A three-dimensional spatially stacked router and methods of designing and manufacturing the same

Abstract

This application belongs to the field of optoelectronic imaging technology and discloses a three-dimensional spatially stacked router and its design and manufacturing method, including: at least one three-dimensional structural unit; the three-dimensional structural unit includes a predetermined number of sub-pixel regions distributed in the horizontal direction; wherein, the sub-pixel regions include a predetermined number of cuboids stacked in the vertical direction. This application can route different wavelengths of light to corresponding sub-pixel regions through spatial beam splitting, reducing photon loss caused by color filtering; it uses a single-step molding process to fabricate the color router, avoiding complex multi-step processes; and it improves the applicability to unpolarized natural light and small-angle oblique incident light through a low-refractive-index, weak-resonance, multi-layer cascaded spatial modulation architecture. Based on the above conditions, incident light of different wavelengths is routed to predetermined spatial sub-pixel regions, realizing multi-band spatial routing within a small pixel area.

Description

Technical Field

[0001] This application relates to the field of optoelectronic imaging technology, and in particular to a three-dimensional spatially stacked router and its design and manufacturing method. Background Technology

[0002] Existing micro / nano color routers or spatial beam splitters mainly include types such as metallic micro / nano structures, high-refractive-index planar dielectric metasurfaces, low-refractive-index planar dielectric structures, and free-topology reverse-design structures. Each of these approaches has its advantages, but the following problems still exist in visible to near-infrared imaging applications.

[0003] First, metallic micro / nanostructures typically rely on surface plasmon polaritons (SPPs) or localized surface plasmon resonances to achieve spectral selectivity and field localization enhancement. However, they suffer from non-negligible ohmic losses and heat dissipation in the visible light band, resulting in low overall transmission efficiency, which is detrimental to high-sensitivity imaging. Second, high-refractive-index planar dielectric metasurfaces offer strong phase modulation capabilities, but usually require high aspect ratios and high-precision etching. Geometric phase-based schemes are sensitive to the incident polarization state and exhibit significant efficiency reduction under natural light conditions; propagation phase or Mie resonance phase-based schemes may cause changes in wave vector matching conditions, effective optical path, or resonant center due to oblique incidence, leading to color crosstalk and routing failure. Furthermore, low-refractive-index planar dielectric structures, such as glass, photoresist, and low-refractive-index polymers, offer high transparency and ease of fabrication, but suffer from low refractive index contrast. If only a single-layer two-dimensional pattern is used, it is generally difficult to simultaneously achieve efficient spatial dispersion of multiple channels in the 400-800 nm wavelength range within a pixel-level scale of approximately 2 μm × 2 μm. Finally, free topology reverse engineering and optimization methods such as the adjoint method can easily result in complex, discontinuous, or sub-resolution structures. If specific fabrication constraints are not incorporated into the design variables, significant deviations can occur between the optimization results and the actual fabrication results, affecting the reproducibility of device performance.

[0004] Therefore, how to achieve multi-band spatial routing within a certain pixel area has become a problem to be solved. Summary of the Invention

[0005] This application provides a three-dimensional spatially stacked router and its design and manufacturing method. It can route different wavelengths of light to corresponding sub-pixel regions through spatial beam splitting, reducing photon loss caused by color filtering. A single-step molding process is used to fabricate the color router, avoiding complex multi-step processes. A low-refractive-index, weak-resonance, multi-layer cascaded spatial modulation architecture improves its applicability to unpolarized natural light and small-angle oblique incident light. The minimum feature size, interlayer continuity, and absence of dangling constraints of two-photon polymerization lithography are pre-defined in the parameter space, allowing the optimization results to directly correspond to a printable solid structure. Based on these conditions, incident light of different wavelengths is routed to preset spatial sub-pixel regions, achieving multi-band spatial routing within a relatively small pixel area.

[0006] In a first aspect, embodiments of this application provide a three-dimensional spatially stacked router, comprising: At least one three-dimensional structural unit; The three-dimensional structural unit includes a predetermined number of sub-pixel regions distributed in the horizontal direction; wherein, the sub-pixel regions include a predetermined number of cuboids stacked in the vertical direction.

