Polarimeters

58
II Escuela de Optica Biomedica, Puebla, 2011 Polarimeters Jessica C. Ramella-Roman, PhD

description

Polarimeters. Jessica C. Ramella-Roman, PhD. Stokes vector formalism. Four measurable quantities (intensities) Characterize the polarization state of light E0x, E0y, Cartesian electric field component d=dx-dy phase difference. Simple Stokes vector polarimeter. Six intensity measurements. - PowerPoint PPT Presentation

Transcript of Polarimeters

Page 1: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Polarimeters

Jessica C. Ramella-Roman, PhD

Page 2: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Stokes vector formalism

• Four measurable quantities (intensities)• Characterize the polarization state of light• E0x, E0y, Cartesian electric field component• d=dx-dy phase difference

S =

IQUV

⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥=

ExEx* +EyEy

*

ExEx* −EyEy

*

ExEy* +EyEx

*

i ExEy* −EyEx

*( )

⎢ ⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥ ⎥=

E0x2 +E0y

2

E0x2 −E0y

2

2E0xE0ycosδ2E0xE0 ysinδ

⎢ ⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥ ⎥

Page 3: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Simple Stokes vector polarimeter

• Six intensity measurements

S=

IQUV

⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥=

IH + IVIH −IVI 45 −I−45I R −IL

⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥

Page 4: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Simple Stokes vector polarimeter

• Horizontal, i.e. parallel to reference frame

S=

IQUV

⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥=

IH + IVIH −IVI 45 −I−45I R −IL

⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥

polarizer

Page 5: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Simple Stokes vector polarimeter

• Vertical, i.e. perpendicular to reference frame

S=

IQUV

⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥=

IH + IVIH −IVI 45 −I−45I R −IL

⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥

polarizer

Page 6: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Simple Stokes vector polarimeter

• Linear polarizer at +/-45o to reference frame

S=

IQUV

⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥=

IH + IVIH −IVI 45 −I−45I R −IL

⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥

polarizer

Page 7: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Simple Stokes vector polarimeter

• Circularly polarized (left and right)Quarter-wave plate

polarizer

S=

IQUV

⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥=

IH + IVIH −IVI 45 −I−45I R −IL

⎢ ⎢ ⎢ ⎢

⎥ ⎥ ⎥ ⎥

Polarizer and quarterwaveplate axis are at 45o to each other

Page 8: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

A Mueller matrix polarimeter

R1 R2PqwpP qwp

Page 9: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

16 measurementsHH -> H source H detectorHV -> H source V detectorHP -> H source P detectorHR -> H source R detector

VH -> V source H detectorVV -> V source V detectorVP -> V source P detectorVR -> V source R detector

PH -> P source H detectorPV -> P source V detectorPP -> P source P detectorPR -> P source R detector

RH -> R source H detectorRV -> R source V detectorRP -> R source P detectorRR -> R source R detector

Page 10: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

16 measurementsHH -> H source H detectorHV -> H source V detectorHP -> H source P detectorHR -> H source R detector

VH -> V source H detectorVV -> V source V detectorVP -> V source P detectorVR -> V source R detector

PH -> P source H detectorPV -> P source V detectorPP -> P source P detectorPR -> P source R detector

RH -> R source H detectorRV -> R source V detectorRP -> R source P detectorRR -> R source R detector

Handbook of opticsVol II

Page 11: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

A Mueller matrix polarimeter

R1 R2PqwpP qwp

Page 12: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

16 measurementsHH -> H source H detectorHV -> H source V detectorHP -> H source P detectorHR -> H source R detector

VH -> V source H detectorVV -> V source V detectorVP -> V source P detectorVR -> V source R detector

PH -> P source H detectorPV -> P source V detectorPP -> P source P detectorPR -> P source R detector

RH -> R source H detectorRV -> R source V detectorRP -> R source P detectorRR -> R source R detector

Page 13: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

A Mueller matrix polarimeter

R1 R2PqwpP qwp

Page 14: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

16 measurementsHH -> H source H detectorHV -> H source V detectorHP -> H source P detectorHR -> H source R detector

VH -> V source H detectorVV -> V source V detectorVP -> V source P detectorVR -> V source R detector

PH -> P source H detectorPV -> P source V detectorPP -> P source P detectorPR -> P source R detector

RH -> R source H detectorRV -> R source V detectorRP -> R source P detectorRR -> R source R detector

Page 15: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

A Mueller matrix polarimeter

R1 R2PqwpP qwp

Page 16: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Special issues in polarimetry

• Spectral stokes vector optimization

• Mueller matrix optimization

Page 17: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Motivations

• Stokes vector polarimeter can be used for• rough surface measurements• characterization of particle size (partial Stokes

vectors, co cross polarization)• Multi-spectral Stokes vector polarimeters are

costly, often we need to sacrifice spectral performance (single wavelengths)

Page 18: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Experimental Layout

LCR1- Liquid Crystal Retarder q = 0o

LCR2- Liquid Crystal Retarder q = 45o

p polarizer

Fiber – 200µm

LED – White LED or Xenon wls

p

Page 19: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Experimental Layout for Mueller M

P

We observe the spectrum between550 and 750 nm

LCR1- Liquid Crystal Retarder q = 0o

LCR2- Liquid Crystal Retarder q = 45o

p polarizer

Fiber – 200µm

LED – White LED or Xenon wls

p

WP

Page 20: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Calibration

• Method was originally proposed by Boulbry et al.* for an imaging system and 3 wavelengths.

