Achromatic Retarders in Polarization Optics: Michael Kraemer Tom Baur

Achromatic retarders in polarization
optics
Michael Kraemer
Tom Baur
Michael Kraemer, Tom Baur, “Achromatic retarders in polarization optics,” Opt. Eng. 58(8), 082406 (2019),
doi: 10.1117/1.OE.58.8.082406.
Downloaded From: https://www.spiedigitallibrary.org/journals/Optical-Engineering on 21 Apr 2019

Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Optical Engineering 58(8), 082406 (August 2019)
Achromatic retarders in polarization optics

Michael Kraemer* and Tom Baur
Meadowlark Optics Inc., Frederick, Colorado, United States
Abstract. Retarders or waveplates are tools for polarization modification in bulk optical systems. These devices
usually have a strong wavelength dependence in their performance, making them suitable for use over a wave-
length band on the order of a few percent of the center wavelength for which they are made. Display and tunable
laser applications are examples that can require consistent polarization modification over a much broader wave-
length range. We discuss new methods and designs for dramatically increasing range of performance and
review older methods as well. We show examples of achievable performance using modern polymer and liquid
crystal materials. © 2019 Society of Photo-Optical Instrumentation Engineers (SPIE) [DOI: 10.1117/1.OE.58.8.082406]
Keywords: ferroelectric liquid crystals; achromatic; retarder; waveplate; liquid crystal; polarization; superachromat.
Paper 181372SS received Sep. 24, 2018; accepted for publication Dec. 26, 2018; published online Jan. 24, 2019.
1 Introduction 4. Requirements for size of the device.

Waveplates or retarders are polarization state modifiers that 5. Requirements for clear aperture of the device.
perform well over only a small range of wavelengths in their 6. Requirements for angular aperture of the device (i.e.,
most common forms. This is because the retardance in waves retardance variation with angle of incidence and
decreases mostly because of the 1∕λ dependence. A wave- angle of azimuth).
length dependence of birefringence also contributes, but
7. Requirement whether the device must be a pure
to a much smaller extent.
retarder, and not a retarder-rotator. (It has been
Polarization modification in a constant manner over
shown by Jones and Hurwitz3 that any combination
wavelength is often desired in systems such as tunable lasers,
of waveplates is equivalent to a retardation and a
imaging systems, and polarimeters. The most common
rotation.)
polarization modifiers are retarders that convert between lin-
ear and circular polarization and between linear polarization 8. Number of materials.
states with different directions of linear polarization. Quarter 9. Number of waveplates.
wave retarders convert linear into circular polarization or, the 10. Environmental requirements (e.g., temperature
reverse, circular into linear polarization. Half wave retarders range, energy flux that the device needs to be able
rotate a plane of linear polarization about an angle dependent to withstand).
on the orientation of the fast axis of the retarder. Half wave
11. Cost of materials, machines and labor for design
retarders also will reverse the handedness of a circularly
modeling, manufacturing, and testing.
polarized beam. Achromatic retarders for other retardance
values are also useful, but the emphasis in this paper is The purpose of this paper is to provide an overview to
on quarter and half wave retarders. help with requirements 1, 2, 3, 6, 8, 9, and 11 and some con-
We describe several designs of quarter and half wave siderations helpful with requirements 4, 5, 7, and 10 as well.
retarders that are much more uniform in their performance
over a broader wavelength range. In some cases, these ach-
romatic retarders perform well over a wavelength range of 3 Types of Achromatic Retarders
several hundred nanometers.1
3.1 Fresnel Rhombs: Achromatic Retardance through
Total Internal Reflection
2 Design Criteria for Achromatic Devices
Retardance, which can be viewed as a phase difference of the
For design and bulding of achromatic devices, the following p (parallel) component relative to the s (perpendicular) com-
criteria must be considered:2 ponent of the electric field of an incoming light ray of a cer-
tain wavelength, can be induced (even in materials without
1. Requirements for retardance achromaticity, i.e., tar-
any birefringence!) by total internal reflection (TIR) at a
get retardance and tolerances in a given wavelength
range. glass–air interface,4–6 as illustrated in Fig. 1, which depicts
a prism called Fresnel rhomb. The introduced phase shift
2. Requirements for fast axis variation over a given δðλÞ for a light ray of wavelength λ depends on the refractive
wavelength range. index nðλÞ of the rhomb material and on the incidence angle
3. Requirements for temperature independence α of the ray on the inside surface (of the rhomb) according to
(athermality) of retardance. the equation:
*Address all correspondence Michael Kraemer, E-mail: mkraemer@

meadowlark.com 0091-3286/2019/$25.00 © 2019 SPIE
Optical Engineering 082406-1 August 2019 • Vol. 58(8)

Kraemer and Baur: Achromatic retarders in polarization optics
Fig. 1 Two separate TIRs in the Fresnel rhomb each contribute a 45-
deg phase change in the p and s components of an incoming light ray,
thereby producing a quarter-wave retarder. Fresnel invented this
prism in 1817 to convert incoming linearly polarized light into outgoing Fig. 3 For internal reflection angle α ¼ 55° 5 0 , the retardance versus
circularly polarized light. wavelength curve for BK7 glass yields the smallest deviation from
90 deg, namely a deviation of 1°31′ over the 400- to 1600-nm wave-
length range, taking dispersion of refractive index according to
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Schott’s data sheets8 into account.
δðλÞ cos α nðλÞ2 · sin2 α − 1
tan ¼− ; (1)
nðλÞ · sin2 α
EQ-TARGET;temp:intralink-;e001;63;548
2
• influence of stress birefringence in the rhomb
for all incidence angles α that obey the TIR condition: material,4
• influence of surface nonuniformities and of surface
1 films,10
arcsin ≤ α ≤ 90 deg : (2)
• lateral displacement of the beam during passage
nðλÞ
through the device (for quarter wave rhomb).11
In the Fresnel rhomb, two phase shifts of 45 deg each are
introduced by two separate TIRs, totaling up to the desired Additional surface films can be deposited on the rhomb
90-deg phase shift of a quarter wave retarder. In Fig. 2, two surface to lower the phase retardation maximum for a single
quarter wave Fresnel rhombs are connected to form a half TIR in Fig. 4 to 45 deg, or to increase retardance achroma-
wave Fresnel rhomb, that is one that shifts the phase of ticity or to decrease the retardance dependence on the angle
an incoming ray by 180 deg. For BK7 glass with a refractive of incidence, see Refs. 4 and 12 and references therein.
index of 1.5151 at 632.8 nm, for example, α ¼ 55° 5 0 produ-
ces the necessary 45-deg phase shift. Figure 3 shows retard-
ance versus wavelength dependence for a 400- to 1600-nm 3.2 N-Crystals Combination Plates: Achromatic
Fresnel rhomb made from Schott N-BK 7® glass, using the Retardance through Dispersion of Birefringence
dispersion formula from Schott.8 The disadvantages of the In birefringent crystals such as quartz or magnesium fluo-
Fresnel rhomb are ride, the birefringence is not constant as a function of wave-
length (which would give rise to a retardance inversely
• size, as compared with fractions of a mm for, e.g., proportional to wavelength), but depends on the wavelength,
polymer stack retarders, too. This phenomenon, which is called dispersion of birefrin-
• sensitive dependence of the resulting phase shift on the gence, can be utilized to construct achromatic retarders by
angle of incidence, which restricts the angular aperture stacking two or more waveplates of different birefringent
considerably,9
• rotation of retarder fast axis with the azimuth angle for
non-normal incidence,
Fig. 2 Doubling the quarter wave Fresnel rhomb from Fig. 1 yields a
half wave retarder with four TIRs, each contributing 45 deg to the Fig. 4 Dependence of rhomb retardance on internal reflection angle
retardance. Note that for the half wave rhomb the outgoing beam for BK7 glass at 632.8 nm (n ¼ 1.5151). Best for increased angular
is collinear with the incoming one (no displacement). For further aperture is TIR retardance graph being tangent to dashed 45-deg
rhomb geometries, see Ref. 7. retardance line.4

