Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

5. INTERFERENCE.

Introduction.
Previously, we have spent time considering the amplitude of an electromagnetic wave
and have focused on its vector quality when discussing polarization. The vector nature
of the electric field is an important property because it is frequently the case that we
have several electromagnetic waves from different sources or more frequently different
parts of the same source, and need to find the resultant field in a particular region of
space by invoking the superposition of fields. In that superposition we obviously need
to take account of the vector nature of the field by formally superposing those fields
using vector addition.
In the previous analysis of polarisation we noted the change in direction of the electric
field vector as a result of superposition and that this lead to different polarization states,
left and right handed elliptically and circularly polarised light and plane polarised light.
In this section our interest is to be the change in amplitude of the electric field vector (or
equivalently light intensity) as a result of superposition or addition of waves from
multiple sources or from one source first split and then later recombined at some new
location. This brings about a variety of effects grouped collectively under the title of
interference; effects including, two slit interference (Young’s Slits), thin film
interference and various interferometers.
Diffraction phenomena are closely related to interference phenomena and are dealt with
using the tools that are developed for interference in what follows. We will however
leave diffraction as a topic to be considered separately.
When considering interference it is convenient to continue using plane waves to
describe the electromagnetic waves as these are more easily manipulated
mathematically. It will later be important to consider how useful this plane wave
approach may be in describing reality and this will bring us on to the concept of
coherence.

121
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

Interference
To establish the mathematical background required to describe interference we continue
to consider the transverse plane wave written as;

   z  
E ( z, t )  E0 cos 2  t   E0 coskz  t  (5.1a)
  
or equivalently

 
E ( z, t )  E0 exp jkz  t  (5.1b)

as the most simple solution to the electromagnetic wave equation yet containing all
essential features.
As before we can establish the electric field in a region of space resulting from two
separate sources of field by adding the two fields

  
E  E1  E2 (5.2)

We note that when discussing optics these fields oscillate at typically 5  1014Hz and
any measurement we make will not be of the instantaneous field but of the light
intensity I which is related to the time averaged square of the electric field as previously
demonstrated

E2
I (5.3)

where  is the impedance of the medium (see previous notes) and the triangular
brackets represent a time average over several cycles.

   
  
 
   
E 2  E  E  E1  E2  E1  E2  E12  E22  2E1  E2 (5.4)

We may take the time average of each side and assume that everything takes place in
the same medium and that therefore  is common to all fields, this will lead to the
cancellation of  in much of what follows.

122
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

NB. It is important to note here that when taking the dot product in 5.4 we are taking
the value of the projection of one field on the other. This automatically takes care of
the possibility that the two fields are of different polarization when adding them to
discover interference effects.

I  I1  I 2  I12 (5.5)

 
E12 E 22
I1  , I2  (5.6)
 

The third term is the interference term and is of interest to us in this section

 
2 E1  E 2
I 12 (5.7)


NB Hecht chooses
(i) to omit the impedance in his derivations, a perfectly reasonable course to
take as the medium doesn’t change and the impedance will cancel when
converting from fields to intensity. I choose to keep the impedance in as
this is formally correct.
(ii) To use the symbol  to represent the additional phase of the plane wave
whereas I will use the symbol  to represent the additional phase and  to
represent the total phase.
(iii) To use  to represent the phase difference between two waves whereas I
will use  to represent the phase difference between two waves.

In order to evaluate the interference term we write our two electric fields as two plane
waves.
We previously called attention to one of the most important properties of the
electromagnetic wave, its phase, and in the case of plane waves travelling in the z
direction the phase is simply the argument of the cosinusoid or exponential,
  kz  t

123
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

and this phase does not depend on x or y once z has been fixed, ie. the phase is the same
at any value of x and y for a given value of z and t, or otherwise stated the phase is
constant over a plane perpendicular to the direction of travel hence the name plane
wave.
When we have two or more plane waves we need to specify the intrinsic phase of each
wave as their peaks and troughs do not generally coincide in time and space. To make
things more general to account for this the phase of a wave traveling in the z direction
becomes   kz  t   .
For a plane wave traveling in an arbitrary direction the information concerning this

direction is represented by the fact that k is a vector (a fact up until now ignored when
the light was assumed propagating in the z direction) and the phase is more precisely
written as
 
  k  r  t  
The two electric fields are then;
 
   
E1  E01 cos k1  r  t  1 (5.8a)

 
   
E2  E02 cos k 2  r  t   2 (5.8b)

NB Whilst the wavevectors k are identified by a subscript 1 or 2 we have no subscript

on . This is because we are using fields oscillating at the same frequency, the
magnitude of the k vectors are the same for both fields and it is the direction of
propagation of the wave (or equivalently wavevector) and the electric field vectors
(polarisation and amplitude) that are different between the two fields.

