17.

Jones Matrices & Mueller Matrices

Jones Matrices

Rotation of coordinates - the rotation matrix

Stokes Parameters and unpolarized light

Mueller Matrices
R. Clark Jones
(1916 - 2004)

Sir George G. Stokes

(1819 - 1903) Hans Mueller
(1900 - 1965)
Jones vectors describe the polarization
state of a wave
Define the polarization state of a field as a 2D vector—
“Jones vector” —containing the two complex amplitudes:

 Ex   1  1  Ex 
E     Ex    E 
E
  y  E y E x  Ex  E y
2 2
 y
(normalized to length of unity)

1 
A few examples: 0   1 
   tan a 
0° linear (x) polarization: Ey /Ex = 0  
linear (arbitrary angle) polarization: Ey /Ex = tan a
right or left circular polarization: Ey /Ex = ±j
1
 j 
 
To model the effect of a medium on light's
polarization state, we use Jones matrices.
Since we can write a polarization state as a (Jones) vector, we use
matrices, A, to transform them from the input polarization, E0, to the
output polarization, E1.
E1  AE0  a11 a12 
A
 a21 a22 
This yields: E1x  a11E0 x  a12 E0 y
E1 y  a21 E0 x  a22 E0 y

1 0 
For example, an x-polarizer can be written: Ax   
 0 0 

1 0  E0 x   E0 x 
So: E1  A x E0    E    
0 0  0 y   0 
 Other Jones matrices
0 0 
A y-polarizer: Ay   
 0 1 

A half-wave plate: 1 0 
A HWP  
 0 1 1 0  1  1 
0 1 1   1
    
A half-wave plate rotates 45-degree- 1 0   1  1
polarization to -45-degree, and vice versa. 0 1  1  1
    

A quarter-wave plate:
1 0  1 0  1  1 
AQWP   0  j  1    j 
 0  j      
The orientation of a wave plate
matters.
0° or 90° Polarizer
Remember that a quarter-wave plate
only converts linear to circular if the
input polarization is ±45°. Wave plate
w/ axes at
If it sees, say, x polarization, 0° or 90°
nothing happens.

 1 0  1  1 
0  j  0   0 
    
AQWP

Jones matrices are an extremely useful way to keep track of all this.
A wave plate example
What does a quarter-wave plate do if the input polarization is linear
but at an arbitrary angle?

1 0   1   1 
0 j   tan a     j tan a 
    

AQWP Ein Eout

For arbitrary a, this is an elliptical polarization.

a = 30°
a = 45°
a = 60°
Jones Matrices for standard components
 Rotated Jones matrices
What about when the polarizer or wave plate responsible for A
is rotated by some angle,  ?

Rotation of a vector by an angle  means multiply by the rotation matrix:

E0 '  R   E0 and E1 '  R   E1

rotated Jones vector
of the input
rotated Jones vector
where: of the output
cos( )  sin( ) 
R     
 sin( ) cos( ) 

Rotating E1 by  and inserting the identity matrix R()-1 R(), we have:

E1 '  R   E1  R   AE0  R   A  R   R   E0
1
 
  R   AR     R   E0    R   AR    E0 '  A ' E0 '
1 1
   

A '  R   AR  
1
Thus:
Rotated Jones matrix for a polarizer
Example: apply this to an x polarizer. A '  R   A R  
1

cos( )  sin( )  1 0  cos( ) sin( ) 

Ax      0 0   sin( ) cos( ) 
 sin( ) cos( )   

cos( )  sin( )  cos( ) sin( ) 

Ax      0 
 sin( ) cos( )  0 

 cos 2 ( ) cos( )sin( ) 

Ax     
cos( )sin( ) sin ( ) 
2

1/ 2 1/ 2  1  
Ax  45     Ax      
for small
1/ 2 1/ 2   0  angles, 
To model the effect of many media on light's
polarization state, we use many Jones matrices.

The aggregate effect of multiple components or objects can be

described by the product of the Jones matrix for each one.

