Chapter 4

Optical materials
and their
properties
4.1. Tensor preliminaries 110
4.1.1 Introduction I l0
4.1.2 Summation notation 110
4.1.3 Tensor transformation rules 110
4.1.4 Tensor description of optical and electrical properties of
materials 111
4.1.5 The influence of crystal symmetry and thermodynamics upon
tensor components llz
4.2 Natural optical birefringence 113
4.2.1 Fresnel's equations 113
4.2.2 Uniaxtal materials 115
4.2.3 Applications of natural birefringence 119
4.3 Controlled birefringence 123
4.3.1 The source and nature of controlled birefringence 123
4.3.2 The electro-optic or Pockels effect 124
4.3.3 The transverse switching modulator L25
4.3.4 The longitudinal switching modulator 129
4.3.5 Bulk phase and amplitude modulators 130
4.3.6 Transit time limitations of bulk modulators 132
4.4 Optical activity 133
4.5 Eigenmodes and eigenvectors of waves propagating in
anisotropic materials 135

109
110 Chapter 4

4.1 TENSOR PRELIMINARIES

4.1.1 lntroduction
Most optical devices such as couplers, modulators, parametric amplifiers, phase-
conjugate mirrors and the like, depend upon anisotropy or non-linearity of transparent
crystalline materials for their operation. The study of these properties is succinctly
described as crystal optics, and any systematic treatment and manipulation of these
properties requires familiarity with the powerful language and techniques of tensor
calculus.l The basic concepts and language of cartesian tensors are presented in
Appendix C, but a brief exposition of the language and the transformation properties
of tensors will be made here to illustrate their applicability in describing birefringence,
both controlled and natural, as well as non-linear coupling in optical crystals.
All that we require for our purposes is an understanding of the summation notation
used in tensor calculus, and the transformation properties under a change of cartesian
coordinates. The reader is asked to accept that the effort required to master the
notation will be more than justified by the ultimate simplicity this will bring to the
understanding of the principles of optoelectronic devices, and the facility it will
provide in choosing optoelectronic materials and their incorporation into optical
devices and systems.

4.1.2 Summation notation

It is often necessary to present a result in the form of a summation. An example, the
relevance of which will soon become apparent, is:

Tit = II
p=l q=l
Lo,ooiro*Trr-
r=1
(4.1)

In the summation convention adopted in tensor calculus, the sigma signs are
omitted and the appearance of a repeated subscript is an indication of summation
from I to n over that suffix. We shall concentrate on cartesian tensors where the
summation is automatically restricted to summation from I to 3, corresponding to
the coordinates of three-dimensional space. Thus (4.1) becomes:

T[* = a;oairamTpq^ (4.2)

with it being implicit that summation over the repeated indices p, q, and ln be from 1 to 3.

4.1.3 Tensor transformation rules

Those readers not familiar with tensor ideas will nonetheless have experience of
manipulating two types of tensor, namely scalars and vectors. These are zero-order
and first-order tensors respectively, and examples relevant to our study ofoptoelectronics
are temperature (zero-order) and electric field (first-order).
The feature of the properties of a tensor that will concern us here is the way
in which it transfglrms under a change of cartesian coordinate system, from
 Optical materials and their properties 11 1

X1,X2;X3 toxl,x'2,x!.. For zero and first-order tensors the ndes for transformation
of the components of these tensors under the prescribed coordinate transformation
are well known and are:

T'=T (zero-order) (4.3)

Ti= a;rT (first-order) (4.4)

where a;, describes the 3x3 direction cosine matrix (Appendix c) with elements
aip = coseipw\".r? 0,0 is the angle between xi and xr.
Tensors of higher order than first-order are defineci as having 3n components which
transform under a change of coordinate sysrem in a manner similar to (4.3) and, (4.4),
namely:

Tii = a;oaioTr, (second order) (4.s)

rio = a;rairap.Tor, (third order) (4.6)

Titon = aipaiqakrar"$qr, (fourth order) (4.7)

Similar extension of the transformation properties of tensors of higher order than

the fourth can be defined, but we will not be concerned with them.

4.1.4 Tensor description of optical and electrical properties of materiats

In explaining the mode of operation and designing optoelectronic devices, we need
to describe how an optical system represented by an optical crystal responds to an
input stimulus. The response might be the electric polarization P, and the input stimulus
might be the input elecric field E. The response of the crystal is described by a constitutive
equation which relates input to output. In the example suggested, the input and
output are both first-order tensors. The quantity that. relates input to output is also a
tensor that describes a material property of the crystal. In this case, the tensor susceptibility
can be shown to qualify as a second-order rensor as defined in (4.5). Generally,
when the coordinates used to describe the optical system are changed, the components
of the stimulus, the response, and the material property relating them, all change
according to the transformarion rules embodied in (4.3) to (4.7), depending upon the
order of the tensors involved.
We now have sufficient command of the tensor language to proceed. Of particular
concern is the influence of the effect of crystal symmetry and thermodynamics upon
those components of the tensor material property that are non-zero and the relationships
between them. This will be demonstrated in the particular optical context already
postulated, in which the input stimulus is E and the response is p or D.
If the material is operating in its linear region (i.e. with E restricted to be much less
than the breakdown field), and the system (crystal) is undisturbed by other stimuli,
such as mechanical stress, or magnetic field, then the relationship between the components
of E and P can be written as:
112 ChaPter 4

(4.8)
,,=#ro=EoxipEp

and eo is
where having components x,, is the second-order susceptibility tensor,
I',
to be D, then the relationship
the permittivity of free sprce. lfltn" response is taken
betweenEandDis:
AD, (4.e)
P1=@= toE;iEi

permittivity
where r,, is the second-order permittivity tensor. whether we use the
;;;;;; #trr" rur..ptibility t"nro. depends upon the application' It is relativelv easy

to show that:
(4.10)
eq = 6;1+ X;1

if i i, and is unity if I = j'

"' 6,, is the Kroneker 6, which is zero
where
of non-zero P for zero
fi #."Jrion, on linearity are relaxed, and the possibility
gives:
E is admitted, then using a Taylor series to describe P

+! ^='^!'r= riEkE,n+ (4.1

aq'i- +++
Lu,.
p. = p;(o)* E ;Er. E 1)
I ae,aro"t"k aEjdEkaEn 3l'

which can be rewritten as:

(4.r2)
P;: P;o+ ;xijE j+ eox;1vEiE1r+ oxijktE jEkEt+
(or Pockels
The two new tensors wiLh components Xiikand X;iy arc the electro-optic
gi;ih of th6ie tensor properties have important
tensor) and the Kerr tensor respectively.
The second-order
applications in the optoelectionic devices to be discussed later.
that appears in (a'8)'
tensor with components xlj is the same susceptibility tensor

4.1.5 The inlluence of crystal symmetry and thermodynamics upon

tensor components
The reader have observed the large number of components of high order [ensors'
will
tensor propelty
Fortunately the actual number of non-zero components of an nlh order
is usually much less than the maximum possible number of 3n.
It is also usual for some
related to each other'
of the non-zero components to be equal to each other or at least
This makes the task of manipulating tensors a lot less formidable than
it might
appear at first glance. There are two reasons for the restricted
number of non-zero
anJthe relationships between them. The first derives from thermodynamics'
"frpon.nr.
and the second from crystal symmetry.
In particular, thermodynamic arguments, outlined in Appendix c, lead to the
in
following relationships between the components of the three tensors appearing
(4.10):.