materials with fast axes either parallel or perpendicular to from 0.25 wavelengths (i.e., quarter wave retardance) in the
each other. The advantages of such a crystal combination range from 400 to 1000 nm. An even better fit for the same
plate are wavelength range, yielding a quarter wave achromat with
λ∕100 tolerance, is a combination of a 104.6-μm-thick mag-
• The stack of crystal plates is typically a few millimeters nesium fluoride MgF2 ) waveplate and a 24.1-μm-thick
thick, less bulky than a prism retarder. ammonium dihydrogen phosphate (ADP) waveplate with
• The direction of the stack optic axis does not vary with parallel optic axes, as shown in the same Fig. 5.
wavelength, as is the case for the Pancharatnam-type However, crystal plates of a thickness of a few hundred
plate combinations described in Sec. 3.3.2. (There microns or less are very fragile and therefore difficult to pol-
are other types of crystal waveplate combinations ish to a given retardance goal. Instead, it is necessary to sub-
(see Ref. 13) where fast or slow axes are not aligned; stitute the 104.6-μm-thick MgF2 waveplate by a pair of two
these show achromaticity over a shorter wavelength thicker MgF2 waveplates, e.g., with thicknesses 2104.6 and
range, however, and experience a variation of the stack 2000 μm, in a manner that their optic axes are perpendicular
fast axis with wavelength.) to each other so that their retardances are subtracted and yield
• Crystal waveplates have a higher threshold to with- a net retardance of the desired 2104.6 − 2000 ¼ 104.6 μm.
stand laser or solar flux than polymer retarders. Similarly, the unrealistically thin 24.1-μm ADP waveplate
• Achromats of arbitrary target retardances can be real- can be replaced by a combination of two thicker ADP plates
ized by increasing or decreasing the thicknesses of all in subtraction mode (i.e., with perpendicular optic axes), e.g.,
plates in a given achromatic crystal stack by the same by a 2024.1 μm and a 2000-μm ADP plate with net retard-
scaling factor. ance 2024.1 − 2000 ¼ 24.1 μm. This method is also appli-
cable to realize small net retardances in the Pancharatnam-
The retardance (in units of wavelength) of a stack of N type combination retarder stacks that will be discussed in
crystal plates at wavelength λ (in microns), assuming normal Sec. 3.3.6 and in the application example in Sec. 4.1.
incidence of the ray on the plates, is given as According to Beckers,14 the substitution of crystal wave-
plates of thickness D by two waveplates of the same material
X
N
ϵ i − ωi in subtraction mode (i.e., with fast axes crossed with thick-
RðλÞ ¼ di ; (3) nesses D1 and D2 such that D ¼ D1 − D2 ) not only allows
λ
i¼0 the use of thicker plates but also provides an opportunity to

widen the angular aperture of the achromatic stack, thereby
where ωi , ϵi , and di are the ordinary refractive index, the reducing the dependence of the retardance on the incidence
extraordinary refractive index, and the thickness (in microns) angle of the light ray on the retarder surface. The generali-
of plate number i. Using the convention that di is positive zation of stack retardance [Eq. (3)] to account for an arbitrary
(negative) if the optic axis of plate number i is parallel incidence angle ϕ and an arbitrary azimuth angle α relative to
(perpendicular) to the stack axis,14,15 the formula accounts the stack axis is, following Evans:16
for the addition and subtraction of the plate retardances qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
N ϵi 1 − a 2 − b2 − ωi 1 − a ωþb
2 2 2 2
depending on the orientation of the plate optic axes. X ωi ϵi 2
As an example, Fig. 5 shows the retardance versus wave- RðλÞ ¼ i
di ; (4)
λ
length graph of a quartz-sapphire quarter wave achromatic i¼0

waveplate consisting of a 1335.2-μm -thick quartz plate
and a 1517.3-μm-thick sapphire plate with parallel optic where a ¼ sin ϕ · cos α and b ¼ sin ϕ · sin α. (Note that in
axes. These thicknesses minimize the deviation of retardance Ref. 16 the derivation is given for the azimuth angle relative
to the fast axis angle of the individual plate instead of the
stack axis!) Taylor expansion of this formula for small inci-
dence angles ϕ yields the approximation, given in Ref. 15:
X
N
εi − ωi ϕ2 cos2 α sin2 α
RðλÞ ¼ di 1 − −
λ ωi εi
i¼0
2ωi
þ Oðϕ4 Þ: (5)
Figure 6 illustrates the dependence of the retardance on

the azimuth angle for fixed incidence angles of 2 deg,
5 deg, and 10 deg at a wavelength of 630 nm for the
quartz-sapphire achromat shown in Fig. 5. With a four-
plate quartz-quartz-sapphire-sapphire combination, where
the two pairs of the same material are in subtraction
mode as explained above, the amplitude of the retardance
Fig. 5 Retardance versus wavelength graphs for two different achro- dependence on the azimuth angle can be greatly reduced,
matic combination waveplates: a quartz-sapphire and a magnesium as is shown in Fig. 7 for the whole wavelength range of
fluoride (MgF2 )-ADP combination. The MgF2 and ADP plates with
thicknesses of 0.104 and 0.024 mm, respectively, are too thin and
achromaticity, namely 400 to 1000 nm. The solution in this
fragile to manufacture (by polishing), but can be replaced by a pair figure uses quartz plate thicknesses of 0.500 and 1.835 mm
of thicker plates of the same material in subtraction mode. and sapphire plate thicknesses of 1.162 and 2.680 mm. If we

