Open Access
Add to BibSonomy
Abstract
The effects of miniaturization on theoretically predicted performance of dual-period guided mode resonance (GMR) transmission filters, which demonstrate immense potential for multispectral imaging in short wave infrared (SWIR), have been compared with experimental findings. With reducing filter size from 112 periods (90 µm) to 12 periods (10 µm), peak transmittance (Tpeak) of simulated and measured filters reduced gradually from 84% to 55% and from 76% to 65%, respectively, with a moderate change of 1 - 3 nm in full width at half maximum (FWHM). For 6 period filters (5 µm), simulations predict drastically reduced Tpeak = 14% accompanied by increase in FWHM by 12 nm. The Tpeak value is theoretically shown to increase to 46% with FWHM reduced by 7 nm upon placing metal reflectors at the optimum positions to increase the optical path length. Our findings indicate that four 5 µm × 5 µm size filters with metal reflectors designed for different resonance wavelengths can be used to form a single, 20 µm × 20 µm mosaic pixel for SWIR multispectral imaging.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Spectral imaging has revolutionized sensing and detection schemes for numerous application fields in the last few decades. It has been employed in a wide range of applications in mineralogy [1,2], agriculture [3–6], object detection [7], atmospheric monitoring for weather forecasting [5], food inspection [6,8], biological process analysis [9,10], medicine development [11,12] and testing [12,13], remote sensing [1,4,7,14] and surveillance [9,15], and others [16–18]. These applications stem from identifying chemical compounds having unique spectral characteristics under electromagnetic excitation [19]. For real-time simultaneous monitoring of multiple sharp spectral features that identify a particular chemical, an imaging array of detectors and sensors with narrow-bandpass filters are required. Such spectral imaging techniques have been largely explored in the visible and near infrared (NIR) ranges; however, despite the immense potential with ample demand in a wide range of applications, development of real-time multi-spectral imaging sensors for the shortwave infrared (SWIR) range has been limited. Notable works found in literature are based on optical filters that rely on absorption [20], interference [21], plasmonic effects [22,23], Fabry-Perot resonance [24,25] and guided mode resonance (GMR) [26–28]. Owing to their sharp resonances, GMR filters offer better potential to demonstrate multispectral imaging requiring simultaneous detection of multiple spectral lines with narrow bandwidths [28–31].
The GMR filters operate on the premise of diffraction of electromagnetic waves incident on a subwavelength grating structure [32]. The diffracted modes propagate laterally through a waveguide layer as leaky modes and interfere with the incident waves. With phase matching conditions fulfilled, constructive interference of the zeroth order diffracted mode resonates at a particular wavelength, forming a narrow transmission window while destructive interference at adjacent wavelengths form wide stopbands on both sides of the transmission peak. With the inclusion of spatial asymmetry [32] in the grating layer, the stopband can be extended further [33]. Recently, we have demonstrated 90 µm × 90 µm size dual-period narrowband transmission GMR filters with 87% transmittance, 15 nm full width at half maximum (FWHM) pass band and 400 nm wide stopbands tunable in the 1200 - 1600 nm wavelength range for TE-polarized normally incident light [33]. A set of four such filters placed into a mosaic pixel of 180 µm × 180 µm can be designed to detect four different center wavelengths. However, for high spatial resolution imaging, further reduction of GMR transmission filter size is required. Although some theoretical investigations are found in literature where finite size effects on single period GMR filters were investigated [34], additional studies are necessary.
In this paper, we report on the performance of TE-mode GMR transmission filters of sizes reduced from 90 µm × 90 µm to 5 µm × 5 µm and explore the effect of pixel size miniaturization on filter characteristics. Further, we investigate the impact of placing metal reflectors on the boundaries of miniaturized filters to recover their performance to the level of large filters and propose their pixel size multi-wavelength mosaic arrangements for multispectral imaging in the SWIR wavelength range.
