r_ - f -. . d ·~ - _~ .. uc•~u ~Jh"
e ongm Ippmg rom pomtmg own to pomtmg up, Wit components,
. '"
JL, = JLo tanh(vt), JLx = J.Lo sech( vt), !Lv =0
where ,_,-l is characteristic of the time taken to flip.
(a) Find the angular distribution of radiation and show that the intensity per unit
frequency interval is
3
dfma• 4 'v\
\ ~} JLo (lb\X/7Tncosecn· x + seen xJJ
dx 3
where x = 7Tw/2v is a dimensionless frequency variable and the quantity in
curly brackets IS the normalized frequency distribution in x. Make a plot of
thj, c]i,tri• " •>n.! fin.! <h~ mPon '"'" ,f ' ;,. ,f .
(b) A_QJ:>]y_ the method of Problem 9.7 to calculate the instantaneous nower anrl
total energy radiated by the flipping dipole. Compare with the answer in
part a.
14.21 Bohr's correspondence principle states that in the limit of large quantum numbers
•L ~- • ,-~ •
" r ' ""u lll CH" " , "'lua. cO we prouUCL Ol me
quantum energy (hw 0 ) and the reciorocal mean lifetime of the transition from
principal quantum number n to (n - 1).
(a) Usmg nonrelallvistlc approximations, show that in a hydrogen-like atom the
tron,itinn nrnh "'" ; · ,J mPo · · f, · · · <. ·
cular orbit of principal qu;;ntum number n to (n - 1) is given classically by
1 4
1 _ 2 e2 Ze 2 mc 2 1
-,- - -r:-
1 · 11. ·
-li -n )
5
(b) For -~ thP rlo"il'ol volnf> fmm nart -"--"lith_ thP

quantum-mechanical results for the mean lives of the transitions 2p ~ 1s
\1.0 ;<. 1U S), 4J ~ Jd (7.3 X 1U s), oh ~ ~g (o.l x JU- S).
.
' • ,.,,., n ·
,vu•~
_,. · .<
.0
·
"~;•v"'
·
mumptes
'"" cv"
-
"v--" "'"
·
ul
·
of the basic frequency of the motion. Appreciable radiation in multioles of the

fundamental can occur because of relativistic effects (Problems 14.14 and 14.15)
'" oJ. lli .£ _, - .1. . "-'-1. • •
- 0 - '!' <J '} HUOVoUU'' V< H ~UH 11 lll"
components of velocity are not sinusoidal, even though periodic. An examole of

Ch.14 Problems 705
this latter motion is an electron undergoing nonrelativistic elliptic motion in a

hydrogen atom.
The orbit can be specified by the parametric equations
X = a( COS U - E)
y = a\ll - .? sin u
where
w0 t = u - E sin u
a is the semimajor axis, E is the eccentricity, w0 is the orbital frequency, and
u is an angle related to the polar angle () of the particle by tan (u/2) =
v ~1 E)l~l + E) tan(IJ/2). In terms of the binding energy Band the angular mo-
mentum L, the various constants are
e2 I 2BU 8B 3
u
2B' • y' me4 '
Wo
m,e4
(a) Show that the power radiated in the kth multinle of w is
Pk =
4e2
,,.
13 (kw 0 )
4 2
a
1
1;2 (h• ( kE)) 2 + e- 2 E
2
h kE)
). 2(
where h(x) is a Bessel function of order k.

(b) Verify that for circular orbits the general result above agrees with part a of
L wuleiiJ l'+.Ll.
14.23 Instead of a single charge e moving with constant velocity w 0 R in a circular path
nf radins:_R_ oo ;n Prnhl~- 1A 1<; 1\T .,,.: <: • • t .• t. · • ·
' · · ' - o· -1.] ;
()1 around the same circle.
(a) ' ..
Show that the nower radiated into the rnth nf "'· is
dPm(N) _ dPm(l) F ( )
dO. - _dQ mN
where dPm(l)!dfl is the result of part a in Problem 14.15 withe--> 1, and

N 2
l,(N) 1.6 j=J

qle I
(b) Show that if the chan,es are all eaual in m~· ~ ~ncl 1
. sn~r·pn
around the circle, energy is radiated only into multiples of Nw 0 , but with an
intensity N 2 times that for a single charge. Give a qualitative explanation of
' " " " ' 1<1\.l>.
(c) For the situation of part b, without detailed calculations show that for non-
relativistic motion the dependence on N of the total power radiated is dom-
· V UWl Hl lU<O HHHl H U<J [ iS emUieU.
(d) By arguments like those of part c show that for N relativistic particles of
__eaual
• 'nl
"nr! svm
21'v'/3y3 .f,
. , "' . . .3 ••
. , . _/V_ ,. .' ' . . r
" ' '
_,_,._ ' . CH\o ' 'HH I 'HV
is emitted.
{e) What relevance have thre rP.snlts nf n~rts r. Hl rl tn th~ rorl;ol;n
. ,. ·r
of a steady current in a loop?
14.24 As an idealization of steady-state currents flowing in a circuit, consider a system
of N identical charges q moving with constant speed v (but subject to accelerations)
in an arbitrary closed path. Successive charges are separated by a constant small
,] A
StartinQ with the Lienard Wiechert ootentials for each narticle onrl
no assumptions concerning the speed v relative to the velocity of light show that,
. . .
111 cue 1111uc n , '1 · v, auu '-' ~ v, vue JY'f • ana qw. constant, no
radiation is emitted by the system and the electric and maQnetic fields of the
are the usual static values.
'~' •L <. .o • • d. · · · · ·
v .. ,v ~ ·~~· '"~ ocuuvu 'Y ~ vc tum 111 u•c ~.:uuuucrors
neutralize the bulk charge density of the movinQ charQes.)
14.25 (a) Within the framework of approximations of Section 14.6, show that, for a
relativistic particle moving in a path with instantaneous radius of curvature
p, the frequency-angle spectra of radiations with positive and negative helic-
e2 I wp), { 1 +
UW UH 01T c c \y /
\ y
[UJ ' W < H "'~ <VUHUWO V< ~~~UVH , . ·.v dHU ~ IT<I<IlJUVe, , • UlC n
of the total radiation emitted as a function of frequency and angle. In par-
ticular, determine the state of polarization at (1) high frequencies (w > w,)
for all angles, (2) intermediate and low frequencies ( w < we) for large angles,
(3) intermediate and low frequencies at very small angles.
'~' <' •L • L . n • n n T • ''~ '~ ~ M e
,~, 'i>"Y U]. , •y· .~, , .. , J J O \ U V V J , W l C A f " "
comparison. See also Handbook on Synchrotron Radiation, (op. cit.), Vol.
lA, p. 139.
14.26 Consider the synchrotron radiation from the Crab nebula. Electrons with energies
-1' 13 . •
UjJ LV i V ~ <HVV~ " ' d ,u~ HC.lU V l LllC-uruer-orTIT -gauss;
{.,\ Fnr P 1nl3 PV R 1 'X ln-4 aomo ~~~~ .1~•~ •h -"• c~; •h <.
damental frequency w0 = c/p, and the critical frequency w,. What is the en-
ergy iiw, in ke V?
(h) Show that for a relativistic electron of energy E in a constant magnetic field
the power spectrum ot synchrotron radmtwn can be written
P(E, w) = const ~) J{ !:!_

I \'
.
WllC.<~ j \A)
.
<0 d '-ULVll W1l'-11Vll
. .
'Eo Ulc-varue-urn:ryar
c
.
V allU
rapidly for x >> 1 [e.g., f = exp( -w/w,)l. and w, = (3/2)( eB/mc)( E/mc2 ) 2 cos ()
where ()is the pitch angle of the helical path. Cf. Problem 14.17a.
(c) If electrons are distributed in energy according to the spectrum N(E) dE
ex: E " dE, show that the synchrotron radiation has the power spectrum
(P(w)> dw ex w-" dw
where a= (n - 1)/2.
(d) Observations on the radiofrequency and optical continuous spectrum from
ol r'nl nl .. l. .. L ol ol • C •AR 1
"1' -] CO ~ 'V 0 LV
15 1
w - 6 X 10 s- the constant a= 0.35. At frequencies above 10 18 s- 1 the
spectrum of radiation falls steeply with a :<: 1.5. Determine the index n for
the electron-energy spectrum, as well as an upper cutoff for that spectrum.
l> Ull> I.ULUU Wllll l i l t : L>-urplll'l<r.
Ia\ "T'h L ,]f_];f, ,f .C. •L
time taken for it (o lose one half of its initial energy. From the result of
Ch.14 Problems 707
Problem 14.9b, find a formula for the half-life of an electron in years when
B is given in milligauss and E in GeV. What is the half-life using the numbers
from part a? How does this compare With the known lifetime ot the Lrao
nf'hnla? Must the enemetic electrons be continuallv renlenished? From what
source?
14.27 Consider the radiation emitted at twice the fundamental frequency in the average
rest frame of an electron in the sinusoidal undulator of Sections 14.7.C and 14.7.D.
The radiation is a coherent sum of E1 radiation from the z'(t') motion and E2
radiation from the x'(t') motion.
(a) Using the techniques and notation of Chapter 9, show that the radiation-zone
ma2:netic induction is given to sufficient accuracy by
n
teK a A
A~f~ ~\1
" ,-r~ •
8 V1 + K 2 /2 "
where k' = 2yk 0 , n is a unit vector in the direction of k', and a factor of
exp[ik'(r' - ct')]ir' is understood.
(b) Show that the time-averaged radiated power in the average rest frame,
summed over outgoing polarizations, can be written
2 K' a,
dP' e c
. S'
d!l' t51T (l + K "IL) b4
WHCIC
S' = k'x 4 + k'y 4 + 18k'x2 k'

y
2
+ 17k'x2 k'
z
2
+ k'y2 k'
z
2
+ 8k' k'x2k'
z
(c) Using the in variance arguments in the text in going from (14.111) to (14.118),
show that the laboratory frequency spectrum of the second harmonic is
dP 3 K2
d.v2 = 6
1
P, (
1 + K 2/2
)" •
2
v (10 - 21 v + 20v 2 - 6v
3
)
" --;;;;}-;;:,ij; onrl P. is the ·in the fundamental 7i4.1171. For the
,
c .,,., . '"m "
·~
.1. ~ onrl orP
•to• '/1
"-1\ L 1/2) dHU "max ~.,,
'fl)•
""-
OHU<
_,
... _, .
••
second harmonic?
In Chapter 14 we discussed radiation by accelerated charges in a general way,
. . . .
amples of radiation by both nonrelativistic and relativistic charged particles in
' . .
netic radiation by charged particles in atomic and nuclear processes.
. .
as described in detail in Chapter 13. In these collisions the particles undergo'

I . . . .
during atomic collisions is customarily called bremsstrahlung (braking radiation)
because it was first observed when hi · ·
metallic target. For nonrelativistic particles the loss of energy by radiation is
ne li ible com ared with the collisional ener loss but for ultrarelativi i
ticles radiation can be the dominant mode of energy loss.
Our discussion be ins with consideration of the radiation s ectrum at ver
low frequencies where a general expression can be derived, valid quantum me-
chanically as well as classically. The angular distribution, the olarization, and
the integrated intensity of radiation emitted in collisions of a general sort are
treated before turning to the specific phenomenon of bremsstrahlung in Coulomb
collSlons. When appropriate, quantum-mechanical modifications are incorpo-
rated by treating the kinematics correctly (including the energy and momentum
o t e p oton . Important quantum effects are included in this way, some-
times leading to the exact quantum-mechanical result. Relativistic effects, which
purposes as the violent acceleration of a charged particle initially at rest to some

. . . . . .
ing on of the charge of the moving particle in the same short interval. We discuss
. . . .
and 15.7.
the major quantum-mechanical effects merely by treating the conservation of

. .
708
Sect. 15.1 Radiation Emitted During Collisions 709
effective momentum transfers. In other processes like radiative beta decay the
quantum effects are more serious. Phase-space modifications occur that have no
classrcal basts. Kadratton rs emrtted m ways that are obscure and not easily related
t~ tJ.. nn ,] ·~•+' ,~ ,J.. r!. ,11, .1o 1: !< ~ <L '
, ')' ,... ·o
of "soft" photons, that is, photons whose energies are small compared to the
tnbl Pnc>rcn, ~v~ibhlP II. t the> 11mwr c>nrl nf tJ..a. ~~n~,n-n" " ~'"
classical e;~ressions can be expected to have only q~alitativ~'"validity.
15.1 Radiation Emitted During Collisions

If a charged particle makes a collision, it undergoes acceleration and emits ra-
-·· 0 T£0 _,.. '.1 0 0 0 0
. u '" 'l'm '""' •o mov a vum 0 ... u F"' Ll<.;lt:, lllt:Y UULll t:lllll ,
and a coherent superposition of the radiation fields must h~ ~ Sine~ th~
amplitude of the radiation fields depends on the charge times the acceleration,
the lighter particle will radiate more, provided the charges are not too dissimilar.
In many applications the mass of one collision partner is much greater than the
uld~' u1 lllt: ulllcl. '""" 10r LUe emission 01 raaiarion ir is surncienr ro uear rne
collision as the interaction of the lighter of the two particles with a fixed field of
£ ., <1.
' V ' ' - " • '"
' .1 .1 1
, ' " " V llll'\ lllUl C lll VUl VCU l.Ocl'C~ LU Lllt:
0
'v 'HH V<HJ UHO OH
problems at the end of the chapter.
A. LOW-L'requency Ltmn
_From (14.6:-J) and (14.66) we see that the intensity of radiation emitted by a
particle of charge ze during the collision can be expressed as
2
d2J z2ez ( d n X (n x 13) iooolo-n.olo\lr\ •
Ul ~lJ.l)
"
I"
dw dfl 47T2 c IJ dt L 1 - n · 13
0
LCL U~ -yy u1a1 un; · ua~ a uu•a,iuu T auring wnicn signincanr acce1-
.
c>r~tinn and th;Jt th~ collision cl- :the> n~rtir],-., vc>lnrih; frnm ~n °
0 0
value cl3 to a final value cf3'. The spectrum of radiation at finite frequencies will
depend on the details of the collision but its form at low freouencies denends
only on the initial and final velocities. In the limit w --'> 0 the exponential factor
in ~D.l) rs equm w unny. 1 nen we rmegrana rs a pertect drtterentral. The spec-
trum of radiation with polarization e is therefore
2
d2J z2ez 13' 13
\
1-'
~ •.. ~() J.~2r * '1 n • f.ll 1
- ,...,
f.l
/1"' ,..,\
\ 0 "J
'
~1- .1/1C',..,\' -1 _1 1. u
' " " ''-OUH \u·~) •o ''-'J cl' WCll i1~
"" auu uv•uo _J
classicallv. To establish the connection to the quantum-mechanical form. we first

convert (15.2) into a spectrum of photons. The energy of a photon of frequency
w is nw. By dividing (15.2) by n2w we therefore obtain the differential number
spectrum per unit energy interval and per unit solid angle of "soft" photons
,-;_ 0 0
~nw V) Ul r lUUll E,
z 2a I 2
d 2N R' R
nm e . (D.J)
hw-->o d(nw) dfl-y 47T2 nw 1 - n . 13' 1-n·l3
710 Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
~ ;
I
X
k k k
' - - y, '
~· 'p~ ~~~·
p
I I
I I I
X X X
Figure 15.1 Quantum-mechanical diagrams describing the scattering of a particle

without photon emission (top) and with the emission of a photon (bottom).
where a = e 2 /he = 11137 is the fine structure constant if e is the proton's charge.
The subscript 'Yon the solid-angle element serves to remind us that it is the solid
angle into which the photon goes. The spectrum (15.3) is to be interpreted as
l'~ll~nw C, th~t tho
.. ;" ~~"oorl hu ~n ·~1 • ,] ~r ~·~
"Yl' 'J r
interaction. Let the cross section for scattering that causes a change in velocity
ell ---" el3' be denoted by duldflP, where p stands for particle. Then the cross
section for scattering and at the same time for producing a soft photon of energy
hw ner umt enemv mterval anrl ner umt solid an~>le IS
3 2
d u [ d N ] du
dflP d(hw) dfly = }~o d(hw) dfly . dflp
(15.4)
~ {1 'i 1) ~~n h" m~dP tn r"l~·;,;ot" >qJlv hv

introducing the energy-momentum 4-vectors of the photon, kfL = (h/e)(w, wn),
..
and of the particle, pfL = Me(')', 'Yil ). It is also valuable to make use of the Lorentz-
. -· _, ""
T
_,3,_11,'U • :. :~ .•,
'J
. .
·~
. ,*
d 3N e2 d 2N ez d2J
- -- (15.5)
(d 3 klk 0 ) hw d(hw) dfl" h(hw? dw dfl"
~· ~ ' r "~ ~-
~UV~H\~J.J),
2
r d3N
--
z2 a E*. p' E*. p
1m 3 - (15.6)
•~o (d klk 0 ) 47T2 k. p' k·p
Wllt:Iet:Ire'
11' pr [In me
~-Vt:lOLUl
gauge, EfL = (0, e)]. That (15.6) emerges from a quantum-mechanical calculation
can be made plausible by considering Fig. 15.1. The upper diagram indicates the
scatterim' nrocess without emission of radiation. The lower three dial'rams have
'-~
'6 "_-::;- "'"~ -!-.~•~ "
'T't
L
.,
-"' -~ 'J.
'T'l .• .•.
cnagrams ontffe lett have the photon emitted by the external lines, that Is, betore
*The fact that w- 2 times d 2 IIdw df! is a Lorentz invariant is not restricted to the limit of w--+ 0. We
find this result useful in some of our later discussions.
Sect. 15.1 Radiation Emitted Dnring Collisions 711
or after the collision· both involve propagators for the particle between the scat-
tering vertex and the photon vertex of the form
1 ~
1
~
(p ± k)z ~ Mz ±2p. k
w ~ v I·" lllC <CU1lll1UUUU11' llUlll UlC'C lWU
HI UlC lllllll
"
diagrams smgular and provide the (hw)-'
lllC'C
m (15.3 ). On the other hand, the dm-

gram on the right has the photon emitted from the interior of the scattering
vertex. Its contribution is finite as w ~ 0 and so is negligible compared to the
first two. The explicit calculation yields (15.4) with (15.6) in the limit that the
energy and the momentum of the photon can be neglected in the kinematics .
.<;;nft :1. Pmi«irm rmh1 frnm thP ,,] linP< in <lnv <lncl io
·,_ J
J '
b' ~ ~;
>L
uw
,1.
·' ·'·
B. Polarization and Spectrum Integrated over Angles

Some limiting forms of (15.2) are of interest. If the particle moves non-
relativistically before and after the collision, then the factors in the denominators
c:m he nut eonlll to 1mitv The rllrliMerl · ~'
..
dzl z z
r!ill NR ~ z e
~--E· I * lll312 (15.7)
<V-->0 dw dfl 47T2C
where lll3 = 13' ~ 13 is the change in velocity in the collision. This is just a dipole
radiation pattern and gives, when summed over polarizations, and integrated
over amdes the total enen1.v radiated oer unit freauencv interval oer nonrelativ-
iotir rnllioinn ,
lim d!NR = 2zzez I ~1312 (15.8)

w--.o dw 37Tc
For relativistic motion in which the change in velocity lll3 is small, (15.2) can be
approximated to lowest order in llB as
')
. a"l z"e" E* • ( Lljj + n X (jj X Llj:i)
hm = - 2- (15.9)
w--.o dw dfl 47T C (1 ~ n · 13? 1
where cl3 is the initial (or average) velocity.

We now consider the explicit forms of the angular distribution of radiation
emrtted wrth a dehmte state ot polanzatron. ln colliSIOn problems rt rs usual that
the direction of the incident narticle is known and the direction of the rllrliMion
is known, but the deflected particle's direction, and consequently that of lll3, are
not known. Consequently the plane containing the incident beam direction and
the direction of the radiation is a natural one with respect to which one specifies
the state of nolarization of the rlldiMion
L".
~ ~· """!
_,. . ._,
"~'"""
• J
• 11 _, "' .. "'
·~
v .
... An •
~~·~~~U~oo ~~
.
UOUo ~ .... W
0 0
mateq pe1 penutclilar 10 llle inciuent uirecnon. rrgure u.L snows tne vecLOriai
relationships. Without loss of generality n, the observation direction, is chosen
in the x-z plane, making an angle (J with the incident beam. The change in velocity
lll3 lies in the x-y plane, making an angle cf> with the x axis. Since the direction
71 '?. rh ont~r 1" '" • • .•. ~ ..r D. o o . D .•••
tlgure l5o2
ot tne scatterea partiCle 1s not ooservea, we w111 average over cf>. The umt vectors
and E.c are polarization vectors parallel and perpendicular to the plane con-
E 11
taining 13 and n.
We leave to Problem 15.6 the demonstration that (15.9) gives the expressions
(averaged over cf>)
~D.lU)
lim dz I'- = zzez I~ lz ______1____
w--.o dw df.l 8'lT 2 c l3 (1 - {3 cos 8?
for the low-frequency limits for the two states of linear polarization. These an-
gular · · 1tinn< are valid for small-:mP"le colli<icm< nf "11 tvnP< ThP nnbri7"-
n '"' ' ~ _, / >2 T >2 T >) >) ' '
LIVHL\V}o ·'"\u'.c u'IIJI\U'.c u'IIJoY ;atu u,ua'a
value of + 1 at cos 8 = {3, and decreases monotonically for larger angles. For
~- 1 · L +L ~- Df/l\ ., 2n21/1 , 4,.4, ~• ' ,. • •
radiation emitted in an arbitrary small-angle collision (~13 small in magnitude

'
a11u jJ"L~
'. -'
tV UL"
. . .
Ull
.
LJ.
.. . .
L'TirT"LauVISIJ:C' lllUllUll lilt:
.
Ul,LH·
.
bution is strongly peaked in the forward direction in the by-now familiar fashion,
with a mean angle of emission of the order of 'Y- 1 = Mc 2 /E. Explicitly, in the
limit 'Y >> 1 we nave
2 2 4 I Aul2 11 4n4'
lim ' 7 (1" 1 1\
w-.o dw d!!
cr>L .l • • £. .> ' o • o • ' o >
elementary mtegratwn rrom ~D.lUJ to De

AT 0 2 2
lim-=--·- '}' 2 1~131 2
w--.o dw 3 'lT c
*Some data for electron bremsstrahlung are given by W. Lichtenberg, A. Przybylski, and M. Scheer,
Phys. Rev. A 11, 480 (1975).
Sect. 15.1 Radiation Emitted Durin2 Collisions 713
For nonrelativistic motion this reduces to (15.8). Since the particle's momentum
is p = yMc~, this result can be written as
. dl 2 z 2e2 2
hm-=--- Q (15.12)
«>->O dw 3rr M 2 c 3
where Q = Ip' - pI is the magnitude of the momentum transfer in the collision.
Equations (15.10) and (15.12) are valid relativistically, as well as nonrelativ-
istically, provided the change in velocity is not too large. For relativistic motion
d. • • •
"'~ <O
lAili ,..-
2 n .-- 'Ju, (1<; 1'l\
I ·r· I
')'
= v '/
This can be seen from (15.2). If the two velocities ~ and W have an angle
ILl~ I!{3 between them of more than 2/y, the two terms in the amplitude will not
. . . . . . .
mLclltac. vv ucu Luc uireuiOn 01 emiSSIOn n IS sucu LlliH one 01 Llleru IS rarge, Llle
other is negligible. The angular distribution will be two searchlight beams, one
centered along ~ and the other along W, each given by the absolute square of
one term. The radiated intensity integrated over angles is then approximately
dl 4z 2 e 2
lim - = - - I n (Q/Mc) (15.14)
«>->0 dw rrc
Q>2Mc
For Q > 2Mc the radiated intensity of soft photons is logarithmically dependent
on Q 2 , in contrast to the linear increase with Q 2 shown by (15.12) for smaller
.&. T:' •1
. . ••
., • "'v U<V
"~'" "~• u~u•~•u.,-.o.,v "'~"~u
"'
always less than the limit of (15.13). The intensity is therefore given by (15.12)
for all momentum transfers.
c. Qualitative Behavior at Finite Frequencies

" &, I. -' •I.
'J
1 &. ,. ,& /1" 1\
\• •• ,.
T<'
time to consider the qualitative behavior of the spectrum at finite frequencies.

The phase factor m(l5.1) controls the behaviOr at timte frequenctes. Appreciable
radiation occurs only when the phase changes relatively little during the collision .
u •• "' h\ .c •• • •1
- . u•v v~~· • \' J ~• uo~ pu• "v'~ '" n u Hvu uo
.( t\
'\'/
.rm -'- (' -llft'\
,~,
Jo •• ,. '
~.,
then :1n:wt from :1 const:1nt the nh:1se of the in· "in (1'i 1) is
rt \
tl>(t) = w( t - n • ) ~(t') dt' )

0
If we imagine that the collision occurs during a time T and that ~ changes rela-
.: , ], thhr frnm itc initi<>l tn lin<>l ']. thP fnr <> ·>hlP r<>cli.
--
. .
atwn IS -- ' -rr
WT~l. 11 • \P IJ -... J. ~LJ.J.J)

where (13) = (1/r) f 13(t) dt is the average value of 13 during the collision. For
nonrelativistic collisions this reduces to
wr< 1
At 1nw frP.nllen,-,if" the nu1iMed · ,;tu i< oiw•n hu (1" 7\ hnt f,.,. ·~ -, 1 ••
oscillating phase factor will cause the integral to be much smaller than when
w - 0. The mtensity Will thus fall rapidly to zero for w > 117. For relativistic
motion the situation is more complex. For small ID.l31 but withy>> 1 the criterion
(15.15) is approximately
WT (1 + ..,lfJl) < 1 (1<; 1t;)

l.y
Now there is angular dependence. For wr < 1, there is significant radiation at all
0 • •
-c llHtl • r Ul WT Ull lllC ~ uu5~, I <-- WT <. 'Y , mere IS appreciaOle radi-
ation onlv out to <~nolP< of the ·-' ·of fl. whPrP A2 1;,.,,.- Fnr •·•~ > ..,,2
' '
(15.16) is not satisfied at any angle. Hence the spectrum of radiation in relativistic
collisrons IS given approximately by(15.11) andl"15.12) provided wr << y 2 , but
modifications occur in the angular distribution as wr approaches y 2 , and the in-
tensity at all angles decreases rapidly for w ;co y 2 /7.
15.2 Bremsstrahlung in Coulomb Collisions

'rha m~et ~~mm ;. ... ,h. ,J.. .c
"' . ·-· ""•o "~
the collision of a fast particle with an atom. Because of its ll"reater chame the
nucleus is more effective at producing deflections of the incident particle than
the electrons. Consequently we ignore the effects of the electrons for the present
and consider the radiation emitted in the collision of a particle of charge ze, mass
M, and initial velocity cl3 with the Coulomb field of a fixed point charge Ze.
The elastic scatterinl! of a chamed narticle bv a static Coulomb field is Piven
by the Rutherford formula (see Section 13.1):
,, 2
dus 2zZe 2 1
~~J.U)
dfl' ' pv ; (2 sin ()' /2) 4
where ()' is the scattering angle of the particle. This cross section is correct non-
rP.htivi<tir<~lhr M <~11 !'lnPlP< <~nd i< trnP nll<~ntl m marh~n;r~1lu f~r tha ral~t;u;et;_,
small-angle scattering of any particle. It is convenient to expr~ss (15.17) as a cross

section for scattering per unit interval in momentum transfer Q. For elastic
""5•
!d 4p Sill~~ tl' /L) - Lp"( 1 cos()') ( 15.11))

With dfl' = d¢' d cos ()' = - Q d¢' dQ/p 2 , integration over azimuth of (15.17)
grves
dus zZeL· 1
dQ = 8'lT (3c-
-
Q3 (15.19)
Sect. 15.2 Bremsstrahlung in Coulomb Collisions 715
In a Coulomb collision with momentum transfer Q the incident particle is

accelerated and emits radiation. From Section 15.1 we know that the angular
distribution is given by (15.10), at least for small deflections, and the integrated
intensity by (15.12). Since the angular distributions have already been discussed,
we focus on the frequency spectrum, integrated over angles. In analogy with
(15.4) we define the differential radiation cross section,
2
d x = dl(w, Q). du, (Q) (
15 20
dw dQ dw dQ · )
where dl( w, Q)ldw is the energy radiated per unit frequency interval in a collision
with momentum transfer Q. The differential radiation cross sectiOn has d1men
sions of (area X energy/frequency X momentum). The cross section for photon
emission per unit energy interval is obtained by dividing by ft 2 w.
The low-frequency radiation spectrum is given by (15.12), provided Q is not
too large. Inserting both (15.12) and (15.19) into (15.20) we obtain
(15.21)
dwdQ
This result is valid at frequencies and momentum transfers low enough to ensure
that the criteria of Section 15.1 are satisfied. The radiation cross section inte-
grated over momentum transfers is
dx
dw
or (15.22)
dx
dw
In summing over momentum transfers we have incorporated the limitations on
the range of validity of (15.21) by means of maximum and minimum values of
Q. At any given frequeney (15.21) describes approximately the differential ra-
diation cross section for only a limited range of Q. Outside that range the cross
section falls below the estimate (15.21) because one or the other of the factors
in (15.20) is much smaller than (15.12) or (15.19) (or zero). This effectively limits
the range of Q and leads to (15.22). Determination of the values of Qmax and
Qmin for different physical Circumstances IS our next task.
A. Class1cal Bremsstrahlung
In our discussiOn of energy loss m Chapter 13 we saw that classical consid
erations were applicable provided
For particles of modest charges this means f3 << 1. In this nonrelativistic limit
the maximum effective momentum transfer is not restricted by failure of (15.12).
The only limit is kinematic. From (15.18) we see that
Qmax 2p 2Jvfu (15.23)
The lower limit on Q is determined classically by the relation between frequency

and collision time that must be satisfied if there is to be significant radiation.
From Section 11.10 and Problem 13.1 we have
l v LZLe
~' !.!
u uu
OV
...
U<Ul
•'-
<U~ ;,:vu ~ ~ ~/T <.0<111 U<;; Wlillt;ll in tt;rlllS 01 1,L aS
2zZe 2 w
Q > Q(c)
mm
= (15.24)
vz
The classical radiation cross section is therefore
2 2 2 2 2
3
dx, = 16 Z e ( z e ) . _!_ . ln ( AMv )
(15.25)
dw 3 c Mc 2 {3 2 zZe 2 w
where A is a number of order unity that takes into account our ignorance of
exactly how the intensity falls to zero around wr = 1. This cross section is mean-
ingful only provided the argument of the logarithm is greater than unity. There
is thus an upper limit w;;;~x on the frequency spectrum. Phrased in terms of a
it i<
~"
'u, 2
li ,,,(c)
m.x
"' (1" ')t:\
\ , "I
'Yl 2
Since TJ is large compared to unity in this classical situation, we find that the range
of photon energies is limited to very soft quanta whose energies are all very small
compared to the kinetic energy of the incident particle. For TJ = 10 the classical
spectrum is shown in Fig. 15.3, with A= 2 (chosen so that for TJ = 1 and w = 0
the classical and quantum-mechanical cross sections agree).
I I
' I I ' I
l"'
r- '' ~
r-, ""' -
~I 3
"<>I 1--\
\
"" ""
Classical Bethe-
1-- '1 10 = Heitler ~
t- -
I I ..J I I I I
0 0.2 0.4 0.6 0.8 1.0
-ffw
E'-----+
r1gure r:>,_, Kamatwn cross sectiOn (energy X area/unit frequency) for nonrelativistic
l'
~·-
· '
op~~••
lS O< 0 f,
uu• •o
•ti~n ~f fr~n"~•wu'
..
lU VCO!J lUW 111
-- '" ito ~f th,
, Ult;
'
~Ul Vt; •uu•MOU

+-
-,
f Dn\
\' .,
newer IS tue quantum-mecnamcal tlorn approxtmatwn result, i.e. (b.29J with A' 1.
Sect. 15.2 Bremsstrahlung in Conlomb Collisions 717
B. Nonrelativistic Bremsstrahlung
Hl UHO
"""' '"" • OJ auu '"" lH UliO. ,_
"'"'" ""'
cons10ereu. A posrenon sucn negwcr was JUSrmea oecause ~D.LO) snows mar tne
spectrum is confined to very low-energy photons. But for fast, though still non-
relativistic, particles with TJ < 1, it is necessary to consider conservation of energy
and momentum including the photon. For scattering by a fixed (or massive) cen-
ter of force, the conservation requirements are
E = E' + liw (15.27)
Q2 = (p - p' - k? = (p - p')2
r2{') u ,_.
• L.
"" ,. 2/') AJ.. rl ""' r
th
·o·
,,. t\,
r
L .£ .
dllU dllCl LllC , nw ctuu K die we """' tsY dllU lLUlll ut we
pnoton, ana 1,J rs tne momentum transterea to tne scattenng center, as Detore.
The reader can verify that the neglect of the photon's momentum kin the second
expression for Q 2 is justified independently of the directions of the momenta
provided the particles are nonrelativistic.
The maximum momentum transfer effective for radiation is again that a!-
lowed by the kinematics. Similarly the minimum effective Q is determined by
the kinematics and not by the collision time.* From the second equation in
~~J.L.t) we OuLam
=
Vrnax p +p ~vn-t- vn nwr
\ LJ.L.O)
Qmin p- p' hw
The second form is obtained by using conservation of energy. The radiation cross
section (15.22) is therefore
dXNR = 16 z2e2 ( z2e2r. I_. In [ A'(v'£ + y E - liw)2l

