MIT OpenCourseWare http://ocw.mit.

edu

Electromagnetic Field Theory: A Problem Solving Approach

For any use or distribution of this textbook, please cite as follows:

Markus Zahn, Electromagnetic Field Theory: A Problem Solving

Approach. (Massachusetts Institute of Technology: MIT
OpenCourseWare). http://ocw.mit.edu (accessed MM DD, YYYY).
License: Creative Commons Attribution-NonCommercial-Share Alike.

For more information about citing these materials or our Terms of

Use, visit: http://ocw.mit.edu/terms.
chapter 7

electrodynamics-
fields and waves
488 Electrodynamics-Fieldsand Waves

The electromagnetic field laws, derived thus far from the

empirically determined Coulomb-Lorentz forces, are correct
on the time scales of our own physical experiences. However,
just as Newton's force law must be corrected for material
speeds approaching that of light, the field laws must be cor-
rected when fast time variations are on the order of the time it
takes light to travel over the length of a system. Unlike the
abstractness of relativistic mechanics, the complete elec-
trodynamic equations describe a familiar phenomenon-
propagation of electromagnetic waves. Throughout the rest
of this text, we will examine when appropriate the low-
frequency limits to justify the past quasi-static assumptions.

7-1 MAXWELL'S EQUATIONS

7-1-1 Displacement Current Correction to Ampere's Law

In the historical development of electromagnetic field

theory through the nineteenth century, charge and its electric
field were studied separately from currents and their
magnetic fields. Until Faraday showed that a time varying
magnetic field generates an electric field, it was thought that
the electric and magnetic fields were distinct and uncoupled.
Faraday believed in the duality that a time varying electric
field should also generate a magnetic field, but he was not
able to prove this supposition.
It remained for James Clerk Maxwell to show that Fara-
day's hypothesis was correct and that without this correction
Ampere's law and conservation of charge were inconsistent:

VxH=JJ V Jf = 0 (1)

for if we take the divergence of Ampere's law in (1), the

current density must have zero divergence because the
divergence of the curl of a vector is always zero. This result
contradicts (2) if a time varying charge is present. Maxwell

_
 Maxwell's Equations 489

realized that adding the displacement current on the right-

hand side of Ampere's law would satisfy charge conservation,
because of Gauss's law relating D to pf (V D = pr).
This simple correction has far-reaching consequences,
because we will be able to show the existence of electro-
magnetic waves that travel at the speed of light c, thus proving
that light is an electromagnetic wave. Because of the
significance of Maxwell's correction, the complete set of
coupled electromagnetic field laws are called Maxwell's
equations:
Faraday's Law

VxE= B E dl= - B *dS (3)

at L d is

Ampere's law with Maxwell's displacement current correction

Vx H = Jf+D H - dl = Jr dS+d D dS (4)

at dte
Gauss's laws

V" D=pf > fs D.sdS= Pf dV (5)

V B=0 B dS=0 (6)

Conservation of charge

V" Jrf+L'=O JfdS+ v pfdV=O (7)

As we have justified, (7) is derived from the divergence of (4)

using (5).
Note that (6) is not independent of (3) for if we take the
divergence of Faraday's law, V - B could at most be a time-
independent function. Since we assume that at some point in
time B = 0, this function must be zero.
The symmetry in Maxwell's equations would be complete if
a magnetic charge density appeared on the right-hand side of
Gauss's law in (6) with an associated magnetic current due to
the flow of magnetic charge appearing on the right-hand side
of (3). Thus far, no one has found a magnetic charge or
current, although many people are actively looking.
Throughout this text we accept (3)-(7) keeping in mind that if
magnetic charge is discovered, we must modify (3) and (6)
and add an equation like (7) for conservation of magnetic
charge.

M = ýýý
490 Electrodynamics--Fiedsand Waves

7-1-2 Circuit Theory as a Quasi-static Approximation

Circuit theory assumes that the electric and magnetic fields

are highly localized within the circuit elements. Although the
displacement current is dominant within a capacitor, it is
negligible outside so that Ampere's law can neglect time vari-
ations of D making the current divergence-free. Then we
obtain Kirchoff's current law that the algebraic sum of all
currents flowing into (or out of) a node is zero:

V.J = 0=>JdS = E ik= (8)

Similarly, time varying magnetic flux that is dominant

within inductors and transformers is assumed negligible
outside so that the electric field is curl free. We then have
Kirchoff's voltage law that the algebraic sum of voltage drops
(or rises) around any closed loop in a circuit is zero:

VxE=O E=-VV* E dl=iO vA =0 (9)

7-2 CONSERVATION OF ENERGY

7-2-1 Poynting's Theorem

We expand the vector quantity

V -(ExH) =H (VxE)-E . (VxH)

= -H. B-_E D- -E *Jr (1)

at at

where we change the curl terms using Faraday's and

Ampere's laws.
For linear homogeneous media, including free space, the
constitutive laws are

D=eE, B=IAH (2)

so that (1) can be rewritten as

2 )
V. (ExH)+t(eE +AH' -E Jf (3)

which is known as Poynting's theorem. We integrate (3) over a

closed volume, using the divergence theorem to convert the
 Conservation of Energy 491

first term to a surface integral:

2
(ExH) -dS+ (E +E
+ H) dV=- E JIdV (4)

I V-(ExH)dV
V

We recognize the time derivative in (4) as operating on the

electric and magnetic energy densities, which suggests the
interpretation of (4) as
dW
Pou,+- = -Pa (5)

where Po., is the total electromagnetic power flowing out of

the volume with density
S = E x H watts/m 2 [kg-s-3] (6)
where S is called the Poynting vector, W is the electromag-
netic stored energy, and Pd is the power dissipated or
generated:

Po.t= (ExH).dS= S dS

2
W= [IeE + tH ] dV (7)

Pd = E -JdV

If E and J, are in the same direction as in an Ohmic conduc-

tor (E •Jr = oE 2), then Pd is positive, representing power dis-
sipation since the right-hand side of (5) is negative. A source
that supplies power to the volume has E and Jf in opposite
directions so that Pd is negative.

7-2-2 A Lossy Capacitor

Poynting's theorem offers a different and to some a

paradoxical explanation of power flow to circuit elements.
Consider the cylindrical lossy capacitor excited by a time
varying voltage source in Figure 7-1. The terminal current
has both Ohmic and displacement current contributions:

eAdv oAv dvv vA I

S+ = C-+- C=-T R= (8)
1 dT I dt R I 'A
From a circuit theory point of view we would say that the
power flows from the terminal wires, being dissipated in the

M
492 Electrodynamics-Fieldsand Waves

2
4 = ra

I~rc
Figure 7-1 The power delivered to a lossy cylindrical capacitor vi ispartly dissipated by
the Ohmic conduction and partly stored in the electric field. This power can also be
thought to flow-in radially from the surrounding electric and magnetic fields via the
Poynting vector S = E x H.

resistance and stored as electrical energy in the capacitor:

2
V d A fI 2)
P= vi= + d(Cv2) (9)
R dt
We obtain the same results from a field's viewpoint using
Poynting's theorem. Neglecting fringing, the electric field is
simply
E, = v/l (10)
while the magnetic field at the outside surface of the resistor
is generated by the conduction and displacement currents:
f 'dl=li- 8Ef,\ dS aAv e dv .
at I/1 dt
where we recognize the right-hand side as the terminal cur-
rent in (8),
H, = il(2ira) (12)

The power flow through the surface at r = a surrounding the

resistor is then radially inward,

S(E x H) dS = - l a ad dz = -vi (13)

. Jis1 2ira

and equals the familiar circuit power formula. The minus

sign arises because the left-hand side of (13) is the power out
of the volume as the surface area element dS points radially
outwards. From the field point of view, power flows into the
lossy capacitor from the electric and magnetic fields outside
 Conservationof Energy 493

the resistor via the Poynting vector. Whether the power is

thought to flow along the terminal wires or from the sur-
rounding fields is a matter of convenience as the results are
identical. The presence of the electric and magnetic fields are
directly due to the voltage and current. It is impossible to have
the fields without the related circuit variables.

7-2-3 Power in Electric Circuits

We saw in (13) that the flux of S entering the surface

surrounding a circuit element just equals vi. We can show this
for the general network with N terminals in Figure 7-2 using
the quasi-static field laws that describe networks outside the
circuit elements:
VxE=OE=-VV
VxE0=>E=-(14)
Vx H = Jf >V - Jf = 0
We then can rewrite the electromagnetic power into a surface
as

Pin=-s ExH*dS

=-IV -(ExH)dV
*v

=V. (VVxH)dV

Figure 7-2 The circuit power into an N terminal network E,..- VAl, equals the
electromagnetic power flow into the surface surrounding the network, -is E XH •dS.
494 Electrodynamics-Fieldsand Waves

where the minus is introduced because we want the power in

and we use the divergence theorem to convert the surface
integral to a volume integral. We expand the divergence term
as
0

V- (VVxH)=H- (VxVV)-VV. (VxH)

=-Jf - VV= -V (JTV) (16)

where we use (14).
Substituting (16) into (15) yields

Pin= - V. (JV)dV

=-JV-dS (17)

where we again use the divergence theorem. On the surface

S, the potential just equals the voltages on each terminal wire
allowing V to be brought outside the surface integral:

N
Pin= I - V• J, . dS
k=I s
N
= Y VAIl (18)
k=1

where we recognize the remaining surface integral as just

being the negative (remember dS points outward) of each
terminal current flowing into the volume. This formula is
usually given as a postulate along with Kirchoff's laws in most
circuit theory courses. Their correctness follows from the
quasi-static field laws that are only an approximation to more
general phenomena which we continue to explore.

7-2-4 The Complex Poynting's Theorem

For many situations the electric and magnetic fields vary

sinusoidally with time:

E(r, t) = Re [E(r) e"']

W .(19)
H(r, t) = Re [H(r) e"']

where the caret is used to indicate a complex amplitude that

can vary with position r. The instantaneous power density is
obtained by taking the cross product of E and H. However, it
is often useful to calculate the time-average power density
<S>, where we can avoid the lengthy algebraic and trig-
onometric manipulations in expanding the real parts in (19).
 Conservation of Energy 495

A simple rule for the time average of products is obtained by

realizing that the real part of a complex number is equal to
one half the sum of the complex number and its conjugate
(denoted by a superscript asterisk). The power density is then
S(r, t)= E(r, t) x H(r, t)
= ¼[f(r)e""+ i*(r) e -""] x [ I(r) e"' + I*(r) e -"]
= f[A(r) x H(r) e2t + E*(r) X I(r) + E(r) x A*(r)
- 2i '
+E*(r) x A*(r) e ] (20)
The time average of (20) is then
<5 > = ¼[k*(r)x 11(r) + A(r) x A*(r)]
= Re [i(r) X A*(r)]
= Re [*(r)xA(r)] (21)
as the complex exponential terms e 2iW" average to zero over a
period T = 2ir/w and we again realized that the first bracketed
term on the right-hand side of (21) was the sum of a complex
function and its conjugate.
Motivated by (21) we define the complex Poynting vector as

= ~(r)x A*(r) (22)

whose real part is just the time-average power density.

We can now derive a complex form of Poynting's theorem
by rewriting Maxwell's equations for sinusoidal time varia-
tions as
V X E(r) = -jwtIH(r)

Vx A(r) = J,(r) + jwe E(r)

V" E(r) = ýf(r)/e
(23)
V B(r) = 0
and expanding the product
•
V. = V * IE(r) x A*(r)] =-1[I*(r) V x E(r) - Ei(r) V x i*(r)]
1 12
= f[-joat I H(r)] +
jwe Ik(r)l 2] -E(r) *Jf (r) (24)
which can be rewritten as
V + 2j,[<w, > - <w,>] = -Pd (25)
where
<w,> =41 IHI(r)l 2
<w,> = I (r)l (26)
id = ~i(t) . jf(r)
496 Electrodynamics-Fieldsand Waves

We note that <w,> and <w,> are the time-average magnetic

and electric energy densities and that the complex Poynting's
theorem depends on their difference rather than their sum.

