Professional Documents
Culture Documents
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
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
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.
97-46873
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. -\, ,], ;,, ~
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" '<
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
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
-· '
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
XV
xvi Contents
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
.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<;
Problems 396
m Wavegm'd e 419
lJ.n Snheric~l Solutions of the Scalar Wave Eouation 425
n ~ 'K. _' ~- £ ,,_ ~' • ~·.'" ,.-,n
-'· •vn.. " ' v• "''-' ~''-''-'" .. w .. • H.-<uo ~~-'
£".. .. ~ I Jn .l.• • • • I o ,.
.... "'• _..~ 1 ·~·· uy II'.Iuvutg \-tturges 00.1
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
lUll,, VV allU L LJU,UI", allU giUUI". lllt: tneureucai u; IUIK " gener-
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
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
4 Introduction and Survey
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
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.
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.
6 Introduction and Survey
Figure 1.1 Cavendish's apparatus for establishing the inverse square law of
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).
8 Introduction and Survey
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
"' \
~ c;;;,:t:l Lr--,. J
\
\
,..-
1
\
'I(-'
~ I 4 MHz
Voltage to I
J
IJ
Calibration
,,.,.,
_j-- 1 ---
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 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"',
10 Introduction and Survey
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
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-
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
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
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
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
- c E
D a-~oa + P,.~ - L...
" aQ~f3 + ...
.L'+
. mtroaucnon aua ;,urvey
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.
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
. .
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.
*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
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.
~
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
16 Introduction and Survey
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.
{ {
'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
n' a unit normal at da pointing in the direction given by the right-hand rule from
H · dl = · n' da (T.15)
c
E • dl = (1.16)
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
. . .
18 Introduction and Survey
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
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
(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
I.
relations of such idealizations to the real world is perhaps worthy of a little dis-
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:
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
20 Introduction and Survey
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
*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."
• • cy'), .1.. c . ·~
simplmes matters!
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
.
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
..
~ . "' • ~ ~
v~y~,'
o"r" ~
~¥"~·"·
~ . ~-
, "''
" __ ,_
'"""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
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.
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)
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
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
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
charge density (1.6) into (1.5) and integration, using the properties of the delta function,
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
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
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)
fs
E · n da = ..!_
Eo
J,v p(x) d x 3
(1.11)
.t..
'i
1 c.
1~1
•• U<v <~<~v Uv< ovvH ~·n:rrg=,
~
. ~ 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
( 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).
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
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<I> 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
<l>(x) = -4-
7TEo
f
S
u(x')
IX - X
'Ida
,
(1.23)
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
+ ···
<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
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.
34 Chapter 1 Introduction to Electrostatics-SI
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)
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
- " f •• \
1 ( p(x') _.13 ' (, 1 '"7\
~,~,
'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
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
-. n ,... ,.
.L.O u . ......... .)' .L /to-., .......
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<P <I n<~rtil'nl<~r ,{, ~n~~lu 1/R 1/lv v'l •v
is the observation point and x' is the integration variable. Further, we put cp = <1>,
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
~
<ll(x) = __!__
47TE0
J.v f:1_XJ
R
d3x' + l Ts
47T s
1
Ran'
a<P - <~> _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).
(1.37)
38 Chapter 1 Introduction to Electrostatics-SI
With the specified properties of U, this reduces (for both types of boundary
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
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
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 (<I>)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 <P v(X, vJ ana 1.fJ !Jfx ,VT, wnerev 1s the mtegratwn
uqr;qhla <;:;n~a tha ~. 7 7 ·
aue Lu a uniL poim source, we symmerrv mererv represents me onvs1ca1
•~nnP~h;J;tu ~f thP c~Hr~P •> 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 '.
-~ (.
~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
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)
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):
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
42 Chapter 1 Introduction to Electrostatics-SI
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 \
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
•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.
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
where the function !fi(x) is well-behaved inside the volume V and on its surface
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)
- a
1 -
pap
a+)
( p- =
ap
-g(p) (1.70)
(1.73)
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>
0.7 I I I I I I I I I
~
.... ....... .
... -~
.,
0.6 f---
..... ":'-.~
.-· ~ ·--~
0.5 f--- ,:;.- -
..
~: ...
F""" .·•
tI
0.3 f- /
.
•
Charge density
•
seal~\'\.• \
I!\
" I!\ -
-
,_
. / (arbitrarv
V.L
I ·, '\:-
<I>( o)
I
.'
"-.. X·.
"'\.·.
i' "·,
0.0
_,
·- ...... __ ......
__..
~ .I
-0.1 f-
.I -
_n?
0.3
0.0
-- 0.1
_,./
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
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
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
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
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,
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
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)
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.
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
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
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).
52 Chapter 1 Introduction to Electrostatics-SI
(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
.
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
(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
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')
~ da ~ da'<T!x\G!x x'\<Tix'\
7s,
c-'[<T] = 7s,
r 2
.. '-
'] ~.,,
- '-' -" _/
,~,
--' .,, .
~
']
,_
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
'
(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
~
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
can oe expressea as
zV V
4 4
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
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
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.
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
q '/47TE0
(2.3)
/p
I
/ -.J
I ~a
t.. //q~
- a
.....
/ I
y
(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.
60 Chapter 2 Boundary-Value Problems in Electrostatics: l-SI
¥ 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
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)
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.
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)
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).
Q141TE0
2 rR cos 11) 112 (2.12)
aQ141TE0
112
a4 2a 2 r
+----cos IJ )
2
R R
where <I> 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:
<I>= _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
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.
64 Chapter 2 Boundary-Value Problems in Electrostatics: l-SI
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.
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
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<I>(a, f)', cp ')
a(x - a )
(x 2 +a-
2 2
2 2
+V--. / ~
/
....... \
::----- - --~
/ ~ y
""""
__/( ~J
""-. _/'--V
~
X t'il!nre :Z.!S
66 Chapter 2 Boundary-Value Problems in Electrostatics: l-SI
From (2.19) the solution for <P(x, {}, <P) is given by the integral:
.? ,,
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 <P = 3Va2/2z 2.
T. .C~L ,J. rl .L ;', +J. 'In :n (111 \ mP r-~n PV.
.,. v· '
"'
pand the denominator in power series and integrate term by term. Factoring out
r
/ 2 2> 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
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
(2.28)
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),
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)
68 Chapter 2 Boundary-Value Problems in Electrostatics: l-SI
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; ~
sion in terms of them heinP" a Fourier series. If the interval in xis r -n/2 n/21 thte
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)
1
L
00
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)
__!__
271'
r
J -oo
e;k(x-x') dk - o(x - x') (2A7)
-+ -+ -= 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
X= sin ax }
Y sin{Jy (2.53)
Z = sinh( V,-a,- 2 z)
2 -:-+---c/3=
To have <P - 0 at x - a andy - b, we must have aa - n7T and f3b - m7T. With
the definitions,
n7T
an=-
a
m7T
f3m = b (2.54)
/, 2 2
'Ynm = 7T ~a2 + b2
we can wnte the partial potentJal <f>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)
(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\
, .. ''
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).
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!
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, 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).
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, <P ), the Laplace equation in two
. .
R dp P dp + 'IF d<f> 2 =
Since the two terms are separately functions of p and <P respectively, each one
p d ( dR)
R dp Pdp = v, -v (2.68)
X
Sect. 2.11 Fields and Charge Densities in Two· Dimensional Corners and Along Edges 77
These are the building blocks with which we construct the potential by linear
..
