1

Lecture notes Part II

1. Fundamental concepts - continued
1.1 Dispersion of light

The index of refraction n of a transparent dielectric medium is defined as the
ratio of the speed of an electromagnetic wave in empty space to its speed in the medium

v
c
n = . (0.1)
It is a measure of the slowing factor for light traveling in that medium. For instance, if the
index of refraction is equal to 2, the wave travels at half the speed it does in empty space.
The refractive index can be expressed as a function of the electric and magnetic
properties of the medium

r r
r r

v
c c
n
c
= = = . (0.2)
For most of the materials that are transparent to visible light

0
= (0.3)
and

r
n = . (0.4)
Hence the magnetic properties have a small effect on the light propagation. When an
electromagnetic wave is incident on a
medium, it electrically polarizes the
molecules. This changes the value of
r

which in its turn, determines the index of
refraction.
The process is wavelength
dependent: different wavelengths will
induce different polarizations of the
molecules and, as a result, and n will
be different. So, the index of refraction
changes with wavelength:





Figure 1. Normal dispersion curve of flint glass



2
( ) n n = . (0.5)
The phenomenon described above is known as the dispersion of light. Equation (0.5) is
the dispersion relation.
For most of the materials, n decreases with the wavelength (see figure1). This
phenomenon is known as normal dispersion. The decreasing is more rapid at shorter
wavelengths. This causes a curved plot of n versus . In the optical region of the
electromagnetic spectrum there exists an empirical formula, known as Cauchy's formula,
for the dependence expressed in (0.5). It reads:

( )
0 2 4

a b
n n

= + + +


, (0.6)
where
0
, , , n a b , are empirical constants whose values depend on the specific material.
For example, glass catalogues indicate ( ) n n = at certain standard wavelengths. The
following spectral lines are particularly relevant:


There also exists a regime of dispersion in which the index of refraction increases
with the wavelength and thus decreases with the frequency. This phenomenon is known
as anomalous dispersion. Anomalous dispersion occurs when the frequency of the
incident light is close to one or another of the resonance frequencies of the atoms or
molecules of a transparent medium. In other words, anomalous dispersion accompanies
the absorption of the incident light.
Please remember that any wavelength dependent phenomenon like refraction,
interference or diffraction, will lead to the dispersion of light.
Dispersion is used in optical instruments to observe and measure light spectra. A
spectroscope is an instrument for visual observation while a spectrometer is designed to
record spectra. In figure 2 a prism spectroscope is presented. Its main component is an
optical prism through which incident light is refracted and thus dispersed. A narrow
entrance slit admits the light from the source then a collimator turns it into a parallel
beam. This beam is incident on the prism and dispersed. Finally, a telescope takes the
parallel beam which exits the prism and creates images of the entrance slit. Due to
dispersion by refraction, they are located at different points for each wavelength present
in the incident light.
These images are called spectral lines. Together they make up the spectrum of
the incident light. A discontinuous (discrete) spectrum is composed of individual colored
lines separated by dark regions while a continuous spectrum is a band in which a
continuum of colors from red to violet is displayed.


3
Figure 2
Occasionally dispersion is an unwanted effect. For instance, single lens
components can produce some coloring which deteriorates the performance of an optical
instrument. This problem is known as chromatic aberration. Solving it means keeping
the red and violet rays (the extremes of the optical range) close together as they travel
through the optical instrument. This is done by using a combination of two lenses, called
an achromatic doublet, instead of single lenses.

1.2 Total internal reflection; evanescent waves

The total internal reflection is a phenomenon which occurs when the refractive
index of the medium in which light is transmitted is lower than the refractive index of the
medium in which light is incident,
t i
n n < . Consider the situation described in figure 3.
As the angle of incidence increases, the transmitted ray bends away more and more from
the normal (rays 1, 2 and 3). At a particular value of the angle of incidence which is
called critical angle (
c
), the angle of refraction becomes as large as 90

and the ray can

no longer be transmitted through the interface into the second medium (ray 4). Rays that
are incident on the boundary surface under angles higher than the critical one
c
will be
back reflected from the interface (ray 5).
4

Figure 3. Total internal reflection of light


The critical angle is obtained by applying Snells law of refraction:

i c t
sin sin90 n n =

. (0.7)
Then

1 t
c
i
sin
n
n

= . (0.8)
If the second medium is air (
t
1 n ), then the critical angle is given by

1
c
i
1
sin
n

= . (0.9)
Notice that due to dispersion, different wavelengths have different critical angles.
For a white light ray (wavelengths in the visible range) incident on a glass/air interface,
the critical angle ranges in an interval ( ) ( )
c c
violet red
,

. So, the smallest critical angle
corresponds to the highest refractive index, which is
violet
n in normal dispersion. On the
other hand, the largest critical angle corresponds to the smallest refractive index, which is
red
n . Then total internal reflection of a white light beam occurs if the angle of incidence is
larger than the largest critical value, that is ( )
c
red
, otherwise some wavelengths will be
transmitted into the second medium.
c

i c
>
t
90 =


2 1 3
4
5
i
n
t i
n n <
S
5
Figure 4. Evanescent waves at the interface between an optically dense
medium and a medium less dense

In the following we will examine the possibility that some fraction of the incident
light intensity is still transmitted in the second medium, in spite of the fact that the angle
of incidence is such that total internal reflection does occur. For this we will consider that
the plane of incidence of the light is Oxy , with Ox lying along the interface (see figure
4). For the time being we will pretend that there exists a transmitted ray which obeys
Snells law of refraction which is in a direction that makes an angle
t
with the normal at
the incidence point (keeping however in mind that
i c
> ). For reasons you will
understand in the following this transmitted light beam is not represented in figure 3.
Then the components of the transmitted wave vector
t
k
,
and those of the position
vector r
,
are

( ) ( )
( )
t t t t t t t
, , 0 sin , cos , 0
, , 0 .
x y
k k k k k
r x y

= =

,
,
(0.10)
The oscillations of the electric field intensity in the transmitted wave are described by an
equation (in the complex representation)

( )
( )
t 0t t 0t t t t t
exp exp sin cos E E i k r t E i k x k y t

= = +


,
,
(0.11)
in which we used the scalar product

t t t t t t t
sin cos
x y
k r k x k y k x k y = + = +
,
,
. (0.12)
As expected, due to the fact that
i c
> , Snells law of refraction gives

t i i i
t i c t
t t t i
sin sin sin 1 sin 1
n n n n
n n n n
= > = = > (0.13)
and accordingly, an imaginary value of the cosine function of the refraction angle:

2 2
2 2 2 i i
t t i i 2 2
t t
cos 1 sin 1 sin sin 1
n n
i
n n
= = = . (0.14)
We will however take this imaginary expression in equation (0.11) which describes the
electric field oscillations in the transmitted wave. Thus
i c
>
i
n
t i
n n <
x
y
6


( )
2
2 i
t 0t t t t i 2
t
2
2 i
0t t i t t 2
t
Amplitude of the evanescent wave
exp sin sin 1
exp sin 1 exp sin .
n
E E i k x i k y t
n
n
E k y i k x t
n


= + =





=




_
(0.15)
This is the equation of a wave which propagates in the Ox direction (meaning along the
interface) whose wavelength is
ev t t
2 sin k = (see figure 4). The amplitude of this
wave is exponentially decreasing in the direction normal to the interface.

2
2 i
ev 0t t i 2
t
exp sin 1
n
E E k y
n


=




(0.16)
Experiments show that this type of waves, which look like the result of pure
mathematical manipulation, does exist in reality. They are called evanescent (meaning
which tend to be vanishing) because of the fast decreasing amplitude.

1.3 Optical fibers

Optical fibers use total internal reflection to transmit light signals and images. A
typical optical fiber is made of a high refractive index core and a low refractive index
coating (cladding), as shown in figure 6. Some fibers are provided with EMA (extra
mural absorption), to suppress evanescent waves. Light needs to be incident
approximately parallel to the fiber axis, so that it can strike the walls at angles higher than
the critical angle of the core/cladding interface and be thus totally reflected. As a result,
light is trapped inside; once it is input, it will continue to reflect almost without loss off
the walls of the fiber (see figure 5). Optical fibers (single or bundles) are the main
components of fiber optic cables which are commonly used in telecommunications.



Figure 5. Optical path of light in an optical fiber

7
As we have already mentioned, for the light to be confined inside the fiber, the
angle of incidence to the core/cladding interface has to be greater than the limiting angle
for total internal reflection. The critical angle corresponds to a refraction angle of 90.



Figure 6. The structure of an optical fiber

In figure 6, the refractive index of the core is denoted by
1
n and the refractive index of
the cladding is
2
n . According to Snells law of refraction, at the point where light strikes
the fiber wall,

2
c
1
sin
n
n
= . (0.17)
To undergo total internal reflection light should be input at an angle less than a maximum
value corresponding to
c
. The value of sin is called numeric aperture NA.
The numeric aperture is easily computed as a function of the refractive indices of
the optical fiber. From Snells law and (0.17):

( )
1 c 1 c
2
2 2 2 2
1 c 1 1 2 2
1
NA sin sin 90 cos
1 sin 1
n n
n
n n n n
n
= = = =
= = =

(0.18)
If
2 1
n n n = , where n is a small quantity, the last member above can be rewritten as

( ) ( )
2 2 2
1 1 1 1
NA 2 2 n n n n n n n = + (0.19)
The value of NA gives the input angle range for a ray of light to be transmitted in the
fiber.
The main applications of fiber optics are in the field of medicine and
communications. In medicine, optical fibers are used to visualize various internal organs.
The fibers are tiny, flexible, have little loss and allow for straightforward construction of
the images. All these qualities make them extremely suitable for medical investigations.
In communications, optical fibers are used to transmit information at a much higher rate
than systems using radio frequencies, since the rate of transmission is related to the signal
frequency. Also, the volume of information is huge: a single fiber the thickness of a hair
can transmit audio information equivalent to 32000 voices speaking simultaneously. If
we add to that the fact that optical fibers are not subject to electrical interference and are
8
long-life devices, you will understand why they are preferred to the classic
telecommunication systems. Original patents on fiber optic transmission were granted to
John Logie Baird in the 1930's.

1.4 Optical path length; the ray model of light

Let us consider that light propagates in space between two points P and Q. The
geometrical path is the actual distance traveled by light, i.e. the distance that separates
the two points, denoted by l . The optical path length is defined as the geometrical path
length times the refractive index of the medium in which light propagates and is denoted
by L ,
L nl = . (0.20)
The optical path length in vacuum is equal to the geometrical path,
vac
L l = . In any other
medium the optical path length is longer.
Light travels at its highest speed c in a vacuum. In a medium, it is slowed down
and it travels at a speed which is n times smaller than c . Say that light takes t seconds
to travel a distance l in a medium of refractive index n . In the same time interval light
would travel a distance n times longer if it propagated in vacuum. Indeed, we have

v
l l nl L
t
c n c c
= = = = . (0.21)
So, the optical path length L is the distance that light would travel in vacuum, in the
time necessary to travel a geometrical path l in a medium whose optical properties are
described by the refractive index n . Obviously, the higher the refractive index, the
smaller the speed is and the optical path length is thus longer.
Note that this physical quantity takes into account both the actual distance
traveled by light and its speed. It is a crucial concept for understanding light propagation
in a dispersive medium. Since any medium is more or less dispersive, we understand that
optical path length is extremely useful for any piece of optics.
The optical path length determines the propagation time - see equation (0.21), the
difference in phase introduced by light traveling a geometrical path l :

vac
vac vac vac
2 2 2 2

kl l l nl L k L
n
= = = = = = (0.22)
and the number N of wavelengths that span a geometrical path l :

vac vac vac

l l nl L
N
n
= = = = . (0.23)
Note that the propagation of light in a medium can be described in terms of propagation
in vacuum, simply by taking the precaution of replacing the geometrical paths by the
corresponding optical paths.