* In be invoked-see Chapter 8
some circumstancfs stronger thermodynamic constraints can
 Optical materials and their properties 113

*
X;i = X
ir (susceptibility tensor)
X;i2 = X
iit (electro-optic tensor)
xito, = x ii*^ (Kett tensor)
which means that the first two indices can be permuted.
Crystalline materials can be divided into 32 crystal classes according to their
symmetry. For example, lithium iodate belongs to a crystal class characGrized by
sixfold rotational symmetry. This means that if the crystal is rotated through 60;
about its axis of symmetry, the atomic and molecular pattern of the crystal is
indistinguishable from the pattern before the rotation. If the crystalline pattern is
unchanged, then the tensor components in the coordinate systems before and after
the rotation are identical. As shown in Appendix C, this leads to a reduction of the
non-zero components for tensor properties exhibited by lithium iodate and all other
crystalline solids having the same symmetry.
Crystals belonging to other crystal classes exhibit other symmetries which may
be rotational, mirror or inversion symmetry, or some combination of these. For each
of the 32 crystal classes, the non-zero components and the relationships between
them can be determined for any order tensor. These relationships are listed for each
crystal class for first, second and third-order tensors in Appendix D. Fourth-order
lensor relationships are also listed for some crystal classes in Appendix D.
We now turn to applying our knowledge of tensor properties to optical properties
and in particular to the topic of optical birefringence.

4.2 NATURAL OPTICAL BIHEFRINGENCE

It has long been known that certain crystalline materials such as calcite (CaCO3) exhibit
birefringence. Essentially, a birefringent material displays two distinct refractive
indices for propagation in any specific direction within the material. The physical
manifestation of this is the observation of two distinct images of an object viewed
through a birefringent material. Not only are there two different refractive indices
for a birefringent material, but also at least one refractive index, and sometimes
both, change with the chosen direction of propagation within the material.
A birefringent material is optically anisotropic, and the electrical susceptibility
measured for optical fields is not desribable by a single scalar, as it is for isotropic
materials like glass, but requires description by the tensor xii, which appears in (4.8).
All crystals with a non-cubic form are optically birefringent, although some are more
birefringent than others; whereas all crystals having a cubic form and all amorphous
materials such as glass are optically isotropic and display only one refractive index.

4.2.1 Fresnel's equation

To provide a quantitative description of birefringence, we need to solve Maxwell's
equations for a propagating uniform plane electromagnetic wave at optical frequencies,
and to introduce the tensor relationships between the optical lields and fluxes e, h, d
1 14 Chapter 4

and b. This will lead to Fresnel's equation, which is the core statement about birefringence
and allows the dependence of the velocity of the two different waves that propagate
in birefringenr materials to be calculated. It also gives an insight into what fraction
of an input wave incident upon a birefringent material will couple into each of the
two waves, and how this depends upon the orientation of the crystal.
In an optically isotropic non-conducting, non-magnetic material, Maxwell's equations
are:

vxh=#=u,r*, and V"*=# =-p,* (4.13)

where the scalar constitutive equations appropriate to an isotropic material have been
used.
For an anisonopic material, the same constitutive equation applies to b and h, but
the tensor form must. be used to relate e and d, so that

d;= e o;iei
and Maxwell's equations become:

Vxh= r,..*, and Vx (4.r4)

"=-r"*
or in component form:

fu, ana (Vxe), ah,

(vxh), = oliil, =-Fo=i (4'15)

We now combine these equations to obtain the wave equation for an anisotropic
rhedium. Using the vector identity:

vxVxe=V(V.e)-V2e
in conjunction with (4.14) gives the wave equation for a monochromatic wave to be

V2e - V(V.e) + a21tooae = O

or alternatively, in terms of the components:

%-*+a2p,,o;;e;=o
&i&i &i&i 'luutrr
(4.t7)

If thewaveisaplanewave,thenaswellasusing dldt=-.lot,wemayalsouse df Ax;= ik;.

Whence substituting in @.17),

trt e; - k;ky i - az p,e =0 (4.18)

oe 4e i
In this equation the ,Q describe the direction of propagation of the wave within the
medium. We may normalize the equation by dividing by to the magnitude of k in vacuum)
which is (2 11,). Qefining ry as (k/k,) and recognizing that:
 Optical materials and their properties 1ls

njnj=nl+n]+nl

nze;-(n;ni - e;)ei (4.1e)

which, when i takes the values 1,2, or 3, defines three scalar equations. Restating
this, using the Kroneker 6, yields Fresnel's equation:

(n26i-n;ni-e;)ei (4.20)

This can be simplified if we choose to describe the crystal in terms of its principal
j
axes for which e, is zero unless i - (see Appendix C). Expressed in matrix terms,
Fresnel's equation takes its most convenient form:

lnz-nl-e 11 -\n2 -ntn3

nz - nl - e22 (4.2r)
| -nn, -nzn3
| -nrn, -nzn3 nz - nl - q3 illll
The condition that the determinant of the square matrix must be zero gives an equation
for n, which can be interpreted as the refractive index, expressed as a function of
direction, relative to principal axes. The most general form occurs when the symmetry
of the crystal is such that, when expressed in principal coordinates, the diagonal
components of the tensor e are all different. Such a material is described as being
biaxial for reasons that will become apparent. Most materials used in optoelectronic
devices have a crystal symmetry such that the diagonal form for the e tensor takes a
simpler form, with two of the diagonal components being equal (uniaxial material),
or all three diagonal components being equal (isotropic material). We shall concentrate
most attention on uniaxial materials.

4.2.2 Uniaxial materials

For a uniaxial material, where e11 is equal to E2, the determinant equation from (4.21)
yields:

lnz - e1ll4nl + e1(nl + n?)*e11e33]= 0 (4.22)