>up"~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
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..
m7T
v = B , m = 1, 2, ...
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
' 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.
78 Chapter 2 Boundary-Value Problems in Electrostatics: l-SI
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
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
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
'
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-
80 Chapter 2 Boundary-Value Problems in Electrostatics: l-SI
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 ,, "
~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
a1 + b 1x 2 + c1y 2 = 0
UJ UJ.>] ' qYJ U
a1 =
2~ e
(x2Y 3 - x,y2)
1
b1 = (h - h)
25e
-1
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
T ·1 -1 tl- D ~on
LIH; ·r ~ -0 u
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
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)
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.
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-
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.
86 Chapter 2 Boundary-Value Problems in Electrostatics: l-SI
(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
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
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 =
<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.
88 Chapter 2 Boundary-Value Problems in Electrostatics: l-SI
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,-, ·
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
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'
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
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
~
"'"'" 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
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,
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 and 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
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.
densities ±A, as shown in the left-hand figure (the right-hand figure shows the elec-
tric field lines).
(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
Hint: You may wish to add an integral that vanishes (associated with the image
point) to the integral for the point inside the circle.
92 Chapter 2 Boundary-Value Problems in Electrostatics: l-SI
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
Va 2 a2
1'-,(Z) (j+-:;
~~ T 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.
94 Chapter 2 Boundary-Value Problems in Electrostatics: l-SI
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
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
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
!! 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.
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
P(x) = xa f
j=O
axi (3.11)
(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\
*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
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)
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
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
(~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)
U1(x) =
'I
1
t -r
2
1
P1(x) (3.22)
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)
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
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
100 Chapter 3 Boundary-Value Problems in Electrostatics: 11-SI
'
,,.., ,,..,
""
~
"'"" Ul<11
(x 2 - 1) dPz - lxP, + lP , = 0
ax
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
(3.33)
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,
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: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)
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
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
<P(z = r) = _q_
47rEo
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.
Figure 3.5
Sect. 3.4 Behavior of Fields in a Conical Hole or Near a Sharp Point 105
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.
~"' ~
' .
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
) = 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
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
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
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·
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
108 Chapter 3 Boundary-Value Problems in Electrostatics: 11-SI
( 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)
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 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 "
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<P
1 {105
I 32
4
~
2 7T S!WtJ COS tJe
l = 3
~
4 ~.~'
2
y 31 = _::. 47T sine ('icos
. e - 1)e'"'
I 7 ,,
~ " cos tJ)
4 7T ~2 cos·"
I 30 2
>.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)
110 Chapter 3 Boundary-Value Problems in Electrostatics: 11-SI
where
Am =
fi;if dD P1( cos 8)g( 8, 4>) (3.60)
<P(r, 8, 4>) = 2: 2:
l=O m=-1
[Aimr 1 + B1mr-(l+l)]Y1m(8, 4>) (3.61)
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
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
-
z
x'
8' Li'Y ~X
I /
~I
y
X Figure 3.7
(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.
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.
(3.71)
112 Chapter 3 Boundary-Value Problems in Electrostatics: 11-SI
7
:
_/ ' '1 "
- - + v2 Q = 0
d¢2
(3.74)
(3.75)
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)
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)
(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).
114 Chapter 3 Boundary-Value Problems in Electrostatics: 11-SI
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
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<" <'.
fl U.
~·
UlCO
. .
Ml<lllVll Milll'
.
W llll Ult;;
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
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:
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 ?
Neumann series:
n=O
2: anlv+n(Z)
n=O
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
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
Xmn
k
a
~ ~
<:fl(p, ¢>, z) Z Z
m=O n=1
lm(kmnP) sinh(kmnZ)(Amn sin m¢
m.n
z
/ t = V(p,c/>)
Ll
R
/
'
t=O- ;...,...
a
~-
~,.~-- ---,
/ ' '-t-0
-r
/ .... c;y.,
~~0
~LLO \_,lldfJltOl CJ • 'ilJU" C Ill
. .....,,.
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
<I>( 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~
<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
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,
[Details of spherical Bessel functions may be found in Chapter 9.] The ortho-
O"Ofl:l 1itv H'>1Mion (:\1 mn t1v 1' ·~
r~ ~
(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
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
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)
' '
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
G(x, x 1
) = L 2:
l=O m=-l
Azm(rlr 1
, 8', c/> 1 )Yzm(IJ, cp) (3.118)
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
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
\ I
\ I /
t \ I /
. ->,111 n;.
r'
122 Chapter 3 Boundary-Value Problems in Electrostatics: 11-SI
Similarly
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 \
) 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.
(3.127)
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
(3.130)
= l{ L P,(O)r~( r'+I
47rE0 t~o
l - ...!...::..._
b2l+J )P,(cos fi)
>
<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 ~ '"
124 Chapter 3 Boundary-Value Problems in Electrostatics: 11-SI
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)
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:
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.
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
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:
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)
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)
1 =-
4 J~ dk cos[k(z - z')]
I
x - x' I 7T o (3.149)
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
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)
r i/l!(x)if!n(x) d x = 3
omn (3.155)
*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
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:
. . . .
1nese eiv ,,Ions nave ueua 1uncuun normanzauun:
1 1 { ,, eik·(x-x')
lx- x'l 27T2 Ju 1\,
e ~.J.~U'+)
!/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
' 32 (3.167)
G(x, X ) = 'TTnhr
m7ry'
l.
X '\' u u u u I \ ~ \ ~ I
LJ
l,m,n=1
z2
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
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
<I>= + (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
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 < p <a). For points on the conducting
>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 < p <a
az z=o+
and (3.171)
;, 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
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)
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,
*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
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
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.
134 Chapter 3 Boundary-Value Problems in Electrostatics: 11-SI
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
for 0 < p <a (and zero, of course, for p >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).
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.
136 Chapter 3 Boundary-Value Problems in Electrostatics: 11-SI
(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) 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:
~here cosy cos 8 cos 8' +sin 8 sin 8' cos(¢ ¢').
(b) <P(x)
(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
/ 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
" ;, (\
'
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
138 Chapter 3 Boundary-Value Problems in Electrostatics: 11-SI
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)
V(¢,z)={-~
for - 7T!2 < ¢ < 7T!2
for 7T/2 < ¢ < 37T/2
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
f(p)
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'. ,].,
·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
ln=-cc
(d) From the last result obtain an integral representation of the Bessel function:
.T,(x)
4 ~
= e sm
L n=l !n=-cc
Ch. 3 Problems 141
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
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
pmm z
•• r_,
"'
n,
" "' T -' .. -• '6'
•• '. ··-
IJ..\
Zo·
f""'oJo,.Jo+a +ha ' ,,4, ... ' .... , .. / o\
'V\1-'J
< ... 'L\~
' -\ +l. L
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'-•
,.
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
'""·
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:
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
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.
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
<I>(x) Eo 1 - p' \1 + 2r3 I
(b) Calculate the Cartesian or cylindrical components of the field, E 2 and EP, and
,], ' ''" ,], ,],
l'
·"" " . ' . ,_ '· . :,
01 p li.J.
HAPTER4
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
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
(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)
' '
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
p
J
f x'p(x') d x' 3
(4 8)
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 <I>(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)
(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
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
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)
(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
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
150 Chapter 4 Multipoles, Electrostatics of Macroscopic Media, Dielectrics-SI
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)
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)
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.