1.5 Fermats principle

9
Fermats principle originally stated that of all the geometrically possible optical
paths that light could take between two points P and Q in a medium, the actual path is
that of least time. In other words, the propagation time has a minimum value on the real
path. The modern formulation of this principle is that of all the geometrically possible
optical paths that light could take between two arbitrary points P and Q in a medium,
the actual optical path is that of a stationary value. By stationary we mean a
minimum or maximum value. According to (0.21) the propagation time is proportional to
the optical path length, so there is no contradiction between the two statements. It is just
that the modern formulation is more general.
If the refractive index is a function of position, ( ) n n l = , then the optical path
length is defined as
d
C
L n l =

, (0.24)
where C is the curve that connects arbitrary P and Q. Fermats principle states that the
real curve
actual
C corresponds to a stationary value of the optical path length integral, that
is

actual
d
0
d
C C
L
l
=
= . (0.25)
Note that, once found, the actual optical path length stays the same for any pair of
points. Hence if light propagates from P to Q on
actual
C , it will take the same path from Q
to P. This is what the principle of reversibility of light rays states. Now we can see that it
results as a direct consequence of Fermats principle.
In the following we will apply Fermats principle to derive the laws of reflection
and refraction. The
demonstrations are simple and
elegant, avoiding cumbersome
geometrical constructions. Firstly
let us consider that light gets from
P to Q after perfect reflection by a
mirror (surface ). There are an
infinity geometrically possible
paths which meet this condition.
However path POQ is the actual
one (see figure 7). Light travels
only in air, whose refractive index
is almost unity. The total optical
path is the sum of the optical path
from P to the plane of the mirror
and the optical path from there to
Q.
Figure 7.

With the denotations in the figure, we have
( )
2
2 2 2
L p x q d x = + + + (0.26)
d

O
P
Q
1
P
1
Q
p
q
x d x

r
i

10
where L is the overall optical path length and
air
1 n was used. For the correct position
of point O, L should exhibit a maximum value, i.e.

( ) ( ) ( )
( )
1/ 2
1/ 2
2
2 2 2
d 1 1
2 2 0
d 2 2
L
x p x d x q d x
x

= + + = . (0.27)
It results

( )
( )
2 2 2
2
d x
x
p x
q d x

=
+
+
. (0.28)
Now we go back to the figure and notice that in
1
PPO . and
1
QQ O . the cosine functions
of the complementary angles of
i
and
r
can be expressed as

( )
1
i i
2 2
OP
cos 90 sin
OP
x
p x
= = =
+

(0.29)

( )
( )
( )
1
r r
2
2
OQ
cos 90 sin
OQ
d x
q d x

= = =
+

(0.30)
Consequently, if we recall equation (0.28), we see that

i r
= (0.31)
which is the law of reflection.
To get a better insight of Fermats principle, we will demonstrate the law of
refraction as well. In this case, P and Q lie in two different dielectric transparent media
separated by the surface (see figure 8), which are described by the indices of
Figure 8.
d

O
P
Q
1
P 1
Q
p
q
x

r

i

i
n
t
n
11
refraction
i
n and
t
n . The total optical path is thus the sum of the optical path
i
L of the ray
in the medium where light is incident and the optical path
t
L in the medium where light
is refracted.
( )
2
2 2 2
i t i t
L L L n p x n q d x = + = + + + . (0.32)
We apply Fermats principle:

( )
( )
( )
t i
2
2
2 2
i t
i t
2 2 2
2
d d d
d d d
d
d
d d
0
L L L
x x x
q d x
p x
n n
x x
d x
x
n n
p x
q d x
= + =
+
+
= + =

= =
+
+
(0.33)
which results in the following equation

i i t t
sin sin n n = . (0.34)
This is the law of refraction. As expected, by applying Fermats principle we obtained
that the actual path of light that crosses the boundary surface between two dielectric
media is subject to Snells law of refraction.

1.6 Huygens principle
In 1670 Christiaan Huygens used wave theory to account for the laws of
geometric optics. He suggested a way of understanding the propagation of waves by
postulating that as a wave propagates through a medium, each point on the advancing
wave front acts as a new point source of the wave and formulated a principle which is
known as Huygens Principle. It states that every point on a primary wave front serves
as a source of secondary spherical wavelets, which advance with a speed v equal to
that of the primary wave front at each point in space; the primary wave front at some
later time is the envelope of these wavelets.
Basically this is a geometrical technique for finding the path of a wave, regardless
its wavelength. All the points on the primary wave front are thus considered sources of
secondary wavelets spreading out in all directions, with a speed equal to the wave speed
in the medium. One can actually draw arcs of a circle with a radius vt , then draw the
surface tangent, or envelope, to find the new wave front at time t (figure 9).




12
Figure 9.
According to Huygens' Principle, the light rays associated with a plane light wave
in empty space propagate in straight lines. It is fairly straightforward to account for the
laws of reflection and refraction using Huygens' Principle. However, the principle fails to
explain why, for example, an expanding spherical wave continues to expand outward
from its source, rather than converging back toward the source. In other words, Huygens
overlooks the fact that the wave front formed by the back half of the wavelets implies a
light disturbance traveling in the opposite direction. Also, it does not account for
diffraction of the light into the region of geometric shadow, when light is incident on
obstacles.
Later on Augustin Fresnel, an adept of Maxwells electromagnetic theory,
amended Huygens Principle to eliminate its above mentioned flaws. The result is known
as the Huygens-Fresnel Principle which states: Every unobstructed point of a wave front
is a source of spherical secondary wavelets with the same frequency as that of the
primary. The amplitude of the resultant wave at any forward point is the superposition
of these wavelets, considering their amplitudes and relative phase. In other words, the
secondary waves mutually interfere and this accounts for light diffraction, as we are
going to show in the following chapters.







13
Figure 10.

According to Huygens-Fresnel Principle, the equation which expresses the
electric field intensity in the spherical wavelet emitted by an arbitrary elementary area
dS of an arbitrary wave front S at a point of observation P, located at the distance r
from dS , at any time t , can be written as
( ) ( ) ( )
0
0
d
d , const cos
E S
E r t t kr
r
= + , (0.35)
where
0
E and
0
are the amplitude and initial phase of the electric field oscillation,
and k are the angular frequency and the wave number of the wavelets (figure 10). The
quantity ( ) const is a constant of proportionality which depends on the angle between
the direction of the normal n
,
to the elementary surface dS and the direction of the
vector position r
,
of the observation point. It describes how waves propagate at
maximum intensity in the forward direction and cannot propagate back to the source.
Thus
( ) const 0 1 = , (0.36)
and

( )
0
const 90 0 = . (0.37)
In equation (0.35) the 1 r factor is due to the stated spherical character of the wavelets.
Their amplitude thus diminishes even when the wave propagates in a nonabsorbing
medium.
The resultant wave at P is obtained by superimposing all the secondary wavelets
generated by the wave front S:
( ) ( )
0
P 0
S S
d
E = d const cos
E S
E t kr
r
= +

. (0.38)
This equation is an analytic expression of the Huygens-Fresnel Principle. It is used to
compute the intensity of the diffracted light at any point in space and any time. The
calculation of the integral is however far from being easy. It presumes knowing the angle
for any infinitesimal surface dS belonging to the primary front wave S, at any point P

dS
r
,

S
P
n
,

14
in space. For an arbitrary surface, the calculations involve sophisticated approximating
techniques. Fortunately, where highly symmetric front surfaces are concerned, the
calculation reduces to a mere algebraic summation, as we are going to see.

2. Polarization of light
2.1 Linear, circular, elliptical polarization
The existence of polarization phenomena is a direct consequence of light being a
transverse wave. In a plane electromagnetic wave, the electric field vector oscillating in
an arbitrary direction can always be resolved into two components parallel to the
coordinate axes in the plane of its oscillation which is perpendicular to the direction of
propagation. Let x label the direction of travel of the light and y and z be directions in
the plane of oscillation. Then the components of the electric field vector of the light wave
will vibrate as
( ) ( )
0
, cos
y y
E x t E kx t = , (0.39)
( ) ( )
0
, cos
z z
E x t E kx t = +

, (0.40)
where a difference in phase between the two perpendicular disturbances was
introduced. The electric field vector is thus
( ) ( ) ( ) ( ) ( )
0 0
, , , cos cos
y y z z y y z z
E x t E x t u E x t u E u kx t E u kx t = + = + +

,
, , , ,
. (41)
According to the ratio of the amplitudes
0
0
y
z
E
E
and the difference in phase between the
component vibrations, an electromagnetic wave can be linearly, circularly or elliptically
polarized.
With linear polarization, the direction of the electric field vector resides in a
fixed plane and it stays constant at any point in space, at any time. The plane of
polarization of the electromagnetic wave is defined by the direction of the electric field
vector and by the direction of propagation. Since it has a fixed position, this type of
polarization is also called plane polarization. The electric field makes a constant angle
Figure 11. The electric field vector in a
plane linearly polarized wave
2 ( 0, 1, 2, ) m m = =
Figure12. The electric field vector in
a plane linearly polarized wave
( ) 2 1 ( 0, 1, 2, ) m m = + =
y
z
0 y
E
0z
E 0
E

0z
E
0 y
E
z
y
0
E
2
15
to the Oy axis, as in the figure 11.
The electric field vector oscillates in space and in time at the frequency of the
light wave. If the direction of E
,
is constant in space, then the changes in E
,
should be in-
step along y and z directions. Thus, for
2 ( 0, 1, 2, ) m m = = (0.42)
we have
( ) ( ) cos cos kx t kx t + =


and, according to (41), the electric field vector oscillates as
( ) ( ) ( ) ( )
0
0 0 0
, cos cos
y y z z
E
E x t u E u E kx t E kx t = + =
,
, ,
, ,
_
(0.43)
in the direction described by the angle

0
0
tan
z
y
E
E
= (0.44)
(see figure 11). Also, the amplitude of the oscillation is given by

2 2 2
0 0 0 y z
E E E = + . (0.45)

Similarly, for
( ) 2 1 ( 0, 1, 2, ) m m = + = (0.46)
the component disturbances are out of phase by radians and the electric field vector of
the linearly polarized wave oscillates as
( ) ( ) ( ) ( )
0 0 0
, cos cos
y y z z
E x t u E u E kx t E kx t = =
, ,
, ,
(0.47)
in a direction which makes an angle 2 with the Oy axis (see figure 12).
In the figures below the time evolution of the electric field intensity vector is
represented for 2 m = (figure 13) and ( ) 2 1 m = + (figure 14), respectively. It is
convenient to use the origin 0 x = of the axis of propagation as an observation point. At
that location, the oscillations in the component disturbances are described by the
equations:

0
0
cos
cos
y y
z z
E E t
E E t
=

for 2 m = (0.48)

and, respectively

0
0
cos
cos
y y
z z
E E t
E E t
=

for ( ) 2 1 m = + . (0.49)

Notice that the two projections
y
E and
z
E acquire simultaneously the maximum, zero
and minimum values, such that the angle
1 0
0
tan
z
y
E
E

=



stays constant at any instant of
time. The time evolution will be similar at any arbitrary point
0
x , since the component
16



disturbances will exhibit there a supplementary phase
0
kx (the same for each of them),
which does not change the value of the difference in phase .
0 y
E

0
E
0z
E

g) 3 / 2 t = h) 7/4 t = i) 2 t =
a) 0 t = b) / 4 t = c) / 2 t =
d) 3 / 4 t =
e) t = f) 5 / 4 t =
0
E
0 y
E
0z
E

0
E
0
E
0 y
E

0 y
E
0 y
E
0z
E
0z
E




0z
E
0 y
E
0 y
E
0z
E 0z
E
0
E
0
E
Figure 13. Time evolution of the electric field components in linearly polarized light
( 2 m = )

17
In conclusion, any arbitrary linearly polarized plane electromagnetic wave will
always be a superposition of two orthogonal linearly polarized waves, which have a
difference in phase given by (0.42) or (0.46). A linear state of polarization is
sometimes designated as a P -state, meaning a plane state of polarization.



