Download Principles of optics

Principles of optics
Electromagnetic theory of propagation,
interference and diffraction of light
MAX BORN
M A , Dr Phil, F R S
Nobel Laureate
Formerly Professor at the Universities of GoÈttingen and Edinburgh
and
E M I L WO L F
P h D, D S c
Wilson Professor of Optical Physics, University of Rochester, NY
with contributions by
A . B. B H AT I A , P. C . C L E M M OW, D. G A B O R , A . R . S TO K E S ,
A . M . TAY L O R , P. A . WAY M A N A N D W. L . W I L C O C K
S E V E N T H ( E X PA N D E D ) E D I T I O N
published by the press syndicate of the university of cambridge
The Pitt Building, Trumpington Street, Cambridge, United Kingdom
cambridge university press
The Edinburgh Building, Cambridge CB2 2RU, UK
40 West 20th Street, New York, NY 10011-4211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
Ruiz de AlarcoÂn 13, 28014 Madrid, Spain
Dock House, The Waterfront, Cape Town 8001, South Africa
http://www.cambridge.org
Seventh (expanded) edition # Margaret Farley-Born and Emil Wolf 1999
This book is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without
the written permission of Cambridge University Press.
First published 1959 by Pergamon Press Ltd, London
Sixth edition 1980
Reprinted (with corrections) 1983, 1984, 1986, 1987, 1989, 1991, 1993
Reissued by Cambridge University Press 1997
Seventh (expanded) edition 1999
Reprinted with corrections, 2002
Reprinted 2003
Printed in the United Kingdom at the University Press, Cambridge
A catalogue record for this book is available from the British Library
Library of Congress Cataloguing in Publication data
Born, Max
Principles of Optics±7th ed.
1. Optics
I. Title
II. Wolf Emil
535
QC351
80-41470
ISBN 0 521 642221 hardback
KT
Contents
Historical introduction
I
Basic properties of the electromagnetic ®eld
1.1 The electromagnetic ®eld
1.1.1 Maxwell's equations
1.1.2 Material equations
1.1.3 Boundary conditions at a surface of discontinuity
1.1.4 The energy law of the electromagnetic ®eld
1.2 The wave equation and the velocity of light
1.3 Scalar waves
1.3.1 Plane waves
1.3.2 Spherical waves
1.3.3 Harmonic waves. The phase velocity
1.3.4 Wave packets. The group velocity
1.4 Vector waves
1.4.1 The general electromagnetic plane wave
1.4.2 The harmonic electromagnetic plane wave
(a) Elliptic polarization
(b) Linear and circular polarization
(c) Characterization of the state of polarization by Stokes parameters
1.4.3 Harmonic vector waves of arbitrary form
1.5 Re¯ection and refraction of a plane wave
1.5.1 The laws of re¯ection and refraction
1.5.2 Fresnel formulae
1.5.3 The re¯ectivity and transmissivity; polarization on re¯ection and refraction
1.5.4 Total re¯ection
1.6 Wave propagation in a strati®ed medium. Theory of dielectric ®lms
1.6.1 The basic differential equations
1.6.2 The characteristic matrix of a strati®ed medium
(a) A homogeneous dielectric ®lm
(b) A strati®ed medium as a pile of thin homogeneous ®lms
1.6.3 The re¯ection and transmission coef®cients
1.6.4 A homogeneous dielectric ®lm
1.6.5 Periodically strati®ed media
II Electromagnetic potentials and polarization
2.1 The electrodynamic potentials in the vacuum
xvi
xxv
1
1
1
2
4
7
11
14
15
16
16
19
24
24
25
25
29
31
33
38
38
40
43
49
54
55
58
61
62
63
64
70
75
76
Contents
2.1.1 The vector and scalar potentials
2.1.2 Retarded potentials
2.2 Polarization and magnetization
2.2.1 The potentials in terms of polarization and magnetization
2.2.2 Hertz vectors
2.2.3 The ®eld of a linear electric dipole
2.3 The Lorentz±Lorenz formula and elementary dispersion theory
2.3.1 The dielectric and magnetic susceptibilities
2.3.2 The effective ®eld
2.3.3 The mean polarizability: the Lorentz±Lorenz formula
2.3.4 Elementary theory of dispersion
2.4 Propagation of electromagnetic waves treated by integral equations
2.4.1 The basic integral equation
2.4.2 The Ewald±Oseen extinction theorem and a rigorous derivation of the
Lorentz±Lorenz formula
2.4.3 Refraction and re¯ection of a plane wave, treated with the help of the
Ewald±Oseen extinction theorem
xvii
76
78
80
80
84
85
89
89
90
92
95
103
104
105
110
III Foundations of geometrical optics
3.1 Approximation for very short wavelengths
3.1.1 Derivation of the eikonal equation
3.1.2 The light rays and the intensity law of geometrical optics
3.1.3 Propagation of the amplitude vectors
3.1.4 Generalizations and the limits of validity of geometrical optics
3.2 General properties of rays
3.2.1 The differential equation of light rays
3.2.2 The laws of refraction and re¯ection
3.2.3 Ray congruences and their focal properties
3.3 Other basic theorems of geometrical optics
3.3.1 Lagrange's integral invariant
3.3.2 The principle of Fermat
3.3.3 The theorem of Malus and Dupin and some related theorems
116
116
117
120
125
127
129
129
132
134
135
135
136
139
IV Geometrical theory of optical imaging
4.1 The characteristic functions of Hamilton
4.1.1 The point characteristic
4.1.2 The mixed characteristic
4.1.3 The angle characteristic
4.1.4 Approximate form of the angle characteristic of a refracting surface of
revolution
4.1.5 Approximate form of the angle characteristic of a re¯ecting surface of
revolution
4.2 Perfect imaging
4.2.1 General theorems
4.2.2 Maxwell's `®sh-eye'
4.2.3 Stigmatic imaging of surfaces
4.3 Projective transformation (collineation) with axial symmetry
4.3.1 General formulae
4.3.2 The telescopic case
4.3.3 Classi®cation of projective transformations
4.3.4 Combination of projective transformations
4.4 Gaussian optics
4.4.1 Refracting surface of revolution
142
142
142
144
146
147
151
152
153
157
159
160
161
164
165
166
167
167
xviii
Contents
4.4.2 Re¯ecting surface of revolution
4.4.3 The thick lens
4.4.4 The thin lens
4.4.5 The general centred system
4.5 Stigmatic imaging with wide-angle pencils
4.5.1 The sine condition
4.5.2 The Herschel condition
4.6 Astigmatic pencils of rays
4.6.1 Focal properties of a thin pencil
4.6.2 Refraction of a thin pencil
4.7 Chromatic aberration. Dispersion by a prism
4.7.1 Chromatic aberration
4.7.2 Dispersion by a prism
4.8 Radiometry and apertures
4.8.1 Basic concepts of radiometry
4.8.2 Stops and pupils
4.8.3 Brightness and illumination of images
4.9 Ray tracing
4.9.1 Oblique meridional rays
4.9.2 Paraxial rays
4.9.3 Skew rays
4.10 Design of aspheric surfaces
4.10.1 Attainment of axial stigmatism
4.10.2 Attainment of aplanatism
4.11 Image-reconstruction from projections (computerized tomography)
4.11.1 Introduction
4.11.2 Beam propagation in an absorbing medium
4.11.3 Ray integrals and projections
4.11.4 The N-dimensional Radon transform
4.11.5 Reconstruction of cross-sections and the projection-slice theorem of
computerized tomography
170
171
174
175
178
179
180
181
181
182
186
186
190
193
194
199
201
204
204
207
208
211
211
214
217
217
218
219
221
223
V Geometrical theory of aberrations
5.1 Wave and ray aberrations; the aberration function
5.2 The perturbation eikonal of Schwarzschild
5.3 The primary (Seidel) aberrations
(a) Spherical aberration (B 6ˆ 0)
(b) Coma (F 6ˆ 0)
(c) Astigmatism (C 6ˆ 0) and curvature of ®eld (( 6ˆ 0)
(d) Distortion (E 6ˆ 0)
5.4 Addition theorem for the primary aberrations
5.5 The primary aberration coef®cients of a general centred lens system
5.5.1 The Seidel formulae in terms of two paraxial rays
5.5.2 The Seidel formulae in terms of one paraxial ray
5.5.3 Petzval's theorem
5.6 Example: The primary aberrations of a thin lens
5.7 The chromatic aberration of a general centred lens system
228
229
233
236
238
238
240
243
244
246
246
251
253
254
257
VI
261
261
263
267
274
Image-forming instruments
6.1 The eye
6.2 The camera
6.3 The refracting telescope
6.4 The re¯ecting telescope
Contents
6.5 Instruments of illumination
6.6 The microscope
VII Elements of the theory of interference and interferometers
7.1 Introduction
7.2 Interference of two monochromatic waves
7.3 Two-beam interference: division of wave-front
7.3.1 Young's experiment
7.3.2 Fresnel's mirrors and similar arrangements
7.3.3 Fringes with quasi-monochromatic and white light
7.3.4 Use of slit sources; visibility of fringes
7.3.5 Application to the measurement of optical path difference: the Rayleigh
interferometer
7.3.6 Application to the measurement of angular dimensions of sources:
the Michelson stellar interferometer
7.4 Standing waves
7.5 Two-beam interference: division of amplitude
7.5.1 Fringes with a plane-parallel plate
7.5.2 Fringes with thin ®lms; the Fizeau interferometer
7.5.3 Localization of fringes
7.5.4 The Michelson interferometer
7.5.5 The Twyman±Green and related interferometers
7.5.6 Fringes with two identical plates: the Jamin interferometer and
interference microscopes
7.5.7 The Mach±Zehnder interferometer; the Bates wave-front shearing interferometer