"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
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
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
M>(x, x') = - 1-
47TE0
I p(x')
Llx - x' I
ilV + P(x') · (x -, x')
lx - x' lo
ilv] (4.30)
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-
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
q
156 Chapter 4 Multipoles, Electrostatics of Macroscopic Media, Dielectrics-SI
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
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:
From the boundary condition at infinity (<P --'> - 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
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
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
. .
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
<'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.
(4.57)
It is constant throughout the volume of the sphere and has a volume integral
(4.58)
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=
classical analysis.
160 Chapter 4 Multipoles, Electrostatics of Macroscopic Media, Dielectrics-SI
~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)
1
E; = - P + Enear (4.63)
--::rEo
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]
(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
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:
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:
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.
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
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
*See, e.g., CRC Handbook of Chemistry and Physics, 78th ed., ed. D. R. Lipe, CRC Press, Boca
Raton, FL (1997-9g).
164 Chapter 4 Multipoles, Electrostatics of Macroscopic Media, Dielectrics-SI
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
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)
7 Polar
t r r T T T. ~
'Ymol ~~~~Nonpolar
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.
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.
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:
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>
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
I = 2
1 r<PV • (D - D 0 ) d-3 x = 0
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
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<P 1 d 3 x (4.98)
2
where 8<1> 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
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
"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-
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
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
the sphere.
4.10 Two concentric conducting spheres of inner and outer radii a and b, respectively,
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
1 1.82
103 1.96
4 X 103 0.796 2.12
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 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
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)
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
(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
(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
bon 5.7.
--
;),.J
OPP
uwer • • p
!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~ •
B(x) = A~ V X
!Lo r I~J(x')~'I d 3 ' X (5.16)
-,-
(5.19)
1 \ I 1 \
ana
I
V X B = JLoJ + !Lo V
47T
JV'lx-x'l
• J(x') d x' 3
(5.21)
" \ \ \ \\ \ \.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.
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
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(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\
*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
'-... <I>A /
I
I
X Figure 5.5
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
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:
k
- 2E(k) l (5.37)
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
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
~"
..) ,~- .. ' , / ..)
~ ~
-=--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 • " •
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.
*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
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
-~ 2: xi }
I 3
(x;Ii - xjl;) d x'
186 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-SI
Then the vector potential from the second term in (5 51) is the magnetic dipole
vector potential,
/.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
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;
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- • • • • ,, • • • ....
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
"' I I
188 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-SI
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).
r; - 2..
jk
Eijk
L
Hk(U) Jf lj(x') d 0
X + J1 (x 1
)X
1
• VB;(O) iFx 1 + ... (5.66)
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 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
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*
*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).
192 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-SI
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
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~
,. _...- I / -
( I I I H
)__/v
I I
., ::
<::R . '~~~n R ;~ o
·o
. ' . ' ·:o .£0T
<~HV
194 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-SI
Figure 5.9
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.
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.
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
matched across the boundary surfaces using the boundary conditions (5.88)
or(<; RQ\
Again, this is a very complicated differential equation unless the medium is linear,
Ill <.:ase llle
.-, I ..,,y., ' n ( <:: ()A\
\I-" M! \' • 'I
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.
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 •
""
I I I I
c L _, .£.
,. " ' u<v V< 'J "' 00
r n• v 1\Jrf, '\
A(x) =j
41T lx- x'l
d 3 x' (5.102)
' '
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.
198 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-SI
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
<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<
<I> 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~·
We note that B;n is parallel to M, while H,n is antiparallel. Outside the sphere,
< U d11U I> , 111<0 jJULC111ldl " UlU'
1 cos 0
'l'M IVI 0 U ().lUbJ
3 r2
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<
~· ,. .£ ~
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'
1 M0
<I>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 <I>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)
200 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-SI
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~- '"'• ·~-
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)
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
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
(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
(5.121)
Sect. 5.13 Effect of a Circular Hole in a Perfectly Conducting Plaue 203
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
. .
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
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
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 < p <a
act>M - 0 at z - 0 fora<p<oo
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)
(5.128)
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
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)
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
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 '] ~
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,
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
(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)
210 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-SI
Figure 5.19
f. [E'
c
k(v X B)] · d! k
r iJB
Js iit • n do (5 138)
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
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)
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)
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
UT:>L
,. .C n '
, •'- •'-. L.
"'"
.,
Af>;U!\- T A~~ >;A. r/1
/
Jc
vector potential is
SW = r SA· J d x 3
(5.144)
.,.,,
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)
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)
214 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-SI
(5.149)
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)
(5.151)
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'
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 ·'
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)
u eilller me conuucwrs carrymg the current are permeable or the medium be-
...
'-'"'-'~
. . '"'" lltal.
lumped circuit equations follows immediately from relating the time derivative
of the linked flux (dF/dt) to the induced emf jg through (5.135).
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
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),
*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_
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~
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
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.
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
i5 = I2
VJ.UTW
(5.165)
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
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
-· ~
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:
~~~
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)
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
h(k) = --;;
H0 f" a e ''Z
.
1 2H0
dz = 'TI"k sin(ka) (5.173)
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
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.
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
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
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
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
~
-=
, ... ... - ..~
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 -,
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
az z)J
[a B,(O,
2 2
p )
B,(p, z) = B,(O, z) - ( + ···
4 2
(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 " ·
(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 ..,.
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
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
(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 =
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-
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
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
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
W = ~0 I 3
H ·H d x = -~0 I 3
M ·H d x
(a) Show that, when it is placed with its flat end against an infinitely permeable
plane surface, it adheres with a force
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
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
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
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
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
234 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-Sf
~-
"'<P uu~ " ' ' w~w• U<:Oll Vall V<;; a l {J u, LJidl lll<O ~Uli<Olll >0 mm
aensuy mswe tne w1re:
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,
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
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>
H UJll JlUW UJl, j VV lUI LllC "b' UL VCII > U<OJ" ~uu ~m, cu~ u~~uy
·'·'· . I kr
~ Sill tJ In e . ]t~tc)}l\ n ux.
(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.
236 Chapter 5 Magnetostatics, Faraday's Law, Quasi-Static Fields-SI
(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
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
237
238 Chaoter 6 Maxwell Eouations Macroscooic F.J, ...
d. d_ '" ,_ - u ' ' ' '
'1" ,, UW< <H'-' '-''-jUUUVHO n u HUH.< lUl lllllC-UC\- ;Ill llClUS,
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
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.
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
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.
*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
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,
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• uv<u "'~
£_ ··-
" ' " ' <uu< uuuo'~"~ >etuwuuu <llCO IOIVCOII uy LIICO
iJA
E = --
iJt (6.31)
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
*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
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,
••
;.~
r .
representations,
r .
27T -oo
rr:. 'n\
. I
'
f(x, t) = -1 f(x, w)e-'w' dw
27T 00
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
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
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)
244 Chapter 6 Maxwell Equations, Macroscopic Electromagnetism, Conservation Laws-SI
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
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
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
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
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
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~•
·~·
<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
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
'
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-
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
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)
1ne eqmvalcnce ot tne two sets or expressions for the fields follows from some
>4'n1 ~l ~L
_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 ,~, ,~, ,~, ·" · ·
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\.
... 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
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
;, r ov I ov\
v' _tl - _j_ "' _l
ax J iJx- \axJ
( 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 . . .
(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
'l7bound = 17,(x, t)
n
(molecules)
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-
252 Chapter 6 Maxwell Equations, Macroscopic Electromagnetism, Conservation Laws-SI
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.