2. Methods
2.1 Simulation
Transmittance spectra of one-dimensional stripe-patterned (for TE-polarized normally incident light) dual period GMR filters were designed with a period, Λ = 800 nm and grating layer depth, dg = 540 nm, using COMSOL Multiphysics. Operation under normal incidence is investigated as this is typical for focal-plane imaging sensor applications. The dual-period filter consisted of two sets of fill factors (FFs) defined as the ratio of the width of a stripe or trench layer to Λ. FFs for the reference design were: FF1(Si) = 10.2%, FF1(Air) = 22.2%, FF2(Si) = 37.6%, FF2(Air) = 30.0%. For the infinitely wide filter designed for a particular wavelength of peak transmittance, the design set of Λ and FFs for a constant dg providing the highest figure of merit was considered the optimum. The Si stripes were assumed to have perfectly vertical sidewalls and sharp edges as shown in Fig. 1(a). To investigate the effects of miniaturization, finite filters with widths of 112 periods (∼90 µm), 75 periods (∼60 µm), 56 periods (∼45 µm), 37 periods (∼30 µm), 19 periods (∼15 µm), 12 periods (∼10 µm), 6 periods (∼5 µm) and 5 Periods (∼4 µm) in the x-direction and infinitely stretched in the z-direction were simulated. Scattering boundary conditions were used at the edges of the structure to mimic a realistic response. For miniaturized filters, the impact of metal reflectors at the boundaries was also explored by using Al layers with thickness more than 5 times the skin depth. The reflector height from the stripe layer was set to 1901 nm, which allows one full wavelength of design (1361 nm) above the stripe layer to be resolved in the simulations. Transmittance response is sustained for 50 nm variation in the reflector height. At the reflector edges, impedance boundary conditions were used, which is equivalent to using an Al layer at least five times the skin depth, ensuring the penetrating waves die off within the conducting layer. The higher reflector compared to the Si stripe thickness allows leaky waves to couple out and reflect from the reflector surfaces and interfere more with the incident waves, enhancing resonance. Away from the stripes, infinitely extended Si and substrate layers were assumed. Absorption in Si [35] and Al [36] was included in the simulations, while constant refractive index of Si (nSi = 3.48) was considered from 1200 nm to 1600 nm, as dispersion is negligible in this range. Dispersion was included for Al as it varies significantly from 1.02 at 1200 nm to 1.40 at 1600 nm [36]. TE-polarized light was chosen, because it provides resonant wavelengths smaller than those with TM-polarized light for the same dg and Λ [32].
Fig. 1. (a) Schematic of a dual-period GMR transmission filter. The structure repeats with a period Λ and fill factors (FFs) define the width of each pillar or trench within a period while dg denotes grating depth. Structures with and without the Al reflectors were simulated. (b) Schematic of transmittance measurement setup.
2.2 Fabrication
A silicon-on-quartz (SoQ) wafer, 540 nm-thick crystalline silicon (c-Si) layer on a 625 µm-thick quartz substrate, was used to pattern the designed stripes forming the filter layer. Thickness of the Si layer was confirmed by ellipsometry. First a ∼400 nm thick positive e-beam resist (ZEP-520A ZEON Co.) was spin-coated on the SoQ wafer. Then, square pattern-fields varying in size from 25 µm2 to 8100 µm2 were written using a Raith Voyager 50 kV EBL system using an optimized exposure dose related to the pattern size. After EBL exposure, the samples were developed in Xylene at room temperature. The silicon layer with the patterned resist acting as a mask was etched in an inductively coupled plasma (ICP) system (RIE-101iPH by SAMCO Inc.) at 20° C using SF6 (flow rate 20 sccm) and C4F8 (flow rate 14 sccm) gas mixture while maintaining the process pressure of 0.5 Pa with bias and ICP power levels of 20 and 60 W, respectively. Finally, the residual e-beam resist was stripped in the H2SO4/H2O2 solution (3:1). Dimensional parameters (i.e., grating width, height etc.) of the fabricated filters were obtained using scanning electron microscopy (SEM).
2.3 Optical characterization
As shown in Fig. 1(b), a quartz tungsten-halogen lamp was used as a light source for transmittance measurements. The light coming from this source was TE-polarized (parallel to the stripes) and collimated. Then the collimated light beam was focused on the sample into a circular spot of 8-9µm diameter at beam-waist by an achromatic objective lens (50x, NA of 0.42). The sample was moved along the direction of propagation to place it at the beam waist ensuring normal incidence. Transmitted light through the sample was further collimated by achromatic lenses and fed through an optical fiber coupled to a 30 cm focal length spectrometer which is equipped with an InGaAs multichannel detector. Experimental optical transmission spectra were compared with simulated responses using the measured structural dimensions.