This equation defines two surfaces. The first is a sphere:

n2=n?+nl+nl
and the second is an ellipsoid:

nl , nl +n]
ll ezr
These surfaces define the refractive indices as seen by propagating waves within
the material. Except for propagation in one specific direction, the refractive index n
116 Chapter 4

t., l*, O*,

Optic axis C)ptic axis Optic axis

x^J x2
x3

x1

Figure 4.1 Refractive index conlours for negative uniaxial material

is multivalued with two distinct values corresponding to two distinct eigenmodes of

the system (crystal) as defined by (a.21). This means that for any given direction,
two distinct waves with differenl velocities can propagate. In other words the material
is birefringent. What this means is best illustrated by examining the intersection of
the sphere and the ellipsoid with planes perpendicular to the principal axes'
This is
shown in Fig 4.1.
Figure 4.1 shows how the refractive index, and hence the velocity, varies with
direction in the planes perpendicular to the principal axes. It can be seen that in the
plane perpendicular to *3 th.r" are two different wave velocities, but the value of the
two velocities is independent of the direction of orientation in the plane' On the
other hand, in the planes perpendicular to the other principal axes one of the waves
has velocity independent of orientation, but the velocity of the other wave depends
upon the angle between the direction of propagation and the x3 axis' The wave that
tras vetocity independent of direction of propagation is called the ordinary wave and
the
wave with a direction dependent velocity is called the extraordinary wave' It will be
seen that for one direction, and one flirection only, the ordinary and extraordinary
waves have the same velocity. This direction is along x3 and this axis is called the
optic axis. The fact that there is only one optical axis is the reason for referring to
materials for which two of the diagonal components of e are equal as uniaxial' lt can
be shown that materials for which all three diagonal components of e are different,
have
two optical axes for which the two propagating modes have the same velocity' This
is the reason for referring to such materials as biaxial.
The two refractive indices along the principal axes perpendicular ta x3 aro no
and n". The contours i1 Figure 4.1 apply to a material where n, < n/ Such a material
is said' pbe negatively uniaxial. The retiactive index contours for posilively uniaxial
materials where n.> no ate shown in Figure 4.2.
 Optical materials and their properties 117

A
t., l*. O*,
Optic axis Optic axis Optic axis

x3 x^J X2

x2
x1

Figure 4.2 Refractive index contours f or positive uniaxial material

The diagrams in Figures 4.1 and 4.2 are drawn relative to the principal axes.
More often than not, the waves traveling in a material or a device are not traveling
along a principal axis, but in some arbitrary direction, as illustrated in Figure 4.3.\t
is a simple exercise in tensor transformations to change the description of a vector
from one cartesian coordinate system to another. If the original coordinate system is
x1,x2,x3, and we choose to rotate the axes through an angle @ about x1, then the components
of a vector transformed from the original coordinates to the new coordinates are
described by:

i = oui

at2
or equivalently azz (4.23)

lll=lii, a3z 1,/.12l=p',',

I cos@ sin @

cos@ (4.24)
h'r=
f ;o 0

Similarly, if we rotate about one of the new axes xl through an angle y, the description
of the vector becomes:

= lx-l = laallx'l (4.2s)

t18 Chapter 4

x2
A

I-----.,
x"z=x'z
3

(a) Original coordinates (b) Rotation about x3 (c) Rotation about x',
through @ through t7

'q sind
lcosd 0l I sin0 0 cos 0l

p"l= l,i, .o,p ,l ,"', = I 0l

lo o rl l- "1,,
0 ,i, Al

Xr = aax xrr = abxt

(d) Matrices and matrix equations describing tranformation from x to x' and from x' [o xtt

Figure 4.3 Refractive indices for uniaxial material with propagation in arbitrary direction

I cos rg 0 sin rgl I sin g 0 cosOl

and
lurl=l o l ol=l o r ol
l-sinrg 0 cosyl l-cos0 0 singl

where 0=tl2-V.
The combination of the two transformations described by laullabl can define propagation
in any arbitrary direction. It can be seen that the choice of @ has no effect on the ordinary
and extraordinary index for uniaxial materials, and such a material is referred to as
being isotropic in the plane perpendicular to x3. On the other hand, the choice of 0
allows the refractive index seen by the extraordinary wave to be varied over the
range from no lo ne. This means that the velocity of the extraordinary wave can be
varied by rotating the crystal about an axis perpendicular to the optic axis. This can
be very useful in matching the velocities of waves that have different temporal
frequencies, as we shall see in Chapter 8.
#
 Optical materials and their properties 119

4.2.3 Applications ol natural birefringence +

Natural birefringence is commonly used as a means of controling the polarization

state of a propagating wave. This is achieved by the use of what are known as half-wave
and quarter-wave plates. These are special examples of a retarder plate, in which the
input optical field couples equally into the ordinary and extraordinary waves in a
uniaxial crystal, and where the ordinary and extraordinary waves that enter the crystal
in the same time phase exit with different time phases because of the different
velocities of the two waves. The operation of a general retarder of arbitrary length is
illustrated in Figure 4.4 for waves propagating in a direction perpendicular to the
optic axis x3. The equal coupling into the ordinary and extraordinary modes is achieved
by rotating the crystal about an axis perpendicular to the optic axis until the linearly
polarized input optical field is oriented at fil4to the optic axis.
The delay in phase of the ordinary wave due to a crystal of thickness L is zmoLf ).r;
remember that L, is the free space wavelength. The delay in phase of the extraoidinary
wave is Zta"Lf ),, so that the difference in time phase between the ordinary and
extraordinary waves emerging from the crystal is the retardance l, where

x2 a

A
k x3 is optic axis

xr
I
Flxr
------+

--) : G-
(a) Propagation direction and crystal axes

Eoa Ecosrol E. Ecosolr'

Ecosror Ecos(cot'-T)
L*z
L_-., Input plane aa' Output plane bb'

(b) Resolution o-f input into (c) Time varying fields at inpur and ourpur planes
and cxtraordinary
fj;xy
Figure 4.4 Operation of a general retarder when a linearly polarized input couples
equally into the ordinary and extraordinary modes
120 Chapter 4

r =2n(no-n")tf 1," (4.26)

It can be seen that if f is nl2, then the difference of path length measured in wavelengths
for ordinary and extraordinary waves is),l4,and such a retarding plate is a quarter-
wave plate. Similarly if L is chosen so that lis n, the retarding plate is a half-wave
plate. The operation of quarter-wave and half-wave plates is illustrated in Figure 4.5.
It is instructive to look at what happens in a uniaxial material when a linearly
polarized wave is incident upon it. This is illustrated in Figure 4.6 for the special
case for which propagation is in a direction perpendicular to the optic axis, and the
crystal is aligned so that the input wave couples equally into the ordinary and exraordinary
modes. The wave within the material starts off as being linearly polarized, but as it
progresses through the material the difference in time phase of the ordinary and
extraordinary and ordinary waves produces elliptical polarization with a major-to-
minor axis ratio, which increases from zero (linear polarization) to unity (circular
polarization) as the retardance (temporal phase difference) l-changes from 0 tonlZ.
The example shown in Figure 4.6 is for a positive uniaxial material. For such a
material the major axis of the ellipse describing the polarization state is oriented
parallel to the optic axis when /-is in the range 0 < f < rt2. As the wave progresses
further through the material, and l-becomes greater than rl2, it again becomes elliptically
polarized, but now with the minor axis of the ellipse oriented parallel to the optic
axis. Eventually, when I-reaches z, the wave becomes linearly polarized again, but
in a direction perpendicular to the direction of polarization of the input wave. For I-
less than z the sense of rotation of the optical electric field that traces out the ellipse
is clockwise. As the wave travels still further within the material it becomes elliptically
polarized again, but with the sense of rotation counter-clockwise. When l-becomes Zfi,
the wave returns to its original input linear polarization. For a negative uniaxial
material the behaviour is very similar, with the difference that l-then becomes negative
as the wave progresses through the material, with a consequent reversal of the sense
of rotation of the elliptical polarization.
From Figure 4.6, the behaviour of quarter-wave and half-wave plates is apparent,
since these correspond to l-equal to rl2 andE respectively. For a quarter-wave plate
a linearly polarized wave of appropriate orientation produces a circularly polarized