(15.29)
dw 3 c Mc 2 {3 2
liw
where again A' is a number expected to be of order unitv. Actuallv. with A' = 1
,. <;_'JQ) ic
(1 d. • thP nn~ntmn_mP..-h~ni..-~1 rPcnlt in thP RArn ,
im~t'An firct
/ ·:rr.
uy auu HIOllliOl \~7J-.)· 11110 , Vl w~:o 1
"' voo
as a function of frequency is shown in Fig. 15.3.
The fact that we have obtained the correct quantum-mechanical Born ap-
proximation cross section by semiclassical arguments in which the quantum as-
pects were included only in the kinematics can be understood from the consid-
erations of Section 15.1, especially Fig. 15.1. In the Born approximation the
scattering vertex, drawn as a blob there to indicate complicated things going on,
;!. ' 1!1. tl, .L. ',1 ' 'T'L tL' ;! '
t , ., t· ·~ 1• ·~ ~ ·~~o. ~ ·~ '""~ ~· •o•
me oonom is aosem. unty we exrerna1 tines raaiare; me amptimae is given oy
(15.6); the exact kinematics and phase space conspire to yield (15.29).
ThP r~rli:>tirm ('ff'\<0 CPdinn av/,-/,,, .1. rl, nn thP · •c nf thP
- l. -' . ',_ - '" . ~z 4,\,z - ,_ ' •'-' ' . c -'~
UJ <VH'-'U JU <H'-' UO ~ .(. <W 0 OHVlHHb <HUl <H'-' Vl lUUJUlWH >O
most important for electrons in materials of high atomic number. The total energy
*For soft photons Qmin ~ p- p' can be approximated by Qmin = 2hw/v, while the classical expression
(15.24) is Q::;/n ~ 271hwlv. With 71 < 1, the quantum-mechanical Qmm is larger than the classical and
so governs the lower cutoff in Q. For more energetic photons (p - p') is even larger. In relativistic
.. ~. ' .,, .. . . ..
: ~rmn r '"'
L'CC \'· ··'-'IJ·
lost in radiation by a particle traversing unit thickness of matter containing ]\l

fixed charges Ze (atomic nuclei) per unit volume is
dE
~=N '
J
wm" dxt w)
dw
dx o dw
Using (15.29) for dxldw and converting to the variable of integration x = (fiw!E),
we can write the radiative energy Joss as
(15 30)
dx
The dimensionless integral has the value unity For comparison we write the ratio
of radiative energy loss to collision energy loss (13.14 ):
(15.31)
For nonrelativistic particles (u << c) the radiative loss is completely negligible

compared to the collision loss. The fine structure constant (e 2 /fic = 1/137) enters
characteristically whenever there is emission of radiation as an additional step
beyond the basic process (here the deflection of the particle in the nuclear
Coulomb field). The factor m!M appears because the radiative loss involves the
acceleration of the incident particle, while the collision loss involves the accel-
eration of an electron.
C. Relativistic Bremsstrahlung
For relativistic pm tides the lilllits obtained from conservation of energy must
be modified. The changes are of two sorts. The first is that the maximum effective
Q value is no longer determined by kinematies. It ·.vas shown in Seetion 15.1 that
(15.12) is valid only for Q <2Mc. For larger Q the radiated intensity is logarith-
mic in Q and given by (15.14). Because of the Q- 3 behavior of (15.19) this means
that Qmax in (15.22) is
Qmax = 2Mc (15.32)
'I he second modification rs that the photon's momentum can no longer be rgnored
in determining the minimum momentum transfer from (15.27). The minimum
clearly occurs when all three momenta are parallel:
'~ k
For relatiVIstic motiOn of the partrcle both imtially and finally (even though the
photon may carry off appreciable energy), we can approximate cp E
2 4 2
1\11 c 72E, cp E 1\11 c'72E', where now E and E' are the total energres. I hen
2EE'
With (15 32) and (15 33), the radiation cross section (15 22) becomes
(15.34)
Sect. 15.2 Bremsstrahlung in Coulomb Collisions 719
with the customary A" of order unity. This result is the same as is obtained quan-
tum mechanically in the relativistic limit, provided the photon energy satisfies
' 2 •
dXR)
( dw Born
=
3 e
2 2
16 Z e (z e:)
Me
2
2
( 1 _ hw +
E 4E
2 2
3h ~ ) [ln ( 2££' ) _
Me 2 hw 2
~J (
15 .35 )
e no e m passmg a smce e, e sma c ange in ve oci y

always lies in the plane perpendicular to the incident direction in a relativistic
collision. The angular distribution of the radiation is thus given by (15.11 ). The
doubly differential radiation cross section for energy radiated per unit frequency
(15.36)
where 8 is the angle of emission of the photon and dxR!dw is given by (15.34).
It is instructive to consider the calculation of relativistic bremsstrahlung from

a somew a i eren poin o view. uppos · ·
frame where the force center is at rest we view the rocess as takin lace in the
rest frame K' of the initial particle. The emission process as it appears in the two
frames is indicated schematically in Fig. 15.4. A small-angle deflection in the
2 ' 2 2
where primes denote quantities evaluated in the frame K'. The change in velocity
. .
change of momentum inK'. For a small deflection in the laboratory, Lip' is per-
Ze v Ze
Laboratory
frame
(nucleus at rest) and the frame K' (incident particle essentially at rest).
K'. Its magnitude is the momentum transfer Q. The radiated energy spectrum
can therefore be written as
2
d2r
=~ 2 2
2
(
-Q ) (1 + cos 2 (:J')
dw' dfl! 81r c Me
The triply differential radiation cross section for emission of radiation per unit
freauencv interval oer unit solid angle and oer unit interval in momentum trans-
'" , '1.
v'
t {1" ')()\
\ ·n
2
a··y· ze u a a-
J ' JA' Jn. "2 I H.
• Jn. • ~l -r cos·tt) ~lJ ..)f)
U<V UH U~ UH ~ \"'~. u~
This is the cross section in frame K'. No primes appear on da-,/dQ or on Q 2

because to the extent that Q is transverse, these quantities are obviously invariant
under Lorentz transformations.*
The emission of radiation in the frame K' appears as simple dipole radiation
in (15.37). To obtain the cross section in the laboratory we must make a Lorentz
transformation. In Section 15.1 we saw that (15.5) is a Lorentz-invariant quantity.
With what has just been said about du5 /dQ, it is clear that the equation relating
the differential cross sections in the two frames is
1 d3x 1 d3x'
- = -w' 2 dw' d[!' dQ (15.38)
w2 dw d[! dQ
1 nus tne tnpry muerenua. cross sec non m we raooraLOry rs

,.j3 ,., 2 2 n ,./, 'l
s
I --"=:. I I (1 + cos 2 1l') (1 'i 19)
dw dil du 37T c Me du 167T w'
The quantities in the square brackets must, of course, be expressed in terms of

(unprimed) laboratory quantities. The relativistic Doppler shift formulas are
w = yw'(1 + {3 cos(:)')
and
w' = yw(l - {3 cos (:))
Combining the two equations we obtain
w 1 L')l
./1 n n\ 1 2n2
~' T\i tJ ~"" vI i
' v
and ( D.4U)
lJ D 1 2[J2
fYIS A'
1 +
'
1 - R cos A 'VL(:JL
*Actually, we can use the manifestly invariant 4-momcntum transfer whose square is given by
Q 2 = -(p 1 - p 2 ) 2 = (p 1 - p2 )'-- (E, - £ 2 )'-/c 2 For clastic scattering by a massive center of force,
E 1 = £ 2 , and for small angles and very high energies, the energy difference term can be neglected
even for inelastic collisions.
Sect. 15.3 Screening Effects; Relativistic Radiative Energy Loss 721
The approximations on the right are appropriate for small angles around the
incident direction in the laboratory. With these approximate forms, the square-
bracketed quantity in (15.39) becomes
I 2 ? 4 ,.,
-' w /1 2nl\ J y ~L y v )
~i ~vo v J
L167T ,w' -
27T (1 + ·le2 t -
which is exactly the normalized expression in (15.36). [Use of the exact forms
from (15.40) leads to the sum of the two terms in (15.10).] With the Rutherford
cross section (15.19), or some other collision cross section for dusfdQ, if appro-
. . . . .
pnme, we ootam rrom ~lJ.jY) me remnv1snc oremssrranmng resuns as oerore.
ThP nr.nnlPr ohift l<>o i1111• "" . 1nt p r.f
hw' in K', e;ditted at essentially any angle in th;t fram;, appear in the laboratory
within the forward cone and with energies of the order of hw ~ yhw'. Thus
energetic photons in the laboratory energy range Mc 2 << hw << yMc 2 come
from soft quanta with hw' << Mc 2 in the rest frame of the incident particle.
15.3 Screening Effects; Relativistic Radiative Energy Loss

In the treatment of bremsstrahlung so far we have ignored the effects of the
atomic electrons. As direct contributors to the acceleration of the incident par-
tide thev can be o"f"l" iQ:nored. since their contribution ner atom is of the ·-'·
£?-1 iL _, n . "" c 1. ££ .L
VL ~ J "" vu 0 u "'~"
'"" VH'-'• .LHU lH'-' H<<V '-' UH
screenmg ot the nuclear charge. The ootentlal energy ot the mc1dent oarticle
in the field of the atom can be approximated by the Yukawa form, V(r) =
(zZe 2 /r) exp(-r/a), with a= 1.4a0 Z- 113 • Instead of (15.17) the scattering cross
section is (13.53) with emin given by (13.55). In terms of momentum transfer
(15.19) is replaced by
?
au zze u
157T ~? ,?,? ~lJ.'Il)
·~
U).t \ fJL ~l.t ' l.ts)
where
Q s -- p8(q) - - ~ -me
n Z 113
mm - a - 192 (15.42)
is me momemum rransrer associmea wim me screemng ramus a. I'lote mat m IS

the electronic mass.
The calculation of bremsstrahlung proceeds as at the beginning of Section
15.2, but with the replacement in (15.22),
(Qmax An (Qmax n3An
I :c:::c ~ I
JQ, u JQ, (UL + O:'Y
With the assumption that Qmax is very large compared with both Qmin and Q"
we find that the logarithm in (15.22) is replaced by
I \ I \
In ~max I ~In ( Qmax