(Q~)af3 =3 (6.77)
Sect. 6.6 Derivation of the Equations of Macroscopic Electromagnetism 253
In terms of these multipole moments the averaged charge density of the nth
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)
'Consider the averaging of the charge density of the nth molecule shown in
" <
·~·
•'-
~ ~ ~·
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)).
Sect. 6.6 Derivation of the Equations of Macroscopic Electromagnetism 255
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)
(6.90)
(molecules)
=p (6.91)
a -uo< a f5 'I'
~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 •
(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
\
\molecules)
(L nL\
\' . J!
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
__ ,"""'
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. • ~· ,, . .
Sect. 6.6 Derivation of the Equations of Macroscopic Electromagnetism 257
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 .•
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 ""'
'
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
1
Jv
_j_ • __E___d_3 y - r
Jv
F. . ru ')( H\ F ..
an
at _
_d3_y_ f{:.
\- .1 fiA\,I
1 '"'
2 \'
..,. n TT\
., {{:. 1 {\{:.\
\ .. 'I
-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
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)
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)
We may tentatively identify the volume integral on the left as the total electro-
magnetic momentum Pfieid in the volume V:
(6.117)
(6.118)
- a T"~
3
-d (Pmcch + Pfield)a -
- "L,
") d X (6.121)
dt ~ v ax~
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
'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).
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)
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
1 [d(wE) ]
(6.126b)
*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
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]
*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
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
can be written in terms of the Poynting vector by using (6.134) applied to all of
space on the outside of S as
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
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
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>
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
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
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
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 " ' ' '
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
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
272 Chapter 6 Maxwell Equations, Macroscopic Electromagnetism, Conservation Laws-SI
1v1axweu equa,wn, v ·" p!E0 , smce Dotn S!Qes must transrorm m the sam"
m · thf'
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
w cm1"'
.ul'u'"'''u'"'u'y l'uwc1
-timP ovm m11ot hP
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
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
*'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)
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).
(6.154)
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.
<;: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
'
*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.. ••• _,
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
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
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 \
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/
~
~
\
\
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,. ' ,.
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.
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.
~
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,
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)
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
(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 ~
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
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
q
A
K ]](Jeer d K
- d' ,.,; 1
"' 4m"o R2 ret c at R2 ret
c2iJ 2 , . .
1
Jrct
(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
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
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<I>. 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
(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
[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
5
3 E-E0 a
<PM = - EoEo<u ( - ) · xz
5 (
E + 2E0 ) r>
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
.'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~
1 I '"
I II
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
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
(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-
where p0 is the n~sistivity in th8 absencg of the magnetic field and R is called
(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)
./ !
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
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)
292 Chapter 6 Maxwell Equations, Macroscopic Electromagnetism, Conservation Laws-SI
(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 <I> 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
m = ~p x v
and
(), - 1(r.,.TJ. + .n.\ ?._ • n,l;
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
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
(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
[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
. . . .
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
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~_
sentially kinematic; those imposed by the Maxwell equations, dynamic. With the
1tion thM thP nh ·~ol -' -•. on.-1 ,f· f>,],-1, -< . ,,-ll ,,-
"'
'"~'~ ""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 )
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
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
"....... /
/
y y
n----
~
\ ~)
/~
E
cB'\ 1\
/ X
~ ~f X
I ~ / \~
\( v \ ~\/
Figure 7.4 Electric field and magnetic induction for an elliptically polarized wave.
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 tiue<u jJUJdJ ;£atiuu
and the circular polarization bases, in terms of the projected amplitudes (7.25)
,.1 ,], o ,Ji~ith 'n 1orn>e 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
*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
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
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
. . .
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.
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
n
/k'
;J.'t' v /
" J
j.IE
/ ~
(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
B'
k'
•
B
(a)
k'
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
306 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-S!
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" '
\ !.'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.
: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
; - 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·
__ T ____ 17\______ _
f 1/ \ '
S·n = -
2wp.!
Re[(n · k') IEbl 2
]
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
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.
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.
are neglected in (7.49). We make the additional approximation that the ampli-
. . _, . u "- ~
1~
p -ex- V·JVj
_, . .,_~_,
molecule, and that, instead of a single binding frequency for all, there are f,
310 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-SI
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)
a
2
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 .
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
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.
Sect. 7.5 Frequency Dispersion Characteristics of Dielectrics, Conductors, and Plasmas 313
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
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
314 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-S!
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.*
*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)
Sect. 7.5 Frequency Dispersion Characteristics of Dielectrics, Conductors, and Plasmas 315
~
~ 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
316 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-51
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!).*
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
e
X= ----~----E (7.65)
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
318 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-S!
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-~-----
ionosphere (schematic).
Sect. 7.7 Magnetohydrodynamic Waves 319
k_ ~ wP
c
I
y w8
w
~
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
. . . . .
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)
-
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'-'
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.
322 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-S!
~
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
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
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
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.
324 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-SI
(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
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
I "gil\ I I
I I I \r 1
~ 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
.£~ • • • • •
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
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)
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
328 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-51
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)
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
(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
"" '""~ 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.
330 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-51
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 ~·-·
With the assumption that the orders of integration can be interchanged, the last
expression can be written as
'
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~
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.
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
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
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
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.
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
p, ,] tJ. ,f tl-
'" . tJ. -'· f, th
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 •
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
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!.*
*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.
. . ' . _,_ . . _, ~.
is the Fourier transform of the real incident electric field u;(x, t) evaluated just
. .
OULsiuemememum,aLx v . '""wavcou1
' " ' .
"~wJ"
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
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
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
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
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).
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).
340 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-S!
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;
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
w ere
2
Sketch log T as a function of (D/8), assuming Rc y- 10
Define "very small thickness."
342 Chapter 7 Plane Electromagnetic Waves and Wave Propagation-51
(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
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
(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
\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-
'"" """'
•
.
n
"
there IS a range or . or 0
• •
:
•
• w1 w w-.u
•
,fl
•
' "
'" f<>r
. '
'", '"'u.,
L £
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 • '
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
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 "'' '
wP2 - wP2 wB b • n
E+/E 1 +
w w
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
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
7.21 Consider the non local (in time) connection between D and E,
(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
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
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-
,~, r
o(o. w)
,mr 7 2o 2 ( rf'n r~ I
I 1 aw w lm 1o( w q. v)
-= f7T
3
'J
2
'V . C\ 'f' ~,
~
_:'_
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:
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
. 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 .
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
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)
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.
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
E - 'J J-LcW (1 -
LIT
i)(n X H,,),-iCI8,ii;/8 (R 10)
ThP<P <nlHtion< frw H ~nrl 1?. · . ,1, thP -'· ·+~v •.•..
y
;~c r!;cm•cc~rl •h~
~y
in Section 5.18: rapid exponential decay, phase difference, and magnetic field
much lars>cr 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)
(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).
356 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
~·· _, . . . . . .
' " ' ' ''-''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
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.
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
Because of the cylindrical geometry it is useful to single out the spatial vari-
[V't + (f.LEW
2
- e)]{:} = 0 (8.19)
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
358 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
(8.26a)
and
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
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 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
/'A
~0
t:JJA
v;;, . .)0)
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
---------------~-~-~-~~
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 ~
>--
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
362 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
"T'L _, • .L .,_
•'- '"'
.. "
_,
~ L ,._ .. ~ ~
~Uv H <HUC
U'v -r vU>Hv' '~ ' v y VJ U'v <HV Jc" '' e
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(
. ~
~n 0 1 2 J
m~
0 2.00 4.00 6.00
1 J.UU L.L'+ '+. l.J
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
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
Poynting vector: --
-- ·oo
' 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.