[0007] Furthermore, the length, width, and height of the three-dimensional structural unit are all 2μm; the preset number is 4; and the height of the cuboid is 500 nm.

[0008] Furthermore, the cuboid is made of a dielectric material; the dielectric material is IP-Dip2 photoresist.

[0009] Furthermore, the three-dimensional structural unit also includes a base; within the three-dimensional structural unit, each cuboid has contact support with its underlying base or cuboid.

[0010] Secondly, embodiments of this application provide a design method for a three-dimensional spatially stacked router, including: A three-dimensional structural unit is established based on the finite-difference time-domain method; the three-dimensional structural unit includes a predetermined number of sub-pixel regions distributed in the horizontal direction; wherein, the sub-pixel regions are composed of a predetermined number of cuboids stacked in the vertical direction; Use the horizontal length and vertical width of the cuboid as parameter vectors; The target focal plane is set based on the plane wave incident light source and preset boundary conditions; a preset number of power monitoring areas are set on the target focal plane to obtain a preset number of target channels; The actual transmission spectrum and the corresponding target bandpass function are obtained for each of the preset number of target channels, and the sum of squared errors is calculated. Based on the optimization algorithm and the sum of squared errors to optimize the parameter vector, the three-dimensional parameters used for preparation are obtained.

[0011] Furthermore, the preset number is 4.

[0012] Furthermore, the preset boundary conditions include: the lateral length is less than or equal to 1000 nm and greater than or equal to 200 nm.

[0013] Thirdly, embodiments of this application provide a method for manufacturing a three-dimensional spatially stacked router, comprising: Obtain the supporting substrate; The support substrate was ultrasonically cleaned with deionized water and isopropanol, and then air-dried with high-purity nitrogen to obtain a cleaned support substrate. Photoresist was drop-coated onto the cleaned support substrate to obtain the first sample; the thickness of the photoresist was greater than the height of the three-dimensional structure to be printed. Obtain the three-dimensional parameters used for preparation and convert them into laser writing paths; The first sample is loaded onto a two-photon polymerization lithography platform and processed according to the laser writing path to obtain the second sample; The second sample was developed in the first reagent, cleaned with the second reagent, and then naturally dried at room temperature in a ventilated environment to obtain a three-dimensional stacked router.

[0014] Furthermore, the first reagent is propylene glycol methyl ether acetate; the second reagent is isopropanol.

[0015] Furthermore, the first sample is loaded onto a two-photon polymerization lithography platform and processed according to the laser writing path to obtain a second sample, including: The first sample was loaded onto a two-photon polymerization lithography platform, and laser direct writing was performed based on the laser writing path using an oil immersion objective. The vertical interlayer movement was controlled by a pressure station, and the horizontal planar scanning was controlled by a high-speed scanning galvanometer to obtain the second sample.

[0016] In summary, compared with the prior art, the beneficial effects of the technical solution provided in this application include at least the following: This application provides a three-dimensional spatially stacked router that can route different wavelengths of light to corresponding sub-pixel regions through spatial beam splitting, reducing photon loss caused by color filtering. A single-step fabrication process is used to prepare the color router, avoiding complex multi-step processes. A low-refractive-index, weak-resonance, multi-layer cascaded spatial modulation architecture improves its applicability to unpolarized natural light and small-angle oblique incident light. The minimum feature size, interlayer continuity, and absence of dangling constraints of two-photon polymerization lithography are pre-defined in the parameter space, allowing the optimization results to directly correspond to a printable solid structure. Based on these conditions, incident light of different wavelengths is routed to preset spatial sub-pixel regions, achieving multi-band spatial routing within a relatively small pixel area. Attached Figure Description

[0017] Figure 1 This is a structural diagram of a three-dimensional spatially stacked router provided as an exemplary embodiment of this application.

