Open Access
Add to BibSonomy
Abstract
Polarization imaging has become a widely-applied detection technique, due to the capabilities of enhanced image contrast and object recognition. Here, we demonstrate 320 × 256 InGaAs focal plane array (FPA) integrated with superpixel-structured subwavelength aluminum grating. An extinction ratio of up to 19:1 at 1310 nm is realized, which indicates a good capability of near-infrared polarization detection. Theoretical simulation shows a fairly high extinction ratio for such superpixel structure. This difference between the actual extinction ratio and the theoretical extinction ratio is further discussed by analyzing the effects of the alignment deviation and structural parameter deviations induced during the actual process. Moreover, the imaging results show that the fabricated polarimetric InGaAs FPA presents a more obvious profile for artificial objects, compared to the conventional detector. Such FPAs integrated with superpixel-structured grating are very promising for high performance polarization imaging in the short wavelength infrared band.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
As a basic property of the wave nature of light, polarization describes the direction of electromagnetic wave oscillating in space-time. When the natural light or the artificial light illuminates objects, the polarization of light caused by reflection from materials contains information about many intrinsic properties of the imaged object, such as the surface roughness [1] and geometry [2]. Therefore, there is an increasing demand in many scientific areas to measure the polarization state of the imaged objects in order to gain rich information. Characterization of the polarization state of a scene in various wavebands provides a more efficient method in identification of camouflaged vehicles [3], material classification [4,5], the detection of cancers [6,7], non-contact fingerprint detection [8,9], classification of pollutants in the atmosphere [10], observation in foggy condition [11] and three-dimensional shape recovery [12].
The polarimetric detectors can be divided into four categories including division-of-time, division-of-amplitude, division-of-aperture and division-of-focal-plane (DoFP) detectors. With the development of focal plane array (FPA) technology, DoFP polarimetric detector is widely used due to its compact structure and real-time detection for different polarization states [13–16]. The FPA of a DoFP detector is subdivided into blocks of superpixels, where each pixel in the superpixel contains a polarizer matched to its pitch [17]. Then Stokes vector can be gotten by the combination of measured signal intensities. However, there are relatively few studies on DoFP polarimetric detector for short wave infrared (SWIR), and the polarization performance needs to be further improved to better meet the demands of polarization detection.
In this paper, in order to detect the polarization information in SWIR, a 320 × 256 InGaAs FPA integrated with the superpixel-structured subwavelength aluminum grating is proposed. The test results of the polarization property of fabricated polarimetric detector show that an extinction ratio up to 19:1 at 1310 nm is realized, which is better than other reported results [18–20]. However, the actual extinction ratio is still obviously lower than the theoretical simulation results for such structure carried out by finite-difference time–domain (FDTD). Therefore, this difference is further discussed by simulating and analyzing the effects of the alignment deviation and the structural parameter deviations induced during actual process. Moreover, an image including SWIR polarization information is recorded by the fabricated polarimetric InGaAs FPA.
2. Design and fabrication of polarimetric InGaAs FPA
Figure 1(a) gives the sketch of designed InGaAs FPA integrated with superpixel-structured subwavelength grating. Each superpixel consists of 2 × 2 adjacent aluminum grating regions with four different angles (0°, 45°, 90°, 135°), as shown in the Fig. 2(b). InGaAs FPA mainly consists of InGaAs photosensitive chip and silicon-based ROIC. The InGaAs photosensitive chip, which is used for converting SWIR signal into electrical signal, is composed of InP substrate, InP buffer, InGaAs absorption region and InP cap. The silicon-based ROIC is electrically interconnected with photosensitive chip via the In bumps in order to read electrical signal.
Fig. 1 (a) Block diagram of the polarimetric InGaAs FPA. The region bounded by the dotted purple line represents a superpixel. (b) Structure of the polarimetric InGaAs FPA. (c) SEM image of the superpixel-structured aluminum grating.