364 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
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)
1 -2 = Vg (8.53)
U W /-LE y;;k 'V w
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
'
1 dP
{3,\ = --- (8.57)
2P dz
_.1 DU . . c±J, . ,,
"'' ·~ •• Jc' ~ n-
~·
H<
• '1
1''-'1 UHH
"_i
"
Vl ll1CO
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:
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
~
...
"t thP rntnff frP()llPnr\1 "n£1 J:. 'YL "rP dimensionleSS numberS Of the order
'J.
1:), 'T'l\"
" • , 'lA
()
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"
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
(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
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)
(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
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
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~
A sm KZ + jj COSKZ
I. 7T f A ~
" l' A \1' v, , L, .•• ) ~ o.1 J 1
~ ' 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
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;,-,.,.
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
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(
Roots of J ,;.(x) = 0
m = 0: xb.. = 3.832, 7.016, 10.173, ...
m = 1: x;., = 1.841, 5.331, 8.536, 0 ••
-"
f. '?()f.
0 ' ...
m 3: x' 4.201 8.015 11 ,,,.,
Sect. 8.8 Power Losses in a Cavity; Q of a Cavity 371
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~'"·
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)
372 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
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,
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):
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 ..•
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
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
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)
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
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.
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.
where the relative permeabilities of the medium between the spheres are taken
as umty. The cp component ot (ts.lUUJ 1s
-
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
*For a spherical geometry the notation TE (TM) indicates the absence of radial electric (magnetic)
field components.
376 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
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
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
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>
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,
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).
378 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
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!
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.
ll=
nf- = 1 (8.1 06)
1
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
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
(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
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\
2
2 dx [n (x)] (8.116)
(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
Sect. 8.10 Multimode Propagation in Optical Fibers 383
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)
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
~ ~ ------,
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
(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.
(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.
386 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
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)
(8.123)
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.
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)
E, WJLoZ X V,h'z]
y
and (8.126)
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
p<a
an
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.
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)
(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)
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
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 • ., •• ,
(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
(8.137) and (8.138) are determined. The completeness of the normal mode ex-
' o 'the> nn;t1 .++L ,,· +. oll
r "t' ~·
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
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'
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_
394 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
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
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
'"-.
<
See also
Galeis
ol.
_, rl' rl. :J'
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
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
398 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
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
(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 ~
k~
r E* · [V X (V X E)] d 3x
I ""* ~ •3....
. "'u
Jv ~
(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 -
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
,, _, ",.], ..
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
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
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
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?
404 Chapter 8 Waveguides, Resonant Cavities, and Optical Fibers-SI
r. '..1.
" -" _,'
" 1''7
'"
~-
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.
= 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
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)
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,
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
. 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:
..!_
n!
I J(x')(kn · x')" d 3 x' (9.10)
A(x) fLolk ;§;, h)Il(kr)Y 1m(IJ, ¢){ J(x')Jt(kr')Y'fm((l', </>') d 3 x' (9.11)
,ffl ,J
(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
*
,. . ~VUL'-'<0 <J VL •
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)
f
p J x' p(x')d· x' (9.17)
~~~ . ~~
---~
. ~· -• . "''~'"'. .... -·· A11
.. 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
ck2 eikr
u (n '>{ nl
47T r (9.19)
E = Z0H X n
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
~~ = ;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\
dP Z 0 I6 ( kd) 2 • 2
dfl = 1287T2 Sill (} (9.28)
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.
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
,.·
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
Then we find
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)
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.
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
~ ~~~-
X~ "
-2~~~~
"' ~
-~-
"~ c. ~
., "-
(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
, 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.
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:
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'
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)
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
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'
'
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)
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
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
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.
(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):
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.
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
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;
the dipole is viewed, as shown in Fig. 9.4. If there are fields on both sides of the
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:
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)
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
":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.
1 D2 1/J
'iPrf! ~ -2 - 2 = o (9.77)
L '"
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
~~ R rnhn Prn. 11? 1< ~0 1 <1 1,; /1 0" \· dll 1f){;Q /1 O'i?\
426 Chapter 9 Radiating Systems, Mnltipole Fields and Radiation-S!
This equation is just the Bessel equation (3.75) with v = l + 1. Thus the solutions
c. c .. \
W< J I""!_!_ U< v
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_}_
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
zL(.cr)
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)
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
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
-~·
..
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,.,.
~
-
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
L = ~ (r x V) (9.101)
l
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 ~ •
(9.105)
where
(9.106)
*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
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
~
.... ;th r\
_\_7oU.~
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
(9.119)
(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)
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!
(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)
(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
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
(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
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)
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~
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
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~~ ~ ~
\"7-.l"t/)
*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)
*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
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
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.
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)
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
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
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
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)
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
aM(l, m) = (
ik/+2
2! + l)!!
c 1r2
-
+
1
- (Mtm + M/m) (9.171)
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..
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
\
,... {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
444 Chapter 9 Radiating Systems, Multipole Fields and Radiation-SI
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
''" • • • 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
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
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)
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,
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
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<:\
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
'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
/(
..,-
.,
......
...._
// '\ ~
(
'
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.
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.
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 _, .
~
eikr
(h{v\ · n • nf1
47TE0r" •
. fLoW eikr
Afv\ ,,,. • • /() 1<0\l
47T r r L ' • ~n
,, .. ,, _,. -'
,..., ("',]
wc;y
"" ..
\">·!I), ctllU UlC lllllC
'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 ·
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
(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,
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
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
'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
(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.
JUUUllUITl!illdUUll.
(h) th<> tntol nnwPr radiated and find a numerical value for the radi-
· ,,
ation resistance.
454 Chapter 9 Radiating Systems, Multipole Fields and Radiation-SI
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=
"
(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
:o or /,
E = ::!:__Z H =-F
y dr /,oP
where m is the index specifying the angular dependence, f3 is the propagation con-
slant, y> = 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
(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
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
,,
. -~ . .
'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
' 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.
""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)
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
\
\
\
.-- ....
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
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.
3n (1{1 1'1\
l' '..-rrq)' lllC \ -;
0/~
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)
'
~ = k4 a 6 [~(1 + cos 2 ()) - cos f)] (10.16)
the same order of magnitude as for the dielectric sphere (10.11) if (e,- 1) is not
omoll
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-
-
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
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 \ -
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
•·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-'
--
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
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:
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
*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)
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
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
Sect. 10.2 Blue Sky, Scattering by Gases and Liquids, Attenuation in Optical Fibers 467
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
468 Chapter 10 Scattering and Diffraction-51
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.
Kondratyev, Radiation in the Atmosphere, Academic Press, New York (1969), Chapter 5.
Sect. 10.2 Blue Sky, Scattering by Gases and Liquids, Attenuation in Optical Fibers 469
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)
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
*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)
"'"''"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.).
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
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.
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
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
~
(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)
. .
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
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)
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
(10.56)
wave
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,
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:
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·
476 Chapter 10 Scattering and Diffraction-51
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
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
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
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)
I v V 12 IV 12
3 {1 -1- ~Ao2tl)
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
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),
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
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
'
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 ' ~
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
, '- £. .L ~
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. /
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)
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 "'"""-
where the obliquity factor 0( 8, 8') is the only point of difference. These factors
are
cos 8 1)
( 1/J approximated on S
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'
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
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
Expansion of the triple cross product and a rearrangement of terms yields the
final result,
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
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,
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
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)
•A :
<lllU ' lllUll 0:, 1 II" >0 101 V <011 U Y
(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
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-
486 Chapter 10 Scattering and Diffraction-Sf
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)
488 Chapter 10 Scattering and Diffraction-SI
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.
