3

Optical Waves D It] v Polarization of Light Waves

Bit,

in Crystals 1
.

The field quantities E and H that describe light waves are vectors. In the
Propagation and Control of previous chapter, when we discussed Gaussian-beam propagation, we used
the scalar-wave approximation and were not concerned with the direction of
Laser Radiation oscillation of the electric field vector, except to note that the electric vector
lies in a plane perpendicular to the direction of propagation. Many cases
involving the propagation of light waves depend crucially on the direction
of oscilhtfion of the electric field. In fact, throughout most of the book, we
deal almost exclusively with the propagation and control of polarized light.
In this chapter we describe many aspects of polarized light and discuss
AMNON YARIV
many of the techniques used to study its propagation.
) Ciilifornia Instiiute of Technology
i
gj) THE CONCEPT OF POLARIZATION
POCHl YEH
Rockwell International Science Center Light waves are electromagnetic fields and require the four basic field
( ' vectors E, H D, and B for their complete description. The electric field
,

vector E is chosen to define the state of polarization of the light waves. This
choice is convenient because, in most optical media, physical interactions
with the wave involve the electric field. The main reason for studying the
polarization of light waves is that in many substances (anisotropic media)
the index of refraction depends on the direction of oscillation of the electric
Claadio Kitano field vector E. This phenomenon can be explained in terms of the motion of
D«pto. Eng." Eletrica electrons which are driven by the electric field of the light waves. To
FEIS - UNESP . illustrate this point assume that the anisotropic material consists of non-
,

spherical, needle-like molecules, and suppose that these molecules are all
aligned with their long axes parallel to one another Consider an electromag-
.

netic wave passing through this substance Because of the structure of the
.

molecule the electrons in the substance are pushed further from their
,

equilibrium positions by electric fields that are parallel to the axes of the
molecules than by those at right angles to the molecular axes We thus.

A Wiley-lnterscience Publication expect a larger induced electronic polarization in the first case than in the
second.
John Wiley & Sons
54
} New York / Chichester / Brisbane / Toronto / Singapore

i
 '
DO POLARIZATION OF LIGHT WAVES
mLAKl AliUN Ut- MUNULHKUMA nC PLANE WAVES 55
r
where A and /!„ are positive numbers, X and 9 are unit vectors. The curve
There are many other physical phenomena which are only associated v

r described by the end point of the electric vector as time evolves can be
with polarized light waves. Before we study these optical phenomena it is ,

obtained by eliminating ut - kz between the equations (3.2-2). After several

important to understand in detail the characteristics of polarized waves. We
r steps of elementary algebra, we obtain
start by reviewing the polarization states of monochromatic plane waves.
r 2

POLARIZATION OF MONOCHROMATIC PLANE WAVES r H -2S sin 25 ' (3-2-4)

The polarization of light waves is specified by the electric field vector E(r r) ,

at a fixed point in space, r, at time /. The time variation of the electric field where

vector E of a monochromatic wave is exactly sinusoidal, that is, the electric

field must oscillate at a definite frequency. If we assume that the light is (3.2-5)
propagating in the z direction, the electric field vector will lie on the xy
plane, Since the x component and the y component of the*field vector can
oscillate independently at a definite frequency, one must first consider the All the phase angles are defined in the range -it < S < tt.
cfleet produced by the vector addition of these two oscillating orthogonal Equation (3.2-4) is the equation of a conic. From (3.2-2) it is obvious that
this conic is confined in a rectangular region with sides parallel to the
components. The problem of superposing two independent oscillations at
coordinate axes and whose lengths are 2A and 2A Therefore, the curve
right angles to each other and with the same frequency is well known and is
.
X y

must be an ellipse. The wave (3.2-1) is then said to be elliptically polarized.

completely analogous to the classical motion of a two-dimensional harmonic
A complete description of an elliptical polarization includes the orientation
oscillator. The general motion of the oscillator is an ellipse, which corre-
of the ellipse with respect to the coordinate axes, the shape, and the sense of
sponds to oscillations in which the x and y components are not in phase.
revolution of E. In general, the principal axes of the ellipse are not in the x
There are, of course, many special cases which are of great importance in
and y directions. By using a transformation (rotation) of the coordinate
optics. We start with a discussion of the general properties of elliptic '
system, we are able to diagonalize Eq, (3.2-4). Let x and y' be a new set of
polarization and follow with a consideration of some special cases.
axes along the principal axes of the ellipse. Then the equation of the ellipse
In the complex-function representation the electric field vector of a
,
in this new coordinate system becomes
monochromatic plane wave propagating in the z direction is given by
{
2
E(z 0 = Re[Aei("'~*')],
> (3.2-!)
(*)" (*) 1, (3.2-6)

where A is a complex vector which lies in the xy plane We now consider the .

nature of the curve which the end point of the electric field vector E where a and b are the principal axes of the ellipse, and E and E are the , ,
x y
describes at a typical point in space. This curve is the time-evolution locus components of the electric field vector in this principal coordinate system.
of the points whose coordinates (Ex, Ey) are Let (O < $ < tt) be the angle between the direction of the major axis x'
and the x axis (see Fig. 3.1) Then the lengths of the principal axes are given
E = A cos(u>t - kz + S )
x x x ,
by
(3.2-2)
'

E
y
= Ayco&(uit - kz + 8 ) ,
y
a1 = Alcos2 + Alsin2 + 2AxAycos 8 cos (/>sin $
where we have defined the complex vector A as (3.2-7)
b2 = A2 sin2$
x + A2cos2<fr - 2AxA>co?. 6 cos <|>sin<|».
,s
A = xA e
x > + $A ei!s
y
>
, (3.2-3) L
 57

'
r
y

r
6=0
5=-ii/2

5= it/2 6= 3nM 6=w

(a}

Figure 3.1. A typical poJarizalion ellipse.

The angle £ can be expressed in terms of A ii Ay, and cos 8 as

r
2 ,
tan 2 = - - cos 6 .
(3.2-8)
y
5 = -Tr/4 5=0
5 -
-
37r/4 5 = -rr/2

The sense of revolution of an elliptical polarization is determined by the

sign of sin S. The end point of the electric vector will revolve in a clockwise
direction if sin fi > 0 and in a counterclockwise direction if sin fi < 0. Figure i
3 2 illustrates how the polarization ellipse changes with varying phase
.

difference 6.
Before discussing some special cases of polarization, it is important to
familiarize ourselves with the terminology. Light is linearly polarized when
ihe tip of the electric field vector E moves along a straight line. When it
5=
describes an ellipse, the light is elliptically polarized. When it describes a S-ir/4 6=n/2

circle, the light is circularly polarized. If the end point of the electric field fW

vector is seen to move in a counterclockwise direction by an observer facing Ex = cos(nf kz),

Figure 3.2. Polarization ellipses at various phase angles 5, where: (a)
the approaching wave, the field is said to possess right-handed polarization. Ev - cos(Wf -kz + S); (b) E = \ cos(Uf - kz) , Ey = cos(wr - kz + S).
L ,

Figure 3.2 also illustrates the sense of revolution of the ellipse. Our
convention for labeling right-hand and left-hand polarization is consistent
with the terminology of modem physics in which a photon with a right-hand
circular polarization has a positive angular momentum along the direction

58

A
 rULAKl A HON OF LIGHT WAVES

r
of propagation (see Table 3.1 and Problem 3.4). However, some optics
"

8 books adopt the opposite convention.

-
a r
T3 9
P
S 9
1 X3 r £3.2.1. Linear and Circular Polarizations
a
a M-1
1)
- J c Two special cases are of significant importance, namely, when the polariza-
tion ellipse degenerates into a straight line or a circle. According to Eq.
2r (3.2-4) the ellipse will reduce to a straight line when
at
5 d - f>x
y
- nrn (m = 0, l). (3.2-9)
CD
s O
V
Here we recall that all the phase angles are defined to be in the range
a) V A
A
O
a. r
- tt < 8 < tt. In this case, the ratio of the components of the electric field
vector is a constant
-1 "I
(

0
(
(3.2-10)
J.
and the light is linearly polarized.
The other special case of importance is that of a circularly polarized
c
C3 (
X \
.

wave. According to Eqs. (3.2-4) and (3.2-7) the ellipse will reduce to a circle
a
\ \
\ r when
\

n
\

3
\ \
r 8 S
y
- 64 = ± iff (3.2-U)
\
\
6 \ C i
a -
and
o

3.
(3.2-12)
on

.
9
According to our convention, the light is right-hand circularly polarized
when 6 = -
"

{tt, which corresponds to a counter-clockwise rotation of the

+
J! electric field vector, and left-hand circularly polarized when 8 = jtt which
,
3 a
3 5 corresponds to a clockwise rotation of the electric field vector (see Table
o
'
£3
n
3 1)..
.

