You are on page 1of 8

PROCEEDINGS OF SPIE

SPIEDigitalLibrary.org/conference-proceedings-of-spie

Depolarization

Russell Chipman

Russell A. Chipman, "Depolarization," Proc. SPIE 3754, Polarization:


Measurement, Analysis, and Remote Sensing II, (25 October 1999); doi:
10.1117/12.366327

Event: SPIE's International Symposium on Optical Science, Engineering, and


Instrumentation, 1999, Denver, CO, United States

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 02 Feb 2023 Terms of Use: https://www.spiedigitallibrary.org/terms-of-use


Invited Paper

Depolarization
Russell A. Chipman

TeraStor Corp., San Jose, CA 95131; Rchipman@terastor.com


Keywords: Polarization, depolarization, Mueller calculus

ABSTRACT
Depolarization is the reduction in degree of polarization which occurs when a light beam interacts with surfaces or optical
elements. Depolarization is a scrambling of polarization states. The depolarization of Mueller matrices is studied here by
mapping the resultant degree of depolarization for all possible polarized incident states (over the surface of the Poincare
sphere). Example Mueller matrices are constructed with a variety of different depolarization properties to demonstrate the
different depolarization degrees of freedom. These properties can then be applied to understand measurements of
depolarization. A classification is presented where depolarization is considered as occurring before diauenuation, in series
with diattenuation, or following diattenuation.

INTRODUCTION
Depolarization is the statistical variation of polarization and resultant reduction in the degree of polarization which results
from stochastic polarization interactions.[ 1 ,2] Depolarization is a phenomena with a large number of degrees of freedom.
The non-depolarizing part of a Mueller matrix can be removed from a Mueller matrix through an algorithm called the polar
decomposition of Mueller matrices leaving a depolarizing residual matrix.[3] The author has measured many depolarizing
samples [3,4,5,6,7] and here comments on the interpretation of depolarization

A completely polarized beam has a predictable distribution of electric field in the transverse plane, be it linear, elliptical, or
circular. An unpolarized beam's transverse electric field randomly varies its polarization state. Most natural beams are
partially polarized, with a distribution of polarization states. The Stokes vector, S, provides a description of the polarization
ellipse and its stochastic behavior. A related description is the Poincare sphere, three-dimensional representation of the last
three Stokes vector elements, {S1, 52, 53 }, where the surface of the unit sphere contains all possible fully polarized states.
For the Stokes vector, the degree of polarization, DoP, is defined as

DoP = /+s:2+s23

where DoP=O indicates unpolarized light and DoP=1 indicates a completely polarized state lying on the surface of the
Poincare sphere.

Polarization properties during light interactions can be represented by Mueller matrices, M, which relate the incident Stokes
vector to the transmitted Stokes vector,

In light's polarization interactions with optical elements and scattering surfaces, three forms of polarization behavior are
generally identified:[2]

1. Diattenuation, the transmitted irradiance is a function of incident polarization state (polarizers and partial polarizers),
2. Retardance, the transmitted phase is a function of incident polarization state (retarders),
3. Depolarization, the degree of polarization is reduced or scrambled.

DEPOLARIZATION
Figure 1 represents a beam of light which is uniformly polarized over an aperture. The ellipses represent the elliptical
polarization state of the beam. The polarization state, which can be represented by the Stokes vector, S[r], is uniform as a
function of position, r. If the polarization state is measured at many positions in the aperture, the standard deviation of the
orientation of the polarization and the ellipticity are zero. This uniform polarization state is the ideal for many optical beams
in most optical systems, especially systems with laser sources.