(10.103)
*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
-/
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:
~
>'U1 lllC
. . . . . . .
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
E - c8 c = E(o) (10.107)
8 + E)c = 8(o)
490 Chapter 10 Scattering and Diffraction-SI
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.*
The subject of diffraction has been extensively studied since Kirchhoff's original
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)
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)
(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,
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.
494 Chapter 10 Scattering and Diffraction-51
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).
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 0 E0 e'k,,·x
Ei -
B, = k 0 X E;fkc
(10.123)
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
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
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
(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 * •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,.
*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.
500 Chapter 10 Scattering and Diffraction-51
"~
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- -
the properties of the illuminated surface, we cannot say how it is divided between
scattering and absorption.
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)
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,
0
0
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)
'<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
(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
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
The scalar product of E;'; with the forward scattering amplitude is then
Sect. 10.11 Ontical Theorem and Related Matters 505
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>
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
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
~·-
'"~ '~"~"
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
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
.. 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 )
€ = ""
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) -
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,
(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
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
510 Chapter 10 Scattering and Diffraction-S!
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
4(z 2 z 4) rz 1 -
JO
cost dt
t
h-,·1 · 1
.ff .· ~ "J f d "· -~r~· (h d"ff d ·~u
,J. 1-.
.. "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:
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•~
(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
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
"''- 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 .
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> ' " ' -
''
•'- .1.
·'
ramauon, ana tne total power T'lflJiaT=.
512 Chapter 10 Scattering and Diffraction-Sf
10 16 (a) Show from (10 125) that the integral of the shadow scattering differential
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
{ 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
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" .\.
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'
514
Sect. 11.1 The Situation Before 1900, Einstein's Two Postulates 515
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
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
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
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
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
L(v) (11 6)
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
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 ~-
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.
•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
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
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
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
*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).
522 Chapter 11 Special Theory of Relativity-G
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.
'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
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~~
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
,,,
,
*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).
524 Chapter 11 Special Theory of Relativity-G
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.
-- 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
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).
526 Chapter 11 Special Theory of Relativity-G
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.
~. ~
. ~~·~·~
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 ,
*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.
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
528 Chapter 11 Special Theory of Relativity-G
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 ,
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
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
'DY · · . c<.[Uutiun> ]lLWJ, t11e reauer can veruy tnat tnere ex1sts a Lorentz transrormanon
with f3 < I provided sh > 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
(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.*
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!
/
--
u; 1
+v
u,
v · u'
1 + -2-
/" ""
\~~.~~,
u~
ut --
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'
· '-
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-'-'
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• '
• < ~
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
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
• .,.
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
c {32)
Uc
M(ua) = +
_ {3 2 M(O) (11.44)
1
2
~
536 Chapter 11 Special Theory of Relativity-G
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.
'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
(11.48)
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
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
2: a
(Po)a - L b
(Po)b = flo
· · "ol fi, o I
(11 ""'
2.. Pa ~ Pb- a \ ->~)
a b
initial final
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'.-
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.
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
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
'"'
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
or, explicitly, by
(11.63)
(11.65)
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·
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··
where the 3-vector A has comnonents A 1 A 2 A 3 ThP '""lor nr.,rhwt (11 flfll of
A
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
iJ"
a = ( axa
= ax" 0 • - v) (11.76)
(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
(11.78)
This is, of course, just the operator of the wave equation in vacuum.
(11.79)
544 Chanter 11 Suecial Theorv of Relativitv- G
_,
' ' a''-' ' ~
-1.
U, , ,L -~ '
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'/
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
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. !. .
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.
546 Chapter 11 Special Theory of Relativity-G
" 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
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
~ = 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
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<;\
548 Chapter 11 Special Theory of Relativity-G
=
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).
fL=--s (11.100)
2mc
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
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)
eE = - - - ( 11.105)
r dr
Then the spin-interaction energy can be written
~ g 1dV
0' s·B + (s · L) (11.106)
(t 0 0 1
(I 1.113)
p+O(J 7'
/op
/ ,.
Figure 1L7
Sect. 11.8 Thomas Precession 551
0 0 0 1
"
\ll.U_.,,
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
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
Osmg (11.117), (11.112), and (11.110) we can express x"' in terms of the labora-
(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)
(11.123)
Sect. 11.9 Invariance of Electric Charl!e' Covariance of Electrodvnamics 553
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.
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
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'
*See, for example, Section. 2.4c of A. Bohr and B. R. Mottclson, Nuclear Structure, Vol. 1, W. A.
Benjamin, New York (1969).
554 Chapter 11 Special Theory of Relativity-G
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 <I> are
1 aLA
2 ~ 2
~?·
v ,....
4'7T
. T
(11.130)
1 a"<P .o
V'l' '+'7TP
2 ;; '2
1 a<I>
-t-V•A u ~ll.l.:ll)
Obviously, Lorentz covanance reqmres that the potentials <P and A form a
4-vector potential,
A" = (<I>, 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<I>
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.)
556 Chapter 11 Special Theory of Relativity-G
1 riA ri<b
E - - - - - - - -la"A'- a'A"l
c dt ax (11.135)
aA aA.
B = _z- - ' = -(azA3- a3Az)
X 0, 0
v~
"J
where the second forms follow from (11.132) and a" = (a/ax 0 , - V). These equa-
t: n• ;~nh, H >t 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:
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
- = m - - - - F"f3U (11.144)
dT dT C I'
(11.146)
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
·"' •
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
B; = 0, B~ = 0, B"J = 0
In terms of the coordinates of K the nonvanishing field components are
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.
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
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 ... -
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.
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
. . .
562 Chapter 11 Special Theory of Relativity-G
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.
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
*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
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
" (
\·" .,
"'"'
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):
/a \
AC::" a 1
I£
0
dS 2 (
-=F+yp S ·dp)
-
dr dr
~nd
""::.'!
Jn
- K -+ -v 2 f
"' . '!
dr dT
:-.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,
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)
"
Lnapter lL.
*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<)()\.
566 Chapter 11 Special Theory of Relativity-G
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
a · b - aaba = a0 b 0 - a•b
P-p+q
or
pa a a
'1
'
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
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.
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
_,
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
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-
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,
(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
570 Chapter 11 Special Theory of Relativity-G
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~
"
(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
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
C1
C2
= [I", 8L] = -
= [I", C 1]
( r
tanh- 1{3
•'-
.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-
• . 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
(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~
or
A 0 = 0 = A'·, A 1 = -2q8(ct - z) V 1 ln(Ar1 )
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 - ~- ;;;; )
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-
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
_,..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-
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
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,
PLAB Sill 4
2
!:J.E=-
2m2
Ch.ll Problems 577
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)
<I>=-- ,
R3 R3
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, <I>' = 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
578 Chapter 11 Special Theory of Relativity-G
(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
-.,
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
TT 1~ J 1\, n2~ 0 ~~
" r \ JL ' ~ t
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
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
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
"- ..
:::!:: = 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 ,,
;-.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~
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
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
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.
me-vlu ua
(f (12.21)
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,
U dUry = 0 (12.24)
u dr
• 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)
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
--rr J,, ·y 6
ds ds c ds \-'<7
\ •I
dT C dT C dT
";nrP dA aldT - ( dx !dT\ a~ A a this eauation can be written as
(12.32)
UT C UT
*The geodesic is the longest path or longest proper lime tor limcl!ke separatmn at evems . .>cc
_, _, ..
cr'l-.