3 3
3
3
The elliptitity of a polarization ellipse is defined as
V5 8 8
8 8-

J3
e
±1. (3.2-13)

where a and b are the lengths of the principal axes The ellipticity is taken as
.

positive when the rotation of the electric field vector is right-handed and
negative otherwise.

59
 the complex plane represents a unique polarization state. Each point on the
fS3\ COMPLEX-NUMBER REPRESENTATION
x axis represents a linearly polarized state with different azimuth angles of
r oscillation. Only two points (0, + 1) correspond to circular polarization.
From the discussion in the previous section we found how the polarization
Each point of the rest of the complex plane corresponds to a unique
state of a light wave can be described in terms of the amplitudes and the r
elliptical polarization state,
phase angles of the x and y components of the electric field vector. In fact, The inclination angle <f> and the eUipticity angle 6 (8 = tan 'e) of the
all the information about the polarization of a wave is contained in the r
polarization ellipse correspond to a given complex number x and are given
complex amplitude A of the plane wave (3.2-1). Therefore, a complex
( by
number x defined as
A
2Re[x]
'fi *.>
( tan 20 (3.3-2)
X = e tan = : V<*, i -lxlz
-

(
and
is suficient to describe the polarization states. The angle is defined to be "

( 21m[x]
between 0 and ir/l. A complete description of the ellipse of polarization, sin 20 = - (3.3-3)
which includes the orientation, sense of revolution, and ellipticify [see Eq. ( 1 + Ixl2
(3.2-13)], can be expressed in terms of 5 and . Figure 3.3 illustrates various
different polarization states in the complex plane. It can be seen from the
figure that all the right-handed elliptical polarization states are in the lower 34
. .
JONES-VECTOR REPRESENTATION
half of the plane, whereas the left-handed elliptical polarization states are in (

the upper half of the plane. The origin corresponds to a linear polarization The Jones vector, introduced in 1941 by R. C. Jones [1], describes eficiently
state with direction of oscillation parallel to the x axis. Thus, each point on C the polarization state of a plane wave. In this representation, the plane wave
(3.2-1) is expressed in terms of its complex amplitudes as a column vector
c

(
J (3.4-1)
(
0
( Notice that the Jones vector is a complex vector, that is, its elements are
complex numbers. J is not a vector in the real physical space; rather it is a
<L vector in an abstract mathematical space. To obtain, as an example, the real
x component of the electric field, we must perform the operation E (t) = x
L
Re[J e ] = R4Axe«u,+8')].
i"<

VV -i-- x

The Jones vector contains complete information about the amplitudes

and the phases of the electric-field-vector components. It thus specifies
T 0 the wave uniquely. If we are only interested in the polarization state of the
L

0 <) "o L .
wave, it is convenient to use the normalized Jones vector which satisfies the
condition that

0 L J* . J = 1 (3.4-2)
L .

where the asterisk (*) denotes complex conjugation Thus, a linearly

.

polarized light wave with the electric field vector oscillating- along a given
Figure 3.3. Each point of the complex plane is associated wiih a polarization siaie
.
C

t _

1
 64 POLARIZATION OF LIGHT WAVES

direction can be represented by the Jones vector

polarizations R and 1- and vice versa. These relations are given by
r
/ cos \
(3.4,3) (3.4-9)
[sin*} '

r
where \p is the azimuth angle of ihe oscillation direction with respect to the v L (3.4-10)
axis. The state of polarization which is orthogonal to the stale representcit (
by Eq. (3.4-3) can be obtained by the substitution of by + lending
10 a Jones vector (
S = - -(R + L) , (3.4-11)

sin \p \
cos ip I
' (3.4-4)
(
(R-L). (3.4-12)

!
Circular polarizations are seen to consist of linear oscillations along the x
For special case when = 0 represents linearly polarized waves whose and y directions with equal amplitude 1/ /2 , but with a phase difference of
electric field vector oscillates along the coordinate axes, the Jones vectors
( jtt. Similarly, a linear polarization can be viewed as a superposition of two >
are given by oppositely sensed circular polarizations.
( We have so far discussed only the Jones vectors of some simple special t

(3.4-5) cases of polarization. It is easy to show that a general elliptic polarization

( can be represented by the following Jones vector:

Jones vectors for the right- and left-hand circulariy polarized light waves are c cos 4" \
J(<M) (3.4-13)
given by
(
»
-*( !). -
(3.4-6) This Jones vector represents the same polarization state as the one repre-
sented by the complex number x ~ e tan . Table 3.2 shows the Jones i

c vectors of some typical polarization states.

The most important application of Jones vectors is in conjunction with
(3.4-7) the Jones calculus. This is a powerful technique used for studying the
propagation of plane waves with arbitrary states of polarizations through an
arbitrary sequence of birefringent elements and polarizers. This topic will be
These two circular polarizations are mutually orthogonal in the sense that considered in some detail in Chapter 5.

R* -L = 0. (3-4-8) PROBLEMS

3 1 Derive Eq. (3.2-4).

Since the Jones vector is a column matrix of rank 2, any pair of orthogonal . .

Jones vectors can be used as a basis of the mathematical space spanned by L 32 . ,

Derive Eqs. (3.2-7) and (3.2-8).
all the Jones vectors. Any polarization can be represented as a superposition
L 33 Show that the end point of the electric vector of an elliptically
of two mutually orthogonal polarizations x and y, or R and L. In particular,
. .

we can resolve the basic linear polarization x and y into two circular polarized light will revolve in a clockwise direction if sin S > 0 and in
a counterclockwise direction if sin 5 < 0 .

i
 6 POLARIZATION OF LIGHT WAVES
Table 3.2. Jones Veciors of Some Typical Polarization States
r
34
. .
(a) A right-hand circularly polarized wave (sin 6 < 0) propagating in
Polarization Ellipse Jones Vector the z direction has a finite extent in the x and y directions.
( Assuming that the amplitude modulation is slowly varying (the
(i) c
wave is many wavelengths broad), show that the electric and
magnetic fields are given approximately by

F(.«:, y, 2,0 = £:0(x, y)(x - ij)

(?) (
+

k
H(jc, y, z,t) ~ i-E(x, y,z,t).
10 ju.

(b) Calculate the time-averaged component of angular momentum

along the direction of propagation ( + z). Show that this compo-
£(-!) nent of the angular momentum is h provided the energy of the
wave is normalized to ftw. This shows that a right-hand circularly
polarized photon carries a positive angular momentum h along
( the direction of its momentum vector (helicity).
(c) Show that the transverse components of the angular momentum
(
vanish.

( 35
. .
Orthogonal polarization states.
(
(a) Find a polarization state which is orthogonal to the polarization
state

id / cos \
JdM)
( \ e
'B
sin ty)
Answer:

sin

e
'
(tr+a)cos )
(b) Show that the major axes of the ellipses of two mutually orthogo-
nal polarization states are perpendicular lo each other and the
senses of revolution are opposite.
36 . .
(a) An elliptically polarized beam propagating in the z direction has
L a finite extent in the x and y directions:

v
.

EU >>, 2, t) = E0{x y){ax + /iy)

, '-
65

if
 67 (d) Show that the major axes of the polarization ellipses are mutually
orthogonal and ihe ellipiticities ate of the same magnitude with
'f'
where a - cos ip, ji = sin ti .
Show thai the electric iiekl nuisi opposite signs.
have a component in the z direction [see Problem 3.4(a)] and r
,

derive the expressions for the electric field and the magneuc lield.
r
(b) Calculate the z component of the angular momentum, assuming
that the total energy of the wave is hv. Answer: /. .
_
=

c REFERENCES AND SUGGESTED READING

- h sin 2 yin 5.

