LECTURE NOTES

OF

PROPAGATION AND RADIATION OF

ELECTROMAGNETIC WAVES

Mário G. Silveirinha

Lisboa, November 2023

 Propagation and Radiation

ii
 Propagation and Radiation

Index

1. The Wave Equation ............................................................................................................. 1

1.1 Maxwell’s Equations in free-space .......................................................................... 1
1.2 Solution of the wave equation .................................................................................. 3
1.3 Nondispersive materials ........................................................................................... 6
2. Plane Waves ........................................................................................................................ 9
2.1 Time-harmonic regime ............................................................................................. 9
2.2 Properties of plane waves ...................................................................................... 10
2.3 Plane waves in nondispersive media ...................................................................... 13
2.4 The wave impedance .............................................................................................. 16
2.5 Time and space evolution of the electromagnetic fields ........................................ 17
3. Propagation in Lossy Media .............................................................................................. 19
3.1 Macroscopic electrodynamics ................................................................................ 19
3.2 Dispersive dielectrics ............................................................................................. 20
3.3 Plane waves in lossy materials ............................................................................... 23
3.4 Time and space evolution of a plane wave in a lossy medium .............................. 24
3.5 Group velocity ....................................................................................................... 26
3.6 Good dielectrics and good conductors ................................................................... 30
4. Wave Polarization ............................................................................................................. 33
4.1 Polarization curve .................................................................................................. 33
4.2 Polarization ratio .................................................................................................... 36
5. Electromagnetic Energy and the Poynting Vector ............................................................ 39
5.1 Energy conservation law ........................................................................................ 39
5.2 Power due to the external excitation ...................................................................... 40
5.3 The Poynting theorem ............................................................................................ 41
5.4 Poynting vector in time-harmonic regime ............................................................. 43
5.5 Poynting vector for a plane wave ........................................................................... 44
6. Reflection and refraction ................................................................................................... 47
6.1 Boundary conditions .............................................................................................. 47
6.2 Normal incidence at a planar interface .................................................................. 48
6.3 Oblique incidence at a planar interface .................................................................. 51
6.4 Snell’s laws ............................................................................................................ 52
iii
 Propagation and Radiation

6.5 Polarization of the reflected and transmitted waves .............................................. 55

6.6 Reflection and transmission coefficients for oblique incidence ............................ 57
6.7 The Brewster angle ................................................................................................ 59
6.8 Total internal reflection .......................................................................................... 62

iv
 Fundamental Concepts

1. The Wave Equation

1.1 Maxwell’s Equations in free-space

Maxwell’s equations are a set of four fundamental equations that form the cornerstone of

classical electromagnetism. They were developed by the Scottish physicist James Clerk Maxwell

in the mid-19th century and represent a unification of various principles governing electric and

magnetic phenomena. In the 1820s and 1830s, Michael Faraday and André-Marie Ampère made

significant contributions to understanding the relationship between electricity and magnetism.

Faraday discovered electromagnetic induction, showing that a changing magnetic field could

induce an electric current, while Ampère formulated Ampère’s law, which relates electric current

to its associated magnetic field. Maxwell’s groundbreaking work not only synthesized existing

knowledge but also paved the way for profound advancements in physics and technology,

leading to the development of radio, television, and wireless communication.

Maxwell’s theory was formulated between 1861 and 1862. The equations that bear his name

describe the behavior of electric E  x, y, z , t  and induction B  x, y, z , t  fields in space and time,

emphasizing the interdependence of these two phenomena. In free-space, the two key equations

read:

 B 
E   B,   j  0 E. (1.1)
t 0 t

Here,  0  8.85 1012  F / m  is the vacuum permittivity and 0  4 107  H / m  is the

vacuum permeability. The equation on the left is known as Faraday’s law and the equation on

the right as Ampère’s law.

1
 Fundamental Concepts

The electromagnetic fields are created by charged particles. These are modelled by a charge

density  , and by a current density j . The current density describes the transport of charge due

to physical motion. Charge conservation requires that j and  are linked through a continuity

equation:

  j  t   0 . (1.2)

We use the shorthand notation  t   /  t . Integrating the continuity equation over a volume V

with boundary surface  one obtains with the help of the divergence theorem that:

dQ
   j  nˆ ds . (1.3)
dt 

Here, Q    dV is the total charge inside volume V and n̂ is the unit normal vector directed to
V

the exterior of  . Thus, the continuity equation in integral form establishes that a variation of the

charge inside V is due to the flux of the current density j through the boundary surface  . In

simple words, a variation of the charge in a volume can always be attributed to an electric

current flowing through the boundary surface.

The Faraday’s and Ampère’s laws are complemented with the Gauss’s law for magnetic and

electric fields:

B  0 , D   , (1.4)

2
 Fundamental Concepts

with D   0 E the electric displacement vector with units of C / m 2 .1 The units of electric field

are V / m , the units of the magnetic induction are Tesla [T] and the units of the current density

are A / m 2 . For convenience, we introduce H  B / 0 which we shall refer to as the magnetic

field. It has units of A / m .

The Maxwell’s equations in free-space are linear. This means that the superposition principle

applies. Thus, the field  E, H  created by a superposition of current distributions ( j  j1  j2 ) is

the superposition of the fields created by the individual current distributions

 E, H    E1  E2 , H1  H 2  .

1.2 Solution of the wave equation

Let us consider the Maxwell’s equations in a free-space (vacuum) region without any field

sources ( j  0 and   0 ). In these conditions, the fields satisfy:

 
  E   0 H,   H  0 E. (1.5)
t t

This determines a set of first-order coupled partial-differential equations in space and time. It is

possible to eliminate the magnetic field and obtain an equation that rules the dynamics of the

electric field alone. To this end, we differentiate in time both sides of the Ampère’s law

H    1
(  0 E ) and use Faraday’s law ( H    E ) to eliminate the magnetic field.
t t t t 0

This yields:

2
    E   0 0 E  0. (1.6)
t 2

1 As a side note, we point out that the Eqs. (1.4) are not fully independent of (1.1). For example, calculating the divergence of both members of
the Faraday’s law (   E    B ) it follows that    B  0 . We took into account that the divergence of the curl of a vector vanishes:
 
t t
    E   0 . Thus, the constraint   B  0 only needs to be enforced at initial time, as then it will be satisfied at any time instant
thereafter. Similar considerations hold true for the Gauss’s law   D   . Due to this reason, the two equations in (1.1) characterize fully the
time evolution of the electromagnetic field, whereas (1.4) may be regarded as initial time boundary conditions.
3
 Fundamental Concepts

To proceed we use the vector identity     E      E    2 E . Taking into account that from

Gauss’s law 0    D      0 E  , it follows that the electric field satisfies the so-called wave

equation:

2
 E   0 0 2 E  0 .
2
(1.7)
t

Note that the wave equation is of second order has it involves two derivatives in the time and

2 2 2
space coordinates. In the above,  2    is the Laplacian. The wave equation may
x 2 y 2 z 2

be spelled out in terms of field components ( E  Ex xˆ  E y yˆ  Ez zˆ ) as follows:

2
 2 Ex   0 0 Ex  0
t 2
2
 2 E y   0 0 2 E y  0
t
...

In some simple cases the wave equation can be integrated explicitly. To illustrate this, let’s

look for solutions of the wave equation of the type:

E  E  z , t  xˆ

This corresponds to a wave that oscillates along the x direction, but propagates along z. Since

 / x   / y  0 the wave equation reduces to (1D wave equation):

2 E 1 2 E
 0. (1.8)
z 2 v 2p t 2

where we introduced the parameter v p  1/  0 0 , which has unities of velocity. The general

solution of the 1D wave equation is well know:

E  z , t   E  z  v p t   E  z  v p t  . (1.9)

4
 Fundamental Concepts

Here, E  u  , E  u  are arbitrary functions of the single parameter u . Next, we check by direct

substitution that (1.9) satisfies the wave equation. Considering for simplicity that E  0 we

find using the chain rule:

E  
 E  z  v p t  u  E  u  , 2 E u
 E  u   E  u 
z    z z 2
z
 u 

E u 2 E u
 E  u   v p E  u  ,  v p E  u   v 2p E  u 
t t t 2
t

The prime denotes a derivative with respect to u. Thus, we have shown that

2 E 1 2 E 1
 2 2  E  2 v 2p E  0
z 2
v p t vp

which confirms that (1.9) satisfies the wave equation for an arbitrary E .

