Open Access
Add to BibSonomy
Abstract
Snapshot multispectral imaging (MSI) has been widely employed in the rapid visual inspection by virtues of the non-invasive detection mode and short integration time. As the critical functional elements of snapshot MSI, narrowband, customizable, and pixel-level multispectral filter arrays (MSFAs) that are compatible with imaging sensors are difficult to be efficiently manufactured. Meanwhile, monolithically integrating MSFAs into snapshot multispectral imagers still remains challenging considering the strict alignment precision. Here, we propose a cost-efficient, wafer-level, and customized approach for fabricating transmissive MSFAs based on Fabry-Perot structures, both in the pixel-level and window-tiled configuration, by utilizing the conventional lithography combined with the deposition method. The MSFA chips own a total dimension covering the area of 4.8 mm × 3.6 mm with 4 × 4 bands, possessing the capability to maintain narrow line widths (∼25 nm) across the whole visible frequencies. After the compact integration with the imaging sensor, the MSFAs are validated to be effective in filtering and target identification. Our proposed fabrication method and imaging mode show great potentials to be an alternative to MSFAs production and MSI, by reducing both complexity and cost of manufacturing, while increasing flexibility and customization of imaging system.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Owing to the non-invasive detection mode, relatively short integration time, and continuing miniaturization, multispectral imaging (MSI) plays particularly significant roles in the fields of bioassay [1], snapshot spectral analysis [2], and portable consumer devices [3]. The past several decades have seen the considerable development of miniaturized optics and downscaled MSI systems [4–6]. Particularly, several snapshot spectral imagers, represented by the VTT and IMEC [7,8], have become commercially available, opening the new fields to portable, integrated, and handheld applications.
Advancements in nanofabrication techniques have provoked revolutionary changes to imaging filters, and facilitated the emergence of a host of micro-nano filtering structures [9–11], including nanohole arrays [12–15], subwavelength gratings [16–19], plasmonic nanoparticles [20–22], and Fabry-Perot (FP) resonant cavities [23–25]. These micro-nano structures provide varieties of methods to manipulate light, presenting unparalleled superiority over conventional filters. More importantly, the small and nearly planar structures enable the direct integration of such filters with detectors. The giants in the electronic industry, Sony [12] and Samsung [13], both focus on the plasmonic RGB color filters based on subwavelength hole arrays that are fabricated by electron beam lithography (EBL), demonstrating the potentials to replace conventional dye-based filters and the feasibility to be integrated onto imaging sensors. In the past two years, innovative approaches have emerged to fabricate multispectral filter arrays (MSFAs) based on FP cavities, such as wafer-level gray-scale lithography [26] and stencil deposition [27]. Although they are aiming to overcome the low-efficiency, high-cost, or the limitations on materials existing in common nanofabrication methods, the proposed approaches still utilize EBL to produce tens of micrometer-scale structures, or employ stainless-steel printed circuit board (PCB) process to manufacture several centimeter-scale stencils. On the other hand, conventional photolithography, one of the fundamental semiconductor technologies, seems to be ignored or not intensively studied. Although the combined etching or deposition techniques have been put forward to manufacture optical filters for several years [28,29], the produced thicknesses of films present evenly progressive trend and arrangement, unable to satisfy the customized design. Accordingly, it is worthwhile to investigate a more versatile and flexible method to achieve efficient fabrication of optical filters.
As for the integration of MSFAs and imaging sensors, there are two mainstream methods: monolithically integration [12,13,19,30] and directly deposition onto pixels of a CMOS sensor [7,8]. The former remains significant challenges in alignment precision, both by hands under microscopy [12] and automatic bonder [30]. The alignment accuracy of the advanced commercial bonder is 0.5 µm in supreme, while the pixels in imaging sensor are only several micrometers or smaller. In the case of directly deposition by IMEC, it is extremely difficult and complex, only suiting for the top-level industry. Nevertheless, these two methods immobilize the filters onto imagers, which restricts the easy, instant, and customized replacement. Therefore, developing an easier and more efficient integration approach is of great advantage to snapshot MSI based on MSFAs.