E0 E1
A1 A2 A3

E1  A3 A2 A1E0

The order may look counter-intuitive, but order matters!

 x
Multiplying Jones Matrices
y z
Crossed polarizers: x-pol
E0
E1  A y A x E0
E1
y-pol

 0 0  1 0   0 0 
Ay Ax        so no light leaks through.
 0 1  0 0   0 0 

rotated
Uncrossed polarizers x-pol
(slightly): E0 E1

0 0   1    0 0  y-pol
A y A x         
0 1   0   0
 Ex   0 0   Ex   0 
A y A x         E    E  So Iout ≈ 2 Iin,x
 y 
E  0  y  x
 Natural light (e.g., sunlight, light bulbs, etc.)
is unpolarized

The direction of the E vector is

randomly changing. But, it is
always perpendicular to the
propagation direction.

polarized light natural light

Light with very complex polarization
vs. position is "unpolarized."
Light that has scattered multiple times, or that has scattered
randomly, often becomes unpolarized as a result.

Here, light from the blue sky is

polarized, so when viewed
through a polarizer it looks
much darker. Light from clouds
is unpolarized, so its intensity is
reduced by only 50%.

If the polarization vs. position is unresolvable, we call this

“unpolarized.” Otherwise, we refer to this light as “locally
polarized” or “partially polarized.”
When the phases of the x- and y-polarizations
fluctuate, we say the light is "unpolarized."


Ex ( z, t )  Re E0 x exp  j  kz  t   x  t    
E ( z, t )  Re E
y 0y exp  j  kz  t   y  t  

where x(t) and y(t) are functions that vary on a time scale slower than
the period of the wave, but faster than you can measure.

The polarization state (Jones vector) is:

 1  In practice, the
E 
 0 y exp  j  t   j  t    amplitudes are also
 E0 x  y x  functions of time!


As long as the time-varying relative phase, x(t)–y(t), fluctuates, the light

will not remain in a single polarization state and hence is unpolarized.
 Stokes Parameters
To treat fully, partially, or unpolarized light, we define "Stokes
parameters."

Suppose we have four detectors, three with polarizers in front of them:

#0 detects total irradiance............................................I0 Note that these

quantities are time-
#1 detects horizontally polarized irradiance..........…...I1 averaged, so even
#2 detects +45° polarized irradiance............................I2 randomly polarized
light will give a well-
#3 detects right circularly polarized irradiance.....…….I3 defined answer.

The Stokes parameters:

S0  I0 S1  2I1 – I0 S2  2I2 – I0 S3  2I3 – I0

Degree of polarization =  S + S + S 
2 1/2
2 2 = 1 for polarized light
1 2 3 / S0 = 0 for unpolarized light
 Interpretation of the Stokes Parameters
The Stokes parameters:

S0  I0 S1  2I1 – I0 S2  2I2 – I0 S3  2I3 – I0

S0 = the total irradiance

S1 = the excess in intensity of light transmitted by a horizontal polarizer
over light transmitted by a vertical polarizer
S2 = the excess in intensity of light transmitted by a 45° polarizer over
light transmitted by a 135° polarizer
S3 = the excess in intensity of light transmitted by a RCP filter over light
transmitted by a LCP filter

What we mean when we say ‘unpolarized light’:

All of these excess quantities are zero
 Interpretation of the Stokes vector
 S0 
We can write the four Stokes parameters in vector form: S 
S   1
 S2 
 
 S3 

The Stokes vector S contain information about both the

polarized part and the unpolarized part of the wave.

S = S(1) + S(2)

unpolarized part: polarized part:

S  S 2  S 2  S 2   S 2 S 2 S 2
 0 1 2 3
  1 2 3

  S   
2 S1
S
1 0
  S 
 0   2 
 0   S3 
Mueller Matrices multiply Stokes vectors

We can define matrices that multiply them,

just as Jones matrices multiply Jones vectors.

Sin Sout
M1 M2 M3

To model the effects of more than one medium on the polarization

state, just multiply the input polarization Stokes vector by all of the
Mueller matrices:
Sout = M3 M2 M1 Sin
Stokes vectors
(and Jones
vectors for
comparison)
 Mueller Matrices
(and Jones
Matrices for
comparison)

With Stokes vectors and

Mueller matrices, we can
describe light with arbitrarily
complicated combination of
polarized and unpolarized light.