Let us now physically interpret the terms E and E of the general solution of the wave

equation. For simplicity, we focus on the solution E and suppose again that E  0 . Then:

E  z , t  0   E  z 

E  z , t  t0   E  z  v p t0 

Thus, the function E defines the profile of the wave along the direction of propagation at

initial time (t=0). As seen, for later time instant ( t0  0 ) the wave suffers a translation in space

5
 Fundamental Concepts

and the profile is recentered at the point z0  v p t0 . Thereby, E describes a wave that propagates

dz0
towards the +z-direction with velocity  v p . A similar analysis shows that E describes a
dt0

wave that propagates along the –z-direction also with velocity v p . Usually the profiles E and

E are determined by the excitation, (e.g., by the format of the voltage signal applied to the

antenna that radiates the wave), and by the reflection of the wave on material bodies (e.g., a

mirror).

The velocity of propagation can be calculated explicitly:

1 1
vp    3 108 m / s  c
 0 0 8.85 10 12
 4 10 7

It is exactly the “speed of light” in the vacuum. Hence, the Maxwell’s equations predict the

propagation of a local disturbance of the electromagnetic field at speed c. As we shall see later in

the course, the analysis can be extended to fully 3D dimensional scenarios, with similar

conclusions.

1.3 Nondispersive materials

In this course, we will restrict our attention to the simplest class of materials: isotropic media.

The strength of the response of an isotropic medium is independent of the orientation of the

applied fields.

For most purposes, a material can be regarded (from an electromagnetic point of view) as a

continuum – without any form of granularity. The medium response is determined by the

relation between the vectors D, B and the E, H fields. Typically, D only depends on the electric

field, whereas B only depends on the magnetic field H .

6
 Fundamental Concepts

For a nondispersive material the response of D to E is instantaneous. In the isotropic case it is

characterized by a scalar: the material permittivity  . Similarly, the response of B to H is

described by the material permeability  . The permittivity and permeability link the D, B fields

with the E, H fields as follows:

D  E, B  H . (1.10)

The permittivity and the permeability model the effect of microscopic currents induced inside the

material by an external excitation. The Maxwell’s equations in a nondispersive material system

read:

 
E   B,   H  jext  D. (1.11)
t t

Here, jext represents the current density due to currents external to the materials.

For a uniform isotropic material (with  and  independent of the space coordinates) the

wave equation becomes:

2
 E   2 E  0
2

t

The propagation velocity in the material is:

1 c
vp   . (1.12)
 n

Here, n   r r is the refractive index of the material. We introduced the relative permittivity

 r   /  0 and the relative permeability r   /  0 .

7
 Fundamental Concepts

The interaction of the fields with the material constituents effectively reduces the wave

velocity. Due to this reason the refractive index of a nondispersive material satisfies n  1 . Thus,

n may be understood as a slow-down factor.

8
 Fundamental Concepts

2. Plane Waves

2.1 Time-harmonic regime

In time-harmonic regime the variation in time of all the field components is of the type

cos t    . Here,   2 f is the oscillation frequency. The period of oscillation is

1 2
T  .
f 

The sinusoidal variation of the fields is typically inherited from the time variation of the

excitation, e.g., the voltage applied to the terminals of an antenna.

In time-harmonic regime the fields are mostly convenient described by complex amplitudes

or phasors. For example, the electric field E  x, y, z , t  is written in terms of a complex amplitude

E  x, y, z  such that:

E  r, t   Re E  r  e jt 

9
 Fundamental Concepts

For convenience, we denote a generic point in space  x, y , z  as r . The bar underneath the

electric field symbol indicates that we refer to the complex amplitude of the field. The above

equation can spelled out in terms of the field components:

Ex  r, t   Re  E x  r  e jt 
E y  r, t   Re  E y  r  e jt 
...

We used E  Ex xˆ  E y yˆ  Ez zˆ and E  E x xˆ  E y yˆ  E z zˆ .

The complex amplitudes of the fields satisfy the time-independent Maxwell’s equations.

These are obtained from (1.11) by replacing the fields by the corresponding phasors ( E  E ,

H  H , etc) and the derivative in time  / t by j :

  E   j B ,   H  jext  j D . (2.1)

As a side note, we point out that the above equations and the continuity equation imply that

  B  0 and   D   ext . Thus, the Gauss’s laws for electricity and magnetism are redundant in

time-harmonic regime, i.e., they are embedded in Faraday’s and Ampère’s laws.

The D, B fields are linked to the E, H fields by material-dependent constitutive relations.

For nondispersive dielectrics Eq. (1.10) holds true so that:

  E   j H ,   H  jext  j E . (2.2)

2.2 Properties of plane waves

Plane waves are the simplest solution of the Maxwell’s equations. By definition, a plane wave

field is described by:

E  E0 e  jk r , H  H 0 e  jk r . (2.3)

The term e  jk r is known as the propagation factor as it determines the spatial variation of the

fields. It is written in terms of the observation point coordinates r   x, y, z  and in terms of the

10
 Fundamental Concepts

wave vector k   k x , k y , k z  . Note that k  r  k x x  k y y  k z z . The vectors E0 , H 0 are the fields

evaluated at the origin ( r  0 ).

It is often convenient to write the wave vector in terms of a unit vector d̂ and a scalar k:

k  kdˆ . The scalar k is the wave number. In standard isotropic media the vector d̂ determines

the direction of propagation of the wave. The propagation factor can be written as:
ˆ
e  jk r  e  jkdr

The phase of the wave is   kdˆ  r . The set of space points wherein   const. is known as a

wavefront. The fields oscillate in phase at any two points on the same wavefront. This means that

the corresponding field oscillations are synchronized, even though they take place at distinct

points of space.

Let us characterize the wavefronts of a plane wave. These correspond to the solutions of

  const. :

kdˆ  r  const.  2 m, m=0,  1,...

We took into account that phases that differ by an integer multiple of 2 are indistinguishable.

For simplicity, we will focus on the solutions of   0 , i.e., on the wavefronts with zero-phase.

The parameter u  dˆ  r is the space coordinate measured along the direction d̂ . Thus, the

wavefronts with zero phase are determined by the conditions:

2
um with m=0,  1,...
k

These are planes perpendicular to the direction of propagation d̂ . The distance between adjacent

equiphase wavefronts is   2 / k , which is known as the wavelength. It determines the spatial

period that separates the wavefronts wherein the electromagnetic field oscillations are

synchronous.

11
 Fundamental Concepts

The wave number has units of inverse length ( m 1 or rad m ). It can be written in terms of

the wavelength as:

2
k . (2.4)


The wave number normalized to 2 gives the “number of waves” in 1 meter:

k 1 1 meter
   number of waves
2  

For example, if k  2 103 m 1 then there are 1000 waves in 1 meter.

12
 Fundamental Concepts

2.3 Plane waves in nondispersive media

Next, we characterize the plane waves in a nondispersive uniform unbounded material, with  , 

independent of r   x, y, z  . The plane waves satisfy the source-free ( jext  0 ) time-independent

Maxwell’s equations

  E   j H ,   H  j E .


 j kx x  k y y  kz z 
The space variation of a plane wave is controlled by the propagation factor e  jk r  e .

The nabla operator for a plane wave reduces to a vector:

  
 xˆ  yˆ  zˆ   jk x xˆ  jk y yˆ  jk z zˆ   jk
x y z

Thus, the Maxwell’s equations become:

k  E   H , k  H   E . (2.5)

The equation on the left implies that H is perpendicular to both k and E . Similarly, the

equation on the right shows that E is perpendicular to both k and H . Thus, the three vectors

are mutually perpendicular and we say that the wave is transverse electromagnetic (TEM). We

can write this formally as:

Ek  0  H k , EH  0 . (2.6)

The vectors E, H, k , or equivalently E, H, dˆ form a right-handed basis of space. The oscillations

of the fields are perpendicular (transverse) to the direction of propagation.

13
 Fundamental Concepts

The equations (2.5) need to be satisfied self-consistently. As shown next, this determines a

constraint on the wave vector. In fact, by eliminating the magnetic field one finds that:

k   k  E    2  E  0 .

Generic vectors A, B, C satisfy the vector identity:

A   B  C  B  A  C  C  A  B  .

Since E  k  0 , the above identity implies that k   k  E    k  k  E . Therefore, we have

shown that

   k  k  E  0
2

which is feasible only if the wave vector is constrained as

 2   k  k  k x2  k y2  k z2 . (2.7)

1
In particular, taking k  kdˆ and noting that   0 0 r r  2 n 2 with n the refractive index of
c

the material, it is found that:


k     n, n  r r . (2.8)
c

Thus, in a nondispersive medium the relation between the oscillation frequency and the wave

number is linear. The free-space wave number is typically denoted with the subscript “0”:

k0   / c .
14
 Fundamental Concepts

The link between angular frequency and the wave number implies that there is also a link

between the time and space periods of the wave. In fact, we can write:

2 2 n n
T =    
time period  k c spatial period c

Thus, it follows that:

 c
  v p  velocity of propagation . (2.9)
T n

The above formula can also be expressed as  f  v p with f the frequency in Hz.


