Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

2. DIELECTRIC INTERFACE & FRESNEL’S EQUATIONS.

Dielectric Interface
The aim of these notes is to establish the following:

(i) Laws of Reflection and Refraction

(ii) Total Internal Reflection, TIR, and the Critical Angle.
(iii) Evanescent Fields
(iv) Transverse Electric, TE, and Transverse Magnetic, TM, Polarisation.
(v) Reflection and Transmission Coefficients.
(vi) Brewster Angle.

Maxwell’s Equations & Boundary Conditions

All of the physics of transmission and reflection of light at the interface between two
dielectrics with differing refractive indices is determined by the boundary conditions on
the electric and magnetic fields at that interface as required by Maxwells equations. We
consider Maxwell’s equations as they appear when written in integral form.

  Q
 E .dA  (2.1)
0
 
 [Link]  0 (2.1a)
  d B
 .dl  
E (2.2)
dt
   d E 
 .dl   0  iC 
B  (2.2a)
 dt 

In particular we focus on just two of these


C E .dl    [Link] (2.2)
t S

and


C H .dl   [Link]
t S

These are the integral forms of the Maxwell equations involving   E and   H using
Stokes theorem and assuming no currents, J = 0.
21
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

l
1, n1
E1tan
l
l
E2tan
2 , n2
l
l//

Consider the boundary between the two dielectrics shown above with the dashed
contour crossing the boundary. As dA can be shrunk to zero by reduction of l .
the integral around the contour can be reduced to

 E  dl E1  l //  E2  l //   E1Tan  E2Tan l //  0 (2.3)

C

E1Tan  E2Tan (2,3a)

And the same can be done with the H field

1, n1
H1 l
l
H2
l 2 , n2
l//

 H  dl H1  l //  H 2  l //   H1Tan  H 2Tan l //  0 (2.4)

C

H1Tan  H 2Tan

In other words the boundary condition we are looking for is:

The tangential H and E fields must be continuous across a boundary.

The other pair of Maxwell equations in integral form can be used to show that the
perpendicular components of D and B at the interface must be continuous but these will
concern us no further.

22
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

Laws of Reflection and Refraction

The first thing we can establish with these boundary conditions are the laws of
reflection and refraction. To deal with this we need to understand the implications of the
boundary conditions
Let the dielectric interface be in the yz plane through x = 0 as shown (below). The
boundary conditions, as previously, require that all tangential fields (E and H ) be
continuous across the boundary at x = 0. Keeping vector notation for the time being we
can write the E fields as:

EI  AI exp  jk I  r 
ER  AR exp  jk R  r  (2.5)
ET  AT exp  jkT  r 

A’s are the amplitudes of the incident, reflected and transmitted electric fields and k’s
are the wavevectors of those waves, denoted I, R and T respectively.
The diagram below shows the situation with an incident, reflected and transmitted wave

kI kR

Sourced region x

z
y
S

U kT

Unsourced region

Please note in the above diagram a sourced region (where the light source is) and an
unsourced region are indicated. Throughout this module the properties of the sourced
region such as refractive index are generally denoted by with the subscript S, eg. S , nS
and the unsourced region by the subscript U, eg. U , nU etc

23
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

The boundary condition then requires:

AI exp jk Iy yexp jk Iz z   AR exp jk Ry yexp jk Rz z Tan  AT exp jkTy exp jkTz z Tan
(2.6)

The important thing to note is that this must be true at all values of y and z and that
matching at one point is insufficient.

This strict requirement that matching occurs everywhere on the interface then translates
as:
k Iy  k Ry  kTy  k y

and that (2.7)

k Iz  k Rz  kTz  k z

i.e. kI , kR and kT must all lie in the same plane known as

the plane of incidence.

if this were not the case the fields whilst satisfying the boundary conditions at x, y, z =
0 (say) would not satisfy the boundary condition at all values of y and z ie.
everywhere on the yz plane at x = 0!!

24
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

It is now possible, without any loss of generality, to rotate the co-ordinate system about
the x axis to cause the xz plane to coincide with the plane of incidence. This will allow a
simpler analysis (and simpler diagrams!).

Sourced region
S , nS
x
kI
kR
 R

Unsourced region
kT
U , nU


With the axis now rotated we can establish some facts concerning reflection and
transmission at oblique incidence using the boundary conditions.