-
-
a) 0 kx t = b) / 4 kx t = c) / 2 kx t =
d) 3 / 4 kx t = e) kx t = f) 5 / 4 kx t =
g) 3 / 2 kx t = h) 7/4 kx t = i) 2 kx t =
Figure 14 Time evolution of the electric field components in linearly polarized light
( ( ) 2 1 m = + )
2- 2-
2- 2- 2-
2-

0
E
0
E
0
E
0
E
0
E
0
E
0 y
E
0 y
E
0y
E
0y
E
0 y
E
0 y
E
0 y
E
2-
0z
E
0z
E
0z
E
0z
E
0z
E
0z
E
0z
E
18
With circular polarization, the direction of the electric field vector is no longer
fixed. In fact, it rotates in time, so that the tip of E
,
sweeps out a circular helix whose axis
coincides to the direction of propagation of the wave. The frequency of rotation is the
wave frequency. In circularly polarized light two different states of polarization are
possible: right and left. If an observer looking toward the source (the wave is approaching
him) can detect a clockwise rotation of E
,
, the wave is right circularly polarized R -
state. If the rotation is counterclockwise, the wave is left circularly polarized L-state.
An arbitrary right circular state of polarization can be expressed in terms of two
orthogonal P-states of equal amplitude
0 0 0 y z
E E E = = , whose difference in phase is
given by

2 ( 0, 1, 2, )
2
m m = + = . (0.50)
Thus the component disturbances are
( ) ( )
0
, cos
y y
E x t u E kx t =
,
,
(0.51)
and
( ) ( )
0
, sin
z z
E x t u E kx t =
,
,
. (0.52)
Let us consider the evolution in time of the component orthogonal oscillations of
the wave at 0 x =
( ) ( )
0 0
0, cos cos
y y y
E t u E t u E t = =
,
, ,
(0.53)
and
( ) ( )
0 0
0, sin sin
z z z
E t u E t u E t = =
,
, ,
. (0.54)

Notice that when ( ) ,
y
E x t
,
reaches its maximum value, ( ) ,
z
E x t
,
is null and vice versa, so
that the tip of the light vector in the R-state of polarization
( ) ( ) ( )
R 0
, cos sin
y z
E x t E u kx t u kx t = +

,
, ,
(0.55)
rotates clockwise on a circle of radius
0
E while traveling in the Ox direction as well (see
figure 15)
The L-state of polarization is a superposition of two orthogonal P-states of equal
amplitudes and a difference in phase

2 ( 0, 1, 2, )
2
m m = + = . (0.56)
Thus
( ) ( )
0
, cos
y y
E x t u E kx t =
,
,
(0.57)
and
( ) ( )
0
, sin
z z
E x t u E kx t =
,
,
. (0.58)
The resultant L-state wave field is a superposition of (0.57) and (0.58)
( ) ( ) ( )
L 0
, cos sin
y z
E x t E u kx t u kx t =

,
, ,
, (0.59)
i.e. a vector of constant amplitude
0
E which rotates counterclockwise at the wave
frequency (see figure 16).
19






b) /4 t = a) 0 t =
d) 3 / 4 t =
c) / 2 t =
e) t = f) 5 / 4 t =
g) 3 / 2 t =
h) 7 / 4 t =
i) 2 t =
Figure 16. Time evolution of the electric field components in right circularly polarized light,

2
2
m = + ; the Oy axis is horizontal, the Oz axis is vertical, the positive Ox direction points
toward the reader and is showing the direction of propagation
20

Figure 17. Time evolution of the electric field components in right circularly polarized light,

2
2
m = + ; the Oy axis is horizontal, the Oz axis is vertical, the positive Ox direction points
toward the reader and is showing the direction of propagation

e) t = d)
3 / 4
f) 5 / 4 t =
a) 0 t = b) / 4 t =
c) / 2 t =
g)
h) 7/4 t =
i) 2 t =
21
It is interesting that by overlying an R-state wave field and an L-state wave field, a P-
state wave field of the same frequency is obtained. Indeed, let us define a superposition
of a right and left state of polarization according to
( ) ( )
R L
1
,
2
E x t E E = +
, , ,
(0.60)
Then

( ) ( ) ( ) ( )
( ) ( )
( )
R L 0
0
0
1 1
, cos sin
2 2
1
cos sin
2
cos
y z
y z
y
E x t E E E u kx t u kx t
E u kx t u kx t
E u kx t
= + = + +

+ =

=
, , ,
, ,
, ,
,
(0.61)
It results that any P-state wave field can be described as a superposition of an R- and L-
state, provided that their amplitudes are equal.
The linear and circular states of polarization are particular cases of the elliptical
polarization of light. In a wave which is elliptically polarized the electric field vector
rotates at the wave frequency thus changing its direction and amplitude (as opposed to
the circular polarization, where the amplitudes remains constant). Its tip sweeps out an
elliptical helix. This state is designated as the E-state of polarization and it can be
described as a superposition of two P-states of the form (0.39) and (0.40), where the
difference in phase does not coincide with the values given by (0.42), (0.46), (0.50) or
(0.56).
To prove that the amplitude varies elliptically we eliminate the time variable
between the equations describing the component disturbances. We have
( )
0
cos
y
y
E
kx t
E
= (0.62)
and

( ) ( )
( )
0
cos cos cos
sin sin .
z
z
kx t kx t
E
kx t
E
+ =
=
(0.63)
By using the fundamental trigonometric identity and (0.62) in ( ), we obtain
( ) ( )
2
2
0
sin 1 cos 1
y
y
E
kx t kx t
E

= =



. (0.64)
Next we take equations (0.64) and (0.62) into the second equality of equation (0.63). It
results:

2
0 0 0
cos 1 sin
y y
z
y y z
E E
E
E E E

=



(0.65)
By isolating the square root and consequently squaring the obtained equation, we get
22

2 2
2
2 2
0 0 0
0 0
1 sin cos
2 cos
y y
z
y y z
y
z
y z
E E
E
E E E
E
E
E E




= +

(0.66)
which yields

2
2
2
0 0 0 0
2 cos sin
y y
z z
y z y z
E E
E E
E E E E


+ =





. (0.67)
This is the equation of an ellipse which is tilted with respect to the
y
E axis at an angle
given by

0 0
2 2
0 0
2
tan 2 cos
y z
y z
E E
E E
=

. (0.68)
For ( ) 2 1
2
m

= + , that is for perpendicularly oscillating electric field
components, equation (0.67) turns into the well-known equation of a right ellipse whose
semi-axes are given by the amplitudes
0 y
E and
0 z
E . Thus, the tip of the resultant field
vector describes an elliptical helix of equation

2
2
2 2
0 0
1
y
z
y z
E
E
E E
+ = . (0.69)

Polarization by reflection

Recall that any state of polarization can be described as a superposition of two
linear polarizations in perpendicular directions. When we discussed the reflection and
refraction of light, we considered the directions of propagation without regard of the
amount of incident light which is reflected and transmitted, respectively. The
corresponding fractions depend on the state of polarization of the incident light. The
coefficients of reflection and transmission at the boundary surface between two different
media are computed in Maxwells classical theory of electromagnetism. They are given
by the Fresnel equations. Two types of coefficients are defined. The amplitude
coefficients of reflection parallel and perpendicular to the plane of incidence are denoted
by r
|
and r

. They are expressed as

( )
i t 0r t i i t
0i t i i t i t
tan cos cos
cos cos tan( )
E n n
r
E n n

= = =

+ +

|
|
(70)

and

( )
( )
i t 0r i i t t
0i i i t t i t
sin cos cos
cos cos sin
E n n
r
E n n

= = =

+ +

, (71)
23
where
i
and
t
are the angle of incidence and the angle of refraction (recall that the
angle of reflection is equal to the angle of incidence according to the law of reflection). In
the previous equations
0i
E and
0r
E denote the disturbance amplitudes in the incident
wave and in the reflected wave. The corresponding state of polarization was designated
by using the indices | and for the parallel and perpendicular directions of polarization
with respect to the plane of incidence. In the last equality of (70) and (71), respectively
Snells law was used to eliminate the refractive indices.
The intensity coefficients of reflection in the parallel and perpendicular directions
are defined as irradiance ratios. They are also named reflectance coefficients or simply
reflectance. The corresponding Fresnel equations are

2
2 0r r
i 0i
E I
R r
I E

= = =


| |
| |
(72)
and

2
2 0r r
i 0i
E I
R r
I E



= = =


(73)
At normal incidence
( )
i t
0 and 0

= = , we obtain

2
n n t i t i
t i t i
n n n n
r R
n n n n

= =

+ +

| |

and

2
n n 2 i t i t
i t i t
n n n n
r R R
n n n n


= = =

+ +

|
,
which means that the reflectance of light whose polarization is parallel to the plane of
incidence is equal to the reflectance of
light with normal polarization. For air-
glass interface
( )
i air t
1and 1.5
glass
n n n n = = = = , the
reflectance coefficients are given by

2
1, 5 1
0, 04 4%
1, 5 1
n n
R R

= = = =

+

|
(74)
Only 4% of the incident light is thus
reflected and the rest is transmitted.
In figure 18, the dependence of the
reflectance on the incidence angle is
presented. The intensity reflectance R

of
the perpendicular state of polarization
increases monotonously from 4%- the
value corresponding to the normal
incidence ( )
i
0 = - to 100% at
0.996
1.395 10
8

rpar x ( )
rperp x ( )
1.57 0 x
0.2
0.4
0.6
0.8
1.0
0.0
20

40

60

80

R
|
Angle of incidence
R
e
f
l
e
c
t
a
n
c
e
Figure 18.
24
grazing incidence, that is
i
90

= .
R
|
has a more complex behavior: for small angles it decreases from 4% to zero,
then increases to 100% as R

does.
Equations (70) and (72) yield that the zero reflectance of the parallel polarization occurs
at a particular angle of incidence
i B
= , where

B t
90 + = (75)
(
B
is called the Brewster
angle of incidence after the
name of the scientist who
discovered this phenomenon).
As a result, light reflected at
this angle will be totally
polarized in a direction which
is perpendicular to the plane
of incidence since ( )
B
0 R

(figure 18). As far as the
transmitted light is concerned,
it is partially polarized,
predominantly in the parallel
direction, as figure 19
indicates.
Let us calculate the angle
B
as a function of the refractive indices of the two
media. According to Snells law
,
where equation (75) was used to
eliminate the angle of refraction
t
.
Then

t
B
i
tan
n
n
= (76)
For air-glass interface Brewsters angle is given by

1
B
tan 1.5 56.3

= =
The polarization of light by reflection is used in some optical devices intended to
obtaining polarized light or to the transmission of polarized light.


