PHYS 263

Physical Optics
Lecture Notes
Jakob J. Stamnes

Department of Physics, University of Bergen, 5007 Bergen.

Tel: 55 58 28 18. Fax: 55 58 94 40. E-mail: JakobJ.Stamnes@.uib.no
Spring 2003
Autumn 2004
PHYS261-OPTICS PART, PART 1
FYS 263 1
Contents
I Elementary electromagnetic waves 3
1 Maxwells equations, the material equations, and boundary conditions 4
1.1 Maxwells equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 The continuity equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 The material equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Poyntings vector and the energy law 8
3 The wave equation and the speed of light 9
4 Scalar waves 11
4.1 Plane waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4.2 Spherical waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.3 Harmonic (monochromatic) waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.4 Complex representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.5 Linearity and the superposition principle . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.6 Phase velocity and group velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.7 Repetition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5 Pulse propagation in a dispersive medium 18
6 General electromagnetic plane wave 21
7 Harmonic electromagnetic waves of arbitrary form - Time averages 24
8 Harmonic electromagnetic plane wave Polarisation 26
9 Reection and refraction of a plane wave 29
9.1 Reection law and refraction law (Snells law) . . . . . . . . . . . . . . . . . . . . . . 29
9.2 Generalisation of the reection law and Snells law . . . . . . . . . . . . . . . . . . . 31
9.3 Reection and refraction of plane electromagnetic waves . . . . . . . . . . . . . . . . 32
9.3.1 Reectance and transmittance . . . . . . . . . . . . . . . . . . . . . . . . . . 35
9.3.2 Brewsters law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
9.3.3 Unpolarised light (natural light) . . . . . . . . . . . . . . . . . . . . . . . . . 38
9.3.4 Rotation of the plane of polarisation upon reection and refraction . . . . . . 39
9.3.5 Total reection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
FYS 263 2
List of Figures
1.1 A plane interface with unit normal n separates two dierent dielectric media. . . . . 7
4.1 A plane wave that propagates in direction s, has no variation in any plane that is
normal to s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
5.1 A plane wave propagates in the positive z direction in a dispersive medium that lls
the half space z 0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
6.1 The vectors E, H, and s for an electromagnetic plane wave represent a right-handed
Cartesian co-ordinate system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
8.1 Instantaneous picture of the electric vector of a plane wave that propagates in the z
direction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
8.2 The end point of the electric vector describes an ellipse that is inscribed in a rectangle
with sides 2a
1
and 2a
2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
9.1 Reection and refraction of a plane wave at a plane interface between two dierent
media. Illustration of propagation directions and angles of incidence, reection, and
transmission. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
9.2 Reection and refraction of a plane wave. Illustration of the co-ordinate system ( n,

b,

t). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
9.3 Generalisation of Snells law and the reection law to include non-planar waves that
are incident upon a curved interface. . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
9.4 Reection and refraction of a plane electromagnetic wave at a plane interface between
two dierent media. Illustration of TE and TM components of the electric eld. . . 35
9.5 Illustration of the angle
q
between the electric vector E
q
and the plane of incidence
spanned by k
q
and e
TMq
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
9.6 Illustration of Brewsters law. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
9.7 Illustration of the refraction of a plane wave into an optically thinner medium, so that

i
<
t
. When
i

ic
, then
t
/2, and we get total reection. . . . . . . . . . . 41
FYS 263 3
Part I
Elementary electromagnetic waves
FYS 263 4
Chapter 1
Maxwells equations, the material
equations, and boundary
conditions
In this course we consider light to be electromagnetic waves of frequencies in the visible range, so
that (4 7.5) 10
14
Hz. Since =
c

, where c is the speed of light in vacuum (c 3 10

8
m/s), we nd that the corresponding wavelength interval is (0.4 0.75) m. Thus, to study
the propagation of light we must consider the propagation of the electromagnetic eld, which is
represented by the two vectors E and B, where E is the electric eld strength and B is the magnetic
induction or the magnetic ux density. To enable us to describe the interaction of the electromagnetic
eld with material objects we need three additional vector quantities, namely the current density J,
the displacement D, and the magnetic eld strength H.
1.1 Maxwells equations
The ve vectors mentioned above are linked together by Maxwells equations, which in Gaussian
units are
H =
1
c

D+
4
c
J, (1.1.1)
E =
1
c

B. (1.1.2)
In addition we have the two scalar equations
D = 4, (1.1.3)
B = 0, (1.1.4)
where is the charge density. Equation (1.1.3) can be said to dene the charge density . Similarly,
we can say that (1.1.4) implies that free magnetic charges do not exist.
1.2 The continuity equation
The charge density and the current density J are not independent quantities. By taking the
divergence of (1.1.1) and using that (A) = 0 for an arbitrary vector A, we nd that
J +
1
4

D = 0,
FYS 263 5
which on using (1.1.3) gives
J + = 0. (1.2.1)
This equation is called the continuity equation, and it expresses conservation of charge. By inte-
grating (1.2.1) over a closed volume V with surface S, we nd
___
V
Jdv =
___
V

t
dv, (1.2.2)
which by use of the divergence theorem gives
_
__
S
J nda =
d
dt
___
V
dv =
d
dt
Q. (1.2.3)
Here n is the unit surface normal in the direction out of the volume V , so that (1.2.3) shows that
the integrated current ux out of the closed volume V is equal to the loss of charge in the same
volume.
Digression 1: Notation
Bold face is used to denote vector quantities, e.g.
E = E
x
e
x
+ E
y
e
y
+ E
z
e
z
,
where e
x
, e
y
, and e
z
are unit vectors along the axes in a Cartesian co-ordinate system.
A dot above a symbol is used to denote the time derivative, e.g.

B =

t
B.
E, B, D, H, , and J are functions of the position r and the time t, e.g.
D = D(r, t).
The connection between Gaussian and other systems of units, e.g. MKS units, follows from
J.D. Jackson, Classical Electrodynamics, Wiley (1962), pp. 611-621. For conversion between
Gaussian units and MKS units, we refer to the table on p. 621 in this book.
1.3 The material equations
Maxwells equations (1.1.1)-(1.1.4), which connect the fundamental quantities E, H, B, D, and J,
are not sucient to uniquely determine the eld vectors (E, B) from a given distribution of currents
and charges. In addition we need the so-called material equations, which describe how the eld is
inuenced by matter.
In general the material equations can be relatively complicated. But if the eld is time harmonic
and the matter is isotropic and at rest, the material equations have the following simple form
J
c
= E, (1.3.1)
D = E, (1.3.2)
FYS 263 6
B = H, (1.3.3)
where is the conductivity, is the permittivity or dielectric constant, and is the permeability.
Equation (1.3.1) is Ohms law, and J
c
is the conduction current density, which arises because
the material has a non-vaninishing conductivity ( ,= 0). The total current density J in (1.1.1) can
in addition consist of an externally applied current density J
0
, so that
J = J
0
+J
c
= J
0
+ E. (1.3.4)
Digression 2: General material considerations
A material that has a non-negligible conductivity is called a conductor, while a material that
has a negligible conductivity is called an insulator or a dielectric.
Metals are good conductors.
Glass is a dielectric; 2.25; = 0; = 1.
In anisotropic media (e.g. crystals) the relation in (1.3.2) is to be replaced by D = E, where
is a tensor, dyadic or matrix.
In a plasma (1.3.1) is to be replaced by J = E, where the conductivity is a tensor.
There are also magnetically anisotropic media, in which (1.3.3) is to be replaced by B = H.
Thus, in this case the permeability is a tensor. Such materials are not important in optics.
In dispersive media is frequency dependent, i.e. = (). Maxwells equations and the
material equations are still valid for each frequency component or time harmonic component
of the eld. For a pulse consisting of many frequency components, one must apply Fourier
analysis to solve Maxwells equations and the material equations separately for each time
harmonic component, and then perform an inverse Fourier transformation.
In non-linear media there is no linear relation between D and E (equation (1.3.2) is not valid).
Most media become non-linear when the electric eld strength becomes suciently high.
1.4 Boundary conditions
Hitherto we have assumed that and are continuous functions of the position. But in optics
we often have systems consisting of several dierent types of glass. At the transition between air
and glass or between two dierent types of glass the material parameters are discontinuous. Let us
therefore consider what happens to the electromagnetic eld at the boundary between two media.
Consider two media that are separated by an interface, as illustrated in Fig. 1.1. From Maxwells
equations, combined with Stokes and Gauss theorems, one can derive the following boundary
conditions
n (B
(2)
B
(1)
) = 0, (1.4.1)
n (D
(2)
D
(1)
) = 4
s
, (1.4.2)
n (E
(2)
E
(1)
) = 0, (1.4.3)
n (H
(2)
H
(1)
) =
4
c
J
s
, (1.4.4)
where n is a unit vector along the surface normal. According to (1.4.1) the normal component of
B is continuous across the boundary, while (1.4.2) says that if there exists a surface charge density
FYS 263 7

