In this paper, a full-Stokes imaging polarimeter operating at 580 nm using an array of elliptical polarizers is presented. The division-of-focal-plane polarimeter utilizes a set of four optimized measurements which represent a regular tetrahedron inscribed in the Poincaré sphere. Results from the device fabrication, instrument calibration and characterization are presented. The performance of the optimized full Stokes polarimeter, as defined by size of the standard deviation of the degree of circular polarization, is found to be approximately five times better than the performance of the simple full-Stokes polarimeter.
© 2014 Optical Society of America
Optical imaging systems have a history of development spanning hundreds of years. Conventional cameras which measure the panchromatic intensity of an optical field and spectral imagers which measure the power spectrum are the common and have become mature and popular tools. In addition to intensity and wavelength, polarization is a third attribute of the optical field which can provide a large amount of important information. An imaging polarimeter can be used for sampling the polarization signature across a scene, and the recorded images can be quantified as Stokes vectors S, which consists of four elements S0, S1, S2, and S3. S0 represents the intensity of an optical field; S1 and S2 denote the affinity towards 0° and 45° linear polarization respectively; S3 expresses the difference between right and left circular polarizations. Using the Stokes vector S, the angle of linear polarization, degree of polarization (DOP), degree of linear polarization (DOLP), and degree of circular polarization (DOCP) across a scene can be derived and investigated. These four quantities have provided valuable information in biomedical imaging [1, 2], remote sensing [3, 4], material sciences , and interferometry .
In general, imaging polarimeters can be categorized as division of time (DoTP), amplitude (DoAmP), aperture (DoAP), and focal-plane (DoFP) [7–10]. A wide variety of configurations exists, and all instruments require a minimum of four measurements to calculate the complete Stokes vector. DoTP polarimeter commonly uses rotating polarization elements in the system, and each measurement is captured at different time frame. Thus, the scene must be stationary in order to avoid temporal blur. DoAmp polarimeter adopts multiple focal plane arrays (FPAs) to capture measurements simultaneously. However, the system is large, complex, and requires a highly rigid mounting structure. DoAP polarimeter consists of additional reimaging optics to form multiple polarization images on a single FPA simultaneously. It allows dynamic measurement, but the cost of reimaging optics is high, and the spatial resolution is reduced.
Compared with other techniques, DoFP polarimeter utilizes a micropolarizer array and has some advantages, such as robust design, small size and dynamic acquisition. The micropolarizer is integrated on top of the FPA, similar to the Bayer filter in color imaging. The polarimeter utilizes an image sensor with neighboring pixels covered by different polarization filters. The DoFP polarimeter has been achieved by incorporating wire-grid , aluminum nanowire , rubbed polyvinyl-alcohol (PVA)  and liquid crystal polymer (LCP) micropolarizers [14, 15] with an image sensor. In most cases, the micropolarizer consists of an array of 0°, 45°, 90°, and 135° linear polarizers; therefore, the polarimeters are only capable of measuring the linear components of the Stokes vector, S1 and S2. Simple full-Stokes DoFP polarimeters were first introduced using a LCP micropolarizer and LCP microretarder [16–18]. However, the accuracy of the DOLP and DOCP were not sufficiently high due to certain aspects of the polarimeter designs that are addressed in this work.
In this paper, an advanced full-Stokes DoFP imaging polarimeter is presented that optimizes the LCP micropolarizer design to improve the accuracy of the DOLP and DOCP. In section 2, the optimization of the four measurements is discussed, and the concepts are applied to the design of a micropolarizer. In section 3, both the fabrication processes and the optical properties of the micropolarizer are investigated. Section 4 presents the prototype and the calibration of the advanced full-Stokes DoFP imaging polarimeter. Finally, the performance evaluation and the measurement results are given in section 5, and the unique properties of the polarimeter are summarized in section 6.