For the back-illuminated InGaAs FPA, InP substrate was first thinned by chemical-mechanical polishing process to optimize the distance between photosensitive elements and polarization gratings, which contributed to reduce the optical crosstalk. Then, the SiO2 dielectric layer with 100 nm thickness was grown on InP substrate by inductively coupled plasma chemical vapor deposition (ICPCVD) process, in order to control the effects of surface plasmon polaritons (SPPs) resonance on transmittance and extinction ratio [21]. Finally, the aluminum grating with 100nm thickness was integrated onto the SiO2 layer by electron beam lithography and thermal evaporation process. The pitch and duty cycle of designed grating were 400 nm and 0.5, respectively. The morphology of fabricated superpixel-structured subwavelength aluminum grating was characterized by SEM as shown in Fig. 1(c). The superpixel-structured polarization grating was successfully integrated onto the InGaAs FPA, but the actual structural parameters were different from designed one due to process deviation, which had effects on the polarization performance of polarimetric InGaAs FPA, as discussed later.
3. Optical measurements of polarimetric InGaAs FPA
The response spectrum of polarimetric InGaAs FPA was characterized by a test system including a monochromator and a collimator. Then, the test result was calibrated by the response spectra of standard detector measured under the same conditions. The polarization performance was characterized by transmittance and extinction ratio. The polarization performance test system consisted of a monochromator, a collimator, a Glan calcite polarizer, a quarter-wave plate and a linear polarizer, as shown in Fig. 2. The incident angle of the linearly polarized light ranged from 0° to 180° by rotating the linear polarizer. The electrical signals collected by non-polarized and polarized photosensitive elements were calculated to obtain the relative transmittance of transverse magnetic (TM) wave and transverse electric (TE) wave. The ratio of maximum to minimum value of electrical signal was the extinction ratio of fabricated polarimetric detector.
The response spectrum of polarimetric InGaAs FPA is shown in Fig. 3(a). The fabricated detector shows a spectral response range from 0.9 μm to 1.7 μm and a transmission of more than 50%. The test results of polarization performance at 1310nm are shown in Fig. 3(b). With the rotation of linear polarizer, it can be seen that the signal received by photosensitive element integrated with grating is periodic, while the response signal of non-polarized photosensitive element is hardly affected, which indicates that InGaAs polarimetric detector has significant characteristics of polarization detection. For a certain incident angle of polarized light, as shown in Fig. 3(c), the four photosensitive elements in one superpixel show different response intensities as expected. In addition, the extinction ratio of polarimetric detector can be calculated by electrical signals collected by polarized photosensitive elements. For the four photosensitive elements corresponding to the aluminum gratings of 0°, 45°, 90° and 135°, the extinction ratios at 1310 nm are 18:1, 19:1, 16:1, 19:1, respectively. The extinction ratios at 1064 nm and at 1550 nm can be reach up to 14:1 and 22:1, respectively. This result is superior to the reported level of polarization detection for similar detectors [18–20], but it is still obviously lower than the simulated results. For the designed detector, the calculated extinction ratios at 1064 nm, 1310 nm and 1550 nm are about 46:1, 102:1 and 140:1, respectively. Such a significant gap can be attributed to the deviation between the fabricated detector and the designed one, therefore it is necessary to further discuss the influence of the process deviation on the polarization performance of fabricated polarimetric InGaAs FPA.
Fig. 3 Response spectrum of polarimetric InGaAs FPA. (b) Electrical signals collected by four neighboring pixels as a function of different angles of linearly polarized light at 1310 nm. (c) Intensity response of polarimetric InGaAs FPA for a certain incident angle of linearly polarized light.