[0018] Figure 2 A three-dimensional tree structure diagram of a three-dimensional spatially stacked router is provided as an exemplary embodiment of this application.

[0019] Figure 3 A flowchart illustrating a design method for a three-dimensional spatially stacked router, provided as an exemplary embodiment of this application.

[0020] Figure 4 This is a schematic diagram of a simulation model of a design method for a three-dimensional spatially stacked router, provided as an exemplary embodiment of this application.

[0021] Figure 5 A flowchart illustrating a method for manufacturing a three-dimensional spatially stacked router, provided as an exemplary embodiment of this application. Detailed Implementation

[0022] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0023] Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] Please see Figure 1 and Figure 2 This application provides a three-dimensional spatially stacked router, including: At least one three-dimensional structural unit; The three-dimensional structural unit includes a predetermined number of sub-pixel regions distributed in the horizontal direction; wherein, the sub-pixel regions include a predetermined number of cuboids stacked in the vertical direction.

[0025] In some embodiments, the length, width, and height of the three-dimensional structural unit are all 2 μm; the preset number is 4; and the height of the cuboid is 500 nm.

[0026] In some embodiments, the cuboid is made of a dielectric material; the dielectric material is IP-Dip2 photoresist.

[0027] In some embodiments, the three-dimensional structural unit further includes a base; within the three-dimensional structural unit, each cuboid has contact support with the base or cuboid below it.

[0028] In one feasible implementation, the three-dimensional stacked router of this application can be a color router. The periodic unit (i.e., the three-dimensional structural unit) of the color router has a size of 2μm×2μm×2μm. This periodic unit corresponds to four sub-pixel regions arranged in a 2×2 pattern in the horizontal direction, and is formed by stacking four layers of low-refractive-index dielectric cuboids in the vertical direction. Each sub-unit contains four layers of cuboids, and the horizontal length and width of each cuboid can be adjusted independently, with the height of a single layer preferably being 500nm. Thus, the four sub-units, four levels, and two horizontal dimensions per layer form a total of 32 optimizable geometric parameters.

[0029] Preferably, the refractive index of the low-refractive-index dielectric material is in the range of n = 1.35-1.70. Suitable dielectric materials include IP-Dip2 photoresist, other two-photon polymeric photoresists, photocurable resins, silicon dioxide, polymers, or other low-loss transparent dielectrics for the target wavelength band. Compared to high-refractive-index strong resonant structures, this material system tends to form weak resonances and multi-layer cascade modulation, which helps reduce polarization sensitivity and incident angle sensitivity.

[0030] Furthermore, the three-dimensional tree-like stacked structure satisfies the solid continuity rule: the upper dielectric block and the lower dielectric block or substrate have contact support, avoiding cantilevered structures and enclosed cavities. This rule serves two objectives simultaneously: first, to ensure the mechanical stability during the development stage after two-photon polymerization lithography; and second, to enable the reverse-designed structure to be directly fabricated without additional support.

[0031] The three-dimensional spatially stacked router provided in the above embodiments can route light of different wavelengths to corresponding sub-pixel regions through spatial beam splitting, reducing photon loss caused by color filtering; it uses a single-step molding process to fabricate the color router, avoiding complex multi-step processes; and it improves the applicability to unpolarized natural light and small-angle oblique incident light through a low-refractive-index, weak-resonance, and multi-layer cascaded spatial modulation architecture. It incorporates the minimum feature size, interlayer continuity, and absence of dangling constraints of two-photon polymerization lithography into the parameter space, allowing the optimization results to directly correspond to a printable solid structure. Based on the above conditions, incident light of different wavelengths is routed to preset spatial sub-pixel regions, achieving multi-band spatial routing within a relatively small pixel area.

[0032] Please see Figure 3 This application provides a design method for a three-dimensional stacked router, including: Step S01: Establish a three-dimensional structural unit based on the finite-difference time-domain method; the three-dimensional structural unit includes a predetermined number of sub-pixel regions distributed in the horizontal direction; wherein, the sub-pixel regions are composed of a predetermined number of cuboids stacked in the vertical direction.

