The planar optical mediums with properties of either birefringence
(i.e., waveplates) or anisotropic absorption (i.e., polarizers) are
well studied. However, how a beam propagates in a birefringent curved
medium with anisotropic absorption, especially for curved-sheet
polarizers, still needs to be investigated. In this paper, we study
optical wave propagation through a curved-birefringent medium with
anisotropic absorption. We built an optical model based on the Mueller
matrix to predict the spatial distributions in light intensity and
polarization when light propagates in a curved-birefringent medium
with anisotropic absorption. To demonstrate how to use the optical
model, the experiments based on ellipsometry are also performed. The
impact of this study is to analyze the light propagation in
birefringent-curved medium with anisotropic absorption, which could
affect the performance of curved liquid crystal devices with curved
polarizers, such as curved liquid crystal displays (LCDs), flexible
LCDs, and flexible LC lenses.
1. Introduction
Crystal optics dealing with optical wave propagations in birefringent and
anisotropic media have been studied theoretically and experimentally
[1]. Usually, researchers discuss a
wave propagation in optical media with properties of anisotropic
absorptions, such as polarizers, in a separate way from a wave propagation
in birefringent optical media. However, anisotropic birefringence as well
as anisotropic absorption for an optical element is unavoidable in many
engineering applications. For example, a polaroid polarizer (sheet
polarizer) with anisotropic absorption encounters
mechanically-induced-birefringence when a sheet polarizer is attached on a
flat or curved substrate under mechanical stresses at high temperature.
Typically, a sheet polarizer consists of several layers: a protective
polymeric film, two tri-acetate cellulose films (so-called TAC films), a
layer of PVA (Polyvinyl alcohol), a layer of adhesive glue, and a release
polymeric film. Usually, the protective polymeric film and the release
polymeric film are removed from the polarizer in order to attach the
polarizer on a substrate, such as a flexible substrate or a glass
substrate. During attachment, mechanically-induced-birefringence of the
polarizer could affect optical performance in many applications, such as
curved liquid crystal displays (LCDs), foldable LCDs, LCDs for vehicle
applications, augmented reality and ophthalmic applications [2–3]. As a result, how to predict light propagation in such a
birefringent-curved medium with an anisotropic absorption, such as spatial
distributions in light intensity and polarization, is necessary. Once we
understand the wave propagation in a birefringent-curved medium with
anisotropic absorption, we are able to improve or compensate the optical
performance of optical systems. In literatures, regarding to curved
birefringent media, researchers proposed or demonstrated ray-tracing
method for uniaxial optical surfaces [4], and explored different methods of measuring phase retardation
and birefringence with bending stress [5–8]. As to curved medium
with anisotropic absorption, researchers proposed a method of coupled-wave
analysis of a curved wire-grid polarizer [9], but the change of polarization due to bending stress was not
considered. To date, studies have yet to report a model describing a
birefringent curved medium with anisotropic absorption, especially for
sheet polarizers. The properties of output light from curved sheet
polarizers, such as polarization state, phase retardation, and intensity,
should be considered. In this paper, we investigate wave propagation in a
birefringent-curved medium with anisotropic absorption. Based on calculus
of Muller matrix (or so-called Stokes calculus method), the optical model
for a birefringent-curved medium with anisotropic absorption is built to
predict the spatial distributions in light intensity and polarization of
an optical element. The impact of this study is to pave a way to analyze
the light propagation in birefringent-curved medium with anisotropic
absorption, such as curved polarizers, which could affect the performance
of curved LCDs, foldable LCDs, curved contact lenses, and curved LC
devices with curved polarizers.
2. Optical model
To describe beam propagation through a birefringent medium, Stokes calculus
method is commonly exploited [1].
Assume a polarized incident plane wave or light ($\vec{E}$) propagating along z-direction is
expressed as:
(1)$$\vec{E} = \hat{x} \cdot
{E_x} \cdot {e^{ - j(k \cdot z + {\delta _x})}} + \hat{y} \cdot {E_y}
\cdot {e^{ - j(k \cdot z + {\delta _y})}},$$
where Ex and
Ey are amplitudes,
δx and
δy are phases, and k
is wave number. A Stoke vector ${\vec{S}_{in}}$ of an incident light consisting of 4
components: S0,
S1, S2, and
S3 is used to describe polarization properties
of a light beam [2–3]. S0 is total intensity (or
irradiance) of light. S1 stands for the excess
of horizontal polarization (i.e. x-direction) over
vertical polarization (i.e. y-direction),
S2 is an intensity difference of
45°-linear-polarization minus 135°-linear-polarization, and
S3 is an intensity difference of
right-circularly-polarization minus left-circularly-polarization.
According to Eq. (1),
the Stoke vector ${\vec{S}_{in}}$ of incident light could be further
expressed as [2–3]: (2)$${\vec{S}_{in}} = \left(
{\begin{array}{c} {{S_0}}\\ {{S_1}}\\ {{S_2}}\\ {{S_3}} \end{array}}
\right) = \left( {\begin{array}{c} {E_x^2 + E_y^2}\\ {E_x^2 - E_y^2}\\
{2E_x^2 \cdot E_y^2 \cdot \cos \delta }\\ {2E_x^2 \cdot E_y^2 \cdot
\sin \delta } \end{array}} \right),$$
where δ equals to
δy-δx.
When light propagates in an optical medium, the output and input Stokes
parameter (${\vec{S}_{out}}$ and ${\vec{S}_{in}}$) are connected by a Mueller matrix
(Ms), as shown in Eq. (3): (3)$${\vec{S}_{out}} = {M_s}
\cdot {\vec{S}_{in}}.$$
In order to analyze optical wave
propagation in a birefringent-curved medium with anisotropic absorption,
we could simplify the medium as a multilayered structure, as depicted in
Fig. 1(a). A
birefringent-curved medium with anisotropic absorption could be treated as
an equivalent medium consisting of an absorptive layer and a birefringent
layer. The birefringent-curved medium is attached on a curved substrate
with a refractive index of ng. The refractive
index of the absorptive layer is set as na.
The birefringent layer is a uniaxial layer with an extraordinary
refractive index ne and an ordinary refractive
index no. We also add a compensated lens for
correcting the propagation direction, which would help to validate with
experiments. The radius of curvature of the compensated lens is identical
to that of the curved substrate with a refractive index of
ng in Fig. 1(a) in order to avoid the divergence of beam which
also benefits the light collimation during measurements.
T1 to T5
represents five interfaces. The detailed beam propagation between
T2 and T3 is
illustrated in Fig. 1(b).