Looking at the above diagram we only have x and z components of k for incident,
reflected and transmitted waves and we have:

k I   x̂k Ix  ẑk Iz

k R  x̂k Rx  ẑk Rz (2.8)

kT   x̂kTx  ẑkTz

Note the sign of the unit vectors x̂ and ẑ which tells us about the direction of the
wavevector.

25
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

Also from examining the diagram;

k Ix  k S cos  I k Iz  k S sin I

k Rx  k S cos  R k Rz  k S sin R (2.9)

kTx  kU cos T kTz  kU sinT

kS and kU are the magnitudes of the wavevectors in the sourced and unsourced regions
respectively

   
kS   nS  nS k 0 and kU   nU  nU k 0 (2.10)
vS c vU c

Comparing the incident and transmitted waves the boundary conditions that the kz
vectors are equal give

k S sin I  kU sinT or nS k0 sin I  nU k0 sinT

In other words nS sin I  nU sinT Snell’s Law (2.11)

Comparing the incident and reflected waves the conditions give

k Iz  k Rz or nS k0 cos  I  nS k0 cos  R

In other words

cos  I  cos  R  I  R Law of reflection (2.12)

In everything that has been done in this section there has been one guiding principle and
that is the boundary condition established early on through Maxwell’s equations that the
tangential fields match at all points along the boundary. We continue using that
principle to obtain several other useful pieces of information.

Internal and external reflection.

We pause here to note that it is useful to distinguish between the situation where the
sourced region has a greater refractive index than the unsourced region and vice versa.
Snells law, 2.11, tells us that when the refractive index of the sourced region is greater
than the unsourced region, the refracted (transmitted) ray is bent away from the normal
compared to the incident ray ie. I < T , whilst the opposite is true when the refractive

26
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

index of the sourced region is less than that of the unsourced region. Bearing this in
mind we can look at the two immediately preceding diagrams and realise that nS < nU.
As far as reflection is concerned it is common to refer to reflection occuring within a
high refractive index region at the boundary with a lower refractive index region as
internal reflection whereas when the reflection occurs within the lower refractive
index region this is called external reflection.

Total Internal Reflection, TIR

Sourced region
S , nS
kI I R kR

kT
T

Unsourced region
U , nU

The construction shown above is known as a phase matching diagram. There is a

boundary between two dielectrics along the z axis. The two concentric semicircles have
radii equal to the magnitudes of the wavevectors, nSk0 and nUk0 in their respective
regions.
Beginning with an incident wave represented by the arrow kI travelling at some angle I
, we can use the law of reflection (I = R ) to construct the arrow representing the
reflected wave, kR . Note the arrows must have a length equal to the magnitude of the

27
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

wavevector in the sourced region i.e. the radius of the semicircle in the sourced region
and thus end on the semicircle.
We now construct the transmitted wave by dropping a perpendicular from the tip of the
reflected arrow onto the circumference of the semicircle in region 2. This ensures that
the z components of the wavevectors are equal in both regions when the transmitted
wavevector is drawn to finish at the point of intersection of the perpendicular with the
semicircle in the unsourced region.
In the example shown it is clear that the refractive index is greater in the unsourced
region as the radius of the semicircle is greater there. More interesting is the situation in
the phase diagram shown below where the refractive index in the sourced region is the
greater.

C

We can now perform the same “phase matching”constructions for the incident
wavevector and the reflected wavevector as shown by the three solid arrows. However
if we make the angle of incidence a little larger (dashed line) we may no longer keep the
z component of the wavevector in the unsourced region equal to that in the sourced
region at least not yet! i.e. there is a critical angle, C beyond which we can no longer

28
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

satisfy the boundary condition with a wave transmitted into the unsourced region. At
this angle (thick arrows) there is a refracted wave travelling along the boundary and at
greater angles (dashed arrows) there is only a reflected wave and no energy is
transmitted into region 2. This phenomena is known as Total Internal Reflection and
is the key to waveguiding and optical fibres.
To find the critical angle consider Snell’s law,

nS sin  I  nU sin T (2.13)

or
nS
sin  I  sin T (2.14)
nU

nS
When nS > nU and  1 , because sinT (the RHS of the equation) can be no larger
nU
than 1, it is possible to find an angle where the LHS of the equation is equal to 1 and the
equation can only just be satisfied. The critical angle, C occurs where T = 900 and
sinT = 1. i.e.

n
C  sin 1 U (2.15)
nS
If we make the angle of incidence any greater then Snell’s law cannot (seemingly) be
satisfied. The problem can be seen to be that the z component of the k vector in the
sourced region is larger than the entire wavevector in the unsourced region,
k Sz  nS k0 sin  I  kU  nU k0

