검색
검색 팝업 닫기

Ex) Article Title, Author, Keywords

Article

Split Viewer

Research Paper

Curr. Opt. Photon. 2023; 7(5): 485-495

Published online October 25, 2023 https://doi.org/10.3807/COPP.2023.7.5.485

Copyright © Optical Society of Korea.

Covered Microlens Structure for Quad Color Filter Array of CMOS Image Sensor

Jae-Hyeok Hwang1, Yunkyung Kim1,2

1Department of ICT Integrated Safe Ocean Smart Cities Engineering, Dong-A University, Busan 49315, Korea
2Department of Electronics Engineering, Dong-A University, Busan 49315, Korea

Corresponding author: *yunkkim@dau.ac.kr, ORCID 0000-0002-4338-7642

Received: April 3, 2023; Revised: June 28, 2023; Accepted: September 8, 2023

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

The pixel size in high-resolution complementary metal-oxide-semiconductor (CMOS) image sensors continues to shrink due to chip size limitations. However, the pixel pitch’s miniaturization causes deterioration of optical performance. As one solution, a quad color filter (CF) array with pixel binning has been developed to enhance sensitivity. For high sensitivity, the microlens structure also needs to be optimized as the CF arrays change. In this paper, the covered microlens, which consist of four microlenses covered by one large microlens, are proposed for the quad CF array in the backside illumination pixel structure. To evaluate the optical performance, the suggested microlens structure was simulated from 0.5 μm to 1.0 μm pixels at the center and edge of the sensors. Moreover, all pixel structures were compared with and without in-pixel deep trench isolation (DTI), which works to distribute incident light uniformly into each photodiode. The suggested structure was evaluated with an optical simulation using the finite-difference time-domain method for numerical analysis of the optical characteristics. Compared to the conventional microlens, the suggested microlens show 29.1% and 33.9% maximum enhancement of sensitivity at the center and edge of the sensor, respectively. Therefore, the covered microlens demonstrated the highly sensitive image sensor with a quad CF array.

Keywords: CMOS image sensor, Covered microlens, FDTD simulation, Microlens, Quad color filter array

OCIS codes: (040.5160) Photodetectors; (130.6010) Sensors; (220.3630) Lenses; (230.0230) Optical devices; (280.4788) Optical sensing and sensors

In the mobile camera market, the demand for high-resolution complementary metal-oxide-semiconductor (CMOS) image sensors has increased [1]. However, pixel pitch has been reduced continuously because of the limited image sensor size [2]. Although recent trends have led to a tendency to reduce pixel pitch, it is hard to reduce the thickness of silicon, color filters (CFs), and microlenses. Silicon has different absorption coefficients depending on the wavelength [3]. Thus, the silicon should be thick enough to absorb light of a long wavelength and compensate for the degradation of sensitivity caused by pixel shrinking. Therefore, the silicon thickness of the photodiode tends to be in the range of 2.5 µm to 3.7 µm [2]. To prevent the transmission of unwanted wavelengths of light into the silicon, CFs between 0.5 µm to 0.8 µm thick are used [4]. A microlens with an appropriate height and radius of curvature (ROC) contributes to improved pixel sensitivity. As the ROC increases, the focal point is lengthened due to flattened microlens. Therefore, optimized microlens height and ROC are significantly related to the optical performance. The pixel pitch has been continuously decreasing, while reducing pixel thickness degrades sensitivity. As a result of pixel shrinking and increased aspect ratio, optical performance, such as sensitivity and crosstalk, has declined. Optical performance is an important factor to implement for the high sensitivity of CMOS image sensors.

Many research efforts have been devoted to increasing sensitivity and decreasing crosstalk. To suppress optical and electrical crosstalk, deep trench isolation (DTI) was proposed [5]. The DTI which is a thin dielectric is placed between neighboring pixels and acts as a total reflection barrier for the complete physical isolation of pixels. However, the dark current occurs because of stress caused by the DTI process. Thus, the metal or doped poly-Si is filled into the DTI as a metal-insulator-silicon capacitor. Crosstalk and dark currents are suppressed by holes accumulated on the DTI surface [6, 7]. Furthermore, a negative DC bias is applied to the poly-Si between the DTI to accumulate more holes [8]. Moreover, to suppress crosstalk, a metal grid is inserted into the passivation layer under the CF [9]. A metal grid is also inserted between each CF to resolve the diffraction limit issue in sub-micron pixels. Moreover, the dielectric grid and air gap grid have been proposed to decrease optical loss from the metal [8, 10]. The DTI and grid act as a total reflection planes. Therefore, the effective depth of silicon is increased so that the sensitivity of long wavelengths is enhanced. However, the DTI and grid reduce the light-receiving area. For the small pixel under the sub-micron, it is hard to reduce the width of DTI for keeping the light-receiving area. Moreover, the DTI must be inserted deep enough to suppress crosstalk. Therefore, pixel shrinking requires a more precise process due to the degraded aspect ratio of the DTI.

To overcome the degraded aspect ratio of pixel structures, advanced nanotechnologies, such as plasmonic CFs and metalenses, have been studied [11, 12]. The plasmonic metal filters, composed of periodic hole arrays, act as optical filters due to the interference of surface plasmon polaritons. By changing the hole size, shape, and separation, a single thin metal layer can control the transmission spectra of the hole array. However, the transmission of plasmonic CFs is relatively low compared to conventional dye-doped filters. The metalens consists of a combination of dielectric nanopillars. The optimized nanopillars are capable of guiding individual primary colors into each pixel. Thus, the metalens gains a large collection area of white light as an integration of the microlens and CF. However, the dispersion of the nanopillars can contribute to crosstalk. Instead of silicon, 2D materials, such as graphene, have been proposed to reduce pixel thickness with increased flexibility [13, 14]. However, the performance and process compatibility for the demonstration of nanoscale structures is still far from commercialization in the mobile camera market.

Various CF patterns and demosaicking have been suggested as alternative methods for sensitivity enhancement [1518]. Moreover, the quad CF array was introduced with pixel binning [1921]. The quad CF array has four photodiodes under the adjacent CFs of the same color. Pixel binning combines spatial resolution to 1/4 of the basic size [22]. In low illuminance conditions, four pixels are combined to process more light. In high illuminance conditions, the pixel binning algorithm improves resolution by processing individual pixels separately. Thus, the quad CF array and pixel binning have the benefit of a high dynamic range regardless of light intensity conditions. However, studies on the optical structure are still needed to increase sensitivity to the conditions of the quad CF.

In this paper, a covered microlens is proposed. This proposed microlens is applied to the quad CF array for the high sensitivity of small pixels in the backside illumination (BSI) structure [5, 9]. In Section II, the concept of the covered microlens pixel is described, and Section III includes the simulation results and a discussion of the sensitivity of the condition with and without in-pixel DTI and the sensitivity of the center and edge of the sensor. Section Ⅳ presents the conclusions of the proposed covered microlens.

In this paper, the previous structures and suggested structures were compared to confirm the improved sensitivity. Figure 1 shows the three types of microlenses, including the suggested covered microlens in the BSI pixel structure. The conventional microlens and 2×2 on-chip lens (2×2OCL) are two types of widely used microlenses with quad CF arrays [23]. In Fig. 1(a), the conventional microlens has the advantage of focusing on the incident light just below the arranged photodiode. However, it is not enough to concentrate the incident light due to the optical loss from the diffraction in the sub-micron pixels. Therefore, a higher and wider microlens is needed for high sensitivity. As shown in Fig. 1(b), the 2×2OCL has enhanced sensitivity because of its wide and high microlens [23]. However, the focal point of the 2×2OCL is between pixels, not at the center of each photodiode as in the beam profile of Fig. 1(b). For the pixel having the quad color filter array, the pixels have four photodiodes under the stacked microlens and color filter. The excited electrons by photons are collected at the photodiode. Therefore, the focal point of 2×2OCL has a limit to collect excited electrons in the four photodiodes of a pixel having the quad CF array. Figure 1(c) shows the suggested pixel structure with the covered microlens. The main feature of the suggested structure is the large lens that covers the four conventional microlenses. The covered microlens enhances the concentration efficiency of the incident light. The base microlens distributes the incident light further into the middle of the four photodiodes as in the beam profile of Fig. 1(c). Thus, the sensitivity is increased due to the two lenses on top of the CF.

Figure 1.The concept drawings for 3D, top, 2D cross-section of a diagonal cut, and beam profile views of pixel structure with the (a) conventional microlens, (b) 2×2 on-chip lens (2×2OCL), and (c) covered microlens.

The key factor of the covered microlens is the difference in the refractive index. To increase the incident light into the photodiode, the reflectance should be minimized [24]. Therefore, the difference in refractive index between mediums should be reduced to minimize the loss of incident light, and the covered microlens pixel should also have a sequentially increasing refractive index from the air to the optical stacks. Moreover, the covered microlens should have a lower refractive index than the base microlens.

Figure 2 shows that the simulation results of reflectance depend on the refractive index of the covered microlens and the base microlens. In the case of typical microlens for CMOS image sensors, the refractive index of materials having around 1.6 has been widely used [25]. Moreover, materials having a high refractive index have been required to concentrate the incident light to the photodiode [26]. To increase sensitivity effectively, materials having the highest refractive index of around 2.1 have been developed at visual light wavelength [26]. In this paper, refractive indices of the covered microlens, ranging from 1.6 to 2.1, were tested. Figure 2(a) shows the covered microlens with a lower refractive index than the base microlens. The reflectance remained almost constant regardless of the refractive index of the base microlens. On the other hand, Fig. 2(b) shows the covered microlens with a higher refractive index than the base microlens. The reflectance increased as the refractive index of the covered microlens increased. The structure in Fig. 2(b) has a relatively large refractive index difference between the air and covered microlens. Therefore, the sensitivity is degraded due to reflection between the air, covered microlens, and base microlens. In the case of the covered microlens having a refractive index of 1.9, the reflectance was suddenly raised due to thin-film interference from two microlenses. However, thin-film interference did not affect for the overall trend. Thus, reflectance shows a tendency to increase as the refractive index of stacked materials increases. Moreover, the oblique incident light enters more obliquely due to refraction, and this issue causes crosstalk.

Figure 2.The simulated reflectance of (a) the covered microlens with a lower refractive index than the base microlens and (b) the covered microlens with a higher refractive index than the base microlens.

The reflection between mediums is one of the important factors causing optical loss. Reflectance increases as the difference in refractive index between mediums increases. To minimize optical loss, the optical stacks were located by gradually increasing refractive indices from air to the base microlens. Therefore, in this paper, the refractive index of the covered microlens was fixed at the lowest value of 1.6, and the base microlens refractive indices varied between 1.7 and 2.1.