7.5.8 The coherence length; the application of two-beam interference to the
study of the ®ne structure of spectral lines
7.6 Multiple-beam interference
7.6.1 Multiple-beam fringes with a plane-parallel plate
7.6.2 The Fabry±Perot interferometer
7.6.3 The application of the Fabry±Perot interferometer to the study of the
®ne structure of spectral lines
7.6.4 The application of the Fabry±Perot interferometer to the comparison of
wavelengths
7.6.5 The Lummer±Gehrcke interferometer
7.6.6 Interference ®lters
7.6.7 Multiple-beam fringes with thin ®lms
7.6.8 Multiple-beam fringes with two plane-parallel plates
(a) Fringes with monochromatic and quasi-monochromatic light
(b) Fringes of superposition
7.7 The comparison of wavelengths with the standard metre
VIII Elements of the theory of diffraction
8.1 Introduction
8.2 The Huygens±Fresnel principle
8.3 Kirchhoff's diffraction theory
8.3.1 The integral theorem of Kirchhoff
8.3.2 Kirchhoff's diffraction theory
8.3.3 Fraunhofer and Fresnel diffraction
8.4 Transition to a scalar theory
8.4.1 The image ®eld due to a monochromatic oscillator
8.4.2 The total image ®eld
xix
279
281
286
286
287
290
290
292
295
296
299
302
308
313
313
318
325
334
336
341
348
352
359
360
366
370
377
380
386
391
401
401
405
409
412
412
413
417
417
421
425
430
431
434
xx
Contents
8.5
Fraunhofer diffraction at apertures of various forms
8.5.1 The rectangular aperture and the slit
8.5.2 The circular aperture
8.5.3 Other forms of aperture
8.6 Fraunhofer diffraction in optical instruments
8.6.1 Diffraction gratings
(a) The principle of the diffraction grating
(b) Types of grating
(c) Grating spectrographs
8.6.2 Resolving power of image-forming systems
8.6.3 Image formation in the microscope
(a) Incoherent illumination
(b) Coherent illumination ± Abbe's theory
(c) Coherent illumination ± Zernike's phase contrast method of
observation
8.7 Fresnel diffraction at a straight edge
8.7.1 The diffraction integral
8.7.2 Fresnel's integrals
8.7.3 Fresnel diffraction at a straight edge
8.8 The three-dimensional light distribution near focus
8.8.1 Evaluation of the diffraction integral in terms of Lommel functions
8.8.2 The distribution of intensity
(a) Intensity in the geometrical focal plane
(b) Intensity along the axis
(c) Intensity along the boundary of the geometrical shadow
8.8.3 The integrated intensity
8.8.4 The phase behaviour
8.9 The boundary diffraction wave
8.10 Gabor's method of imaging by reconstructed wave-fronts (holography)
8.10.1 Producing the positive hologram
8.10.2 The reconstruction
8.11 The Rayleigh±Sommerfeld diffraction integrals
8.11.1 The Rayleigh diffraction integrals
8.11.2 The Rayleigh±Sommerfeld diffraction integrals
IX
The diffraction theory of aberrations
9.1 The diffraction integral in the presence of aberrations
9.1.1 The diffraction integral
9.1.2. The displacement theorem. Change of reference sphere
9.1.3. A relation between the intensity and the average deformation of
wave-fronts
9.2 Expansion of the aberration function
9.2.1 The circle polynomials of Zernike
9.2.2 Expansion of the aberration function
9.3 Tolerance conditions for primary aberrations
9.4 The diffraction pattern associated with a single aberration
9.4.1 Primary spherical aberration
9.4.2 Primary coma
9.4.3 Primary astigmatism
9.5 Imaging of extended objects
9.5.1 Coherent illumination
9.5.2 Incoherent illumination
436
436
439
443
446
446
446
453
458
461
465
465
467
472
476
476
478
481
484
484
489
490
491
491
492
494
499
504
504
506
512
512
514
517
518
518
520
522
523
523
525
527
532
536
538
539
543
543
547
Contents
X Interference and diffraction with partially coherent light
10.1 Introduction
10.2 A complex representation of real polychromatic ®elds
10.3 The correlation functions of light beams
10.3.1 Interference of two partially coherent beams. The mutual coherence
function and the complex degree of coherence
10.3.2 Spectral representation of mutual coherence
10.4 Interference and diffraction with quasi-monochromatic light
10.4.1 Interference with quasi-monochromatic light. The mutual intensity
10.4.2 Calculation of mutual intensity and degree of coherence for light from
an extended incoherent quasi-monochromatic source
(a) The van Cittert±Zernike theorem
(b) Hopkins' formula
10.4.3 An example
10.4.4 Propagation of mutual intensity
10.5 Interference with broad-band light and the spectral degree of coherence.
Correlation-induced spectral changes
10.6 Some applications
10.6.1 The degree of coherence in the image of an extended incoherent
quasi-monochromatic source
10.6.2 The in¯uence of the condenser on resolution in a microscope
(a) Critical illumination
(b) KoÈhler's illumination
10.6.3 Imaging with partially coherent quasi-monochromatic illumination
(a) Transmission of mutual intensity through an optical system
(b) Images of transilluminated objects
10.7 Some theorems relating to mutual coherence
10.7.1 Calculation of mutual coherence for light from an incoherent source
10.7.2 Propagation of mutual coherence
10.8 Rigorous theory of partial coherence
10.8.1 Wave equations for mutual coherence
10.8.2 Rigorous formulation of the propagation law for mutual coherence
10.8.3 The coherence time and the effective spectral width
10.9 Polarization properties of quasi-monochromatic light
10.9.1 The coherency matrix of a quasi-monochromatic plane wave
(a) Completely unpolarized light (natural light)
(b) Complete polarized light
10.9.2 Some equivalent representations. The degree of polarization of a light
wave
10.9.3 The Stokes parameters of a quasi-monochromatic plane wave
XI
Rigorous diffraction theory
11.1 Introduction
11.2 Boundary conditions and surface currents
11.3 Diffraction by a plane screen: electromagnetic form of Babinet's principle
11.4 Two-dimensional diffraction by a plane screen
11.4.1 The scalar nature of two-dimensional electromagnetic ®elds
11.4.2 An angular spectrum of plane waves
11.4.3 Formulation in terms of dual integral equations
11.5 Two-dimensional diffraction of a plane wave by a half-plane
11.5.1 Solution of the dual integral equations for E-polarization
11.5.2 Expression of the solution in terms of Fresnel integrals
11.5.3 The nature of the solution
xxi
554
554
557
562
562
566
569
569
572
572
577
578
580
585
590
590
595
595
598
599
599
602
606
606
609
610
610
612
615
619
619
624
624
626
630
633
633
635
636
638
638
639
642
643
643
645
648
xxii
Contents
11.6
11.7
11.8
11.9
11.5.4 The solution for )-polarization
11.5.5 Some numerical calculations
11.5.6 Comparison with approximate theory and with experimental results
Three-dimensional diffraction of a plane wave by a half-plane
Diffraction of a ®eld due to a localized source by a half-plane
11.7.1 A line-current parallel to the diffracting edge
11.7.2 A dipole
Other problems
11.8.1 Two parallel half-planes
11.8.2 An in®nite stack of parallel, staggered half-planes
11.8.3 A strip
11.8.4 Further problems
Uniqueness of solution
XII Diffraction of light by ultrasonic waves
12.1 Qualitative description of the phenomenon and summary of theories based on
Maxwell's differential equations
12.1.1 Qualitative description of the phenomenon
12.1.2 Summary of theories based on Maxwell's equations
12.2 Diffraction of light by ultrasonic waves as treated by the integral equation
method
12.2.1 Integral equation for E-polarization
12.2.2 The trial solution of the integral equation
12.2.3 Expressions for the amplitudes of the light waves in the diffracted and
re¯ected spectra
12.2.4 Solution of the equations by a method of successive approximations
12.2.5 Expressions for the intensities of the ®rst and second order lines for
some special cases
12.2.6 Some qualitative results
12.2.7 The Raman±Nath approximation
XIII Scattering from inhomogeneous media
13.1 Elements of the scalar theory of scattering
13.1.1 Derivation of the basic integral equation
13.1.2 The ®rst-order Born approximation
13.1.3 Scattering from periodic potentials
13.1.4 Multiple scattering
13.2 Principles of diffraction tomography for reconstruction of the scattering
potential
13.2.1 Angular spectrum representation of the scattered ®eld
13.2.2 The basic theorem of diffraction tomography
13.3 The optical cross-section theorem
13.4 A reciprocity relation
13.5 The Rytov series
13.6 Scattering of electromagnetic waves
13.6.1 The integro-differential equations of electromagnetic scattering
theory
13.6.2 The far ®eld
13.6.3 The optical cross-section theorem for scattering of electromagnetic
waves
XIV Optics of metals
14.1 Wave propagation in a conductor
652
653
656
657
659
659
664
667
667
669
670
671
672
674
674
674
677
680
682
682
686
686
689
691
693
695
695
695
699
703
708
710
711
713
716
724
726
729
729
730
732
735
735
Contents
14.2 Refraction and re¯ection at a metal surface