Substituting k     n into Eq. (2.5) and using k  kdˆ , we obtain the following
c

explicit relation between the fields and the direction of propagation of the wave:

1ˆ
H dE , E   H  dˆ . (2.10)


The vector product effectively rotates the relevant field by 90º. In the above,    /  is the

intrinsic impedance (or wave impedance) of the material, which will be discussed in detail in the

following section. It is implicit that the fields are given by [Eq. (2.3)]:
ˆ ˆ
E  E0 e  jkdr , H  H 0 e  jkdr

with k    . The complex amplitudes E0 and H 0 determine the fields at the origin ( r  0 ).

Note that E0 , H 0 , dˆ must be orthogonal.

15
 Fundamental Concepts

2.4 The wave impedance

As E, H, dˆ are mutually perpendicular, it follows from the properties of the vector product that

the complex field amplitudes satisfy:

E  H. . (2.11)

Since the electric field has units of V/m and the magnetic field has units of A/m, it should be

clear that  has units of impedance. For a nondispersive dielectric it is a real-number:

 0  r r
   0 .
 0 r r

The parameter 0  0 /  0 is known as the free-space (vacuum) impedance. It is

approximately given by:

0
0   377  120    .
0

The wave impedance relates the amplitudes of electric and magnetic fields of a plane wave.

In circuit theory a real-valued impedance occurs in dissipative scenarios when the energy is

absorbed in the form of heat. For example, the impedance of a resistor is positive and real-

valued. In contrast, the wave impedance  is unrelated to Joule-type dissipation. In fact, it turns

out that nondispersive dielectrics do not absorb electromagnetic energy and are conservative

systems.

What is then the explanation for a real-valued  ? A circuit analogue is shown in the figure

below. The impedance  may be understood as some sort of “radiation” impedance. It describes

the fact that the energy stored in some small material volume dV will be transported away from

dV due to the propagation of the wave. Such a process (from the point of view of the volume

dV ) is indistinguishable from conventional Joule-type dissipation.

16
 Fundamental Concepts

2.5 Time and space evolution of the electromagnetic fields

It is instructive to analyze the space-time variation of the electromagnetic fields. To do this, we

consider an example:

E  E0 xˆ e  jk0 z , with k0    0 0   / c .

It corresponds to a plane wave that propagates along the +z-direction ( dˆ  zˆ ) in free-space. The

electric field oscillates along the x-direction. The instantaneous electric field can be found using

E  Re Ee jt  , so that

E  z , t   E0 xˆ cos t  k0 z 

For simplicity, it was assumed that E0 is a real-valued number. The figure below shows several

time snapshots of the wave along the z-axis. As seen, in one time period the wave advances

exactly one wavelength towards the +z-direction. This type of wave is known as travelling wave.

The propagation velocity is:

distance  2   1
vp      .
time T T 2 k 

This result agrees with the formula for the wave velocity obtained from the wave equation.

17
 Fundamental Concepts

1 ˆ
The magnetic field of the plane wave can be found with the help of H  d  E . For
0

nondispersive dielectrics the formula can used both for the complex amplitudes and

instantaneous fields. It gives:

1 E0
H zˆ   E0 xˆ cos t  k0 z    yˆ cos t  k0 z  .
0 0

The ratio of the field components is determined by the intrinsic impedance of the vacuum:

Ex / H y  0 . A 3D representation of the plane wave fields along the z-axis is shown below.

18
 Fundamental Concepts

3. Propagation in Lossy Media

3.1 Macroscopic electrodynamics

A natural material is formed by a very large number of atoms, which can interact with a field

applied to the material. Such an interaction originates microscopic currents, which need to be

taken into account in the Maxwell’s equations. For this purpose, it is convenient to split the

electric current density as a sum of material (microscopic) currents and external (macroscopic)

currents:

j  jmic  jext . (3.1)

The microscopic currents are determined by distinct physical mechanisms:

P
jmic  
E   
 M . (3.2)
conduction current t
 magnetization current
polarization current

The first mechanism consists of the transport of free-electrons. The conduction electrons are

free to drift through the material under the influence of an applied field. The corresponding drift

(conduction) current is determined by the electric conductivity of the material  with units of

S/m.

The second mechanism is determined by the excitation of bound electrons. Different from

the free electrons, the bound electrons are tightly attached to the atoms. The effect of an applied

field is to slightly separate the bound electrons from the atomic nucleus, i.e., to separate the

negative and positive charges, generating an electric dipole moment. A time-varying field will

induce a time-varying dipole moment and thereby a nontrivial polarization current.

Finally, the third mechanism is related to magnetization currents, which are associated with

time-varying magnetic dipoles. This mechanism will not be analyzed in detail in this course.

19
 Fundamental Concepts

The polarization and magnetization currents are described by the polarization and

magnetization vectors P and M , which enter into the macroscopic Maxwell’s equations through

the auxiliary (macroscopic) electric displacement and magnetic fields:

B
D   0E  P , H M .
0

These fields satisfy the macroscopic Maxwell’s equations

 D
E   B,   H  jext   E  . (3.3)
t t

The material (dipolar) response is “hidden” in the polarization and magnetization vectors P and

M , respectively.

3.2 Dispersive dielectrics

As already discussed in Sect. 1.3, the simplest material model is based:

D  x, y , z , t    E  x, y , z , t  , B  x, y , z , t    H  x, y , z , t  .

These constitutive relations assume that the response of the medium is instantaneous. For

example, a change in the electric field implies an instantaneous change in the polarization vector

P . Note that D   0 E  P implies that P      0  E , and thereby the polarization vector

20
 Fundamental Concepts

immediately follows any variations in time of the electric field. Such a behaviour is physically

unrealistic. In fact, the polarization vector is proportional to the mean separation r0 between the

positive and negative charges of the atoms. As the electrons are massive particles their motion is

constrained by inertia. Thus, P cannot respond instantaneously to E .

A better physical model for the electric dipole oscillations is based on the differential

equation:

2P P
 02 P   e2 0 E . (3.4)
t 2  t 
restoring force  electric force
damping force

It is known as the Drude-Lorentz oscillator model. The dynamics of P can be understood in

terms of Newtonian mechanics. As P is proportional to the displacement r0 , the term on the

left-hand side is proportional to the electron acceleration. The acceleration is determined by the

total force acting on the electron. It consists of three terms: a restoring force that acts to bring the

electron back to the equilibrium position ( r0  0 ), a damping force proportional to the velocity

which models the energy transferred to the ionic lattice, and the electric force due to the applied

field that drives the oscillation. The parameters 0 , e ,  have units of frequency.

For a time-harmonic oscillation, the polarization vector and the electric field are described by

complex vector amplitudes. Using  / t  j it is easily seen that the complex amplitudes

satisfy:

e2
P  E.
02   2  j 0

Substituting this result into D   0 E  P , we find that the electric displacement vector and the

electric field are linked in the frequency domain by a complex permittivity function:

D     E . (3.5)

For the Drude-Lorentz oscillator model, the electric permittivity is given by:

21
 Fundamental Concepts

 e2 
     0 1  . (3.6)
 0    j 
2 2

When   0 the permittivity is nearly constant and identical to    0 1  e2 / 02  . However,

for high frequencies, particularly near the resonance   0 the permittivity is sensitive to

frequency variations. The imaginary part of the permittivity is linked to the damping parameter

 . As we shall see, it leads to the absorption of an electromagnetic wave propagating in the

material. The sign of the imaginary part of    is always negative.

Similar to D and E , the induction and magnetic fields are linked by a complex magnetic

permeability    such that B     H . Substituting D     E and B     H into the

macroscopic Maxwell’s equations [Eq. (3.3)] and switching to the frequency domain, one finds

that:

  E   j   H ,   H  jext  j e   E , (3.7)

where  e   is an equivalent complex permittivity that includes the contributions from both the

conduction and polarization currents:


 e        . (3.8)
j

22
 Fundamental Concepts

It is often convenient to express the equivalent permittivity in terms of its real and imaginary

parts as  e     j  . As previously mentioned, for a passive material Im  e   0 , and thereby

   0 . The quantity

 
tan  


is known as the loss tangent of the material. One may write  e    1  j tan   . A small (large)

tan  implies a weakly (strongly) dissipative response.

3.3 Plane waves in lossy materials

The analysis of Sect. 2.3 can be readily extended to the case of lossy media described by the

parameters  e   and    . For this purpose, it is convenient to write the propagation factor

of the plane waves as

ˆ ˆ
e  jkdr  e  dr

where   jk is the propagation constant. Thus, the fields of a plane wave are

ˆ ˆ
E  E0 e  dr , H  H 0 e  dr

being d̂ the direction of propagation. Similar to Sect. 2.3, in order that the plane-wave fields

satisfy the source-free Maxwell’s equations in the lossy medium [Eq. (3.7) with jext  0 ] it is

necessary that:

1ˆ
H dE, E   H  dˆ .