In this work, we propose a low-cost, wafer-scale approach for producing FP transmissive MSFAs which are narrowband, customizable, and easily assembly with CMOS sensors. The conventional lithography and customized deposition series are applied for personalized definition of the positions and thicknesses of FP resonant cavities to acquire various central wavelengths across the whole visible frequencies. On the basis of parametric investigation of FP and contrastive analyses of two dielectric materials, two palettes of the transmissive colors are generated by simply changing the thickness of different layers constructing FP structures. MSFA chips, both in the pixel-level and window-tiled configuration, are efficiently fabricated. The tiled MSFAs are compactly integrated with the imaging sensor to realize filtering and imaging. The simple and effective imaging mode can open their wide applications in the fields of easy and cheap assembling, as well as plug and play.
2. Results and discussions
2.1 Fabry-Perot color filters
Our proposed MSFAs, schematically shown in Fig. 1(a), consist of n by n channels. Each channel incorporates three layers—an amorphous silicon (a-Si) or aluminum oxide (Al2O3) dielectric layer sandwiched between the top and bottom silver (Ag) layers deposited on a glass substrate. The filter arrays are patterned using an optimized set of lithography while the layers are deposited by electron beam (E-beam) evaporation (see the Supplement 1 for details). Here, silver is selected as the top and bottom layer material, as it is an excellent and widely used material across the visible with relatively low absorption loss and high reflectivity. Al2O3 is selected for its obvious merits in low-loss, low-cost, chemically stable, and compatible with semiconductor processes; while a-Si reveals a strong absorption property in the visible frequency range. The thicknesses of both the top and bottom Ag layers (t and b) are set to be equal and optically thin. As shown in Fig. 1(b), with the fixed thickness of Al2O3 (h=100 nm), the FWHM obviously gets narrower with the increase of t and b. The calculated transmittance as a function of the Al2O3 thickness h and wavelength for the FP cavities are mapped in Fig. 1(c), illustrating that the transmission peaks continuously and nearly linearly shift to longer wavelength by gradually increasing h. Figure 1(d) shows the detailed representative data in Fig. 1(c), demonstrating the decreasing transmittance and FWHM with the increase of h when fixing t and b.
Fig. 1. Schematic diagram, fabrication, and measurement of the transmissive color filter arrays. (a) Schematic illustration of the MSFAs with the zoomed-in inset showing the FP structures. The thicknesses of the top and bottom Ag layers are set to be t (nm) and b (nm), respectively, while the middle dielectric layer (a-Si or Al2O3) is varying in thickness h. (b)(c) Simulated transmittance as the functions of wavelength and structural parameters, taking the Al2O3 layer as an example. (d) Detailed representative data in (c), showing the increasing central wavelength, decreasing transmittance and FWHM, by gradually increasing h. Color palettes and spectra measurements of the FP filters: transmission (marked T) and reflection (marked R) microscopy images of (e) Ag-(a-Si)-Ag and (f) Ag-Al2O3-Ag configurations, showing the tunability of color filtering. Measured transmittance and reflectance of (g) the Ag-(a-Si)-Ag filters and (h) the Ag-Al2O3-Ag filters. The resonant peaks shift from 400 to 700 nm with the increase of h.
Based on the above designs and analyses, two series of FP filters, with a-Si or Al2O3 dielectric layers, were fabricated. The spectral properties of each filter were characterized using a spectrometer (NOVA-EX) coupled to a microscopy system (see the Supplement 1 for detailed methods). For comparison, both the bright-field transmission and reflection microscopy images are illustrated in Figs. 1(e) and 1(f), respectively, showing that the transmissive colors generated from the filters display much higher saturation and wider gamut than the reflection colors. The peak positions in the measured transmittance and reflectance of the filters, plotted in Figs. 1(g) and 1(h), continuously redshift with the increase of h both for a-Si and Al2O3, whereas the ranges of h for the two materials are distinct. Owing to the relatively higher refractive index of a-Si (over 3.4 in the visible), the thicknesses of a-Si layers are relatively thin, adding the difficulty in E-beam evaporation. As for the Ag-Al2O3-Ag configuration, the FP resonances and spectra are less sensitive to the Al2O3 thickness compared to a-Si, so the precision and deviation of h are much easier to be controlled.