Fig. 8 Assuming upper bounds of 2-nm retardance error and 0.25-

deg fast axis clocking error for each plate, computer simulations for
Fig. 6 Retardance versus azimuth graph at 630 nm for the quartz/ all possible combinations of these worst-assumed plate errors yield
sapphire combination waveplate from Fig. 5. The azimuth depend- an estimate for achievable manufacturing tolerances. Ellipsometer
ence is periodic in 180 deg, and its amplitude increases with the measurements on the finished part stay well within these worst-
angle of incidence, i.e., angle between light ray and surface normal. case modeling tolerances, compare Fig. 9.
Here, the graphs are shown for 2 deg, 5 deg, and 10 deg incidence.
produce birefringence in the films. This stretching aligns

the long polymer chain molecules in a common direction,
and the resulting anisotropy causes the films to be birefrin-
gent. The birefringence increases with the stretch distance
but the resulting retardance also depends on the film thick-
ness, which is reduced by stretching. (The film thickness
varies from about 10 to 150 μm.) These competing effects
limit the maximum achievable retardance in an individual
polymer film to a few thousand nanometers. This is more
than enough retardance for the designs discussed in this
paper for wavelengths from visible to the near IR.
Most polymer retarders do not transmit well in the UV and
also show absorption bands in the IR, especially at wave-
lengths longer than 2 μm. The transmitted wavefront quality
of these films is poor but improves dramatically, typically to
Fig. 7 The dependence of the retardance on incidence angle for the
quarter wave or better at visible wavelengths, when the films
quartz-sapphire combination waveplate in Fig. 6 can be greatly are laminated between optically flat windows with an adhe-
reduced, if we split, following Beckers,14 the quartz plate into a suit- sive that is a good index of refraction match to the polymer.
able combination of two quartz plates in subtraction mode, and like- Examples of polymers that can be stretched to make retarders
wise the sapphire plate into two suitable sapphire plates in subtraction are polyvinyl alcohol, polycarbonate, polyimide, polysty-
mode. Shown here for this four-plate combination is the range of
retardances for incidence angles up to 10 deg and arbitrary azimuth rene, and polymethyl methacrylate.17 Birefringent reactive
angles. mesogen polymer films can also be made by spin coating
onto a glass or fused silica substrate.18
At Meadowlark Optics retardance tolerances of λ0 ∕350 at
limit the feasible solutions by imposing a lower bound on the the given design frequency λ0 are achieved for individual
plate thicknesses (because of technical limitations in polish- polymer waveplates. The rotational errors for the fast axis
ing the plates to optimal thickness), the angular aperture for directions of the parallel waveplates in a retarder stack are
a given wavelength range depends on the lower bound for <0.25 deg in practice. Computer simulations (using Mueller
the plate thickness, if we define the angular aperture as matrices to represent retarders and Mueller matrix multiplica-
the maximum possible incidence angle so that the retardance tion to model their combination) that allow for the com-
deviation does not surpass a given tolerance (such as, for pounding of the retardance and angle errors of the individual
example, 0.005 waves). plates yield upper bounds for the retardance deviation and
fast axis variation over a given wavelength range. Typical
3.3 Pancharatnam-Type Combination Plates: simulation graphs for a quarter wave superachromat in the
Achromatic Retardance through Fast Axis 370- to 1100-nm wavelength range are shown in Fig. 8.
Rotation of Same-Material Plates Ellipsometer measurements on the manufactured part confirm
3.3.1 Materials for multilayer film retarders these error modeling results (see Fig. 9).
Most multilayer achromatic retarder designs are best manu-
factured using birefringent polymer retarders for the individ- 3.3.2 Pancharatnam’s three-layer design
ual component retarders. Usually cast polymer films are Pancharatnam19 introduced an achromatic retarder formed of
heated to a softening temperature and then stretched to three parallel waveplates (elements) of the same material,

Fig. 9 Measured retardance versus wavelength dependence for a quarter wave five-layer stretched
polymer superachromat for 380 to 1100 nm (measured data points as circles) that was manufactured
in October 2016 at Meadowlark Optics, compared with the theoretical curve obtained by modeling with
Mueller matrices, taking dispersion of birefringence into account.

where the first and third waveplates are of equal retardance sin f · π2
2δ1 with fast axes parallel, whereas the center plate has sinðf · 2δ1 Þ ¼ sin 2δ1 ; (8)
cos Δ
retardance 2δ2 and is rotated relative to the outer plates

by an angle c. Title20 proved, using Jones calculus,3,21
that these assumptions, which describe the symmetry of whereas for half wave achromats (2Δ ¼ π), we have
the waveplate combination about the center plane, are nec- 2δ1 ¼ π, i.e., the outer plates are half wave retarders at
essary so that the combination of the three plates is a pure the design wavelength λ0 .
retarder, i.e., does not produce any rotation of the polariza- 2. 2δ2 ¼ π, i.e., the center plate has half wave retardance
tion axis of incident light that is polarized along the fast or at the design wavelength λ0 for both half wave and
slow axis of the stack. (The decomposition theorem that a nonhalf wave achromats.
combination of retarders can always be represented as a com- 3. The center plate must have a rotation angle c (relative
bination of a retarder followed by a rotator is proven in to the outer plates) that solves the equation:
Ref. 3.) Pancharatnam derives the resulting stack retardance

2δ to be
tan f · π2
cos 2c ¼ − : (9)
cos δ ¼ cos 2δ1 cos δ2 − sin 2δ1 sin δ2 cos 2c: (6) tanðf · 2δ1 Þ
The corresponding stack fast axis angle c1 is given as

3.3.3 Title’s three-layer design
EQ-TARGET;temp:intralink-;e007;63;311 cot 2c1 ¼ csc 2cðsin 2δ1 cot δ2 þ cos 2δ1 cos 2cÞ: (7)
Instead of imposing equal retardance and equal fast axis
By a suitable choice of the degrees of freedom δ1 , δ2 , and angle boundary conditions at a wavelength below and a
c Pancharatnam achieves an achromaticity of the stack that is wavelength above the design wavelength λ0 , Title20 deter-
characterized as mined a three-layer achromat design by setting the deriva-
tives of both the retardance and the fast axis angle equal to
• the desired retardance 2Δ in the vicinity of the design zero at the design wavelength and obtained the following
wavelength λ0 at which the plate retardances are equations, written here in Pancharatnam’s notation from
measured, the previous Sec. 3.3.2 for comparison:
• a dimensionless constant f such that the stack retard-
1. The retardance 2δ1 of the outer plates is for nonhalf
ance is equal to the desired retardance 2Δ at the two
wave achromats (2Δ ≠ π), the implicit solution of
wavelengths λ0 ∕ð1 þ fÞ and λ0 ∕ð1 − fÞ,
• the fast axis orientation angles of the stack being equal π
(to each other) at these two wavelengths. 2δ1 ¼ 2
sin 2δ1 ; (10)
cos Δ
Assuming constant birefringence as a function of wave-

length, Pancharatnam derives, using Poincaré sphere consid- whereas for half wave achromats (2Δ ¼ π), we have
2δ1 ¼ π as in Pancharatnam’s design.
erations for which he credits Pockels.22
2. 2δ2 ¼ π, i.e., the center plate has half wave retardance
1. 2δ1 is, for nonhalf wave achromats (2Δ ≠ π), the at the design wavelength λ0 for both half wave and
implicit solution of the transcendental equation: nonhalf wave achromats.