The optical performance of the covered microlens pixels in the sub-micron was simulated using a TCAD Sentaurus for considering diffraction and refraction [27]. The finite-difference time-domain (FDTD) method is widely used to solve Maxwell’s equation for numerical analyses of optoelectronic devices. In the TCAD simulation, the absorbed photon density is calculated by solving the temporal evolution of electromagnetic waves in the pixel structure [28]. Moreover, the beam profile visually shows the absorbed photon density of the pixel as in Fig. 1. To investigate sensitivity, the wavelengths of incident light used were 650, 540, and 450 nm. RGB Bayer was used as a quad CF array. The sensitivity which is one of the most important performance measures is defined as the ratio of the absorbed light to the incident light [3]. On the condition of the same intensity of the incident light, the sensitivity is proportional to the absorbed photon density, which means the absorbed light. Therefore, the high absorbed photon density indicates high sensitivity because the electrons are generated by absorbed light in silicon. In this paper, the sensitivity is the sum of absorbed photon density by 16 pixels under the red, green, and blue CFs. The absorbed photon density (Aopt) is calculated by the absorbed power density divided by the photon energy as in Eq. (1) [28, 29].

Aopt=·Savhv=12hvσE2.

Sav, σ, and are noted with time-averaged Poynting vector, non-zero conductivity, and photon energy, respectively. Ε is noted with the electric field by the incident photon’s energy. As parameters, non-zero conductivity includes a complex permittivity and a complex refractive index. The absorbed photon density is calculated in the silicon area of the pixel.

The pixels less than 1.0 μm are critically affected by pixel shrinking. For high resolution image sensors, the 0.5 μm pixel was reported as the smallest pixel [30]. Therefore, the conventional microlens, 2×2OCL [23] and suggested covered microlens structure were simulated with 1.0 µm to 0.5 µm pixel pitches. For optimization of the microlens, ROC and height varied in the steps of 0.1 μm, respectively. Moreover, the refractive index of the base microlens located under the covered microlens was determined between 1.7 and 2.1. The parameters of optimized microlenses and DTI showing the most enhanced sensitivity are shown in Table 1. From the result, the optimized microlenses show irregular patterns in parameters as the pixel pitch changes due to strong diffraction causing limited spot size [3133]. Moreover, the pixel has a smaller width of DTI as the pixel size gets smaller [34]. For the simulated pixels in this paper, the heights of the silicon, CF, and passivation layer were 3.0 μm, 0.6 μm, and 0.2 μm which is generally used in CMOS image sensors [2, 4]. The red, green, and blue CF have a refractive index of 1.9 at the wavelength of 650 nm, 1.6 at the wavelength of 540 nm, and 1.6 at the wavelength of 450 nm, respectively [35]. Also, the metal grid and passivation layer consisted of tungsten and HfO2 which have been conventionally used [36].

TABLE 1 Parameters of optimized microlens and deep trench isolation (DTI) for each pixel pitch

Pixel Pitch (µm)1.00.90.80.70.60.5
Conventional MLheight [µm]0.60.40.40.30.20.1
ROC [µm]0.60.90.50.40.40.9
Refractive Index (@650 nm)1.6
2×2OCLHeight [µm]1.41.30.70.51.00.7
ROC [µm]1.41.61.21.20.90.7
Refractive Index (@650 nm)1.6
Covered MLBase MLHeight [µm]0.40.80.40.30.30.1
ROC [µm]0.40.70.40.40.30.6
Refractive Index (@650 nm)1.71.81.71.91.81.9
Covered MLHeight [µm]1.41.30.70.51.00.7
ROC [µm]1.41.61.21.20.90.7
Refractive Index (@650 nm)1.6
DTIWidth [nm]1008065453525
Refractive Index (@650 nm)1.9


To confirm the optical performance of the suggested covered microlens, we compared with the 2×2OCL and conventional microlens. Moreover, the suggested structure was simulated according to the location of the pixels having different chief ray angles (CRAs) in the image sensor [37], and also with and without in-pixel DTI. The in-pixel DTI has a function to separate four pixels under the same color of quad CF array [23, 38]. Depending on the pixel location in the image sensor, the main ray of incident light that enters each pixel from the camera module has different angles. The CRA represents the incident angle of the main ray. In this paper, we consider that the CRA of the edge pixel is 35° and the CRA of the center pixel is 0° [39]. At 0°, the incident light enters the middle of the pixels. On the other hand, at 35°, the incident light enters obliquely from the lens of the camera module to the microlens of the image sensor. Therefore, the alignment of the optical stacks was significant as the image sensor size increased. To guide the oblique incident light into the photodiode, the pixels at the CRA 35° were simulated by shifting the microlens, and CF regardless of in-pixel DTI.

The in-pixel DTI indicates the DTI for separating the four photodiodes under the same CF. In the case of the CRA 35°, the sensitivity disparity of adjacent pixels under the same CF array occurred due to the absorption coefficient of silicon. The in-pixel DTI between the four photodiodes distributes the incident light equally into each pixel. To resolve the sensitivity disparity, the in-pixel DTI has been used in the high-performance image sensor. Therefore, in this paper, with and without in-pixel DTIs were investigated with CRA 0° and CRA 35°.

3.1. Sensitivity at a CRA 0°

In this section, we investigated the sensitivity of the suggested structure at CRA 0°. At first, the conventional microlens, 2×2OCL, and covered microlens were compared without in-pixel DTI. Figure 3(a) shows the simulated pixel with the covered microlens without in-pixel DTI at a CRA 0°. Figure 3(b) shows the normalized sensitivity to the incident light at 0° from 1.0 µm to 0.5 µm pixel pitches. To compare the optical performance of the suggested structure, the sensitivity of the 1.0 µm pixel with the covered microlens and the in-pixel DTI, which shows the highest sensitivity, was set as 1. From the simulation results, the covered microlens pixels were more sensitive than the pixels of the 2×2OCL and conventional microlens, regardless of pixel pitch. Moreover, the difference in sensitivity between the covered microlens and the conventional microlens increased as the pixel pitch decreased. Figure 3(c) shows the sum of the sensitivities when the incident light enters the pixels at 0°, 10°, and 20°. Compared to the conventional microlens and 2×2OCL, the covered microlens pixel indicated higher sensitivity, regardless of pixel pitch and incident angle. Compared to the conventional microlens pixel and the covered microlens pixel, the pixel having the covered microlens with a 0.5 μm pixel pitch showed the highest enhancement of sensitivity, and the sensitivity increased by 28.7% and 22.0% at the incident angle of 0° and the sum of 0°, 10°, and 20°, respectively.

Figure 3.Simulation results without in-pixel DTI at a CRA 0° of (a) the pixel structure, (b) the sensitivity of the incident angle at 0°, and (c) the sum of sensitivities of the incident angles at 0°, 10°, and 20°. The results of the suggested covered ML are compared with 2×2OCL, and conventional ML from 1.0 µm to 0.5 µm pixel pitches. DTI, deep trench isolation; CRA, chief ray angle; 2×2OCL, 2×2 on-chip lens; ML, microlens.

Next, the conventional microlens, 2×2OCL, and covered microlens were compared with in-pixel DTI. Figure 4(a) shows the suggested pixel with in-pixel DTI at CRA 0°. Compared to Fig. 3(a), the pixel has an in-pixel DTI that is located in the middle of each pixel. Figures 4(b) and 4(c) show the sensitivity of the incident angle at 0° and the sum of 0°, 10°, and 20°. Figures 4(b) and 4(c) have a similar tendency to Figs. 3(b) and 3(c). Regardless of pixel pitch, the covered microlens pixels had the highest sensitivity compared to the conventional microlens and 2×2OCL. Compared to the conventional microlens pixels in the simulations, the sensitivity of suggested pixels with 0.5 µm pixel pitch increased the maximum by 29.1% and 22.2% at the incident angle of 0° and the sum of 0°, 10°, and 20°, respectively. Simulation results show a high sensitivity improvement rate in small pixels. As pixel sizes get smaller, the sensitivity is limited due to the diffraction limit. Therefore, the sensitivity is most improved with 0.5 μm pixels having the covered microlens of large width.

Figure 4.Simulation results with in-pixel DTI at a CRA 0° of (a) the pixel structure, (b) the sensitivity of the incident angle at 0°, and (c) the sum of sensitivities of the incident angles at 0°, 10°, and 20°. The results of the suggested covered ML are compared with 2×2OCL, and conventional ML from 1.0 µm to 0.5 µm pixel pitches. DTI, deep trench isolation; CRA, chief ray angle; 2×2OCL, 2×2 On-chip lens; ML, microlens.

3.2. Sensitivity at a CRA 35°

In this section, the sensitivity of the suggested structure was investigated at CRA 35°. Before comparing with the conventional and the suggested structures, the shift distance of microlens and a CF at a CRA 35° is optimized for the incident light to enter the photodiode without loss. The shifted microlens and CF were simulated as moving in the unit of 0.01 μm to determine the exact shift distance. Thus, the incident light enters the photodiode as a minimized degradation of optical performance when the incident light is at 35°. Figure 5 shows the simulation results of the shift distance and the relationship between the microlens height and the shift distance. Figure 5(a) shows the simulation results of the conventional microlens. As the pixel pitch decreased, the microlens height and the shift distance decreased. Figures 5(b) and 5(c) show the shift distance of the 2×2OCL and covered microlens. The simulated results show that the shift distance is determined by the microlens height. For instance, in the case of the 2×2OCL and covered microlens, the optimized 0.6 µm pixel has a microlens height of 1.0 µm, and a 0.7 µm pixel has a microlens height of 0.5 µm. Compared to the 0.7 µm pixels, the microlens of the 0.6 µm pixel with a high microlens is shifted more. Therefore, we confirmed that the shift distance of the covered microlens depends on the microlens height at a CRA 35°.

Figure 5.The microlens and CF shift distance according to the microlens height of the (a) conventional microlens, (b) 2×2OCL and (c) covered microlens at a CRA 35°. CF, color filter; 2×2OCL, 2×2 on-chip lens; CRA, chief ray angle.

Figure 6(a) shows the pixel with the covered microlens without in-pixel DTI. The sensitivity was also set to normalize based on 1.0 µm pixel having the covered microlens and in-pixel DTI at a CRA 0° in Fig. 4. Figure 6(b) shows the simulation results of the sensitivity at the 35° incident light. Compared to the conventional microlens pixel, the 0.5 µm pixel having the covered microlens shows a maximum improvement of 33.9%. Figure 6(c) shows the sum of the sensitivity graph at 35°, 45°, and 55° incident angles. The enhancement of suggested pixels with 0.7 µm pixel pitch increased to 18.6%.

Figure 6.Simulation results without in-pixel DTI at a CRA 35° of (a) the pixel structure, (b) the sensitivity of the incident angle at 35°, and (c) the sum of sensitivities of incident angles 35°, 45°, and 55° at a CRA 35°. The results of the suggested covered ML are compared with 2×2OCL, and conventional ML from 1.0 µm to 0.5 µm pixel pitches. DTI, deep trench isolation; CRA, chief ray angle; 2×2OCL, 2×2 On-chip lens; ML, microlens.

