Checkerboard type metasurface efficient spectrum router based on RGB-NIR imaging

Abstract

The invention provides a chessboard type metasurface efficient spectrum router based on RGB-NIR imaging, which comprises a chessboard type metasurface and an imaging plane, and after incident light passes through the metasurface, photons of RGB and NIR channels are guided to color pixels which are positioned on the imaging plane and respectively correspond to the four channels. According to the invention, through accurate light splitting and photon redirection, high-efficiency light splitting of four wavebands of R, G, B and NIR is realized under the pixel size of 1 [mu] m, and the efficiencies are respectively 41.6%, 42.2%, 35.9% and 39.5%. A simulation result shows that the design not only significantly improves the brightness of RGB-NIR imaging, but also shows excellent robustness.

Description

Technical Field

[0001] The present invention relates to the field of optical computing technology, and in particular to a chessboard-format metasurface high-efficiency spectral router based on RGB-NIR imaging. Background Art

[0002] In the field of modern high-tech imaging, RGB-NIR image sensors play a vital role. With the growing demand for high-resolution images, the pixel size of image sensors has gradually shrunk to the sub-wavelength level. However, the efficiency limitations of the filter array in traditional image sensors and the diffraction effect caused by the reduction in pixel size have seriously affected the imaging quality under small pixels. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings and defects in the prior art and provide a checkerboard metasurface efficient spectral router based on RGB-NIR imaging. Through precise light splitting and photon redirection, the efficient light splitting of four bands of R, G, B and NIR is achieved at a pixel size of 1μm, with efficiencies of 41.6%, 42.2%, 35.9% and 39.5% respectively. The simulation results show that the design not only significantly improves the brightness of RGB-NIR imaging (increased by 1.75 times), but also shows excellent robustness.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0005] A checkerboard-format metasurface efficient spectral router based on RGB-NIR imaging, characterized in that the router comprises a checkerboard-format metasurface and an imaging plane, and after the incident light passes through the metasurface, the photons of the four channels of RGB and NIR are guided to the color pixels corresponding to the four channels respectively on the imaging plane.

[0006] According to the above technical means, in order to ensure the light transmittance of the color routing and its compatibility with the CMOS process, the present invention selects the metasurface material so that the metasurface is consistent with the CMOS parameters.

[0007] Furthermore, the material of the router's metasurface is Si 3 N 4 The substrate material is SiO with a thickness of 2 μm. 2 , wherein the distance between the metasurface and the imaging plane is 2 μm.

[0008] Based on the above technical means, checkerboard metasurfaces can provide the required degrees of freedom: by achieving multiple scattering of incident light through different sizes and arrangements of microstructures, these metasurfaces can realize complex spectral routing functions.

[0009] Furthermore, the checkerboard-format metasurface is composed of a 16×16 grid, and the size of each grid unit is 125nm×125nm.

[0010] Furthermore, a microstructure with a height of 500 nm is placed in each of the grids, and the material of the microstructure is Si 3 N 4 .

[0011] According to the above technical means, in practical applications, there are multiple design schemes for the color pixel arrangement of the RGB-NIR image sensor. The present invention adopts a variant based on the Bayer array: replacing one G pixel in the Bayer array with a NIR pixel.

[0012] Furthermore, the color pixels located on the imaging plane include R pixels, G pixels, B pixels, and NIR pixels.

[0013] According to the above technical means, due to the small spacing between the microstructures of the checkerboard metasurface, there is a complex coupling effect between the electromagnetic fields generated after the light is scattered by the microstructure, making it impossible to use the traditional method of establishing a microstructure phase library for design. In addition, there are 2256 possible microstructure arrangements for the checkerboard metasurface, which makes the brute force search method unrealistic. Therefore, the present invention adopts a random search global optimization algorithm based on the concepts of evolution and natural selection, which can efficiently find the optimal solution in a huge design space.

[0014] Furthermore, the metasurface is optimized by the following method:

[0015] The checkerboard microstructure and air lattice are set to be represented by "1" and "0" respectively, and the metasurface structure is abstracted into a binary matrix;

[0016] The binary matrix is ​​converted into nanostructure modeling through FDTD simulation software, and the spectral transmittance curve received by each pixel is extracted to calculate the pre-set evaluation function F. The optimized metasurface is determined through iterative calculation, wherein the evaluation function F is expressed by the following formula:

[0017] F=ΣT-σ(T)

[0018] Wherein, T represents a set including the splitting efficiency of each channel, Σ represents the sum, and σ() represents the standard deviation.