'"b' ' •• (1,., '2'2\
_'- ' 'I
••• (!')
\ '
1~\
'/'
..
~: = :~ = ! (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.
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
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
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)
~
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
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
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.
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 )
vc = (v,> = -
B0
-
ago
w0 X ((x0 )j_(n · x0 )> (12.52)
((xa)jn x0 )) H (12.53)
2
(12.54)
2 B0 ag o
(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.
590 Chapter 12 Dynamics of Relativistic Particles and Electromagnetic Fields-G
./
VB
Fignre 12.2 Drift of charged particles
due to transverse gradient of magnetic
field.
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)
(12.58)
(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
.. 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
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
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
(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)
1 = f -ymv.L • dl + ~ f A · dl (12.65)
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)
(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:
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
594 Chapter 12 Dynamics of Relativistic Particles and Electromagnetic Fields-G
/ 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:::-----
~
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
596 Chapter 12 Dynamics of Relativistic Particles and Electromagnetic Fields-G
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).
we are
particle in external electromagnetic fields described by the vector and scalar po-
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
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
598 Chapter 12 Dynamics of Relativistic Particles and Electromagnetic Fields-G
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.
'
£~( (:~)
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
A = JJ9:, d x dt = J9:, d x
3 4
(12.84)
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
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
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)
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
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)
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-
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.
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
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.
11..-
.
· vuttun Laws
"'
£.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
606 Chapter 12 Dynamics of Relativistic Particles and Electromagnetic Fields-G
~ 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
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
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'
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.)
Lfield
1
41TC
J x >< (E x B) tPx
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
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 \
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
3
ag - ~ -a
o = a 0>"; = -' r<MJ
a- ,1t . ,1r. l]
J
. f3
()a(")"
1
= -:;-- rJIL(FILJAf3) + ~ a13 (FILAF~'-A)
1
-- - (a~'-F
!LA
)F'f3 + F!LA u~'-Pf3 + -21 F!LA af3FILA
47T
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)
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
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
D(z) (12.128)
!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
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 ~
Then, since the theta functions select one or the other of the two terms, we have
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
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
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.
" 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.
618 Chapter 12 Dynamics of Relativistic Particles and Electromagnetic Fields-G
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 '
ct = R V1 + A 2 sinh <I>
p
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
"""" ""'u ~.
(a) Show that the difference f1 of the spin precession frequency and the orbital
gyration frequency is
eBa
!1=-
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
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
(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-
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
622 Chapter 12 Dynamics of Relativistic Particles and Electromagnetic Fields-G
. .
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
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
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,
':':"": = 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
.,]
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
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
cldent particles oy nuclei ana multiple scattenng are presented. Fmally, we treat
624
"· ... " . • "J
T · u . ., ro.~"' · /,;.,.::;:
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
~ ~ ~
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
' ~
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
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)
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 ''-~
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
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.
(13.18)
smaller than the ionization potential, indeed, smaller than the minimum possible
. . .
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
(13.19)
Sect. 13.3 Density Effect in CoUisional Energy Loss 631
(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.
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
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
'f'hP iJ;pl, c( ,,,\ ~nnpgro "h~ :~n llu :n ~Ao;t:Ano rl:~tnt ,;1 \.
• Tr 'J t" ')
.T(x t) - vn(x t\
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 ().).)
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
\'·'·"'')
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~)"£
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
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
~ (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
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
'"'•
'
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
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
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
'
'-
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,
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
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)
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
(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'"'"'
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
.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
. . .
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)
"" '
!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
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.
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
J 1:12 aa dfl
dfl
(1:12) = (13.60)
Jdfl
dad
fl
ur;.~ ol-. .+ (' 1 ,., "- ,]..
~1'1'
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~·
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
/ '- I /
X
'--¥
Incident
direction
n Ncrt (13.64)
(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 ,
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 ()'
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
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
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"
646 Chapter 13 Energy Loss, Scattering; Cherenkov and Transition Radiation-G
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).
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
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 .
*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
648 Chapter 13 Energy Loss, Scattering; Cherenkov and Transition Radiation-G
.'
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
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..,
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'
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
:}", ,.
,. .,\ A3 v ;n tho uAlnmo
• u "<
. -/
1
m e ,. • r1
R , ... , ..~·)
I
Ubrad A ... u A
650 Chapter 13 Energy Loss, Scattering; Cherenkov and Transition Radiation-G
Emd = -
e
r c
E(w) - 1 kz
47T
J. (kz'.>O
X E;) X ke-ik·x' d 3 x'
e'h 1 -w~ l (
v flr v 1<'.\
..,
y lr -ik·x' ,.J3v'
0 {1'1 '70\
r 47TC
2 Jz'>O ,. \ • Jj
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
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
~-
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
F = Ea
2
- {( dx dye ikxsin fl Ko ( -w \lx2 + l
\
v w ' I
v --kcosel
1!
(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.
652 Chapter 13 Energy Loss, Scattering; Cherenkov and Transition Radiation-G
In the approximation of relativistic motion ( y >> 1), small angles (IJ << 1), and
high frequencies (w >> wp), this becomes
F (13.83)
(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--
""\\ '\
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!
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•
654 Chapter 13 Energy Loss, Scattering; Cherenkov and Transition Radiation-G
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<
..
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
656 Chapter 13 Energy Loss, Scattering; Cherenkov and Transition Radiation-G
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,
ze
<P(w, x) = - (-)
VE W
h'iM 11 '( gnrJ
.. ion _(
I·'"'·
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
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.
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
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
I 1T C
(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
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
where D,(x x') is the retarded Green function (12 133) and
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
"
= -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
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
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,
~
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)
~
/ \
/
/~
// /
/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
eyb
E2 = (b2 + lv2t2)312 (14.17a)
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.
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<
..
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)
*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.
666 Chapter 14 Radiation by Moving Charges-G
/v
I
v
/
/
/
/ Figure 14.3
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)
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-
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.>
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
668 Chapter 14 Radiation by Moving Charges-G
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.
~ {_.!_ ~]
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')
(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-
--
--- J--
for "' 21 havinf! been reduced bv a
factor ~ 102 for the same acceleration.
670 Chapter 14 Radiation by Moving Charges-G
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~
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)
'-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.
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,
._,
!''"'"'"
. ~·
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
,... _,_. ,_ .
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 •
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
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)
A(w) = - 1 f~ A(t)eiwt dt
(14.54)
v27T -w
A(t) = ~
1 1- A(w)e-iwt dw (14.55)
v L7r J X
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
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)
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
A(w) =
Since the
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
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
(14.70)
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 ..
n x (n x 13) - 1{ € 1
sin(;) + e~ cos(;) sine] (14 71)
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
678 Chapter 14 Radiation by Moving Charges-G
.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:
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
*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 •'-
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 ·~
(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
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
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
?;
"'
0.4
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)
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)
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
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-
. . . .
684 Chapter 14 Radiation by Moving Charges-G
~ ~ ~ ~ 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)
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 \
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.
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)
(14.99)
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' =
686 Chapter 14 Radiation by Moving Charges-G
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.
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
Sect. 14.7 Undulators and Wigglers for Synchrotron Light Sources 687
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
or
(} -k
y 0 ct '
688 Chapter 14 Radiation by Moving Charges-G
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-
~
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£ •
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 "'~
~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.