(c) Decompose elliptically polarized light inlo a linear superposition l W. A. ShurclifT, Polarized Light, Harvard University Press, Cambridge, 1966.
of right-hand and left-hand polarized states R and L i.e., if J is ,

2 M. Born and E. Wolf, Principles of Optics, MacMillan, New York, 1964.

the Jones vector of the polarization state in (a), find r and / such .

that 3 .
R M. A. A2zam and N. M Bashara, EUtptornei/y and Polarized Light. Norlh-Holland.
Amsterdam, \911.
J rR + /L.
4 EE. Wahlstrom, Optical Cn-stalhgrapity, Wiley, New York, 1969.
(d) If r and / are the. probability amplitudes that jhe photon is
right-hand and left-hand circularly polarized respectively, show ,

that the angular momentum can be obtained by evaluating

\r\2-\l\2:
Lz = h{\r?-\1?) .

(e) Express the angular momentum in terms of the ellipticity angle d. (

37
. . Denve Eqs. (3.3-2) and (3.3-3). ( l
38 Orthogonal polarization slates. Consider two monochromatic plane
(
. .

waves of the form

*'
E z /) = RelAe""" 1]
-

, ,

.
£ (2,0 = Re[Be'l-"-*£)].

The polarization states of these two waves are orthogonal that is, ,

A* * B = 0.

(a) Let 50, S;, be the phase angles defined in Eq. (3.2-2). Show that
S
fl
-

Sfi= ±ir.

(b) Since 5fl) 5,, are all in the range -tt < S it, show thai

$A < o.

(c) Let xu, Xb the complex numbers representing the polarization

slates of these two waves. Show that

XuXb
-
1
.

v
 122
JONES CALCULUS AND ITS APPLICATION
axes pj the crystal for that
"

orthogonal and are called the "slow" and "fast

direction of propagation. Retardation plates are usually cut in such a way
tion
that the c axis lies in the plane of the plate surfaces. Thus the propaga
direction of normally incident light is perpendicular to the c axis.
5 Retardation plates (also called wave plates) and phase shifters are
polarization-stale converters, or transformers. The polarizat
ion state of a
c light beam can be converted to any other polarization state by means of a
suitable retardation p late. In formulating the Jones matrix .method, we

Jones Calculus and its assume that there is no reflection of light from either surface o
f the plate
In practice,
Application to Birefringent aPiL the lifcht is totally transmitted through the plate surfaces.
f there is reflection, though most retardation plates are coated so as to reduce
Optical Systems the surface reflection loss. The Fresnel ref lections at the plate surfaces not
c
only decrease the transmitted intensity, but also affect the fine structure of
the spectral transmittance because of multiple-reflection interference (see
Many sophisticated birefringent optical systems, such as wide-angle electro- Section 5.5). Referring to Fig. 5.1, we consider an incident light beam with
optic modulators [1], Lyot filters [2-5] and Sole filters [6, 7] involve the
polarization state described by the Jones vector
passage of light through a train of polarizers and retardation plates. The
effect of each individual element, either polarizer or retardation plate, on
i
the polarization state of the transmitted light can be easily pictured without V (5-1-1)
the aid of any matrix algebra. However, when an optical system consists of V '
C \ y}
many such elements, each oriented at a different azimuth angle, the calcula-
tion of the overall transmission becomes complicated and is greatly facili- (
tated by a systematic approach. The Jones calculus, invented in 1940 by where V and V are two complex numbers. The x and y axes are fixed
R C. Jones (8], is a powerful 2 X 2-matrix method in which the state of
.
f laboratory axes. To determine how the light propagates in the retardat.on
polarization is represented by a two-component vector (see Section 3.4)
*
while each optical element is represented by a 2 X 2 matrix. The overall y
matrix for the whole system is obtained by multiplying all the matrices, and
c f
the polarization state of the transmitted light is computed by multiplying
the vector representing the input beam by the overall matrix. We will first L
derive the mathematical formulation of the Jones matrix method, and then
apply it to some birefringent filters. v _

I
.

5 1 JONES MATRIX FORMULATION

. .

We have shown in the previous chapter that light propagation in a birefrin-

gent crystal consists of a linear superposition of two eigenwaves. These
eigenwaves have well-defined phase velocities and directions of polarization. L
The birefringent crystals may be either uniaxial or biaxial. However, most
commonly used materials such as calcite and quartz are uniaxial. In a
uniaxial crystal, these eigenwaves are the ordinary and (he extraordinary
c Figure 5.1. A retardation plate with azimuth angle ty.
wave. The directions of polarization for these- eigenwaves are mutually
-

-!
121 L
 JONES CALCULUS AND ITS APPLICATION
JONES MATRIX FORMULATION 3 r
The Jones vector of the polarization state of the emerging beam in the xy
plate, we need to decompose the light imo a linear combination of ihc coordinate is given by transforming back from the crystal sf coordinate
"
fast" and "slow" eigenwaves of the crystal. This is done by the coordinate system:
transformation r
cos V> - sin i//
IV\ l cos i// sin \lVx\ r
I Vx \ r sin i// cos \p
(5-1-7)

By combining Eqs. (5.1-2) (5.1-6), and (5.1-7), we can write the transforma-
,

V . is the slow component of the polarization vector V, whereas Vf is the fast tion due to the retardation plate as
and "fast" axes are fixed in the crystal. These two
" "

component. The slow

components are eigenwaves of the retardation plate and will propagate with
cheir own phase velocities and polarizations. Because of the difference in
phase velocity, one component is retarded relative to the other. This
c

(
( J = /U-mK<*)( ), (5.1-8)
retardation changes the polarization state of the emerging beam.
and be the refractive indices for the "slow" and "fast"
where R( ) is the rotation matrix and WQ is the Jones matrix for the
Let ns (
retardation plate These are given respectively, by
components, respectively. The polarization state of the emerging beam in
.
,

the crystal sf coordinate system is given by

cos sini/A
(5.1-9a)
/ \ I ( sin \p cos i// /
exp(-m /) , 0 (5.1-3) and

\
0 exp -w /j / t
where / is the thickness of the plate and co is the frequency of the light beam.
'
W"
r "
"

i'T A)- (5-'-9b)

The phase retardation is given by the difference of the exponents in (5.1-3)
and is equal to TAe Phase ctor
_-
.
can be neglected if interference effects are not
ilDPPI i QLJ3£LobserraWe. A retardation plate is characterized by its
v _

c
(5.1-4) phase retardation F and its azimuth angle ty, and is represented by the
product of three matrices (5.1-8):
Notice that the phase retardation T is a measure of the relative change m W= R{-i>)W0R{i>). (5 )-10)
phase, not the absolute change. The birefringence of a typical ciyst;il
.

retardation plate is small, that is, |rtf - n nj. Consequently, the Note that the Jones matrix of a wave plate is a unitary matrix that is,
absolute change in phase caused by the plate may be hundreds ofjjmes
,

greater than the phase retardation. Let $ be the mean absolute phase W*W*= 1,
change,
r where the dagger1 means Hermitian conjugate The passage of a polarized
.

(5.1-5) light beam through a wave plate is mathematically described as a unitary

transformation .