2. Full-Stokes imaging polarimeter design
To calculate the complete Stokes vector, a minimum of four measurements are needed, and the four measurements cannot be coplanar as plotted on the Poincaré sphere. Sabatke, et. al.  have shown that, for a DoTP polarimeter, the optimized four measurements should form a regular tetrahedron inscribed in the Poincaré sphere [20, 21]. Since the function of a micropolarizer in a DoFP polarimeter is analogous to a spinning polarization element of a DoTP polarimeter, the same strategy is applicable in the design of the micropolarizer. The optimized design can be achieved using the combination of a uniform vertical polarizer and a microretarder, which consists of a set of four pixels. Each retarder has 132° retardance with fast axis angles of ± 15.1° and ± 51.7° respectively.
The Poincaré sphere representations for the two full-Stokes DoFP polarimeter designs are shown in Fig. 1. The linear micropolarizer of 0°, 45°, 90°, and 135° linear polarizers has four measurements on the equator of the Poincaré sphere. The design of the simple full-Stokes DoFP polarimeter is shown in Fig. 1(a). The combination of a micropolarizer with a microretarder allows three measurements on the equator and one measurement at the north pole. The complete Stokes vectors can be derived, but the design is unbalanced which results in varying signal to noise ratios (SNRs) for each of the four Stokes vector components . In general, the SNR of a polarimeter is proportional to the volume that the inscribed measurement points encompass. The unbalanced design of the simple full-Stokes DoFP polarimeter only occupies one fourth of the whole Poincaré sphere. Figure 1(b) shows the new optimized design on the Poincaré sphere, and the four measurements are located at the vertices of a regular tetrahedron which encompasses a much larger volume. This configuration affords a factor of 1.5 theoretical improvements in SNR and secondly the SNR is balanced for each component of the vector. It is important to note that more points on the Poincaré sphere would allow a greater volume to be encompassed along with a concomitant performance increase, but since more pixel elements would need to be dedicated for each polarization measurement, the DoFP polarimeter spatial resolution would suffer. On the other hand, the pixels for the polarization measurement sample a slightly different part of the scene; therefore, systematic errors can increase due to the differential gains of more pixels .
The four measurements represented by the regular tetrahedron can be realized by four elliptical micropolarizers, consisting of a uniform polarizer and a patterned microretarder layer. The fabrication process is simplified, and defect density of the micropolarizer decreases because only one patterned polarization layer, instead of two, is needed. For the elliptical micropolarizer fabrication, the LCP polarizer is adopted instead of the wire grid polarizer for two reasons. First, LCP polarizer requires fewer lithographic steps and tools compared with the fabrication of the wire grid polarizer. Secondly, the topography of the wire grid polarizer prevents the uniform coating and fabrication of an additional microretarder layer.
An optimized polarimeter design is shown in Fig. 2. Figure 2(a) shows the multilayer structure of the elliptical micropolarizer to be installed over the FPA sensor. The incident light enters through a transparent substrate on which the layers of the micropolarizer are formed. The FPA is comprised of a glass substrate, a microretarder, an isolation layer, and a uniform polarizer on top of a sensor. The orientations of the polarization elements are shown in Fig. 2(b). The orientation of the uniform polarizer is in vertical direction. The microretarder is comprised of a 2 × 2 macro pixel array. Each macro pixel has four 132° retarders with fast axis angles of 15.1°, −15.1°, 51.7°, −51.7° with respect to vertical direction as shown in the Fig. 2(b). The resultant elliptical micropolarizers A and B have DOLP of 0.928 while C and D have DOLP of 0.691. The angle of linear polarization of elliptical micropolarizers A, B, C, and D are −64.3°, 64.3°, 16.5°, and −16.5°; the DOCP are 0.383, −0.383, 0.808, and −0.808. The target operating wavelength is 580 nm. Each elliptical micropolarizer transmits a different elliptical polarization state that is captured as an intensity signal by its corresponding pixel on the sensor. The resultant elliptical micropolarizer design is shown in Fig. 2(c).