We can then find the interference term as follows;

   

 
  

E1  E2  E01  E02 cos k1  r  t  1  cos k 2  r  t   2 
(5.9)

 
We are going to want the time average of E1  E2 and recognizing this we rewrite the
equation with the time dependence separated out using the trigonometric identity
cos A  B  cos A cos B  sin A sin B

124
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

   
 
 
  
 
E1  E 2  E01  E02 cos k1  r  1 cost   sin k1  r  1 sint  

cosk2  r  2 cost   sink2  r  2 sint 

(5.10)

We have now separated the time dependent variable (represented in the cosinusoids)
 
making the time average E1  E2 easier to evaluate. The time average only applies to

the sin2(t) terms and the cos2(t) terms and averaging over a time period, T, much
2
greater that the period of the cosinusoids,   , they both average to:

1
cos 2 t  sin 2 t  .
2
Also the time average of the product of the sin and cosine is zero;
cost  sint   0 .

This leads to a tremendous simplification in the time average of 5.10 to

  1  
  
 
 

 
 
E1  E2  E01  E02 cos k1  r  1 cos k 2  r   2  sin k1  r  1 sin k 2  r   2
2

(5.11)

Using further trigonometric identities;

cos A cos B 
1
cos A  B   cos A  B  sin A sin B 
1
cosA  B  cosA  B
2 2
cos A cos B  sin A sin B  cos A  B

we can simplify further and 5.11 becomes;

 
  1      
E1  E2  E01  E02 cos k1  r  1  k 2  r   2 (5.12)
2

125
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

 
2 E1  E2 E01  E02
I 12  cos  (5.13)
 

   
  k1  r  k 2  r  1   2 (5.14)


is the total phase difference between the two plane waves at the position, r , where the
interference is to be determined. This phase difference is the sum of the phase
difference due to different optical path lengths traversed by each wave and due to the
initial phases of the two waves, 1 and 2.

In a common situation, where the polarization of the two interfering waves are identical,
ie. the electric fields are parallel;

E01E02
I12  cos  (5.15)


We can write I12 in terms of intensities by using the earlier expressions relating fields
and intensities
2 2
E01 2
E01 E02 2
E02
I1   I2   (5.16)
 2  2

I12  2 I1I 2 cos  (5.17)

And the total irradiance (power per unit area or intensity) is then

I  I1  I 2  2 I1I 2 cos  (5.18)

The total irradiance then varies from point to point in space as cos varies between +1
and -1
The maximum irradiance
I Max  I1  I 2  2 I1I 2 (5.19a)

Is obtained when the phase difference is   2m m  0,1,2,3

126
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

In this case of total constructive interference the two waves are in phase.
And the minimum irradiance

I Min  I1  I 2  2 I1I 2 (5.19b)

is obtained when   2m  1 m  0,1,2,3

In this case of total destructive interference the two waves are 1800 out of phase.

Another commonly encountered situation is where the electric fields have not only the
same polarization but also the same intensity, I1 = I2 = I0 . In this case the total
irradiance may be written


I  2 I 0 1  cos   4 I 0 cos 2 (5.20)
2

Where the trigonometric identity 1  cos   2 cos 2 has been used.
2
From this we have;
IMin = 0, IMax = 4I0

Identical arguments and results apply to the interference of two spherical waves
emanating from two point sources S1 and S2 that overlap at a point P.

r1 r2

S1 S2

127
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

We have seen previously that the spherical waves propagating in free space may be
written as


E1 (r1 , t )  01 exp  j k 0 r1  t  1 
A
(5.21a)
r1


E2 (r2 , t )  02 exp  j k 0 r2  t   2 
A
(5.21b)
r2

r1 and r2 are the distances of point sources 1 and 2 from the point of overlap, P or
equivalently the radii of curvature of the two spherical waves.
The phase in this case depends only on the distance r from the point sources and the
surface of any sphere centered on a point source is a surface of constant phase.
We can easily identify the phase difference between the two waves as

  k0 r1  r2   1   2  (5.22)

Using the expression for total irradiance as found previously for plane waves, 5.20, and
the new phase difference

1 
I  4 I 0 cos 2  k 0 r1  r2   1   2  (5.23)
2 
The condition for maxima and minima are as before ie.  = 2m for maxima and 
= (2m + 1) for minima where m = 0, 1, 2, 3, ……..
Using the expression for  we obtain constructive interference when

r1  r2    2m   2  1  (5.24a)

k0
And destructive interference when

r1  r2    2m  1   2  1  (5.24b)

k0

128
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

r1 r2

S1 S2

Both equations 5.24a and b describe a family of hyperboloids as depicted in the above
figure where the hyperboloid surfaces represent the points P where
constructive/destructive interference occur. If we hold the phases from the two sources,
S1 and S2 , to be equal, ie, 1 = 2 and choose them to be zero then 5.24 may be
rewritten as

r1  r2    2m  m 0 (5.25a)

k0

r1  r2    2m  1   m  1  0 (5.25b)

k0  2

Each value of m is represented by a hyperboloid with the positive values on the right
hand side of the mid-line with r1 > r2 and the midline (or in 3D a plane perpendicular to
the page) where r1 = r2 and the left hand side hyperboloids for r1 < r2 . We can imagine
a screen placed as a plane intersecting this set of hyperboloids where at the points of
intersection there is constructive interference. Any point, P, on this midpoint plane has
r1 = r2 and therefore represents 5.25a with m = 0, the zeroth order interference fringe.