Part of the SPIE Conference on Polarization: Measurement, Analysis,


14 and Remote Sensing II. Denver, Colorado • July 1999
SPIE Vol. 3754 • 0277-786X1991$1O.OO

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 02 Feb 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
For comparison, Figure 2 shows a beam, which has been depolarized in space, a randomly polarized field. The orientation
and ellipticity vary randomly and rapidly over the aperture. When the polarization state of such a beam is measured with a
polarimeter, the Stokes vector is dependent on which area is measured. When the measurement integrates the Stokes vector
over a large enough region, the polarization state is measured to be nearly unpolarized yielding the Stokes vector for
unpolarized light, Savg 10 [1,0,0,0]. Figure 3 depicts a beam which has been only slightly depolarized.

VIEWGRAPH PROJECTOR SCREENS AS EXAMPLE DEPOLARIZERS


When a uniformly polarized beam is incident on an optical element and a randomly polarized beam exits, the optical element
is said to have "depolarized" the beam of light; the degree of polarization has been reduced. Such an element is called a
depolarizer. An example of a good depolarizer is the typical viewgraph projector screen. Illuminate the screen with
horizontal linearly polarized light. The incident beam can be well extinguished with a vertical linear polarizer placed before
the screen. Next examine the scattered light from the screen a rotating linear polarizer. The beam can no longer be
extinguished; in fact the intensity is nearly constant as the polarizer is rotated. This is consistent with the scattered light
being unpolarized light. To eliminate the possibility that the light is circularly polarized, a retarder can be rotated in front of
a polarizer. Again, the transmitted intensity is nearly constant, confirming that the scattered light is nearly unpolarized and
that the screen acts as a depolarizer.

THE IDEAL DEPOLARIZER AND PARTIAL DEPOLARIZERS.


An ideal depolarizer acts upon an incident beam of arbitrary polarization state and the exiting beam (reflected, refracted,
diffracted, or scattered, depending on the element or surface under consideration) is unpolarized. The Mueller matrix for an
ideal depolarizer is

1000
0000
IdealDepolarizer =
0000
0000
T is the intensity transmission of the device. This matrix represents an interaction where all incident polarization states have
the same transmission because the matrix elements in the 0th row, mOl=m02=m03=O. The transmitted beam is unpolarized
for all incident polarization states because all the elements in rows 1 , 2, and 3 are zero. Thus when ID operates on an
arbitrary Stokes vector, the elements, 5' 1 , S'2, and S'3, of the exiting Stokes vector are zero. Examples of ideal depolarizers
include integrating spheres, screens for viewgraph projectors, and transmission of light through optically thick milk and
similar colloids.

Mueller matrices for partial depolarizers represent interactions where the exiting light has a degree of polarization between 0
and 1 when the incident light is completely polarized. For example the Mueller matrix

1000
PartialDepolarizer =
. OaOO
OOaO
000a
represents an element which reduces the degree of polarization of any fully polarized incident beam to a.

NONUNIFORM DEPOLARIZERS
Depolarizing interactions usually depolarize different polarization states by different amounts. The next Mueller matrix
represents a matrix which depolarizes horizontal and linearly polarized light to an exiting degree of polarization of a, 45 and
135 degree linearly polarized light to an exiting degree of polarization of b, and incident circular polarized light to a degree of
polarization of c,

15

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 02 Feb 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
of PoIarition Map

OGOGOOD o
GGDGOO
GGOOGO3 0.5
Figure 1. Uniformly polarized wavefront.

1 0 25 50 75
Onelabcn
1QO 125 150 175

Figure 4. Map olt the exiting degree of


polarization as a tunction of incident polarization
state over the Poihcare sphere for an aperture half
filled with horizoflal linear polarizers, and half
filled with vertic4l linear polarizers.

\Q Degree of Polarization Map

Figure 2. A depolarized wavefront with a


randomly varying polarization state.

/////////
/////////
_.:

/////////
/////////
-Ji'
0 25 50 75
Orientation
100 125 150 175

/////////
Figure 3. A wavefront with a small amount of
depolarization.
Figure 5. Map of the degree of polarization
exiting an array of horizontal, vertical, 45,
135 degree polarizers which subtend equal
areas. Linearly polarized light is partially
depolarized (1/2) while circularly polarized
light is completely depolarized.

