Color router and its metasurface structure design method and image sensor

Abstract

This application relates to a color router and its metasurface structure design method and image sensor. The method includes: establishing a color router model, which includes a substrate and a multilayer metasurface structure; each layer of the metasurface structure includes multiple dielectric microstructures arranged according to a preset lattice; determining a target wavelength band and a target color pixel corresponding to each target wavelength band; determining the variables to be optimized in the color router model, including the cross-sectional geometric parameters of the dielectric microstructures; setting a target reference surface on the emission side of the multilayer metasurface structure in combination with the target wavelength band and the target color pixel, and constructing a target near field on the target reference surface; performing forward electromagnetic simulation on the color router to obtain the simulated near field on the target reference surface; constructing a near field matching representation function based on the target near field and the simulated near field, and inversely updating the variables to be optimized based on the near field matching representation function to improve routing efficiency.

Description

Technical Field

[0001] This application relates to the field of image sensor technology, and in particular to a color router and its metasurface structure design method and image sensor. Background Technology

[0002] As imaging devices develop towards miniaturization, high pixel count, high resolution, and high integration, the pixel size of image sensors in imaging devices continues to decrease, and the number of photons that can be received per pixel decreases accordingly. This also leads to higher requirements for the light utilization efficiency, signal-to-noise ratio, dynamic range, and low-light imaging capabilities of image sensors.

[0003] If an image sensor uses an RGB (red, green, blue) color filter array for color selection, it will result in significant light energy loss. For example, for red, green, and blue light uniformly incident on the RGB Bayer color units of the RGB color filter array, the theoretical reception efficiency is limited by the area ratio of the corresponding color pixels. The theoretical reception efficiencies for the red, green, and blue bands are approximately 25%, 50%, and 25%, respectively, resulting in significant light energy loss.

[0004] Compared to RGB color filter arrays, color routers suffer less light energy loss and theoretically have higher reception efficiency. Color routers do not achieve color selection by absorbing non-target wavelengths; instead, they modulate the wavefront of incident light, guiding different wavelengths to the corresponding color pixels in the image sensor. Photons that might otherwise be absorbed by the RGB color filter array can be redistributed to the corresponding pixels by the color router, thereby improving color detection efficiency and low-light imaging performance.

[0005] One important technology for implementing color routers is the metasurface composed of subwavelength-scale artificial microstructures. This metasurface can spatially redirect light of different wavelengths using micro / nano structures, thereby controlling the phase, amplitude, polarization, and propagation direction of incident light. When pixel sizes shrink to the micrometer scale, devices need to simultaneously control the light field distribution of multiple wavelengths (red, green, and blue) within a limited lateral space. Single-layer structures often cannot provide sufficient degrees of freedom for light field control.

[0006] Improving the degree of freedom in controlling the light field of color routers has become a pressing technical challenge. Summary of the Invention

[0007] Based on this, a color router, an image sensor, and a design method thereof are provided to improve the degree of freedom of light field control in a color router.

[0008] In a first aspect, embodiments of this application provide a method for designing a metasurface structure for a color router, including:

[0009] A color router model is established, which includes a substrate and a multi-layer metasurface structure.

[0010] Each layer of metasurface structure includes multiple dielectric microstructures arranged according to a pre-defined lattice.

[0011] Determine the target band and the target color pixels corresponding to each target band;

[0012] The variables to be optimized in the color router model are determined, including the cross-sectional geometric parameters of the dielectric microstructure.

[0013] By combining the target band and the target color pixel, a target reference surface is set on the emission side of the multi-layer metasurface structure, and a target near field is constructed on the target reference surface.

[0014] A forward electromagnetic simulation of the color router is performed to obtain the simulated near field on the target reference surface;

[0015] A near-field matching representation function is constructed based on the target near-field and the simulated near-field, and the variable to be optimized is updated in reverse based on the near-field matching representation function.

[0016] In one embodiment, the variables to be optimized also include at least one of the following parameters: the gap width between adjacent dielectric microstructures, the height of the dielectric microstructure, the number of layers of the metasurface structure, and the auxiliary band.

[0017] In one embodiment, constructing the target near field on the target reference plane includes:

[0018] Focus the light of the target wavelength band to the center position of the corresponding target color pixel, and obtain the complex near field constructed by the hyperbolic phase distribution, which is used as the target near field;

[0019] The near-field matching representation function constructed based on the target near field and the simulated near field includes:

[0020] Based on the target band and multiple polarization states, at least one of the following is used as the near-field matching characterization function: crosstalk penalty term, polarization equalization term, incident angle robustness and processing constraint penalty term obtained by matching between the target near field and the simulated near field.

[0021] In one embodiment, the inverse update of the cross-sectional geometric parameters of the dielectric microstructure based on the near-field matching characterization function includes:

[0022] Electromagnetic field distribution under incident light is obtained based on forward electromagnetic simulation;

[0023] The adjoint field distribution of the inverse definition of the near-field matching characterization function is obtained based on adjoint simulation.

[0024] By combining the electromagnetic field distribution and the accompanying field distribution, the gradient of the cross-sectional geometric parameters of the medium microstructure is determined;

[0025] The cross-sectional geometric parameters of the medium microstructure are updated based on the gradient.