3. Results and discussion
Characteristics obtained from the simulated transmission spectra of filters without reflectors are listed in Table 1. As shown in Fig. 2(a), the design yielded a transmission peak at ∼1361 nm, with 89% peak transmittance, ∼10 nm FWHM and wide stopbands on both sides of the peak within the wavelength range 1200 nm to 1600 nm. With reducing size, peak transmittance noticeably decreased gradually from 84% to 14% only for filters smaller than 30 µm, while the full width at half maximum (FWHM) of the transmission peak remained nearly the same at 10 - 12 nm for all but doubled for the smallest size filter. This is owing to the fact that, the larger size of the filters allows the guided waves to traverse longer optical paths to interfere constructively resulting in a stronger resonance at the transmission peak. The width of the stopband, defined for transmittance less than 1%, increased slightly from 251 nm to 286 nm with decreasing filter size also only beyond 30 µm, when the resonant peak became broader. While overall variations in the transmission spectra were clearly observed for sizes smaller than 30 µm, drastic changes occurred for 5 µm wide filters. The peak transmittance dropped to ∼14% with FWHM increasing to 22 nm, stopband width broadened to 286 nm, and the transmission peak blueshifted by 5 nm to 1355 nm. From the inset of Fig. 2(a), the resonance occurring at the transmission dips creating the stopbands, become weaker with decreasing size particularly below 12 periods [34]. The gradual reduction in peak transmittance due to decreasing filter size could be attributed to mode-splitting [37] and asymmetric lateral leakage of light [38]. More than two orders of magnitude difference between the transmission peak and the adjacent transmission dip are sustained up to 6 periods forming a stopband (T < 1%). For further reduction in size, the difference of magnitude becomes even smaller, as evident from Fig. 2(a) inset. Hence, 6 periods is chosen as the smallest size of consideration in our exploration.
Fig. 2. (a) Simulated transmission spectra of design GMR transmission filters of different sizes; inset: transmittance spectra in semi-log scale for filters of 12, 6 and 5 periods with the same legends. The broadened GMRs for filter with 5 periods yield narrow stopbands, making the filter with 6 periods as the smallest finite size of consideration in our exploration. (b) Experimental transmission spectra of GMR transmission filters of different sizes.
Table 1. Simulated peak wavelength, peak transmittance (Tpeak), full width at half maximum (FWHM), stopband width, and figure-of-merit (FOM) for different size filter designs
For multispectral detection of chemical signatures, it is desirable to have peak transmittance as high, FWHM as small and stopband as wide as possible. Hence, to quantify the overall quality of a GMR transmission filter, we have defined a figure-of-merit (FOM) parameter as the ratio of stopband (Δλ) to the FWHM (δλ) times the peak transmittance (Tpeak). As shown in Table 1, with reducing filter size beyond 30 µm, FOM significantly decreases indicating a gradual deterioration of the filter performance.
Based on the one-dimensional design in Fig. 1(a), square shaped filters of multiple sizes ranging between 90 µm × 90 µm to 10 µm × 10 µm were fabricated. 5 µm × 5 µm filters were excluded as they would have been smaller than the lowest attainable beam spot diameter (8 µm - 9 µm) in our setup. A beam spot larger than the filter size would result in Fabry-Perot oscillations from the surrounding unpatterned Si surface hindering the individual filter response. Experimental responses from the fabricated filters of different sizes show excellent agreement in peak position [Fig. 2(b)]. The slight redshift of < 6 nm with the filter size reducing from 90 µm × 90 µm to 10 µm × 10 µm as well as a small blueshift of 5 - 11 nm with respect to the simulated spectra in Fig. 2(a) (∼1360 nm) is attributed to an FF variation of ∼0.5% in fabricated structures due to the 10 nm resolution of our EBL system. The transmittance spectra follow a general trend where the stopband on the short wavelength side is retained at ∼11% transmittance down to the 15 µm × 15 µm size filter but becomes slightly larger for the 10 µm × 10 µm size filter. Choosing the 11% transmittance level as reference, the FWHM and the overall stopband within which the transmission peak is present were determined. From 90 µm × 90 µm to 10 µm × 10 µm size filters, peak transmittance decreased from 76% to 65% and the transmission FWHM varied between 25 nm and 28 nm. Compared to the simulation values, FWHM was wider by 15 - 16 nm. Stopbands were also wider for the fabricated filters compared to the design values and increased from 278 nm to 292 nm with decreasing filter size from 90 µm × 90 µm to 15 µm × 15 µm. However, the stopband width drastically reduced to 177 nm for the 10 µm × 10 µm filter. FOM values for the fabricated filters, which are less than half of those for the simulated designs, are listed in Table 2 along with the Tpeak, δλ & Δλ obtained from the transmission spectra.