16

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 02 Feb 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
1000
NonunzformDepolarzzer =
OaOO
OObO
000c
For the six basis polarization states, horizontal, vertical, 45, 135 degrees, right, and left circular, the polarized part of the
exiting state is equal to the incident state, i. e. the polarized portions of the Stokes vectors are like eigenvectors of the matrix.
For other incident polarization states, the polarized portion of the exiting beam is not equal to the polarized portion of the
incident beam.

EXAMPLES OF NONUNIFORM DEPOLARIZERS


Depolarizing surfaces and optical elements commonly depolarized different incident polarization states by different amounts,
what is known as nonuniform depolarization. Does this commonly occur with optical surfaces? Yes. To understand, first
consider a piece of fabric arranged with cylindrical threads aligned in only the x and y-axes. X-polarized normally incident
light is s-polarized on the horizontal threads and p-polarized at the vertical threads. At some viewing angle, depolarization
occurs which is dependent upon its angle of incidence and orientation distributions. By symmetry, y-polarized light should
have comparable depolarization. But linearly polarized light with a 45 degree orientation is not in either an s or p
eigenpolarization and will undergo a different depolarization, probably greater. Similarly circularly polarized light may have
a still different depolarization.

Or consider a randomly rough surface that is modeled as a distribution of smooth facets randomly oriented. Here each facet
has linearly polarized eigenpolarizations, so that linearly polarized light will always be an eigenpolarization for a small set of
facets. All linearly polarized states will be depolarized approximately equally. Circularly polarized light is never an
eigenpolarization of any facets and consequently undergoes a larger depolarization than linearly polarized light.

Mapping the Exiting Degree of Polarization


The depolarizing characteristics of a partial depolarizer are well represented by mapping the degree of polarization of the
exiting light as a function for all fully polarized incident states, i. e. over the Poincare sphere. Figure 4 maps the exiting
degree of polarization for PD[l,O.5,O]. This map flattens the Poincare sphere onto a plane with right circularly polarized
incident light at the top, left circularly polarized light at the bottom, and linearly polarized light along the middle with
horizontal linearly polarized incident light at both the right and left edges.

EXAMPLES OF IDEAL DEPOLARIZERS


An ideal depolarizer is an optical element for which the exiting polarization state is unpolarized independent of the
polarization of the incident state. Besides common projection screens, integrating spheres are useful as essentially ideal
depolarizers. Transmission through optically thick samples of milk and other colloids are also highly depolarizing. Similarly
clouds are depolarizing.

A frequent occurrence in optical systems is that the degree of polarization is only slightly reduced upon interaction with
optical elements. Figure 3 shows a beam of light with small variations of the polarization state. This beam has only been
slightly depolarized. For example, a surface may be slightly rough or have fingerprints or grime, a lens may have residual
strain birefringence, or a coating may be delaminating and peeling. Many imperfections and problems cause small amounts
of depolarization.

NONDEPOLARIZING INTERACTIONS
Most optical elements and polarization elements are intended to be nondepolarizing, so that a fully polarized incident beam
exits as a fully polarized beam. The elements of the Mueller matrix for a nonpolarizing element satisfy the relation

17

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 02 Feb 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Tr[M Mn]= M2,1 =4M0.
1,j=O

The Mueller matrices for all ideal polarizers and retarders are nondepolarizing.

VISUALIZING DEPOLARIZATION, SERIAL AND PARALLEL POLARIZATION ELEMENTS


A helpful model to visualize depolarization is to picture light passing through polarization elements arranged in serial and in
parallel. Elements in serial represent an aperture with different polarization elements over different parts of the aperture so
that different parts of the beam undergo different polarization transformations and different Stokes vectors will emerge from
the various parts of the aperture. Elements in parallel are encountered sequentially, one after another.