2
^
1
,
2
n
,
Figure 1.1: A plane interface with unit normal n separates two dierent dielectric media.

s
at the boundary, then the normal component of D is changed by 4
s
across the boundary
between the two media. According to (1.4.3) the tangential component of E is continuous across
the boundary, while (1.4.4) implies that if there exists a surface current density J
s
at the boundary,
then the tangential component of H, i.e. of n H, is changed by
4
c
J
s
.
FYS 263 8
Chapter 2
Poyntings vector and the energy
law
The electric energy density w
e
and the magnetic energy density w
m
are dened by
w
e
=
1
8
E D, (2.1)
w
m
=
1
8
H B, (2.2)
and the total energy density is the sum of these, i.e.
w = w
e
+ w
m
. (2.3)
The energy ux of the eld is represented by Poyntings vector S, given by
S =
c
4
EH. (2.4)
Here S represents the amount of energy that per unit time crosses a unit area that is parallel with
both E and H.
In a non-conducting medium ( = 0) we have the conservation law
w
t
+ S = 0, (2.5)
which expresses that the change of the energy density in a small volume is equal to the energy ux
out of the same volume [cf. (1.2.2) and (1.2.3)]. In optics the Poynting vector is very important,
because its absolute value is proportional to the light intensity, i.e.
[S[ light intensity. (2.6)
The direction of the Poynting vector, dened by the unit vector
s =
S
[S[
, (2.7)
points in the direction of light propagation.
FYS 263 9
Chapter 3
The wave equation and the speed
of light
The electric and magnetic elds E and H are connected through Maxwells equations (1.1.1)-(1.1.4),
which are simultaneous, rst-order partial dierential equations. But in those parts of space where
there are no sources, so that J = 0 and = 0, we can through dierentiation obtain second-order
partial dierential equations that E and H satisfy individually. We assume that the medium is
non-dispersive, so that D = E, where = 0, and B = H, where = 0. Then we have from (1.1.1)
and (1.1.2)
H =
1
c

D =
1
c

E, (3.1)
E =
1
c

B =
1
c

H. (3.2)
Next, we assume that the medium is homogeneous, so that and do not vary with position. By
taking the curl of (3.2) and combining the result with the time derivative of (3.1), we nd that
(E) =
1
c

H =
1
c

1
c

E =

c
2

E. (3.3)
Now we use the vector relation
(A) = ( A)
2
A, (3.4)
which applies to an arbitrary vector A, to obtain
( E)
2
E =

c
2

E, (3.5)
which since E = 0, gives

2
E

c
2

E = 0. (3.6)
In a similar manner we nd

2
H

c
2

H = 0. (3.7)
By comparing these results with the scalar wave equation

2
V
1
v
2

V = 0, (3.8)
we see that in a source-free region of space each Cartesian component of E and H satises the scalar
wave equation with phase velocity
FYS 263 10
v =
c

. (3.9)
Note that this derivation is valid only in a non-dispersive medium in which both the permittivity
and the permeability do not depend on the frequency.
FYS 263 11
Chapter 4
Scalar waves
Scalar waves are solutions of the scalar wave equation (3.8), which is given by

2
V (r, t)
1
v
2

2
t
2
V (r, t) = 0. (4.0.1)
4.1 Plane waves
Any solution of (4.0.1) of the form
V (r, t) = V (r s, t), (4.1.1)
is called a plane wave, since V at any time t is constant over any plane
r s = constant, (4.1.2)
which is normal to the unit vector s (see Fig. 4.1).
To show that (4.1.1) is a solution of (4.0.1), we introduce a new variable
= r s = xs
x
+ ys
y
+ zs
z
, (4.1.3)
so that

x
= s
x
;

y
= s
y
;

z
= s
z
. (4.1.4)
Further we nd that
V
x
=
V

x
= s
x
V

. (4.1.5)

2
V
x
2
=

x
_
s
x
V

_
= s
x

x
_
V

_
= s
x

_
V

_

x
= s
2
x

2
V

2
. (4.1.6)
In a similar way we nd

2
V
y
2
= s
2
y

2
V
y
2
;

2
V
z
2
= s
2
z

2
V
z
2
. (4.1.7)
When we substitute (4.1.6) and (4.1.7) in (4.0.1) and take into account that s
2
x
+s
2
y
+s
2
z
= 1, since
s is a unit vector, the wave equation becomes

2
V

2

1
v
2

2
V
t
2
= 0. (4.1.8)
FYS 263 12
n
^

y
x
z
e
e
e
^
^
^

r
^
z
e
e
e

r
^
^
^

n
^
n
Figure 4.1: A plane wave that propagates in direction s, has no variation in any plane that is normal
to s.
By introducing two new variables p and q, dened by
p = vt ; q = + vt, (4.1.9)
we nd (Exercise 2) that the wave equation in (p, q) variables can be written

2
V
pq
= 0. (4.1.10)
This equation has the following general solution
V = V
1
(p) + V
2
(q), (4.1.11)
where V
1
and V
2
are arbitrary functions. By substitution from (4.1.3) and (4.1.9) in (4.1.11), we
nd the following general plane-wave solution
V (r, t) = V
1
(r s vt) + V
2
(r s + vt). (4.1.12)
Note that
vt = + v v(t + ), (4.1.13)
so that
V
1
(, t) = V
1
( + v, t + ). (4.1.14)
Equation (4.1.14) shows that during the time , V
1
is displaced a length s = v in the positive
direction, i.e. V
1
propagates with velocity v in the positive direction. The conclusion is that
V ( vt) represents a plane wave that propagates at velocity v in the positive direction (lower
sign) or in the negative direction (upper sign).
4.2 Spherical waves
Consider now solutions of the scalar wave equation (4.0.1) of the form
V = V (r, t), (4.2.1)
where
r = [r[ =
_
x
2
+ y
2
+ z
2
, (4.2.2)
FYS 263 13
is the distance from the origin (0, 0, 0). Since we have no angular dependence in this case, the
Laplacian operator has the following form in spherical coordinates (Exercise 3)

2
V =
1
r

2
r
2
(rV ), (4.2.3)
which upon substitution in the wave equation (4.0.1) gives

2
r
2
(rV )
1
v
2

2
t
2
(rV ) = 0. (4.2.4)
Since (4.2.4) is of the same form as (4.1.8), the solution becomes (cf. (4.1.12))
rV = V
1
(r vt) + V
2
(r + vt). (4.2.5)
Thus, we have obtained the following result:
V (rvt)
r
represents a spherical wave that converges
towards the origin (upper sign) or diverges away from the origin (lower sign). Thus,
V (r+vt)
r
prop-
agates towards the origin with velocity v, whereas
V (rvt)
r
propagates away from the origin with
velocity v.
4.3 Harmonic (monochromatic) waves
At a given point r in space the solution of the wave equation is a function only of time, i.e.
V (r, t) = F(t), (4.3.1)
where F(t) can be an arbitrary function. If F(t) has the simple form
F(t) = a cos(t ), (4.3.2)
then we have a harmonic wave in time. The quantities in (4.3.2) have the following meaning: a is the
amplitude (positive), is the angular frequency, and t is the phase. A harmonic wave is also
called a monochromatic wave because it consists of only one frequency or wavelength component.
The frequency and the period T follow from
=

2
=
1
T
. (4.3.3)
The harmonic wave in (4.3.2) has period T because
F(t + T) = a cos((t + T) ) = a cos(t + 2) = F(t). (4.3.4)
From (4.1.12) we see that the general expression for a wave that propagates in the s direction
can be written
V = V
1
(r s vt) = V
1
_
v
_
t
r s
v
__
= V

1
_
t
r s
v
_
, (4.3.5)
where both V
1
and V

1
are arbitrary functions. By replacing t with t
rs
v
in (4.3.2) we get a harmonic
plane wave
V (r, t) = a cos
_

_
t
r s
v
_
+
_
= a cos[kr s t + ], (4.3.6)
where
k =

v
(4.3.7)
is the wave number. We see that that V (r, t) remains unchanged if we replace r s with r s + n,
where n = 1, 2, . . ., and is given by
FYS 263 14
=
2
k
= v
2

