Optics I 2017 Chapter4 PDF
Optics I 2017 Chapter4 PDF
Optics I 2017 Chapter4 PDF
Chapter 4
Polarization
Most of the properties of light we have considered so far apply to each component of the
electric field vector, and hence a scalar treatment of the wave functions has mostly been
appropriate. However, the electric field being a vector quantity implies that a precise
description of light should include its state of polarization, i.e., the orientation of the electric
field in the plane normal to the propagation direction. This chapter defines and classifies the
various possible states of polarization of a light beam. Techniques for generating and
analyzing specific polarization states and their manipulations are then described.
where ε defines the phase difference between the two components at a given position and
time. Each component represents a linearly polarized plane wave. The resultant electric field
is given by:
By eliminating the term (kz − ωt) from both field components, we obtain a
relationship between the amplitudes of the field in the x-y plane:
1
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
2
E E 2 E E
y + x − 2 x y cos ε = sin 2 ε
E0 y E0 x E0 x E0 y
This equation describes an ellipse (see Fig. 4.1) with principal axes rotated with respect to the
x-axis by an angle α that is given by the relation:
2E 0x E 0y cosε
tan 2α = 2 2
E 0x − E 0y
In a given polarization state, the electric field vector E always points to this ellipse. The
different classes of polarization states can then be defined using the parameters of the
polarization ellipse, as further described below.
Figure 4.1: The polarization ellipse. The electric field vector points to the elliptic curve,
defining the polarization state.
E x (t) E 0x e iϕ x E0 x eiϕ x
˜
E = = iϕ
E=
E y (t) E 0y e y E eiϕ y
0y
E˜ ⋅ E˜ * = 1
2
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
a
A non-normalized Jones vector E˜ = can be normalized by scalar multiplication with an
b
appropriate factor:
1 a
E˜ → E˜ '=
aa* + bb* b
ε = lπ ; l = 0,±1,±2,...
E 0y
Ey = ± E
E 0x x
The head of the electric field vector thus traces a linear trajectory in the x-y plane (see Fig.
4.2). It stays in the odd or the even quadrants depending on the relative sign of the field
components. €
Consider some examples of the Jones vectors of linearly polarized fields. The Jones
vector for linear polarization in the horizontal (h) x-direction and the vertical (v) y-direction
are given by:
1 0
E˜ h = E˜ v =
0 1
Normalized Jones vectors for light linearly polarized at 45° and -45° angle with respect to the
x-axis are written:
€
1 1 1 1
E˜ 45 = E˜ −45 =
2 1 2 −1
The orthogonality relations for these vectors are easily obtained by the scalar products:
€
3
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
E˜ h ⋅ E˜ *v = 0
and
E˜ 45 ⋅ E˜ −45
*
=0
Figure 4.2: The electric field vector for a linearly polarized plane wave.
In this case, the tip of the electric field vector exhibits a circular motion in the x-y
plane (see Fig. 4.3). We distinguish two types circular polarization, termed R- and L-states:
π
ε = − 2 + 2lπ ; l = 0,±1,±2,... R - polarized
ε = + π + 2lπ ; l = 0,±1,±2,... L - polarized
2
€
4
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
Figure 4.3: The electric field vector of a right-circular polarized plane wave.
The electric field of a right- (left-) circular polarized wave rotates clockwise
(counterclockwise) when the wave propagates towards the observer.
Note that a linear superposition of R- and L-polarized waves of identical amplitudes
yields a P-polarized wave, since
ER ( z, t ) + E L ( z, t ) =
= 2E0 ê x cos ( kz − ω t )
The normalized Jones vectors of the two types of circular polarization are written:
1 1 1 1
E˜R =
;
E˜ L =
2 −i 2 i
E˜ R ⋅ E˜ *L = 0
Using the Jones representation, it is easy to see that the linear superposition of right- and left-
circular waves yields a linearly polarized wave:
€
5
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
E˜ R + E˜ L = 2 E˜ v
1 a
E˜ =
a2 + 1 ±i
Natural Light
The superposition of independent wave trains with well-defined states of polarization (as
defined above), but finite lengths, yields randomly polarized light. In this case, the well-
defined polarization states add up with random relative phases because the different wave
trains do not have well defined mutual phase relations. This polarization state is also referred
to as natural light, and can be thought of as a non-polarized state of light.
