CHAPTER

TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 1
ELECTROMAGNETIC FIELDS AND
PLANE WAVES
2.1 MAXWELLS EQUATIONS
The understanding of any field of physics or electrical engineering requires a suitable
theoretical basis. In optics, we are fortunate that two highly developed and accurate theories
are available. In the older theory - often described as the 'classical theory' - the behaviour of
light is described in terms of electromagnetic fields and waves. This is particularly
appropriate for the analysis of passive devices, where the absorption and emission of
radiation is unimportant, and consequently where the interaction of a wave with matter may
represented in a somewhat phenomenological way.
In the newer theory - the quantum theory - a different model is employed. Light is
considered to be composed of photons, which are the elementary units or quanta of
radiation. The interaction of light and matter is then understood in terms of exchanges of
energy between photons and electrons. For example, the generation of a photon may be
identified with the transition of a single electron between two energy levels. Quantum theory
is therefore directly applicable to active optical devices. The development of this alternative
model, and of its later incarnation, quantum electrodynamics, occupied the first half of the
century, and involved many of the world's foremost physicists.
Quantum theory may also describe situations that do not involve matter at all, but are still
not accurately represented by classical theory. One example is provided by low light levels,
where the photon flux may be extremely small and the arrival of radiation in discrete units is
important. However, for a high enough photon flux, the two theories are equivalent. We
shall encounter both of them in this book. Since our early discussions will centre on passive
devices, we shall begin with a classical approach, turning only to the quantum theory at a
much later stage (Chapter 11).
In classical theory, the laws of electricity and magnetism are described by Maxwell's
equations. These represent the result of a synthesis of several existing theories and
experimental observations by James Clerk Maxwell (1831-1879). In effect, Maxwell's
equations are a set of relations linking the values of a number of quantities that describe
electric and magnetic fields. These are the electric flux density D, the magnetic flux
density B, the electric field strength E, the magnetic field strength H, and the current
density J. All are vector quantities, and are functions not only of the three spatial
coordinates x, y and z but also of time t. Of these parameters, D, B, E and H are the most
important to high frequency electromagnetic theory. In this regime, they are essentially
'inventions'; they are not directly observable, but are linked by a self-consistent set of
equations, which correctly predict the magnitudes of other measurable quantities (like the
flow and distribution of power).
Often, different representations are used for the fields; for example, it is common to separate
their time- and space-variation. The electric field strength E may therefore be written
alternatively as:
E(x, y ,z, t) = E(x, y, z) f(t)
2.1-1
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 2
Here E accounts for the spatial variation of the field, and f(t) for temporal changes. Note the
use of a bold-faced type for the complete field E, and lighter type for the time-
independent field E.
It is also common to refer to individual elements of the vectors concerned. In cartesian
coordinates, we may write the time-dependent electric field E as:
E(x, y, z, t) = [ E
x
(x, y, z, t) , E
y
(x, y, z, t) , E
z
(x, y, z, t) ]
2.1-2
Alternatively, we could write for the time-independent field E:
E(x, y, z) = [ E
x
(x, y, z) , E
y
(x, y, z) , E
z
(x, y, z) ]
2.1-3
Here E
x
refers to the x-component of E, and so on.
Because of the nature of the fields, Maxwell's equations are written in terms of vector
calculus. Though this is hard for the novice to begin with, the notation rapidly becomes a
useful (not to say essential) tool-of-the-trade in electromagnetic theory. Unfortunately, for
reasons of space, we must assume here some familiarity with the basic techniques involved.
The equations can be written in two different forms. From experimental observations, by
and large carried out in the previous century, the common integral version was obtained.
This is generally presented in the form of a series of laws. The first pair of these (Gauss'
law, and its magnetic equivalent) describe relations which are most important for static
fields.
1. GAUSS' LAW
Consider a closed surface, surrounding a volume containing electric charges, as shown in
Figure 2.1-1. Gauss' law (named after the German mathematician Karl Gauss, 1777-1855)
states that the electric flux flowing out of the surface is equal to the charge enclosed inside
the volume. This assertion can be written mathematically in terms of the vector fields
involved in Maxwell's theory in the following way.
Figure 2.1-1 Geometry for illustration of Gauss Law
Firstly, we define a vector da as having magnitude equal to the area da of a small element of
the surface, and a direction normal to that element. Secondly, we introduce a new operation,
the scalar or dot product between two vectors F and G. This is written as F . G, and is
defined in cartesian coordinates as:
F . G = F
x
G
x
+ F
y
G
y
+ F
z
G
z
2.1-4
Simple trigonometry may then be used to show that Equation 2.1-4 can also be written as:
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 3
F . G = F G cos()
2.1-5
Where F and G are the moduli or lengths of the two vectors, and is the angle between
them.
Consequently, if D is the electric flux density, the term D . da represents the product of the
component of D normal to the small surface element da and the area of the element. When
integrated over the whole of the surface, this will give the net outward normal electric flux.
Similarly, if we define the scalar term as the charge density, the integral of over the
whole volume must give the charge enclosed. We may therefore write Gauss' law for vector
fields as:

A
D . da =
V
dv
2.1-6
Here the left-hand integral is a surface integral, taken over the whole of the closed surface,
while the right-hand one is a volume integral, over the volume enclosed.
2. MAGNETIC EQUIVALENT OF GAUSS' LAW
We can do the same thing for magnetic flux density. If we now consider the magnetic flux
flowing out of a closed surface, we get a similar picture, shown in Figure 2.1-2.
Figure 2.2 Geometry for illustration of the magnetic equivalent of Gauss Law.
Comparison with Figure 2.1-1 shows that the resulting integral equation must have a similar
form. However, no magnetic monopoles (the magnetic equivalent of electric charges) have
ever been found experimentally, despite extensive searches. Evidence of this curious fact is
provided by the simple bar magnet, which has both north and south poles. However, if the
magnet is divided into two, each half will also have two poles, and no amount of further
subdivision will isolate a monopole. In this case, therefore, the right-hand side of the
equation is zero, giving:

A
B . da = 0
2.1-7
Where B is the magnetic flux density.
The second pair of laws (Faraday's and Ampre's laws) describe relations which are of
greater significance for time-varying fields.
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 4
3. FARADAY'S LAW OF MAGNETIC INDUCTION
Consider a time-varying magnetic flux passing through a closed loop L, defining the rim of
an open surface, as shown in Figure 2.1-3.
Figure 2.1-3 Geometry for illustration of Faradays Law.
Now, the flux of magnetic induction
B
through the open surface is:

B
=
A
B . da
2.1-8
The electromotive force E induced round the loop is therefore:
E = -
B
/t
2.1-9
However, we know from the integral relationship between electric potential and field
strength that this can also be written as a line integral, in the form:
E =
L
E . dL
2.1-10
Here E is the electric field strength, dL is a small element of the closed loop in Figure 2.1-3,
and the integral is taken round the whole loop. By comparing Equations 2.1-9 and 2.1-10,
and using Equation 2.1-8 we obtain Faraday's law, named after Michael Faraday (1791-
1867):

L
E . dL = -
A
B/t . da
2.1-11
Faraday's law implies that a time-varying magnetic field must have an electric field
associated with it. This feature is of great importance for electromagnetic waves, as we shall
see later on.
4. AMPRE'S LAW
Now consider the flow of current through a closed loop, of the same geometry, as shown in
Figure 2.1-4. Ampre's law (named after Andr Ampre, 1775-1836) states that:

L
H . dL =
A
J . da
2.1-12
where H is the magnetic field strength and J is the current density. This implies that moving
charges give rise to a magnetic field, a notion in pleasing symmetry with Faraday's law.
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 5
Figure 2.1-4 Geometry for illustration of Ampres Law.
In addition to these laws, there are a set of relationships known as the material equations,
which link field strengths with flux densities, through a set of material coefficients that are
representative of the bulk properties of matter. There are three of them, written:
J = E
D = E
B = H
2.1-13
Here is the conductivity, is the permittivity, and is the permeability. The physical
origin and significance of some of these quantities will be discussed in Chapter 3. Note that
the first equation is a vectorial form of Ohm's law, commonly encountered in electrical
engineering. For the dielectric materials used to guide high frequency electromagnetic waves
(which we will mainly consider here) the conductivity is typically zero, while the
permeability is that of free space. The former implies that the current density J is zero; there
are also no free charges, so that is zero.
However, this is not quite the end of the story, because experiments showed that magnetic
fields can also be measured in free space - for example, between the plates of a capacitor,
while it is being charged. Consequently, electromagnetic theory was modified by Maxwell
(in what amounts to a stroke of genius) to cope with this observation.
5. GENERALISED FORM OF AMPRE'S LAW
Consider the parallel-plate capacitor shown in Figure 2.1-5, which is linked to a circuit by
wires carrying a current I. Clearly, time-varying currents can somehow travel round the
circuit, despite the absence of conducting material in the region between the plates.
Figure 2.1-5 Geometry for derivation of Maxwells displacement current.
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 6
To account for the flow of current across this apparent break in the circuit, Maxwell
suggested the existence of a new type of current. This is known as the displacement
current, and is calculated as follows. If A is the area of each plate, and Q is the charge on it,
then the electric field E between the plates is:
E = Q/A
2.1-14
As the charge varies, the electric field changes, so that dE/dt = I/A is effectively a current
density. Maxwell therefore defined a vector displacement current density J
D
as:
J
D
= E/t = D/t.
2.1-15
This displacement current must be added into any calculation involving the 'normal'
conduction current. The only example we have come across so far is Ampre's law.
Including the displacement current, Ampre's law must be restated as:

L
H . dL =
A
[J + D/t] . da
2.1-16
This simple modification was of great historical importance, since it showed that time-
varying electric fields can exist even in the absence of normal currents (i.e. when J = 0). It
allowed Maxwell to justify theoretically the electromagnetic waves discovered
experimentally by Heinrich Hertz in 1888, thus demonstrating at a stroke that static electric
and magnetic fields, and radio and light waves, are all part of the wider phenomenon of
electromagnetism.
2.2 THE DIFFERENTIAL FORM OF MAXWELL'S
EQUATIONS
We can use two standard vector theorems to transform the integral equations above into an
alternative differential form, which will be of great use in later calculations. The theorems are
described in detail in many mathematics texts, and so will simply be stated here.
1. GAUSS' THEOREM
The first is Gauss' theorem (not to be confused with Gauss' law), which states that for any
vector field F the following relation holds:

A
F . da =
V
div(F) dv
2.2-1
Here the divergence of a vector field F is an important new quantity, a scalar term, which is
defined in cartesian coordinates by:
div(F) = F
x
/x + F
y
/y + F
z
/z
2.2-2
div(F) is also often written as . F, where the vector operator (or 'del') is defined in
cartesian coordinates as:
= i /x + j /y + k /z
2.2-3
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 7
Here i, j and k are unit vectors in the x-, y- and z-directions, respectively. Hence, Gauss' law
becomes:

A
D . da =
V
div(D) dv =
V
dv
2.2-4
By examining the latter part of Equation 2.2-4, we see that an integral-free relation between
D and can be obtained, in the form:
div(D) =
2.2-5
Similarly, we can show from the magnetic equivalent of Gauss' law that div(B) = 0.
2. STOKES' THEOREM
A similar type of theorem, due to Stokes, states that for any vector field F:

L
F . dL =
A
curl(F) . da
2.2-6
Here we have introduced another new quantity, the curl of a vector field F. This is itself a
vector, and is defined in cartesian coordinates as:
curl(F) = i {F
z
/y - F
y
/z} + j {F
x
/z - F
z
/x} + k {F
y
/x - F
x
/y}
2.2-7
Equation 2.2-7 is a rather long-winded expression. A more compact version (which is also
rather easier to remember) can be written in the form of a determinant, as shown below:
i j k
curl(F) = /x /y /z
F
x
F
y
F
z

2.2-8
Here M represents the determinant of M, where M is a general matrix. Curl(F) is also
often written as x F, where the symbol x denotes a further new operation, the vector
product.
The vector product F x G between two general fields F and G is defined as:
F x G = i { F
y
G
z
- F
z
G
y
} + j { F
z
G
x
- F
x
G
z
} + k { F
x
G
y
- F
y
G
x
}
2.2-9
This can also be written in determinantal form, as:
i j k
F x G = F
x
F
y
F
z

G
x
G
y
G
z

2.2-10
Note that the vector product (unlike the scalar product) is not commutative, so that F x G
G x F. However, with the definition given above, x F reduces to both Equations 2.2-7 and
2.2-8.
Using Stokes' theorem, Faraday's law transforms to:

L
E . dL =
A
curl(E). da = -
A
B/t . da
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 8
2.2-11
Again, examining the latter part of Equation 2.2-11, we see that:
curl(E) = -B/t
2.2-12
Similarly, we can show from Ampre's law that:
curl(H) = J + D/t
2.2-13
We have now derived the differential form of Maxwell's equations. Because they contain
only differential operators, they are often much easier to manipulate than the integral form.
For completeness, we show all the new equations grouped together below:
div(D) = (1)
div(B) = 0 (2)
curl(E) = -B/t (3)
curl(H) = J + D/t (4)
2.2-14
Note that when = 0 and J = 0 (which is the case for electromagnetic waves in dielectric
media) the equations for electric and magnetic quantities appear interchangeable.
2.3 HARMONICALLY-VARYING FIELDS AND THE
WAVE EQUATION
One striking success of Maxwell's equations is to predict the existence of harmonically-
varying fields, otherwise known as electromagnetic waves. We shall now perform a
similar demonstration, with the following assumptions: a) We restrict ourselves to non-
magnetic materials, so that =
0
, where
0
is the permeability of free space (known from
electrostatic experiment to have the value 4 x 10
-7
m kg/C
2
). b) We assume that there are
no currents flowing, and no charges present, so that J and are both zero.
THE WAVE EQUATION
We start by deriving a suitable wave equation. If we put D = E and B =
0
H in
Equation 2.2-14, then Equations 3 and 4 contain only the two variables E and H. We can
therefore eliminate one or other by direct manipulation. Taking the curl of Equation 3 gives:
curl [ curl(E)] = - curl [B/t]
= -
0
/t [curl(H)]
= -
0

2
D/t
2
2.3-1
We now simplify Equation 2.3-1 using a standard vector identity:
curl [curl(F)] = grad [ div(F)] -
2
F
2.3-2
Here we have introduced two new operators. The first is the gradient of a scalar function
(which in the case of Equation 2.3-2 is div(F)). This is defined in cartesian coordinates as:
grad() = = /x i + /y j + /z k
2.3-3
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 9
The second is the operator
2
(known as the Laplacian), defined in cartesian coordinates
by:

2
=
2
/x
2
+
2
/y
2
+
2
/z
2
2.3-4
Using this identity, Equation 2.3-1 can be reduced to:
grad [ div(E)] -
2
E = -
0

2
(E)/t
2
2.3-5
This is now a form of wave equation. This be simplified considerably, by making
assumptions and approximations that are frequently valid:
a) Since, with no charges present, div(D) = 0, it follows that div(E) = 0. If the medium is
homogeneous and isotropic, then is independent of position and direction, so div E = 0. If
varies only slowly with position, we can still put div(E) = 0 to a reasonable approximation.
Consequently, the first term in Equation 2.3-5 is often zero.
b) If is not time-dependent, then
2
(E)/t
2
=
2
E/t
2
. This is normally a good
assumption; even if varies with time, it needs to do so significantly in a period of
oscillation of the field before it matters. Non-linear optics is concerned precisely with
circumstances in which varies with t, but high fields are required for significant effects to
occur.
Assuming approximations a) and b), Equation 2.3-5 becomes:

2
E =
0

2
E/t
2
2.3-6
This is also a vector wave equation. In cartesian coordinates, it can be written as a set of
three independent scalar equations - one for each co-ordinate. For the E
x
component, for
example, we have:

2
E
x
/x
2
+
2
E
x
/y
2
+
2
E
x
/z
2
=
0

r

2
E
x
/t
2
2.3-7
Here we have written the permittivity as a product, in the form:
=
0

r
2.3-8
Here
0
is the permittivity of free space, again known from experiments to have the value
8.85 x 10
-12
s
2
C
2
/m
3
kg, while
r
is the relative permittivity or relative dielectric
constant of the material concerned.
Equation 2.3-7 is now in the form of a simple classical scalar wave equation, which is
adequate for introducing many aspects of electromagnetic wave propagation. It should be
compared with the one-dimensional wave equation for waves on a string, which is often
written as:

2
y/x
2
= (1/c
2
)
2
y/t
2
2.3-9
Here y is the displacement of the string, x is the distance along it, and c is the wave velocity.
Both equations are essentially similar in character, involving second derivatives of some
quantity with respect to space (on the left-hand side) and second derivatives with respect to
time (on the right). The main difference is that electromagnetic waves are not constrained to
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 10
travel in any particular direction, so that all three spatial coordinates appear in Equation 2.3-
7.
Note that an exactly similar equation to 2.3-6 can be obtained for the magnetic field H:

2
H =
0

r

2
H/t
2
2.3-10
This follows from the 'interchangeability' of the equations for electric and magnetic field
quantities discussed earlier. There will be occasions when it is easier to work with Equation
2.3-6 than 2.3-10, and vice versa.
THE TIME-INDEPENDENT WAVE EQUATION
Now, the electric field E is, in general, a function of x, y, z and t. However, it is often the
case that all the fields involved will be harmonically varying, at some single angular
frequency . This will occur in many situations involving monochromatic light. It is then
convenient to eliminate any time- dependence from the problem. One possibility is to
assume cosinusoidally-varying solutions to the wave equation, but it is generally far more
convenient to use the complex exponential form:
E(x, y, z, t) = E(x, y, z) exp(jt)
2.3-11
Here the function E(x, y, z) accounts for the spatial dependence of the field, while the
exponential exp(jt) describes the time-variation. The use of such complex notation is
standard in electromagnetic theory, but it is important to note that ultimately only the real
parts of Equation 2.3-11 are significant. Generally, entire calculations are performed in
books with the simple understanding that real parts are implied throughout. With this
assumption, we can find time-derivative terms as:
E/t = j E ;
2
E/t
2
= -
2
E
2.3-12
and so on, so that the wave equation reduces to:

2
E = -
2

r
E
2.3-13
Equation 2.3-13 is a time-independent vector wave equation, which is valid for fields
oscillating at a single angular frequency . We will now try to find some solutions to it.
2.4 PLANE WAVES
The simplest form of solution is a plane wave, i.e. a wave whose surfaces of constant phase
are infinite planes, perpendicular to the direction of propagation. Figure 2.4-1 shows the
geometry for a wave travelling in the +z-direction.
Clearly, in this case no field quantities can vary with the transverse coordinates x and y; the
only spatial variation is caused by changes in z. Considering the electric field to begin with,
we must therefore have:
E/x = E/y = 0
2.4-1
so that E must be a function of z only. We may therefore replace E/z by dE/dz, and so on.
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 11
Figure 2.4-1 A plane wave.
For simplicity, we now consider the electric field to have one single component, say in the x-
direction. The vectorial wave equation 2.3-13 then reduces to the scalar equation:
d
2
E
x
/dz
2
+
2

r
E
x
= 0
2.4-2
We now guess that a possible solution has the form:
E
x
= E
x+
exp(-jkz)
2.4-3
where E
x+
is a constant. Direct substitution into 2.4-2 shows that this solution is valid,
provided:
k = (
0

r
)
2.4-4
Hence, the full solution, including the time variation, is:
E
x
= E
x+
exp[j(t - kz)]
2.4-5
Equation 2.4-5 represents a travelling wave, moving in the +z-direction. E
x+
is clearly the
wave amplitude. What do the other constant terms represent? Well, = 2f is the angular
frequency of the wave; typically, the corresponding temporal frequency f is 10
15
Hz for
light waves. We could also write = 2/T, where T is the period of the oscillating field. By
analogy, we could write k = 2/, introducing the new quantity , the spatial wavelength.
This is the distance separating planes of equal phase, as shown in Figure 2.4-1. Typically,
lies in the approximate range 0.4 - 0.8 m for visible light; the lower limit corresponds to
the ultra-violet or blue end of the spectrum, while the latter corresponds to deep red or near
infrared wavelengths. k is an important new parameter, known as the propagation
constant, which we will refer to often later on.
Finally, the velocity of the wave (known as the phase velocity) is given by:
v
ph
= /k.
2.4-6
This can also be written in the more familiar form:
v
ph
= f
2.4-7
From Equation 2.4-4, we can find the phase velocity as:
v
ph
= 1/(
0

r
)
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 12
2.4-8
In a vacuum,
r
= 1, so v
ph
= 1/(
0

0
). Putting in the numbers for
0
and
0
, we get:
v
ph
= 1 / (8.85 x 10
-12
x 4 x 10
-7
) 3 x 10
8
m/s.
2.4-9
This is the velocity of light, written as c, and one of the major successes of Maxwell's
equations was the discovery that this directly measurable quantity could be derived so
simply. In 1849, Fizeau measured c as 3.153 x 10
8
m/s, while in 1983 the value was fixed at
2.99792458 x 10
8
m/s. In other media,
r
is normally greater than unity, so that v
ph
< c; light
therefore generally travels slower in matter than in free space. The refractive index n (a
useful parameter in optics) is then defined as:
n = c/v
ph
=
r
2.4-10
Often, quantities measured in a particular material are referred to those measured in free-
space. For example, the spatial wavelength could be related to the free-space wavelength

0
by:
=
0
/n
2.4-11
Similarly, the propagation constant k could be related to the free-space propagation constant
k
0
by:
k = nk
0
2.4-12
Typically, the relative permittivity
r
and the refractive index n are both functions of
frequency , as we shall see in the next chapter.
THE TRANSVERSE NATURE OF ELECTROMAGNETIC WAVES
We will now consider the other field components that must accompany the solution we have
found for the electric field. First, we note that with the assumptions we have made so far,
Maxwell's equations can be rewritten in the following time-independent form:
div(E) = 0 (1)
div(
0
H) = 0 (2)
curl(E) = -j
0
H (3)
curl(H) = j E (4)
2.4-13
Now, our solution has so far contained only an x-component of the electric field. In this
case:
curl(E) = j E
x
/z - k E
x
/y
2.4-14
Given that E
x
= E
x+
exp(-jkz), we find:
curl(E) = -jk E
x+
exp(-jkz) j
2.4-15
Now, from Equation (3) in 2.4-13, we must have curl(E) = -j
0
H . Hence, the magnetic
field accompanying our solution only has a component in the y-direction. Writing this as:
H = H
y+
exp(-jkz) j
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 13
2.4-16
we can obtain the following constant relation between the electric and magnetic field
amplitudes:
H
y+
/E
x+
= k/
0
2.4-17
The solution therefore really consists of two travelling waves - an electric and a magnetic
component. Both are in-phase, but the field directions are at right angles to each other. We
can represent the complete solution at any given instant in time as in Figure 2.4-2, which
shows the real parts of the two components together. Both exhibit similar cosinusoidal
variations with distance.
Figure 2.4-2 A plane electromagnetic wave.
Is this solution the only one possible? It would seem reasonable to repeat the analysis,
starting with the assumption that the electric field only has a component in the y-direction.
In this case, we find that if E = E
y+
exp(-jkz) j, then H = H
x+
exp(-jkz) i, so the magnetic
field now only has a component in the x-direction. As before, we can find a relation between
the two field amplitudes. This time, we get:
H
x+
/E
y+
= -k/
0
2.4-18
Apart from the minus sign, the amplitude ratio is as before.
What happens if we assume instead that the electric field only has a component in the z-
direction? Well, in an isotropic medium, div(E) = div(E), so Equation (1) in 2.4-13 must
reduce to div(E) = 0. Remember that we can expand this as:
E
x
/x + E
y
/y + E
z
/z = 0
2.4-19
However, since we have already assumed that E/x and E/y = 0, it follows that E
x
/x =
E
y
/y = 0. Hence, E
z
/z must be zero, so E
z
must be a constant independent of z. We
therefore do not find travelling-wave solutions for E
z
, and a similar argument can be used to
show that there are no wave solutions for H
z
. Plane electromagnetic waves are therefore
strictly transverse. They are therefore often described as TEM (standing for Transverse
ElectroMagnetic) waves.
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 14
OPTICAL POLARIZATION
We now consider some of the wider properties of the solutions found so far, beginning with
the important feature of optical polarization. We start by noting that the two independent
travelling wave solutions discussed above can be combined into a more general solution, in
the form:
E = E
x+
exp[j(t - kz +
x
)] i + E
y+
exp[j(t - kz +
y
)] j
2.4-20
where
x
and
y
are arbitrary (but constant) phase factors. The nature of the resulting wave
then depends on the values of E
x+
, E
y+
,
x
and
y
. Several combinations are particularly
important.
(i) If
x
=
y
, the solution can be written as:
E = [E
x+
i + E
y+
j] exp[j(t - kz + )]
= E
0
exp[j(t - kz + )]
2.4-21
In this solution, the direction of the electric field vector is independent of time and space,
and is defined by a new vector E
0
, which is the vectorial sum of E
x+
i and E
y+
j as shown in
Figure 2.4-3. This type of wave is known as a linearly polarized wave, and the direction
of the electric field vector E
0
represents the direction of polarization. Linearly polarized
light is particularly important in engineering optics. It can be produced from natural light
(which has random polarization) by passing it through a polarizer. More importantly, it is
emitted directly by many types of laser.
Figure 2.4-3 Construction of the polarization vector.
(ii) If E
x+
= E
y+
, and
y
=
x
/2, the solution can be written as:
E = E
0
exp[j(t - kz + )] i + E
0
exp[j(t - kz + /2)] j
2.4-22
Or, alternatively, as:
E = E
0
(i j j) exp[j(t - kz + )]
2.4-23
Now, ultimately, we are interested in the real part of E. This is given by:
Re{ E } = E
0
{cos(t - kz + ) i sin(t - kz + ) j}
2.4-24
In this case, the amplitude of the electric field vector is still constant (and equal to E
0
), but
the direction of polarization is not. Instead, it rotates as a function of space and time. This
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 15
solution is known as a circularly polarized wave, because the locus traced of the electric
field vector as a function of time (at a given point) is a circle, as shown in Figure 2.4-4a.
Right- and left-hand circular polarizations are both possible, depending on the sign of the
/2 phase-shift. If E
x+
E
y+
, the locus becomes an ellipse, and the wave is described as
being elliptically polarized (Figure 2.4-4b).
Figure 2.4-4 Loci of the electric field vector for a) circular and b) elliptic polarization.
OBLIQUE WAVES
Simple solutions can also be found for plane waves travelling in different directions. For
example, Figure 2.4-5 shows a wave, travelling in the x - z plane at an angle to the z-axis.
If the wave is linearly polarized in the y-direction, i.e. perpendicular to the plane of the
figure, the time-independent electric field is given simply by:
E = j E
y+
exp[-j (kx sin() + kz cos())]
2.4-25
This result is obtained simply by rotating the co-ordinate system of Figure 2.4-1 about the
y-axis. We can check that it is correct by putting = 0, whereupon Equation 2.4-25 reduces
to our original solution, E = j E
y+
exp(-jkz). Similarly, if = /2, we get E = j E
y+
exp(-jkx) -
this is also as expected, as the wave is now travelling in the +x direction.
Figure 2.4-5 An obliquely-travelling plane wave.
However, we note that in the time it takes the wave to travel a distance in the direction of
propagation, the phase-fronts have advanced further in the z-direction, a distance ' = /
cos(). The effective phase velocity in the z-direction is therefore greater than v
ph
, by the
factor 1/cos(). Some thought is required fully to appreciate this point, since it implies faster
motion of the disturbance in a direction off-axis to the direction of propagation.
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 16
IMPEDANCE
We can find the impedance of a medium through which a wave is propagating, as follows.
Since /k = 1/(
0