1. Brestwers angle polarizer is a device which is used to obtain totally polarized
light. It is composed of a stack of glass plates disposed so that the natural light is incident
at Brewsters angle (see figure 20). The light reflected by the first plate is totally
polarized in a direction perpendicular to the plane of incidence. The transmitted light is
partially polarized in a direction parallel to the plane of incidence. It is subsequently
incident on the second plate at the Brewster incidence, such that the light transmitted by
the second plate will exhibit a higher degree of polarization in a direction parallel to the
plane of incidence. The process is repeated at the incidence on the next plate. As a result,

B

i
n
t
n
Figure 19.
( )
i B t t t B t B
sin sin sin 90 cos n n n n

= = =
25
the light transmitted by the stack of plates is practically totally polarized in a direction
parallel to the plane of
incidence. The intensity of that
light is however much lower,
due to the fact that some
energy is lost at each
reflection.













2. Bresters window is a device which transmits 100% polarized light, without
losses by reflection. It is composed of a glass plate placed at Brewsters angle with
respect to the vertical direction. It is used in laser systems as an exit window for polarized
laser light. As you can see in figure 21, laser light which is vertically polarized in the
plane of incidence impinges on the exit window of a laser system at Brewsters angle.
Since there is no light polarized in the horizontal direction, the entire incident light is
transmitted into the plate and subsequently exits the window, while maintaining the
incident vertical state of polarization.
B

glass
n
B

Figure 21.

glass
n
B

Figure 20.
26
2.3 Polarization by absorption. Malus law
2.4 Optical activity. The liquid crystal display




















27
2.5 Polarization by refraction birefringence

As we could see, the propagation of light in transparent materials and at interfaces
depends on the optical properties of the materials and on the state of polarization of the
incident light. The optical properties are described by the refractive index. If a certain
material has a refractive index which is the same in any direction for a monochromatic
wave, then the medium is said to be optically isotropic. The velocity of the wave is thus
identical in any direction. Examples of optically isotropic media are: unstressed
amorphous solids, some highly symmetrical crystalline substances, liquids, and most of
the gasses.
Many crystals are however optically anisotropic. The optical properties of an
anisotropic crystal depend on the direction of propagation of the electromagnetic waves.
In other words, the refractive index which is seen by the light depends on the direction
of propagation in the crystal. A uniaxial anisotropic crystal has a single symmetry axis
which is called the optic axis and two principal indices of refraction. The optic axis is
actually defining a direction in the crystal.
Calcite(
3
CaCO , Iceland spar) is a typical anisotropic crystal. Let us consider an
arbitrarily linearly polarized wave incident on a calcite crystal.
Each ray in the incident beam is split into two rays which propagate differently in the
crystal. One ray, named the ordinary ray, obeys Snells law of refraction while the other,
named the extraordinary ray does not. The two rays propagate in the crystal at different
velocities. The optic axis is defined as the direction in which the velocities of the two
rays are equal in the material. This property is called birefringence, meaning double
O E O E
Optical axis
28
refraction. A birefringent crystal which has only one optic axis is called uniaxial. There
also exist biaxial crystals.
As a result of its birefringence, a calcite crystal will produce two different images
of an object which is placed right under it. The image which is due to the ordinary wave
will be located, as usual, right above the object. The second image, which is produced by
the extraordinary wave, is laterally shifted. If one rotates the crystal about the direction of
the incident light, the shifted image will rotate as well. Another thing that can be
observed is that a polarizer placed on top of the crystal can be oriented so that only one of
the images is transmitted at a time through it. This proves that the planes of polarization
are reciprocally perpendicular in the ordinary and extraordinary rays.
In order to explain this peculiar behavior of light propagating in an optically
anisotropic medium, let us imagine a point source of monochromatic light which is
imbedded in a uniaxial crystal. The source produces two different types of wave fronts
an ordinary wave front, which is spherical, and an extraordinary wave front, which is
ellipsoidal.
As we said, it is customary to describe the optical properties in terms of the
refractive index. A uniaxial birefringent medium has two indices of refraction: the
ordinary refractive index, and the extraordinary refractive index. The ordinary index of
refraction has a single value, which is called the principal ordinary refractive index and
is denoted by
o
n . As a result, the disturbance in the ordinary wave propagates at the same
velocity in any arbitrary direction. This explains the spherical geometry of the wave
front.
The situation is more complex where the extraordinary wave is concerned. The
refractive index is different in different directions and so is the velocity. As a result, the
geometry of the wave front is ellipsoidal, meaning that in an arbitrary time interval, the
disturbance in the extraordinary wave touches the surface of an ellipsoid. The major
semi-axis corresponds to the direction of the fastest motion and, consequently, to the
smallest value of the extraordinary index of refraction. The minor semi-axis corresponds
to the slowest motion in the direction described by the largest extraordinary index.
It results that the extraordinary refractive index takes all the values within
( ) ( )
e e
min max
, n n

. One of the limits of this range will always coincide to the ordinary
refractive index
o
n . This situation occurs along the optic axis, a direction in which the
ordinary wave and extraordinary wave travel together, at the same speed. The other limit
is named the principal extraordinary refractive index and is denoted by
e
n . This is the
value which is given in tables containing information on the optical properties of uniaxial
birefringent crystals. A quantity which is very important for the purpose of classifying
birefrigent materials is the difference n between the two principal refractive indices,
which is also named birefringence:

e o
n n n =
If
e o
, n n < then 0 n < and the crystal is negative uniaxial. If
e o
, n n > then 0 n > and
the crystal is positive uniaxial. In table 1 some uniaxial crystals data are presented.



o
n
e
n
n
29
Ice 1.309 1.313 0,004
Quartz 1.544 1.553 0,009
Calcite 1.658 1.486 -0,172
Tourmaline 1.669 1.638 -0,032
Sapphire 1.768 1.759 -0,008
Table 1

1. Negative uniaxial crystals

In such birefringent media E propagates faster than O. The figure shows the two
corresponding wave fronts. By - the direction polarization of O is indicated (it is
perpendicular to the plane of incidence). By the direction of polarization of E is
indicated (it is parallel to the plane of incidence).
According to the figure, the velocity of O is the same in any direction. We
designate it by
o
v (which stands for
ordinary
v ). The implication is that the refractive index
seen by O is also the same in any direction. The velocity of E,
e
v (which stands for
extraordinary
v ) changes with the direction of propagation. Its maximum value
( )
extraordinary e
max
v v = is reached when E propagates normally to the optic axis. The
minimum value
( )
extraordinary o
min
v v = corresponds to the propagation along the optic axis.
Thus,
negative uniaxial
extraordinary o e
v [v , v ] .
The refractive index seen by E,
extraordinary
n , is no longer constant. In fact, it is
maximum in the direction where
extraordinary
v is minimum, meaning along the optic axis,
where the refractive index is
o
n . The minimum refractive index corresponds to the
maximum velocity of E, i.e. in a direction perpendicular to the optic axis. It is denoted by
e
n . Thus,
negative uniaxial
extraordinary e o
[ , ] n n n . The two principal refractive indices can be found in
tables (see table 1). Examples of negative uniaxial crystals are: calcite, sapphire,
tourmaline.

Optic axis (dotted vertical line)
E
O
30
2. Positive uniaxial crystals










Positive uniaxial crystals are birefringent media in which the extraordinary wave
propagates slower than the ordinary wave (see the figure below). The velocity of the
extraordinary wave is thus smaller than that of the ordinary wave,
e o
v v < . This is due to
the fact that the principal extraordinary refractive index is larger than the principal
ordinary refractive index,
e o
n n > . Again the ordinary wave sees the same refractive
index in any direction. For positive uniaxial the range of the refractive index is
uniaxial positive
extraordinary o e
[ , ] n n n , meaning that the principal value along the optic axis is minimum
while the principal value in the direction perpendicular to the optic axis is maximum, as
opposed to the negative uniaxial crystals. Examples of positive uniaxial crystals are
quartz and ice. As we have already mentioned, biaxial birefringent crystals also exist:
mica, turquoise, topaz a. o. They have two symmetry axes, but their study is beyond the
goal of this course.
The properties of birefringent uniaxial crystals are used in several optical devices
to produce the splitting of the incident beam followed by its polarization (the Wollaston
polarizing beam splitter, the Nicol prism) and the rotation of incident linearly polarized
light (the half wave plate). They are also used to obtain circularly polarized light (the
quarter wave plate) or ellipsoidal polarized light (arbitrary wave plates).
It is obvious that the propagation of light in birefringent crystals depends on the
direction of polarization and on the direction of propagation with respect to the optic axis.
We shall first consider the case where natural light is incident on a plate made of a
Optic axis (dotted vertical line)
O
E
Incident wave front
O + E wave front
O
p
t
i
c

a
x
i
s

31
negative uniaxial crystal in the direction of the optic axis.

As you can see, the ordinary and the extraordinary wave fronts are traveling
together in the plate as a single wave. The separation of the different states of polarization
does not occur and the propagation in the direction parallel to the optic axis is similar to
the propagation in an isotropic material.
The situation is completely changed if the optic axis is lying in a plane
perpendicular to the direction of propagation (see the figure).
Again the direction of propagation of O and E are the same. However they
propagate at different velocities
o
v and
e
v , respectively. As a result, although initially
the two are in-phase, O will be delayed with respect to E (in a negative uniaxial crystal
plate, as calcite). If the crystal is positive uniaxial then E will lag behind O.
Such plates are called retarders. They introduce a difference in-phase between the
ordinary and extraordinary waves and thus, a phase shift. After crossing the plate, the
superimposed ordinary and extraordinary waves that are shifted in phase will form a
single wave whose state of polarization is different from the initial one. In the following
we will consider some particular plates made of calcite.