Many physical properties are invariant under unitary trans-

formations; these include the orthogonal relation between the Jones
Then Eq. (5.1-3) can be written in terms of $ and F as
v and the magnitude of the Jones vectors Thus if the polarization states of
.
vectors

\ two beams are mutually orthognna they _

will ,
remain orthoRona] after
P§ing through an arbitrary wave glate.
V
_
JONES MATRIX FORMULATION 125

126 ; JONES CALCULUS AND ITS APPLICATION

The Jones matrix of an ideal homogeneous lineur sheet polanzer oriunled
with its transmission axis parallel to the laboratory x axis is r
(5.1-9) and (5.1-10):
f
-

i<t>
M
10
0\
0/'
(5.1-11)
r
w
(1 !)( o 2 (-i i) (-/ o) -

(5.1-14)
where is the absolute phase accumulated due to the tiniie optical thickness r
of the polarizer. The Jones matrix of a polarizer rotated by an angle \p about The Jones vector for the emerging beam is obtained by multiplying Eqs.
(
z is given by (5.M4) and (5.1-13); the result is
c
(5.1-12) ( Mi)-
"

o
(5.1-15)
(
Thus, if we neglect the absolute phase 4> the Jones matrix representations of
} This is horizontally polarized light. The effect of the half-wave plate is to
the polarizers transmitting light with electric field vectors parallel to the x rotate the polarization by 90 It can be shown that for a general azimuth
°
.

and y axes, respectively, are given by angle the half-wave plate will rotate the polarization by an angle 2 (see
Problem 5.1). In other words, linearly polarized light remains linearly
polarized, except that the plane of polarization is rotated by an angle of 2i/>.
*
lo o;
and P,
y il ?). When the incident light is circularly polarized, a half-wave plate will
( convert right-hand circularly olariz|d light into left-hand circularly
To find the effect of a train of retardation plates and polarizers on the polarized light and vice versa, regardless of the azimuth angle. The proof is
polarization state of a polarized light, we write down the Jones vector of the c left as an exercise (see Problem 5.1). Figure 5.2 illustrates the effect of a
incident beam, and then write down the Jones matrices of the various half-wave plale.
elements. The Jones vector of the emerging beam is obtained by carrying c
out the matrix multiplication in sequence. 5 12
. . . Example: A Quarter-Wave Plate
(

5 1 1
A quarter-wave plate has a phase retardation of F = {tt. If the plate is made
. . . Example: A Half-Wave Retardation Plate c
of an x-cut (or j>-cut) uniaxially anisotropic crystal, the thickness is i =
\/4(n - n } (or odd multiples thereof). Suppose again that the azimuth
A half-wave plate has a phase retardation of Y = tt According, lo Eq. e 0

angle of the plate is \p ~ 45 and the incident beam is vertically polarized.

.
°

(5.1-4) an x-cut* (or y-cut) uniaxial crystal will act as a half-wave plate The Jones vector for the incident beam is given again by Eq. (5.1-13). The
provided the thickness is i = \/2(ne - n0). We will determine the effect of
Jones matrix for this quarter-wave plate, according to Eq. (5.1-10), is
a half-wave plate on the polarization state of a transmitted light beam The .

L '

azimuth angle of the wave plate is taken as 45° and the incident beam as 1 -
:)(
if/4 o
) Lf i
-

vertically polarized. The Jones vector for the incident beam can be written
_

1 0
as

L
V
(?) (5.1-13)
L
fii-i l)' (5.1-16)

The Jones vector for the emerging beam is obtained by multiplying Eqs.
and the Jones matrix for the half-wave plate is obtained by using Eqs .

(5.1-16) and (5.1-13) and is given by

*
A crystal plate is called x-cut if its faces are perpendicular to the principal x axis V (5.1-17)
.
 128 JONES CALCULUS AND ITS APPLICATION i

r
45°

r
plate
2

plate
4

c
plats
2

(
plate
4
Figure 5.2. The efl'eci of a half-wave plate on the polarization state of a beam .

c
This is left-hand circularly polarized light The effect of a 450-oriented
.
Figure 5.3. The elTect of a quarter-wave plale on the polarization sfate of a linearly polarized
beam.
(
quarter-wave plate is to convert vertically polarized light into left-hand .

circularly polarized light If the incident beam is horizontally polarized the

.

(
emerging beam will be right-hand circularly polarized The effect of (his .

analyzer is usually required. An analyzer is basically a polarizer. It is called

quarter-wave plate is illustrated in Fig. 5,3.
an analyzer simply because of its location in the optical system. In most
birefringent optical systems, a polarizer is placed in front of the system in
order to prepare a polarized light. A second polarizer (analyzer) is placed
" "

52
. . INTENSITY TRANSMISSION
at the output to analyze the polarization state of the emerging beam.
So far our development of the Jones calculus was concerned with the
Because the phase retardation of each wave-plate is wavelength-dependent,
the polarization state of the emerging beam depends on the wavelength of
polarization state of the light beam. In many cases ,
we need to determine L
the light. A polarizer at the rear will cause the overall transmitted intensity
the transmitted intensity A narrowband filter for example
.
transmits
, ,

L lo be wavelength-dependent.
radiation only in a small spectral regime and rejects (or absorbs) radiation The Jones vector representation of a light beam contains information
at other wavelengths. To change the intensity of the transmitted beam ,
an
about not only the polarization state but also the intensity of light. Let us.
L
-

-!
now consider the light beam after it passes through the polarizer. Its electric

L (
 JONES CALCULUS AND ITS APPLICATION
vector can be written as a Jones vector
r

E
(-:) (5.2-1) r

( Z1
The intensity is calculated as follows:
( s
45°
/ = Et-E = \E \2 + X
(5.2-2) -

Analyzer

where the dagger indicates the Hermitian conjugate. If the Jones vector of c
the emerging beam after it passes through the analyzer is written as
(
Birefringent
plaie
(5.2-3) (

(
the transmissivity of the birefrmgent optical system is calculated as
{ Polatizer

i i2 + \e: |2 A birefnngent plate sandwiched bci n a pair of parallel polar-zers.

'

T ' (5.2-4) ( Figure 5.4.

| l2 + |£/
f

52 1
. . . Example: A Birefrmgent Plate Sandwiched between Parallel ( Jones vector:
Polarizers
(5.2-7)
Referring to Fig. 5.4 we consider a birefringent plale sandwiched between a
,

(
pair of parallel polarizers. The plate is onented so thai the "slow" and (w thP intensity of the incident beam is unity and only
"

fast" axes are at 45° with respect to the polarizer. Let the birefringence be where we assume that the mtensi y oi represen-
n - n and the plate thickness be d. The phase retardation is then given by
c .

e 0

c follows:
I = 2'rr{n e
-

n
0 ) t (5.2-5)
(0 0\[ cos F -isinirU [o\
and the corresponding Jones matrix is according to tiq. (5.1-10),
,
ijl-csinir cos L J72\l|
-

(Sinir\
(5.2-6)
L
= j_| cos0 r \ (5.2-8),
\ -i'sin4r cos T j \

Lei the incident beam be unpolarized so that after it passes through the
,
The transmitted beam is vertically (,) polarized with an intensity given by
front polarizer, the electric field vector can be represented by the following (5.2-9)
/ = icosur = icoS2[ x !'

-
:

f
 FULAKIZA 1 ION IN 1 tKMlKtNCt flLltRb I J I

It can be seen from Eq. (5.2-9) that the transmitted intensity is a sinusoidal JONES CALCULUS AND ITS APPLICATION
function of the wave number and peaks at A = (n - n )d, (n - nil)cl/2, , 0 r Table 5.1. The Folded Sole Filter
(ne - n0)d/3,... . The wave-number separation between transmission max-
ima increases with decreasing plate thickness. r Element Azimuth

r Front polarizer 0
°

522
. . . Example; A Birefnngent Plate Sandwiched belween a Pair of Crossed
Plate 1 P
Polarizers
r Plate 2 -

P
Plate 3 P
If we rotate the analyzer shown in Fig. 5.4 by 90°, then the input and
output polarizers are crossed. The transmitted beam for this case is obtained
as follows: Plate If
r
Rear polarizer 90°

'
0 cos F /siller
-

(
E
0 0 cos r ft \ i
(
gent crystal plates (wave plates) and polarizers. The two basic versions of
-

i 1 sini? these birefringent filters are Lyot-Ohman filters [2-5, 121 and Sole filters [6,
(5.2-10)
72 0 7]. They are based on the interference of polarized light, which requires a
(
.
phase retardation between the components of the light polarized parallel to
The transmitted beam is horizontally (x) polarized with an intensity given the fast and slow axes of the crystal when radiation passes through it. Since
by ( the phase retardation introduced by a waveplate is proportional to the
birefringence of the crystal, it is desirable to have crystals with large
(
. 4r- [ p ]
/ .
(.5.2-11) birefringence (n ~ n ) for filter construction. Currently, the most com-
e 0

monly used materials are quartz, calcite, and (NH4)H2PO<, (ADP).