Evanescent fields
There is in fact a way that the boundary conditions can be achieved for angles of
incidence greater than the critical angle. This does not lead to energy propagating away
from the interface into region 2 but gives a field whose amplitude is exponentially
decaying away from the interface in region 2. Such evanescent fields are of some
technological interest and we examine them here.
NB The evanescent field is related to quantum mechanical tunneling!
The boundary condition is that the z components of the wavevectors match at the
interface between sourced and unsourced regions. In other words given that;

29
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

2 2
k Sz  k Sx  k S2 and 2
kUz 2
 kUx  kU2

The boundary condition may be written as

k Sz  k S2  k Sx
2
 kU2  kUx
2
 kUz  k z (2.16)

This condition on the z components of k may be seen alternatively as a condition on the

x component of the wavevector in the unsourced region by rewriting 2.16;

kUx  kU2  k z2  kU2  k S2 sin 2  I (2.17)

Recall that the problem of satisfying the boundary condition arises because above the
critical angle the z component in the sourced region is greater than the entire
wavevector in the unsourced region i.e.

k S sin  I  kU

Inspection of 2.17 then shows that kUx is given by the square root of a negative
number. In other words we have managed to satisfy the boundary condition by allowing
kUx to be imaginary i.e.

kUx   j k S2 sin 2  I  kU2   j (2.18)

Where we have written

  k S2 sin 2  I  kU2  k 0 nS2 sin 2  I  nU2 (2.19)

This gives us a spatially varying electric field in the unsourced region when this kU
inserted back into our usual plane wave field description, exp(-jk.r) ,as follows

E( x, z)  A exp  j k z z  jx   A exp( jk z z) exp(x) (2.20)

In our case (see diagram and axes) the unsourced region is in the negative x co-ordinate
region and so we need to choose the positive solution in order to avoid an unphysical

30
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

solution with an exponentially increasing field away from the boundary into the
unsourced region.
E( x, z)  A exp(x) exp( jk z z) (2.21)
This is a field with an exponentially decaying amplitude in the x direction away from
the interface that is propagating in the z direction as a wave. The amplitude has dropped
1 1
to of its value at the boundary after travelling a distance x  away from the
e 
boundary into the unsourced region and the field is called an evanescent field. and
exists only;

i) when there is total internal reflection

which only happens

ii) when the sourced region has a higher refractive index than the unsourced
region
and

iii) when the angle of incidence is greater than the critical angle.

The situation is analogous to an electron wave hitting a potential step where the
potential is greater than the electron’s total energy. The electron is reflected back with a
partial tunnelling of the electron into the classically forbidden region.

A/e Field Amplitude

-1

evanescent
wave

31
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

Reflection and Transmission at Normal Incidence

Consider an EM wave incident normally upon a dielectric interface. There are two semi-
infinite dielectrics, S and U, to the left and right in the figure below. Region S is the
sourced region (there is a source of EM waves in this region).

Sourced Unsourced
region 1 region 2

HR ER EI ET
u u
I I
u
HI HT
I I

The above diagram shows the incident, reflected and transmitted waves at normal
incidence including the E and H vectors and u the unit vector in the direction of
propagation. We note that all the fields are polarised in the plane of the interface i.e.
with reference to our earlier discussion on boundary conditions, all of the fields are
tangential to the interface and must therefore be continuous at the interface.
The sourced medium has a dielectric constant, S and unsourced medium a dielectric

 S 0 U  0
constant U. Their electromagnetic impedances are  S  and U 
 S0 U  0
respectively. The distinction between the sourced and unsourced regions will later be
important and should be noted.
Noting that the relation between E and H is through the impedance as

H  E (2.22)

Using the boundary conditions we have:

EI  ER  ET (2.23)
and
H I  H R  HT (2.24)

32
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

Relating the H fields to the E fields through the impedance we have

EI E E
 R  T (2.25)
S S U
rewritten

To find the amplitude transmission coefficient, t  ET , we can use 2.23 to eliminate

EI

E R from 2.25 and we obtain

EI ET  EI  E
  T (2.26)
S S U
Re arranging to get all E I on one side of the equation and all ET on the other

  
2 E I  1  S  ET (2.27)
 U 

Re-arranging

2U
ET  EI (2.28)
 S  U

and
E 2U
t T  (2.29)
E I U   S

E
To find the amplitude reflection coefficient r  R , we can use 2.23 to eliminate
EI