= vT =
v

. (4.3.8)
The quantity is called the wavelength. Note that for t = constant, V (r, t) in (4.3.6) is periodic
with wavelength , i.e.
V (r s, t) = V (r s + n, t) ; n = 1, 2, 3, . . . . (4.3.9)
Now we introduce the wave vector or propagation vector k, dened by
k = ks. (4.3.10)
so that the expression (4.3.6) for a plane, harmonic wave can be written
V (r, t) = a cos(k r t + ). (4.3.11)
In a similar way the expression for a converging or a diverging harmonic spherical wave becomes
V (r, t) = a
cos(kr t + )
r
, (4.3.12)
where the upper sign corresponds to a converging spherical wave and the lower sign to a diverging
spherical wave.
Consider now a plane, harmonic wave that propagates in the positive z direction, so that [cf.
(4.3.11)]
V (z, t) = a cos(kz t + ). (4.3.13)
A wave front is dened by the requirement that the phase shall be constant over it, i.e.
= kz t + = constant. (4.3.14)
Hence it follows that on a wave front we have
z = vt + constant ; v =

k
. (4.3.15)
Thus, the wave front propagates at the velocity
v =

k
, (4.3.16)
which is called the phase velocity.
4.4 Complex representation
Alternatively we can express (4.3.11) and (4.3.12) in the following way
V (r, t) = ReU(r)e
it
, (4.4.1)
where Re. . . stands for the real part of . . ., and where the complex amplitude U(r) is given by
U(r) = ae
i(kr+)
, (4.4.2)
for a plane wave, and by
U(r) =
a
r
e
i(kr+)
, (4.4.3)
for a diverging (upper sign) or converging (lower sign) spherical wave.
FYS 263 15
Note that when we perform linear operations, such as dierentiation, integration or summation,
we can drop the Re symbol during the operations, as long as we remember to take the real part of
the result in the end.
By substituting
V (r, t) = U(r)e
it
, (4.4.4)
in the wave equation (4.0.1), we get
(
2
+ k
2
)U(r) = 0, (4.4.5)
which shows that the complex amplitude U(r) is a solution of the Helmholtz equation.
4.5 Linearity and the superposition principle
For any linear equation the sum of two or several solutions is also a solution. This is called the
superposition principle. Since Maxwells equations are linear, the superposition principle is valid
for electromagnetic waves as long as the material equations are linear. The superposition principle
implies that we can construct general solutions of the wave equation or Maxwells equations by
adding elementary solutions in the form of harmonic plane or spherical waves. We will discuss this
in detail in Part II.
4.6 Phase velocity and group velocity
Consider a harmonic wave of the form [cf. (4.3.11)]
V (r, t) = Re
_
U(r)e
it

, (4.6.1)
where the complex amplitude U(r) is a solution of the Helmholtz equation (4.4.5), i.e.
(
2
+ k
2
)U(r) = 0. (4.6.2)
The wave number k can be written
k =

v
=

c
_
c
v
_
= k
0
n, (4.6.3)
where k
0
is the wave number in vacuum, i.e.
k
0
=

c
, (4.6.4)
and n is the refractive index given by
n =
c
v
=

. (4.6.5)
A general wave V (r, t) can always be expressed as a sum of harmonic components. We will return
to this later. If depends on , i.e. = (), then the phase velocity also will depend on , since
v =
c
n
= v(). This means that dierent harmonic components will propagate at dierent phase
velocities. A polychromatic wave or a pulse, which is comprised of many harmonic components,
therefore will change its shape during propagation, and the energy will not propagate at the phase
velocity, but at the group velocity, which is dened as
v
g
=
d
dk
. (4.6.6)
If n() = constant, we have a non-dispersive medium. Since
= vk, (4.6.7)
FYS 263 16
where the phase velocity v =
c
n
now is constant, we have in this case
v
g
=
d
dk
(vk) = v.
Thus, the phase velocity and the group velocity are equal in a non-dispersive medium where n =
constant. In dispersive media we have
v
g
=
d
dk
(vk) = v + k
dv
dk
= v
dv
d
= v +
dv
d
, (4.6.8)
where the last two results follow from the relation k =
2

=

v
.
4.7 Repetition
From Maxwells equations in source-free space (J = 0 ; = 0) we nd

2
E

c
2

E = 0 ;
2
H

c
2

H = 0. (4.7.1)
Comparison of (4.7.1) with the scalar wave equation

2
V
1
v
2

V = 0, (4.7.2)
shows that any Cartesian component of E and H satises the scalar wave equation with phase
velocity v given by
v =
c

=
c
n
. (4.7.3)
The scalar wave equation (4.7.2) has simple solutions in the form of plane waves or spherical waves.
Plane waves
For a plane wave V is given by
V (r, t) = V
1
(r s vt) + V
2
(r s vt), (4.7.4)
where V ( vt) represents a plane wave that propagates in the positive direction (upper sign) or
in the negative direction (lower sign).
Spherical waves
For a spherical wave V is given by
V (r, t) =
V
1
(r vt)
r
+
V
2
(r + vt)
r
, (4.7.5)
where
V (rvt)
r
represents a spherical wave that propagates away from the origin (upper sign) or
towards the origin (lower sign).
Harmonic (monochromatic) waves
A plane harmonic wave that propagates in the direction k = ks is given by
V (r, t) = a cos(k r t + ), (4.7.6)
and the corresponding spherical wave is
V (r, t) =
a
r
cos(kr t + ), (4.7.7)
where the upper sign represents a diverging spherical wave and the lower sign represents a converging
spherical wave.
FYS 263 17
Complex representation of harmonic waves
In complex notation we have
V (r, t) = Re[U(r)e
it
]. (4.7.8)
For a plane wave the complex amplitude U(r) is given by
U(r) = ae
i(kr+)
,
and for a diverging or converging spherical wave it is given by
U(r) =
a
r
e
i(kr+)
.
By substituting (4.7.8) into the wave equation (4.7.2), we nd that U(r) satises the Helmholtz
equation, i.e.
(
2
+ k
2
)U(r) = 0. (4.7.9)
FYS 263 18
Chapter 5
Pulse propagation in a dispersive
medium
z=0
z
n( )
Figure 5.1: A plane wave propagates in the positive z direction in a dispersive medium that lls the
half space z 0.
Consider a polychromatic, plane wave that propagates in the positive z direction in a linear,
homogeneous, isotropic, and dispersive medium that lls the half space z > 0 (Fig. 5.1). The
polychromatic, plane wave u(z, t) is comprised of dierent harmonic components, which implies that
we can represent u(z, t) by the following inverse Fourier transform
u(z, t) =
1
2
_

u(z, )e
it
d, (5.1)
where the frequency spectrum u(z, ) is given as the Fourier transform of u(z, t), i.e.
u(z, ) =
_

u(z, t)e
it
dt. (5.2)
Thus, u(z, t) and u(z, ) constitute a Fourier transform pair. Since u(z, ) can be any Cartesian
component of the frequency spectrum of the electric or magnetic eld, it satises the Helmholtz
equation, i.e.
[
2
+ k
2
()] u(z, ) = 0, (5.3)
where
k() =

v()
=

c
c
v()
=

c
n(). (5.4)
FYS 263 19
Suppose now that u(z, t) is known for all values of t in the plane z = 0, and that u(0, t) vanishes for
t < 0.
Since there is no variation in the x and y directions, the Helmholtz equation (5.3) can be written
as
_
d
2
dz
2
+ k
2
()
_
u(z, ) = 0, (5.5)
which has the following general solution
u(z, ) = u
+
()e
ik()z
+ u

()e
ik()z
. (5.6)
If we consider propagation in the positive z direction only, then u

() = 0, so that (5.1) gives

u(z, t) =
1
2
_

u
+
()e
i(k()zt)
d. (5.7)
Now we put z = 0 i (5.7), take an inverse Fourier transform, and use (??) to obtain
u
+
() =
_

u(0, t)e
it
dt = u(0, ), (5.8)
so that (5.7) gives
u(z, t) =
1
2
_

u(0, )e
i(k()zt)
d, (5.9)
or
u(z, t) =
1
2
_

u(0, )e
i
z
c
f()
d, (5.10)
where
f() = [n() ] ; =
ct
z
. (5.11)
Consider rst the special case in which n() =
c
v()
= constant, which implies that we have a
non-dispersive medium. Since k =

v
, where v now is constant, we have from (5.1) and (5.9)
u(z, t) =
1
2
_

u(0, )e
i(
z
v
+t)
d = u
_
0, t
z
v
_
. (5.12)
This result shows that the pulse propagates in the positive z direction at velocity v without changing
its shape.
Suppose now that the medium is dispersive and that the frequency spectrum g() of the pulse
in (5.10) does not contain singularities and that it is suciently wide. Then the main contribution
to the pulse in (5.10) comes from frequencies
s
for which the phase f() in (5.11) is stationary, i.e.
from
s
that satisfy the equation
f