Table 2. Experimental peak wavelength, peak transmittance (Tpeak), full width at half maximum (FWHM), stopband width, and figure-of-merit (FOM) for different size filters fabricated
To understand the discrepancy between design and experiment, the structural parameters, i.e., Λ & FFs were measured using SEM [Fig. 3(a)]. No variation in Λ = 800 nm was observed. FFs for the fabricated filters matched reasonably well with each other but deviated significantly from the design values. The FFs for the fabricated filters were: FF1(Si) = 13.0%, FF1(Air) = 19.6%, FF2(Si) = 40.2%, FF2(Air) = 27.2%. These values have an uncertainty of up to ∼5% arising from the ∼40 nm accuracy in determining the stripe widths from the SEM images. Moreover, as shown in Fig. 3(b), the smaller width trenches did not etch all the way to the Quartz substrate primarily due to the etch rate variation with trench width and the proximity effect during the EBL step [39]. Within the mentioned error margin, a more accurate set of Λ and FF values for the fabricated filter structures was obtained using COMSOL simulations to fit the experimental responses [Fig. 3(c)-Fig. 3(h)]. This set is defined as the ‘Fabricated’ (Fab.) parameters in Fig. 3, where Λ = 800 nm, FF1(Si) = 10.2%, FF1(Air) = 20.2%, FF2(Si) = 37.4%, FF2(Air) = 32.2% with a residual Si layer thickness of 30 nm [for the trench with FF1(Air)]. The residual Si layer reduces the effective depth of the narrow trench from dg. Along with FF variation, the under-etched Si layer broadens and blueshifts the transmission peak [32]. This is further confirmed by simulation using the Fab. Parameters, which included the residual Si layer of ∼33 nm showing a peak shift of up to 11 nm and FWHM increase of 15 nm to 17 nm. Also, slightly curved sidewalls observed in the fabricated structures [Fig. 3(b)] compared to the vertical and sharp-edged geometries used in our COMSOL simulations may have also contributed to this broadening. The lower peak transmittance is attributed to sidewall roughening which enhances scattering, which is excluded in simulations, reducing the zeroth order mode transmittance.
Fig. 3. SEM images: (a) Mosaic of plan-view SEM images of different size filters from 90 µm × 90 µm to 10 µm × 10 µm (b) Angled-view SEM image of the 90 µm × 90 µm filter’s cross-section. Comparison between experiment and simulations with design and fabricated filter parameters with and without Al reflectors (experimental filters without reflectors) of sizes (c) 90 µm (d) 60 µm (e) 45 µm (f) 30 µm (g) 15 µm (h) 10 µm. (i) Comparison between simulations with design and fabricated filter parameters for a 5 µm wide filter with and without Al reflectors.