ET from 2.25 and we obtain

EI E E  ER
 R  I (2.30)
S S U
Re arranging to get all E I on one side of the equation and all E R on the other

EI EI ER E
   R (2.31)
S U U S

33
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

 U   S     S 
  E I   U  E R (2.32)
 U  S   U  S 
and
E   S
r R  U (2.33)
EI U   S

Where t and r are the electric field amplitude transmission and reflection coefficients
respectively.
We may note in passing the similarity between the result for r and the reflectivity of a
voltage pulse on a transmission line of impedance Z0 when terminated in a load
impedance ZL which is given by

Z  Z0
r L (2.34)
Z L  Z0

The coefficients may be written in terms of the refractive indices of the two media if
they are not magnetic and  S  U  1

0
nS   S S  (2.35)
 S0

Similarly for the unsourced medium.

0
nU  U U  (2.36)
U  0
Therefore we can write

0 0
S  and U  (2.37)
nS nU

the reflection and transmission coefficients for non-magnetic media are therefore;

n  nU
r S (2.38)
nS  nU

34
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

and
2n S
t (2.39)
nS  nU
NB
(i) The equations for reflection and transmission coefficients are not symmetric
in either n nor , i.e. it does matter which is the sourced region (as noted
earlier). Eg. light incident normally on water from air will reflect differently
to light incident normally on air from water. This is reflected in the
expressions for r and t which are not symmetric wrt sourced and unsourced
regions. However, we have found that the intensity, I, is proportional to the
time average of the square of the field and the quantity of experimental and
practical interest is not the amplitude reflection and transmission coefficients

r and t but rather the intensity reflection coefficient R  r 2 and the intensity

transmission coefficient T  t 2 . We will get precise expressions for these

quantities presently.
(ii) All this was done for normal incidence where importantly all the fields were
tangential to the interface and the boundary conditions derived earlier can be
appled in a straightforward manner for all fields.
(iii) When writing in terms of n rather than  the subscripts in the expressions for
r and t are interchanged in the new equations.

We have now obtained: reflection and transmission coefficients at normal incidence,

Snell’s law, the law of reflection, total internal reflection and evanescent fields. All have
been derived from the simple boundary condition that tangential electric and magnetic
fields match either side of a dielectric interface. We have only used tangential magnetic
fields in the case of deriving the reflection and transmission coefficients.
Things get a little more tricky when deriving the reflection and transmission
coefficients, r and t, for oblique incidence.

35
 Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

Reflection and Transmission at Oblique Incidence;

The Fresnel Equations.

The Fresnel equations are a set of equations that describe reflection and transmission
coefficients for EM waves incident at an interface between two homogeneous, isotropic
dielectrics, eg glass, water, for arbitrary angle of incidence.

HI nS EI
x HR x ER
EI ER HI nS
HR
z z
HT ET
nU nU

ET HT

Transverse Electric, TE Transverse Magnetic, TM

EI  ER  ET H I  H R  HT

EI E E
 R  T
S S U

H I cos  I  H R cos  I  HT cos T E I cos  I  E R cos  I  ET cos T

EI E E
cos  I  R cos  I  T cos T
S S U

TE and TM polarisations
To begin to tackle this problem it is to be noted that a wave of arbitrary polarisation will
need to be resolved into components where the electric field vector is normal to the
plane of incidence, xz plane in the above figures, (TE polarisation) and in the plane of
incidence (TM polarisation). The transmission and reflection coefficients will then need
to be found for each seperately. The above diagram shows the situation and in
constructing this diagram the right hand rule has been applied in determining the

36
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

directions of H from direction of propagation and E. ie. index finger-E, middle finger-H,
thumb-S(u or k). The magnetic field will be in the plane of incidence (TE polarisation)
or normal to the plane of incidence (TM polarisation). These polarisations are often
refered to in scientific literature as the s polarisation (TE) (from the German word for
perpendicular, senkrecht) and p polarisation (TM) (from the German word for parallel,
parallel) the polarisation of the electric field being perpendicular to and parallel to the
plane of incidence respectively.
We make this distinction between the two polarisations because the boundary
conditions require matching of the tangential components of the electric and magnetic
fields. In the case of the TE polarisation the electric field is completely tangential to the
interface whereas in the case of TM polarisation it is the magnetic field that is entirely
tangential to the interface. The boundary conditions for E and H are also shown in the
above diagrams.
We will now find the TE and TM reflection and transmission coefficients separately.