(
s
) = n(
s
) +
s
n

(
s
) = 0. (5.13)
A model that is commonly used to study propagation in dispersive media, is the so called Lorentz-
medium. For such a medium with one single resonance frequency the refractive index n() is given
by the following expression
n() =
_
1
b
2

2
0
+ 2i
_
1/2
, (5.14)
where b is a constant,
0
is the resonance frequency, and represents the damping (attenuation) in
the medium.
FYS 263 20
Equation (5.9) shows that when the medium is dispersive, then u(z, t) (for any z > 0) is a sum
of harmonic plane waves of the form u(0, )exp[i(k()z t)] = u(0, )exp[k
i
()z]exp[i(k
r
()z
t)], where k
r
() and k
i
() are the real and the imaginary part, repectively, of k(). Thus, the
amplitude u(0, )exp[k
i
()z], is damped exponentially as z increases, and the phase velocity is
given by v() =

k
r
()
, where k() = (/c)n() = (/c)[n
r
() + in
i
()] = k
r
() + ik
i
(). Since
the phase velocity v depends on the frequency , plane waves of dierent frequencies will arrive at
a given position z at dierent times and thus cause a distortion of the pulse, i.e. the shape of the
pulse will get changed. Also, the damping factor k
i
() depends on , so that dierent frequency
components will have dierent amplitudes when they arrive at a given position z.
FYS 263 21
Chapter 6
General electromagnetic plane
wave
A general electromagnetic plane wave can be written in the form
E = E(k r t) ; H = H(k r t), (6.1)
where k = ks, with s pointing in the direction of propagation. We introduce a new variable u =
k r t, so that
u
x
= k
x
;
u
y
= k
y
;
u
z
= k
z
;
u
t
= . (6.2)
In source-free space Maxwells equations (1.1.1)-(1.1.2) are given by
H =
1
c

D =

c

E, (6.3)
E =
1
c

B =

c

H. (6.4)
By using the chain rule, we nd that the x component of E can be expressed as follows
(E)
x
=
y
E
z

z
E
y
=
E
z
y

E
y
z
=
dE
z
du
u
y

dE
y
du
u
z
= E

z
k
y
E

y
k
z
= (k E

)
x
=

v
(s E

)
x
, (6.5)
where E

=
dE
du
. By proceeding in a similar manner, we nd that
E =

v
s E

, (6.6)
H =

v
s H

, (6.7)
where
E

=
dE
du
; H

=
dH
du
. (6.8)
Further, we have

E =
E
t
=
dE
du
u
t
= E

;

H = H

. (6.9)
FYS 263 22
By substitution of (6.6)-(6.9) into Maxwells equations (6.3)-(6.4) the result is
s H

=

c
(v)E

c
c

=
_

, (6.10)
s E

c
(v)H

=
_

, (6.11)
where we have used (3.9). Thus, we have
E

=
_

s H

; H

=
_

s E

. (6.12)
By integrating over u in (6.12) and setting the integration constant equal to zero, we get
E =
_

s H ; H =
_

s E. (6.13)
Scalar multiplication of the equations in (6.13) with s gives
s E = s H = 0, (6.14)
which shows that both E and H are transverse waves, i.e. both E and H are normal to the propa-
gation direction s, as illustrated in Fig. 6.1. Thus, the vectors s, E, and H represent a right-handed
Cartesian co-ordinate system.
^
s
E
H
Figure 6.1: The vectors E, H, and s for an electromagnetic plane wave represent a right-handed
Cartesian co-ordinate system.
For the electric and the magnetic energy density we nd
w
e
=
1
8
E D =

8
E
2
; E = [E[, (6.15)
w
m
=
1
8
B H =

8
H
2
; H = [H[. (6.16)
Since

H =

E (cf. (6.13)), we get w

e
= w
m
, and the total energy density becomes
w = w
e
+ w
m
= 2w
e
=
1
4
E
2
= 2w
m
=
1
4
H
2
, (6.17)
and the Poynting vector (2.4) becomes
S =
c
4
EH =
c
4
EHs =
c
4
E
_

Es =
_
1
4
E
2
__
c

_
s = wvs. (6.18)
FYS 263 23
Thus, we have
S = wvs, (6.19)
which shows that the Poynting vector represents the energy ow, both with respect to absolute value
and direction. A dimensional analysis of (6.19) shows that
[[S[] = [w][v] =
Energi
m
3

m
s
=
Energi
m
2
s
=
W
m
2
. (6.20)
Thus, S represents the amount of energy per unit time that passes through a unit area of the plane
that is spanned by E and H, as asserted previously in chapter 2.
FYS 263 24
Chapter 7
Harmonic electromagnetic waves of
arbitrary form - Time averages
The E and H elds for a harmonic wave of arbitrary form can be written
E = Re
_
E
0
(r)e
it
_
; H = Re
_
H
0
(r)e
it
_
, (7.1)
where E
0
(r) and H
0
(r) are complex vectors. Thus, we have
E
0
(r) = E
R
0
(r) + iE
I
0
((r), (7.2)
H(r) = H
R
0
(r) + iH
I
0
(r), (7.3)
where E
R
0
, E
I
0
, H
R
0
, and H
I
0
are real vectors.
Since optical frequencies are very high ( 10
15
s
1
), we can only observe averages of w
e
, w
m
,
and S, taken over a time interval T

t T

, where T

is much larger than the period T =

2

.
For the time average of the electric energy density we have [cf. (2.1)]
w
e
) =
1
2T

_
T

8
[E[
2
dt. (7.4)
For any complex number z, we have Rez =
1
2
(z + z

), where z

is the complex conjugate of z.

Therefore, we may write
E = Re[E
0
(r)e
it
] =
1
2
[E
0
e
it
+E

0
e
+it
],
so that we get
[E[
2
= E E =
1
4
[E
0
e
it
+E

0
e
it
] [E
0
e
it
+E

0
e
it
] =
1
4
[E
2
0
e
2it
+2E
0
E

0
+E
2
0
e
2it
]. (7.5)
Further, we have
1
2T

_
T

e
2it
dt =
1
2T

_
e
2it
2i
_
T

=
1
2T

_
1

_
sin(2T

) =
1
4
T
T

sin(2T

). (7.6)
Since T

T, the integral that includes the factor e

2it
can be neglected. Similarly, the integral
that includes the factor e
2it
can be neglected, and we get
w
e
) =

16
E
0
E

0
. (7.7)
FYS 263 25
By proceeding in a similar manner, we nd that the time average of the magnetic energy density
becomes
w
m
) =

16
H
0
H

0
. (7.8)
The time average of the Poynting vector is given by [cf. (2.4)]
S) =
1
2T

_
T

c
4
(EH)dt, (7.9)
where EH can be written
EH =
1
2
[E
0
e
it
+E

0
e
it
]
1
2
[H
0
e
it
+H

0
e
it
]
=
1
4
E
0
H
0
e
2it
+E
0
H

0
+E

0
H
0
+E

0
H

0
e
2it
. (7.10)
By substituting (7.10) into (7.9) and performing time averaging, we nd that the time average of
the Poynting vector becomes
S) =
c
16
E
0
H

0
+E

0
H
0
=
c
8
Re(E
0
H

0
). (7.11)
FYS 263 26
Chapter 8
Harmonic electromagnetic plane
wave Polarisation
For an electromagnetic plane wave that is time harmonic, each Cartesian component of E and H is
of the form
a cos( + ) = Re[ae
i(+)
] ; a > 0, (8.1)
where
= k r t. (8.2)
Let the z axis point in the s direction. Then only the x and y components of E and H are non-zero,
since the electromagnetic eld of a plane wave is transverse.
Now we want to determine that curve which the end point of the electric vector describes during
propagation. This curve consists of points that have co-ordinates (E
x
, E
y
) given by
E
x
= a
1
cos( +
1
) ; a
1
> 0, (8.3)
E
y
= a
2
cos( +
2
) ; a
2
> 0, (8.4)
E
z
= 0. (8.5)
In order to determine that curve which E() describes (Fig. 8.1), we eliminate from (8.3)-(8.4).
We let = +
1
and get
E
x
= a
1
cos , (8.6)
E
y
= a
2
cos( + ) = a
2
[cos cos sin sin ], (8.7)
where =
2

1
. We substitute from (8.6) into (8.7) and get
E
y
a
2
=
E
x
a
1
cos

1
_
E
x
a
1
_
2
sin , (8.8)
which upon squaring gives
_
E
x
a
1
_
2
+
_
E
y
a
2
_
2
2
E
x
a
1
E
y
a
2
cos = sin
2
. (8.9)
This is the equation of a conic section. The cross term implies that it is rotated relative to the
co-ordinate axes (x, y). By letting =