According to Fig. 1(a), the
Mueller matrix of the whole system M is: (4)$$M = {M_{T5}} \cdot {M_{T4 -
T5}} \cdot {M_{T4}} \cdot {M_{T3 - T4}} \cdot {M_{T3}} \cdot {M_{T2 -
T3}} \cdot {M_{T2}} \cdot {M_{T1 - T2}} \cdot {M_{T1}},$$
where MTi
(i=1∼5) represents the corresponding
Mueller matrix of an interface,
MTi-Ti+1
(i=1∼4) represents the corresponding
Mueller matrix as light propagates from Ti to
Ti+1.When light propagates normally from one isotropic medium to another
isotropic medium, a polarization of light remains unchanged. As a result,
the Mueller matrix is a unit matrix:
(5)$${M_{T1 - T2}} = {M_{T3 -
T4}} = {M_{T4 - T5}} = \left[ {\begin{array}{cccc}
1&0&0&0\\ 0&1&0&0\\ 0&0&1&0\\
0&0&0&1 \end{array}} \right].$$
The isotropic medium between
T3 and T4 surface
is air. According to the Fresnel equations, the transmittance (A) could be
estimated at a normal incidence as $A = 1 - {[({n_t} -
{n_i})/({n_t} + {n_i})]^2}$, where ni and
nt are refractive indices of medium that light
is passing out of and passing into, respectively. Since the optical media
that light is passing out of and passing into are the air and the glass
substrate,
nt = ng = 1.5,
and ni = 1. As
a result, we could simplify MT1 and
MT5 to a transmittance
AT1 = AT5
∼0.96 since only the transmittance changes but the polarization
does not change at T1 and
T5. Thus, Eq. (4) turns out: (6)$$M = {A_{T1}} \cdot {A_{T5}}
\cdot {M_{T4}} \cdot {M_{T3}} \cdot {M_{T2 - T3}} \cdot
{M_{T2}}.$$
Next, we discuss the optical propagation
between T2 an T3
as depicted in Fig. 1(b).
Here, we start from extended Jones matrix and then convert the extended
Jones matrix back to Mueller matrix. Assume the coordinates between
T2 and T3 is
(x’, y’,
z’) which may differ from the coordinate of
(x, y, z) as light
travels in air (before T1 interface). Both
y-z plane and
y’-z’ plane are the plane
of incidence. When light obliquely impinges into an interface
T2 with a wave vector of ${\vec{k}_o}$, the polarization of s-wave (i.e.
x-linearly polarized light) is parallel to the
polarization of s’-wave (i.e. x’-linearly
polarized light). p-wave and p’-wave could be expressed as a
function of incident angle in terms of unit vectors of $\hat{y}^{\prime}$ and $\hat{z}^{\prime}$ in Eq. (7). (7)$$\left\{ \begin{array}{l}
\hat{s} = \hat{s}^{\prime} = \hat{x}^{\prime}\\ \hat{p} = \cos
{\theta_i} \cdot \hat{y}^{\prime} - \sin {\theta_i} \cdot
\hat{z}^{\prime}\\ \hat{p}^{\prime} = \cos {\theta_r} \cdot
\hat{y}^{\prime} - \sin {\theta_r} \cdot \hat{z}^{\prime} \end{array}
\right.,$$
where θi
is incident angle to T2-interface, and
θr is the refraction angle of
T3-interface (Fig. 1(b)). Two unit vectors, ${\vec{c}_1}$ and ${\vec{c}_2}$, could be expressed as: (8)$${\vec{c}_1} = \sin [{\theta
_{pol}}] \cdot \cos [{\phi _{pol}}] \cdot \hat{x}^{\prime} + \sin
[{\theta _{pol}}]\sin [{\phi _{pol}}] \cdot \hat{y}^{\prime} + \cos
[{\theta _{pol}}] \cdot \hat{z}^{\prime},$$
(9)$${\vec{c}_2} = \sin [{\theta
_c}] \cdot \cos [{\phi _c}] \cdot \hat{x}^{\prime} + \sin [{\theta
_c}] \cdot \sin [{\phi _c}] \cdot \hat{y}^{\prime} + \cos [{\theta
_c}] \cdot \hat{z}^{\prime},$$
where
θpol and
ϕpol are the polar and the azimuthal
angle of ${\vec{c}_1}$. θc
and ϕc are polar and azimuthal angle of ${\vec{c}_2}$. From directions of ${\vec{k}_{o1}}$, ${\vec{k}_{e1}}$, ${\vec{k}_{o2}}$, ${\vec{k}_{e2}}$, ${\vec{c}_1}$, and ${\vec{c}_2}$, we could obtain the polarization
directions of o-wave and e-wave: (10)$$\left\{ \begin{array}{l}
{{\hat{o}}_1} = \frac{{{{\vec{k}}_{o1}} \times
{{\vec{c}}_1}}}{{|{{{\vec{k}}_{o1}} \times {{\vec{c}}_1}}
|}},{{\hat{e}}_1} = \frac{{{{\vec{o}}_1} \times
{{\vec{k}}_{e1}}}}{{|{{o_1} \times {k_{e1}}} |}}\\ {{\hat{o}}_2} =
\frac{{{{\vec{k}}_{o2}} \times {{\vec{c}}_2}}}{{|{{{\vec{k}}_{o2}}
\times {{\vec{c}}_2}} |}},{{\hat{e}}_2} = \frac{{{{\vec{o}}_2} \times
{{\vec{k}}_{e2}}}}{{|{{{\vec{o}}_2} \times {{\vec{k}}_{e2}}} |}}
\end{array} \right.,$$
where ${\hat{o}_1}$, ${\hat{e}_1}$, ${\hat{o}_2}$, and ${\hat{e}_2}$ are unit vectors standing for
polarization directions of o-wave or e-wave in either the anisotropic
absorptive layer or the birefringent layer in Fig. 1(b). In Fig. 1(b), θo and
θe stands for the refractive angles of
o-wave and e-wave when the light is propagating in an optical anisotropic
medium. The curved substrate is usually a glass substrate with refractive
index of ng ∼1.5. The anisotropic
absorptive layer is a polarizer
(no1 = ne1
∼1.5). The birefringence layer is a wave plate and we assume that
ne2 > no2
∼1.5. Since ng ∼
no1 = ne1,
we do not consider the refraction of light at T2-interface, and
that indicates
θo∼θi
(in Fig. 1(b)). Thereafter,
the incident angle to the birefringence layer is also
θo, and ${\vec{k}_{o1}} =
{\vec{k}_{o2}}\sim {\vec{k}_o}$ in Eq. (10). Assume
ne2 -
no2 <<
no2,
θe∼θo.