TE Polarisation.

At the boundary the BCs require:

EI Tan  ERTan  ETTan  EI  ER  ET (2.40)

Following the diagram (for the directions of the H fields) and using the BCs for H

H I Tan  H RTan  HTTan  H I cos  I  H R cos  I  HT cos T (2.41)

E
We may as usual use the relation between E and H H  to obtain the H field

boundary conditions in terms of E

EI E E
cos  I  R cos  I  T cos T (2.42)
S S U

We may now use equ 2.40 to eliminate ET from 2.42

E I ER  E  E R 
cos  I  I cos T
S U
(2.43)

Grouping EI and ER on either side of a re-arranged equ 2.43

E I U cos  I  S cos T   E R U cos  I  S cos T  (2.44)

37
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

Finally we obtain the electric field amplitude reflection coefficient for the TE
polarisation

E R U cos  I   S cos T
rETE   (2.45)
EI U cos  I   S cos T

By dividing these Fresnel equations top and bottom by cosIcosT we may obtain the
Fresnel equations in a slightly different form as follows;

 sec T   S sec  I
rETE  U (2.45a)
U sec T   S sec  I

This is the first of four Fresnel equations.

The last operation (2.45  2.45a) was performed for reasons of symmetry, the angles
associated with sourced and unsourced media are now compounded with the
impedances associated with those media and we can compare 2.45a with the equivalent
expression previously found for normal incidence, 2.33.

We may find the transmission coefficient in similar fashion for the TE polarisation
using equation 2.40 to eliminate ER in equ 2.42

ER  ET  EI (2.47)

1
2E I  ET cos  I  1 ET cos T (2.48)
S U

Grouping terms containing ET and EI

 1   1 1 
E I 2 cos  I   ET  cos  I  cos T  (2.49)
 S   S U 

Multiply both sides by SU

E I 2U cos  I   ET U cos  I  S cos T 

We obtain the amplitude transmission coefficient for the TE polarisation

38
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

ET 2U cos  I
t TE
E   (2.50)
E I U cos  I   S cos T

Again we dividing these Fresnel equations top and bottom by cosIcosT to obtain the
Fresnel equation in a more symmetric fashion as

2U sec T
t TE
E 
 S sec  I  U sec T
(2.50a)

This is the second Fresnel equation. We now need to deal with the TM polarisation in a
similar treatment to obtain the final two Fresnel equations.
NB. We can write the Fresnel equations in a number of different ways eg, 2.45 and
2.45a and 2.50 and 2.50a. Which we use is a matter of convenience, I have written
2.45a and 2.50a purely as a matter of aesthetics!

TM Polarisation.
Looking at the diagram for the TM polarised field to obtain the correct signs for the
fields we find that continuity of the tangential E field at the interface now requires

E I cos  I  E R cos  I  ET cos T (2.51)

For this polarisation the entire magnetic field is tangential to the interface thus the BC is

H I  H R  HT (2.52)

Writing 2.52 in terms of E

EI E E
 R  T (2.53)
S S U

Use 2.53 to find ET


ET  U E I  E R  (2.54)
S

39
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

And eliminating ET from 2.51


E I cos  I  E R cos  I  U E I  E R  cos T (2.55)
S

Grouping EI and ER either side of a re-arranged equ 2.55

    
E I cos  I  U cos T   E R  U cos T  cos  I 
 S   S 
(2.56)

We thus obtain the electric field amplitude reflection coefficient for the TM polarisation


cos  I  U cos T
E S  cos  I  U cos T
rETM  R   S (2.57)
EI U  S cos  I  U cos T
cos T  cos  I
S

NB we could have found the reflection and transmission coefficients for the magnetic
fields, H, and this might be considered more natural for the TM polarisation as the
complete magnetic field is tangential to the boundary plane between the two dielectrics.
The reflection coefficients would not be change for the magnetic field amplitudes as the
reflected wave and incident wave are propagating in the same medium and E and H are
related by the appropriate impedance.