(
The basic principle of the Lyot-Ohman filter is very simple and is
This is again a sinusoidal function of the wave number. The transmission ( outlined in Problem 5.5.
spectrum consists of a series of maxima at X = 2(« - n d 2{ne - 1, , The Jones calculus developed in the previous two sections will now be
n )d/3
0 These wavelengths correspond to phase retardations of v, 3i7t C applied to study the transmission characteristics of the Sole filters. Sole
577,..., that is, when the wave plate becomes a half-wave plate or odd filters, named after their inventor [6, 7], are composed of a pile of identical
integral multiple of a half-wave plate. c birefringent plates each oriented at a prescribed azimuth angle. The azimuth
angle of each* plate is measured from the transmission axis of the front
( _

polarizer. The whole stack of birefringent plates is placed between a pair of

53 POLARIZATION INTERFERENCE FILTERS polarizers.
. .

( .
.

Spectral filters can be based on the interference of polarized light. These L 531
. . .
Folded Sole Filters
1
filters play an important role in many optical systems where filters of
extremely narrow bandwidth with wide angular fields or tuning capability L
There are two basic types Of Sole filters: folded and fan filters. The folded
are required. In solar physics, for example, the distribution of hydrogen may Sole filter works between crossed polarizers. The azimuth angles of the .

be measured by photographing the solar corona in the light of the H a

individual plates are prescribed in Table 5.1. The geometrical arrangement
(A = 6563 A) line. In view of the large amount of light present at neighbor-
L of a six-plate Sole filter is sketched in Fig. 5-5. As described in Table 5.1, the
ing wavelengths, a filter of extremely narrow bandwidth (~ 1 A) is required
if reasonable discrimination is to be attained. Such filters consist of birefrin-
front polarizer has its transmission axig parallel to the x axis; and the rear
C polarizer parallel to the -axis. The overall Jones matrix for these N plates is
-

-!

f
 i

POLARlZtR 134 : -

JONES CALCULUS AND ITS APPLICATION

C
with
(

(
A = (cos F - /cos2psin r)2 + sin22psin4r,

f
B = Rin4psin2-5r,
(5.3-3)
C= -B
( ,

r D = (cos T + icos2psin r)2 + sin22psin24r,

Y r
c where Y is the phase retardation of each plate. Note that this matrix is
unimodular (i.e., AD - BC = 1), since all the matrices in Eq. (5.3-1) are
( unimodular. Equation (5.3-2) can be simplified by using Chebyshev s iden-
'

F Y
(
tity [9] to
m
A

s
{
[ c I)
( \
A sin mKA - sin(m - ])KA sin mKA
L - s
sin KA
g
sin kA
x (
sin mKA Dsin mKA ~ sin(m - \)KA
C
( .
sin KA sin KA

( (5.3-4)
POLARIZER

K .

with

Figure S S.
.

A su-siage folded Sole filler. K

KA = co$-][\{A + D)}. (5.3-5)
( "
.
_

We use the notation KA for the purpose of comparing this result with
given by
{ thai obtained from coupled-mode theory (see Section 6.4).
The incident wave and the emerging wave are related by

where we assume that the number of plates is an even number N = 2m

L (5.3-6)

L
The emerging beam is polarized in the y direction with a field amplitude
{cA ,

M
(5.3-2) L given by

L E; = M E
,
2] X (5.3-7)
-
j

L ' If the incident light is linearly polarized in the x direction the transmissivity
,

L
 POLARIZATION INTERFERENCE FILTERS
1 15
.

of this filter is "

( 136 JONES CALCULUS AND ITS APPLICATION

(5.3-8) ( azimuth angle -p, making an angle of 3p with respect to the polarization
From Eqs direction of light incident on it. The polarization direction at its output face
.

(5.3-2) and (5.3-3) , we obtain ( will be rotated by 6p and oriented at the azimuth angle -4p (see Fig. 5.6).
The plates are oriented successively at +p, - p, +p, -p,..., while the
,2
polarization direction at the exit of the plates assumes the values 2p, - 4p,
2 sin KA (5-3-9) 6p, ~8p,... , The final azimuthal angle after N plates is thus 2jVp. If this
with final azimuth angle is 90° (i.e., 2Np = {w), then the light passes through the
( rear polarizer without any loss of intensity. Light at other wavelengths ,

where the plates are not half-wave plates does not experience a 90° rotation
,

cos*A - J - 2cos22pSin2ir .

(5 3.10) ( of polarization and suffers loss at the rear polarizer.

The transmissivity Tis often expressed in terms of a The Sole filter can also be viewed as a periodic medium as far as wave
by new variable x , defined ( propagation is concerned. The alternating azimuth angles of the crystal axes
constitute a periodic perturbation to the propagation of both eigenwaves .

This perturbation couples the fast and slow eigenwaves. Because these
KA = tt - 2x .

(5.3-11) waves propagate at different phase velocities complete exchange of electro-

,

(
In terms of this new variable x the transmissivity is
,
magnetic energy is possible only when the perturbation is periodic so as to
maintain the relations necessary to cause continuous power transfer from
(
2 the fast to the slow wave and vice versa. This is the first manifestation of the
r = |tan2/>cos X
iHZx
sinx (5.3-12) r principle of phase matching by a periodic perturbation, to which we will
return in subsequent chapters. The basic physical explanation is as follows:
with ( If power is to be transferred gradually with distance from modey4 to mode
B by a static perturbation then it is necessary that both jvaves travel with
,

{ the same phase velocity. If the-phase velocities are not equal, the incident
cosx - cos2psii r .

(5.3-13) wave A gets progressively out of phase with the wave B into which it
(
couples. This limits the total fraction of power that can be exchanged. This
Cwhen
hTtthe
te phase
nhf retardat
C5t,2)ion of each platetra- ty becomes T = sin>2N
is r = vr li W th , t
-
a situation can be avoided if the sign of the perturbation is reversed whenever
atwhen (
each plate becomes a haJf-wave olate Th , , .
.

the phase mismatch (between the coupled field and the field into which it
azimuth angle p is such that transnnsstvuy is 100 if the couples) is equgi to tt. This reverses the sign of the coupled power and thus
,

maintains the proper phase for continuous power transfer. A coupled mode
IT theory of the folded Sole filter is given in Section 6.5.
P
AN' (5.3-14) The transmission characteristics around the peak and its sidelobes in a
i Sole filter are interesting and deserve some investigation. Assume that each
The transmission under these conditions can be easily understood if plate is characterized by refractive indices ne and n0 and thickness d. Let \v
examine the polarization state after passing through each we denote the wavelength at which the phase retardation is {2v + The
§olc filter We recall that in passing through
.
plate within the
a half wave (T = tt 37r,... ) phase retardation at a general wavelength is
plate the azimuthal angle between the polarization
,
vect
-

L
slow) axis of the cr or and the fast (or
linearl ystal changes sign. Pas* the front poiartzer, the light is (5.3-15)
y polarized in the x direction (azimuth = 0) Since the first plate is
at the azimuth angle p, the emerging beam af
.

plate is linearly polarized at = 2p The second ter passage through the first If X is slightly away from \ [i.e., (X - X„)
„ X„], F can be approximately
plate is oriented at the
.

given by
T = {2v + l)7r + AT
 to ui o
f JONES CALCULUS AND ITS APPLICATION
{ 7 ? 138

r where

r
Ar_ ( Jl5(x _

u . (5.3-16)
in "> in (
LI

a
tj
i
We assume further that the azimuth angle of the plate obeys the condition
s
H a
(5.3-14), and N is much larger than 1. Under these conditions, the trigono-
a
.
metric functions in Eq. (5.3-13) can be expanded to yield
CO
I 1/2
-
g ja
X (5.3-17)
2AM \
O

Substituting x into Eq. (5.3-12), we obtain

.