MAPPING DEPOLARIZATION
The amount of depolarization associated with a Mueller matrix is quantified by describing the reduction in the degree of
polarization, which occurs when light interacts with the corresponding optical element. As will be shown, depolarization is a
mathematically complex phenomenon with many degrees of freedom. The depolarization characteristics for specified
Mueller matrices can than be calculated. A complete description is given by calculating the degree of polarization in the
exiting beam for all incident polarization states. The set of incident states can be parameterized as the normalized Stokes
over the Poincare sphere. The surface of the Poincare sphere can be parameterized in polarization orientation and ellipticity,

. Cos(9)Cos(Ø)
PozncareSuiface=
Szn(6)Cos(çb)
Sin(O)

EXAMPLE DEPOLARIZING MUELLER MATRICES


It is instructive to examine the depolarizing properties of some serial polarization elements. The incident light will be treated
as polychromatic incoherent light, and the results can be readily generalized to coherent light along the lines presented.

Parallel Horizontal and Vertical Polarizers


Consider an aperture half-filled with horizontal linear polarizers and half filled with vertical linear polarizers yielding a
Mueller matrix

1000
0100
HLP+VLP =
0000
0000
Figure 5 maps the degree of polarization of the exiting beam over the incident Poincare sphere flattened out onto a plane.
This system transmits incident horizontally linearly polarized light as horizontal with DoP=1, i. e. horizontal is an eigenstate.
Similarly, vertical linearly polarized light is transmitted as vertical, another eigenstate. All other polarized incident states are
transmitted in a combination of horizontal and vertically polarized light. When equal amounts of horizontal and vertical
polarized light are present, the exiting beam is unpolarized. This occurs for the great circle of incident states on the Poincare
sphere which run through 45, R, 135, and L. All incident states have an equal transmission of as this system has no
diattenuation.

Such a form of depolarization likely occurs from structures arranged like fabric with horizontal and vertical scattering
cylinders. The incident light is only aligned with the s and p planes of the cylinders when the orientation is horizontal or

18

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 02 Feb 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
vertical. Light at 45 degrees and 135 degrees is an equal mixture of these two eigenstates and will be more highly
depolarized.

Four Parallel Linear Polarizers


Next consider an aperture equally divided between linear polarizers with four orientations separated by 45 degrees, for
example, horizontal, 45, vertical, and 135 degrees. The corresponding Mueller matrix is

2000
0100
HLP +VLP+ LP[45]+ LP[135] =
0010
0000
This Mueller matrix depolarizes all incident linearly polarized states equally to a degree of polarization of DoP=l/2. Right
and left circularly polarized light are completely depolarized. Elliptical polarized light is depolarized by the amount of the
degree of circular polarization.

It is instructive that this Mueller matrix is the same as is obtained by averaging over linear polarizers at all orientations
equally distributed. Thus mathematically, once we reach four linear polarizers, there is no further change in using six or
more symmetrically arrayed linear polarizer matrices. This matrix has no diattenuation and all incident states are transmitted
with equal intensity.

Such depolarization behavior is seen from many samples. Measurements of paint Mueller matrices have consistently shown
more depolarization of circularly polarized light than linearly polarized light.

Six Parallel Polarizers


An isotropic depolarizer is obtained by adding two circular polarizers to the previous example, so that horizontal, vertical, 45,
135 degree, left, and right polarizers cover six equal areas

3000
HLP+VLP+LP[45]+LP[135]+RCP+LCP =
0100
0010
0001
This depolarizer transmits all incident polarizations with DoP=l/3. Similarly, any six polarizers located at the vertices of an
octahedron on the Poincare sphere yield this isotropic depolarizer, as well as larger numbers of symmetrically located
polarizers. In fact if the general Mueller matrix for elliptical polarizers is integrated over the Poincare sphere, this matrix
results. To bring the average depolarization below 1/3 requires a more complex configuration than serial depolarizers.