4. Analysis of difference in polarization performance of polarimetric InGaAs FPA
4.1 Effects of alignment deviation
First, the alignment deviation between photosensitive elements and grating regions was discussed. For the superpixel structure designed in this paper, each photosensitive element should have a one-to-one correspondence with the grating region, and the spacing between every two adjacent grating regions is 2 μm. The photosensitive area S of each photosensitive element can be approximately 30 μm × 30 μm, and the corresponding grating area Sp is 28 μm × 28 μm as shown in Fig. 4(a). However, in the actual process, the thickness of InP substrate was still close to 100 μm after thinning and polishing. A certain deviation was introduced during the flip-chip bonding and the optical lithography process, which resulted in the alignment deviation between photosensitive elements and grating regions. The photoelectric performance test of the 320 × 256 InGaAs detector had been carried out. The results show that the operable pixel factor is more than 99% and inoperable pixels are randomly distributed. Therefore, alignment deviation of 0-4 μm could be considered during actual process, and the relative rotation angle between photosensitive elements and grating regions was extremely small, which could be negligible. The effect of alignment deviation on the polarization performance of polarimetric InGaAs FPA was analyzed as followed.
Within one superpixel, there may be four kinds of alignment deviation as shown in Figs. 4(b)-4(e). It is assumed that I is the electrical signal of non-polarized photosensitive element caused by linearly polarized light. For the photosensitive element corresponding to 0° polarization grating, its electrical signal Itol theoretically consists of two parts: the electrical signal Iunp generated by the light directly irradiating the non-grating region and the electrical signal Ip generated by the light transmitting through grating region on the photosensitive element. Therefore, for these four kinds of alignment deviation, the electrical signals Itol collected by one photosensitive element corresponding to the 0° polarization grating region at normal incidence can be calculated as:
where, Sunp is the area occupied by the non-grating region in this photosensitive element, and Si (i = 0, 45, 90, 135) is the area occupied by grating regions with different angles. Ti is the transmittance of metal grating polarizer. Ii is the electrical signal generated the light through grating regions with different angles.Based on the simulation results of FDTD solutions, the transmittance of metal grating at 1550 nm was extracted to analyze the influence of alignment deviation. When the angle θ between the linearly polarized light and the 0° metal grating was 90° and 0°, the influence of alignment deviation was calculated by the electrical signal of photosensitive element corresponding to 0° grating region, as shown in Fig. 5.
Fig. 5 The electrical signal Ip for (a) θ = 90° and (b) θ = 0° as a function of alignment deviation between the photosensitive elements and the polarization gratings.
When the angle θ is 90° the electrical signal Ip slowly reduces to 93% as the deviation increases along x-axis and the y-axis. However, the electrical signal Ip gradually increases to 10 times when the angle θ was 0°, which indicates that the electrical signal Ip corresponding to the transmission of TE wave is more sensitive to the alignment deviation. When the actual alignment deviation () exceeds 2 μm, the extinction ratio of InGaAs polarimetric detector significantly decreases (more than 50%). In order to reduce the influence of alignment deviation on the polarization performance of the detector, it is first necessary to ensure that photosensitive chip has good flatness during flip chip bonding, thereby reducing the alignment deviation between chip and ROIC to less than 1 μm. Secondly, the growth process of In bumps needs to be improved to ensure that the In bumps on both photosensitive chip and ROIC have good consistency, which is beneficial to avoid secondary displacement after bonding.
4.2 Effects of structural parameter deviations
The structural parameter deviations induced by the actual process also affect the performance of polarimetric InGaAs FPA, including pitch, thickness, duty cycle, angle, shape of the grating, etc. In order to find the main structural parameters that result in the degraded polarization performance, these structural parameter deviations were simulated by FDTD solutions to analyze their influence on transmittance and extinction ratio. Here, the extinction ratio based on the simulation results is the ratio of TM wave transmittance to TE wave transmittance. A broad-band plane with 0.9-1.7 μm wave source was launched normally to the aluminum gratings on the SiO2 layer. The frequency-dependent dielectric constant functions of aluminum and InP were fitted by multi-coefficient models (MCMs) [22]. Periodic boundary condition was used in x-y plane and perfectly matched layer (PML) boundary condition was chosen in z axis. A uniform cell with space step 1 nm was used in the range −0.2 μm<z<0.2 μm (including grating and SiO2 layer) and a nonuniform cell was used outside this range. A power monitor was inserted into InP layer directly below gratings to obtain transmission spectrum.