Figure 7(a) shows the pixel with the covered microlens with in-pixel DTI. Figure 7(b) shows the graph of sensitivity at a 35° incident angle. Compared to the conventional microlens pixel and suggested pixel, 0.5 µm pixel having the covered microlens shows a maximum enhancement of 33.5%. Figure 7(c) shows the sum of sensitivities at 35°, 45°, and 55° incident angles; The maximum improvement of the 0.7 µm pixel having the covered microlens was 20.5%.

Figure 7.Simulation results with in-pixel DTI of (a) the pixel structure, (b) the sensitivity of the incident angle at 35°, and (c) the sum of sensitivities of incident angles 35°, 45°, and 55° at a CRA 35° from 1.0 µm to 0.5 µm pixel pitches. DTI, deep trench isolation; CRA, chief ray angle.

Compared with the conventional microlens and 2×2OCL, the sensitivity of pixels with the covered microlens was improved at a CRA 35°. Therefore, the covered microlens demonstrated enhanced sensitivity at the center and edge of the sensor both.

3.3. Characteristics of the Covered Microlens

From the simulation results from Figs. 37, the trends of sensitivity with pixel pitch is investigated. For CRA 0° and CRA 35°, the sensitivities of pixels having the covered microlens show a similar characteristic. Figure 8 shows the sensitivity enhancement rate associated with pixel pitch. Figure 8(a) shows the incident angles of 0° and 35° at CRA 0° and CRA 35°, respectively. The sensitivity enhancement rate of the pixels with the covered microlens increased as pixel pitch decreased compared to the conventional microlens. The 0.5 µm pixels achieved a sensitivity improvement of over 28.7%, regardless of the CRA and in-pixel DTI. The pixels with the small microlens were strongly affected by the diffraction limit. Therefore, the incident light can be effectively focused into the pixels without optical loss due to the large covered microlens. The 0.5 µm pixel pitch also showed the highest enhancement of sensitivity. Figure 8(b) shows the sums of the sensitivity enhancement rate of incident angles 0°, 10°, 20°, and 35°, 45°, 55° at CRA 0° and CRA 35°, respectively. In the case of the sum of sensitivities with oblique incident angles, the enhancement rate increased according to the decreased pixel pitch at a CRA 0°. Among them, the sensitivity of the 0.5 µm pixel increased to 22.2%. In the case of the CRA 35°, the 0.7 µm pixel indicated the highest improvement by 20.5%.

Figure 8.Simulation results between the covered microlens and the conventional microlens for sensitivity improvement of 1.0 µm to 0.5 µm pixel pitch at chief ray angle (CRA) 0° and CRA 35°: (a) Incident angles 0° and 35°, (b) sums of incident angles 0°, 10°, 20° and 35°, 45°, 55°.

3.4. Discussion

In this paper, the covered microlens pixels showing improved sensitivity for the quad CF array were implemented in the optical simulation. As result, improved sensitivity was shown in the small pixels. As a tendency of the image sensor, the pixels with the quad CF array have emerged as elements of advanced pixels. Moreover, the pixel pitch has been reduced, and the image sensor size has increased. On the other hand, degraded sensitivity is inevitable due to pixel shrinking; Overcoming this issue is necessary for future studies. Considering this trend, the covered microlens for a quad CF array is an appropriate structure for small pixels and highly sensitive CMOS image sensors.

When considering the fabrication to use the process of the thermal reflow which is one of the representative methods for microlens processing, the deformation of the base microlens must be minimized during the process of the covered microlens. In other words, meticulous thermal profile control is required for the fabrication of the covered microlens. To make the desired shapes, the etch-back which has the benefit of providing a wide selection of microlens materials is also required after reflow processing [40]. Therefore, we expect that the proposed structure is possible to be manufactured with the meticulous processing of the thermal reflow and etch-back.

The suggested structures were investigated for sensitivity which is the amount of absorbed light in the silicon area of a pixel. The covered microlens and the 2×2OCL show almost identical sensitivity. However, the covered microlens shows better sensitivity than the 2×2OCL for all simulated structures regardless of pixel pitch. As one of the major advantages of the covered microlens, guided incident light to the photodiodes minimizes the loss of the electron by recombination. Therefore, the rate of increase in quantum efficiency is expected to be conspicuously improved more than the rate of increase in sensitivity. In the future work, quantum efficiency simulation is required to apply the suggested structure for fabrication. Moreover, meticulous fabrication is needed for aligning the base microlens and the covered microlens.

We suggested the covered microlens pixel on the quad CF arrays. The microlens height, ROC, and refractive index of the covered microlens were optimized according to the pixel pitch from 1.0 µm to 0.5 µm. Furthermore, the optimized microlens pixels were simulated with and without in-pixel DTI at CRA 0° and CRA 35° of the image sensor. The suggested structure was compared to the 2×2OCL and conventional microlens to evaluate its optical performance. As a result, the sensitivity of the covered microlens pixel increased compared to the 2×2OCL and conventional microlens, regardless of the pixel pitch and CRA. Compared to the conventional microlens at a CRA 35°, the 0.5 µm pixel with the covered microlens showed the highest sensitivity enhancement, reaching up to 33.9%. Therefore, the covered microlens pixel can contribute as one of the prototypes for pixels having the quad CF array in CMOS image sensors.

Data underlying the results presented in this paper are not publicly available at the time of publication, but may be obtained from the authors upon reasonable request.