Table 1 Comparison of three-layer quarter wave achromat designs

by Pancharatnam19 and Title20 and a five-layer quarter wave achro-
mat design by the authors. The corresponding graphs for retardance
and fast axis angles in the one-octave wavelength range of 400 to
800 nm are shown in Fig. 10.
QW achromat design
400 to 800 nm
(1 octave) const. No. of Design Retardance FA
birefringence plates parameters deviation variation
Title, conditions at λ0 3 λ0 ¼ 567.0 nm, 4°26′ 6°16′

retardance
94°26′
Pancharatnam, 3 λ0 ¼ 547.2 nm, 5°3′ 4°18′

conditions at ð1 f Þλ0 f ¼ 0.2296
Meadowlark, global 5 Proprietary 0°36′ 3°53′

min. of retardance
deviation
half wave achromat meets very narrow tolerances for both

the retardance deviation and the fast axis variation, as
shown in Tables 1 and 2 and illustrated in Fig. 12. An
Fig. 10 Of the two three-stack quarter wave achromat designs the
even better nine-stack to minimize retardance deviation in
one by A. Title has slightly less retardance deviation over the one- a given wavelength range is achieved by stacking three
octave 400- to 800-nm wavelength range, but Pancharatnam’s has Pancharatnam-design three-stacks with the same 60-deg
less fast axis angle variation, see comparison in Table 1. These are rotation of the center three-stack against the parallel outer
modeled graphs assuming constant birefringence as function of three-stacks.
wavelength.
3.3.5 How to account for dispersion of birefringence

3. The center plate must have a rotation angle c (relative in Pancharatnam-type combination waveplates
to the outer plates) that solves the equation:
In the their pioneering papers Pancharatnam19 and Title20
π
assumed the birefringence to be constant, that means inde-
cos 2c ¼ − 2
: (11)
pendent of wavelength. While some polymer sheet materials
2δ1
such as polyvinyl alcohol20 possess a nearly constant bire-
fringence for visible light, in general the dispersion of
By comparison of Eq. (8) with Eq. (10) and of Eq. (9) birefringence may lead to less achievable achromaticity in
with Eq. (11), we see that in the limit f → 0 Pancharatnam’s a given wavelength range and must be taken into account
equations go over into Title’s equations, so that Title’s ach- when designing achromats. With the following mathematical
romat can be considered a limit case of the Pancharatnam transformation, all achromat models may be generalized to
achromat for small wavelength ranges. Figure 10 shows
the stack retardance and stack fast axis angle graphs for
Table 2 Retardance deviation (in degrees) and fast axis variation for
both approaches for the fit with least retardance deviation the three-half wave achromat designs shown in Fig. 11 for a 400- to
from 0.25 waves (quarter wave achromat) over the wave- 800-nm wavelength range (i.e., one octave).
length range. As a third solution, we show the one obtained
by numerical optimization for a five-plate-stack designed by
the authors. HW achromat design
400 to 800 nm
(1 octave) const. No. of Design Retardance FA
3.3.4 N-layer designs
Title, conditions at λ0 3 λ0 ¼ 533.3 nm 14°22′ 5°27′
By stacking an odd number of waveplates symmetric about
the center plate (as for plate retardances and plate fast axis Pancharatnam, 3 λ0 ¼ 533.3 nm, 4°25′ 7°8′
orientations), a tighter tolerance can be realized for quarter conditions at ð1 f Þλ0 f ¼ 0.2851
wave, half wave, and other achromats. The authors have Meadowlark, global 5 proprietary 0°16′ 1°41′
min. of retardance
designed and manufactured five-plate polymer film achro- deviation
mats for quarter wave, third wave, and half wave require-
ments. In Ref. 20, Title proposed and provided error Title, 3 × 3 9 λ0 ¼ 533.3 nm, 0°13′26″ 0°23′5″
analysis for a half wave achromat obtained by stacking combination retardance
three of his three-plate achromats with parallel fast axes 92°13′
for the outer three-stack elements and with fast axis of the
Pancharatnam, 9 λ0 ¼ 533.3 nm, 0°0′23″ 0°2′12″
center three-stack rotated by 60 deg with respect to the 3 × 3 combination f ¼ 0.2851
fast axis of the outer three-stack elements. This nine-stack

Table 3 Retardance deviation (in degrees) and fast axis variation for
the three half wave achromat designs shown in Fig. 12 for a 400- to
1600-nm wavelength range (i.e., two octaves). The nine-stack
consisting of three Pancharatnam three-stacks provides the least
retardance deviation over the two octave range, while the stacking
of three title three-stacks yields better achromaticity near the design
wavelength.
HW achromat design
400 to 1600 nm
(two octaves) const. No. of Design Retardance FA
Meadowlark, global 5 Proprietary 5°3′ 6°33′

min. of retardance
deviation
Title, 3 × 3 9 λ0 ¼ 640.0 nm 17°5′ 6°1′

combination
Pancharatnam, 9 λ0 ¼ 640.0 nm, 2°1′ 1°37′

3 × 3 combination f ¼ 0.4942
λ0 bðλÞ
δðλÞ ¼ δðλ0 Þ · · : (12)
λ bðλ0 Þ
Since the birefringence function bðλÞ in this derivation

always occurs as a ratio, divided by its value at the design
Fig. 11 Of the two three-stack half wave achromat designs the one by
Pancharatnam has less retardance deviation over the one-octave wavelength λ0 , it may be expressed in any units. This ratio is
400- to 800-nm wavelength range, see comparison in Table 2. a normalized birefringence with value 1 at the design wave-
length λ0 .
account for an arbitrary birefringence versus wavelength
function bðλÞ: 2. We can reduce the retardance function δðλÞ to the one
corresponding to constant birefringence by dividing
1. In a given wavelength range λmin ≤ λ ≤ λmax , the plate the wavelength λ by the normalized birefringence
retardance at wavelength λ can be expressed in terms bðλÞ∕bðλ0 Þ and obtaining a transformed “zero-
of the birefringence bðλÞ as dispersion” wavelength λ 0 given as
λ bðλ0 Þ
λ0 ¼
bðλÞ
¼λ· ; (13)
bðλÞ
bðλ0 Þ
which transforms the retardance function in Eq. (12)

into one without birefringent dispersion:
λ0
δ 0 ðλ 0 Þ ¼ δðλ0 Þ · ; (14)
λ0
where the transformed wavelength λ 0 runs from λmin