7-3 TRANSVERSE ELECTROMAGNETIC WAVES

7-3-1 Plane Waves

Let us try to find solutions to Maxwell's equations that only

depend on the z coordinate and time in linear media with
permittivity e and permeability M. In regions where there are
no sources so that pf=-0, Jr =0, Maxwell's equations then
reduce to
aE, aE,. aH
_iX + l, -y A (1)
az - z at
aH,. cH, aE
-- i. + =- (2)
az az at
aE,
e-= o (3)

aH,
a- = 0 (4)

These relations tell us that at best E, and H, are constant in

time and space. Because they are uncoupled, in the absence
of sources we take them to be zero. By separating vector
components in (1) and (2) we see that E2 is coupled to H, and
E, is coupled to H,:
aE, aH, aE, aH.
az at Oz at
(5)
aH, aE , aH, aE,
az at az at
forming two sets of independent equations. Each solution has
perpendicular electric and magnetic fields. The power flow
S= E X H for each solution is z directed also being perpendic-
ular to E and H. Since the fields and power flow are mutually
perpendicular, such solutions are called transverse elec-
tromagnetic waves (TEM). They are waves because if we take
a/az of the upper equations and a/at of the lower equations
and solve for the electric fields, we obtain one-dimensional
wave equations:
2
E_ 1 aRE_ R_ 1 a2E_
IX 9 r II
Z c Oat z c Oat-
 Transverse Electromagnetic Waves 497

where c is the speed of the wave,

1 1 3x 10 8
=-
c = - n
m/sec (7)

In free space, where e, = 1 and I, = 1, this quantity equals the

speed of light in vacuum which demonstrated that light is a
transverse electromagnetic wave. If we similarly take a/at of
the upper and a/az of the lower equations in (5), we obtain
wave equations in the magnetic fields:

a2 H, 1 a2 H, aH 1 a2H.
2=a2 a2 2 at2 (8)

7-3-2 The Wave Equation

(a) Solutions
These equations arise in many physical systems, so their
solutions are well known. Working with the E, and H, equa-
tions, the solutions are

E.(z, t)= E+(t-z/c)+E_(t+z/c)

H,(z, t) = H+(t- z/c) + H_(t + z/c)

where the functions E+, E_, H+, and H_ depend on initial

conditions in time and boundary conditions in space. These
solutions can be easily verified by defining the arguments a
and P with their resulting partial derivatives as

z aa aa 1
a = t---= 1,
c at az c
(10)
P=t+z=a= 1, a= I
c at az c

and realizing that the first partial derivatives of E,(z, t) are

aE. dE+aa dE_ ap

at da at dp at
dE+ dE-
da dO
(11)
aE_ dE+ aa dE_ ap
+-
az da az d1 az
1( dE+ dE+
c da dp
498 Electrodynamics-Fieldsand Waves

The second derivatives are then

a2 E, dE+
2 aa d 2E. ap
77 at dS at
d 2E+ d2 E

I 2E. 11 dE+ a d2E-aI(

z c 1 da2 az + d-
2 2
1 /d E+ d E-\ 1 a2E,

which satisfie's the wave equation of (6). Similar operations

apply for H,, E,, and H..
In (9), the pair H+ and E+as well as the pair H- and E. are
not independent, as can be seen by substituting the solutions
of (9) back into (5) and using (11):
8E.
az
=-- aH
at
- 1c ---
da
+-- =-,(dH+
dE+ dE.
-+-dH
\ da d
1
(13)
'
The functions of a and P must separately be equal,
-(E+ - AcH+) = 0, (E- + icH-)= 0 (14)
da dp
which requires that

E+ = uicH+= H+, E- = -cH_ =- HH (15)

where we use (7). Since / has units of Ohms, this quantity

is known as the wave impedance ?,

S= j - 120irj (16)

and has value 120ir 377 ohm in free space (I, = 1, e, = 1).
The power flux density in TEM waves is

S =ExH = [E+(t-z/c)+E-(t+z/c)]ix
x [H+(t- z/c) + H.(t + z/c)]i,
= (E+H++ E-H- + E-H++ E+H.)i, (17)

Using (15) and (16) this result can be written as

s, = (E+ -EP-) (18)

n1
 Transverse Electromagnetic Waves 499

where the last two cross terms in (17) cancel because of the
minus sign relating E- to H_ in (15). For TEM waves the total
power flux density is due to the difference in power densities
between the squares of the positively z-directed and nega-
tively z-directed waves.

(b) Properties
The solutions of (9) are propagating waves at speed c. To
see this, let us examine E+(t - z/c) and consider the case where
at z = 0, E+(t) is the staircase pulse shown in Figure 7-3a. In
Figure 7-3b we replace the argument t by t-z/c. As long as
the function E, is plotted versus its argument, no matter what
its argument is, the plot remains unchanged. However, in
Figure 7-3c the function E+(t -z/c) is plotted versus t result-
ing in the pulse being translated in time by an amount z/c. To
help in plotting this translated function, we use the following
logic:

(i) The pulse jumps to amplitude Eo when the argument is

zero. When the argument is t - z/c, this occurs for t = z/c.
(ii) The pulse jumps to amplitude 2Eo when the argument
is T. When the argument is t - z/c, this occurs for t =
T+ z/c.
(iii) The pulse returns to zero when the argument is 2 T. For
the argument t - z/c, we have t = 2 T+ z/c.

E, (t), = 0 E, (tQ--

(a) (b)
E+ (t )
C

I' -. . . .

(d)

Figure 7-3 (a) E+(t) at z = 0 is a staircase pulse. (b) E+(4) always has the same shape as
(a) when plotted versus 0, no matter what 0 is. Here .4 = t - z/c. (c) When plotted versus
t, the pulse is translated in time where z must be positive to keep t positive. (d) When
plotted versus z, it is translated and inverted. The pulse propagates at speed c in the
positive z direction.
500 Electrodynamics-Fieldsand Waves

Note that z can only be positive as causality imposes the

condition that time can only be increasing. The response at
any positive position z to an initial E, pulse imposed at z = 0
has the same shape in time but occurs at a time z/c later. The
pulse travels the distance z at the speed c. This is why the
function E,(t - z/c) is called a positively traveling wave.
In Figure 7-3d we plot the same function versus z. Its
appearance is inverted as that part of the pulse generated first
(step of amplitude EO) will reach any positive position z first.
The second step of amplitude 2Eo has not traveled as far
since it was generated a time T later. To help in plotting, we
use the same criterion on the argument as used in the plot
versus time, only we solve for z. The important rule we use is
that as long as the argument of a function remains constant,
the value of the function is unchanged, no matter how the
individual terms in the argument change.
Thus, as long as
t - z/c = const (19)
E+(t - z/c) is unchanged. As time increases, so must z to satisfy
(19) at the rate
z dz
t - = const - =c (20)
c dt
to keep the E, function constant.
For similar reasons E_(t + z/c) represents a traveling wave at
the speed c in the negative z direction as an observer must
move to keep the argument t + z/c constant at speed:
z dz
t +- = const • = -c (21)
c dt
as demonstrated for the same staircase pulse in Figure 7-4.
Note in Figure 7-4d that the pulse is not inverted when
plotted versus z as it was for the positively traveling wave,
because that part of the pulse generated first (step of ampli-
tude Eo) reaches the maximum distance but in the negative z
direction. These differences between the positively and nega-
tively traveling waves are functionally due to the difference in
signs in the arguments (t- z/c) and (t + z/c).

7-3-3 Sources of Plane Waves

These solutions are called plane waves because at any

constant z plane the fields are constant and do not vary with
the x and y coordinates.
The idealized source of a plane wave is a time varying
current sheet of infinite extent that we take to be x directed,
 Transverse Electromagnetic Waves 501

E (t).z=0 E (t + )

(a) (b)

(c) (d)

Figure 7-4 (a) E_(t) at z = O0 is a staircase pulse. (b) E-(4) always has the same form of
(a) when plotted versus 4. Here 46 = t + z/c. (c) When plotted versus t, the pulse is
translated in time where z must be negative to keep t positive. (d)When plotted versus z,
it is translated but not inverted.

as shown in Figure 7-5. From the boundary condition on the

discontinuity of tangential H, we find that the x-directed
current sheet gives rise to a y-directed magnetic field:

H,(z = 0+)- H,(z = 0_) = -K,(t)

In general, a uniform current sheet gives rise to a magnetic

field perpendicular to the direction of current flow but in the
plane of the sheet. Thus to generate an x-directed magnetic
field, a y-directed surface current is required.
Since there are no other sources, the waves must travel
away from the sheet so that the solutions on each side of the
sheet are of the form

SH(t - z/c) S)H+(t - z/c), z>0

H_(t + z/c) -rqH_(t + z/c), z<0
(23)
For z > 0, the waves propagate only in the positive z direction.
In the absence of any other sources or boundaries, there can
be no negatively traveling waves in this region. Similarly for
z <0, we only have waves propagating in the -z direction. In
addition to the boundary condition of (22), the tangential
component of E must be continuous across the sheet at z = 0

H+(t)- H_(t) = -K,(t) = H -Kt(t)

7l[H+(t)+H-(t)]=0 (24)2
 502 Electrodynamics-Fieldsand Waves

K (t - c )
E, (z, t) = Ct
+ ) Ex (z, t) = -
2 2
S
S *..
Kx2t +
Hy (z, t) K, (t -- )
2 H, (z, t)= 2- c

K, (t)

2Ko
Ko
E, (z, t)
T 2T

-E L _LJ - 7LK
o

Hy (z, t)

-Ko

t~l . ct
. •tz

- -Ko L-r-j
S, (z, t) = ExHy

(b)

Figure 7-5 (a) A linearly polarized plane wave is generated by an infinite current sheet.
The electric field is in the direction opposite to the current on either side of the sheet.
The magnetic field is perpendicular to the current but in the plane of the current sheet
and in opposite directions as given by the right-hand rule on either side of the sheet. The
power flowS is thus perpendicular to the current and to the sheet. (b) The field solutions
for t > 2 T if the current source is a staircase pulse in time.
 Transverse Electromagnetic Waves 503

so that the electric and magnetic fields have the same shape as
the current. Because the time and space shape of the fields
remains unchanged as the waves propagate, linear dielectric
media are said to be nondispersive.
Note that the electric field at z = 0 is in the opposite direc-
tion as the current, so the power per unit area delivered by
the current sheet,

-E(z = 0, t) . K.(t) = ) (25)

2

is equally carried away by the Poynting vector on each side of

the sheet:

iz, z>0
S(z - 0)=Ex H iK (t) (26)

4
7-3-4 A Brief Introduction to the Theory of Relativity

Maxwell's equations show that electromagnetic waves

propagate at the speed co = 1/,,/io in vacuum. Our
natural intuition would tell us that if we moved at a speed v we
would measure a wave speed of co - v when moving in the same
direction as the wave, and a speed co + v when moving in the
opposite direction. However, our intuition would be wrong,
for nowhere in the free space, source-free Maxwell's equa-
tions does the speed of the observer appear. Maxwell's equa-
tions predict that the speed of electromagnetic waves is co for
all observers no matter their relative velocity. This assump-
tion is a fundamental postulate of the theory of relativity and
has been verified by all experiments. The most notable
experiment was performed by A. A. Michelson and E. W.
Morley in the late nineteenth century, where they showed
that the speed of light reflected between mirrors is the same
whether it propagated in the direction parallel or perpendic-
ular to the velocity of the earth. This postulate required a
revision of the usual notions of time and distance.
If the surface current sheet of Section 7-3-3 is first turned
on at t = 0, the position of the wave front on either side of the
sheet at time t later obeys the equality

z 2 - C2 =0 (27)

Similarly, an observer in a coordinate system moving with

constant velocity ui, which is aligned with the current sheet at
504 Electrodynamics-Fieldsand Waves

t = 0 finds the wavefront position to obey the equality

z ' -cot= 0 (28)

The two coordinate systems must be related by a linear

transformation of the form
zr = a1z +a 2 t, t'= b1z +b 2 t (29)
The position of the origin of the moving frame (z'= 0) as
measured in the stationary frame is z = vt, as shown in Figure
7-6, so that a, and a2 are related as

O= alvt + a 2 t av+a2 = (30)

We can also equate the two equalities of (27) and (28),
2 22= 2_ 2,2. = )2_C2
z -cot =z-c•t =(az +a 2 t)2-ct(b 1 z +b 20t) 2 (31)
so that combining terms yields
2
2(l2+C2 2)
2t2 2 2 2
z (-al +cobi)-cot\ 1+- - 2) -2(aia 2 -cobib 2 )zt= 0
co
(32)
Since (32) must be true for all z and t, each of the coefficients
must be zero, which with (30) gives solutions
1 = -v/c2 2
bl =1 -(v/co)
1 - (V/io)
(33)
-V bs= 1 2
1a2
-(v/co)
11 -(v/co)

-g

Figure 7-6 The primed coordinate system moves at constant velocity vi, with respect
to a stationary coordinate system. The free space speed of an electromagnetic wave is co
as measured by observers in either coordinate system no matter the velocity v.