[0026] In one embodiment, the method further includes:

[0027] After updating the cross-sectional geometric parameters of the medium microstructure, repeat the following steps until the near-field matching characterization function satisfies the preset convergence condition or the number of repetitions reaches the preset number of iterations:

[0028] A forward electromagnetic simulation was performed on the color router to obtain the simulated near field on the target reference surface;

[0029] A near-field matching representation function is constructed based on the target near-field and the simulated near-field, and the optimization variables of the medium microstructure are updated in reverse based on the near-field matching representation function.

[0030] In one embodiment, the method further includes:

[0031] After updating the cross-sectional geometric parameters of the medium microstructure, repeat the following steps until the near-field matching characterization function satisfies the preset convergence condition or the number of repetitions reaches the preset number of iterations:

[0032] By combining the target band, auxiliary wavelength, and the target color region of the color router to be prepared, a target reference plane and a target near field matching the target reference plane are set.

[0033] A forward electromagnetic simulation was performed on the color router to obtain the simulated near field on the reference plane;

[0034] A near-field matching representation function is constructed based on the target near-field and the simulated near-field, and the optimization variables of the medium microstructure are updated in reverse based on the near-field matching representation function.

[0035] In one embodiment, setting the target reference plane and the target near field matching the target reference plane includes:

[0036] For each target band, set the focal point at the center of the corresponding target color area;

[0037] Based on the hyperbolic phase distribution corresponding to the focal point, the distance from the preset reference surface to the photosensitive element is determined;

[0038] By combining the preset refractive index of the coating medium, the distance from the preset reference surface to the photosensitive element, the coordinates of the reference point of the preset reference surface, and the coordinates of the sampling point of the photosensitive element in the color router model, the phase compensation amount at different wavelengths is determined.

[0039] Using the phase compensation amount as a phase term with a complex exponent, a unit amplitude complex field carrying the distribution of the phase term is generated as the target near field matching the target reference surface.

[0040] Secondly, embodiments of this application also provide a color router, comprising:

[0041] Substrate and multilayer metasurface structure;

[0042] Each layer of metasurface structure includes multiple dielectric microstructures;

[0043] Each layer of metasurface structure is filled with a transparent coating medium between it and between the topmost metasurface structure and the preset detection surface.

[0044] In some embodiments, the color router further includes a CMOS photosensitive element, each color unit corresponding to the CMOS photosensitive element comprising 4 color pixels, wherein the size of a single color pixel is 1.12μm × 1.12μm, and the size of a single color unit is 2.24μm × 2.24μm.

[0045] In some embodiments, the color router has a rectangular columnar medium microstructure with a minimum cross-sectional dimension of not less than 100 nm and a minimum gap of not less than 50 nm between adjacent medium microstructures; the distance from the incident top surface of the uppermost metasurface structure to the preset detection surface includes 3.9 μm.

[0046] Thirdly, embodiments of this application also provide an image sensor, including a color router, wherein the metasurface structure of the color router is designed using the method described above.

[0047] This application provides a method for designing a metasurface structure for a color router. The color router model includes a multi-layer metasurface structure. Compared to a single-layer metasurface, the multi-layer metasurface structure can comprehensively utilize intra-layer scattering, inter-layer coupling, and propagation phase accumulation to achieve multi-band color routing. Therefore, it has a higher degree of freedom in light field control. The matching degree between the target near field and the simulated near field guides the updating of cross-sectional geometric parameters and other optimization variables. It is suitable for efficiently optimizing a large number of continuous variables to be optimized in the multi-layer metasurface. Compared to the theoretical receiving efficiency of the traditional RGB Bayer color filter, the routing efficiency can be improved by at least half. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of this application or the conventional technology, the drawings used in the description of the embodiments or the conventional technology will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 This is a schematic diagram of the structure of a color router provided in an embodiment of this application;

[0050] Figure 2 This is a schematic diagram of a supercell representing a metasurface structure provided in an embodiment of this application.

[0051] Figure 3 This application provides a schematic diagram of a metasurface structure design method for a color router according to an embodiment of the present application.

[0052] Figure 4 This is an example of the energy distribution of incident light at different polarization states and a target wavelength of 620 nm on a predetermined detector surface according to an embodiment of this application.

[0053] Figure 5 This is an energy distribution diagram of incident light with a target wavelength of 520nm in different polarization states on a predetermined detector surface according to an embodiment of this application;

[0054] Figure 6 This is an example of the energy distribution of incident light at a target wavelength of 450 nm under different polarization states on a predetermined detector surface according to an embodiment of this application.

[0055] Figure 7 A schematic diagram of the RGB routing efficiency spectrum curve of the color router with a three-layer metasurface structure provided in this application embodiment;

[0056] Figure 8 A schematic diagram illustrating the routing efficiency variation of the color router with the three-layer metasurface structure provided in this application embodiment under different incident angles;

[0057] Figure 9 A schematic flowchart illustrating the method for optimizing the cross-sectional geometric parameters of a color router model in the design method provided in this application embodiment;

[0058] Figure 10 This is a schematic diagram of the structural distribution of the color router using a three-layer metasurface structure provided in the embodiments of this application;

[0059] Figure 11 This application also provides a schematic flowchart of a method for preparing a color router. Detailed Implementation

[0060] To facilitate understanding of this application, it will be described in more detail below. However, it should be understood that this application can be implemented in many different forms and is not limited to the embodiments or examples described herein. Rather, these embodiments or examples are provided to provide a more thorough and complete understanding of the disclosure of this application.