2.2 Contrastive analyses of the two dielectric materials
To further understand the distinctions between a-Si and Al2O3 for light manipulation in FP cavities, the material properties, experimental results, and calculated light field characteristics are systematically and comparatively investigated. Firstly, the transmissive colors and spectra depicted in Fig. 1 are converted as the discrete points in the CIE 1931 chromaticity coordinates, plotted in Fig. 2(a). The positions of colors, both for Ag-(a-Si)-Ag and Ag-Al2O3-Ag configurations, generate a circle, respectively, demonstrating the capability of the FP filters for widely ranged color tunability. The circles generated by the Ag-Al2O3-Ag configuration are bigger and closer to the achromatic point, producing a larger degree of visible colors than the Ag-(a-Si)-Ag configuration.
Fig. 2. Comparisons between a-Si and Al2O3 dielectric layers. (a) Measured results in the CIE 1931 chromaticity coordinates for the two groups of filters in Fig. 1. The white dashed lines with arrows represent the evolution trends for the colors when h increases. (b) The complex refractive index (n, k) of a-Si and Al2O3 materials as a function of wavelength, adopted from Palik [31]. (c)(d) Simulated electric field profiles and absorbed power distributions for the FP filters at the maximum electric and absorbed peak wavelength. Comparisons of measured transmission spectra for (e) different metallic thicknesses and (f) different dielectric materials.
Secondly, the complex refractive index (n, k) of a-Si and Al2O3 materials as a function of wavelength, adopted from Palik [31], are depicted in Fig. 2(b). The real part of refractive index of Al2O3 [n(Al2O3)] is obviously lower and flatter in the visible than that of a-Si, and the k(Al2O3) can be ignored, proving the lower refraction and lower absorption loss of Al2O3, respectively. Furthermore, as plotted in Figs. 2(c) and 2(d), the electric fields at the transmission peak are highly confined at the dielectric layers, where standing waves are formed due to the constructive interference of the incoming and reflected lights. However, absorbed power distributions for a-Si and Al2O3 are distinctly different. At the absorption resonant peak for the Ag-(a-Si)-Ag structure, most of the optical power is absorbed inside the a-Si layer and bottom Ag film. While for the Al2O3 layer, absorption is approximately only produced at the metal−dielectric interfaces, owing to its low extinction coefficient k. Experimental results support the above analyses. As illustrated in Fig. 2(e), the transmittance and FWHM of Ag-Al2O3-Ag structure both decrease when the metallic thickness turns from 25 nm to 40 nm, identical to the tendency in Figs. 1(b)–1(d). Additionally, to produce the same resonant peaks, thinner a-Si films are needed, approximately one-third of the Al2O3 films, while the transmittance is evidently declining due to the absorption loss, as shown in Fig. 2(f).
2.3 Batch fabrication of multispectral filter arrays
In order to efficiently fabricate the large-scale MSFAs, a host of manufacturing procedures have been researched. Apart from the EBL and focused ion beam (FIB) milling, which are high-cost and non-scalable, the combined etching/deposition patterning processes have been proposed for long years [28,29]. However, fabricating MSFAs which are pixel-level, customizable, commensurate with the CMOS sensor area, and across the whole visible frequency, still remains challenging. Here, we are devoted to addressing the above problems through iteratively and precisely defining the deposition positions and thicknesses only involving E-beam evaporation and lithography. The whole process parameters and flow are customized by the filter designing and channel arrangement, just like building Lego bricks, as schematically depicted in Fig. 3. The bottom Ag film is deposited on the glass substrate, followed by the alternative lithography and E-beam evaporation for four turns, yielding sixteen filtering channels after the top Ag film deposition.
Fig. 3. Schematic batch fabrication process for MSFAs. (a) The bottom Ag film deposition by E-beam evaporation. (b)-(e) Alternative processes of lithography and E-beam evaporation to produce dielectric layers with a customized definition of regions, materials, and thicknesses. (f) The final top Ag film deposition.