0 ¼
0Þ 0 bðλ0 Þ
min½λ · bðλ
bðλÞ to λmax ¼ max½λ · bðλÞ .
3. We find a suitable achromat design in which the trans-
formed retardance δ 0 ðλ 0 Þ is close to a given target
retardance in the zero-dispersion wavelength range
0
from λmin 0 .
to λmax
4. We transform back to the wavelength λ in the interval
λmin ≤ λ ≤ λmax to obtain the achromatic retardance

bðλ0 Þ
δðλÞ ¼ δ 0 λ · :
EQ-TARGET;temp:intralink-;e015;326;147 (15)
Fig. 12 Title20 suggested a stacking of three identical three-layer half
bðλÞ
wave achromats with parallel outer achromats and center achromat
rotated by 60 deg with respect to the outer achromats; this yields a
nine-layer half wave achromat with far less retardance deviation than
The assumption that the birefringence keeps its sign over
a single three-layer half wave achromat, see comparisons in Tables 2 the wavelength range of consideration does not require the
and 3. birefringence to be monotone in this range. Few materials

have a birefringence increasing with wavelength; an example Here, c is a compensation parameter to be chosen so that
is apophyllite,7,23,24 a mineral with a “natural” retardance the retardance versus wavelength function of the pair is
achromaticity in the visible, which is, however, “rare in usa- close to constant, thus making the pair as retardance achro-
ble optical quality specimens.”23 Certain stretched polymer matic as possible. If we then symmetrically stack three
films have increasing birefringence, also known as reverse such achromatic waveplate pairs of suitable retardances
or abnormal wavelength dispersion and can in some cases to form a 3 × 2 Pancharatnam combination waveplate, the
be used as single-layer achromatic retarder.25,26 achromaticity is far better than for the simple Pancharatnam
As an example, we will use Pancharatnam’s design for a three-stack of material A. As an example, we want to build
quarter wave retarder using three single magnesium fluoride a mixed dispersion 3 × 2 Pancharatnam-type achromat for
(MgF2 ) plates in the wavelength interval from 2 to 5 μm. the wavelength range 400 to 1600 nm (two octaves). With
According to Ref. 27, a best fit to measured data for MgF2 design wavelength λ0 ¼ 1000 nm, we transform the wave-
birefringence as function of wavelength λ (in microns) is length λ to the wavelength λ 0 in which the birefringence
obtained in the model equation: dispersion appears constant:
Iλ2 Jλ2 λ
bðλÞ ¼ H þ þ ; (16) λ0 ¼ : (17)
λ2 − G λ2 − L bðλÞ
c þ 1 − c bðλ
with coefficients H ¼ −19.364 · 10−3 , I ¼ 30.992 · 10−3 , For a particular choice of polymers for materials A and B,
G ¼ 2.3253 · 10−3 , J ¼ 40.060 · 10−3 , and L ¼ 388.37. 0 ∕λ 0
the wavelength interval ratio λmax min obtains its minimum
With λmin ¼ 2, λmax ¼ 5, and the design wavelength at approximately c ¼ 2.76, yielding the wavelength trans-
λ0 ¼ 3.5, we transform to zero-dispersion wavelength formation depicted in Fig. 15 with values between λmin 0
¼
bðλ0 Þ
λ 0 in the range from λmin
0
¼ λmin · bðλ min Þ
0
¼ 1.84 to λmax ¼ 0
755.5 nm and λmax ¼ 1461.6 nm. So we see that the origi-
bðλ0 Þ nal two octaves of 400 to 1600 nm have shrunk down to
λmax · bðλ max Þ
¼ 5.82 (see Fig. 13), and, choosing f ¼ 0.29
less than one octave in terms of zero-dispersion wavelength,
in Pancharatnam’s method, we obtain the retardance and
see Fig. 16. The “doubling” of the shape is a result of the fact
fast axis graphs shown in Fig. 14 for zero-dispersion
that the wavelength transformation is not one-to-one, see
wavelength λ 0 and for actual wavelength λ.
Fig. 15.
3.3.6 Mixed dispersion Pancharatnam-type six-plate
combinations 3.3.7 Effect of oblique incidence on retardance and
Instead of a single waveplate of material A with birefringence fast axis direction of a Pancharatnam-type
function bðλÞ and retardance (in waves) 2δ at the design combination plate
wavelength λ0 , one can consider a pair (of same resulting Consider a stack of N parallel waveplates of the same
retardance 2δ) consisting of a waveplate of a zero-dispersion uniaxially birefringent material with positive birefringence
material B and retardance ðc þ 1Þ · 2δ in subtraction mode βðλÞ that is assumed to be very small compared with the
with a waveplate of material A with retardance c · 2δ. ordinary refractive index (small-birefringence approxima-
tion).28 Let the normal-incidence plate retardances be
r1 ðλÞ; r2 ðλÞ; : : : ; rN ðλÞ at a certain wavelength, and let
the fast axes be in the surface plane oriented at angles
α1 ; α2 ; : : : ; αN relative to a reference axis in the surface
plane. Since the birefringence is much smaller than the ordi-
nary refractive index, the ordinary and extraordinary ray are
parallel in zero order approximation, so that a linearly polar-
ized beam that is obliquely incident on the waveplate stack
will transverse the stack in a straight line with an internal
incidence angle ϕ determined from the external incidence
angle with Snell’s law and with an azimuth angle α to
the reference axis in the surface plane that is unaffected
by refraction. According to Hale and Day,28 the phase
delay of the extraordinary wave relative to the ordinary
wave in plate no. i in the small-birefringence aproximation
is given as
Fig. 13 By a wavelength transformation, the achromat modeling sin2 θi

problem for same-material plate combinations is reduced to an equiv- ri ¼ ri · ; (18)
cos ϕ
alent problem with constant birefringence (i.e., zero dispersion of

birefringence), which is both easier to solve and independent of the
retarder material. The recipe for this transformation technique is where θi is the angle between the direction of propagation of
described in Sec. 3.3.5. The graph shows the wavelength transforma- the beam and the optic axis of plate no. i. By placing
tion to reduce the MgF2 Pancharatnam stack to the equivalent stack a Cartesian x − y − z coordinate system such that the z-axis
for a zero dispersion material. Both the transformation and the sub-
sequent back transformation of the zero-dispersion solution to actual is normal to the waveplates and the x-axis is aligned with
wavelength are straightforward and do not require the numerical the reference axis in the surface plane, the propagation
solution of implicit equations. direction is given by unit vector:

Fig. 14 In terms of the zero-dispersion wavelength λ 0 , the “classic” Pancharatnam three-stack solution
(i.e., assuming constant birefringence) in the range from 1.84 to 5.82 μm is equivalent to the original
problem for λ between 2 and 5 μm.
and therefore the angle θi between them is
EQ-TARGET;temp:intralink-;e021;326;391 cos θi ¼ k^ · Ω
^ i ¼ sin ϕ sinðα − αi Þ: (21)
Substitution into Eq. (18) yields the oblique incidence

retardance of plate no. i as
1 − sin2 ϕ sin2 ðα − αi Þ
ri ¼ ri · : (22)
cos ϕ
The corresponding fast axis of the beam propagating