14.3 Elementary electron theory of the optical constants of metals
14.4 Wave propagation in a strati®ed conducting medium. Theory of metallic
®lms
14.4.1 An absorbing ®lm on a transparent substrate
14.4.2 A transparent ®lm on an absorbing substrate
14.5 Diffraction by a conducting sphere; theory of Mie
14.5.1 Mathematical solution of the problem
(a) Representation of the ®eld in terms of Debye's potentials
(b) Series expansions for the ®eld components
(c) Summary of formulae relating to the associated Legendre functions and to the cylindrical functions
14.5.2 Some consequences of Mie's formulae
(a) The partial waves
(b) Limiting cases
(c) Intensity and polarization of the scattered light
14.5.3 Total scattering and extinction
(a) Some general considerations
(b) Computational results
XV Optics of crystals
15.1 The dielectric tensor of an anisotropic medium
15.2 The structure of a monochromatic plane wave in an anisotropic medium
15.2.1 The phase velocity and the ray velocity
15.2.2 Fresnel's formulae for the propagation of light in crystals
15.2.3 Geometrical constructions for determining the velocities of
propagation and the directions of vibration
(a) The ellipsoid of wave normals
(b) The ray ellipsoid
(c) The normal surface and the ray surface
15.3 Optical properties of uniaxial and biaxial crystals
15.3.1 The optical classi®cation of crystals
15.3.2 Light propagation in uniaxial crystals
15.3.3 Light propagation in biaxial crystals
15.3.4 Refraction in crystals
(a) Double refraction
(b) Conical refraction
15.4 Measurements in crystal optics
15.4.1 The Nicol prism
15.4.2 Compensators
(a) The quarter-wave plate
(b) Babinet's compensator
(c) Soleil's compensator
(d) Berek's compensator
15.4.3 Interference with crystal plates
15.4.4 Interference ®gures from uniaxial crystal plates
15.4.5 Interference ®gures from biaxial crystal plates
15.4.6 Location of optic axes and determination of the principal refractive
indices of a crystalline medium
15.5 Stress birefringence and form birefringence
15.5.1 Stress birefringence
15.5.2 Form birefringence
xxiii
739
749
752
752
758
759
760
760
765
772
774
774
775
780
784
784
785
790
790
792
792
795
799
799
802
803
805
805
806
808
811
811
813
818
818
820
820
821
823
823
823
829
831
833
834
834
837
xxiv
Contents
15.6 Absorbing crystals
15.6.1 Light propagation in an absorbing anisotropic medium
15.6.2 Interference ®gures from absorbing crystal plates
(a) Uniaxial crystals
(b) Biaxial crystals
15.6.3 Dichroic polarizers
840
840
846
847
848
849
Appendices
853
I
The Calculus of variations
1 Euler's equations as necessary conditions for an extremum
2 Hilbert's independence integral and the Hamilton±Jacobi equation
3 The ®eld of extremals
4 Determination of all extremals from the solution of the Hamilton±Jacobi equation
5 Hamilton's canonical equations
6 The special case when the independent variable does not appear explicitly in the integrand
7 Discontinuities
8 Weierstrass' and Legendre's conditions (suf®ciency conditions for an extremum)
9 Minimum of the variational integral when one end point is constrained to a surface
10 Jacobi's criterion for a minimum
11 Example I: Optics
12 Example II: Mechanics of material points
853
853
855
856
858
860
861
862
864
866
867
868
870
II
Light optics, electron optics and wave mechanics
1 The Hamiltonian analogy in elementary form
2 The Hamiltonian analogy in variational form
3 Wave mechanics of free electrons
4 The application of optical principles to electron optics
873
873
876
879
881
III
Asymptotic approximations to integrals
1 The method of steepest descent
2 The method of stationary phase
3 Double integrals
883
883
888
890
IV
The Dirac delta function
892
V
A mathematical lemma used in the rigorous derivation of the LorentzULorenz formula
(}2¯W¯2)
898
VI
Propagation of discontinuities in an electromagnetic ®eld (}3¯1¯1)
1 Relations connecting discontinuous changes in ®eld vectors
2 The ®eld on a moving discontinuity surface
901
901
903
VII
The circle polynomials of Zernike (}9¯2¯1)
1 Some general considerations
2 Explicit expressions for the radial polynomials Rm m (r)
905
905
907
VIII
Proof of the inequality jì12 (í)j < 1 for the spectral degree of coherence (}10¯5)
911
IX
Proof of a reciprocity inequality (}10¯8¯3)
912
X
Evaluation of two integrals (}12¯2¯2)
914
XI
Energy conservation in scalar wave®elds (}13¯3)
918
XII
Proof of Jones' lemma (}13¯3)
921
Author index
Subject index
925
936
I
Basic properties of the electromagnetic ®eld
1.1 The electromagnetic ®eld
1.1.1 Maxwell's equations
The state of excitation which is established in space by the presence of electric charges
is said to constitute an electromagnetic ®eld. It is represented by two vectors, E and B,
called the electric vector and the magnetic induction respectively.
To describe the effect of the ®eld on material objects, it is necessary to introduce a
second set of vectors, viz. the electric current density j, the electric displacement D,
and the magnetic vector H.
The space and time derivatives of the ®ve vectors are related by Maxwell's
equations, which hold at every point in whose neighbourhood the physical properties
of the medium are continuous:{
1_
4ð
Dˆ
j,
c
c
1
curl E ‡ B_ ˆ -,
c
curl H
(1z
()z
the dot denoting differentiation with respect to time.
In elementary considerations E and H are, for historical reasons, usually regarded as the basic ®eld vectors,
and D and B as describing the inBuence of matter. In general theory, however, the present interpretation is
compulsory for reasons connected with the electrodynamics of moving media.
The four Raxwell e±uations (1zE(4z can be divided into two sets of e±uations, one consisting of two
homogeneous e±uations (rightWhand side zeroz, containing E and B, the other of two nonhomogeneous
e±uations (rightWhand side different from zeroz, containing D and H. If a coordinate transformation of
space and time (relativistic 1orentz transformationz is carried out, the e±uations of each group transform
together, the e±uations remaining unaltered in form if j=c and r are transformed as a fourWvector, and each
of the pairs E, B and D, H as a sixWvector (antisymmetric tensor of the second orderz. 9ince the
nonhomogeneous set contains charges and currents (which represent the inBuence of matterz, one has to
attribute the corresponding pair (D, Hz to the inBuence of matter. It is, however, customary to refer to H
and not to B as the magnetic ®eld vector6 we shall conform to this terminology when there is no risj of
confusion.
{ The soWcalled Xaussian system of units is used here, i.e. the electrical ±uantities (E, D, j and rz are
measured in electrostatic units, and the magnetic ±uantities (H and Bz in electromagnetic units. The
constant c in (1z and ()z relates the unit of charge in the two systems6 it is the velocity of light in the
vacuum and is approximately e±ual to L 3 1-1- cm=s. (q more accurate value is given in }1.).z
1
)
I Masic properties of the electromagnetic ®eld
They are supplemented by two scalar relations:
div D ˆ 4ðr,
(Lz
div B ˆ -:
(4z
'±. (Lz may be regarded as a de®ning e±uation for the electric charge density r and (4z
may be said to imply that no free magnetic poles exist.
Orom (1z it follows (since div curl -z that
div j ˆ
1
_
div D,
4ð
or, using (Lz,
@r
‡ div j ˆ -:
@t
(Sz
My analogy with a similar relation encountered in hydrodynamics, (Sz is called the
equation of continuity. It expresses the fact that the charge is conserved in the
neighbourhood of any point. Oor if one integrates (Sz over any region of space, one
obtains, with the help of XaussN theorem,
d
r dV ‡ j . n dS ˆ -,
(Dz
dt
the second integral being tajen over the surface bounding the region and the ®rst
throughout the volume, n denoting the unit outward normal. This e±uation implies that
the total charge
e ˆ r dV
(Pz
contained within the domain can only increase on account of the Bow of electric
current
J ˆ j . n dS:
(Fz
If all the ®eld ±uantities are independent of time, and if, moreover, there are no
currents (j ˆ -z, the ®eld is said to be static. If all the ®eld ±uantities are time
independent, but currents are present (j 6ˆ -z, one speajs of a stationary ®eld. In
optical ®elds the ®eld vectors are very rapidly varying functions of time, but the
sources of the ®eld are usually such that, when averages over any macroscopic time
interval are considered rather than the instantaneous values, the properties of the ®eld
are found to be independent of the instant of time at which the average is tajen. The
word stationary is often used in a wider sense to describe a ®eld of this type. qn
example is a ®eld constituted by the steady Bux of radiation (say from a distant starz
through an optical system.