[0033] Step S02: Use the horizontal length and vertical width of the cuboid as parameter vectors.

[0034] Step S03: Set the target focal plane based on the plane wave incident light source and preset boundary conditions; set a preset number of power monitoring areas on the target focal plane to obtain a preset number of target channels.

[0035] Step S04: Obtain the actual transmission spectrum and the corresponding target bandpass function for each of the preset number of target channels, and calculate the sum of squared errors.

[0036] Step S05: Based on the optimization algorithm and the sum of squared errors to optimize the parameter vector, the three-dimensional parameters used for preparation are obtained.

[0037] In some embodiments, the preset number is 4.

[0038] In some embodiments, the preset boundary conditions include: the lateral length is less than or equal to 1000 nm and greater than or equal to 200 nm.

[0039] In one feasible implementation, parametric reverse design is employed instead of unconstrained free topology optimization. First, the lateral dimension of each cuboid layer is used as a design variable, and this variable is limited to a range that can be stably formed by two-photon polymerization lithography. In this embodiment, the lower bound of the lateral dimension (i.e., the lateral length) is preferably 200 nm, and the upper bound is preferably 1000 nm. This lower bound corresponds to the minimum voxel size for laser direct writing and the structural fidelity requirements after development, while the upper bound corresponds to the sub-pixel scale.

[0040] Four power monitoring regions are set on the target focal plane, corresponding to blue, green, red, and near-infrared sub-pixels, respectively. The four target passbands are 400-500 nm, 500-600 nm, 600-700 nm, and 700-800 nm, respectively. For the Xth channel, its target bandpass function can be defined as: Where X represents the sub-pixel category, including four types: blue, green, red, and near-infrared; λ is the wavelength, ranging from 400 to 800 nm; passband is the passband, which for blue, green, red, and near-infrared sub-pixels are 400-500 nm, 500-600 nm, 600-700 nm, and 700-800 nm, respectively; and stopband is the stopband, which is the wavelength range outside the passband but within the 400-800 nm range.

[0041] The figure of merit (FoM) can be defined as the actual transmission spectrum of each channel. With the corresponding target bandpass function Sum of squared errors between: In a preferred embodiment, the FoM values ​​under different incident polarization states and different incident angles can be further weighted and summed to ensure that the final structure not only has high splitting efficiency under normal incidence but also maintains relatively stable routing performance under P / S polarization and a certain range of oblique incidence. The optimization algorithm can employ particle swarm optimization (PSO), or other methods such as genetic algorithms, simulated annealing, Bayesian optimization, adjoint optimization, or machine learning-assisted optimization.

[0042] In one feasible implementation, when using PSO, the velocity and position of the i-th particle can be updated in the following form: in, Let t be the position of the particle in generation t, which is a set of three-dimensional structural geometric parameters; Let be the velocity of the particle in generation t; This represents the historical optimal solution for the individual in generation t. Let ω be the global historical optimal solution for generation t; ω is the inertia weight. , These are cognitive constants and social constants, respectively. , The coefficients are random. After approximately 500 iterations, the FoM in this embodiment converges to a local minimum of 174; and 32 sets of lateral dimension parameters that can be directly used for fabrication are derived, i.e., the three-dimensional parameters used for fabrication.

[0043] Specifically, the 32 sets of lateral dimension parameters are the output results of the particle swarm optimization algorithm, and their iterative update method is shown in the formula above. In this example, the algorithm iterates for 500 generations until the FoM value reaches a local minimum, at which point the algorithm ends, and the 32 variables output by the algorithm are the selected values. Furthermore, in the reverse design process, the design variables are restricted to the range that can be stably formed by two-photon polymerization lithography. In this embodiment, the lower bound of the lateral dimension is preferably 200 nm, and the upper bound is preferably 1000 nm. Therefore, the lateral dimensions after algorithm optimization are all greater than the minimum lateral feature size of two-photon polymerization lithography (approximately 160 nm). In addition, the three-dimensional tree-like stacked structure satisfies the solid continuity rule: there is contact support between the upper dielectric block and the lower dielectric block or substrate, avoiding cantilevered arms and closed cavities, thus satisfying the stability of the structure. In summary, since the structural features and the range of variables in the algorithm are defined in advance, it is assumed that the structure corresponding to the parameters obtained by the algorithm satisfies the two points of "size greater than the minimum feature size of the process" and "stable and not collapsing", and therefore can be directly used for fabrication without additional judgment methods.