r
), the ratio E
x+
/H
y+
can be found as (
0
/
0

r
). Since this ratio has the
dimensions of ohms, it is called the characteristic impedance Z of the medium. Thus, we
may put:
Z = (
0
/
0

r
)
2.4-26
For free space,
r
= 1, so that Z
0
= (
0
/
0
) = 377 . Anyone owning a radio tuner with an
external aerial socket should note the input impedance - it will be around this value. The
impedance of any other material may be related to the impedance of free space using:
Z = Z
0
/
r
= Z
0
/n
2.4-27
Impedance is therefore inversely proportional to refractive index.
WAVES IN LOSSY DIELECTRICS
We may also extend the analysis to describe the behaviour of plane waves propagating in
slightly lossy dielectric media, as follows. We begin by including loss in a
phenomenological way, by assuming that the relative dielectric constant of the material is
complex-valued. We shall justify this more rigorously in Chapter 3, but for the time being
we will simply take
r
to be defined by:

r
=
r
' - j
r
''
2.4-28
For x-polarized waves travelling in the +z-direction, the scalar wave equation we must solve
can be found by substituting Equation 2.4-28 into Equation 2.4-2. We get:
d
2
E
x
/dz
2
+
2

0
[
r
' - j
r
''] E
x
= 0
2.4-29
Assuming a solution in the form used previously, namely E
x
= E
x+
exp(-jkz), we find that
the propagation constant k is now given by:
k
2
=
2

0
[
r
' - j
r
"]
2.4-30
Or:
k = (
0

r
') [1 - j
r
"/
r
']
2.4-31
Making the additional assumption that the loss is small, so that
r
"
r
', we may use a
binomial approximation for Equation 2.4-31, which gives:
k = (
0

r
') [1 - j
r
"/2
r
']
= k' - jk"
2.4-32
where the real and imaginary parts of k are given by:
k' = (
0

r
') , and k" = k'
r
"/2
r
'
2.4-33
Since the propagation constant is complex, we can now rearrange the solution as the product
of two exponentials, as:
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 17
E
x
= E
x+
exp(-k"z) exp(-jk'z)
2.4-34
Equation 2.4-34 has the form of a plane wave, whose amplitude decays exponentially with
distance z. The real part of the propagation constant, k', defines the phase variation of the
wave, while the imaginary part k" defines the amplitude variation. k" is known as the
absorption constant, and is often given the symbol . The presence of even a small
amount of loss in a material (which is often unavoidable) therefore causes the exponential
decay of a propagating wave.
We may also hypothesise the existence of media with a negative value of
r
". In this case,
the solution corresponds to an exponentially growing wave, of the form:
E
x
= E
x+
exp(+gz) exp(-jk'z)
2.4-35
where g, the gain constant of the medium, is defined by g = -k". The growth of an optical
wave as it passes through a medium with gain is the key to the operation of all laser devices.
WAVES IN METALS
Finally, we can extend the analysis to include materials with non-zero conductivity - for
example, metals. All that is required is to work through Equations 2.3-1 to 2.3-13 again,
assuming to start with that curl(H) = J + D/t, rather than simply D/t. If this is done, a
slightly revised time-independent wave equation can be obtained for monochromatic waves:

2
E +{
2

0
- j
0
} E = 0
2.4-36
Note that this has exactly the same form as Equation 2.3-13, but the term
2

0
has been
replaced here by
2

0
- j
0
. Since the former was previously interpreted in terms of the
propagation constant by putting
2

0
= k
2
, it is reasonable to write in this case:
k
2
=
2

0
- j
0

=
2

0
( - j/)
2.4-37
If we wish, we can interpret all the terms inside the bracket in Equation 2.4-37 as a modified
dielectric constant, given by:
=

- j /
2.4-38
where

is the value as tends to infinity. If this is done, we see that the effective dielectric
constant is complex once again. However, we cannot immediately equate the real and
imaginary parts of to

and -/, since this implies an assumption that the conductivity

is real. While this is justified for low values of , it is certainly not the case at optical
frequencies. We shall see why this should be so in Chapter 3.
2.5 POWER FLOW
It is clearly important to form an expression for power flow in terms of E and H, since this
is something we can measure. This is done by considering the work done on charges in a
volume, as a combination of a change in stored energy and a flow of power into or out of
the volume. We shall consider initially time-dependent fields.
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 18
The rate of work done by the fields per unit volume is J . E , by analogy with W = I V in
electrical circuits. From Equation 4 in 2.2-14 we get:
curl(H) - D/t = J
2.5-1
We can then write:
J . E = E . curl (H) - E . D/t
2.5-2
We now introduce a further vector identity, again described more fully in mathematics texts:
div(E x H) = H . curl(E) - E . curl(H)
2.5-3
We also need another relation, from Equation 3 in 2.2-14:
curl(E) = -B/t
2.5-4
Combining Equations 2.5-2 to 2.5-4, we get:
J . E = - H . B/t - E . D/t - div (E x H)
2.5-5
We can now rewrite this in the following standard form:
U/t + . S + J . E = 0
2.5-6
Where:
U/t = [ E . D/t + H . B/t ]
2.5-7
And:
S = E x H
2.5-8
Equation 2.5-6 (known as Poynting's Theorem, after John Poynting, 1852-1914) is
effectively a power conservation relation, since it relates the rate of change of stored energy
to the outward energy flow and the energy dissipated. U is the density of the energy stored
in the electromagnetic fields (measured in J/m
3
). This can be divided conveniently into two
components: U
E
= 1/2 E . D represents the electrically-stored energy, while U
M
= 1/2 H . B
is the energy in the magnetic field. Similarly the vector S, known as the Poynting vector,
describes the power flow (measured in W/m
2
). However, it should be noted that S is an
extremely fast-varying function. In fact, S contains components at 2 (or about 10
30
Hz for
optical waves) which are clearly not measurable by any practical technique. It is therefore
convenient to define an alternative quantity related to power, that is directly measurable.
IRRADIANCE
For oscillating fields, we have been using the alternative notation E(x, y, z, t) = E(x, y, z)
exp(jt), where the real part is implied. In this case, we can calculate an associated time-
averaged Poynting vector or irradiance S. This is the mean of S, over many oscillations,
defined as:
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 19
S = (1/T)
T
S dt
2.5-9
where T is large compared with the period of the oscillations. Substituting harmonic
solutions for E and H into Equation 2.5-9, we obtain:
S = (1/T)
T
Re{ E exp(jt) } x Re{ H exp(jt) } dt
2.5-10
It is then simple to show that the irradiance is given by:
S = 1/2 Re [ E x H* ]
2.5-11
The expression for S therefore contains no time variation, as required.
TIME-AVERAGED POWER
Irradiance is still not a directly measurable quantity. However, we can measure the time-
averaged power P flowing through a given surface. This is found as the integral of the
normal component of S over the surface, as shown in Figure 2.5-1. Using Equation 2.5-11,
P may be evaluated as:
P = 1/2 Re{
A
[ E x H* ] . da }
2.5-12
Time-averaged power is extremely important, since it is one of the few parameters of a high-
frequency electromagnetic field that can actually be measured. Suitable detectors are the
human eye, solar cells and semiconductor p-n junction photodiodes.
Figure 2.5-1 Geometry for calculation of power flow.
DESIGN EXAMPLE
As an example, we shall calculate the power carried by a plane wave travelling in the +z-
direction. Assuming that components of both polarizations are present, the electric field can
be written as:
E = [ i E
x+
+ j E
y+
] exp (-jkz)
2.5-13
The corresponding magnetic field can be found from Equations 2.4-17 and 2.4-18 as:
H = [E
y+
(-k/
0
) i + E
x+
(k/
0
) j] exp(-jkz)
2.5-14
The time-averaged power flow in the z-direction, per unit area, is therefore:
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 20
P = 1/2 Re [ E x H* ] . k
= 1/2 Re [i {E
y
H
z
* - E
z
H
y
*} + j {E
z
H
x
* - E
x
H
z
*} + k {E
x
H
y
* - E
y
H
x
*}] . k
2.5-15
which in this case reduces to:
P = 1/2 Re [ E
x
H
y
* - E
y
H
x
* ]
2.5-16
Substituting in the necessary values, we get:
P = (1/2) (
0

r
/
0
) [E
x+
2
+ E
y+
2
]
2.5-17
This expression allows us to relate electric field strengths in V/m to power density in W/m
2
.
Now, each of the two terms above clearly corresponds to one polarization component.
Writing E
2
= E
x+
2
+ E
y+
2
, where E is the amplitude of the combined electric field, and using
the definitions of impedance given in Equations 2.4-26 and 2.4-27 we obtain:
P = 1/2 E
2
/ Z
= 1/2 n E
2
/ Z
0
2.5-18
The power carried by a plane wave is therefore proportional to the product of E
2
and the
refractive index of the medium.
2.6 THE PROPAGATION OF GENERAL TIME-VARYING
SIGNALS
So far, we have concentrated on the behaviour of single-frequency electromagnetic fields,
which in optics correspond to monochromatic light. Throughout this book, however, we will
be interested in the use of optical devices in an engineering environment. One of the most
obvious applications is a communications system, which might loosely be defined as an
arrangement for the transmission of information between different points. Unfortunately, a
perfectly monochromatic wave (which, strictly speaking, must exist for all time without
changing its frequency or amplitude) cannot carry any information. Only the modulation of
such a wave - for example, by switching it on and off - can do so.
In general, therefore, we will be interested in the transmission of time-varying signals. A
suitable framework for their analysis is provided by Fourier transform theory, which states
that any signal f(t) may be decomposed into an infinite sum of single-frequency terms.
Conventionally, this relationship is written as an integral transformation, of the form:
f(t) = (1/2)
-