129
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

These equations can be used to describe the appearance of interference fringes between
two line or point sources such as Young’s slits. We now need to explore some examples
of the ways in which interference effects are manifested.

Interference Effects in Practice

(i) Interference by Division of Wavefront.
Young’s Slits
Thomas Young, in 1801, was one of the first people to demonstrate the wave nature of
light carrying out what has since become a classic experiment that has been applied to
particles to demonstrate matter waves as well as light. In Young’s original experiment
he used a pinhole with a monochromatic light source behind it to define a point source
of spherical waves with two further pinholes in a screen at a distance from the source
pinhole much greater than the wavelength of light. The two pinholes in the screen act as
two separate but related light sources that have been derived from an original light
source by division of the wavefront of that original source. That the two secondary
sources are related allow interference effects to be observed by placing a second screen
at a similar distance from the two secondary light sources.
To find the light intensity at the second screen we need to add the electric fields of the
two light sources taking into account the difference in phase between those two fields as
usual. If the two holes providing the interfering sources are of the same size then we can
make the approximation that the intensities of the two sources are equal and use
equation 5.20 to find the intensity at any point P on the screen;


I  4 I 0 cos 2 (5.20)
2

Before continuing further, it is necessary, for purposes of simplification, to keep the

observation point P close to the centre of the screen and the diagram below is
exaggerated for demonstration purposes. In fact we are interested in small values of .

130
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

r2
y
S2
r1

 
S O
a

S1
d
s

It remains to establish the phase difference,  , at the observation point, P.

The two point sources act as sources of spherical waves and we have the phase
difference earlier in 5.22

  k0 r1  r2   1   2   k0 (r1  r2 ) (5.22)

If the screen with the two point sources is far enough from the original point source, S,
(orders of magnitude larger than a wavelength) then the spherical wave at S 1 and S2 will
be approximately plane and there will be no other source of phase difference apart from
the path difference and the term (1 - 2) is zero. With the aid of the figure above we
may find the path difference which is the same as the optical path difference, , (as the
waves propagate in air with n = 1) and hence the phase difference   k 0  .

  S1P  S 2 P  r1  r2 (5.26)

131
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

We may try to get a simplified expression for r1 – r2 beginning with the law of cosines,

c2  a 2  b2  2ab cos C
as applied to the triangle S1S2P

r2

S2

r1


a

90 - 

S1

r22  a 2  r12  2ar1 cos(90  )  a 2  r12  2ar1 sin  (5.27)

r22  r12  a 2  2ar1 sin  (5.28)

r2  r1 r2  r1   r2  r1 2r1  a 2  2ar2 sin  (5.29)

  r1  r2  a sin   a (5.30)
Where we used the fact that a << r1 and r1  r2 to simplify 5.29
And a further simplification
y
 (5.31)
s
may be used to obtain

132
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

y 2 y
  k 0   k 0 a  a (5.32)
s 0 s
Equation 5.25a and b gave us conditions for maxima and minima respectively and using
this result and 5.25 we find the condition for a bright fringe

r1  r2   ay  m 0 (5.33)

s
Or in terms of position on the screen, y

s
ym  m  0 (5.34)
a

Gives the position of the mth bright fringe on the screen. The fringes are equally spaced
with a separation

s
y  y m  y m1  0 (5.35)
a

Now that we have  we can write the intensity as a function of y by using this in 5.20
 ay   as sin  
I  4 I 0 cos 2    4 I 0 cos 2    4 I 0 cos 2  (5.36)
 0s   0s 
With

a a
 sin   k 0 sin  (5.37)
0 2

y = 0s/a

y
0
133
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

 is the phase difference between rays emanating from each slit when the screen is at
infinity (ie the two rays are parallel) as shown in the diagram below.

  k 0 S1B (5.38)


S2

S1 B

We have found from this analysis that;

(i) to have well separated fringes we need a small separation, a, between

pinholes or slits and that
(ii) longer wavelengths will give rise to broader fringes.

We have seen that Young had some stringent requirements on the geometry of his
experiment in order to see the interference fringes;

(i) The slits cannot be too far apart and generally a << s

134
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

(ii) The original source must be at a distance, d, from the screen such
that the spherical wave approximates a plane wave in order that 1   2
(iii) Fringes will only be observed near the centre of the screen where
r1  r2 is not too large.