1.1.2 Material equations
The Raxwell e±uations (1zE(4z connect the ®ve basic ±uantities E, H, B, D and j. To
allow a uni±ue determination of the ®eld vectors from a given distribution of currents
1.1 The electromagnetic ®eld
L
and charges, these e±uations must be supplemented by relations which describe the
behaviour of substances under the inBuence of the ®eld. These relations are jnown as
material equations (or constitutive relationsz. In general they are rather complicated6
but if the ®eld is timeWharmonic (see }1.4.Lz, and if the bodies are at rest, or in very
slow motion relative to each other, and if the material is isotropic (i.e. when its
physical properties at each point are independent of directionz, they taje usually the
relatively simple form{
j ˆ ó E,
(Iz
D ˆ åE,
(1-z
B ˆ ìH:
(11z
Here ó is called the speci®c conductivity, å is jnown as the dielectric constant (or
permittivityz and ì is called the magnetic permeability.
'±. (Iz is the differential form of GhmNs law. 9ubstances for which ó 6ˆ - (or more
precisely is not negligibly small6 the precise meaning of this cannot, however, be
discussed herez are called conductors. Retals are very good conductors, but there are
other classes of good conducting materials such as ionic solutions in li±uids and also
in solids. In metals the conductivity decreases with increasing temperature. However,
in other classes of materials, jnown as semiconductors (e.g. germaniumz, conductivity
increases with temperature over a wide range.
9ubstances for which ó is negligibly small are called insulators or dielectrics. Their
electric and magnetic properties are then completely determined by å and ì. Oor most
substances the magnetic permeability ì is practically unity. If this is not the case, i.e.
if ì differs appreciably from unity, we say that the substance is magnetic. In particular,
if ì . 1, the substance is said to be paramagnetic (e.g. platinum, oxygen, nitrogen
dioxidez, while if ì , 1 it is said to be diamagnetic (e.g. bismuth, copper, hydrogen,
waterz.
If the ®elds are exceptionally strong, such as are obtained, for example, by focusing
light that is generated by a laser, the rightWhand sides of the material e±uations may
have to be supplemented by terms involving components of the ®eld vectors in powers
higher than the ®rst.{
In many cases the ±uantities ó, å and ì will be independent of the ®eld strengths6 in
other cases, however, the behaviour of the material cannot be described in such a
simple way. Thus, for example, in a gas of free ions the current, which is determined
There is an alternative way of describing the behaviour of matter. Instead of the ±uantities å ˆ D=E,
ì ˆ B= A one considers the differences D E and B H6 these have a simpler physical signi®cance and
will be discussed in Khapter II.
{ The more general relations, applicable also to moving bodies, are studied in the theory of relativity. Ye
shall only need the following result from the more general theory: that in the case of moving charges there
is, in addition to the conduction current óE, a convection current rv, where v is the velocity of the moving
charges and r the charge density (cf. p. Iz.
{ Vonlinear relationship between the displacement vector D and the electric ®eld E was ®rst demonstrated
in this way by U. q. Oranjen, q. '. Hill, K. Y. Ueters and X. Yeinrich, khysB LevB Cett., 7 (1ID1z, 11F.
Oor systematic treatments of nonlinear effects see V. Mloembergen, ®onlinear .ptics (Vew 2orj, Y. q.
Menjamin, Inc., 1IDSz, U. V. Mutcher and 7. Kotter, Mhe Elements of ®onlinear .ptics (Kambridge,
Kambridge 3niversity Uress, 1II-z or Â. Y. Moyd, ®onlinear .ptics (Moston, qcademic Uress, 1II)z.
4
I Masic properties of the electromagnetic ®eld
by the mean speed of the ions, depends, at any moment, not on the instantaneous value
of E, but on all its previous values. qgain, in soWcalled ferromagnetic substances
(substances which are very highly magnetic, e.g. iron, cobalt and nicjelz the value of
the magnetic induction B is determined by the past history of the ®eld H rather than by
its instantaneous value. The substance is then said to exhibit hysteresis. q similar
historyWdependence will be found for the electric displacement in certain dielectric
materials. Oortunately hysteretic effects are rarely signi®cant for the highWfre±uency
®eld encountered in optics.
In the main part of this booj we shall study the propagation in substances which
light can penetrate without appreciable weajening (e.g. air, glassz. 9uch substances are
said to be transparent and must be electrical nonconductors (ó ˆ -z, since conduction
implies the evolution of 8oule heat (see }1.1.4z and therefore loss of electromagnetic
energy. Gptical properties of conducting media will be discussed in Khapter :I/.
1.1.3 Boundary conditions at a surface of discontinuity
RaxwellNs e±uations were only stated for regions of space throughout which the
physical properties of the medium (characterized by å and ìz are continuous. In optics
one often deals with situations in which the properties change abruptly across one or
more surfaces. The vectors E, H, B and D may then be expected also to become
discontinuous, while r and j will degenerate into corresponding surface ±uantities. Ye
shall derive relations describing the transition across such a discontinuity surface.
1et us replace the sharp discontinuity surface M by a thin transition layer within
which å and ì vary rapidly but continuously from their values near M on one side to
their value near M on the other. Yithin this layer we construct a small nearWcylinder,
bounded by a stocjade of normals to M6 roofed and Boored by small areas äQ1 and
äQ) on each side of M, at constant distance from it, measured along their common
normal (Oig. 1.1z. 9ince B and its derivatives may be assumed to be continuous
throughout this cylinder, we may apply XaussN theorem to the integral of div B tajen
throughout the volume of the cylinder and obtain, from (4z,
div B dV ˆ B . n dS ˆ -6
(1)z
the second integral is tajen over the surface of the cylinder, and n is the unit outward
normal.
9ince the areas äQ1 and äQ) are assumed to be small, B may be considered to have
constants values B(1z and B()z on äQ1 and äQ), and (1)z may then be replaced by
n12
δA2
n2
δh
ε2, µ2
ε1, µ1
δA1
n1
T
Oig. 1.1 7erivation of boundary conditions for the normal components of B and D.
1.1 The electromagnetic ®eld
B(1z . n1 äQ1 ‡ B()z . n) äQ) ‡ contribution from walls ˆ -:
S
(1Lz
If the height äh of the cylinder decreases towards zero, the transition layer shrinjs into
the surface and the contribution from the walls of the cylinder tends to zero, provided
that there is no surface Bux of magnetic induction. 9uch Bux never occurs, and
conse±uently in the limit,
(B(1z . n1 ‡ B()z . n) zäQ ˆ -,
(14z
äQ being the area in which the cylinder intersects M . If n1) is the unit normal pointing
from the ®rst into the second medium, then n1 ˆ n1) , n) ˆ n1) and (14z gives
n1) . (B()z
B(1z z ˆ -,
(1Sz
i.e. the normal component of the magnetic induction is continuous across the surface
of discontinuity.
The electric displacement D may be treated in a similar way, but there will be an
additional term if charges are present. In place of (1)z we now have from (Lz
div D dV ˆ D . n dS ˆ 4ð r dV :
(1Dz
qs the areas äQ1 and äQ) shrinj together, the total charge remains ®nite, so that the
volume density becomes in®nite. Instead of the volume charge density r the concept
^ must then be used. It is de®ned by
of surface charge density r
^ dQ:
lim r dV ˆ r
(1Pz
ä h!-
Ye shall also need later the concept of surface current density ^j, de®ned in a similar
way:
lim j dV ˆ ^j dQ:
(1Fz
ä h!-
If the area äQ and the height äh are tajen suf®ciently small, (1Dz gives
D(1z . n1 äQ1 ‡ D()z . n) äQ) ‡ contribution from walls ˆ 4ð^
r äQ:
The contribution from the walls tends to zero with äh, and we therefore obtain in the
limit as äh ! -,
n1) . (D()z
D(1z z ˆ 4ð^
r,
(1Iz
^ on the surfacex the normal
i.e. in the presence of a layer of surface charge density r
component of the electric displacement changes abruptly across the surfacex by an
amount equal to 4ð^
r.
Oor later purposes we note a representation of the surface charge density and the surface current density in
terms of the 7irac delta function (see qppendix I/z. If the e±uation of the surface of discontinuity is
w(x, y, Tz ˆ -, then
^jgrad wjä(wz,
rˆr
(1Paz
j ˆ ^jjgrad wjä(wz:
(1Faz
These relations can immediately be veri®ed by substituting from (1Paz and (1Faz into (1Pz and (1Fz and
using the relation dw ˆ jgrad wjdh and the sifting property of the delta function.
D
I Masic properties of the electromagnetic ®eld
Vext, we examine the behaviour of the tangential components. 1et us replace the
sharp discontinuity surface by a continuous transition layer. Ye also replace the
cylinder of Oig. 1.1 by a 5rectangularN area with sides parallel and perpendicular to M
(Oig. 1.)z.