• Calibration does not require ANY knowledge of LCR retardation or orientation

• There is a linear transformation between a set of measurements and the Stokes vector

*B. Boulbry, J.C. Ramella-Roman, T.A. Germer, Applied Optics, 46, pp. 8533–8541, 2007.

Page 21: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Theta is the orientation angle of the polarizerwith respect to the reference plane, 0 to 180o

Six spectra Ii , are acquired for each theta for differentLCR retardation

pWP achromatic ¼wave plate

Polarizer after wave plate

Page 22: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Polarizer before wave plateTheta is the orientation angle of the polarizerwith respect to the reference plane, 0 to 180o

Six spectra Ii , are acquired for each theta for differentLCR retardation

pWP achromatic ¼wave plate

Page 23: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Calibration cnt.

• The calibration polarizer and wave plate ideally create the Stokes vectors

Sbefore=MwavepM polarizerq( )Sunpol BEFORE

Safter=M polarizerq( )MwavepSunpol AFTER

M Mueller matrices S Stokes vectors

Page 24: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Calibration cnt.

• The Stokes vectors are related to the measured values Ii through the data reduction matrix W for which

• W is finally calculated using the SVD of I

SbeforeSafter[ ] =W IbeforeIafter[ ]

W = S[ ] I[ ]−1

Page 25: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Calibration cnt.

• Once W is know only 6 I measurements are necessary to build the full Stokes vector

• This is true at every wavelength.

S[ ] =W I[ ]

Page 26: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Results - Incident [1 -1 0 0]

90o

Page 27: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Results - Incident [1 0 0 1]

45o

wp

Page 28: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Is chicken a perfect wave-plate?

Angle

Wav

elen

gth Transmitted

degree of polarization

[1 1 0 0]

P

Page 29: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011OASIS 2011

Chicken muscle ~ cylinder scattering + Rayleigh scattering

DLP

Real

Simulated

DCP

Page 30: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

More on chicken and polarization on

Page 31: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

The same layout & calibration can be used to build a Mueller matrix polarimeter

45o

wp

Page 32: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Mueller matrix of air45o

wp

Page 33: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Mueller matrix of air45o

wp

Page 34: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Conclusions

• Stokes vector polarimeter is fiber based and usable between 550-750 nm

• Point measurements of small scatterers• Miniaturizing the system

Page 35: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Optimization of Mueller Matrices measurements

• The classic Mueller matrix polarimeter

• Previous work on optimizing a polarimeter

• Mueller matrix polarimetry with SVD

Page 36: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Dual rotating retarder polarimeter

R1 R2

R1 : 5 R2

R1R2

Page 37: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Pq =AqT MSq

D. B. Chenault, J.L. Pezzaniti, R.A. Chipman, “Mueller matrix algorithms,” in D. Goldstein and R. Chipman (eds.) , “Polarization analysis and measurement,” in Proc. Soc. Photo-Opt. Instrum. Eng. V. 1746, pp. 231-246 (1992)

Measured flux

Analyzing vector

Source vector

Sample Mueller matrix

Pq =AqTMSq

Calculation of Mueller Matrix

Page 38: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Pq measured flux for q source detector retarders combinationSq source vector (Stokes vectors of source polarizing elements)Aq detectors vector (Stokes vectors of detector analyzing elements)

= aq,0 aq,1 aq,2 aq,3⎡⎣ ⎤⎦

m 00 m 01 m 02 m 03

m10 m11 m12 m13

m 20 m 21 m 22 m 23

m 30 m 31 m 32 m 33

⎢⎢⎢⎢⎢

⎥⎥⎥⎥⎥

sq,0sq,1sq,2sq,3

⎢⎢⎢⎢⎢

⎥⎥⎥⎥⎥

Pq =AqT MSq

Pq =AqTMSq =WqM

Calculation of Mueller Matrix

Page 39: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Pq =AqT MSq =WqM

Flattened Mueller Matrix

Measurement matrix

Pq =AqTMSq =WqM

• The qth measurement

Calculation of Mueller Matrix

aqsq

Page 40: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

M =W −1P

For 16 measurements

• W is square with a unique inverse – (if W non singular)