·I_ __
 Sinusoidal Time Variations 505

The transformations of (29) are then

z - vt t - vz/c
z= t'= (34)
'
1- (V/co) 11 - (V/co)(
and are known as the Lorentz transformations. Measured
lengths and time intervals are different for observers moving
at different speeds. If the velocity v is much less than the
speed of light, (34) reduces to the Galilean transformations,

lim z'-z-vt, t' t (35)

vlc<< 1

which describe our usual experiences at nonrelativistic

speeds.
The coordinates perpendicular to the motion are
unaffected by the relative velocity between reference frames
x'= x Y,y'= y (36)

Continued development of the theory of relativity is be-

yond the scope of this text and is worth a course unto itself.
Applying the Lorentz transformation to Newton's law and
Maxwell's equations yield new results that at first appearance
seem contrary to our experiences because we live in a world
where most material velocities are much less than co.
However, continued experiments on such disparate time and
space scales as between atomic physics and astronomics verify
the predictions of relativity theory, in part spawned by Max-
well's equations.

7-4 SINUSOIDAL TIME VARIATIONS

7-4-1 Frequency and Wavenumber

If the current sheet of Section 7-3-3 varies sinusoidally with

time as Re (Ko e"i'), the wave solutions require the fields to
vary as ei'' t- Z1
C)
and ei'(t+ic) :

2
H,(z, t) z<0
Re +K eio2'"+c),

(1)
Re(- e-'-', z>0
t)
Ex (z,
Re( - +z~c)),z <0
eitKo
506 Electrodynamics-Fieldsand Waves

At a fixed time the fields then also vary sinusoidally with

position so that it is convenient to define the wavenumber as

2k
Ac acc
where A is the fundamental spatial period of the wave. At a
fixed position the waveform is also periodic in time with
period T:
"
1 2.
T=1=
f a
where f is the frequency of the source. Using (3) with (2) gives
us the familiar frequency-wavelength formula:

O = kc _fAý = c (4)
Throughout the electromagnetic spectrum, summarized in
Figure 7-7, time varying phenomena differ only in the scaling
of time and size. No matter the frequency or wavelength,
although easily encompassing 20 orders of magnitude, elec-
tromagnetic phenomena are all described by Maxwell's equa-
tions. Note that visible light only takes up a tiny fraction of the
spectrum.

6 4 2 - 2 - a - 12
Xmeters 3x1 f 3x10 3x10 3 3x10 3x10-4 3x1076 3x10 3x10-' 3x10
f(Hz)
4
0 102 10 106 108 1010 1012 1014 1016 10la 1020
I I I I I I I I I I I
Power Radio and television Infrared Visible Ultraviolet X-rays Gamma
(heat) light Red (700nm) rays
AM FM
AM FM Orange (650nm)
Yellow (600nm)
Green (550nm)
Circuit theory Microwaves Blue (450nm)
Violet (400nm)

Figure 7-7 Time varying electromagnetic phenomena differ only in the scaling of time
(frequency) and size (wavelength). In linear dielectri,ymedia the frequency and
wavelength are related as fA = c (to= kc), where c = 1/Ve'g is the speed of light in the
medium.
 Sinusoidal Time Variations 507

For a single sinusoidally varying plane wave, the time-

average electric and magnetic energy densities are equal
because the electric and magnetic field amplitudes are related
through the wave impedance 77:
<w,,> = <w,>= 2 = 1 El2HI K 2eEL (5)

From the complex Poynting theorem derived in Section

7-2-4, we then see that in a lossless region with no sources for
Iz >0 that Pd = 0 so that the complex Poynting vector has
zero divergence. With only one-dimensional variations with z,
this requires the time-average power density to be a constant
throughout space on each side of the current sheet:
<S > = - Re [i(r)X A*(r)]
(K~it, z>0
= 1 z< (6)

The discontinuity in <S > at z = 0is due to the power output of

the source.

7-4-2 Doppler Frequency Shifts

If the sinusoidally varying current sheet Re (Ko e i w')moves

with constant velocity vi,, as in Figure 7-8, the boundary
conditions are no longer at z = 0 but at z = vt. The general
form of field solutions are then:

Re (-±+ ei•,+-(zc)), z > vt

Re (H- e i ' - '
+ /c)) z <vt
(7)
-/ z> vt
ei'
[Re (nH.+ ),
Re (-i7/1-
ed-('+~c)), z <vt

where the frequencies of the fields w+ and w- on each side of

the sheet will be different from each other as well as differing
from the frequency of the current source w. We assume
v/c << I so that we can neglect relativistic effects discussed in
Section 7-3-4. The boundary conditions
' " -
'
E,.(z = vt) = E,_(z = vt)> A+ ei+ /c = _--_
e-'"'+v'c

H,,(z = vt) - H, (z= vt) = -K, (8)

=,+ ee,t(-c) -
(c) = -Ko ed" t

must be satisfied for all values of t so that the exponential time

factors in (8) must all be equal, which gives the shifted
508 Electrodynamics--Fieldsand Waves

,I
Re(Koe wt )

E = Re [--- e '

E,=Re[-l -e Ko ij, (-L)

HY =Re[ -- y +

Ko i•w- (t+ 1) I
Hy =Re [ ] O++
0(1+) C

I oI t

Figure 7-8 When a source of electromagnetic waves moves towards an observer, the
frequency is raised while it is lowered when it moves away from an observer.

frequencies on each side of the sheet as

- d1-Hv/c/
1-
+ Ko
2+-H-'
1+v/c 2
where v/c << 1. When the source is moving towards an obser-
ver, the frequency is raised while it is lowered when it moves
away. Such frequency changes due to the motion of a source
or observer are called Doppler shifts and are used to measure
the velocities of moving bodies in radar systems. For v/c << 1,
the frequency shifts are a small percentage of the driving
frequency, but in absolute terms can be large enough to be
easily measured. At a velocity v = 300 rm/sec with a driving
frequency of f- 10'0 Hz, the frequency is raised and lowered
on each side of the sheet by Af= +f(v/c) = -10 4 Hz.

7-4-3 Ohmic Losses

Thus far we have only considered lossless materials. If the

medium also has an Ohmic conductivity o, the electric field

_·
 Sinusoidal Time Variations 509

will cause a current flow that must be included in Ampere's

law:
oE, OH,
Oz at
(10)
aH, aE. aE, (10)
S- - e-- = - aE,-
-
at
az at
where for conciseness we only consider the x-directed electric
field solution as the same results hold for the E,, H, solution.
Our wave solutions of Section 7-3-2 no longer hold with this
additional term, but because Maxwell's equations are linear
with constant coefficients, for sinusoidal time variations the
solutions in space must also be exponential functions, which
we write as
E,(z, t)= Re (Eo e i'(t-k))
'

H,(z, t) = Re (H 0o ei(~t - k))

h

where Eo and H 0oare complex amplitudes and the wavenum-

ber k is no longer simply related to w as in (4) but is found by
substituting (11) back into (10):
-jkEo = -jwioHo
(12)
-jkHo = -jWe (1 + ojwe )Eo
This last relation was written in a way that shows that the
conductivity enters in the same way as the permittivity so that
we can define a complex permittivity E as
S= e(1+ o/ljW) (13)
Then the solutions to (12) are

---- j k2 = W2,AE = oW2

Ho 0 Wo
t -A
k hk
we 2 a,2Ms(1+cr
ws (14)

which is similar in form to (2) with a complex permittivity.

There are two interesting limits of (14):

(a) Low Loss Limit

If the conductivity is small so that o/aoe << 1, then the solution
of (14) reduces to

lim k= =h& e(1+ 2-) (- (15)

/,.e4Cx
I c 2 E)
where c is the speed of the light in the medium if there were
no losses, c = 1/IE. Because of the spatial exponential
dependence in (11), the real part of k is the same as for the
510 Electrodynamics-Fieldsand Waves

lossless case and represents the sinusoidal spatial distribution

of the fields. The imaginary part of k represents the
exponential decay of the fields due to the Ohmic losses with
exponential decay length •lo, where 71 = -4Ie is the wave
impedance. Note that for waves traveling in the positive z
direction we take the upper positive sign in (15) using the
lower negative sign for negatively traveling waves so that the
solutions all decay and do not grow for distances far from the
source. This solution is only valid for small o"so that the wave is
only slightly damped as it propagates, as illustrated in Figure
7-9a.

Low loss limit

6 6 5
eV/ e2/6 eP' e-'

_____ ,,g. z

Large lass limit

Figure 7-9 (a) In a slightly lossy dielectric, the fields decay away from a source at a slow
rate while the wavelength is essentially unchanged. (b) In the large loss limit the spatial
decay rate is equal to the skin depth. The wavelength also equals the skin depth.
 Sinusoidal Time Variations 511

(b) Large Loss Limit

In the other extreme of a highly conducting material so
that o/wse >> 1, (14) reduces to

lim k 2 =-joAcrk=>k (I-j), 8= (16)

where 8 is just the skin depth found in Section 6-4-3 for

magneto-quasi-static fields within a conductor. The skin-
depth term also arises for electrodynamic fields because the
large loss limit has negligible displacement current compared
to the conduction currents.
Because the real and imaginary part of k have equal
magnitudes, the spatial decay rate is large so that within a few
oscillation intervals the fields are negligibly small, as illus-
trated in Figure 7-9b. For a metal like copper with A= o=
41r x 10- 7 henry/m and o - 6 x 107 siemens/m at a frequency
of 1 MHz, the skin depth is 8 -6.5 x 10-5 m.