I ,..,, r12
Q?
r12\ (15.43)
,~min V l.tmin ' l.ts ~\~min ' ~s}
_
_..,..
~~~ '-"I1UpLef 1J D£.,111SMr , ,..~muu 01 VIrtUal t,lUama, n.HUIHliV" D""' rro.:.,sses -\_>
For Qmin >> Qs the effects of screening are unimportant and the results of the
preceding section are unaffected. But for Qmin :S Q, important modifications
occur.
From (15.23), (15.28), and (15.32) we see that Qmax can be written in all
circumstances as
n 'JU,
,.
/1<: A A\
::;:,max TJ
while from (15.28) and (15.33) we find Qm;n values,

')'I;
nUVR) - n - n' ~
' ' u
(15.45)
Q . -~
( R) _!L{J)_ -~ hw2
mm 2yy'c 2y c
The approximatiOns on the nght are applicable tor soft photons. (Note that, up
to factors of 2 in the logarithms, a universal formula for Qmin for soft photons is
2
Qmin = hwly u.) Since both values of Qmin are proportional tow for soft photons,
;t ;< rlP~r th~t thoro will nlwm<e ho n +.. •r" holm, .1~'-L. ,;,.,, rr.
-, o -, -, ·o -""
important. With Qs given by (15.42), the ratio of Qmin to Qs for nonrelativistic

bremsstrahlung is
,{
\.!'min
"'" CJO'+ nw 1':1L-1Yl p nw
Qs zv3 muc mzir3 (hw)max
where (hw)mex = Mu 2!2. Except for extremely slow speeds, the frequency at
mh;l'h n(NR) < n ;< ~ t;m, c_ ' r>f thP m~· ' H'r>r ,], m;th 1 ()() ]rp '\1
eteCLrons on a goiU LargeL--

~uuu ~,-
•. . .
~"' t'7 ), on1y tor W/Wmax <- u.ll'+ IS screenmg ImponanL.
For particles heavier than electrons the factor M/m makes screening totally in-
significant in nonrelativistic bremsstrahlung.
For ·' · is tic brem: -' "' in o- -~~- . can be· It.
'T'L .< n n·
• u v '""~ ~· ~min w ~s '" "~"
Q~?, 96hw 96M hw

\Ls yymc L ymL ~nW)max
where (hw)max = yMc 2 . The presence of the factory' in the denominator implies
that at sufficiently high energies Q~~J, can be less than Qs for essentially the whole
ranQe of freauencies fif w/w . x. then v' (1 xhl Then the · is
. " _,
O<UU
L
<V Uv -' 'T'L
•y'Uv . • Hv '" 0 ' vHvL " "- "
Lv0 LVL •<v ovL' <Uv uo
E >> E, where the critical energy Es is

l,n,u'
L/~HL
J;' 1Mr 2 (1 <;,J.hl

s mzl13 - '
For energies large compared to £ 5 , Qmin can be neglected compared to Qs in

(15.43) at all freauencies excent the verv tin of the snectrum. The radiation cross
d.
.'
UL ULv
""''
LHHH LO <UUO <H'-' WLU'-'<
2 1
dx 16 Z 2 e 2 z2 e 2 233M'
uW c
I
VYL<-
? In I (15.47)
I'"~
J
Sect. 15.3 Screening Effects; Relativistic Radiative Energy Loss 723
I
I
----
I
1
1.00
tI 10.75
Bethe-Heitler--"'
d)( Figure 15.5 Radiation cross section in
. . .
--=- tu~ ~. ~,-m:IDT. 1 ne
,-u;
constant value is the semiclassical
result. The curve marked "Bcthe-
0 w Heitler" is the ouantum-mcchanical
w-----;.. Born approximation.
The numerical coefficient in the logarithm is subject to some uncertaintv, of

course. Be the and Heitler found a result with 183 instead of 233 in the logarithm
r1 , . , +), -' ',1 (1 l; .ID •n2 2!A 172\ .£ " " ~<'\ -'· ' · - ' - ' '
. Y "J \ I \ • .. J J ·'1-''Y "5 "·
For electrons, E, - 42 MeV in aluminum (Z = 13) and 23 MeV in lead
(Z = 82). The corresponding values for mu mesons are 2 X 106 MeV and 106
MeV. Because of the factor Mlm, screening is important only for electrons. When
E > E, the radiation cross section is given by the constant value (15.47) for all
frequencies. Figure 15.5 shows the radiation cross section (15.47) in the limit of
complete screening, as well as the corresponding Bethe-Heitler result. Their
.- ·r
>+ trP ' thP e1 .,1
-, 7
' -'
"'D""""F "?"
'·'
, '
changes from unity at w = 0 to 0.75 at w = Wmax· For cosmic-ray electrons and

tor electrons rrom most mgn-energy electron accelerators, the bremsstrahlung ts
in the comolete screening limit. Thus the ohoton soectrum shows a tvnical (fiw)-l
behavior.
. ..
rrL
H<v •
~·
v . 0 , '<voo nao
~
~-
··-
tvu <a U<v m. Ill Cldll V I>lliC 1lllllll11
' • • n
15.2.B and was found to be negligible compared to the energy loss bv collisions.
For ultrarelativistic particles, especially electrons, this is no longer true. The ra-
rl:n+,.,n · 1. ' ' ' ,) ' +h 1' ' -· 1 L
r
2r rMc2 In
"bJ b' -l"'r ']
dE
~ = -16 N z2 2 (
_e_ 2
3:._!!_ 1n ( Qmax ) dw
2
dx 3 c Mc o VQ;;,in + Q;
For negligible screening we find approximately
dE
~
dx
=-
16 Z 2e 2 z 2 e 2
3
N -- I
fie
I -
Mc 2
r In (Ay)yMc 2
For higher energies where complete screening occurs this is modified to

2 1
dErad 16 Z 2 e2 z2e2
In 233M ) I 'VMr 2
N {1 'i 4R)
ax j fi.C \MC ,z m
'
'"6 "tuat c " tuc 1
'v" "y•vpv•
11y 110 llle pantcle 'euergy.
1 vc
The comparison of radiative loss to collision loss now becomes

I _,
1
233M 1
dE cad 4 1 Zz 21 m Z 113 m 1
1 \ "
dbcoll 37T IJ/ M ln Hq
'With the Bethe-Heitler energy dependence shown in Fig. 15.5, the coefficient 16/3 is replaced by 4;
-= '"'" , tHO Ul D l ·r uy L\• """TJ.
lct~lUl ~
The value of y for which this ratio is unity depends on the particle and on Z. For
electrons it is y ~ 200 for air and y ~ 20 for lead. At higher energies, the radiative
energy loss is larger than the collision loss and for ultrarelativistic particles is the
dominant loss mechanism.
At energies where the radiative energy loss is dominant, the complete screen-
ing result (15.48) holds. Then it is useful to introduce a unit of length X 0 , called
the radiation length, which is the distance a particle travels while its energy falls
to e- 1 of its initial value. By conservation of energy, we may rewrite (15.48) as
dE E
dx X0
with solution
E(x) = Eoe-x!Xo
where the radiation length X 0 (including quantum corrections, lac. cit.) is
2
r 7(7 1\ 2 I 2 2 /"""u' -l
Y. 4N ~ l~ l r1 <:;liD\
he Me" zwm
the passage of cosmrc-rav or man-made high-enemoartrcles throu!!h matter thP-

,-1: ,.' ], .,. V ' -' .L -' d. ~'
energy ross is governea oy ir, om arso me pro ucuon ot electron-positron Patrs
MlUWt:I.
15.4 Weizsiicker-Williams Method of Virtual Quanta

The emission of bremsstrahlung and other processes involving the electromag-
netic interaction of relativistic particles can be viewed in a way that is very helpful
in providing physical insight into the processes. This point of view is called the
method of virtual quanta. It exploits the similarity between the fields of a rapidly
moving charged particle and the fields of a pulse of radiation (see Section 11.10)
and correlates the effects of the collision of the relativistic charged particle with
some system with the corresponding effects produced by the interaction of ra-
diation (the virtual quanta) with the same system. The method was developed
independently by C. F. Weizsacker and E. J. Williams in 1934. Ten years earlier
Enrico Fermi had used essentially the same idea to relate the energy loss by
ionization to the absorption of x-rays by atoms (see Problem 15.12).
In any given collision we define an "incident particle" and a "struck system."
The perturbing fields of the incident particle are replaced by an equivalent pulse
of radiation that is analyzed into a frequency spectrum of virtual quanta. Then
the effects of the quanta (either scattering or absorption) on the struck system
are calculated. In this way the charged-particle interaction is correlated with the
photon interaction. Table 15.1 lists a few typical correspondences and specifies
the incident particle and struck system. From the table we see that the struck
system is not always the target in the laboratory. For bremsstrahlung the struck
Sect. 15.4 Weizsacker-Williams Method of Virtual Quanta 725
Table 15.1 Correspondences Between Charged Particle Interactions

and Photon Interactions
. .. .
. - •
.
~"
"'""'-"
"'J'"'""'
- .. -
' I lU'-""
•
Umin
ill' "
..
LdiiiUll!\ Ill n UI,;H;U' LUll ,:)1-<tllCOlHl!\ Ul V 11 LUdl
electron (light (light photons of nuclear

particle )-nucleus particle Coulomb field by
collision mass M) the electron (light
particle)
Collisional ionization Incident Atom Photoej ection of a
of atoms (in distant particle atomic electrons by
collisions) virtual quanta
. . ..
~·
~ --==rrurr -y --. ll~!\ldllUU Ul
of nuclei (mass m) nuclei by virtual Larger

quanta of
Production of pions in Electron Nucleus Photoproduction of hlymv
electron-nuclear (mass m) pions by virtual and R
collisions quanta interactions
with nucleus
system rs the lighter of the two collision partners, since its radiation scattering
power is greater. For bremsstrahlung in electron-electron collision it is necessary
from symmetry to take the sum of two contributions where each electron in turn
is the struck system at rest initially in some reference frame.
The chief assumption in the method of virtual quanta is that the effects of
the various frequency components of equivalent radiation add incoherently. This
... .
wrrrue-rrm:-pr
. .
l\.oU~
. . . .
~" , dilU IS ;Lt:lll
With our assumption m :SectiOn D.L.V that the motion ot the particle m the trame
K' was nonrelativistic throughout the collision.
It is convenient in the discussion of the Weizsacker-Williams method to use
the language of impact parameters rather than momentum transfers in order to
make use of results on the Fourier transforms of fields obtained in previous
chapters. The connection between the two approaches is via the uncertainty-
principle relation,
l;
b --
{2
V'V1llTll110 e A Jll Or \Lmax In Ult:lllSSll I ~1:1.'1'1) I , we see rnm rne mimmum

Impact parameter effective m producing radiation is
h h
b.=-=- (15.50)
mm Qmax 2Mv
~c ];ct~rl ;n T~hlo 1 <: 1 Tl.o ' mm ;m~~~t •~t~rc ,rJ;n~ t~ tl.o
"'
~min
_, .£"
H \
~ U\ "
LJ, tJ) UV UVl ll\.o\.oU lV U\.o
_, '- • "T' • r; •
~· ·:: .
lll<O OjJ\.o'-UUHl Vl VlllUUl'{U<llllCl
-'
auTomaticarry incorporaTeslile cuwrr equivarenTTo'1!min-
The snectrum of eauivalent radiation for an indenendent narticle of chan>e
q, velocity P r, passing a struck systemS at impact parameter b, can be found

from the fields of Section 11.1 0:
E 2 (t) q
E,(t)
for f3- I the fields E 2 (t) and B 3 (t) are completely eqmvalent to a pulse of plane-
polarized radiation P 1 incident on S in the x 1 direction, as shown in Fig. 15.6.
There is no magnetic field to accompany E 1 (t) and so form a proper pulse of
radiation P2 incident along the x 2 direction, as shown Nevertheless, if the motion
of the charged particles in S is nonrelativistic in this coordinate frame, we can
add the necessary magnetic field to create the pulse P2 without affecting the
physical problem because the particles in S respond only to electric forces. Even
If the particles m 5 are mftuenced by magnetic forces, the additional magnetic
field implied by replacing E 1 (t) by the radiation pulse P 2 is not important, since
the pulse P 2 will be seen to be of minor importance anyway.
From the discussion Section 14 5 especially equations (14 51), (14 52), and
(14.60), it is evident that the equivalent pulse PI has a frequency spectrum (en-
ergy per unit area per unit frequency interval) d11 (w, b)!dw given by
(15.51a)
where E 2 (w) IS the Founer transform (14.54) of E 2 (t). S1milarly the pulse P 2 has
the frequency spectrum
d12 ( b) (15.51b)
dw w,
The Fourier integrals, calculated in Chapter 13, are given by (13.80). The two
frequency spectra are
2
d11 (w, b) wb _, wb
I I Aj
~2 ~2 (;r ~2
dw
= \yv' z'yv/ , (15.52)
d!I(w, b)
-1 wb) ~~ wb l
dw "2
"'' K o "''
fi ~s pl
~s
/ "
~
/
'
X
q
Figure 15.6 Relativistic charged paiticlc passing the stluek systemS and the
equivalent pulses of radiation.
Sect. 15.4 Weizsacker-Williams Method of Virtual Quanta 727
We note that the intensity of the pulse P 2 involves a factor y- 2 and so is of little
imnortance for ultrarelativistic narticles. The shaves of these soectra are shown
in Fig. 15. 7. The behavior is easily understood if one recalls that the fields of
]JUIM: r 1 <tu:: ueu- -r m lllllt: WHH " WIUlll '-"' uryu. 1 uu' LHe Ht:L[I -,
spectrum will contain all frequencies up to a maximum of order Wmax ~ l!D.t. On

the other hand, the fields of pulse P 2 are similar to one cycle of a sine wave of
•• ,jl, r' . +'· •
;n •+ . _, ~ -+ -~
'"1 'J I ' T -J "Y 'J 'b'
of frequencies centered around -yv!b.

In collision problems we must sum the trequency spectra (D.:JL) over the
' 11< nnccihlP imn~r-t n~r~mPtPr< Thio rrivP< thP PnPr<nT nPr 11nit .frPnllPnr-v
· fl e ld . A s a Jways m sue h pro bl ems we
0
mterval present m the eqmvalent ra d'1atwn

' r ' -r r' , '
musT specny a minimum impacT parameTer Dm;ll' 1ne memoo or vinua1 q1

will be useful only if bmin can be so chosen that for impact parameters greater
than bmin the effects of the incident particle's fields can be represented accurately
by the effects of equivalent pulses of radiation, while for small impact parameters
the effects of the particle's fields can be neglected or taken into account by other
means. Setting aside for the moment how we choose the orooer value of bm; In
general [(15.50) is valid for bremsstrahlung], we can write down the frequency
')Jt:Cll Ulll .
0
idlCU UVt:l r -r JRU ,. '·
a1 a1 1 ( a1 2
;!,,, (w) = 27r ;!,,, w, b) + ;!,,, (w, b) b db (15.53)
·'- "
.L'
~ '" • +.'
•• y ·" D
'
rl D
L'
cr'l- .1+ •
o I \ 2
;cz x [K~(x) -
0
lf_
-(w)=-- ~I 2
K~(x)]
Ul L
xK0(x)K1(x) - (15.54)
dw '71' r v
where
wbmin
x=-- (15.55)
yv
[,
-........--------~- lo=
I
q" ( c
rr2c v r 1
b2
I
-s-
'-;-,-Q
01~
"' \
\:
:
I
I
\J
!\
~
I
I 1
--In
T.
. r
0.001 0.01 0.1 1 10
wb
yv
.
- · .1,;:,,1 rlt:L[Ut:m;y >jJClOlld Ul LUC LWU '-'"!1 r Ul 1
For low frequencies (w << yulhm;,) the energy per unit hequency interval re-
duces to
(15.56)
whereas for high frequencies ( w >> yvlbmm.) the spectrum falls off exponentially
as
(15.57)
Figure 15.8 shows an accurate plot of J(w) (15.54) for v = c, as well as the low-
frequency approximation (15.56). We see that the energy spectrum consists pre-
dominantly of low-frequency quanta with a tail extending up to frequencies of
min·
The number spectrum of virtual quanta N(nw) is obtained by using the
relation
(w) dw liwN(Iiw) d(liw)

dw
Thus the number of virtual quanta per unit energy interval in the low-frequency
limit is
(15.58)
I \ I
3t-
\
\
-
t \
\
2- -
(w) ·,\
q2/7rc
-"'~""
approximation ~
\ .
\
. \
\\ \
\.
\.
0
,I
0.1
I '
0.5
\ I I""--
I 5
Figure 15.8 Frequency spectrum of virtual quanta for a relalivistjc particle, wi•h the
energy per unit frequency d!( w)ldw in units of q 217rc and the frequency in units of
yvlbmin- The number of virtual quanta per unit energy interval is obtained by dividing
b h 2 w.
Sect. 15.5 Bremsstrahlung as the Scattering of Virtual Quanta J.:.'J
The choice of minimum impact parameter bmin must be considered. In brems-

strahlung, bmin = fi/2Mv, where M is the mass of the lighter particle, as already
rl;c~nccPrl Pnr · ;,.,nol ;,.,n;7ot;r>n nf •+- h . n tlw ;~ rorl;>>< m;th
"""
closer impacts treated as collisions between the incident particle and free elec-
trnn• Tn f>lf>rtrnn ni•intf>OTMinn nf nnrlf>i nr f>lf>rtrnn nrnnndinn nf :frnm
nuclei, the wave nature of the particle whose fields provide the virtual quanta
sets the effective minimum imoact parameter. In these circumstances b ~
h!yMv or bmin = R, the nuclear radius, whichever is larger. The values are sum-
marized in Table 15.1.
ThP cm~nt11m-mP.C'h~nir~l * nf thP. yu,;7<~rkP.r-Wil'' <>- ~ of
virtual quanta for ultrarelativistic spin ~ electrons ({3 ~ 1) replaces the square-
bracketed auantitv in (15 58)--"the lo2:arithm"-bv
1 1
E 2 + E' 2 \ 2EE' \ (E + E') 2 E + E'\ E'
L = ( 2£2 I In ( rnc2hw I ~ 4£2 In ( hw
~-
2E
(15.59)
where E and E' = E ~

hw are the initial and final energies of the electron. In
+~ 1-' ;+ ¥, ~...- v'<L"+h
" ,-11. +.
'
1 Ez ,
L = (1 ~ hw!E) In (E!rnc 2 ) ~ 7 + 0 (h;t, (h;t In mr 2 /i,,,)
which is consistent with (15.58) with bmin = c!w =X, the wavelength (divided by
. . . .
L.7TJ 01 we V!nua1 pn~)lon, lite quanwm- ca1 v llllU~ ~ ap-
plication in the so-called two-photon processes in electron-positron collisions.'
15.5 Bremsstrahlung as the Scattering of Virtual Quanta

The emission of bremsstrahlung in a collision between an incident relativistic
1-'a"""'" 01 cua• ge ze <HlU ma" JVl auu au uuueu' ut cuaJ 5 ~ Le can ue
viewed as the scattering of the virtual quanta in the nuclear Coulomb field by
the incident particle in the coordinate system K', where the incident particle is
at rest. 1 ne spectrum ot virtual quanta ar (w' )law IS given Dy ( D.)4) with q ze.
The minimum impact parameter is h!2Mv, so that the frequency spectrum ex-
1enus up w w "f1Y1C In.
The virtual quanta are scattered by the incident particle (the struck system
inK') according to the Thomson cross section (14.125) at low frequencies. Thus,
in the frame K' and for frequencies small compared to Mc"!h, the differential
radiation cross section is approximately
2 2>\ 2
dx' (ze - dl'
,-~, ' riO' = Mr2 I . ~(1 + cos20') . ri. '
/
Since the spectrum of virtual quanta extends up to yMc 2/h, the approximation
(15.56) can be used for dl'(w')!dw' in the region w' << Mc 2/h. Thus the radiation
cross section in K' becomes
1 72 2 I 2 2\
2
rl ' ' '"'"2 \
f.":_ I (1 -1- ~nc 2 1l'\ . ln I (1<; hn\
dw' dO.' 7T c McLJ hw' ' / /
*R. H. Dalitz and D. R. Yennie, Phys. Rev. 105, 1598 (1957).

1H. Terazawa, Rev. Mod. Phys. 45, 615 (1973).
..,.,
'--'"
.....
LJ , 1nemoo 01 v •nua1 vuanta, Kamanve nem rrocesses -lJ
The cross section in the laboratory can be obtained in the same fashion as in
Section 15.2.D. Using (15.38) and the Doppler formulas (15.40) we find
2 I
dx 16 Z 2 e 2 z 2 e2' 2Jv-/Mc 2 \
3·i(1 + ·le•)
1
,/, riO __'2
\
u.2
I
111
"
'\
./1 ' 2n2\
r " ' -
'>
~ . .LL
,~
~
r " '
~D.bl)
This is essentially the same cross section as (15.36). Upon integration over angles
of emission, it yields an expression equal to the soft-photon limit of (15.34).
Equations (15.60) and (15.61) are based on the Thomson scattering cross
section and so are restricted to w' < Mc 2 /n in the rest frame K'. Of course, as
-' A ;n <;:.,,-.t;nn 1 <; ') n
' . . ' -·· _, -·
~l-
r
-~
·~ .... l-''
_,
in Lne .aoorawry. nUL Lne spectrum ot virtual quanta contams trequencies unto
w' = yMc 2/n. For such frequencies the scattering of radiation is not given by the
Thomson cross section, but rather by (14.127) for spinless struck particles or the
Klein-Nishina formula for particles of spin ~- The angular distribution of scat-
tering of such photons is altered from the dipole form of (15.60), as is shown in
Fur. 14.lt>. More Imnortant the tnhl C'rn« fnr rl. ><P< r~n;.-n.,
--r ')
•• j', _L u, H 2uc "A~-'I'L' ' L ~-
'1' ' " , uo uvu. >u.o -Lila!
~uu u~ o~~u \L•-~~VJ·
in the frame K' the bremsstrahlung quanta are confined to a frequency range
0 < w' :'S Mc 2 /n, even though the spectrum of virtual quanta in the nuclear
Coulomb field extends to much higher frequencies. The restricted spectrum in
K' is required physically by conservation of energy, since in the laboratory system
Wll~HO
W yw w" u ucy speCLrum is 1imi1eu LOu<.. w <.. \ ')'lVlC 1 n I· ~ aetaileel
tr~Mm~nt th~ o "'•lor rl;ctr;hnt;nn nf <<'ottc>r;n<T frnm the> !Zl ' "T." l- .
formula yields a brems;trahlung cross section in co;plete agreement with the

Bethe-Heitler formulas (Weizsacker, 1934).
The effects of screening on the bremsstrahlung spectrum can be discussed in
terms of the Weizsacker-Williams method. For a screened Coulomb potential
f}. ,j' 'rf,( ' -1!<!.-1~. 'CCL\'T'L '
~y \'J.JVJ- ... Ul Lilt; HVH> ~
wgarimm IS c:Il_ang<:_Cl_lo a constant as CIIscusseCI m SectiOn 12_._:2.

Further applications of the method of virtual quanta to such problems as
collisional ionization of atoms and electron disintegration of nuclei are deferred
to the problems at the end of the chapter.
1" ,; u. J." ..... Jl ... htorl n . o~ .. n

••o ~~·~
In the process of beta decay an unstable nucleus with atomic number Z trans-
forms spontaneously into another nucleus of atomic number (Z ::':: 1) while emit-
tinP ~~~ ~1 _L-o.el_arui o nc>ntr;nn The> ;c ,.. -"rml ,1' ,11.
. r -, -,
7
,.
/7~1\'
'! ~
'
' /~
\.lJ,U"-)
The energy released in the decay is shared almost entirely by the electron and
the neutrino, with the recoiling nucleus getting a completely negligible share
because of its very large mass. Even without knowledge of why or how beta decay
takes place, we can anticipate that the sudden creation of a rauidlv moving
<'h~r<Tc>rl n~r<' ,], ;n J. • _, 1-.
•'- .c . rl.' _,
·o r T 'J VL
• "'" u
iu Luc iuuuuuc,ion, eh11er we can rmnK or rne e1ecuon mitiauy at rest anel bemg
accelerated violently during a short time interval to its final velocity, or we can
imagine that its charge is suddenly turned on in the same short time interval. The
Sect. 15.6 Radiation Emitted During Beta Decay 731
acceleration and so does not contribute to
at the origin with a constant velocity v = cj3. Then from (15.1) or (15.2) the
(15.63)
from this value at frequencies that violate the condition (15.15). Although it is
difficult to be precise about the value of (13) that appears there, if the formation
process is imagined to involve a velocity-versus-time curve, such as is sketched
in Fig. 15.9, the value of ({3) should not be greater than~- In that case, the criterion
(15.15) is equivalent to wr < 1. The formation time T can be estimated from the
(15.64)
since in the act of beta an electron of total E is created.

This estimate of r implies the frequencies for appreciable radiation are limited
. . .
d2I e2 sin2 0
- - = -- 2
{32 ----~ (15.65)
dw dfl 47T C (1 - {3 COS 0?
while the total intensity per unit frequency interval is
dl e2 1 1 + {3
-(w)=- -ln -2 (15.66)
dw 7TC 1
number of photons per unit energy range given by
t~ Figure 15.9
/j~ Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
It sometimes bears the name inner bremsstrahlung to distinguish it from brcms-

srranmng cminea oy me same oera panicle in passing through matter.
The total energy radiated is approximately
("""" dl ez 1 + (3 I
17 f J 1 ' ~ ~
rad
Jo dw ,~, u~
7Tfie (3 Hl 1 - (3 I "" ].W ~U.OOJ
ror very rasr oeta panicles, me rado or energy going inTo r:IT!iarion to the particle
io
2 e2 2E'
Erad
E 7T lie
...
L
,me 2 1
'
--r ~J..J.U~J
This shows that the radiated energy is a very small fraction of the total energy
rPl -' in hPt" ifpr"" . PVI'n fnr thP mnot P • h~to f Jl
~ ~
2
30me ). Nevertheless, the radiation can be observed and provides useful infor-
mahan tor nuclear phystctsts.
In the actual beta process the energy release is shared by the electron and
the neutrino so mat tne electron has a whole spectrum Of energtes up to some
' "'m Th~n th~ rorl;ot;An ' {1 <; {;{;\ ~ •o+ h _,- • ••
. . . . -, ·~, ~- . ·w
dtstnbutwn of the beta particles. Furthermore, a quantum~mcchamcal treatment
leads to modifkatinn< nP"r thP. Pnil nf th~ -r-' --r
ThM0 "'n
important details for quantitative comparison with experiment. But the origins
of the radiatwn and 1ts semtquanhtahvc description are given adequately by our
classical calculation.
15.7 Radiation Emitted During Orbital-Electron Capture:

Disappearance of Charge and Magnetic Moment
In beta emission the sudden creation of a fast electron gives rise to radiation. In
orbital-electron capture the sudden disappearance of an electron does likewise.
Orbital-electron capture is the process whereby an orbital electron around an
unstable nucleus of atomic number Z is captured by the nucleus, which is trans-
formed into another nucleus with atomic number (Z- 1), with the simultaneous
emission of a neutrino that carries off the excess energy. The process can be
,, . ~ "·
. oym "J uo
Z + e- ~ (Z- 11 + v (15.701
~· nll
"J
~
-'-' ... "Y
... ' -·o """'5J
'£
11
'""'"
no radiation, the spectrum of photons accompanying orbital-electron capture is

Af·oNr~·•t; T ton0n ;n u;nJrl;nN ' ot;A ~• •+ +C
J
nl
~ •
As a simplified model we consider an electron moving in a circular atomic

orbit of radius a with a constant angular velocity w0 . The orbit lies in the x~y
plane, as shown in Fig. 15.10, with the nucleus at the center. The observation
dtrectton n ts defined by the polar angle 0 and lies in the x~z plane. The velocity
· ' +l. .1. .
v(t) = -e 1 w0 a sin(w0 t +a)+ EoWoa cos( wet+ a) (15.71)

Sect. 15.7 Radiation Emitted During Orbital-Electron Capture 733
I \ I
r-" I \ Ze ---....._I
"-= / _/ y
V -e
/
X Figure 15.10
where a is an arbitrary pliase angle. If the electron vamsfies at t 0, the frequency

spectrum of emitted radiation (14.67) is approximately*
(15.72)
dwdfl
where we have assumed that (wale) << 1 (dipole approximation) and put the
retardation factor equal to unity. The integral in (15.72) can be written
ro
(15.73)
where
and EJ., E, are unit polarization vectors perpendicular and parallel to the plane
1
containing nand the z axis. The integrals are elementary and lead to an intensity
distribution,
(15.75)
Since the electron can be captured from any position around the orbit, we av-
erage over the phase angle a. Then the intensity distribution is
(15.76)
The total energy radiated per unit frequency interval is

dl(w)
(15.77)
dw
*To conform to the aclmonition following (14.97), we should multiply tile veloeity (15.71) by a faeto1
such as (I - e'h)8( -t) in order to bring the velocity to zero continuously in a short time T near
t 0. The reader may verif) that in the limit w0 « 1 and WJ « 1 the tesults given ~Jelow eme1ge.
J
while the number of photons per unit energy interval is

\ 2
e2 ' 2 w0 a w2 ( w6 + w2 ) 1
N(hw) = - - (15.78)
37T he c ( w 2 - w~) 2 hw
For w >> w0 the square-bracketed quantity approaches unity. Then the spectrum
is a typical bremsstrahlung spectrum. But for w = w0 the intensity is very large
(infinite in our approximation). The behavior of the photon spectrum is shown
in Fig. 15.11. The singularity at w %may seem alarming, but it is really quite
natural and expected. If the electron were to keep orbitmg forever, the rad1atwn
spectrum would be a sharp line at w = w0 . The sudden termination of the periodic
motion produces a broadening of the spectrum in the neighborhood of the char-
acteristic frequency.
Quantum mechaaically, the radiatioH arises wh ea an I 1 eleetr on (mainly
from the 2p orbit) makes a virtual radiative transition to an I = 0 state, from
which it can be absorbed by the nucleus. Thus the frequency w0 must be identified
with the frequency of the characteristic 2p -'> 1s x-ray, hw0 = (3Z 2 e 2 /8a0 ). Simi-
larly the orbit radius is actually a lnmsitional dipole moment With the estimah~
N(hw) =
3 2
321f
rr
a = a0 /Z, where a0 is the Bohr radius, the photon spectrum (15.78) is
2(ehc2 1
li.w (a/
2
w (w +
2
w6) 2
w6)] (15.79)
The essential characteristics of this spectrum are its strong peaking at the x-ray
energy and its dependence on atomic number as Z 2 •
So far we have considered the radiation that accompanies the disappearance
of the charge of an orbnal electron m the electron-capture process. An electron
possesses a magnetic moment as well as a charge. The disappearance of the
magnetic moment also gives rise to radiation, but with a spectrum of quite dif-
ferent character. The intensity distribution in angle and frequency for a point
magnetic moment in nomcidtivistic motion is given in PIOblcm14.19a. The elec-
tromc magnetic moment can be treated as a constant vector in space until its
I
I
f--
I -
I
I
I
i
1 f-
l
-
"''~"'~ I
f-- -
f--j
1/
I
I
I
I
I
I
\ I
..___
I I
-
"0 &lo
Jiw-
Figure 15.11 Spectrum of photons emitted in orbital-electron capture because of
disappearance of the charge of the electron
~
. ·--
..,, - .. ·- ... .uun -r
,_,_,
""""'
disappearance at t = 0. Then, in the dipole approximation, the appropriate in-

tensity distribution is
w4 2
d21 -
dwdfl
- --
47T2C3
n X IJ. eiwt dt foo
(15.80)
wmcngives
a·J w 2 . z~
'• r
~~J.U~)
n•'
dw dfl 4 7Tz c3 f" '"' "
where El is the angle between IJ. and the observation direction n.

In a semiclassical sense the electronic magnetic moment can be thought of
as having a magnitude f.L = V3(eli/2mc), but being observed only through its
projection 11-z = ±(eli/2mc) on an arbitrary axis. The moment can be thought of
"'-'-ry
-rmn-orraver
.. "P
..... •'-.. . ....
as precessing around the axis at an angle a= tan- 1 \12, so that on the average
.
>J. •t;o
equat w Its av-
.. -·
\JVCrTlilK jJ 1 ~)
erage value ot 3, Independent ot observatiOn directiOn. Thus the angular and
frequency spectrum becomes
d2I
dw dfl
The total energy radiated per unit frequency interval is
=
e2
8-n2c (r
liw
mc 2
(15.82)
di = ~ n~ 1
'
l
' 2
(15.83)
uw L7TC V""
~~•~ " oL
"" ~ ' . c L ' . ~
"!" ==~'-'"'·"5Y """' •m"

e2 liw
N(liw) = 2f:
7T c (me 2?
(15.84)
These spectra are very different in their frequency dependence from a brems-
strahlung spectrum. They increase with increasing frequency, apparently without
limit. Of course, we have been forewarned that our classical results are valid only
in the low-frequency limit. We can imagine that some sort of uncertainty-
jJl "!" . 'uc11 a' wa' u'eu 111 I LJ.O lUI l Ive oeta uecay IIOIU'
here and that conservatiOn of energy, at least, IS guaranteed. Actually, modifi-
~-;;f;nnc ~r;c~h~~~,c~ ~ n~Hlr;nn ;c ohuouc ~m;lt~rl ;n thP ·' -~ontHrP
·-r ~ r
The probability of emission of the neutrino can be shown to depend on the square
of its energy E". When no photon is emitted, the neutrino has the full decay
energy Ev = E 0 • But when a photon of energy liw accompanies it, the neutrino's
energy is reduced to E~ = E 0 - liw. Then the probability of neutrino emission is
reduced by a factor,
I , 2 I li 2
I Ev \ _ 1- ~~ (1 'i R'i\
\Ev "Eo
1 Ill' Ul<O<lHO lllaL UUI ra ~LJ.Cc:>J anu ~tJ.C'+) mu'L ue currecLeu oy
muTtipTicatwn w1lli (15.85) to take mto account the kmematics of the neutrino
;"n TJ;"; m~ rlc•cc;~.,~ntnn llm io
e2
N(liw) = - -
liw
27Tiic (mc 2 ) 2 ( r liw
1 - -
E0
(15.86)
I
.'
~ow-frequency
I- -
I limit
I
I- I -
t / / ~
i/
·'·
lVfnW}
-
I
I / "' -
I
1- I. -
I I
nw l!.Q
Figure 15.12 Spectrum of photons emitted in orbital-electron capture because of

disappearance of the magnetic moment of the electron.
This is essentially the correct quantum-mechanical result. A comparison of the

' '' '',ntinn (1 'i Rh\ ~nrl thP rh«ir~l ~nP {1 <; RA\ ;" ehmun ;n H;n 1 <; 1 "'!
.. , ' , ' :/ ' , ·o· , ' '
'u~ lllY lllC UCUlllUV- , crucra1 m oorammg Ine proper Hl j-11
behavior of the photon energy spectrum. For the customary bremsstrahlung spec-
tra such correction factors are less important because the bulk of the radiation
is emitted in photons with energies much smaller than the maximum allowable
value.
The total radiation emitted in orbital-electron capture is the sum of the con-
tr;h,t;~no fr~m thP r1;oo >nrP ~f f~ ' ,1, rl ,f <J..
0
'•e-rr
n vc11av;v,, v1 \,LJ./:7) anu \l.J.COJ we see Lnm Ine
"· lllC
upper end of the spectrum will be dominated by the magnetic-moment contri-
N(liw)
t
I
I
1\ , ___ - - ~
/ -........
\/
~
A
/ \
/
I, ' -.... ----
•u "'0
Fieure 15.13 Typrcal photon spectrum for radiative orbital-electron caoture with
energy release Eo, showing the contributions from the disappearance of the electronic
charge and magnetic moment.
Ch. 15 Problems 737
bution unless the energy release is very small, whereas the lower end of the
spectrum will be dominated by the clcctric-chargc term, especially for high z.
tumre l.'J.U shows a tvmcal combmed soectrum tor z LU-JU. ObservatiOns
on a number of nuclei confirm the general features of the spectra and allow
determination of the energy release £ 0 •

Classical bremsstrahluno- is discussed brieflv bv
Landau and Lifshitz, Classical Theory of Fields, Section 9.4
-· .
A semiclassical discussion analogous to ours, but using impact parameters throughout and
. '
"""KUSSr,
':n"
. " .
. L.lL
Bremsstrahlung can be described accurately only by a proper quantum-mechanical

treatment. The standard reference is
TT o 0
The method of virtual quanta (Weizsacker-Williams method) has only one proper
reference, the classic article by
Williams
Short discussions appear in
Heitler, Appendix 6
~ .. _,_ .<.. " . ,.
••. ' £. .£ d.
". ,.
" "
Terazawa (op. cit.)
The work of Fermi that predates by 10 years the Weizsacker-Williams treatment is
E. Fermi, Z. Phys. 29, 315 (1924).
"""Tirclimirlll:imrs or J:<ermi s approacn to energy Joss were c1arinea rnrougn me nse or qnan-
tum mechanics by
E. J. Williams, Proc. R. Soc. London Al39, 163 (1933).
.., .. 'i' ·' .. ' . ' "" .. 'l' w "5 -
pari sons wlnr experimemin some cases, aretlfoseoy
C. S. W. Chang and D. L. Falkoff, Phys. Rev., 76,365 (1949).
P. C. Martin and R. J. Glauber, Phys. Rev., 109, 1307 (1958).
-siegrrann, Lnapter 201. ( 111) oy L. :-.. wu.
An important topic not mentioned so far is the production of coherent bremsstrah-
lnng beams from high-energy electrons on crystals. Under proper conditions roughly
monochromatic, highly polarized photon beams can be produced. A review of the theory
arur experimem rs grvenoy
G. Diambrini Palazzi, Rev. Mod. Phys. 40, 611 (1968).
15.1 In radiative collision problems it is useful to have the radiation amplitude ex-
..
-prt»cu -asarr ""lTIVUIVIIIg'Ie or rn e f'"rricJes, as in
(14.65). for examole. In the nonrelativistic limit oarticles do not move raoidlv or
far during the period of acceleration; only the lowest order velocity and retardation
effects need be kept in an approximate description.
I ,jlj Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
(a) Show that the integral (times e) in (14.65), which is, apart from an inessential
puao._, CH UW'-0 UHO ~ UUU'-> U >W U lilt: !U- VV lCl'llCll "~vvCU~ Tie](!
at distance R, can be expanded in inverse powers of c (remembering that
R nfr onrl lr ,,,{,·\ "' fnllnw<·
ur
where !3: = P · n, r = r · n, and

1 11
P~ = (n X P) X n. Neglected terms are of
oroer tic ana mgner.
(b) Show that the first term in part a corresponds to the electric dipole approx-
tmatwn, while the next terms are the magnetJcclipme allif the electric quad-
rnnniP rnntri . L'<nmP in .. '"-h .. ~.,rtc
r
.,.a ra~n;ra..1
. ..,
m'tJ.
vergence factor e-<lrl to give meaning to the integrals, as discussed following

(14.67).] For a group of charges, show that the generalization of part a can
be written as
J = -1 J dt eiwr
,,(d2p~ (t) + d·2m(t)
2
X n
3
+ -1 --~
d Q
3
(n, t) + · · ·. )
c df dt 6c dt
and the radiated intensitv oer unit solid an~>le and ner unit frenuencv "'
uw au '+7TT J \ ur w oc ar· ;
wnere
Qa~ = 2: qj(3xj,.xj~ - rf 8a~)

1
and the vector Q(n) has components, Q" = 2-~Q.~n~. Relate to the treatment
of multipole radiation in Sections 9.2 and 9.3 and Problem 9.7.
15.2 A nonrelativistic particle of charge e and mass m collides with a fixed, smooth,
hard sphere of radius R. Assuming that the collision is elastic, show that in the
dipole approximation (neglecting retardation effects) the classical differential
cross section for the emission of photons per unit solid angle per unit energy
interval is
2 22f: 2
d a R e I _v l 1 · (2 + 3 sin 2 tll
where tits measured relative to the incident direction. Sketch the angular distri-
bution. Integrate over angles to get the total bremsstrahlung cross section. Qual-
. . .
·•y,
'
>HUC LQCCU>
.
\U~ -,
. .
t;VV'-''"~--upper--um:rr LU lllC
spectrum?
15.3 Treat Problem 15.2 without the assumption of nonrelativistic motion, using (15.2)
and assuming the elastic impact is of negligible duration. Show that the cross sec-
tion for photon emission is now
Ch. 15 Problems 739
15.4 A g1oup of charged pmticles with charges ej and coordinates tj(t) undergo inter-
actions and are accelerated only during a time -7/2 < t < 7/2, during which their
velocities change from cJ3 to cJ3'.
(a) Show that for w7 << I the intensity of radiation emitted with polarization e
per unit solid aHgle afl d uHit frequeHey iHten al is
E = 2.:,. ei(-1_13_::;__ - Iii )e-,wn·<,(O)!c

- n· Iii I - n· 131
(b) An w0 meson of mass 784 MeV decays into 1T+1T- and e 'e- with branching
ratios of 1.3 x 10 2 and 8 X 10--s, respectively. Show that for both decay
modes the frequency spectrum of radiated energy at low frequencies is
dl
dw
= £1TG [C +
{3
/32) In (~)
1 {3
- 2] = 4e2
1TG
[In (M,")
m
-!]2
where M," is the mass of the w0 meson and m is the mass of one of the decay
products. Evaluate approximately the total energy radiate(} in each decay by
integrating the spectrum up to the maximum frequency allowed kinemati-
cally. \Vhat fraction of the rest energy of the aP is it in each decay?
15.5 A situation closely related to that of Problem 15.4b is the emission of radiation
caused by the disappearance of charges and magnetic moments in the annihilation
of electrons and positrons to form hadrons in high-energy storage ring experi-
ments. If the differeHtial cross sectioH for the process e+ e- ' hadrons is dtJ'0 ,
without the emission of photons, calculate the cross section for the same process
accompamed by a soft photon (hw ~ 0) 10 the energy mterval d(hw) around hw.
Compare your results with the quantum-mechanical expressions:
d!J. d(hw)
+ h w
2
s' + 2m 2
2
[1 (1 + 131
- l n - - - 1 ]}
Bl - B
where s = (p 1 + p 2 ?, s' = (p 1 + p 2 - kj2, f3 = electron vic in c.m. frame. Neglect
the emJssJon of radiation by any of the hadrons, all assumed to be much heavier
than the electrons.
The factors proportional to ,,)2 in the numerators of these expressions can be
attributed to the disappearance of the magnetic moments. If you have not included
such con!Iibutions in yow semiclassical calculation, you may consider doing so.
15.6 For the soft-photon limit of radiation emitted when there is a small change a(3 of
velocity (15.9) applies, with convenient polarization vectors shown in Fig !52
(a) Show that
El. [ali+ n X (13 X aJ3)] = laJ3I(J3- cos fJ) cos</>
E • [3.13 + n X (13 X 3.13)] IZS.131 (I J3 cos e) Slll > 1 and e << 1
~o dw dft 2
27T C (1 + y 2 02 ) 1 (1 + y 2 ti") 2
where the first (second) term in square brackets corresponds to the parallel
(perpendicular) polarization. This expression leads immediately to (15.11).
(c) Show that the res all of part b gives the exp1 essiou for 1'(0) given followmg
(15.10).
(d) Show that the angular integral of (15.11) or the answer in part b can be
written
. d!
inn
~o dw 7TC
. JI y4
(2 2y + y')
3 7TC
13.7 Consider the radiation emitted in nuclear fission by tile sudden creation of two
fragments of charge and mass (Z 1e, A 1m) and (Z2 e, A 2 m) recoiling in opposite
directions with total c.m.s. kinetic energy E. Treat the nuclei as point charges and
their motion after the very short initial period of acceleration is nonrelativistic,
bat keep terms ap to second mde1 in lie, as in Problem 15.1. For simplicity, assume
that the fragments move with constant speeds in opposite directions away from
the origin for t > 0. The relative speed is c{3.
(a) Using the appropriate generalization of Problem 15.1a, show that the in ten-
sity of radiation per unit solid angle and per unit photon energy in the c.m.
2
d(hw) dfl 4~
where e is the angle between the line of recoil and the direction of obser-
vation, and
n q
A,+ A,
Show that the radiated energy per unit photon energy is
dl
d(hw)
where the first term is the eleetrie dipole and the second the quadmpole
radiation.
(b) As an example of the asymmetric fission of 235 U by thermal neutrons, take
Z 1 = 36, A 1 = 95 (krypton), Z 2 = 56, A 2 = 138 (barium) (three neutrons are
emitted during fission), withE- 170 MeV and mc2 - 931.5 MeV. What are
the values of p 2 and q 2 ? Determine the total amount of energy (in MeV)
radiated by this "inner bremsstrahlung" pwcess, with the substitution, {3 2 =>
{3 2 (1 - hw!E), as a crude way to incorporate conservation of energy. What
are the relative amounts of energy radiated in the dipole and quadrupole
modes? In actual fission, toughly 7 fvfeV of electr mnagnetic energy is radi-
8
ated w1thm 10 s. How does your esllmate compare'? Il1t 1s much smaller
or larger, attempt to explain
15.8 Two identical point particles of charge q and mass m interact bv means of a short-
range repulsive interaction that is to a hard sphere of radius R in their
eqnhmlept
relative separation. Neglecting the electromagnetic interaction between the two
parucles, determme the rad1at10n cross sectwn m the center-of -mass system for a
rh 1<: n . .... 7&1
rnllisinn L thP"' i..1Pntiro 1 '·]p, tn thP L i'hinn " \0.
tion. Show that the differential cross section for emission of photons pe; ~nit solid
() [ ( ) ( r(
angle per unit energy interval is
q2 f3 2 R 2 1
d'-u
d(hw) dO
= -
he 481T hw
2
5
wR
· - - · - {32 - + 3 sin 2 11 cos 2 11 + -
c
3
1 - :_ sin 4 11
8 )J
where IJ is measured relative to the incident direction and cf3 is the relative speed.
By integration over the angles of emission, show that the total cross section for
.auiation per uui, puo,on energy nw is
q~~ ·f3- + ( ~~~2

2 2
_ f-
-du- - 1
R- · - {32
U\ftW) ,nc; .lJ nw \ c I
Compare these results with that of Problem 15.2 as to frequency dependence,
. ' .
•~w•n~
.
nuu~, ~cv.
15.9 A particle of charge ze, mass m, and nonrelativistic velocity u is deflected in a

1ed C:oulomh fielrl V( r) 77P 2 P-a'lr ~nrl con• 1tlv emits '"'' ion. Dis-
cuss the radiation with the approximation that the particle moves in an almost
straight-line trajectory past the force center.
(a) Show that, if the impact parameter is b, the energy radiated per unit fre-
quency interval is
? ?
aJ (w, b)= <; ~

dw 31T c
e (;mce;) I~)u a?KT(ab)
2
tor w << u!b, and negligible for w >> ulb.

(b) Show that the radiation cross section is
= _16Z e (z e\(~J -x2 K6(x) - KT(x) + 2Ko(x)K,(x) '

2 2 2
dx(w)
2
dw 3 c mc v 2 X x,
where X1 = abmim Xz = abmax·
(c) With bmin = h!mu, bmox = vlw, and a- 1 = 1.4a 0 z-w, determine the radiation
cross sec,ion in Llle LWO .imi,s, x 2 '--'-- • anu x 2 __.,__., 1. ~~ompare your resuns
with the "screening" and "no screening" limits of the text.
1<:;_111 A nnrtirlP of rhnrnP 7P mn" m nn.-1 "Pinrih; n is ~ .. n •d in • h"nc•rholir nnlh
by a fixed repulsive Co~lomb po~ential, V(r) = Zze /r. Assume the ';onrelati~istic 2
dipole approximation (but no further approximations).

(a) Show that the energy radiated per unit frequency interval by the particle
when initially incident at impact parameter b is
., -
c -1
-d l( w, b) = -8 (zeaw)" e-(~wlw,,) K'iwlw.,(
WE
+ Kiw!w 0 ( : )
dw 31T c' 0
Wo E2
(b) Show that the radiation cross section is

d .J .. ,\ 16 (zeav )2 o" (rrwlwo) (J) 1< ( w_ \ (_w\
dw . ' 3 c' Wv
IW Wo
Wv/ "'
tllJt(<Jo\ ~
Wo
(c) Prove that the radiation cross section reduces to that obtained in the text for
c1ass1ca1 bremsstrahlung tor w << Wv· What IS the limiting form for w >> w0 ?
(d) What modifications occur for an attractive Coulomb interaction?
1ne nyperoouc patn may oe aescnoea oy
x = a(E + cosh t), y = -b sinh t. w0 t = (t + E sinh t)
where a = Zze'-!mv 2 , E = V 1 + (b!a)'-, ""' = via.
15.11 Using the method of virtual quanta, discuss the relationship between the cross
section for photodisintegration of a nucleus and electrodisintegration of a nucleus.
(a) Show that, for electrons of energy E = ymc 2 >> mc 2 , the electron disinte-
grat10n cross sectiOn is approximately:
Uet(E) =-
2-he2JF:i' Uphutu(w) In
(k h2me2) -d w
y
7TCuJT W W
where hwT is the threshold energy for the process and k is a constant of order
unity.
(b) Assuming that upb,,,(w) has the resonance shape:
A e2 f
uphoto(w) = 27T Me (w- %)2 + (fl2f
where the width r is small compared to ( Wo - wT ), sketch the behavior of
ue 1(E) as a function of E and show that forE>> hw0 ,
(c) In the limit of a very narrow resonance, the photonuclear cross section can
be written as uphoto( w) = (Ae 2 /Mc) o( w- w0 ). Then the result of part b would
represent the electrodisintegration cross section for E > hw0 • The corre-
sponding bremsstrahlung-induced cross section is given jn the same approx-
imation by (15.47), multiplied by (Ae 2h'ddiwo), where Z is the atomic number
of the rad~ator. Compansons of the electron- and bremsstrahlung-mduced
disintegration cross sections of a number of nuclei are given by E. Wolynec
et a!. Phys. Rev. C 11, I 083 ( 1975). Calculate the quantity called F as a func-
tion of E (with a giant dipole resonance energy hw0 = 20 MeV) and compare
its magnitude and energy dependence (at the high energy end) with the data
in Figures 1-5 of Wolynec et a!. The comparison is only qualitative at E =
hw0 because of the breadth of the dipole resonance. [Pis the ratio of the
bremsstrahlung-induced cross section in units of Z 2d to the electrodisinte-
gration cross section.]
15.12 A fast particle of charge ze, mass M, and velocity u, collides with a hydrogen-like
atom with one electron of charge -e, mass m, bound to a nuclear center of charge
Ze. The collisions can be divided into two kinds: close collisions where the particle
passes through the atom (b < d), and distant collisions where the particle passes
by outside the atom (IJ > d). The atomic "radius" d can be taken as a 0 /Z. For the
close collisions the interaction of the incident patticle and the electron can be
treated as a two-body collision and the energy transfer calculated from the Ruth-
erford cross section. For the distant collisions the excitation and ionization of the
atom can be considered the result of the photoelectric effect by the virtual quanta
of the incident particle's fields.
For simplicity assume that for photon energies Q greater than the ionization
potential I the pllotoeleetrie cross seetion is
O'y(Q) =
8w" a0
137 (z )2( I
Q
)3
(This obeys the empirical Z 4 A3 law for x-ray absorption and has a coefficient ad-
justed to satisfy the dipole sum rule,] u (Q) dQ 27fie 2 hlmc.)
Ch.lS Problems 743
(a) Calculate the differential cross sections daldQ for energy transfer Q for close
and ~' collisions (write them as fnn~tirms of Oil as far as · lP and
in units of 21Tz e /mv P). Plot the two distributions for Q!I > 1 for non-
2 4 2
. . .+ •'- -~ ·~1. ~ 1 2 "'''

~ 2
(b) Show that the number of distant collisions measured by the integrated cross
~h 1 +h th ·'- ,f .J, ,n; · h th ,, th
·o· ' -o,
transfer per collision is much smaller. Show that the energy loss is divided
approximately equally between the two kinds of collisions, and verify that
your total energy loss IS in essential agreement With Bethe's result (U.l4) .
•.,.J W UHO UvlOaJ <H a f'' <H '""a H>U auu a UlOUUWV a•lO '-''-d<lOU. "''-
total kinetic energy available is (m~- m'")c 2 = 34 MeV. The mu meson has a
kinetic energy of 4.1 MeV. Determine the number of quanta emitted per unit
energy interval because of the sudden creation of the moving mu meson. Assuming
that the photons are emitted perpendicular to the direction of motion of the mu
meson (actually it is a sin 2 11 distribution), show that the maximum photon energy
is 17 MeV. Find how many quanta are emitted with energies greater than one-
tenth ot tt1e maxtmum, ana compare your result w1tn tne ooservea ratio or radi-
ative pi-mu decays. [1.24 ± 0.25 X w-• for muons with kinetic energy less than
i 4 MPV -""" "kn H Prim"knff Phm· RPv !l4 1?~~ 119~1\1
15.14 In internal conversion, the nucleus makes a transition from one state to another
and an orbital electron is ejected. The electron has a kinetic energy equal to the
transition energy minus its binding energy. For a conversion line of 1 MeV deter-
. +'- ,].. ·' . • ... .+tl. ~~- . .
of the electron. Wha~-fraction of tl~:electrons ;ill have energies less than -99% of
the total energy? Will this low-energy tail on the conversion line be experimentally
observable?
-'-"•-'-" VllC Ul lllC UC~ay Ul a I\. 1' lllC llHCC-JllUH UClOilJ, I\. ~'II 'II 'II •
The energy release is 75 MeV, small enough that the pions can be treated non-
relativistically in rough approximation.
(a) Show that the differential spectrum of radiated intensity at low frequencies
in the K meson rest frame is approximately
dwd!i ~c m c2 ~
. "" "
. . f h h 1 _'bJ h J: f h . . 'b'
emiSSIOn o t e p oton re attve to t e momentum o t e negative pwn.
(b) Estimate the branching ratio for emission of a photon of energy greater than
.:1 relative to the nonradiative three-pion decay. What is its numerical value
. .
wr u liVtev. 1u IVtev. ~Otupare Wlw expenmem \ -,:_ ,.._ 1u tOr u 11
MeV) .
•,,.U VH<O Ul LH<O U<OlOaJ v• LH<O lOHaliO,<OU n. \'"K +7J. '"'-'} >0 •~
0
1r+ 1r(Mrr = 139.6 MeV, Mrro = 135.0 MeV). Inner bremsstrahlung is emitted by
the creation of the positive pion. A study of this radiative decay mode was made
by Edwards et a!. [Phys. Rev. DS. 2720 (1972)].
(aJ Calculate the classical aistnoutwn m angle ana rrequency or sort pnotons
and compare with the data of Fig. 6 of Edwards et a!. Compute the classical
<ii· "1"' fnr R 0 71 rnrrtesnnn<iin" to a ~hor<Jte.-1 ninn nf kin~etic
energy 58 MeV, and compare.
(b) Estimate the number of radiative decays for charged pion kinetic energies
on the interval, 55 MeV :S Trr :S 90 MeV, as a fraction of all K' decays (the
1T+1To decay mode is 21% of all decays). You can treat the kinematics, in-
cluding the photon, correctly, or you can approximate reality with an ideal-
ization that has the neutral pion alnray~ urith the ~arne momentum and th€
photon and the charged pion with parallel momenta (see part a for justifi-
cation of this assumption). This idealization permits you to correlate directly
the limits on the charged pion's kinetic energy with that of the photon. Com-
pare your estimate with the experimental >alue for the branching ratio fm
0 4
1T' 1r y (wnh the hn11tcd range of 1T energies) of (2. 75 ± 0.15) X 10 .
CHAPTER 16
Radiation Damping, Classical Models

of Charged Particles
16.1 Introductory Considerations

In the preceding chapters the problems of electrodynamics have been divided
into two classes: one in which the sources of charge and current are specified and
the resulting electromagnetic fields are calculated. and the other in which the
external electromagnetic fields are specified and the motions of charged particles
or currents are calculated. Waveguides, cavities, and radiation from prescribed
multipole sources are examples of the first type of problem, while motion of
charges in electric and magnetic fields and energy-loss phenomena are examples
of the second type. Occasionally, as in the discussion of bremsstrahlung, the two
problerns are contbined. But the treatrnent is a stepwise one first the nrotion of
the charged particle in an external field is determined, neglecting the emission
of radiation; then the radiation is calculated from the trajectory as a given source
distribution.
It is evident that this manner of handling problems in electrodynamics can
be of only approximate validity. The motion of charged particles in external force
fields necessarily involves the emission of radiation whenever the charges are
accelerated. The emitted radiation carries off energy, momentum, and angular
momentum and so must influence the subsequent mot ton of the charged particles.
Consequently the motion of the sources of radiation is determined, in part, by
the manner of emiSSIOn of the radtatwn. A correct treatment must mclude the
reaction of the radiation on the motion of the sources.
Why is it that we have taken so long in our discussion of electrodynamics to
face this fact? Why is it that many answers calculated in an apparenlly erroneous
way agree so well with experiment? A partial answer to the first question lies in
the second. There are very many problems in electrodynamics that can be put
with negligible error into one of the two categories described in the first para-
graph. Hence it is worthwhile discussing them without the added and unnecessary
complication of including reaction effects. The remaining answer to the first quos-
tion is that a completely satisfactory classical treatment of the reactive effects of
radiation does not exist. The difficulties presented by this problem touch one of
the liiOSt fundamental aspects of physics, the natme of an elelllentaty patticle.
Although partial solutions, workable within limited areas, can be given, the basic
problem remains unsolved.
In quantum mechanics, the situation at first appeared worse, but develop-
ment of the renormalization program of quantum field theory in the 1950s led to
a consistent theoretical description of electrodynamics (called quantum electro-
745
746 Chapter 16 Radiation Damping, Classical Models of Charged Particles-G
dynamics or QED, the interaction of clectrom and positrons •.vith elcctromag

netic fields) in terms of observed quantities such as mass and static charge. A
weak-coupling theory (a = 1/137), QED has proven remarkably successful in
explaining to amazing accuracy the tiny radiative corrections observed in preci-
sion atomic experiments (Lamb shift, anomalous magnetic moments. etc.) by
calculating to higher and higher orders in perturbation theory. More recently,
the success has been extended to weak and strong interactions as well within the
standard model, sketched briefly at the beginning of the Introduction. Unfortu-
nately, the strong interactions are not really amenable to accurate calculations
vra perturbation theory.
In this chapter we address only some of the classical aspects of radiation
reaction.
The question of why many problems can apparently be handled neglecting
reactive effects of the radiation has the obvrous answer that such eiiects must be
of negligible importance. To sec qualitatively when this is so, and to obtain
semiquantitative estimates of the ranges of parameters where radiative effects
arc or arc not important, we need a simple criterion. One such criterion can be
obtained from energy considerations. If an external fm ce field causes a par tide
of charge e to have an acceleratiOn of typrcal magmtude a for a period of time
T, the energy radiated is of the order of
from the Larmor formula (14.22). If this energy lost in radiation is negligible
compared to the relevant energy E 0 of the problem, we can expect that radiative
effects will be unimportant. But If E,aa > ~.the effects of radiation reaction will
be appreciable. The criterion for the regime where radiative effects are ttnim-
portant can thus be expressed by
(16.2)
The specification of the relevant energy Eo demands a little care We distin-
guish two apparently different situations, one in which the particle is initially at
rest and is acted on by the applied force only for the finite interval T, and one
where the particle undergoes continual acceleration, e.g., in quasiperiodic motion
at some charactenstrc frequency w0 • For the part1cle at rest rmtrally, a typ1cal
energy is evidently its kinetic energy after the period of acceleration. Thus
E 0 ~ m(aT) 2
The criterion (16.2) for the unimpm tancc of 1 adiative effects then becomes
It is useful to define the characteristic time in this relation as
(16.3)
Sect. 16.2 Radiative Reaction Force from Conservation of Ener2v 747
Then the conclusion is that for time T long compared to r radiative effects are
unimportant. Only when the force is applied so suddenly and for such a short
time that T ~ r will radiative effects modify the motion appreciably. It is useful
to note that the longest characteristic time r for charged particles is for electrons
aJ:lllluat 1ts vame IS r o.26 X w-L· s. This is of the ~thP. timP. fnr
15
light to travel 10 m. Only for phenomena involving such distances or times
will we expect radiative effects to play a crucial role.
If the motion of the charged particle is quasiperiodic with a typical amplitude
d and characteristic frequency w0 , the mechanical energy of motion can be iden-
titiecl with V. ~nrl io nf thP nrrlPr ,J'
Eo mw6d 2
The accelerations are typically a ~ w6d, and the time interval T ~ (llw0 ). Con-
sequently criterion (16.2) is
L.C w0a
"--'-- mw 0 u
"J~ 3 wo
or (16.4)
W0T << 1
w_,._ere r is given uy ~i0.-'_1. ~nee Wn IS a t1me appnmriate to the mPrh:mical
motion, again we see that, if the relevant mechanical time interval is long com-
pared to the characteristic time r (16.3), radiative reaction effects on the motion
will be unimportant.
The examples of the last two paragraphs show that the reactive effects of
radiation On the motion Of~ ch• <lPrl . -l r~n hP t • l ,t; f
__tl ol£ ...1~ :; <' < ' ','
, ~·v u~v. . . . . ~, . . ul'l'''-'dctuq Ill Limes Ot tHe
order of r or over distances of the order of cr. This is a general criterion within
the framework of classical electrodynamics. For motions less violent, the reactive
effects are small enough to have a negligible effect on the short-term motion.
~lr.nn_tMm rnmnl·>t' .n. • ,], ' ' ·
.' . Ul'l''
•v~•>< " ' UH
"""
way, a> We see JmmeUiaLely.
16.2 Radiative Reaction Force from Conservation of Energy

-':.."':':_'::""",,iun now arises as _to _ll_OW w mcmae tne reactive enects ot radiation in
theeou~tinn•·nfmntinnfnr~ "~~rti~!... 'X/,.1- · ·n .· , :1. -1 · ;•:.:
1
argument based on conservation ;f energy for a no~;~lativistic charged particl~.
A more fundamental derivation and the incorporation of relativistic effects are
deferred to later sections.
If the emission of radiation i' n · ' --' ~ r.h~ruP.cl · n]p r.J' m occ ,]
.ch ,] • l ,]. "' ~ -1: •

~J 'LeX(H<~oVU U~~~~ n) "'"'nc;VVlUll c;qua-
t!On 0 mouon:
mv = Fext (16.5)
Since the particle is accelerated it rarlia1ion at_a_rate_.ill,vPn hv thP T -~rmnr
r
~. ,1, /1
\ . --,,
"Vl\
,., .. 2
P(t) - ~ ': ___3 (v) 2 (16.6)
3c
.. , 0
"TU -r >U ..... a ...pmg, uass1cat tnouets or \..-nargeu rantctes -~
To accmmt for this rarli~tivP · ln" ~nr1 it< ,u. On thP mnJ;An nf tho
particle we modify the Newton equation (16.5) by adding a radiative reaction
force I<'rad:
mv = Fext + Fract (16.7)

While Fract is not determined at this stage, we can see some of the requirements
it "must" satisfy:
Fmct ''must" (1) vanish if v - 0, since then there is no radiation;
• 0 • •
~"'-! uco piupur tu e , ~mce ~a) llle r . puwn lS pro-
oortional to e 2 anrl (h) the siPn of thte char Pte r~nnnt PntPr
in radiative effects;
p J m ract mvolve the charactenst1c time T (lo.3 ), smce that is
-'J..Omrontln tho nnlu '~r oot ,f ,· .Lfi :1,1-1
' '. -J -J r , ~ .
We determine the form of Frad by demanding that the work done by this
force on the particle in the time interval t1 < t < t2 be equal to the negative of
the energy radiated in that time. Then energy will be conserved, at least over the
interval (t 1 , t2 ). With the Larmor result (16.6), this requirement is
r
Jil
F . v dt -
Jtl
r, 2 e2
j c
v. v dt
r
The second integral can be integrated by parts to yield
r 2
'1
e2i
Frad · v dt = -2 """"'
3 c ,,
v · v dt - -2 3e
3c
2
(v · v)
'
lj
If the motion is periodic or such that (v · v) = 0 at t = t 1 and t = t2 , we may write

('2 ,., 2 '
J,l Frad- J 3c V . v dt - ()
Then it is oermissible to id~entifv th~e rarliativP. ne~ction force as

? ,2
F,·act -
3 c3
v- mTV (16.8)
The modified equation of motion then reads

m(v - TV) = Fext (16.9)
Equation (16.9) is sometimes called the Abraham-Lorentz equation ofmo-
twn. 1t can oe conswerea as an equatiOn that mcludes m some approximate and
t;mP- ~ mmr thP ·~t;n~ orr ot. ,f.~ . . ,f rl" .t· . TJ.. '1 .
.
can be crit~ized ~n the grounds that it is second order in time, rather than first
and therefore runs counter to the well-known requirements for a dynamical equa-
,;vu v< · . "'~' · • 1 rnaniteM~ it~eu · ·latety iu · runaway so!U-
tions. If the external force is zero, it is obvious that (16.9) has two possible
~· ,
u
v(t) = flpt/T
'
whPrP. <1 is thP "rrPlPr~t;nn gt t {) nn "J
lu thP (;rof . ''· '1- I.
The method of derivation shows that the second solution is unaccCfll.able, since
Seet. 16.2 Radiative Reaction Foree from Consenation of Energy 749
(i · \) of= 0 at t 1 and t2 • It is dear that the equation is useful only in the domain
where the reactive term is a small correction. Then the radiative reaction can be
treated as a perturbation producing slow or small changes in the state of motion
of the particle.
An alternative to (16.9) can be obtained by using the zeroth-order equation
ol motiOn, mv Fext• to evaluate the radiatiOn reachon term. The resulting
equation,
mv = Fexl + T
dFexl
dt = Fext +
[fJFext
T ----;;( + (v. V)Fextjl (16.10)
is a valid equation of motion without runaway solutions or acausal behavior. It

is a sensible alternative to the Abraham Lorentz equation for the classical re
gime of small radiative effects. It also emerges from a different starting point-
see G. W. Ford and R. F. O'Connell [Phys. Lett. A, 157,217 (1991)]. Relativistic
generalizations of (16.9) and (16.10) can be constructed-see Problems 16.7 and
16.9.
To illustrate the use of (16.10) to account for small Jadiative effects, we
consider a particle moving in an attractive, conservative, central force field. In
the absence of radiation reaction the particle's energy and angular momentum
are conserved and determine the motion. The emission of radiation causes
changes in these quantities. Provided the accelerations are not too violent, the
energy and angular momentum will change appreciably only in a lime interval
that is long compared to the characteristic period of the motion. Thus the motion
will instantaneously be essentially the same as in the absence of radiative reac
lion. The long-term changes can be described by averages over the particle's
unperturbed orbit.
If the conservative central force field is described by a potential V(r ), the
acceleration, neglecting reactive effects, is
(16.11)
By conservation of energy, the rate of change of the particle's total energy is

given by the negative of the Larmor power:
dE
dt
With the defiml!on oi r (16.3) this can be wntten
dE
(16.12)
dt
Since the change in energy is assumed to be small m one cycle of the orbit, the
right-hand side may be replaced by its time-averaged value in terms of the
Ne ~toni an orbit. Then we obtain
(16.13)
•u:::l\ ., . "' n. .
•
C' •.
.. .,.. _,_,_
__
-~ no 0 .
'" "~' oo·
~
The secular chanl!e in anl!ular momentum can be found bv considerin<> th"