  1. S. Yoon, J. Yun, Y. Jung, I. Jeong, J. Choi, W. Choi, J. Lee, H. Lee, and J. Ko, “World largest mobile image sensor with all directional phase detection auto focus function,” in Proc. IEEE Hot Chips 33 Symposium-HCS (Palo Alto, CA, USA, Aug. 22-24, 2021), pp. 1-22.
    CrossRef
  2. R. Fontaine, “The state-of-the-art of smartphone imagers,” in Proc. International Image Sensors Workshop (Snowbird, Utah, USA, Jun. 23-27, 2019), p. Article no. R01.
  3. J. Nakamura, Image Sensors and Signal Processing for Digital Still Cameras (CRC press, USA, 2017), pp. 53-93.
    KoreaMed CrossRef
  4. B. C. Hseih, H. Siddiqui, J. Luo, T. Georgiev, K. Atanassov, S. Goma, H. Y. Cheng, J. J. Sze, R. J. Lin, K. Y. Chou, C. Chao, and S. G. Wuu, “New color filter patterns and demosaic for sub-micron pixel arrays,” in Proc. International Image Sensor Workshop (Vaals, Netherlands, Jun. 8-11, 2015), pp. 8-11.
  5. B. J. Park, J. Jung, C. R. Moon, S. H. Hwang, Y. W. Lee, D. W. Kim, K. H. Paik, J. R. Yoo, D. H. Lee, and K. Kim, “Deep trench isolation for crosstalk suppression in active pixel sensors with 1.7 µm pixel pitch,” Jpn J. Appl. Phys. 46, 2454 (2007).
    CrossRef
  6. Y. Kitamura, H. Aikawa, K. Kakehi, T. Yousyou, K. Eda, T. Minami, S. Uya, Y. Takegawa, H. Yamashita, Y. Kohyama, and T. Asami, “Suppression of crosstalk by using backside deep trench isolation for 1.12 μm backside illuminated CMOS image sensor,” in Proc. IEEE International Electron Devices Meeting-IEDM (San Francisco, CA, USA, Dec. 10-13, 2012), pp. 24.2.1-24.2.4.
    CrossRef
  7. Y. Kim, W. Choi, D. Park, H. Jeoung, B. Kim, Y. Oh, S. Oh, B. Park, E. Kim, Y. Lee, T. Jung, Y. Kim, S. Yoon, S. Hong, J. Lee, S. Jung, C.-R. Moon, Y. Park, D. Lee, and D. Chang, “A 1/2.8-inch 24Mpixel CMOS image sensor with 0.9 μm unit pixels separated by full-depth deep-trench isolation,” in Proc. International Solid-State Circuits Conference-ISSCC (San Francisco, CA, USA, Feb. 11-15, 2018), pp. 84-86.
    CrossRef
  8. S. Park, C. Lee, S. Park, H. Park, T. Lee, D. Park, M. Heo, I. Park, H. Yeo, Y. Lee, J. Lee, B. Lee, D.-C. Lee, J. Kim, B. Kim, J. Pyo, S. Quan, S. You, I. Ro, S. Choi, S. I. Kim, I. S. Joe, J. Park, C. H. Koo, J. H. Kim, C. K. Chang, T. Kim, J. Kim, J. Lee, H. Kim, C. R. Moon, and H. S. Kim, “A 64Mpixel CMOS image sensor with 0.56 μm unit pixels separated by front deep-trench isolation,” in Proc. Solid-State Circuits Conference-ISSCC (San Francisco, CA, USA, Feb. 20-26, 2022), pp. 1-3.
    CrossRef
  9. Y. W. Cheng, T. H. Tsai, C. H. Chou, K. C. Lee, H. C. Chen, and Y. L. Hsu, “Optical performance study of BSI image sensor with stacked grid structure,” in Proc. International Electron Devices Meeting-IEDM (Washington, DC, USA, Dec. 7-9, 2015), pp. 30.4.1-30.4.4.
  10. I. S. Joe, Y. Lee, H. Y. Park, J. U. Kim, D. Kang, T. Kim, M. Kim, K. Lee, M. Heo, I. Ro, J. Kim, I. Park, S. Kwon, K. Yoon, D. Park, C. Lee, E. Jo, M. Jeon, C. Park, K. R. Byun, C. K. Chang, J. Hur, K. Yoon, T. Jeon, J. Lee, J. Park, B. Kim, J. Ahn, H. Kim, C. R. Moon, and H. S. Kim, “Development of advanced inter-color-filter grid on sub-micron-pixel CMOS image sensor for mobile cameras with high sensitivity and high resolution,” in Proc. Symposium on VLSI Technology (Kyoto, Japan, 13-19, Jun. 2021), pp. 1-2.
  11. S. Yokogawa, S. P. Burgos, and H. A. Atwater, “Plasmonic color filters for CMOS image sensor applications,” Nano Lett. 12, 4349-4354 (2012).
    Pubmed CrossRef
  12. B. H. Chen, P. C. Wu, V. C. Su, Y. C. Lai, C. H. Chu, I. C. Lee, J. W. Chen, Y. H. Chen, Y. C. Lan, C. H. Kuan, and D. P. Tsai, “GaN metalens for pixel-level full-color routing at visible light,” Nano Lett. 17, 6345-6352 (2017).
    Pubmed CrossRef
  13. S. Goossens, G. Navickaite, C. Monasterio, S. Gupta, J. J. Piqueras, R. Pérez, G. Burwell, I. Nikitskiy, T. Lasanta, T. Galán, E. Puma, A. Centeno, A. Pesquera, A. Zurutuza, G. Konstantatos, and F. Koppens, “Broadband image sensor array based on graphene-CMOS integration,” Nat. Photonics 11, 366-371 (2017).
    CrossRef
  14. C. Choi, M. K. Choi, S. Liu, M. S. Kim, O. K. Park, C. Im, J. Kim, X. Qin, G. J. Lee, K. W. Cho, M. Kim, E. Joh, J. Lee, D. Son, S. H. Kwon, N. L. Jeon, Y. M. Song, N. Lu, and D. H. Kim, “Human eye-inspired soft optoelectronic device using high-density MoS2-graphene curved image sensor array,” Nat. Commun. 8, 1664 (2017).
    Pubmed KoreaMed CrossRef
  15. B. E. Bayer, “Color imaging array,” U. S. Patent US3971065A (1976).
  16. L. Condat, “Color filter array design using random patterns with blue noise chromatic spectra,” Image Vis. Comput. 28, 1196-1202 (2010).
    CrossRef
  17. J. Wang, C. Zhang, and P. Hao, “New color filter arrays of high light sensitivity and high demosaicking performance,” in Proc. IEEE International Conference on Image Processing (Brussels, Belgium, Sep. 11-14, 2011), pp. 3153-3156.
    Pubmed CrossRef
  18. C. Kwan and B. Chou, “Further improvement of debayering performance of RGBW color filter arrays using deep learning and pansharpening techniques,” J. Imaging 5, 68 (2019).
    Pubmed KoreaMed CrossRef
  19. Y. Kim and Y. Kim, “High-sensitivity pixels with a quad-WRGB color filter and spatial deep-trench isolation,” Sensors 19, 4653 (2019).
    Pubmed KoreaMed CrossRef
  20. Z. Zhimin, B. Pain, and E. R. Fossum, “Frame-transfer CMOS active pixel sensor with pixel binning,” IEEE Trans. Electron Devices 44, 1764-1768 (1997).
    CrossRef
  21. H. Y. Huang, P. A. Conge, and L. W. Huang, “CMOS image sensor binning circuit for low-light imaging,” in Proc. IEEE Symposium on Industrial Electronics and Applications (Langkawi, Malaysia, Sep. 25-28, 2011), pp. 586-589.
    CrossRef
  22. D. Park, S. W. Lee, J. Han, D. Jang, H. Kwon, S. Cha, M. Kim, H. Lee, S. Suh, W. Joo, Y. Lee, S. Nah, H. Jeong, B. Kim, S. Jung, J. Lee, Y. Kim, C. R. Moon, and Y. Park, “A 0.8 µm smart dual conversion gain pixel for 64 Megapixels CMOS image sensor with 12k e-full-well capacitance and low dark noise,” in Proc. International Electron Devices Meeting-IEDM (San Francisco, CA, USA, Dec. 7-11, 2019), pp. 16.2.1-16.2.4.
    KoreaMed CrossRef
  23. T. Okawa, S. Ooki, H. Yamajo, M. Kawada, M. Tachi, K. Goi, T. Yamasaki, H. Iwashita, M. Nakamizo, T. Ogasahara, Y. Kitano, and K. Tatani, “A 1/2inch 48M all PDAF CMOS image sensor using 0.8 µm Quad Bayer coding 2×2OCL with 1.0 lux minimum AF illuminance level,” in Proc. International Electron Devices Meeting-IEDM (San Francisco, CA, USA, Dec. 7-11, 2019), pp. 16.3.1-16.3.4.
    CrossRef
  24. E. Hecht, Optics (Pearson Education, UK, 2017), pp. 129-133.
  25. J. K. Lee, A. Kim, D.W. Kang, and B. Y. Lee, “Efficiency enhancement in a backside illuminated 1.12 μm pixel CMOS image sensor via parabolic color filters,” Opt. Express 24, 16027-16036 (2016).
    Pubmed CrossRef
  26. J. James, A. B. Unni, K. Taleb, J. P. Chapel, N. Kalarikkal, S. Varghese, G. Vignaud, and Y. Grohens, “Surface engineering of polystyrene-cerium oxide nanocomposite thin films for refractive index enhancement,” Nano-Struct. Nano-Objects 17, 34-42 (2019).
    CrossRef
  27. Synopsys Inc., “TCAD Product Family,” (Synopsys Inc.), https://www.synopsys.com/silicon/tcad.html (Accessed Date: May. 30, 2023).
  28. Sentaurus Device Electromagnetic Wave Solver User Guide, Version L-2016.03, (Synopsys Inc., CA, USA), pp. 1-68.
  29. A. Prajapati, Y. Nissan, T. Gabay, and G. Shalev, “Broadband absorption of the solar radiation with surface arrays of subwavelength light funnels,” Nano Energy 54, 447-452 (2018).
    CrossRef
  30. T. Jung, M. Fujita, J. Cho, K. Lee, D. Seol, S. An, C. Lee, Y. Jeong, M. Jung, S. Park, S. Baek, S. Jung, S. Lee, J. Yun, E. S. Shim, H. Han, E. Park, H. Sul, S. Kang, K. Lee, J. C. Ahn, and D. Chang, “A 1/1.57-inch 50Mpixel CMOS image sensor with 1.0 μm all-directional dual pixel by 0.5 μm-pitch full-depth deep-trench isolation technology,” in Proc. International Solid-State Circuits Conference-ISSCC (San Francisco, CA, USA, Feb. 20-26, 2022), pp. 102-104.
    CrossRef
  31. Y. Li, “Dependence of the focal shift on Fresnel number and f number,” J. Opt. Soc. Am 72, 700-774 (1982).
    CrossRef
  32. Y. Huo, C. C. Fesenmaier, and P. B. Catrysse, “Microlens performance limits in sub-2μm pixel CMOS image sensors,” Opt. Express 18, 5861-5872 (2010).
    Pubmed CrossRef
  33. H. Bak, H. Lee, W. J. Kim, I. Choi, H. Kim, D. Kim, H. Lee, S. Han, K.I. Lee, Y. Do, M. Cho, M. S. Baek, K. Kim, W. Park, S. H. Kang, S. J. Hong, S. J. Hong, and C. Song, “Advanced color filter isolation technology for sub-micron pixel of CMOS image sensor,” in Proc. International Electron Devices Meeting-IEDM (San Francisco, CA, USA, Dec. 3-7, 2022), pp. 37.6.1-37.6.4.
  34. H. Kim, J. Park, I. Joe, D. Kwon, J. H. Kim, D. Cho, T. Lee, C. Lee, H. Park, S. Hong, C. Chang, J. Kim, H. Lim, Y. Oh, Y. Kim, S. Nah, S. Jung, J. Lee, J. Ahn, H. Hong, K. Lee, and H.-K. Kang, “5.6 A 1/2.65in 44Mpixel CMOS image sensor with 0.7 µm pixels fabricated in advanced full-depth deep-trench isolation technology,” in Proc. International Solid-State Circuits Conference-ISSCC (San Francisco, CA, USA, Feb. 16-20, 2020), pp. 104-106.
    CrossRef
  35. H. Taguchi, “Technology of color filter materials for image sensor,” in Proc. International Image Sensor Workshop (Jun. 8-11, 2011, Hokkaido, Japan), pp. 34-37.
  36. R. Fontaine, “The evolution of pixel structures for consumer-grade image sensors,” IEEE Trans. Semiconduc. Manuf. 26, 11-16 (2013).
    CrossRef
  37. G. Agranov, R. Mauritzson, J. Ladd, A. Dokoutchaev, X. Fan, X. Li, Z. Yin, R. Johnson, V. Lenchenkov, S. Nagaraja, W. Gazeley, J. Bai, H. Lee, A. D'Anna, and G. De-Amicis, “Pixel continues to shrink, pixel development for novel CMOS image sensors,” in Proc. International Image Sensor Workshop (Bergen, Norway, Jun. 22-28, 2009), pp. 69-72.
  38. S. Choi, K. Lee, J. Yun, S. Choi, S. Lee, J. Park, E. S. Shim, J. Pyo, B. Kim, M. Jung, Y. Lee, K. Son, S. Jung, T. Wang, Y. Choi, D. Min, J. Im, C. Moon, D. Lee, and D. Chang, “An all pixel PDAF CMOS image sensor with 0.64 μm × 1.28 μm photodiode separated by self-aligned in-pixel deep trench isolation for high AF performance,” in Proc. IEEE Symposium on VLSI Technology (Kyoto, Japan, Jun. 5-8, 2017), pp. T104-T105.
    CrossRef
  39. V. Blahnik and O. Schindelbeck, “Smartphone imaging technology and its applications,” Adv. Opt. Technol. 10, 145-232 (2021).
    CrossRef
  40. K. Uehira, “Microlens and method of manufacturing microlens,” U. S. Patent US20160181309A1 (2016).

Article

Research Paper

Curr. Opt. Photon. 2023; 7(5): 485-495

Published online October 25, 2023 https://doi.org/10.3807/COPP.2023.7.5.485

Copyright © Optical Society of Korea.

Covered Microlens Structure for Quad Color Filter Array of CMOS Image Sensor

Jae-Hyeok Hwang1, Yunkyung Kim1,2

1Department of ICT Integrated Safe Ocean Smart Cities Engineering, Dong-A University, Busan 49315, Korea
2Department of Electronics Engineering, Dong-A University, Busan 49315, Korea

Correspondence to:*yunkkim@dau.ac.kr, ORCID 0000-0002-4338-7642

Received: April 3, 2023; Revised: June 28, 2023; Accepted: September 8, 2023

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

The pixel size in high-resolution complementary metal-oxide-semiconductor (CMOS) image sensors continues to shrink due to chip size limitations. However, the pixel pitch’s miniaturization causes deterioration of optical performance. As one solution, a quad color filter (CF) array with pixel binning has been developed to enhance sensitivity. For high sensitivity, the microlens structure also needs to be optimized as the CF arrays change. In this paper, the covered microlens, which consist of four microlenses covered by one large microlens, are proposed for the quad CF array in the backside illumination pixel structure. To evaluate the optical performance, the suggested microlens structure was simulated from 0.5 μm to 1.0 μm pixels at the center and edge of the sensors. Moreover, all pixel structures were compared with and without in-pixel deep trench isolation (DTI), which works to distribute incident light uniformly into each photodiode. The suggested structure was evaluated with an optical simulation using the finite-difference time-domain method for numerical analysis of the optical characteristics. Compared to the conventional microlens, the suggested microlens show 29.1% and 33.9% maximum enhancement of sensitivity at the center and edge of the sensor, respectively. Therefore, the covered microlens demonstrated the highly sensitive image sensor with a quad CF array.