To demonstrate the feasibility and the versatility of our fabrication methods, different types of MSFAs constructed with various material configurations and spatial arrangements were manufactured at wafer level. As shown in Figs. 4(a) and 4(b), more than 300 filter chips were manufactured on a four-inch glass wafer, each of which possesses a total dimension covering 4.8 mm × 3.6 mm, slightly larger than the window area of the commercial CMOS imager. The filtering channels, configured in Ag-(a-Si)-Ag structures, are 6 µm × 6 µm squares, each matching with one pixel of the imaging sensor and forming 4 × 4-band arrays. Diverse deposition regions and thicknesses are customized and presented in Fig. 4(c), in which the areas with red dotted borders are the deposited positions corresponding to Fig. 3. Through calculation and optimization, we can flexibly choose the regions, thicknesses, and materials to be deposited, differentiating it from the uniform strategies which produce evenly progressive thicknesses. After finishing the top Ag film deposition, both the reflection and transmission characteristics were measured by the microscopy camera and spectrometer, respectively, as shown in Figs. 4(d) and 4(e). In addition, as shown in Figs. 4(f)–4(h), another Ag-Al2O3-Ag filter array was fabricated in the similar way (see the Supplement 1 for details).
Fig. 4. Batch fabrication of pixel-level MSFAs. (a) A filter chip with Ag-(a-Si)-Ag structures manufactured on (b) a four-inch glass wafer. (c) Microscopy images monitoring the processing flow. The regions with red dotted borders are the areas to be deposited, corresponding to Fig. 3. (d-i) Reflection and (d-ii/iii) transmission microscopy images after the top Ag deposition, with the measured transmittance shown in (e) (selecting seven curves here for clarity). (f) Ag-Al2O3-Ag filter chips on a four-inch wafer. (g-i) Reflection and (g-ii) transmission microscopy images of the Ag-Al2O3-Ag filter structures after the top Ag deposition, with the measured transmittance shown in (h). Scale bars: 50 µm for (d-iii), 20 µm for (g), and 10 µm for the others.
2.4 Imaging experiment with window-tiled multispectral filter arrays
Apart from the above pixel-level MSFAs, enlarged MSFAs which can be tiled on the window of imaging sensors are also fabricated by the similar method. Further, differing from the imagers based on pixel-level MSFA, we demonstrate a more simplified imaging mode snapshot MSI. It is realized by inserting a window-tiled MSFA chip between the commercial monochrome CMOS sensor (Sony IMX226) and lens. The tiled MSFAs were mounted upside down to minimize the effect of substrate thickness, into a customized holder which matches the size of imaging sensor, schematically depicted in Fig. 5(a). Benefiting from the compact assembly of the filter chip and the imaging sensor, with no interference to other components, the lens can be normally installed by the threads [Fig. 5(b)]. As shown in Fig. 5(c), MSFAs containing 4 × 4-band filtering channels were fabricated on a four-inch glass wafer, configured in 40 nm Ag-(65–140 nm) Al2O3−40 nm Ag. The channels are 1.2 mm × 0.9 mm, tiled on a 4.8 mm × 3.6 mm chip, exhibiting the central wavelengths (λc) from 406 nm to 688 nm and FWHM of ∼25 nm in average. As illustrated in Figs. 5(d)–5(g), the filtering channels (numbered in accordance with the increasing order of λc from small to large) are set in a ‘rectangular’ arrangement [32,33] to fulfill spectral consistency and spatial uniformity and minimize crosstalk among adjacent channels (see the Supplement 1 for details). Owing to the separation of the filter chip and sensor, this integration method and imaging mode are potentially more superior than pixel-level snapshot MSI in the situation of easy and cheap assembling, as well as plug and play.
Fig. 5. Integration of the tiled MSFAs on the imaging sensor. (a) Schematic of the filter chip mounted in a customized holder. The channels are 1.2 mm × 0.9 mm, tiled on a 4.8 mm × 3.6 mm chip. (b) Compact assembly of the filter and the imaging chip. (c) 16-band tiled MSFAs manufactured on a four-inch glass wafer. (d)-(g) Measured transmittance of every 2 × 2 unit in the filter chip, with the detailed channel arrangements illustrated in the insets.