7-4-4 High-Frequency Wave Propagation in Media

Ohm's law is only valid for frequencies much below the

collision frequencies of the charge carriers, which is typically
on the order of 1013 Hz. In this low-frequency regime the
inertia of the particles is negligible. For frequencies much
higher than the collision frequency the inertia dominates and
the current constitutive law for a single species of charge
carrier q with mass m and number density n is as found in
Section 3-2-2d:
aJ,/at= O~E (17)

where w( = ,•-n/ me is the plasma frequency. This constitutive

law is accurate for radio waves propagating in the ionosphere,
for light waves propagating in many dielectrics, and is also
valid for superconductors where the collision frequency is
zero.
Using (17) rather than Ohm's law in (10) for sinusoidal time
and space variations as given in (11), Maxwell's equations are

-jk°o = 8
az S at '
depende(18)
frequency
isnow
permittivity
effective
The
ýH, aE. W P

The effective permittivity is now frequency dependent:

A= e(1-w
~ /o2•
512 Electrodynamics-Fieldsand Waves

The solutions to (18) are

0o O k 2 2 . CO WP
-.. g = 2 (20)
Ho k we c

For w > wp, k is real and we have pure propagation where the
wavenumber depends on the frequency. For w <we,, k is
imaginary representing pure exponential decay.
Poynting's theorem for this medium is

V
a
S-+ (½sIEI + h&IH62) =-E JI=---21 • oJf
-
at woe at

a 1- 2 (21)

Because this system is lossless, the right-hand side of (21)

can be brought to the left-hand side and lumped with the
energy densities:

V.S+- [ EI 2+•lHI2+1 I j112=0 (22)

at 2 woe
This new energy term just represents the kinetic energy
density of the charge carriers since their velocity is related to
the current density as
11 J [
Jr=qnv= 2 " J--• =mnnvI (23)

7-4-5 Dispersive Media

When the wavenumber is not proportional to the

frequency of the wave, the medium is said to be dispersive. A
nonsinusoidal time signal (such as a square wave) will change
shape and become distorted as the wave propagates because
each Fourier component of the signal travels at a different
speed.
To be specific, consider A stationary current sheet source at
z = 0 composed of two signals with slightly different frequen-
cies:
K(t) = Ko[cos (wo + Ao)t + cos (woo- A)t]
= 2Ko cos Awt cos wot (24)

With Aw << w the fast oscillations at frequency wo are modu-

lated by the slow envelope function at frequency Aw. In a
linear dielectric medium this wave packet would propagate
away from the current sheet at the speed of light, c = 1/v-e).

·
 Sinusoidal Time Variations 513

If the medium is dispersive. with the wavenumber k(w) being

a function of w, each frequency component in (24) travels at a
slightly different speed. Since each frequency is very close to
w0 we expand k(w) as
dk
k(jo+ w) k(wo)+-, A,&
(25)

k(&Oo-Aw)= k(wo)-d-- Ao

where for propagation k(wo) must be real.

The fields for waves propagating in the +z direction are
then of the following form:

E,.(z, t)= Re Eo exp I(woo+w)t - (wo)+k

+z AW

+exp [ (woJ-Aw)t- k(wodw)A z

=Re (. exp {j[wot -k(wo)z]}j exp j AW t- A ]

+exp -j Am t - z

- k(wo)z) cosw
= 2Eo cos (wotk At - z (26)

where without loss of generality we assume in the last relation

that E0 = Eo is real. This result is plotted in Figure 7-10 as a
function of z for fixed time. The fast waves with argument
wot -k((oo)z travel at the phase speed vp = wo/k(wo) through
the modulating envelope with argument Aw(t-dk/dwooz).
This envelope itself travels at the slow speed
dk dz d(
t- z= const =v =d (27)

known as the group velocity, for it is the velocity at which a

packet of waves within a narrow frequency band around wo
will travel.
For linear media the group and phase velocities are equal:

wo= kc > v, = =

(28)
Vd
do
514 Electrodynamics-Fieldsand Waves

E4 (s, t = 0) E-a •nl It -A--)1 cosrAwit---ls

,°

Modulating
envelope cos [A(t- )]

Figure 7-10 In a dispersive medium the shape of the waves becomes distorted so the
velocity of a wave is not uniquely defined. For a group of signals within a narrow
frequency band the modulating envelope travels at the group velocity v,. The signal
within the envelope propagates through at the phase velocity v,.

while from Section 7-4-4 in the high-frequency limit for

conductors, we see that

W2 = k2C2++ 2W2 =VW

k 2 (29)
dw
V, = - = -C
dk w
where the velocities only make sense when k is real so that
a > 04. Note that in this limit
VvJ, = C 2 (30)
Group velocity only has meaning in a dispersive medium
when the signals of interest are clustered over a narrow
frequency range so that the slope defined by (27), is approxi-
mately constant and real.

7-4-6 Polarization

The two independent sets of solutions of Section 7-3-1 both

have their power flow S = E x H in the z direction. One solu-
tion is said to have its electric field polarized in the x direction
 Sinusoidal Time Variations 515

while the second has its electric field polarized in the y direc-
tion. Each solution alone is said to be linearly polarized
because the electric field always points in the same direction
for all time. If both field solutions are present, the direction
of net electric field varies with time. In particular, let us say
that the x and y components of electric field at any value of z
differ in phase by angle 4:
E = Re [Eoi. + E, e'i,] e• ' = Eo cos wti. + E, cos (ot + 4))i,
(31)
We can eliminate time as a parameter, realizing from (31) that
cos wt = EE,
(32)
(32)
sin w cos at cos 4 - EE, = (EJEI,) cos 4)- EE,
sin 4 sin 4
and using the identity that
sin wt + cos 2 (Ot

= 1(E ) (EJEn)2 cos 2 4)+ (E/E,) 2 - (2E.E/EýoE,0 ) cos 4

I Eo. sin 2 40
(33)
to give us the equation of an ellipse relating E, to E,:
(E +( E, 2 2EE, 2
S cos4 = sin2 4 (34)
(E,. 0 E E.E,
as plotted in Figure 7-11 a. As time increases the electric field
vector traces out an ellipse each period so this general case of
the superposition of two linear polarizations with arbitrary
phase 4 is known as elliptical polarization. There are two
important special cases:

(a) Linear Polarization

If E. and E, are in phase so that 4 = 0, (34) reduces to
(E.E,\ 2
E, E
E,, E•0 = tanO=-• (35)
Ex E.K E, E,.
The electric field at all times is at a constant angle 0 to the x
axis. The electric field amplitude oscillates with time along
this line, as in Figure 7-1 lb. Because its direction is always
along the same line, the electric field is linearly polarized.

(b) Circular Polarization

If both components have equal amplitudes but are 90* out
of phase,
E.o= Eo Eo, 4 = fir/2
 516 Eletrodynamics-Fieldsand Waves

E=Exoix +Eoe ioi

( Ex)2 +
xo fyo -xo Yo0

Figure 7-11 (a)Two perpendicular field components with phase difference 46 have the
tip of the net electric field vector tracing out an ellipse each period. (b) If both field
components are in phase, the ellipse reduces to a straight line. (c) If the field
components have the same magnitude but are 90* out of phase, the ellipse becomes a
circle. The polarization is left circularly polarized to z-directed power flow if the electric
field rotates clockwise and is (d) right circularly polarized if it rotates counterclockwise.

(34) reduces to the equation of a circle:

E2 +E2 = E0 (37)
The tip of the electric field vector traces out a circle as time
evolves over a period, as in Figure 7-11 c. For the upper (+)
sign for 4 in (36), the electric field rotates clockwise while the
negative sign has the electric field rotating counterclockwise.
These cases are, respectively, called left and right circular
polarization for waves propagating in the +z direction as
found by placing the thumb of either hand in the direction of
power flow. The fingers on the left hand curl in the direction
of the rotating field for left circular polarization, while the
fingers of the right hand curl in the direction of the rotating
field for right circular polarization. Left and right circular
polarizations reverse for waves traveling in the -z direction.

7-4-7 Wave Propagation in Anisotropic Media

Many properties of plane waves have particular appli-

cations to optics. Because visible light has a wavelength on the
order of 500 nm, even a pencil beam of light 1 mm wide7 is
2000 wavelengths wide and thus approximates a plane wave.

1
 Sinusoidal Time Variations 517

,•wt= 0, 2r
= Exo coswt
= EO coswt

(b)

- )
2

Ex
Wt = r 0

Left circular polarization Right circular polarization

(c)

Figure 7-11

(a) Polarizers
Light is produced by oscillating molecules whether in a
light bulb or by the sun. This natural light is usually
unpolarized as each molecule oscillates in time and direction
independent of its neighbors so that even though the power
flow may be in a single direction the electric field phase
changes randomly with time and the source is said to be
incoherent. Lasers, an acronym for "light amplification by
stimulated emission of radiation," emits coherent light by
having all the oscillating molecules emit in time phase.
A polarizer will only pass those electric field components
aligned with the polarizer's transmission axis so that the
transmitted light is linearly polarized. Polarizers are made of
such crystals as tourmaline, which exhibit dichroism-the
selective absorption of the polarization along a crystal axis.
518 Electrodynamics-Fieldsand Waves

The polarization perpendicular to this axis is transmitted.

Because tourmaline polarizers are expensive, fragile, and
of small size, improved low cost and sturdy sheet polarizers
were developed by embedding long needlelike crystals or
chainlike molecules in a plastic sheet. The electric field
component in the long direction of the molecules or crystals is
strongly absorbed while the perpendicular component of the
electric field is passed.
For an electric field of magnitude Eo at angle 4 to the
transmission axis of a polarizer, the magnitude of the trans-
mitted field is
E, = Eo cos 4 (38)

so that the time-average power flux density is

<S > = I Re [i(r) x A*(r)]1

=- cos' 4 (39)
2 71

which is known as the law of Malus.

(b) Double Refraction (Birefringence)

If a second polarizer, now called the analyzer, is placed
parallel to the first but with its transmission axis at right
angles, as in Figure 7-12, no light is transmitted. The
combination is called a polariscope. However, if an anisotro-
pic crystal is inserted between the polarizer and analyzer,
light is transmitted through the analyzer. In these doubly
refracting crystals, light polarized along the optic axis travels
at speed c1u while light polarized perpendicular to the axis
travels at a slightly different speed c,. The crystal is said to be
birefringent. If linearly polarized light is incident at 450 to the
axis,
E(z = 0, t) = Eo(i, + i,) Re (edw) (40)

the components of electric field along and perpendicular to

the axis travel at different speeds:

E,(z, t) = Eo Re (ei<(ct-hII)), kt = a/c1

E,(z, t) = Eo Re (eijt-h''z), k± = wo/c (41)

After exiting the crystal at z = 1,the total electric field is

E(z = i, t) = Eo Re [ei'(e-iLi,,+ e-i'i,)]

= Eo Re [ei"(-l-')(i, + eihlr-k)'i,)] (42)

 Sinusoidal Time Variations 519

Crossed polarizer
(analyzer)

Incident field at = 0.

at

Elliptically
polarized wave

Complex electric field vector rotates

clockwise along crystal

Doubly refracting
(birefringent) medium

Waves polarized along this

Linearly axis travel at speed cq
polarized
wave

Transmission axis

Polarizer

Figure 7-12 When a linearly polarized wave passes through a doubly refracting
(birefringent) medium at an angle to the crystal axes, the transmitted light is elliptically
polarized.

which is of the form of (31) for an elliptically polarized wave

where the phase difference is

S= (kll- kJ)1 = ol 1 1
cli c-L

When 4 is an integer multiple of 27r, the light exiting the

crystal is the same as if the crystal were not there so that it is
not transmitted through the analyzer. If 45 is an odd integer
multiple of 7r, the exiting light is also linearly polarized but
perpendicularly to the incident light so that it is polarized in
the same direction as the transmission axis of the analyzer,
and thus is transmitted. Such elements are called half-wave
plates at the frequency of operation. When 4 is an odd
integer multiple of ur/2, the exiting light is circularly
 5• 0 Electrodynamics-Fiedr s and Waves

polarized and the crystal serves as a quarter-wave plate.