Keywords: CMOS image sensor, Covered microlens, FDTD simulation, Microlens, Quad color filter array

I. INTRODUCTION

In the mobile camera market, the demand for high-resolution complementary metal-oxide-semiconductor (CMOS) image sensors has increased [1]. However, pixel pitch has been reduced continuously because of the limited image sensor size [2]. Although recent trends have led to a tendency to reduce pixel pitch, it is hard to reduce the thickness of silicon, color filters (CFs), and microlenses. Silicon has different absorption coefficients depending on the wavelength [3]. Thus, the silicon should be thick enough to absorb light of a long wavelength and compensate for the degradation of sensitivity caused by pixel shrinking. Therefore, the silicon thickness of the photodiode tends to be in the range of 2.5 µm to 3.7 µm [2]. To prevent the transmission of unwanted wavelengths of light into the silicon, CFs between 0.5 µm to 0.8 µm thick are used [4]. A microlens with an appropriate height and radius of curvature (ROC) contributes to improved pixel sensitivity. As the ROC increases, the focal point is lengthened due to flattened microlens. Therefore, optimized microlens height and ROC are significantly related to the optical performance. The pixel pitch has been continuously decreasing, while reducing pixel thickness degrades sensitivity. As a result of pixel shrinking and increased aspect ratio, optical performance, such as sensitivity and crosstalk, has declined. Optical performance is an important factor to implement for the high sensitivity of CMOS image sensors.

Many research efforts have been devoted to increasing sensitivity and decreasing crosstalk. To suppress optical and electrical crosstalk, deep trench isolation (DTI) was proposed [5]. The DTI which is a thin dielectric is placed between neighboring pixels and acts as a total reflection barrier for the complete physical isolation of pixels. However, the dark current occurs because of stress caused by the DTI process. Thus, the metal or doped poly-Si is filled into the DTI as a metal-insulator-silicon capacitor. Crosstalk and dark currents are suppressed by holes accumulated on the DTI surface [6, 7]. Furthermore, a negative DC bias is applied to the poly-Si between the DTI to accumulate more holes [8]. Moreover, to suppress crosstalk, a metal grid is inserted into the passivation layer under the CF [9]. A metal grid is also inserted between each CF to resolve the diffraction limit issue in sub-micron pixels. Moreover, the dielectric grid and air gap grid have been proposed to decrease optical loss from the metal [8, 10]. The DTI and grid act as a total reflection planes. Therefore, the effective depth of silicon is increased so that the sensitivity of long wavelengths is enhanced. However, the DTI and grid reduce the light-receiving area. For the small pixel under the sub-micron, it is hard to reduce the width of DTI for keeping the light-receiving area. Moreover, the DTI must be inserted deep enough to suppress crosstalk. Therefore, pixel shrinking requires a more precise process due to the degraded aspect ratio of the DTI.

To overcome the degraded aspect ratio of pixel structures, advanced nanotechnologies, such as plasmonic CFs and metalenses, have been studied [11, 12]. The plasmonic metal filters, composed of periodic hole arrays, act as optical filters due to the interference of surface plasmon polaritons. By changing the hole size, shape, and separation, a single thin metal layer can control the transmission spectra of the hole array. However, the transmission of plasmonic CFs is relatively low compared to conventional dye-doped filters. The metalens consists of a combination of dielectric nanopillars. The optimized nanopillars are capable of guiding individual primary colors into each pixel. Thus, the metalens gains a large collection area of white light as an integration of the microlens and CF. However, the dispersion of the nanopillars can contribute to crosstalk. Instead of silicon, 2D materials, such as graphene, have been proposed to reduce pixel thickness with increased flexibility [13, 14]. However, the performance and process compatibility for the demonstration of nanoscale structures is still far from commercialization in the mobile camera market.

Various CF patterns and demosaicking have been suggested as alternative methods for sensitivity enhancement [1518]. Moreover, the quad CF array was introduced with pixel binning [1921]. The quad CF array has four photodiodes under the adjacent CFs of the same color. Pixel binning combines spatial resolution to 1/4 of the basic size [22]. In low illuminance conditions, four pixels are combined to process more light. In high illuminance conditions, the pixel binning algorithm improves resolution by processing individual pixels separately. Thus, the quad CF array and pixel binning have the benefit of a high dynamic range regardless of light intensity conditions. However, studies on the optical structure are still needed to increase sensitivity to the conditions of the quad CF.

In this paper, a covered microlens is proposed. This proposed microlens is applied to the quad CF array for the high sensitivity of small pixels in the backside illumination (BSI) structure [5, 9]. In Section II, the concept of the covered microlens pixel is described, and Section III includes the simulation results and a discussion of the sensitivity of the condition with and without in-pixel DTI and the sensitivity of the center and edge of the sensor. Section Ⅳ presents the conclusions of the proposed covered microlens.

II. STRUCTURE OF THE COVERED MICROLENS

In this paper, the previous structures and suggested structures were compared to confirm the improved sensitivity. Figure 1 shows the three types of microlenses, including the suggested covered microlens in the BSI pixel structure. The conventional microlens and 2×2 on-chip lens (2×2OCL) are two types of widely used microlenses with quad CF arrays [23]. In Fig. 1(a), the conventional microlens has the advantage of focusing on the incident light just below the arranged photodiode. However, it is not enough to concentrate the incident light due to the optical loss from the diffraction in the sub-micron pixels. Therefore, a higher and wider microlens is needed for high sensitivity. As shown in Fig. 1(b), the 2×2OCL has enhanced sensitivity because of its wide and high microlens [23]. However, the focal point of the 2×2OCL is between pixels, not at the center of each photodiode as in the beam profile of Fig. 1(b). For the pixel having the quad color filter array, the pixels have four photodiodes under the stacked microlens and color filter. The excited electrons by photons are collected at the photodiode. Therefore, the focal point of 2×2OCL has a limit to collect excited electrons in the four photodiodes of a pixel having the quad CF array. Figure 1(c) shows the suggested pixel structure with the covered microlens. The main feature of the suggested structure is the large lens that covers the four conventional microlenses. The covered microlens enhances the concentration efficiency of the incident light. The base microlens distributes the incident light further into the middle of the four photodiodes as in the beam profile of Fig. 1(c). Thus, the sensitivity is increased due to the two lenses on top of the CF.

Figure 1. The concept drawings for 3D, top, 2D cross-section of a diagonal cut, and beam profile views of pixel structure with the (a) conventional microlens, (b) 2×2 on-chip lens (2×2OCL), and (c) covered microlens.

The key factor of the covered microlens is the difference in the refractive index. To increase the incident light into the photodiode, the reflectance should be minimized [24]. Therefore, the difference in refractive index between mediums should be reduced to minimize the loss of incident light, and the covered microlens pixel should also have a sequentially increasing refractive index from the air to the optical stacks. Moreover, the covered microlens should have a lower refractive index than the base microlens.

Figure 2 shows that the simulation results of reflectance depend on the refractive index of the covered microlens and the base microlens. In the case of typical microlens for CMOS image sensors, the refractive index of materials having around 1.6 has been widely used [25]. Moreover, materials having a high refractive index have been required to concentrate the incident light to the photodiode [26]. To increase sensitivity effectively, materials having the highest refractive index of around 2.1 have been developed at visual light wavelength [26]. In this paper, refractive indices of the covered microlens, ranging from 1.6 to 2.1, were tested. Figure 2(a) shows the covered microlens with a lower refractive index than the base microlens. The reflectance remained almost constant regardless of the refractive index of the base microlens. On the other hand, Fig. 2(b) shows the covered microlens with a higher refractive index than the base microlens. The reflectance increased as the refractive index of the covered microlens increased. The structure in Fig. 2(b) has a relatively large refractive index difference between the air and covered microlens. Therefore, the sensitivity is degraded due to reflection between the air, covered microlens, and base microlens. In the case of the covered microlens having a refractive index of 1.9, the reflectance was suddenly raised due to thin-film interference from two microlenses. However, thin-film interference did not affect for the overall trend. Thus, reflectance shows a tendency to increase as the refractive index of stacked materials increases. Moreover, the oblique incident light enters more obliquely due to refraction, and this issue causes crosstalk.

Figure 2. The simulated reflectance of (a) the covered microlens with a lower refractive index than the base microlens and (b) the covered microlens with a higher refractive index than the base microlens.

The reflection between mediums is one of the important factors causing optical loss. Reflectance increases as the difference in refractive index between mediums increases. To minimize optical loss, the optical stacks were located by gradually increasing refractive indices from air to the base microlens. Therefore, in this paper, the refractive index of the covered microlens was fixed at the lowest value of 1.6, and the base microlens refractive indices varied between 1.7 and 2.1.

III. EXPERIMENTAL RESULTS AND DISCUSSION

The optical performance of the covered microlens pixels in the sub-micron was simulated using a TCAD Sentaurus for considering diffraction and refraction [27]. The finite-difference time-domain (FDTD) method is widely used to solve Maxwell’s equation for numerical analyses of optoelectronic devices. In the TCAD simulation, the absorbed photon density is calculated by solving the temporal evolution of electromagnetic waves in the pixel structure [28]. Moreover, the beam profile visually shows the absorbed photon density of the pixel as in Fig. 1. To investigate sensitivity, the wavelengths of incident light used were 650, 540, and 450 nm. RGB Bayer was used as a quad CF array. The sensitivity which is one of the most important performance measures is defined as the ratio of the absorbed light to the incident light [3]. On the condition of the same intensity of the incident light, the sensitivity is proportional to the absorbed photon density, which means the absorbed light. Therefore, the high absorbed photon density indicates high sensitivity because the electrons are generated by absorbed light in silicon. In this paper, the sensitivity is the sum of absorbed photon density by 16 pixels under the red, green, and blue CFs. The absorbed photon density (Aopt) is calculated by the absorbed power density divided by the photon energy as in Eq. (1) [28, 29].

Aopt=·Savhv=12hvσE2.

Sav, σ, and are noted with time-averaged Poynting vector, non-zero conductivity, and photon energy, respectively. Ε is noted with the electric field by the incident photon’s energy. As parameters, non-zero conductivity includes a complex permittivity and a complex refractive index. The absorbed photon density is calculated in the silicon area of the pixel.

The pixels less than 1.0 μm are critically affected by pixel shrinking. For high resolution image sensors, the 0.5 μm pixel was reported as the smallest pixel [30]. Therefore, the conventional microlens, 2×2OCL [23] and suggested covered microlens structure were simulated with 1.0 µm to 0.5 µm pixel pitches. For optimization of the microlens, ROC and height varied in the steps of 0.1 μm, respectively. Moreover, the refractive index of the base microlens located under the covered microlens was determined between 1.7 and 2.1. The parameters of optimized microlenses and DTI showing the most enhanced sensitivity are shown in Table 1. From the result, the optimized microlenses show irregular patterns in parameters as the pixel pitch changes due to strong diffraction causing limited spot size [3133]. Moreover, the pixel has a smaller width of DTI as the pixel size gets smaller [34]. For the simulated pixels in this paper, the heights of the silicon, CF, and passivation layer were 3.0 μm, 0.6 μm, and 0.2 μm which is generally used in CMOS image sensors [2, 4]. The red, green, and blue CF have a refractive index of 1.9 at the wavelength of 650 nm, 1.6 at the wavelength of 540 nm, and 1.6 at the wavelength of 450 nm, respectively [35]. Also, the metal grid and passivation layer consisted of tungsten and HfO2 which have been conventionally used [36].

TABLE 1. Parameters of optimized microlens and deep trench isolation (DTI) for each pixel pitch.