~~~rl,~t ~· (11'. 1{)\ ·t 1-- ol- . rl.' "' •'- .I.
r ' - ., '5'
is L = mr X v, we find
dL rfF
dt - r X l<'ext + 7£ X - 7f X--cv • V'JFext (16.14)
dt
where the second torm results because the force is central and time indenend,nt
With (16.11), the right-hand side of (16.14) is found to be
I
u v 1 av •
f.
\ ''"""
• /.._ext .. r dr I .c
7
m r dr.., ~
/ ~-
.LU • .L,,}
..,.....
¥V Jlll llHO a n.-oab'-'
r
V.L lllD lV11.jUv UV\01 lH\0 MVWJY -ururr, lll e. rate
ot change ot angular momentum can be written as
I AT\ /1 r/17\
( l ( \J (11> 11>\
-,
\ dt; m 1r drJ
NntP thM this r.esnll fnr lhP d.e<C~V nf thP. nnrti:·la'e ~nmd~r 1tnm is L'
..
wco ucot;all vco v1
' ..
'"" """ vuco "a'" · 1V1 lll\0 < llUill
~J
lll
electric dipole radiation (Problem 9.9).
r;
"1
•• ·
. f1t: 1~\
\ . "!
·' f1t: 1t:\
\ . ~, ·uo
< .. . '
U1'-' !'a1U'-'1'-' v• u n
__ , . _, _
'ao a
tunctwn ot time because ot radiative reaction. Although the detailed behavior
depends on the specific law of force, some qualitative statements can be made.
1t the charactensnc trequency ot mouon IS Wo. llie average vaTue m (16.16) can
be written
/, JT rl
I~ ~ • 2 2
71) 7lJ
m \r dr m
,.,__ rl.' -- ·"" .. .r .,_ - ,_
,1.
' .£
UUJl J , ~·-·
UUO OUV >0 ·' '"
mat me charactensnc ume over whiCh me angular momentum changes Is of the
order of 1/(w0 T)w0 . This time is very long compared to the orbital period 27rlw0 ,
l'ruviueu w0 T ...__...__ L .:.imi1ar arguments can De maoe wnn tne energy equations.
These equations including radiative effects can be used to discuss practical
problems such as the moderation time of a mu or pi meson in cascading from an
orbit of verv larl!e OLI~ntnm number arounrl ~ "';-;;~,," -:,- tn thP lnw.hrin<>
orbits. Over most of the time interval the quantum numbers are sufficiently large
that the classical descrintion of continuous mntinn is ~n ~dPnn~tP. ~nn · •~tinn
...,.. .f' • • .£ ... • . •. .
'JcH~O VL UHO ""'~
..1 •
LO
• .£•
L~Jl LV lll tiT
..
.l O.J- •
.f:lv• ~
•
~ .. -Lu. LntZ n valUation OJ me ~elJ-1' orce
1ne uerivation 01 tne rauiation reacuon wrce m :section lb.L, wnile ptausJble 1s
certainly not rigorous or fundamental. The problem is to give a satisfactory ac-
count of the reaction back on the charged particle of its own radiation fields.
Thns ~nv ;,.. rlisrnooin.-, mnst rnnsirlPr th,--:::ho.oo ;:-;:::;-;:;-,., nfthP .t'ol
and its self-fields. Abraham (1903) and Lorentz (1904) made the first att~mpt at
such a treatment bv trvin<> to make a nnrelv .e],. ..: ' nf" --'
'_, T.
! ' " " " " ' ' - ' · ' " LH~
d. L
u 0, VlU "' D jJ<ll a:ner-.:rra:T');l VCH uy LUI t;;JILZ
in his book, Theory of Electrons (note 18, p. 252).

Sect. 16.3 Abraham-Lorentz Evaluation of the Self-Force 751
Let us consider a single charged particle of total charge e with a sharply

localized charge density p(x) in the particle's rest frame. The particle is in external
electromagnetic fields, Ee,,(x, t), Bcx,(x, t). We have seen in Sections 6.7 and 12.10
that the rate of change of mechanical momentum plus electromagnetic momen-
tum in a given volume vanishes, p10vidcd theiC is no flow of momentum out
of or mto the volume. Abraham and Lorentz proposed that the apparently
mechanical momentum of a charged particle is totally electromagnetic in
origin. Here we take the more conservative position that the particle's momen-
tum is partly mechanical, but with an electromagnetic contribution. Then, if G is
the total electromagnetic momentum, the conservatiOn of momentum reads,
Cp)
dt mech
+ dG- 0
dt
or equivalently in terms of the Lorentz force density (12.121),
tdp)
dt mech
f (pE + ~J X 8) 3
d x (16.17)
In this equation the fields are the total fields, and the integration is over the
volume of the particle.
In order that (16.17) take on the form of the Newton equation of motion
dpdt -= (dp) (dp)

dt mech + dt em = Fext
(16.18)
we decompose the total fields into the external fields and the self-fields E,, B,
due to the particle's own charge and current densities, fJ and J:
E : Eext + E_,}
B - Bcxt + B,
Then (16.17) can be written as the Newton equations of motion, with the external
force as
Fext = ] (pEext + ~J X Bext) d x

3
(16.19)
and the electromagnetic contribution to the rate of change of momentum of the
(dp)
~ -_- I (pE,. + -1J X B, )d x 3 (16.20)
dt em C
Provided the external fields vary only slightly over the extent of the particle, the
external force (16.19) becomes just the ordinary Lorentz force on a particle of
charge g and velocity v
To calculate the self-force [the integral on the right hand side of (16.20)] it
is necessary to have a model of the charged particle. We will assume for simplicity
that:
the particle is instantaneously at rest;
the charge distribution is rigid and spherically symmetric.
Our results will then necessarily be restricted to nonrelativistic motions and will
lack some Lorentz transformation properties.
-~~
/;;J~ io n.amauon uampmg, Ltasstcat !Vlodets ot Lhargea rarttctes -l.
For a particle instantaneously at rest (16.20) becomes

A~ r
- - - J p(x, t)Es(x, t) d 3 x (16.21)
dt jem
The self-field can be expressed in terms of the self-potentials, A and , so that
r/n
.......
r 1 .:!A
dt
I - J p(x, t) V(x, t) +
c at
(x, t) d 3x (16.22)
em
The potentials are given by A"= (, A):

1 r rl"(v' t'\1
A "(x, t) d~x•
~) R
(16.23)
withr = (cp,J) and R = x- x'.

T,
\
(1 t:.. '1'1\ th
~ e I
A .+
' ~· . ,. "' L .1. ~
uc
...
cu~ •~cuo u~u
~ .
u u w • • dHO UHLC"
trom tne ume loy a nme ot the order ot f).t (ale), where a IS the dimensiOn of
the particle. For a highly localized charge distribution this time interval is ex-
tremely short. During such a short time the motion of the particle can be assumed
to chan{Je onlv sli{JhtlV. r'on<POlHen11v it i< nMnr~l to m~kf'. >1 ~• •
/1 L "'"'\
, '" \'"·~·'} u+vo lHC UHlC t t. 0111<.-CL" • "Jret ar
t' = t - (Ric), any retarded quantity has the expansion
~ (-1Y IR\ n ()"
_1· ·L = ), - I- l -
n 0 n. c dt
r· · ·l· (1fi ?4)
With this expansion for the retarded 4-current in (16.23), expression (16.22)
becomes
I J. = I n r r n -1
~ ~ \ • j 1 _d_3 I A3 ' ,(. ,\ u u.J\X,t)
. J ~ ,~ ' • J at"
J ' ,\ 't7 r;,n 1 "
dt em~o nl en J
r\ ' 'I c2 at
Consider the n = 0 and n = 1 terms in the scalar potential part (the first term in
the square bracket) of the right-hand side. For n = 0 the term is proportional to
J d 3x j d x' p(x, t)p(x', t) V( ~)

3
•I
This is just the electrostatic self-force. For spherically symmetric charge distri-
hutions it vanishes. The n 1 tPrm i< i, llv 7Prn <inrf'. it · <MUD" 1
Thus the first nonvanishing contribution from the scalar potential part comes
from n = 2. This means that we can change the summation indices so that the
cl.
= r
"· I
"" r tl
u_t'
dt en1
--
L
n=O
1
nJ
'
Cn+
2• J d x
3
J d'x' p(x, t)R" 1 ;tn+l {·. ·}
where (16.25)
rin '\!Rn+l
_i·. "1 _.11X t) + (X t)
dl \n -t 1 J\n -t L)It
With the continuity equation for charge and current densities, the curly
hr"rkPt in (1 fi ?'i\ r~n hP ,,.;H~~
I
{· · ·} = J(x', t) - n : V' · J(x', t)

2
Sect. 16.3 Abraham-Lorentz Evaluation of the Self-Force 753
In the integral over d 3 x' we can integrate the second term by parts. We then
have
I .J3
. ' vn 1 -x- ...,, T
T I ,3 'IT ..-r•\ vn ln
\.
J n +? 11 + ? ) I
'
=
n + 2
Jd x'R"- (J+(n-1)J·KRI
·l
R
3 1
2
This means that the curly bracket in (16.25) is effectively equal to
{·. ·) = (n + 1)J(x', t) _ (n- 1) (J · R)R (16.26)

n + 2 n + 2 R2
P~~ ~ ~;a;.-1 Ah~~an rJ;ct~;h,t;~~ th~ ~"~~n-t ;"
~
-:u;-. 7\ - --;;r;; Av7f)

If the charge distribution is spherically symmetric, the only relevant direction in
the problem is that of v(t). Consequently in the integration over d 3x and d 3x'
only the component of (16.26) along the direction of v(t) survives. Hence (16.26)
is equivalent to
{· · ·) = p(x', t)v(t)
Furthermore all directions of Rare enuallv nrobable. This means that the second
r
n+1
-
n-1 R·v
-
n + 2 n + 2 Rv ( rl
·~·~ .::;:::~"" ~~- --;:::~ ~·-1" '"'' ' ;, . ]. •• 1
j•
"T'. . 1. .-1 .
•• "'· "1
-, ' '"7 .
T
' · ''JH11 v'e vu' \-UllJ " f
. ,11 ~ l "·"'-·' }·
"'::':
(· · ·) = ~p(x', t)v(t) (16.27)

With (16.27) in (16.25) the self-force becomes, apart from neglected nonlinear
terms in time derivatives of v (which appear for n > 4),
( dp ') =
dt em
.i: (-1)" _3__
n~o 3n!
cn+l
a" 'tv
atn+ 1
Jd3x' Jd3x p(x')Rn-tp(x) (16.28)
10 proceeo runner it is convenient to introouce l:'ouner transrorms m ume

for the external force the velocitv. and the self-force.* The Fourier transform of
the velocity v( w) is defined by
v(t) = -1- Jv(w)e-'"''. dw

v27T
and its inverse, and similarly for the others. If (dpldt)mech = m 0 (dvldt), the Fourier
transform of the force equation (16.18) is
i W/VJ(WJ'VC W} - "'1:' exiT WJ ~ 1b.L'J)
,, " ee " -.. I /
W 1110' 10 tllC c VIO 11la00 1¥<~ W) 10
,., t· \n r r ~
JWW\- m" + - ) 3' UZ ~-;pr ld3 r' n(x\R" 1 n(x')

Jc n~o n! c J J
*Here we parallel quantum-mechanical discussions of radiation reaction in the correspondence limit:

Nonrelativistic theory, E. J. Moniz and D. E. Sharp, Phys. Rev. D 10,1133-1136 (1974); fully rcla-
tivistic quantum theory (QED) of electrons and positrons, F. E. Low, Ann Phy. (N.Y.). 265, No. 2
(1998).
The sum over n can be recognized as eiwR!c/R, the outgoing wave Green function.
Hence M(w) can be written
2 ( ,, ( .~ eiwR/c
1Y1 ~ w} " fflQ T C J U .< J U. X p~X) p~x) ~lO,..)U)
3 2 R
The spherically symmetric average of e'wRk/R is
\ ~ ) = i w io( wr < h&l) wr>)

R c c c
For some specific spherically symmetric charge distributions, the spatial integrals
in (16.30) may be performed to give an explicit closed form for M(w). [See Prob-
lem 16.4.]
Alternatively, we can introduce the spatial Fourier transform (form factor)
~+tho~' ~rbne;l t. -'- · rl' ~ ~ · C Hi\ " .1
·e "!" '\'UJ• ~r ''~!-'-
resentation" familiar in quantum mechanics. We define the form factor f(k)

through the three-dimensional transform
r
p(x) = - -
(21T)3
J d 3 k f(k)e'k·x (16.31)
where e is the total charge. For a point charge, f(k) -- 1. By straightforward

. .
'
>UUMHUllUll
' .
<lllU
. .
~· dllUll, ~ lV.JV) I> LO
M(w) = m 0 + _e- J d 3 k IN<) (16.32)

31T2C2 k2 - (w!c?
where w has a small positive imaginary part.
Equations (16.29) and (16.32) are an almost complete solution for the clas-
sica! nonrelativistic motion of an extended charged particle, including radiation
reactwn. L"Almost, because we neglected small nonlmear terms in higher pow-
MO ~f thP uP' · ~nrl mo ~oonm d • ,t · ,] •l T. •'- 1: ;, f\
~, 'T "J 'J•J '
(16.32) is M(O) = m, the physical mass of the particle, including the contribution
rof thP sp]f_fip]rk
2 r I r11. \ 12
3
m mo +
31T
2 2
c
J d k k2 (16.33)
In terms of m, the effective mass M(w) can be written
M(w) = m +
2
e w
31T2c4
2
J d3k z.lr£-2-
if(k)
2
1
(,,.fr'\21 (16.34)
WP nnw lPnt ron thP <nln1iron \XlP h 'o ~hto;norl +~• tho ~r '"
I Cll<il!SCU pal LlClt:, ~-d. r auiadon reaction:
1. 1ne sen-new comnounon to me mass m (16.33) diverges lmearly at large k
mithront thP frorm f~r-1nr ,1'1 inn thP f~r-t tho+ +ho eoll' f;o],j,' ,]
' ,o '
uuMatic cHCI~SY ut we uruer ur e w, wnere a is a scale parameter oetermmmg
thP sizP of thP rh~r<>P distrih11tion
2. The frequency-dependent integral in (16.34) is more convement bv a factor

~+ /,2 lhnn lh '1 . ( 1 h '1'1\ r1 1 '~ ;c r t L \
'I D '5' ·s~
.~
'
'' ~HHJ\RJ
(point charge).
Sect. 16.4 Relativistic Covariance; Stability and Poincare Stresses 755
3. For a point charge, the integral in (16.34) can be performed easily by contour
[M(w)]point = m(1 + iWT) (16.35)
form, leads back to the Abraham-Lorentz equation, (16.9). The zero in the
upper half-complex-w plane at wT = i in (16.35) signals the runaway solutions
of that equation.
4. For a sufficiently convergent form factor, the integrals in (16.33) and (16.34)
5. While the nonrelativistic approximation causes conceptual difficulties-the

2 .
self-energy, rather than 1/c 2 times it-these are removable by more careful
arguments. [An early relativistic treatment was given by Fermi*; a covariant
6. A uantum-mechanical treatment of a nonrelativistic extended char ed ar-

tide in interaction with electromagnetic fields gives essentially the same re-
sults, (16.29) and (16.32), for the expectation value of the appropriate op-
. .
tions occur. Moniz and Sharp endorse (16.10) as the most sensible form of a
classical equation of motion with radiation reaction, to be considered ap-
roximatel valid when the reactive effects are small.
ovarzance; omcare resses

So far our discussion of the Abraham-Lorentz model of a classical char
tide has been nonrelativistic, with apologies for the paradox of different electro-
magnetic "masses" from electrostatic and Lorentz force (dynamic) consider-
netic stress tensor (12.113). In contrast to source-free fields, the stress tensor ®"13
(16.36)
if the 4-divergence of a stress tensor vanishes everywhere do the spatial integrals
*E. Fermi, Z. Phy. 24, 340 (1922), or Atti. Accad. Nazi. Lincei Rend. 31, 184, 306 1922).
oi 8"0 transform as a 4-vector. Thus the usual spatial integrals at a fixed time of
the energy and momentum densities,
1
g =-(Ex B) (16.37)
may be used to discuss conservation of electromagnetic energy or momentum in

a given inertial frame, but they do not transform as components of a 4-vector
unless the fields are source-free.
As Poincare observed in 1905 1906.* a deficiency of the purely electromag-
netic classical models is their lack of stability. Nonelectromagnetic forces are
necessary to hold the electric charge in place. Poincare therefore proposed such
forces, described by a stress tensor P" 13 to be added to the electromagnetic 8"13
to give a total stress tensor 5"13 ,
The particle's total 4-momentum is then defined to be
where the integral is over all3-space at a fixed time. The right-hand side of (16.38)
transforms as a 4-vector provided
or equivalently, provided
(16.40)
with i, j 1, 2, 3, and the superscript (0) denoting the rest ftame (P 0).
Condition (16.40) is just the statement that the total self-stress (in the three-
dimensional sense) must vanish-the condition for mechanical stability.
Poincare's solution provides stability and also, because of the generality of
the postulates of special relativity, g'larantees the proper Lorentz transformation
properties for the now stable charged particle. A criticism might be that Poincare
stresses are not known a priori in the way that ®"13 is known for the fields. If we
think, however, of macroscopic charged objects, for example a dielectric sphere
with charge on its surface. we know that there are "nonelectromagnetic"
forces polarization and quantum mechanical exchange forces (actually electro-
magnetic at the fundamental level) that bmd the charge and giVe the whole
system stability. It is not unreasonable then to include Poincare stresses in our
classical models of charged particles, or at least to remember that care must be
taken in discussion of the purely electromagne1ic aspects of such models
It is of interest to note that for strongly interacting elementary par tides one
has a concrete reahzat10n of the Pomcare stresses through the gluon field. Con-
*H. Poincare, Cornptes Rendue 140, 1504 (1905); Rendiconti del Circolo Matematico di Palermo 21.
129 (1906). The second reference is translated. with modern notation in H M Schwartz Am T
Phy' 39 1287 ( 1971) 40 862 1282 (1 Y72)
Sect. H>.5 l.:ovanant Uenmuons 01 t<,tecrrornagneuc "nergy ano ~~•omemum """'""
/.J I
sider the proton, for example. Its three charged quarks are bound together by
the gluon field in a stable entity with an extended charge distribution. Setting
•> -~rl. +ho ;n+orn~l drn~hr ~nd obhilitv nf thP nn~rl<< thPm<PlVeS the eleCtfO-
magnetic stress tensor eafl must be co~bined with the "Poincare stress" tensor
e;/3 of the gluon field to give a divergence-free total stress tensor. The mam part
;: . .• c.. <1.. • • • •• c. +h ,J.
UL lllC HW'' Vl liLt: p<V' · uvm cu~ o v 'IS ,
troma2:netic contribution to the self-energy--the neutron and proton have the
same internal strong interactions, but different electromagnetic; their masses dif-
fer by only 0.14% (and in the opposite from expected sense).
In the next section we examine covariant definitions of the total energy and
. • " ~L
momentum or e1ectromagneuc llelUs, even m wc: pu vt . . , u"'""'
~- L'. h~ -~
;;: ~.' HTPO
·a
mhPn nnrP]U
r
pjp, >~<TnPtir- io"lf" ;)fp con-
sidered, but in general the nonelectromagnetic forces or stresses must not be

forgotten.
16.5 Covarrant llepnitzons OJ btecTromagneuc .l!.nergy

ana JVJom~
As emphasized by Rohrlich, even if the electromagnetic stress tensor eafl is not
divergence less, it is possible to give covariant definitions of the total electromag-
netic energy and momentum of a system of fields. The expressions
1 r
}'__'~ S7T J (I!:' + H' c) a· X
,..
~J.V."<.l.}"
nl
1 ( T' I Dl J3 I
~
•e
47TC j ~
1
can be considered to deline the energy and momentum at a fixed time t in some
~
. -·· ,] or :n •r+;~J <'r~~o v• , tn hP on' "' ~ ohnrtlu ThP in• . in (1 11.41) >Ire
elements of the second-rank tensor ®all_ Evidently we must contract one of the
1
tensor indices with a 4-vector, and the 4-vector must be such as to reduce to d"x
. . .
m tne mernarrrame-x- . vv c: lll(; <-t-vc;uv<,
du 10 - n 10 d'u
where d 3 u is an invariant element of three-dimensional "area" on a spacelike

hvnPrnlane in four dimensions. The normal to the hvoerolane n 13 has components
3
(1, 0, 0, 0) in K The invariant d 3 u is evidently d 3 cr = n 13 du 13 = d x If the
1 1
• •
mertlal trame K' moves wttn velOcity cjj wnn respect to an inertial rrame A, lllen
• V +h A /3 ~"
ni3 = ( y, y~) (16.42)
A nonor~l clpr.n;;;~n ~f thP .1.
~-
.. : 4 .. mnmP.ntum in _any c. is
therefore
r r
cP; - J 8"13 du13 - J ®"iln 13 d'u (16.43)
..,,., .
JoJU >V '""U<dUUU ,
~-
~mssica1 InooeJS 01 ~nargeo rarncJes -'l>
In K', n 13 has only a time component. With d\r = d 3 x', this covariant expression
reduces to (16.41 ). But in the frame K, n 13 = ( y, -y~) and the covariant definition
has time and ~ace com_Q_onents
r
cf' e - 1' J (u v . g) d· (T
~10.44)
_c_Pi '
1
{ {
J ,-o
r~i -' r(M) Qi\ A3
'] /
where T)J<Il is the 3 X 3 Maxwell stress tensor (6.120). If desired, the invariant
volume element d 3 u = d 3 x' can be suppressed in favor of the volume element
3
d x in the frame K by means of d 3 x' = y d 3 x (integration at fixed time t).
The definitions (16.43) or (16.44) of the electromagnetic 4-momentum afford
a covariant definition starting from the naive expressions (16.4 1) in any frame
''" . · - •~ea• - , v• <H<' uame 1\. teau 10 · '+·Vectors, 01 course, Dut
that is no cause for alarm.* Th~erP ;, " nMnr"l "hoi"" nf th~e fr'lmte T<' ;f thP
electromagnetic mass of the fields is nonvanishing, namely, the rest frame in
Wl1!Cn
1
J E(o) x B(o) d 3x(o) = 0
41TC
We denote this frame where the total electromagnetic momentum P; is zero as
K '"' and attacn superscripts zero on quantities in that frame to make it clear that
;,•, ·,,,,. ,C £. TTJ ,. r ·
·r ·-- ~• '"" H O>Hv H. , 0 UU<t; LV \ l V , ' H j U>v , ~ ;LI(;
rest energy is then

1 (
·'"'
~e mec
0
87T ) L~ -t D -J a x UtJ.4))
In the frame K the electromagnetic energy and momentum are given by (16.44)
where now v is the velocity of the rest frame K(o) in K.
For electromagnetic configurations in which all the charges are at rest in some
frame (the Abraham-Lorentz model of a charged particle is one example), the
aPnPr~l fnrmnl~, "~" hP ~- ~ •- · .-1 '-
·r • •
Clearly the frame where all the charges are at rest is K<"l, since there all is
electrostatic and the magnetic field vanishes everywhere in 3-space. For such
.1 . ·" . " . ' ' ' . ' •
v<vv• _(_ • ~i""• Ul<O , HIOlU 1, t',lV<OH VVOLHUUL ·nc iil
the frame K l;Jy 11.150 :
B-RxF
The mtegrand m tne nrst equation of (16.44) is thus
(u - v · g) = _!_ (E 2 + B2 ) - _!_ B · (E x B)
~'IT '+'IT
' '
= ' (E 2 + B 2 ) - ' (~ x E) · B
R,. 4,.
1
87T ~~- D-)
*One possible choice for K' is the "laboratory" where the observer is at rest. The discussion of the
CVHOC1 oaHUH 1m <0 r v may oe inLerpre.eu m LOIS way.
Sect. 16.6 Covariant Stable Charged Particle 759
a Lorentz invariant. Thus the energy in K is given by
(16.46)
(1 f. A'7\
c.o e Yl-' Q
With the invariant integrand (E 2 - B2 ) it is clear that we have a 4-vector

P; = (ymec, ymev), where the electromagnetic mass is
me = _1_
8 'lTC 2
I (E2 - B2) d3u = ~I
8 'lTC
E(0)2 d3x(O) (16.48)

The equation (16.46) for the energy has been used by Butler* to discuss the
Trouton-Noble experiment, a test of special relativity involving the question of
a torque on a charged suspended capacitor moving with respect to the ether.
th~t ;nr-luriPe ~c " ]p thP nr thte ~"ertion of Prohlem 12.18

+ • • ·' ~ " ' ' TL
1 _J -u:a:s-re' lllCO -r ll VL lllC, HVULVll' rur:m;:c -r "' "~
stresses that the removal of the paradox requires consideration of the nonelec-
tromagnetic forces for stability, but that it is a matter of choice whether the
balancing of electromagnetic and nonelectromagnetic forces is done in a mani-
festly covariant way or not. All that matters is that the total stress tensor S"' 13 be
divergenceless.
16.6 Covariant Stable Charged Particle

~ 'T'L • • , '
" .
• '" . e oL '-l .C C'. 1 L A rl 1 f. <; '
1">.11 Ill' lllll~ "'l""' ~• "'~ ..
nrovided bv a model of Schwm2:er+ for a classical stable smnless char2:ea panicle.
With its consideration of the Poincare stresses needed for stability, it may also
be viewed as a prototype for the discussion of macroscopic charged mechanical
systems. The model is, in fact, a modern generalization of Poincare's work 77
years earlier [see the middle paper of Schwartz's translation (op. cit.)]. In the rest
frame K' of the particle, the 4-vector potential is defined as
A'= 0
With f(rL) an armtraiVwell-behaved lUnCt!On bUt with theTimffingTorm nr) --7
1/, •· -' ·"- oJ.. • •••1 ,J..~rn~ ~f thP ' oc o '\Tp n~m r-~ne;dPr ~ l:
. -e:: ,. . .
trame K m Which the oarncte moves Wlln vetocHy v anu Lne ... -v• .y
'J. w. tlutler,Am. J. n1ys. Jll, ~jo \IYOCJ.
1J. Schwinger, Found. Phys. 13, 373 (1983).

760 Chapter 16 Radiation Damping, Classieal Models of Cmnged Particles 6
(dlVlded by c), v" (y, yl$) 0"/c (11.36), With v"v, 1. We mtroduce a
4-vector coordinate !;"perpendicular to v",
(16.49)
Then we define the invariant coordinate variable z by
Z = -e = -{; · {; = -x+ (u · x?
·X (16.50)
In the rest frame K', /; 0 = 0, ~ = x, and z becomes z = r2 •
The covariant generalization of the rest-frame potentials is
(16.51)
To evaluate the fields we need
that with a f3 (and summed) we obtain the Loren:~: condition on the potentials
because {; u 0.] The field slt cngth tensor is
F" 13 = -2e(!;'V - !;13 v")f' (16.52)
The current density is obtained from the Maxwell equations,
c ec
1 13 =- a,F"13 = -- [3f + 2zf"]vll (16.53)
47T 27T
B. The Electromagnetic and Poincare Stress Tensors; Arbitrariness

The symmetric stress-energy-momentum tensor (12.113) is easily found to
be
( 16.54)
and its dive1gence (16.36) is
(16.55)
The Lorentz force density [negative of the right-hand side of (16.55)] must
be balanced by Poincare stresses for stability. Schwinger, noting the derivative
relation a"G(z) = -2/;"G', defines a function t(z) whose derivative is
dt(z) e2
t'(z) = - = -- [3(!') 2 + 2zf'f"] (16.56)
He then defines the Poincare stress tensor to be
the right-hand side of (16 55)1 We thus have

< af3
"
The total stress tensor S"fl = @"f3 + P"f3 is divergence less; the spatial integrals of
S"0 transform as a 4-vector. The model is covariant and stable.
~ ....
Before proceeding, we note an arbitrariness in the Poincare stresses. Any
nonelectromagnetic stress tensor with a vanishing divergence may be added to
paf3 RPr~n~e v · 1:- 0 it follows that a::Tv"v 13 S7z1T- 0. This means that we may
add !!,.paf3 = v"vf3 s(z), with s(z) arbitrary, without changing the stability or co-
variance of the model. We will, of course, change the energy and momentum of
the particle, as is illustrated below for our special choice of the additional term,
(16.58)
w1th h constant. ~chwmger discusses the two cases, n u anu n

components of the total stress tensor S"0 are explicitly
soo = eoo + (1 + h·l)t
~
'"
a '"
-r- rry 0 p t
showing that when h = 0, S 00 = (8° 0 + t) and Sw = ew in all frames. When

h = -1, S 00 = 8° 0 in the particle's rest frame. Schwinger's original choice of
Poincare stresses (16.57) is in some sense the minimal and natural choice, tied
directly to the electromagnegic field configuration. Note that the terms propor-
. 0 • · L <L .L'''
L10na1 LO v v conLnuuLe LV LHe 111 Lue 1eM 11 ""'"• " u ' au• cv cu~,.
forces (from the space parts of ll"P). Pomcare had a spnencat snell or cnarge
with an arbitrary "pressure" inside, equivalent to our arbitrary s(z) above.
· ol
• •vm •a~ "'[' ' ·· -; r -o
ment that t(z) vanish at infinity, an integration by parts leads to
(16.59)
For specific forms of the potential function f(z) it is a straightforward matter of

integration to find t(z). It is left as an exercise to show for a spherical shell of
charge of radius a and a uniform volume distribution of charge of the same radius
that
e shell or charge
1 1
u uniform density
The shell of charge provides the most dramatic illustration of the stabilizing effect
of the Poincare stresses. They exist only inside the sphere. Because there are no
helds ms1de the sphere, the electromagnetic stress ex1srs onty oursiae me spnere
4
and !!ives a desl ili"imr outward force ner unit area at r a+ equal to e"I87Ta
in the rest frame. At r =a-, the Pioncare stress provides the stabilizing inward
force-the surface layer of charge feels no net force. Continuity across an inter-
face of the total stress tensor contracted with the unit normal is the more general
rritPrinn fnr no nP.t fnrrP. ~t thf'. in1 Prf~rf'
• u~ '", '5' , , r '"" Tn n

tram ~ 1O.'li'\) mrecuy or rrom enner rne resr -11 ctluc . o , u1 " ' v1 J ./'1. , Lc. 111
0 2 2 2
the first way, we need £C l = 4e • z(f') . Then we find
e I z "U') az ~lO.OU)
JU
The contribution to the rest mass from the Poincare stresses is
Integration over angks, then substitution of (16.59) for t(z) and an integration
by pm ts leads to the Jesuit,
(16.61)
The total mass is therefore
M =me + mP = ~(4 + h)me (16.62)
Note that when h = 0 the mass is 4me/3, the "dynamic" result, while if h = -1,
M = me, the electrostatic result. On the other hand, if h >> 1, most of the mass
is of nonelectromagnetic origin. Neither the 4/3 nor the unity proves anything
about the covariance of the energy and momentum of the particle. This property
is guaranteed by the divergence-free saf3, as we now demonstrate.
D. Demonstration of the Covariance of the Particle's Energy

and Momentum
I fie evaluatwn of the spat1al mtegrals of (3 00 • ®' 0 and Il 00 • n.o and their sums
at fixed time YP in the laboratory frame illustrates the conspiracy between the
electromagnetic and Poincare stresses to assure the proper Lorentz transforma-
tion properties. We begin with ®00 :
®
00
=-: [ -(t
0 2
) + ·h- ~ z]cf'?
Since we are to integrate (<) 00 over 3-space at fixed time in K, we need (t0 ) 2 and
z evaluated explicitly in K. If we take the 3-axis parallel to (3, from the definition
(16.49) and v · t = 0, we find go= f3e and e = y(x' - f3x 0). With (t'? + (ef
= (x ) + (x ?- p , we have
1 2 2 2
z = p2 + ·/(x3 _ f3xo)2
If we define x~ = y(x 3 - f3x 0), which is just the 3-coordinate in K', the volume
element d'x can be wntten d"x d·'x' /y. Putting the pieces together we have
the electromagnetic pr.rt of the energy as
Averaging over angles introduces a factor of [(~)y2 f3 2 + ~]r' 2 instead of the square
bracket. With the definition of me through (16.60), we obtain
Ee = ( ~ y - 3
1
y )mec 2
(16.63)
A corresponding computation of the integral of ®30 gives the electromagnetic

momentum
!16.64)
Clearly the electromagnenc contnbutwns alone do not transform properly.