ER
TM H  S  S cos  I  U cos T
rM  R   (2.57a)
HI EI  S cos  I  U cos T
S
Finally by eliminating ER from equ 2.51 using equ 2.53 rewritten as

E R  S ET  E I
U
Substituting in 2.51

2 E I cos  I  S ET cos  I  ET cos T (2.58)
U
Grouping the fields

40
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

 
2 E I cos  I  ET  S cos  I  cos T  (2.59)
 U 
Giving us the final Fresnel equation for the electric field amplitude coefficient of
transmission for the TM polarisation

ET 2 cos  I 2U cos  I

t TM
E    (2.60)
EI S  S cos  I  U cos T
cos  I  cos T
U
To write the transmission coefficient in terms of the magnetic field requires account to
be taken of the fact that the incident and transmitted waves travel in media of different
impedance and

TM ET
 HT  U  S ET 2 S cos  I
  TM
 tM   
 HI  EI U E I  S cos  I  U cos T
S
Equations 2.45, 2.50, 2.56 and 2.59 are the Fresnel equations.

THE FRESNEL EQUATIONS (Part i)

 sec T   S sec  I
rETE  U
U sec T   S sec  I

2U sec T
t TE
E 
 S sec  I  U sec T

E R  S cos  I  U cos T
rETM  
EI  S cos  I  U cos T

ET 2U cos  I
t TM
E  
E I  S cos  I  U cos T

Also as most dielectrics in which we are interested are non-magnetic 1 = 2 = 1 we

 
may use the fact that  S  0 and U  0 and write the Fresnel equations in terms
nS nU
of refractive indices rather than impedances as follows.

41
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

THE FRESNEL EQUATIONS (Part ii)

n cos  I  nU cos T
rETE  S (2.60a)
nS cos  I  nU cos T

2nS cos  I
t TE
E  (2.60b)
nS cos  I  nU cos T

n cos  I  nS cos T
rETM  U (2.60c)
nU cos  I  nS cos T

2nS cos  I
t TM
E  (2.60d)
nU cos  I  nS cos T

In the case of normal incidence where  I  T  0 there is no distinction between TE

and TM and we can check this by verifying that we get the same reflection coefficient
for TE and TM waves at normal incidence. At 00 the secant is 1 and the cosine is 1

n cos  I  nU cos T nS  nU
rETE  S  (2.61a)
nS cos  I  nU cos T nS  nU

n cos  I  nS cos T nU  nS
rETM  U  (2.61b)
nU cos  I  nS cos T nU  nS

Thus we have identical absolute magnitudes for the amplitude reflection coefficients but
a change in sign. This should concern us as the TE and TM waves are equivalent in this
case. The problem occurs at the start of this section on page 36 where the TE and TM
waves were drawn in order to obtain the boundary conditions of equs 2.29 and 2.40. In
constructing those figures the assumptions were that the right hand rule applies in
determining the directions of H from direction of propagation and E. ie. right hand,
index finger-E, middle finger-H, thumb-S(u or k). This procedure is of course correct in
principle but something else occurs at the interface which we have ignored.
We made a further assumption that the electric field doesn’t change direction in the
reflected wave for TM. In fact the E field reverses upon reflection. This implies a phase
change upon reflection which we have ignored. It also alters the outcome of applying
the boundary conditions. Apart from this phase change the technique is correct and as

42
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

we want the intensity reflection coefficients the phase change plays no role. To find the
intensity reflection coefficients we will square the amplitude reflection coefficient and
therefore the sign is irrelevant.

To obtain intensity reflection coefficients and transmission

coefficients requires a bit more work. So far we have been interested in the electric
field or amplitude reflection coefficients. As we have already learned the intensity is
related to the electric field by

E 2 1  0 2 1  1 n 2
I  E0  n 0 E02  E0 (2.62)
2 2  0 2 0 2 0

AI AR

I R

T

AT

43
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

For the reflection coefficients the situation is simple as the reflected and incident waves
are in the same medium and n has the same value when converting EI and ER and
therefore the reflection coefficient is simply

E R2
A
PR I R  AR 2 S R
R    r2 (2.63)
PI I I  AI 2
EI
A
2 S R
for both TE and TM.

2
TE  n S cos  I  nU cos T 
RE   (2.64a)
 nS cos T  nU cos  I 

2
TM  n cos  I  nS cos T 
RE  U  (2.64c)
 nU cos  I  nS cos T 

For the transmission coefficients the incident and transmitted waves are travelling in
different media and the refractive indices will differ. This must be accounted for along
with the fact that it is the power flow normal to the direction of propagation that we are
interested in. This introduces a cosine factor as the area normal to the direction of
propagation

   
ANorm  A  S  A  ki  A cos  (2.65)

 
is altered by this factor. A is the area on the interface, k and S are the unit wavevector
and Poynting vector respectively.

