We report an imaging sensor capable of recording the optical properties of partially polarized light by monolithically integrating aluminum nanowire optical filters with a CCD imaging array. The imaging sensor, composed of 1000 by 1000 imaging elements with 7.4μm pixel pitch, is covered with an array of pixel-pitch matched nanowire optical filters with four different orientations offset by 45°. The polarization imaging sensor has a signal-to-noise ratio of 45dB and captures intensity, angle and degree of linear polarization in the visible spectrum at 40 frames per second with 300mW of power consumption.
©2010 Optical Society of America
Solid-state imaging devices have become the prevalent choice in the multi-billion dollar industry of digital photography. Consumer demands for high pixel count imaging sensors have set trends for small pixel pitch sensors, high signal-to-noise ratios, high dynamic range and better color replication of the imaged scene among others . These trends have resulted in the proliferation of imaging sensors in our daily lives that capture two properties of light, namely the intensity and the wavelength, and encode this information into perceptual quantities of brightness and color. The third property of light, polarization [2,3], has been ignored by solid-state imaging sensors partially due to the human inability to discriminate polarization information.
Capturing the optical properties of partially polarized light has proven to be very useful in gaining additional visual information in optically scattering environments, such as target contrast enhancement in hazy/foggy conditions , depth map of the scene in underwater imaging  and in normal environmental conditions, such as classifications of chemical isomers , classifications of pollutants in the atmosphere [7,8] and non-contact fingerprint detection  among others. In addition, polarization imaging sensors have found a niche in many biomedical applications, such as imaging for early skin cancer detection  and retinal surgery .
In order to capture the optical properties of partially polarized light, three parameters are of importance: the intensity of the light wave, the angle of polarization (AoP) and the degree of linear polarization (DoLP). There are different ways of computing DoLP and AoP of the electric-field vector, one of which is presented by Eq. (1) and (2).
In Eqs. (3) through (5), I(0°) is the intensity of the E-vector filtered with a 0 degree linear polarization filter; I(45°) is the intensity of the E-vector filtered with a 45 degree linear polarization filter and so on. In order to compute the first three Stokes parameters, the incoming light wave must be filtered with four linear polarization filters offset by 45°. Hence, an imaging sensor capable of characterizing partially polarized light has to employ four linear polarization filters offset by 45° together with an array of imaging elements.
Recovering polarization information in the visible spectrum has typically been implemented using standard CMOS or CCD imaging sensors coupled with electrically controlled polarization filters . These imaging systems, known as division of time polarimeters, filter the imaged environment with four time multiplexed polarization filters offset by 45°, record the total intensity of the filtered light wave and extract polarization information, such as angle and degree of linear polarization, using standard digital image processing algorithms. Although this is the most prevalent sensory approach toward polarization imaging today, the reduction of frame rate by a factor of three and inaccuracy of extracted polarization information due to motion in the scene are major shortcomings.
Monolithic integration of micropolarizaiton filters with CMOS imaging sensors is currently a subject of intense research [12–16]. These sensors, known as division of focal plane polarimeters, include silicon photodetectors and micropolarization filters on the same substrate, where the incoming light wave is filtered with spatially distributed micropolarization filters over neighborhoods of pixels. The amplitude of the filtered light wave is recorded by the underlying photodetectors and polarization parameters are computed based only on measured intensity information. Incorporating pixel-pitch-matched polarization filters at the focal plane for visible spectrum has been explored with birefringent materials  and thin-film polarizers [14–16]. However, the small imaging array size , large pixel size  and small extinction ratios  have limited the applications for these polarimeters. Another main disadvantage of the thin-film polymer polarization imaging sensors has been the complex fabrication techniques required to merge CMOS technology with polymer filters. These fabrication steps are prone to misalignment errors, which can degrade the sensing capabilities of the polarization imaging sensor. Furthermore, the thick multilayer filter array, which is around 10 micron of thickness, is prone to optical crosstalk and limits the extinction ratios of the sensors .