Ecos<rY' Ecosorl' uD total Ecosat'

Locus of
+
total field I
or equivalently
I

L---------
Ecos(o:l '-a) -Ecosolt'
Ecos(rot'-ttlZ )

(a) Quarter-wave plate (b) Half-wave plate

Figure 4.5 The output f or quarter-wave and half-wave plates with a linearly polarized
input coupling equally into the ordinary and extraordinary modes
 Optical materials and their properties 121

+
A
-t
a:l-t
6rl at,
o
o_
f,
o
o
tr
I
o_

so
o c
a. o
c
0)
o
() =
(d
o
(U

o E
o C
o o
o o)
c
.C

E
o
-o
o. o)
C)
o
o
=U)
o
o
(5
o
O)
t.E
5- =
o
(g 1a
N
EE ;.s
-=o
Eb oE
F

o=
o:
o.
ir
:(6
'o7
=ao o.S
>D
=N otr (6:
oc
a =H
oE
o6. o
G E6
9o-
bo
d gE
g(d
OH
o* (,>,
!)
d EE
B irP
oZ .N6
p.E (U(l)
=9
Od
os
o-o
o.N E
6

0)
= s5
I
EE
o
d
tro 6r
o. q
\l
0,)
o
Q no 5
d p .9
l!
,J.

i
{,

i
122 Chapter 4

output with a sense of rotation depending upon whether the material is positive or
negative uniaxial. It is easily seen that a three-quarter wave plate behaves in a
similar manner, but with a reversed sense of rotation. For a half-wave plate, the
output is linearly polarized perpendicular to the input.
We have chosen to assume that the input is linearly polarized. If the input is a
clockwise circularly polarized wave, Figure 4.6 still describes the performance of
quarter-wave and half-wave plates, provided the plane aa' is taken to be the input
plane. It can be seen that a quarter-wave plate converts a circularly polarized wave
into a linearly polarized output, whilst a half-wave plate provides a circularly polarized
output with the opposite sense of rotation to the input. For a counter-clockwise
circularly polarized input, Figure 4.6 describes the conditions within the material if
plane bb' is viewed as the input plane.
More generally, if the input wave is linearly polarized but does not couple equally
into the ordinary and extraordinary modes, the output from a half-wave plate is still
linearly polarized, but its polarization is not perpendicular to the input. In particular,
if the input wave is polarized at an angle @ with respect to the optic axis the output
polarization is rotated through an angle 2Qwith respectto the input, as is illustrated
in Figure 4.1 .lt is left to the reader to demonstrate that a quarter-wave plate will not
produce a circularly polarized wave unless @is equal tozr. It is also left to the reader
to demonstrate that if the input to a quarter-wave or half-wave plate is elliptically
polarized the output is also elliptically polarized (see Problems 4.1 and 4.2).
It is clear that quarter-wave and half-wave plates are useful passive components
that can transform an input of known polarization into an output of some other
desired polarization. There is another very useful application of the quarter-wave
plate, where it is employed as an isolator between a source and a load. This is illustrated
in Figure 4.8. The operation as an isolator requires the positionin g of a polarizer between
the source and the quarter-wave plate. The polarizer is a device that will pass the
component of a light wave only in a particular linear polarization. In the isolator
shown in Figure 4.8 the polarizer has been aligned so that its output field is at an

Input Output
------>
k
-->k

lnput field
0utput field

Figure 4.7 Rotation of polarisationby M2 plate when input does not couple equally
into ordinary and extraordinary modes
 I
Optical materials and their properties 129
Polarizer

No reflected
output from
load
Polarizer

+-

Figure 4.8 Application ol quarler-wave plate as isolator

Wave reflected
from load
<+

I
Nore. Orientation of crystal is such thar input couples equally into ordinary and extraordinary modes

angle rl4 relative to the optic axis. This means that the wave that emergos from the
quarter-wave plate is circularly polarized. Any reflected wave from the load passes
back through the quarter-wave plate, so that the reflected wave that fro1n
"rerg"i
the front surface of the wave plate will have effectively traveled through a hilf-wave
plate, and has a linear polarization orthogonal to that accepted by the polarizer.
Consequently, no reflected wave emerges from the front surface of the polaii zer, and
the load is isolated from reflections from the source.

4.3 CONTROLLED BIREFRINGENCE

Natural birefringence provides passive optical devices. We now turn to consider
birefringence which is controlled by some external field or variable. It will be seen
that this provides the basis for a number of active devices and optical transducers.

4.3.1 The source and nature of controlled birefringence

Natural birefringence is a consequence of the low symmetry of crystalline form of
some materials. This is essentially a linear phenomena. It is possible to reduce the
natural symmetry of crystals by subjecting them to an external influence such as
stress or electric field or magnetic field. The consequence of this is that in the
presence of the external influence some materials previously optically isotropic become
birefringent, whilst other materials that were naturally birefringent exhibit a change
in their birefringence. In naturally birefringent materials this can be observed as a
change in the ordinary and extraordinary refractive indices, whilst in naturally isotropic
materials it can be detected by the appearance of extraordinary and ordinary *aues
where only ordinary waves previously existed. The physical basis for this phenomena
124 Chapter 4

is the distortionbf the crystal lattice by the external influence. The manifestation of
the phenomena is most succinctly described by Equation (4.12), which was discussed
earlier in this chapter.
Examination of @.12) reveals that the relationship between P (or D) and the applied
electric field is non-linear. The non-linear terms become significant when high fields
are present. Several additional terms can be added to (4.12) if the influence of other
external influences besides the electric field are included, but examination of the
implications of this are deferred until Chapter 8.
There are two different contexts in which non-linearity with respect to electric
field are exploited. The first of these, which will be developed here, is the case
where the controlling field (external influence) is restricted to low temporal frequencies
well below the optical spectrum. The other context, in which the controlling fields
are at optical frequencies, is deferred until Chapter 8.

4.3.2 The electro-optic or Pockels elfect

The third term in (4.12) describes the quadraric dependence of P upon the applied electric
field E. This can be restated conveniently as a linear dependence of electric susceptibility
upon the components of E, i.e.:

r; = to(xEe1+ xUkE jEk)= e o(xi + xijkEk)E j

l.e. P; = eox;1(E)E (4.27)

1

where xij@)= x;(0)+ xE*E* (4.28)

Similarly, we may express D as

D; = eoe4(E)81 (4.2e)

where e,i (E) = e;1(0) + e

4r,E *
(4.30)

and where it follows, from (4.10) that e;,0= Xiik.