The degradation of the filter response with reducing size, particularly below 30 µm, can be mitigated by using metal reflectors placed at the edges of the filter [see Fig. (1)], which ideally could recover the response to that of an infinite size filter. The reflectors increase the optical path length of the propagating guided modes which would enhance resonance at the transmission peak. Metals with high infrared reflectivity i.e., Al, Au and Ag were initially considered as the reflector material. Despite Au and Ag’s slightly higher reflectivity, Al is chosen because it adheres well to Quartz and Si compared to Au or Ag, which would require an additional adhesion layer of Cr, Ni or Ti, which would diminish the reflectivity. Hence, Al deposition will be simpler and also cost effective. A reflector height of dAl = 1901 nm allows the out-coupling leaky modes from the stripe layer to reflect back better from the Al surface and interact with the incident waves further. This is chosen from adding the grating height (540 nm) to the resonant wavelength of design (1361 nm), to resolve the wave-interactions above the stripe layer in simulation. As the reflectors are adjacent to the unpatterned Si layer away from the stripes, this structure can be fabricated by etching a ∼1.9 µm thick Si layer down first in the stripe region by the difference between Al reflector depth and grating layer depth, (dAl – dg) = 1361 nm, and then patterning to create the stripes in the desired filter region. On the other hand, the reflectors need to be placed at the nodes of the guided mode to generate the same GMRs. The electric field profile for an infinitely wide filter with the design values [Fig. 4(a)] shows that there is no symmetry plane due to the structural asymmetry of the dual period design. Thus, the field profile cannot be reproduced by placing a mirror plane at a field node. This is apparent also for the finite filter with 6 periods [Fig. 4(b)] where the electric field strength decreases from the center towards the edges as opposed to the infinite filter structure. Through parametric sweep simulations, it was found that when Al reflectors are placed at 60 nm and 100 nm outwards from the left and right edge, respectively, in the trenches, the obtained electric field profile [Fig. 4(c)] matches better with the infinite size filter profile in Fig. 4(a), compared to the finite filter of the same size but without reflectors. Similarly, for the fabricated filter parameter sets, the electric field profiles and filter responses were investigated to find the optimum positions to be 60 nm inwards and 100 nm outwards from the left and right edges, respectively. For the design set, peak transmittance (80 - 85%) does not change much when using Al reflectors for the larger filters (90 µm to 30 µm) but increases significantly for filters smaller than 30 µm [Fig. 3(c)–3(h)]. Particularly for 5 µm filters, peak transmittance increases from 14% to 46% with the Al reflectors [Fig. 3(i)]. The observed trend for the fabricated set is similar for sizes down to 10 µm [Fig. 3(c)–3(h)] while 5 µm filters show significant enhancement as peak transmittance increases from 28% to 57% while FWHM reduces by 7 nm [Fig. 3(i)]. The slight blueshift in peak position can be adjusted by redshifting the center wavelength, λ0 in design to obtain the transmission peak at the desired wavelength. This can be easily achieved by adjusting the FFs [29,30]. It should be noted that the Al reflectors could also be placed near the field nodes inside the Si stripes instead of near those inside the trenches.
Fig. 4. Electric field profile simulated for filters having design parameters [FF1(Si) = 13.0%, FF1(Air) = 19.6%, FF2(Si) = 40.2%, FF2(Air) = 27.2%] and (a) infinite width, (b) finite width with 6 periods and no Al reflector, and (c) finite width with 6 periods with Al reflectors at the optimum positions. The trenches in (c) are extended by 60 nm to the left and 100 nm to the right to make sure reflectors are placed at the nodes.
For comparison FOM values of the experimental response, design and fabricated parameter sets with and without Al reflectors are plotted in Fig. 5. For each set, FOM gradually decreases with reducing filter size. With or without the reflectors, FOM values for the design set show similar trend with respect to size variation. The FOM variation from 112 to 6 periods is 21.1 to 1.8 without reflectors while it is 22.1 to 8.1 with Al reflectors placed at the optimum positions. With the reflectors in place, FOM increases marginally (< 1) for larger filters from 112 to 37 periods. For the smaller filters from 19 periods down to 6 periods, the improvement is significant as evident from Fig. 5. For 6 periods, the Al reflectors improve the FOM significantly to 8.1 from 1.8. For the fabricated set, the Al reflectors increase FOM values marginally down to 19 periods. The smaller filters with 12 and 6 periods show visible improvement in FOM, particularly for the latter case where FOM is increased to 5.8 from 1.8. This significant improvement in FOM is possible because of the metal reflectors which re-navigate the incident light back to the stripe layer which further interacts with the guided modes, depicting an infinite size behavior. This phenomenon leads to stronger resonance at the peak improving the FOM. Thus, use of metal reflectors for the smaller size filters is needed to recover the transmittance response while it is redundant for the larger size filters. With this observation in mind, it would be possible to detect narrow spectral lines in the SWIR wavelength range with 5 µm size filters if metal claddings were placed on both sides of the stripe layer. Four such filters of 5µm could be placed in a mosaic to form a pixel for multispectral imaging. Then, multiple chemical signature lines can be detected in the SWIR range with 20 µm × 20 µm size mosaic pixels.
