Jones Framework

Derivation of the calibrated sample Jones matrix

For an ideal linear polarizer at angle θ\theta, define the Jones vector

e(θ)=(cos⁡θsin⁡θ).\mathbf e(\theta)= \begin{pmatrix} \cos\theta \\ \sin\theta \end{pmatrix}.

For a measured 2×22\times2 transfer matrix

T=(t00t01t10t11),\mathbf T= \begin{pmatrix} t_{00} & t_{01} \\ t_{10} & t_{11} \end{pmatrix},

the complex field measured with the analyzer at α\alpha and the input polarizer at β\beta is

mα,β=eT(α) T e(β).m_{\alpha,\beta} = \mathbf e^{\mathsf T}(\alpha)\, \mathbf T\, \mathbf e(\beta).

The four measured configurations yield

M0=m0∘,0∘=t00,M90=m90∘,90∘=t11,M45,0=m45∘,0∘=t00+t102,M45,90=m45∘,90∘=t01+t112.\begin{aligned} M_0 &=m_{0^\circ,0^\circ}=t_{00},\\ M_{90} &=m_{90^\circ,90^\circ}=t_{11},\\ M_{45,0} &=m_{45^\circ,0^\circ} =\frac{t_{00}+t_{10}}{\sqrt 2},\\ M_{45,90} &=m_{45^\circ,90^\circ} =\frac{t_{01}+t_{11}}{\sqrt 2}. \end{aligned}

Therefore, the two unmeasured matrix elements are

t10=2 M45,0−M0,t01=2 M45,90−M90.t_{10}=\sqrt 2\,M_{45,0}-M_0, \qquad t_{01}=\sqrt 2\,M_{45,90}-M_{90}.

The complete transfer matrix measured with the sample installed is therefore

Tsam=(M02 M45,90−M902 M45,0−M0M90).\mathbf T_{\mathrm{sam}}= \begin{pmatrix} M_0 & \sqrt 2\,M_{45,90}-M_{90}\\ \sqrt 2\,M_{45,0}-M_0 & M_{90} \end{pmatrix}.

The reference measurements give the analogous matrix

Tref=(R02 R45,90−R902 R45,0−R0R90).\mathbf T_{\mathrm{ref}}= \begin{pmatrix} R_0 & \sqrt 2\,R_{45,90}-R_{90}\\ \sqrt 2\,R_{45,0}-R_0 & R_{90} \end{pmatrix}.

In the convention used here, the reference (system) response acts on the left of the sample Jones matrix:

Tsam=Tref Jsam.\mathbf T_{\mathrm{sam}} = \mathbf T_{\mathrm{ref}}\, \mathbf J_{\mathrm{sam}}.

Provided that

det⁡ ⁣(Tref)≠0,\det\!\left(\mathbf T_{\mathrm{ref}}\right)\neq0,

solving the matrix equation from the left yields

Jsam=Tref−1Tsam.\boxed{ \mathbf J_{\mathrm{sam}} = \mathbf T_{\mathrm{ref}}^{-1} \mathbf T_{\mathrm{sam}} }.

Numerically, the notebook evaluates this as

LinearSolve[Treference, Tsample]\boxed{ \texttt{LinearSolve[Treference, Tsample]} }

rather than explicitly forming the inverse. This is both clearer and numerically more stable. The order is essential because Jones matrices generally do not commute:

Tref−1Tsam≠TsamTref−1.\mathbf T_{\mathrm{ref}}^{-1}\mathbf T_{\mathrm{sam}} \neq \mathbf T_{\mathrm{sam}}\mathbf T_{\mathrm{ref}}^{-1}.

Let's load the Jones matrix component:

PacletDirectoryLoad[NotebookDirectory[]]; <<JerryI`THzTDSTools`; <<JerryI`THzTDSTools`Jones`;

Linear Birefringence

Orienting an unknown tetragonal sample with strong linear birefringence

Load the test data measured in four different polarizer–analyzer configurations:

testData = NotebookRead[NotebookStore["bootes-1e7"]]; {listSam, listRef} = testData; listSam // Shallow listRef // Shallow {{Times[1/4, Pi], 0} -> QuantityArray[...], {Times[1/2, Pi], Times[1/2, Pi]} -> QuantityArray[...], {Times[1/4, Pi], Times[1/2, Pi]} -> QuantityArray[...], {0, 0} -> QuantityArray[...]} {{Times[1/4, Pi], 0} -> QuantityArray[...], {Times[1/4, Pi], Times[1/2, Pi]} -> QuantityArray[...], {Times[1/2, Pi], Times[1/2, Pi]} -> QuantityArray[...], {0, 0} -> QuantityArray[...]}