Si
' 2

0
I / sin /lTTiVAT/V)
« -
T (5.3-18)
2
?
o
p
(
.
a .2 /
3 §
i! This approximate expression for the trans mi ssivity is valid provided N » 1
c
and (A. - \ ) « \v. From (5.3-18), the full bandwidth at half maximum
y

« « o i 1 (FWHM) of the main transmission peak is given approximately by Ar]/2 =

1 . 607r/Ar, which in terms of the wavelength is

1 « II
aV = '-6o[ I_]. (5.3-19)

i "

3 .a
[2

Thus, to build a narrowband Sole filter with a bandwidth of 1 A to observe

the H line (A0 *= 6563 A) requires the number of half-wave {v = 0) plates
0
_

It be approximately 104. The transmission spectrum consists of a main peak at

\0 and a series of sidelobes around it. According to Eq. (5,3-18), these
01 secondary maxima occur approximately at
w co -p -fa

I 1 11 2/+ 1, /= 1,2,3,..., (5.3-20)

L
with transmissivity given by
137 1
L 2
"
(5.3-21)
(2/+ 1)
 u ) i I m r-i I <

f 140 JONES CALCULUS AND ITS APPLICATION

0 8
.
C Fig. 5.8. According to the Jones matrix method formulated in the previous
section, the overall matrix for these N plates is given by
(
06
.
M /?(-J2ir + pKK(iir-p)
« (

ra
X/?(-5p) (5p)fi(-3p) (3p)i?(-p) 0/?(p)
h - 0 4
.

= + p)| K(2p)r*(-p),
o (5.3-22)
0 2
.

r where we have used the following identity for the rotation matrix:

0 0 1 U
r /{(p,)* ,)- (p, + p2). (5.3-23)
.

0 0
.
05.
1 .
0 1.5 2.0 2.5 3.0
P/tt ( Notice that the last plate always appears first in the product (5.3-22).
Figure 5,7. A calculated transmission spectrum of a folded Sole filter. By using the Chebyshev identity (5.3-4) and carrying out the matrix
multiplication in Eq. (5.3-22), we obtain

sin iVx
M sin 2pcos r
A calculated transmission spectrum is shown in Fig. 5.7. Notice that the sin x
(
bandwidth is inversely proportional to the total number of plates.
( sin jVx
M 12 -

cos - /sin r
532 Fan Sole Filters sinx
. . .

(5.3-24)
(
A fan Sole filter also consists of a stack of identical birefringenl plates, each
(
M21. cos Nx - f sin
'

r- ? *
oriented at a prescribed azimuth. A brief description of the basic type of fan sm x
Sole filter is given in Table 5.2. The geometrical arrangement is sketched in ( M 12 M

c with
Table 5.2. The Fan Sole Filter
( .
cosx == cos2pcos r. (5.3-25)
Element Azimuth

°
( ,
These are the elements of the overall Jones matrix not including the
Front polarizer 0
polarizers.
Plate 1 P
Plate 2
The incident wave E and the emerging wave E' are thus related by
Plate 3 5p L

Plate N (IN - \)p= {it- p

(5H::)(:;;:;:)(: :)(*;) <>«
Rear polarizer 0U
The emerging beam is horizontally polarized (x) with amplitude given by

i
E' = MUE
X X
. (5.3-27)
 142 JONES CALCULUS AND ITS APPLICATION
s f
If the incident wave is Hnearly polarized in the x direction, the transmis-
sivity is given by

l
r = )Mn)2. (5.3-28)

i
From Eq. (5.3-24), we have the following expression for the transmissivity:
2
sin A x
T tan 2 p cos x (5.3-29)
sinx
b (
with

( cos x = cos2pcos r. (5.3-30)

I
in
oO
( Notice that the transmission formula (5.3-29) is formally identical to Eq.
3
(5.3-12). Maximum transmissivity (T = 1) occurs when T = 0, 2tt, Air,...,
a
.

and p = -n/AN. This unity transmission results simply from the fact that at
<
fhese wavelengths the plates are full-wave plates. The light will remain
{
linearly polarized in the x direction after going through each plate and will
suffer no loss at the rear polarizer. Light at other wavelengths, where the
a (
E
,
plates are not full-wave plates, does not remain hnearly polarized in the x
direction and suffers loss at the rear polarizer. Let \ be the wavelength at y
1
which the phase retardation F = 2vu If A differs slightly from \ (i.e.,
.
B

( A - A <«; A ) F is given approximately by

t, ,, ,

(
F = 2pit + AF = Ivir - 4~-(a - A ) „ , (5.3-31)
(

where v = 1 2 3
, , ,... . The case v = 0 occurs only when the birefringence
vanishes at some particular wavelength A0 This special case has some .

important applications in wide-angle narrowband filters and deserves spe-

{ .

cial attention (see Problem 5.8). If we now further assume that N is much
larger than 1 and follow the same procedure as in Eq. (5.3-17), we obtain
\
the following approximate expression for the transmissivity:
\ 2

sin 7r/l + {mY/itf

L r I (5.3-32)
\ i
L

L which is identical to Eq (5.3-18). The FWHM AAl/2 of the transmission

.

maxima is again given approximately by

L
!4l AA
1/2
1 60
.
v 1 2 3
, , (5.3-33)
 1 0
.
r i-i-i-i-n-i-n i i i i i i r i r !-i-i-i-i-i-r-r~i

144 JONES CALCULUS AND ITS APPLICATION

0 8
.
phase retardation and an azimuth angle. The overall Jones matrix can then
( be obtained by multiplying together all the matrices associated with these
plates. 1

We will limit ourselves to the case when the twisting is linear and the
§ 0.6
.
S azimuth angle of the axes is

K 0.4
where z is the distance in the direction of propagation and a is a constant.
Let F be the phase retardation of the plate when it is untwisted. In
0 2
. particular, for the case of nematic liquid crystal with c axis parallel to the r

plate surfaces, T is given by

o o 2tt , ,
.

0 0
. 0 5
.
1 0
.
1.5 2.0 2.5 3 0 .
r (5.4-2) .
F/ir

Figure 5.9. A calculated Iransmission spectrum of a fan Sole filter. where / is the thickness of the plate. The total twist angle is
( (5.4-3)

The transmission spectrum of the fan Sole filter is identical to that of the ( To derive the Jones matrix for such a structure, we need to divide this plate
folded Sole filter except that the curves are shifted by T = ir. In other into N equally thick plates. Each plate has a phase retardation of T/N. The
( plates are oriented at azimuth angles p, 2p, 3p,...,(A - l)p Np with
r
words, the transmissivity of the fan Sole filter at a phase retardation F is ,

identical to that of the folded Sole filter at phase retardation T + tt This '

p = Q/N. The overall Jones matrix for these N plates is given by

.

(
can also be seen from the expression for the transmissivity in Eqs (5.3-12) .
N
and (5.3-29). A calculated transmission spectrum of a fan Sole filter is M = O £(mp)W0 R(-mp). . (5.4-4)
shown in Fig. 5.9. m= 1

Sole filters play an important role in many modern optical devices such ( v

as electro-optic tunable filters [10, U] and wide- field-of-view narrowband It is important to remember that in the above matrix product, m = 1
filters. More information on Sole filters can be found in [12] .
( appears at the right-hand end. Following the procedure (5.3-23), this matrix
can be writtea
( "'

M = 7?( )[ 0i?(-|]]
.

> (5.4-5)
54
. . LIGHT PROPAGATION IN TWISTED ANISOTROPIC MEDIA c
where
In this section the propagation of electromagnetic radiation through a
slowly twisting anisotropic medium is described by the Jones calculus. The (5.4-6)
transmission of light through a twisted nematic liquid crystal is a typical L - 0 \ 0 e'rH '

example. This situation is similar to a fan-type Sole filter structure with the Using Eqs. (5.1-9a) and (5.4-6) we obtain
t
,

number of plates, N, tending to infinity and the plate thickness tending to

zero as l/N. In fact, we are going to subdivide the twisted anisotropic \ N

medium into N plates and assume that each plate is a wave plate with a N
'

N
M (5.4-7)
sin N e ' "

' 1 /2"
\
N /

r
Equation (5.4-7) can be further simphfied by using Cliebyshev s idcntiiy '

146 JONES CALCULUS AND ITS APPLICATION

(5.3-4). In the limit when N tends to infinite (N ~* oo), the result is given by
(see Problem 5.10) gives a phase shift to the light beam and leaves the polarization state
( unchanged. Second, the operation of R{<t>) is (o rotate the Jones vector by
/ v . F sin X sin X \ an angle <f>. For linearly polarized light, this rotation makes the polarization
cos X - i - * vector parallel to the principal axes al the exit face of the plate. Thus, if the
-
~

2 X
M = R{<$>) (5.4-8) incident light is polarized along the direction of the normal modes at
sip X . F sin X
the input plane (z =0), the polarization vector of the light wave will follow
„

4> cos X + i- --~

\ X
l A j
the rotation of pnncipal axes and remain parallel to the local slow (or fast)
where
i
axis provided the twist rate is small. This is called adiabatic following. This
i
phenomenon has an important application in liquid-crystal light valves. The
principle of operation of these light valves is discussed below.
X
(5.4-9) We consider the case of twisted nematic liquid crystal with a quarter turn
f
(<f> = {it) If this layer is placed between a pair of parallel polarizers with
.