However, only that polarization of light along the trans-
mission axis of the analyzer is transmitted.
Double refraction occurs naturally in many crystals due to
their anisotropic molecular structure. Many plastics and
glasses that are generally isotropic have induced birefrin-
gence when mechanically stressed. When placed within a
polariscope the photoelastic stress patterns can be seen. Some
liquids, notably nitrobenzene, also become birefringent when
stressed by large electric fields. This phenomena is called the
Kerr effect. Electro-optical measurements allow electric field
mapping in the dielectric between high voltage stressed elec-
trodes, useful in the study of high voltage conduction and
breakdown phenomena. The Kerr effect is also used as a light
switch in high-speed shutters. A parallel plate capacitor is
placed within a polariscope so that in the absence of voltage
no light is transmitted. When the voltage is increased the light
is transmitted, being a maximum when 4 = w. (See problem
17.)

7-5 NORMAL INCIDENCE ONTO A PERFECT CONDUCTOR

A uniform plane wave with x-directed electric field is

normally incident upon a perfectly conducting plane at z = 0,
as shown in Figure 7-13. The presence of the boundary gives
rise to a reflected wave that propagates in the -z direction.
There are no fields within the perfect conductor. The known
incident fields traveling in the +z direction can be written as
Ei(z, t) = Re (Ei eik1t-'i)
(1)

Hi(z, t)= Re (eitm-2)i(1

while the reflected fields propagating in the -z direction are
similarly
E,(z, t) = Re (P, ei"c+Ai.)

+• i, (2)
H,(z, t) = Re ( - 7 ei

where in the lossless free space

710 = v4"o/eo, k = ,"ego (3)
Note the minus sign difference in the spatial exponential
phase factors of (1) and (2) as the waves are traveling in
opposite directions. The amplitude of incident and reflected
magnetic fields are given by the ratio of electric field ampli-
tude to the wave impedance, as derived in Eq. (15) of Section

I
 Normal Incidence onto a Perfect Conductor 521

eo, AO (170=v/-o)

Er = Re(E, e j(r+kX)i. )

H, (z,
H,, t)=
(, 0 ,o j(ot+kz)
= -2E
?0
coskzcoswt -710
H, = Re(-L e
Ex(s,

Figure 7-13 A uniform plane wave normally incident upon a perfect conductor has
zero electric field at the conducting surface thus requiring a reflected wave. The source
of this reflected wave is the surface current at z = 0, which equals the magnetic field
there. The total electric and magnetic fields are 900 out of phase in time and space.

7-3-2. The negative sign in front of the reflected magnetic

field for the wave in the -z direction arises because the power
flow S, = E, x H, in the reflected wave must also be in the -z
direction.
The total electric and magnetic fields are just the sum of
the incident and reflected fields. The only unknown
parameter E, can be evaluated from the boundary condition
at z =0 where the tangential component of E must be
continuous and thus zero along the perfect conductor:
Ei+E,== 0 2=> (4)
The total fields are then the sum of the incident and reflected
fields
E.(z, t)= Ei(z, t) + E,(z, t)
= Re [Ei(e i- ~ -e+3k '
) eic]
= 2E, sin kz sin wt
H,(z, t)= Hi(z, t)+H,(z, t)

(e-jkz + e +L') eiw (5)

= Re

2Ei
= - cos kz cos ot
7lo
522 Electrodynamics-Fiedsand Waves

where we take Ai = Ei to be real. The electric and magnetic

fields are 90* out of phase with each other both in time and
space. We note that the two oppositely traveling wave solu-
tions combined for a standing wave solution. The total solu-
tion does not propagate but is a standing sinusoidal solution
in space whose amplitude varies sinusoidally in time.
A surface current flows on the perfect conductor at z = 0
due to the discontinuity in tangential component of H,
2E,
K, = H,(z= O)=-cos t (6)
11o
giving rise to a force per unit area on the conductor,
F = 2K x oH = p0oH, (z = 0)i = 2eoE? cos 2 Wti, (7)
known as the radiation pressure. The factor of 2 arises in (7)
because the force on a surface current is proportional to the
average value of magnetic field on each side of the interface,
here being zero for z = 0+.

7-6 NORMAL INCIDENCE ONTO A DIELECTRIC

7-6-1 Lossless Dielectric

We replace the perfect conductor with a lossless dielectric

of permittivity e2 and permeability l2, as in Figure 7-14, with
a uniform plane wave normally incident from a medium with
permittivity el and permeability j1. In addition to the
incident and reflected fields for z < 0, there are transmitted
fields which propagate in the +z direction within the medium
for z > 0:
Ei(z, t) = Re [4i ei-A)i,], ki = W ,L
HE(z, t)= Re[ ei(ut-kI')i, , Al=

1 <0
E,(z, t)= Re [- • ei"*+'A)i,]

1(1)

E,(z, t) =Re [E ej"')i.•, k2 = 0 ]

H,(z, t)= Re [E itu*-k+:i], = (12
1

It is necessary in (1) to use the appropriate wavenumber

and impedance within each region. There is no wave travel-
ing in the -z direction in the second region as we assume no
boundaries or sources for z > 0.
 Normal Incidence onto a Dielectric 523

Selp e 2 , P2 (2 C2
VEIA E2

E i = Re(Eie i)(

ki = kli
j( 8-
Et = Re(Et ei k2s) is)
j(e t-kx i)
H -i y
I -i)e

k, = k2 i
2 = is

Er = Re( re i (Q'I+kIdi)
H =- Re(-t- e hjt-k2ziy)

=
Hr Re(-- e-lRt )

kr.= -ki is is

Figure 7-14 A uniform plane wave normally incident upon a dielectric interface
separating two different materials has part of its power reflected and part transmitted.

The unknown quantities E, and E, can be found from the

boundary conditions of continuity of tangential E and H at
z = 0,

1 r2

from which we find the reflection R and transmission T field

coefficients as

R=-=-
E. ml+?h
Ei 712+711

E, 2712
T= -X=
Ei 72+ 2 11
E, 112+111
where from (2)
1+R=T

If both mediums have the same wave impedance, II1 = 12,

there is no reflected wave.
 524 Electrodynamics-Fieldsand Waves

7-6-2 Time-Average Power Flow

The time-average power flow in the region z <0 is

<S&>= 2 Re [ .(z)A (z)]

= •-
Re [E ,h '"
+ ,e+i*'"][E
1 e+ik' - E e-ihI

[1 '1

2
+-Re [PA* ee+ iz- P ie-20k 1z
27 a (5)

The last term on the right-hand side of (5) is zero as it is the

difference between a number and its complex conjugate,
which is pure imaginary and equals 2j times its imaginary
part. Being pure imaginary, its real part is zero. Thus the
time-average power flow just equals the difference in the
power flows in the incident and reflected waves as found
more generally in Section 7-3-2. The coupling terms between
oppositely traveling waves have no time-average yielding the
simple superposition of time-average powers:

<s, > = 2[I •1-il 2-111]

= [1 -R 2] (6)
2n,
This net time-average power flows into the dielectric
medium, as it also equals the transmitted power;
2 2 2
I-'= 1I1 T
1I 1
<S,> = 2712 2n2 271
[I-R'] (7)

7-6-3 Lossy Dielectric

If medium 2 is lossy with Ohmic conductivity o, the solu-

tions of (3) are still correct if we replace the permittivity 62 by
the complex permittivity S,.

2=e 1+(8)

so that the wave impedance in region 2 is complex:

'12 = 'IA/

I ____
 Normal Incidence onto a Dielectric 525

We can easily explore the effect of losses in the low and large
loss limits.

(a) Low Losses

If the Ohmic conductivity is small, we can neglect it in all
terms except in the wavenumber k2 :

lim k2-1E -2 (10)

/W624C1 2 82

The imaginary part of k 2 gives rise to a small rate of

exponential decay in medium 2 as the wave propagates away
from the z = 0 boundary.

(b) Large Losses

For large conductivities so that the displacement current is
negligible in medium 2, the wavenumber and impedance in
region 2 are complex:

(k= 1-ij Q 2
lim (11)
o,_ ,*'2 1 +1

The fields decay within a characteristic distance equal to the

skin depth 8. This is why communications to submerged
submarines are difficult. For seawater, ~2 = 0=
41rX 10-7 henry/m and o-4 siemens/m so that for 1 MHz
signals, 8-0.25m. However, at 100Hz the skin depth
increases to 25 meters. If a submarine is within this distance
from the surface, it can receive the signals. However, it is
difficult to transmit these low frequencies because of the large
free space wavelength, A-3 106 m. Note that as the
conductivity approaches infinity,

- I
lim (12)
-( 1o12 = 0 (T=0

so that the field solution approaches that of normal incidence

upon a perfect conductor found in Section 7-5.

EXAMPLE 7-1 DIELECTRIC COATING

A thin lossless dielectric with permittivity e and permeabil-

ity M is coated onto the interface between two infinite half-
spaces of lossless media with respective properties (E , p1r) and
(e6, Ip), as shown in Figure 7-15. What coating parameters e
and Ct and thickness d will allow all the time-average power
526 Electrodynamics-Fieldsand Waves

= I_-AI

E2

Y' "e- E2 k2
d

H1 H2

Region 1 Region 2

No reflections
if d -, n = 1,3,5...
4
and 7 = V71
1-2 ,where = 2•r is

measured within the coating

Figure 7-15 A suitable dielectric coating applied on the interface of discontinuity

between differing media can eliminate reflections at a given frequency.

from region 1 to be transmitted through the coating to region

2? Such coatings are applied to optical components such as
lenses to minimize unwanted reflections and to maximize the
transmitted light intensity.

SOLUTION

For all the incident power to be transmitted into region 2,

there can be no reflected field in region 1, although we do
have oppositely traveling waves in the coating due to the
reflection at the second interface. Region 2 only has positively
z-directed power flow. The fields in each region are thus of
the following form:
Region 1
Ex= Re [E 1 ei("-k&)ix],

HI=Re Ee -k,>i, , ki =
Si
 Normal Incidence onto a Dielectric 527

Coating
E+ = Re [E+eit""-a'i,], k = o/c = wE1

H+= Re [E, e' t"+ni,] 1

EH- = Re ei(+')i,]
e[E

Region 2
E2 = Re [P 2 ei(kt-k)ix], k2 = ICw(62/
=

H2 = Re [E2 e••M )i, , *92

Continuity of tangential E and H at z = 0 and z = d requires

1 =t++L-, E, E+-E-

' + - e+ iu = E2 e-isd
P+e-i
P+ e-"d -- e+i e-ikgd
71 712

Each of these amplitudes in terms of E 1 is then

E.= - 1+-

~ = ej'gdE+e- +L~ e+iv ]

=12 ei d[t+ e-iud

- _e+id]
77

Solving this last relation self-consistently requires that

.+e-'( 1- +L e" 1+ =0

Writing ~+and .. in terms of t 1 yields

(I+I)i I 72)+e2ji1 +1) (1 _j =0

Since this relation is complex, the real and imaginary parts

must separately be satisfied. For the imaginary part to be zero
requires that the coating thickness d be an integral number of
 528 Electrodynamics--Fieldsand Waves

quarter wavelengths as measured within the coating,

2kd = nar d = nA/4, n = 1, 2,3,...

The real part then requires

1+1+ 1 n even
) ]1 n odd
For the upper sign where d is a multiple of half-wavelengths
the only solution is

P12=-1 (d=nA/4, n=2,4,6,...)

which requires that media 1 and 2 be the same so that the

coating serves no purpose. If regions 1 and 2 have differing
wave impedances, we must use the lower sign where d is an
odd integer number of quarter wavelengths so that

712#1
2 ==1=
fq2 (d=nA/4, n =1,3,5,...)
Thus, if the coating is a quarter wavelength thick as measured
within the coating, or any odd integer multiple of this thick-
ness with its wave impedance equal to the geometrical average
of the impedances in each adjacent region, all the time-
average power flow in region 1 passes through the coating
into region 2:

<S,> . . .
2 , 2
712

( e +e+'hz-
(*E e'
-'•
=2Re (, e-" +- e+·)

271

Note that for a given coating thickness d, there is no reflection

only at select frequencies corresponding to wavelengths d =
nA/4, n = 1,3,5,.... For a narrow band of wavelengths
about these select wavelengths, reflections are small. The
magnetic permeability of coatings and of the glass used in
optical components are usually that of free space while the
permittivities differ. The permittivity of the coating e is then
picked so that

and with a thickness corresponding to the central range of the

wavelengths of interest (often in the visible).