An Ideal Depolarizer from Serial and Parallel Polarizers


Several simple arrangements yield an ideal depolarizer, the basic idea being to convert the output from each serial element
into two orthogonal polarizations. Consider the serial horizontal and vertical polarizer example. If a half wave linear retarder
oriented at 45 degrees is placed behind one half of each polarizer, an ideal depolarizer is obtained. Similarly but less
efficiently, a 45 degree and 135 degree polarizer can be placed behind each polarizer for an all polarizer design. All nature of
partial depolarizers can now be formed by serially combining polarizers and depolarizers.

POLARIZANCE AND DIATTENUATION


Depolarization can cause the polarizance vector (first column,{M1O, M20, M30)) and diattenuation vector (first row {MO1,
M02, M03}) of a Mueller matrix to differ. Placing a polarizer (or diattenuator) after an ideal depolarizer yields an element
which transmits all polarization states equally due to the depolarizer, but which transmits a polarized beam. For example an
ideal depolarizer followed by a horizontal linear polarizer has the Mueller matrix

19

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 02 Feb 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
1000
1000
HLP.ID=
0000
0000
Since only one polarization state is transmitted, the element is a polarizer. It is not an analyzer since all incident states are
transmitted equally.

Conversely, a polarizer followed by an ideal depolarizer, acts as an analyzer but not a polarizer, since unpolarized light is
always transmitted, such as in the following example matrix,

1 100
0000
ID.HLP=
0000
0000
Thus, differences between the polarizance vector and diattenuation vector can be due to diattenuation occurring before or
after the depolarization, for example in scattering from a rough surface or paint.

The polarizance vector represents the average polarization state transmitted by a Mueller matrix in the following sense. The
lower right 3x3 submatrix of the Mueller matrix transforms the surface of the Poincare sphere into an ellipsoid centered at the
polarizance vector. All transmitted polarization states lie within this ellipsoid.

CONCLUSION
The scrambling of polarized light into partially polarized light occurs in a myriad of ways. Simple sequences of Mueller
matrices, especially matrices in parallel, provide insight into the degrees of freedom associated with depolarization
phenomena.

ACKNOWLEDGEMENTS
Shih-Yau Lu shared much insight into the structure of the Mueller calculus. Larry Pezzaniti provided extensive Mueller
matrix measurements of depolarizing samples.

REFERENCES
1. Fundamentals of Polarized Light, A Statistical Optics Approach, C. Brosseau, section 4.1 .5, John Wiley & Sons,
New York, (1998).
2. Polarimetry, R.A. Chipman, in "Handbook of Optics," M. Bass, ed. Optical Society ofAmerica, Washington (2nd
Edition, 1995).
3. An interpretation of Mueller matrices based upon the polar decomposition, S.Y. Lu, R.A. Chipman, Journal of the
OpticalSociety ofAmerica A , vol. 13, no. 5, pp.1- 8 (May 1, 1995).
4. Mueller matrix imaging polarimetry, J.L. Pezzaniti and R.A. Chipman, Optical Engineering, vol. 34, no. 6, pp. 1558
(June 1, 1995).
5. Mueller matrix scatter polarimetry of a diamond turned mirror, J.L. Pezzaniti and R.A. Chipman, Optical
Engineering, vol. 34, no. 6, pp. 1593-1598 (June 1, 1995).
6. Depolarization measurements of an integrating sphere, S.C. McClain, C.L. Bartlett, J.L. Pezzaniti, and R.A.
Chipman, Applied Optics, vol. 34, no. 1, pp. 152-154. (January 1, 1995).
7. Depolarization in liquid crystal TV's, J.L. Pezzaniti, S.C. McClain, R.A. Chipman, and S.Y. Lu, Optics Letters, vol.
18, pp. 207 1-2073 (Dec. 1993).

20

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 02 Feb 2023


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

You might also like