Pixel Pitch (µm)1.00.90.80.70.60.5
Conventional MLheight [µm]0.60.40.40.30.20.1
ROC [µm]0.60.90.50.40.40.9
Refractive Index (@650 nm)1.6
2×2OCLHeight [µm]1.41.30.70.51.00.7
ROC [µm]1.41.61.21.20.90.7
Refractive Index (@650 nm)1.6
Covered MLBase MLHeight [µm]0.40.80.40.30.30.1
ROC [µm]0.40.70.40.40.30.6
Refractive Index (@650 nm)1.71.81.71.91.81.9
Covered MLHeight [µm]1.41.30.70.51.00.7
ROC [µm]1.41.61.21.20.90.7
Refractive Index (@650 nm)1.6
DTIWidth [nm]1008065453525
Refractive Index (@650 nm)1.9


To confirm the optical performance of the suggested covered microlens, we compared with the 2×2OCL and conventional microlens. Moreover, the suggested structure was simulated according to the location of the pixels having different chief ray angles (CRAs) in the image sensor [37], and also with and without in-pixel DTI. The in-pixel DTI has a function to separate four pixels under the same color of quad CF array [23, 38]. Depending on the pixel location in the image sensor, the main ray of incident light that enters each pixel from the camera module has different angles. The CRA represents the incident angle of the main ray. In this paper, we consider that the CRA of the edge pixel is 35° and the CRA of the center pixel is 0° [39]. At 0°, the incident light enters the middle of the pixels. On the other hand, at 35°, the incident light enters obliquely from the lens of the camera module to the microlens of the image sensor. Therefore, the alignment of the optical stacks was significant as the image sensor size increased. To guide the oblique incident light into the photodiode, the pixels at the CRA 35° were simulated by shifting the microlens, and CF regardless of in-pixel DTI.

The in-pixel DTI indicates the DTI for separating the four photodiodes under the same CF. In the case of the CRA 35°, the sensitivity disparity of adjacent pixels under the same CF array occurred due to the absorption coefficient of silicon. The in-pixel DTI between the four photodiodes distributes the incident light equally into each pixel. To resolve the sensitivity disparity, the in-pixel DTI has been used in the high-performance image sensor. Therefore, in this paper, with and without in-pixel DTIs were investigated with CRA 0° and CRA 35°.

3.1. Sensitivity at a CRA 0°

In this section, we investigated the sensitivity of the suggested structure at CRA 0°. At first, the conventional microlens, 2×2OCL, and covered microlens were compared without in-pixel DTI. Figure 3(a) shows the simulated pixel with the covered microlens without in-pixel DTI at a CRA 0°. Figure 3(b) shows the normalized sensitivity to the incident light at 0° from 1.0 µm to 0.5 µm pixel pitches. To compare the optical performance of the suggested structure, the sensitivity of the 1.0 µm pixel with the covered microlens and the in-pixel DTI, which shows the highest sensitivity, was set as 1. From the simulation results, the covered microlens pixels were more sensitive than the pixels of the 2×2OCL and conventional microlens, regardless of pixel pitch. Moreover, the difference in sensitivity between the covered microlens and the conventional microlens increased as the pixel pitch decreased. Figure 3(c) shows the sum of the sensitivities when the incident light enters the pixels at 0°, 10°, and 20°. Compared to the conventional microlens and 2×2OCL, the covered microlens pixel indicated higher sensitivity, regardless of pixel pitch and incident angle. Compared to the conventional microlens pixel and the covered microlens pixel, the pixel having the covered microlens with a 0.5 μm pixel pitch showed the highest enhancement of sensitivity, and the sensitivity increased by 28.7% and 22.0% at the incident angle of 0° and the sum of 0°, 10°, and 20°, respectively.

Figure 3. Simulation results without in-pixel DTI at a CRA 0° of (a) the pixel structure, (b) the sensitivity of the incident angle at 0°, and (c) the sum of sensitivities of the incident angles at 0°, 10°, and 20°. The results of the suggested covered ML are compared with 2×2OCL, and conventional ML from 1.0 µm to 0.5 µm pixel pitches. DTI, deep trench isolation; CRA, chief ray angle; 2×2OCL, 2×2 on-chip lens; ML, microlens.

Next, the conventional microlens, 2×2OCL, and covered microlens were compared with in-pixel DTI. Figure 4(a) shows the suggested pixel with in-pixel DTI at CRA 0°. Compared to Fig. 3(a), the pixel has an in-pixel DTI that is located in the middle of each pixel. Figures 4(b) and 4(c) show the sensitivity of the incident angle at 0° and the sum of 0°, 10°, and 20°. Figures 4(b) and 4(c) have a similar tendency to Figs. 3(b) and 3(c). Regardless of pixel pitch, the covered microlens pixels had the highest sensitivity compared to the conventional microlens and 2×2OCL. Compared to the conventional microlens pixels in the simulations, the sensitivity of suggested pixels with 0.5 µm pixel pitch increased the maximum by 29.1% and 22.2% at the incident angle of 0° and the sum of 0°, 10°, and 20°, respectively. Simulation results show a high sensitivity improvement rate in small pixels. As pixel sizes get smaller, the sensitivity is limited due to the diffraction limit. Therefore, the sensitivity is most improved with 0.5 μm pixels having the covered microlens of large width.

Figure 4. Simulation results with in-pixel DTI at a CRA 0° of (a) the pixel structure, (b) the sensitivity of the incident angle at 0°, and (c) the sum of sensitivities of the incident angles at 0°, 10°, and 20°. The results of the suggested covered ML are compared with 2×2OCL, and conventional ML from 1.0 µm to 0.5 µm pixel pitches. DTI, deep trench isolation; CRA, chief ray angle; 2×2OCL, 2×2 On-chip lens; ML, microlens.

3.2. Sensitivity at a CRA 35°

In this section, the sensitivity of the suggested structure was investigated at CRA 35°. Before comparing with the conventional and the suggested structures, the shift distance of microlens and a CF at a CRA 35° is optimized for the incident light to enter the photodiode without loss. The shifted microlens and CF were simulated as moving in the unit of 0.01 μm to determine the exact shift distance. Thus, the incident light enters the photodiode as a minimized degradation of optical performance when the incident light is at 35°. Figure 5 shows the simulation results of the shift distance and the relationship between the microlens height and the shift distance. Figure 5(a) shows the simulation results of the conventional microlens. As the pixel pitch decreased, the microlens height and the shift distance decreased. Figures 5(b) and 5(c) show the shift distance of the 2×2OCL and covered microlens. The simulated results show that the shift distance is determined by the microlens height. For instance, in the case of the 2×2OCL and covered microlens, the optimized 0.6 µm pixel has a microlens height of 1.0 µm, and a 0.7 µm pixel has a microlens height of 0.5 µm. Compared to the 0.7 µm pixels, the microlens of the 0.6 µm pixel with a high microlens is shifted more. Therefore, we confirmed that the shift distance of the covered microlens depends on the microlens height at a CRA 35°.

Figure 5. The microlens and CF shift distance according to the microlens height of the (a) conventional microlens, (b) 2×2OCL and (c) covered microlens at a CRA 35°. CF, color filter; 2×2OCL, 2×2 on-chip lens; CRA, chief ray angle.

Figure 6(a) shows the pixel with the covered microlens without in-pixel DTI. The sensitivity was also set to normalize based on 1.0 µm pixel having the covered microlens and in-pixel DTI at a CRA 0° in Fig. 4. Figure 6(b) shows the simulation results of the sensitivity at the 35° incident light. Compared to the conventional microlens pixel, the 0.5 µm pixel having the covered microlens shows a maximum improvement of 33.9%. Figure 6(c) shows the sum of the sensitivity graph at 35°, 45°, and 55° incident angles. The enhancement of suggested pixels with 0.7 µm pixel pitch increased to 18.6%.

Figure 6. Simulation results without in-pixel DTI at a CRA 35° of (a) the pixel structure, (b) the sensitivity of the incident angle at 35°, and (c) the sum of sensitivities of incident angles 35°, 45°, and 55° at a CRA 35°. The results of the suggested covered ML are compared with 2×2OCL, and conventional ML from 1.0 µm to 0.5 µm pixel pitches. DTI, deep trench isolation; CRA, chief ray angle; 2×2OCL, 2×2 On-chip lens; ML, microlens.

Figure 7(a) shows the pixel with the covered microlens with in-pixel DTI. Figure 7(b) shows the graph of sensitivity at a 35° incident angle. Compared to the conventional microlens pixel and suggested pixel, 0.5 µm pixel having the covered microlens shows a maximum enhancement of 33.5%. Figure 7(c) shows the sum of sensitivities at 35°, 45°, and 55° incident angles; The maximum improvement of the 0.7 µm pixel having the covered microlens was 20.5%.

Figure 7. Simulation results with in-pixel DTI of (a) the pixel structure, (b) the sensitivity of the incident angle at 35°, and (c) the sum of sensitivities of incident angles 35°, 45°, and 55° at a CRA 35° from 1.0 µm to 0.5 µm pixel pitches. DTI, deep trench isolation; CRA, chief ray angle.

Compared with the conventional microlens and 2×2OCL, the sensitivity of pixels with the covered microlens was improved at a CRA 35°. Therefore, the covered microlens demonstrated enhanced sensitivity at the center and edge of the sensor both.

3.3. Characteristics of the Covered Microlens

From the simulation results from Figs. 37, the trends of sensitivity with pixel pitch is investigated. For CRA 0° and CRA 35°, the sensitivities of pixels having the covered microlens show a similar characteristic. Figure 8 shows the sensitivity enhancement rate associated with pixel pitch. Figure 8(a) shows the incident angles of 0° and 35° at CRA 0° and CRA 35°, respectively. The sensitivity enhancement rate of the pixels with the covered microlens increased as pixel pitch decreased compared to the conventional microlens. The 0.5 µm pixels achieved a sensitivity improvement of over 28.7%, regardless of the CRA and in-pixel DTI. The pixels with the small microlens were strongly affected by the diffraction limit. Therefore, the incident light can be effectively focused into the pixels without optical loss due to the large covered microlens. The 0.5 µm pixel pitch also showed the highest enhancement of sensitivity. Figure 8(b) shows the sums of the sensitivity enhancement rate of incident angles 0°, 10°, 20°, and 35°, 45°, 55° at CRA 0° and CRA 35°, respectively. In the case of the sum of sensitivities with oblique incident angles, the enhancement rate increased according to the decreased pixel pitch at a CRA 0°. Among them, the sensitivity of the 0.5 µm pixel increased to 22.2%. In the case of the CRA 35°, the 0.7 µm pixel indicated the highest improvement by 20.5%.

Figure 8. Simulation results between the covered microlens and the conventional microlens for sensitivity improvement of 1.0 µm to 0.5 µm pixel pitch at chief ray angle (CRA) 0° and CRA 35°: (a) Incident angles 0° and 35°, (b) sums of incident angles 0°, 10°, 20° and 35°, 45°, 55°.