I _
 Uniform and Nonuniform Plane Waves 529

7-7 UNIFORM AND NONUNIFORM PLANE WAVES

Our analysis thus far has been limited to waves propagating

in the z direction normally incident upon plane interfaces.
Although our examples had the electric field polarized in the
x direction., the solution procedure is the same for the y-
directed electric field polarization as both polarizations are
parallel to the interfaces of discontinuity.

7-7-1 Propagation at an Arbitrary Angle

We now consider a uniform plane wave with power flow at

an angle 0 to the z axis, as shown in Figure 7-16. The electric
field is assumed to be y directed, but the magnetic field that is
perpendicular to both E and S now has components in the x
and z directions.
The direction of the power flow, which we can call z', is
related to the Cartesian coordinates as
z'=x sin O+z cos 0
so that the phase factor kz' can be written as
kz' = kx + k,z, k, = k sin 0
k, = k cos 0

where the wavenumber magnitude is

scos0
0 i,+ sin Oi l

E = Re(Ee

Figure 7-16 The spatial dependence of a uniforiN plane wave at an arbitrary angle 0
can be expressed in terms of a vector wavenumber k as e-ik' ,where k is in the direction
of power flow and has magnitude co/c.
530 Electrodynamics-Fieldsand Waves

This allows us to write the fields as

E = Re [ ee•t-1kx-h"z•i,]
A (4)
H= Re [E(-cos Oi, +sin Oi,) ei''-t--."-h.4

We note that the spatial dependence of the fields can be

written as e-i'kL ,where the wavenumber is treated as a vector:
k = ki + ki +ki, (5)
with
r = xix +yi, +zi, (6)
so that
kr=
r=+k,y
+ky +kz (7)
The magnitude of k is as given in (3) and its direction is the
same as the power flowS:

IAI2
S= ExH = - (cos Oi, +sin cos2 (wt -k - r)
ix,)

1EI 2k
- - cos 2 (wt - k r) (8)

where without loss of generality we picked the phase of f to

be zero so that it is real.

7-7-2 The Complex Propagation Constant

Let us generalize further by considering fields of the form

' - '
E = Re [E e"' e- '
]= Re [E e( tkr) e-"I]
'- ''] - k
H = Re [Hi e' e = Re [i e(t r) e-"a] (9)
where y is the complex propagation vector and r is the posi-
tion vector of (6):
- = a + jk = y,i, + yi, + y i, (10)
Y' r = yx + y,y + y,z
We have previously considered uniform plane waves in
lossless media where the wavenumber k is pure real and z
directed with a =0 so that y is pure imaginary. The
parameter a represents the decay rate of the fields even
though the medium is lossless. If a is nonzero, the solutions
are called nonuniform plane waves. We saw this decay in our
quasi-static solutions of Laplace's equation where solutions
had oscillations in one direction but decay in the perpendic-
ular direction. We would expect this evanescence to remain at
low frequencies.
 Uniform and Nonunifonm Plane Waves 531

The value of the assumed form of solutions in (9) is that the

del (V) operator in Maxwell's equations can be replaced by the
vector operator -y:

V= -i, -+-i, -iz

ax ay az

= -Y (11)

This is true because any spatial derivatives only operate on

the exponential term in (9). Then the source free Maxwell's
equations can be written in terms of the complex amplitudes
as

- x fi =ij•- (12)

-Y -,tI= 0

The last two relations tell us that y is perpendicular to both

E and H. If we take y x the top equation and use the second
equation, we have
-7 x (y x •) = -jot (y X HI)= -jay (-jweE)
-_= ACet (13)

The double cross product can be expanded as

-- x(,y X f) = -y(y •I) + (,y y)i
= (,y ./)j = -_oWCE (14)

The y --i term is zero from the third relation in (12). The
dispersion relation is then

y*y= (ar- k +2j k)= -W9Ie (15)

For solution, the real and imaginary parts of (15) must be

separately equal:
'2 - k2 = --- P_ E (16)
at k=0

When a= 0, (16) reduces to the familiar frequency-

wavenumber relation of Section 7-3-4.
The last relation now tells us that evanescence (decay) in
space as represented by a is allowed by Maxwell's equations,
but must be perpendicular to propagation represented by k.
532 Electrodynamics-Fieldsand Waves

We can compute the time-average power flow for fields of

the form of (9) using (12) in terms of either E or H as follows:

<S > = - Re ( x ii*) (17)

- JRe(r
-'.Re (y*) x fi*) j
k
2I WE ymp
<S> *)
write the (powerflow
to in terms of either E or(17)
Re jWR

k ( 9 1 H
(I_*)
ReH| - `9Re

Although both final expressions in (17) are equivalent, it is

convenient to write the power flow in terms of either E or H.
When E is perpendicular to both the real vectors a and 0,
defined in (10) and (16), the dot product y* • E is zero. Such a
mode is called transverse electric (TE), and we see in (17) that
the time-average power flow is still in the direction of the
wavenumber k. Similarly, when H is perpendicular to a and
13, the dot product y H* is zero and we have a transverse
magnetic (TM) mode. Again, the time-average power flow in
(17) is in the direction of k. The magnitude of k is related to w
in (16).
Note that our discussion has been limited to lossless
systems. We can include Ohmic losses if we replace E by the
complex permittivity E of Section 7-4-3 in (15) and (17).
Then, there is always decay (a 4 0) because of Ohmic dis-
sipation (see Problem 22).

7-7-3 Nonuniform Plane Waves

We can examine nonuniform plane wave solutions with

nonzero a by considering a current sheet in the z = 0 plane,
which is a traveling wave in the x direction:

K,(z = 0) = Ko cos (wt - kx) = Re (Koei"j- ' ='

kx))

_I··I
 Uniform and Nonuniform Plane Waves 533

The x-directed surface current gives rise to a y-directed

magnetic field. Because the system does not depend on the y
coordinate, solutions are thus of the following form:
'
"), z > 0O
H,=,=Re (H1e e
- "' ), z<0
Re (H eJ 2 e
A. (19)
--
Re 7X Ho i, e ir"e' , z>O

Re -Y2 iie 'W

e"'r], z<0

where ,yand y2 are the complex propagation vectors on each

side of the current sheet:
yi = yliix + yiz
(20)
7Y2 = Y2A x + Y2 (0
The boundary condition of the discontinuity of tangential H
at z = 0 equaling the surface current yields
-
-AI e~~,X + A2 e- ^2- = Ko e-ik
"
(21)
which tells us that the x components of the complex prop-
agation vectors equal the trigonometric spatial dependence of
the surface current:
Y.l = 2x. = jk. (22)
The z components ofy'1 and y2 are then determined from (15)
as
2 2 21
yx + Y2 =-••2 = + -•
-We)1/•2 (23)
2
If k2 <Co e, y,is pure imaginary representing propagation
and we have uniform plane waves. If k2 >W 2e6/, then y, is
pure real representing evanescence in the z direction so that
we generate nonuniform plane waves. When w = 0, (23) cor-
responds to Laplacian solutions that oscillate in the x direc-
tion but decay in the z direction.
The z component of y is of opposite sign in each region,
Yl = --Y2 •= +(k -_W2ey)1/2 (24)
as the waves propagate or decay away from the sheet.
Continuity of the tangential component of E requires
YlI = H22 H 2 = -H = Ko/2 (25)
If k.= 0, we re-obtain the solution of Section 7-4-1.
Increasing k. generates propagating waves with power flow in
the ki,+ ksi. directions. At k = w 2 , kz = 0so that the power
flow is purely x directed with no spatial dependence on z.
Further increasing k. converts k,to a, as y, becomes real and
the fields decay with z.
 534 Electrodynamics-Fieldsand Waves

7-8 OBLIQUE INCIDENCE ONTO A PERFECT CONDUCTOR

7-8-1 E Field Parallel to the Interface

In Figure 7-17a we show a uniform plane wave incident

upon a perfect conductor with power flow at an angle 0i to
the normal. The electric field is parallel to the surface with
the magnetic field having both x and z components:
Ei = Re [Ei ei('tL'-.x-kz)i,]
j( - . -k. Z)
Hi=Re [ (-cos Oii. +sin Oii.) eit -

where
k,i = k sin Oi
k=<
0 eA, T1 1
ki = k cos Oi

0= -

H,
a=-
E,
)P~kor

OH

E,

Hj

(b)
Figure 7-17 A uniform plane wave obliquely incident upon a perfect conductor has its
angle of incidence equal to the angle of reflection. (a) Electric field polarized parallel to
the interface. (b) Magnetic field parallel to the interface.

I
 Oblique Incidence onto a Perfect Conductor 535

There are no transmitted fields within the perfect conductor,

but there is a reflected field with power flow at angle 0,from
the interface normal. The reflected electric field is also in the
y direction so the magnetic field, which must be perpendic-
ular to both E and S= E x H, is in the direction shown in
Figure 7-17a:
E, = Re [E, ei("-"k-,+",z)i,]

H, = Re [-(cos O,i, + sin 0,i,) eit" - , + ',

where the reflected wavenumbers are

kx,= k sin 0,
k,,=k cos 0,(4)
At this point we do not know the angle of reflection 0,or
the reflected amplitude E,.They will be determined from the
boundary conditions at z = 0 of continuity of tangential E and
normal B. Because there are no fields within the perfect
conductor these boundary conditions at z = 0 are
4 e --'z + 4,e-ir"= 0
(5)
i'" -
-(Ei sin Oi e- +E sin 0,e "'') = 0

These conditions must be true for every value of x along z = 0

so that the phase factors given in (2) and (4) must be equal,
= ký,,= O = 0,= 0
kx. (6)
giving the well-known rule that the angle of incidenceequals the
angle of reflection. The reflected field amplitude is then
t = -i (7)
with the boundary conditions in (5) being redundant as they
both yield (7). The total fields are then:
' i( - ]
E,= Re [Ei(e -ik' -- e+ik ) e a kx)
= 2Ej sin k,z sin (wt - kx)

H=Re E[cos O(-e-j '-e+k-')i,+sin O(e-'

-e +jk.)i] ej(-t-k.x] (8)

= 2E[-cos 0cos kAz cos (wt- kx)i,

+ sin 0 sin k,z sin (wt- kAx)i

where without loss of generality we take ei to be real.
536 Electrodynamics-Fieldsand Waves

We drop the i and r subscripts on the wavenumbers and

angles because they are equal. The fields travel in the x
direction parallel to the interface, but are stationary in the z
direction. Note that another perfectly conducting plane can
be placed at distances d to the left of the interface at
k,d = nir (9)
where the electric field is already zero without disturbing the
solutions of (8). The boundary conditions at the second
conductor are automatically satisfied. Such a structure is called
a waveguide and is discussed in Section 8-6.
Because the tangential component of H is discontinuous at
z = 0,a traveling wave surface current flows along the inter-
face,
2E,
K,= -H,(z = 0) = -cos cos (wt - kx) (10)

From (8) we compute the time-average power flow as

<S > = 1 Re [E(x, z) x I*(x, z)]
2E'
= E2-sin 0 sin kzi, (11)

We see that the only nonzero power flow is in the direction

parallel to the interfacial boundary and it varies as a function
of z.