690 Chapter 14 Radiation by Moving Charges-G
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€
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.
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
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-
Sect. 14.7 Undulators and Wigglers for Synchrotron Light Sources 691
" 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_ • -'
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.
277 2
Ny = "i aK (14.119)
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
,- . '
-- ..... _ ,
.....
.....
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
694 Chapter 14 Radiation by Moving Charges-G
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)
~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)
,.~
-
,~
~~\·
,,,
i\\~
ol
!G-
~ 0.5 I-
I\\
I \ \
II \ \\
\\ \\ \ ~ _,/
/
-- --
>ei
b
"'
..
.!I.
\ • \ \ .....<:
~- ' ...
\\ ' ":.:....
0.25 ..... ~:·
..... ··~ ...........
-
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.
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
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
(}"
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
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
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 ,,;
~-.
••
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
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
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~• ~ • -u
2
'HU()I ~'-'t
2•
" ' " ' " auu u
.. .,. .,.
or cwsesr approacn in a nearr=urr
700 Chapter 14 Radiation by Moving Charges-G
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
(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
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
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.
(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:
2 . •'-
14.15 A o: ._,_ nf P
~.- .. , -c -, ~a·
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 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
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 •
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
(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
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
Pk =
4e2
,,.
13 (kw 0 )
4 2
a
1
1;2 (h• ( kE)) 2 + e- 2 E
2
h kE)
). 2(
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.] ;
(a) ' ..
Show that the nower radiated into the rnth nf "'· is
dPm(N) _ dPm(l) F ( )
dO. - _dQ mN
(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-
· <I 2N • • · · • · •
• fJ , >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
706 Chapter 14 Radiation by Moving Charges-G
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(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
(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.
. .
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.
. 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
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
~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
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)
..
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
b' ~ ~;
>L
uw
,1.
·' ·'·
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, ~• ' ,. • •
*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.
.( t\
'\'/
.rm -'- (' -llft'\
,~,
Jo •• ,. '
~.,
then :1n:wt from :1 const:1nt the nh:1se of the in· "in (1'i 1) is
rt \
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
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
-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.
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;_,
dus zZeL· 1
dQ = 8'lT (3c-
-
Q3 (15.19)
Sect. 15.2 Bremsstrahlung in Coulomb Collisions 715
(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)
716 Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
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;
'
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 .£ .
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
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·
718 Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
(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)
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 )
(15.36)
where 8 is the angle of emission of the photon and dxR!dw is given by (15.34).
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).
720 Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
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~
1 d3x 1 d3x'
- = -w' 2 dw' d[!' dQ (15.38)
w2 dw d[! dQ
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.
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)
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
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 -""
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
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
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.
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
'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 ~
724 Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
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
MlUWt:I.
"'""'-"
"'J'"'""'
- .. -
' I lU'-""
•
Umin
ill' "
..
LdiiiUll!\ Ill n UI,;H;U' LUll ,:)1-<tllCOlHl!\ Ul V 11 LUdl
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
"'
~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
726 Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
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 -,
·'- "
.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
728 Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
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
(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
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)
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-
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' ' / /
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- .
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
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)
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
t~ Figure 15.9
/j~ Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
("""" 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.
Z + e- ~ (Z- 11 + v (15.701
~· nll
"J
~
-'-' ... "Y
... ' -·o """'5J
'£
11
'""'"
I \ I
r-" I \ Ze ---....._I
"-= / _/ y
V -e
/
X Figure 15.10
(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)
*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
734 Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
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
,_,_,
""""'
wmcngives
a·J w 2 . z~
'• r
~~J.U~)
n•'
dw dfl 4 7Tz c3 f" '"' "
"'-'-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 ' . ~
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)
736 Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
I
.'
~ow-frequency
I- -
I limit
I
I- I -
t / / ~
i/
·'·
lVfnW}
-
I
I / "' -
I
1- I. -
I I
nw l!.Q
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 •
-· .
A semiclassical discussion analogous to ours, but using impact parameters throughout and
. '
"""KUSSr,
':n"
. " .
. L.lL
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
Panofsky and Phillips, Section 18.5
~ .. _,_ .<.. " . ,.
••. ' £. .£ 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
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
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
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 <P
leading to (15.10) after averaging over ,P.
740 Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
~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
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
(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.
742 Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
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
(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.
744 Chapter 15 Bremsstrahlung, Method of Virtual Quanta, Radiative Beta Processes-G
(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
745
746 Chapter 16 Radiation Damping, Classical Models of Charged Particles-G
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
(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.
,., .. 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:
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
J,l Frad- J 3c V . v dt - ()
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)
(16.11)
dE
dt
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·
~
..,.....
¥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
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
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
(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.
dt
I - J p(x, t) V<I>(x, t) +
c at
(x, t) d 3x (16.22)
em
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
•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
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
{· · ·) = 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 "·"'-·' }·
"'::':
( dp ') =
dt em
.i: (-1)" _3__
n~o 3n!
cn+l
a" 'tv
atn+ 1
Jd3x' Jd3x p(x')Rn-tp(x) (16.28)
,., t· \n r r ~
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
2 r I r11. \ 12
3
m mo +
31T
2 2
c
J d k k2 (16.33)
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
(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
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)
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
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.
netic stress tensor (12.113). In contrast to source-free fields, the stress tensor ®"13
(16.36)
*E. Fermi, Z. Phy. 24, 340 (1922), or Atti. Accad. Nazi. Lincei Rend. 31, 184, 306 1922).
756 Chapter 16 Radiation Damping, Classical Models of Charged Particles-G
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)
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-
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
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(;
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
(16.46)
(1 f. A'7\
c.o e Yl-' Q
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.
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
(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
( 16.54)
(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)
· ol
• •vm •a~ "'[' ' ·· -; r -o
ment that t(z) vanish at infinity, an integration by parts leads to
(16.59)
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'
e I z "U') az ~lO.OU)
JU
762 Chapter 16 Radiation Damping, Classical Models of Charged Particles-G
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.
®
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)
!16.64)
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
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
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."'
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).
764 Chapter 16 Radiation Damping, Classical Models of Charged Particles-G
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
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..
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
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
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
e 1
.1'-md 2 ]D X lD X XJrct
c r
or
(16.75)
\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.
768 Chapter 16 Radiation Damping, Classical Models of Charged Particles-G
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
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
t
r ~f) ro ~L~CT)"
T
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
(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.
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
(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
(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
r o
0"1 (w) dw = 2rrze
-
me
-
2
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 .
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
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
(A.5)
(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)
2
k_k a
Combining (A 5) with (A JO), we find
a
an eqnality holding for both magnitude and dimensions
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
(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)
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
).
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.
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
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
~~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.
~}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
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•
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
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(';~)
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
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
' <u
nL
".J"'""' "
~u~
~ ~' . , uvau-
. , 1->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
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·
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
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
'
'~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
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
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.,
T"
'6J
.' , n ·•'
llJUla lUl, OL I
,
..,. . _, <. 1:" -~"
_..,_ cry
,u Hl "'
,JTT;:Y7U
.7CJ• L
. "Y'"
" -
'JO
~, , LA , L• , L">
L ' f C<dUUH, Ul WHH l• , no , //1-L /'HI-~,
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
. .
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
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.• , .,~,
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
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
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
ORTHOGONALFUNCT~EXP~N~IIJNS
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
1 a ( r 2 -alji) = -
[ Note that--
r 2 ar ar
1 a
- (rlji).
r ar 2
2
J