The JonesMatrix constructor accepts two lists containing the sample and reference data sets:

d = JonesMatrix[listSam, listRef] (*VB[*)BoxForm`temporalStorage$436481(*,*)(*"1:eJxdlMeO42YQhAcOgOGnsPe0OzTAKIoy4IMYxZyD/oUPzKQYfmaKeju/mWd89KVRXahTofv7PYFO8ePb29v868cQsnqBU1Dnu/vPT29vCuzzWY+XqX5+//oesN/fv/329f2P929f37/gf+bKgbdBZ0Hsbj+PAjjCEDau3Na11saFVT6p7nqa96RihBvBPhLCuvopn3eJTSc7wSHu+TRJ/fyan52x6O0o01K9jhDW2/kSont8as7YKbV4G20uehQ70YsV0hs9pgodYGBGGHt+ust5zMENkZ1MbEzMiQwwSJUTCZBGaGlBZOG+MeMeg56/wxbP2FDiWqgjN1m0y7rkHsd+o1xL01fxcRWsXZLt2XFFgsX8YYXoLnpCp43L+UUXgVYULxY3mQJdKXvsND+xmDr3Gu6AnL4059q6Qxk9r8i2nhHa2Gqn5SIzECOUTAvF7hntTLdDjd+JAbaYiaGKJbBwp+6SE0nRVV1t5l7bLiTxWjo1Lnu9848wGBok2phgvCJthqVVRV0lH49XVhti31L6Lnkp6g5loN+nnBrXhLRkrBxoUgdeGetJ3/dCYk8G1m0oq5lWMGpdqbmT6oK2Bgnpn+3pow2ibvnIYd3y7MbAC+7GiwL2eOk5AB6vUcSNdlgq3aDt7aYZV+7lbUYtpXqrOqkjPsabyZLryoZ7lm06fcRUrrnolFI9Tno3nBYY5mJxewBuVMwC957Vq9pyczBr4qsyplA1RcBoHn3zHysud13cEBft8FYyNDLLEZ3z9Q4/LoUwsFjSNK15Jue6spIk66dNIYBwRDifNVEXHswelsbUZHRWPByTf67l1HatI7vxJr1sFqKpv6OXyvN6q5jJKqxx3Wsj0jOP+socx4kc1Qk2CRr4OtoMh6gfC7QokxNvys5C7VhhCcNSvbJAhun4lDHPIw8X0k90Qx0/4Jf1rtJqKWlHb05q3RLiUA4UVc0bIUuUuV/S0GsUwgMaZoCRx9zBwOBS9AqP0v6l21k+Ix8KS4KFEhbPJoxn7l4Yxsbl9rEqeZDITRTPyf6kj8jX5Fo7XVZrzM2nfnIuTzTAQgKNML25NqAJfe5VXlPsasi9cNFuL6XEq5IzyfCjQNriC/bgtfxiourY8XDxw4WtMZO0QSZij7gCnT+xnjvHHF3Wma4EgBQDqurkgtXuCGI+6iw2yzMIuBM5NUoR9yUVEegS0XlFmnCNMn4imQpmmZp7U/7szfnIHoXk8gnWJxaXveLh4/s7EgQekSiVhMi8rhryYDfARg/Ap00WW2lZANeAKBLfWxwlt1NZqpmk0Ssiwz0IdB3pRr2ymc0TIuqWXZT0ry+ffPo7YN+//V388Em2D5K9OWubu798ijzOzL49/nO9ac3/l/nEoJu3ebrESZvPP3+sYtzO+b+45KzI"*)(*]VB*)

Now, d contains a series of Jones matrices—one for each measured frequency—rather than a single matrix. We can extract the matrix at a specific frequency:

d[Quantity[1.25, "Terahertz"]] // MatrixForm (*GB[*){{0.5963015865631917` -0.05208772491777156` I(*|*),(*|*)0.09143847094395247` -0.03082360560322146` I}(*||*),(*||*){-0.06687365647459285`+0.09671165603605036` I(*|*),(*|*)-0.203333895979706`-0.7882630234360012` I}}(*||*)(*1:eJxTTMoPSmNkYGAo5gUSYZmp5S6pyflFiSX5RcFsQBHfxJKizAoAs04KOA==*)(*]GB*)

Alternatively, we can use the derived Jones vectors to plot the polarization eigenstates of light inside the sample directly:

Graphics[{ PolarizationEllipse[d[Quantity[1.25, "Terahertz"]]] }, PlotRange->1.2, ImagePadding->1, ImageSize->250, AxesOrigin->{0,0.001}, Axes->True, Ticks->{{}, {}} ] (*VB[*)(Graphics[{{{RGBColor[0.9335296737704388, 0.30680945503457624, 0.30680945503457624], Rotate[Circle[{0, 0}, {0.9944172044476516, 0.10551977775998972}], 0.015392970363389945, {0, 0}]}, {RGBColor[0.4302543466037613, 0.6201695644025075, 1.], Rotate[Circle[{0, 0}, {0.997036518384493, 0.07692971472538067}], 1.613074266747015, {0, 0}]}}}, PlotRange -> 1.2, ImagePadding -> 1, ImageSize -> 250, AxesOrigin -> {0, 0.001}, Axes -> True, Ticks -> {{}, {}}])(*,*)(*"1:eJxTTMoPSmNkYGAoZgESHvk5KWnsIB4HkHAvSizIyEwuRsj7ZBaXpDHh4THD9Aa5Oznn5+QXFQkJhs2vvP/WvmhWxz35I4suIxgQxWwgxfkliSWpEKNAfOfMouScVFSjM4E0A5hAFS/yuVbK7nLpvX3RlqJ+qQjm3fZFbBUpTZFt/fbEGYDFzXwML9o8O2/bF61fteh67o3H9kUMYPCBWm6ervhGYedzoJvfc18rWLVxs31Rpmb0N7XzP4lwc1BpTmoxJ5ARkJNfEpSYl55aZAwGn+3RFPEAGZ65iempAYkpKZl56ZmM2A0CqwnOrErN/IWpgAvIcKxILfYvykzPzMPivKI/Kz9e8k0KQLedBaoRzAgpKk1Fk2cFCWcmZxejRQcDbh4ATwGmkA=="*)(*]VB*)