3.4. Discussion

In this paper, the covered microlens pixels showing improved sensitivity for the quad CF array were implemented in the optical simulation. As result, improved sensitivity was shown in the small pixels. As a tendency of the image sensor, the pixels with the quad CF array have emerged as elements of advanced pixels. Moreover, the pixel pitch has been reduced, and the image sensor size has increased. On the other hand, degraded sensitivity is inevitable due to pixel shrinking; Overcoming this issue is necessary for future studies. Considering this trend, the covered microlens for a quad CF array is an appropriate structure for small pixels and highly sensitive CMOS image sensors.

When considering the fabrication to use the process of the thermal reflow which is one of the representative methods for microlens processing, the deformation of the base microlens must be minimized during the process of the covered microlens. In other words, meticulous thermal profile control is required for the fabrication of the covered microlens. To make the desired shapes, the etch-back which has the benefit of providing a wide selection of microlens materials is also required after reflow processing [40]. Therefore, we expect that the proposed structure is possible to be manufactured with the meticulous processing of the thermal reflow and etch-back.

The suggested structures were investigated for sensitivity which is the amount of absorbed light in the silicon area of a pixel. The covered microlens and the 2×2OCL show almost identical sensitivity. However, the covered microlens shows better sensitivity than the 2×2OCL for all simulated structures regardless of pixel pitch. As one of the major advantages of the covered microlens, guided incident light to the photodiodes minimizes the loss of the electron by recombination. Therefore, the rate of increase in quantum efficiency is expected to be conspicuously improved more than the rate of increase in sensitivity. In the future work, quantum efficiency simulation is required to apply the suggested structure for fabrication. Moreover, meticulous fabrication is needed for aligning the base microlens and the covered microlens.

IV. CONCLUSION

We suggested the covered microlens pixel on the quad CF arrays. The microlens height, ROC, and refractive index of the covered microlens were optimized according to the pixel pitch from 1.0 µm to 0.5 µm. Furthermore, the optimized microlens pixels were simulated with and without in-pixel DTI at CRA 0° and CRA 35° of the image sensor. The suggested structure was compared to the 2×2OCL and conventional microlens to evaluate its optical performance. As a result, the sensitivity of the covered microlens pixel increased compared to the 2×2OCL and conventional microlens, regardless of the pixel pitch and CRA. Compared to the conventional microlens at a CRA 35°, the 0.5 µm pixel with the covered microlens showed the highest sensitivity enhancement, reaching up to 33.9%. Therefore, the covered microlens pixel can contribute as one of the prototypes for pixels having the quad CF array in CMOS image sensors.

ACKNOWLEDGMENT

The EDA tool was supported by the IC Design Education Center (IDEC), Korea.

FUNDING

This work was supported by Dong-A University Foundation Grant in 2022.

DISCLOSURES

The authors declare no conflicts of interest.

DATA AVAILABILITY

Data underlying the results presented in this paper are not publicly available at the time of publication, but may be obtained from the authors upon reasonable request.

Fig 1.

Figure 1.The concept drawings for 3D, top, 2D cross-section of a diagonal cut, and beam profile views of pixel structure with the (a) conventional microlens, (b) 2×2 on-chip lens (2×2OCL), and (c) covered microlens.
Current Optics and Photonics 2023; 7: 485-495https://doi.org/10.3807/COPP.2023.7.5.485

Fig 2.

Figure 2.The simulated reflectance of (a) the covered microlens with a lower refractive index than the base microlens and (b) the covered microlens with a higher refractive index than the base microlens.
Current Optics and Photonics 2023; 7: 485-495https://doi.org/10.3807/COPP.2023.7.5.485

Fig 3.

Figure 3.Simulation results without in-pixel DTI at a CRA 0° of (a) the pixel structure, (b) the sensitivity of the incident angle at 0°, and (c) the sum of sensitivities of the incident angles at 0°, 10°, and 20°. The results of the suggested covered ML are compared with 2×2OCL, and conventional ML from 1.0 µm to 0.5 µm pixel pitches. DTI, deep trench isolation; CRA, chief ray angle; 2×2OCL, 2×2 on-chip lens; ML, microlens.
Current Optics and Photonics 2023; 7: 485-495https://doi.org/10.3807/COPP.2023.7.5.485

Fig 4.

Figure 4.Simulation results with in-pixel DTI at a CRA 0° of (a) the pixel structure, (b) the sensitivity of the incident angle at 0°, and (c) the sum of sensitivities of the incident angles at 0°, 10°, and 20°. The results of the suggested covered ML are compared with 2×2OCL, and conventional ML from 1.0 µm to 0.5 µm pixel pitches. DTI, deep trench isolation; CRA, chief ray angle; 2×2OCL, 2×2 On-chip lens; ML, microlens.
Current Optics and Photonics 2023; 7: 485-495https://doi.org/10.3807/COPP.2023.7.5.485

Fig 5.

Figure 5.The microlens and CF shift distance according to the microlens height of the (a) conventional microlens, (b) 2×2OCL and (c) covered microlens at a CRA 35°. CF, color filter; 2×2OCL, 2×2 on-chip lens; CRA, chief ray angle.
Current Optics and Photonics 2023; 7: 485-495https://doi.org/10.3807/COPP.2023.7.5.485

Fig 6.

Figure 6.Simulation results without in-pixel DTI at a CRA 35° of (a) the pixel structure, (b) the sensitivity of the incident angle at 35°, and (c) the sum of sensitivities of incident angles 35°, 45°, and 55° at a CRA 35°. The results of the suggested covered ML are compared with 2×2OCL, and conventional ML from 1.0 µm to 0.5 µm pixel pitches. DTI, deep trench isolation; CRA, chief ray angle; 2×2OCL, 2×2 On-chip lens; ML, microlens.
Current Optics and Photonics 2023; 7: 485-495https://doi.org/10.3807/COPP.2023.7.5.485

Fig 7.

Figure 7.Simulation results with in-pixel DTI of (a) the pixel structure, (b) the sensitivity of the incident angle at 35°, and (c) the sum of sensitivities of incident angles 35°, 45°, and 55° at a CRA 35° from 1.0 µm to 0.5 µm pixel pitches. DTI, deep trench isolation; CRA, chief ray angle.
Current Optics and Photonics 2023; 7: 485-495https://doi.org/10.3807/COPP.2023.7.5.485

Fig 8.

Figure 8.Simulation results between the covered microlens and the conventional microlens for sensitivity improvement of 1.0 µm to 0.5 µm pixel pitch at chief ray angle (CRA) 0° and CRA 35°: (a) Incident angles 0° and 35°, (b) sums of incident angles 0°, 10°, 20° and 35°, 45°, 55°.
Current Optics and Photonics 2023; 7: 485-495https://doi.org/10.3807/COPP.2023.7.5.485

TABLE 1 Parameters of optimized microlens and deep trench isolation (DTI) for each pixel pitch

Pixel Pitch (µm)1.00.90.80.70.60.5
Conventional MLheight [µm]0.60.40.40.30.20.1
ROC [µm]0.60.90.50.40.40.9
Refractive Index (@650 nm)1.6
2×2OCLHeight [µm]1.41.30.70.51.00.7
ROC [µm]1.41.61.21.20.90.7
Refractive Index (@650 nm)1.6
Covered MLBase MLHeight [µm]0.40.80.40.30.30.1
ROC [µm]0.40.70.40.40.30.6
Refractive Index (@650 nm)1.71.81.71.91.81.9
Covered MLHeight [µm]1.41.30.70.51.00.7
ROC [µm]1.41.61.21.20.90.7
Refractive Index (@650 nm)1.6
DTIWidth [nm]1008065453525
Refractive Index (@650 nm)1.9