As a result, Eq. (10) could be expressed as: (11)$$\left\{ \begin{array}{l}
{{\hat{o}}_1} = \frac{{{{\vec{k}}_0} \times
{{\vec{c}}_1}}}{{|{{{\vec{k}}_0} \times {{\vec{c}}_1}} |}},{{\hat{\rm
e}}_1} = \frac{{{{\hat{o}}_1} \times {{\vec{k}}_0}}}{{|{{{\hat{o}}_1}
\times {{\vec{k}}_0}} |}}\\ {{\hat{o}}_2} = \frac{{{{\vec{k}}_0}
\times {{\vec{c}}_2}}}{{|{{{\vec{k}}_0} \times {{\vec{c}}_2}}
|}},{{\hat{\rm e}}_2} = \frac{{{{\hat{o}}_2} \times
{{\vec{k}}_0}}}{{|{{{\hat{o}}_2} \times {{\vec{k}}_0}} |}} \end{array}
\right..$$
The relation between linear polarizations
(Es, Ep) before
T2-interface and
(Es’,
Ep’) after
T3-interface could be written according to the
extended Jones matrix method: (12)$$\left( {\begin{array}{cc}
{{E_s}^{\prime}}\\ {{E_p}^{\prime}} \end{array}} \right) = {J_{T3}}
\cdot {J_{T2 - T3}} \cdot {J_{T2}}\left( {\begin{array}{cc} {{E_s}}\\
{{E_p}} \end{array}} \right),$$
where
JT2,
JT3, and
JT2-T3 are
extended Jones matrices for two interfaces T2
and T3, and between those two interfaces,
respectively. Although e-wave and o-wave light might split up, the
splitting angle of two waves is small (<1 degree) in fact, and the
detective area (∼1 cm) of a detector is larger than two
slightly splitting light spots. Thus, we could assume the measured light
intensity is a sum of two light intensities: ${|{{E_s}^{\prime}} |^2} +
{|{{E_p}^{\prime}} |^2}$. Under a small angle approximation, we
could ignore the reflection between the absorptive layer and the
birefringent layer. Equation (12) could be further extended as:
(13)$$\begin{array}{l} \left(
{\begin{array}{@{}cc@{}} {{E_s}^{\prime}}\\ {{E_p}^{\prime}}
\end{array}} \right) = \left[ {\begin{array}{@{}cccc@{}}
{\hat{s}^{\prime} \cdot {{\hat{e}}_2} \cdot
{t_s}^{\prime}}&{\hat{s}^{\prime} \cdot {{\hat{o}}_2} \cdot
{t_s}^{\prime}}\\ {\hat{p}^{\prime} \cdot {{\hat{e}}_2} \cdot
{t_p}^{\prime}}&{\hat{p}^{\prime} \cdot {{\hat{o}}_2} \cdot
{t_p}^{\prime}} \end{array}} \right] \cdot \left[
{\begin{array}{@{}cc@{}} {{e^{ - j \cdot {k_{ez}} \cdot d}}}&0\\
0&{{e^{ - j \cdot {k_{oz \cdot }}d}}} \end{array}} \right] \cdot
\left[ {\begin{array}{@{}cccc@{}} {{{\hat{e}}_1} \cdot
{{\hat{e}}_2}}&{{{\hat{o}}_1} \cdot {{\hat{e}}_2}}\\
{{{\hat{e}}_1} \cdot {{\hat{o}}_2}}&{{{\hat{o}}_1} \cdot
{{\hat{o}}_2}} \end{array}} \right] \cdot \left[
{\begin{array}{@{}cccc@{}} {{t_e}}&0\\ 0&{{t_o}} \end{array}}
\right]\\ \cdot \left[ {\begin{array}{@{}cccc@{}} {{{\hat{e}}_1} \cdot
\hat{s} \cdot {t_s}}&{{{\hat{e}}_1} \cdot \hat{p} \cdot {t_p}}\\
{{{\hat{o}}_1} \cdot \hat{s} \cdot {t_s}}&{{{\hat{o}}_1} \cdot
\hat{p} \cdot {t_p}} \end{array}} \right] \cdot \left(
{\begin{array}{@{}c@{}} {{E_s}}\\ {{E_p}} \end{array}} \right),
\end{array}$$
where te and
to are absorption coefficients as the
oscillation direction of the linear polarization is parallel and
perpendicular to the transmissive axis in the absorptive layer,
respectively [2–3]. koz and
kez are wave numbers of o-wave and e-wave in
the birefringent layer [1,10]. ts and
tp are transmission coefficients for s-wave
and p-wave when light propagates through T2-interface.
ts’ and
tp’ are transmission coefficients for
s’-wave and p’-wave when light propagates through
T3-interface. d is the
thickness of the birefringent layer, and λ is the
wavelength. In Eq. (13), ts,
tp, ts’,
and tp’ are listed in Eq. (14). (14)$${t_s} = \frac{{2{n_g} \cdot
\cos {\theta _i}}}{{{n_g} \cdot \cos {\theta _i} + {n_o} \cdot \cos
{\theta _o}}},$$
(15)$${t_p} = \frac{{2{n_g} \cdot
\cos {\theta _i}}}{{{n_g} \cdot \cos {\theta _o} + {n_o} \cdot \cos
{\theta _i}}},$$
(16)$${t_s}^{\prime} = \frac{{2
\cdot {n_o} \cdot \cos {\theta _i}}}{{{n_o} \cdot \cos {\theta _i} +
\cos {\theta _r}}},$$
(17)$${t_p}^{\prime} = \frac{{2
\cdot {n_e} \cdot \cos {\theta _i}}}{{{n_o} \cdot \cos {\theta _r} +
\cos {\theta _i}}},$$
In Eq. (14)-(17),
ts = tp = 1
due to
ng = no
and
θi = θo.
kez and koz in
Eq. (13) could be
solved from Eq. (18). (18)$${k_{ez}} = \frac{{v + \sqrt
{{v^2} - 4 \cdot u \cdot w} }}{{2 \cdot u}},$$
(19)$${k_{oz}} = k \cdot \cos
{\theta _i},$$
where $w = \frac{{{k_d}^2{{\sin
}^2}{\theta _c} + {k_{eb}}^2}}{{{n_e}^2}} + \frac{{{k_d}^2\sin {\theta
_c}^2}}{{{n_o}^2}} - {\left( {\frac{{2\pi }}{\lambda }}
\right)^2}$, $u = \sin {({\theta
_c})^2}/{n_e}^2 + \cos {({\theta _c})^2}/{n_o}^2$, $v = {k_d} \cdot
\sin(2{\theta _c}) \cdot ({1/{n_e}^2 - 1/{n_o}^2} )$, ${k_d} = \frac{{2\pi
}}{\lambda }\sin {\theta _i}\sin {\phi _c}$, and ${k_{eb}} = \frac{{2\pi
}}{\lambda }\sin {\theta _i}\cos {\phi _c}$.As a result, Eq. (13)
could be calculated and then expressed as a general form.
(20)$$\left[ {\begin{array}{c}
{{E_s}^{\prime}}\\ {{E_p}^{\prime}} \end{array}} \right] = \left[
{\begin{array}{cc} {{J_{11}}}&{{J_{12}}}\\
{{J_{21}}}&{{J_{22}}} \end{array}} \right] \cdot \left[
{\begin{array}{c} {{E_s}}\\ {{E_p}} \end{array}} \right].$$
By means of a conversion from extended
Jones Matrix to Muller matrix as well as a consideration of modified term ${{{\cos {\theta _r}} /
{{n_i} \cdot \cos \theta }}_i}$ due to a variation of the projected area
from oblique incidence, ${M_{T3}} \cdot {M_{T2 - T3}}
\cdot {M_{T2}}$ is [2–3]: (21)$$\begin{array}{l} {M_{T3}}
\cdot {M_{T2 - T3}} \cdot {M_{T2}}\\ = \frac{{\cos {\theta
_r}}}{{{n_i} \cdot \cos {\theta _i}}} \cdot \left(
{\begin{array}{@{}cccc@{}} 1&0&0&1\\ 1&0&0&{ -
1}\\ 0&1&1&0\\ 0&i&{ - i}&0 \end{array}}
\right) \cdot \left( {\begin{array}{@{}cccc@{}} {{J_{11}}{J_{11}}^\ast
}&{{J_{11}}{J_{12}}^\ast }&{{J_{12}}{J_{11}}^\ast
}&{{J_{12}}{J_{12}}^\ast }\\ {{J_{11}}{J_{21}}^\ast
}&{{J_{11}}{J_{22}}^\ast }&{{J_{12}}{J_{21}}^\ast
}&{{J_{12}}{J_{22}}^\ast }\\ {{J_{21}}{J_{11}}^\ast
}&{{J_{21}}{J_{12}}^\ast }&{{J_{22}}{J_{11}}^\ast
}&{{J_{22}}{J_{12}}^\ast }\\ {{J_{21}}{J_{21}}^\ast
}&{{J_{21}}{J_{22}}^\ast }&{{J_{22}}{J_{21}}^\ast
}&{{J_{22}}{J_{22}}^\ast } \end{array}} \right) \cdot
\frac{1}{2}\left( {\begin{array}{@{}cccc@{}} 1&1&0&0\\
0&0&1&{ - i}\\ 0&0&1&i\\ 1&{ -
1}&0&0 \end{array}} \right). \end{array}$$
Before we combine Eq. (21) with Eq. (6), a coordinate transformation
(i.e. a rotational matrix MR) is needed in
order to transform
x-y-z coordinate to
x’-y’-z’
coordinate. (22)$$\begin{array}{l}
M{^{\prime}_{T2 - T3}} = {M_{T3}} \cdot {M_{T2 - T3}} \cdot {M_{T2}}
\cdot {M_R}(\phi )\\ = {M_{T3}} \cdot {M_{T2 - T3}} \cdot {M_{T2}}
\cdot \left[ {\begin{array}{cccc} 1&0&0&0\\ 0&{\cos
(2\phi )}&{\sin (2\phi )}&0\\ 0&{ - \sin (2\phi
)}&{\cos (2\phi )}&0\\ 0&0&0&1 \end{array}}
\right], \end{array}$$
where $\phi$ is the rotational angle between
x-y-z coordinate and
x’-y’-z’
coordinate, and $\phi$ depends on the location of the wave and
radius of curvature of the substrate. Thereafter, Eq. (6) turns out: (23)$$M = {A_{T1}} \cdot {A_{T5}}
\cdot {M_{T4}} \cdot M{^{\prime}_{T2 - T3}}.$$
In Eq. (23),
MT4 is still unknown.