(iv) The value of s must be large if fringe separation is to be large

a
enough to observe due to the small size of the wavelength of visible light.

All of the above apart from (iv) result from the lack of coherence in the sources
available to Young. We return to the question of coherence after we examine the second
type of interference.

135
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

Interference by Division of Amplitude.

An important way of achieving the conditions for exhibiting interference effects, ie.
obtaining two or more oscillating electric fields, sending them on different paths before
recombining them to observe interference is to split a primary wave into two separate
secondary waves by partial reflection of the primary and using the reflected and
transmitted waves as the secondary waves to be recombined. This may be achieved in
many ways, sometimes coming about in a quite natural manner while often achieved
with a particular optical arrangement.

Dielectric Thin Film Interference.

One of the most commonly observed examples of interference through division of
amplitude occurs where light incident on a thin layer of dielectric undergoes reflections
from the top and bottom surface of the layer and under the right conditions constructive
or destructive interference occurs. Examples of this are the colour effects seen when a
thin layer of oil is floating on water and the colours seen in a soap bubble. The
interference effects caused by multiple reflections in the thin scales of a butterflies
wings give rise to their spectacular iridescence. The effect is also of use in many
technological applications where thin layers are designed and constructed with the
intention of creating thin dielectric film interference.
The diagram below indicates schematically how interference effects are produced as a
result of partial reflection/transmission at a thin dielectric film. For purposes of this
analysis the dielectric film of refractive index, nF , is standing on a substrate of
refractive index, nS, and light is incident from a medium of refractive index, n0
(typically air). There will in general be multiple reflections at the air/film interface and
at the film/substrate interface. Transmission will also occur into the substrate. We can
use equations 5.19 to discover the conditions for constructive or destructive interference
by first calculating the phase difference between the two rays, 1 and 2, shown
propagating upwards from the top of the film in the figure.
The optical path is the real space distance traveled multiplied by refractive index of the
medium and therefore the optical path difference, , between rays 1 and 2 after their
separation on arrival at A is
  nF  AB  BC   n0 AD (5.39)

136
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

And the phase difference due to the optical path difference is the optical path difference
multiplied by the magnitude of the wavevector in free space (vacuum).

2
OPD  k 0   k 0 nF  AB  BC   n0 AD   nF  AB  BC   n0 AD  (5.40)
0

1
2

I

n0 D

A C

d nF

T
B
nS

137
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

NB. For the total phase change we need to include the phase change that occurs on
reflection which may be different for each ray as we recall from the Fresnel
equations. Thus,

2
  k 0 nF  AB  BC   n0 AD    2  1   nF  AB  BC   n0 AD   2  1 
0
(5.41)
From geometry and Snell’s law we obtain everything in terms of T,
d n
AB  BC  AC  2d tan T AD  AC sin  I  2d tan T F sin T
cos T n0

2d n sin 2 T
AB  BC  AD  2d F
cos T n0 cos T
whence

 
2  2nF d

 0  cos T
 
1  sin 2 T    2  1  
4nF d
0
cos T   2  1  (5.42)


It is often convenient to have the phase difference in terms of the angle of incidence and
we can use Snell’s law to achieve this

4nF d 4d
  1  sin 2 T   2  1   nF2  n02 sin 2  I   2  1  (5.43)
0 0

Before we address ourselves to the outstanding question of the reflection phase shifts 1
and 2 we note that nF may be greater than or less than n0 and nS, eg an air gap between
two parallel separated glass slides or a freestanding soap film in air respectively. We
also recall that there are two types of reflection namely internal reflection where
the refractive index of the sourced region is greater than that of the unsourced region
and external reflection where the refractive index of the unsourced region is
greater than that of the sourced region. With these possible types of reflection in mind
we can identify several possibilities that apply at near normal incidence,   300;

138
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

(i) External reflection plus internal reflection

An example is the soap film/bubble free standing in air, nF >n0 = nS = nAir . In this case
for near normal incidence,   300 , we will have  2  1    and the total phase
difference is

4nF d
  cos T   (5.44)
0

(ii) Internal reflection plus external reflection.

An example of this is an air gap between two glass slides. Again for near normal
incidence 2  1    and the total phase difference is

4nF d
  cos T   (5.45)
0

(iii) Both reflections are internal (or external)

If the reflection at the top and bottom interfaces are both internal reflections (or external
reflections) the factor  2  1   0 and

4nF d
  cos T (5.46)
0

(iv) Reflections from partially metallised surfaces

Examples are a thin metallic film or two partially metallised surfaces separated by a
dielectric (eg. air gap between two partial mirrors. In this circumstance then 1   2  0
as there is no phase shift upon reflection at a metal where boundary conditions require
that the light electric field is zero and again

4n F d
  cos T (5.47)
0
So far nothing has been said about the intensities of the two combining beams, I1 and I2
and so nothing is known about the actual irradiance, IR , achieved in the reflected beam.
Neither has the possibility of multiple reflections been considered nor the intensity of

139
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

the final transmitted beam, IT. We return to this when multiple beam interference and
the Fabry Perot interferometer are discussed.