[0061] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the specification of this application is for the purpose of describing particular implementations or embodiments only and is not intended to be limiting of this application. The optional scope of the term "and / or" as used herein includes any one of two or more of the related listed items, as well as any and all combinations of the related listed items, including any two related listed items, any more related listed items, or a combination of all related listed items.

[0062] If a color router is built using a single-layer metasurface, it is difficult to achieve high routing efficiency if the requirements of multi-wavelength, broadband, polarization insensitivity and low crosstalk are met. Furthermore, if a topology-optimized structure is used for the single-layer metasurface, the processing feature size is small, the process window is narrow, and integration is difficult.

[0063] To solve the above technical problems, see Figure 1 As shown, this application embodiment provides a color router, including:

[0064] Substrate 101 and multilayer metasurface structure 102;

[0065] Each layer of metasurface structure 102 includes multiple dielectric microstructures 103;

[0066] A transparent covering medium 105 is filled between each layer of metasurface structure 102 and between the topmost metasurface structure 1021 and the preset detection surface 104.

[0067] In optional embodiments of this application, the substrate may be, but is not limited to, D263TECO glass. The transparent coating medium may be, but is not limited to, at least one of photoresist, polymer, transparent resin, silicon dioxide, or aluminum oxide.

[0068] The dielectric microstructure of the metasurface structure 102 may, but is not limited to, employ at least one of titanium dioxide, silicon nitride, gallium nitride, silicon, hafnium oxide, and tantalum pentoxide.

[0069] The shape of the dielectric microstructure can include, but is not limited to, at least one of the following: rectangular cylinder, cylindrical cylinder, elliptical cylinder, polygonal cylinder, and annular cylinder. Assuming the dielectric microstructure is rectangular cylinder, its cross-sectional geometry can include length and width.

[0070] In an optional embodiment of this application, the color router further includes a CMOS image sensor. Each color unit corresponding to the CMOS image sensor contains four color pixels; for example, each color unit contains one red pixel, one blue pixel, and two green pixels. Assuming the size of a single color pixel area is P×P, then the size of the color unit can be 2P×2P.

[0071] For example, the size of a single color pixel can be, but is not limited to, 1.12μm × 1.12μm, and the size of a single color unit can be, but is not limited to, 2.24μm × 2.24μm. The minimum cross-sectional size of the dielectric microstructure is not less than 100nm, and the minimum gap between adjacent dielectric microstructures is not less than 50nm; the distance from the incident-side top surface of the uppermost metasurface structure (i.e., the top surface of the incident-side uppermost dielectric microstructure) to the preset detection surface can be, but is not limited to, 3.9μm.

[0072] In an optional embodiment of this application, the number of layers in the multilayer metasurface structure is 3, and the dielectric microstructure unit of each layer of the metasurface structure shares the same preset lattice.

[0073] A multilayer metasurface structure is disposed on one side of the substrate, and a transparent coating medium fills the internal gaps of the multilayer metasurface structure and its outgoing propagation space (i.e., between the uppermost metasurface structure and the preset detection surface). After being modulated by the substrate and the multilayer metasurface structure, the incident light forms an energy distribution corresponding to different target wavelengths on the predetermined detection surface, thereby routing the light of different target wavelengths to the corresponding color pixels. Figure 2 A schematic diagram of a supercell representing a metasurface structure provided in an embodiment of this application. See also... Figure 2 As shown, each color unit corresponds to a supercell, and each supercell contains 4 color pixels. Assuming a supercell has a size of 2.24μm × 2.24μm, and contains 8 × 8 dielectric microstructures, with a preset lattice period of 280nm.

[0074] In an optional embodiment of this application, the multilayer metasurface structure of the color router can be implemented using the following design method. See [link to documentation]. Figure 3 As shown in the figure, this application provides a method for designing a metasurface structure for a color router, including:

[0075] S301: Establish a color router model, which includes: substrate and multilayer metasurface structure;

[0076] S302: Each layer of metasurface structure includes multiple dielectric microstructures arranged according to a preset lattice.

[0077] S303: Determine the target band and the target color pixel corresponding to each target band;

[0078] S304: Determine the variables to be optimized in the color router model, including the cross-sectional geometric parameters of the medium microstructure;

[0079] S305: Combining the target band and target color pixels, a target reference surface is set on the emission side of the multi-layer metasurface structure, and a target near field is constructed on the target reference surface;

[0080] S306: Perform forward electromagnetic simulation on the color router to obtain the simulated near field on the target reference surface;

[0081] S307: Construct a near-field matching representation function based on the target near-field and the simulated near-field, and update the variable to be optimized in reverse based on the near-field matching representation function.

[0082] For example, assuming the dielectric microstructure is rectangular columnar, the length and width of the rectangular columns at different locations in each layer of the metasurface structure can be optimized individually based on the metasurface structure design method. For instance, to ensure structural manufacturability, the minimum cross-sectional dimension of the rectangular column is set to be no less than 100 nm, and the minimum gap between adjacent rectangular columns is set to be no less than 50 nm. The dielectric microstructure unit can also be a cylinder, elliptical cylinder, polygonal cylinder, annular cylinder, or a combination thereof, as long as its cross-sectional geometric parameters can be determined by the above design method and the incident light field can be controlled.