Fig. 5. Figure-of-merit (FOM) of experimental (Exp.) responses, and simulations with fabricated (Fab.) and design parameters with and without Al reflectors. When Al reflectors are placed, the 5 µm filters show almost 3 times improvement in FOM for the fabricated parameters and 4.5 times for the design parameters.
4. Conclusion
We have explored GMR filters as viable candidates to form pixels for multispectral imaging in the SWIR region. We have experimentally demonstrated that GMR transmission filters can be miniaturized from 90 µm × 90 µm to 10 µm × 10 µm with reasonable agreement in transmittance spectra between filters of different sizes. The 10 µm × 10 µm filter can sustain 65% peak transmittance with 26 nm FWHM, yielding a 13% drop in peak transmittance compared to the 90 µm × 90 µm filter while FWHM changes by 2 - 3 nm. FOM gradually reduces from 8.5 to 4.4 with decreasing filter size from 90 µm × 90 µm to 10 µm × 10 µm. Metal reflectors can be integrated at the boundaries of a 5 µm wide filter to enhance peak transmittance by 32% and consequently increase FOM value by more than 4.5 times, which would be sufficient to detect narrow spectral lines in the SWIR range. With this capability within a wide stopband up to 281 nm, a mosaic can be formed by placing 5 µm × 5 µm size filters of different resonance wavelengths with metal reflectors to form a single pixel of 20 µm × 20 µm size that would pave the way for tunable multispectral imaging in the SWIR range. This would further provide the platform to realize polarization independent SWIR imagers utilizing two-dimensional (2D) gratings.
Funding
Night Vision and Electronic Sensors Directorate (W909MY-17-1-001).
Acknowledgments
This research was funded by the U.S. Army Communications, Electronics, Research, Development and Engineering Center Night Vision & Electronic Sensors Directorate. The authors thank Shin-Etsu Chemical, Co., Ltd., Japan, for providing the SOQ wafers used in this research. The filter structures were fabricated in the C. Kenneth and Dianne Harris Wright Virginia Microelectronics Center (VMC) at Virginia Commonwealth University.
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 request.
References
1. R. G. Vaughan, W. M. Calvin, and J. V. Taranik, “SEBASS hyperspectral thermal infrared data: surface emissivity measurement and mineral mapping,” Remote Sens. Environ. 85(1), 48–63 (2003). [CrossRef]
2. F. A. Kruse, J. W. Boardman, and J. F. Huntington, “Comparison of airborne hyperspectral data and EO-1 Hyperion for mineral mapping,” IEEE Trans. Geosci. Remote Sens. 41(6), 1388–1400 (2003). [CrossRef]
3. D. P. Roy, M. A. Wulder, T. R. Loveland, C. E. Woodcock, R. G. Allen, M. C. Anderson, D. Helder, J. R. Irons, D. M. Johnson, R. Kennedy, and T. A. Scambos, “Landsat-8: Science and product vision for terrestrial global change research,” Remote Sens. Environ. 145, 154–172 (2014). [CrossRef]
4. Z. Zhu and C. E. Woodcock, “Object-based cloud and cloud shadow detection in Landsat imagery,” Remote Sens. Environ. 118, 83–94 (2012). [CrossRef]
5. R. Lu and Y. R. Chen, “Hyperspectral imaging for safety inspection of food and agricultural products,” Proc. SPIE 3544, 121–133 (1999). [CrossRef]
6. E. Ben-Dor, S. Chabrillat, J. A. M. Demattê, G. R. Taylor, J. Hill, M. L. Whiting, and S. Sommer, “Using imaging spectroscopy to study soil properties,” Remote Sens. Environ. 113, S38–S55 (2009). [CrossRef]