In order to verify the effectiveness and performance of the proposed method, the MSFAs and compact CMOS sensor were used for two kinds of testing scenes: snapshot imaging of a series of color paperboards covering all filtering channels, and multiple exposures for imaging relatively small objects.
Firstly, paperboards with different colors [A4-sized, shown in Fig. 6(a)], as well as the standard white plate, were captured by the imaging sensor through our MSFAs. The arrangement of each channel’s central wavelength is illustrated in Fig. 6(a-iii), which is rotated 180° compared with that of MSFA chips. After the white calibration, the gray levels of sixteen channels are used to characterize the spectral responses [Figs. 6(a-iv/v)]. As illustrated in Fig. 6(b), fourteen paperboards are snapshot imaged, each of which yielding a tiled mosaicked multispectral image with sixteen bands. Thus, the 16-band information can be achieved in one image. Then the gray levels of each region corresponding to filtering channels are plotted as the function of λc, forming a spectral response curve of each paperboard. In Fig. 6(c), three paperboards (#2/6/13) are selected for demonstration and compared with the measured results of a commercial hyperspectral imaging (HSI) system (GaiaField-V10, 400∼1000 nm with spectral resolution of 3.5 nm). The dashed lines with data markers are the fitting curves of our MSI system’s results, which present the similar trends, peaks, and peak positions as HSI imager. Here, we utilize fourteen bands (467 nm to 688 nm) for calculation and fitting due to the relatively larger errors of shorter-wavelength channels (406 nm and 433 nm). Several deviation dots and fluctuations may result from the relatively fewer bands, larger FWHM, and manufacturing errors of the MSFAs. The compared results prove that our MSI system owns the capability to measure the spectral responses of imaging objects, although the numbers of channels are less than one-tenth of the HSI imager. Based on the measurement of spectral responses, color reconstruction was conducted to further evaluate and check the effectiveness of our system. As shown in Fig. 6(d), the spectral responses of paperboard #5 measured by the MSFAs are used to calculate XYZ tristimulus values in CIE 1931 color space under D65 standard illuminant, and then converted to sRGB components [34]. The reconstructed color is very close to the actual color of the paperboard [Fig. 6(e)].
Fig. 6. Snapshot imaging through the MSFAs. (a) Experiment flow of snapshot imaging: (i) a standard white plate and color paperboards are captured by (ii) our MSI system. (iii) The arrangement of λc in the multispectral image. (iv) Snapshot image of paperboard #5 after white calibration, with the calculated gray levels shown in (v). (b) Snapshot images of 14 color paperboards. (c) The spectral responses of three picked paperboards calculated by our MSI system and measured by a commercial HSI system. (d) The calculated relative reflectance of paperboard #5 is integrated and converted to the CIE 1931 color space to (e) reconstruct the color.
Secondly, the MSFAs and imaging sensor were used to capture the same scenes through different filtering channels by moving the camera for multiple exposures. As shown in Fig. 7(a), a cat assembled by the seven-piece puzzle, which is a Chinese traditional game, was captured by our MSI system. The blue, red, and yellow pieces remarkably change in signal intensities through different filtering channels, showing the capability of MSFAs and imaging system to identify different targets that are characteristic in spectra. Three representative bands are demonstrated in Figs. 7(b)–7(d) for comparison with the spectral slices by the commercial HSI system. The single-band images of seven-piece puzzle by our system and HSI imager present similar signal intensities. While looking closely to the images by the HSI imagers, the edge lines of the cat’s head and ears (indicated by the red arrows) are slightly out of shape, which are caused by the scanning errors of the HSI imager, proving the advantage of snapshot spectral imaging in shape preserving.
Fig. 7. Imaging experiments and spectral measurement. (a) A cat assembled by seven-piece puzzle is captured through different channels of the MSFAs and a commercial HSI system, (b)-(d) three channels selected here for comparison. The edge lines of the cat’s head and ears (indicated by the red arrows) are slightly out of shape, which are caused by the scanning errors of the HSI imager. (e) Seven billiard balls are captured through 16 channels of the MSFAs. (f) Color images of the billiard balls taken by a color camera. (g) A spectral slice of balls #1 and #5 by a commercial HSI system at 565 nm, comparing with the image at the same wavelength in (e). (h) Spectral responses of the two dots marked as stars in (g), which are calculated by our MSI system and measured by the commercial system respectively.