[0083] The predetermined detection surface is located on the emission side of the multilayer metasurface structure and may correspond to the photosensitive surface of a CMOS image sensor or its equivalent position. For example, the CMOS image sensor uses the distance from the top surface of the uppermost dielectric microstructure unit on the incident side to the predetermined detection surface as the working distance, which may be, but is not limited to, 3.9 μm. A transparent coating medium covers the propagation space within this working distance range, allowing the light field modulated by the multilayer metasurface structure to propagate to the predetermined detection surface in the transparent medium environment.

[0084] The variables to be optimized also include at least one of the following parameters: the gap width between adjacent medium microstructures, the height of the medium microstructure, the number of layers of the metasurface structure, and the auxiliary band.

[0085] After passing through the color router, red band incident light is routed to red pixel regions, green band incident light is routed to two green pixel regions, and blue band incident light is routed to blue pixel regions. The routing efficiency of a target band can be defined as the proportion of energy received by the corresponding target color pixel region on the predetermined detector surface to the total incident energy of that target band when the target band is incident. For example, the routing efficiency of the red band is the ratio of the energy received by the red pixel region to the total incident energy of the red band; the routing efficiency of the blue band is the ratio of the energy received by the blue pixel region to the total incident energy of the blue band; and the routing efficiency of the green band is the ratio of the sum of the energy received by the two green pixel regions to the total incident energy of the green band.

[0086] When the red band is incident, the non-target color pixel region includes two green pixel regions and a blue pixel region; when the green band is incident, the non-target color pixel region includes a red pixel region and a blue pixel region; when the blue band is incident, the non-target color pixel region includes a red pixel region and two green pixel regions.

[0087] See Figures 4 to 6 As shown, the energy distribution of incident light with different polarization states and different target wavelengths is different on the predetermined detection surface. Figure 4 , Figure 5 and Figure 6 These are schematic diagrams showing the energy distribution of incident light of different target wavelengths polarized in the X and Y directions on a predetermined detection surface, respectively, according to embodiments of this application. Figure 4 This is an energy distribution diagram of incident light at a target wavelength of 620nm in different polarization states on a predetermined detection surface according to an embodiment of this application. Figure 5 This is an energy distribution diagram of incident light at a target wavelength of 520nm in different polarization states on a predetermined detection surface according to an embodiment of this application. Figure 6 This is an energy distribution diagram of incident light at a target wavelength of 450nm in different polarization states on a predetermined detection surface according to an embodiment of this application.

[0088] In an optional embodiment of this application, a color router model is established, comprising a substrate and a multilayer metasurface structure. Each layer of the metasurface structure includes periodically repeating supercells, each supercell corresponding to a color unit, and each supercell includes multiple dielectric microstructures arranged according to a preset lattice. The number of layers in the metasurface structure may be, but is not limited to, two, three, or four layers. Figure 7This is a schematic diagram of the RGB routing efficiency spectrum curve of the color router with the three-layer metasurface structure provided in this embodiment. As can be seen from the figure, the color router exhibits high routing efficiency at the RGB peak wavelengths, with routing efficiencies of approximately 60%, 80%, and 60% for the red, green, and blue bands at their respective peak wavelengths. Compared to the area ratio limitation of each color pixel region in a traditional RGB Bayer color filter, the color router in this embodiment can redirect light that might otherwise be absorbed or reflected by the color filter back to the corresponding color pixel region, thereby improving the color detection efficiency of the image sensor. Figure 8 The routing efficiency of the color router in this embodiment was evaluated under different incident angles. The results show that the color router can still maintain effective color routing functionality within a certain angle range near normal incidence. In one specific embodiment, when the incident angle is within approximately ±12°, the color router still has effective RGB color routing capability; when the incident angle is within approximately ±6°, the routing efficiency of each color channel does not show a significant decrease. This indicates that the color router in this embodiment has a certain degree of incident angle robustness and is suitable for paraxial imaging scenarios in practical image sensors.

[0089] The number of layers in the metasurface structure, the gap width between adjacent medium microstructures, the height of the medium microstructure, the cross-sectional geometric parameters of the medium microstructure, and auxiliary bands, etc., can all be adjusted and updated during the iteration process.

[0090] Assuming the dielectric microstructure uses rectangular pillars of uniform height, oriented along two orthogonal X and Y directions, the cross-sectional length and width of the rectangular pillars are considered as variables to be optimized. Multilayer metasurface structures share the same pre-defined lattice, but the cross-sectional length and width of the rectangular pillars at different layers and locations can be optimized independently. Processing constraints can be incorporated into the optimization process. For example, if the variables to be optimized are the cross-sectional geometric parameters (length and width) of the rectangular pillars, processing constraints could include a minimum cross-sectional dimension of no less than 100 nm for the rectangular pillars and a minimum gap of no less than 50 nm between adjacent rectangular pillars.

[0091] This application provides a method for designing the metasurface structure of a color router, see [link to relevant documentation]. Figure 9 As shown, the process of optimizing the cross-sectional geometric parameters of the color router model in this design method may include the following steps:

[0092] S901: Determine the cross-sectional geometry, target band, and auxiliary band.

[0093] Cross-sectional geometry parameters can include the cross-sectional length and width of the dielectric microstructure.