We interpret e;;(E) as relating to th"e retrdctive index that a propagating optical
wave will see in paising through the material in the presence of the external controlling
electric field. To determine the refractive indices for any given propagation direction
we inserl. ea(E) in placc of e; in Fresnel's equation, which is of the lorm of (a.2D
if the material is uniaxial in the presence of thc ext"ernal iicld, and the principal axes
are known.
There may appear to be a contradiction in restricting attention to low-frequency
electric fields, and writing equations in terms of the refractive index, which is a
property defined for fields at optical frequencies. The total field is of course the sum
of the optical field due to the propagating wave and the external control field. There
is no contradiction provided the optical field is small and if only the external field is
permitted to take on large values.
 Optical materials and their properties 125

We have already seen that crystal symmetry and thermodfnamics restrict how
many non-zero tensor coefficients a material may display. It will now be demonstrated,
as an example, that a material that has a crystal structure with a center of symmetry
cannot display any electro-optic effect because, for such materials, ;i1, is zero for ail
ijk.The lal matrix describing a center of symmetry (inversion sy*meiiyy, is:

l-r o ol
e;rld = lo -1 o
lo o -{
I

so that under this symmetry transformation (4.6) gives 4t = ai^ainekomno

i.e: e4t =(l)3;i1r=0 Vijk

Of the 32 crystal classes 1l are centro-symmetric. One further crystal class that is
not cenEo-symmetric has no non-zero electro-optic coefficients for other symmetry
reasons, so that about a third of the crystal classes do not display the electro-optic
effect. The remaining 20 crystal classes do exhibit the electro-optic effect, with
various constraints upon the non-zero coefficients, depending upon the symmetry
transformation characterizing the class.
Most electro-optic devices based on the application of controlled birefringence
belong to one of three crystal classes. Typical examples of materials belonging to
these crystal classes are: LiNbo3 (3m symmetry); KDp (42m symmetry); and GaAs
( 432 symmetry). The non-zero electro-optic coefficients for these materials and a
number of commonly used isomorphs are listed in Appendix D.

4.3.3 The transverse switching modulator

An important application of the electro-optic efTect is the electro-optic modulator,
examples of which are shown in Figures 4.9,4.10 and 4.11. Figure 4.9 shows the
simplest form, in which the electro-optic material is LiNbOr. For this material the crystal
is cut with its edges parallel to the crystal axes, which are also principal axes. In the
absence of any external controlling field the total retardance is set to be an integral
multiple of zr. This is done either by making the crystal thickness L correcr by grinding
and polishing its surface, or by incorporating a compensating plate made from a
naturally birefringent material that adds to the natural retardance of the LiNbO3
crystal to produce the required total retardance. The output polarizer, or analyzer, ii
crossed with respect to the input polarizer, which means that if the light incident on
the analyzer is linearly polarized in the same direction as the input, the output from
the modulator is zero. Thus if the total system retardance is an even multiple of zr, the
modulator is switched o1T, whilst if it is an odd multiple of z, the modulator is switched
on. The system operates as a switch when an external electric field is applied which,
through the operation of the electro-optic effect, changes the retardance by z. If the total
retardance in the absence of the external field is an even multiple of z the switch is initially
OFF. If it is an odd multiple of a rhe swirch is inirially ON.
To determine what external field is required to produce a change of state of the
switch from ON to OFF, the dependence of the ordinary ancl extraordinary refractive
126 Chapter 4

+Vmod
Polarizer Compensating
plate Crossed
polarizer
I (if necessary)

Input
------>
Input to crystal
couples equally
into the ordinary
and extraordinary
modes

*, II
!
ffil

x ,x2,x
t
tl

lt

: are crystal coordinates

output
+

x,(lHx,
Figure 4.9 The transverse electro-optic modulalor using LiNbO3

+vmod
Polarizer
Crossed
polarizer
1

lnput
-----+ output
Input to crystal
couples equally
into the ordinary
il
and extraordinary
modes

Crystal coordinates are x I ,x2,x 3 and coordinates x'l,x'2,x'3 are

those obtained by rotation through n /4 about the optic axis x3.
xt x'z
I

Figure 4.10 The transverse electro-optic modulator using GaAs

indices upon the external field E must be calculated. The non-zero electro-optic coefficients
for LiNbO. are (see Appendix D):
2tt = t1t = tt1 = -0ZZZ) t311 = \221 1g -.223, ca33

In the arrangement in Figure 4.9, the field is parallel to x3, so that

Et= Ez=0, and4*0
 Optical materials and their properties 127

and F
e;(E)=e,;(0)+ eqrLt=e;;(0)+e;zh
i.e. the s matrix in the presence of the external field is:

which, using e,1 =r22, gives

ll ie11+e,,rr'3 0 0
|e|=l 0 err+e1pE3 0
I

rr I o o ey+errrorl
I

It can be seen that this is a diagonal matrix, which indicates that the original crystal
axes remain principal axes in the presence of an external field. Figure 4.9 shows the
direction of propagation of the optical wave to be parallel to x2, so that the refractive
indices for the ordinary and extraordinary and ordinary waves are: no = .{in , n, = ,
^[E3
in the absence of a field; and n'o =
1811 ll3E3 , n; = 143 +
+ qx\, in the presence
of the field.
The products e;pE, are very small, so that the first two rerms of a binomial
expansion can be used to approximate to n'o and nl":

etJlEt
+ @-
Zno, and ni = n" +
nL = no
2n,

For switching to occur it is necessary for the difference in retardance with and
without the switching field be z:

f'*l=
zn((ni -,;)-(n" - ,,))
2tu

i.e. v=n4Llqn-g*l-'
A, lr, no
(4.31)
I

we know that vmod = -E3d, so the switching condition can be expressed in terms of
Vr*o as:

)."d1 e6
ttmod _- -;-l ,r3 l-'
Y

, ln" no )
-
It is common to calculate this switching voltage for the particular case of a cube of
the electro-optic material, and use it as a figure of merit for the material,2 which allows
128 Chapter 4

comparisons to tie made between alternative materials. The switching voltage for a
cube of material is known as the half-wave voltage, where it can be seen that:

v,.tz=^"lT-T)' (4.32)

As a comparison we now examine the use of a different material, namely GaAs,

as shown in Figure 4.10. Gallium arsenide is a cubic material, and is optically
isotropic in the absence of an external field. When an external field is applied
parallel to one of the crystal axis it becomes biaxial, and the crystal axes are no
longer principal axes. To see this we again express e(E) in terms of e;i and e;ip,
using the symmetry properties of GaAs. The non-zero electro-optic coefficients for
GaAs expressed in terms of crystal coordinates are:
tl13 = tl31 = 213 = 231 = t312 = 321
With the field E directed along the x, axis, so that Et = Ez = 0, the e(E) matrix is:

lr,, o ol lo \ztlt ol lt,, rzt4t 0

:
I

r,,"l:13 ,;, rr o I

,1,1.1,,,j, :l=[,,,8, o errl

which is not diagonal, indicating that the crystal axes are no longer principal axes.
Finding the coordinate system in which the le(E)l matrix is diagonal requires finding
the eigenvalues of Fresnel's equation, as described in Appendix E. However, this
particular case has a well known solution, and the coordinate transformation required
is a rotation of the crystal axes about x3 through rl4. The direction cosine matrix that
provides this rotation is:

tlJ, llJ, ol

1a1=llJi -tl.t, ,l
lo o ,l

which, when used in (4.6) gives:

lui, 0 0l lu,,*tztEt 0 0l

l0 o eirl | 0 6 errl

This shows that the material is biaxial in the presence of the external field. In Figure
4.10 the crystal is shown as cut with its edges parallel to the principal axes and
propagation is in the direction parallel to xi . The two refractive indices seen by a
wave traveling in this direction are:

nr = .Fr r zlrd, n2 = e11- egjE3

 Opticat materiats and their properties 129

In the absence of the external field, the retardance -l-providedoby the crystal is zero,
so that no compensating plate or grinding the crystal to size is necessary. The condition
for switching is that the retardance l'in the presence of the switching field must be
zr, which using a binomial expansion for n2and recognising that nl is equal to
z (the
refractive index in the absence of the field) leads to:

nelyE3L
la-
--L
flLu

and the half-wave voltage is:

rl nL, (4.33)
v)rtz=7
- Ll23

4.3.4 The longitudinal switching modulator

A lower half-wave voltage can be obtained, using any chosen material, if an arrangement
such as that illustratedin Figure 4.11 is used, where the controlling field is
parallel
I
to the direction of propagation of the optical wave. This can only be done at the
expense of a more complex technology, since it is necessary to use transparent
I

electrodes. This is commonly done by depositing a film of conducting tin oxide on

the front and back surfaces of the crystal. This configuration, shown for a KDP
crystal in Figure 4. 1 I , is described as a longitudinal modulator '
The material KDP is uniaxial in the absence of an external field, with the crystal
axes being principal axes. In the presence of an external field parallel to the
original
no longer principal
optic axis x3, the material becomes biaxial and the crystal axes are
axes. The non-zero components of e4lfor KDP are:

Polarizer Tiansparent Compensating Crossed

elecffodes Plate Polarizer

I ilj*
(if necessary)

Input

->
Input to crystal couPles
equally into ordinary and
extraordinary modes

tV-od

xt
3
Crystal coordinates are x t,x2,x 3 and coordinates x'r,x'r,x', are
those obtained by rotation through n i4 about the optic axis x r'
xt *'2
1

Figure 4.11 The longitudinal electro-optic modulator using KDP

130 Chapter 4

\yj = b21. i e23l = 321 = Et32 = 3tz

Itcan be shown that the principal axes in the presence of the switching field are
obtained by rotating the crystal axes through nl4 abot x3, and that le'(E)l in the new
coordinate system xl,xi,xi is:

lr,, + e1yE3 0
lr'(r)l=l 0 trr - snzEt
00 fl
so that the values of n1 and rr2 for propagation along xl are:

fho= ! ep3E3

and consequently the half-wave voltage is:

t/ il" v
'Ll2-- LLl23

which is half the value for the same material used in transverse mode.

4.3.5 Bulk phase and amplitude modulators

The switching modulator is a pulse modulator. The phenomena of electro-optic control
of phase delay exploited in the switching modulator can also be used to produce a
phase modulator or a quasi-linear amplitude modulator.
The phase modulator is provided by a very simple arrangement. If an input
optical wave propagates along one of the principal axes of a uniaxial material and
couples into only one of the modes of the crystal (i.e. the ordinary or extraordinary
wave) then applying a modulating field to control the refractive index seen by the
wave will directly control the phase of the emerging output wave. Thus for LiNbO3
the modified form of Figure 4.9, shown in Figure 4.12, will provide phase modulation.
If the input modulating field is E^t$) and the optical input couples only into the
extraordinary wave, the output optical wave is :

E o,a3u) = 4, *r(( ,P\( n"L

'[( Au A, - c -
%33!E
2cn" ll)) ,.
m3(t)
. (4.34)

so that the modulation of phase is

a(0=w (4.3s)

lf frequency modulation rather than phase modulation is required, this is achieved by

including an integrator between the modulation source and the modulator plates.
To achieve amplitude modulation a little more subtlety is needed. The arrangement
 Optical materials and their properties 131

+Vmod.=E mt'-(tld i
'

Polarizer i
I AmPlitude

rnpur [l!\l I'"-l *Txl""o t

_> I I ------+ d

rnpur ro
on'ly couplLs
crysrar I \l
into l-l
II I
, I t
I

exraordinary
the I
mode I
a
0

*, II xt,x2,x3 are crystal coordinates

I

*,(H*,I

Figure 4.12 The electro-optic phase modulator using LiNbO3

is shown in Figure 4.13. The LiNbO3 crystal is of a thickness to produce a retardance

of 2rn in the absence of any modulating field, and a modulating voltage much less
than the half-wave voltage is applied to the modulator plates. The quarter-wave plate
is included to provide the offset of total retardance that is required in order to get a
quasi-linear amplitude response from the modulator.
The optical field at the input passes through a polarizer set to couple equally into
the ordinary and extraordinary waves in LiNbO3. The fields of the ordinary and
extraordinary waves are in temporal phase at the input. Upon emerging from the
crystal there is a temporal phase difference 0(r) between them where:

*=u#(t t)
After passing through a quarter-wave plate the temporal phase difference increases
to 0(0 + ttl2. The analyzer accepts the projection of the ordinary and extraordinary
fields, and the output field aligned in the direction of the analyzer is:

Eo,r =f (."r{or"* eg)+ x /2)+ cos(ar"))

which is

to* = "lit(cos(ar"
+0{t)I2+ rI4)cos(e@Iz+ fl+))

so that the intensity is

t = cos(zat'+0(t)+ rlz))(r- sina(r)))

+ff+
132 Chapter 4

V*a = E-s (t)d

)"14

#
Polarizer plate Analyzer

A
rnpur
------> [l +; I

I\l
Input to crystal
couples equally
the ordinary and
extraordinary modes
into f!
I \l
t_-l
I0
! D
Amplitude
modulated Y
output
I

+-L-+
x^J
x1,x2,x1 are crystal coordinates

*1 xz

E cos tot

E cos at, E cos at"

Field at plane A Fieid atplane B Field at plane C Field ar plane D

Figure 4.13 The electro-optic quasi-linear amplitude modulator using LiNbo3

and the envelope of the optical intensity is

E2
1o"Q)= sin0(r))
TA-
When 0(r) is much less than unity, sin A(l) 0(r), and
=

u' qll-- 9*l'l

"dO (
TTLV!
i o,U) = l, - Lrd \n,
(4.36)
4( no).)
so that the output optical intensity is amplitude modulated.