Similarly, seven billiard balls with different colors were also imaged in sixteen bands through our MSI system by multiple exposures, as illustrated in Fig. 7(e). Ball #1 and ball #5, whose actual colors are shown in Fig. 7(f), are selected for demonstration. These two balls present obvious diversities at 565 nm. The single-band image taken by our system is consistent with the commercial HSI imager [Fig. 7(g)]. Moreover, spectral responses of the two dots marked as stars in Fig. 7(g) are calculated by our MSI system and measured by the HSI system respectively. As plotted in Fig. 7(h), our measured results are also consistent with the HSI imager, further validating the effectiveness and accuracy of our MSFAs and imaging system.
Although the imaging object or partial region here is not captured simultaneously through different channels compared to pixel-level snapshot systems, we believe that the proposed system and the imaging mode own its special advantages and suit for specific applications. By moving the objects or the imager, the whole object can be scanned through more bands, whether to generate images or to acquire spectral data. The relative movements of object and imager would be implemented by head-mounted display devices, random scanning and image fusion, or conveyor belts. Moreover, the tiled MSFAs and the imaging mode can overcome the expensiveness and complexity of scientific snapshot MSI systems, providing an alternative method in manufacture and assembly. The simple, cheap, and flexible system is capable to open a new route for rapid visual inspection, target identification, and cost-effective optoelectronic devices.
3. Conclusions and outlook
In summary, through the use of an optimized set of lithography and E-beam evaporation, we have achieved the customized wafer-level fabrication of MSFAs, and the compact integration of MSFAs with imaging sensor to realize target identification and spectral measurement. Based on the parametric influences of FP resonant structures and contrastive analyses of the two dielectric materials (absorbing a-Si or lossless Al2O3), the filtering properties of the microstructure arrays can be randomly adjusted, across the whole visible frequencies. The high-efficiency and low-cost fabrication techniques avoid the EBL or FIB process and provide versatile approaches to massively fabricate the MSFAs. Through the ‘plug and play’ assembly, our imaging system further retains the advantages of compactness and an easier operating mode. Different from the pixel-level snapshot spectral imaging cameras, our imaging system mainly focuses on the simple and cheap assembly, and rapid identification of objects. This peculiarity not only reduces the complexity and cost of MSI system, but also allows for consumer applications such as wearable electronics, rapid food detection, and smart mobile diagnosis.
Funding
Science, Technology and Innovation Commission of Shenzhen Municipality (JCYJ20180508151936092); National Natural Science Foundation of China (51975483); Natural Science Foundation of Ningbo (202003N4033); Key Research and Development Projects of Shaanxi Province (2020ZDLGY01-03); The Peak Experience Program of Northwestern Polytechnical University (201912); Open Foundation Project of the Key Laboratory of Spectroscopic Imaging of the Chinese Academy of Sciences (LSIT201912W).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
Supplemental document
See Supplement 1 for supporting content.