MT4 indicates the effects of
refraction and reflection from the compensated lens. We assume that the
incident angle at the T4-interface is ${\theta
_i}^{\prime}$ and refraction angle is ${\theta
_r}^{\prime}$. By considering interfacial refraction
and reflection using Fresnel’s equations [1,10], the
calculated Mueller matrix MT4 is
[10–12]: (24)$${M_{T4}} = \left[
{\begin{array}{cccc} {\frac{{{\tau_s} +
{\tau_p}}}{2}}&{\frac{{{\tau_s} - {\tau_p}}}{2}}&0&0\\
{\frac{{{\tau_s} - {\tau_p}}}{2}}&{\frac{{{\tau_s} +
{\tau_p}}}{2}}&0&0\\ 0&0&{{{({{\tau_s}{\tau_p}}
)}^{1/2}}}&0\\ 0&0&0&{{{({{\tau_s}{\tau_p}} )}^{1/2}}}
\end{array}} \right],$$
where τs
and τp are effective transmission
coefficients and could be expressed as: (25)$${\tau _s} = \left(
{\frac{{\tan {\theta_i}^{\prime}}}{{\tan {\theta_r}^{\prime}}}}
\right){\left( {\frac{{2\sin {\theta_r}^{\prime}\cos
{\theta_i}^{\prime}}}{{\sin ({\theta_i}^{\prime} +
{\theta_r}^{\prime})}}} \right)^2}.$$
(26)$${\tau _p} = \left(
{\frac{{\tan {\theta_i}^{\prime}}}{{\tan {\theta_r}^{\prime}}}}
\right){\left( {\frac{{2\sin {\theta_r}^{\prime}\cos
{\theta_i}^{\prime}}}{{\sin ({\theta_i}^{\prime} +
{\theta_r}^{\prime})\cos ({\theta_i}^{\prime} -
{\theta_r}^{\prime})}}} \right)^2}.$$
After the
T4-interface, light is collimated, we
transform the
x’-y’-z’
coordinate to the x-y-z
coordinate: $M{^{\prime}_{T4}} = M( -
\phi ) \cdot {M_{T4}}$. Equation (22) turns out: (27)$$M = {A_{T1}} \cdot {A_{T5}}
\cdot M{^{\prime}_{T4}} \cdot M{^{\prime}_{T2 - T3}}.$$
Equation (27) is able to describe the optical wave
propagation through a birefringent-curved medium with anisotropic
absorption and a compensated lens.From the model described above, we could know the wave propagation at each
layer by calculation of Muller matrix from input and output stokes vectors
as long as we know the radius of curvature of the substrate and the
position of light at the substrate. Moreover, this optical model could
help analyzing anisotropic absorption and birefringence of a
birefringent-curved media which may be fabricated by stretching or molding
under heat and pressure.
To use the model with experiments, first, the Mueller matrix of overall
system shown in Fig. 1(a)
should be obtained. Next, we need to evaluate the relationship between
te and to by
assuming MT2-T3 is
a polarizer
(MT2-T3 = Mpol)
and inversing the matrix of Eq. (27) as
(28)$$\begin{array}{l} M( - \phi
) \cdot {M_{pol}} \cdot M(\phi ) = \\ {({{A_{T5}} \cdot M( - \phi )
\cdot {M_{T4}} \cdot {M_{T3}} \cdot M(\phi )} )^{ - 1}} \cdot {M_{\exp
}} \cdot {({M( - \phi ) \cdot {M_{T2}} \cdot M(\phi ) \cdot {A_{T1}}}
)^{ - 1}}, \end{array}$$
where Mexp is
the Mueller matrix of whole system.
AT1 and
AT5 can be obtained from
Fresnel’s equations; MT2,
MT3, and
MT4 can be calculated using
Eq. (24). With
Eq. (28), we can get
relationship between te and
to as ${M_{pol}}(1,1) = ({t_e}^2 +
{t_o}^2)/2$. Next, we consider the birefringence part
and inveres the Eq. (27) again using Eq. (29). (29)$${M_{T3}} \cdot {M_{T2 -
T3}} \cdot {M_{T2}} = {({{A_{T5}} \cdot M( - \phi ) \cdot {M_{T4}}}
)^{ - 1}} \cdot {M_{\exp }} \cdot {({M(\phi ) \cdot {A_{T1}}} )^{ -
1}}.$$
Because we have birefrignce now,
MT2 and
MT3 can not be calcuated with
Eq. (24) any more.
At this point, experimental result of Eq. (29) is compared with therotical value
stated in Eq. (21)
and relationship ${M_{pol}}(1,1) = ({t_e}^2 +
{t_o}^2)/2$ with numeriacal approach. In numerical
approach, te, to, ${\phi _{pol}}$, ${\phi _{wp}}$, and phase retardation of birefrigent
layer should be obtained by fitting.3. Experimental validation of optical model
To show how to use the optical model in the experiments, we setup an
ellipsometry for measurement of Mueller matrix and stokes vectors [3–8], as depicted in Fig. 2. In experiments, we used a laser (unpolarized light,
wavelength = 632.8 nm, Newport, Model:
N-LHR-151), three sheet polarizers (P1, P2, and
P3 in Fig. 3),
two quarter-wave plates (Q1 and Q2, Newport 10RP54-1
¼ waveplate for 400-700 nm) mounted to motorized rotation
stages (Newport PR50CC) connected to a 3-axes motion controller (Newport
Model ESP301) in order to adjust the fast axes of quarter-wave plates, a
beam expander (Newport 10x Beam Expander), two optical power detectors
(Newport 918D-SL-0D3R) connected to a dual channel power meter (Newport
Model 2936-R) as illustrated in Fig. 2. The ellipsometry consists of two parts: one is
polarization state generation (PSG) and the other is polarization state
detection (PSD). By means of adjusting the angle (${\alpha _{pol1}}$) of transmissive axis of P1
with respect to x-axis and the angle (${\alpha _Q}$) of the fast axis of Q1 with
respect to x-axis, PSG could generate arbitrary
polarization state of light. As to PSD, by means of measuring of light
intensity at different angle (${\alpha _{pol2}}$) of transmissive axis of P2
with respect to x-axis and different angle ($\alpha
{^{\prime}_\textrm{Q}}$) of the fast axis of Q1 with
respect to x-axis, PSD could help to obtain Stokes
vectors of light. The optical power detector #2 in
Fig. 2 is used to monitor
the light intensity of input light. We changed the angle of fast axes of
two quarter-wave plates pairly ($\alpha
{^{\prime}_\textrm{Q}} = 5{\alpha _\textrm{Q}}$) and recorded the corresponding optical
powers. Thereafter, we could calaulate Mueller matrix of the sample.