( their transmission axes (x) parallel to the c axis of the liquid crystal at the
Here we have an exact expression for the Jones matrix of a linearly twisted entrance plane (z = 0), the Jones vector representation of the wave im-
anisotropic plate. { mediately after passage through the first polarizer can be written
Let V be the initial polarization state the polarization state V after
,
/

exiting the plate can be written

(5.4-12)
10/
V = M\. (5.4-10) (
54 1
. . .
Adiabatic Following ( The rotation matrix £(</>) in Eq. (5.4-8) can be written

If often happens especially in twisted nematic liquid crystals that the phase (
,
, (5.4-13)
retardation T is much larger than the [wist angle 0 For example consider a
.

liquid-crystal layer 25 thick with a twist angle of The birefringence

of liquid crystal is typically n e
- n0 = 0.1. For wavelength X = 0 5 /xm .
,
we After passing through the liquid crystal, the polarization state of the beam
becomes
have r/<f> = 20. This number can be even bigger if the layer is thicker .
If we according to Eqs. (5.4-8), (5.4-12), and (5.4-13),
assume F » <$> the overall Jones matrix (5 4-8) becomes
, .

sin X
«*>-
'

0 \ IT
M
(5.4-11) (5.4-14)
\ o eir/2j -

v
cos X
JsmX
i
£
If the incident light is linearly polarized along either the slow or the fast axis K
at the entrance plane, then according to Eq (5.4-11), the light will remain
..

The y component will be blocked by the second polarizer. The transmissiv-

linearly polarized along the "local" slow or fast axis .
In a sense ,
the ity of the whole structure is thus given by
polarization vector follows the rotation of the local axes, provided the
polarization vector is along one of the axes. The operation of the Jones
matrix on any polarization vector can be divided into two parts First, a sin 2 { Wl + (r/Tr)2)L
phase-retardation matrix operates on the Jones vector of the incident wave
.

T= -

; ,
(5.4-15)
1 +(r/w)
.

For light Unearly polarized along one of the principal axes this matrix only ,

where we have used Eq. (5.4-9) for X and =

THE PROBLEM OF FRtSNEL REFLECTION AND PHASfc SHUT M7
"

JONES CALCULUS AND ITS APPLICATION

f MB
For a cryslal layer with thickness large enough, the phase relaalaiion is
much larger than w (i.e., F » n). The transmissivity is virtually iero, (
'

according to Eq. (5.4-15). This is a result of the adiabatic following. The

polarization vector follows the rotation of the axes and therefore is rotated f
l Oh
by an' angle = ir equal to the twist angle. Since this direction is
.

\
orthogonal to the transmission axis of the analyzer, the transmission is zero. \
In many twisted liquid crystals, the c axis can be forced to align along a 0 8
.

given direction by the application of an electric field (or stress) (see Section
7 7). The application of an electric field along the z direction will destroy
I
.

this twisted structure (see Section 7.7). This leads to T = 0 and to total in CL6

transmission of light, according to Eq. (5.4-15). When the electric held is S

1 1

removed, the liquid crystal recovers its twisted structure arjd the light ts <

extinguished. This is the basic principal of liquid-crystal light valves.

0 3
.

55
. .
THE PROBLEM OF FRESNEL REFLECTION AND PHASE
SHIFT
0
( 0 0
.

AK
In the Jones-calculus formulation, the reflections of light from the surface of
the wave plate are neglected. These reflections will normally decrease the (
throughput of electromagnetic energy. However, if the plate surfaces are
(
optically fiat, the interference effect can result in a decrease or increase in
the transmissivity, depending on the optical path lengths.
PROBLEMS
The exact approaches derived from the electromagnetic theory involve
the use of a 4 X 4-matrix method [13-15]. The light wave of a beam is
5 1 we plate. A half wave plate
Half-wave
-
has a phase retardation of
represented mathematically by a column vector that consists of the complex . .

oriented so that the azimuth angle

Fr = it. Assume that the plate is
amphtudes of the incident wave and the reflected wave. Each wave lias a
(i.e., the angle between the x
axis and the slow axis of the plate) is ty.
slow and a fast component. (The details of this method are beyond the of the transmitted beam, assuming
(a) Find the polarization state tion.
scope of this book; interested readers are referred to the cited references.)
that the incident beam is linearly polarized in the)' direc
This exact approach gives the exact transmission spectrum of the bire- will convert right-hand circularly
fnngent filter, because the interference effects due to the reflected waves are (b) Show that a half-wave plate circularly
left-hand polarized light, and vice
accounted for. The practical calculation of the transmission requires a polarized light into
computer. The calculated transmission spectrum of a Sole Filter (see Fig. versa, regardless of the azimuth angle of the plate. 391
5 10) shows fine structure due to interference. This fine structure should be
.
(c) Lithium tantalate(LiTa03) is a uniaxial crystal with n0 = 2.1
half-wave-plate thick-
experimentally observable if the birefnngent plates are thin and optically and ne = 2 1432 at X « 1 ym. Find the
ness at this wavelength, assuming the plate is cut in such incipal
a way
.

flat. This approach also predicts the existence of a superfine structure that
comes from the Fabry-Perot interference fringes of the whole stack of that the surfaces are perpendicular to the x axis of the pr
crystals rather than those of the individual plates; however, this superfine L coordinate (i.e., x-cut).
structure is beyond the resolution limit of the plotter at the scale shown in
L 52 Quarter-wave plate. A quarter-wave plate has a phase retardation
tion with
of F = tt. Assume that the plate is oriented in a direc
. .

Fig. 5,10. The figure also compares the Jones-calculus method, which
neglects reflected waves, and the exact 4 X 4-matrix1method. '
azimuth angle
v

(a) Find the polarization state of the transmitted beaiji* assuming

ion.
that the incident beanj
'

is polarized in the / direct

PROBLEMS

(b) If the polarization state resulting from (a) is represenied by a 150 JONES CALCULUS AND ITS APPLICATION
*
complex number on the complex plane show that the locus of r
,

these points as varies from 0 to {it is a branch of a hyperbola. (b) Show that if E, is the amplitude of the relative field of the
r
Obtain the equation of the hyperbola .
electric field, the amplitude of the beam after it passes through
(c)
-

Quartz (a-Si02) is a uniaxial crystal with = t 53283 and . r the polarizer is given by
n
e
= 1 54152 at X - 1 1592 jinx Find the thickness of an jc-cut
.
.
P(p-Et),
quartz quarter-wave plate at this wavelength.
53
where p is the unit vector along the transmission axis of the
. .
Polarization transformation by a wave plate A wave plate is char-
.

polarizer.
acterized by its phase retardation Y and azimuth angle .

If the incident beam is vertically polarized (i.e., E, = $E0), the

(c)
(a) Find the polarization state of the emerging beam assuming that ,
polarizer transmission axis is in the x direction (i.e., p = X). The
the incident beam is polarized in the x direction .

transmitted beam has zero ampUtude, since X * f = 0. However,

(b) Use a complex number to represent the resultiag polarization if a second polarizer is placed in front of the first polarizer and
state obtained in (a) .
is oriented at 45° with respect to it, the transmitted amplitude is
(c) The polarization state of the incident x-polarized beam is not zero. Find this amphtude.
represented by a point at the origin of the complex plane Show .
(d) Consider a series of N polarizers with the first one oriented at
that the transformed polarization state can be anywhere on the i//, = tt/2N, the second one at \p2 = 2(tt/2N) the third one at
,

complex plane provided Y can be varied from 0 to 2tt and

,
3 = Xtt/IN),..., and the nth one at N {tr/lN). Let the
can be varied from 0 to \-n Physically, this means that any
.
incident beam be horizontally polarized. Show that the trans-
polarization state can be produced from linearly polarized light ( mitted beam is vertically polarized with an ampUtude of
,

provided a proper wave plate is available.