through plate no. i is obtained as the intersection of the
plane normal to its optic axis Ω ^ i and the plane normal to
the beam and is therefore aligned with the unit vector:
!
cos ϕ cos αi
Fig. 15 The wavelength transformation for a Pancharatnam three-
stack of a retarder material with birefringent dispersion is not one-
f^ i ¼ Ω
^ i × k^ ¼ cos ϕ sin αi
EQ-TARGET;temp:intralink-;e023;326;251 : (23)
to-one. This is an advantage, because it results in the “doubling” of − sin ϕ cosðαi − αÞ
the Pancharatnam solution upon back transformation as shown in
Fig. 16. As expected, this fast axis direction for oblique incidence
goes over into the fast axis direction ðcos αi ; sin αi ; 0ÞT for
! normal incidence in the plane of the film in the limit for
sin ϕ cos α
k^ ¼ sin ϕ sin α ; (19) ϕ → 0. The three-dimensional fast axis direction f^ i can be
uniquely characterized by its angle ωi with a reference direc-
cos ϕ
tion in the plane normal to the beam. We suggest to define ωi
as the polar angle in the s–p coordinate system in the beam
the optic axis of plate no. i is given by the unit vector: normal plane, where the unit vector s^ ¼ ð− sin α; cos α; 0ÞT
! is both normal to the plane of incidence and normal to
− sin αi the beam, and the unit vector p^ ≔ k^ × s^ is in the plane of
î ¼
Ω
EQ-TARGET;temp:intralink-;e020;63;112 cos αi ; (20) incidence. With this convention, the waveplate no. i under
0 oblique incidence is equivalent to a waveplate normal to

Fig. 16 In the transformed wavelengths λ 0 , the original two-octave (400 to 1600 nm) Pancharatnam prob-
lem is reduced to one with less than an octave of wavelength (from 755.5 to 1461.6 nm) if the wavelength
transformation to zero-birefringence wavelength is not one-to-one, see Fig. 15.
the beam with the normal-incidence retardance ri in Eq. (22) variation of the stack fast axis over a given wavelength
and the fast axis orientation βi given as range, depend on incidence and azimuth angles of the
beam relative to the compound retarder. For compound
half wave retarders with three layers (by Pancharatnam19),
cos βi ¼ f^ i · s^ ¼ cos ϕ sinðαi − αÞ;

five layers (by authors), seven layers (by authors), or nine
sin βi ¼ f^ i · p^ ¼ − cosðαi − αÞ: (24) layers (by Title20), we modeled the retardance deviation
and fast axis variation (worst-case over all wavelengths).
It follows that For the five-layer HW achromat, the contour plots are
shown in Fig. 17. Taking—for a fixed incidence angle—the
sin βi 1 tan γ i worst possible retardance deviation/fast axis variation over
tan βi ¼ ¼− ¼ ; (25)
cos βi cos ϕ · tanðαi − αÞ cos ϕ
all azimuth angles we obtain for the four achromat designs

under consideration, the graphs of the respective error versus
where γ i ≔ αi − α − π∕2 is the angle of the normal- incidence angle shown in Fig. 18. Additionally—but not
incidence fast axis of plate no. i with the s-direction. We illustrated here—we must be aware that with the incidence
have shown that the given waveplate stack under oblique angle not only the fast axis variability with wavelength
incidence with plate retardances ri and plate fast axis angles increases (implying less achromaticity), but even for a fixed
αi can be replaced by an equivalent stack of waveplates nor- wavelength and a fixed azimuth the stack fast axis is shifted
mal to the beam with retardances ri in Eq. (22) and fast axis as a function of the incidence angle alone, similar to the
angles βi in Eq. (25) relative to the s-direction of the beam. effect on a single waveplate given in Eq. (25).
The compound retardance, the compound fast axis, and the
compound optical rotation of this equivalent waveplate stack 4 Applications
under normal incidence can be determined with the usual
methods, such as Mueller calculus,5 Jones calculus,5 or— 4.1 DKIST Infrared Superachromatic Retarder for of
for symmetric stacks—explicit computation with (recur- the Calibration of the Cryogenic Nea Infrared
sively applied) formulas given by Pancharatnam.19 The Spectropolarimeter
theory developed in the last section can be applied to calcu- The Daniel K. Inouye Solar Telescope (DKIST) that is cur-
late for achromatic Pancharatnam-type retarder stacks how rently under construction on the Haleakalā mountain on the
the achromaticity performance, namely the deviation of Maui Island, Hawai’i, will be “the world’s largest solar tele-
the stack retardance from a desired target value and the scope with wavelength coverage from 380 nm to 28 μm and

advanced adaptive optics enabling the highest spatial

resolution of the solar atmosphere yet achieved.”29 One of
the instruments of DKIST, the Cryogenic near-infrared
spectro-polarimeter (Cryo-NIRSP), is dedicated to the meas-
urement of solar coronal magnetic fields with 1 arc sec res-
olution at near- and thermal-infrared wavelengths, and will
be able to measure the full polarization state of spectral lines
in the wavelength range from 1 to 5 μm. The calibration
retarder of Cryo-NIRSP, the so-called superachromatic
retarder (SAR), is required to be a zero-order retarder within
33% of a quarter wave in the 2- to 5-μm range. Because of
the high solar flux in the application crystalline materials are
required, and for high transmission out to 5 μm magnesium
fluoride (MgF2 ) was selected as suitable. In order to meet the
achromaticity requirement (90 deg 30 deg retardance),
a three-element Pancharatnam design was selected. A zero-
order MgF2 half wave retarder, as a building block in the
Pancharatnam design, would have a thickness of <100 μm,
less than any vendor would have confidence to polish, the
aspect ratio (diameter divided by thickness) being on on
the order of 1000:1 for the 12-cm diameter plate. Therefore,
each of the the three Pancharatnam elements was built at
Fig. 17 For an ideal (modeled) Pancharatnam-type half wave achro- Meadowlark Optics30 as a pair of about 2-mm thick MgF2
mat in a 400- to 800-nm wavelength range, the dependence of retard-
ance deviation and fast axis variation (worst case over wavelength
plates in subtraction mode, whose >40 order retardances
range) on both incidence and azimuth angles are shown as contour subtract to second or third order at the design wavelength
graphs. of 633.443 nm, resulting in a zero-order stack retardance
Fig. 18 For ideal (modeled) Pancharatnam-type half wave achromats in a 400- to 800-nm wavelength
ranges of 3, 5, 7, and 9 layers, the dependence of retardance deviation and fast axis variation (i.e., differ-
ence of maximum and minimum compound fast axis angle in the wavelength range) are compared, for
each incidence angle taking the maximum (worst case) over all wavelengths and all azimuth angles.