Sect. 16.7 Line Breadth and Level Shift of a Radiating Oscillator 763
The nonelectromagnetic contributions to the energy and momentum are
EP = J Il d x = (1 + h·-/) J t(z) y
3
00 3 1 ( -I
d x' = 3 + yh mec2
'Y (16.65)
r ,3 ,
cPP =
'
J Il 30 d 3 x = hy (3
2
J t(z) "; = i
'
hyf3mec
2
>.L :+L -'· .t.. ~ . ' 'tPn<nr

'.: +
... '
-... ._,. Tho •~to 1 et
contributions, the sums of the separate contributions, do, however, yield a proper
relativistic energy and momentum:
E = yMc 2 , cP = y(3Mc 2 with M = ~(4 + h)me (16.66)
the same rest mass as found above. Schwinger's choices of h = 0 and h = -1
were made to illustrate that etther the electrostatiC mass orlile lTynamic mass
can serve as "the mass" when the charge is stabilized by the Poincare stresses.
Other choices of h are possible and, as noted, above, other totally arbitrary con-
tributions to the mass can be introduced without affecting the question of the
covariance of the model.
a lthrmnh mo oobhl;d,prl thte 4- · nMllrP. nf PnPr<Jv >Jnrl momentum usin!!.
the conventional definitions of the total energy and momentum by taking 3-space
integrals at fixed time x 0 in the laboratory frame K, it is of interest to see how
"'~ ~' · • ~' •e tl. -1 .<C ' ' -~C. ,t." 1<:" · ;olrl ·~
<H'-' " " " . ~0 • • J
variant expressions for the separate contributions. The appropriate quantities,

according to (16.43), 16.54) and (16.58), are
2
cP~ = J f<VI3v 13 d a = Z7r

e va
3
J z(f'? d u 3
(16.67)
cP; = J IIaf3v 13 d 3 a = (1 + h)va j t(z) d u 3

(16.68)
<;c;n~P tho ;ntoo
-co
"rp T.nrP.nt7 ;, . to Wf' m"v PvqJuate the
'. onrl . .
integrals in the rest frame. From (16.60) and below, we see that
cP~ = (mec 2 )va and cP; = W + h)(mec )ua 2
are separately 4-vectors by construction, with a sum equal to (16.66). The sim-
plicity and elegance of the use of the manifestly covariant (16.43) is apparent.
lhe results are, or course, me same eiwer way.
The Poincare-Schwinger model of a stable charged particle addresses the
issue of the Lorentz transformation properties of the particle's energy and mo-
mentum, but does not attack the question of radiation reactiOn. For the sphencal
shell model, this problem has been treated in detail by Yaghjian* who also treats
thte Poincare stresses and stabilitv. See also Rohrlich."'
111 7 T.inP ~ -l~l. nnrl Tmonl .f\hift nf a Rrul:afinp Or.:ri/lr~~

v
The effects of radiative reaction are of great importance in the detailed behavior
• o• • • 0 .0 oL ,1,
Ul • OJ JUbH a 'I"" o HPV'~ •u•~
*A. D. Yaghjian, Relativistic Dynamics of a Charged Sphere, Lecture Notes in Physics mll, Springer-
Verlag, l:lerlm, New York (f99Z).
'F. Rohrlich, Am. J. Phys. 65, 1051-1057 (1997).
formalism of quantum electrodynamics, the qualitative features are apparent

from a classical treatment As a typical example we consider a charged particle
bound by a one-dimensional linear restoring force with force constant k - mw~,
In the absence of radiation damping, the pm tide oscillates with constant ampli-
tude at the characteristic frequency w 0 , When the reactive effects are included,
the amplitude of oscillation gradually decreases, since energy of motion is being
converted into radiant energy, Thrs rs the classrcal analog ol spontaneous emis-
sion in which an atom makes a transition from an excited state to a state of lower
energy by emission of a photon,
If the displacement of the charged particle from equilibrium is x(t) and
Fext = -mw~x, (16,10) becomes
mx = - mw 02 x - mw 02 7X. (16.69)
Because of the expected decay of the amplitude, we assume a solution of the
form
x(t) = x0e at (16.70)
where a should have a positive real part and an imaginary part close to w0 if the
radiative dampmg effects are small. The ansatz leads to a quadratic equation
for a,
a2 - 2a
7w 0 + w 2_0
0 -
with roots
In the last form we have expanded to order 7 2 in the real parL The real part of
a is f/2, where r is known as the decay constant and the change ~w in the
imaginmy pm t &om w0 is known as the level shift*:
1 3 2
f= ~w = -swoT (16.71a)
The alert reader will rightly question the legitimacy of keeping terms of order 7 2
in the solution of an equation that is an approximation valid only for small 7 (see
Pwblem 16.10b). In fact, if the Abraham Lorentz equatwn (16.9) rs used mstead
of (16 10), the resulting cubic equation in a yields, to order 7 2 , the same r, but
(16.71b)
The important message here is that the classical level shift A txJ is one power higher
order in Wo7 than the decay COnstant f.
The energy or the oscillator decays exponentially as e 1 ' because of radiation
damping. This means that the emitted radiation appears as a wave train with
effective length of the order of elf. Such a finite pulse of radiation is not exactly
monochromatic but has a frequency spectr am covering an inter val of order f.
The exact shape of the frequency spectrum is given by the square of the Fomier
'I he reader rs mvrted to pause at this point and consider the decay constant r from various points
of view. One is to use the Larmor power formula (16 6) and conservation of energy directly to relate
the time-averaged radiated power F(t) to the total energy of the oscillator E(t). Another is to ask fur
thg initial energy and amplitt:Ide x 0 of the oscillator s tteh that F f'ffcw0 , CUI 1 espunding to the emission
of a single photon of energy htv0 . These can then be compared to the values for a quantum-mechanical
oscillator in its nth quantum state.
Sect. 16.7 Line Breadth and Level Shift of a Radiating Oscillator 765
transform of the electric field or the acceleration. Neglecting an initial transient

(of duration T), the amplitude of the spectrum is thus proportional to
E( w) oc Ce-"'e'"'' dt =
1
cr~
v'
~: ~
.. :t •• ..,
-
,]" <L .c.
.Hf . . \ T' 1
~::!. = lo- (16.72)
dw 27T (w - w0 - ~w) 2 + (f/2) 2
where 10 is the total energy radiated. This spectral distribution is called a resonant
line shape. The width of the distribution at half-maximum intensity is called the
hal{width or line breadth and is equal to f. Shown in Fig. 16.1 is such a spectral
line. Because of the reactive effects of radiation the line is broadened and shifted
in frequency.
The classical line breadth for electronic oscillators is a universal constant
.... ..
.. L
~
•
.1 •
H> ·~HHO
'
V>
.c
·~·
_,
v
. t..
~A= 27T..:::._ f = 27TCT = 1.2 X 10 4

A
wi;
Quantum mechanically the natural widths of spectral lines vary. To establish a
connection with the classical treatment, the quantum-mechanical line width is
UHWH
.
~0
'
T' £ T'
q J I)
where j,1 is the "oscillator strength" of the transition (i ~ j). Oscillator strengths
vary considerably, sometimes being nearly unity for strong single-electron tran-
sitions and sometimes much smaller. For optical transitions, A ~ 4-8 X 103 A.
Thus ~A/A :S 3.5-1.5 X 10- 8 and Wo'T = 0(10- 8 ).
The classical level shift ~w is smaller than the line width r by a factor
w0 T << 1. Quantum mechanically (and experimentally) this is not so. The reason
is that in the quantum theory there is a different mechanism for the level shift,
~ lthmwh dill involvino thP Pl. l~O'nPtir fiPlrl 'PvPn in thP ~L of -L
the qu~ntized radiati;n field has non vanishing expectation values of th~ squares
01 we e1ecLromagnenc nelG suengws ~vacuum nucwauons )· 1 nese nucwaung
),
......
/: \
.! iLr.J ~
:;;I.e
/: I ....I\_
.., ~
_/ I "-..._
I
w-
,,
·' ,.,, ,.;
u' 1" 1 u "· ,., '<•. '1 1: h 'T'h
r~:~nant line shape has width f. The level shift is ~w.

/00 Chapter 16 Radiation Damping, Classical Models of Charged Particles-G
fields (along with vacuum fluctuations in the electron-positron field) act on the
charged particle to cause a shift in its energy. The quantum-mechanicallevel shift
for an oscillator is of the order of
11wq
- - - W0 Tlog
I mc 2
Wo 1 nw0
as compared to the classical shift due to emission of radiation,

I I
lL.lW<:J
- - - (w 0 T)
2
~u
The quantum-mechanicallevel shift is seen to be comparable t~ orgreater than

the line width. The small radiative shift of energy levels of atoms was first ob-
served by Lamb in 194 7* and is called the Lamb shift in his honor.
16.8 Scattering and Absorption of Radiation by an Oscillator

'rh ,£ _. .,, _. .
d,'. _,_. -"' 1. ~- ,1.
-o J J • s~v pm "~'"" '" ... ,, .0.
We now consider the scattering and absorption of radiation by bound charges,
in n~rtirnl~r thP o • · ·~ ~• ~ X · ,f f, _ . ~ · -' -' · · ·
r c '1 -J JJ ·~ •~wunou-.,
particle of mass m and charge e bound by a spherically symmetric linear restoring

force mwtx. The total force acting on the particle is (neglecting the magnetic field
term because of the assumption of nonrelativistic motion)
F = -mw6x + eeEaeik·x ;'"'
where £ 0 is the magnitude and e the polarization vector of the incident electric
held. We mtroduce a resistive term mf'v in the equation of motion to allow for
other dissipative processes, corresponding quantum mechanically to other modes
or aecay nes10es pnoton re emiSSIOn. W1th this addition, substitution into (16.10)
leads in the electric dipole approximation to the equation of motion,
.. eEn
A T ~' 1 JX -t WaX E~l IWT)e (lb. fj)
m
Here we have neglected the (v · V) term for the incident field because it leads to
.!. . 'r~ ·-'. _, ..
-,w,.._,
"
eE" (1 - iwT)e-;'"'
X e 2 (16.74)
m w 0 - w 2 - iwf t
where 1, - 1 + f' is the total decay constant or total width at resonance.

_,
crt
... ~
_.
o ~·~v U o V U ' 1;;1 V<;;~ 11~(; lU 1 11 , 1;;1V<OH uy ~l.'+.li:\),
e 1
.1'-md 2 ]D X lD X XJrct
c r
*W. E. Lamb and R. C. Retherford, Phys. Rev. 72, 241 (1947).

Sect. 16.8 Scattering and Absorption of' Radtation by an Osctllator 767
The scattenng amphtude for scattered radtahon ol polanzatwn ~· ts
or
(16.75)
The differential scattering cross section is the absolute square off:
:~ (~:2)Tcw6 _w~: + wzn] I~'*· ~1 2

=
(16.76)
\Ve have omitted the factor of (1 I w2 1 2 ) 1 in the numerator because the cross
section ts already proportiOnal to (cTf I he total scattenng cross section can be
written
(16.77)
Here X = c/w0 is the wavelength divided by 27T at resonance and r = w6T is the
resonant scattering width or radiative decay constant.
The scattering cross section exhibits a resonance at w = w0 with a peak value
of u~~~' = 61rX6(f/f,) 2 . It is proportional to w 4 at very low frequencies-Ray-
leigh's law of scattering, discussed in Chapter 10. At very high frequencies
( w >> w0 , r,), it approaches the Thomson scattering cross section for a free
particle. Figure 16.2 shows the scattering cross section over the whole classical
range of frequencies.
The sharply resonant scattering at w w0 is called resonance fluorescence.
Quantum mechanically it cm responds to the absorption of radiation by an atom,
molecule, or nucleus m a transitiOn from tts ground state to an exCited state wtth
the subsequent re-emission of the radiation in other direclions in the process o[
de excitation. The factor 61rl\6 in the peak cross section is replaced quantum
mechanically by the statistical factor,
2 2 2Jex + 1
61rX 0 ~ 47TX 0 Z(Zlg + 1)
-----7!1----6...1-o2 (r)
2
r,
i
3
I
:
.,';/, I
I
,.,
:...!.
\
_,.,4
-,. -r / I !
I
I
' ......._
0
-u
"' --
Figure 16 2 Total cross section for the scattering of radiation by an oscillator as a
function of frequency. uT is the Thomson free-particle scattering cross section.
where .Tg and .T,. are the angular momenta of the ground and excited slates, and
4 rrK6 is the maximum allowable scattering fot any single quantum stdte. The
remammg factors represent a sum over all final magnetic substates and an av-
erage over initial ones, the factor 2 being the statistical weight associated with
the incident radiation's polarizations. The classical result corresponds to .Tg = 0
and .lex = 1.
The total cross section, scattering plus absorption, is obtained from the scat-
tering amplitude (16 75), including the numerator factor (1 i&n) neglected in
(16.76), by means of the optical theorem (10.139):
u, =
411'
k Im[j(E' = E, k' =
2
[ w f(f' + w f/w )
2
k)] = 67Tll.6 (w _ w 2 ) 2 + w 2 ~~
2
J (16.78)
6
The structure of the numerators in the scattering and total cross sections has a
simple interpretation In (16 78) there is one factor of r corresponding to the
incident radiation being absmbed. This is multiplied by the sum of widths for all
possibilities in the final state-the elastic scattering and the absorptive pro-
cesses-because it is the total cross section. For the elastic scattering cross section
(16.77) there are two factors of r, one for the initial and one for the final state.
Note that, while the elastic scattering and total cross sections approach the Thom-
son hm1hng form at h1gh energies. the melaslic or absorpl!ve cross sectiOn has
only the resonant shape, vanishing as llw 2 at high energies provided f' is energy
independent.
Just as was done in Section 7.5 in the discussion of the atomic contributions
to the polarization and dielectric constant, we can generalize the one-oscillator
model to something closer to reality by assuming that there are a number of
oscillators with 1 esonant frequencies w;, 1 adiative decdy constants f; f 1 w] T and
absorptive widths fj. Then the total cross section, for example, becomes
2 2 2 2
~ w f(f'
2 1 } + w f 1 /w 0 )]
O'total = 611' L.. J\.1 [ ( 2 _ 2)2 + 21~2
1 (j.JJ w w ]J
With the appropriate definitions off, f 1, and w1 , this result is almost the correct
quantum-mechanical expression. Lackmg are the mterference terms from over
lapping resonances. The quantum-mechanical scattering amplitude is a coherent
superposition of the contributions of all the intermediate states allowed by the
selection rules. Usually the states are narrow and separated by energy differences
large compared to their widths. Then the interference terms can be ignored. In
special situations they must be included, however

The history of th<e attempts at classical models of charged particles and associated
questions is treated in interesting detail by
Whittaker
The ideas of Abraham, Lorentz, Poincare, and others arc presented by
Lorentz, Sections 26-37, 179-184, Note 18
Clear, if brief, treatments of self-energy effects and radiative reaction are given by
Abraham and Becker, Band II, Sections 13, 14, 66
Landau and Lifshitz, Classical Theory ofFzelds, Section 9.9
Ch. 16 Problems 769
Panofsky and Phillips, Chapters 20 and 21

Sommerfeld, Electrodynamics, Section 36
A relativistic classical point electron theory was first developed by
P. A, M. Dirac, Proc. R. Soc. London, A167, 148 (1938).
'rl. .T TY >L
. )
.
. ~- ~ "
tlarur
Rohrlich
_, .. -~. ~-
••L
'" ·''"
~·-
~·
"~U
') "
A succinct discussion of "classical electron theory from a modern standpoint" is given
by
S. Colem"n. in P/, · Paths to Research ed. D. M. Teolitz Plenum
Press, New York (1982), pp. 183-210.
A useful review of all aspects of classical electron models is
P. n in ~· umet;_,·m: Paths to Research ed. D. M. Tenlitz Plenum
Press, New York (1982), pp. 211-295.
The most detailed and explicit discussion of radiation reaction for a charged spherical
shell 1s that or
Yaghjian (op. cit.)
Problems
16.1 A nonrelativistic particle of charge e and mass m is bound by a linear, isotropic,
restoring force with force constant mw6.
Usin2: (16.13) and (16.16) of Section 16.2 show that the ener2:v and angular
momentum of the particle both decrease exponentially from their initial values as
e-rt, where r = w6T. Quantum mechanically, the mean excitation energy of an
oscillator decavs in exactlv the same wav because the total radiative transition
probability for a state with quantum numbers n 0 , [ 0 is f(n 0 , l0 ) = n 0 f. The decay
of the angular momentum approaches the classical law only for [0 >> 1.
.. c _, '- " .
~u•• ~ ~~~~U VH Vl ~um b~ c aw 100
Coulomb potential (- · Ze"/r) moves m a circular orb1t m the absence ot rad1at10n

reaction.
(,.) .'>hom thM hoth tlw "n.-1 "nanbr-mnmPntnm P<lll"tion< (1 h 11\ "n.-1
(16.16) lead to the solution for the slowly changing orbit radius,
t
r ~f) ro ~L~CT)"
T
where r0 1s the value ot r(t) at t - U.

(b) For circular orbits in a Bohr atom the orbit radius and the principal quantum
-'- n "r" '- ' hv r n2 a,J7. Tf the ''i"n nroh"hilitv for transi-
c.. I.
\'
1\"
•J •
- J..u. -'- •'-
~-
· ~
·' y ...
'
agrees with that found in Problem 14.21.
(c) From part a calculate the numerical value of the times taken for a mu meson
of m"<< m - ?07m to fall from a circular orbit with nrincinal nuantum num-
ber n1 = 10 to one with n 2 = 4, and n 2 = 1. These are reasonable estimates
of the time taken for a mu meson to cascade down to its lowest orbit after
caoture bv an isolated atom.
776 Chapter 16 Radmhon Dampmg, Classical Models of Charged Particles G
16.3 An electron moving in an attractive Coulomb field ( -Ze 2 /r) with binding energy
E and angular momentum L has an elliptic orbit,
2 2
-1 = -Ze 2-m [ 1 + /1 - 2EL cos(ll - 11 ) ]
2 4 0
r L \1 Zem
The eccentricity of the ellipse is given by the square root mnltjplying the co~ine.
(a) By performing the appropriate time averages over the orbit, show that the
secular changes in energy and angular momentum are
dE =
312
2 Z em
3 8 112
E
312
(:1 _ 2EU )
dt 3 c3 L5 - Z 2e4 m
dL 25!2 ze4 E312
dt
(b) If the initial values of E and L are q, and l r~o show that
E(L) =
Z 2 e4 m
2L
2
[1 - (-LL )3] + -Lo L
0
Eo
Calculate the eccentricity of the ellipse, and show that it decreases from its
initial value as (UL 0 ) 312 , showing that the orbit tends to become circular as
time goes on.
(c) Compare your results here to the special case of a circular orhjt of Problem
Hint: In performing the time averages make use of Kepler's law of equal areas
(dt = mr 2 d()/L) to convert time integrals to angular integrals.
16.4 A classical model of an electron is a spherical shell of charge of radius a and total
charge e.
(a) Using (16.30) for the "mass" M(w) and the angular average of e'wR1'/R, show
that
where g= 2walc, and m = m 0 + 2e 2 /3ac 2 is the physical mass of the electron.

(b) Expand in powers of the frequency (g) and show that, to lowest nontrivial
order, M( w) has a zero in the upper half-plane at wr = i, where r = 2e 2!3mc3 .
What is the physical significance of such a zero?
(c) For the exact result of part a, show that the zeros of wM( w), II any, are defined
by the two simultaneous equations proportional to the real and imaginary
e-y sin x - x(1 - a/cr) = 0
where x = Re g and y = Im g. Find the condition on the radius a such that

wM(w) has no zeros in the upper half-w-plane. Express the condition also in
terms of the mechanical mass mo for fixed physical mass. What about zeros
and lor singularitie~ in the lower half plane'?
16.5 The particle of Problem 16.4 is initially at rest in a spatially uniform but time-
varying electric field E(t) E 0 @(t).
Ch. 16 Problems 771
(a) Show that J!s speed m tlie d1rection of the field is given by the integral
where T = ct/2a.
(b) From the analytic properties of M({;) established in part c of Problem 16.4,
show that v(t) = 0 fort :S 0 (no preacceleration).
16.6 A particle of bare mass m 0 and chargee has a charge density, p(x) = e e-d"l47ra 2 r.
1
(a) Show that the charge form factor IS J(k) - (1 + k 2a2 )
(b) Show that the mass, (16.33), is
m- m 0 +
2a
(c) Show that the zeros of M( w), (16.34), in the complex w plane, are given by
112
wT = -i(cT!a)[1 ± (1 - 2a!cT) ]
(d) Find the trajectories of the roots in the complex w plane for m 0 > 0 and
m 0 < 0. Find the limiting form for the roots when a!cT << 1 and alcT >> 1.
Discuss.
16.7 The Dirac (1938) relativistic theory of classical point electrons has as its equation
of motion.
dp I" = pexl + F'·ad

T
whete p~ is the particle's 4-tnonrenturn, r is t11e par tide's p1oper tirne, and p;d is
the covariant generalization of the radiative reaction force (16.8).
Using the requirement that any force must satisfy F,.p'" - 0, show that
16.8 (a) Show that for relativistic motion in one dimension the equation of motion of
Problem 16.7 can be written in the form,
. 2 c? ( .. PP"2 )
p - 3mc3 p - p 2 + m 2 c2
where p is the momentum in the direction of motion, a dot means differen-

tiation with respect to proper time, and f( T) is the ordinary Newtonian force
as a function of proper time.
(b) Show that the substitution of p = me sinh y reduces the relativistic equation
to the Abraham Lorentz form (16.9) in y and T. W1ite down the geneia!
solution for p( T), with the initial condition that
p(T) =Po at T=O

16.9 (a) Show that the radiation reaction force in the Lorentz-Dirac equation of
Problem 16.7 can be expressed alternatively as
(h) The relativistic generalization of (16 10) can be obtained by replacing

d 2p"ldT 2 by gvAdF~x'ldT in the expression for F;;ct. Show that the spatial
part of the generahzatwn of (16.10) becomes
dp
dt
y
2
dv X (v X F) J
where F is the spatial part of F:;'x,ly. For a charged particle in external electric
and magnetic fields F is the Lorentz force.
Reference: G. W. Ford and R. F. O'Connell, Phys. Lett. A 174, 182 (1993).
16.10 l he Abraham-Lorentz equatwn of motion (16.9) can be replaced by an integra-
differential equation if the external force is considered a function of time.
(a) Show that a first mtegral of (16.9) that ehmmates the possibility of "runaway"
(b) Show that a Taylor series expanswn ol the force for small T leads to
n d"F(t)
n=O dt"
The approximate equation (16.10) contains the first two terms of the infinite
senes.
(c) For a step-function force in one dimension, F(t) = F0 0(t), solve the
integra differential equation of part a for the acceleration and 'eloeity fot
t < 0 and t > 0 for a particle at rest at t = - oo. Plot mal F0 and mvl f~T in
units of tiT. Compare with the solution from (16.10). Comment
16.11 A nonrelativistic particle of chargee and mass m is accelerated in one-dimensional
motion across a gap of width d by a constant electric field. The mathematical
idealization is that the particle has applied to it an external force ma while its
comdinate lies in the inlet val (0, d). Without radiation damping the particle, hav-
ing initial velocity u 0 , is accelerated uniformly for a time T = (-vola) +
\/(v5!cl) + (2dla), emerging at x = d with a final velocity u1 = Vv6 + 2ad.
With radiation damping the motion is altered so that the particle takes a time
1 ' to cross the gap and emerges w1th a velocity v;.
(a) So1 ve the integro-diffet ential equation of rnotion, including danrping, ass urn-
ing T and T' large compared to T. Sketch a velocity-versus-time diagram for
the motion with and without damping.
(b) Show that to lowest order in T,
T'=
(c) Verify that thg sHrn of thg energy radiated and the ctlange in the particle's
kinetic energy is equal to the work done by the applied field.
16.12 A classical model for the description of collision broadening of spectral lines is
that the oscillator is interrupted by a collision after oscillating for a time T so that
the coherence of the wave train is lost.
(a) Taking the oscillatm used in Section 16.7 and assuming that the probability
that a collision will occur between time Tand (T + dT) is (ve-" 1 dT), where
Ch.16 Problems 773
vis the mean collision frequency, show that the averaged spectral distribution
IS
dl(w)
dw
--
-
21r
Io
( w - w0 )"
r+
+ (r
2v
r
2+ v
so that the breadth of the line is (2 v + f).

(b) For the sodium doublet at 5893 A the oscillator strength is f = 0.975, so that
the natural width is essentially the classical value, D. A = 1.2 X 10- 4 A. Esti-
mate the Doppler width of the line, assuming the sodium atoms are in ther-
mal equilibrium at a temperature of 500K, and compare it with the natural
• • . . . -lh -? • . .
W!Ulll. ;a _ClOSS 11 lJl .V UU , - CHv vv.,ooov"
breadth ot the sodmm dolil5Iet as a Tuncrion or me pressure or me sooium

vapor. For what pressure is the collision breadth equal to the natural
hrPo<lth'l ThP. OnnnlPT hrParith?
. .
unuerrne-
. .
.LO •.L-' A smgu,, parncte -ura:!rdfl}'ll<OU <OI<OUUv UUU DQC UdO
a dipole moment given Dy

iwt
I' o"
In\ <:hmu thot thP tAtol ,. 0
~TACO OP<',;An ~On 1-
-r
"'
21T
O't\ w) L lWtx(W) -t C.C.J
(h) T kino rmlv thP fRrt< thRt oll thP I rnA.IP · of oscillation must have some
. -•
.l
.• , "b ?
.-l •• - ·
?
••
-·'"' ..1- .\
., ~J '
ot
<Lrr
omorh thP frpp_
0
. ·;, <L
va1uc 1 -e 1mw ) "' 1111;11 '""'f ,--:rnu " ' " uov vwoo
dipole sum rule,
r o
0"1 (w) dw = 2rrze
-
me
-
2
(The discussion of Kramers-Kronig dispersion relations in Chapter 7 lS

clearly relevant.)
Appendix on Units and Dimensions
The question of units and dimensions in electricity and magnetism has exercised
a great number of physicists and engineers over the years. This situation is in
. •t r1 '" ;th ·~ ,J, ,] ·~ . } ;t. ,f 1. .t}
(centimeter or meter), mass (gram or kilogram), and time (mean solar second).
'T'} · •rh•on•' ' th >t th m ·~h~n· ·~I nn;tc mPrP rJ,finPrl mhPn thP ;rJ,~ nf
. r -r
auSOIULe scanuarus was a nove• concepL IJUSL utaure rovv), auu urey were urgeu
on the professional and commercial world by a group of scientific giants (Borda,
Ld)JidU:;, dllU • vy LllC LllllC LJIC !JJ_UUU:;lll Ul ClC\.-Ll' ;[1\.- Ullll> dlV>C
there were (and still are) many experts. The purpose of this appendix is to add
as little heat and as much light as possible without belaboring the issue .
.I unus ana utmenswns; nastc unus ana uerivea unit~
The arbitrariness in the number of fundamental units and in the dimensions of

any physical quantity in terms of those units has been emphasized by Abraham,
Planck, Bridgman,* Birge,t and others. The reader interested in units as such will
rln mPll tn hm;Jior mith thP PvrPliPnt nf "rtirll'< hv Rinrl'
The desirable features of a system of units in any field are conve~ience and
claritv. For examnle theoretical ohvsicists active in relativistic auantum field
theory and the theory of elementary particles find it convenient to choose the
universal constants such as Planck's quantum of action and the velocity of light
+. ~ J' • ,], r1 ,f . d. 'T'h .... ,(
units (called "natural" units) has only one basi; unit, ~ustomarily ;ho~en to be
A 11 · · ·1. no>th · t;~ •r fnr~. w >t~
, , , •m ooPrl
-~·
in terms of this one unit and have dimensions that are powers of its dimension.
There is nothing contrived or less fundamental about such a system than one
JUVUIVHig Llle lllt:Lt:l, Ult: . 0 1 illll, illlU Lllt: d> Ud>li.O UUll>. H 1> lllt:lt:Ly d
matter of convenience!
A J J. l. 'J .l. 1- J. J. 'J. J -
"' ~·~ w ~~ ou•~ u~~u u •o•~ u • ~ , ·~ ·~
pendent quantities, and derived umts or standards, which are dehned m both
• .-1. " "' • +h ~· th • r1 • • . ,( th 1- •
'b' 'J T ' '
umts. lfadltiOn reqmres that mass ~m), 1engm ~t), ana time ~t) De treatea as
basic. But for electrical quantities there has been no compelling tradition. Con-
siaer, ror exampre, me uni1 or currem. r ne imernaLionar ampere ~ror a 10ng
w. tlnagman, Ulmenswnal AnaLysis, r aJe umversJiy rress, 1'ew Haven, L 1 l' ~-" J.
'1'.
'R. T. Birge, Am. Phys. Teacher (now Am. J. Phys.), 2, 41 (1934); 3, 102, 171 (1935).
'In quantum field theory, powers of the coupling constant play the role of other basic units in doing
mmenswna1 analySis.
775
776 Appendix
period the accepted practical unit of current) is defined in terms of the mass of
silver deposited per unit time by electrolysis in a standard silver voltameter. Such
a unit of current is properly considered a basic unit, independent of the mass,
length, and time units, since the amount of current serving as the unit is found
from a supposedly reproducible experiment in electrolysis.
On the other hand, the presently accepted standard of current, the "abso-
lute" ampere "is that constant current which, if maintained in two ~traight par
allel conductors of infinite length, of negligible circular cross section, and placed
one metre apart in vacuum_ would produce between these conductors a force
equal to 2 10- 7 newton per metre of length." This means that the "absolute"
ampere IS a denved unit, since its definition is in terms of the mechanical force
between two wires through equation (A.4) below.* The "absolute" ampere is,
by this definition, exactly one-tenth of the em unit of current, the abampere.
Since 1948 the internationally accepted system of electromagnetic standards
has been based on the meter, the kilogram, the second, and the above definition
of the absolute ampere plus other derived units for resistance, voltage, etc. This
seems to be a desirable state of affairs. It avoids such difficulties as arose when,
in 1894, by act of Congress (based on recommendations of an international com-
mission of engineers and scientists), independent basic units of current, voltage,
and resistance were defined in terms of three independent experiments (silver
voltameter, Clark standard cell, specified column of mercury):'. Soon afterward,
because of systematic errors in the experiments outside the claimed accuracy,
Ohm's law was no longer valid, by act of Congress I
The Systcme International d'Unit6s (SI) has the unit of mass defined since
1889 by a platmum-mdmm kilogram prototype kept m Sevres, France. In 1967
the SI second was defined to be "the duration of 9 192 631 770 periods of the
radiation corresponding to the transition between the two hyperfine levels of the
ground state of the cesium-133 atom." The General Conference on Weights and
Measures in 1983 adopted a definition of the meter based on the speed of light,
namely, the meter is "the length of the distance traveled in vacuum by light during
a time 1/299 792 458 of a second." The speed of light is therefore no longer an
experimental number; it is, by definition of the meter, exactly c = 299 792 458
m/s. For electricity and magnetism, the Systcme International adds the absolute
ampere as an additional unit, as already noted. In practice, metrology laborato-
ries around the world define the ampere through the units of electromotive force,
the volt, and resistance, the ohm, as determined experimentally from the
Josephson effect (2e/h) and the quantum Hall effect (hle 2 ), respectively "
*The proportionality constant k 2 in (A.4) is thereby given the magnitude k 2 - 10 in the SI system.
The dimen£iOn£ of the ''absolute" ampere as distinct from its magnitude depend on the dimensions
assigR€Hi k2 . In the cmweRtior~.al ~I syf'tem of electromagH@tic uHits, @}@ctric curnmt (!)is arbitrarily
chosen as a fourth basic dimension. Consequently charge has dimensions It, and k 2 has dimensions
of mlr't- 2 If k 2 is taken to be dimensionless, then current has the dimensions m 1121112t- 1 • The ques-
tion of whether a fourth basic dimension like current is introduced or whether electromagnetic quan-
tities have dimensions given by powers (sometimes fractional) of the three basic mechanical dimen-
sions is a purely subjective matter and has no fundamental significance.
'See, for example, F. A. Laws, Electrical Measurements, McGraw-Hill, New York (1917), pp. 705-706.
'For a general discussion of SI units in electricity and magnetism and the use of quantum phenomena
to define standards, see B. W. Pelley, in Metrology at the Frontiers of Physics and Technology. eds.
L. Corvini and T. J. Quinn, Proceedings of the International School of Physics "Enrico Fermi," Course
CX, 27 June I July 1989, North-Holland. Amsterdam (1992), pp. 33 61.
Sect. 2 Electromagnetic Units and Equations 777
2 Electromagnetic Units and Equations

In discussing the units and dimensions of electromagnetism we take as our start-
ing point the traditional choice of length (l), mass (m), and time (t) as indepen-
dent, basic dimensions. Furthermore, we make the commonly accepted definition
of current as the time rate of change of charge (!- dq!dt). This means that the
dimension of the ratio of charge and current is that of time.* The continuity
equation for charge and current densities then takes the form:
To simplify matters we initially consider only electromagnetic phenomena in free

space. apart from the presence of charges and currents.
The basic physical law governing electrostatics is Coulomb's law on the force
between two point charges q and q', separated by a distance r. In symbols this
law is
The constant k 1 is a proportionality constant whose magnitude and dimensions

either are determined by the equation (if the magmtude and d1mens1ons of the
unit of charge have been specified independently) or are chosen arbitrarily in
order to define the umt of charge. W1thm our present framework all that IS de-
termined at the moment is that the product (k1 qq') has the dimensions (mz3t 2 ).
The electnc field E IS a denved quantify, customarily defined to be the force
per unit charge. A more general definition would be that the electric field be
numerically proportional to the force per unit charge, with a proportionality con-
stant that is a universal constant perhaps having dimensions such that the electric
field is dimensionally different from force per unit charge. There is, however,
nothing to be gained by this extra freedom in the definition of E, since E is the
first derived field quantity to be defined. Only when we define other field quan-
lilies may 1t be convement to msert d1menswnal proportwnahty constants m the
definitions in order to adjust the dimensions and magnitude of these fields relative
to the electnc field. Consequently, With no s1gmficant loss of generality the elec-
tric field of a point charge q may be defined from (A.2) as the force per unit
c arge,
q
E = k1 2 (A.3)
All systems of units known to the author use this definition of electric field.
For steady state magnetic phenomena Ampere's observations form a basis
for specifying the interaction and defining the magnetic induction. According to
Ampere, the force per unit length between two infinitely long, parallel wires
separated by a distance d and carrying currents i and r is
*From the point of view of special relativity it would be more natural to give current the dimensions
of charge divided by length. Then current density J and charge density p would have the same di-
mensions and would fu1m a natm al'' 4-vectOI. This is the choice made in a modiftcd Gaussian system
(see the footnote to Table 4, below).
778 Appendix
The constant k 2 is a proportionality constant akin to k 1 in (A.2). The dimension-

less number 2 is inserted in (A.4) for later convenience in specifying k 2 • Because
of our choice of the dimensions of current and charge embodied in (A.l), the
dimensions of k 2 relative to k 1 are determined. From (A.2) and (A.4) it is easily
found that the ratio k 1lk2 has the dimension of a velocity squared Wt 2 ). Fur-
thermonl, by comparison of the magnitude of the two mechanical forces (A.2)
and (A.4) for known charges and currents, the magnitude of the ratio k 1 /k 2 in
free space can be found. The numerical value is closely given by the square of
the velocity of light in vacuum. Therefore in symbols we can wnte
(A.5)
where c stands for the velocity of light in magnitude and dimension.