7. D. J. Eisenstein, D. H. Weinberg, E. Agol, H. Aihara, C. A. Prieto, S. F. Anderson, and J. A. Arns, “SDSS-III: Massive spectroscopic surveys of the distant universe, the Milky Way, and extra-solar planetary systems,” The Astron. J. 142(3), 72 (2011). [CrossRef]
8. S. Lohumi, S. Lee, H. Lee, and B. K. Cho, “A review of vibrational spectroscopic techniques for the detection of food authenticity and adulteration,” Trends Food Sci. Technol. 46(1), 85–98 (2015). [CrossRef]
9. R. M. Levenson and J. R. Mansfield, “Multispectral imaging in biology and medicine: slices of life,” Cytometry, Part A 69A(8), 748–758 (2006). [CrossRef]
10. T. Zimmermann, J. Rietdorf, and R. Pepperkok, “Spectral imaging and its applications in live cell microscopy,” FEBS Lett. 546(1), 87–92 (2003). [CrossRef]
11. H. Liu, X. Lv, J. Qian, H. Li, Y. Qian, X. Wang, X. Meng, W. Lin, and H. Wang, “Graphitic Carbon Nitride Quantum Dots Embedded in Carbon Nanosheets for Near-Infrared Imaging-Guided Combined Photo-Chemotherapy,” ACS Nano 14(10), 13304–13315 (2020). [CrossRef]
12. J. Depciuch, E. Kaznowska, I. Zawlik, R. Wojnarowska, M. Cholewa, P. Heraud, and J. Cebulski, “Application of Raman spectroscopy and infrared spectroscopy in the identification of breast cancer,” Appl. Spectrosc. 70(2), 251–263 (2016). [CrossRef]
13. X. Gao, Y. Cui, R. M. Levenson, L. W. K. Chung, and S. Nie, “In vivo cancer targeting and imaging with semiconductor quantum dots,” Nat. Biotechnol. 22(8), 969–976 (2004). [CrossRef]
14. F. D. Van Der Meer, H. M. A. Van Der Werff, F. J. A. Van Ruitenbeek, C. A. Hecker, W. H. Bakker, M. F. Noomen, M. Van Der Meijde, E. J. M. Carranza, J. B. D. Smeth, and T. Woldai, “Multi-and hyperspectral geologic remote sensing: A review,” ITC J. 14(1), 112–128 (2012). [CrossRef]
15. T. Slonecker, G. B. Fisher, D. P. Aiello, and B. Haack, “Visible and infrared remote imaging of hazardous waste: a review,” Remote Sens. 2(11), 2474–2508 (2010). [CrossRef]
16. W. Y. Jang, Z. Ku, A. Urbas, J. Derov, and M. J. Noyola, “Plasmonic superpixel sensor for compressive spectral sensing,” IEEE Trans. Geosci. Remote Sens. 53(6), 3471–3480 (2015). [CrossRef]
17. A. M. Valm, S. Cohen, W. R. Legant, J. Melunis, U. Hershberg, E. Wait, A. R. Cohen, M. W. Davidson, E. Betzig, and J. Lippincott-Schwartz, “Applying systems-level spectral imaging and analysis to reveal the organelle interactome,” Nature 546(7656), 162–167 (2017). [CrossRef]
18. Z. Zhong, J. Li, Z. Luo, and M. Chapman, “Spectral–spatial residual network for hyperspectral image classification: A 3-D deep learning framework,” IEEE Trans. Geosci. Remote Sens. 56(2), 847–858 (2018). [CrossRef]
19. B. H. Cheong, O. N. Prudnikov, E. Cho, H. S. Kim, J. Yu, Y. S. Cho, H. Y. Choi, and S. T. Shin, “High angular tolerant color filter using subwavelength grating,” Appl. Phys. Lett. 94(21), 213104 (2009). [CrossRef]
20. A. Kuze, H. Suto, M. Nakajima, and T. Hamazaki, “Thermal and near infrared sensor for carbon observation Fourier-transform spectrometer on the Greenhouse Gases Observing Satellite for greenhouse gases monitoring,” Appl. Opt. 48(35), 6716–6733 (2009). [CrossRef]
21. A. Piegari and E. F. Flory, Optical Thin Films and Coatings: From Materials to Applications (Woodhead Publishing, 2018).