[0094] Assume the target wavelength bands can include a red wavelength of 620nm, a green wavelength of 520nm, and a blue wavelength of 450nm. To improve the operating bandwidth, auxiliary wavelengths are set near the peak wavelengths to participate in optimization. These auxiliary wavelengths include 665nm, 650nm, 635nm, 565nm, 550nm, 535nm, 465nm, 435nm, and 420nm. These target and auxiliary wavelengths form the set of wavelengths participating in optimization and are used to constrain the color routing performance in different wavelength bands during the iteration process.

[0095] S902: Determine the region where the target color pixel is located.

[0096] The area where the target color pixel is located can be the target color pixel area where the CMOS sensing element senses RGB colors.

[0097] S903: Determine the preset reference surface and construct the target near field.

[0098] For example, a preset reference plane can be set at 300 nm on the emission side of the uppermost metasurface structure. For each target wavelength, a target focal point is set at the center of the corresponding target color pixel region, and a complex amplitude target near field is constructed according to the hyperbolic phase distribution corresponding to the target focal point position.

[0099] When there are two green pixels in the same color unit, the green band can adopt a dual-focus target near field method to construct the target near field corresponding to the center of the two green pixel regions respectively, and then combine the two target near fields with weights.

[0100] For example, for each target wavelength, a focal point is set at the center of the corresponding target color pixel region. And based on the hyperbolic phase distribution corresponding to the focal point:

[0101]

[0102] in This represents the hyperbolic phase distribution corresponding to the focal point set at the center of the target color pixel region corresponding to the target wavelength. Indicates the target wavelength. Indicates the refractive index of the coating medium. This indicates the distance from the preset reference plane to the CMOS image sensor. This represents the x-coordinate of the focal point set at the center of the corresponding target color pixel region. The vertical coordinate of the focal point set at the center of the corresponding target color pixel area.

[0103] Based on this, the target's near field can be constructed:

[0104]

[0105] in This represents the hyperbolic phase distribution corresponding to the focal point set at the center position of the target color pixel region corresponding to the target wavelength, where j represents the imaginary unit, and satisfies j 2 =-1, The near-field of the target wavelength is represented. Indicates the target wavelength.

[0106] S904: Perform forward electromagnetic simulation on the color router model to obtain the electromagnetic field distribution.

[0107] Electromagnetic field distribution of dielectric microstructure under incident light express.

[0108] S905: Construct a near-field matching representation function based on the target near field and the simulated near field.

[0109] The electromagnetic field distribution of S904 on a preset reference plane is taken as the simulated near field. A near field matching characterization function is constructed based on the degree of matching between the target near field and the simulated near field on the preset reference plane. Assuming the target band involved in the optimization includes a wavelength set Λ and a polarization state set P, when light of target wavelength λ and polarization state p is incident, the simulated near field on the preset reference plane is E(λ, p), and the corresponding target near field is... .

[0110] The degree of matching between the target near field and the simulated near field on the preset reference surface can be expressed as the real part of the overlap integral F(λ,p) in the following formula:

[0111]

[0112] Where λ represents the target wavelength, p represents the polarization state, and Ω represents the preset reference plane. Re denotes complex conjugation, Re represents the operation of finding the real part of the complex number, and E(λ, p) characterizes the simulated near field of the medium microstructure on a pre-defined reference surface under the action of incident light. Characterizes the near field of the target on the preset reference surface. The area element on the preset reference surface is represented. The larger F(λ,p) is, the closer the simulated near field is to the target near field.

[0113] Based on the above formula, the near-field matching representation function can be derived:

[0114]

[0115] Where L represents the near-field matching characterization function, λ represents the target wavelength, and p represents the polarization state. The weighting coefficients for different wavelengths and polarization states are represented by F(λ,p), which indicates the degree of matching between the target near field and the simulated near field on the preset reference plane. Λ represents the set of wavelengths included in the target band participating in the optimization, and P represents the set of polarization states participating in the optimization.

[0116] In an optional embodiment of this application, an additional penalty term may also be set in the near-field matching representation function to prevent the variable to be optimized from exceeding the normal setting range:

[0117] Where L represents the near-field matching characterization function, λ represents the target wavelength, and p represents the polarization state. The weighting coefficients for different wavelengths and polarization states are represented by F(λ,p), which indicates the degree of matching between the target near field and the simulated near field on the preset reference plane. Λ represents the set of wavelengths included in the target band participating in the optimization, and P represents the set of polarization states participating in the optimization. This represents an additional optional penalty term, such as a penalty term for processing constraints, and α represents the weight coefficient corresponding to the additional optional penalty term.

[0118] S906: The adjoint field distribution is defined inversely from the near-field matching characterization function using the adjoint method, and the gradient of the cross-sectional geometric parameters of each medium microstructure is calculated inversely based on the electromagnetic field distribution and the adjoint field distribution.

[0119] Assuming that for each target wavelength and polarization state involved in the optimization, the electromagnetic field distribution under the action of incident light is obtained through forward electromagnetic simulation using S304, and then an adjoint source is constructed based on the near-field matching characterization function, and adjoint simulation is performed to obtain the electromagnetic field distribution of the dielectric microstructure under the adjoint source. .

[0120] Assume the variables to be optimized include:

[0121]

[0122] in Characterizing the gap width of adjacent medium microstructures, , and Indicates the first The distances from the center of each dielectric microstructure to the top, bottom, left, and right edges of the rectangular pillars of the dielectric microstructure. For example, assuming a supercell includes all cells within the entire 2×2 dielectric microstructure (RGGB) CMOS pixel region, a supercell can include 8×8 cells.