Half wave plates
After traversing a half-wave plate of thickness d, the extraordinary wave will lead
in phase the ordinary wave by an angle of :
Incident wave front
E wave front
O wave front
Optic axis
32

e o
e o e o
0
2( )
( )

n n
k k d d

= = = = , (77)
where
0
is the wavelength of the incident monochromatic wave in vacuum. This
corresponds to a difference in optical path length of half a wave (/2) and this is where
the plate gets its name from. So, the result of incident light traversing the half wave plate
is a rotation of the direction of polarization by an angle 2.
Notice that the condition for a half-wave plate to work with some specific
monochromatic incident light involves the wavelength. Indeed we have:


( )
e o
0
2

n n
= (78)
This means that half-wave plates are designed specifically for particular wavelengths.
Let us find the minimum thickness d of a half wave plate of calcite which is used
with light at
0
589nm = in air. The number of waves which span d in the extraordinary
wave is

e
e
e 0

d n d
N

= = (79)

and, correspondingly, the number
o
N is

o
o e
o 0

d n d
N N

= = > (80)

(recall that O lags behind E).
The difference between those numbers should be 1 2 for the half wave condition
to be met. Thus

o e
1
2
N N = (81)
or, equivalently

( )
o e
0
1
2
d n n
= . (82)
It results

( ) ( )
9
6 0
o e
589 10 m
1.7 10 m
2 2 1.6584 1.4864
d
n n

= =

, (83)

which is too small to obtain in practice. Fortunately the condition
0
( ) 1
2
o e
d n n

= also
works if we add multiples of 2 to the difference in phase. The next thickness is thus
obtained from the condition

( )
1 o e
0
1 5
2
2 2
d n n
= + = . (84)
Then the next possible thickness of the half-wave plate is
33

( ) ( )
9
6 0
1
o e
5 5 589 10 m
8.5 10 m
2 2 1.6584 1.4864
d
n n

= =

, (85)
which is still too small (about a tenth of the thickness of human hair), but we can
continue the procedure until we find a practicable thickness.

Quarter wave plates
Similarly, quarter wave plates introduce a difference in phase of

2
between
ordinary and extraordinary waves or, equivalently, a difference in optical path of

4
.
Recall that by superimposing linearly polarized waves of the same amplitude which are
out-of-phase by

2
, circularly polarized light is obtained. So, a quarter wave plate
produces circularly polarized light on two conditions:
1) the incident light is linearly polarized;
2) the amplitudes of the oscillations in the extraordinary and the ordinary wave
are equal (this can be done by taking

4
= ). A device designed to circularly polarize
light is composed of a linear polarizer and a quarter wave plate.
The birefringence of crystals is used in several devices as the Nicol prism. In the
figure below, a Nicol prism is presented.



It is made of calcite and consists of a rhombohedral crystal which is split
diagonally. The parts are consequently joined together using Canada balsam ( 1.53 n = ).
The ordinary wave is incident at the calciteCanada balsam interface under an angle
higher than the critical angle
c
for total internal reflection (recall that ( )
o
calcite
1.658 n = ).
The O-ray is thus totally internally reflected. Because it sees a refractive index which is
less than 1.53, (recall that ( )
e
calcite
1.486 n = ) the E-ray is transmitted at the calciteCanada
balsam interface and leaves the Nicol prism through the side opposite to the entrance
side. It will have a vertical state of polarization.

E-ray
O-ray
34
2.4 Polarization by scattering


The sunlight is unpolarized. However, when it crosses atmosphere sunlight gets
partially or totally polarized. This happens because incident unpolarized light sets the
molecules in charges to reradiate light with directions of polarization related to the
direction of acceleration of charges.

As you can see from figure above, light which is scattered at an angle of /2 is
totally polarized. Light scattered at angles smaller than /2 is partially polarized. Light
scattered in the forward direction is unpolarized.



As a consequence, light coming from the sun at dawn or at sunset is unpolarized.
Light above the head of the observer at daytime is totally horizontally polarized. Light
which comes from the sun is partially polarized. This phenomenon causes the polarized
Totally polarized
Unpolarized light
y
x
z
Totally polarized
Unpolarized light
35
sunglasses to be more effective than ordinary unpolarized glasses. The direction of
transmission of polarized sunglasses is always vertical.

3. Interference of light

3.1 Introduction

Optical interference is the phenomenon of interaction of two or more coherent
light waves in which the resultant intensity differs from the simple sum of the intensities
of the superposing waves. We have already discussed the superposition of scalar
electromagnetic waves. However, in a general approach, we need to take into
consideration their vector nature. Suppose that several linearly polarized harmonic waves
described by the electric field intensity vectors
1
E
,
,
2
E
,
, coexist in a region of space.
Thus
( )
( )
0 0
, cos
i i i i
E r t E k r t = +
, , ,
, ,
, (0.86)
where 1, 2, i = . The frequency of the disturbance in the individual waves is identical,
yet they propagate in different directions, with different propagation vectors
i
k
,
.
Let us examine the electric field vector E
,
of the resultant disturbance. According
to the principle of superposition it is equal to the vector sum of the components
( ) ( )
( )
0 0
, , cos
i i i i
i i
E r t E r t E k r t = = +

, , , ,
, , ,
. (0.87)
The purpose of the following calculations is to find out the light intensity distribution in
the interference pattern. For the sake of simplicity we will restrict them to the case of two
light sources

( ) ( ) ( )
( ) ( )
1 2
01 1 01 02 2 02
, , ,
cos cos
E r t E r t E r t
E k r t E k r t
= + =
= + + +
, , ,
, , ,
, , , ,
, ,
. (0.88)
It was already shown that the light intensity of a wave is proportional to the time
average of the magnitude of the electric field squared


2
I E

,
(0.89)
where the angular brackets denote, as usual, the time average. and are the electric
and magnetic constants of the medium where light propagates. By using (0.88) in (0.89),
we get

( )
2
2 2
1 2 1 1 1 2
2 I E E E E E E

= + = + +

, , , , , ,
, (0.90)
which can be re-written as

1 2 12
I I I I = + + . (0.91)
36
In the previous equation
12
I is a denotation for the term which involves the two
overlapping disturbances. It is known as the interference term:

12 1 2
2 I E E

, ,
. (0.92)
To understand the significance of this term, we need to pursue further
calculations.
First we will consider the scalar product
1 2
E E
, ,
and we will separate the spatially-
dependent term of the argument in the cosine functions from the time-dependent one:

( ) ( )
( ) ( )
1 2 01 1 01 02 2 02
01 02 1 01 2 02
cos cos
cos cos .
E E E k r t E k r t
E E k r t k r t
= + + =

= + +

, , , , , ,
, ,
, , , ,
, ,
. (0.93)
By using a well known trigonometric identity ( ) cos cos cos sin sin = + in the
previous expression, we get

( ) ( )
( ) ( )
1 2 01 02 1 01 1 01
2 02 2 02
cos cos sin sin
cos cos sin sin .
E E E E k r t k r t
k r t k r t

= + + +


+ + +

, , , , , ,
, ,
, ,
, ,
. (0.94)
Next we perform the multiplications

( ) ( )
( ) ( )
( ) ( )
( ) ( )
2
1 2 01 02 1 01 2 02
1 01 2 02
2
1 01 2 02
1 01 2 02
[cos cos cos
cos sin cos sin
sin sin sin
sin cos sin cos ].
E E E E k r k r t
k r k r t t
k r k r t
k r k r t t
= + + +
+ + + +
+ + + +
+ + +
, , , , , ,
, ,
, ,
, ,
, ,
, ,
, ,
, ,
. (0.95)
Now we are ready to calculate the time average of this scalar product. Recall that

2 2
1
cos sin
2
cos sin 0
t t
t t
= =
=
(0.96)
Thus, by using (0.95) and (0.96), we obtain

( ) ( )
( ) ( )
( ) ( )
{ }
( )
( )
{ }
1 2 01 02 1 01 2 02
1 01 2 02
01 02 1 01 2 02
01 02 1 2 01 02
1
[cos cos
2
sin sin ]
1
cos
2
1
cos
2
E E E E k r k r
k r k r
E E k r k r
E E k k r
= + + +
+ + + =

= + + =


= +

, , , , , ,
, ,
, ,
, ,
, , , ,
, ,
, , , ,
,
. (0.97)
or

1 2 01 02
1
cos
2
E E E E =
, , , ,
(0.98)
In the last equality we used to denote the difference in phase between the superposing
37
waves:

( ) ( ) ( )
( )
1 2 1 01 2 02 1 2 01 02
k r t k r t k k r

= = + + = +

, , , ,
, , ,
.(0.99)
depends both on the difference in optical path between the light waves at the point
were they superpose and on the initial difference in phase.
The final expression of the interference term is obtained by taking (0.98) in (0.92)

12 01 02
cos I E E

, ,
(0.100)
The result (0.100) shows that the magnitude of the interference term depends on
the electric and magnetic properties of the material in which light propagates (through the
constants , ) and on the phase difference between the components, which is
assumed to be constant in time. It also depends on the relative orientation of the direction
of vibration of the electric field intensity vector in one component with respect to the
other.
When the directions of vibration are perpendicular,
01 02
E E
, ,
, and
12
0 I = . In
other words, electromagnetic waves which oscillate in perpendicular directions will never
interfere. When such waves are superposed, the state of polarization of the resultant wave
is changed (it may be rotated or turned into circularly or elliptically polarized) but its
intensity will always be equal to the simple sum of the individual intensities.
The case of parallel directions of vibration,
01 02
E E
, ,
| , is commonly met in the
theoretical approach of the interference phenomenon. The interference term is thus

12 01 02
cos I E E

(0.101)
and, keeping in mind the fact that the individual intensities can be expressed as

2 2
1 01 2 02
1 1
,
2 2
I E I E

= =

,
it can be written in the form

12 1 2
2 cos I I I = . (0.102)
Then the light intensity distribution is given by

1 2 1 2
2 cos I I I I I = + + . (0.103)
According to this result, the light intensity exhibits dark and bright regions which
we call fringes of interference. They are not localized, meaning that they span the whole
region where the individual waves are superposed. The exact geometry of the fringes
depends on the geometry of the light sources.
If at a certain point in space the component disturbances present a constant
difference in phase, given by

max
2 , 0, 1, 2, m m = = , (0.104)
then a maximum of interference will occur at that point. We say that the waves interfere
constructively. The resultant intensity in a maximum is

( )
2
max 1 2 1 2 1 2
2 I I I I I I I = + + = + . (0.105)
On the other hand, if the difference in phase is given by
38
( )
min
2 1 , 0, 1, 2, m m = + = , (0.106)
then a maximum of interference occurs. The waves interfere destructively and the
resultant intensity is

( )
2
min 1 2 1 2 1 2
2 I I I I I I I = + = . (0.107)
The extreme values between which the intensity of light varies in the interference
pattern are used to define a parameter called the visibility V . The visibility of fringes was
first defined by Michelson as

max min
max min
I I
V
I I

=
+
. (0.108)
As we are going to see, this quantity is a measure of the degree of coherence of the
interfering waves. The visibility takes value between 0 for non-coherent light, and 1
for totally coherent light.
A particular case is that of two identical sources, such that
1 2 0
I I I = = . The light
intensity will present again minima and maxima according to the formula:
( )
2
0 0
2 1 cos 4 cos
2
I I I

= + = . (0.109)
The intensity corresponding to totally destructive interference is
min
0 I = , while the
intensity corresponding to totally constructive interference is
max 0
4 I I = . Ideally, the
pattern of interference consists in an infinite alternating sequence of dark fringes
(minima) and bright fringes of fairly equal intensity (maxima), whose visibility is
maximum 1 V = . However, as we are going to see in the following, the number of fringes
that can be observed is finite and the maximum intensity is slowly decreasing. The
appearance of the actual interference pattern strongly depends on the coherence of the
light sources.