One of the main challenges in realizing division of focal plane polarimeters for the visible spectrum is the design of a micropolarization array. The purpose of the micropolarization array is to filter the incoming light wave prior to the absorption by the silicon photodiode. A successful candidate for a micropolarization array must satisfy four basic criteria. First, the array must be able to filter the incoming light wave with four polarization filters offset at 45°. Second, the pixel-pitch of the micropolarization filter has to match the pixel-pitch of the underlying photodetector array. Third, the extinction ratios of the polarization filter i.e. the ratio of the parallel polarized light to cross polarized light has to be 20 or higher in order to be able to extract useful polarization information [4–9]. Finally, the crosstalk between neighboring pixels, i.e. mixing of polarization information between neighboring pixels has to be minimized. If a filtered light wave from one pixel of the micropolarization array is recorded by a neighboring photodiode of the image sensor, then the extinction ratios are degraded and polarization information is potentially lost.
We sought an approach for an integrated polarization imaging sensor that would satisfy the four criteria mentioned above. We hypothesized that high accuracy of the extracted polarization information can be achieved by directly depositing aluminum nanowires on top of the imaging sensor as close as possible to the photosensitive element. Fabricating these metallic nanowires with sub-wavelength dimensions has become feasible with the maturity of nanofabrication technologies and techniques . In this paper, we present a high resolution division of focal plane polarization imaging sensor. The imaging sensor monolithically combines CCD imaging elements with aluminum nanowire polarization filters.
2. Imaging sensor overview
We present a block diagram of the integrated CCD polarization imaging sensor in Fig. 1 . The imaging sensor contains an array of 1000 by 1000 buried channel vertical CCDs, a 1000 buried channel horizontal CCD register and a charge to voltage conversion amplifier. Each pixel consists of a photodiode with a well capacity of 20K electrons and two light shielded buried channel CCDs. The read-out noise of the CCD imaging sensor is 16e- and has conversion gain of 30μV/e-. An array of pixel-pitch matched aluminum nanowire polarization filters covers the CCD array of photo elements. Figure 1(a) shows the pattern of the micropolarization filter array, which contains 4 distinct filters offset by 45°. The aluminum nanowires in each individual filter of the micropolarization array are 70nm wide, 70nm high and have a pitch of 140nm. Figure 1(b) presents an SEM image of the aluminum nanowires oriented at 45°.
The nanofabrication of the optical nanowire polarization filter is achieved by first depositing a 70nm thin film of aluminum followed by 30nm thin film deposition of SiO2 using e-beam evaporation. A 100nm thin layer of photoresist S-1805 is spin coated at 3000 rpm and baked at 115°C for 60 seconds. A continuous-wave single frequency Nd:YAQ laser with 532 nm wavelength is used together with frequency-doubler to produce coherent light waves at 266 nm wavelength. Two continuous-waves with 266 nm wavelength are aligned to interfere at 110 degree and produce an interference pattern with a period of 140nm. The interference pattern is transferred to the photoresist by exposing the sample for 40 seconds. After developing the photoresist, the pattern is transferred to the SiO2 using standard RIE/ICP etching recipe for SiO2. The SiO2 is used as a hard mask for etching the aluminum. The aluminum is etched for 150 seconds using 30sccm BCl3, 15sccm Cl2, 10mTorr pressure, 70°C temperature, 100W RIE power and 150W ICP power. The procedure is repeated four times, where the sample is rotated by 45° each time in order to produce nanowires with 4 different orientations.
3. Optical measurements of the polarization imaging sensor
We characterized the CCD polarization imaging sensor for responsivity, accuracy of the measured degree and angle of polarization and extinction ratios. The imaging sensor is tested with a uniform and collimated light, where the degree and angle of linear polarization of the incident light is controlled via a computer.
The optical performance of the imaging sensor is evaluated at 625nm ± 5nm, 515 ± 5nm and 460 ± 5nm. A 4” integrating sphere with three ports is used to produce uniform light intensity. Two arrays of narrow band and high intensity LEDs are used as inputs to the integrating sphere. A 100μm pinhole is placed at the output port of the integrating sphere and a 40mm condensing lens is used to produce collimated light. The imaging sensor is placed on a computer-controlled rotating stage and is rotated between 0° and 30° in increments of 2°. Hence, the incident angle of the incoming light to the surface of the imaging sensor is varied in a systematic manner. For every incident angle, the angle of polarization of the incident light is swept from 0° to 180° in increments of 1° via a computer controlled linear polarization filter. The degree of polarization of the incident light is modulated via a computer controlled variable liquid crystal phase retarder between 0 (unpolarized light) to 1 (linearly polarized light) in increments of ~0.1. The reference angle and degree of linear polarization is obtained with a single calibrated photodetector and mechanically rotating linear polarization filter.