23
 Fundamental Concepts

Note that these formulas imply that the fields are perpendicular to d̂ . For a dispersive material

the intrinsic impedance is given by:

     
  . (3.9)
 e      

j

Moreover, now the propagation constant satisfies  2  k 2   2     e   (compare with Eq.

(2.8)). This leads to the expression:

  
  j     e    j        . (3.10)
 j 

For lossy systems both  and  are complex-valued. A complex-valued  implies that the

fields E and H do not oscillate in phase. On the other hand, as we shall discuss in detail in the

following subsection, a complex  implies the attenuation of the wave along the direction of

propagation. It is convenient to write  in terms of its real and imaginary parts:

    j . (3.11)

The parameter  is known as the attenuation constant and the parameter  is the wave number

(or phase constant). Both parameters have units of inverse distance ( m 1 ). Similar to the

nondispersive case, the wave number  determines the distance between adjacent wavefronts,

i.e. it determines the wavelength:

2
 .


3.4 Time and space evolution of a plane wave in a lossy medium

In order to characterize the time and space variation of a wave in a lossy medium we consider an

example. For simplicity, let us take the direction of propagation along z so that dˆ   zˆ and

r  dˆ  z . For an electric field directed along x, the electric field associated with the plane wave is

of the type:
24
 Fundamental Concepts

E  xˆ E0 e  z  xˆ E0 e  z e  j  z .

The field amplitude is given by1:

E  E0 e  z .

For simplicity, the parameter E0 is taken as a positive number. The amplitude E determines the

peak oscillation amplitude at a fixed observation point:

 E  z   E  z , t   E  z  with E  z   E0 e  z .

The peak amplitude decays exponentially along the direction of propagation due to the

absorption of the wave energy by the material. The absorbed energy is dissipated as heat.

The skin depth or penetration depth  is defined as the distance that the wave needs to

propagate in order that its amplitude decays to 37% of the original amplitude: E z   0.37 E z 0 .

Taking into account that 0.37  e 1 , it follows that

E0 e   e 1 E0 .

Thus, the skin depth is given by:

25
 Fundamental Concepts

1
 . (3.12)


The skin depth provides a rough estimate of the distance that a wave can penetrate inside a lossy

material before being completely absorbed.

The electric field in the time domain is given by:

Ex  z , t   Re  E0 e  z e  j  z e jt   E0 e  z cos t   z  .

A few time snapshots of the electric field are shown in the figure below. The wavelength

  2 /  determines the distance between adjacent wavefronts, i.e., the surfaces where the

electric field oscillates in phase.

The phase velocity is determined by the dynamics of the wavefronts. It is given by:

distance  2  
vph       .
time T T 2   

For dispersive materials, the phase velocity depends on the oscillation frequency.

3.5 Group velocity

In general, in a dispersive system the phase velocity, i.e., the velocity of the wavefronts, is not

coincident with the propagation velocity. This is because the shape of a signal that propagates in

2
1 The amplitude of a complex vector is defined as E  E  E*  Ex  Ey  Ez .
2 2

26
 Fundamental Concepts

a dispersive material changes with the propagation distance. This is illustrated in the figure

below, tracking the energy associated with a certain wave “parcel” (shaded in blue) in a medium.

In the nondispersive case, the shape of the wave parcel remains unchanged after propagation.

In contrast, in a dispersive medium the shape of the parcel can change. In the example of the

figure, the field associated with the parcel is weaker at t  t0 than at t  0 . The change in

amplitude of the parcel is due to the fact that the wave exchanges continuously energy with the

medium. Hence, at some points, it gives away energy to the medium, while at other points, it

receives energy that had been previously stored in the medium at an earlier time instant. Thus,

the shape of a certain energy parcel changes over time, even in a time-harmonic regime. For this

reason, in a dispersive medium, the wave peaks are not good reference points to calculate the

propagation velocity.

27
 Fundamental Concepts

The propagation velocity in the dispersive medium is determined by the so-called group

velocity. To introduce this concept, consider an amplitude modulated signal described by:

Ein  t   E0  t  cos c t  .

The real-valued amplitude E0  t  (the envelope) varies slowly in time. The envelope is the

information signal. The carrier frequency is c .

Suppose that this signal propagates a distance z inside a dispersive material. We want to

determine the corresponding output signal Eout  t  . For convenience, we write the input signal

as:

Ein  t   Re  Ein,C  t  , with Ein,C  t   E0  t  e jct .

It is convenient to switch to the Fourier domain. The Fourier transform of Ein,C  t  is written in

terms of the Fourier transform of E0  t  as Ein,C    E0   c  . Thus, the spectrum of

Ein,C  t  is peaked at the carrier frequency. After propagation of a distance z, the output signal is

ruled by the propagation factor of the material:

Eout,C    e Ein,C    e Ein,C  

    z     z  j     z
e

As the spectrum Ein,C   is peaked at the carrier frequency   c , we can use

d
       c  ,        c      c   c  .
d

28
 Fundamental Concepts

Note that we use a zero-order approximation for the attenuation constant and a more refined

first-order Taylor approximation for the phase constant. This is justified by the fact that the

parameter  controls the propagation time. Using the above formulas, one gets:

d
 j  c  z
Eout,C    e  c z  j  c z
e e d
Ein,C   ,

with the shorthand notations  c   c  and  c   c  . Next, we switch back to the time-

domain by applying the inverse Fourier transform. The result is:

  
 j   c   c

z
  z   c  z  j  c  z j t  z 
Eout,C  t   e
 c  z vg
e  
Ein,C  t    e e e c E0  t   ,
 v  v
 g   g 

where the parameter

1
d  d
vg     , (3.13)
 d  d

is known as the group velocity. The group velocity is given by the slope of the phase constant

     .

For a real-valued input signal, Ein  t   Re  Ein,C  t  , the corresponding output is simply

Eout  t   Re  Eout,C  t  . Thus, the previous analysis yields:

 z 
Eout  t   e c z cos c t   c z  E0  t   . (3.14)
 v
 g 

The above equation confirms that the shape of the input pulse changes with the propagation

distance. In fact, Eout  t  is not simply a time-delayed attenuated version of the input field

Ein  t  . This is so because the carrier signal ( cos c t   c z  ) is controlled by the phase velocity

vph  c /  c , whereas the envelope ( E0 ) is controlled by the group velocity vg . As the profile of

the field is ruled by two different velocities, it changes with the propagation distance. The

29
 Fundamental Concepts

propagation time of the envelope (i.e., the information signal) is determined by the group delay

z / vg .

The group velocity is forcibly less than the speed of light in vacuum c. In contrast, the

amplitude (and sign) of the phase velocity are unconstrained because the spatial variations of the

fields in the medium are dictated not only by the time profile of the input field but also by the

exchange of energy between the wave and the medium, as already discussed in the beginning of

this subsection. This means that there is no exact correspondence between adjacent wave peaks

in the time domain at a fixed z and adjacent wave peaks in the space domain at a fixed t. Thus,

for dispersive materials, the phase velocity may be rather different from the actual propagation

speed. For nondispersive materials, the group and phase velocity are coincident.

3.6 Good dielectrics and good conductors

Let us consider the particular case where the dispersion of    ,    is negligible so that the

propagation constant and intrinsic impedance of the material can be written as:

    
  j      j r r 1 
 j  c j

30
 Fundamental Concepts

  1
  ,
  
 1
j j


As seen, both the impedance and the propagation constant are ruled by the parameter 1  .
j

The dissipation is fully controlled by the material conductivity. If   0 the material becomes

nondispersive and is said to be a “perfect dielectric”.

 x
When   0 but  1 the material is a “good dielectric”. Using 1  x  1  one finds
 2

that for a good dielectric:

      
 good  j r r 1  j  j r r ,
dielectric c  2  2  c

   
 good  1  j ,
dielectric   2 

Glass and wood are examples of good dielectrics.


Conversely, when  1 the material is said to be a good conductor. Metals are the typical


example of a good conductor. For a good conductor, we have

    j 4 
1   e  1  j  .
j j  2

This yields the following approximate formulas for the propagation constant and impedance:


 good  1  j 
conductor 2


 good  1  j 
conductor 2

The skin depth for a good conductor is given by:

1 2
 good   .
conductor  

31
 Fundamental Concepts

A perfect conductor corresponds to an idealized material with    . Perfect conductors are

impenetrable by light,   0 , and behave as perfect mirrors, i.e., perfect reflectors. The intrinsic

impedance of a perfect conductor vanishes:  perfect 0.

conductor

32
 Fundamental Concepts

4. Wave Polarization

4.1 Polarization curve

An oscillation is typically characterized by a frequency  , which determines the time period of

the oscillation, and by an amplitude, which determines how strong is the oscillation. For

example, an acoustic wave in air is due to a local perturbation of the density of air molecules.