in the application wavelength range of 1 to 5 μm.27 The

design parameters for the three pairs of plates are shown
in Table 4, and the corresponding retardance versus wave-
length and fast axis angle versus wavelength graphs are
shown below in Fig. 19 for normal incidence and accounting
for dispersion of birefringence as given in Ref. 27.
Upon closer inspection, one notices that the three-element
design in Table 4 is not a classic quarter wave Pancharatnam
achromat design as outlined above, but one that ensures
tighter retardance tolerance (here about 1.5 deg for 2500-
to 3900-nm wavelength). In the selected design, the center
plate is not a half-wave element at the wavelength
2621 nm, where 90-deg retardance is reached for the first
time with descending wavelength, but is a three-quarter-
wave element, whereas the outer elements are half wave ele-
ments. Such a 2:3:2 design (denoting ratios of the element
retardances here), could be called an inverted Pancharatnam
design and leaves the fast axis orientation of the center
element relative to the outer elements as parameter to be var-
ied to achieve the desired retardance achromaticity. Varying
this clocking angle yields a family of curves that intersect at
the wavelengths (∼2022 and 2621 nm), where the element
retardances are 0.5, 0.75, and 0.5 waves, respectively.
Figure 19 shows retardance and fast axis angle versus wave-
length for the 107.75-deg clocking angle that was chosen for
the design.27 Fig. 19 Retardance and fast axis orientation versus wavelength for
the Pancharatnam design used for the SAR of the DKIST Cryo-
NIRSP.
4.2 Achromatic Ferroelectric Liquid Crystal Shutter
(or Light Valve) the incident horizontally polarized (0 deg) light passes
An FLC waveplate is made from a birefringent material through the retarder with unchanged polarization state and
whose fast axis can have one of two orientation angles is then blocked by the vertically (90 deg) oriented output
(their difference is called switching angle), depending on polarizer, resulting in a closed state. If the FLC device is
an applied voltage. A switching angle of 45-deg is useful placed between parallel polarizers (oriented at 0 deg), the
for constructing an optical shutter that will open and close light valve function is reversed, that is we have a closed
as the applied voltage is changed. The principle is illustrated state for FLC at 45 deg and an open state for FLC at 0 deg;
in Fig. 20: an FLC device with half wave retardance at a cer- however, the operation of the light valve between crossed
tain design wavelength, when placed at 45 deg between two polarizers is preferred, leading to a higher contrast ratio of
crossed polarizers oriented at 0 deg and 90 deg, respectively, transmitted intensity for open and closed state. The function
turns the horizontally (0 deg) polarized light coming from the of this light valve consisting of one single FLC cell has the
input polarizer into vertically (90 deg) polarized light, which following disadvantage:
is passed through unattenuated by the output polarizer (also The change of the retardance as measured in waves is
90 deg), resulting in an open state. If the FLC retarder fast inversely proportional to the wavelength (in absence of
axis is switched from 45 deg to 0 deg by an applied voltage, dispersion of birefringence), which lowers the contrast ratio.
Table 4 Design retardance and fast axis orientation for the three
pairs of Pancharatnam elements that form the SAR (in the 2 to
5 μm wavelength range) for calibration of the Cryo-NIRSP of the
DKIST, currently being built in Hawai’i.
Element Design Design retardance Fast axis

number thickness (mm) (waves at 633.443 nm) orientation (deg)
1 2.27314 42.230 0.00
1 2.15311 40.000 90.00
2 2.33321 43.346 107.75
2 2.15311 40.000 197.75

Fig. 20 Open and closed states are obtained by placing a single FLC
3 2.27314 42.230 0.00 cell with half wave retardance at 45 deg between crossed polarizers.
By application of a suitable voltage to the FLC cell, its fast axis is
3 2.15311 40.000 90.00 rotated by ∼45 deg, switching an open state to a closed state or
vice versa.

polarizers. The indicated angles are approximate, depending

on the design wavelength, the wavelength range for shutter
operation, the FLC material (birefringence and switching
angle), and the material of the outer waveplates. Typical
closed state and open state retardances and fast axis angles
are shown in Fig. 22 and resulting transmitted intensity
curves in Fig. 23, whereas Fig. 24 shows measured data
for an achromatic FLC device built by Meadowlark Optics.
A different achromatic FLC design to achieve a switchable
quarter wave FLC retarder for use in low-coherence interfer-
ence microscopy is shown in Ref. 32.
Fig. 21 Open and closed states obtained by placing a Pancharatnam-
stack half wave achromat with FLC cell as center half wave plate
between crossed polarizers. This is the achromatic equivalent to 5 Controlling Temperature Dependence of
Fig. 20. The indicated fast axis rotation angles of the individual plates Retardance
are approximate and depend on the wavelength range for operation
as well as the birefringent dispersion and the switching angle of the
By the same principles to control birefringence dispersion by
FLC material. using waveplate pairs of two different materials in addition
or subtraction mode, with the goal to achieve retardance
achromaticity in a given wavelength range about a design
wavelength, one can achieve retardance independence of
temperature changes by the use of waveplate pairs of two
materials with different temperature coefficients, see Guimond
and Elmore.33 Defining the temperature coefficients of the
two materials as
dR1 ∕dT dR2 ∕dT

γ 10 ¼
EQ-TARGET;temp:intralink-;e026;326;498 ; γ 20 ¼ ; (26)
R1 R2
where T is the temperature and R1 and R2 are the respective

retardances of the two materials in any units. Imposing the
athermality constraint
Fig. 22 Modeled retardance graph (left) and fast axis graph (right) for γ 1 R1 þ γ 2 R2 ¼ 0;
EQ-TARGET;temp:intralink-;e027;326;422 (27)
open and closed state of half wave retarder between crossed
polarizers. we ensure that the change of retardance of material 1 due to
a temperature gradient is compensated by an equal, but of
opposite sign, change of retardance of material 2. The con-
Following Hariharan,31 this issue can be relieved by
dition that the two retardances R1 and R2 must add up to a
making the FLC half wave retarder achromatic by placing
given target retardance yields a second equation to calculate
it between two nonswitchable half wave polymer retarders the two unknown plate retardances R1 and R2 . Achromaticity
with parallel fast axes to form a Pancharatnam-type half (independence of retardance from wavelength) and athermal-
wave achromat, see Fig. 21 for operation between crossed ity (independence of retardance from temperature) cannot, in
Fig. 23 Comparison of modeled closed state (a) and open state (b) polarized transmission intensities for
achromatic Pancharatnam-stack (solid lines) and nonachromatic single (dashed lines) FLC light valves
between crossed polarizers. Since for the closed state, the fast axis of the single-layer FLC retarder is
aligned with the axes of the crossed polarizers, the closed state transmission is zero over the entire
wavelength range (for perfect polarizers).