Figure 2 illustrates the responsivity of four neighboring pixels with different polarization filters as a function of the polarization angle of the incident linearly polarized light.
All four pixels follow Malus’ law for polarization irradiance and are offset by 45° due to the physical orientations of the aluminum nanowires. Hence, the maximum and minimum transmissions between two neighboring pixels are shifted by 45°. The maximum transmission for the polarization pixels occurs when parallel polarized light illuminates the photodiode and corresponds to 64% of the total incident intensity. The minimum transmission is 1.1% of the total incident intensity and is recorded when cross polarized light illuminates the photodetector.
Figure 3(a) and Fig. 3(b) depict the computed angle and degree of linear polarization respectively, as a function of the polarization properties of an incident reference light wave. Both angle and degree of linear polarization are computed across the entire imaging array on a neighborhood of 2 by 2 pixels, and the mean values as well as its standard deviations are presented. The linear fit error for the computed angle and degree of linear polarization are 0.2% and 0.5%, respectively. The root-mean-square value of the standard deviations for each angle and degree of linear polarization computed on all pixels across the imaging array are 0.08° and 0.02 respectively.
The extinction ratios across the imaging array, defined as the ratio of the maximum to the minimum transmission, are evaluated and shown in Fig. 4 . Figure 4(a) presents the measured extinction ratios for three different wavelengths as a function of integration time. The maximum extinction ratio is achieved for an incident light wavelength of 625 ± 5nm and is 58 ± 1.2 across the imaging array. The extinction ratios for green (515 ± 5nm) and blue (460 ± 5nm) wavelengths are 44 ± 0.7 and 30 ± 0.5 respectively. Furthermore, the extinction ratios for each wavelength are constant for integration periods between 3 msec and 24 msec.
Figure 4(b) presents the measured extinction ratios as a function of the incident angle of the reference light wave on the surface of the imaging sensor. The incident angle of the impingent light wave is swept between 0° and 30° in increments of 2°. For incident light (625 ± 5nm wavelength) perpendicular to the surface of the imaging sensor, the extinction ratios are maximum and equal to 58. The extinction ratio remains constant as the incident angle of the light wave is increased to 5° and rapidly decreases for higher incident angles. For incident angles greater than ~20°, the polarization information is completely lost due to the pixel crosstalk.
4. Application results from the polarization imaging sensor
The indices of refraction from five flat surfaces are estimated using optical information recorded with the CCD polarization imaging sensor. The five materials under test are: fused silica, polymethyl methacrylate, silicon, gallium arsenide and silicon carbide. The degree of polarization of the reflected light wave from the surface of the test material is a function of the index of refraction and the incident angle. The maximum degree of linear polarization for a given material with index of refraction n1 is achieved at the incident angle known as the Brewster angle. Once the Brewster angle (ΘΒ) is determined, the refractive index of the material can be computed as n1 = tanΘΒ.
In order to determine accurately the Brewster angle, all five samples have flat surfaces and are mounted on a calibrated tilting stage. As the samples are tilted between 0° and 180° in increments of 0.1°, the polarization imaging sensor records the degree of linear polarization of the reflected light wave. The Brewster angle is equal to the incident angle of light when the degree of linear polarization recorded by the CCD polarization sensor is 1. Table 1 presents the measured and theoretical values of the Brewster angles for the five test samples. The deviations of the experimental value from the theoretical values are less than 3.8° and should be further explored and compared to other measuring techniques of the index of refraction published in the literature .
Sample images recorded from the polarization CCD imaging sensor are presented in Fig. 5 (Media 1). The imaged scene is composed of six linear polarization filters offset by 30° and a black plastic horse figure against a white background. The intensity variations along the plastic figure are minimal and the shape of the figure is difficult to determine in the intensity image presented in Fig. 5(a).
The degree of linear polarization image is presented in Fig. 5(b), where bright and dark values indicate high and low degrees of linear polarization in the scene respectively. The linear polarization filters have high degree of polarization due to the intrinsic properties of the filters. Figure 5(c) presents the angle of linear polarization of the imaged scene. The angles of polarization for the six linear polarization filters are presented in different colors offset by 30°. The angle and degree of linear polarization along the plastic figure exhibit smooth variations due to the curvature of the figure, i.e. surface normal and intrinsic properties of the object, i.e. index of refraction of the object. This sensor captures intrinsic information about the imaged environment, i.e. surface curvature and index of refraction, represented in polarization data space in high resolution and in real-time, which can bring advancements in many research areas.