3. Micropolarizer fabrication and characterization
LCP is used for the fabrication of the uniform polarizer layer and the microretarder layer of the elliptical micropolarizer. Here, the LCP is photo-aligned using a process capable of producing linear polarization devices with feature sizes as small as 4 µm . The photo-alignment is achieved by using linearly polarized ultraviolet illumination (LPUV) to anisotropically cross-link the linearly photopolymerizable polymer (LPP). The orientation of LPUV defines the alignment direction of LPP. Patterned alignment domains can be defined using traditional photolithography techniques with a tunable linear polarized source. A Karl Suss MA6 Mask Aligner is used to align and to pattern the LPP layer. When aligning the LPP layer, a Bolder Vision Optics dichroic UV polarizer is adopted for generation of the LPUV illumination.
In this paper, the LPP material is supplied by Rolic Technologies Ltd. The LCP is model RMM141C manufactured by EMD Chemicals and is a dry powder that is added to chloroform (CHCl3) at a 20% weight-to-weight (w/w) ratio. The LCPs are reactive mesogens which can be permanently fixed as a solid plastic film by ultraviolet (UV) exposure. For polarizer fabrication, a dichroic dye is added to the LCP/solvent mixture at 12.5 mg/ml to introduce diattenuation in the LCP. The dichroic dye is model G-241 made by Hayashibara Biochemical Laboratories, Inc., and has peak absorption at 580 nm. The substrates are 100 mm diameter 0.5 mm thick fused silica double side polished wafers.
Kodak KAI-2020 monochromatic CCD (microlens version) which is comprised of a 1608 × 1208 7.4 μm square pixel array is utilized in the DoFP polarimeter, and therefore the size of each elliptical micropolarizer is also 7.4 μm square. The process begins with the fabrication of chromium alignment marks using an Edwards EB3 e-beam evaporator and a standard lift-off process. These alignment marks are critical for accurately defining each of the four unique microretarder orientation and for dicing of the finished wafers.
Figure 3 shows a schematic diagram of the steps required to fabricate the LCP elliptical micropolarizer. The microretarder layer is fabricated first. The coating of LPP layer consists of three steps: (1) dispensing LPP onto the wafer, (2) spin coating at 2500 rpm for 1 minute, and (3) soft baking at 170 þC for 5 minutes. The LPP patterning requires four selective LPUV exposures to register the four different domains, and each exposure generates one quadrant of the macro-pixel. Each LPUV exposure dose is 150 mJ/cm2 at 365 nm. After patterning the LPP layer, LCP is coated above the LPP layer. The method for coating LCP is (1) spin coating at 3750 rpm for 25 seconds, (2) hard baking at 55 þC for 2 minutes to remove residual solvent, and (3) curing with UV exposure of 200 mJ/cm2 at 365 nm. After curing, a solid plastic microretarder film is formed. Next, Norland optical adhesive NOA-81 is spin coated at 4000 rpm and UV cured as an isolation layer. Above the isolation layer, a uniform vertical polarizer is coated using LPP and LCP doped with dichroic dye. The second layer of LPP is again coated, baked, and exposed to the uniform LPUV. LCP doped with dichroic dye is then coated above the LPP layer. The spin coating speed is 2000 rpm which results in a thicker layer with higher extinction ratio. After 2 minutes of hard baking at 55 þC and UV curing with 3 J/cm2 at 365 nm, a uniform vertical polarizer film is fabricated above the microretarder. The resulting multilayers make up the elliptical micropolarizer.
After fabrication, polarization properties of the elliptical micropolarizer are characterized by a calibrated Mueller matrix imaging DoTP polarimeter with spatial resolution of 1 µm . The measured linear diattenuation, linear diattenuation orientation, and circular diattenuation at 580 nm are shown in Fig. 4 along with two sets of cross-section data . The dashed line passes through elliptical micropolarizers A and C while the solid line passes through elliptical micropolarizers B and D. The linear diattenuation alternates between 0.25 and 0.75, and the magnitude of circular diattenuation alternates between 0.85 and 0.45. The diattenuation orientation of dashed cross-section alternates between 65° and −15°, and solid cross-section alternates between −65° and 15°. The fabricated elliptical micropolarizer is measured to be more circular polarized than the target design. The differences are attributed mainly to the deviation from the 132° retardance and secondly to the orientation misalignments in the LCP of the microretarder layer. The magnitude of diattenuation is reduced by the presence of depolarization which comes from scattering in the LCP layer and stress birefringence from the isolation layer.