[0019] Furthermore, the microstructured material Si 3 N 4 The width is 125nm and the thickness is 500nm.

[0020] Beneficial effects of the present invention:

[0021] The present invention proposes a checkerboard metasurface color routing, which can achieve efficient RGB and NIR four-channel spectrometry at a pixel size of 1μm. The arrangement of the single-layer metasurface microstructure is optimized by genetic algorithm, which significantly improves the spectroscopic performance and robustness of the metasurface while ensuring the feasibility of processing. Compared with the maximum spectroscopic efficiency of the filter array of only 25%, the spectroscopic efficiency of the metasurface color routing of the present invention on the four channels of RGB and NIR reaches 41.6%, 42.2%, 35.9% and 39.5% respectively, achieving an energy enhancement of about 60%. The simulation results show that the imaging system based on the color routing can achieve excellent imaging effects without complex color processing algorithms, and the image brightness is improved by 1.75 times compared with the traditional architecture. In addition, the present invention also takes into account the feasibility of large-scale production. The microstructure unit of the metasurface is a Si3N4 regular tetrahedron with a width of 125nm and a height of 500nm. This structure is very suitable for batch manufacturing using nanoimprinting technology. With these characteristics, the metasurface color routing technology of the present invention has shown a wide range of application prospects in the field of RGB-NIR imaging. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 Schematic diagram of the traditional RGB-NIR image sensor architecture;

[0023] Figure 2 A schematic diagram of the metasurface color routing architecture proposed by the present invention;

[0024] Figure 3 Schematic diagram of the 3D spectroscopic analysis of the metasurface proposed in the present invention;

[0025] Figure 4 It is a schematic diagram of the Si3N4 microstructure of the present invention;

[0026] Figure 5 The super surface structure of the present invention is abstracted into a binary matrix schematic diagram;

[0027] Figure 6 It is a schematic diagram of the RGB and NIR four-channel light-splitting efficiency curves displayed when evaluating the performance of the metasurface structure in the present invention;

[0028] Figure 7 This is a schematic diagram of the energy density distribution of the incident light after passing through the metasurface when evaluating the performance of the metasurface structure in the present invention. DETAILED DESCRIPTION

[0029] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention, not for limiting the scope of protection of the present invention.

[0030] In this embodiment, a checkerboard-format metasurface efficient spectral router based on RGB-NIR imaging is proposed, including a checkerboard-format metasurface and an imaging plane. After the incident light passes through the metasurface, the photons of the four channels of RGB and NIR are guided to the color pixels corresponding to the four channels respectively on the imaging plane.

[0031] like Figure 1 As shown, Figure 1 The traditional RGB-NIR image sensor architecture is demonstrated. The sensor consists of a microlens array, a color filter, and a photodetector. The function of the microlens array is to focus the incident light onto the photodetector, while the color filter is used to separate light of different wavelengths, thereby providing the photodetector with information of a specific color. However, there is a problem with this sensor architecture: while the color filter allows photons of a specific wavelength to pass through, it will reflect or absorb photons of other wavelengths, resulting in energy loss. Figure 1 As shown by the colored arrows in the figure, in the traditional architecture, the splitting efficiency of each channel is only 25% at most. Here, the splitting efficiency is defined as the ratio of the energy of light in a single band reaching the corresponding detection pixel to its total incident energy.

[0032] In this embodiment, a checkerboard-format metasurface efficient spectral router based on RGB-NIR imaging is proposed. Figure 2 As shown, unlike the traditional routing architecture, after the incident light passes through the metasurface, the photons of the four channels of RGB and NIR are efficiently guided to their corresponding pixels. The technical solution of this embodiment theoretically enables all incident light to be received by the photodetector, thereby significantly improving the light splitting efficiency. In addition, this architecture integrates the functions of the microlens array and color filter into the metasurface, thereby simplifying the overall structure of the image sensor.

[0033] In practical applications, there are many design schemes for arranging color pixels of RGB-NIR image sensors. This embodiment adopts a variation based on the Bayer array: a G pixel in the Bayer array is replaced by a NIR pixel. Figure 3A 3D schematic diagram of this modified color pixel layout and a spectral router based on a metasurface are shown. Since the arrangement of CMOS pixels is periodic, the design of the metasurface only needs to be optimized for a periodic unit containing four pixels, such as Figure 3 as shown in the illustration on the right.