[0123] Assume the first The cross-sectional geometric parameters of the dielectric microstructure are: ,in, The gradient of the cross-sectional geometric parameters of each medium microstructure, calculated inversely based on the electromagnetic field distribution and its adjoint field distribution, is expressed as follows: The calculation formula can be as follows:

[0124]

[0125] in, Let λ represent the near-field matching characterization function, λ represent the target wavelength, and p represent the polarization state. The weighting coefficients represent different wavelengths and polarization states; F(λ,p) represents the degree of matching between the target near field and the simulated near field on the preset reference plane; Re represents the operation of finding the real part of the complex number; j represents the imaginary unit; and j satisfies the condition that... 2 =-1, where c represents the speed of light in a vacuum. Λ represents the set of wavelengths included in the target band participating in the optimization, and P represents the set of polarization states participating in the optimization. and Let D represent the dielectric constant of the dielectric microstructure and the dielectric material surrounding the dielectric microstructure, respectively, and let D represent the electric displacement vector. Indicates the first Cross-sectional geometric parameters of a medium microstructure It is the first A slice of the near-field region corresponding to each dielectric microstructure represents the structural parameter of the distance from one side of a rectangular column of the dielectric microstructure to the center. ⊥ and ⊥ represent the field components that are parallel and perpendicular to the slice plane of the near-field region, respectively. This represents the positive simulated electric field intensity component parallel to the slice plane in the near-field region. This represents the accompanying simulated electric field intensity component parallel to the slice plane in the near-field region. This represents the positive simulated electric displacement vector component perpendicular to the slice plane of the near-field region; This represents the accompanying simulated electric displacement vector component perpendicular to the slice plane of the near-field region.

[0126] S907: The gradient calculated in conjunction with S906 is updated using the gradient descent method or its variant to update the cross-sectional geometric parameters of each medium microstructure.

[0127] For example, execute S307 according to the following formula:

[0128]

[0129] in, Let μ represent the cross-sectional geometry parameters at the k-th iteration, and μ represent the step size for updating the cross-sectional geometry parameters in a single iteration. denoted by , Proj represents the gradient of the near-field matching characterization function with respect to the cross-sectional geometric parameters. The constraint projection is used to ensure that the dimensions of the rectangular pillars and the adjacent gaps in the updated dielectric microstructure meet the processing constraints. For example, the minimum cross-sectional dimension of the rectangular pillars in the dielectric microstructure is not less than 100 nm, and the minimum gap between adjacent rectangular pillars is not less than 50 nm.

[0130] S908: Iterate through S903 to S907, or iterate through S904 to S907, until the near-field matching representation function meets the preset convergence condition or the number of repetitions reaches the preset number of iterations.

[0131] In an optional embodiment of this application, during the above iteration process, only the cross-sectional geometric parameters of the dielectric microstructure can be optimized. The cross-sectional geometric parameters of the dielectric microstructure may include the length, width, and gap between adjacent rectangular pillars.

[0132] For example, the color router designed using the above design method has a three-layer titanium dioxide metasurface structure for its medium microstructure, a D263TECO glass substrate, and a homogenized medium with a refractive index of about 1.52 for its transparent coating medium.

[0133] Figure 10 A schematic diagram illustrating the structural distribution of the three-layer metasurface structure used in the color router provided in this embodiment of the application. See also... Figure 10 As shown, assuming the color router adopts a three-layer metasurface structure, all three layers share the same 8×8 preset lattice with a lattice period of 280nm; the height of the dielectric microstructure unit in each layer of the metasurface structure is 450nm; the size of a single-color pixel in the CMOS photosensitive element is 1.12μm×1.12μm, and the size of a color unit is 2.24μm×2.24μm; the distance from the top surface of the uppermost dielectric microstructure on the incident side to the preset detection surface is 3.9μm; the minimum cross-sectional size of the dielectric microstructure is not less than 100nm; and the minimum gap between adjacent dielectric microstructures is not less than 50nm. A supercell includes all units within the entire 2×2 dielectric microstructure (RGGB) CMOS pixel region, and thus a supercell can include 8×8 units.

[0134] See Figure 11 As shown in the embodiments of this application, a method for preparing a color router is also provided, including:

[0135] S1101: Clean and surface treat the substrate.

[0136] S1102: Deposit the first dielectric thin film.

[0137] For example, a titanium dioxide thin film is deposited on one side of a substrate. The thickness of this titanium dioxide thin film corresponds to the height of a dielectric microstructure unit, which can be 450 nm. The titanium dioxide thin film can be formed by atomic layer deposition, sputtering, evaporation, or other thin film deposition processes.

[0138] S1103: Forms an electron beam adhesive layer and a conductive adhesive layer.

[0139] For example, a double layer of MMA, PMMA electron beam adhesive, and conductive adhesive are sequentially spin-coated onto a titanium dioxide film. MMA is a methyl methacrylate electron beam adhesive, and PMMA is a polymethyl methacrylate electron beam adhesive. The conductive adhesive is used to reduce charge accumulation during the electron beam writing process.

[0140] S1104: Electron beam direct writing and development.

[0141] For example, electron beam direct writing exposure is performed based on the target pattern of the first layer metasurface structure, and development is performed to form a resist pattern corresponding to the dielectric microstructure to be prepared on the first layer metasurface structure.

[0142] S1105: Deposit a chromium layer and then peel it off.