1et b be the unit vector perpendicular to the plane of the rectangle. Then it follows
from ()z and from 9tojesN theorem that
1 _ .
curl E . b dS ˆ E . dr ˆ
B b dS,
()-z
c
the ®rst and third integrals being tajen throughout the area of the rectangle, and the
second along its boundary. If the lengths k1 S1 (ˆ äs1 z, and k) S) (ˆ äs) z are small, E
may be replaced by constant values E (1z and E ()z along each of these segments.
9imilarly B_ may be replaced by a constant value. '±. ()-z then gives
1_ .
B b äsäh,
c
E (1z . t1 äs1 ‡ E ()z . t) äs) ‡ contribution from ends ˆ
()1z
where äs is the line element in which the rectangle intersects the surface. If now the
height of the rectangle is gradually decreased, the contribution from the ends k1 k) and
S1 S) will tend to zero, provided that E does not in the limit ac±uire suf®ciently sharp
singularities6 this possibility will be excluded. qssuming also that B_ remains ®nite, we
obtain in the limit as äh ! -,
(E (1z . t1 ‡ E ()z . t) zäs ˆ -:
())z
If t is the unit tangent along the surface, then (see Oig. 1.)z t1 ˆ
t) ˆ t ˆ b 3 n1) , and ())z gives
b . Qn1) 3 (E ()z
tˆ
b 3 n1) ,
E (1z zJ ˆ -:
9ince the orientation of the rectangle and conse±uently that of the unit vector b is
arbitrary, it follows that
n1) 3 (E ()z
E (1z z ˆ -,
()Lz
i.e. the tangential component of the electric vector is continuous across the surface.
Oinally consider the behaviour of the tangential component of the magnetic vector.
The analysis is similar, but there is an additional term if currents are present. In place
of ()1z we now have
n12
t2
P2
µ2
ε 2,
µ1
ε 1,
t1
Q2
Q1
t
T
b
P1
Oig. 1.) 7erivation of boundary conditions for the tangential components of E and H.
1.1 The electromagnetic ®eld
P
1
4ð
H (1z . t1 äs1 ‡ H ()z . t) äs) ‡ contribution from ends ˆ D_ . b äs äh ‡ ^j . b äs:
c
c
()4z
Gn proceeding to the limit äh ! - as before, we obtain
n1) 3 (H ()z
H (1z z ˆ
4ð ^
j:
c
()Sz
Orom ()Sz is follows that in the presence of a surface current of density ^jx the
tangential component (considered as a vector quantity) of the magnetic vector changes
abruptlyx its discontinuity being (4ð=cz^j 3 n1) .
qpart from discontinuities due to the abrupt changes in the physical properties of
the medium, the ®eld vectors may also be discontinuous because of the presence of a
source which begins to radiate at a particular instant of time t ˆ t- . The disturbance
then spreads into the surrounding space, and at any later instant t1 . t- will have ®lled
a wellWde®ned region. qcross the (movingz boundary of this region, the ®eld vectors
will change abruptly from ®nite values on the boundary to the value zero outside it.
The various cases of discontinuity may be covered by rewriting RaxwellNs e±uations
in an integral form. The general discontinuity conditions may also be written in the
form of simple difference e±uations6 a derivation of these e±uations is given in
qppendix /I.
1.1.4 The energy law of the electromagnetic ®eld
'lectromagnetic theory interprets the light intensity as the energy Bux of the ®eld. It is
therefore necessary to recall the energy law of RaxwellNs theory.
Orom (1z and ()z it follows that
E . curl H
H . curl E ˆ
4ð .
1
1
_
j E ‡ E . D_ ‡ H . B:
c
c
c
()Dz
qlso, by a wellWjnown vector identity, the term on the left may be expressed as the
divergence of the vector product of H and E:
E . curl H H . curl E ˆ div (E 3 Hz:
()Pz
Orom ()Dz and ()Pz we have that
1 . _
_ ‡ 4ð j . E ‡ div (E 3 Hz ˆ -:
(E D ‡ H . Bz
c
c
()Fz
Yhen we multiply this e±uation by c=4ð, integrate throughout an arbitrary volume and
apply XaussNs theorem, this gives
1
c
_
_
.
.
.
(E D ‡ H BzdV ‡ j E dV ‡
(E 3 Hz . n dS ˆ -,
()Iz
4ð
4ð
where the last integral is tajen over the boundary of the volume, n being the unit
outward normal.
The relation ()Iz is a direct conse±uence of RaxwellNs e±uations and is therefore
9ee, for example, q. 9ommerfeld, Electrodynamics (Vew 2orj, qcademic Uress, 1IS)z, p. 116 or 8. q.
9tratton, Electromagnetic Mheory (Vew 2orj, RcXrawWHill, 1I41z, p. D.
F
I Masic properties of the electromagnetic ®eld
valid whether or not the material e±uations (IzE(11z hold. It represents, as will be
seen, the energy law of an electromagnetic ®eld. Ye shall discuss it here only for the
case where the material e±uations (IzE(11z are satis®ed. Xeneralizations to anisotropic
media, where the material e±uations are of a more complicated form, will be considW
ered later (Khapter :/z.
Ye have, on using the material e±uations,
1
_ ˆ 1 E . @ (åEz ˆ 1 @ (åE ) z ˆ 1 @ (E . Dz, (E . Dz
4ð
4ð
@t
Fð @ t
Fð @ t
(L-z
1
_ ˆ 1 H . @ (ìHz ˆ 1 @ (ìH ) z ˆ 1 @ (H . Bz: (H . Bz
4ð
4ð
@t
Fð @ t
Fð @ t
9etting
we ˆ
and
()Iz becomes,
1 .
E D,
Fð
wm ˆ
1 .
H B,
Fð
V ˆ (we ‡ wm zdV ,
dV
c
.
‡ j E dV ‡
(E 3 Hz . n dS ˆ -:
dt
4ð
(L1z
(L)z
(LLz
Ye shall show that V represents the total energy contained within the volume, so that
we may be identi®ed with the electric energy density and wm with the magnetic energy
density of the ®eld.
To justify the interpretation of V as the total energy we have to show that, for a
closed system (i.e. one in which the ®eld on the boundary surface may be neglectedz,
the change in V as de®ned above is due to the worj done by the ®eld on the material
charged bodies which are embedded in it. It suf®ces to do this for slow motion of the
material bodies, which themselves may be assumed to be so small that they can be
regarded as point charges eZ (Z ˆ 1, ), . . .z. 1et the velocity of the charge eZ be v Z
(jv Z j cz.
The force exerted by a ®eld (E, Bz on a charge e moving with velocity v is given by
the soWcalled CorentT law,
1
Fˆ e E‡ v3B ,
(L4z
c
which is based on experience. It follows that if all the charges eZ are displaced by äx Z
(Z ˆ 1, ), . . .z in time ät, the total worj done is
In the general case the densities are de®ned by the expressions
1
1
E . dD,
wm ˆ
H . dB:
we ˆ
4ð
4ð
Yhen the relationship between E and D and between H and B is linear, as here assumed, these expressions
reduce to (L1z.
1.1 The electromagnetic ®eld
äQ ˆ
F Z . äx Z ˆ
Z
ˆ
Z
eZ E Z . äx Z ˆ
Z
eZ
I
1
E Z ‡ v Z 3 B . äx Z
c
eZ E Z . v Z ät,
Z
since äx Z ˆ v Z ät. If the number of charged particles is large, we can consider the
distribution to be continuous. Ye introduce the charge density r (i.e. total charge per
unit volumez and the last e±uation becomes
äQ ˆ ät rv . E dV ,
(LSz
the integration being carried throughout an arbitrary volume. Vow the velocity v does
not appear explicitly in RaxwellNs e±uations, but it may be introduced by using an
experimental result found by Âo;ntgenÈ, according to which a convection current (i.e. a
set of moving chargesz has the same electromagnetic effect as a conduction current in
a wire. Hence the current density j appearing in RaxwellNs e±uations can be split into
two parts
j ˆ jc ‡ jv ,
(LDz
where
jc ˆ ó E
is the conduction current density, and
jv ˆ rv
represents the convection current density. (LSz may therefore be written as
äQ ˆ ät jv . E dV :
(LPz
1et us now de®ne a vector S and a scalar S by the relations
c
Sˆ
(E 3 Hz,
4ð
S ˆ j c . E dV ˆ ó E ) dV :
Then by (LSz and (LDz
(LFz
(LIz
j . E dV ˆ S ‡ jv . E dV
ˆ S‡
äQ
,
ät
(4-z
where the second function is not, of course, a total derivative of a spaceWtime function.
'±. (LLz now tajes the form
È Y. K. Âo;ntgen, QnnB dB khysiZ, 35 (1FFFz, )D46 40 (1FI-z, IL.