[0034] In this embodiment, in order to ensure the light transmittance of the color routing and its compatibility with the CMOS process, this embodiment selects Si 3 N 4 As the material of the super surface, the substrate material is SiO with a thickness of 2 μm. 2 , which keeps the distance between the metasurface and the imaging plane at 2 μm, consistent with commercial CMOS parameters. In addition, the design target of this embodiment is 1 μm pixel size CMOS, which means that the size of one periodic unit is 2 μm.

[0035] To achieve independent control of multiple channels, the metasurface must have a high degree of design freedom. Traditional checkerboard metasurfaces can provide the required degrees of freedom: by achieving multiple scattering of incident light through different sizes and arrangements of microstructures, these metasurfaces can achieve complex spectral routing functions. In this embodiment, the material of the microstructure is Si 3 N 4 The width W is 125nm, the thickness H is 500nm, and the SiO 2 The height of the substrate H s 500nm

[0036] Therefore, this embodiment adopts a checkerboard-style metasurface as a design solution. In the embodiment, the 2μm×2μm design plane is divided into a 16×16 grid, and the size of each grid unit is 125nm×125nm. In each grid unit, you can choose whether to place a Si with a height of 500nm. 3 N 4 Microstructure, specific microstructure parameters such as Figure 4 shown.

[0037] Since the microstructure spacing of the checkerboard metasurface is small, there is a complex coupling effect between the electromagnetic fields generated after light is scattered by the microstructure, making it impossible to design it using the traditional method of establishing a microstructure phase library. In addition, the checkerboard metasurface has 2 256 There are many possibilities for microstructure arrangement, which makes the brute force search method unrealistic. Therefore, this embodiment adopts a genetic algorithm. This is a random search global optimization algorithm based on the concepts of evolution and natural selection, which can efficiently find the optimal solution in a huge design space. The specific process is as follows:

[0038] First, mathematical modeling is performed on the optimization object to facilitate subsequent calculation and optimization. In this embodiment, the checkerboard microstructure and air grid are represented as "1" and "0" respectively, thereby abstracting the supersurface structure into a binary matrix, such as Figure 5 In order to comprehensively evaluate the spectral uniformity of the metasurface in four channels and its response to the polarization state of the incident light, an evaluation function F is predefined:

[0039] F = ΣT-σ(T).

[0040] Wherein, T represents a set including the splitting efficiency of each channel, Σ represents the sum, and σ() represents the standard deviation.

[0041] In this embodiment, the splitting efficiency of each channel is ensured to be as uniform as possible. In order to consider the influence of the polarization state, the influence of the incident light in the X and Y polarization states is also considered in the T set. The specific expression is:

[0042] T={T x,i ,T y,i |i∈{R,G,B,NIR}}

[0043]

[0044] T x,NIR =f x,NIR (850)

[0045] Among them, T x,i With T y,i represents the splitting efficiency of each channel under different polarization states of incident light, f x,i represents the spectral transmittance curve received by each pixel, λ represents the wavelength, and all the numbers in the formula are in nm. Since the expressions for X and Y polarized light are similar, only the formula for X polarized light is shown here.

[0046] Finally, the optimization content is clear as follows: Use the genetic algorithm to find a 16×16 binary matrix (representing the microstructure arrangement of the metasurface) so that the evaluation function F reaches the maximum value. First, according to the principle of the genetic algorithm, the initial 100 groups of genes are generated, each gene is a 16×16 binary matrix. Then, the commercial software lumerical FDTD is used to convert the binary matrix into nanostructure modeling, and the spectral transmittance curve received by each pixel is extracted through FDTD simulation to calculate the evaluation function F; then, evaluate whether F reaches the target value; if not, the gene is updated, and the above process is repeated (except for generating the initial gene) until the stopping criterion of the genetic algorithm is reached. In this embodiment, the stopping criterion is that the number of iterations reaches 200 times.

[0047] After multiple iterations of optimization, the structure of the metasurface was finally determined. Figure 3 Right picture.