In Fig. 2, the laser emits
unpolarized light with a stokes vector of ${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} _{i,PSG}}$. ${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} _{f,PSG}}$ represents a stokes vector after light
passes through P1 and Q1 and could be calculated as
(30)$$\begin{array}{l}
{{\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} }_{f,PSG}} = {W_{\textrm{Q}1}} \cdot {W_{P1}} \cdot
{{\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} }_{i,PSG}}\\ = \left( {\begin{array}{@{}cccc@{}}
1&0&0&0\\ 0&{{{\cos }^2}2{\alpha_Q} + \cos {\Gamma _Q}
\cdot {{\sin }^2}2{\alpha_Q}}&{(1 - \cos {\Gamma _Q}) \cdot \sin
2{\alpha_Q} \cdot \cos 2{\alpha_Q}}&{ - \sin {\Gamma _Q} \cdot
\sin 2{\alpha_Q}}\\ 0&{(1 - \cos {\Gamma _Q}) \cdot \sin
2{\alpha_Q} \cdot \cos 2{\alpha_Q}}&{{{\sin }^2}2{\alpha_Q} + \cos
{\Gamma _Q} \cdot {{\cos }^2}2{\alpha_Q}}&{\sin {\Gamma _Q} \cdot
\cos 2{\alpha_Q}}\\ 0&{\sin {\Gamma _Q} \cdot \sin
2{\alpha_Q}}&{ - \sin {\Gamma _Q} \cdot \cos
2{\alpha_Q}}&{\cos {\Gamma _Q}} \end{array}} \right)\\ \cdot
\left( {\begin{array}{@{}cccc@{}} A&{B \cdot \cos
2{\alpha_{pol1}}}&{B \cdot \sin 2{\alpha_{pol1}}}&0\\ {B \cdot
\cos 2{\alpha_{pol1}}}&{A \cdot {{\cos }^2}2{\alpha_{pol1}} + C
\cdot {{\sin }^2}2{\alpha_{pol1}}}&{(A - C) \cdot \sin
2{\alpha_{pol1}} \cdot \cos 2{\alpha_{pol1}}}&0\\ {B \cdot \sin
2{\alpha_{pol1}}}&{(A - C) \cdot \sin 2{\alpha_{pol1}} \cdot \cos
2{\alpha_{pol1}}}&{A \cdot {{\sin }^2}2{\alpha_{pol1}} + C \cdot
{{\cos }^2}2{\alpha_{pol1}}}&0\\ 0&0&0&C \end{array}}
\right) \cdot \left[ {\begin{array}{@{}c@{}} 1\\ 0\\ 0\\ 0
\end{array}} \right]. \end{array}$$
In Eq. (30), A,
B, and C are: (31)$$\left\{ \begin{array}{l} A
= (t_x^2 + t_y^2)/2\\ B = (t_x^2 - t_y^2)/2\\ C = {t_x} \cdot {t_y}
\end{array} \right.,$$
WQ1
and WP1 represent Muller matrices
of Q1 and P1, respectively. ${\Gamma _Q}$ is phase retardation of Q1.
tx and ty are
absorption coefficients when oscillation of linear polarization is
parallel to x-axis and y-axis,
respectively. We set the transmissive axis of the polarizer P1
parallel to x-axis (i.e. ${\alpha _{pol1}}\textrm{ =
0}$), and the transmissive axis of the
polarizer P2 is set to be parallel to y-axis
(i.e., ${\alpha _{pol2}}\textrm{ =
}\pi \textrm{/2}$). Assume that
tx >> ty and
tx ∼1. As a result, ${S_{f,PSG}}$ could be obtained from Eq. (30) after putting parameters: ${\alpha _{pol1}} =
0$, ${\Gamma _Q} = \pi
/2$ radians, ${t_x} \approx 1$,and ${t_y} \approx 0$. (32)$${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} _{f,PSG}} = \frac{1}{2}\left[ {\begin{array}{c} 1\\ {{{\cos
}^2}2{\alpha_Q}}\\ {\cos 2{\alpha_Q} \cdot \sin 2{\alpha_Q}}\\ {\sin
2{\alpha_Q}} \end{array}} \right].$$
According to Eq. (32), arbitrary polarization state could be
obtained by means of rotating Q1. After light passes through a
sample, a Stokes vector ${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} _{f,PSG}}$ changes to another Stokes vector ${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} _{i,PSD}}$. After keeping propagating to
Q2 and P2, a Stokes vector ${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} _{f,PSG}}$ could be calculated as: (33)$${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} _{f,PSD}} = {W_{P2}} \cdot {W_{Q2}} \cdot
{\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} _{i,PSD}} = \frac{1}{2}\left( {\begin{array}{cccc}
1&{{{\cos }^2}2\alpha_Q^{\prime}}&{\sin 2\alpha_Q^{\prime}
\cdot \cos 2\alpha_Q^{\prime}}&{ - \sin 2\alpha_Q^{\prime}}\\
1&{{{\cos }^2}2\alpha_Q^{\prime}}&{\sin 2\alpha_Q^{\prime}
\cdot \cos 2\alpha_Q^{\prime}}&{ - \sin 2\alpha_Q^{\prime}}\\
0&0&0&0\\ 0&0&0&0 \end{array}} \right) \cdot
\left[ {\begin{array}{c} {{S_0}}\\ {{S_1}}\\ {{S_2}}\\ {{S_3}}
\end{array}} \right].$$
In Eq. (27), ${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} _{i,PSD}}$ is arbitrary polarization state with
Stokes components of S0,
S1, S2,
S3. Assume the Muller matrix of a sample is
Ms. From Eqs. (32)-(33), the output Stokes vector of the
ellipsometry ${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} _f}$ can be expressed
as:(34)$$\begin{aligned}
&{{\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} }_f} = \frac{1}{4}\left( {\begin{array}{cccc} 1&{{{\cos
}^2}2\alpha {^{\prime}_\textrm{Q}}}&{\sin 2\alpha
{^{\prime}_\textrm{Q}}\cos 2\alpha {^{\prime}_\textrm{Q}}}&{ -
\sin 2\alpha {^{\prime}_\textrm{Q}}}\\ 1&{{{\cos }^2}2\alpha
{^{\prime}_\textrm{Q}}}&{\sin 2\alpha {^{\prime}_\textrm{Q}}\cos
2\alpha _Q^{\prime}}&{ - \sin 2\alpha {^{\prime}_\textrm{Q}}}\\
0&0&0&0\\ 0&0&0&0 \end{array}} \right) \cdot
{\textrm{M}_\textrm{s}} \cdot \left[ {\begin{array}{cccc} 1\\ {{{\cos
}^2}2{\alpha _Q}}\\ {\cos 2{\alpha _Q}\sin 2{\alpha _Q}}\\ {\sin
2{\alpha _Q}} \end{array}} \right]\nonumber\\ &\textrm{ =
}\frac{1}{4}\left( {\begin{array}{cccc} 1&{{{\cos }^2}2\alpha
{^{\prime}_\textrm{Q}}}&{\sin 2\alpha {^{\prime}_\textrm{Q}}\cos
2\alpha {^{\prime}_\textrm{Q}}}&{ - \sin 2\alpha
{^{\prime}_\textrm{Q}}}\\ 1&{{{\cos }^2}2\alpha
{^{\prime}_\textrm{Q}}}&{\sin 2\alpha {^{\prime}_\textrm{Q}}\cos