140
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

Multiple Beam Interference.

When considering the thin dielectric film interference it was recognized that multiple
reflections could occur and this possibility is now explored.

r r/
t t/

n1 nF n1

Consider the above system of a thin dielectric film of thickness d and refractive index nF
embedded in a medium of refractive index n1. Further, we consider the film to be non
absorbing at wavelengths of interest. An electromagnetic wave impinges from the left
hand side and is partially transmitted and partially reflected undergoing a series of
subsequent reflections and transmissions at the left hand and right hand interface. To
find the total reflected or transmitted fields we need to add all of the reflected or
transmitted waves taking into account phase shifts between the waves as usual. We
denote the fraction of the electric field amplitude transmitted on entering the film as t
and on leaving the film as t/ and the fraction reflected at the n1/nF interface as r and at
the nF/n1 interface as r/. These quantities were discussed earlier when the Fresnel
equations were established and they depend, in general on the angle of incidence and
the difference in refractive indices at the interfaces. In particular we established earlier,
using an argument due to Stokes, that r(I)= -r/(T) and t  I t / T   1  r 2 where I and

T come as a pair of angles related by Snell’s law.

141
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

tt/r/9E0
tt/r/8E0expj(t-4)

tt/r/7E0
tt/r/6E0expj(t-3)
tt/r/5E
0
tt/r/4E0expj(t-2)
tt/r/3E
0
/ /
tt/r/2E0expj(t-)
tt r E0

tr/2E0
rE0 tt/E0expj(t)
/
tr E0
tE0 T
I

E0 d

The above diagram establishes the sequence of reflected partial amplitudes and of
transmitted partial amplitudes.
We can find the sum of the partial transmitted amplitudes giving the transmitted electric
field amplitude E0t.

E0t  tt / E0e jt  tt / r / 2 E0e j (t   )  tt / r / 4 E0e j (t 2 )  ...  tt / r /(2 N 1) E0e j (t ( N 1) )
(5.48)

 
E0t  tt / E0 e jt 1  r / 2 e  j  r / 4 e  j 2  .r / 6 e  j 3 .  .r / 8 e  j 4 .  ........ (5.49)

NB.  is the phase difference between two adjacent secondary waves.

Recalling that an infinite geometric progression is
1
1  a  a 2  a 3  a 4  ..... 
1 a

142
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

The geometric progression in square brackets on the RHS has a  r / 2 e  j  r 2 e  j
again and 5.72 may be compactly rewritten as
 tt / 
E 0t  E 0 e jt  2  j 
(5.50)
1  r e 

Taking the time average of E0t multiplied by it’s complex conjugate and dividing by 
gives the intensity as usual and for the transmitted intensity we obtain

E2
I 0t  0
 
tt /
2


(1  r 2 ) 2
 
2 1  r 2 e  j 1  r 2 e  j
I i
 
1  r 4  2r 2 cos  
I 0t  I I
1  R 2 (5.51)
1  R 2  2 R cos 


We finally use the trigonometric identity, cos   1  2 sin 2 to get an expression for
2
the fractional transmitted intensity, .

I
T  0t 
1  R 2 (5.52)
II   
1  R  2R 1  2 sin2
2

 2 

I
T  0t 
1  R 2 
1  R 2 (5.53)
 
II 2
1  R  2R  4R sin 2
1  R 2
 4R sin 2
2 2
1
T (5.54)
4R  2
1 sin
1  R 2 2

We may now introduce a quantity that will appear frequently called the coefficient of
finesse, F which we will see is a commonly used figure of merit for a Fabry Perot
device
2
 2r  4R
F    (5.55)
 1  r2  1  R 2
and obtain a compact form for the fractional transmitted intensity;

143
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

1
T (5.56)

2
1  F sin
2
It is now a simple task to find the fractional reflected intensity, R, as in the absence of

absorption R + T = 1 and thus;


F sin2
1 2
R  1  (5.57)
2  2 
1  F sin 1  F sin
2 2

Maximum transmitted intensity will obtain under the same conditions that give
minimum reflected intensity. It is clear that to maximize the transmitted intensity the
denominator in 5.56 needs to be minimized, ie. the sinusoid in the denominator needs to
be zero and therefore

 m (5.58)
2
for maximum transmission (minimum reflection). Bearing in mind the relation 5.42
2
between phase difference and wavelength,   2n F d cos T , this is also a
0
condition on wavelength;
By rearrangement of 5.42 with the condition 5.58 we obtain this condition

 
 2n F d cos T  m
2 0

0 
m  m F  d cos T
2n F 2

where we have taken 1   2  0 for simplicity.