ET
Using t TE TM
E ,t E 
EI

ET2
P I T AT I T A cos T 2U cos T
TETE , TETM  T    (2.66)
PI I I AI I I A cos  I E I2 cos  I
2 S

44
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

 S cos T n cos T
TETE , TETM  t 2  t2 U
U cos  I nS cos  I
Using the expression for tTE from equ 2.50b in 2.56

2
 2nS cos  I  nU cos T
TETE    (2.66b)
 nS cos  I  nU cos T  nS cos  I

Using the expression for tTM from equ 2.50d in 2.56

2
 2nS cos  I  nU cos T
TETM    (2.66d)
 nU cos  I  nS cos T  nS cos  I

We should note that all of the Fresnel equations, as shown, involve both the incident
angle, I , and the transmitted angle, T. To obtain reflection and transmission
coefficients in terms of I alone it would be necessary to use Snell’s law to eliminate T.

Conservation of Energy
To conserve energy at the boundary we are considering, the power incident must be , in
the absence of absorption, equal to the sum of the power reflected and the power
transmitted
PI  PR  PT (2.68)

Using the definitions of the reflectance and transmittance

P P
R R T T (2.69)
PI PI
therefore
1 R T (2.70)

45
 Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

The Stokes Relations

Fermat’s principle of least action requires that, in the absence of dissipation in the form
of absorption, any viable ray diagram (such as the usual diagram shown in figure a) of
a ray incident at a dielectric interface undergoing refraction and reflection) will also
work with all ray directions reversed as in figure b). An argument due to Stokes
invokes Fermat to demonstrate a relation between the amplitude transmission and
reflection coefficients, r and t.

r(rEi) t/(tEi)
Ei rEi Ei rEi rEi

I I

tEi tEi tEi

r/(tEi)

T t(rEi)

a) b) c)
Clearly, in figure b) there is an internal reflection missing. Also the reflected ray in the
upper half of the diagram should be a composite of a refracted ray from the incident ray
in the lower half and the incident ray in the upper half. These component rays have been
added in figure c). Figures b) and c) must be equivalent.
If the configuration in figures c) and b) are equivalent then it follows that

tt / Ei  rrEi  Ei (2.71)

and

rtEi  tr / Ei  0 (2.72)

46
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

From 2.71

tt /  1  r 2 (2.73)

From 2.72

r  r / (2.74)

We need to be a bit circumspect and note that the amplitude reflection and transmission
coefficients depend on angle and the equations 2.73 and 2.74 need to include this fact
explicitly and so are written

r ( I )  r / (T ) (2.73a)

t ( I )t / (T )  1  r 2 ( I ) (2.74a)

Equations 2.73 and 2.74 are known as the Stokes relations

Of course we bear in mind that I and T always come as a pair of angles related
through Snell’s law.

47
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

Brewster’s Angle
One consequence of the tunability (by angle) of the impedance of a dielectric medium is
that it is possible in certain circumstances to find an angle where the impedance of the
unsourced medium matches that of the sourced medium and at that angle the reflectivity
becomes zero.
For TM waves, the angle at which r becomes zero is known as Brewster’s angle.

We can find this angle by considering the Fresnel equation for amplitude reflection
coefficient

TM  cos  I  U cos T
rM  S
 S cos  I  U cos T

In non-magnetic media this may be rewritten;

TM n cos  I  nS cos T
rM  U
nU cos  I  nS cos T

TM
We see that rM  0 when nS cos T  nU cos  I

i.e. at some angle of incidence when I = B the reflectivity becomes zero.

Re-arranging we have

2
n 
cos T   U
2  cos 2  B (2.75)
 nS 

From Snell’s law

n
sin T  S sin  B
nU

Now we can use a trigonometric identity to write

2 2
n  n 
cos T  sin T  1   U
2 2  cos 2  B   S  sin 2  B  cos 2  B  sin 2  B (2.76)
 nS   nU 

Collecting sin and cos terms together

48
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

 n 
2   n 
2
 S   1 sin 2  B  1   U   cos 2  B (2.77)
 nU     nS  
   

2 n S2  nU2
1  nU 
 n S  n S2 n2
tan 2  B    U (2.78)
2 2 2
 nS   1 n S  nU n S2
 n 
 U nU2

The Brewster angle for TM waves is given by

n
tan  B  U (2.79)
nS

When I = B for a TM wave then rM

TM
 0  rETM

NB. It is important that the equation is used the correct way around with the
refractive index of the unsourced region appearing in the numerator! This is
another example of the importance of noting the difference between the sourced
and unsourced regions when using the equations derived here.