2\alpha {^{\prime}_\textrm{Q}}}&{ - \sin 2\alpha
{^{\prime}_\textrm{Q}}}\\ 0&0&0&0\\ 0&0&0&0
\end{array}} \right) \cdot \left( {\begin{array}{cccc}
{{M_{11}}}&{{M_{12}}}&{{M_{13}}}&{{M_{14}}}\\
{{M_{21}}}&{{M_{22}}}&{{M_{23}}}&{{M_{24}}}\\
{{M_{31}}}&{{M_{32}}}&{{M_{33}}}&{{M_{34}}}\\
{{M_{41}}}&{{M_{42}}}&{{M_{43}}}&{{M_{44}}} \end{array}}
\right)\nonumber\\ &\cdot \left[ {\begin{array}{cccc} 1\\ {{{\cos
}^2}2{\alpha _Q}}\\ {\cos 2{\alpha _Q}\sin 2{\alpha _Q}}\\ {\sin
2{\alpha _Q}} \end{array}} \right]. \end{aligned}$$
In Eq. (34), the
components of ${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} _f}$ are basically trigonometric functions
which could be expanded as Fourier series. Under change of variables, we
set: ${p_1} = 1$, ${p_2} = {\cos ^2}2{\alpha
_Q}$, ${p_3} = \cos 2{\alpha _Q}
\cdot \sin 2{\alpha _Q}$, ${p_4} = \sin 2{\alpha
_Q}$, ${h_1} = 1$, ${h_2} = {\cos ^2}2\alpha
{^{\prime}_\textrm{Q}}$, ${h_3} = \sin 2\alpha
{^{\prime}_\textrm{Q}} \cdot \cos 2\alpha
{^{\prime}_\textrm{Q}}$, ${h_4} ={-} \sin 2\alpha
{^{\prime}_\textrm{Q}}$. Eq. (34) could be expressed as:
(35)$${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}}
\over S} _f} = \frac{1}{4}\left( {\begin{array}{cccc}
{{h_1}}&{{h_2}}&{{h_3}}&{{h_4}}\\
{{h_1}}&{{h_2}}&{{h_3}}&{{h_4}}\\ 0&0&0&0\\
0&0&0&0 \end{array}} \right)\left( {\begin{array}{cccc}
{{M_{11}}}&{{M_{12}}}&{{M_{13}}}&{{M_{14}}}\\
{{M_{21}}}&{{M_{22}}}&{{M_{23}}}&{{M_{24}}}\\
{{M_{31}}}&{{M_{32}}}&{{M_{33}}}&{{M_{34}}}\\
{{M_{41}}}&{{M_{42}}}&{{M_{43}}}&{{M_{44}}} \end{array}}
\right)\left[ {\begin{array}{c} {{p_1}}\\ {{p_2}}\\ {{p_3}}\\ {{p_4}}
\end{array}} \right] = \frac{1}{4}\left[ {\begin{array}{c}
{\sum\limits_{i,j = 1}^4 {{h_i}{m_{ij}}{p_j}} }\\ {\sum\limits_{i,j =
1}^4 {{h_i}{m_{ij}}{p_j}} }\\ 0\\ 0 \end{array}} \right].$$
We then define
uij: (36)$$\sum\limits_{i,j = 1}^4
{{h_i}{p_j}{m_{ij}}} \equiv \sum\limits_{i,j = 1}^4 {{u_{ij}}{m_{ij}}}
.$$
From Eqs. (35)– (36), uij
is: (37)$$\begin{aligned}{u_{ij}}& = \frac{1}{4}\left(
\begin{array}{@{}cccc@{}}1&{{{\cos
}^2}2{\alpha_Q}}&{\sin 2{\alpha_Q} \cdot \cos
2{\alpha_Q}}\\{{{\cos }^2}2\alpha {^{\prime}_\textrm{Q}}}&{{{\cos
}^2}2{\alpha_Q} \cdot {{\cos }^2}2\alpha
{^{\prime}_\textrm{Q}}}&{\sin 2{\alpha_Q} \cdot \cos 2{\alpha_Q}
\cdot {{\cos }^2}2\alpha {^{\prime}_\textrm{Q}}}\\{\sin 2\alpha
{^{\prime}_\textrm{Q}} \cdot \cos 2\alpha
{^{\prime}_\textrm{Q}}}&{{{\cos }^2}2{\alpha_Q} \cdot \sin 2\alpha
{^{\prime}_\textrm{Q}} \cdot \cos 2\alpha
{^{\prime}_\textrm{Q}}}&{\sin 2{\alpha_Q} \cdot \cos 2{\alpha_Q}
\cdot \sin 2\alpha {^{\prime}_\textrm{Q}} \cdot \cos 2\alpha
{^{\prime}_\textrm{Q}}}\\{ - \sin 2\alpha
{^{\prime}_\textrm{Q}}}&{ - {{\cos }^2}2{\alpha_Q} \cdot \sin
2\alpha {^{\prime}_\textrm{Q}}}&{\sin 2{\alpha_Q} \cdot \cos
2{\alpha_Q} \cdot \sin
2\alpha_Q^{\prime}}\end{array}\right.\nonumber\\&\qquad\qquad\qquad\qquad\qquad\qquad\qquad\quad\left.\begin{array}{@{}cccc@{}}{\sin
2{\alpha_Q}}\\{\sin 2{\alpha_Q} \cdot {{\cos }^2}2\alpha
{^{\prime}_\textrm{Q}}}\\{\sin 2{\alpha_Q} \cdot \sin 2\alpha
{^{\prime}_\textrm{Q}} \cdot \cos 2\alpha {^{\prime}_\textrm{Q}}}\\{ -
\sin 2{\alpha_Q} \cdot \sin 2\alpha {^{\prime}_\textrm{Q}}}
\end{array}\right).\end{aligned}$$
In the ellipsometry system, we actually
measure light intensity which is a function of angles of fast axes of two
quarter-wave plates in Eq. (37). This also indicates that we could calculate the Mueller
matrix of the sample by manipulating angles of two quarter-wave plates. As
a result, we further expand each element in Eq. (37) using Fourier series [2,13–14]:
(38)$$\frac{I}{{{I_0}}} =
\frac{{{a_0}}}{4} + \frac{1}{4}\sum\limits_{i = 1}^{12} {({a_i} \cdot
\cos{\alpha _i} + {b_i} \cdot \sin{\alpha _i})} ,$$
where I is the measured
light intensity after light passes through PSD, and ${I_o}$ is maximum light intensity at the
measurement. Both of ai and
bi are coefficients of Fourier series
expansion. In Eq. (38), The relations for ${\alpha _i}$, ${\alpha _Q}$ and $\alpha
{^{\prime}_\textrm{Q}}$ are:${\alpha _1} = 2{\alpha
_Q}$, ${\alpha _2} = 4{\alpha
_Q}$, ${\alpha _3} = 2\alpha
{^{\prime}_\textrm{Q}} - 4{\alpha _Q}$, ${\alpha _4} = 2\alpha
{^{\prime}_\textrm{Q}} - 2{\alpha _Q}$, ${\alpha _5} = 2\alpha
{^{\prime}_\textrm{Q}}$, ${\alpha _6} = 2\alpha
{^{\prime}_\textrm{Q}} + 2{\alpha _Q}$, ${\alpha _7} = 2\alpha
{^{\prime}_\textrm{Q}} + 4{\alpha _Q}$, ${\alpha _8} = 4\alpha
{^{\prime}_\textrm{Q}} - 4{\alpha _Q}$, ${\alpha _9} = 4\alpha
{^{\prime}_\textrm{Q}} - 2{\alpha _Q}$, ${\alpha _{10}} = 4\alpha
{^{\prime}_\textrm{Q}}$, ${\alpha _{11}} = 4\alpha
{^{\prime}_\textrm{Q}} + 2{\alpha _Q}$, and ${\alpha _{12}} = 2\alpha
{^{\prime}_\textrm{Q}} + 4{\alpha _Q}$. When $\alpha
{^{\prime}_\textrm{Q}} = 5{\alpha _Q}$, Eq. (38) could be simplified to: (39)$$\frac{I}{{{I_0}}} =
\frac{{{a_0}}}{4} + \frac{1}{4}\sum\limits_{k = 1}^{12} {[{a_{2k}}
\cdot \cos(2 \cdot {\rm k} \cdot {\alpha _Q}) + {b_{2k}} \cdot \sin(2
\cdot {\rm k} \cdot {\alpha _Q})} ].$$
When we have the measured light intensity
I using the photodetector, we can accuquire the Fourier