References

  1. S. Yoon, J. Yun, Y. Jung, I. Jeong, J. Choi, W. Choi, J. Lee, H. Lee, and J. Ko, “World largest mobile image sensor with all directional phase detection auto focus function,” in Proc. IEEE Hot Chips 33 Symposium-HCS (Palo Alto, CA, USA, Aug. 22-24, 2021), pp. 1-22.
    CrossRef
  2. R. Fontaine, “The state-of-the-art of smartphone imagers,” in Proc. International Image Sensors Workshop (Snowbird, Utah, USA, Jun. 23-27, 2019), p. Article no. R01.
  3. J. Nakamura, Image Sensors and Signal Processing for Digital Still Cameras (CRC press, USA, 2017), pp. 53-93.
    KoreaMed CrossRef
  4. B. C. Hseih, H. Siddiqui, J. Luo, T. Georgiev, K. Atanassov, S. Goma, H. Y. Cheng, J. J. Sze, R. J. Lin, K. Y. Chou, C. Chao, and S. G. Wuu, “New color filter patterns and demosaic for sub-micron pixel arrays,” in Proc. International Image Sensor Workshop (Vaals, Netherlands, Jun. 8-11, 2015), pp. 8-11.
  5. B. J. Park, J. Jung, C. R. Moon, S. H. Hwang, Y. W. Lee, D. W. Kim, K. H. Paik, J. R. Yoo, D. H. Lee, and K. Kim, “Deep trench isolation for crosstalk suppression in active pixel sensors with 1.7 µm pixel pitch,” Jpn J. Appl. Phys. 46, 2454 (2007).
    CrossRef
  6. Y. Kitamura, H. Aikawa, K. Kakehi, T. Yousyou, K. Eda, T. Minami, S. Uya, Y. Takegawa, H. Yamashita, Y. Kohyama, and T. Asami, “Suppression of crosstalk by using backside deep trench isolation for 1.12 μm backside illuminated CMOS image sensor,” in Proc. IEEE International Electron Devices Meeting-IEDM (San Francisco, CA, USA, Dec. 10-13, 2012), pp. 24.2.1-24.2.4.
    CrossRef
  7. Y. Kim, W. Choi, D. Park, H. Jeoung, B. Kim, Y. Oh, S. Oh, B. Park, E. Kim, Y. Lee, T. Jung, Y. Kim, S. Yoon, S. Hong, J. Lee, S. Jung, C.-R. Moon, Y. Park, D. Lee, and D. Chang, “A 1/2.8-inch 24Mpixel CMOS image sensor with 0.9 μm unit pixels separated by full-depth deep-trench isolation,” in Proc. International Solid-State Circuits Conference-ISSCC (San Francisco, CA, USA, Feb. 11-15, 2018), pp. 84-86.
    CrossRef
  8. S. Park, C. Lee, S. Park, H. Park, T. Lee, D. Park, M. Heo, I. Park, H. Yeo, Y. Lee, J. Lee, B. Lee, D.-C. Lee, J. Kim, B. Kim, J. Pyo, S. Quan, S. You, I. Ro, S. Choi, S. I. Kim, I. S. Joe, J. Park, C. H. Koo, J. H. Kim, C. K. Chang, T. Kim, J. Kim, J. Lee, H. Kim, C. R. Moon, and H. S. Kim, “A 64Mpixel CMOS image sensor with 0.56 μm unit pixels separated by front deep-trench isolation,” in Proc. Solid-State Circuits Conference-ISSCC (San Francisco, CA, USA, Feb. 20-26, 2022), pp. 1-3.
    CrossRef
  9. Y. W. Cheng, T. H. Tsai, C. H. Chou, K. C. Lee, H. C. Chen, and Y. L. Hsu, “Optical performance study of BSI image sensor with stacked grid structure,” in Proc. International Electron Devices Meeting-IEDM (Washington, DC, USA, Dec. 7-9, 2015), pp. 30.4.1-30.4.4.
  10. I. S. Joe, Y. Lee, H. Y. Park, J. U. Kim, D. Kang, T. Kim, M. Kim, K. Lee, M. Heo, I. Ro, J. Kim, I. Park, S. Kwon, K. Yoon, D. Park, C. Lee, E. Jo, M. Jeon, C. Park, K. R. Byun, C. K. Chang, J. Hur, K. Yoon, T. Jeon, J. Lee, J. Park, B. Kim, J. Ahn, H. Kim, C. R. Moon, and H. S. Kim, “Development of advanced inter-color-filter grid on sub-micron-pixel CMOS image sensor for mobile cameras with high sensitivity and high resolution,” in Proc. Symposium on VLSI Technology (Kyoto, Japan, 13-19, Jun. 2021), pp. 1-2.
  11. S. Yokogawa, S. P. Burgos, and H. A. Atwater, “Plasmonic color filters for CMOS image sensor applications,” Nano Lett. 12, 4349-4354 (2012).
    Pubmed CrossRef
  12. B. H. Chen, P. C. Wu, V. C. Su, Y. C. Lai, C. H. Chu, I. C. Lee, J. W. Chen, Y. H. Chen, Y. C. Lan, C. H. Kuan, and D. P. Tsai, “GaN metalens for pixel-level full-color routing at visible light,” Nano Lett. 17, 6345-6352 (2017).
    Pubmed CrossRef
  13. S. Goossens, G. Navickaite, C. Monasterio, S. Gupta, J. J. Piqueras, R. Pérez, G. Burwell, I. Nikitskiy, T. Lasanta, T. Galán, E. Puma, A. Centeno, A. Pesquera, A. Zurutuza, G. Konstantatos, and F. Koppens, “Broadband image sensor array based on graphene-CMOS integration,” Nat. Photonics 11, 366-371 (2017).
    CrossRef
  14. C. Choi, M. K. Choi, S. Liu, M. S. Kim, O. K. Park, C. Im, J. Kim, X. Qin, G. J. Lee, K. W. Cho, M. Kim, E. Joh, J. Lee, D. Son, S. H. Kwon, N. L. Jeon, Y. M. Song, N. Lu, and D. H. Kim, “Human eye-inspired soft optoelectronic device using high-density MoS2-graphene curved image sensor array,” Nat. Commun. 8, 1664 (2017).
    Pubmed KoreaMed CrossRef
  15. B. E. Bayer, “Color imaging array,” U. S. Patent US3971065A (1976).
  16. L. Condat, “Color filter array design using random patterns with blue noise chromatic spectra,” Image Vis. Comput. 28, 1196-1202 (2010).
    CrossRef
  17. J. Wang, C. Zhang, and P. Hao, “New color filter arrays of high light sensitivity and high demosaicking performance,” in Proc. IEEE International Conference on Image Processing (Brussels, Belgium, Sep. 11-14, 2011), pp. 3153-3156.
    Pubmed CrossRef
  18. C. Kwan and B. Chou, “Further improvement of debayering performance of RGBW color filter arrays using deep learning and pansharpening techniques,” J. Imaging 5, 68 (2019).
    Pubmed KoreaMed CrossRef
  19. Y. Kim and Y. Kim, “High-sensitivity pixels with a quad-WRGB color filter and spatial deep-trench isolation,” Sensors 19, 4653 (2019).
    Pubmed KoreaMed CrossRef
  20. Z. Zhimin, B. Pain, and E. R. Fossum, “Frame-transfer CMOS active pixel sensor with pixel binning,” IEEE Trans. Electron Devices 44, 1764-1768 (1997).
    CrossRef
  21. H. Y. Huang, P. A. Conge, and L. W. Huang, “CMOS image sensor binning circuit for low-light imaging,” in Proc. IEEE Symposium on Industrial Electronics and Applications (Langkawi, Malaysia, Sep. 25-28, 2011), pp. 586-589.
    CrossRef
  22. D. Park, S. W. Lee, J. Han, D. Jang, H. Kwon, S. Cha, M. Kim, H. Lee, S. Suh, W. Joo, Y. Lee, S. Nah, H. Jeong, B. Kim, S. Jung, J. Lee, Y. Kim, C. R. Moon, and Y. Park, “A 0.8 µm smart dual conversion gain pixel for 64 Megapixels CMOS image sensor with 12k e-full-well capacitance and low dark noise,” in Proc. International Electron Devices Meeting-IEDM (San Francisco, CA, USA, Dec. 7-11, 2019), pp. 16.2.1-16.2.4.
    KoreaMed CrossRef
  23. T. Okawa, S. Ooki, H. Yamajo, M. Kawada, M. Tachi, K. Goi, T. Yamasaki, H. Iwashita, M. Nakamizo, T. Ogasahara, Y. Kitano, and K. Tatani, “A 1/2inch 48M all PDAF CMOS image sensor using 0.8 µm Quad Bayer coding 2×2OCL with 1.0 lux minimum AF illuminance level,” in Proc. International Electron Devices Meeting-IEDM (San Francisco, CA, USA, Dec. 7-11, 2019), pp. 16.3.1-16.3.4.
    CrossRef
  24. E. Hecht, Optics (Pearson Education, UK, 2017), pp. 129-133.
  25. J. K. Lee, A. Kim, D.W. Kang, and B. Y. Lee, “Efficiency enhancement in a backside illuminated 1.12 μm pixel CMOS image sensor via parabolic color filters,” Opt. Express 24, 16027-16036 (2016).
    Pubmed CrossRef
  26. J. James, A. B. Unni, K. Taleb, J. P. Chapel, N. Kalarikkal, S. Varghese, G. Vignaud, and Y. Grohens, “Surface engineering of polystyrene-cerium oxide nanocomposite thin films for refractive index enhancement,” Nano-Struct. Nano-Objects 17, 34-42 (2019).
    CrossRef
  27. Synopsys Inc., “TCAD Product Family,” (Synopsys Inc.), https://www.synopsys.com/silicon/tcad.html (Accessed Date: May. 30, 2023).
  28. Sentaurus Device Electromagnetic Wave Solver User Guide, Version L-2016.03, (Synopsys Inc., CA, USA), pp. 1-68.
  29. A. Prajapati, Y. Nissan, T. Gabay, and G. Shalev, “Broadband absorption of the solar radiation with surface arrays of subwavelength light funnels,” Nano Energy 54, 447-452 (2018).
    CrossRef
  30. T. Jung, M. Fujita, J. Cho, K. Lee, D. Seol, S. An, C. Lee, Y. Jeong, M. Jung, S. Park, S. Baek, S. Jung, S. Lee, J. Yun, E. S. Shim, H. Han, E. Park, H. Sul, S. Kang, K. Lee, J. C. Ahn, and D. Chang, “A 1/1.57-inch 50Mpixel CMOS image sensor with 1.0 μm all-directional dual pixel by 0.5 μm-pitch full-depth deep-trench isolation technology,” in Proc. International Solid-State Circuits Conference-ISSCC (San Francisco, CA, USA, Feb. 20-26, 2022), pp. 102-104.
    CrossRef
  31. Y. Li, “Dependence of the focal shift on Fresnel number and f number,” J. Opt. Soc. Am 72, 700-774 (1982).
    CrossRef
  32. Y. Huo, C. C. Fesenmaier, and P. B. Catrysse, “Microlens performance limits in sub-2μm pixel CMOS image sensors,” Opt. Express 18, 5861-5872 (2010).
    Pubmed CrossRef
  33. H. Bak, H. Lee, W. J. Kim, I. Choi, H. Kim, D. Kim, H. Lee, S. Han, K.I. Lee, Y. Do, M. Cho, M. S. Baek, K. Kim, W. Park, S. H. Kang, S. J. Hong, S. J. Hong, and C. Song, “Advanced color filter isolation technology for sub-micron pixel of CMOS image sensor,” in Proc. International Electron Devices Meeting-IEDM (San Francisco, CA, USA, Dec. 3-7, 2022), pp. 37.6.1-37.6.4.
  34. H. Kim, J. Park, I. Joe, D. Kwon, J. H. Kim, D. Cho, T. Lee, C. Lee, H. Park, S. Hong, C. Chang, J. Kim, H. Lim, Y. Oh, Y. Kim, S. Nah, S. Jung, J. Lee, J. Ahn, H. Hong, K. Lee, and H.-K. Kang, “5.6 A 1/2.65in 44Mpixel CMOS image sensor with 0.7 µm pixels fabricated in advanced full-depth deep-trench isolation technology,” in Proc. International Solid-State Circuits Conference-ISSCC (San Francisco, CA, USA, Feb. 16-20, 2020), pp. 104-106.
    CrossRef
  35. H. Taguchi, “Technology of color filter materials for image sensor,” in Proc. International Image Sensor Workshop (Jun. 8-11, 2011, Hokkaido, Japan), pp. 34-37.
  36. R. Fontaine, “The evolution of pixel structures for consumer-grade image sensors,” IEEE Trans. Semiconduc. Manuf. 26, 11-16 (2013).
    CrossRef
  37. G. Agranov, R. Mauritzson, J. Ladd, A. Dokoutchaev, X. Fan, X. Li, Z. Yin, R. Johnson, V. Lenchenkov, S. Nagaraja, W. Gazeley, J. Bai, H. Lee, A. D'Anna, and G. De-Amicis, “Pixel continues to shrink, pixel development for novel CMOS image sensors,” in Proc. International Image Sensor Workshop (Bergen, Norway, Jun. 22-28, 2009), pp. 69-72.
  38. S. Choi, K. Lee, J. Yun, S. Choi, S. Lee, J. Park, E. S. Shim, J. Pyo, B. Kim, M. Jung, Y. Lee, K. Son, S. Jung, T. Wang, Y. Choi, D. Min, J. Im, C. Moon, D. Lee, and D. Chang, “An all pixel PDAF CMOS image sensor with 0.64 μm × 1.28 μm photodiode separated by self-aligned in-pixel deep trench isolation for high AF performance,” in Proc. IEEE Symposium on VLSI Technology (Kyoto, Japan, Jun. 5-8, 2017), pp. T104-T105.
    CrossRef
  39. V. Blahnik and O. Schindelbeck, “Smartphone imaging technology and its applications,” Adv. Opt. Technol. 10, 145-232 (2021).
    CrossRef
  40. K. Uehira, “Microlens and method of manufacturing microlens,” U. S. Patent US20160181309A1 (2016).