4.3.6 Transit time limitations of bulk modulators

There are two factors that limit the bandwidth of bulk modulators. One is the inherent
charging time required to charge the modulator plates, which depends upon the
modulator capacitance, and the internal resistance of the electronic source driving
the modulator. The other is the transit time taken by the optical wave to pass through
the modulator.
For the bulk amplitude modulator with a sinusoidal modulation source, there will
be no modulation of the output if the period ol the modulating source is equal to the
 Optical materials and their properties 133

transit time. The frequency response of the modulator can be calculated by integrating
t}re instantaneous retardance to obtain the total retardance produced by the modulation
source as the optical wave passes through the modulator. Thus for a modulation
input 7- exp(j2tf the effective retardance for a transverse modulator is:
^t),
r @ = r" li exp(i z4
-,, ^r)ar
where /, is the transit time clnL and f, is the retardance when the modulating
frequency is zero, which for the LiNbO3 transverse modulator is:

e,,,, )
T)
Thus r(r)= ,",",(*)*r(irt^(, *)) (4.37)

This means that the modulation of the optical intensity envelope is delayed by half
the transit time, and the amplitude of the modulation of the envelope is reduced as f^
is increased, which imposes a limitation on the modulation frequencies that can be
used, both as an amplitude modulator and also as a switching modulator.
Some reduction of the transit time effect can be obtained for transverse modulators
by launching a modulating electrical wave along the electrode structure in synchronism
or close synchronism with the optical wave. This technique of traveling wave madulation
is also used with success with optical waveguide modulators. Further discussion of
traveling wave modulators is deferred until Chapter 9 when optical waveguide modulators
are discussed.

4.4 OPTICAL ACTIVITY

In developing an expression for the influence of the electric lield upon the polarization
P in (4.11) we overlooked one additional term which is present for materials belonging
to i0 of the 32 crystal classes. This additional term does not depend upon E or on powers
of E but depends upon the gradient of E. It can be observed as an effect present only
when an electromagnetic wave is passing through the material. The gradient of the
field becomes large only when the wavelength of the electromagnetic wave is small,
because that gradient is directly dependent upon the propagation vector k, which has
a magnitude 2r/),. Because of this wavelength dependence the phenomena is detectable
only at. wavelengths of the order of 1 pm or less. The phenomena is important in some
optoelectronic applications and is termed optical activiry.3'4
The modified form of (4.11), which includes optical activity and neglects the
electro-optic and Kerr effects is:

4 =4(o) .#u,.fru,r (4.38)

where E'is a second-order tensor with components

134 Chapter 4

' aE,
r
E'i2=Y;Ei - axk

By using tensor notation (4.38) can be written as

Pi= Pio+eoX;1E1+ X.";fiU'it @.39)

where X,ffi are the cpmponents of the third-order optical activity tensor. The total
optical frequency polarization may be written as

Pltyi = e oX i" j + eoTfifte'i* = P ; + Pfo

The tensor product eoX?;ft,e';* describes rhe polarization due to optical activity,
which is oscillating at opticdl fiequencies. It includes information about the relative
phase of optical electrical field e and the optical activity polarization paa. This phase
information is conveyed by treating the fields and polarizations as phasors, as was
done when considering Jones vectors in Chapter 1. Landau and Lifshitz have shown
that the components of Xffi are imaginary (implying a nl2 phase difference between
e and paa), and are anti-simmetrical in j,t , i.e. Xfrft = -Xifir.
For a uniform plane wave represented by
e(r : r) = (t7Z)r exp(-;(ar - k.r)) + cc

wehave 3:0,,
dxt ' +=
dxz - !=ke,
k,,
dxl -
ie 3=0,.
d*j r
Thus the components of the total polarization p, can be expressed in terms of an effective
susceptibility XU, with componenrs:

X1t1t^= X6+ jT,ffik"

Consequently, using (4.10) and the anti-symmetry of Xoa, the total permittivity
tensor ft1; that includes the effecr of optical activity has components:

I r,, en+ jYltnkn ep + jLf{,k,1

.:lrr, - iXft^k, ezz ey + jXi!,k^l (4.40)

ler, -
jt,ff"kn - il(.rt"k" %z
32
I

This simplifies, if the zxeS x1,x2,x3 are principal axes, to:

I r,, jLit,k, jYff,k^l

.=l-ixiL^*, ezz jxrg,k^l
l-ixig^t^ -irftnkn sr: I

A specific example of a crystal class that is optically acrive is the cubic class 28 for
which it can be shown (see Problem 4.4) that the non-zero components of the optical
activity tensor are:
{
 Optical materials and their properties 135

)(ffi = Lffir= )Gfz= 4fftz= -I\tz= -26h= o'

where a is a constant. The permittivity tensor for such a material is very simple, and
is:

I r,, jakt jakz

. =l-ior,, rr iakr
l-i,k - ja\ tn
If the wave is directed along the x3 axis, then:

I t,, jakt 0
trr
|

.=l-io*, o (4.41)
I o o errl
I

allows the polarization state of optical electromagnetic waves traveling through an

optically active material to be determined. To obtain this interpretation we need to
examine what are referred to as the eigenmodes associated with wave propagation.

4.5 EIGENMODES AND EIGENVECTORS OF

WAVES PROPAGATING IN ANISOTROPIC
MATERIALS
We have already used the idea of resolving waves propagating through materials or
devices into orthogonal components that transmit through the material or device
without change in the nature or direction of polarization. Such a description constitutes
analysing the system into its characteristic waves or eigenmodes. We now determine
the eigenmodes as expressed in terms of the permittivity tensor and apply them to
anisotropic materials in general, and to optically active materials in particular.
Fresnel's equation given by $.19):

nzr;-(n;ni-eq)e,=o (4.1e)

can be rearranged in matrix form for an arbitrary orthonormal set of axes *i,xi,4
AS:

In?-ri,
ln1n2 - e;1
- ti1
n1n2

"3
nznt
-
-
eLz
ei2

- eiz \?;'Alil
.lil (4.42)

lnrry
This is the so-called eigenvalue equation. If we choose the direction of propagation
to be along xi , one of the arbitrary orthonormal axes, then the values of n; associated
with the other axes become zero, and (4.42) is simplified to be:
136 Chapter 4

lei'- 12

| '5, (4.43)

| 'i, "i,i*:l[il H
Note that for a [ransverse plane wave, the value of ei is zero, so that the equations
for the eigenvalues become the roots of:

(4.44)
l'",i: ,ri':,,\?rl=lt
which, taking the determinant of the matrix to be zero, gives the eigenvalues to be:

(ei, + e'rr)t
n2=

If the axes are changed to the principal axes xX,z the matrix becomes diagonal, and

(4.4s)
lt';" ,rri,'lEl=l3l
so that the eigenvalues are 11 and e22. This is no surprise since we already know
that plane waves propagating along one principal axis with their linear field polarization
aligned along another principal axis are propagated without change in the polarization
srare wirh velocity ,lrfr" o, cf
$n, depending upon which of the principal axes
the field is aligned with.
when the eigenvalues are known they can be inserted in (a.a$ or (4.45) to
obtain the two eigenvectors. For the principal axis system of (a.a5), the eigenvectors
are:

(4.46)
l;;l= ll
- l3l

where a and b are constants. It is common to normalize these eigenvectors to have

unit amplitude, so that (4.46) becomes:

(4.41)
l;l=l?l,,lll
These will be recognized as the normalized Jones vectors used in Chapter 1 to
describe the polarization states of orthogonal linearly polarized waves. For the arbitrary
coordinate system xi,x'2,x|, with propagation along xi, the normalized eigenvectors
describing the polarization states that are propagated without change are more complicated,
and will provide orthogonal Jones vectors that are elliptically polarized. To obtain
the eigenvectors requiles the solution to (4.44).
 Optical materials and their properties 131

We now return to discuss optical activity, and the pafticular case of a cubic
material belonging to crystal class 28, where the permittivity tensor is given by (a.a1),
for propagation along the x1 axis. The eigenvalue equation for this material is thus:

l'
: ;;',,::!,,11::l=l?,1
so that the eigenvalues are n2 = e11* ak3, and the normalized eigenvectors or Jones
vectors are:

lll=#li,l* ^E-t-l:
These will be recognised as the Jones vectors for two orthogonal circularly polarized
waves. Thus circularly polarized waves propagating in this optically active material
have their polarization preserved as they transit through the material and are eigenwaves
of the system. The refractive indices that determine the velocity with which the two
waves travel within the material are

n!= - akz and nl, = + ak3 (4.48)

from which it can be seen that the two waves travel with different velocities. If a is positive
then the clockwise wave is faster than the counter-clockwise wave.
We have characterized some crystalline materials as being linearly birefringent.
This is describable in terms of different refractive indices (and velocities) associated
with ordinary and extraordinary linearly polarized waves. A material that is optically
active is referred to as being circularly birefringent, with the birefringence being described
by the difference in refractive indices for circularly polarized waves of opposite
rotation.
It will not have escaped the reader's notice that materials belonging to some
crystal classes display both linear birefringence and circular birefringence. Strictly
speaking such materials display elliptic birefringence, but in practice the difference
between no and n, is much larger than the difference between n, arrd n"". What this means
is that for linearly birefringent materials, the effect of optical activiry can be observed
only when the wave travels along or close to the optic axis of the material where the
effect of linear birefringence is small. We shall concentrate on propagation in materials
that are not linearly birefringent but are optically active, or to propagation in a
direction close to the optic axis of linearly birefringent materials. From Appendix D,
the only materials exhibiting circular birefringence and no linear birefringence are
cubic materials having 23 or 432 symmetry.
The difference in velocity of the two circularly polarized waves in optically
active materials means that waves of opposite rotation that are in temporal phase at
the input plane of a slab of optically active material will have different temporal
phases at the output of the slab. We know that a linearly polarized input wave can be
resolved into two equal amplitude circularly polarized waves. The effect of circular
birefringence is that the sum of the two circularly polarized waves at the output add
together to produce a linearly polarized output, which has been rotated in space.
Thus a slab of optically active material has the function of a rotator of a linearly polarized
138 Chapter 4

wave input. This is illustrated in Figure 4.14 for a material in which the clockwise
wave is the slow wave.
To avoid confusion between temporal phase and spatial phase the time of observation
of the output from the slab is chosen to maintain the same temporal phases at input
and output. The time of ransit of the fast wave through the slab of thickness s is t,1 = sn y f c
and the time of transit of the slow wave is t,s = sn, f c. Thus the emerging waves have
a difference of spatial phase of 6, where:

Two oppositely
rotahng vectors

Linearlv
oolariz-6d
bscillating
vector Linearlv
oolariz6d
bscillating
'. Two oppositely vectol
rotatmg vectors t=t'

Clockwise wave is slow;

counter-clockwise wave is fast
i.e. n"r< n,

it=t'+tz
Input plane Output plane

Note:To simplify the diagram, the time of observation at the outpui plane has been chosen to give the
same temporal phase at the outpur and at the input.

Figure 4.1 4 Clockwise rotation of linear polarization vector at output ol slab of

optically {active material with nc > ncc
 Optical materials and their properties 139
6

. ('" -
"r)t (4.49)
2Lu

which for the material in Figure 4.14 is

* ('" - nrr)t
2hu

It can be seen t}tat output is a linearly polarized wave that has been rotated an
angle d in the same sense as the polarization of the slow wave. Thus for the material
in Figure 4.14 the linear polarization direction has been rotated through an angle of 6
in the clockwise direction. It is clear that a material for which the counter-clockwise
wave is the slow wave produces counter-clockwise rotation of a linearly polarized
input.
It can be seen that a slab of optically active material provides a linear rotator with
its sense of rotation determined by the sign of a in (4.41), which is a property of the
material chosen. Rotation through an angle 6 can be ascribed a Jones matrix M. To
determine M for a counter-clockwise rotator we use the simple transformation rules
for the rotation of coordinates. If the input linear polarization is aligned at an angle 0
with the x axis, the input and output Jones vectors are:

cos(o + 6)l

li:A= :l,i;l-,1';;:l= sin(a + 5)l

and
Hl,:3ll=H T:)li?:e

whence M=llcosd -sindl I

lsind cosd I

as listed in Table 1.2.

In the next chapter we turn from the plane wave propagation of optical waves to
the propagation of optical waves in guiding structures.

Problems
Note: The values of electro-optic tensor components are listed in Appendix D.
4.1 If aclockwiseellipticallypolarizedwaveof ellipticity e ispassedthroughaquarter-
wave plate, what is the polarization state of the output?

4.2 A KDP crystal in the form of a cube is connected in the longitudinal modulator
mode shown in Figure 4.11.
(a) Determine the half-wave voltage VLp if the optical beam being modulated
has a free space wavelength of 633 nm.
140 ChaPter 4

(b) the crossed polarizer is removed, what is the polarization state of the
If
output when the modulation voltage is reduced to VlP f 2 '
(c) Show that if the modulating voltage is reduced to vlp f 4,the output is elliptically
polarized, and calculate the ellipticity e'

Answers
(a) 9.37 kV.
(b) The output is circularly polarized.
(c) The ellipticity is e = tan(nlS) = 0.+t+.

4.3 A phase modulator of the form shown in Figure 4.12 uses a LiNbo3 crystalrrith
voltage
a llngth of 10 cm and a cross-section of I mm x 1 mm. The modulating
optical beam being modulated has a free space
has a peak value of 10 V, and the
wavelength of 633 nm.
(a) Determine the peak phase modulation achieved at low modulating frequencies'
fall to zero?
iUi nt what moduiating frequency will the peak phase modulation
Answers
(a) 1.65 radian (b) 1.36 GHz

4.4 Show that the non-zero components of the optical activity tensor for BSO, which
is a class 28 crYstal, are:

X(tz = Xffir = Ltiz = -xi?z = -X\fz = -X?ft = o

References
A. R. Billings,Tensor Properties of Materials (London: wiley Interscience,
1969).
1.
2. A. Yariv, optical Electronics,3rd edn (New York: HoIt Rinehart and winston, 1985)'
continuous media (oxford: Pergamon
3. L. D. Landaur and E. M. Lifshitz, Electrodynamics of
Press, 1960), P.337.
4. A. Nussbaum and R. A. Phillips, contemporary optics for scientists and Engirzers (Englewood
Cliffs N.J.: Prentice Hall, 1976), p.382'