coefficients a2k and
b2k through fitting. From
Eqs. (36) and (39),
the inversion of these relation gives the element of
Ms in terms of Fourier coefficients using
ai and bi : (40)$${M_s} = \left[
{\begin{array}{cccc} {({a_0} - {a_2} - {a_{10}} + {a_8} +
{a_{12}})}&{2({a_2} - {a_8} - {a_{12}})}&{2({b_2} + {{\rm
b}_8} - {b_{12}})}&{({b_1} + {{\rm b}_9} - {b_{11}})}\\
{2({a_{10}} - {a_8} - {a_{12}})}&{4({a_8} +
{a_{12}})}&{4({b_{12}} - {b_8})}&{2({b_{11}} - {b_9})}\\
{2({b_{10}} - {b_8} - {b_{12}})}&{4({b_8} +
{b_{12}})}&{4({a_8} - {a_{12}})}&{4({a_9} - {a_{11}})}\\
{({b_3} - {b_5} + {b_7})}&{ - 2({b_7} + {b_3})}&{2({a_7} -
{a_3})}&{({a_6} - {a_4})} \end{array}} \right].$$
Here, Ms is
Mexp in Eq. (28). After we obtain
Ms (or Mexp), we
could first meausred te and to by giving e-wave and
o-wave to targeted sample. Second, ${\phi _{pol}}$, ${\phi _{wp}}$, and phase retardation of birefrigent
layer can be further meaused and characterized by data fitting based on
Eqs. (38)-(40).First, we did not place any sample and then we measured optical power in a
unit of watt as a function of αQ, as shown in
Fig. 3. From
Fig. 3, we calculated
Mueller matrix of air according to Eqs. (39)-(40). The measured Mueller matrix of
air show in Eq. (41):
(41)$${\textrm{M}_{air}} = \left[
{\begin{array}{cccc} {0.99734}&{0.00002}&{\textrm{ -
0}\textrm{.00249}}&{\textrm{ - 0}\textrm{.0008}}\\ {\textrm{ -
0}\textrm{.00008}}&{\textrm{0}\textrm{.99882}}&{\textrm{0}\textrm{.005}}&{\textrm{0}\textrm{.00124}}\\
{\textrm{0}\textrm{.00061}}&{\textrm{ -
0}\textrm{.00753}}&{\textrm{0}\textrm{.99835}}&{\textrm{0}\textrm{.00114}}\\
{\textrm{ -
0}\textrm{.00092}}&{\textrm{0}\textrm{.00409}}&{\textrm{0}\textrm{.00273}}&{\textrm{0}\textrm{.99802}}
\end{array}} \right].$$
Theoretically, Mueller matrix of air should
be an unit matrix. Mair in Eq. (41) is closed to an unit matrix.
As a result, the accuracy of our measurement is around 3 decimal digits.
Before using the proposed model for measuring a birefringent optical
medium with absorption anisotropic, we verified the accuracy of the
proposed method using Muller matrix. First, we used a liquid crystal cell
with a tunable phase retardation as one of the testing sample, and the
phase retardation was pre-determined by a typical measurement method using
a pair of crossed polarizers. The difference (or said variance of
measurement using Muller matrix method) of phase retardation between the
measured data using Muller matrix method and the pre-determined data is
less than 0.0174 radians. Second, the accuracy of absorption coefficients
measurement was performed by comparing the measured data of a commercial
linear polarizer and its pre-determined spec. The variances of two
absorption coefficients tx and
ty in Eq. (31) could be less than 0.01 and 0.035,
respectively. There is high accuracy of measurement using proposed Muller
matrix on birefringence and absorption coefficient.By using the optical model, we developed and the ellipsometry we setup, we
could measure the birefringence and anisotropic absorption of a polarizer
with stress-induced birefringence. The polarizer, provided by General
Interface Solution Holding Ltd., consists of a protective film,
antireflection layer, a hard coating layer, one iodine-doped PVA layer,
two TAC layers, a layer of adhesive glue, and a release film. After we
removed protective film and release film, the polarizer was heat up to
around 90 degree Celsius and then was pressed to attach the polarizer on a
curved glass substrate with a radius of curvature of 52 mm
(R = 52 mm), as illustrated
in Fig. 4(a). The aperture
size of the curved glass substrate was 5 cm. Similarly, we prepsed
another sample with a radius of curvature of 77 mm
(R = 77 mm). The whole
attachment process was done in General Interface Solution Holding Ltd.
After high temperature process and pressure applying process, the
polarizer could have induced birefringence and the change of anisotropic
absorption. Such induced birefringence might result from the internal
structure change of the materials under stress and the change of
anisotropic absorption might result from the orientation change of local
transmissive axes of the iodine-doped PVA layer under stress [3]. In our experiments, we measured 17
points for r = 1.2 cm and
2 cm at different $\psi$ are measured to demonstrate the concept
(Fig. 4(b)). We attached a
black paper on each sample and punched 17 small holes to assist
observation and measurement. Figures 4(c)–4(d) show two samples under two attached with black paper. We also
observed a bubble defect at
r = 2 cm and $\psi$ = 270 degree for the
sample of R = 52 mm which
resulted from attachment process (Fig. 4(c)). This indicates the mechanically-induced
birefringence for the sample with small radius of curvature
(Fig. 4(b)) is larger and
also has obvious certain distribution than the one with large radius of
curvature (Fig. 4(d)). The
outer region of each samples has larger mechanically-induced birefringence
compared to the region in the center of each sample.