F() exp(jt) d
2.6-1
where F() represents the amplitude of the component at angular frequency , which itself
may be found from the signal using the inverse transform:
F() =
-

f(t) exp(-jt) dt
2.6-2
DESIGN EXAMPLE
As an example, we may compute the frequency spectrum of a signal consisting of a short
burst of a monochromatic carrier, of unity amplitude and angular frequency
c
, as in Figure
2.6-1.
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 21
Figure 2.6-1 A burst of single-frequency tone.
Assuming that the duration of the burst is T, Equation 2.6-2 reduces to:
F() =
-T/2

T/2
exp(j
c
t) exp(-jt) dt
2.6-3
Evaluation of the integral then gives:
F(') = T sinc('T/2)
2.6-4
where ' =
c
- and sinc(x) = sin(x) / x. Figure 2.6-2 shows a plot of the frequency
spectrum; this peaks at =
c
and decays away on either side of this point with a typical
filter envelope.
Figure 2.6-2 Frequency spectrum of a burst of a single tone.
We may obtain an estimate of the width of this frequency spectrum by noting that the first
zeros in Equation 2.6-4 are reached when 'T/2 = . This allows the definition of an
approximate signal bandwidth (which is then about half the width of the main lobe) in
the form:
T = 2
2.6-5
The signal bandwidth is therefore inversely proportional to the duration of the burst. This
implies that high bit-rate data transmission will involve large bandwiths.
DISPERSION
The Fourier relations above may be used to analyse communication channels, as follows.
Given a specified input f(t), Equation 2.6-2 may be used to identify the frequency
components of the signal and their corresponding amplitudes F(). These components may
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 22
then mentally be passed through the channel in turn; naturally, a component of angular
frequency will propagate as a travelling wave of the same angular frequency. On arrival at
the far end of the channel, the amplitudes of the components may well have changed, so that
an amplitude F() might be received as F'(). However, the total received signal f'(t) may be
reconstructed from these modified constituents by using a simple adaptation of Equation
2.6-1, namely:
f'(t) = (1/2)
-

F'() exp(jt) d
2.6-6
By comparing the received signal f'(t) with the transmitted signal f(t), the effect of the
channel may be assessed. For example, it will be important to know in advance what type of
signals may be passed through the channel and still arrive in recognisable form. Before this
can be done, however, we must identify the signal distortions that are possible. By and large,
there are just two. Firstly, the relative amplitudes of the frequency components may alter.
This could occur in a channel with frequency- dependent attenuation. Secondly, the relative
phases of the components might change, if the phase velocity of the channel is frequency-
dependent. Generally, the latter effect is the most significant. It is known as dispersion, and
we will now consider some of its features.
We start by returning to Section 2.4, where the phase velocity of a wave of angular
frequency travelling in a homogeneous medium was defined as v
ph
= c/n. From this, we
may infer that the medium will be dispersive if there is any dependence of the refractive
index on frequency. In fact, this is the case in all matter. The result is therefore normally
described as material dispersion (to distinguish it from other effects that occur in more
complicated transmission channels, particularly waveguides). We will examine the
underlying reasons for this dependence in Chapter 3. However, we also obtained the
alternative definition v
ph
= /k. Consequently, in a dispersive medium we would expect a
more complicated relation between and k than just /k = constant.
One of the simplest examples of a dispersive medium is provided by an ionised medium
(such as the ionosphere), for which it can be shown that:
= [
p
2
+ c
2
k
2
]
2.6-7
where
p
is a constant, the plasma frequency, which will be introduced properly in
Chapter 3. This type of relation can be represented as a plot of against k called a
dispersion diagram, as shown in Figure 2.6-3. In this case, the diagram shows that
tends to
p
for small values of k, while for large k, tends to the dashed line = ck.
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 23
Figure 2.6-3 -k diagram for an ionized medium.
The phase velocity /k may then be found either from the dispersion diagram, or directly
from Equation 2.6-7. The latter process yields:
v
ph
= [
p
2
/k
2
+ c
2
]
2.6-8
From this, it can be seen that the phase velocity is not constant; it tends to infinity as k tends
to zero, and to c as k becomes large. To assess the effect of this variation, we shall consider
the propagation of an elementary compound signal, consisting of components at just two
distinct angular frequencies + d and - d, where d is small. For simplicity, we take
the amplitudes of the two waves to be the same. However, we assume that the phase
velocities at the two frequencies are unequal, so that the corresponding propagation
constants must be written as k + dk and k - dk. Assuming y-polarization and z-propagation,
the electric field of the signal might then be written:
E
y
= E
y+
[ exp{j (( + d)t - (k + dk)z)} + exp{j(( - d)t - (k - dk)z)} ]
2.6-9
Figure 2.6-4 shows a snapshot of the combined field at t = 0. Note that the two waves sum
together to give what amounts to a carrier of constant period, modulated by an envelope
(shown dashed).
Figure 2.6-4 The sum of two waves of a similar frequency.
We may illustrate this mathematically, by combining the two components slightly
differently. If the common factor exp{j(t - kz)} is taken out, we get:
E
y
= 2E
y+
exp{j(t - kz)} cos{d t - dk z}
2.6-10
This suggests that we may view propagating signals in two different ways. We may either
regard them as a sum of a number of separate travelling waves (as in Equation 2.6-9) or as a
single modulated wave (Equation 2.6-10). However, the latter viewpoint shows clearly that
the information-carrying component of the signal - the modulation envelope - is also
propagating as a travelling wave, defined by the term cos(d t - dk z).This envelope must
therefore also have a velocity of propagation, which in general is distinct from that of the
carrier wave. Since it refers to a group of waves, rather than a single wave, it is known as the
group velocity v
g
, and is defined as:
v
g
= d/dk
2.6-11
For the ionosphere, we may find the group velocity by differentiating Equation 2.6-7, to get:
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 24
v
g
= c
2
/ [
p
2
/k
2
+ c
2
]
2.6-12
This is clearly different from the expression derived earlier for the phase velocity. However,
for large k (and thus very high frequencies), v
g
tends to v
ph
.
The analysis above can be used to find the velocity of information carried by groups of
waves of similar frequency. However, for groups comprising a wider range of frequencies,
v
g
may not be considered constant, so different groups of a signal will travel at different
speeds. This can result in a damaging effect, known as pulse broadening, which limits the
rate of data transmission. We can illustrate this by considering the problem of transmitting
data over a distance L via successive bursts of a high-frequency carrier. Each 'one' in the
data stream corresponds to a burst of duration T, and the separation between successive
bursts is T. Figure 2.6-5a shows the modulation envelope of a typical section of a message.
Figure 2.6-5 A short section of a message, a) as sent and b) after travelling some distance
in a dispersive medium.
From our earlier example, we may estimate the required bandwidth to be , so that the
frequencies comprising the signal range approximately from
1
=
c
- /2 to
2
=
c
+
/2. At these extremes, the group velocity may have different values - say, v
g1
and v
g2
.
Consequently, different constituents of the signal must arrive at the far end of the channel at
times ranging from t
1
= L/v
g1
to t
2
= L/v
g2
. Assuming that t
2
> t
1
, this spread in arrival times
will amount to a lengthening of the information-carrying envelope from T to T + t,
where t = (t
2
- t
1
). If L is large enough, we may even find that t T. Information from
successive bursts will then become mixed (as is beginning to happen in Figure 2.6-5b) so
that the message is beyond recovery by simple means.
Now, the effective broadening of the signal may be calculated by writing:
t = L{1/v
g2
- 1/v
g1
} L d(1/v
g
)/d
2.6-13
Since 1/v
g
= dk/d, this yields:
t L d
2
k/d
2
2.6-14
Clearly, pulse broadening may be minimised by ensuring that the term d
2
k/d
2
is as small
as possible, and preferably zero. We may see how this relates to material parameters by
noting that k = n/c. Consequently, for a dispersive medium we can write:
dk/d = 1/c {n + dn/d}
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 25
2.6-15
Similarly, since
0
= 2c/, we may put:
dn/d = (dn/d
0
) (d
0
/d) = -2c/
2
dn/d
0
2.6-16
so that:
dk/d = 1/c {n -
0
dn/d
0
}
2.6-17
Differentiating a second time, we get:
d
2
k/d
2
= {d(dk/d)/d
0
} (d
0
/d) = -
0
/c (d
2
n/d
0
2
) (d
0
/d)
2.6-18
Finally, if we put d
0
/d
0
/, where
0
is the wavelength range of the signal, we can
obtain a variant of Equation 2.6-14 that emphasises the contribution of the medium to pulse
broadening:
t = -(L
0