7-8-2 H Field Parallel to the Interface

If the H field is parallel to the conducting boundary, as in

Figure 7-17b, the incident and reflected fields are as follows:
' ' -
Ei = Re [Ei (cos Oii, -sin 0ii,)
ei(t~- 4z)]
k=

E, = Re [E, (-cos Ori, -sin O1,i) . - x

ei (t-h k'
)] (12)

H, = Re e

The tangential component of E is continuous and thus zero

at z = 0:
AEcos e-ik' - o 0,e-i ,"= 0
c0
cos (13)

There is no normal component of B. This boundary condi-

tion must be satisfied for all values of x so again the angle of

__I
 Oblique Incidence onto a Perfect Conductor 537

incidence must equal the angle of reflection (Oi = 0,) so that

E£ = P, (14)

The total E and H fields can be obtained from (12) by adding

the incident and reflected fields and taking the real part;

E = Re {ti [cos 0(e - ij' - e+ijh")ix

i' "
-sin 0(e - kz + e+jk")i,] eij(W -kX

= 2E {cos 0 sin kz sin (wt - kx)i,

(15)
- sin 0 cos kz cos (wt - k~)i,}

H= Re (eikz e+jhz) ej(.t -k.x)

2E,
=- E cos kzz cos (wtot - kxx)i,

The surface current on the conducting surface at z = 0 is

given by the tangential component of H

2E,
K.(z = 0) = H,(z = 0) = - cos (ot- kx) (16)

while the surface charge at z = 0 is proportional to the normal

component of electric field,

tr,(z = 0) = -eE(z = 0) = 2eEi sin 0 cos (wt - k~x) (17)

Note that (16) and (17) satisfy conservation of current on the

conducting surface,

V" K + =
=0• + 0 (18)
at ax at
where

Vx =- i + i,
Ox ay

is the surface divergence operator. The time-average power

flow for this polarization is also x directed:

<S> = 1 Re (E x AI*)
2 2
= • sin 0 cos2 k,zi, (19)
71
5.8 Electrodynamics-Fieldsand Waves

7-9 OBLIQUE INCIDENCE ONTO A DIELECTRIC

7-9-1 E Parallel to the Interface

A plane wave incident upon a dielectric interface, as in

Figure 7-18a, now has transmitted fields as well as reflected
fields. For the electric field polarized parallel to the interface,
the fields in each region can be expressed as

Ei = Re [E, ei "'" z'i,]

•- -, ]
Hi = Re [(-cos 0i. +sin Oi,) ei Aa s

E, = Re [E( ei(,-.x+k-) i,l

H, = Re [E(cos ,Pi,+sin Oi.)) e

i (• -' + ] ()

E, = Re [E1 e •k, k, ,]Z

i - k.,=- ,z)
2) t
H= Re [.(-cos O,i+sinOie A.

where 8i, 0,, and 0, are the angles from the normal of the
incident, reflected, and transmitted power flows. The
wavenumbers in each region are

k• = kAsin 0i, kx,= k1 sin 0, , =,,

k2 sin 0,
(2)
k = k cos 8, k cos 0,, k, = k 2 cos 0,
where the wavenumber magnitudes, wave speeds, and wave
impedances are

ki k2 CI •
7 1=- 1 2 =-,
2 E 1(3)C =
= c= 1-
1I'1 , 2a= ,

The unknown angles and amplitudes in (1) are found from

the boundary conditions of continuity of tangential E and H
at the z = 0 interface.
ei -i.k-i + re- L =4,e - "

- i cos 0 i e -j'kix + E, cos Or e -jkS , cos 0, e -ikr,,

(4)

These boundary conditions must be satisfied point by point

for all x. This requires that the exponential factors also be
 Oblique Incidence onto a Dielectric 539

E2 - A2

1 S 1

C2 , U2
1E 1,

.q
- u tr - C1 .. i

Figure 7-18 A uniform plane wave obliquely incident upon a dielectric interface also
has its angle of incidence equal to the angle of reflection while the transmitted angle is
given by Snell's law. (a) Electric field polarized parallel to the interface. (b) Magnetic
field parallel to the interface.

equal so that the x components of all wavenumbers must be

equal,

k.i = k., = kR, > kl sin Oi = ki sin 0, = k2 sin 0,

which relates the angles as

0, = 8,

sin 01 = (c2/ci) sin Oi

540 Electrodynamics-Fieldsand Waves

As before, the angle of incidence equals the angle of

reflection. The transmission angle obeys a more complicated
relation called Snell's law relating the sines of the angles. The
angle from the normal is largest in that region which has the
faster speed of electromagnetic waves.
In optics, the ratio of the speed of light in vacuum, co =
1/ ,e-oo, to the speed of light in the medium is defined as the
index of refraction,
ni = co/c, n 2 = Co/IC (8)
which is never less than unity. Then Snell's law is written as
sin 0, = (n1 /n 2 ) sin Oi (9)
With the angles related as in (6), the reflected and transmitted
field amplitudes can be expressed in the same way as for
normal incidence (see Section 7-6-1) if we replace the wave
impedances by 71 -* 17/cos 0 to yield

712 711
E, cos 0, cos 0i 12 os O - 711 cos 0
Ei 712 11i co
+12 s
+i0+cosOt
cos 6, cos 0,
(10)
S2112 2 O cosi
1 0
cos o ( 72+ . '2cos 0i+lcos
cos 0, +
cos 0, cos Os
In (4) we did not consider the boundary condition of
continuity of normal B at z = 0. This boundary condition is
redundant as it is the same condition as the upper equation in
(4):
-'(Pi+4r) sin 0i = L-4 sin 0, > (1i + r) = (11)
711 712
where we use the relation between angles in (6). Since

711 c1 712 c2

the trigonometric terms in (11) cancel due to Snell's law.

There is no normal component of D so it is automatically
continuous across the interface.

7-9-2 Brewster's Angle of No Reflection

We see from (10) that at a certain angle of incidence, there

is no reflected field as R = 0. This angle is called Brewster's
angle:
R = 0='712 cos 0i = 71 cos Ot
 Oblique Incidence onto a Dielectric 541

By squaring (13), replacing the cosine terms with sine terms

(cos2 0 = 1- sin' 0), and using Snell's law of (6), the Brewster
angle On is found as

sin2 OB -E 2 /(EL 2 ) (14)

1 -(_O/•s)2
There is not always a real solution to (14) as it depends on the
material constants. The common dielectric case, where 1~1 =
P,2 - j but I # e2, does not have a solution as the right-hand
side of (14) becomes infinite. Real solutions to (14) require the
right-hand side to be between zero and one. A Brewster's
angle does exist for the uncommon situation where e1 = E2
and P 1 #I 2:

sin 2 B= 1 tan O8 = (15)

1+A/II 2 A1

At this Brewster's angle, the reflected and transmitted power

flows are at right angles (On + 0, = ir/2) as can be seen by using
(6), (13), and (14):
cos (On + 80)= cos OB cos 0, - sin On sin 0,

= cos 2 sin22 On A
2A
-1 + A22

= - sin2 e~(J + = (16)

7-9-3 Critical Angle of Transmission

Snell's law in (6) shows us that if c2 >CI, large angles of

incident angle Oi could result in sin 0, being greater than
unity. There is no real angle 0, that satisfies this condition.
The critical incident angle 0c is defined as that value of Oi that
makes 0, = ir/2,
sin 0c= C1/c2 (17)

which has a real solution only if cI <c2. At the critical angle,

the wavenumber k., is zero. Lesser incident angles have real
values of k,. For larger incident angles there is no real angle 0,
that satisfies (6). Snell's law must always be obeyed in order to
satisfy the boundary conditions at z =0 for all x. What
happens is that 0, becomes a complex number that satisfies
(6). Although sin 0, is still real, cos 0, is imaginary when sin 0,
exceeds unity:

cos 0, = 41-sin 0,
542 Elecrodynamics-Fieldsand Waves

This then makes k,, imaginary, which we can write as

ka, = k2 cos 0, = -ja (19)

The negative sign of the square root is taken so that waves

now decay with z:
E, = Re t ei[-( . ,"e-i(,])
(20)
= ~
H, = Re [(-cos Oi +sin
+ ,in) ei("*-,
x e-a

The solutions are now nonuniform plane waves, as discussed

in Section 7-7.
Complex angles of transmission are a valid mathematical
concept. What has happened is that in (1) we wrote our
assumed solutions for the transmitted fields in terms of pure
propagating waves. Maxwell's equations for an incident angle
greater than the critical angle require spatially decaying
waves with z in region 2 so that the mathematics forced k=. to
be imaginary.
There is no power dissipation since the z-directed time-
average power flow is zero,
<S,> = -I Re [E,H]

- Re (-cos 0,)* e-I=- (21)

because cos 0, is pure imaginary so that the bracketed term in

(21) is pure imaginary. The incident z-directed time-average
power is totally reflected. Even though the time-averaged
z-directed transmitted power is zero, there are nonzero but
exponentially decaying fields in region 2.

7-9-4 H Field Parallel to the Boundary

For this polarization, illustrated in Figure 7-18b, the fields

are
Ej = Re [Ei (cos O8i. -sin Oii.) e i(t-k.Xk
-k
)]

Hi = Re [ L e
i i(L-k.-hi,]

E, = Re [E, (-cos ,i. -sin O,i,) ei (' * - '

.,x+k,)]
(22)
:--~
H, = Re [Leit +k')i ,

E, = Re [tE (cos 0,ix -sin 0,i,) eit(m' -k x,- ~'

)]

H,= Re [L eiY-k.,=-.,Ci]
7L2
 Oblique Incidence onto a Dielectric 543

where the wavenumbers and impedances are the same as in

(2) and (3).
Continuity of tangential E and H at z = 0 requires
Ei cos 0i e-"*-* -~, cos 0, e-i"-' =, cos 0, e-"--'
4, e-'ix"+4, e-i.,x 4, e-i'x (23)

Again the phase factors must be equal so that (5) and (6) are
again true. Snell's law and the angle of incidence equalling
the angle of reflection are independent of polarization.
We solve (23) for the field reflection and transmission
coefficients as
E, nl cos Oi - 12 COS 0,
R = -= (24)
Ei 72 cos , a COS 0
cos

,
T =--=
2712 cos OG
(25)
Ei '/2 COs Ot + ~ cos 0i
Now we note that the boundary condition of continuity of
normal D at z = 0 is redundant to the lower relation in (23),
EIEisin O9+EI, sin 0, = E2 E, sin 0, (26)

using Snell's law to relate the angles.

For this polarization the condition for no reflected waves is
R = 0> 7q2 cos O1 = rl cos Oi (27)

which from Snell's law gives the Brewster angle:

I- e sp2/(e2/z,)
sin2 On = 1(21L1) (28)
1-(e /E2)
There is now a solution for the usual case where /A. =
=2 but
El # E2:

sin 2 OB = 1 tan O = (29)

l+EII/2 81

At this Brewster's angle the reflected and transmitted power

flows are at right angles (OB + 0,) = r/2 as can be seen by using
(6), (27), and (29)
cos (OB + 0,) = cos OB cos 0, - sin OB sin 0,

= cos2 OG -lsin' eG
&1 E2

= j -sin 2 0. (V + r•)= 0 (30)

"el e•
E2
544 Electrodynamics--Fieldsand Waves

Because Snell's law is independent of polarization, the

critical angle of (17) is the same for both polarizations. Note
that the Brewster's angle for either polarization, if it exists, is
always less than the critical angle of (17), as can be particularly
seen when A =-L2 for the magnetic field polarized parallel to
the interface or when 81 = e2 for the electric field polarized
parallel to the interface, as then
1 1
=i + 1 (31)
sin eB sin O+

7-10 APPLICATIONS TO OPTICS

Reflection and refraction of electromagnetic waves

obliquely incident upon the interface between dissimilar
linear lossless media are governed by the two rules illustrated
in Figure 7-19:

(i) The angle of incidence equals the angle of reflection.