From Muller matrix and stokes parameters we measured at each location and
the optical model we developed, we calculated angle of transmissive of
axis, absorption coefficients (to and
te), mechanically-induced birefringence
(Δn) as a function of $\psi$ for two samples, shown in in
Figs. 5(a)-(d) and
Figs. 6(a)-(d). From
Fig. 5(a), the angle
variation is around ±1.5 degrees at
r = 2 cm and
-0.6∼+0.3 degrees at
r = 1.2 cm. The outer region
has larger variation in the angle of the transmissive axis than the inner
region. From Fig. 6(a), the
variation is similar in both of the outer and the inner region
(-0.7∼+0.4 degrees). Compared Figs. 5(a) and 6(a), the sample with large R has smaller angle
variation. This also means the transmissive axis of the sample with
R = 77 mm is almost along
x-axis after attaching into a curved substrate with a
large radius of curvature. The variation of local transmissive axes are
because of uneven stress force and the attached process on a curved
substrate. The outer region experiences larger stress force. In addition,
the small transmissive axis angle could result in light leakage in LCD
[15]. The information we obtained
about the spatial distribution of the transmissive axis of the samples
could help us to exam the process and then to adjust the mechanically
attached process.
Figures 5(b), 5(c), 6(b), and 6(c) show
absorption coefficients (te and
to) as a function of $\psi$. Once again,
te and to are
absorption coefficients as the oscillation direction of the linear
polarization is parallel and perpendicular to the transmissive axis of the
polarizer, respectively. When
R = 52 mm,
te is around 0.002 at
r = 0 cm and
0.001∼0.003 at
r = 1.2 cm and 2 cm.
to is around 0.932 at
r = 0 cm, 0.887∼0.930
at r = 1.2 cm, and
0.906∼0.926 at
r = 2 cm. When
R = 77 mm,
te is around 0.002 at
r = 0 cm, 0.002∼0.003
at r = 1.2 cm, and
0.00056∼0.0003 at
r = 2 cm.
to is around 0.937 at
r = 0 cm, 0.900∼0.926
at r = 1.2 cm, and
0.926∼0.955 at
r = 2 cm. For
R = 52 mm and
R = 77 mm, the absorption
coefficient to at
r = 1.2 cm and 2 cm is
around 2%-3% variation compared to the one at the central
part, but to could change 100%. This
also indicates the anisotropic absorption changes under stress process.
Figures 5(d) and 6(d) depict mechanically-induced
birefringence (Δn) as a function of $\psi$. Δn for both
samples at the central region are around zero. For
R = 52 mm,
Δn changes almost periodically between 0 and $2 \times {10^{ -
5}}$ at
r = 1.2 cm, and Δn
changes between 0 and $4.5 \times {10^{ -
5}}$ at
r = 2 cm. The precocity and
zero at $\psi$ = 90, 180, and 270
degrees indicate the distribution of mechanically-induced birefringence
exhibits radiative distribution. At
r = 2 cm, we observed bubbles
at $\psi$ = 270 degree due to
attachment process which also affects the measuring value of
Δn. For
R = 77 mm, Δn is
around zero at r = 0 and
1.2 cm, and it changes periodically between 0 and $2 \times {10^{ -
5}}$ at
r = 2 cm. In
Fig. 6(d),
Δn at
r = 2 cm is zero at $\psi$ = 90, 180, and 270
degrees. In Figs. 5(d) and
6(d), the absolute value of
Δn increases with r. Smaller
radius of curvature of the substrate, larger Δn.
We hope that the information provided in Figs. 5 and 6 could
help manufacturing companies to design compensation films or to adjust the
attached process in order to improve performance of the LCD. Also,
according to the results, the stress-induced birefringence is uneven due
to the three-dimension deformation; therefore, if a curved-polarizer
exists, it may experience less stress during thermal attaching.
4. Summary
We built up a model for light propagation in a birefringent-curved medium
with anisotropic absorption based on calculus of Muller matrix. The
proposed model could help us predict the transmittance, absorption,
birefringence, phase retardation, and polarization after beam propagates
through the optical medium. The method in this paper could be further
extended to multi-layered as well as curved-optical components. As long as
we know how beam propagates in the birefringent-curved medium with
anisotropic absorption, we could develop the corresponding compensation
components or devices to improve the performance in the optical systems.
We believe the study could benefit to researchers and optical engineers to
know better about optical wave propagation in a birefringent-curved medium
with anisotropic absorption and to easily manage the polarization state of
output light for further optimizing the performance of optical
systems.
Funding
General Interface Solution Holding (GIS)
Ltd; Ministry of Science and
Technology, Taiwan
(110-2112-M-A49-024).
Acknowledgments
The authors are indebted to Mr. Jun-Lin Chen for technical assistance.
Disclosures
YHL: GIS Ltd (F), YJW: GIS Ltd (F), HCL: GIS Ltd (F), MLL: GIS Ltd (F),
PLC, GIS Ltd (E).
Data availability
Data underlying the results presented in this paper are not publicly
available at this time but may be obtained from the authors upon
reasonable request.
References
1. P. Yeh and C. Gu, Optics of Liquid Crystal
Display (Wiley,
2010).
2. D. H. Goldstein, Polarized
light (CRC,
2003).
3. R. A. Chipman and W. S. T. Lam, Polarized light and optical
systems (CRC,
2019)
4. Q. T. Liang and X. D. Zheng, “Ray-tracing
calculations for uniaxial optical components with curved
surfaces,” Appl. Opt. 30(31),
4521–4525
(1991). [CrossRef]
5. T. A. El-Dessouki, N. I. Hendawy, and A. A. Zaki, “Measuring birefringence
of curved sheet and single crystals by double exposure speckle
photography,” Optics and Lasers in
Engineering 47(6),
622–628
(2009). [CrossRef]
6. G. Singh and D. S. Mehta, “Measurement of change
in refractive index in polymeric flexible substrates using wide field
interferometry and digital fringe analysis,”
Appl. Opt. 51(35),
8413–8422
(2012). [CrossRef]
7. A. Sato, T. Ishinabe, and H. Fujikake, “Evaluation of curving
characteristics of flexible liquid crystal displays fabricated using
polycarbonate substrates,” Jpn. J.
Appl. Phys. 55(1),
011701 (2016). [CrossRef]
8. S. Honda, T. Ishinabe, Y. Shibata, and H. Fujikake, “Evaluation of phase
retardation of curved thin polycarbonate substrates for wide-viewing
angle flexible liquid crystal displays,”
IEICE Trans. Electron. E100.C(11),
992–997
(2017). [CrossRef]
9. D. Kim and E. Sim, “Segmented coupled-wave
analysis of a curved wire-grid polarizer,”
J. Opt. Soc. Am. A 25(3),
558–565
(2008). [CrossRef]
10. E. Collett, “Mueller-Stokes matrix
formulation of Fresnel's equations,”
Am. J. Phys. 39(5),
517–528
(1971). [CrossRef]
11. E. Collett, “The Description of
Polarization in Classical Physics,”
American Journal of Physics 36(8),
713–725
(1968). [CrossRef]
12. E. Garcia-Caurel, A. De Martino, J. P. Gaston, and L. Yan, “Application of
spectroscopic ellipsometry and Mueller ellipsometry to optical
characterization,” Appl.
Spectrosc. 67(1),
1–21 (2013). [CrossRef]
13. R. M. A. Azzam, “Photopolarimetric
measurement of the Mueller matrix by Fourier analysis of a single
detected signal,” Opt. Lett. 2(6),
148–150
(1978). [CrossRef]
14. P. S. Hauge, “Mueller matrix
ellipsometry with imperfect compensators,”
J. Opt. Soc. Am. 68(11),
1519–1528
(1978). [CrossRef]
15. J. S. Hsu, Y. H. Lin, H. C. Lin, and K. H. Yang, “Thermally induced light
leakage in in-plane-switching liquid crystal
displays,” J. Appl. Phys. 105(3), 033503
(2009). [CrossRef]