(
(d) Show that the locus of these points in the complex plane cos - !
V 2A';j
'

obtained by rotating a wave plate from = 0 to is a C

hyperbola. Derive the equation of this hyperbola .
Evaluate the amplitude for N - 1,2,3,..., 10. Show that in the
(e) Show that the Jones matrix W of a wave plate is unitary ,
thai is, limit of N ~* oo, the amplitude becomes one. In other words, a
series of polarizers oriented like a fan can rotate the polariza-
W W = 1 , tion of the light without attenuation.

where the dagger indicates Hermitian conjugation 55

. . Lyot-Ohman filter. In solar physics, the distribution of hydrogen in
the solar corona is measured by photographing at the wavelength of
.

(f) Let andbe the transformed Jones vectors from Vl and V2 ,

H line J\ = 6563 A). To enhance the signal-to-noise ratio, a filter of
respectively Show that if V1 and V2 are orthogonal so are V[ Q

extremely narrow bandwidth (~ 1 A) is required The polarization

.

,
.

and \{.
filter devised by Lyot and Ohman consists of a set of birefringent
54
. .

Polarizers and projection operators. An ideal polarizer can be con- plates separated by parallel polarizers. The plate thicknesses are in
sidered as a projection operator which acts on the incident polariza- geometric progression, that is, d, 2d, Ad, %d,... . All the plates are
tion state and projects the polarization vector along the transmission oriented at an azimuth angle of 45
°
.

axis of the polarizer. (a) Show that if n0 and ne are the refractive indices of the plates,
(a) If we neglect the absolute phase factor in Eq. (5 1-11) .
,
show that L the transmissivity of the whole stack of N plates is given by

Po2 = PQ and P2 = P .
L T = cos2 cos22xcos4x cos22A' 'x
with
L .

Operators satisfying these conditions are called projection oper- TTd{n ~n

e
)
0
7rd{ne - njv
x
ators in linear algebra. c
 151

(b) Show that, the transmission can be written JONES CALCULUS AND IT? APPLICATION
152
<
2
1 f sin2wx\
T = -
E

2 \ 2 "sin x J O

(
(c) Show thai the transmission bandwidth (FWHM) of the whole
system is governed by that of the bands of the thickest plate,
that is,
c

1Nd(n t
-n )
e

and the free spectra) range Ac is governed by that of the bands

of the thinnest plate, that is,
c
Ac ~
s
l Figure 5.U. OfT-a s ray .

The finesse J17 of the system, defined as A;'/Act/2, is then i

F~ 2N .
( c wis and the .angenUal
lhat the phase relardat.on can be expressed
J Z of S and *
(d) Design a filter with a bandwidth of 1 A at the H,, line, using ( as
quartz as the birefringent material. Assume that n0 = 1.5416
and n = 1.5506 at X = 6563 A. Find the required thickness of
e ( 2n j rf in Ti
r _

the thickest plate. X

(
(e) Show that according to (b) the bandwidth (FWHM) is given by
c
Ac 0 886
1/2 . '

2Nd(n e
-n )
0
(

Off-axis effect. A wave plate made of uniaxial crystal with its c axis aller than ni and
(c) Show that in caseiion
when rti sm
56
. .

parallel to the plate surfaces has a phase retardation of 2w(/il. -

n )d/\ for a normal incident beam. For an off-axis beam, the light
the phase retarda can be wnueo
0
"
will see different birefringence because the refractive index for the
"

( . sin2* _cos \
extraordinary wave is dependent on the direction of the beam. In r
X
addition, the path length in the crystal plate is no longer d (see Fig.
5.
11).
(a) Let Q0, 8e be the refractive angles for the ordinary and extraor- L
dinary waves, respectively. Show that the phase retardation is
L lQ± AA for off-axis light. Show that with «eH n0 <>
given by
Xsin2f
AX i
2n eff

J1
 v
'
jj i_ j„jvi j>

15.1
154 JONES CALCULUS AND ITS APPLICATION
(e) Show thai a narrowband Lyot-Ohman titter has n lim 59 Field of view of iso-index filters. Use the result obtained in Problem
i
aperture of ilod r
. .

5 6(c)
. to study the field of view of the iso-index Lyot-Ohman filter.
r (a) Show that the passband is independent of the angle of inci-
dence.
\ A /
(b) Derive an expression for the bandwidth for off-axis light as a
where we assume n - « n
e
0
.

function of 0 and <p.

57 .

. o/c y?/wrj The transmissivily near a transmissi

.

(c) Show that for the most extreme angle of incidence, the band-
written as on peak can be
width is increased or decreased by a factor of (1 + \/2n2 ) 0
.

sin Tr/TT 7 5 .
10. Twisted nematic liquid crystals.
1 + x 2 (a) Use Chebyshev's identity (5.3-4) to simplify Eq. (5.4-7).
with (b) Derive Eq. (5.4-8).

x = A/AF/tt .

(a) Show that r = 0 5 .

at x = 0 8 and derive the ex REFERENCES
for the FWHM pression (5.3-19)
.
,

(b) Find the wavelengths for which the transmissivily vanishes. H A, West, "Exlcnding the field of view of KD"? electro-pptic modulators," Appl. Opl.
.

17. 3010 (1978).

(c) Find the peak fransmissivity of the sidelobes . 2 .
B Lyol, "Optical apparatus with wide field using interference of polarized light," C. R.
(d) Esdmate the integrated transmissivily over all the sid
.

Acad. Sci. (Paris) 197, 1593 (1933).

compare it with the are elobes and
a under the main peak. 3 . B Lyot, "Filter monochromatique polarisant et ses applications en physique solaire,"
.

58 Ami. Astrophys. 7, 3! (1944).

.

. Iso~index Lyot-Ohman filters. Consider a L

lem 5 5) made .

of iso-index crystals (i e . .,
yot-Ohman filter.(Prob-
crystals with =n
4

5
. Y

Y
. Ohman, "A new monochromator," Nature 41, 157, 291 (1938).
Ohman, "On some new birefringent filter for solar research," Ark. Astron. 2, 165
certain wavelength Ar). at . .

(1958).
(a) Derive an expression for the bandwidth (FWHM) as a function 6 .
I .
Sole, Ceskostou. Casopis pro Fysiku 3, 366 (1953); Csek. Cas. Fys. 4, 607, 699 (1954); 5,
of the plate thickness and the rate of change a of the biref 114 (1955).
gence . rin- 7 . I . Sole, "Birefringent chain filters," J. Opt. Soc. Am. 55, 621 (1965).
(b) Derive an expression for the free spectral ran 8 .
R . C. Jones, " New caJculus for the treatment of optical systems," /. Opt. Soc. Am. 31, 488
a and the plate thickness d .
ge as a function of (1941).
9 See, for example, P. Yeh, A. Yariv, and C. S. Hong, "Electromagnetic propagalion in
(c) Design an iso-index Lyot-Ohman filter using CdS at 5245 .

"

periodic stratified media. I. General theory, /. Opt. Soc. Am 61, 423 (1977).
with a bandwidth of 0 1 A and a free s A
10 A pectral range of at least 10. D A. Pinnow et al., "An electro-optic tunable filter," Appl. Phys. Lett. 34, 392 (1979).
.

11. H . A. Tarry, "Electrically tunable narrowband optical filter," Electronics Lett. 11, 471
(d) Explain why the total thickness required for a (1975).
A X1 given bandwidth
.

/2 is usually smaller than in the conventional Lyot-Ohman

filter made of quartz or
12. J .
W. Evans, /. Opt. Soc. Am. "The birefringent filter," 39, 229 (1949); "Sole birefringent
calcite filter," 48 142 (1958),
.
Show that this is true only when ,

a satisfies 13 P Yeh, "Electromagnetic propagation in birefringent layered media," J. Opt. Soc. Am.
.

69.742 (1979).
a > 14. P .
Yeh, "Transmission spectrum of a Sole filter," Opt. Comm. 29, (1979).
\ 1

L 15. P .
Yeh, "Optics of anisotropic layered media: A new 4x4 matrix algebra," Surface
where „ n0 are the refractive indices of calcue or quartz
,,
Science 96, 41-53 (1980).
. L