2
, we get
FYS 263 27
E ( )
x
E ( )
y
E( )
x
y
z
Figure 8.1: Instantaneous picture of the electric vector of a plane wave that propagates in the z
direction.
_
E
x
a
1
_
2
+
_
E
y
a
2
_
2
= 1, (8.10)
which shows that the equation describes an ellipse. In a co-ordinate system (, ), which coincides
with the axes of the ellipse, the equations for the eld components become
E

= a cos( +
0
), (8.11)
E

= b sin( +
0
), (8.12)
which upon squaring gives
_
E

a
_
2
+
_
E

b
_
2
= 1. (8.13)
When +
0
= 0, we have E

= a; E

= 0, and when +
0
=

2
, we have E

= 0; E

= b. This
shows that when the upper or lower sign in (8.12) applies, the electric vector rotates against or with
the clock, respectively, if we view the xy plane from the positive z axis. Rotation against the clock
is called left-handed polarisation, and rotation with the clock is called right-handed polarisation.
The relation between the two co-ordinate systems (x, y) and (, ) is shown in Fig. 8.2, where
(cf. Exercise 7)
a
2
+ b
2
= a
2
1
+ b
2
2
, (8.14)
tan 2 = tan(2) cos ; tan =
a
2
a
1
(0

2
), (8.15)
sin 2 = sin(2) sin ; tan =
b
a
. (8.16)
Since sin < 0 when the upper sign in (8.12) applies, we have left-handed polarisation when sin < 0.
We consider now some special cases of (8.6)-(8.7).
Linear polarisation. If the phase dierence is equal to an integer times , i.e. if
= m (m = 1, 1, 2, . . .), (8.17)
then we get from (8.6)-(8.7)
E
x
= a
1
cos , (8.18)
FYS 263 28
2a
1
2a
2

x
y
b
a

Figure 8.2: The end point of the electric vector describes an ellipse that is inscribed in a rectangle
with sides 2a
1
and 2a
2
.
E
y
= a
2
cos( + m) = a
2
(1)
m
E
x
a
1
, (8.19)
which shows that the ellipse degenerates into a straight line, i.e.
E
y
E
x
= (1)
m
a
2
a
1
. (8.20)
Circular polarisation. If the amplitudes are equal and the phase dierence is

2
plus a multiple
of 2, i.e. if
a
1
= a
2
, (8.21)
=

2
+ 2m (m = 0, 1, 2, . . .), (8.22)
then the ellipse in (8.6)-(8.7) degenerates into a circle, i.e.
E
x
= a cos , (8.23)
E
y
= a cos
_
+ 2m

2
_
= a sin . (8.24)
By squaring these two equations, we get
E
2
x
+ E
2
y
= a
2
. (8.25)
We have right-handed circular polarisation when E
y
= a sin and left-handed circular polarisation
when E
y
= +a sin .
FYS 263 29
Chapter 9
Reection and refraction of a plane
wave
k
i
k
r
k
t
n = e
^ ^
z

r
1
,
1
,
2 2
Figure 9.1: Reection and refraction of a plane wave at a plane interface between two dierent
media. Illustration of propagation directions and angles of incidence, reection, and transmission.
We let a plane wave be incident upon a plane interface between two dierent media, as shown in
Fig. 9.1. The incident wave gives rise to a reected wave and a transmitted wave, which we assume
are plane waves as well. Thus, each component of E or H can be written
A
q
j
= Rea
q
j
e
i(k
q
rt)
(j = x, y, z) , (9.0.1)
where A stands for E or H and q = i, r, t, so that k
i
, k
r
, and k
t
are the wave vectors of the incident,
reected, and transmitted waves, respectively.
9.1 Reection law and refraction law (Snells law)
The existence of continuity conditions that E and H must satisfy at the interface between the two
media in Fig. 9.1, implies that when r represents a point at the interface, the argument in the
exponential function in (9.0.1) must be the same for the reected and transmitted waves as for the
incident wave. Thus, we have
FYS 263 30

i
k
k
i
i
___________
| |
^
^
^
b =
n
n
x
x
^ ^ ^
t n b = x
k
i
^
n
Figure 9.2: Reection and refraction of a plane wave. Illustration of the co-ordinate system ( n,

b,

t).
k
i
r t = k
r
r t = k
t
r t, (9.1.1)
or
k
i
r = k
r
r = k
t
r. (9.1.2)
Now we introduce a Cartesian co-ordinate system in which the unit vectors n,

b, and

t represent
a right-handed system (Fig. 9.2). Let n point along the interface normal into the medium of the
refracted wave, and let

b and

t be dened by

b =
k
i
n
[k
i
n[
;

t = n

b. (9.1.3)
In this co-ordinate system we have
k
i
= k
i
t

t + k
i
n
n ; k
i
b
= k
i

b = 0, (9.1.4)
k
r
= k
r
t

t + k
r
n
n + k
r
b

b, (9.1.5)
k
t
= k
t
t

t + k
t
n
n + k
t
b

b, (9.1.6)
r = r
t

t + r
b

b. (9.1.7)
Note that the co-ordinate system is dened such that k
i
has no component along

b, i.e.

b is normal
to the plane of incidence, which is spanned by the vectors k
i
and n.
Since
k
i
r = (k
i
t

t + k
i
n
n) (r
t

t + r
b

b) = k
i
t
r
t
, (9.1.8)
k
r
r = (k
r
t

t + k
r
n
n + k
r
b

b) (r
t

t + r
b

b) = k
r
t
r
t
+ k
r
b
r
b
, (9.1.9)
k
t
r = (k
t
t

t + k
t
n
n + k
t
b

b) (r
t

t + r
b

b) = k
t
t
r
t
+ k
t
b
r
b
, (9.1.10)
it follows from the continuity condition (9.1.2) that
k
i
t
r
t
= k
r
t
r
t
+ k
r
b
r
b
= k
t
t
r
t
+ k
t
b
r
b
. (9.1.11)
FYS 263 31
But since (9.1.11) shall apply to any point at the interface, i.e. to all values of r
t
and r
b
, we must
have
k
r
b
= k
t
b
= 0. (9.1.12)
Thus, both k
r
and k
t
must lie in the plane of incidence spanned by k
i
and n. Therefore, we have
k
i
t
= k
r
t
= k
t
t
= k
t
, (9.1.13)
which implies that the components of k
i
, k
r
, and k
t
parallel to the interface are equal. By using
n k
q
= n (k
t

t + k
q
n
n) =

bk
t
; q = i, r, t, (9.1.14)
we nd that
n k
i
= n k
r
, (9.1.15)
n k
t
= n k
i
. (9.1.16)
Further, we use the relation [a b[ = [a[[b[ sin , where is the angle between the vectors a and b.
Thus, we nd from (9.1.15) and Fig. 9.1 that
k
i
sin
i
= k
r
sin
r
. (9.1.17)
Also, we know that k
i
= k
r
= n
1
k
0
, where n
1
is the refractive index in medium 1, and k
0
is the
wave number in vacuum. The reection law therefore becomes

i
=
r
, (9.1.18)
which in (9.1.15) is given in vectorial form.
From (9.1.16) and Fig. 9.1 we get the refraction law or Snells law
k
i
sin
i
= k
t
sin
t
. (9.1.19)
which by using k
i
= n
1
k
0
and k
t
= n
2
k
0
, becomes
n
1
sin
i
= n
2
sin
t
. (9.1.20)
Equation (9.1.16) represents Snells law in vector form. Note that (9.1.15) and (9.1.16) contain more
information than (9.1.18) and (9.1.20). From the vector equations it is clear that k
r
and k
t
lie in
the plane of incidence.
9.2 Generalisation of the reection law and Snells law
The reection law and Snells law (the refraction law) can be generalised to include non-planar waves
that are incident upon a non-planar interface. This is illustrated in Fig. 9.3, where the eld from a
point source propagates towards a curved interface. Suppose now that the distance from the point
source to the interface is much larger than the wavelength. Then at each point on the interface we
may consider the incident wave to be a plane wave locally, and we may replace the interface locally
by the tangent plane through the point in question. Then we can use Snells law and the reection
law as derived for a plane wave that is incident upon a plane interface, as illustrated in Fig. 9.3.
FYS 263 32
k
k
k

