Optical devices and methods of manufacture, metasurface structures and methods of design, imaging systems

Abstract

The application discloses an optical device and a preparation method thereof, a metasurface structure and a design method thereof, and an imaging system. The optical device comprises a detector array including a plurality of detection pixels, which is used for receiving incident light and converting the incident light into an electrical signal; and a metasurface structure arranged on the light incidence side of the detector array, wherein the metasurface structure comprises a plurality of light filtering units, and the plurality of light filtering units are arranged in correspondence with the plurality of detection pixels in an optical path; wherein each light filtering unit is configured to transmit light of a target waveband corresponding to the light filtering unit in the incident light to a corresponding detection pixel along a first direction, and deflect at least part of light other than the target waveband in the incident light along a second direction different from the first direction, so that the part of light is incident to an area other than the corresponding detection pixel. The application can realize spectral resolution and color imaging without sacrificing image resolution.

Description

Technical Field

[0001] This invention relates to the field of semiconductor photoelectric detection, and particularly to an optical device and its fabrication method, a metasurface structure and its design method, and an imaging system. Background Technology

[0002] Among all information detection methods, the acquisition and identification of target optical information provides the largest and most comprehensive amount of information, making it the most important detection method. By leveraging optimized material design and substrate removal processes, the response band of standard InP / InxGa1-xAs short-wave infrared detectors can be extended to the visible light spectrum, achieving broadband composite information detection encompassing both visible and short-wave bands. Furthermore, visible light color recognition can be further achieved through algorithmic reconstruction or the use of filter films.

[0003] In terms of algorithmic restoration, current color restoration methods are mainly based on visible light color classification or human visual acceptance for pseudo-color restoration. However, this approach has problems such as complex algorithms, color information not being able to intuitively reflect the spectral information of the target or scene, and the potential for target feature loss or indistinguishability.

[0004] In the application of optical filter films, for example, linear variable filters (LVFs) deposited using high-energy physical vapor deposition have a center wavelength that varies linearly with position between 900nm and 1700nm. They can be combined with InGaAs focal plane arrays to detect the visible light band by selecting and transmitting different wavelengths of light. However, this method sacrifices the effective array size by using pixel fusion, which requires merging multiple pixels to achieve wavelength resolution, resulting in a significant decrease in image resolution. Summary of the Invention

[0005] In view of the shortcomings of the prior art, the technical problem to be solved by the present invention is to provide an optical device and its preparation method, metasurface structure and design method, and imaging system that can achieve spectral resolution and color imaging without sacrificing image resolution.

[0006] To solve the above-mentioned technical problems, one technical solution adopted by the present invention is to provide an optical device, comprising: A detector array, comprising multiple detector pixels, for receiving incident light and converting it into electrical signals; and A metasurface structure is disposed on the light incident side of the detector array. The metasurface structure includes multiple filter units, which are correspondingly disposed with the multiple detector pixels in the optical path. Each of the filter units is configured to: transmit light of the target wavelength band corresponding to itself in the incident light along a first direction to the corresponding detector pixel, and deflect at least a portion of the incident light other than the target wavelength band along a second direction different from the first direction, so that the portion of light is incident on a region other than the corresponding detector pixel.

[0007] To solve the above-mentioned technical problems, another technical solution adopted by the present invention is to provide a metasurface structure, comprising: Substrate; and A metasurface functional layer disposed on the substrate, the metasurface functional layer comprising a plurality of filter units arranged in an array; Each of the aforementioned filter units comprises multiple subwavelength structural units, and each of the aforementioned subwavelength structural units is configured to generate a preset phase modulation of the incident light, the phase modulation being wavelength-dependent; In the same filter unit, the phase modulation of multiple subwavelength structural units works together to make the filter unit exhibit transmission characteristics for light in the target wavelength band and deflection characteristics for light in non-target wavelength bands.