[0044] Please see Figure 4In one feasible implementation, Figure 4 This is a schematic diagram of a simulation model of a design method for a three-dimensional spatially stacked router, provided as an exemplary embodiment of this application.

[0045] First, a simulation model is established. A 2μm × 2μm × 2μm three-dimensional structural element is created in the Finite-Difference Time-Domain (FDTD) software Lumerical. This element contains four sub-elements arranged in a 2×2 pattern. Figure 4 The blue structure in the image represents a subunit composed of four layers of low-refractive-index cuboids stacked together, each layer being 500 nm high. The x-direction length (lateral length) and y-direction width (vertical width) of each cuboid layer are used as optimization variables to form a 32-dimensional parameter vector.

[0046] Secondly, the light source and boundary conditions are set. A plane wave incident light source covering the 400-800 nm wavelength band is used. Figure 4 The gray rectangle below consists of P-polarized and S-polarized components of the same amplitude. The vertical z-direction uses perfectly matched layers (PML) boundary conditions to absorb the light field propagating out of the simulation region; the x and y directions use periodic boundary conditions to simulate the periodic structure of the array.

[0047] Furthermore, monitors and objective functions are configured. Four 1μm × 1μm power monitoring areas are set at the focal plane 1μm away from the structure along the z-direction on the router's output side, corresponding to the four sub-pixel channels respectively. Figure 4 (The four yellow rectangular areas at the top). The sum of the squared errors between the actual transmission spectrum and the target bandpass spectrum for each channel is taken as FoM. The lower the FoM, the higher the transmittance in the target passband and the lower the crosstalk in the stopband.

[0048] Finally, optimization was performed. A particle swarm optimization algorithm was used to search within the 32-dimensional parameter space. Based on the minimum feature size of two-photon polymerization lithography, each lateral parameter was limited to the range of 200-1000 nm. After approximately 500 iterations, the set of geometric parameters with the lowest FoM was selected as the final device structure, and its robustness was further verified under P / S polarization and different incident angles.

[0049] Please see Figure 5 This application provides a method for manufacturing a three-dimensional stacked router, including: Step S11: Obtain the support substrate.

[0050] Step S12: The support substrate is ultrasonically cleaned with deionized water and isopropanol, and then air-dried with high-purity nitrogen to obtain a cleaned support substrate.

[0051] Step S13: Photoresist is drop-coated onto the cleaned support substrate surface to obtain the first sample; the thickness of the photoresist is greater than the height of the three-dimensional structure to be printed.

[0052] Step S14: Obtain the three-dimensional parameters used for preparation and convert them into laser writing paths.

[0053] Step S15: Load the first sample onto the two-photon polymerization lithography platform and process it according to the laser writing path to obtain the second sample.

[0054] Step S16: The second sample is placed in the first reagent for development, cleaned with the second reagent, and naturally dried at room temperature in a ventilated environment to obtain a three-dimensional stacked router.

[0055] In some embodiments, the first reagent is propylene glycol methyl ether acetate; the second reagent is isopropanol.

[0056] In some embodiments, loading a first sample onto a two-photon polymerization lithography platform and processing it according to a laser writing path to obtain a second sample includes: loading the first sample onto a two-photon polymerization lithography platform, performing laser direct writing based on the laser writing path using an oil immersion objective, controlling interlayer movement in the vertical direction using a pressure station, and controlling planar scanning in the horizontal direction using a high-speed scanning galvanometer to obtain the second sample.