The sound pressure describes the deviation from the equilibrium atmospheric pressure. The

oscillation at a certain point in space is fully characterized by the pressure amplitude and by the

oscillation frequency.

Different from acoustic waves in air, an oscillation of the electromagnetic field is not fully

determined by the field amplitude. Due to the vector nature of the field, there is an additional

degree of freedom related to the direction of space along which the field oscillates. Such degree

of freedom is known as the field polarization.

In time-harmonic regime, the instantaneous electric field at a certain point of space is related

to the field complex amplitude as:

E  t   Re E e jt  .

The closed curve determined by the time evolution of E  t  in a full time cycle (with period

T  2 /  ) is the polarization curve. It is simple to show that the polarization curve is always

planar as it lies in the plane generated by the vectors E  Re E and E  Im E . In general,

the polarization curve depends on the observation point.

The polarization curve is generically an ellipse, with the particular (degenerate) cases of a

circumference and of a line segment. The polarization of the wave at a certain point P of space is

33
 Fundamental Concepts

said to be elliptical (circular, linear) when the corresponding polarization curve is an ellipse

(circumference, line segment).

For the cases of elliptical and circular polarization, there are two different polarization states for

the same polarization curve. Indeed, there are two possible trajectories for the electric field,

corresponding to clockwise or anticlockwise rotations.

For plane wave, the field polarization is independent of the observation point. Furthermore,

as the plane waves in isotropic media are transverse waves ( E  dˆ  0 ), the polarization curve is

perpendicular to the direction of propagation ( d̂ ). In this case, the sense of rotation of the

electric field is classified with respect to d̂ , as either to the “left” or to the “right”. The sense of

rotation is found using the right-hand rule by imitating the trajectory of the electric field with the

“right-hand” and finding its orientation with the “thumb”. When the “thumb” direction is

coincident with the direction of propagation ( d̂ ), the field is said to rotate to the “right”.

Otherwise, it rotates to the “left”.

For example, the first panel of the figure below corresponds to a “right” elliptical

polarization, whereas the second panel to a “left” circular polarization. A simple way to find the

sense of rotation is to determine the time evolution (trajectory) of the electric field on the

polarization curve, e.g., by calculating E  t  0  and E  t  0  .

34
 Fundamental Concepts

To illustrate this, we consider an example corresponding to a plane wave propagating in

vacuum along the z-direction. Suppose that the complex field vector is:

E  E0  xˆ  2 jyˆ  e  jk0 z .

From the propagation factor it is evident that the direction of propagation is along –z: dˆ  zˆ .

The instantaneous field is E  t   Re E e jt  so that:


E  t   Re E0  xˆ  2 jyˆ  e
j   t  k0 z 
  E cos t    xˆ  2E cos  t  2    yˆ ,
0 0

where we denoted   k0 z and assumed that E0  0 . The instantaneous electric field lies in the

xoy plane perpendicular to d̂ . The x and y components of the field may be written as:

Ex  E0 cos t   
E y  2 E0 sin t   

The trajectory of the field can be found by eliminating t. This can be done by noting that:
2 2
 Ex   E y 
    1,
 E0   2 E0 

which corresponds to the equation of an ellipse with major axis along y and minor axis along x.

Thus, the field polarization is elliptical. The sense of rotation of the curve is determined from the

dynamics in time of the electric field. For convenience, let us consider the time instant for which

35
 Fundamental Concepts

t    0 and another slightly larger time instant for which t    0 . The electric field

evaluated at the two neighbouring time instants is given by:

 Ex  E0  Ex  E0 cos 0  E0
t    0   , t    0  


 E y  0
 
 E y  2 E0 sin 0  0

The figure below shows the polarization curve and represents the electric field vectors at the two

relevant time instants. The sense of rotation of the polarization curve is indicated by the blue

arrows. The right-hand rule yields a direction that is anti-parallel to d̂ . Thus, the wave is

elliptically polarized to the left.

4.2 Polarization ratio

Let us write the complex amplitude of the electric field as:

E  E1uˆ 1  E 2uˆ 2 , (4.1)

where uˆ 1 , uˆ 2 form an orthogonal basis ( uˆ 1  uˆ 2  0 ) of the plane perpendicular to the direction of

propagation with dˆ  uˆ 1  uˆ 2 .

36
 Fundamental Concepts

We define the polarization ratio as:

E2
p  p e j . (4.2)
E1

The amplitude ( p ) and phase  of the polarization ratio determine completely the wave

polarization according to the following rules:

 If   90º and p  1 , the wave is circularly polarized.

 If   0º or   180º or p  0 or p   , the wave is linearly polarized.

 If none of the above holds true, then the polarization state is elliptical.

The above properties are equivalent to state that:

 When E1  E 2 and arg E1  arg E 2  90º the polarization state is circular.

 When E1  0 or E 2  0 or arg E1  arg E 2 or arg E1  arg E 2  180º the polarization

state is linear.

 If none of the above holds true, then the polarization state is elliptical.

Furthermore, the angle  controls the sense of rotation, such that for 0º    180º the wave is

polarized to the left, whereas for 180º    0º the wave is polarized to the right. It is

underlined that this result holds true only when the direction of propagation is dˆ  uˆ 1  uˆ 2 .

37
 Fundamental Concepts

38
 Fundamental Concepts

5. Electromagnetic Energy and the Poynting Vector

5.1 Energy conservation law

Consider some volume V enclosed by its boundary surface  . The region inside V contains

arbitrary materials. The volume V may enclose some field sources, modelled by an external

current density jext .

Let us define EM  t  as the stored energy of the electromagnetic field in volume V at time t.

The energy EM  t  may vary in time due to different gain/dissipation mechanisms:

d EM
 pext  pdis  p . (5.1)
dt

The first term, pext , describes the power pumped into the volume V due to the external sources

within V, and contributes to increase the energy within V. The second term pdis describes the

energy dissipated (per unit of time) in the materials in the form of heat. Finally, p models the

net flow of power from V to the outside region through the boundary  . If p is positive there is

a net flow of energy exiting V, whereas if p is negative there is a net flow of energy entering

volume V. In the following subsections, we will use the Maxwell’s equations to identify the

different terms.

39
 Fundamental Concepts

5.2 Power due to the external excitation

Let us find the power pumped into the volume V due to the excitation currents jext . To begin let

us consider a single point charge with velocity v . In the presence of a field, the particle is

subject to the Lorentz force F  q  E  v  B  . From classical mechanics, the instantaneous

power transferred from the field to the charge is given by F  v :

pfield charge  qE  v

Let us now consider some volume dV in the source region. The total charge inside dV is given

by dQ  ext dV with ext the electric charge density. Hence, the power transferred from the

charges in dV to the field is ( dpext  dpcharge in dV field ):

dpext  dQ E  v  dV E  ext v .

The current density is related to the charge density as jext  ext v . This shows that

dpext  dV E  jext . The total power transferred from the external sources to the electromagnetic

field is obtained by integrating dpext  dV E  jext over the entire volume:

pext  t     dV E  jext . (5.2)

V

40
 Fundamental Concepts

5.3 The Poynting theorem

Next, we use the Maxwell’s equations to identify the remaining terms of the power balance

equation [Eq. (5.1)]. For simplicity, it is assumed that all the materials are nondispersive (  , 

are frequency independent), but we allow for conduction currents characterized by the electric

conductivity  . Thus, the time evolution of the electromagnetic field is ruled by:

H E
  E   ,   H  jext   E.
t t

The material parameters  ,  ,  may be position dependent.

The trick to obtain an energy conservation law is to evaluate the divergence of the E  H .

This vector was originally introduced by J. H. Poynting and bears his name, the “Poynting

vector”. It is denoted by the letter S:

S  E H . (5.3)

In order to calculate   S we use the auxiliary vector identity,

   E  H   H   E  E   H ,

which holds true for any two vectors. Substituting the Maxwell’s equations into the vector

identity, it is found that:

41
 Fundamental Concepts

H E
  S   H  E  E  jext   E  E
t t
. (5.4)

  WEM  E  jext   E  E,
t

with

1 1
WEM   E  E   H  H . (5.5)
2 2

Next, we integrate both members of Eq. (5.4) over the volume V. This gives:

d
   S dV   dt  W EM dV  pext  t     E  E dV ,
V V V

where we took into account Eq. (5.2).