[0008] Furthermore, the plurality of filter units are divided into plurality of filter unit groups, each filter unit group containing the same number of filter units as the total number of target wavelengths. Different filter units in the same filter unit group correspond to different target wavelengths, thereby forming a superpixel capable of responding to multiple target wavelengths simultaneously; and / or Each of the subwavelength structural units consists of several coding units arranged discretely in the horizontal direction. The distribution of each coding unit constitutes a coding pattern, and the coding pattern defines whether a high refractive index medium material exists at each coding unit.

[0009] Furthermore, the encoding pattern is a binary matrix, where "1" represents the presence of a high-refractive-index medium material and "0" represents the absence of a high-refractive-index medium material; the encoding pattern is generated by a reverse design algorithm based on each target wavelength and the size of the filter unit. The metasurface structure also includes a protective layer covering the surfaces of the metasurface functional layer and the substrate.

[0010] To solve the above-mentioned technical problems, another technical solution adopted by the present invention is: a design method for a metasurface structure, used to design at least one filter unit in the metasurface structure, comprising the following steps: Obtain the target design parameters, which include the target wavelength band and the size of the filter unit; Based on the target design parameters, the target phase value corresponding to each subwavelength structural unit in each target band is determined, forming a multi-dimensional target phase vector for each subwavelength structural unit. Based on the multi-dimensional target phase vector, an inverse design algorithm is used to generate an encoding pattern corresponding to each subwavelength structural unit in the filter unit. The encoding pattern is used to define the physical distribution of the corresponding subwavelength structural units.

[0011] Furthermore, the reverse design algorithm includes: A forward prediction model is constructed, which is used to predict the corresponding optical response based on the input coded pattern. The optical response includes the transmission phase and the transmission amplitude. Construct a reverse generation model, which is used to output a corresponding encoded pattern based on the input target phase vector; Using the multi-dimensional target phase vector as input to the inverse generation model, an initial coding pattern is generated. The initial coding pattern is then optimized based on the forward prediction model to obtain an optimized coding pattern.

[0012] Furthermore, methods for constructing forward prediction models include: Generate multiple training encoding patterns under the minimum feature size constraint; Calculate the transmission phase and transmission amplitude of each of the trained encoded patterns at each target wavelength to construct a training dataset; and The forward prediction model is trained based on each of the aforementioned training datasets; and / or In the step of optimizing the initial coding pattern based on the forward prediction model, a genetic algorithm is used, including the following sub-steps: An initial coding library containing multiple candidate coding patterns is generated based on the initial coding pattern; The fitness of each candidate coding pattern is calculated using the forward prediction model as the fitness evaluation function. Based on the fitness, candidate coding patterns are selected and selection, crossover, and mutation operations are performed to generate new candidate coding patterns. Return to the steps for calculating the fitness of each candidate coding pattern until the convergence condition is met.

[0013] To solve the above-mentioned technical problems, another technical solution adopted by the present invention is: to provide a method for fabricating an optical device, comprising the following steps: A detector array is provided, the detector array comprising a plurality of detector pixels; Using the aforementioned metasurface structure design method, a coding pattern for the metasurface structure is obtained; A metasurface structure is formed according to the coded pattern, the metasurface structure including multiple filter units; and The metasurface structure is aligned with the light incident side of the detector array, so that the plurality of filter units and the plurality of detector pixels are correspondingly arranged in the optical path.

[0014] Furthermore, the step of forming a metasurface structure based on the coded pattern includes: Depositing a high-refractive-index dielectric thin film on a substrate; A patterned photoresist layer corresponding to the encoded pattern is formed on the surface of the high refractive index dielectric film; Using the patterned photoresist layer as a mask, the high refractive index dielectric film is etched to form the metasurface functional layer.

[0015] To solve the above-mentioned technical problems, another technical solution adopted by the present invention is: to provide an imaging system, comprising: The optical device is used to receive incident light and output electrical signals; A signal processing module, electrically connected to the optical device, is used to process the electrical signals to form an image; and A display module or storage module is connected to the signal processing module and is used to display or store the image.