When the condition is fulfilled
T=1
The physical meaning becomes clear when we consider normal incidence on the film
and the equation becomes
F
m d
2

144
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

This is the requirement that half integer wavelengths of the wave fit between the two
inner surfaces of the film, ie. that when standing waves are supported there is maximum
transmission and minimum reflection.

145
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

m=1

m= 3

m=2

Unsurprisingly we obtain maximum reflection where transmission is minimum. We will

now examine this more closely.
Minimum transmission occurs where the denominator of 5.56 is as large as possible

which is when sin 2  1 . When this condition holds
2

 2m  1   2m
2

And the value of the minimum transmitted intensity using sin 2  1 in 5.56 is;
2
 
 
  I i  1  R  1  R 2
 2   
I t Min  Ii
1
 Ii 
1
  I  
1 F    1  R   4 R 
i
1  2 R  R  4 R 
4R 2 2
1 
 1  R 2 
 
(5.59)

I t Min  I i 1  R 2
2
(5.60)
1  R 

And the maximum reflected intensity occurs when sin 2  1 and is
2

146
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

 4R 
 
 1  R 2 
I r Max  I i F
 Ii    Ii
4R
(5.61)
1 F  4R 
2
1  R 2
1   
2 
 1  R  
The fractional transmission and reflection are plotted below.

147
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

Transmission through thin film vs

phase shift, r = 0.2, 0.5, 0.9

0.8
It (fraction)

0.6

0.4

0.2

0
-4 -2 0 2 4
Phase shift  ( rads)

Reflection from thin film vs phase

difference, r = 0.2, 0.5, 0.9

0.8
Ir (fraction)

0.6
0.4

0.2

0
-4 -2 0 2 4
Phase shift  ( rads)

148
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

The transmission/reflection intensities from a thin dielectric film as described by

equations 5.56 and 5.57 are shown in the two graphs above where three values of r are
used, the sharpest minima/maxima occur for the largest values of r.
There is a maximum in the transmission where  is an integer multiple of 2 and as r
is increased this maximum becomes increasingly sharp around these values. The inverse
is true of the reflection. Such behavior suggests potential use as a tuned bandpass filter.
It is important to pause and to recall once again what it is that  represents physically.
It represents the difference in phase between adjacent reflected (or transmitted) waves.
and that phase difference is determined by the extra distance traveled by one wave with
respect to the other and so varies with angle of incidence and film thickness. It also

depends inversely on the wavelength of the light within the film,   0 . This
nF

means that the horizontal axis, , in the above graphs could as easily be a plot of either
inverse wavelength or of angle of incidence. We see this when writing the phase

difference explicitly as   2dnF 2 cos T .

0
In other words there are several variables that could be plotted on the x axis in place of
1
, namely by holding T constant the inverse wavelength, 01 (or frequency   c
0
) could be plotted on the x axis with a series of wavelengths/frequencies at which sharp
resonant transmission occurs given by;

2 2dnF  m 0 
  2m  2dnF cos T , 0  cos T , T  cos 1 
0 m  2dnF 
NB the resonant frequencies are equally spaced the wavelengths are not!
With modern technology allowing control over very small distances it is also possible to
adjust d and allow this as the variable. Finally, as we have seen, the refractive index of
certain media may be altered by application of an electric field allowing another way to
alter  and thus the transmission.
In the above equations the integer m is also known as the order of the transmission (or
reflection). An extended white light source would then be split into transmitted spectral
components where the wavelength criteria is satisfied. Depending on the order, m, this
would be a wavelength band around some peak that is transmitted whilst the other
wavelengths are reflected and lost until a second order (third order etc) transmission

149
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

peak allows further transmission of another band of wavelengths around a second (third
etc.) wavelength. The peak transmission (or reflection) will also depend on the viewing
direction and it is this that gives rise to the colours observed in a thin oil film floating on
water or the colours observed in a soap bubble.
Looking at the above graph for transmission through the thin film we see an
increasingly narrow range of  (or T or 0 ) over which transmission occurs as r is
increased. Ie. the transmission of the thin film may be highly tuned. This highly defined
directionality and wavelength range, comes about due to the large number of coherent
sources that contribute to the overall beam.
The thin film dielectric with multiple interference clearly demonstrates useful properties
and the potential for constructing a useful device. The Fabry Perot interferometer is
an engineered structure using the basic principles that have just been discussed and
finds many uses in optics from spectroscopy through high resolution optical filters to
laser resonators. For this reason it is worth examining in some detail.

b) Fabry Perot Interferometer.

One of the simplest realizations of an optical structure, using the principles of the
multiple reflection interference of the thin film previously discussed, is to take a pair of
parallel partially reflecting surfaces separated by a careful engineered spacer.