3.2 Youngs device

As we have already mentioned, the interference phenomenon is occasionally
produced in white light and can be observed in day-to-day life. In the laboratory, the
interference of light is used in interferometer devices which can be grouped according to
the type of sources of coherent light they use: wave front splitting devices and amplitude
splitting devices.
In wave front splitting devices the secondary sources of coherent light originate in
the same primary wave front and they are obtained either by passing the incident light
through a system of two or more slits thus selecting different regions of the incident wave
front (Youngs device), or by using optical devices which yield virtual secondary light
sources (Fresnels double prism, Lloyds mirror, etc.)
Amplitude-splitting devices divide the incident light beam into two secondary
light beams which subsequently travel different optical paths before they recombine to
produce interference (the thin film, the optical wedge, Michelson interferometer, etc.)
39
Youngs device is an optical system which is basically composed of two identical
slits (circular or rectangular). It also contains an individual slit which represents the
primary source. By an adequate selection of the distance between it and the double slit
system, a plane wave front is incident on the latter. The slits act as secondary sources of
coherent light (the frequency is the same and the difference in phase is maintained
constant in time).
We denote by S the primary source and by
1
S ,
2
S the secondary sources.
Each of the sources is presumed to be a point source, regardless the actually geometry of
the opening. The distance between the two slits is d . The distance L from the double slit
plane to the plane of observation needs to be much longer than d , for the following
approximations to hold. We will denote by
1
L and
2
L the optical paths of light from the
secondary sources to an arbitrary point P where they overlap to give a maximum or a
minimum of interference.
Notice that interference occurs at any point in the region between the double slit
and the plane of observation which was considered. L cannot be however as long as
possible, because of the requirements for coherence of the overlapping waves (see the
detailed discussion of the coherence concept). We say that the fringes are not localized.
The angle between the two rays originating in
1
S and
2
S is very small since
they do not bend away too much from the direction of the incident light. If we denote by
x the coordinates of the points along the observation direction, the following
approximations hold:

1 2
, , sin , sin tan
L x
L d L L
d L

= > | , (0.110)
where L is the difference in optical path.
The interference pattern consists in a sequence of maxima and minima, according
to the difference in phase between the overlapping waves at each point of observation.
Let us compute the coordinates of the bright fringes (maxima). The condition for
constructive interference is given by the equation (0.104). The difference in phase
max
is
due to the difference in optical path at the points of coordinates
max
x , where the maxima
occur. Thus

max max max
2 2
sin 2 , 0, 1, 2, L d m m

= = = =

, (0.111)
and the condition for constructive interference is that the difference in optical path
between the interfering light waves should be an integer multiple of the wavelength

max
sin , 0, 1, 2, d m m = = . (0.112)
In the previous equation
max
denotes the angle under which the mth-order bright
fringe is formed on the screen. We can easily find the corresponding coordinate
max
x , by
considering the approximation (0.111).

max
max
sin
x
d d m
L
= = (0.113)
and consequently

max
, 0, 1, 2,
L
x m m
d

= = . (0.114)
40
The condition for destructive interference is that the difference in optical path
between the interfering light waves should be an odd integer multiple of half-waves
( )
min
sin 2 1 , 0, 1, 2,
2
d m m

= + = , (0.115)
where
min
is the angle under which the mth-order dark fringe is formed on the screen. In
a similar way, the coordinates
min
x of the points where minima occur, are obtained as

min
1
, 0, 1, 2,
2
L
x m m
d

= + =


. (0.116)
According to the equations (0.114) and (0.116), the separation x of two consecutive
maxima or minima is the same,

L
x
d

= . (0.117)
The light intensity distribution in the interference pattern is given by the equation
(0.109). For Youngs device, this can be also expressed as

2 2
0 0
4 cos 4 cos sin
2
d
I I I

= =

, (0.118)
or, equivalently
( )
2
0
4 cos
d
I x I x
L

=

. (0.119)
This expression does not describe accurately the light intensity distribution function
which can be experimentally observed. This is because in our calculations we considered
ideal point sources. In other words, we neglected the finite dimensions of the slits which
are used to produce the sources. Diffraction effects that originate in the finite size of the
slits make the actual interference pattern drop off with distance, symmetrically with
respect to Youngs device axis.
3.3 Thin film interference

The coloring you can see on the surface of a soap bubble or that of a puddle is a
result of light interference on thin films. In these examples the thin films are of
soap/water solution or of oily substances on water.
Let us consider a thin film of an arbitrary transparent material, say glass, with a
refractive index
f
n , on which a white light beam is incident. The thickness d of the film
is assumed to be constant and small. The incident light is partially reflected and partially
refracted at the air/glass interface. The refracted beam reaches the lower surface of the
film where it subsequently undergoes reflection and refraction. Light reaching the
observer is a superposition of the coherent light beams reflected from the upper
(air/glass) interface and lower (glass/air) interface - see the figure below.

41



The superposition results in an interference pattern. According to the difference in
optical path length between the light beams reflected from the top and bottom interfaces,
some wavelengths reinforce each other and thus interfere constructively. Other
wavelengths undergo destructive interference and cannot be seen. For interference to
occur, the difference in optical path length should be within the coherence length of
incident light. This explains why the optical film needs to be thin.
Notice that the difference in optical path length between ray 1 and ray 2 is almost
equal to
f
2n d , since the plate is so thin. However, when we compute the difference in
phase, we need to recall adding radians. This is the shift in phase which the beams
reflected by the upper (air/glass) interface undergo upon reflection. Note that the rays
which are reflected by the lower (glass/air) are not shifted in phase.
The observers eye collects light from different points of the film. As a result, he
sees a wide range of colors. Consider that the reflected beams originating in region A
interfere destructively for red wave lengths, for instance. Then A looks blue-greenish. On
the other hand, the difference in optical path length of the reflected rays originating in
region B is shorter so the bluegreen wavelengths interfere destructively and B looks red.
In practice all the rainbow colors can be seen.
The irregular shape of the bright interference fringes is explained by the
following:
1) the thickness of the film may be different at different points;
2) the refractive index be variable (according to the concentration of the solution,
oil, etc).
The interference of light by thin films has several important applications. It is
used, for instance, to get rid of the unwanted reflection of light in optical instruments,
where it yields light losses at lens surfaces and image distortion by multiple internal
reflections. A solution to this problem is provided by antireflection coatings. An
antireflection coating is obtained by vaporizing a dielectric substance (magnesium
fluoride or thorium fluoride) and subsequently allowing it to deposit onto an optical
substrate (glass, plastic etc.) in vacuum. The composition and the thickness of the film is
chosen such that its refractive index is a geometric average of the refractive indices of the
two media with which the film is in contact.

f air substrate
n n n = , (0.120)
Then the amplitudes of the two reflected waves that interfere destructively are equal,
resulting in a zero minimum.
Let us calculate the minimum thickness of an antireflection coating of magnesium
fluoride (
f
1.38 n = ) on flint glass ( 1.80 n = ) which will annihilate the reflection of white
42
light (in our calculations we will consider
0
550nm = , which is the average value of the
visible wavelength range). For the sake of simplicity, we will consider that the incidence
of light is normal to the film.
Note that the refractive indices are chosen such that both the reflected waves
undergo a phase shift of radians (recall that this phase shift occurs only when the
refractive index of the incident medium is less than the refractive index of the medium
where light can be transmitted). Consequently, the difference in phase between the beams
reflected by the upper and lower interfaces is due only to the difference in optical path
length:

f
f
f 0
2 2
2 2

n
k l d d = = =

, (0.121)

The condition for a minimum of interference is
( ) 2 1 , 0, 1, 2, m m = + = (0.122)
so that
( )
f
0
2
2 1 2

n
m d + = , (0.123)

and thus
( )
0
f

2 1
4
d m
n
= + . (0.124)
Notice that the thickness of the thin antireflection coating can be an odd number
of the smallest thickness

0
min
f
550nm
99.6nm
4 4 1.38
d
n
= =

, (0.125)
i.e. it can be 298.8nm, 896.4nm, etc. If the minimum value is too small to be obtained in
practice with minimum expenses, another value can be chosen from those which meet the
condition for destructive interference.
Also, you should observe that antireflection coatings are designed for definite
incident wavelengths. This is why antireflection coatings that are intended to annihilate
reflection in the whole visible range are multiplelayered.

3.4 Thin wedge interference

A thin wedge is an optical device made of two thin plates of a transparent
dielectric material (glass, for example) that are in contact at one end (the vertex of the
wedge) and slightly separated at the opposite end. The figure is an exaggerated
representation. In reality, the separation is of the order magnitude of a hair or a sheet of
paper.


43


The figure

Consider a beam of incident monochromatic light. An observer looking from
above the wedge will notice an interference pattern whose fringes are parallel to each
other and perpendicular to the plane of incidence of the light. This is the result of
superposing two coherent waves:
1) the wave reflected by the lower surface of the upper plate (glass/air interface);
2) the wave reflected by the upper surface of the lower plate (air/glass interface).
We will designate by ( ) d x the variable thickness of the wedge and we will
measure the distance x in the horizontal Ox direction, starting from the vertex of the
wedge. Then ( ) 0 0 d = (the plates are in contact at that point) and ( )
max
d L d = (the plates
furthest apart there), where L is the length of the plates.
Since the angle of inclination of the plates which make up the wedge is very
small, we can write:
( ) tan d x x x = , (0.126)
The difference in phase between ray 1 and ray 2 is:
( ) ( ) ( ) ( )
( )
0
0 0 0
2 2
2 4

d x
x k l x l x d x = + = + = + = + , (0.127)
In writing the above expression we took into account the phase shift that ray 2
undergoes upon reflection. We also approximated the difference in path lengths between
the two waves by ( ) 2d x which is valid for such small inclinations of the upper plate.
Let us find the position of the minima and maxima along the wedge. The
condition for a minimum of interference is given by(0.122). By using it in conjunction
with the expression (0.127) of the difference in phase between the interfering beams, we
get

( )
( )
( )
min min
0 0
4 2 1 4 2

d x d x
m m + = + = (0.128)
or
( )
min 0
2 d x m = (0.129)

44
This is the condition for a minimum. To find the position
min
x of the minima
along the wedge, we use the expression (0.126) of d as a function of the angle of
inclination in the condition (0.129). Thus:

min 0
2 x m = (0.130)
and

0
min

2
m
x = (0.131)
Note that the first minimum occurs at the vertex of the wedge. We expected this
result since the difference in optical path length is null there and the corresponding
difference in phase is just (the phase shift of ray 2).
The condition for a maximum is

( )
( )
max
0
max
0
4 2 , 0, 1, 2,

4 2

d x
m m
d x
m
+ = =
+ =

(0.132)
Next we divide by 2 and rearrange the terms we find
( )
max 0
1
2
2
d x m