[0016] The optical device and its fabrication method, the metasurface structure and its design method, and the imaging system of the present invention have at least the following beneficial effects: (i) By utilizing the wavelength-selective response characteristics of the metasurface structure, each filter unit exhibits transmission characteristics for light in the target band and deflection characteristics for light in non-target bands, enabling simultaneous detection and imaging of multiple bands without sacrificing pixel resolution. This solves the problem of sacrificing the effective array size by pixel fusion in traditional filter film methods, and can significantly improve the amount of information and recognition rate of the detected target.

[0017] (ii) Free-form metasurfaces designed based on the generalized Snell's law can provide preset differentiated transmission phase gradients for different center wavelengths, realize pixel-level wavelength splitting, multi-band multiplexing and optical field encoding, achieve pixel focusing of wavelength multiplexing, and finally realize spectral resolution and color imaging.

[0018] (iii) The present invention has near-infrared band sensing capability, which can detect more spectral characteristics of the target to enhance imaging quality; compared with traditional lenses, metasurface lenses have a smaller structural volume, which can reduce the volume of the imaging system and is easy to integrate; at the same time, the metasurface structure is made of all dielectric materials, which is compatible with standard semiconductor processes, the manufacturing process is mature, and it is suitable for large-scale mass production.

[0019] (iv) By establishing a forward prediction model, rapid prediction from the coded pattern to the optical response is realized, which greatly reduces the dependence on full-wave simulation in the design process and significantly shortens the design cycle. By combining the inverse generation model with the genetic algorithm, end-to-end automatic generation from the target phase distribution to the coded pattern is realized, avoiding the tediousness and inefficiency of traditional manual design. The introduction of the genetic algorithm can break out of local optima and search for coded patterns with higher global performance in a wide design space, ensuring that the optical performance of the final design is close to the theoretical limit.

[0020] (v) By discretizing the target phase distribution into each subwavelength structural unit to form a multi-dimensional target phase vector, each subwavelength structural unit can be independently reverse-engineered and optimized, which greatly reduces the complexity of the single optimization problem and facilitates parallel computing, further improving design efficiency.

[0021] (vi) By performing manufacturing feasibility correction steps, the generated coding pattern is ensured to meet the design rules of the photolithography process, thus improving the manufacturability of the design; by performing tolerance analysis, the design is ensured to have stable performance under actual manufacturing conditions, thus improving the reliability and yield of the device.

[0022] (vii) The coding pattern adopts a binary matrix form, where “1” and “0” represent the presence or absence of high refractive index medium material, respectively. This discretization representation is naturally compatible with processing technology such as electron beam direct writing lithography and can be directly used for mask pattern generation, avoiding the accuracy loss and process deviation caused by intermediate format conversion.

[0023] (viii) This invention is universal and can be flexibly adapted to the application requirements of different target band numbers and different pixel sizes. It is suitable for a variety of application scenarios such as multi-band and hyperspectral imaging, and has broad application prospects in many fields such as reconnaissance and identification, low-light night vision, security and fire protection, and machine vision. Attached Figure Description

[0024] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram of the structure of an embodiment of the optical device of the present invention.

[0025] Figure 2 This is a schematic diagram of the arrangement of filter units in a filter unit group according to an embodiment of the present invention.

[0026] Figure 3 This is a schematic diagram showing the distribution of subwavelength structural units in a filter unit according to an embodiment of the present invention.

[0027] Figure 4This is a schematic diagram of the subwavelength structural unit in one embodiment of the present invention.

[0028] Figure 5 This is a schematic diagram of the encoding pattern of a subwavelength structural unit in one embodiment of the present invention.

[0029] Figure 6 This is a flowchart of one embodiment of the design method for metasurface structures of the present invention.

[0030] Figure 7 This is a flowchart of a reverse design algorithm in one embodiment of the present invention.