The magnetic induction B is derived from the force laws of Ampere as being
numerically proportional to the force per unit current with a proportionality
constant a that may have cer Lain dimensions chosen for convemence. I hus for a
long straight wire carrying a current T the magnetic induction B at a distance d
has the magnitude (and dimensions)
(A.6)
The dimensions of the ratio of electric field to magnetic induction can be found
&om (A.l), (A.3), (A.5), and (A.6). The result IS that (EIB) has the d1menswns
The third and final relation in the specification of electromagnetic units and
dimensions IS Faraday's law of mduction, which connects electric and magnetic
phenomena. The observed law that the electromotive force induced around a
circuit is proportional to the rate of change of magnetic flux through it takes on
the differential form
aB
V X E + k3 - = 0 (A.7)
· at
where k 3 IS a constant of proportionality. Since the dimensions of E relative to
B are established, the dimensions of k 3 can be exp1essed in terms of previously
defined quantities merely by demanding that both terms in (A.7) have the same
dimensions. Then it is found that k 3 has the dimensions of a 1 • Actually, k 3 is
1
equal to a • This is established on the basis of Galilean mvanance m SectiOn
5.15. But the easiest way to prove the equality is to write all the Maxwell equa-
tions in terms of the fields defined here.
(A.8)
V X E + k3 -- 0
· at
V ·B = 0
Sect. 3 Various Systems of Electromagnetic Units 779
Then for source-free regions the two curl equations can be combined into the
wave equation,
-o (A 9)
Tht:! vt:!locity of propagation of thg waves de>cribed by (A 9) is related to the
2
k_k a
Combining (A 5) with (A JO), we find
a
an eqnality holding for both magnitude and dimensions
3 Various Systems of Electromagnetic Units

The various systems of electromagnetic units differ in their choices of the mag
mtudes and drmenswns of the vanous constants above. Because of relatiOns (A.5)
and (A.ll) there are only two constants (e.g., k 2 , k,) that can (and must) be
chosen arbnranly. It is convenient, however, to tabulate all four constants
(kh kz, a, k.~) for the commoner systems of units These are given in Table 1 We
note that, apart from dimensions, the em units and SI units are very similar,
differing only in various powers of 10 in their mechanical and electromagnetic
units. The Gaussian and IIeaviside Lorentz systems differ only by factors of4lf.
Table 1 Magnitudes and Dimensions of the Electromagnetic Constants

for Various Systems of Units
The dimensions are given after the numerical values. The symbol c stands for the
velocity of light in vacuum (c = 2.998 X 10 10 cm/s = 2.998 X 108 m/s). The first four
systems of units use the centimeter, gram, and second as their fundamental units of
length, mass, and time (1, m, t). The Sl system uses the meter, kilogram, and second,
plus current (I) as a fourth dimension, with the ampere as unit
1 2
Electrostatic ( esu) c 2 (t 2 l 2) I 1
ElectronragtJetic (etna) c2(12 l 2) 1 1 1
GaYs sian 1 c 2(t21 2) c(lt 1) ('
1
(tl I)
1
Heaviside-Lorentz - (tll-2) c(lt- 1) c-l(tl-')
47T 47Tc 2
SI
1
-- = w-7c2 /Lo = w-7 1
47TEo 47T
(mt't- 4 r 2 ) (mlt- 2 r 2 )
780 Appendix
Only in the Gaussian (and Heaviside-Lorentz) system does k 3 have dimensions.

It is evident from (A.7) that, with k 3 having dimensions of a reciprocal velocity,
E and B have the same dimensions. Furthermore, with k 3 = c-r, (A.7) shows
that for electromagnetic waves in free space E and B are equal in magnitude as
we.
For SI units, (A.10) reads 1/(JLoEo) = c 2 . With c now defined as a nine-digit
number and k2 JLoi47T = 10- 7 Him, also by definition, 107 times the constant
k 1 in Coulomb's law is
10
= c2 = 89 875 517 873 681 764
4 7TEo
an exact 17-digit number (approximately 8.9876 X 1016). Use of the speed of light
without error to define the meter in terms of the second removes the anomaly
in Sl units of having one of the fundamental proportionality constants Eo with
experimental errors. Note that, although the right-hand side above is the square
of the speed of light, the dimensions of c0 (as distinct from its magnitude) are not
seconds squared per meter squared because the numencal factor on the left has
the dimensions of JL0 1 . The dimensions of 1!E0 and fLo are given in Table 1. It is
conventional to express the dimensions of Eo as farads per meter and those of fLo
as henrys per meter. With k 3 = 1 and dimensionless, E and cB have the same
dimensions in SI units, for a plane wave in vacuum they a1 e equal in magnitude.
Only electromagnetic fields in free space have been discussed so far. Con-
sequently only the two fundamental fields E and B have appeared. There remains
the task of defining the macroscopic field variables D and H. If the averaged
electromagnetic properties of a material m<Jdium ar<J d<Jscrib<Jd by a macroscopic
polarization J> and a magnetization M, the general form of the defimuons of D
and Hare
(A.l2)
H = - B - A'M
fLo
where E0 , fLo, A, A' are proportionality constants. Nothing is gained by making D
and P or Hand M have different dimensions Consequently a and A' are chosen
as pure numbers (A = A' = 1 in rationalized systems, A = A' = 47T in unration-
alized systems). But there is the choice as to whether D and P will differ in
dimensions from E, and H and M differ from B. This choice is made for conve-
nience and simplicity, usually to make the macroscopic Maxwell equatiOns have
a relatively simple, neat form. Before tabulating the choices made for different
systems, we note that for linear, isotropic media the constitutive relations are
alwa s written
(A.13)
relative permittivity of a substance (often called the dielectric constant) is defined

as the dimensionless ratio (E/E0 ), while the relanve permeability (often called the
permeability) is defined as (JLI JLo)-
Table 2 displays the values of Eo and fL<h the defining equations forD and H,
the macroscopic forms of the Maxwell equations, and the I ,orentz force equation
Table l De:Rnitiefls ef eo. ;±o. 0, H, Maereseerie Max nell Egtiatiens, and Lmentz Fmec Egtintioll in Vmious Systems of Units
Wht.'r~ H~G~ssar) tbe dimsnsifiAS ef EjHaAtitier are given in flarentbeses. The SjmBel e stands l"or the 'el5eit} ef light in , aet1t1m ••ith diJJJCJJSions (h 1
).
")" --v;-n- """Macroscopic Maxwell Equations umt Lnarge

' "
aD aB
Electrostatic 1 c-2 D~E I 41rP V·D = 41Tp V X H = 47TJ + - VxE+-::.;_0 V·B ~ 0 E+vXB
(esu) (r'z-') H""' c2 B- 47TM at at
1 aD aB
Electromagnetic c-2
I D =--;:;- E + 47TP V·D = 47Tp V X H = 47TJ + dt VxE+-=0 V·B ~ 0 E+vXB
2 c· or
(emu) (Fl- )
H=B-41TM
47T 1 aD
Gaussian I I D = E + 47TP V · D = 41Tp VXH=-J+-- VxE+!JB=O V·B ~ 0 E+.!xB
c c Jl c c
H = B- 47TM "
v
Hcavisidc- 1 I D=E+P V·D ~ p VxH=-1 ( J +aD)
- V X E +.!_DB= 0 V·B ~ 0 E +- x B
Lorentz H~B-M
c at r at c
J(f aD
S1 --
2 47TX 10 ' D=f0 E+P V·D ~ p V X H- J I - VxE+JB=c-0 V·B ~ 0 E+vxB
or
41TC
(Jlt4m 'I ') (mll- 2 r 1
) H--B-M
I "
"'
782 Appendix
m ille live common systems of umts of Table 1. For each system of units the
continuity equation for charge and current is given by (A.l), as can be verified
from the first pair of the Maxwell equations in the table in each case.* Similarlv.
in all systems the statement of Ohm's law is J = aE, where a is the conductivity.
4 Conversion of Equations and Amounts Between Sf Units

and Gaussian Units
The two systems of electromagnetic units in most common use today are the SI
and Gaussian systems. The SI system has the virtue of overall convenience in
Table 3 Conversion Table for Symbols and Formulas
The symbols for mass, length, time, force, and other not specifically electromagnetic
quantities are unchanged. To convert any equation in SI variables to the corresponding
" · in r."""i"n nnont;tiP< nn hnth, oic1~< nf thP PnnMinn thP rPIPvont
symbols listed below under "SI" by the corresponding "Gaussian" symbols listed on
the left. The reverse transformation is also allowed. Residual powers of fLoEo should be
,];m;notPrl ;n fovM nf thP <nPPrl nf Enht I ,,2 < 1\ <;:;n~P th·, lPnnth onrl t;mP
are unchanged, quantities that differ dimensionally from one another only by powers of
length and/or time are grouped together where possible.
Quantity Gaussian SI
Velocity of light c (fLo Eo) - 112

Electric field (potential, voltage) E(<t>, V)/Y47TE0 E(<t>, V)
Displacement ~D D
~L
'6'
. ". ' , , " . 'J,
,~-.I
on~·
• r
, , n\
' J
. I T
, f ~\
, ' J
current, polarization) "'~'
Magnetic induction ~B B
Magnetic field HtV47TfLo H
Magnetization ~M M
~ . ..
-=:rmrm:=vrry "<'lltou u
Dielectric constant EoE E

Magnetic permeability ILoiL u
Resistance (impedance) R(Z)i47TE0 R(Z)
Inductance Li47fEo L
CaPacitance 47TEnC c
c - 2.997 924 58 X 108 m/s

to O.OJ · J0 I Q ••• A lU rm
1-'o = 1.256 637 0 ... X 10 6
Him
I
I -~ .. :-rr
'I <o
ou ocu u >y>ecJJJ Ul uum HJ vwcu 1> uy '

(1/c)(dq/dt). Then the current density J in Table 2 must be replaced by cJ, and the continuity equation
is V • J + (l!c)(ap/at) = 0. See also the footnote to Table 4.
Sect. 4 Conversion of Equations and Amounts Between SI Units and Gaussian Units 783
~~kO.
,_,. _,
• ,-.
•v ~~·~· £
.£ ~'-
'"y.
•u~·~ ~ ~~ -~
·~~·~ w• ~ '-' 'Y
The table is arranged so that a given amount of some physical quantity, expressed as so
manv SI or Gaussian units of that nuantitv. can be exnressed as an enuivalent number
of units in the other system. Thus the entries in each row stand for the same amount,
expressed in different units. All factors of 3 (apart from exponents) should, for accurate
work, be replaced by (2.997 924 58), arising from the numerical value of the velocity of
li<>ht. For examnle in the row for disnlacement (D) the entrv (12'1T X 105 ) is actuallv
(2.997 924 58 X 4'7T X 105 ) and "9" is actually 10- 16 c 2 = 8.987 55 .... Where a name
for a unit has been agreed on or is in common usage, that name is given. Otherwise,
one merely reads so many Gaussian units, or SI units.
Physical Quantity Symbol SI Gaussian
T Pnath I 1 meter ( m) 102 · 1eters (em)

3
Mass m 1 kilogram (kg) 10 grams (g)
Time t 1 second (s) 1 second (s)
Frequency v 1 hertz (Hz) 1 hertz (H;)
FnrrP ·"· IN\ 1 n5 rlunP<
" 1
~}p
Work
1 joule (J) 10" ergs
Energy
Power 1 watt (W) Hf ergs s- 1
Charge q 1 coulomb (C) 3 X 109 statcoulombs
Charge density p 1 cm- 3 3 X 103 statcoul em_,
Current I 1 ampere (A) 3 X 109 statamperes
Current densitv J 1 Am -L 3 X 100 statamn em -
Electric field E 1 volt m- 1 (Vm- 1 ) l3 X 10-4 statvolt em - 1
Potential <!>, v 1 volt (V) 1
300 statvolt
Polarization p 1 cm- 2 3 X 105 dipole moment cm- 3
Displacement D 1Cm -L
12'1T X 10 statvolt em
(statcoul cm- 2 )
Conductivity (]' I siemens m - 1 9 X 109 s-1
KeS!Stance K. 1 onm \") 9 X. lU s em
Capacitance c 1 farad (F) 9 X 10 11 em
Magnetic flux cf>,F 1 weber (Wb) 10" gauss cm2 or maxwells
. . -'- n '~' n4 ',-.'
"'~5"~"c ~
' oc"a \ ' J •v 0 nuoo \ ~)
JVIagneuc uelll n lAm '+'1T X. lU · oersreo \ ue J
Magnetization M 1 Am-' w-3 magnetic moment cm- 3
Inductance* L 1 henry (H) ~X 10- 11
*There is some confusion about the unit of inductance in Gaussian units. This stems from the use
by some authors of a modified system of Gaussian units in which current is measured in
'
"' ' '5' m \'
;,;
JWC~ Co oO ueuu~u :o WC wuucou "b' > L\"''"'J UO WO CHC\!;J U zLo , WC
choice of current defined in Section 2 means that our Gaussian unit of inductance is equal in
magnitude and dimensions (t 2 / - ' ) to the electrostatic unit of inductance. The electromagnetic
current I is related to our Gaussian current I bv the relation I ~ {1/c)I. From the enerev
definition of inductance, we see that the electromagnetic inductance Lm is related to our Gaussian
inductance L through L'" ~ c2 L. Thus L'" has the dimensions of length. The modified Gaussian
.c .c' ·" ' ~c .c -'•
'b'
'
UCLWOOU UUH> U\
\CdUO \Lm!C)\Ulm!U<)- U\C
"
1 henry ~ ~ X 10~ 11 Gaussian (es) unit~ 10 9 emu
784 Appendix
practical, large-scale phenomena, especially in engineering applications. The

Gaussian system is more suitable for microscopic problems involving the elec-
trodynam1cs of mdividual charged particles, etc. Previous editions have used
Gaussian units throughout, apart from Chapter 8, where factors in square brack-
ets could be omitted for the reader wishing Sl units. In this edition SI units are
employed exclusively in the first 10 chapters. For the relativistic electrodynamics
of the latter part of the book, we retain Gaussian units as a matter of convenience.
A rPmin<lPr-nf thP nnit, L • "'"<l ~- ot th~ tn~ nf loft_honrl ~on'
' ' • JC •J " -o '
with the designation, Chapter Heading-SI or Chapter Heading-G. Some may
feel it awkward to have two systems of units in use, but the reality is that scientists
L " • T ~T • ' ·, r
,,..n_..-,uu.L ~u- .UU.U.LJ .tU.L.L5uu5'"-'• -uo~ U.L.U I dlC: ldlC:lJ U:SC:U .lUI C::lC:l,lJ -
netic interactions in quantum mechanics, but atomic or Hartree units are, and
similarly in other fields.
1 ables :; and 4 are designed tor general use m conversiOnrrom one system
to the other. Table 3 is a conversion scheme for symbols and equations that allows
the reader to convert any equation from the Gaussian svstem to the SI svstem
and vice versa. Simpler schemes are available for conversion only from the SI
system to the Gaussian system, and other general schemes are possible. But by
L · :1ll mc>rh~nirlll n11:1ntiti1"' llnrh:mu<e<l th<> · in T~hl<> '1. ollmuo th~
straightforward conversion of quantities that arise from an interplay of electro-
magnetic and mechanical forces (e.g., the fine structure constant e 2 /hc and the
-1, , r 2 A 2/,_\ ·,1.. · · -' ._, ~-'-' '·
'i' -, p •; Lauvuo. >auH..,-. "a
conversion table for units to allow the reader to express a given amount of any
physical entity as a certain number of SI units or cgs-Gaussian units.
Bibliography
ABRAHAM, M., AND R. BECKER, Electricity and Magnetism, Blackie, London (1937),
.L . ,_ O<L ro .. .c 7'1. -1 • . . .. D .A T
'
, Theorie der F/ektrizitdt, Band II, Elektronentheorie, Teubner, Leipzig (1933).
ABRAMOWITZ, M., AND L A STEGlJN, eds., Handbook of Mathematical Functions, U.S.
~ v . •L '>nU\
<
.C 0o
"
UUO¥UUVO UCUHUUOUJ_\'"'~}o " r~V•¥<
0 <0¥ •vu' \""~j•
ADLER, R. B., L J. CHu, AND R. M. FANO, Electromagnetic Energy, Transmission and

Radiation, Wiley, New York (1960).
. ..T T1. l'
'£
. • 1 7'1.
'"•J
• ~D .
"J
..
. .
~- " "' . •
-' T T•.
'J
.n.
ford (1965).
ALFVEN, H., AND C.-G. FALTHAMMAR, Cosmical Electrodynamics, 2nd edition, Oxford
' L ' n. "nL~\
'J ~· ' \· -~,.
ANDERSON, J. L, Principles of Relativity Physics, Academic Press, New York (1967).

ARFKEN, G., AND H. J. WEBER, Mathematical Methods for Physicists, 4th edition, Aca-
"· . . D. "-L V, ,1, II 00';\
' ' ' ' • J'
ARGENCE, E., AND T. KAHAN, Theory of' Waveguides and Cavity Resonalors, Blackie,
London (1967).
A U' D .I. , . . vo D • f10U\
L' ,, ' 0 ' ' • :· -' !'
, Relativistic Point Dynamics, Pergamon Press, Oxford (1972).
AsHCROFT, N. W., AND N.D. MERMIN, Solid State Physics, Holt, Rinehart, and Winston,
"-L v. "'' {107t;\
BAKER, B. B.~ AN~·E. T. COPSON, Mathematical Theory of Huygens' Principle, 2nd edi-
tion, Oxford University Press, Oxford (1950).
BALDIN, A. M., V. 1. GoL'DANSKII, AND 1. L ROZENTHAL, Kinematics of Nuclear Re-
actions, Pergamon Press, New York (1961).
BARUT, A. 0., Electrodynamics and Classical Theory of Fields and Pariicles, Macmillan,
New York (1964); Dover reprint (1980).
BATEMAN MANUSCRIPT PROJECT, Higher Transcendental Functions, 3 vols., edited by
A. Erdelyi, McGraw-Hill, New York (1953).
, Tables of Integral Transforms, 2 vols., edited by A. Erdelyi, McGraw-Hill, New
York (1954).
BEAM, W. R., Electronics of Solids, McGraw-Hill, New York (1965).
BEKEFI, G., Radiation Processes in Plasma, Wiley, New York (1966) .
. .
, r. U., .,.., vudLttUft tU tft~ 1 lt~Uf y VJ ·n -uau, -o·
Cliffs NJ (1942): Dover reorint (1976).
BrEBERBACH, L, Conformal Mapping, Chelsea, New York (1964).
, "-· J., r. J. , ANJ.J ~. VV. <KU' ·, 1 It~£- unu "" "'"cu'
Solution of Electric and Ma~netic Fields Wiley, New York (1992).
BLATT, J. M., AND V. F. WEISSKOPF, Theoretical Nuclear Physics, Wiley, New York
\D.J"-)•
BOHM D. The Soecial Theorv of Relativitv. Benjamin New York (1965); Addison-
Wesley, Reading, MA (1989).
, "·· • !llUll ut : ~u~u¥•¥·' uuuugu , 1\./;l• ~u"""' V
Mat.-fys. Medd. XVIII No.8 (1948).

7!01l0:
786 Bibliography
u. . _,
u
-·
n.•
·'t Vj
1'~
·r
''L "' . ' .
n
JO '"' , V<' ~vu.
"""' ""w
I Oft<.~"""')·
BOTTCHER C. J. F. Theorv of FINtri,· Pnlnri7ntinn ~~
n ;
'
TT
., ' r R A , n • ,.-, T "
, DL
·o '"
Scattering by Simple Shapes, North-Holland, Amsterdam (1969); reprinted (1987).
-·
BRILLOUIN, L., Wave Propagation and Group Velocity, Academic Press, New York
(1960).
BUDDEN, K. G., The Propagation of Radio Waves, Cambridge University Press, New York
(1985); paperback, corrected (1988).
BYERLY, W. E., Fourier Series and Sphericalllarmonics, Ginn, Boston (1893); also Dover
reprint.
CAIRO L. AND T. KAHAN Variational Terhnim1P' in Til. T nn..lnn
11 Q(';~)
~nA 1 rnysics, university 01 <.-nicago rress, <.-nicago (1'1oU).

l ..:J,, ..._H.<-..>
CHAKRAVARTY, A. S., Introduction to the Magnetic Properties of Solids, Wiley, New York
(1980).
CHURCHILL, R. V., AND J. W. BROWN, Fourier Series and Boundary Value Problems, 5th
edition, McGraw-Hill, New York (1993).
CLEMMOW, P. C., The Plane Wave Spectrum Representation of Electromagnetic Fields,
Pergamon Press, Oxford (1966); reissued in IEEE/OUP series (1996).
CLEMMOW, P. C., AND J. P. DOUGHERTY, Electrodynamics of Particles and Plasmas,
Addison-Wesley, Reading, MA (19!\9); reprinted (1990).
~ -=
, , .. ~., • •c•u • "cv'_J' UJ
. YYUVtS, LnU euuJOn,
.. HODL rreSS, r!SCaiaway,
. I~J
(1'1'11).
1=< TT '~"' J..l' , ..... ., lJ .
" '/, J
VJ
DC
··r
.., ... ,,.1' •
"
TT'n ~L .U. _L ' '
" " ' • "~ •vu> \'"V:.!'
CORBEN, H_ C., AND P. STEHLE, Classical Mechanics, 2nd edition, Wiley, New York
(1960); reprinted (1974).
COURANT, R., AND D. HILBERT, Methods of Mathematical Physics, 2 vols., Wiley-
Intcrscience, New York (1962).
CowLING, T. G., Magnetohydrodynamics, 2nd edition, Hilger, London (1976).
CRAIK, D. J., Magnetism: Principles and Applications, Wiley, New York (1995).
CULLWTCK, E. G., Electromagnetism and Relativity, 2nd edition, Longmans, London
II 0~9\
~LD • L , . ,, 'V•w H>' , lJVVt:l, l~t:W I Ull'. ~l':l'tJ).
UF.'lNERY, r., AND A. ]',RZYWICKI, Mathematics tor Physicists Harper & Row New York
(1967).
DURAND, E., Electrostatique et Magm?tostatique, Masson, Paris ( 1953).
EINSTEIN, A., H. A. LORENTZ, H. MINKOWSKI, AND H. WEYL, the Principle of Relativity.
Collected papers, with notes by A. Sommerfeld, Dover, New York (1952).
FABELINSKII, I. L., Molecular Scattering of Light, Plenum Press, New York (1968).
FANO, R. M., L. J. CHU, AND R. B. ADLER, Electromagnetic Fields, Energy, and Forces,
Wiley, New York (1960).
FEYNMAN, R. P., R. B. LEIGHTON, AND M. SANDS, The Feynman Lectures on Physics, 3
VOIS., AOOlSOn~eSJ_ey,_JS_eadmg, MA (1'163).
FRENCH A P Sm>rin NP V mlc 11 Q(';R\
D D' .-~ '
'
., ·~
,'~I
:"1' • VJ HH<OJ, ""W llJ!l'.

'
uover repn~-
\l'I:JO);
FROHLICH H. Theorv of' iolo·····"" nyfnrrl Tln. · PrP« I (1 Q4Q\
GALEJS, J., Terrestrial Propagation of Long Electromagnetic Waves, Pergamon Press, Ox-
ford (1972).
GIBBS, W. J., Conformal Transformations in Electrical Engineering, Chapman & Hall,
London ( 1958).
GoLDSTEIN, H., Classical Mechanics, 2nd edition, Addison-Wesley, Reading, MA (1980).
Bibliography 787
GRADSHTEYN, L S., AND L M. RYZHIK, Tables of Integrals, Series, and Products, 4th
edition, prepared by Yu. V. Geronimus and M. Yu. Tseytlin, corr. enlarged edition by
A. Jeffrev. Academic Press New York (1980\.
nFr.Rnn-r S R ThP u. .11 r .
:in ·1l MPrh"nir' Vnl TV - ~;:, ~:
U~ll· --1 A
_, 11 n,;n\
, \ '!'
wo H, '-'• n .. , '"'"'-' L, U. C>UJ JUKr," UJ
- ' l'VJlJJ· ,
Amsterdam (1972).
HADAMARD, J., Lectures on Cauchy's Problem, Yale University Press, New Haven, CT
(1923); Dover reprint (1952).
HAGEDORN, R., Relativistic Kinematics, Benjamin, New York (1963, 1973).
HALLEN, E., F:lectromagnetic Theory, Chapman & Hall, London (1962).
HARRINGTON, R. F., Time-Harmonic F:lectromagnetic Fields, McGraw-Hill, New York
(1961).
HAus, H. A., AND J. R. MELCHER, Electromagnetic Fields and Energy, Prentice-Hall,
Englewood Cliffs, NJ (1989).
BEITLER, W., Quantum Theory of Radiation, 3rd edition, Oxford University Press, Oxford
(1 05Ll\
u. 1:7 D
., .I r'-'- .L
"'- ''FF·" .'
.... ~ ~-· . ,n . _TT-11
, ,-,_ -
• 'J
5'~ vUU \.AIH>, l ' J ~DIU).
lDA, N., AND J. 1'. A. l:IASTOS, t.tectromagnettcs and Catcutatwn o.t J•tetds, :Spnnger-
Verlag, New York (1992).
lWANENKO, D., AND A. SOKOLOW, Klassische Feldtheorie, Akademic-Verlag, Berlin
(1953), translated from the Russian edition (1949).
JAHNKE-EMDE-LOscH, Tables of lligher Functions, 6th edition, revised by F. Losch,
Teubner, Stuttgart; McGraw-Hill, New York (1960).
JEANS, J. H., Mathematical Theory of Electricity and Magnetism, 5th edition, Cambridge
University Press, Cambridge (1925); reprinted (1958).
JEFFREYS, H., AND B. S. JEFFREYS, Methods of Mathematical Physics, 3rd edition, Cam-
bridge University Press, Cambridge (1956); reprinted (1972).
JEFIMENKO, 0. D., Electricity and Magnetism, Appleton-Century Crofts, New York
(1966); 2nd edition, Electret Scientific, Star City, WV (1989).
JoHNSON, C. C., Field and Wave Electrodynamics, McGraw-Hill, New York (1965).
JONES, D. S., The Theory of Electromagnetism, Pergamon Press, Oxford (1964).
JORDAN, E. C. AND K. G. BALMAIN Electromagnetic Waves and RadiatinJ< Systems 2nd
·•rlitinn PncnticcP.-H"II Fn<>l, " ~I iff, NT (I 9nX)
v, , T
~
{<;:,
,. UT TT ,,\ D.
.,,/' n ··r <;'/, . _, "'-
.... " .. ;, 0~':,\
.,. '
' '
T
"· \
. . . .
, m., '"~ ,,cauenng OJ 1Agru anu J/trtEY Lt~Cuo ··c· lC 1\UU/UUOn, M.CctueuuC
Press, New York (1969).
KELLOGG, 0. D., Foundations of Potential Theory, Springer-Verlag, Berlin (1929); re-
printed by Ungar, New York and by Springer-Verlag (1967).
KILMISTER, C. W., Special1heory of Relativity, Selected Readings in Physics, Pergamon
Press, Oxford (1970).
KING, R. W. P., AND T. T. Wu, Scattering and Diffraction of Waves, Harvard University
Press, Cambridge, MA (1959).
KITTEL, C., Introduction to Solid State Physics, 7th edition, Wiley, New York (1996).
KRAUS J.D. Antennas 2nd edition McGraw-Hill New York (1988).
T ANnAT! T n ANn F M T ThP r l
""'- nf Fio>ld< Lith --1 ,-,_ . L L
"
..
,
~l
-'- --1 L 'Lf. • !..
' ·o
n ,,.;;
o• ~K
', " ""' Ul.,-1 'J' ,
.
_':-'' "m~ _\vu' }• .. . .
, ~tecrroaynamtcs OJ conunuous 1V1ecua, Lnu eumon, AuuJson-vvcsJCy, Keaumg,
MA (1984).
LIGHTHILL M. J. Introduction to Fourier Analysis and Generalised Functions Cambridge
University Press, Cambridge (1958).
788 Bibliography
• ,
r
•
~
.,
m
' Lll>LCLU<UH
"
~DV'J·
LIVINGSTON, M.S., AND J.P. BLEWETT, Particle Accelerators, McGraw-Hill, New York
(1 Qli?\
LORENTZ, H. A., Theory of Electrons, 2nd edition (1915); reprint, Dover, New York
(1952).
Low F. E. Classical Field Theorv: Electromn· nnrl r;, Wil.ev N .ew York
(1997).
MAGNUS, W., F. 0BERHETTINGER, AND R. P. SON!, Formulas and Theorems for the
Snecial Functions of' Mathematical Phvsics Sorinrrer-Verlao New York (1966\.
MARCUVITZ, N., Waveguide llandbook, M.I.T. Radiation Laboratory Series, Vol. 10,
McGraw-Hill, New York (1951).
MASON, M., AND W. WEAVER, T11e Electromagnetic Field, University of Chicago Press,
Chicago (1929); Dover reprint (1952).
MAXWELL, J. C., Treatise on Electricity and Magnetism, 3rd edition (1891), 2 vols., reprint,
Dover, New York (l\l:'l4).
MERMIN, N.D., Space and Time in Special Relativity, McGraw-Hill, New York (1968);
"' .~n. r .u·~' TTf1nom
M0LLER, C., The lheo~; of Rel~~ivi;y, 2nd edi;ion, Clarendon Press, Oxford (1972).
MONTGOMERY, C. G., R. H. DICKE, AND E. M. PURCELL, Principles of Microwave Cir-
· T ~ T> ~'
, H .. ' ' , T -' • ('
'J ,
'
, H -1 , ' =
u .. 1
\
1 n H>' " Q «'H .
~ ·~u;. ~.ov
available as a Dover reprint.

MoRSE, P. M., AND H. FESHBACH, Methods of Theoretical Physics, 2 Pts., McGraw-Hill,
"T. ' V ,.rlr (1 O<;~\
NORTHROP, T. G., The Adiabatic Motion of Charged Particles, Wiley-lnterscience, New

York (1963).
PANOFSKY, W. K. H., AND M. PHILLIPS, Classical Electricity and Magnetism, 2nd edition,
Addison-Wesley, Reading, MA (1962).
PAULI, W., Theory of Relativity, Pergamon Press, New York (1958), translated from an
article in the Encyklopedie der mathematischen Wissenschaften, Vol. V19, Teubner,
Leipzig (1921), with supplementary notes by the author (1956).
!'ENFIELD, 1'., AND t1. A. t1AUS, ctectroaynamzcs ii[7VTovmg"Mema, M.l.l'. Press, Cam-
bridge, MA (1967).
PERSICO, E., E. FERRARI, AND S. E. SEGRE, Principles of Particle Accelerators, Benjamin,
l~CW I UfK \""0")·
P6LYA, G., AND G. SzEG6, Isoperimetric Inequalities in Mathematical Physics, Annals of
Mathematics Series No. 27, Princeton University Press, Princeton, NJ (1951 ).
L V' • 1 L., '-'• ' . 'L£U'""'"- L, e>. n,, UC
KD~~, , AND VV. "· VJOTT' , 1v urnencat
Recipes, The Art of Scientific Computing, Cambridge University Press, Cambridge
(1 ORt>\
RAMO, S., J. R. WHINNERY, AND T. VAN DUZER, Fields and Waves in Communication
Electronics, 3rd edition, Wiley, New York (1994).
~ r PR w ~ .· 1 '' . • • SnPrinl r;, -1 nwl r, . • TPV ?n.-1 Pri;t;nn
~ . u ..1. "n~n
u 0 , ~··~
" -1
J}" u0< •v•a\v'~J·
RoBINSON, F. N.H., Macroscopic Electromagnetism, Pergamon Press, Oxford (1973).