=


(0.133)
The same condition can be written as

max 0
1
2 .
2
x m

=


(0.134)
In the previous equation the value 0 m = of the order parameter is not possible
from a physical point of view. Indeed it would mean that the maximum of order zero
occurs at a point which does not belong to the wedge. In order to get a condition which is
correct for the limit value 0 m = , i.e.
( ) 0
0
max

2
2
x = , we can translate m by a unity, and
consequently obtain a condition for the maxima

max 0
1
2
2
x m

= +


(0.135)
which is valid for 0, 1, 2, m = . Then the position of the maxima
max
x is given by

0
max
1
, 0, 1, 2,
2 2
x m m

= + =


(0.136)
According to (0.131) and (0.136), the spacing of two consecutive minima or
maxima is the same, that is

0

2
i = (0.137)
As you can notice, it varies directly with the wavelength
0
of the incident light and
inversely with the angle of inclination of the wedge. This is why an optical wedge can
be used to measure very accurately the diameter of thin wires or thicknesses of very fine
objects. Indeed, since
0
is known, by measuring the spacing i , one can easily find
and subsequently
max
d .
45
Thin wedges are also used for checking the flatness of surfaces. To this effect an
optical flat is used in conjunction with the work piece being inspected. An optical flat is a
disk of highgrade quartz glass which is approximately 2cmthick and has one side
polished, with a deviation in flatness which is less than 50nm. The optical flat and the
inspected surface touch at only one point thus forming an air wedge. If the surface is
perfectly plane, a pattern of equally spaced dark and bright fringes is obtained when the
system is illuminated with monochromatic light. If the surface being inspected is not
plane, the interference fringes exhibit irregularities which are determined by the variable
thickness of the work piece.



3.5 Newtons Rings

Newtons rings this is the name which was given to the interference pattern that
is obtained when a device consisting of an optical flat and a plane-convex lens is
illuminated by monochromatic light. Between the spherical surface of the lens and the
upper side of the optical flat an air wedge is formed (see the figure):


The figure
Contact point
of the optical
flat and probe
Dark fringes
46


The interference pattern consists in consecutive dark and bright consecutive
circular fringes, each corresponding to a certain thickness d of the wedge. In the
following we will compute 2d (the difference in optical path length between the
reflected interfering waves) as a function of the lens radius R and the distance r from the
symmetry axis to the point where d is computed. This distance is actually Newtons ring
radius.
In OAB, we have
( )
2
2 2 2 2 2
2 R R d r R d Rd r = + = + + (0.138)
which results in

2 2
2 2
2 2
d r
Rd d r d
R R
= + = + (0.139)
Since the radius of the plane-convex lens is much larger than the current thickness
of the optical wedge, R d > , the term
2
d
R
in the previous equation is quite small and it
can be thus neglected. Then the difference in optical path length between the reflected
interfering waves is given by

2
2
r
d
R
= (0.140)
Next we apply the conditions for maxima and minima of interference:

2
max
0
1
2 (maxima)
2
r
d m
R

= + =


(0.141)

2
min
0
2 (minima)
r
d m
R
= = (0.142)
Where
max
r and
min
r represent the radii of the bright and dark rings and 0,1, 2, m =
represents the order of the maxima and minima, respectively. Then the radii of the bright
and dark fringes of interference are given by

max 0
1

2
r m R

= +


(0.143)
and

min 0
r m R = (0.144)
Notice that they increase as m , which makes the rings to be more and more
closely spaced as the interference order increases (see the figure). So we cannot speak
about a constant spacing.







47
However the area between two consecutive rings of the same type (bright or dark) is
constant since it varies proportionally to
2
r , i.e. it varies proportionally to
( )
2
m m = .
Also observe that you will always get a minimum in the centre of the interference pattern.
Indeed,

( ) ( ) 0 0
0
min max

0 and .
2
R
r r = = (0.145)

3.6 The coherence of light
Coherence is a very important concept in optics and it is related to the ability of
light to yield interference and diffraction effects. The coherence term is used to describe a
light field which exhibits a fixed phase relationship between the electric field values at
different locations in the field and/or at different times. In other words it describes the
correlation between the phases of the electric field values. When the light waves behave
in a completely predictable way, as in a plane harmonic light wave, then the coherence is
complete.
Since waves are oscillatory motions in time and space, two types of coherence can
be defined: temporal coherence and spatial coherence. Temporal coherence describes
the correlation between the phases at different times, when observed at the same location.
As we are going to see in the following, temporal coherence is related to the spectral
purity of light. Ideal monochromatic light waves exhibit complete temporal coherence.
On the other hand, spatial coherence is a measure of the correlation between the
phases of the light waves at different locations, when observed at the same time. Spatial
coherence of light is related to the dimension of the light source. Ideal point sources
exhibit complete spatial coherence.
We will start this discussion of the coherence concept with the temporal
coherence of light. As we have already seen, plane harmonic electromagnetic waves
spanning the whole space at a certain frequency represent an idealized oscillatory
process. All real waves propagate as wave trains, i.e. pulses of light with a finite
extension in time and space.
Such a pulse is obviously a nonperiodic function. There exists a very powerful
mathematical tool for analyzing such functions: Fourier analysis. In Fourier analysis it is
proven that any nonperiodic function ( ) f t can be represented as a superposition of
harmonic components:
( ) ( ) d
i t
f t g e
+

, (0.146)
where the coefficient spectrum is given by the equation
( ) ( )
'
1
' d '
2
i t
g f t e t
+

. (0.147)
We discussed the concept of beats as a result of the superposition of two waves of
a slightly different frequency: at the points where the waves fall out-of-phase, a minimum
of the envelope function is produced and when the waves are back in-phase, the envelope
function exhibits a maximum. We can imagine that if several waves of slightly different
48
frequency are packed in, a longer distance in space will be required for all of them to be
in phase again and to produce the next maximum of the envelope function. Consequently,
the maxima of the envelope function will be separated by longer distances.
Now, the more frequencies are added, the longer the separation is. Thus, the
superposed wave can be thought of as a sequence of isolated wave trains. Such a function
describes a wave process which is turned on at some arbitrary moment and which is
turned off after a certain period of time, . It is overall nonperiodic (however, keep in
mind that inside the range of seconds the wave behaves, acting as a harmonic wave
at the carrier frequency). The purpose of the next calculations is to find out the
frequencies that mainly contribute to a certain wave train. Let us consider the electric
field intensity in a wave train of the form:
( )
0
0
2
0
2
i t
E e t
E t
t

>

, (0.148)
where
0
is the denotation we have already used for the frequency of the carrier wave. It
is a nonperiodic function which can be written as a superposition of harmonic
components
( ) ( ) d
i t
E t E e
+

, (0.149)
The coefficient spectrum describes the amplitude of the electric field intensity of the
resultant disturbance as a function of the frequency. According to (0.147), it is given by

( ) ( )
'
1
' d '
2
i t
E E t e t
+

, (0.150)
where ' t is the auxiliary integration variable. By taking ( ) E t defined by (0.148) into
(0.150), we get


( )
( )
( )
( )
( )
( )
( )
( )
( ) ( )
0 0
0
0 0
0
/ 2 / 2
' ' ' 0
0
/ 2 / 2
/ 2
'
0
0
0 / 2
/ 2 / 2
/ 2
'
0 0
/ 2
0 0
1
d ' d '
2 2
d '
2
2 2
i t i t i t
i t
i i
i t
E
E E e e t e t
E
e i t
i
E E e e
e
i i
+ +


+

= = =

= =

= =

, (0.151)

The coefficient spectrum can be rewritten, by using the Euler formula

sin
2
i i
e e
i

= , (0.152)
to obtain the final expression
49

( )
( )
( )
( )
( )
0
0
0
0
0
0
sin / 2
sin / 2
.
2 / 2
E
E
E
= =

=

(0.153)
In the previous equation, one can identify a function which is frequently used in
optics, known as sincu . By definition

sin
sinc
u
u
u
= . (0.154)
A representation of this function is given in figure Its central peak is given by

0
limsinc 1
u
u

= (0.155)
Also, the zeroes of this function occur at the points where sin 0 u = , i.e.
sinc 0, for , 0, 1, 2, u u m m = = = (0.156)
In the following we will use the definition (0.154) to get the frequency spectrum
of the electric field amplitude coefficients. Thus
( ) ( )
0
0
sinc / 2
2
E
E

=

(0.157)
Its maximum value is proportional to the amplitude which, for simplicity, we assumed to
be constant and to the lifetime of the wave train.

0
max
2
E
E

=

(0.158)
The maximum value corresponds to
0
= .
0
is a symmetry point for the frequency
spectrum of the electric field amplitude coefficients.
The zero-amplitude points of the wave train occur at the frequencies

0
2
, 0, 1, 2,
m
m

= + =

(0.159)
Note that the secondary maxima and minima of the frequency have much smaller
amplitudes than the central maximum.
Let us recall that the energy which is carried by a wave is proportional to the
squared amplitude of the electric field magnitude. It means that the squared amplitude
spectrum is actually a measure of spectral distribution of the energy per unit time per unit
surface carried by the wave train. That is why the quantity ( )
2
E is called the power
spectrum. You can examine a graph representation of the amplitude and of the power
spectrum in the figure. Notice that the largest amount of energy is transported at the
central frequency
0
and at the nearby, symmetrically disposed frequencies, i.e. it lies
within the central maximum of the power spectrum. The width of this maximum is
considered between the two zeros of the first order on the right and on the left of
0
, that
is
4

. In signal analysis the half-width of the central maximum

2
=

(0.160)
50
is chosen as an indicative of the frequencies having the most important contribution to the
considered wave train. Bear however in mind the fact that the actual wave train is a
superposition of much more plane harmonic components; it is just that the energy they
carry is insignificant as compared to the energy of the components whose frequencies are
lying within the range, which is called frequency bandwidth of the harmonic wave
train.
Notice that varies inversely to the wave train lifetime . In the limit
the bandwidth tends to zero and the wave train is reduced to a single monochromatic
plane component which propagates from to + at the central frequency
0
. On the
other hand, in the limit of extremely short lifetime wave trains, 0 , the frequency
bandwidth formally becomes infinity, , which means that the number of spectral
components with an important contribution to the wave train is extremely high. In-
between these limit cases, we should be aware of the fact that long light pulses are made
up of a small number of spectral components, i.e. they exhibit a high spectral purity.
Laser light is a good example for this situation. On the other hand, we expect white light
to propagate as very short wave trains, with a large number of significant components.
The time of coherence
c
of light from a monochromatic source is given by the
average lifetime of the emitted wave trains. According to (0.160), the time of coherence
varies inversely with the average frequency bandwidth:

c
2 1
f

= =

. (0.161)

The time of coherence is thus longer for spectrally pure light sources as lasers. Let us
express the time of coherence as a function of the wavelength bandwidth . First we
differentiate the equation which relates the frequency to the wavelength,
c
f =

:

2
c
f =

, (0.162)
then we take this into the expression of the coherence time (never mind the minus sign, it
simply shows that an increase in frequency is equivalent to a decrease in wavelength):

2
c
1
f c

= =

. (0.163)
This is the equation we were looking for. It shows that the time of coherence is
proportional to the average wavelength in the wave train but it varies inversely with the
spectral width.
The distance traveled by light in a time interval equal to the time of coherence is
called the length of coherence. Over a distance equal to the length of coherence, the
waves behave predictably and there is a fixed phase relationship between the disturbances
at different locations. In other words, light is coherent and it can produce interference or
diffraction effects. At a distance longer than that, the waves undergo arbitrary shifts in
phase, thus becoming incoherent. Interference effect can no longer be observed.
The length of coherence is given by

2
c c
c
l c
f

= = =

. (0.164)
51
Let us compute the length of coherence for some typical light sources:
- white light: 550nm , 300nm ,
( )
2
c
550
nm 1000nm 2
300
l =
- gas discharge lamps
green line of Hg lamp, 546nm , , 0.025nm

( )
2
c
546
nm 1.2mm
0.025
l =
orange line of Kr 86 lamp, 606nm , , 0.00047nm

( )
2
c
606
nm 78cm
0.00047
l =
- lasers

2
CO laser, 10.6 m = , ,
5
10 nm

c
11km l
As you can notice, the spectral purity does indeed determine the coherence of light.