[0031] Figure 8 for Figure 7 The flowchart of step S231.

[0032] Figure 9 This is a flowchart illustrating the optimization of the initial coding pattern based on a forward prediction model in one embodiment of the present invention.

[0033] Figure 10 This is a flowchart of one embodiment of the method for preparing the optical device of the present invention.

[0034] Figure 11 for Figure 10 Flowchart of step S300. The meanings of the labels in the attached diagram are as follows: Metasurface structure 100, substrate 110, metasurface functional layer 120, protective layer 130, detector array 200, readout circuit 300. Detailed Implementation

[0035] The following disclosure provides various different embodiments or examples for implementing different features of the invention. Specific embodiments of components and arrangements will be described below to simplify the invention. Of course, these are merely embodiments and are not intended to limit the invention. For example, in the following description, forming a first component above or on a second component may include embodiments where the first and second components are in direct contact, or embodiments where other components may be formed between the first and second components such that the first and second components are not in direct contact. Additionally, reference numerals and / or characters may be repeated in various instances of the invention. Such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or configurations.

[0036] Furthermore, spatial relation terms such as "below," "under," "below," "above," and "above" may be used herein to readily describe the relationship between one element or component and another element (or component) or component (or component) as shown in the figure. In addition to the orientations shown in the figure, spatial relation terms will encompass various different orientations of the device in use or operation. The device may be positioned in other ways (rotated 90 degrees or in other orientations) and will be interpreted accordingly through the spatial relation descriptors used herein.

[0037] Although the numerical ranges and parameter settings presented in this invention are approximations, the numerical settings in specific instances are reported as precisely as possible. Any numerical value, however, inherently contains certain inevitable errors arising from the standard deviation found in the respective test measurements. Similarly, as used herein, the term "about" generally refers to within 10%, 5%, 1%, or 0.5% of a given value or range. Alternatively, the term "about" means within an acceptable average standard error that can be conceived by one of ordinary skill in the art. Except in instances of operation / work, or unless expressly stated otherwise, all numerical ranges, totals, values, and percentages, such as those for material quantities, durations, temperatures, operating conditions, amounts, and other similarities disclosed herein, should be understood to be modified by the term "about" in all cases. Therefore, unless otherwise stated, the numerical parameter settings set forth in this invention and the appended claims are approximations that can be changed upon request. At a minimum, each numerical parameter should be interpreted based on the number of significant figures reported and the application of ordinary rounding techniques. A range herein may be expressed as from one endpoint to another or between two endpoints. All scopes disclosed herein include endpoints unless otherwise stated.

[0038] Furthermore, the technical parts described in this invention and the appended claims are primarily the improved technical parts of this invention, and do not limit the object protected by this invention to only having these technical parts. Other known essential components (structures and / or methods) and / or non-essential components of the object protected, besides the technical parts described in this invention and the appended claims, are not included in this invention and the appended claims because they do not fall within the scope of improvements of this invention; however, this does not mean that the object protected by this invention does not possess these known components.

[0039] Please see Figure 1One embodiment of the optical device of the present invention includes a detector array 200 and a metasurface structure 100. The detector array 200 includes a plurality of detector pixels for receiving incident light and converting it into electrical signals. The metasurface structure 100 is disposed on the light-incident side of the detector array 200, and the metasurface structure 100 includes a plurality of filter units, which are correspondingly disposed with respect to the plurality of detector pixels in the optical path. Each filter unit is configured to: transmit light of a target wavelength band corresponding to itself in the incident light along a first direction to the corresponding detector pixel, and deflect at least a portion of the incident light other than the target wavelength band along a second direction different from the first direction, so that this portion of light is incident on a region other than the corresponding detector pixel.