To our knowledge, in this paper we report the first high resolution, real-time division of focal plane polarization imaging sensor in the visible spectrum. The imaging sensor presented in this paper synergistically combines nanowire technology with CCD imaging technology in order to realize a new type of sensor capable of recording the optical properties of partially polarized light. Table 2 presents a summary of the imaging sensor characteristics.
The integrated polarization sensor computes angle, degree of linear polarization and intensity information at 40fps and consumes 300mW of power. The measured signal-to-noise ratio of the polarization imaging sensor is 45dB and the dynamic range is 65dB. The fixed pattern noise without polarization filters is 1.9% of saturated level and with polarization filter is 2.4% of the saturated level. The polarization filters with extinction ratios of 58 can be used to enhance the intensity information of the imaged environment as well as determine the index of refraction of flat surfaces.
The work described in this paper was supported by National Science Foundation NSF grant number 0905368, Air Force of Scientific Research grant number FA9550-10-1-0121 and the Center for Material Innovation at Washington University in St. Louis.
References and links
1. E. R. Fossum, “CMOS image sensors: electronic camera-on-a-chip,” IEEE Trans. Electron. Dev. 44(10), 1689–1698 (1997). [CrossRef]
2. D. Goldstein, “Polarized Light,” (Marcel Dekker: New York, NY, 2003).
4. S. Shwartz, E. Namer, and Y. Schechner, “Blind Haze Seperation,” Proc. IEEE Comp. Vision and Pat. Recog. 2, 1984–1991 (2006).
6. W. Plucknett and R. Dowd, “Refraction and Polarization Properties of Binary Solutions of the Nitrotoluene Isomers with the Xylene Isomers, Chloroform and Cyclohexane,” J. Chem. Eng. Data 8(2), 207–210 (1963). [CrossRef]
7. J. L. Deuzé, F. M. Bréon, C. Devaux, P. Goloub, M. Herman, B. Lafrance, F. Maignan, A. Marchand, F. Nadal, G. Perry, and D. Tanré, “Remote sensing of aerosols over land surfaces from POLDER-ADEOS-1 polarized measurements,” J. Geophys. Res. 106(D5), 4913–4926 (2001). [CrossRef]
8. E. Puttonen, J. Suomalainen, T. Hakala, and J. Peltoniemi, “Measurement of Reflectance Properties of Asphalt Surfaces and Their Usability as Reference Targets for Aerial Photos,” IEEE Trans. Geosci. Rem. Sens. 47(7), 2330–2339 (2009). [CrossRef]
9. S. S. Lin, K. M. Yemelyanov, E. N. Pugh Jr, and N. Engheta, “Polarization-based and specular-reflection-based noncontact latent fingerprint imaging and lifting,” J. Opt. Soc. Am. A 23(9), 2137–2153 (2006). [CrossRef]
10. E. Salomatina-Motts, V. A. Neel, and A. N. Yaroslavskaya, “Multimodal polarization system for imaging skin cancer,” Opt. Spectrosc. 107(6), 884–890 (2009). [CrossRef]
11. B. I. Gramatikov, O. H. Zalloum, Y. K. Wu, D. G. Hunter, and D. L. Guyton, “Directional eye fixation sensor using birefringence-based foveal detection,” Appl. Opt. 46(10), 1809–1818 (2007). [CrossRef] [PubMed]
12. A. Andreou and Z. Kalayjian, “Polarization imaging: principles and integrated polarimeters,” IEEE Sens. J. 2(6), 566–576 (2002). [CrossRef]
14. T. Tokuda, S. Sato, H. Yamada, K. Sasagawa, and J. Ohta, “Polarisation-analysing CMOS photosensor with monolithically embedded wire grid polarizer,” Electron. Lett. 45(4), 228–230 (2009). [CrossRef]
15. X. Zhao, F. Boussaid, A. Bermak, and V. G. Chigrinov, “Thin Photo-Patterned Micropolarizer Array for CMOS Image Sensors,” IEEE Photon. Technol. Lett. 21(12), 805–807 (2009). [CrossRef]
17. J. Wang, F. Walters, X. Liu, P. Sciortino, and X. Deng, “High-performance, large area, deep ultraviolet to infrared polarizers based on 40 nm line/78 nm space nanowire grids,” Appl. Phys. Lett. 90(6), 061104 (2007). [CrossRef]