[0048] In this embodiment, in order to evaluate its performance, the spectroscopic performance of the metasurface was analyzed using a FDTD monitor under normal incidence conditions. Figure 6 The spectral efficiency curves of the four channels of RGB and NIR are shown. The curve data are the average values ​​under the conditions of x and y polarization state incidence, and the band range of each channel is marked with a rectangular color box. The results show that the spectral efficiency of the metasurface in the four channels of RGB and NIR is 41.6%, 42.2%, 35.9% and 39.5%, respectively. Compared with the efficiency limit of 25% of traditional filter sensors, the efficiency of each channel has been improved by at least 60%. In addition, the total energy utilization of the metasurface averages 96.9% in the RGB-NIR band, indicating that most of the incident light can effectively pass through the metasurface and reach the image plane.

[0049] In order to show the distribution state of the incident light after it reaches the image plane, this embodiment further simulates the energy density distribution of incident light with wavelengths of 450nm, 550nm, 650nm and 850nm after passing through the metasurface. The yellow dotted lines and letters indicate the pixel positions of the four channels, such as Figure 7 As shown. The results show that light of each wavelength is mainly concentrated in its corresponding pixel area. However, the energy distribution is not uniform and is not completely focused into a single point. This phenomenon can be attributed to the fact that the optimization process focuses on the total energy of the pixel area rather than the specific distribution pattern of the energy. Nevertheless, this uneven energy distribution does not significantly affect the pixel's detection of light energy. The photoelectric conversion efficiency of the photodetector mainly depends on the proportion of photons reaching the photosensitive area and the spatial uniformity of the quantum efficiency, and has nothing to do with the distribution pattern of light. In short, despite the uneven energy distribution, most of the energy can still effectively reach the corresponding pixel area, which shows that the metasurface can effectively achieve multi-wavelength spectrometry.

[0050] Figure 6 and Figure 7 It also reveals that there is a certain degree of crosstalk between channels, especially the NIR channel has a significant interference with the RGB channel, while the crosstalk between the RGB channels is relatively small. Despite the existence of this crosstalk, its impact can be effectively mitigated by applying color correction algorithms or integrating metasurfaces with CFA.

[0051] The above embodiments are only preferred embodiments for fully illustrating the present invention, and the protection scope of the present invention is not limited thereto. Any equivalent substitution or change made by a person skilled in the art based on the present invention is within the protection scope of the present invention.

Claims

1. A highly efficient spectral router based on a chessboard metasurface based on RGB-NIR imaging, characterized by: The router includes a checkerboard-shaped metasurface and an imaging plane. After the incident light passes through the metasurface, photons of four channels, RGB and NIR, are guided to color pixels corresponding to the four channels respectively on the imaging plane.

2. The checkerboard-format metasurface efficient spectral router based on RGB-NIR imaging according to claim 1 is characterized in that: The material of the super surface of the router is Si3N4, and the substrate material is SiO2 with a thickness of 2 μm, wherein the distance between the super surface and the imaging plane is 2 μm.

3. The checkerboard-format metasurface efficient spectral router based on RGB-NIR imaging according to claim 1 is characterized in that: The checkerboard-format metasurface consists of a 16×16 grid, and the size of each grid unit is 125nm×125nm.

4. The chessboard-format metasurface efficient spectral router based on RGB-NIR imaging according to claim 3 is characterized in that: A microstructure with a height of 500 nm is placed in each of the grids, and the material of the microstructure is Si3N4.

5. The checkerboard-format metasurface efficient spectral router based on RGB-NIR imaging according to claim 1, characterized in that: The color pixels located on the imaging plane include R pixels, G pixels, B pixels, and NIR pixels.

6. The checkerboard-format metasurface efficient spectral router based on RGB-NIR imaging according to claim 1, characterized in that: The metasurface is optimized by the following method: The checkerboard microstructure and air grid are set to be represented by "1" and "0" respectively, and the metasurface structure is abstracted into a binary matrix; The binary matrix is ​​converted into nanostructure modeling through simulation software, and the spectral transmittance curve received by each pixel is extracted to calculate the pre-set evaluation function F. The optimized metasurface is determined through iterative calculation, wherein the evaluation function F is expressed by the following formula: F=∑T-σ(T) Wherein, T represents a set including the splitting efficiency of each channel, Σ represents the sum, and σ() represents the standard deviation.

7. The checkerboard-format metasurface efficient spectral router based on RGB-NIR imaging according to claim 4 is characterized in that: The width of the microstructure material Si3N4 is 125nm and the thickness is 500nm.

Images