For the deviation range of grating height h1 within ± 10 nm, the simulation result is shown in Fig. 6(a). When h1 <100 nm, the transmittance of TM wave is less affected, but the transmittance of TE wave obviously increases. In the case of h1 = 90 nm, the transmittance of TE wave increases by about 13%, resulting in a decrease of more than 20% at the extinction ratio. Figure 6(b) shows the influence of angular deviation α of the grating on the polarization property. The results show that when│α│<1° the changes in transmittance and extinction ratio are both less than 3%. However, as │α│continues to increase, the TE transmission of wave increases rapidly. The longer the wavelength is, the more obvious the influence is, which leads to a reduction (about 10%) of extinction ratio. Meanwhile, for the superpixel structure, this influence of angular deviation can be transmitted to the Stokes vector, which affects the polarization degree of the detector. Hence, the influence of angular deviation needs to be corrected by data processing [23].
Fig. 6 The relationships between polarization performance and (a) height deviation, (b) angular deviation, (c) pitch deviation, and (d) duty cycle deviation. The first column on the left is the sketch for simulation, the second column and the third column are the extinction ratio and the transmittance, respectively.
The simulation was performed for the case where the pitch w1 deviation of grating is ± 40 nm, as shown in Fig. 6(c). The results show that when w1 >400 nm, the significant increase of TE wave transmittance leads to the decrease of extinction ratio (about 33% at w1 = 440 nm). While the duty cycle (w2/w1) is determined by width w2 and pitch w1 of the grating. Its deviation ranging from 0.45 to 0.55 was simulated by changing w2, as shown in Fig. 6(d). As w2 decreases, the transmittances of TM wave and TE wave increase simultaneously, but the change of TE wave is more noticeable, which causes the extinction ratio to rapidly decrease. When the duty cycle is 4.5, the extinction ratio reduces to about 65%. Figure 7 shows electric field distribution and Poynting vector distribution for TE wave at 1550 nm for three different structures. It can be found that when the spacing between two adjacent gratings (w3 = w1-w2) is larger than designed parameter (200 nm), the transmission of TE wave significantly increases, which is the major reason of the reduction of extinction ratio.
Fig. 7 Electric field distribution (upper part) and Poynting vector distribution (lower part) for TE wave at 1550nm for (a) designed structure, (b) duty cycle = 0.45, and (c) pitch = 440nm.
Moreover, since the metal grating was formed by thermal evaporation and lift-off process after electron beam lithography, its cross-sectional shape is difficult to present the standard rectangle. Therefore, it is necessary to simulate the influence of cross-sectional shape on the polarization performance. When the top of grating cross-section is curved and the bottom is rectangular, the influence of the height h2 of the curved top within 0-10 nm was simulated, as shown in Fig. 8(a). The increase of h2 causes the extinction ratio to gradually decrease. The decline of extinction ratio is about 11% when h2 is 10 nm. When the shape of the grating is trapezoid the sides of grating is inclined, and the influence on the polarization performance was simulated, as shown in Fig. 8(b). As changing the bottom width w3 of grating (w4 = 400-w3), there is less effect on the transmittance of TM wave and TE wave, which eventually leads to a less 5% decrease of the extinction ratio.
Fig. 8 The relationships between polarization performance and (a) height of the curved top and (b) bottom width of grating. The first column on the left is the sketch for simulation, the second column and the third column are the extinction ratio and the transmittance, respectively.
The actually thickness t1 of SiO2 dielectric layer grown by ICPCVD process is given in Table 1, and ± 10 nm is considered as the deviation range of t1. The simulation results, shown in Fig. 9, indicate that the extinction ratio is gradually reduced as the decreases of t1. When t1 is 90 nm, the extinction ratio is reduced by less than 6%.