Fig. 24 Measured open and closed state transmissions and contrast ratio for an achromatic FLC device
built by Meadowlark Optics compared with the measured contrast ratio of a single layer FLC device.
The fringing oscillations seen in the open state transmission graph are due to reflections from the
glass windows that encase the FLC center waveplate.
general, be satisfied simultaneously. However, the relative References

thicknesses of the two materials can in some cases be chosen 1. M. Kraemer and T. Baur, “Achromatic devices in polarization optics,”
to meet given tolerances for both requirements. Proc. SPIE 10528, 105281C (2018).
2. R. E. Fischer, B. Tadic-Galeb, and P. R. Yoder, Optical System Design,
McGraw Hill, New York (2008).
3. H. Hurwitz and R. C. Jones, “A new calculus for the treatment of optical
6 Conclusion systems II. Proof of three general equivalence theorems,” J. Opt. Soc.
Am. 31, 493–499 (1941).
There are several choices for achromatic polarization control 4. R. J. King, “Quarter-wave retardation systems based on the Fresnel
with retarders. Most recently, birefringent polymers and rhomb principle,” J. Sci. Instrum. 43, 617–622 (1966).
liquid crystals are more often applied as alternatives to tradi- 5. E. Collett, Polarized Light: Fundamentals and Applications, Marcel
Dekker, New York, Basel (1993).
tionally used birefringent crystals, such as quartz, magne- 6. E. Collett, Field Guide to Polarization, SPIE Press, Bellingham (2005).
sium fluoride, and sapphire. Methods for achieving half and 7. J. M. Bennett, “A critical evaluation of rhomb-type quarterwave retard-
quarter wave retardance include ers,” Appl. Opt. 9, 2123–2129 (1970).
8. “Schott optical glass collection data sheets,” http://www.us.schott
.com/d/advanced_optics/102fefee-c1cb-4772-a784-1ef2e328eb4c/1.3/
1. Using TIR, schott-optical-glass-collection-datasheets-english-us-17012017.pdf,
(27 December 2017).
2. using a combination of birefringent materials of differ- 9. H. Fabricius, “Achromatic prism retarder for use in polarimetric
ent birefringent dispersions, and sensors,” Appl. Opt. 30, 426–429 (1991).
10. R. W. Ditchburn and G. A. J. Orchard, “The polarization of totally
3. using a stack of retarders of one common material but reflected light,” Proc. Phys. Soc. B 67, 608–614 (1954).
different fast axis directions. 11. C. M. McIntyre and S. E. Harris, “Achromatic wave plates for the visible
spectrum,” J. Opt. Soc. Am. 58, 1575–1580 (1968).
12. C. J. Koester, “Achromatic combinations of half-wave plates,” J. Opt.
The latter two methods can also be combined for further Soc. Am. 49, 405–409 (1959).
improved performance and to meet multiple simultaneous 13. C. F. Buhrer, “High-order achromatic quarterwave combination plates
requirements (e.g., tighter retardance tolerance with respect and tuners,” Appl. Opt. 27, 3166–3169 (1988).
14. J. M. Beckers, “Achromatic linear retarders,” Appl. Opt. 10, 973–975
to changes of wavelength or temperature, greater angular (1971).
aperture, or less fast axis angle variation with wavelength). 15. J. M. Beckers, “Achromatic linear retarders with increased angular
aperture,” Appl. Opt. 11, 681–682 (1972).
Each of these methods has some drawbacks. TIR devices 16. J. W. Evans, “The birefringent filter,” J. Opt. Soc. Am. 39, 229–242
are bulky and must be mounted carefully to avoid stress bire- (1949).
fringence. Devices using crystals of different birefringent 17. V. A. Kucherov, V. S. Samoilov, and V. A. Chernavin, “Polymethyl
methacrylate achromatic phase plates,” Sov. J. Opt. Technol. 57(1),
dispersions are expensive in small quantities and the 66–67 (1990).
retardance can have a strong angle of incidence and angle 18. B. Tran and T. Baur, “Reactive mesogen retarders and applications,”
of azimuth dependence. Retarder stacks have some small Proc. SPIE 8489, 84890B (2012).
19. S. Pancharatnam, “Achromatic combinations of birefringent plates,”
wavelength dependence of the fast axis direction and have Indian Acad. Sci. Proc. 41, 137–144 (1955).
a lower laser damage threshold if made using polymers. 20. A. M. Title, “Improvement of birefringent filters. 2: Achromatic wave-
plates,” Appl. Opt. 14, 229–237 (1975).
Detailed discussion with a polarization optics manufacturer 21. R. C. Jones, “A new calculus for the treatment of optical systems I.
or reseller can be helpful in making the best choice for a Description and discussion of the calculus,” J. Opt. Soc. Am. 31,
particular application. 488–493 (1941).

22. F. Pockels, Lehrbuch der Kristalloptik, Teubner, Leipzig and Berlin 31. P. Hariharan and P. Ciddor, “Achromatic switchable polarization
(1906). rotators,” Opt. Eng. 36, 952–956 (1997).
23. A. M. Title and W. J. Rosenberg, “Achromatic retardation plates,” Proc. 32. P. Hariharan, P. Ciddor, and M. Roy, “Improved switchable achromatic
SPIE 307, 120–125 (1981). phase shifters, part 2,” Opt. Eng. 44, 105603 (2005).
24. M. Françon, S. Mallick, and J. Vulmière, “Use of apophyllite in polari- 33. S. Guimond and D. Elmore, “Polarizing views,” oemagazine (2004),
zation interferometers—achromatic half-wave plate,” J. Opt. Soc. Am. http://spie.org/newsroom/polarizing-views?SSO=1 (03 January 2017).
55, 1553–1553 (1965).
25. A. Uchiyama and T. Yatabe, “P-5: characteristics and applications of
new wide-band retardation films,” in SID 01 Digest, pp. 566–569 Michael Kraemer received his PhD in mathematics from the
(2001). University of Montana–Missoula in 2001 and his PhD in physics from
26. C. D. West and A. S. Makas, “The spectral dispersion of birefringence, Colorado State University in 2006. His current research interests
especially of birefringent plastic sheets,” J. Opt. Soc. Am. 39, 791–794 include numerical computations in polarization optics, optical meas-
(1949). uring techniques, and achromatic devices.
27. S. R. Sueoka, “Polarization optical components of the Daniel K. Inouye
Solar Telescope,” PhD thesis, University of Arizona (2016). Tom Baur is the founder and CTO of Meadowlark Optics. He has
28. P. D. Hale and G. W. Day, “Stability of birefringent linear retarders been developing polarization-related products for the company for
(waveplates),” Appl. Opt. 27, 5146–5153 (1988).
29. D. F. Elmore et al., “The Daniel K. Inouye Solar Telescope first light almost 40 years. He is the 2017 winner of the SPIE G.G. Stokes
instruments and critical science plan,” Proc. SPIE 9147, 9147071 Award and has a dual career in precision optics and cattle ranching
(2014). on the eastern plains of Colorado.
30. W. H. Schubert, E. Petrak, and T. G. Baur, “Measurement of polariza-
tion assemblies for the Daniel K. Inouye Solar Telescope,” Proc. SPIE
9369, 93690N (2015).


Achromatic Retarders in Polarization Optics: Michael Kraemer Tom Baur

Uploaded by

Document Informationclick to expand document informationAchromatic retarders in polarization optics

Document Informationclick to expand document information

Copyright:

Available Formats

Achromatic Retarders in Polarization Optics: Michael Kraemer Tom Baur

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Achromatic Retarders in Polarization Optics: Michael Kraemer Tom Baur

Uploaded by

Copyright:

Available Formats

Achromatic retarders in polarization

Downloaded From: https://www.spiedigitallibrary.org/journals/Optical-Engineering on 21 Apr 2019