1-
I Masic properties of the electromagnetic ®eld
dV
ˆ
dt
äQ
ät
S
S . n dS:
(41z
Oor a nonconductor (ó ˆ -z we have that S ˆ -. qssume also that the boundary
surface is so far awaythat we can neglect the ®eld on it, due to the electromagnetic
processes inside6 then S . n dS ˆ -, and integration of (41z gives
V ‡ Q ˆ constant:
(4)z
Hence, for an isolated system, the increase of V per unit time is due to the worj done
on the system during this time. This result justi®es our de®nition of electromagnetic
energy by means of (L)z.
The term S represents the resistive dissipation of energy (called 1oule's heatz in a
conductor (ó 6ˆ -z. qccording to (41z there is a further decrease in energy if the ®eld
extends to the boundary surface. The surface integral must therefore represent the Bow
of energy across this boundary surface. The vector S is jnown as the koynting vector
and represents the amount of energy which crosses per second a unit area normal to
the directions of E and H.
It should be noted that the interpretation of S as energy Bow (more precisely as the
density of the Bowz is an abstraction which introduces a certain degree of arbitrariness.
Oor the ±uantity which is physically signi®cant is, according to (41z, not S itself, but
the integral of S . n tajen over a closed surface. Klearly, from the value of the integral,
no unambiguous conclusion can be drawn about the detailed distribution of S, and
alternative de®nitions of the energy Bux density are therefore possible. Gne can always
add to S the curl of an arbitrary vector, since such a term will not contribute to the
surface integral as can be seen from XaussN theorem and the identity div curl -.
However, when the de®nition has been applied cautiously, in particular for averages
over small but ®nite regions of space or time, no contradictions with experiments have
been found. Ye shall therefore accept the above de®nition in terms of the Uoynting
vector of the density of the energy Bow.
Oinally we note that in a nonconducting medium (ó ˆ -z where no mechanical worj
is done (Q ˆ -z, the energy law may be written in the form of a hydrodynamical
continuity e±uation for noncompressible Buids:
@w
‡ div S ˆ -,
@t
(w ˆ we ‡ wm z:
(4Lz
q description of propagation of light in terms of a hydrodynamical model is often
helpful, particularly in the domain of geometrical optics and in connection with scalar
diffraction ®elds, as it gives a picture of the energy transport in a simple and graphic
manner. In optics, the (averagedz Uoynting vector is the chief ±uantity of interest. The
magnitude of the Uoynting vector is a measure of the light intensity, and its direction
represents the direction of propagation of the light.
qccording to modern theories of ®elds the arbitrariness is even greater, allowing for alternative expressions
for both the energy density and the energy Bux, but consistent with the change of the 1agrangian density
of the ®eld by the addition of a fourWdivergence. Oor a discussion of this subject see, for example, X.
Yentzel, Suantum Mheory of wields (Vew 2orj, Interscience Uublishers, 1I4Iz, especially }) or 8. 7.
8acjson, 9lassical Electrodynamics (Vew 2orj, 8. Yiley and 9ons, )nd ed. 1IPSz, 9ec. 1).1-, especially p.
D-).
1.) The wave e±uation and the velocity of light
11
1.- The Xave eTwation and the velocitR oD light
RaxwellNs e±uations relate the ®eld vectors by means of simultaneous differential
e±uations. Gn elimination we obtain differential e±uations which each of the vectors
must separately satisfy. Ye shall con®ne our attention to that part of the ®eld which
contains no charges or currents, i.e. where j ˆ - and r ˆ -.
Ye substitute for B from the material e±uation }1.1 (11z into the second Raxwell
e±uation }1.1 ()z, divide both sides by ì and apply the operator curl. This gives
1
1
_ ˆ -:
curl
curl E ‡ curl H
(1z
ì
c
Vext we differentiate the ®rst Raxwell e±uation }1.1 (1z with respect to time, use the
_ between the resulting e±uation
material e±uation }1.1 (1-z for D, and eliminate curl H
and (1z6 this gives
1
å 
curl
curl E ‡ ) E
ˆ -:
()z
ì
c
If we use the identities curl uv ˆ u curl v ‡ (grad uz 3 v and curl curl ˆ grad div
()z becomes
åì 
=) E
E ‡ (grad ln ìz 3 curl E grad div E ˆ -:
c)
=),
(Lz
qlso from }1.1 (Lz, using again the material e±uation for D and applying the identity
div uv ˆ u div v ‡ v . grad u we ®nd
å div E ‡ E . grad å ˆ -:
(4z
Hence (Lz may be written in the form
åì 
=) E
E ‡ (grad ln ìz 3 curl E ‡ grad (E . grad ln åz ˆ -:
c)
In a similar way we obtain an e±uation for H alone:
åì 
=) H
H ‡ (grad ln åz 3 curl H ‡ grad (H . grad ln ìz ˆ -:
c)
(Sz
(Dz
In particular, if the medium is homogeneous, grad log å ˆ grad ln ì ˆ -, and (Sz and
(Dz reduce to
åì 
åì 
=) E
E ˆ -,
=) H
H ˆ -:
(Pz
c)
c)
These are standard e±uations of wave motion and suggest the existence of electroW
magnetic waves propagated with a velocity
p
v ˆ c= åì:
(Fz
The concept of a velocity of an electromagnetic wave has actually an unambiguous meaning only in
connection with waves of very simple jind, e.g. plane waves. That v does not represent the velocity of
propagation of an arbitrary solution of (Pz is obvious if we bear in mind that these e±uations also admit
standing waves as solutions.
In this introductory section it is assumed that the reader is familiar with the concept of a plane wave,
and we regard v as the velocity with which such a wave advances. The mathematical representation of a
plane wave will be discussed in }1.L and }1.4.
1)
I Masic properties of the electromagnetic ®eld
The constant c was ®rst determined by Â. Zohlrausch and Y. Yeber in 1FSD from
the ratio of the values of the capacity of a condenser measured in electrostatic and
electromagnetic units, and it was found to be identical with the velocity of light in free
space. 3sing this result, Raxwell developed his electromagnetic theory of light,
predicting the existence of electromagnetic waves6 the correctness of this prediction
was con®rmed by the celebrated experiments of H. Hertz (see Historical introductionz.
qs in all wave theories of light, the elementary process which produces the optical
impression is regarded as being a harmonic wave in spaceWtime (studied in its simplest
form in }1.L and }1.4z. If its fre±uency is in the range from 4 3 1-14 s 1 to
P:S 3 1-14 s 1 (approximatelyz it gives rise to the psychological impression of a
de®nite colour. (The opposite, however, is not true: coloured light of a certain
subjective ±uality may be a composition of harmonic waves of very different
fre±uency distributions.z The actual connection between colour and fre±uency is very
involved and will not be studied in this booj.È
The ®rst determination of the velocity of light{ was made by Âo;mer in 1DPS from
observations of the eclipses of the ®rst satellite of 8upiter and later in a different way
(from aberration of ®xed starsz by Mradley (1P)Fz.
The ®rst measurements of the velocity of light from terrestrial sources were carried
out by Oizeau in 1F4I. It is necessary to employ a modulator, which marjs off a
portion of the beam{ and for this purpose Oizeau used a rotation wheel. 1ater methods
employed rotating mirrors or electronic shutters. The rotating mirror method was
suggested by Yheatstone in 1FL4 and was used by Ooucault in 1FD-. It was later
systematically developed over a period of many years by Richelson. The average
value based on about )-- measurements by Richelson gave c as )II,PID jm=s. qn
optical shutter method employing a Zerr cell was developed by Zarolus and RittelW
staedt (1I)Fz, qnderson (1ILPz and Hu;ttel (1I4-z. The values of c obtained from these
measurements are in excellent agreement with those based on indirect methods, such
as determinations from the ratio of an electric charge measured in electrostatic and
electromagnetic units6 for example Âosa and 7orsey (1I-Pz in this way found c as
)II,PF4 jm=s. Reasurements of the velocity of electromagnetic waves on wires
carried out by Rercier (1I)Lz gave the value of c e±ual to )II,PF) jm=s. The value
adopted by the Oifteenth Xeneral Konference of Yeights and Reasures} is
c ˆ )II,PI):4SF jm=s:
(Iz
The close agreement between the values of c obtained from measurements of very
different jinds (and in some cases using radiation whose fre±uencies differ by a factor
of hundreds of thousands from those used in the optical measurementsz gives a strijing
con®rmation of RaxwellNs theory.
The dielectric constant å is usually greater than unity, and ì is practically e±ual to
È The sensitivity of the human eye to different colours is, however, brieBy discussed in }4.F.1.
{ Oor a description of the methods used for determination of the velocity of light, see for example, '.
Mergstrand, Encyclopedia of khysics, ed. 9. Olu;gge, /ol. )4 (Merlin, 9pringer, 1ISDz, p. 1.
7etailed analysis of the results obtained by different methods is also given by Â. T. Mirge in LepB krogrB
khysB (1ondon, The Uhysical 9ocietyz, v (1I41z, I-.
{ 9uch determinations give essentially the group velocity (see }1.L.4z. The difference between the group
velocity and the phase velocity in air at standard temperature and pressure is about 1 part in S-,---.
} Konfèrence Xènèrale des Uoids et Resures, :/, Uaris, 1IPS, Komptes Âendus des 9èances (Uaris, Mureau
International des Uoids et Resures, 1IPDz.