[0143] For example, a chromium layer is deposited by electron beam evaporation, followed by a stripping process to remove excess chromium, forming a chromium hard mask on the surface of a titanium dioxide film.

[0144] S1106: Etching the first layer of metasurface structure.

[0145] For example, using a chromium hard mask as an etching mask, inductively coupled plasma etching (ICP) is performed on a titanium dioxide thin film to form a titanium dioxide dielectric microstructure in the first metasurface layer. This etching method can employ ICP (Inductively Coupled Plasma).

[0146] S1107: Remove chrome hard mask.

[0147] After etching, the residual chromium hard mask is removed to obtain the first layer of titanium dioxide dielectric microstructure array.

[0148] S1108: Transparent coating medium filling and planarization.

[0149] A transparent coating medium with a refractive index of approximately 1.52 is spin-coated onto the first layer of titanium dioxide dielectric microstructures to fill the lateral gaps between the microstructures and form a planarized surface for subsequent fabrication. Alternatively, the planarization effect can be improved by combining baking, reflow, etching retreat, or chemical mechanical planarization.

[0150] S1109: Prepare the second layer metasurface structure and the third layer metasurface structure.

[0151] A titanium dioxide thin film is deposited again on the planarized transparent coating medium, and steps S1103 to 1108 are repeated to form a second metasurface structure; then the corresponding steps are repeated again to form a third metasurface structure. When forming the second and third metasurface structures, alignment marks on the substrate or the already formed metasurface structure can be used for interlayer alignment to ensure that the microstructure of each layer of the medium corresponds to the preset lattice.

[0152] S1110: Fill the propagation space.

[0153] After forming the three-layer metasurface structure, the propagation space between the three-layer metasurface structure and the predetermined detection surface is filled by a transparent coating medium, so that the working distance from the top surface of the uppermost medium microstructure unit on the incident side to the predetermined detection surface is 3.9 μm.

[0154] In this embodiment, each metasurface structure is first formed into a dielectric microstructure through thin film deposition, photolithography patterning, hard mask transfer, and etching. Then, a transparent coating medium is used for filling and planarization. This provides a relatively flat processing surface for the fabrication of the next metasurface structure, reducing the difficulty of continuous fabrication of multilayer structures. The transparent coating medium fills the gaps between the dielectric microstructure units, stabilizing the refractive index environment around each layer and reducing the impact of air gaps or surface undulations on light field propagation. Furthermore, it achieves interlayer isolation and planarization, which is beneficial for the alignment and pattern transfer of subsequent layers.

[0155] S1111: Integrated with CMOS image sensor.

[0156] The prepared color router is placed on the photosensitive side of the CMOS image sensor, so that the predetermined detection surface corresponds to the photosensitive surface of the CMOS image sensor, and the supercell of the color router corresponds one-to-one with the color unit in the CMOS image sensor. An anti-reflection film, a planarization layer, a residual color filter array, or a microlens array may also be placed between the color router and the CMOS image sensor.

[0157] In the above embodiments, the image sensor may include a color router, a CMOS photosensitive element, and an anti-reflection film disposed between the color router and the CMOS photosensitive element, which can significantly reduce the reflection loss of light at the interface, increase the visible light transmittance to more than 95%, allow more light to reach the CMOS photosensitive element, improve the imaging brightness and contrast, and at the same time allow different wavelengths of colored light to pass through more uniformly, reducing color cast problems.

[0158] A planarization layer can be set between the color router and the CMOS photosensitive element, which can significantly improve the uniformity of the film thickness of the device, solve the process bottleneck of stacked CMOS sensors, effectively improve the sensor's photosensitivity, signal-to-noise ratio and dynamic range, and also improve the device production yield.

[0159] A residual color filter array can also be set between the color router and the CMOS image sensor to selectively filter light of specific wavelengths, reduce ineffective light loss, significantly enhance light sensitivity in low-light environments, and optimize low-light imaging performance.

[0160] A microlens array can also be placed between the color router and the CMOS photosensitive element to focus light that would otherwise fall on the non-photosensitive area onto the photosensitive diode, thereby improving light sensitivity, enhancing the image quality at the sensor edge, and reducing color channel differences.

[0161] In an optional embodiment of this application, the color router may also include a CMOS image sensor. By integrating the CMOS image sensor into the color router, signal processing, analog-to-digital conversion, and other functions can be integrated onto a single chip, eliminating the need for multiple power supplies and complex driving circuits, significantly reducing device size. Furthermore, relying on mature semiconductor manufacturing processes, overall manufacturing costs can be effectively reduced. It can directly output digital signals without additional complex signal reconstruction processes, enabling rapid integration with backend image processing systems, offering greater scalability, and adapting to the color recognition needs of various consumer electronics and industrial vision applications.

[0162] Each color unit of a CMOS image sensor contains four color pixels, with each color pixel measuring 1.12μm × 1.12μm and each color unit measuring 2.24μm × 2.24μm.

[0163] The electron beam writing, chromium hard mask, and ICP etching processes described above are only preferred fabrication methods. In alternative embodiments of this application, the patterned mask can also be formed by deep ultraviolet lithography, extreme ultraviolet lithography, nanoimprint lithography, or other micro / nano fabrication methods.