Spacer

Spacer

Extended
source d Screen

The above diagram shows such a structure, known as a Fabry Perot interferometer or
Fabry Perot etalon, with a pair of metallised partially reflecting surfaces, in this case

150
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

evaporated onto two transparent substrates. The reflecting surfaces are held precisely
parallel to one another and usually the substrates will be slightly wedged on the non-
metallised surface in order to suppress the formation of secondary parallel reflecting
systems that would interfere with the operation of the primary system. The two
reflecting surfaces of the primary system are held at a precise separation, d. There is in
the example shown an extended light source to the left and a screen to the right.
Choosing a ray from a given portion of the extended source traveling at an arbitrary
angle, its progress is followed as it undergoes multiple reflection/transmission events
before the transmitted rays from this one coherent point source ar collected by a second
lens and brought to a focus at some point P on a viewing screen (photographic plate,
retina etc.). Of course, any ray traveling at the same angle from an equivalent point on
the extended source (at the same distance from the system axis) would have been
brought to a focus on the screen at the same distance from the system axis resulting in
the appearance of a series of concentric rings centered on that axis.

This system has the essential characteristics of the previous situation examined with the
exception that;
(i) The metallised reflecting surfaces will be a source of dissipation/absorption and
this will mean that the Stokes relations that were used frequently in the previous
discussion no longer hold ie.

Ii  I r  It T R 1 and tt /  r 2  1

Rather these are modified to account for the absorbance, A, in the following way;

Ii  I r  It  I A T R A 1 (5.62)
Where 5.62 again represents the conservation of energy but with the dissipation
(absorption) term included.
(ii) The gap, d, is much larger than in our thin film scenario and can be from microns
up to centimeters as it is now a factor under control of engineering.
and
(iii) Metallic films will introduce a phase shift, (), upon reflection which may not
be zero or  and that may depend on the angle of incidence, T.
Now, the phase difference between two successive waves is as usual

151
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

4nF d
  cos T  2 (5.63)
0

A complete analysis with absorbance included gives

2
 A  1
T  1  
 1  R  1  F sin2 
2
With maximum transmission
2
 A 
TMax  1  
 1 R 

We will ignore all these differences and use what we have found already from analysis
of multiple reflections in thin films.

It 1
 (5.64)
Ii 2
1  F sin
2

This expression of 5.64 makes it much easier to understand what is happening with the
transmission as it did with the thin film analysis as we simply get maxima when the
sinusoid in the denominator goes to zero
Two graphs shown below demonstrate the effect of the absorbance. The effect of A is to
reduce the maximum value of the transmission. The graphs show the transmission of the
Fabry Perot with an absorbance, A = 0.1 as a function of phase difference (equivalently,
 or 0 ). Also shown is the normalised transmission. Note in the first that the maximum
is well below 1 as a result of the absorbance. In those two graphs the amplitude
reflectance is changed from 0.9 to 0.7 and this has big effects on the transmission with
the width of the transmission getting much larger and the transmission never dropping
to zero in the case of a reflectance of 0.7.

152
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

Transmission of a Fabry Perot with r= 0.9, A = 0.1

0.25
Transmission of Fabry Peroy

0.2

0.15

0.1

0.05

0
-4 -2 0 2 4
Phase Difference  (units of )

Fabry Perot Transmission Normalised to the

Maximum Transmission, r = 0.9, 0.7
F = 22, 1.66
1
Normalised Transmission.

0.75

0.5

0.25

0
-4 -2 0 2 4
Phase difference,  (units of )

153
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

Properties of the Fabry Perot.

When using the Fabry Perot as a filter or monochromator (wavelength selective device
cf prisms and gratings) several important questions arise concerning the resolution of
the Fabry Perot, ie. how well defined is the transmitted wavelength/frequency?
To answer this we need to look at a number of things.

i) The free spectral range. This is the separation of the transmission peaks of
adjacent orders which occur at  = 2m so peaks are separated in phase by 2. The
separation of the peaks of adjacent orders in terms of wavelength is called the free
spectral range of the Fabry Perot and indicates the wavelength range over which the
interferometer may be operated without adjacent orders mixing. We see how this
matters by considering operating the interferometer as a variable plate separation device
to find an unknown wavelength 1. As d is altered  will change with it until at some
point  satisfies the condition for maximum transmission at the specific wavelength of
interest and the light is transmitted. We know d and  and should therefore know 1.
But there is a problem as we don’t know m. There is a second transmission peak in the
neighbourhood of this one where the order number is m + 1 and we may have a
wavelength corresponding to m+1 and not m. To avoid any such confusion we need to
know how far apart the two transmission windows are in terms of wavelength (or
frequency). The peak separation or free spectral range is simply found as
2nF d cos T 2nF d cos T 2nF d cos T
 FSR   m   m 1   
m m 1 m(m  1)
2nF d cos 
We may write this in a different way using  m
m T