n n
t
r
1 2
^
i
r
t
i
n
local tangent plane
Point source
Figure 9.3: Generalisation of Snells law and the reection law to include non-planar waves that are
incident upon a curved interface.
9.3 Reection and refraction of plane electromagnetic waves
Note that the reection law and the refraction law apply to all types of plane waves, i.e. to acoustic,
electromagnetic, and elastic waves. In the derivation we have only used that k
q
r t (q = i, r, t)
shall be the same for q = i, q = r, and q = t. Now we take a closer look at the reection and
refraction of plane electromagnetic waves in order to determine how much of the energy in the
incident wave that is reected and transmitted.
We know that a plane electromagnetic wave is transverse, i.e. that both E and B = H are
normal to the propagation direction k = ks. In Fig. 9.1 we have chosen the z axis in the direction of
the interface normal. If E is normal to the plane of incidence, we have s polarisation (from German,
Senkrecht) or TE polarisation (transverse electric relative to the plane of incidence or the z
axis). And if E is parallel with the plane of incidence, we have p polarisation or TM polarisation,
since in this case B is normal to the plane of incidence or the z axis; hence the use of the term TM
or transverse magnetic.
A general time-harmonic, plane electromagnetic wave consists of both a TE and a TM compo-
nent. With the time dependence e
it
suppressed, we have for the spatial part of the eld
E = E
TE
+E
TM
; B = B
TE
+B
TM
, (9.3.1)
E
TE
= E
TE
k
t
e
z
k
t
e
ikr
, (9.3.2)
E
TM
= E
TM
k (k
t
e)
z
kk
t
e
ikr
, (9.3.3)
B
TE
=
1
k
0
k E
TE
= E
TE
k (k
t
e
z
)
k
0
k
t
e
ikr
, (9.3.4)
B
TM
=
1
k
0
k E
TM
= E
TM
1
k
0
kk
t
k [k (k
t
e
z
)]e
ikr
. (9.3.5)
But since k [k (k
t
e
z
)] = k[k (k
t
e
z
)] (k
t
e
z
)k k = k
2
k
t
e
z
, we get
B
TM
=
k
k
0
E
TM
k
t
e
z
k
t
e
ikr
. (9.3.6)
FYS 263 33
Note that the vectors
e
TE
=
k
t
e
z
k
t
; e
TM
=
k (k
t
e
z
)
kk
t
, (9.3.7)
are unit vectors in the directions of E
TE
and E
TM
, respectively.
We represent each of the incident, reected, and transmitted elds in the manner given above,
so that (q = i, r, t)
E
q
= E
TEq
+E
TMq
; B
q
= B
TEq
+B
TMq
, (9.3.8)
E
TEq
= E
TEq
k
t
e
z
k
t
e
ik
q
r
, (9.3.9)
E
TMq
= E
TMq
k
q
(k
t
e
z
)
k
q
k
t
e
ik
q
r
, (9.3.10)
B
TEq
=
k
q
k
0
E
TEq
k
q
(k
t
e
z
)
k
q
k
t
e
ik
q
r
, (9.3.11)
B
TMq
=
k
q
k
0
E
TMq
k
t
e
z
k
t
e
ik
q
r
, (9.3.12)
where
k
i
= k
t
+ k
z1
e
z
; k
t
= k
x
e
x
+ k
y
e
y
, (9.3.13)
k
r
= k
t
k
z1
e
z
; k
t
= k
t
+ k
z2
e
z
, (9.3.14)
k
q
=
_
k
1
= n
1
k
0
for q = i, r
k
2
= n
2
k
0
for q = t.
(9.3.15)
The continuity conditions that must be satised at the interface z = 0 are that the tangential
components of E and H =
1

B be continuous, i.e.
e
z

_
E
TEi
+E
TEr
E
TEt
+E
TMi
+E
TMr
E
TMt
_
= 0, (9.3.16)
e
z

_
1

1
_
B
TEi
+B
TEr
_

2
B
TEt
+
1

1
_
B
TMi
+B
TMr
_

2
B
TMt
_
= 0. (9.3.17)
Further, we have
e
z
[k
q
(k
t
e
z
)] = (k
q
e
z
) e
z
k
t
, (9.3.18)
e
z
(k
t
e
z
) = k
t
. (9.3.19)
By substituting from (9.3.9)-(9.3.12) into the boundary conditions (9.3.16)-(9.3.17) and using (9.1.2)
and (9.3.18)-(9.3.19), we get
k
t
_
E
TEi
+ E
TEr
E
TEt
_
+ e
z
k
t
_
k
z1
k
1
E
TMi

k
z1
k
1
E
TMr

k
z2
k
2
E
TMt
_
= 0, (9.3.20)
e
z
k
t
_
1

1
_
k
z1
k
0
E
TEi

k
z1
k
0
E
TEr
_

2
k
z2
k
0
E
TEt
_
+k
t
_
1

1
_
k
1
k
0
E
TMi

k
1
k
0
E
TMr
_

k
2
k
0
E
TMt
_
= 0. (9.3.21)
FYS 263 34
Since k
t
and e
z
k
t
are orthogonal vectors, the expression inside each of the parentheses in
(9.3.20) and (9.3.21) must vanish, i.e.
E
TEi
+ E
TEr
= E
TEt
, (9.3.22)
k
z1

2
_
E
TEi
E
TEr
_
= k
z2

1
E
TEt
, (9.3.23)
k
z1
k
2
_
E
TMi
E
TMr
_
= k
z2
k
1
E
TMt
, (9.3.24)
k
1

2
_
E
TMi
E
TMr
_
= k
2

1
E
TMt
. (9.3.25)
Now we dene reection and transmission coecients as
R
TE
=
E
TEr
E
TEi
; T
TE
=
E
TEt
E
TEi
, (9.3.26)
R
TM
=
E
TMr
E
TMi
; T
TM
=
E
TMt
E
TMi
, (9.3.27)
so that (9.3.22)-(9.3.25) give
1 + R
TE
= T
TE
; 1 R
TE
=
k
z2

1
k
z1

2
T
TE
, (9.3.28)
1 R
TM
=
k
z2
k
1
k
z1
k
2
T
TM
; 1 + R
TM
=
k
2

1
k
1

2
T
TM
. (9.3.29)
The two equations in (9.3.28) have the following solution
R
TE
=

2
k
z1

1
k
z2

2
k
z1
+
1
k
z2
; T
TE
=
2
2
k
z1

2
k
z1
+
1
k
z2
, (9.3.30)
whereas the two equations in (9.3.29) give
R
TM
=
k
2
2

1
k
z1
k
2
1

2
k
z2
k
2
2

1
k
z1
+ k
2
1

2
k
z2
; T
TM
=
2k
1
k
2

2
k
z1
k
2
2

1
k
z1
+ k
2
1

2
k
z2
. (9.3.31)
The interpretation of the reection and transmission coecients follow from (9.3.26)-(9.3.27). Thus,
the reection coecient represents the amplitude ratio between the reected and the incident E
eld, whereas the transmission coecient represents the amplitude ratio between the transmitted
and the incident E eld.
Note that (9.3.22)-(9.3.23) and (9.3.28) contain only TE quantities, whereas equations (9.3.24)-
(9.3.25) and (9.3.29) contain only TM quantities. This implies that these two wave types are
independent or de-coupled upon reection and refraction. Thus, an incident TE plane wave produces
a reected TE plane wave and a transmitted TE plane wave, whereas an incident TM plane wave
produces a reected TM plane wave and a transmitted TM plane wave. Upon reection and
refraction there is no coupling between TE and TM waves.
From Fig. 9.1 it follows that
k
z1
= k
i
e
z
= k
1
cos
i
; k
z2
= k
t
e
z
= k
2
cos
t
, (9.3.32)
so that if
1
=
2
= 1 the reection and transmission coecients become
T
TM
=
2n
1
cos
i
n
2
cos
i
+ n
1
cos
t
; R
TM
=
n
2
cos
i
n
1
cos
t
n
2
cos
i
+ n
1
cos
t
, (9.3.33)
T
TE
=
2n
1
cos
i
n
1
cos
i
+ n
2
cos
t
; R
TE
=
n
1
cos
i
n
2
cos
t
n
1
cos
i
+ n
2
cos
t
, (9.3.34)
FYS 263 35
n
1
n
2
e
^
z
= = =
z = 0

i
=
i r

t
^ TMi
e
e
^ TMr
e
^ TMt
k
i
e
^ TEi
e
^ TEr
e
^ TEt
k
r
k
t
k
k
t
i
^ TE
e
Figure 9.4: Reection and refraction of a plane electromagnetic wave at a plane interface between
two dierent media. Illustration of TE and TM components of the electric eld.
These expressions are called the Fresnel formulas. By using Snells law (9.1.20), we can rewrite them
as (Exercise 9)
T
TM
=
2 sin
t
cos
i
sin(
i
+
t
) cos(
i