(ii) Waves incident from a medium of high light velocity
(low index of refraction) to one of low velocity (high
index of refraction) are bent towards the normal. If the
wave is incident from a low velocity (high index) to high
velocity (low index) medium, the light is bent away from
the normal. The incident and refracted angles are
related by Snell's law.

El

1:

Figure 7-19 A summary of reflection and refraction phenomena across the interface
separating two linear media. When 90=-0 (Brewster's angle), there is no reflected ray.
When 0, > 0, (critical angle), the transmitted fields decay with z.
 Applications to Optics 545

Most optical materials, like glass, have a permeability of

fr~ee space o0 . Therefore, a Brewster's angle of no reflection
only exists if the H field is parallel to the boundary.
At the critical angle, which can only exist if light travels
from a high index of refraction material (low light velocity) to
one of low index (high light velocity), there is a transmitted
field that decays with distance as a nonuniform plane wave.
However, there is no time-average power carried by this
evanescent wave so that all the time-average power is
reflected. This section briefly describes various applications of
these special angles and the rules governing reflection and
refraction.

7-10-1 Reflections from a Mirror

A person has their eyes at height h above their feet and a

height Ah below the top of their head, as in Figure 7-20. A
mirror in front extends a distance Ay above the eyes and a
distance y below. How large must y and Ay be so that the
person sees their entire image? The light reflected off the
person into the mirror must be reflected again into the
person's eyes. Since the angle of incidence equals the angle of
reflection, Figure 7-20 shows that Ay = Ah/2 and y = h/2.

7-10-2 Lateral Displacement of a Light Ray

A light ray is incident from free space upon a transparent

medium with index of refraction n at angle 0,, as shown in
Figure 7-21. The angle of the transmitted light is given by
Snell's law:
sin 0, = (1/n) sin Oi (1)

Ah
Ah Ay=- 2

ror

Figure 7-20 Because the angle of incidence equals the angle of reflection, a person can
see their entire image if the mirror extends half the distance of extent above and below
the eyes.
 546 Electrodynamics-Fieldsand Waves

-' dsin(6i-6,)
&A.- lI COSU0r

6
62 Bi

-<--d-->-

Figure 7-21 A light ray incident upon a glass plate exits the plate into the original
medium parallel to its original trajectory but laterally displaced.

When this light hits the second interface, the angle 0, is now
the incident angle so that the transmitted angle 0 2 is again
given by Snell's law:

sin 02 = n Sin 0, = sin Oi (2)

so that the light exits at the original incident angle Oi.

However, it is now shifted by the amount:

d sin (0i - 0,)

cos 0,

If the plate is glass with refractive index n = 1.5 and thickness

d = 1 mm with incident angle Oi = 30*, the angle 0, in the glass
is
sin 0,= 0.33= 0,= 19.50 (4)

so that the lateral displacement is s = 0.19 mm.

7-10-3 Polirization By Reflection

Unpolarized light is incident upon the piece of glass in

Section 7-10-2 with index of refraction n = 1.5. Unpolarized
light has both E and H parallel to the interface. We assume
that the permeability of the glass equals that of free space and
that the light is incident at the Brewster's angle OB for light
polarized with H parallel to the interface. The incident and

I
 Applications to Optics 547

transmitted angles are then

tan Os = EIE =n 0 = 56. 3 °
tan 0, = IEo/e = 1/n = 0, = 33.70 (5)

The Brewster's angle is also called the polarizing angle

because it can be used to separate the two orthogonal
polarizations. The polarization, whose H field is parallel to
the interface, is entirely transmitted at the first interface with
no reflection. The other polarization with electric field
parallel to the interface is partially transmitted and reflected.
At the second (glass-free space) interface the light is incident
at angle 0,. From (5) we see that this angle is the Brewster's
angle with H parallel to the interface for light incident from
the glass side onto the glass-free space interface. Then again,
the H parallel to the interface polarization is entirely trans-
mitted while the E parallel to the interface polarization is
partially reflected and partially transmitted. Thus, the
reflected wave is entirely polarized with electric field parallel
to the interface. The transmitted waves, although composed
of both polarizations, have the larger amplitude with H

S H

. E
zed
Polarized ligi ed
Ilel
(E parallel to inte lel
ce)
:e)

Unpolarized
light
(E and H parallel
to interface)

Figure 7-22 Unpolarized light incident upon glass with A = A-o can be polarized by
reflection if it is incident at the Brewster's angle for the polarization with H parallel to
the interface. The transmitted light becomes more polarized with H parallel to the
interface by adding more parallel glass plates.
 548 Electrodynamics-Fieldsand Waves

parallel to the interface because it was entirely transmitted

with no reflection at both interfaces.
By passing the transmitted light through another parallel
piece of glass, the polarization with electric field parallel to
the interface becomes further diminished because it is par-
tially reflected, while the other polarization is completely
transmitted. With more glass elements, as in Figure 7-22, the
transmitted light can be made essentially completely
polarized with H field parallel to the interface.

7-10-4 Light Propagation In Water

(a) Submerged Source

A light source is a distance d below the surface of water
with refractive index n = 1.33, as in Figure 7-23. The rays
emanate from the source as a cone. Those rays at an angle
from the normal greater than the critical angle,
sin O,= 1/n > 0, = 48.80 (6)
are not transmitted into the air but undergo total internal
reflection. A circle of light with diameter
D = 2d tan Oc - 2.28d (7)
then forms on the water's surface due to the exiting light.

(b) Fish Below a Boat

A fish swims below a circular boat of diameter D, as in
Figure 7-24. As we try to view the fish from the air above, the
incident light ray is bent towards the normal. The region
below the boat that we view from above is demarcated by the
light rays at grazing incidence to the surface (0i = 1r/2) just
entering the water (n = 1.33) at the sides of the boat. The
transmitted angle of these light rays is given from Snell's law
as
sin O 1
sin 0, = sin 1 = - = 48.8" (8)
n n

Figure 7-23 Light rays emanating from a source within a high index of refraction
medium are totally internally reflected from the surface for angles greater than the
critical angle. Lesser angles of incidence are transmitted.

I
 Applications to Optics 549

-
Z
vtzk ýD-
PWWW))~;)~W) HWIYrYlur

D
2
Y=- tanO 1

Figure 7-24 A fish cannot be seen from above if it swims below a circular boat within
the cone bounded by light rays at grazing incidence entering the water at the side of the
boat.

These rays from all sides of the boat intersect at the point a
distance y below the boat, where
D D
tan 0t =-- y= 0.44D
2y 2 tan 0,
If the fish swims within the cone, with vertex at the point y
below the boat, it cannot be viewed from above.

7-10-5 Totally Reflecting Prisms

The glass isoceles right triangle in Figure 7-25 has an index

of refraction of n = 1.5 so that the critical angle for total

- - = [n( n+' I 2z[

22
<Ss,>
Figure 7-25 A totally reflecting prism. The index of refraction n must exceed 2 so
that the light incident on the hypotenuse at 450 exceeds the critical angle.
 550 Electrodynamics-Fieldsand Waves

internal reflection is
1 1
sin oc =- 0c = 41.80 (10)
n 1.5
The light is normally incident on the vertical face of the
prism. The transmission coefficient is then given in Section
7-6-1 as
E, 2n 2/n 2
T =-=-= 0.8 (11)
Ei i7+tjo 1+1/n n+1

where because the permeability of the prism equals that of

free space n = ve/Eo while 1/1o0 = VE••e = 1/n. The transmitted
light is then incident upon the hypotenuse of the prism at an
angle of 450, which exceeds the critical angle so that no power
is transmitted and the light is totally reflected being turned
through a right angle. The light is then normally incident
upon the horizontal face with transmission coefficient:

E2 2/0 2 2n
T2 ------.
= = 1.2 (12)
0.8Ei 7 + o l/n + 1 n+l
The resulting electric field amplitude is then

P2 = TIT 2 E, = 0.96Ei (13)

The ratio of transmitted to incident power density is
<S> 21|Pi21
<S>= 217o,12/7o 24 22 -0.92
1tl 2 (24 (14)
<s1> |/ o(25 1
This ratio can be increased to unity by applying a quarter-
wavelength-thick dielectric coating with index of refraction
ncoating= as developed in Example 7-1. This is not usually
-nh,
done because the ratio in (14) is already large without the
expense of a coating.

7-10-6 Fiber Optics

(a) Straight Light Pipe

Long chin fibers of transparent material can guide light
along a straight path if the light within the pipe is incident
upon the wall at an angle greater than the critical angle
(sin 0, = 1/n):
sin 02 = cos 0, - sin 0~ (15)
The light rays are then totally internally reflected being
confined to the pipe until they exit, as in Figure 7-26. The

I
 Applications to Optics 551

no =1

Figure 7-26 The index of refraction of a straight light pipe must be greater than /2 for
total internal reflections of incident light at any angle.

incident angle is related to the transmitted angle from Snell's

law,
sin 0, = (1/n) sin Oi (16)
so that (15) becomes
cos 0 = -sin = %11-(1/n 2 ) sin" - 1/n (17)
which when solved for n yields
n 2 - 1 +sin2 0i (18)
If this condition is met for grazing incidence (i0 = ar/2), all
incident light will be passed by the pipe, which requires that
n2>-2*n - r2 (19)

Most types of glass have n - 1.5 so that this condition is easily

met.

(b) Bent Fibers

Light can also be guided along a tortuous path if the fiber is
bent, as in the semi-circular pipe shown in Figure 7-27. The
minimum angle to the radial normal for the incident light
shown is at the point A. This angle in terms of the radius of
the bend and the light pipe width must exceed the critical angle
R
sin OA =- sin 0c (20)
R+d
+d

+d

Figure 7-27 Light can be guided along a Circularly bent fiber if R/d > 1/(n - 1) as then
there is always total internal reflection each time the light is incident on the walls.
552 Electrodynamics-Fieldsand Waves

so that
R/d
Rd 11- (21)
R/d + n

which when solved for Rid requires

R 1
-a--
d n-I (22)

PROBLEMS

Section 7-1
1. For the following electric fields in a linear media of
permittivity e and permeability Cj find the charge density,
magnetic field, and current density.
(a) E = Eo(xi. +yi,) sin wt
(b) E = Eo(yi, -xi,) cos wt
(c) E= Re[Eo e" \-•--&)i,]. How must k,, k,, and o be
related so that J = 0?
2. An Ohmic conductor of arbitrary shape has an initial
charge distribution po(r) at t = 0.
(a) What is the charge distribution for all time?
(b) The initial charge distribution is uniform and is
confined between parallel plate electrodes of spacing d. What
are the electric and magnetic fields when the electrodes are
opened or short circuited?
(c) Repeat (b) for coaxial cylindrical electrodes of inner
radius a and outer radius b.
(d) When does a time varying electric field not generate a
magnetic field?
3. (a) For linear media of permittivity e and permeability /,
use the magnetic vector potential A to rewrite Faraday's law
as the curl of a function.
(b) Can a scalar potential function V be defined? What is
the electric field in terms of V and A? The choice of V is not
unique so pick V so that under static conditions E = -V V.
(c) Use the results of (a) and (b) in Ampere's law with
Maxwell's displacement current correction to obtain a single
equation in A and V. (Hint: Vx (Vx A) = V(V - A) -V 2A.)
(d) Since we are free to specify V *A, what value should we
pick to make (c) an equation just in A? This is called setting
the gauge.
(e) Use the results of (a)-(d) in Gauss's law for D to obtain a
single equation in V.