[0057] In one feasible implementation, two-photon polymerization (2PP) lithography is used to achieve single-step molding of three-dimensional structures. This method uses a high numerical aperture objective lens to focus a femtosecond laser inside the photoresist, and utilizes the nonlinear absorption at the focal point to induce local polymerization, solidifying the three-dimensional structure point by point, line by line, and layer by layer according to a preset trajectory.

[0058] Compared to multilayer planar overlay and dry etching methods, two-photon polymerization lithography does not require interlayer overlay alignment or plasma etching of the solidified structure. Therefore, it avoids the additional scattering losses caused by interlayer alignment errors and etching sidewall roughness. In this embodiment, IP-Dip2 photoresist, a 63x high numerical aperture (NA=1.4) oil immersion objective, and a femtosecond laser source with a center wavelength of approximately 780 nm are preferably used. Three-dimensional scanning is achieved through a piezoelectric radio and a high-speed scanning galvanometer.

[0059] To ensure the fabricated morphology matches the reverse engineering model, the optimized 3D structural parameters are converted into a General Writing Language (GWL) file (i.e., laser writing path). The preferred slice distance is 250 nm, the profile line distance is 100 nm, and the exposure paths are arranged in ascending order of the z-axis coordinate and ordered by the y-axis coordinate within the same layer. The preferred process window is a laser power of 45 mW and a scanning speed of 10000 μm / s; in practical applications, these can be adjusted equivalently based on the photoresist, objective lens, laser repetition frequency, and development conditions.

[0060] In one feasible implementation, substrate cleaning is performed first. A commercially available fused silica or a detector substrate with a transparent passivation layer is selected as the support substrate, and ultrasonically cleaned sequentially with deionized water and isopropanol for 10 min each, followed by drying with high-purity nitrogen.

[0061] Next, photoresist is applied. IP-Dip2 photoresist is dropped onto the substrate surface to cover the area to be processed, ensuring that the photoresist thickness is greater than the height of the three-dimensional structure to be printed.

[0062] Further, GWL file generation is performed. The optimized 3D cuboid parameters (i.e., the 3D parameters used for fabrication obtained in the design method) are converted into laser writing paths. The structure is layered along the z-axis with a slicing distance of 250 nm; each layer is divided into multiple parallel scan lines with a section line distance of 100 nm. The command to scan a line segment can be represented in the format "x1y1 z1 \n x2 y2 z2 \n Write", where the start and end coordinates define the spatial position of the scan line.

[0063] Next, 3D printing is performed. The sample is loaded onto a two-photon polymerization lithography platform, and immersion laser direct writing is performed using a 63x NA=1.4 oil immersion objective. The interlayer movement in the z-axis is controlled by a voltage generator, and the scanning of the xy-plane is controlled by a high-speed scanning galvanometer. A femtosecond laser with a center wavelength of 780 nm is preferred, with a laser power of 45 mW and a scanning speed of 10000 μm / s.

[0064] Finally, development and drying are performed. After exposure, the sample is developed in propylene glycol methyl ether acetate (PGMEA) for 10 min to remove unexposed photoresist; then it is cleaned with isopropanol (IPA) for 5 min and allowed to air dry at room temperature in a ventilated environment to obtain a cured three-dimensional color router array, i.e., a router stacked in three-dimensional space.

[0065] In summary, firstly, the simulation and characterization results of this application show that, under normal incidence conditions, the peak transmittance of the passband for the four sub-pixel channels (blue, green, red, and near-infrared) in this embodiment reaches 57.6%, 35.3%, 40.6%, and 59.9%, respectively, all exceeding the 25% efficiency limit of traditional color filter architectures. Compared with traditional absorptive color filters, this scheme achieves photon redistribution through spatial routing rather than absorption filtering, which is beneficial for improving the effective incident light utilization rate of the image sensor front end.

[0066] Secondly, for both P-polarized and S-polarized incident light, the spectral response curves of the four sub-pixel regions show a high degree of overlap, indicating that the device does not rely on a single polarization state to operate and can better adapt to natural light and multipath scattered light fields. This performance stems from the weak resonance characteristics of the low-refractive-index dielectric structure and multilayer cascade modulation, rather than the selective response of the geometric phase to a specific circular polarization state.