4. Optimized full-Stokes DoFP imaging polarimeter
Figure 5(a) shows a micrograph of the elliptical micropolarizer illuminated with vertically polarized light and imaged using a 100X 0.4 NA microscope objective. The microscope optics has inherent diattenuation, especially at this high NA, and the exact transmission is not representative. The micrograph here can serve as a good reference for defect inspection and alignment quality. Imperx ICL-B1620 camera is adopted for construction of the polarimeter. The camera body with the KAI-2020 CCD is shown in Fig. 5(b) and operates at a resolution of 1608 × 1208 pixels (14-bit digitization) and at a speed of 20 frames/second. After the finished wafer is diced, the individual die is bonded to a mounting frame, and a 6-axis stage with polarized light illumination is used to align the mounted die to the CCD array. When the die is aligned properly, UV curing epoxy is applied to fix the whole package. The mounted elliptical micropolarizer is shown in Fig. 5(c). Once the micropolarizer and camera body are fully assembled, a one inch diameter 580 nm bandpass filter with 10 nm FWHM is installed between the C-mount lens and the micropolarizer. The optimized full-Stokes DoFP imaging polarimeter attached with a C-mount lens (Computar H6Z0812 8-48 mm f/1.2 6X) is shown in Fig. 5(d).
The polarimeter must first be calibrated at the target wavelength of 580 nm. The optical setup for the calibration is shown in Fig. 6. A 580 nm collimated light source is generated using a tungsten lamp, a 580 nm bandpass filter, and a collimating lens. The polarization of the collimated light is modified using a linear polarizer and a linear retarder of 89.6° retardance at 580 nm. Different polarizations of light at normal incidence illuminate the DoFP polarimeter without lens. The captured images are utilized for computing the calibration matrices. The conventional polarimetric data reduction matrix method is adopted here for the calibration [25, 26].
With the DoFP polarimeter, the image field is divided into the discrete array elements that require calibration. The calibration matrices are functions of position with each pixel having a discrete position corresponding to the CCD array. A macro pixel comprising the four neighboring pixels, , , , and , is used to derive the corresponding calibration matrix. In general, for the polarization image Stokes vectors S, the measured flux as a function of discrete position , is the result of the inner productFig. 6, 0°, 45°, 90°, 135°, nearly right and left circular uniformly polarized images can be generated, and can be calculated from a flux matrix of calibration as
After is obtained, the pseudo inverse of can be utilized to calculate the estimated incident polarization of a measured image asEquation (4) with a complete set of matrices is then utilized to calculate the polarization signature of an optical field in the following sections.
In this work, the calibration of the polarimeter assumes most of the light rays are at normal incidence to the micropolarizer array. This is an accurate description for long focal length lenses and telecentric lenses operating at large f-number. Alternatively, a more sophisticated and precise calibration method will include the effects of light rays at different angles. The camera attached to a lens can be calibrated and is measured for different focal length, f-number and image distance.
5. Polarimeter characterization
An estimation of the accuracy and precision of the DOLP and DOCP can be made by uniformly illuminating the polarimeter with carefully controlled polarized light. The results can be used to evaluate the performance of the optimized full-Stokes DoFP polarimeters and to compare with the unbalanced design of the simple full-stokes DoFP polarimeter (Fig. 7). The test optical setup is identical to the calibration configuration in Fig. 6, and the ellipticity of the input polarization is varied by rotating an 89.1° retarder in front of a 0° linear polarizer at 580 nm. The retarder rotates from 0° to 180° with 5° increment. The average DOLP (black) and DOCP (red) of the uniformly polarized input with one standard deviation error bars are shown in Fig. 7(a). The solid lines represent the theoretical DOLP (black) and DOCP (red) for the corresponding configuration. The error in the accuracy of the simple and optimized polarimeters is shown in Fig. 7(b). The maximum errors of DOLP and DOCP of the optimized full-Stokes polarimeter are 7%. For comparison, the maximum errors of DOLP and DOCP of the simple full-Stokes polarimeter are about 23% . The improvement in accuracy is attributed to the optimized elliptical micropolarizer design and the reduced defect density of the micropolarizer resulting from the simplified fabrication process.