[0164] Employing chromium hard masks and ICP etching processes can improve the accuracy of titanium dioxide microstructure pattern transfer and the control over etching depth, making the resulting structure closer to the reverse engineering result. By setting minimum structure size and minimum gap, and using interlayer alignment marks in the process, the fabrication tolerance and repeatability of the device can be improved. The above process route is also compatible with existing micro / nano fabrication and CMOS image sensor integration processes, thus facilitating the fabrication of multilayer dielectric metasurface color routers and subsequent device integration.

[0165] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0166] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application.

Claims

1. A method for designing a metasurface structure of a color router, characterized in that, include: A color router model is established, which includes a substrate and a multilayer metasurface structure. Each layer of metasurface structure includes multiple dielectric microstructures arranged according to a pre-defined lattice. Determine the target band and the target color pixels corresponding to each target band; The variables to be optimized in the color router model are determined, including the cross-sectional geometric parameters of the medium microstructure. By combining the target wavelength and the target color pixel, a target reference surface is set on the emission side of the multilayer metasurface structure, and a target near field is constructed on the target reference surface. A forward electromagnetic simulation was performed on the color router to obtain the simulated near field on the target reference surface; A near-field matching representation function is constructed based on the target near-field and the simulated near-field, and the variable to be optimized is updated in reverse based on the near-field matching representation function.

2. The design method according to claim 1, characterized in that, The variables to be optimized also include at least one of the following parameters: the gap width between adjacent medium microstructures, the height of the medium microstructure, the number of layers of the metasurface structure, and auxiliary bands.

3. The design method according to claim 1, characterized in that, Constructing the target near field on the target reference plane includes: Focus the light of the target wavelength band to the center position of the corresponding target color pixel, and obtain the complex near field constructed by the hyperbolic phase distribution, which is used as the target near field; The near-field matching representation function constructed based on the target near field and the simulated near field includes: Based on the target band and multiple polarization states, at least one of the following is used as the near-field matching characterization function: crosstalk penalty term, polarization equalization term, incident angle robustness and processing constraint penalty term obtained by matching between the target near field and the simulated near field.

4. The design method according to claim 1, characterized in that, The reverse updating of the cross-sectional geometric parameters of the medium microstructure based on the near-field matching characterization function includes: Electromagnetic field distribution under incident light is obtained based on forward electromagnetic simulation; The adjoint field distribution of the inverse definition of the near-field matching characterization function is obtained based on adjoint simulation. By combining the electromagnetic field distribution and the accompanying field distribution, the gradient of the cross-sectional geometric parameters of the medium microstructure is determined; The cross-sectional geometric parameters of the medium microstructure are updated based on the gradient.

5. The design method according to claim 1, characterized in that, The method further includes: After updating the cross-sectional geometric parameters of the medium microstructure, repeat the following steps until the near-field matching characterization function satisfies the preset convergence condition or the number of repetitions reaches the preset number of iterations: A forward electromagnetic simulation was performed on the color router to obtain the simulated near field on the target reference surface; A near-field matching representation function is constructed based on the target near-field and the simulated near-field, and the optimization variables of the medium microstructure are updated in reverse based on the near-field matching representation function.

6. The design method according to claim 1, characterized in that, The method further includes: After updating the cross-sectional geometric parameters of the medium microstructure, repeat the following steps until the near-field matching characterization function satisfies the preset convergence condition or the number of repetitions reaches the preset number of iterations: By combining the target band, auxiliary wavelength, and the target color region of the color router to be prepared, a target reference surface and a target near field matching the target reference surface are set. A forward electromagnetic simulation was performed on the color router to obtain the simulated near field on the reference plane; A near-field matching representation function is constructed based on the target near-field and the simulated near-field, and the optimization variables of the medium microstructure are updated in reverse based on the near-field matching representation function.

7. The design method according to claim 1, characterized in that, The setting of the target reference plane and the target near field matching the target reference plane includes: For each target band, set the focus point at the center of the corresponding target color area; The corresponding hyperbolic phase distribution is determined based on the distance from the preset reference surface to the CMOS photosensitive element and the focal point; By combining the preset refractive index of the coating medium, the distance from the preset reference surface to the photosensitive element, the coordinates of the reference point of the preset reference surface, and the coordinates of the sampling point of the photosensitive element in the color router model, the phase compensation amount at different wavelengths is determined. Using the phase compensation amount as a phase term with a complex exponent, a unit amplitude complex field carrying the distribution of the phase term is generated as the target near field matching the target reference surface.

8. A color-coded router, characterized in that, include: Substrate and multilayer metasurface structure; Each layer of metasurface structure includes multiple dielectric microstructures; Each layer of metasurface structure is filled with a transparent coating medium between it and between the topmost metasurface structure and the preset detection surface.

9. The color router according to claim 8, characterized in that, The color router also includes a CMOS image sensor, and each color unit corresponding to the CMOS image sensor contains 4 color pixels, wherein the size of a single color pixel is 1.12μm×1.12μm, and the size of a single color unit is 2.24μm×2.24μm.

10. The color router according to claim 8, characterized in that, The medium microstructure is rectangular columnar, with a minimum cross-sectional dimension of not less than 100 nm and a minimum gap of not less than 50 nm between adjacent medium microstructures; the working distance between the incident top surface of the uppermost metasurface structure and the preset detection surface includes 3.9 μm.

11. An image sensor, characterized in that, The invention includes a color router, wherein the metasurface structure of the color router is designed using the method described in any one of claims 1 to 7.

Images