1.) The wave e±uation and the velocity of light
1L
unity for transparent substances, so that the velocity v is then according to (Fz smaller
than the vacuum velocity c. This conclusion was ®rst demonstrated experimentally for
propagation of light in water in 1FS- by Ooucault and Oizeau.
The value of v is not as a rule determined directly, but only relative to c, with the
help of the law of refraction. qccording to this law, if a plane electromagnetic wave
falls on to a plane boundary between two homogeneous media, the sine of the angle è1
between the normal to the incident wave and the normal to the surface bears a constant
ratio to the sine of the angle è) between the normal of the refracted wave and the
surface normal (Oig. 1.Lz, this constant ratio being e±ual to the ratio of the velocities
v1 and v) of propagation in the two media:
sin è1 v1
ˆ :
sin è) v)
(1-z
This result will be derived in }1.S. Here we only note that it is e±uivalent to the
assumption that the waveWfront, though it has a jinj at the boundary, is continuous, so
that the line of intersection between the incident wave and the boundary travels at the
same speed (v9, sayz as the line of intersection between the refracted wave and the
boundary. Ye then have
v1 ˆ v9 sin è1 ,
v) ˆ v9 sin è) ,
(11z
from which, on elimination of v9, (1-z follows. This argument, in a slightly more
elaborate form, is often given as an illustration of HuygensN construction (}L.Lz.
The value of the constant ratio in (1-z is usually denoted by n1) and is called the
refractive index, for refraction from the ®rst into the second medium. Ye also de®ne
an 5absolute refractive indexN n of a medium6 it is the refractive index for refraction
from vacuum into that medium,
c
nˆ :
(1)z
v
If n1 and n) are the absolute refractive indices of two media, the (relativez refractive
index n1) for refraction from the ®rst into the second medium then is
n1) ˆ
n) v 1
ˆ :
n1 v )
Refracted
wave-front
(1Lz
θ2
v2, n2
θ1
v1, n1
Incident
wave-front
Oig. 1.L Illustrating the refraction of a plane wave.
14
I Masic properties of the electromagnetic ®eld
Table 1.1. Lefractive indices and static dielectric constants of
certain gases
qir
Hydrogen H)
Karbon dioxide KG)
Karbon monoxide KG
n (yellow lightz
p
å
1.---)I4
1.---1LF
1.---44I
1.---L4-
1.---)IS
1.---1L)
1.---4PL
1.---L4S
Table 1.). Lefractive indices and static dielectric constants of
certain liquids
Rethyl alcohol KHL GH
'thyl alcohol K) HS GH
Yater H) G
n (yellow lightz
p
å
1.L4
1.LD
1.LL
S.P
S.I.-
Komparison of (1)z and (Fz gives RaxwellNs formula:
p
n ˆ åì:
(14z
9ince for all substances with which we shall be concerned, ì is effectively unity
(nonmagnetic substancesz, the refractive index should then be e±ual to the s±uare root
of the dielectric constant, which has been assumed to be a constant of the material. Gn
the other hand, wellWjnown experiments on prismatic colours, ®rst carried out by
Vewton, show that the index of refraction depends on the colour, i.e. on the fre±uency
of the light. If we are to retain RaxwellNs formula, it must be supposed that å is not a
constant characteristic of the material, but is a function of the fre±uency of the ®eld.
The dependence of å on fre±uency can only be treated by tajing into account the
atomic structure of matter, and will be brieBy discussed in }).L.
RaxwellNs formula (with å e±ual to the static dielectric constantz gives a good
approximation for such substances as gases with a simple chemical structure which do not
disperse light substantially, i.e. for those whose optical properties do not strongly depend
on the colour of the light. Âesults of some early measurements for such gases, carried out
by 1. Moltzmann, are given in Table 1.1. '±. (14z also gives a good approximation for
li±uid
p
hydrocarbons6 for example benzene KD HD has n ˆ 1:4F) for yellow light whilst
å ˆ 1:4FI. Gn the other hand, there is a strong deviation from the formula for many
solid bodies (e.g. glassesz, and for some li±uids, as illustrated in Table 1.).
1.3 Scalar XaveS
In a homogeneous medium in regions free of currents and charges, each rectangular
component V (r, tz of the ®eld vectors satis®es, according to }1.) (Pz, the homogeneous
wave e±uation
1. Moltzmann, VienB BerB, OC (1FP4z, PIS6 koggB QnnB, 155 (1FPSz, 4-L6 VissB QbhB khysiZ0technB
LeichsanstB, 1, Vr. )D, SLP.
1.L 9calar waves
1S
1 @) V
ˆ -:
v) @ t )
=) V
(1z
Ye shall now brieBy examine the simplest solution of this e±uation.
1.3.1 Plane waves
1et r(x, y, Tz be a position vector of a point k in space and S(sx , sy , sT z a unit vector in
a ®xed direction. qny solution of (1z of the form
V ˆ V (r . S, tz
()z
is said to represent a plane wave, since at each instant of time V is constant over each
of the planes
r . S ˆ constant
which are perpendicular to the unit vector S.
It will be convenient to choose a new set of Kartesian axes .î, .ç, .æ with .æ in
the direction of S. Then (see Oig. 1.4z
r . S ˆ æ,
(Lz
and one has
@
@
ˆ sx ,
@x
@æ
@
@
ˆ sy ,
@y
@æ
@
@
ˆ sT :
@T
@æ
Orom these relations one easily ®nds that
@) V
,
@æ)
(4z
1 @) V
ˆ -:
v) @ t )
(Sz
=) V ˆ
so that (1z becomes
@) V
@æ)
If we set
æ
vt ˆ p,
æ ‡ vt ˆ q,
ζ
z
P
y
η
(Dz
s
r
ξ
O
Oig. 1.4 Uropagation of a plane wave.
x
1D
I Masic properties of the electromagnetic ®eld
(Sz tajes the form
@) V
ˆ -:
@ p@q
(Pz
The general solution of this e±uation is
V ˆ V1 ( pz ‡ V) (qz
ˆ V1 (r . S
vtz ‡ V) (r . S ‡ vtz,
(Fz
where V1 and V) are arbitrary functions.
Ye see that the argument of V1 is unchanged when (æ, tz is replaced by (æ ‡ vô,
t ‡ ôz, where ô is arbitrary. Hence V1 represents a disturbance which is propagated
with velocity v in the positive æ direction. 9imilarly V) (æ ‡ vtz represents a disturW
bance which is propagated with velocity v in the negative æ direction.
1.3.2 Spherical waves
Vext we consider solutions representing spherical waves, i.e. solutions of the form
V ˆ V (r, tz,
(Iz
where r ˆ jrj ˆ x ) ‡ y ) ‡ T ) .
3sing the relations @=@x ˆ (@ r=@xz(@=@ rz ˆ (x=rz(@=@ rz, etc., one ®nds after a
straightforward calculation that
=) V ˆ
1 @)
(rV z,
r @ r)
(1-z
so that the wave e±uation (1z now becomes
@)
(rV z
@ r)
1 @)
(rV z ˆ -:
v) @ t )
(11z
Vow this e±uation is identical with (Sz, if æ is replaced in the latter by r and V by rV.
Hence the solution of (11z can immediately be written down from (Fz:
Vˆ
V1 (r
vtz
r
‡
V) (r ‡ vtz
,
r
(1)z
V1 and V) being again arbitrary functions. The ®rst term on the rightWhand side of (1)z
represents a spherical wave diverging from the origin, the second a spherical wave
converging towards the origin, the velocity of propagation being v in both cases.
1.3.3 Harmonic waves. The phase velocity
qt a point r- in space the wave disturbance is a function of time only:
V (r- , tz ˆ w(tz:
(1Lz
qs will be evident from our earlier remarjs about colour, the case when w is periodic
is of particular interest. qccordingly we consider the case when w has the form
w(tz ˆ a cos(ùt ‡ äz:
(14z
1.L 9calar waves
1P
Here a (. -z is called the amplitude, and the argument ùt ‡ ä of the cosine term is
called the phase. The ±uantity
íˆ
ù
1
ˆ
)ð M
(1Sz
is called the frequency and represents the number of vibrations per second. ù is called
the angular fre±uency and gives the number of vibrations in )ð seconds. 9ince w
remains unchanged when t is replaced by t ‡ M, M is the period of the vibrations.
Yave functions (i.e. solutions of the wave e±uationz of the form (14z are said to be
harmonic with respect to time.
1et us ®rst consider a wave function which represents a harmonic plane wave
propagated in the direction speci®ed by a unit vector S. qccording to }1.L.1 it is
obtained on replacing t by t r . S=v in (14z:
r.S
V (r, tz ˆ a cos ù t
‡ä :
(1Dz
v
'±. (1Dz remains unchanged when r . S is replaced by r . S ‡ ë, where
ëˆv
)ð
ˆ vM :
ù
(1Pz
The length ë is called the wavelength. It is also useful to de®ne a reduced wavelength
ë- as
ë- ˆ cM ˆ në6
(1Fz
this is the wavelength which corresponds to a harmonic wave of the same fre±uency
propagated in vacuo. In spectroscopy one uses also the concept of a wave number k,
which is de®ned as the number of wavelengths in vacuo, per unit of length (cmz:
kˆ
1
í
ˆ :
ë- c
(1Iz
It is also convenient to de®ne vectors L- and L in the direction S of propagation,
whose lengths are respectively
)ð ù
ˆ ,
ëc
()-z
)ð nù ù
ˆ
ˆ :
ë
c
v
()1z
Z - ˆ )ðk ˆ
and
Z ˆ nZ - ˆ
The vector L ˆ ZS is called the wave vector or the propagation vector in the medium,
L- ˆ Z - S being the corresponding vector in the vacuum.