t
)
; R
TM
=
tan(
i

t
)
tan(
i
+
t
)
, (9.3.35)
T
TE
=
2 sin
t
cos
i
sin(
i
+
t
)
; R
TE
=
sin(
i

t
)
sin(
i
+
t
)
. (9.3.36)
At normal incidence where
i
=
t
= 0, we get from (9.3.33) and (9.3.34)
T
TE
= T
TM
=
2
n + 1
; R
TM
= R
TE
=
n 1
n + 1
; n =
n
2
n
1
. (9.3.37)
The fact that R
TM
= R
TE
at normal incidence follows from the way in which E
TE
and E
TE
are dened. From Fig. 9.4 we see that these two vectors point in opposite directions at normal
incidence.
9.3.1 Reectance and transmittance
Fig. 9.4 shows the polarisation vectors e
TMq
(q = i, r, t) and e
TE
for TM and TE polarisation.
These unit vectors are parallel with the electric eld and follow from (9.3.9)-(9.3.12)
e
TEi
= e
TEr
= e
TEt
= e
TE
=
k
t
e
z
k
t
; [ e
TE
[ = 1, (9.3.38)
e
TMq
=
k
q
(k
t
e
z
)
k
q
k
t
; [ e
TMq
[ = 1. (9.3.39)
Let the angle between E
q
and the plane of incidence spanned by k
q
and e
TMq
, be
q
[see Fig. 9.5],
so that
E
q
= e
TE
E
q
sin
q
+ e
TMq
E
q
cos
q
. (9.3.40)
FYS 263 36
E
e
q
E
E
e
^ TMq
^ TEq
TEq
k
q
TMq

q
Figure 9.5: Illustration of the angle
q
between the electric vector E
q
and the plane of incidence
spanned by k
q
and e
TMq
.
Further, we let J
i
, J
r
, and J
t
denote the energy ows of respectively the incident, reected, and
transmitted elds per unit area of the interface. Then we have
J
pq
= S
pq
cos
q
; p = TE, TM ; q = i, r, t, (9.3.41)
where S
pq
is the absolute value of the Poynting vector, given by
S
pq
=
c
4
[E
pq
H
pq
[ =
c
4
E
pq
H
pq
=
c
4
_

q
(E
pq
)
2
. (9.3.42)
Here we have used the relation

q
E
pq
=

q
H
pq
. The reectance
p
(p = TE, TM is the ratio
between the reected and incident energy ows. From (9.3.41)-(9.3.42) we have

TM
=
J
TMr
J
TMi
=
[E
TMr
[
2
[E
TMi
[
2
= (R
TM
)
2
. (9.3.43)

TE
=
J
TEr
J
TEi
=
[E
TEr
[
2
[E
TEi
[
2
= (R
TE
)
2
, (9.3.44)
Thus, the reectance
p
is equal to the square of reection coecient R
p
.
The transmittance T
p
(p = TE, TM) is the ratio between the transmitted and incident energy
ows, and (9.3.41)-(9.3.42) give
T
TM
=
J
TMt
J
TMi
=
n
2
n
1

2
cos
t
cos
i
(T
TM
)
2
, (9.3.45)
T
TE
=
J
TEt
J
TEi
=
n
2
n
1

2
cos
t
cos
i
(T
TE
)
2
. (9.3.46)
Thus, the transmittance T
p
is proportional to the square of the transmission coecient T
p
(p =
TE, TM). When
2
=
1
= 1, we nd on substitution from (9.3.35)-(9.3.36) into (9.3.43)-(9.3.46)
the following expressions for the reectance and the transmittance

TM
=
tan
2
(
i

t
)
tan
2
(
i
+
t
)
, (9.3.47)
FYS 263 37

TE
=
sin
2
(
i

t
)
sin
2
(
i
+
t
)
, (9.3.48)
T
TM
=
sin 2
i
sin 2
t
sin
2
(
i
+
t
) cos
2
(
i

t
)
, (9.3.49)
T
TE
=
sin 2
i
sin 2
t
sin
2
(
i
+
t
)
. (9.3.50)
By use of these formulas one can show that

TM
+T
TM
= 1 ;
TE
+T
TE
= 1, (9.3.51)
so that for each of the two polarisations the sum of the reected energy and the transmitted energy
is equal to the incident energy.
From (9.3.41) and (9.3.42) we have
J
pi
=
c
4
_

1
[E
pi
[
2
cos
i
, (9.3.52)
which by the use of (9.3.40) gives
J
TEi
= cos
i
c
4
_

1
[E
TEi
[
2
= cos
i
c
4
_

1
E
2
sin
2

i
, (9.3.53)
J
TMi
= cos
i
c
4
_

1
[E
TMi
[
2
= cos
i
c
4
_

1
E
2
cos
2

i
. (9.3.54)
But since the total incident energy ow is given by
J
i
= cos
i
c
4
_

1
E
2
, (9.3.55)
we nd
J
TEi
= J
i
sin
2

i
; J
TMi
= J
i
cos
2

i
. (9.3.56)
Thus, we have
=
J
r
J
i
=
J
TMr
+ J
TEr
J
i
=
J
TMr
J
TMi
cos
2

i
+
J
TEr
J
TEi
sin
2

i
, (9.3.57)
which gives
=
TM
cos
2

i
+
TE
sin
2

i
, (9.3.58)
and similarly we nd
T = T
TM
cos
2

i
+T
TE
sin
2

i
. (9.3.59)
At normal incidence,
i
=
t
= 0, and the distinction between TE and TM polarisation disappears.
From (9.3.43)-(9.3.46) combined with (9.3.33)-(9.3.34), we nd (when
1
=
2
= 1)
=
TM
=
TE
= (R
TE
)
2
=
_
R
TM
_
2
=
_
n 1
n + 1
_
2
; n =
n
2
n
1
, (9.3.60)
T = T
TM
= T
TE
=
_
T
TE
_
2
=
_
T
TM
_
2
=
4n
(n + 1)
2
; n =
n
2
n
1
. (9.3.61)
When n 1, we see that 0 and T 1, as expected. Similarly, we nd from (9.3.47)-(9.3.50)
that

0,

0, T

1, T

1 when n 1.
FYS 263 38
tB
1
2
k
k
k

i
r
t

iB iB
Figure 9.6: Illustration of Brewsters law.
9.3.2 Brewsters law
From (9.3.47) it follows that
TM
= 0 when
i
+
t
=

2
, since then tan(
i
+
t
) = . We call this
particular angle of incidence
iB
and the corresponding refraction or transmission angle
tB
. By
using Snells law (9.1.20), we nd
n
2
sin
tB
= n
2
sin
_

2

iB
_
= n
2
cos
iB
= n
1
sin
iB
, (9.3.62)
so that
TM
= 0 when
i
=
iB
, where
iB
is given by
tan
iB
=
n
2
n
1
= n. (9.3.63)
The angle
iB
is called the polarisation angle or the Brewster angle. When the angle of incidence
is equal to
iB
, the E vector of the reected light has no component in the plane of incidence (Fig.
9.6). This fact is exploited in sunglasses with polarisation lter. The lter is oriented such that
only light that is polarised vertically (Fig. 9.6) is transmitted. Thus, one avoids to a certain degree
annoying reections from e.g. a water surface.
Note that k
r
k
t
= 0, i.e. k
r
and k
t
are normal to one another when
i
=
iB
, as shown in
Fig. 9.6.
9.3.3 Unpolarised light (natural light)
For natural light, e.g. light from an incandescent lamp, the direction of the E vector varies very
rapidly in an arbitrary or irregular manner, so that no particular direction is given preference. The
average reectance is obtained by averaging over all directions . Since the average value of both
sin
2
and cos
2
is
1
2
, we nd from (9.3.56) that
J
TMi
= J
i
cos
2

i
= J
TEi
= J
i
sin
2

i
=
1
2
J
i
. (9.3.64)
For the reected components we nd
J
TMr
=
J
TMr
J
TMi
J
TMi
=
J
TMr
J
TMi

1
2
J
i
=
1
2

TM
J
i
, (9.3.65)
FYS 263 39
J
TEr
=
J
TEr
J
TEi

1
2
J
i
=
1
2

TE
J
i
, (9.3.66)
which shows that the degree of polarisation for the reected light can be dened as
P
r
=

TM

TE

TM
+
TE

=
[J
TMr
J
TEr
[
J
TMr
+ J
TEr
. (9.3.67)
The average reectance is given by
=
J
r
J
i
=
J
TMr
+ J
TEr
J
i
=
J
TMr
2J
TMi
+
J
TEr
2J
TEi
=
1
2
_