ROHRLICH F. Classical Charrzed Particles ku· -WP,lPv n ~- MA (1Qii~ 1QQO\
RosENFELD, L., Theory of Electrons, North-Holland, Amsterdam (1951).
Rossi, B., lligh-Tonergy Particles, Prentice-Hall, Englewood Cliffs, NJ (1952).
ROSSI B. AND S. OLBERT Introduction to the Phvsics of Snace McGraw-Hill New York
(1970).
ROTHE, R., F. 0LLENDORFF, AND K. POLHAUSEN, Theory of Functions as Applied to
ftnJZineerinJZ f'roDtems, Technology Press Cambrids:>e MA 11933). Dover reorint !1961 ).
SADIKU, M. N. 0., Numerical Techniques in Electromagnetic.1·, CRC Press, Boca Raton,
''nM
•
~T
~ ~>77~,.
Bibliography 789
SARD, R. D., Relativistic Mechanics, Benjamin, New York (1970).

co " A A"· .I A ThPnru WilPv NPm V "rl< (1 0~?\
------,Applied Mathematics for Engineers and Scientists, 2nd edition, Van Nostrand,
New York (1965).
, ~tectromagnettc tzetas, t11aJSOeii,J"ew rorK (1'10..lJ.
SCHELKUNOFF, S. A., AND H. T. FRIIS, Antennas, Theory and Practice, Wiley, New Y ark
(1952).
ScHOTT, G. A., Electromagnetic Radiation, Cambridge University Press, Cambridge
(1912).
. .
J L t l VY JH<. 1 L, H . Jn., H " ' w .,. •c<w'" ty, -u111, nco JUJJ\. ~17UO)•
~EGRE, n., eu., r.xpertmenratJvuctear rnyszcs, vo1. 1, vvi1ey,1"ew rorK ]1'1:J-'J·

SJEGBAHN, K., ed., Beta- and Gamma-Ray Spectroscopy, North-Holland, Amsterdam;
Interscience New York (1955).
SiLVER, S., ed., Microwave Antenna Theory and Design, M.I.T. Radiation Laboratory
Series, Vol. 12, McGraw-Hill, New York (1949).
'-'L~JG~,J.~.,mte<V>VUVC , 'HH<,<''-' '"'
1U11\.~17JV}·
SLATER, J. C., AND N.H. FRANK, Electromagnetism, McGraw-Hill, New York (1947).
SMITH, J. H., Introduction to Special Relativity, Benjamin, New York (1965).
SMYTHE, W. R., Static and Dynamic Electricity, 3rd edition, McGraw-Hill, New York
(1969); reprinted (1989).
c. "'"" n A , m,,. ' ,;,.< A~orlPmi~ PrM< NPm VMlr (10~?\
' '
, Optics, Academic Press, New York (1904).
, Partial Differential Equations in Physics, Academic Press, New York (1964).
SoPER, D. E., Classical Field Theory, Wiley, New York (1976).
SPITZER, L., Physics of Fully Ionized Gases, 2nd edition, Interscience, New York (1962).
STONE, J. M., Radiation and Optics, McGraw-Hill, New York (1963).
"TRAN(; G Tm -'- . '" tn A ·" J " ,,_ . WPllP<lPv- 'laP PrP<" w .. n,.,Jpv
MA (1986).
STRATTON, J. A., Electromagnetic Theory, McGraw-Hill, New York (1941 ).
!AYLOR, J:C. r., AND J. A. WHEELER, ;-,paceume rnysics, Lna eaiiion, rreeman, :>an t'ran-
cisco (1992).
THIRRING W.E. Princinles of Quantum Electrodvnamics translated bv J. Bernstein
Academic Press, New York (1958).
THOMSON, J. J., Recent Researches in Electricity and Magnetism, Clarendon Press, Oxford
•1 on1\
\~v•~ J•
TITCHMARSH, E. C., Introduction to the Theory of Fourier Integrals, 3rd edition, Oxford
University Press, Oxford (1986).
TONNELAT, M.-A., Electroma!lnetic Theorv and Relativitv, Reidel, Holland (1966).
TRANTER, C. J., Integral Tran1jorms in Mathematical Physics, 3rd edition, Methuen, Lon-
don (1966).
"· n, ' .,
T
~ ·o
r-:.u.
-
'
u;n , " "
v .,,
v • • 1. f1 n£:A\.
~K.~
'
(1985).
VAN VLECK, J. H., Theory of Electric and Magnetic Susceptibilities, Oxford University
Press, Oxford (1932).
WAIT, J. R., Electromagnetic Waves in Stratified Media, Pergamon Press, Oxford (1962);
. in T'H'H'H /()T JP <PriP< (1 QQ;;\
. w. .I D, . ,h ,1.1
"' '
D A ., T'lo
•I ._,+ r-. '"·" m. ··o· ' "· ~-
'
London (1970).
WATSON, U. 1'1., 1 neory OT JJesset runctwns, LnO eaition, LamDriage University l'ress,
Cambridge (1952); reprinted (1995).
WERT, C. A., AND R. M. THOMSON, Physics of Solids, 2nd edition, McGraw-Hill, New
V ()TJ< (1 Q7n\
WHITTAKER, E. T., A History of the Theories of Aether and Electricity, 2 vols., Nelson,
790 Bibliography
London, Vol. 1, The Classical Theories (1910, revised and enlarged 1951), Vol. 2, The
I\1()denz Thet}lies 19()() 1926 (1953); reprinted hy Hatper Torehhooks, New York
(1960); Dover repnnt (1989).
WHITTAKER, E. T., AND G. N. WATSON, A Course in Modern Analysis, 4th edition,
Cambridge University Press, Cambridge (1950).
WILLIAMS, E. J., Correlation of certain collision problems with radiation theory, Kg!.
Danske Videnskab. Selskab Mat.-fys. Medd., XIII, No.4 (1935).
WooTEN, F., Optical Properties of Solids, Academic Press, New York (1972).
ZHao, P.-B., i\Untetical Analysis ofEteczronragnelic FieidJ, Springcr-\rcrlag, New York
(1993).
ZIMAN, J. M., Principles of the Theory of Solids, 2nd edition, Cambridge University Press,
Index
Abraham-Lorentz equation of motion, 748 electromagnetic, of electric charge and mag-

Dirac's relativistic generalization of, 771 netic monopole, 277
workable approximation to, 749, 772 of bound particle, slow change caused by radi-
Abraham-Lorentz model of electron, 750f ation damping, 750
difficulties with, 754-5 of multipole llelds, 432f
form factor in, 754 rate of radiation of, by oscillating electric di-
see also Classical charged particle pole, 451-2
Absorption, resonant, 310 Angular momentum density of the electromag-
of radiation, by earth's atmosphere, 467 netic field, covariant form of the conser-
by oscillator, 655, 768 vation law for, 608, 610
Absorption coefficient, definition, 310 Angular momentum operator, L, 428
of ideal gas, 466 commutation relations for, 429
of liquid water as a function of frequency, and other vector differential operators, identi-
315 ties involving, 428, 432, 441, 472
of sea water at low frequencies, 315 and vector spherical harmonics, 431
Acceleration, relativistic transformation of, Anomalous drsperswn, 310
569 Anomaly, of magnetic moment of electron and
Acceleration fields of charge in arbitrary mo- muon, 565
tion, 664 Antenna, as a boundary-value problem, 418
Action, Lorentz invariance of, 580 center-fed, linear, 416f
Addrhon of velocrhes, relahvrshc, SJOf Imcar, radrauon from m terms of multrpoles,
Addition theorem for spherical harmonics, 444f
110-1 radiation resistance of, 412
Adiabatic invariance, of flux through particle's short, linear, 412
orbit, 592f in wave guide, 392f, 404, 405
of magnetic moment of particle, 593 Apcr Lute in wave guide or cavity, effective di-
Admrttance, field defimtwn of, 288 pole moments of, 421f
Advanced Green function for wave equation, Arrival of a signal in a dispersive medium, 335f
245 Associated Legendre functions, 108
invariant expression for, 613-4 Attenuation, in optical fibers, 470-1
Airy integrals, in terms of Bessel functions, Rayleigh scattering limit for, 470
678 in resonant cavities, 3/lf
Alfven velocrty, 321 treatment by perturbalion of boundary condr-
Alfven waves, 319f tions, 366f, 374, 401-2
Ampere's law, 179 in wave guides, 363f
Angles of incidence, reflection and refraction, Attenuation cogfficient, see Absorption
Angular drstnbuuon of radmuon, from oscrllat- Attenuation length, for visible light in the atmo-
ing dipole, 411, 438 sphere, 467
from oscillating quadrupole, 415-6, 438 Averaging procedure, to define macroscopic
from relathriitic accelerated charge, 668f, 678 fields, 249, 251
.1ee alse Bremsstrahlung; Multipole radiation; Axial vector, definition of, 270
and Radiation Azimuthal sytiiiiielty, poteiitial problems with,
Angular momentum, electromagnelic, m Circu- IO If
larly polarized plane wave, 350
electromagnetic, expansion of, in plane waves, Babinct's principle, 488f
-"'~
/ 7 .. moex
Bessel functions. 112f in magnetostatics, methods of solving, 194f

connection with Airv integrals 678 in rectangular coordinates 70f
dual integral equations involving, 132, 205 in spherical coordinates, 95f
expansions involving, 115, 118, 119, 126, 138, in two dimensions, 72f
occ uoov
. . .~
'
'~v, •~•, 'u~, 'v~ ~uua~uvu,
~· 'Y•
of first kind, 1 "' 1 13 Wave guide
J:< ouner-J::>esse• senes, 1 D, u11 oren mreracnon, :l~11
Fourier-spherical Bessel series, 119 Bremsstrahlung, 714f
~='· · nf K and K 6~ti an11ular distribution of 712
integral relations involving, 118, 126, 132, 140, as scattering of virtual quanta, 729-30
142,205,493 Bethe-Heitler formula for, 717, 719
~'."'-'lo;ta< ('.,- >UVH OH 0 ( .. V ,.u
Kapteyn series of, 116, 702, 705 frequency spectrum, 716, 717, 719
leading behavwr tor large and small argu- at low frequenCies, 711
ments, 114, 116, 427 in Coulomb collisions, 714f
mnrliiiPrl T K 11 h innPr TO
Neumann series of, 116 maximum effective momentum transfer in,
orthogonality, on finite interval, 114-5, 138 713, 715, 718
on infinite interval, 118, 140 nonrelativistic, 717-8
recursion formulas, 113, 427 polarization of, 712
Schlomilch series of, 116 relativistic, 718f
of second kind, N,., 113 screening effects in, 721f
Oo · •o f, I 11'\ D. 'o on rrlP Cl(l,;.:_ 7
"'
spherical, 426-7 Brillouin precursor, 338
..
HHHLW!;',.W<W>, ..L/
Wronskians, 427 Canonical stress tensor, 605-6
of third kind, Jl\}>, H\~1, 113 for electroma.gnetic fields, 606
zeros of .Tm(x), 114 Capacitance, definition of, 43
zeros of .r,;,(x), 370 of a circular disc, Cavendish's value for, 19
R,,,, ,.J, • • ,f ,.J; • ,.], • 7Cl(lf ;~"'" f, -~
-:' ' -
Bethe-Heitler formula for bremsstrahlung, 717,
. ,,;
'
'1
Cauchy boundary conditions, 38

.
-
IJvo' U ilHU n, .:u,;,
""' 785f
Bibliography, Causal Green function, 614
Biot and Savart law 175f Causality, 330f
J efimenko generalization of, 24 7 consequences in dispersion, 334
Birefringence of the ionosphere, 317 in Coulomb gauge, 242, 291
Bistatic cross section, definition of, 457 lack of, with radiation reaction, 748, 772
Blue sky, Rayleigh's explanation of, 465f in special relativity, 528
J::>IVI 1 equatiOn wr spm, Jo 11 Lavenmsn s apparams ror mverse square mw,
Boost parameter, relativistic, 526 6-7
Rnrn "' · '"tinn in scatterin" 464 5 Cavitv resonant see Resonant cavitv
Born-lnfeld nonlinear electrodynamics, 10 Center of mass, of electromagnetic energy, 622
n. r1 · ·,,, AA< "- ,,.,,; .. ,; "' . ,f ,;:;,
Boundary conditions, ~~ inte~fa;e between me-

'
Characteristic time, in radiation damping, 746
dia, 18, 154, 194 Charge, discreteness of, 4-5
Cauchy, lJmchlet, and Neumann, 37-~ electnc, 25
for dielectric wave guide 388-9 electronic and protonic. equality of magni-
'•"·nc\1 in Kirchhoff'< fnr tOirlf'< """
diffraction, 480 invariance of, 554
. . "
'
vo, '~n,
~A
~v. VL, U'-'-UO uo:•ss ~· ~~N
COO ~H a~, ~ OJO
perturbation of, 366f, 374, 401-2 radiation emitted by sudden creation or disap-
at surface of, good conductor, 353 pearance of, in beta processes, 730f, 732f
scatterer, in terms of surface impedance, 475 in uniform motion in vacuum, fields of, 559-
fnr TF "nrl TM w"v"s in w""" <mirlP. 1~9 liO
D
"· _ duP nrnhiPmo ~ .~f.. rl-oorn0 ;nn 17< 'J'lQ 777
~~thod of solution, 38f Charge density, and current density as 4-vector,

-'-'
j~ ~ielectrics, 154f ''"' and current density of charged particle, covar-
1mage method of solulion, 571 1ant expresswn for, 615
Index 793
effective magnetic, 196-7 Conservation, of angular momentum of particles

induced by point charge ncar conducting and fields, 288
,,_ ,
~,
, . <;Q
, .
,f
~o·
, ,, ,
'
., 17<; 'J'!Q 777
, ' , '
~ ~
at sharp corners, edges, and points, 78, 106-7 of electromagnetic angular momentum, in co-
at sunace 01 conuucwr, quanmm-mecuamcat, vananL wrm, ow>
21 . _, "-
of energy of particles and fields, 258f, 611
Chon>ed 1ics. S79f of field.enen>v and in covariont
Lorentz force on, 3, 260, 579 form, 607, 609 . _,
<: ., L1. "OM
, ,- .£
~.
.£
pu ~
clr.-U .
, .-..:A , t:.11
in uniform static magnetic field, 585 Constitutive relations, 14
classical model of, 759f Continuity, at interface, of tangential E and nor-
Poincare stresses in, 755-7, 760f mal B, 18
dohilitv oncl rm;orionrc- r.f <>nc-nm oncl mr. r. 1itv c-nnotir.n , fr.r -'· oncl lt '
mentum, 762-3 175,238,777
different contributions to mass, 761-2 in covariant form, 555, 610
,\ee atso Amanam-Loremz moue! or ror e1ecrromagnet1c energy now, L;)~
electron in dissipative media 264
Cherenkov angle, 638 Contraction of length, see FitzGerald-Lorentz
Cherenkov radiation, 63 7f contraction ..
~H~UWl lVVP; H~lUO Vl, lUH ~uuu a vouom , v•, J·,v
Classical electron radius, 604, 695, 755 Convective derivative, in Faraday's law, 210
Classical limit, of angular momentum in multi- in fluid flow, 320
pole fie Ids, 435 Conversion table, for equations in Gaussian and
,f _,: >t; hPlrl, , '!_A "' ;,. , 7Q'J
Clausius-Mossotti relation, 162 for given amounts in Gaussian and SI units, 783
~1osure,see ~omp,c,eness ~orresponuence pnnc1p1e 01 oour, ,v.,
Coherence, of scattering by collection of scatter- Coulomb gauge, definition of, 241

im> centers. 41i1 2 and ro -" . 242. 291
Coherence
.. volume in transition radiation, 649 Coulomb's law, 24
,
"· , 1
,]. 1
, "'""' '-"· -'· ,1:
.£ , " · ' " ' '
radiatio'n emitted duri~;,J 709f Coupling const~~t, running, 12
Collision time, for fields of relativistic particle, Covariance, of electrodynamics under Lorentz
560 transformations, 553f
rr.mmlltotinn TP.lotinns fm infinitf'simol T .nrf'nt'7 nf ·'- 1l lows nndP.r T lronef, -
transformation generators, 548 tions, 517, 540
of angular momentum operator, L, 429 Covariant expressions, for electromagnetic en-
Lomp1ememary screen, aenmnon or, '11;1; ergy ana momenmm, D II
Completeness of set of orthogonal functions for equation of motion for spin 561f
68 for Lorentz force equation, 557, 580
Completeness relation, 68 for Maxwell equations, 557
'- . . . " . • • . £.
lUl UH HH~l HH, 1 1/' lUl
-~ . w•c~, '' 1
140 Covariant vector, definition of, 541

for complex exponentials, 70, 125 Lritlcal frequency, as upper hilllt of frequency
for spherical harmonics, 108 spectrum of radiation emitted by relativ-
-r
>ffPrt , ..1:<:. • ,f .t. ;~t;r nort;rT. f.Tl. (,70
.L: ' - '
tering
. . by, 696-7 . . Critical opalescence,
.
469-70
.
'-'onuUC110n m a movmg meu1um, .JLV, - " L '-'ross sec,Jon, ueLHe-11e1uer, ,or uremsstrau-
Londuct1v1ty, ettect ot, on quas1stat1c helds, lung, II I
218f classical for bremsstrahlun2:. 716
relation to complex dielectric constant, 312 classical particle scattering, relation to impact
".
, • •
, ., 1 .,- """
boundary conditions at, 352f definltion of, f~r scattering of electromagnetic
definition of, 50 waves, 457, li94
diffusion of fields in, 221f Rutherford, !125, 714
fiPlds ol "JrforP nf 'l~?f fnr · w oncl o'- nf rarl; .t· hv
fields inside, 220, 354 harmonically bound charge, 766f
penetration or skin depth in, 220, 354 by large conducting sphere, 499-500
surrace 1mpeoance or, .Do oy smau conoucung spnere, 'IOU, '+ 1 1
Conical hole or point fields near 104f by small dielectric sphere 458-9
--
,,0 Index
Energy (Continued) Faltung theorem of Fourier integrals ::no

-··
in rliPIPrtrir mPrlio 1 (iq
Mlf
Fororlov
p, rl· ,
·--~~•
1· _,. :;'" -, ,
-' '<Ai<
,·,.0,
hyperfine int~raction,' 190-1 in differential form, 211
magnetic, 212f for moving circuit, 209-10
of magnetically permeable body, 214 Ferromagnetism, definition of, 15-fi
relativistic, of a particle, 537-8 Feynman-Heaviside expressions for fields, see
self-, 42, 754 Heaviside-Feynman expressions for fields
Energy conservation between particles and Field, electric, see Electric field
fields, 258f, 610-1 magnetic, see Magnetic field
Energy density, electromagnetic, 259 Fields, of charge in arbitrary motion, 664
electromagnetic, as (0,0) element of symmet- of charge moving uniformly, in dielectric,
ric stress tensor, 609 Fourier transforms of, 633-4
effective, in dissipative media, 263 in vacuum, 559
electrostatic, 40f in vacuum, Fourier transforms of, 650, 656
Energy flow, 259 of relativistic charge, equivalence of, to pulse
veracity or, jLi, jL) or raatatwn,:5l5IT,7221f
m wave guide. 3631 Ficld-strene:th tensors 556
Ener11v-lcvel shift due to radiative reaction Finite di ::;: -::>-.nf
''7t;'<f p; . • .. 1 • _, · f+,n A \ ""'
T"
'6J
.' , n ·•'
llJUla lUl, OL I
,
..,. . _, <. 1:" -~"
_..,_ cry
,u Hl "'
,JTT;:Y7U
.7CJ• L
. "Y'"
" -
'JO
tionr s c1asstca1 rormu1a ror, OLI J:< mcruatwns,

'
m denstty of tlmd, and scattenng,
density effect in, 631f 468-9
in electronic plasma, 656-7 in energy loss, 631
fluctuations in, 631 Force, between charge and image charge, 60, 61,
by magnetic monopole, 658 62
radiative, in collisions, nonrelativistic, 718 on charged surface of conductor, 42-3
in collisions, relativistic, 723-4 Coulomb's law of, 24
per revolution in circular orbit, 667 between current -carrying circuits, 178, 777
Energy-momentum 4-vector, 538 on current distribution in magnetic field, 188f
Energy radiated, by accelerated charge, angular on dielectric body, 167, 169
and frequency distribution of, 673f Lorentz, 3, 260, 579
by accelerated magnetic moment, angular and in covariant form, 557, 580
frequency distribution of, 704 on magnetically permeable body, 214
Energy transfer, in Coulomb collision, 625f on magnetic dipole in nonuniform field, 189
fA]:;' • ],; L ~ .L • h~~ nn •wrmo~a~t ~""~~to ,~(I
.. ;.,
~
,f o1 '"' ;~" 1
... .~
, ""(\ t .'
~
.t
'3:
~
~, , LA , L• , L">
L ' f C<dUUH, Ul WHH l• , no , //1-L /'HI-~,
/'+()· -.JU, //1-L

'" ' ''ueLween I
LWO parallel wues, 1115, 111

Ether, 5l:l Force density, and divergence of stress tensor
Ether drift exneriments 519 22 fill
Fvon
_, . in WOVP tmirlf• i/i(l nn ~rln~t<>r 'lOt;
,'
Excitation of waves in wave guide, by localized Force equation, Abraham-Lorentz, 748
source, 392f, 419f approximation to, 749
Expansion, of arbitrary fields in wave guide in with radiative reaction, Dirac's relativistic, 711
normal modes, j'Jl approximation to, 772
of circularly polarized vector plane wave in integrodifferential, 772
multipole fields, 471f Forward scattering amplitude, relation of, to the
of Green function, e'kR/R, in spherical waves, total cross section, 502
428 relation to dielectric constant, 504
of lx - x'l-'. in cylindrical coordinates, 126, 4-current, 554
140 4-divergence, 543
in plane waves, 128 4-Laplacian, 543
in spherical coordinates, 102, 111 4-tensors of rank k, 540-1
of scalar plane wave, in spherical harmonics, 4-vector, contravariant and covariant, definitions
471 of, 540-1
.t:xpanswns, see urtnonormal cxpanswns 4-vector potenttal, 555
Extinction coefficient see Absorption coefficient 4-velocitv. of narticle 532
Index 797
Fourier integrals, 69, 243, 253, 330 Goos-Hanchen effect, 308, 342
Fourier series, tiS Gradient, contravariant and covariant in special
summation of, example, 74-5 relativity, 543
Fourier-Bessel series, 115, 138 of electric field, and force on electric dipole,
Fourier transform, of exponentially damped 171
wave, 372, 765 and quadrupole interaction, 150-1
of fields of charge in uniform motion, in di- of magnetic induction, and associated particle
electric, 633-4 drift velocity, 588-9
in vacuum, 650, 656 and force on magnetic dipole, 189
of wave packet, 323, 327, 336 in rectangular, cylindrical, and spherical coor-
Ftaunhofet dif£1 action, definition of, 491. See dinates, end papns
also Diffraction m sphencal form wilh L, 472
Frequency distnbutJOn of radiatiOn, emitted by Green function, defimt10n of, m electrostalics,
electron in classical equivalent of hydro- 38
gen ~tom, 704 5 eikRfR, spheric~! wave exp~nsion of, 428
emitted by extremely relativistic charge, quali for Helmholtz waye equation, 244
!alive aspects of, 671 3 invariant, fur wave equation, 612f
emitted by relativistic charge in instanta- retarded and advanced, 244, 614
neously circular motion, 681-2 for scalar diffraction by plane screen, 480
explicit formulas for, 675-6 symmetry of electrostatic, 40
from the sun, 467 for time-dependent wave equation, 245
of transition radiation, 653 Green function in electrostatics, 38-40
from undulators and wigglers, 691, 693 for concentric spheres, 122
Frequency shift, llw, in resonant cavity, 374 for cylindrical box, 143
of resonant line of oscillator with radiation examples of use of, 64-5, 122f, 141, 142
damping, 763f expansion of, in cylindrical coordinates, 125f
Frequency spectrum, see Frequency distribution in Legendre polynomials, 102
of radiation in eigenfunctions, 127f
Fresnel diffraction, definition of, 491. See also in spherical coordinates, 111, 119f
Diffraction for Neumann boundary conditions, example
Fresnel formula for speed of light in moving of, 144
media 570 symmetry of 40 52
Galerkin's method, in FEA, 79f for two parallel grounded planes, 140-1
Galilean invariance, 515 Green's first identity, 36, end papers
and Faraday's law, 209-10 Green's reciprocation theorem, 52
of phase of wave, 519 Green's theorem, 36, end papers
Galilean relativity, 515-6 use of, in diffraction, 479
Galilean transformation of time and space coor- vector equivalent of, 482f
dinates, 515 Ground, concept of, 19-20
Gauge, Coulomb, 241 Group velocity, 325
Coulomb and causality, 242, 291 and phase velocity, 325
Lorentz, 241 in wave guide, 364
radiation or transverse, 241 Guides, see Wave guide
Gauge invariancc, 240 Gyration frequency, of particle in magnetic
Gauge transformation, 181, 240 field, 317, 585
and Dirac string, 279-80 Gyration radius, 586
and partiele Lagrangian, 583 Gyrotropie media, phenomenological expression
fm polarization in, 273
Half-width, of resonant line shape, 765

in 781 See a/5o I In its relation to Q value of resonant cavity 372
Gauss's law, 27 9 Hall dfect, ph@nom@nological @Xpression for,
diffeiCntial form of, 29 289
g-factm, of electron, 548 Hamiltonian, of charged particle in extemal
and muon, numencal values of, 565 fields, 582, 585
798 Index
. .
r 'J WI WOIU>, it> \ll,ll) I,;UIIIjJO- anu magnenc energy, LD
ncnt of symmetric stress tensor, 609 mutual, of two current loops, 234
Hankel function of order v, 113. See also Bessel self, of transmission lines, 232
fnnctinn 1mitc ,f 7R1
Wo~lrnl t , 11 Q T, rl,
, r. ' ' .c •'HWC
,
Heaviside-Feynman expressions for fields, 248, Infinitesimal generators of the Lorentz group,
284 546, 548
Helical path of particle in magnetic field, 586 Inhomogeneous plane wave, 298
svnchrotron radiation associated with 703 Inner b 'hl .. ~ o 71?
HPlicitv cnnnPctinn to c.ircnler · 1tinn 100 11 pm•oti~nc ~f';hn firct l-inrl , cl.,ol , 11?
.
Hemispheres, at different potentials, 65f, 101 Interaction energy, see Energy
Hertz vectors, 280f Interface between two media, boundary condi-
Hidden momentum, 189, 618 tions at, 16f
High-frequency behavior of dielectric constant, Internal field at oosition of molecule in dielec-
313 333 tric u;n
Hole, circular, in conducting plane, electric Invariance, see Adiabatic invariance, Relativistic
fields near, 134 in variance
effective dipole moments of, 133, 205, 422-4 Inverse distance between two points, expansion
magnetic fields near, 206 in Bessel functions, 126, 140
Huygens s pnnc!ple, 4 n; expansion in Lel'endre nolvnomials 102
Hydrodynamic equations for conducting fluid, in spherical harmonics, 111
319-20 Fourier integral representation of, 128
Hyperfine interaction energy, 190-1 Inverse square law, precise verification of, Sf
Hysteresis, magnetic, 193 Inversion, see Spatial inversion
Ionosphere, propagatiOn Of waves m, 316f, 34!\-7
ldeahzatJons m electromagnetism, 19f lrrotational vector, definition of, 242
Image charges, see Images
Images, method of, 57f Jacobian, in Lorentz transformation of 4-dimen-
for conducting sphere in uniform field, 62f sional volume element, 555
. .
IV! , U.J--1 in Lranswrmation or coorainates tor aena
ror magnencauy permeao1e mema, LL'-J runctron, 12U
for point charge near conducting sphere, 58f Jefimenko's expressions for fields, 246f
for two charged spheres, 86-7
Impact parameter, and scattering angle, 655
'
, r"' .l :L =~~·
.CC. '
, ' ' Kinematics, relativistic, examples of, 573f
mnuuv" '' , .>U. -u
OLO, OL/ !'..ircnnon aiuraCfion, see"'JTTfiactwn
minimum effective, in Coulomb collisions, Kirchhoff's integral, in diffraction, 479
626,629 vector equivalents of, 482f
in method of virtual quanta, 725, 729 Klein-Nishina formula, 697
•nieol -h?O .cc. ,f . ··- _; .c _,
",' ·~M
, _'-I: , ~on
-
' If c c
' ~~~, v• u~~ opa~~, L/
·~· UHH;O> o· ": re1auons, .:l.:l.:ll, .1'+1>-Y
surface, 355, 475
of two-terminal device, general definition of, Lagrangian, Darwin, for charged particle inter-
264f actions, 597
"'""P in "'""" nnirlP 1<;0 f. nl -•' ; , , u , a>O>
T
''b
+- -" -+ n.
, +.
·p
' '
'vvv
/AAC
-=«LL<OI COt>, w,:, 1v1 l<:owLivioti~,;c-crrargcrrp,uLiue in external

Index of refraction, 296, 303 fields, 579f
analytic properties of, 337 Lagrangian density, for continuous fields, 598-9
and nhase and o-roun velocities 32S I .em h shift 7,;,;
rPbtinn In I crottPrinu "mnlitnrlP :;n.:1 f onloc,• Pnl'"'' •~ , 1A
•c ' +. _, +• ::. ' ..
~·
L
see also 1eI ectnc I)'

. constant''1' 'J, . OUVII> W I , ,
in cylindrical coordinates, lllf

Inductance, 215f general solution of, in cylindrical coordinates,
accurate result for circular loop, 234 117 9
coefficients of mutual and SPlf 215 -li in rPrbnno.lor . '"'"" '71 ?
0eti~otinn nf cplf 71 h ,.
:"'-' ' ' ;"
high-frequency compared to low-frequency, in t~o-dimensional coo;dinates, 77, 89
218 in rectangular coordinates, 70f
Index 799
in spherical coordinates, 95 I orentz invariant differential photon spectrum,

in two dim!msions, Cartesian coordinates, 72f 710, 720
in two-dimcusious, polm eumdinates, 76 Lmentz-Lorenz relation, 162
uniqueness of solution of, 37 1l Lorentz tr ansfur rna lion, 524f
Laplace transform, usc of, 222 of electnc and magneflc fields, 558
Laplacian operator. and angular momentum op-
in four dimensions, 543

La1 mo1 fo1 ntula fen radiated powet, 665 matiix nprese11tation of, 543f
relativistic generalization of, 666 noncommutativity of, 548
Legendre differential equatton, 96 proper and 1m proper, defimflons of, 544
Legendre functions, associated, l 08 of time and space coordinates, 525
asymptotic form for large v, in terms of Bessel see also Relativistic transformation
function, 1On Loss, see Attenuation, Power loss
of order v, 105
Legendre polynomials, 97 Macroscopic averages, 249-50
expansion of inverse distance in, 102 Macroscopic electromagnetic quantities, 250f
explicit forms of, 97 Macroscopic equations, elementary derivation
integrals of products of, with powers of cos e, of, for electrostatics, l5lf
100-1 for magnetostatics, 191f
orthogonality of, 99 Macroscopic fields, 13, 250, 255-6
recurrence relations for, 100 Macroscopic Maxwell equations, 13, 238
Rodrigues's formula for, 98 derivation of, 248f
see also Spherical harmonics Magnet, permanent, 16, 200
Lenz's law, 209 energy of, in external field, 190
Level shift, of oscillator frequency, from radia- method of treating magnetostatic boundary
tion damping, 763f value problems involving, 196
Lienard's generalization of Larmor power for- Magnetic dipole, see Dipole fields; Dipole
mula, 666 moment
Lienard-Wiechert potentials, 661f Magnetic charge and current densities, 273-4
Lifetime, of pi mesons in relativistic motion, transformation properties of, 274
529 Magnetic diffusion, 221-3
see Transition probability Magnetic field, H, boundary conditions on, 18,
Light, speed of, 3, 776 194
Light cone, 527 definition of, 14, 192, 257
Limiting speed, 518 see also Magnetic induction
Linear superposition, of electric fields, 26 Magnetic flux density, see Magnetic induction
of plane waves, 296, 322f Magnetic induction, B, boundary conditions on,
of potentials, example of, 61 18, 194
valid1ty of, 9f of charge Ill umform relaflv1sflc moflon, 559
Line breadth, of oscillator with radiation damp- of charge, limiting form as v -> c, 573
ing, 763f of circular loop 182f
Localized source, in wave guide, 392 3 of current element, 175
,\ee aim Dipole, Multipolc, and Multipole definition of, 174, 178
moment of long straight wire, 176
London penetration depth, 604 of magnettzed sphere, 198
I ongitudinal vector, definition of, 242 of nonrelativistic moving charge, 176, 560
Lorenz condition, 240, 294, 555 relativistic transformation of 558
Loreutz condition. See Lorenz condition. Magnetic moment, adiabatic invariancc of,
Lorentz force, 3, 260, 553, 579 592f
m covanant form, 55 I, 563, 580 anomalous, of the clcclron and muon, 565
density, 611 dens1ty, 186, 256
Lorenz gauge, 241 effective, of hole in conductor, 205
Lorentz group, 540 of electron, 548
infinitesimal generators of, 546, 548 force on, in nonuniform magnetic field, 189
Lorentz in variance, of electric charge, experi- intrinsic, caused by circulating currents, 191
mental evidence for, 554 radiation emitted, in disappearance of, 735-6
of radiated power, 66n when in motion, 704
see also Relativistic invariance radiation from time-varying, 413f, 442
800 Index
Magnetic moment (Continued) Microwaves see Diffraction- Resonant cavitv·

scattering by induced, 457, 460 and Wave guide
torque on, 17 4, 189-90 Mirror, magnetic, 595-6
see also Dipole moment Mixed boundary conditions, 38, 129f, 205
Magnetic monopole, 273f MKSA units, see Units
auu yuan · · 1 ur t:H;~u;~. ~~large, L ':li JYlOOes, m cynnonca1 cavny, .iOIII
vector potential of, 278, 290-1 in cylindrical dielectric wave guide, 388-9,
Magnetic multi pole fields see Multi poles Multi- 404
pole fields normal, in wave guide, 389f
Magnetic permeability, 14-5, 193 propagating and cutoff or evanescent, 360
Magnetic polarization, 13-4 TE and TM, in wave guide, 359
Magnetic pressure, 320 in slab dielectric guide, 385f
tv1agneLic scamr potemiw, 1 ou, lY.J m spnencal geometry, aennJtJOn or, 375
Magnetic shielding, 201 f, 228-9 in wave guide, 360
Magnetic tension 320 "-' ·nlor __252 ?~h
"Magnetic" waves, 359, 430 Molecular polarizability, 151
~A, ,,; ,,; rl.-fi, ;,· ,f U~£
, '""' , ~u .L L ,:
of current d~nsity distribution, 186 • '

Momentum, canonical, for particle in external
divergence of, as effective magnetic charge fields, 582
oensny, lYO conservation of, between particles and fields,
effective current density of, 192 261, 611
r"rli"tinn hv timP.-v"rvin<J 4i9f p],,ct ,,;
'· -,co
Magnetized sphere, 198f hidden mechanical, 189, 618 _, '
• •fiol.-1 '1()(\ 1
, -' . . .
'vwu uouv, v• P_~•
.
Hv>v, •~v
Magnetohydrodynamics, equations of, 320 Momentum density, electromagnetic, 262

Magnetostatics, basic equations of, 180, 194 and co-moving mechanical momentum in di-
methods of solvmg boundary-value problems electric, 262, 294
in, 194f as part of covariant symmetric stress tensor,
Mass of photon, consideration of, using Proca 609
Lagrangian, 600f in macroscopic media, 262
·r v• wvaouuu5, uow5 • ''>" 1 cu- auw LlallSHOf, in \..UU!OUH) COl is ion, ()L)
cuits, 601-2 maximum effective, in bremsstrahlung, 713,
limits on, 7-9 715 718
modification of the earth's dipole field by, 621 minimum effective, in bremsstrahlung, 716,
M"ximnm onrl mi . I imn"ct PtPrc onn '71'7 '71Q .,.,, ')
Impact parameters Monopole radiation fields, absence of, 410