Such random polarization state can also be represented as a superposition of two
mutually incoherent, orthogonal P-states with random relative phase εrandom (t) :
E x = E x 0 cos( kz − ωt )
E y = E y 0 cos( kz − ωt + εrandom€(t))
Partially polarized light can be obtained by superposing polarized and natural (non-polarized)
light. When polarization analysis is performed on such light (see section 4.2), it will indicate
that some of the intensity is not in a well-defined polarization state.
Note that three parameters are needed to specify the state of polarized light: the two
orthogonal components of the electric field and their relative phase ε . For the more general
state of partially polarized light, a fourth parameter describing the fraction of unpolarized
component (the admixture of polarized and non-polarized components) is also required.
€
6
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
Figure 4.4: Two ways of describing natural (non-polarized) light: (a) superposition of two
orthogonal linearly polarized fields with random mutual phase; (b) superposition of many
linearly polarized waves with random polarization direction and different amplitudes.
The concept of circular polarization is related to the angular momentum state of light.
This angular momentum can be evaluated by considering the effect of the electric field
carried by a circularly polarized wave on a charged particle.
The electric force associated with the electric field of the circularly polarized light
exerts a torque τ on a charged particle. The corresponding power of this torque is given by
dε
τω = P =
dt
where ω is the angular frequency of the light and ε is the energy. Since the angular
momentum L is related to the torque by
dL
€ τ =€
dt
we have
dLω = dε
and hence
L=
ε = ω =
ω ω
7
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
R - state : −
L - state : +
(where the rule of a right-handed screw is applied). Other polarization states are described by
a proper (quantum mechanical) superposition of photons in R- and L-states.
These mechanisms are described in some detail in the next section. We start, however, with
the description of a technique for analyzing the state of polarization of a light, formulated by
Malus’s law.
Figure 4.5: A linear polarizer generating linearly polarized light with plane of
polarization oriented at an angle θ with respect to the y-axis.
8
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
exactly along the y-axis (see Fig. 4.6.). The analyzer transmits, by definition, only electric
field components that are parallel to its transmission axis. Denoting the amplitude of the
polarized light generated by the first polarizer by E 0 , the amplitude transmitted by the
analyzer is thus E 0 cosθ . The intensity transmitted by the analyzer hence reads:
Figure 4.6: Set up for deriving Malus’s law: natural light is polarized by a linear
polarizer, and is subsequently analyzed by another linear polarizer (analyzer).
Dichroism
9
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
Birefringence
Consider the harmonic oscillator model introduced in Chapter 2 for describing the dispersion
properties of an optical medium. The frequency dependence of the refractive index n(ω ) is
determined by the oscillator resonance frequencies. Therefore, if the harmonic oscillators
display a spatial anisotropy (that is, if the corresponding spring constant is different for two
perpendicular directions in space), it follows that the refractive index will show a similar
anisotropy. This is illustrated schematically in Fig. 4.8, which shows the anisotropic
harmonic oscillator model and the corresponding frequency dependence of the refractive
index. For light linearly polarized in the x- and y-directions, two different resonance
frequencies exist, and thus for given frequency two different refractive indices will apply.
This phenomenon is referred to as birefringence. In a birefringent medium, different
refractive indices apply for two perpendicularly polarized P-states. In the particular case
when the light frequency overlaps with one of the corresponding absorption lines, the optical
medium will be dichroic. Otherwise, the birefringence will result in two different phase
velocities for the x- and y-polarizations.
Figure 4.8: Model of birefringence: anisotropic harmonic oscillator system (left) and the
resulting frequency dependence of the refractive index .
The phenomenon of birefringence can be utilized to separate the optical paths of light
that is€linearly polarized at different orientations. Birefringent crystals, such as calcite, are
characterized by an optic axis, around which the crystal shows rotational symmetry. Consider
10
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
the propagation of rays in one of the principal sections passing through such an optic axis
(see Fig. 4.9). The propagation of such ray will depend dramatically on its polarization. For
light polarized normal to the plane of the principal section, i.e. along the optic axis,
propagation proceeds like in an isotropic medium. The corresponding ordinary ray will thus
propagate straight when incident perpendicular to the crystal facet. On the other hand, rays
polarized in the plane of the principal section will have polarization components parallel and
perpendicular to the optic axis. Such an extraordinary ray will experience propagation in a
birefringent medium, with phase velocities that are different along the optic axis and
perpendicular to it. This results in a deflection of the incident ray, and thus a separation of the
two perpendicular incident polarizations (see Fig. 4.9). This process is also referred to as
double refraction.