TM
+
TE
_
, (9.3.68)
so that the degree of polarisation becomes
P
r
=
1

1
2
[
TM

TE
[, (9.3.69)
where [
TM

TE
[ is called the polarised part of the reected light.
Similarly, we nd for the transmitted light
T =
1
2
(T
TM
+T
TE
) ; P
t
=
1
T
1
2
[T
TM
T
TE
[. (9.3.70)
9.3.4 Rotation of the plane of polarisation upon reection and refraction
Note that if the incident light is linearly polarised, then also the reected and the transmitted light
will be linearly polarised, since the phases only change by 0 or . This follows from the fact that the
reection and transmission coecients are real quantities [cf. (9.3.33)-(9.3.36)]. But the planes of
polarisation for the reected and the transmitted light are rotated in opposite directions relative to
the polarisation plane of the incident light. The angles
i
,
r
, and
t
that the planes of polarisation
of the incident, reected, and transmitted light form with the plane of incidence, are given by [cf.
Fig. 9.5]
tan
i
=
E
TEi
E
TMi
, (9.3.71)
tan
r
=
E
TEr
E
TMr
=
E
TEr
E
TEi
E
TMr
E
TMi
E
TEi
E
TMi
=
R
TE
R
TM
tan
i
, (9.3.72)
tan
t
=
E
TEt
E
TMt
=
E
TEt
E
TEi
E
TMt
E
TMi
E
TEi
E
TMi
=
T
TE
T
TM
tan
i
. (9.3.73)
By use of the Fresnel formulas (9.3.35)-(9.3.36) we can write
tan
r
=
cos(
i

t
)
cos(
i
+
t
)
tan
i
, (9.3.74)
tan
t
= cos(
i

t
) tan
i
. (9.3.75)
Since 0
i

2
and 0
t

2
, we get
[ tan
r
[ [ tan
i
[, (9.3.76)
[ tan
t
[ [ tan
i
[. (9.3.77)
In (9.3.76) the equality sign applies at normal incidence (
i
=
t
= 0) and at grazing incidence (
i
=

2
), whereas in (9.3.77) the equality sign applies only at normal incidence. These two inequalities
FYS 263 40
imply that upon reection the plane of polarisation is rotated away from the plane of incidence,
whereas upon transmission it is rotated towards the plane of incidence. Note that when
i
=
iB
,
so that
iB
+
tB
=

2
, then tan
r
= . Thus, we have
r
=

2
in accordance with Brewsters law.
9.3.5 Total reection
Snells law (9.1.20) can be written in the form
sin
t
=
sin
i
n
; n =
n
2
n
1
=
_

1
. (9.3.78)
Hence, it follows that if n < 1, then we get sin
t
= 1 when
i
=
ic
, where
sin
ic
= n. (9.3.79)
This implies that when
i
=
ic
, we get
t
=

2
, so that the transmitted light propagates along the
interface. If
i

ic
, we have total reection, i.e. no light will pass into the other medium. All
light is then reected. There exists a eld in the other medium, but there is no energy transport
through the interface. When
i
>
ic
, then sin
t
> 1, which means that
t
is complex. We have
from (9.3.78)
cos
t
=
_
1 sin
2

t
= i

sin
2

i
n
2
1 =
i
_
sin
2

i
n
2
n
. (9.3.80)
The lower sign in (9.3.80) must be discarded. Otherwise the eld in medium 2 would grow expo-
nentially with increasing distance from the interface. The electric eld in medium 2 is
E
pt
= T
p
E
pi
e
pt
e
i(k
t
rt)
(p = TE, TM), (9.3.81)
where
k
t
r = k
x
x + k
y
y + k
z2
z, (9.3.82)
with (cf. Fig. 9.1 and (9.3.80) with upper sign)
k
z2
= k
2
cos
t
= ik
2
1
n
_
sin
2

i
n
2
; n =
n
2
n
1
, (9.3.83)
so that
e
ik
t
r
= e
i(k
x
x+k
y
y)
e
|k
z2
|z
; [k
z2
[ =
k
2
n
_
sin
2

i
n
2
. (9.3.84)
We see that E
pt
represents a wave that propagates along the interface and is exponentially damped
with the distance z into medium 2.
From D
t
=
2
E
t
= 0 it follows that
k
t
E
t
= 0, (9.3.85)
which gives
E
t
z
=
(k
x
E
t
x
+ k
y
E
t
y
)
k
z2
. (9.3.86)
If we let the plane of incidence coincide with the xz plane, we have (cf. Fig. 9.7)
k
x
= k
1
sin
i
; k
y
= 0, (9.3.87)
E
t
y
= E
TEt
e
i(k
x
xt)
e
|k
z2
|z
= T
TE
E
TEi
e
i(k
x
xt)
e
|k
z2
|z
, (9.3.88)
FYS 263 41
i
r
z
x
= =

t
E
TMt
E
TMr

i
k
k
k
t

t
i
E
TMi
k
i E
TEi
E E
TEr TEt
Figure 9.7: Illustration of the refraction of a plane wave into an optically thinner medium, so that

i
<
t
. When
i

ic
, then
t
/2, and we get total reection.
E
t
x
= E
TMt
cos
t
e
i(k
x
xt)
e
|k
z2
|
= T
TM
E
TMi
cos
t
e
i(k
x
xt)
e
|k
z2
|z
, (9.3.89)
E
t
z
=
k
x
k
z2
E
t
x
=
k
x
k
2
T
TM
E
TMi
e
i(k
x
xt)
e
|k
z2
|z
. (9.3.90)
From these expressions for the components of E
t
and corresponding expressions for the components
of H
t
one can show (Exercise 11) that the time average of the z component of the Poynting vector
is zero, which implies that there is no energy transport through the interface, as asserted earlier.
The reection coecients in (9.3.35)-(9.3.36) can be written as follows
R
TM
=
sin
i
cos
i
sin
t
cos
t
sin
i
cos
i
+ sin
t
cos
t
, (9.3.91)
R
TE
=
sin
i
cos
t
sin
t
cos
i
sin
i
cos
t
+ sin
t
cos
i
. (9.3.92)
By combining Snells law (9.3.78) and (9.3.80) with the upper sign with (9.3.91)-(9.3.92), we get
R
TM
=
n
2
cos
i
i
_
sin
2

i
n
2
n
2
cos
i
+ i
_
sin
2

i
n
2
, (9.3.93)
R
TE
=
cos
i
i
_
sin
2

i
n
2
cos
i
+ i
_
sin
2

i
n
2
. (9.3.94)
Since both reection coecients are of the form z/z

, where z is a complex number, it follows that

[R
TM
[ = [R
TE
[ = 1, (9.3.95)
FYS 263 42
which shows that for each polarisation the intensity of the totally reected light is equal to the
intensity of the incident light.
But the phase is altered upon total reection. Letting
R
p
=
E
pr
E
pi
= e
i
p
=
z
p
z
p

= e
2i
p
(p = TE, TM), (9.3.96)
where [cf. (9.3.93)-(9.3.94)]
z
TM
= n
2
cos
i
i
_
sin
2

i
n
2
= [z
TM
[e
i
TM
, (9.3.97)
z
TE
= cos
i
i
_
sin
2

i
n
2
= [z
TE
[e
i
TE
, (9.3.98)
we nd
tan
TM
= tan
_
1
2

TM
_
=
_
sin
2

i
n
2
n
2
cos
i
, (9.3.99)
tan
TE
= tan
_
1
2

TE
_
=
_
sin
2

i
n
2
cos
i
. (9.3.100)
The relative phase dierence
=
TE

TM
, (9.3.101)
is determined by
tan
_
1
2

_
=
tan
_
1
2

TE
_
tan
_
1
2

TM
_
1 + tan
_
1
2

TE
_
tan
_
1
2

TM
_, (9.3.102)
which upon substitution from (9.3.99)-(9.3.100) gives
tan
_
1
2

_
=
cos
i
_
sin
2

i
n
2
sin
2

i
. (9.3.103)
We see that = 0 for
i
=

2
(grazing incidence) and
i
=
ic
(critical angle of incidence). Between
these two values there is an angle of incidence
i
=
im
which gives a maximum phase dierence
=
m
, where
im
is determined by
d
d
i

im
= 0. (9.3.104)
From (9.3.104) we nd
sin
2

im
=
2n
2
1 + n
2
, (9.3.105)
which upon substitution in (9.3.103) gives (Exercise 10)
tan
_
1
2

m
_
=
1 n
2
2n
. (9.3.106)
If the phase dierence is equal to

2
and in addition E
TMi
= E
TEi
, the totally reected light
will be circularly polarised. By choosing the angle
i
between the polarisation plane and the plane
of incidence equal to 45

, we make E
TMi
equal to E
TEi
. In order to obtain =

2
we must have

m
2


4
. This means that tan
_

m
2
_
tan
_

4
_
= 1, which according to (9.3.106) implies that
n
2
+ 2n 1 0. (9.3.107)
By completing the square on the left-hand side of (9.3.107), we nd that
FYS 263 43
n

2 1 ;
1
n
=
n
1
n
2

2 + 1 = 2.41. (9.3.108)
Thus,
n
1
n
2
must exceed 2.41 in order that we shall obtain a phase dierence of

2
in one single
reection.