Maximum and. minimum
.
scattering angles, in Mossbauer effect, use in ether drift experiments,
~ldMlC o~:ed oarticle in uniform 'htir
derived from a Lagrangian, 599-600 magnetic field, 585-6
in rliff..r .. nt .-.f nnit< 75<1 ,1., ~D. 'ol, •
macroscopic, 13f, 238, 248f Moving charge, fields of, 549, 573, 664
plane wave solutions of, 295f Moving circuits and law of induction, 209-10
•spuericat wave somLions 01, 'tLO>l JYlU!tlple scattenng or particles oy atoms, 643f
Maxwell stress tensor, 261 Multioole electrostatic 146
Mean-square angle of scatterin_g, 643 4 el 1tic. C:"rtesi"n ~
Mean-value theorem, for electric field, 149 spherical, 146
f.-.r . . ,..,
, .
, 10"
for magnetic field, 188 radiating, general aspects of, 407f
>v .>o lt:l t:Ht:U, UVJ nme-varymg, 4U /I, 4~
acma1 msumunon 01 cnarge at surtace
lVleLa~s, see also Dipole moment; Magnetic moment;
of, 21 and Multioole moment
ultraviolet transparency of, 314 Multipole expansion, of electromagnetic fields,
Method of Images, see Images 429f
" . ' ,f ·~I ,_t_,.; ili ~ .£ -'
v n•v.• , .,~,
Michelson-Morle'i experiment, 51? of Green function for wave equation, 428

mouern successors 10, wr ue1ecuon o ether or mteract1on energy, DU
drift, 51l)f of localized source in wave guide, 419f
of radiation by linear antenna, 444f Operator relations, see Gradient; Laplacian
of scalar plane wave, 471 Optical fibers, attenuation in, 383, 470-1
of vector plane wave 4Tif cjrcular 387f
Multipole fields, 429f eikonal approximation for, 380f
angular momentum of, 433-5 graded index for, 380, 3S4, 402
connection to sources, 439f meridional and skew rays in, 381
electric and magnetic, 430-1 modal dispersion in, 383, 403
energy of, 433 modes in, 3S5f
expamion of arbitrary source free fields in, 431 propagation in, 373f, 3il5f
near-zone properties of, 432 slab, 385f
parity properties of, 436 Optical theorem, proof of, 500f
radiation-zone properties, 433 Orthogonal functions and expansions, 67f
use of, in description of scattering, 473f Orthogonality, of Bessel functions on finite in-
Multipolc moment, electrostatic, 146 7 terval, 115, 13ll
estimates of, for radiating atoms and nuclei, of Bessel functions on infinite interval, 118,
of linear center-fed antenna, 446 of complex exponentials on infinite interval,

magnetostatic, 184f 69
of oscillating source, exact expressions for, of Legendre polynomials, 99
440 of sines and cosines, 6S
long-wavelength approximations for, 441-2 of spherical harmonics, 108
see also Dipole moment; Magnetic moment; of vector sphencal harmomcs, 431
and Quadrupole moment Orthogonal transformations, 268
Multipolc radiation, angular distributions of, Orthonormal, definition of, 67
437f Orthonormal expansions, 67f
by atoms and nuclei, 442f Fourier, on finite interval, 68
by linear center-fed antenna, 444f on infinite interval, 69
lowest order, elementary discussion of, 410f Fourier-Bessel, 115, 11S, 13S-9
quantum-mechanical selection rules for, 436 on infinite interval, 11S-9
sources of, 439f Legendre, 99
total power radiated in, 439 spherical Bessel function, 119
spherical harmonic, 110
Neumann boundary conditiOns, defimhon of, 3 I vector, for fields m wave gmde, 390-1
use of, in generalized Kirchhoff diffraction Oscillations. see Radiation: Waves
theory, 480-1 Oscillator, absorption of energy by, 655
Neumann function, 113. See also Bessel in model for dielectric constant, 162, 309
functions with radiation damping, 763f
Nonlmear electrodynamic effects, IOf scattenng and absorptiOn of radiatiOn by,
Nonlinear optics, 16 766f
Non locality, in time, in connection between D Oscillator strength, 310, 627, 634, 765
and E, 330
in time and space, 14-5, 331-2 Paramagnetism, definition of, 15
Normalization of fields in wave guide, 391 Parity, of multi pole fields, 436. See also Spatial
Normal mode expansion of fields in wave guide, Inversion
3S9f Parseval's theorem example of 674
Nuclear quadrupole moment, 151 Particle motion, in crossed E and B, 586f
interactiofl eflerg} of, 150, 171 in dipole field of earth, 619
Nuclei, estimates of multipole transition rates in, in external fields, 579f
442f m mhomogeneous B, 588±, 592f
Numerical methods. in electrostatics, 47f, 79f with radiation reaction, 74S 50, 769, 771 2
in magnetostatics, 206f in uniform static B, 585-6
Penetration depth, see Skin depth
Obliquity factm in diffJ action, 482 in superconductivity, 604
Obstacles in wave guides, 394 Perfect conductor, definition of, for magnetic
Ohm's law, 14, 219, 312, 356 fields, 204
covariant generalization of 572 Permanent magnetization, 16
in moving medium, 320 Permeability, incremental, 193
nonlocality of, in conductors at high frequen- n1agnctic, 14, 193
cies, 332 Permittivity, electric, 14
802 Index
Perturbation of boundary conditions, method of, Polarization effects in energy loss, 63lf
366£, 374, 401 Polarization of radiation, hy reflection, 307
for degenerate modes, 402 from accelerated charges 665 676 678 706
Phase difference and elliptic polarization, 299 in brcmsstrahhmg, 712
bgtwggn E and Bin conductor, 221, 354 Cherenko', 639
Phase of plane wave, relativistic invariance of, circular, elliptical, linear, 299f
519, 529 left- and right-handed, definition of, 300
Phase shift for scattering by sphere, 476 from multipoles, 411, 414, 437
Phase velocity, and group velocity, 325-6 scattered by atmosphere, 468
of Alfven waves, 321 scattered by small conducting sphere, 460-1
and group velocity, in wave guide, 364 scattered by small dielectric sphere, 459
of whistlers, 319 Stokes parameters for description of state of,
of plane waves, 296 301-2
in wave guide, 361 from synchrotron, 678-9, 706
Photon, angular momentum of multi pole, 435-6 in Thomson scattering, 695
Photon mass, effective, in superconductors, 604 x-ray 712
limits on 7 9 Polarization potentials, 230f
trgatmgnt of, using Proca Lagrangian, 600f Polarization 'eetot s, 297, 299f
Photon speetnun, emitted during collisious, 709 Polar sa bs lances, 164
10 Polar vector, definition of, 270
Lorentz-invariant expression for, 710 Potential, electrostatic, 30
see also Bremsstrahlung; Radiation of dipole layer, 33
Plane wave, electromagnetic, 295f expansion, in Bessel functions, 118
inhomogeneous, 298 in Legendre polynomials, 101f
magnetohydrodynamic, 321 in spherical harmonics, 110
reflection and refraction of, 302f of line charge in cylindrical coordinates, 127
scalar, propagation in dispersive medium, near small hole in conducting plane, 133
322f, 326f of point charge, between grounded planes,
expansion of, in spherical harmonics 471 141 2
vector expansion of, in spherical harmonics, in cylindrical box, 143
473 expansion i11 cylindrical coordinates, 126
Plastna, confinetnent of, by tnagnctic niirrors,
,
energylossin,656-7 Fourier integral representation of, 128
transverse waves in, 313 in rectangular hox, 128-9
in external magnetic field, 316f, 347 polarization (Hertz vectors), 280f
Plasma frequency, and first precursor, 338 in rectangular hox, 71-2
of dielectric medium, 313 scalar and vector, for time-varying fields, 239
sum rule for, 335 in two dimensions, 72f
Poincare stresses, 755f, 760f vector, see Vector potential
Poisson equation, 34 Potential energy, see Energy
equivalent integral equation, 36-7 Power, instantaneous radiated, by accelerated
examples of solution of, 123, 124, 137 3 charge, 665 6, 701
formal solution with Green funetioH, 38f radiated by, charge in at biLiary peiiodic nro-
uniqueness of solution of, 37 8 tion, 702
see also Green function in electrostalics charged particles m Imear and cucular ac-
PolanzabJhty, electromc, 163, 309-10 celerators, 667
effective, of aperture in conducting plane, electrons in undulators 689 91
423 4 linear antcmm, angular distributions of,
models of, 162f 417,447
molecular, 162f (1, m) multipole, 437
orientation, 164 oscillating dipole, 411-2, 437
Polarization, charge density, 153, 156 oscillating quadrupole, 414-5, 437
macroscopic, electric, 14, 152, 255 Power flow, see Energy flow
magnetic, 14, 192, 256 Power loss, because of finite conductivity, 221,
magnetic, see Magnetization 355-6
surfaee-eharge density, 156, 159 in ICSOIIdiiL cavity, 371£
transition radiation ftorn, h47, 649£ in wave guide, J6Jf
of vacuum, I I-13 per umt area, at surface of conductor, 356
see also Magnetization see also Attentuation
Index 803
Poynting's theorem, 258f from electric quadrupole, 414-5

covariant generalization of, 607, 610 from linear antenna, 4 ](if, 444f
fat dispetsive and dissipative media, 262f from localized source, 407f, 439f
for harmonic fields, 2Mf from magnetic dipole, 413-4
Poynting vector, 259 multipolc, see Multipole radiation
for plane wave, 298 by relativistic charge in instantaneously crrcu-
uniqueness, 259 lar orbit, 676f
in wave guide, 363 invariant and coordinate-free form, 702
Precession, of spin, Thomas, 548f, 564 from short antenna, 412
PrecessiOn frequency, of particle in magnetic synchronon, 681 3
lleld, 317, 585 transition, see Transition radiation
Precursor, Brillouin (second), 338 from undulators and wigglers, 683f
Sommerfeld (llrst), 338 Radiation condition for asymptotic fields, 479
Pressure, radiation, 288 Radiation cross section, definition of, 715
Proca Lagrangian, 000 f01 classical br emsstr ahltmg, 716
Propagation, in anisotropic dielectric, 346 in complete screening limit, 722
in dispersive medium, 322f, 326f for nonrclativistic bremsstrahlung, 717
in ionosphere, 316f, 345 for relativistic bremsstrahlung, 718-9
of signal incident on dispersive medium, 335f Radiation damping, see Radiative reaction
.<ee alstJ Signal propagation Radiation fields, 4Qg 9
Proper time, 528 of charge in arbitrary motion, 664
Pseudoscalar, -tensor, -vector, dellnitions of, 270 of(!, m) multipole, 433
Radiation length, 724
Q_ of resonant cavity definition of Radiation pressure. 288
371 Radiation resistance, 267
physical interpretation of, 373 of short linear antenna, 412-3
of right circular cylindrical cavity, 373 Radiation zone, 408
of Schumann resonances, 377, 399 in diffraction, 491
of spherical cavity, 455 Radiative energy loss, in accelerators, 667-8
Quadrupole moment, electrostatic, 146 in collisions, nonrelativistic, 718
interaction of, with lleld gradient, 150, 171 relativistic, 723
nuclear, 151 Radiative reaetioA, 745f
of oscillating source, 414 characteristic time T of, 746
see also Mullipole moment effective force of, 748, 749
Quantization of charge, Dirac's argument for, equation of motion including, Duac's, 771
275f jntegrodifferential equation of motion includ-
Quantum-mechanical modifications, in brems- ing, 772
strahlung, 717 and line breadth, 763f
m elaslic scattenng, 641 and shift of oscillator frequency, 763f
in energy loss, 629-30 simple equation of motion including, 748,
Quasi-static approximation, 218 749
slow changes of energy and angular momen-
Radiated electromagnetic energy, Lorentz trans- tum from, 749-50, 7!\9-70
forntcttion propel ties of, 017 Radius, classical, of electron, 604, 695, 755
Radiation, by accelerated charge, angular distri- gyration, of particle in magnetic field, 586
bution of, 668f Rapidity, definition of, 526
Larmor formula for power, !\65-6 use of, in relativistic kinematics, 539
angular distribution of, for relativistic parti- Ray, in optical fibers, 378f
cles, 668f meridional and skew, 3g1
angular and frequency distribution of, for Rayleigh's, approximation in diffraction, 481
charge m penodrc motwn, /02 approxrmation in scattering, 464
for charge in arbitrary motion, 675 6 explanatiOn of blue sky, 462f
for movjng magnetic moment 704 Jaw of scattering. 457
for relativistic charge, qualitative aspects of, Rayleigh scattering, 466
671-3 Reactance, definition of, in terms of fields, 267
in undulators. 689 91 Reciprocation theot ern of Green, 52
emitted in, beta decay, 730f Reflection, from sphere, in diffraction, 497-9
collisions, 709f. See also Bremsstrahlung of charged particle from region of high mag-
orbital electron capture, 732f netic field, 595
from electric dipole, 411 of plane wave at interface, 302f
804 Index
Kenectwn ( contmuea) Resonant cavitv. 368f

polarization bv 306 7 earth ancl ionosnhere__as._31M_
r>f rorlir> hv inn<"nhf'rP ':\ 17.-R , 'lT!
t. ol ' _, '?.!17
,. ,;"'" .<
·'" . ~ . ' .>V7
-J ou~>,
'""• "' yoau~ "a'~ ac uu~ua~~, ~·

. ·'"" I~
also Index of refraction power losses in walls of, 373

Relativistic addition of velocities, 530f Q of, 371f, 455
Relativistic effects in angular and frequency dis- resonant frequencies of, 369
tributions of radiation, 669-70, 672-3, spherical, 455
678f, 686f, 701f, 712 spherical concentric shell, 399
Relativistic covariance, of electrodynamics, Resonant frequency, in cavity, shift of, because
553f of power loss, 374
of physical laws under Lorentz transforma- of atomic oscillator, 162, 309, 627, n55, 764
tions, 517, 540 shift of, by radiative reaction, 764-6
Relativistic invariance, of action integral, Resonant line shape, 372, 765
580 Retarded Green function, 245, 614
of electric charge, 554 Retarded time, 245, 662-3
of 4-dimensional Laplacian, 543 Rodrigues's formula for Legendre polynomials,
of 4-dimensional volume element 555 98
of 4- scalar nroducts 52n. 541 D"'t' lS ?f.7f
,f . 1," ,-,_ A.
,- • T ,,, "u -,
.< _, '~ ·~ '.
-< -·
:" .': , -"1" '""''~uo uoauvu J-"U]J~' u~o u> r 4 U<illll-
Ul Ulllt:l Cllllil1op~~•l UIH, I ~U ues unaer, L 11
or ramatea power, ooo Kuthertord scattering, connection between an-
of radiation cross section 720 Qcle and imnact · in _655_
of speed of light, experiment on, 522-3 cross section, 625
Relativistic kinematics, notation and units for, in terms of momentum transfer, 625, 714
565-6
Relativistic transformation, of acceleration, Scalar, under Lorentz transformations, 540
569 under ordinary rotations, 268
of charge and current densities, 554 Scalar potential, 30
from em system to laboratory, 575 connection to work done, 30
of coordinates, 525 in magnetostatics, 196
of electromagnetic fields, 558, 586-7 for time-varying fields, 239
of charge in uniform motion, 559, 572 see also Potential, electrostatic
of 4-vectors and tensors, 52n, 540f Scalar product of two 4-vectors, 527, 541
of 4-velocity, 532 Scattering amplitude, forward, relation to dielec-
of momentum and energy, 533f tric constant, 504
of potentials, 555 relation to total cross section, 502
. .
U> ~JJ,:',' VC .... LVl, JV.C unegra1 expressiOn wr,__':"'-'_
anu uwmas precessiOn, J'toi muJtJpole expansiOn or, )U'J
ot velocities, 530f Scattering cross section for__12_articles classical
of wave vector ancl freouencv. 530 rlPfinitinn nf in tPrms r>f · n
r
-
'
<PP nhn f
..
r>rPnf7 tro .< Uinn_ t. , h""
--
n .1.
·;n
. _, '" .<, "' I+
w> wuwu"' , u~•
..
ouuu u~, -,_,, U7'
~At'~' C~O<WC" J W> see atso "cattenng 01 partiCles; .>Cauenng or
mamemauca1 srrucmre ~ radiatiOn
postulates of 517 8 Scatterin<> of oarticles bv atoms 640f
Relaxation method, in electrostatics, 47f effects of, electronic screening on, 641
in magnetostatics, 200f f1nite nuclear size on, 641-2
Resistance, definition of, in terms of fields, 266- mean square angle of, 643
7. See also Conductivity; Ohm's law; Ra- multiple, 643f
diation resistance; Surface resistance Rutherford, 625
Resonance fluorescence, 767 single, tail on multiple scattering distribution,
Resonance, in cavity, 372 645
Schumann, 376 total atomic cross section for, 643
width r of, definition of, 372 Scattering of radiation, at long wavelengths, 456f
Resonant absorption, 310, 768 coherent and incoherent, 462
and anomalous dispersion, 310, 334 Comoton. and Klein-Nishina formula 696 7
Index 805
Delbriiek, 10 Spatial inversion, 269-70

by density fluctuations of fluid, 468f opposite behaviOr of clectnc and magnetic
nf li<rht h~ li<rht 1n "h"r"" rlPnsities under 274
multipole description of, 473f transformation properties of physical quanti-
va . ~ -,.,~;
L
VJ
0>'
""'
• • •
''" , vv• ,
perturbation treatment of, 462f Special theory of relativity, see Relativity
oy pomt cnarge, n'J4T ;:,peeo or ugnt, expenmentai constancy, moepen-
mt"ntum-mechanical modifications of dent of frenuencv. 523-4
696-7 experimental constancy, independent of mo-
n .. 1
"'J
·-L
'6' ,
~ +L LL
'"
-'··']> >LL ~
' -< , """ '~-
resonant, 767-8 numerical value of, 3, 776
shadow, 496-7 Sphere, conducting, and point charge, 58, 60, 61
at short wavelengths, by sphere, 495f electrostatic Green function for, 64-5
by small conducting sphere, 459f, 477 general solution for potential in, 65, 122f
by small dielectric sphere, 457f with hemispheres at different potentials, 65f
Thomson, 694f scattering of radiation by, 457-8, 459-61, 473f,
Scattering phase shift, 476 495f
Schumann resonances, 374f in uniform electnc field, 62f, 157f
~
in" hv "tnmir p], •t· " pffprt nf nn nnifnrmlv mo<rnPti7P.d l9Rf
bremsstrahlung, 721f in external field, 200f
c<. ·" _, _, ,. ~-
• _,
,. n
£. _, <. •
vuv-. v•, vu """'',' ~u
0 +v vwouv uu 0 , v~+
Sea water, attenuatiOn constant of, 316 Spherical coordinates, 95

SelectiOn rules tor muthpole transtttons, 4_ib delta tunctwn m, lLU
.<::Plf-PnPr<rv · p], 1tir 754 Lanlace enuation in 95
Self-energy and momentum, 755f Spherical harmonics, Y 1m, 107f
' .l C ' ' < ~C~< ~~;, +L <. , ' " ' '
vVl~U~H<
v•, ·~ "'
Self-force, Abraham-Lorentz evaluation of, 750f and angular momentum, 428-9
Self-stresses and Poincare stresses, 755-7 completeness relation for, 108
Separation of variables, 70 explicit forms of, 109
in cylindrical coordinates, 112 orthogonality of, 108
in rectangular coordinates, 70 raising and lowering operators for, 428
in spherical coordinates, 95-6 sum rule for, 111
Shielding, magnetic, with permeable shell, 201f vector, see Vector spherical harmonic
of two-wire cable, with iron pipe, 228-9 Spherical wave, scalar, 425f
<;iio-nol nmno' · in o ~' ' r1; 11q 4?llf
Brillouin percursor in, 338 Spherical wave expansion, of, electromagnetic
cuv•uo,
." '" "
~VHHHv"v'U J-:'' vvU' OVl w, JJU •J
steady-state signal in, 338 of Green function, e'kRjR, 428

Signal velocity, upper limit on, 337 ot scalar plane wave, 4 I 1
.<::kin rlPnth ??n ~~4 nf vector nlane wave 473
and Q of cavity, 373 Spin, -orbit interaction, 552
. '- . ' ' ' < au
auu o• /
'vOlOlUHvv, JJV "' " ~· ,
L
in plasma, 313 Thomas precession of, 548f, 563-4

Smythe-Kirchhoff integral for diffraction of vee- Thomas's relativistic equation of motion for,
tor fields, 487 564
approximation for diffraction by circular Stability, of classical charged particle, and Poin-
,'
'
'
'~M
'/V'
•
, ~ca,-~c~<
Snell's law, 303 Standards, units and, 775-6
Soft photon emission, 709-10 Standing waves in resonant cavity, 368
equality of classical and quantum-mechanical Stationary phase, method of, 338
· fnr 71 ll~1 <;itpn fn · (<){ t\ ,jpfi, · · r.f 77? 111
Solenoid, 225-6 Stokes parameters, 301
.. "JL, CHU' J-'Uf/C"
l~UVL, v•, "-.."- ~CVhvO 0 CHvVL~W,
Source, localized, in wave guide, 392f Stress tensor, and conservation laws, 261, 606,
of multipole radiation, 439f 609, 611
<;inorP.lil.-P ' r.ftwr. nr.int< in onorP. ronnnirol in 4 n< ,;n,;
'
time, S28 Maxwell, 261
. ' -" .£ _; '
' ~"'
,__,v, ~/M
ovu, v• 1 •v
c.pa~vuuu.-, w ~pvvw• ,v,auvuy, " ' " ' " " '
"'
description of, 539f symmetric, in 4 dimensions, 608-9
Structure factor, for scattering by collection of Thomas factor, 552
scatterers, 461 Thomas precession, 548f, 563-4
Sturm-Liouville equation, 126 Thomas's relativistic equation for motion of
Sum rule, dipole, for oscillator strengths, 310 spin, S64
for plasma frequency, 33S Thom~on cross section, 695
Summation convention, for repeated indices, Thomson scattering, 694f
540-1 Thomson's theorem, 53
Superconvergence relation for dielectric con- Time dilatation, 527f
stant, 335 experimental verification of, 529
Superposition principle, see Linear Time-like and space-like separation, 527 8
superposition Time reversal, ;no
Surface-charge density, and discontinuity of nor- transformation of physical quantities under,
malE and D, 18, 31, 154 271
on conducting sphere, 59, 64 Torque, on current distribution, 178
effective magnetic, 197 on magnetic dipole. 174, 190
and force on surface of conductor. 43 on spin 549
near circular hole in conducting plane, 143 on spin, relativistic equation for, 561f
ncar conical hole or point, 106 Total cross section and optical theorem, 502
near edge or corner in 2 dimensions, 78 Transformation, see Galilean transformation,
polarization, 157, 159 Lorentz transformation, Relativistic
potential of, 32 transformation
on sphere with line charge inside, 124 Transformation propertie~ of physical quantities
Surface current, and discontinuity of H, 18, 194, under rotations, spatial reflections, and
353 time reversal, 267f
effective, 221, 356 table of, 271
effective magnetic, 197 Transition probability, 442
Surface distribution, of charge, 31 2 estimates of, in atoms and nuclei 44?f
of electric dipole moment. 32 4 in hydrogen-like atoms, 704, 769 70
Surface impedance, definition of, 356 Transition radia lion, 646f
use of as boundary condition in scattering, angular and frequency distribution of, 652
475 effects of foil thickness and multiple foils on,
Surface of conductor, charge density at, 21 658-9
Surface resistance, of good conductor, 356 formation length for 649
Susceptibility. electric, 154, 158f qualitative considerations of, 646 9
Synchrotron light source, 661, 683f Transmission coefficient, for diffraction by circu-
examples of, 692-3 lar aperture, 493
typical photon energy spectra from, 693 Transmission line, dominant mode in, 358
undulators and wigglers in, 683f examples of, 397-8
Synchrotron radiation, 676f relation between T and C fur 232
angular and frequency distribution of, 678, Transparency, of water in the visible region, 315
680,682 ultraviolet, of metals, 314
astrophysical examples of, 681-2 Transverse Doppler shift, 530
by charge in helical path, 703 Transverse electric (TE) waves, attenuation of,
polarization of, 678-9, 706 in wave guide, 365 6
Systeme International (SI), standards of mass, connection of, wjtb muJtipole moments, 441
length and time, 776 cylindrical, in wave guide, 359
electromagnetic units, 776, 779 in dielectric wave guide, 388
in rectangular wave guide, 361-2, 391
Tensor, electromagnetic angular momentum, spherical, 430
2SS, 608, 6 10 in concentric sphere cavity, ::ps
electromagnetic field-strength, 556 Transverse electromagnetic (TEM) waves, 358
dual, 556 ahsence of in hollow wave guides, 358
Lorentz transformation properties of, 541 Transverse magnetic (TM) waves, attenuation
Maxwell stress, 261 of, in wave guide, 365-6
wtational definition of, 268 connectwn of, with multipole moments, 441
stress, m 4 d1menswns, see Stress tensor cylindrical, in wave guide 359
Test function, in finite element analysis 79 in cylindrical cavity, 369
Theorems from vector calculus, end papers in dielectric wave guide, 388
Index 807
in rectangular wave guide, 391 of oscillating electric dipole, 410

spher ieal, 430 1 of oscillating electric quadrupole, 414
in earth-ionosphere cavity, 375 of oscillating magnetic dipole, 413
I ransverse vector held, del!mtwn of, 242 on surface of linear antenna, boundary cond1
Transverse waves, in magnetohydrodynamics, tion for, 418
321 Vector spherical harmonics, absolute square of,
, ,
inhomogeneous, 298 definition of, 431
Traveling wave solutions, 296, 324, 327, 348 orthogonality properties of, 431, 472
for signal propagation in dispersive medium, sum rule for, 438
335f Vector theorem, divergence, 29
in wave guide, 357 Green's, 482-3
Two-dimensional corners and edges, fields and Stokes's, 31
surface charge densities near, 78 involving surface and volume integrals, 482-3
Two-dimensional potentials, 7:5f involving vector spherical harmonics, 472
see end a ers
Uncertainty principle, 324, 329 Velocity, addition of, in special relativity, 530f
me of, in collision problems to obtain quan- E X R drift 586
tum-mechanical modifications of, 629, 4-vector, 532
641,642,725 of light, see Speed of light
Undulators and wigglers, K parameter of, 685 relativistic transformation law of, 531
angular and frequency spectra from, 689-93 VeloCity fields, of charge m arflttrary motwn,
average rest frame, dynamics and radiated 6fi3
power in, 687-9 Velocity selector, 587-8, 617
see also Synchtrotron light sources Virtual quanta, method of, 724f
Uniqueness theorem, for solutions of Poisson or spectrum of, for point charge, 727-8
Laplace equation, 37-8 quantum-mechanical form of, 729
usc of, with Legendre polynomial expansion, treatment of relativistic bremsstrahlung, 729f
102, 103 4, 234 use of, examples in atomic and nuclear calli-
Units, and relative dimensions of electromag- sions, 742-3 .
netic quantities, 777f Visible region, of frequency spectrum, reason
appendix on, 775f for, 314-5
baste versus denved, I 75 6
conversion between Gaussian and SI, 782-3 Water, index of refraction and absorption coeffi-
Maxwell and other equations, in different sys- cien t of, 315
tems of, 781 Wave equation, 240, 243
table for conversion of, 782-3 covariant form of, 555, 612
variant of electwmagne tic system of, 782 3 Green functions for, 243f
Helmholtz, 243
Vacuum polarization, 11-2 for photons with mass, 601
contribution to atomic potential, 12 solutions of, in covariant form, 614-5
Van Allen belts, of Jupiter, synchrotron radia- spherical wave solutions of, 425f
tion from, 682-3 transverse two-dimensional, in wave guide,
problems illustrating principles of, 619 357,360
Variational principle, for capacitance, 53 Wave guide, 356f
in electrostatics, 43f attenuation in, 3fi3f, 3fi7-8
for wave guides and cavities, 400 "hound" state in 405 6
Vector, under rotations, definition of, 268 boundary conditions in, 359
'Jv'ector field, decornposition of~ into longitudinal cutoff Lequcney in, 360, 362
and transverse parts, 242 dielectric, 385f. See Optical fibers
Vector Green's theorem, 482-3 modes, propagatmg and evanescent, 360
Vector plane wave, spherical wave expansion of, in rectangular, 361£, 391
473 obstacles in, 394
Vector potential, for time varying fields, 239 orthonormal fields in rectangular, 391
of localized oscillating source, 408f sources in, 392f
of magnetic dipole, 186 TE and TM modes in, 359
of magnetic monopole, 278, 290-1 variational methods in, 400
in magnetostatics, 180, 195 Wavelength in wave guide, 361
808 Index
Wave number, and freauencv. as 4-vector 530 Whistlter 110 1dO

connection of with fr""""nrv ~' d- ur;~nl~r . T T. ,1.1
'
"
A
'Jl){; 10d ,;:,--..,-, (l :,,.,., n;o';' '~• . . ·oo =

vu-..,' -uJ lllill CllCI gy, -:\tT
imaginary part of, because of losses, 310, 363, Work function of metal and image charges, 61
367 World line, 527
spread of, in wave packets, 324 Wronskian, definition of, 126
in wave guide, 360, 364, 367 of Bessel functions, 126
Wave oackets in nnP. r1; · 1?.?f 1dR Af -' • '1 D , 1 +, A'J"7
w~tinn in rl' . "~ ~n~;,. '1')')+ n" _,.,T "'~ . " " '
;. .+ . ~~/r ' au, HUH ' rum or cavny
Ul 11' ~~
"!' '5 = , JLVl

, wiw1osscs, jfL
Waves, Alfven, 321 finite, of frequency spectrum of oscillator with
in ionosphere, 316f damping, 764
magnetohydrodynamic, 319f
magnctosonic, 322 x-rays, polarization of, in bremsstrahlung, 712
onn n/on l>l,n, n ' .1 " .
•-r '"' = 11vm >yucrrr -ugrn ~ource~, o~
Weizsacker-Williams method, see Virtual Yukawa potential, for scalar potential if photon
quanta, method of has mass, 601
"11""I7• ... •IDII...... .,.
--- •
IVI~•Prl~l
.
"" n~rP ••• APV
.J ---
on Special Functions
SPHERICAL
Legendre polynomials P_dxl _21_-_101
Associated Legendre functions P7'(x) 108
Spherical harmonics Y 1m(e, 4>) 108-9
CYLINDRICAL
Bessel functions JJx), N ,(x) 113-4
l\Jfr.rlifiPrl RP<:<:Pl f1mrtir.n<: f {y) k' fv\ 11 h
C'. • L · .1 n ' l "· ·7.~ : ' : · ' , 11?.)/ ' f
·r _~v~~v~ <' '}/~~}, fl{~A), "t ~A) '-tLV· ·t
Roots of lm(x) = 0 114

Roots of1~,(x) = 0 370
Identities involving Bessel functions 126, 132, 140, 205
Airy integrals, connection to Bessel functions 678
ORTHOGONALFUNCT~EXP~N~IIJNS
Bessel function (finite interval in p) 114-5, 138-9

Bessel function (infinite interval in p) 118
Eigenfunction, of Green function 127-8
Fourier series 68
Fourier integral 69-70
T <>nPnrlrP nr.l ;.,] nn
JJ
" ;" • -' ,'" 'J

l l \ - a t HUL
·r-- .lU';t
Sphencal tlessel tunctwn 119

Explicit Forms of
Vector Operations
Let e 1 , e 2 , e 3 be orthogonal unit vectors associated with the coordinate directions

specified in the headings on the left, and A 1 , A 2 , A 3 be the corresponding com-
ponents of A. Then
V. A-
3
V·A
V ,,, =
1
+ e 2 ~ alji + e 3 1 aijJ
e aijJ
'I' ar r ae r sin e a¢>
1a 2 1 a. 1
V ·A = 2 - (rAt) + . - (sm8A2) + .
r ar r sm e a8 r sm e
V x A = e1 1. [a
-
r sm 8 a8
(sin 8A 3 ) - aA 2 ]
-
a¢>
+ e2 [ 1. -aA1 - -
r sm 8 a¢>
1-a (rA 3 )
r ar
J+ e 3
r
[a
-1 - ( r2A ) aA~J
ar
--
a8
v2 1/J =1- a ( r2 -alji) +

- 1 -a ( sme-
. alji) + 1 -alijJ2
2 2
r ar ar r sine a8 a8 r sin e a¢>
2 2
1 a ( r 2 -alji) = -
[ Note that--
r 2 ar ar
1 a
- (rlji).
r ar 2
2
J

Classical Electrodynamics (Jackson) (3rd Ed. 1999)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Classical Electrodynamics (Jackson) (3rd Ed. 1999)

Uploaded by

Copyright:

Available Formats

Vector Formulas

V X (a X b) a(V · b) b(V ·a) + (b · V)a (a· V)b

John David Jackson

JOHN WILEY & SUNS, INC.

This book is printed on acid~free paper. (OOi

Copyright© 1999 John David Jackson. All nghts reserved.

No part oi this publication may be reproduced, stored in a retrieval system or transmitted

To order books or for customer service please, call I (800)-CALL-WILEY (225-5945).

Library of Congress Cataloging-in-Publication Data

Jackson, John David, 1925-

Printed in the United States of America

relativity and relativistic electrody,namics than SI. As a reminder of the sys-

and models of classical char11:ed narticles as well as chan11:ed emphasis.

Berkeley, California, 1998, 2001

magnetism needs almost a book in itself.

mostly ot a macroscomc sort. Plane waves m different media mcludmg plasmas

classical in nature. even though involving smalJ-scale col!jsions A student who

Introduction and Survey 1

Chapter 1/ Introduction to Electrostatics 24

Chapter 2/ Boundary- Value Problems in Electrostatics: I 57

2.8 Orthogonal Functions and Expansions 67

Chapter 3 I Boundary- Value Problems in Electrostatics: II 95

3.2 Legendre Equation and Legendre Polynomials 96

Y,,.(e. q,) l07

Hole 129 - , ~vuu• I ld11C W ll11 d ~11 C,Uldl

n -•· no -/ (", ,. -1."

Chapter 4 I Multipoles, Electrostatics of Macroscopic Media,

4.7 El lVl. E ''. D' I ''_Y M

5.3 Differential Equations of Magnetostatics and Ampere's Law 178

5.6 Magnetic Fields of a Localized Current Distribution, Magnetic

References and Suggested Reading 223

Chapter 6/ Maxwell Equations, Macroscopic Electromagnetism,

Chapter 7/ Plane Electromagnetic Waves and Wave Propagation 295

7.7 Magnet~hydr~dynamic Waves 319

f'l lil I U!. . -1. n ,, r. .. nnrl fl ,,,- rl v:J. '210")

8.1 Fields at the Surface of and Within a Conductor 352

8.12 Expansion in Normal Modes; Fields Generated by a Localized

'1 'b J7J

Chapter 9 I Radiating Systems, Multipole Fields and Radiation 407

9.3 Magnetic Dipole and Electric Quadrupole Fields 413

9.tl Properties of Multipole Fields Energy and Angular Momentum

~-~ Angular Uistnbution ot Mult!po!e KaOlation 45/

Chapter 10 I Scattering and Diffraction 456

Chapter 11/ Special Theory of Relativity 514

Chapter 12 I Dynamics of Relativistic Particles

12.1 Lagrangian and Hamiltonian for a Relativistic Charged Particle

12.3 Motion in Combined, Uniform, Static Electric and Magnetic

References and Suggested Reading 615

Chapter 13 I Collisions, Energy Loss, and Scattering of Charged Particles,

13.1 Energy Transfer in Coulomb Collision Between Heavy Incident Particle

L~.~ ..,""""u"1S v• , "" '"'""'1S"'u , "' """'" uy fiLum~

14.1 Lienard-Wiechert Potentials and Fields for a Point Charge 661

'"'·-' · · JuLion in rrequency anu fingw 01 l::,nergy KaciJatea

Chapter 15 I Bremsstrahlung, Method of Virtual Quanta,

Chapter 16/ Radiation Damping, Classical Models

Appendix on Units and Dimensions 775

aescnpuon ot electromagnetic, weak, ana strong mteractwns based on tunda-

ated throu!'h nrin, · of continuous (rP.:<llv nh~cp) in· of thP

From the nomt of v1ew of the standard model. classical electrodvnam1cS IS a

Despite the presence of a rather large number of quantities that must be

1.1 Maxwell Equations in Vacuum, Fields, and Sources

continuous distributions in x, although we consider from time to time localized

course, some e 1cate macroscopic or a most macroscopic expenments m w 1c