Table 1. Actual Thickness of SiO2 Dielectric Layer
Fig. 9 (a) Sketch for simulation of thickness deviation of SiO2 layer. The relationships between thickness deviation of SiO2 layer and (b) the extinction ratio and (c) the transmittance.
From the above, for the structural parameter deviations induced during the actual process, the spacing w3 between two adjacent gratings is the major factor causing the reduction of polarization performance of InGaAs FPA. When w3 is larger than 200 nm, the increase of transmission of TE wave leads to the obvious reduction of extinction ratio.
5. Imaging result of polarimetric InGaAs FPA
Sample images recorded from the three kinds of imaging sensors are presented in Fig. 10. The imaged scene is mainly composed of plastic car models and plants. For these models made by artificial materials, the fabricated polarimetric InGaAs FPA presents more obvious profile compared to the two other, which contributes to enhance the image contrast.
Fig. 10 Sample images recorded from (a) CMOS imaging sensor, (b) conventional InGaAs FPA, and (c) fabricated InGaAs FPA with superpixel-structured grating.
6. Conclusion
In this paper, a 320 × 256 InGaAs FPA integrated with the superpixel-structured subwavelength aluminum grating was fabricated. The test results show that the polarimetric detector has a spectral response range from 0.9 μm to 1.7 μm and responds well to linearly polarized light with the angle from 0° to 180°. The extinction ratios of the detector at 1064 nm, 1310 nm and 1550 nm reach up to 14:1, 19:1 and 22:1, respectively, which indicates that it has a good detection capability of near-infrared polarization information. Meanwhile, the several factors of the process and structure were investigated in detail in order to improve the actual extinction ratio of the integrated polarization grating. It is found that the alignment deviation between photosensitive elements and metal grating regions mainly affects the transmittance of TE wave. When the actual alignment deviation exceeds 2 μm the transmittance of TE wave increases by nearly 3 times, which leads to the significant reduction of extinction ratio. In terms of the structural parameter deviations, the spacing w3 between two adjacent gratings is the major factor causing the reduction of polarization performance of InGaAs FPA. When w3 is larger than designed value, the increase of transmission of TE wave leads to the obvious reduction of extinction ratio. Moreover, the imaging results show that the fabricated InGaAs FPA presents more obvious profile for artificial objects compared to the conventional detector, which contributes to enhanced the image contrast.
Funding
National Key R&D Program of China (2016YFB0402401); National Natural Science Foundation of China (61704180, 61604159).
References
1. S. Hong, “Surface roughness and polarization ratio in microwave remote sensing,” Int. J. Remote Sens. 31(10), 2709–2716 (2010). [CrossRef]
2. D. K. Beamer, U. Abeywickrema, and P. Banerjee, “Polarization vector signatures for target identification,” Proc. SPIE 10407, 104070T (2017).
3. J. S. Tyo, D. L. Goldstein, D. B. Chenault, and J. A. Shaw, “Review of passive imaging polarimetry for remote sensing applications,” Appl. Opt. 45(22), 5453–5469 (2006). [CrossRef] [PubMed]
4. T. V. Thilak Krishna, C. D. Creusere, and D. G. Voelz, “Passive polarimetric imagery-based material classification robust to illumination source position and viewpoint,” IEEE Trans. Image Process. 20(1), 288–292 (2011). [CrossRef] [PubMed]
5. M. Sarkar, D. S. S. San Segundo Bello, C. van Hoof, and A. Theuwissen, “Intergrated polarization analyzing CMOS image sensor for material classification,” IEEE Sens. J. 11(8), 1692–1703 (2011). [CrossRef]
6. S. L. Jacques, R. Samatham, S. Isenhath, and K. Lee, “Polarized light camera to guidesurgical excision of skin cancers,” Proc. SPIE 6842, 68420I (2008). [CrossRef]
7. R. Patel, A. Khan, R. Quinlan, and A. N. Yaroslavsky, “Polarization-Sensitive Multimodal Imaging for Detecting Breast Cancer,” Cancer Res. 74(17), 4685–4693 (2014). [CrossRef] [PubMed]
8. 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] [PubMed]
9. L. Zhang, H. W. Yuan, and X. M. Li, “Active polarization imaging method for latent fingerprint detection,” Opt. Quantum Electron. 50(9), 353 (2018). [CrossRef]
10. D. Li, N. Zeng, Y. F. Wang, D. S. Chen, Y. R. Chen, and H. Ma, “Study on Polarization features of carbonaceous particles in atmosphere pollutants,” Proc. SPIE 9979, 99790H (2016).