For TE waves we may follow the same methodology;

n cos I  nU cos T
rETE  S
nS cos I  nU cos T

For  I   B rETE  0

nS cos B  nU cos T (2.80)

Re-written

nS
cos B  cos T (2.81)
nU

Following the same procedure

49
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

2 2
n  n 
cos T  sin T  1   S
2 2  cos 2  B   S  sin 2  B  cos 2  B  sin 2  B (2.82)
 nU   nU 

Again collecting sin and cos terms

 n 
2   n 
2
 S   1 sin  B  1   S
 2   cos 2  B (2.83)
 nU     nU  
   

2 nU2  n S2
1   S 
n
2  nU  nU2
tan  B    1 (2.84)
2 2 2
 nS   1 n S  nU
 n 
 U
nU2

While tanB may take on any value from - to  this would require tanB to be
imaginary!

Clearly the condition for rETE  0 cannot be satisfied and there is NO

Brewster angle for the TE polarisation.

We are able to deduce the expression for the Brewster angle geometrically as follows

nS
B
B

B
T
nU T

Recalling that the polarisation is perpendicular to the direction of propagation and that
the E polarisation of the medium gives rise to the secondary waves at that polarisation
and frequency that are the reflected waves, we can see that when the refracted ray is at

50
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

right angles to the direction the reflected wave should propagate in according to laws
of reflection then, in the case of TM waves there can be no reflected wave.

The polarisation induced by the incident wave would lie in the same direction as the
reflected wave which is not allowed. Therefore at this particular angle of incidence
where the reflected ray would be at right angles to the transmitted ray there is zero
reflection coefficient. This is therefore the Brewster angle.

We find the Brewster angle as follows from the ray diagram

nS sin B  nU sinT  nU sin( 90   B )  nU cos  B

n
tan  B  U
nS

This is the same as the answer we found previously.

NB.

1) The result gives the tangent of the Brewster angle as

index of the unsourced region divided by index of the sourced region.

2) The Brewster angle exists only for the TM wave a similar angle cannot be
described for a TE wave. It will always be possible to derive a secondary
(reflected) wave from the polarisations induced by a TE wave without going
against any fundamental principle.

3) The Brewster angle exists irrespective of the relative size of the refractive
indices (unlike the critical angle which exists only when nS > nU ).
We note (as is evident in the construction) that for nU > nS , B is greater than 450 and
for nU < nS , B is less than 450.

51
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

The Brewster angle is used to produce polarising elements in laser tubes where
Brewster windows allow TE polarised waves to suffer greater losses (due to reflection)
than TM waves.

52
Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

Highlights of Part 2
Law of Reflection I  R

Law of Refraction (Snell’s Law) nS sin  I  nU sin T ***

n
Critical Angle C  sin 1 S nS  nU
nU

n
Brewster Angle  B  tan 1 U
nS

Evanescent wavevector is   k12 sin 2  I  k 22  k 0 n12 sin 2  I  n22

E R U   S nS  nU
Reflection coeff, normal incidence r  
EI U   S nS  nU

E 2U 2n S
Transmission coeff, normal incidence t T  
E I U   S nS  nU

*** The subscripts S and U used here refer to sourced and unsourced regions
respectively.
The subscripts I , R and T refer to incidence, reflection and transmission
angles respectively.

53
 Electromagnetic Waves & Optics: Lecture Notes © Kevin Donovan

Highlights of Part 2 cont.

E R nS cos  I  nU cos T
rETE  
EI nS cos  I  nU cos T

ET 2nS cos  I
t TE
E  
EI nS cos  I  nU cos T
The Fresnel equations (non magnetic)
E R nU cos  I  nS cos T
The amplitude coefficients rETM  
EI nU cos  I  nS cos T

ET 2nS cos  I
t TM
E  
EI nU cos  I  nS cos T

2
 n cos  I  nU cos T 
RTE   S 
 nS cos T  nU cos  I 

2
 2nS cos  I  nU cos T
TTE   
 nS cos  I  nU cos T  nS cos  I

The Fresnel equations (non magnetic)

The intensity coefficients
2
TM  n cos  I  nS cos T 
R  U 
 nU cos  I  nS cos T 

2
TM  2nS cos  I  nU cos T
T  
 nU cos  I  nS cos T  nS cos  I

54