0
/c) d
2
n/d
0
2
2.6-19
This suggests that dispersion can be minimised by operating near a wavelength where
d
2
n/d
0
2
= 0. In silica (the main constituent of most optical fibres) this occurs at
0
1.27
m. Consequently, optical communications systems almost always operate at near infrared
wavelengths.
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 26
PROBLEMS
2.1. Starting from the differential form of Maxwell's equations, derive a time-dependent
vector wave equation, valid for a uniform dielectric medium and containing only the
magnetic field H.

2.2. A plane electromagnetic wave oscillates at an angular frequency of 2.978 x 10
15
rad/sec.
Find its wavelength in (a) free space, and (b) a material of relative dielectric constant
2.25.
[(a) 0.633 m; (b) 0.422 m]

2.3. A plane electromagnetic wave is defined by E = E
yo
exp{-jk
0
(ax + bz)} j, where k
0
is
the propagation constant of free space, and the constants a and b are given by a = 1, b =
3. Find (a) the wave amplitude, (b) the direction of polarization, (c) the direction of
travel, and (d) the refractive index of the medium.
[(c) 30
o
to the z-axis, in the x - z plane; (d) 2]

2.4. The expression E
y
= E
y0
exp(-jkz) is a solution to the scalar equation
2
E
y
+
2

0
E
y
= 0 that represents a plane wave travelling in the z-direction. What is k? Show that the
inhomogeneous wave E
y
= E
y0
exp(x) exp(-jz) is also a solution. What relation must
be satisfied by and ?

2.5. Sketch the loci of (a) constant phase, and (b) constant amplitude for the inhomogeneous
wave in Question 2.4. What direction is it travelling in? Does it travel faster or slower
than the plane wave? What do you think it represents?

2.6. A optical wave of wavelength
0
= 0.633 m is travelling in a material of complex
relative dielectric constant
r
= 2.25 - j10
-8
. What are the values of the propagation
constant and the absorption coefficient? Find the distance the wave must travel in the
medium before the power it carries decays to 1/e of its initial value.
[14.89 x 10
6
m
-1
; 0.0331 m
-1
; 15.11 m]

2.7. Show that the irradiance S = 1/2 Re [E x H*] does indeed represent the time average of
the Poynting vector S = E x H, for harmonically varying fields.

2.8. Determine the z-component of the irradiance, for the wave in Question 2.3. What is the
direction of maximum irradiance?

2.9. The time-averaged power density of a linearly polarized plane wave in free space is 100
mW/cm
2
. Find the peak amplitude of the electric field.
[868 V/m]

2.10. The relationship between and k for electromagnetic waves in the ionosphere may be
shown to have the form = [
p
2
+ c
2
k
2
], where c is the velocity of light and
p
is the
plasma frequency. Sketch the variation of the phase and group velocity with k. Find the
value of v
g
when
p
. What is the significance of this result? Find a solution for k
when <
p
. What form of wave do you think that this solution describes?
CHAPTER
TWO
R.R.A.Syms and J.R.Cozens Optical Guided Waves and Devices 27
SUGGESTIONS FOR FURTHER READING
Born M., Wolf E. "Principles of optics", 6
th
Edn., Pergamon Press, Oxford (1980), Chapter
1
Hecht E., "Optics", 2nd Edn., Addison-Wesley Publishing Co., Reading MA (1987);
Chapters 1, 2, 3, & 8.
Ramo S., Whinnery J.R., van Duzer T. "Fields and waves in communication electronics" 2
nd
Edn., John Wiley and Sons, New York (1984); Chapter 3.