11. C. L. Li, W. J. Lu, S. Xue, Y. C. Shi, and X. N. Sun, “Quality Assessment of Polarization Analysis Images in Foggy Conditions,” in Proceedings of IEEE Conference on Image Processing (IEEE, 2014), pp. 551–555. [CrossRef]
12. A. Sohaib, A. R. Farooq, J. Ahmed, L. N. Smith, and M. L. Smith, “3D reconstruction of concave surfaces using polarisation imaging,” J. Mod. Opt. 62(11), 927–932 (2015). [CrossRef]
13. V. Gruev, R. Perkins, and T. York, “CCD polarization imaging sensor with aluminum nanowire optical filters,” Opt. Express 18(18), 19087–19094 (2010). [CrossRef] [PubMed]
14. E. Gilboa, J. P. Cunningham, A. Nehorai, and V. Gruev, “Image interpolation and denoising for division of focal plane sensors using Gaussian processes,” Opt. Express 22(12), 15277–15291 (2014). [CrossRef] [PubMed]
15. D. Vorobiev and Z. Ninkov, “Design, fabrication and characterization of a polarization-sensitvie focal plane array,” Proc. SPIE 9403, 94030A (2015).
16. A. Abubakar, X. J. Zhao, S. T. Li, M. Takruri, E. Bastaki, and A. Bermak, “A Block-Matching and 3-D Filtering Algorithm for Gaussian Noise in DoFP Polarization Images,” IEEE Sens. J. 18(18), 7429–7435 (2018). [CrossRef]
17. T. York and V. Gruev, “Characterization of a visible spectrum division-of-focal-plane polarimeter,” Appl. Opt. 51(22), 5392–5400 (2012). [CrossRef] [PubMed]
18. X. Y. Hu, X. Wang, Z. Y. Ding, X. Lin, D. L. Zhang, and W. P. Wang, “Research Progress in Integrated Polarization Infrared Detector and Image Processing,” Proc. SPIE 10697, 106974I (2018).
19. G. Baker, M. Wilson, and P. Coulter, “Development and results of NIR polarization camera,” Proc. SPIE 6567, 65671L (2007). [CrossRef]
20. S. Ikeda, E. Higurashi, T. Suga, and T. Oguchi, “Miniaturized polarization sensors integrated with wire-grid polarizers,” in Proceedings of IEEE Conference on Electronics Packaging (IEEE, 2014), pp. 376–379. [CrossRef]
21. R. Wang, T. Li, X. Shao, X. Li, X. Huang, J. Shao, Y. Chen, and H. Gong, “Subwavelength Gold Grating as Polarizers Integrated with InP-Based InGaAs Sensors,” ACS Appl. Mater. Interfaces 7(26), 14471–14476 (2015). [CrossRef] [PubMed]
22. E. D. Palik, Handbook of Optical Constants of Solids (Academic, 1998).
23. F. Huang, F. M. Li, S. Y. Wang, C. S. Chen, and Y. B. Shi, “Calibration method for NIR polarization detector,” Hongwai Yu Haomibo Xuebao 37(3), 302–306 (2018).