22. D. Inoue, A. M. Daisuke, T. Nomura, H. Fujikawa, K. Sato, N. Ikeda, D. Tsuya, Y. Sugimoto, and Y. Koide, “Polarization independent visible color filter comprising an aluminum film with surface-plasmon enhanced transmission through a subwavelength array of holes,” Appl. Phys. Lett. 98(9), 093113 (2011). [CrossRef]
23. Q. Chen and D. R. S. Cumming, “High transmission and low color cross-talk plasmonic color filters using triangular-lattice hole arrays in aluminum films,” Opt. Express 18(13), 14056–14062 (2010). [CrossRef]
24. Y. T. Yoon and S. S. Lee, “Transmission type color filter incorporating a silver film based etalon,” Opt. Express 18(5), 5344–5349 (2010). [CrossRef]
25. K. T. Lee, S. Seo, J. Young Lee, and L. Jay Guo, “Ultrathin metal-semiconductor-metal resonator for angle invariant visible band transmission filters,” Appl. Phys. Lett. 104(23), 231112 (2014). [CrossRef]
26. E. H. Cho, H. S. Kim, B. H. Cheong, P. Oleg, W. Xianyua, J. S. Sohn, D. J. Ma, H. Y. Choi, N.-C. Park, and Y. P. Park, “Two-dimensional photonic crystal color filter development,” Opt. Express 17(10), 8621–8629 (2009). [CrossRef]
27. Y. Kanamori, M. Shimono, and K. Hane, “Fabrication of transmission color filters using silicon subwavelength gratings on quartz substrates,” IEEE Photonics Technol. Lett. 18(20), 2126–2128 (2006). [CrossRef]
28. M. Niraula, J. W. Yoon, and R. Magnusson, “Single-layer optical bandpass filter technology,” Opt. Lett. 40(21), 5062–5065 (2015). [CrossRef]
29. R. C. Ng, J. C. Garcia, J. R. Greer, and K. T. Fountaine, “Miniaturization of a-Si guided mode resonance filter arrays for near-IR multi-spectral filtering,” Appl. Phys. Lett. 117(11), 111106 (2020). [CrossRef]
30. B. R. Green, V. Avrutin, Ü Özgür, N. K. Dhar, and E. Topsakal, “Finite size narrow-band transmission filters for real-time short wave IR spectroscopy and imaging,” Proc. SPIE 10656, 106561U (2018). [CrossRef]
31. Y. Khorrami, D. Fathi, and R. C. Rumpf, “Guided-mode resonance filter optimal inverse design using one-and two-dimensional grating,” J. Opt. Soc. Am. B 37(2), 425–432 (2020). [CrossRef]
32. S. S. Wang and R. Magnusson, “Theory and applications of guided-mode resonance filters,” Appl. Opt. 32(14), 2606–2613 (1993). [CrossRef]
33. M. Mamun, R. Sayeed, M. Gigante, Ü Özgür, and V. Avrutin, “Dual-period guided-mode resonance filters for SWIR multi-spectral image sensors,” Opt. Lett. 46(9), 2240–2243 (2021). [CrossRef]
34. J. O. Grepstad, M. M. Greve, B. Holst, I. R. Johansen, O. Solgaard, and A. Sudbø, “Finite-size limitations on Quality Factor of guided resonance modes in 2D Photonic Crystals,” Opt. Express 21(20), 23640–23654 (2013). [CrossRef]
35. C. Schinke, P. Christian Peest, J. Schmidt, R. Brendel, K. Bothe, M. R. Vogt, I. Kröger, S. Winter, A. Schirmacher, S. Lim, H. T. Nguyen, and D. MacDonald, “Uncertainty analysis for the coefficient of band-to-band absorption of crystalline silicon,” AIP Adv. 5(6), 067168 (2015). [CrossRef]
36. K. M. McPeak, S. V. Jayanti, S. J. P. Kress, S. Meyer, S. Iotti, A. Rossinelli, and D. J. Norris, “Plasmonic films can easily be better: Rules and recipes,” ACS Photonics 2(3), 326–333 (2015). [CrossRef]
37. Z. Wang, L. Ni, H. Zhang, H. Zhang, J. Jin, C. Peng, and W. Hu, “Mode splitting in high-index-contrast grating with mini-scale finite size,” Opt. Lett. 41(16), 3872–3875 (2016). [CrossRef]
38. T. Sang, X. Yin, H. Qi, J. Gao, X. Niu, and H. Jiao, “Resonant excitation analysis on asymmetrical lateral leakage of light in finite zero-contrast grating mirror,” IEEE Photonics J. 12(2), 1–11 (2020). [CrossRef]
39. E. H. Eriksen, A. Nazir, P. Balling, J. Vester-Petersen, R. E. Christiansen, O. Sigmund, and S. P. Madsen, “Dose regularization via filtering and projection: An open-source code for optimization-based proximity-effect-correction for nanoscale lithography,” Microelectron. Eng. 199, 52–57 (2018). [CrossRef]