In the last step, we use Gauss’ divergence law to transform the volume integral on the left-

hand side into an integral over the surface boundary:

   S dV   S  nˆ dA .
V 

Here, n̂ is the unit vector normal to the surface and directed to the exterior of V. Combining the

last two equations and rearranging the terms we obtain the “Poynting theorem”:

d
dt V
WEM dV  pext    E  E dV   S  nˆ dA . (5.6)
V 

Interestingly, Poynting’s result has exactly the same structure as the power balance law

discussed in the beginning of this Chapter [Eq. (5.1)]. Thus, we may identify the electromagnetic

field energy with,

EM   WEM dV , (5.7)

V

the instantaneous power dissipated in the materials with,

pdis    E  E dV . (5.8)
V

and the net power flowing through the boundary with,

42
 Fundamental Concepts

p   S  nˆ dA . (5.9)


1 1
In particular, it is seen that WEM   E  E   H  H may be understood as the density of
2 2

electromagnetic energy in the medium. It has units of Joule m 3 . Furthermore, as expected, the

dissipated power is ruled by the electric conductivity.

Remarkably, the net power exiting volume V ( p ) is given by the flux of the Poynting vector

through the boundary [Eq. (5.9)]. Thus, S has a rather powerful physical meaning: it determines

how the electromagnetic energy flows in space, i.e., it is some sort of electromagnetic energy

“current”. The units of S are Watt m 2 . When S is uniform and perpendicular to the surface, the

power that transverses the surface  is given by p  S  A with A the area of the surface.

5.4 Poynting vector in time-harmonic regime

Let us now consider time-harmonic excitations so that the electric field can be expressed as:

1
E  Re E e jt   E e jt  E* e  jt 
2 

The symbol “*” denotes complex conjugation. Similarly, the magnetic field can be written as:

43
 Fundamental Concepts

1
H  H e jt  H* e  jt 
2 

Thus, the instantaneous Poynting vector S  E  H is:

1 1
S E e jt  E* e  jt    H e jt  H* e  jt 
2   2 

Expanding the product one obtains 4 terms. Two terms are independent of time, whereas the

other two terms oscillate with a frequency 2 :

1 1
S E  H*  E*  H   E  He j 2t  E*  H*e  j 2t  . (5.10)
4   4 

The time-averaged Poynting vector is obtained by integrating the Poynting vector over one

oscillation cycle ( T  2 /  ):
T
1
S av   S  t  dt . (5.11)
T 0

The time-averaged Poynting vector determines the time-averaged power flow. For example, the

time-averaged power that transverses a certain surface  is given by:

p ,av   S av  nˆ dA . (5.12)


1
From Eq. (5.10) it is evident that S av  E  H*  E*  H  , which may also be written as:
4 

S av 
1
2
 *
Re E  H .  (5.13)

5.5 Poynting vector for a plane wave

For an electromagnetic plane wave in an isotropic medium the fields are linked to the direction

of propagation d̂ as [Eq. (2.10)]:

1ˆ
E   H  dˆ , H dE.


Thus, the time-averaged Poynting vector can be expressed as:

44
 Fundamental Concepts

1  1 ˆ * 
S av  Re E   * d  E   .
2   

Using the vector identity A   B  C   B  A  C   C  A  B  and recalling that the field is

1 ˆ 1 * 
transverse ( E  dˆ  0  E  dˆ ) one finds that S av 
*
Re d E  E  , which can also be expressed
2  * 

as:

1 1  2 ˆ 1
Re   E d  Re  H dˆ .
2
S av  (5.14)
2   2

We took into account that Re 1/   Re 1/  * . The second identity is a consequence of the

relation E   H . For lossless materials the intrinsic impedance is real-valued and thereby the

Poynting vector can be simply expressed as:

11 2ˆ 1
E d   H dˆ ,
2
S av  (lossless materials). (5.15)
2 2

As intuitively expected, the Poynting vector is aligned with d̂ . Thus, in isotropic materials the

energy flows along the direction normal to the wavefronts. The Poynting vector lines may be

identified with “light rays”.

45
 Fundamental Concepts

46
 Fundamental Concepts

6. Reflection and refraction

6.1 Boundary conditions

In this Chapter, we shall study the reflection and refraction of waves at an abrupt interface of two

different media. As is well-known, a material discontinuity may cause a discontinuity of the

some components of the electromagnetic field, typically the components perpendicular to the

interface. On the other hand, for standard materials the electric and magnetic field components

tangential to the interface remain continuous. This means that the fields on the two sides of a

material interface are linked as:

E1,tan  E2,tan , H1,tan  H 2,tan , (at the interface). (6.1)

The subscript “tan” refers to the tangential fields. For example, E tan can be expressed in terms of

the normal to the interface n̂ as:

E tan  E   E  nˆ  nˆ .

Note that the second term on the right-hand side is the normal component of the field.

47
 Fundamental Concepts

It is relevant to note that the continuity of the tangential fields at an interface guarantees the

continuity of the normal component of the Poynting vector at the interface:

S1  nˆ   E1  H1   nˆ   E1,tan  H1,tan   nˆ   E2,tan  H 2,tan   nˆ  S 2  nˆ .

Thus, the interface cannot absorb the wave energy.

It is relevant to analyze the boundary conditions at an interface with a perfect electric

conductor (PEC). A perfect electric conductor is an idealized material with infinite electric

conductivity    . As the conduction current is jcond   E , the electric field inside a perfect

conductor must vanish. Thus, an ideal PEC material behaves as a perfect mirror and is

impenetrable by light. Consequently, the electric field at the interface with a PEC material must

vanish:

E1,tan  0  E2,tan , (PEC interface). (6.2)

The tangential magnetic field H1,tan (evaluated immediately outside the material) does not need

to vanish. It determines a surface current density on the PEC boundary, given by K s  nˆ  H1 .

6.2 Normal incidence at a planar interface

Next, we consider the problem where a plane wave illuminates a planar interface that separates

two different media. For now, we restrict our attention to the case where the direction of

propagation of the incoming wave ( dˆ i ) is perpendicular to the interface, i.e., to the case of

“normal incidence”. Without loss of generality, the interface is taken as the plane z  0 and

dˆ i  zˆ . The fields of the incoming wave can be written as:

1 ˆi
 E0 e 1z , d  E0 e 1z .
inc inc inc inc
E H 
1

We used the formula H  dˆ  E /  that links the electromagnetic fields of a plane wave. In the

above,  1  j 11 and 1  1 / 1 are the propagation constant and intrinsic impedance of

region 1. Note that the permittivity and permeability may be complex-valued and depend on
48
 Fundamental Concepts

frequency. If the electric conductivity is nontrivial then the permittivity 1 must be replaced by

1  1   1 / j .

As intuitively expected, the incoming wave will generate a transmitted wave in region 2 and

a reflected wave in region 1, with directions of propagation dˆ t  zˆ and dˆ r  zˆ , respectively.

The fields associated with these waves can be explicitly written as:

1 ˆr
 E0 e   1 z , d  E0 e  1z .
ref ref ref
E H ref 
1

1 ˆt
E  E0 e  2 z , d  E0 e  2 z ,
tx tx tx
H tx 
2

where  2  j  2 2 and 2  2 /  2 .

The complex amplitudes of the transmitted and reflected fields are determined by the

boundary conditions [Eq. (6.1)] at the interface ( z  0 ). The total fields in regions 1 and 2 are:

 Einc  Eref , z0 H  H ,

inc ref
z0
E   tx , H   tx
 E , z0 H , z0

Note that the field in region 1 is a superposition of the incident and reflected waves. Since the

plane waves are transverse, both E and H are parallel (tangential) to the interface. Thus, Eq.

49
 Fundamental Concepts

(6.1) reduces to E  z  0   E  z  0  and H  z  0   H  z  0  . These two constraints yield

the following equations:

inc
E0  E0  E0 ,
ref tx
zˆ 
1

E1
inc
0  E0
ref
  zˆ  1 E
2
tx
0 .

The cross-product by ẑ may be dropped in the second equation. Thus, the problem reduces to:

inc ref
E0  E0  E0 ,
tx 1

E
1
inc
0  E0
ref
  1 E 2
tx
0 . (6.3)

Solving this system of equations with respect to the reflected and transmitted fields, one easily

finds that:

E0   E0 , E0   E0 ,
ref inc tx inc
(6.4)

with the reflection and transmission coefficients  and  ruled by the contrast of the intrinsic

impedances of the two materials:

2  1 22
 ,   1   . (6.5)
2  1 2  1

inc ref tx
Note that E0 , E0 , E0 are precisely the fields evaluated at the interface ( z  0 ).

When the impedances of the two materials are identical (2  1 ), there is no reflected wave

and the incoming wave is fully transmitted to the second medium. In such a case, one says that

the materials are “matched”.

The opposite situation occurs when the medium 2 is a perfect electric conductor (PEC). As

the intrinsic impedance of a perfect conductor vanishes (PEC  2  0 ), the reflection coefficient

in such a case becomes  PEC  1 . This confirms that a perfect conductor behaves as a perfect

mirror.

As a final observation, we note that for normal incidence the reflection and transmission

coefficients are independent of the wave polarization.