In Fig. 7(c), the standard deviations of DOLP and DOCP of the optimized full-Stokes polarimeter, which represent the uncertainty of the polarimeter, are shown to be less than 5%. As expected, the simple full-Stokes polarimeter has a larger variation in the standard deviation, caused mainly by the reduced coverage of the Poincaré sphere as shown in Fig. 1. The performance of the optimized full Stokes polarimeter, as defined by the size of standard deviation of DOCP, is approximately five times better than the performance of the simple full-Stokes polarimeter. The optimized design has higher SNR than the imbalanced design. The remaining standard deviation is attributed to cross-talk between different microretarder elements, the deviation of the fabricated elliptical polarizer from a perfect elliptical polarizer, and the finite bandwidth of the 580 nm bandpass filter which cannot be easily removed by calibration. In addition, the accuracy of the calibration as well as DOLP and DOCP measurements are affected by orientation errors in the calibration, since the rotation of the polarizer and retarder (Fig. 6) are performed manually with an approximate accuracy of about 0.5°. For some applications such as microscopy, the performance of the polarimeter can be further improved by illuminating the object of interest by a monochromatic laser. In this case, the bandpass filter can be removed from the polarimeter setup.
An image of a calcite crystal and a Plusiotis optima beetle is captured with the optimized full-Stokes DoFP imaging polarimeter using a Computar lens at f/5.6. The measurement results are shown in Fig. 8. The birefringent calcite crystal spatially separates the two eigen-polarizations resulting in image splitting of the printed letters. The Plusiotis optima beetle has an external surface consisting of chitinous cuticle that possesses microscopic molecular-level helicity . The helicity of the beetle’s exoskeleton results in the circular polarization properties which have been studied by using different polarized light sources and observation angles . From the plots of S2 and linear angle, the two orthogonal eigen-polarizations of the calcite crystal can be observed and analyzed. In the plot of S3 and DOLP, the Plusiotis optima beetle shell has a DOCP up to 0.8, and the sides of the beetle have a DOLP up to 0.4 due to the polarization sensitivity of large angle Fresnel reflections.
The optimized full-Stokes DoFP polarimeter can provide a 1608 × 1208 pixel polarization image at 20 frames/second. Figure 9 shows one frame from a polarization video which is recorded using a Computar C-mount lens at f/5.6. The test target is a beam chopper with inner and outer open windows. Linear polarizers oriented radially are installed in the outside windows, while right and left circular polarizers are installed in the inner windows. The polarimeter successfully measures the complete polarization signature of the moving target. The noise inside the uniform polarizers is attributed to fabrication defects and LCP transition areas in the micropolarizer which cannot be removed by calibration.
In this paper, a visible full-Stokes DoFP imaging polarimeter using an optimized LCP elliptical micropolarizer design is demonstrated and its fabrication and calibration process described. The optimized elliptical micropolarizer is comprised of a substrate, a patterned LCP microretarder layer, an isolation layer, and a uniform LCP polarizer. The macro pixel element utilizes a set of four optimized elliptical polarizers which form a regular tetrahedron inscribed in the Poincaré sphere. Compared to the simple full-Stokes design, this optimized design affords a theoretical improvement factor of 1.5 in SNR. The fabrication process for the optimized design is simpler resulting in a lower defect density. The optical properties of the elliptical micropolarizer designs are characterized by a Mueller matrix imaging polarimeter.
The assembly, calibration, and measurement results of the novel optimized polarimeter are discussed. The optimized full-Stokes DoFP polarimeter provides a resolution of 1608 × 1208 pixels at 20 frames/second. The errors of DOLP and DOCP are less than 7%, and the uncertainties, defined by the standard deviations of DOLP and DOCP, are less than 5%. The optimized polarimeter design performs about five times better than the simple polarimeter design. To the best of our knowledge, the optimized polarimeter is the first prototype utilizing an all elliptical micropolarizer design. Future improvement in SNR can be achieved by incorporating a high performance bandpass filter with a narrower bandwidth, by reducing fabrication defects, and by decreasing the microretarder element size to lower pixel to pixel cross-talk. In our experiment, the conventional polarimetric data reduction matrix method is adopted for our calibration. Data quality and resolution can be further improved by using interpolation [29, 30] and deconvolution techniques .