4. Lasers
The word LASER is an acronym for Light Amplification by Stimulated Emission
of Radiation. By extension it became a noun which designates a source of optical
radiation. The laser is thus an optical device which performs light amplification by using
a radiative process called stimulated emission. It provides an intense, highly collimated
beam of coherent light. The first laser was built by T. Maiman in 1960.
The story of lasers started in 1916, when A. Einstein predicted the stimulated
absorption and emission processes and computed their probabilities. Although the
physics was all there, it took some time before C.H. Townes and co-workers developed a
microwave amplifier (maser) in 1954. After A. Schawlow and C.H. Townes extended its
working principle to the visible domain of the electromagnetic spectrum (1958), the
advent of the first laser was finally made possible. Maimans device was a solid state
(ruby) laser, operating at 694.3 nm.
Type Wavelength Power/energy Output
Ruby (solid) 694.3 nm 0.03-100 J/pulse pulsed
Nd:YAG (solid) 1.064 m 0.04-600 W cw
Nd-glass (solid) 1.06 m 0.15-100 J/pulse pulsed
Ho-YAG (solid) 2097 nm 6.4 W cw
He-Ne (gas) 632.8 nm 0.1-50 mW cw
CO
2
(gas) 10.6 m 3-100 W cw
N
2
(gas) 337 nm 1-300 mW/pulse pulsed
Dye (liquid) 400-900 nm 20-800 mW cw/pulsed
HF (chemical) 2.6-3 m 0.01-150 W cw/pulsed
ArF (excimer) 193 nm Up to 10 W pulsed
GaAs (semiconductor 780-900 nm 1-40 mW cw/pulsed
52
diode)
After 1960, the development of lasers was extremely fast and successful. Even
today new laser media are discovered; new wavelengths and ranges of power become
available. In the table above some of the most common lasers are presented. As for the
field of applications, we can say without fear of exaggeration that it is huge.
As we have already mentioned, stimulated emission is a process in which an atom
which lies in a higher energy state (
2
E ) decays to a lower energy state (
1
E ) by emitting a
photon of frequency
2 1
E E
f
h

= when a photon of resonant frequency f impinges upon

it. The emitted photon will be a twin brother of the incident photon, meaning it will
have the same energy, direction, phase and polarization. One photon is incident on the
atom and the stimulated emission results in two identical photons. This suggests that
amplification of incident radiation is possible.
One of the practical problems which arise is how to obtain more atoms on the
higher level than on the lower level, otherwise atoms in
1
E state would absorb the
incident photons and no amplification would occur (Einstein proved theoretically that the
two processes: stimulated emission and absorption have the same probability). Such a
situation is unnatural; you need to remember that most of the atoms lie in the ground state
or in low energy states at room temperature. For laser effect to occur a population
inversion has to exist between the two laser levels. In other words, atoms should be
pumped to the higher energy level. This is done by using an external source of energy, as
we shall see in the following. Another problem is to maintain the population inversion
long enough for the stimulated emission processes to multiply the input photons. If
2
E
and
1
E were ordinary excited states, it would take only
8
10 s

for the atoms to decay to

lower levels. The solution is to find a type of atom where the upper level of the laser
transition is actually a metastable state. We will see that this requirement is met by
selecting certain amplifying media.
Supposing the previously problems are solved, we are still faced with the need to
get as many stimulated decay processes as possible, i.e. to make the path of photons as
long as possible. This is done by placing the amplifying medium between two highly
reflecting mirrors actually an optical resonator.
The argument above was meant to present the three main components which
should be found in any type of laser system: 1) pump (an external energy source); 2)
amplifying (gain) medium; 3) optical cavity (resonator). In the following we will present
some of their characteristics.
53

The pump is an external source which provides the necessary energy to obtain the
population inversion. Its nature depends on the type amplifying medium a certain laser
uses; it can be optical, electrical or chemical (there is a type of laser whose pumping
energy is the result of a nuclear fusion reaction, but that is quite exotic). Optical pumping
is done by using flashtubes, arc lamps or even laser diodes. It is a type of pumping
suitable for solid state and liquid state lasers. By absorbing the incident photons, the
atoms or molecules in the gain medium make the transition to high energy levels, called
pumping levels, then get to populate the upper metastable laser level by fast nonradiative
transitions. Electrical pumping is adequate for gas, semiconductor and some excimer
lasers. It is done by using continuous electric discharges (gas lasers), electric currents
(semiconductor lasers) or electron beams (excimer lasers). Again the laser atoms or ions
use the external electric energy to populate the upper laser level. In chemical lasers the
population inversion is obtained by using the energy resulted from a chemical reaction.
The process is known as chemical pumping.
The amplifying media are extremely diverse (it is significant that more than half
of the known chemical elements were proved to have laser levels). The wavelength of the
laser output is essentially dependent on the gain medium it uses. Because of its
importance, most of the commonly used lasers are named after the amplifying medium.
Solid, liquid and gaseous gain media are available. As a general rule, the amplifying
medium is complex. It is composed of a laser host medium and laser atoms / ions. The
role of the host medium is to take over the excess energy of the pumped laser atoms.
This makes the required population inversion possible. The laser atom/ion needs to have
specific metastable laser levels which represent upper laser levels (if they were short
lived, the gain medium could not sustain the population inversion long enough for
amplification).
The optical cavity is essentially an optical resonator made of two mirrors, one
perfectly reflecting (as close as possible to 100% reflectivity) and another weakly
transmitting (99% reflectivity). The space between the two mirrors is fully or partially
filled with the amplifying medium. The role of the resonator is to direct the useful
54
photons in a motion back and forth along the axis of the laser system, allowing for the
build up of the laser output (which consist in photons that escape through the partially
transmitting mirror). The geometry and alignment of the optical cavity have a crucial
effect on the exact distribution of the intensity pattern across the output laser beam.
For historical and pedagogical reasons we begin the presentation of the working
principle of lasers with the first laser the ruby laser. In a ruby laser the population
inversion is achieved by optical pumping. A helical flash lamp (with Xenon, for example)
surrounds the amplifying medium consisting in a ruby rod. The short (
3
10 s

) bursts of
intense light supply the necessary energy of excitation. Ruby is a red, transparent crystal
of aluminum tri-oxide (
2 3
Al O ) which contains about 5% of chromium ions (
3+
Cr ). The
simplified energy level patterns of
3+
Cr (laser ions) are presented in the figure below.
As you can see, the
3+
Cr ions have two absorption bands in the green and blue
regions of the visible range (this explains why the ruby is red) which can be used as
pump levels. Below those there exist two metastable states, collectively labeled
2
E in the
figure. Because the pumping levels are short-lived, the laser ions which are pumped there
will soon decay to
2
E , by giving off some of their energy to the host medium (
2 3
Al O ) in
fast, radiationless relaxation processes. The population of level
2
E thus builds up and a
population inversion between
2
E and
1
E (the ground state) occurs. The above mentioned
levels are in fact the upper and the lower levels of the laser transition. Since it basically
involves three levels (the ground state, the pump level and the metastable laser level) the
ruby laser belongs to the category of three-level lasers.
55
Some of the atoms lying in the energy state
2
E will spontaneously emit photons
of frequency
2 1
E E
f
h

= and thus decay to the lower level. As we have already

mentioned, the direction of the photons which are obtained by spontaneous emission is
arbitrary, so we expect that a number of them will be lost through the side walls of the
laser system. However, there will always be some photons which are emitted in the right
direction, along the axis of crystal. We call them useful photons because they initiate
stimulated emission processes when passing by
3+
Cr ions lying in the upper laser level.
The optical cavity is created by silvering the flat ends of the ruby bar. As long as the
pump maintains the population inversion, the photons traveling back and forth along the
axis of the optical cavity will be multiplied by further stimulated emission processes. A
small fraction of them escape the cavity through the weakly transmitting mirror. They
build up the output laser beam which consists of short pulses (a few milliseconds) with
energies up to 100 J in modern ruby lasers.
The population inversion in a three level laser requires very intense pumping that
is high input external energy because the lower laser level is the ground state which
naturally tends to be highly populated. A different type of laser where the lower laser
level is actually an excited state also exists. Such a laser works on a four-level scheme.
An example is the helium- neon laser. In the figure below a simplified diagram of the
energy levels involved in its working is presented. Helium is the host medium while the
neon atoms constitute the laser medium. The pump energy is produced by an electrical
discharge in the low-pressure gas mixture. The helium atoms lying in the ground state are
excited to the high pump energy level denoted by
2He
E . Because neon happens to have an
56
energy level
3Ne
E close to it, excited atoms of helium give off their energy in resonant
collisions to unexcited neon atoms and then return to their ground state
1He
E . The process
of pumping is continued as long as the discharge exists.
As for the atoms of neon which are excited to
3Ne
E (which is a metastable upper
laser level), they build up the population inversion between
3Ne
E and
2Ne
E . The latter acts
as a lower laser energy level and it is also an excited state. Consequently, it is fast
depleted in spontaneous decay processes. The laser transition occurs between
3Ne
E and
2Ne
E . It gives rise to photons with 632.8nm = , which represents the red radiation of
what is perhaps the most common type of laser. The He-Ne laser is classified as a four
level laser because it essentially involves the following four levels: the ground energy
level (close together for both types of atoms), the pump level of the helium atoms, the
upper and lower laser levels belonging to the neon atoms.
The properties which make lasers a unique source of light consist in
monochromaticity, high coherence, directionality and high levels of intensity. The
remarkable number of laser applications is due to these unmatched properties of laser
light. The interaction of laser light with matter is used in materials processing (cutting,
drilling, heat treating, welding etc), medicine (cutting, vaporizing, cauterizing,
coagulating, destroying unwanted cells, healing, treating ocular disorders), applied
science (laser spectroscopy, imagery, laser-driven nonpolluting sources of energy, etc).
Another wide range of applications use lasers to transmit, detect, store and process
information. Telecommunications, information processing, optical sensing, optical
storing, metrology, holography are only some of these fields.