50
 Fundamental Concepts

6.3 Oblique incidence at a planar interface

Next, we consider the general case where the incident wave illuminates the interface along an

oblique direction. The incident angle i is determined by the normal to the interface n̂ and by

the direction of propagation of the incoming wave dˆ i , as illustrated in the figure. It can be found

 
using dˆ i  nˆ  cos i . By definition, the plane generated by dˆ i and n̂ is designated as the

incidence plane.

Since the fields of the incident wave oscillate in a plane transverse to dˆ i ( E  dˆ i  0 ), it is

inc

convenient to introduce two unit vectors, uˆ  and uˆ ||i , that generate this plane. By construction,

the unit vectors are such that:

uˆ ||i  uˆ   dˆ i . (6.6)

Note that the vector uˆ  is perpendicular to the incidence plane, whereas the vector uˆ ||i is parallel

to the incidence plane, consistent with the notations. The incident field can be written in terms of

uˆ ||i and uˆ  as follows:

51
 Fundamental Concepts

inc i i
E  E ||uˆ ||i  E uˆ  . (6.7)

i i
The coefficients E || , E  are referred to as the parallel and perpendicular field components,

respectively.

Similar, to the case of normal incidence, the scattering of the incoming wave by the interface
ref
generates a reflected plane wave, described by the field E , and a transmitted or refracted

tx
wave, described by the field E . It will be demonstrated in the next subsection that the

directions of propagation of these waves, dˆ r and dˆ t , are also contained in the incidence plane.

The reflection and transmission angles are determined by the conditions:

dˆ r  nˆ  cos  r ,  dˆ   nˆ  cos .

t
t

Similar to the incident wave, it is convenient to introduce unit vectors uˆ ||r and uˆ ||t such that:

uˆ ||r  uˆ   dˆ r , uˆ ||t  uˆ   dˆ t . (6.8)

The reflected and transmitted electric fields can be written in terms of its parallel and

perpendicular components as follows:

ref r r tx t t
E  E || uˆ ||r  E uˆ  , E  E ||uˆ ||t  E uˆ  . (6.9)

In subsection 6.5, we shall see how to characterize these fields.

6.4 Snell’s laws

Next, we turn our attention to the angles of reflection  r and refraction (transmission) t .

These angles can be found by imposing the boundary conditions [Eq. (6.1)] at the material

interface. For this purpose, we write explicitly the fields associated with the incident, reflected

and transmitted plane waves as

inc i i ref r r tx t t
E  E0 e  jk r , E  E0 e  jk r , E  E0 e  jk r .

52
 Fundamental Concepts

where k i , k r , k t are the wave vectors associated with the different waves. For convenience, we

express the wave vectors in terms of their Cartesian components:

k l   k xl , k yl , k zl  , l=i,r,t.

Imposing the continuity of the electric field components tangent to the interface it is found

that:

 Einc  Eref   tˆ  Etx  tˆ .

 

where t̂ is a generic direction parallel to the interface. This formula is equivalent to:

 Eint e  jk r  E0 e  j k r  E0 e  jk r   tˆ  0 .
i ref r tx t
(6.10)
 0 

It must hold true for generic interface point r . To illustrate the discussion, let us suppose that the

interface is the z  0 plane, so that r is a point of the form r   x, y, 0  . Then, the boundary

condition can be written explicitly as:

 Eint e  j  kxi x  k iy y   Eref e  j  kxr x  k yr y   Etx e  j  kxt x  k ty y    tˆ  0 . (6.11)

 
0 0 0

The only way that the above equation can be satisfied (for nontrivial fields) for all the points x

and y is if all the exponentials are identical! Thus, it follows that the projections of the wave

vector on the interface are the same for the three waves:

k xi  k xr  k xt , k yi  k yr  k yt . (6.12)

To see the implications of this result let us write the wave vectors in terms of the directions of

propagation:

k i  k1dˆ i , k r  k1dˆ r , k t  k2dˆ t . (6.13)

For simplicity, we restrict our attention to lossless materials. The propagation constants are

k1   11 and k2    2 2 . Furthermore, we suppose that dˆ i is a vector in the xoz plane, as

illustrated in the figure below. Clearly, the result (6.12) implies that dˆ r and dˆ t cannot have a y-

53
 Fundamental Concepts

component, and thereby dˆ i , dˆ r and dˆ t must all lie on the same plane, i.e., they must lie on the

incidence plane (in this example the xoz plane). The result is known as the Snell’s first law.

The directions of propagation can be written in terms of the angles shown in the figure as:

dˆ i  sin i xˆ  cos i zˆ , dˆ r  sin  r xˆ  cos  r zˆ , dˆ t  sin t xˆ  cos t zˆ . (6.14)

Using this result and Eq. (6.13) in k xi  k xr  k xt [Eq. (6.12)], it follows that:

k1 sin i  k1 sin  r  k2 sin t .

It is convenient to write k1  n1 / c and k2  n2 / c with n1   r1r1 and n2   r 2 r 2 the

refractive indices of the materials. This simplification yield the Snell’s sine (second) law:

n1 sin i  n1 sin  r  n2 sin t .

Snell’s sine law establishes that the incidence and reflected angles are identical and that the

refraction angle is controlled by the contrast of the refractive indices of the two media:

i   r , n1 sin i  n2 sin t . (6.15)

n1
From sin t  sin i it is clear that t  i when n1  n2 and that t  i when n1  n2 . The two
n2

situations are depicted in the figure below.

54
 Fundamental Concepts

6.5 Polarization of the reflected and transmitted waves

For oblique incidence, the polarization of the reflected and transmitted waves is usually different

from the polarization of the incident wave. For example, an incident circularly polarized wave

typically originates elliptically polarized reflected and transmitted waves. Moreover, the

handedness (i.e., the sense of rotation) of the field typically changes after the wave interacts with

the interface. This last property holds true even for normal incidence. For example, a left

circularly polarized wave that illuminates a dielectric interface along the normal direction

originates a reflected wave that is polarized “to the right”.

There are two particular polarization states that are preserved by the interactions with the

interface. For these special cases, the polarization of the reflected and transmitted waves is

identical to the polarization of the incoming wave. The two eigen-polarizations are referred to as

parallel (||) and perpendicular (). By definition, a wave has parallel (perpendicular) polarization

when the incident electric field is parallel (perpendicular) to the incidence plane. When the

incident wave has parallel (perpendicular) polarization, the corresponding reflected and

transmitted waves have the same property1. A wave with parallel (perpendicular) polarization is

1The perpendicular polarization is also known as the transverse electric (TE) polarization, whereas the parallel
polarization is also referred to as the transverse magnetic (TM) polarization.
55
 Fundamental Concepts

characterized by components E ||  0 and E   0 ( E   0 and E ||  0 ), respectively [see Eqs.

(6.7) and (6.9)].

It should be evident that the parallel and perpendicular polarizations correspond to a linear

polarization such that the electric field either oscillates along a direction parallel or perpendicular

to the incidence plane.

The complex amplitudes of the || and  components of the scattered fields may be related to

the || and  components of the incident wave through reflection and transmission coefficients:

E ||  || E || , E ||   || E || ,
r i t i
(at the interface) (6.16a)

E    E  , E   E ,
r i t i
(at the interface). (6.16b)

The reflection and transmission coefficients depend on the polarization and will be determined in

the next subsection.

A generic incident wave can always be regarded as a superposition of a wave with parallel

polarization with a wave with perpendicular polarization. Thereby, in the general case the

incident, reflected and transmitted fields can be written as:

56
 Fundamental Concepts

E
inc
 i
 i
 E 0||uˆ ||i  E 0uˆ  e  jk1d r ,
ˆi

   E uˆ   E uˆ  e  jk1dˆ r r
ref i r i
E || 0|| ||  0  , (6.17)

  E uˆ   E uˆ  e  jk2dˆ t r
tx i t i
E || 0|| ||  0  .

Without loss of generality, it is supposed that the origin of the coordinate system is contained at

the material interface.

6.6 Reflection and transmission coefficients for oblique incidence

Similar to Sect. 6.2, the reflection and transmission coefficients for oblique incidence can be

found by matching the field projections at the interface. To illustrate the idea, let us focus on a

perpendicular polarization described by the following fields:

ˆi ˆr ˆt
   E 0 uˆ  e  jk1d r , E    E 0 uˆ  e  jk2d r .
inc i ref i tx i
E  E 0uˆ  e  jk1d r , E

The electric fields are tangent to the interface. Hence, the boundary condition

 Einc  Eref   Etan

tx
reduces to:
  tan

1      .