References
1. F. Yesilkoy, E. R. Arvelo, Y. Jahani, M. Liu, A. Tittl, V. Cevher, Y. Kivshar, and H. Altug, “Ultrasensitive hyperspectral imaging and biodetection enabled by dielectric metasurfaces,” Nat. Photonics 13(6), 390–396 (2019). [CrossRef]
2. J. Wu, B. Xiong, X. Lin, J. He, J. Suo, and Q. Dai, “Snapshot hyperspectral volumetric microscopy,” Sci. Rep. 6(1), 24624 (2016). [CrossRef]
3. S. M. Park, M. A. Visbal-Onufrak, M. M. Haque, M. C. Were, V. Naanyu, M. K. Hasan, and Y. L. Kim, “mHealth spectroscopy of blood hemoglobin with spectral super-resolution,” Optica 7(6), 563–573 (2020). [CrossRef]
4. P. J. Lapray, X. Wang, J. B. Thomas, and P. Gouton, “Multispectral filter arrays: recent advances and practical implementation,” Sensors 14(11), 21626–21659 (2014). [CrossRef]
5. Z. Yang, T. Albrow-Owen, W. Cai, and T. Hasan, “Miniaturization of optical spectrometers,” Science 371(6528), eabe0722 (2021). [CrossRef]
6. J. Bao and M. G. Bawendi, “A colloidal quantum dot spectrometer,” Nature 523(7558), 67–70 (2015). [CrossRef]
7. B. Geelen, C. Blanch, P. Gonzalez, N. Tack, A. Lambrechts, G. von Freymann, W. V. Schoenfeld, R. C. Rumpf, and H. Helvajian, “A tiny VIS-NIR snapshot multispectral camera,” Proc. SPIE 9374, 937414 (2015). [CrossRef]
8. P. Gonzalez, K. Tack, B. Geelen, B. Masschelein, W. Charle, B. Vereecke, A. Lambrechts, M. A. Druy, and R. A. Crocombe, “A novel CMOS-compatible, monolithically integrated line-scan hyperspectral imager covering the VIS-NIR range,” Proc. SPIE 9855, 985501 (2016). [CrossRef]
9. N. I. Zheludev and Y. S. Kivshar, “From metamaterials to metadevices,” Nat. Mater. 11(11), 917–924 (2012). [CrossRef]
10. I. Kim, G. Yoon, J. Jang, P. Genevet, K. T. Nam, and J. Rho, “Outfitting next generation displays with optical metasurfaces,” ACS Photonics 5(10), 3876–3895 (2018). [CrossRef]
11. S. Daqiqeh Rezaei, Z. Dong, J. You En Chan, J. Trisno, R. J. H. Ng, Q. Ruan, C. Qiu, N. A. Mortensen, and J. K. W. Yang, “Nanophotonic structural colors,” ACS Photonics 8(1), 18–33 (2021). [CrossRef]
12. S. P. Burgos, S. Yokogawa, and H. A. Atwater, “Color imaging via nearest neighbor hole coupling in plasmonic color filters integrated onto a complementary metal-oxide semiconductor image sensor,” ACS Nano 7(11), 10038–10047 (2013). [CrossRef]
13. Y. Horie, S. Han, J. Lee, J. Kim, Y. Kim, A. Arbabi, C. Shin, L. Shi, E. Arbabi, S. M. Kamali, H. Lee, S. Hwang, and A. Faraon, “Visible wavelength color filters using dielectric subwavelength gratings for backside-illuminated CMOS image sensor technologies,” Nano Lett. 17(5), 3159–3164 (2017). [CrossRef]
14. J. Zhao, X. Yu, X. Yang, Q. Xiang, H. Duan, and Y. Yu, “Polarization independent subtractive color printing based on ultrathin hexagonal nanodisk-nanohole hybrid structure arrays,” Opt. Express 25(19), 23137 (2017). [CrossRef]
15. Z. Wang, S. Yi, A. Chen, M. Zhou, T. S. Luk, A. James, J. Nogan, W. Ross, G. Joe, A. Shahsafi, K. X. Wang, M. A. Kats, and Z. Yu, “Single-shot on-chip spectral sensors based on photonic crystal slabs,” Nat. Commun. 10(1), 1020 (2019). [CrossRef]
16. A. F. Kaplan, T. Xu, and L. Jay Guo, “High efficiency resonance-based spectrum filters with tunable transmission bandwidth fabricated using nanoimprint lithography,” Appl. Phys. Lett. 99(14), 143111 (2011). [CrossRef]
17. K. Lee, J. Jang, S. J. Park, C. Ji, S. Yang, L. J. Guo, and H. J. Park, “Angle-insensitive and CMOS-compatible subwavelength color printing,” Adv. Opt. Mater. 4(11), 1696–1702 (2016). [CrossRef]