2nF d cos T 
 FSR   m (5.65)
m(m  1) m 1

The free spectral range in terms of frequency is

c c c(m  1) cm c 1
 m 1   m      
 m 1  m 2nF d cos T 2nF d cos T 2nF d cos T 
(5.66)

154
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

c c
 FSR   (5.66a)
2nF d cos T m(m  1) FSR

Generally the device is used with an air gap, nF = 1, and at normal incidence in which
case 5.65 simplifies to

2d 
 FSR   m
m(m  1) m  1

And 5.66 simplifies to

c c 1
 FSR   
2d m(m  1) FSR 

Where  is the time taken for the wave to travel from one reflecting surface to the other
and back again, also known as the round trip time within the interferometer.
If we make the free spectral range larger than the interval over which we are looking for
any emission the problem of confusing orders can be avoided.

ii) How wide is the window of transmission

As a measure of the width of the transmission window we look for the value of 
where the maximum transmission has dropped by half,  1 and the width of the
2

transmission window is defined as twice this ie Full Width at Half Maximum,

FWHM  2   1 , a commonly employed criterion.
2

1
The curve is simply T  and we need to know where this has dropped by

2
1  F sin
2
50%.
1
 0.5 (5.67)
 1
1  F sin2 2
2

155
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

Re arranging to make  1 the subject of the equation

2

 1
2 2 1
F sin (5.68)
2
 1
2  1
sin (5.69)
2 F

1
 1  2 sin1 (5.70)
2 F

For large coefficient of finesse, F, the approximation

1 1
sin1 
F F
may be used

1
 1  2 (5.71)
2 F

And the width of the peak, at half the maximum is

4
FWHM  2 1 
2 F

It is the case that the phase cannot be measured and only relative phases have any
practical effect. We are more interested in asking questions about wavelength or
frequency when discussing light waves. These are, of course, related to the phase as
2
seen many times,   2dnF cos T  m2 for maximum transmission of the mth
m

order wavelength m. If we know about the phase but wish to know about the
wavelength we proceed as follows;

d  1
 4nF d cos T (5.72)
d 0 2

156
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

Where the device is used at m

d  1
 4nF d cos T (5.72a)
d 0 2m

 FWHM  2m
1
FWHM   2m 1 1
(5.73)
4nF d cos T nF d cos T F

Using the previously established condition for maximum transmission of mth order
2nF d cos T
m 
m

We may simplify 5.73

2 1
 FWHM   m (5.74)
m F

An important property, the resolution of the Fabry Perot, used in mth order is defined as

m m
 F (5.75)
 FWHM 2

and serves as a dimensionless figure of merit.

We may want this in terms of frequency where we use the relationship between
frequency and wavelength, m m  c for the mth order. Using a procedure identical to
that used to translate from phase to wavelength we translate from wavelength to
frequency as follows;

d c

d 2

At the specific frequency, m where  = m we have

c c  FWHM 2 1
 FWHM    FWHM     m (5.76)
2m m m m F

157
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

Therefore the resolution in mth order is as previously found for this quantity in terms of
wavelength
m m
 F (5.77)
 FWHM 2
NB the minus signs that occur in these equations when differentiating has been
subsequently ignored as the meaning is simply that for example on going from  to 
+  the frequency will go from  to  -  and in this context it carries no
significance.

iii) What is a good figure of merit when defining the performance of a Fabry Perot?
Comparing the separation of peaks or free spectral range with the width of a single peak
we may find a figure of merit for the resolution of the Fabry Perot.
The ratio of peak separation to peak width is known as the finesse, F ,

FSR 2  F
F    (5.78)
FWHM 4 2
F

We see by comparing the frequency resolution expression 5.77 and the definition of
finesse, F that the frequency resolution of the mth order may also be written as

m m
 F  mF (5.79)
 FWHM 2
and
m m
 F mF (5.79b)
 FWHM 2

NB The finesse and the coefficient of finesse , while closely related are not the same
thing!!

iv) What is the resolving power, R , of the Fabry Perot?

This is a question that asks, “ if we have a source of two wavelengths very close
together what is the minimum wavelength separation, Min that a Fabry Perot would
be able to “see”?

158
Electromagnetic Waves & Optics: Lecture Notes ©Kevin Donovan

Min

1.0

1 2
0.5

The criteria for being able to resolve two slightly different wavelengths is that their half
maximum height values as measured by the Fabry Perot just cross as shown in the
diagram above where monochromatic wavelengths 1 and 2 are measured as finite
width features due to the finite width of the transmission window of the Fabry Perot.
1  1 1  2  2 1
2 2

Therefore

1 1
 Min   2  1  2 1   FWHM  2 m
2 m F
The resolving power is simply defined as

 m m F
R    mF
Min Min 2

Finally, if we know F or F we can find the minimum resolvable wavelength difference

for a given order, m

2m m
Min  
m F mF

159