We evaluated the field at the origin ( r  0 ) which is assumed to lie on the interface.

i
Furthermore, for simplicity we suppose that E 0  1 [V/m]. On the other hand, taking into

account the geometric relation between the electric and magnetic fields (see the figure on the

previous page), one sees that the second boundary condition  H  H   H tan reduces to:
inc ref tx
tan

1 1 1
cos i    cos  r    cos t
1 1 2

The cosine terms appear in the formula because of the projection of the magnetic field on the

interface. Solving the two previous equations with respect to   ,  and using  r  i one

readily finds that:

2 cos i  1 cos t
  ,    1   . (6.18)
2 cos i  1 cos t
57
 Fundamental Concepts

Thus, the reflection and transmission coefficients depend on the intrinsic impedances and on the

incidence angle. In the limit i  0 , one recovers the normal incidence result [Eq. (6.5)] because

t  0 .

The analysis for parallel polarization is completely analogous. In this case, the boundary

condition  E  E   Etan leads to:

inc ref tx
tan

cos i  || cos  r   || cos t ,

whereas the boundary condition  H  H   H tan gives:

inc ref tx
tan

1 1 1
 ||   || .
1 1 2

By combining the two equations one finds that1:

1 cos i  2 cos t 2
|| 
1 cos i  2 cos t
,  || 
1
1  ||  . (6.19)

It is interesting to analyze how the reflection coefficients || and   vary with the incidence

angle. For this purpose, we consider an interface between air (medium 1) and a lossless dielectric

with relative permittivity 10 (medium 2). As seen in the figure below, the amplitude of the two

reflection coefficients is identical for normal incidence i  0 , but follows different trends for

the two polarizations. Whereas for perpendicular polarization the reflection coefficient increases

monotonically with i , for parallel polarization one observes a reduction of the reflection level

up to some angle  B , known as the Brewster angle. At the Brewster angle the reflection

coefficient || is exactly zero (in the absence of material loss). For i   B , the coefficient ||

1 We note in passing that || i  0º      i  0º  . The reason for the awkward minus sign is related to the fact that

uˆ ||i and uˆ ||r become anti-parallel ( uˆ ||r  uˆ ||i ) in the limit i  0 . In that case || relates field components along anti-

parallel directions!

58
 Fundamental Concepts

increases monotonically with i . For grazing incidence ( i  90º ), both || and   approach

the unity so that the wave is fully reflected by the interface. In general, the level of reflections

increases as the refractive index of medium 2 increases (not shown). The Brewster angle is

further discussed in the following subsection.

It is useful to characterize || and   when the medium 2 is a perfect electric conductor. As

previously discussed, a perfect conductor has impedance PEC  2  0 . Thereby, for a perfect

conductor the two reflection coefficients are independent of the incidence angle:

  PEC  1 , || PEC  1 . (6.20)

6.7 The Brewster angle

As seen in the previous subsection, for dielectric materials with a trivial magnetic response

(   0 ), the reflection coefficient for parallel polarization vanishes for the Brewster angle. An

explicit formula for  B can be found by solving ||  0 with respect to i . From Eq. (6.19),

1 cos i  2 cos t  0 .

59
 Fundamental Concepts

For media with   0 the intrinsic impedance can be written in terms of the refractive index n

as   0 / n . Thus, the condition that determines the Brewster angle reduces to:

1 1
cos i  cos t .
n1 n2

Evidently, the transmission and incidence angles are linked by Snell’s second law:

n1 sin i  n2 sin t . By squaring both sides of the previous two formulas and using the

fundamental theorem of trigonometry one obtains a system of equations:

1 1
2
cos 2 i  2 cos 2 t , n12 1  cos 2 i   n22 1  cos 2 t  .
n1 n2

Eliminating the term cos 2 t it is found after some algebra that:

n122  1 n2
cos 2 i   2 12 ,
1 n12  1
n122  2
n12

n12 1
where n122   . Thus, our analysis shows that the Brewster angle is such that:
n22  2

1
cos 2  B  .

1 2
1

1
Finally, using the trigonometric relation  1  tan 2  , the Brewster angle can be expressed
cos 
2

as:

2
tan  B  . (6.21)
1

One curious geometric relation is that for i   B the vectors dˆ r and uˆ ||t are parallel, as

illustrated in the figure below. This can be verified noting that Brewster’s condition may be

expressed as n1 sin  B  n2 cos  B . From Snell’s second law, n1 sin  B  n2 sin t . Thereby,

60
 Fundamental Concepts

Brewster’s condition implies that cos  B  sin t , which is possible only if  B  90º t . Since

the vector uˆ ||t can be explicitly written as

uˆ ||t  cos t xˆ  sin t zˆ  sin  B xˆ  cos  B zˆ  sin  r xˆ  cos  r zˆ  dˆ r

we conclude that dˆ r  uˆ ||t for the Brewster angle. The intermediate steps use  B  90º t and

i   r   B .

The geometric condition leads to an interesting physical interpretation. The transmitted wave

generates polarization currents inside medium 2, sketched with gray arrows in the figure. The

reflected wave can be attributed to the radiation from these polarization currents. Interestingly, as

we shall see in a later Chapter, a point electric dipole cannot radiate along the direction of the

oscillation. As the electric field inside medium 2 oscillates along uˆ ||t , the polarization currents

cannot emit electromagnetic energy along that direction, i.e., cannot emit radiation along dˆ r .

This property explains why for the Brewster angle it is impossible to generate a reflected wave.

61
 Fundamental Concepts

6.8 Total internal reflection

As seen in Sect. 6.4, when a wave is refracted from a “dense” medium (e.g., water) to a less

dense medium (e.g., air), n1  n2 , the transmission angle is greater than the incidence angle,

t  i . This raises two interesting questions: Is there an angle ( i   c ) for which the

transmitted ray propagates along a direction parallel to the interface, t  90º ? If so, what

happens when the incidence angle exceeds i   c ?

The answer to the first question is affirmative. The incidence angle i   c for which

t  90º is designated as the critical angle. It can be readily found with the help of Snell’s

second law:

n1 sin  c  n2 sin 90º .

Solving for  c , one obtains:

n2
 c  arcsin , (for n1  n2 ). (6.22)
n1

Let us now turn to the second question. For i   c the transmission angle is determined by:

62
 Fundamental Concepts

n1 n
sin t  sin i  1 sin  c  1 .
n2 n2

We used the fact that sin i is a monotonically increasing function in the interval 0  i  90º .

The above expression shows that for i   c , the sine of the transmission angle exceeds 1! Thus,

t does not have an immediate geometrical meaning, it is actually a complex imaginary number.

Furthermore, the cosine of t is given by:

n12
cos t  1  sin 2 t  1  sin 2 i .
n22

For i   c , the radicand is a negative number. This implies that for an incidence angle that

exceeds the critical angle cos t is a pure imaginary number:

n12
cos t   jC , with C  sin 2 i  1  0 .
n22

ˆt
To make sense of this, let us analyze the propagation factor of the refracted plane wave e  jk2d r .

From the geometry of the figure below, dˆ t  r  sin t x  cos t z . Hence, the propagation factor

satisfies:
ˆt  jk2  sin t x  cost z 
e  jk2d r  e  e  jk1 sin i x e  jk2 cost z .

We used Snell’s second law in the middle identity. Here, k1  n1 / c and k2  n2 / c are the

wave numbers in the medium 1 and 2, respectively. Clearly, when i   c the variation of the

field along z exhibits a (real) exponential behaviour because cos t   jC . The sign that

corresponds to an exponential growth with z must be rejected on physical grounds. Thus, the

correct sign choice is such that cos t   jC .

In summary, for i   c the propagation factor in medium 2 is of the type:

ˆt n12
e  jk2d r  e  jk1 sin i x e  k2 C z , with C sin 2 i  1  0 . (6.23)
n22
63
 Fundamental Concepts

In particular, it follows that the amplitude of the transmitted field decays exponentially with

z:

E
tx
E
tx
 z  0  e k C z .
2
(6.24)

Due to the exponential decay, the wave in the medium 2 is said to be evanescent. An evanescent

plane wave is associated with trivial Poynting vector: there is no power flow from medium 1 to

medium 2. Even though, medium 2 is “transparent” (in the sense that it supports wave

propagation) it is impossible to transmit a wave from medium 1 to medium 2 when the incidence

angle exceeds the critical angle! The wave suffers total internal reflection, so that all the

incoming energy is fully reflected back to medium 1. The superposition of the incoming wave

with the reflected wave in medium 1 creates a stationary wave pattern where the wave maxima

and minima alternate.

For i   c , the reflection coefficients (for lossless materials) have exactly unity amplitude

for both polarizations:

   ||  1 , (total internal reflection). (6.25)

64
 Fundamental Concepts

Let us demonstrate this result for perpendicular polarization. Substituting cos t   jC in Eq.

(6.18) one finds that:

2 cos i  1 jC
  .
2 cos i  1 jC

As the numerator and denominator of the fraction are related by complex conjunction it is

immediate that    1 .

65
 Fundamental Concepts

66