Instead of the constant ä one also uses the concept of path length l, which is the
distance through which a waveWfront recedes when the phase increases by ä:
lˆ
v
ë
ëäˆ
äˆ
ä:
)ðn
ù
)ð
())z
Ye shall refer to k as the 5spectroscopic wave numberN and reserve the term 5wave numberN for Z or Z,
de®ned by ()-z and ()1z, as customary in optics.
1F
I Masic properties of the electromagnetic ®eld
1et us now consider timeWharmonic waves of more complicated form. q general
timeWharmonic, real, scalar wave of fre±uency ù may be de®ned as a real solution of
the wave e±uation, of the form
V (r, tz ˆ a(rz cosQùt
g(rzJ,
()Lz
a (. -z and g being real scalar functions of positions. The surfaces
g(rz ˆ constant
()4z
are called cophasal surfaces or wave surfaces. In contrast with the previous case, the
surfaces of constant amplitude of the wave ()Lz do not, in general, coincide with the
surfaces of constant phase. 9uch a wave is said to be inhomogeneous.
Kalculations with harmonic waves are simpli®ed by the use of exponential instead
of trigonometric functions. '±. ()Lz may be written as
V (r, tz ˆ Rf± (rze
iùt
g,
()Sz
where
± (rz ˆ a(rzei g(rz ,
()Dz
and R denotes the real part. Gn substitution from ()Dz into the wave e±uation (1z, one
®nds that ± must satisfy the e±uation
=) ± ‡ n) Z - ) ± ˆ -:
()Pz
± is called the complex amplitude of the wave. In particular, for a plane wave one has
r.S
g(rz ˆ ù
ä ˆ Z(r . Sz ä ˆ L . r ä:
()Fz
v
If the operations on V are linear, one may drop the symbol R in ()Sz and operate
directly with the complex function, the real part of the ®nal expression being then
understood to represent the physical ±uantity in ±uestion. However, when dealing with
expressions which involve nonlinear operations such as s±uaring, etc. (e.g. in calculaW
tions of the electric or magnetic energy densitiesz, one must in general taje the real
parts ®rst and operate with these alone.{
3nlije a plane harmonic wave, the more general wave ()Sz is not periodic with
respect to space. The phase ùt g(rz is, however, seen to be the same for (r, tz and
(r ‡ dr, t ‡ dtz, provided that
ù dt (grad gz . dr ˆ -:
()Iz
If we denote by T the unit vector in the direction of dr, and write dr ˆ T ds, then ()Iz
gives
ds
ù
ˆ .
:
dt T grad g
(L-z
This expression will be numerically smallest when T is the normal to the cophasal
surface, i.e. when T ˆ grad g=jgrad gj, the value then being
In the case of a plane wave, one often separates the constant factor e iä and implies by complex amplitude
.
only the variable part aei Z r .
{ This is not necessary when only a time average of a ±uadratic expression is re±uired Qsee }1.4, (S4zE(SDzJ.
1.L 9calar waves
v( pz (rz ˆ
ù
:
jgrad gj
1I
(L1z
v( pz (rz is called the phase velocity and is the speed with which each of the cophasal
surfaces advances. Oor a plane electromagnetic wave one has from ()Fz that
grad g ˆ L, and therefore
ù
c
v( pz ˆ ˆ p
,
L
åì
because of ()1z. Oor waves of more complicated form, the phase velocity v( pz will in
p
general differ from c= åì and will vary from point to point even in a homogeneous
medium. However, it will be seen later (}L.1.)z that, when the fre±uency is suf®ciently
p
large, the phase velocity is approximately e±ual to c= åì, even for waves whose
cophasal surfaces are not plane.
It must be noted that the expression for ds=dt given by (L-z is not the resolute of the
phase velocity in the T direction, i.e. the phase velocity does not behave as a vector.
Gn the other hand its reciprocal, i.e. the ±uantity
dt T . grad g
ˆ
,
(L)z
ds
ù
is seen to be the component of the vector (grad gz=ù in the T direction. The vector
(grad gz=ù is sometimes called phase slowness.
The phase velocity may in certain cases be greater than c. Oor plane waves this will
p
be so when n ˆ åì is smaller than unity, as in the case of dispersing media in
regions of the soWcalled anomalous dispersion (see }).L.4z. Vow according to the
theory of relativity, signals can never travel faster than c. This implies that the phase
velocity cannot correspond to a velocity with which a signal is propagated. It is, in
fact, easy to see that the phase velocity cannot be determined experimentally and must
therefore be considered to be void of any direct physical signi®cance. Oor in order to
measure this velocity, it would be necessary to af®x a marj to the in®nitely extended
smooth wave and to measure the velocity of the marj. This would, however, mean the
replacement of the in®nite harmonic wave train by another function of space and time.
1.3.4 Wave packets. The group velocity
The monochromatic waves considered in the preceding section are idealizations never
strictly realized in practice. It follows from OourierNs theorem that any wave V (r, tz
(provided it satis®es certain very general conditionsz may be regarded as a superW
position of monochromatic waves of different fre±uencies:
1
V (r, tz ˆ
aù (rz cosQùt g ù (rzJ dù:
(LLz
-
The problem of propagation of electromagnetic signals in dispersive media has been investigated in classic
papers by q. 9ommerfeld, QnnB dB khysiZ, 44 (1I14z, 1PP and by 1. Mrillouin, ibid., 44 (1I14z, )-L.
'nglish translations of these papers are included in 1. Mrillouin, Vave kropagation and 2roup Velocity
(Vew 2orj, qcademic Uress, 1ID-z, pp. 1P, 4L. q systematic treatment of propagation of transient
electromagnetic ®elds through dielectric media which exhibit both dispersion and absorption is given in Z.
'. Gughstun and X. K. 9herman, Electromagnetic kulse kropagation in 9ausal Dielectrics (Merlin and
Vew 2orj, 9pringer, 1II4z.
)-
I Masic properties of the electromagnetic ®eld
It will again be convenient to use a complex representation, in which V is regarded as
the real part of an associated complex wave:
1
V (r, tz ˆ R aù (rze iQù t gù (rzJ dù:
(LLaz
-
q wave may be said to be 5almost monochromatic,N if the Oourier amplitudes aù differ
appreciably from zero only within a narrow range
ù
1
)Äù
< ù < ù ‡ 1)Äù
(Äù=ù 1z
around a mean fre±uency ù. In such a case one usually speajs of a wave group or a
wave pacZet.{
To illustrate some of the main properties of a wave group, consider ®rst a wave
formed by the superposition of two plane monochromatic waves of the same
amplitudes and slightly different fre±uencies and wave numbers, propagated in the
direction of the TWaxis:
V (T, tz ˆ ae
i(ù t ZTz
‡ ae
iQ(ù‡äùz t ( Z‡ä ZzTJ
:
(L4z
The symbol R is omitted here in accordance with the convention explained earlier. '±.
(L4z may be written in the form
1
V (T, tz ˆ aQe)i( täù‡Tä Zz ‡ e
ˆ )a cosQ1)(täù
1
)i( täù
TäZzJe
Tä Zz
Je
i(ù t Z Tz
i(ù t Z Tz
,
(LSz
where
ù ˆ ù ‡ 1)äù,
Z ˆ Z ‡ 1)äZ
(LDz
are the mean fre±uency and the mean wave number respectively. '±. (LSz may be
interpreted as representing a plane wave of fre±uency ù and wavelength )ð=Z
propagated in the T direction. The amplitude of this wave is, however, not constant, but
varies with time and position, between the values )a and - (Oig. 1.Sz, giving rise to the
wellWjnown phenomenon of beats. The successive maxima of the amplitude function
are at intervals
ät ˆ
4ð
(with T fixedz
äù
or
äT ˆ
4ð
(with t fixedz
äZ
(LPz
from each other, whilst the maxima of the phase function are at intervals
ät ˆ
)ð
(with T fixedz
ù
or
äT ˆ
)ð
(with t fixedz:
Z
(LFz
Hence, since äù=ù and äZ=Z are assumed to be small compared with unity, the
amplitude will vary slowly in comparison with the other term.
Orom (LSz it follows that the planes of constant amplitude and, in particular, the
maxima of the amplitude are propagated with the velocity
Oor a fuller discussion of the complex representation of real polychromatic waves see }1-.).
{ 9trictly speajing, in order that V should exhibit properties commonly attributed to a wave group, one
should also assume that over the effective fre±uency range the phase function gù can be approximated by
a linear function of ù.