[0067] Furthermore, within an incident angle range of ±8°, the transmission response of each target channel maintains its dominant position, and the field of view (FoM) remains relatively stable with the incident angle. Further analysis using the "transmittance-wavelength-incident angle" heatmap of the four sub-pixels reveals that when the absolute value of the incident angle exceeds a certain range, crosstalk in non-target bands increases, and routing selectivity decreases. Therefore, under the performance criterion adopted in this embodiment, ±8° can be considered the effective field of view range of this structure.

[0068] Furthermore, the ratio of the average transmittance within the passband to the average transmittance within the stopband of a target channel is defined as the signal-to-noise ratio (SNR). in, The target passband bandwidth is 100 nm in this embodiment; The total stopband bandwidth is 300 nm in this embodiment. The SNR of the four channels—blue, green, red, and near-infrared—are 3.99, 3.06, 1.95, and 3.33, respectively, indicating that the device can suppress color crosstalk between adjacent sub-pixels to a certain extent.

[0069] Furthermore, the three-dimensional structure can be formed in a single-step direct-write process using two-photon polymerization lithography, eliminating the need for multi-layer overlay, nanoscale interlayer alignment, and plasma etching. Low-refractive-index photoresists or transparent polymer materials have lower processing temperatures and can serve as candidate solutions for back-end integration of CMOS image sensors after the front-end circuitry and metal interconnects are completed. It should be noted that actual CMOS process compatibility still needs further verification based on specific photoresist residue, developer, thermal budget, packaging conditions, and reliability testing.

[0070] This application also has the potential for parallel fabrication. Although the single-focus two-photon direct writing speed is limited, this approach can be combined with diffractive optical elements (DOE) or spatial light modulators (SLM) to form multi-focus arrays, and can be used in conjunction with wafer-level step-scan processes to improve array fabrication efficiency.

[0071] In some feasible implementations, as long as the core idea of ​​"three-dimensional multi-layer cascaded stacked structure under manufacturability constraints" is retained, genetic algorithms, simulated annealing, Bayesian optimization, adjoint method, neural network prediction, or multi-algorithm hybrid optimization can all be used as equivalent alternatives.

[0072] In some feasible implementations, IP-Dip2 photoresist is a preferred embodiment material, but this is not a limitation. Transparent dielectric materials with a refractive index in the range of about 1.35-1.70 and low absorption in the target wavelength band are all candidates, including but not limited to silicon dioxide, magnesium fluoride, PMMA, PDMS, polyimide, low refractive index fluorinated polymers, other photocurable resins, and two-photon polymerizable materials.

[0073] In some feasible implementations, the above embodiments use cuboids as basic voxels. However, under the conditions of no suspension, printability, and low refractive index cascade modulation, the basic structural unit can also be replaced with a cylinder, elliptical cylinder, frustum, trapezoid, polygonal cylinder, continuous curved surface, or an equivalent solid structure formed by combining multiple voxels. The number of layers can be 3, 4, 5, 6, 8, or more; the height and lateral span of a single layer can be scaled according to the pixel size, target wavelength, and process resolution.

[0074] In some feasible implementations, the above embodiments demonstrate four-channel spatial spectral dispersion in the 400-800 nm range, encompassing blue, green, red, and near-infrared wavelengths. By adjusting the geometry and objective function, this scheme can be extended to pure visible light, ultraviolet, short-wave infrared, mid-infrared, or specific communication bands. The channel arrangement can also be changed from a 2×2 four-channel configuration to a one-dimensional two / three-channel configuration, a 3×3 nine-channel configuration, a 4×4 sixteen-channel configuration, or other multispectral / hyperspectral arrays.