This work is funded by the Tech Launch Arizona Proof of Concept Award and the Arizona Technology Research Infrastructure Fund (TRIF). The authors thank Prof. Russell Chipman’s and Prof. Nassar Peyghambarian’s research groups for the use of their equipment.
References and links
1. K. M. Twietmeyer, R. A. Chipman, A. E. Elsner, Y. Zhao, and D. VanNasdale, “Mueller matrix retinal imager with optimized polarization conditions,” Opt. Express 16(26), 21339–21354 (2008). [CrossRef]
2. J. Soni, S. Chandel, J. Jagtap, A. Pradhan, and N. Ghosh, “Mueller matrix polarimetry in fluorescence scattering from biological tissues,” in Frontiers in Optics 2013, I. Kang, D. Reitze, N. Alic, and D. Hagan, eds., OSA Technical Digest (online) (Optical Society of America, 2013), paper FW5A.3.
4. W. L. Hsu, S. Johnson, and S. Pau, “Multiplex localization imaging and sub-diffraction limited measurement,” J. Mod. Opt. 60(5), 414–421 (2013). [CrossRef]
5. S. Tominaga and A. Kimachi, “Polarization imaging for material classification,” Opt. Eng. 47(12), 123201 (2008). [CrossRef]
6. M. Novak, J. Millerd, N. Brock, M. North-Morris, J. Hayes, and J. Wyant, “Analysis of a micropolarizer array-based simultaneous phase-shifting interferometer,” Appl. Opt. 44(32), 6861–6868 (2005). [CrossRef]
7. Y. Zhang, H. Zhao, and N. Li, “Polarization calibration with large apertures in full field of view for a full Stokes imaging polarimeter based on liquid-crystal variable retarders,” Appl. Opt. 52(6), 1284–1292 (2013). [CrossRef]
9. F. Afshinmanesh, J. S. White, W. Cai, and M. L. Brongersma, “Measurement of the polarization state of light using an integrated plasmonic polarimeter,” Nanophotonics 1(2), 125–129 (2012). [CrossRef]
17. G. Myhre, W. L. Hsu, A. Peinado, C. LaCasse, N. Brock, R. A. Chipman, and S. Pau, “Liquid crystal polymer full-Stokes division of focal plane polarimeter,” Opt. Express 20(25), 27393–27409 (2012). [CrossRef]
18. X. Zhao, A. Bermak, F. Boussaid, and V. G. Chigrinov, “Liquid-crystal micropolarimeter array for full Stokes polarization imaging in visible spectrum,” Opt. Express 18(17), 17776–17787 (2010). [CrossRef]
19. D. S. Sabatke, M. R. Descour, E. L. Dereniak, W. C. Sweatt, S. A. Kemme, and G. S. Phipps, “Optimization of retardance for a complete Stokes polarimeter,” Opt. Lett. 25(11), 802–804 (2000). [CrossRef]
23. J. Pezzaniti and R. Chipman, “Mueller matrix imaging polarimetry,” Opt. Eng. 34(6), 1558–1568 (1995). [CrossRef]
24. S. Y. Lu and R. Chipman, “Interpretation of Mueller matrices based on polar decomposition,” J. Opt. Soc. Am. A 13(5), 1106–1113 (1996). [CrossRef]
25. R. Chipman, “Polarimetry,” in OSA Handbook of Optics (McGraw-Hill, 1995).
28. H. Arwin, R. Magnusson, J. Landin, and K. Järrendahl, “Chirality-induced polarization effects in the cuticle of scarab beetles: 100 years after Michelson,” Philos. Mag. 92(12), 1583–1599 (2012). [CrossRef]