In this particular case, the sample was mounted at a small angle relative to the laboratory coordinate system. Because the mirrors rotate the polarization, the eigenvectors also exhibit slight ellipticity. We can inspect these vectors directly:

OpticalAxisAnalysis[d[Quantity[1.25, "Terahertz"]]]; %[["Eigenpolarizations", All, {"JonesVector", "AzimuthDeg"}]] // TableView (*VB[*)(FrontEndRef["F19f09a28-0a21-4427-9484-a20231c0437c"])(*,*)(*"1:eJxTTMoPSmNkYGAoZgESHvk5KRCeEJBwK8rPK3HNS3GtSE0uLUlMykkNVgUJG1qmGVgmGlnoGiQaGeqamBiZ61qaWJjoJhoZGBkbJhuYGJsnAwCFpRS5"*)(*]VB*)

We can also average the Jones matrix over a frequency interval, which may provide a better estimate:

OpticalAxisAnalysis[d, Quantity[Interval[{1.15, 1.5}], "Terahertz"]]; %[["Eigenpolarizations", All, {"JonesVector", "AzimuthDeg"}]] // TableView (*VB[*)(FrontEndRef["F6998f186-9499-42a3-8a93-a2277308d6ce"])(*,*)(*"1:eJxTTMoPSmNkYGAoZgESHvk5KRCeEJBwK8rPK3HNS3GtSE0uLUlMykkNVgUJm1laWqQZWpjpWppYWuqaGCUa61okWhrrJhoZmZsbG1ikmCWnAgCF/xUd"*)(*]VB*)

Extracting Material Parameters

Assuming that the ellipticity is caused by optical misalignment, we can neglect it and use linearized eigenvectors:

basis = OpticalAxisAnalysis[d, Quantity[Interval[{1.15, 1.5}], "Terahertz"]]["LinearizedEigenvectors"] {{0.9997855134235536`,0.020710556448375564`},{-0.040682908928530645`,0.9991721077577741`}}

When expressed in basis, our 2×22\times 2 transmission matrix becomes nearly diagonal. Repeating this process at every frequency allows us to extract the sample transmission for each optical eigenmode:

t1122 = OpticalAxesTransmission[d, basis]; FrequencySpectrumTrim /@ %; ListLinePlot[%[[All, All, {1, 2}]], PlotRange -> {{0.25, 1.55}, {0, 1}}, Frame -> True, FrameLabel -> {"\[Omega] (THz)", "t(\[Omega])"}, PlotLegends -> {"Polarization 1", "Polarization 2"} ] (*VB[*)(Legended[ToExpression[FrontEndRef["Fbf1786b5-d46a-4e36-9fd8-c722891df115"], InputForm], Placed[LineLegend[{Directive[PointSize[0.006944444444444445], RGBColor[0.24, 0.6, 0.8], AbsoluteThickness[2]], Directive[PointSize[0.006944444444444445], RGBColor[0.95, 0.627, 0.1425], AbsoluteThickness[2]]}, {"Pol 1", "Pol 2"}, LegendMarkers -> {{False, Automatic}, {False, Automatic}}, Joined -> {True, True}, LabelStyle -> {}, LegendLayout -> "Column"], After, Identity]])(*,*)(*"1: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"*)(*]VB*)

Here is what these two components look like in the time domain:

BandPassFilter[Quantity[0.7, "Terahertz"], 2.5] /@ FD2TD /@ t1122; ListLinePlot[%[[All, ;; 600]], PlotRange->Full, Frame->True, FrameLabel->{"\[Tau] (ps)", "t(\[Tau])"}, PlotLegends->{"Pol 1", "Pol 2"}] (*VB[*)(Legended[ToExpression[FrontEndRef["F6d2b35cb-7fbe-4295-85bc-8d94648e3baf"], InputForm], Placed[LineLegend[{Directive[PointSize[0.006944444444444445], RGBColor[0.24, 0.6, 0.8], AbsoluteThickness[2]], Directive[PointSize[0.006944444444444445], RGBColor[0.95, 0.627, 0.1425], AbsoluteThickness[2]]}, {"Pol 1", "Pol 2"}, LegendMarkers -> {{False, Automatic}, {False, Automatic}}, Joined -> {True, True}, LabelStyle -> {}, LegendLayout -> "Column"], After, Identity]])(*,*)(*"1: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"*)(*]VB*)