18. J. Zhao, X. Yu, X. Yang, T. C. Th, W. Yuan, and Y. Yu, “Polarization-independent and high-efficiency broadband optical absorber in visible light based on nanostructured germanium arrays,” Opt. Lett. 44(4), 963–966 (2019). [CrossRef]
19. Y. Ohtera and K. Shinoda, “NIR spectrum estimation utilizing a photonic crystal distributed passband-type multiple filter array,” Appl. Opt. 58(12), 3166–3173 (2019). [CrossRef]
20. H. Lee, H. Ahn, J. Mun, Y. Y. Lee, M. Kim, N. H. Cho, K. Chang, W. S. Kim, J. Rho, and K. T. Nam, “Amino-acid- and peptide-directed synthesis of chiral plasmonic gold nanoparticles,” Nature 556(7701), 360–365 (2018). [CrossRef]
21. E. Panchenko, L. Wesemann, D. E. Gómez, T. D. James, T. J. Davis, and A. Roberts, “Ultracompact camera pixel with integrated plasmonic color filters,” Adv. Opt. Mater. 7(23), 1900893 (2019). [CrossRef]
22. A. Leitis, A. Heßler, S. Wahl, M. Wuttig, T. Taubner, A. Tittl, and H. Altug, “All-dielectric programmable huygens’ metasurfaces,” Adv. Funct. Mater. 30(19), 1910259 (2020). [CrossRef]
23. Z. Yang, Y. Chen, Y. Zhou, Y. Wang, P. Dai, X. Zhu, and H. Duan, “Microscopic interference full-color printing using grayscale-patterned Fabry-Perot resonance cavities,” Adv. Opt. Mater. 5(10), 1700029 (2017). [CrossRef]
24. J. Zhao, M. Qiu, X. Yu, X. Yang, W. Jin, D. Lei, and Y. Yu, “Defining deep-subwavelength-resolution, wide-color-gamut, and large-viewing-angle flexible subtractive colors with an ultrathin asymmetric Fabry-Perot lossy cavity,” Adv. Opt. Mater. 7(23), 1900646 (2019). [CrossRef]
25. W. Yan, V. R. Shresha, Q. Jeangros, N. S. Azar, S. Balendhran, C. Ballif, K. Crozier, and J. Bullock, “Spectrally selective mid-wave infrared detection using Fabry-Pérot cavity enhanced black phosphorus 2D photodiodes,” ACS Nano 14(10), 13645–13651 (2020). [CrossRef]
26. C. Williams, G. S. D. Gordon, T. D. Wilkinson, and S. E. Bohndiek, “Grayscale-to-color: scalable fabrication of custom multispectral filter arrays,” ACS Photonics 6(12), 3132–3141 (2019). [CrossRef]
27. X. Li, Z. J. Tan, and N. X. Fang, “Grayscale stencil lithography for patterning multispectral color filters,” Optica 7(9), 1154 (2020). [CrossRef]
28. S. Wang, X. Chen, W. Lu, L. Wang, Y. Wu, and Z. Wang, “Integrated optical filter arrays fabricated by using the combinatorial etching technique,” Opt. Lett. 31(3), 332–334 (2006). [CrossRef]
29. S. Wang, C. Xia, X. Chen, W. Lu, M. Li, H. Wang, W. Zheng, and T. Zhang, “Concept of a high-resolution miniature spectrometer using an integrated filter array,” Opt. Lett. 32(6), 632–634 (2007). [CrossRef]
30. X. He, Y. Liu, K. Ganesan, A. Ahnood, P. Beckett, F. Eftekhari, D. Smith, M. H. Uddin, E. Skafidas, A. Nirmalathas, and R. R. Unnithan, “A single sensor based multispectral imaging camera using a narrow spectral band color mosaic integrated on the monochrome CMOS image sensor,” APL Photonics 5(4), 046104 (2020). [CrossRef]
31. E. D. Palik, Handbook of optical constants of solids (Academic, 1998).
32. R. Shrestha, J. Y. Hardeberg, and R. Khan, “Spatial arrangement of color filter array for multispectral image acquisition,” Proc. SPIE 7875, 787503 (2011). [CrossRef]
33. L. Miao and H. Qi, “The design and evaluation of a generic method for generating mosaicked multispectral filter arrays,” IEEE Trans. on Image Process. 15(9), 2780–2791 (2006). [CrossRef]
34. K. Shinoda, Y. Ohtera, and M. Hasegawa, “Snapshot multispectral polarization imaging using a photonic crystal filter array,” Opt. Express 26(12), 15948–15961 (2018). [CrossRef]