[0075] In some feasible implementations, the preferred process is two-photon polymerization lithography single-step direct writing. If other three-dimensional micro / nano fabrication methods can form the same or equivalent low-refractive-index three-dimensional cascaded stacked structures, such as multiphoton parallel direct writing, grayscale lithography combined with transparent medium molding, nanoimprint replication, or template transfer, they can also be used as variations of this application.

[0076] In some feasible implementations, in addition to the CMOS image sensor front end, this application can also be used in on-chip multispectral detection, miniature spectrometers, optical communication wavelength division / mode demultiplexing, fluorescence imaging channel separation, polarization / mode / orbital angular momentum sorting and other scenarios.

[0077] 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.

[0078] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, 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. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A three-dimensional spatially stacked router, characterized in that, include: At least one three-dimensional structural unit; The three-dimensional structural unit includes a predetermined number of sub-pixel regions distributed in the horizontal direction; wherein, the sub-pixel regions include a predetermined number of cuboids stacked in the vertical direction.

2. The three-dimensional spatially stacked router according to claim 1, characterized in that, The length, width, and height of the three-dimensional structural unit are all 2 μm; the preset number is 4; and the height of the cuboid is 500 nm.

3. The router with three-dimensional spatial stacking according to claim 1, characterized in that, The cuboid is made of a dielectric material; the dielectric material is IP-Dip2 photoresist.

4. The three-dimensional spatially stacked router according to claim 1, characterized in that, The three-dimensional structural unit also includes a base; within the three-dimensional structural unit, each of the cuboids has contact support with the base or cuboid below it.

5. A design method for a three-dimensional stacked router, characterized in that, include: A three-dimensional structural unit is established based on the finite-difference time-domain method; the three-dimensional structural unit includes a predetermined number of sub-pixel regions distributed in the horizontal direction; wherein, the sub-pixel regions are composed of a predetermined number of cuboids stacked in the vertical direction; The horizontal length and vertical width of the cuboid are used as parameter vectors; A target focal plane is set based on a plane wave incident light source and preset boundary conditions; a preset number of power monitoring areas are set on the target focal plane to obtain a preset number of target channels; Based on the preset number of target channels, the actual transmission spectrum and the corresponding target bandpass function are obtained respectively, and the sum of squared errors is calculated. The parameter vector is optimized based on the optimization algorithm and the sum of squared errors to obtain the three-dimensional parameters used for preparation.

6. The design method of a three-dimensional spatially stacked router according to claim 5, characterized in that, The preset number is 4.

7. The design method of a three-dimensional spatially stacked router according to claim 5, characterized in that, The preset boundary conditions include: the lateral length is less than or equal to 1000 nm and greater than or equal to 200 nm.

8. A method for manufacturing a three-dimensional stacked router, characterized in that, include: Obtain the supporting substrate; The support substrate was ultrasonically cleaned with deionized water and isopropanol, and then dried with high-purity nitrogen to obtain a cleaned support substrate. Photoresist was drop-coated onto the cleaned support substrate surface to obtain the first sample; The thickness of the photoresist is greater than the height of the three-dimensional structure to be printed; Obtain the three-dimensional parameters used for preparation and convert them into laser writing paths; The first sample is loaded onto a two-photon polymerization lithography platform and processed according to the laser writing path to obtain the second sample; The second sample was developed in the first reagent, cleaned with the second reagent, and then naturally dried at room temperature in a ventilated environment to obtain a three-dimensional stacked router.

9. The method for manufacturing a three-dimensional spatially stacked router according to claim 8, characterized in that, The first reagent is propylene glycol methyl ether acetate; the second reagent is isopropanol.

10. The method for manufacturing a three-dimensional spatially stacked router according to claim 8, characterized in that, The process of loading the first sample onto a two-photon polymerization lithography platform and processing it according to the laser writing path to obtain the second sample includes: The first sample is loaded onto a two-photon polymerization lithography platform, and laser direct writing is performed based on the laser writing path using an oil immersion objective. The interlayer movement in the vertical direction is controlled by a pressure station, and the horizontal planar scanning is controlled by a high-speed scanning galvanometer to obtain the second sample.

Images