Figure 4.9: Double refraction in calcite. The plane represents a principal section of
calcite, defined by the optic axis and the normals of two opposite planes.
Scattering
Scattered light retains the polarization of the incident beam in the forward direction, as
illustrated in Fig. 4.10. Laterally scattered waves are polarized due to the transverse nature of
the electromagnetic wave. Hence, scattering of natural light results in linearly polarized with
polarization orientation dependent on the scattering direction.
11
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
Reflection
As shown in Section 2.3, Fresnel’s equations indicate that natural light is polarized when
reflected at the Brewster angle (see also Fig. 4.11). At this angle of incidence θ i , the
reflectivity for the p- and s-waves is given by
R// = 0 p - wave
2
€
sin (θ − θ )
R⊥ = 2 i t s - wave
sin (θ i + θ t )
Ii I
Ir // = Ir // = R// i €
2Ii // 2
and similarly
€
12
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
Ii
Ir⊥ = R⊥
2
Hence,
€ I I +I 1
R = r = r // r⊥ = ( R// + R⊥ )
Ii Ii 2
4.3.1 Retarders
If a birefringent crystal like calcite is shaped into a plate such that the optic axis is parallel to
its interface, an incident plane wave will “see” a different refractive index depending on its
polarization. An ordinary wave, linearly polarized perpendicular to the optic axis, will have a
smaller phase velocity and hence a larger refractive index than that of the extraordinary
wave:
no > ne
Therefore, the optical path will have a different length for plane waves traversing the plate
with polarization parallel or perpendicular to the optic axis. The difference in this optical path
length is: €
Λ = d( n o − n e )
€ 13
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
Δφ = k 0 Λ = k 0 d ( n o − n e )
Half-Wave Plate
In this case, the thickness of the retarder is designed so that a phase shift Δφ = π is
introduced at the selected wavelength of light. Thus, we have
d ( n o − n e ) = (2m + 1) λ0 /2
€
The effect of such retarder on a linearly polarized light whose plane of polarization makes an
angle θ with the optic axis is illustrated in Fig. 4.12. The emerging wave is linearly polarized
€
at an angle -θ with respect to the optic axis. In addition, a half-wave plate will also change
right-circular light into left-circular light, and vice versa.
Quarter-Wave Plate
In this case, a phase shift of Δφ = π /2 is introduced between the two axes, and the optical
thickness difference is given by
d ( n o − n e ) = ( 4m + 1) λ0 /4
€
This phase shift can convert linearly polarized light into elliptical light, and vice versa. When
the electric field vector is at 45 degrees angle with respect to the principal axes of the quarter-
wave plate, a circularly€ polarized light will emerge (see Fig. 4.13). A quarter-wavelength
plate would also convert a circularly polarized light into linearly polarized one.
14
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
Figure 4.13: Use of quarter-wave plates (90° retarders) for transforming between linearly
and circularly polarized light.
15
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
˜˜ = M11
M
M12
M 21 M 22
Thus, a linear polarizer producing horizontal or vertical P-states is represented by the Jones
matrices
1 0 0 0
M˜ h = M˜ v =
0 0 0 1
1 1 1 ˜ = 1 1 −1
M˜ 45 = M −45
2 1 1 2 −1 1
16
E. Kapon/Optics I (2016/2017)/ Chapter 4: Polarization
4.4 Summary
2 2
Ey Ex E E
-Polarization ellipse + − 2 x y cosε = sin 2 ε
Ey 0 Ex 0 E x0 E y 0
˜ E x (t) E 0x e iϕ x
-Jones vectors E = = iϕ
E y (t) E 0y e y
1 0
-Linear polarization E˜ h = E˜ v =
0 1
1 1 1 1
-Circular polarization E˜ R = E˜ L =
€ 2 −i 2 i
a
1
-Elliptic polarization E˜ =
€a + 1 ±i
2
€
2
-Malus’s law I(θ ) = I (0) cos θ
€
-Retarder Δφ = k 0 d ( n o − n e )
-Jones matrices:
˜˜ = M11
€ M
M12
M 21 M 22
E˜ out = M˜ E˜ in
€
17