M =W −1P

Page 41: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

M =(W TW )−1W TPM =WP

−1P 

For more than 16 measurements

• W is not square so to calculate the Mueller matrix

*M. Smith ‘Optimization of the dual-rotating-retarder Mueller matrix polarimeter”, Applied Optics, V. 41, No. 13 (2002)

M =W −TW( )−1W TP (*)

M =WP−1P

Page 42: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Two main issues

• Which wave-plates are best for Mueller matrix calculation

• number of measurements to calculate the Mueller matrix

Page 43: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Which retarder are best for Mueller matrix calculation*

*M. Smith ‘Optimization of the dual-rotating-retarder Mueller matrix polarimeter”, Applied Optics, V. 41, No. 13 (2002)

Change retardation of R1 and R2, (R1,R2 have same retardation)200 measurements to calculate WR1:R2, 1:5 ratio

Calculate cond( W)

Page 44: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Condition number

cond( A) = A A−1

A → max norm A

A =max1≤i≤n

aijj=1

n

conδ A( )= A A−1

A → max norm A

A =max1≤i≤n

aijj=1

n∑

1 / cond( A)how  close  is  A to a sin gular  matrix1/cond(A) how close is A to a singular matrix

Page 45: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Best retardance 127

Minima at 127 and 233

Page 46: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

n of measurements*

*M. Smith ‘Optimization of the dual-rotating-retarder Mueller matrix polarimeter”, Applied Optics, V. 41, No. 13 (2002)

Angular increments of source and detectorretarders are varied

Angular increments 0:60

Fixed retardance 127o

16 measurements30 measurements Calculate cond( W)

Page 47: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Cond(W)

• Over-determined system are better

16 measurements 30 measurements

Page 48: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Svd vs pseudo-inverse

1Air2 Linear P3 qwp

Page 49: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Modeled error

error 2 = Mi

reconstructeδ −Miiδeal

( )i=1

16∑( )2

Mi

reconstructed =W TW( )−1W TP

Mi

reconstructeδ =V S−1U TP         [U ,S,V ] =svδ(W )

Mireconstructeδ=W −TW( )

−1W TP (*)

Mireconstructeδ=VS−1U TP V ,S,U[ ] =svδW( )

error2 = Mireconstructeδ−Mi

iδeal( )

i=1

16

∑ ⎛

⎝ ⎜ ⎜

⎠ ⎟ ⎟

2

Page 50: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Modeled error

SVD

Smith

SVD gives low level error for broader range of retardances

SVD

pseudoinverse

Page 51: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Does the sample Mueller M bias results?

1Air2 Linear P3 qwp

Page 52: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Generating a sample Mueller matrix

• Generate 4 different Mueller matrices with 2,4,6,and 8 degrees of freedom (100 MM total)

• Check for physical plausibility of Mueller matrix (Handbook of Optics Vol II)

Tr( MM T ) ≤4m oo2

moo − m o12 +m o2

2 +m o32

( )2

≥ m o, j− m j,k

m o, j

m o12 +m o2

2 +m o32

k=1

3

∑⎛

⎝⎜⎜

⎠⎟⎟

j=1

3

m oo2 ≥ m o1

2 +m o22 +m o3

2 ; m oo ≥m ij

Tr MMT( )≤4m oo

2

m oo− m o12 +m o2

2 +m o32 ⎛

⎝ ⎜ ⎞ ⎠ ⎟2

≥ m o,jj=1

3

∑ − m j,km o,j

m o12 +m o2

2 +m o32

k=1

3

Page 53: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

SVD

Smith

Error with reconstructed Mueller matrices

SVD gives low level error for broader range of retardances

SVD

pseudoinverse

Page 54: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Conclusions

• Using SVD a broader range of retarders may be used in DRR polarimetry

• Several numerical programs (such as Matlab) use SVD in their pseudo-inverse algorithms

• In most cases the error due to use of SVD is minimal compared to instrumental errors

Page 55: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Tomorrow

• Monte Carlo modeling basics

• Monte Carlo with Meridian planes

Page 56: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

Condition number

cond( A) = A A−1

A → max norm A

A =max1≤i≤n

aijj=1

n

conδ A( )= A A−1

A → max norm A

A =max1≤i≤n

aijj=1

n∑

1 / cond( A)how  close  is  A to a sin gular  matrix1/cond(A) how close is A to a singular matrix

Page 57: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

SVD

Ax =bSolution this is obtained minimizing the error

Ax −b

Fundamental property of SVD

A=USV T

Ax=b

Ax=b

Ax=USV T

Page 58: Polarimeters

II Escuela de Optica Biomedica, Puebla, 2011

SVD

U and V are respectively mxm and nxn unitary matrices

and is a diagonal whose elements are the singular value of the matrix A.

is a mxn matrix, the singular values of A are written along the main diagonal often in descending order.

The columns of U are eigenvectors (left singular vectors) of AAT and the columns of V are eigenvectors of ATA (right singular vectors).

S

SS

S

S