[0040] It should be understood that the second direction refers to a direction different from the first direction, and is not limited to all wavelengths of light being deflected along the same direction. It can be a single direction or multiple different directions. In a four-band scenario, one filter unit corresponds to one target wavelength. When it first deflects the light of the other three wavelengths, the different wavelengths can be deflected along different second directions. The deflection direction of each wavelength is determined by the phase gradient distribution determined by the coding pattern of multiple subwavelength structural units in the filter unit (as described in the metasurface structure below), and can be preset according to actual design requirements. More specifically: the multiple filter units are divided into multiple filter unit groups, and the number of filter units in each filter unit group is consistent with the total number of target wavelengths. Taking four target wavelengths as an example, each filter unit group consists of four filter units, corresponding to the first, second, third, and fourth wavelengths, respectively. In the same filter unit group, each filter unit transmits the light of its corresponding target wavelength band along a first direction to the corresponding detector pixel, while deflecting the light of the other three wavelength bands along different second directions to the detector pixels corresponding to other filter units in the same filter unit group. Specifically, the filter unit corresponding to the first wavelength band deflects the light of the second, third, and fourth wavelength bands to the detector pixels corresponding to the second, third, and fourth wavelength bands, respectively; the filter unit corresponding to the second wavelength band deflects the light of the first, third, and fourth wavelength bands to the detector pixels corresponding to the first, third, and fourth wavelength bands, respectively; and so on, thereby achieving the separation and parallel detection of multi-wavelength light, and completing the synchronous acquisition of multi-wavelength spectral information without sacrificing pixel resolution.

[0041] Specifically, the detector array 200 can be an InGaAs detector array, a CMOS image sensor array, a CCD image sensor array, or other suitable types of detector arrays. In this embodiment, the detector array 200 is an InGaAs focal plane array, and the target wavelength band includes the visible light band and the short-wave infrared band. The InGaAs detector array is preferably a visible-extended InGaAs focal plane array after removing the InP substrate, to achieve a broad spectral response from visible light to short-wave infrared. The size of the detector pixel in the detector array 200 can be selected according to actual application requirements; for example, the size of the detector pixel is 15μm × 15μm.

[0042] The metasurface structure 100 is aligned and disposed on the light-incident side of the detector array 200 by flip-chip bonding or adhesive bonding. The metasurface structure 100 and the detector array 200 can be bonded together using ultraviolet-cured adhesive (UV adhesive) to ensure a one-to-one correspondence between each filter unit and each detector pixel. For a detailed description of one embodiment of the metasurface structure 100, please refer to the following description of the metasurface structure 100.

[0043] In a preferred embodiment, the optical device further includes a readout circuit 300. The readout circuit 300 is disposed on the side of the detector array 200 opposite to the metasurface structure 100, and is used to read out and process the electrical signals output by the detector array 200. The readout circuit 300 can collect, amplify, and output the weak photocurrent signals generated by each detector pixel of the detector array 200 to an external signal processing module in a predetermined order.

[0044] Please see Figure 1 One embodiment of the metasurface structure 100 of the present invention includes a substrate 110 and a metasurface functional layer 120 disposed on the substrate 110. The metasurface functional layer 120 includes a plurality of filter units arranged in an array. Each filter unit is composed of a plurality of subwavelength structural units, and each subwavelength structural unit is configured to generate a preset phase modulation of incident light, the phase modulation being wavelength-dependent. In the same filter unit, the phase modulation of the plurality of subwavelength structural units works together to make the filter unit exhibit transmission characteristics for light in the target wavelength band and deflection characteristics for light in non-target wavelength bands.

[0045] Specifically, the substrate 110 can be made of alumina, quartz, or other materials with high transmittance in the target wavelength band. In this embodiment, the substrate 110 is an alumina substrate with a thickness of approximately 200 to 350 micrometers. Alumina substrates have excellent light transmittance and chemical stability, making them suitable as the supporting substrate for the metasurface structure 100. The material of the subwavelength structural unit is a high-refractive-index dielectric material. In this embodiment, the high-refractive-index dielectric material is titanium dioxide. It should be understood that the high-refractive-index dielectric material is not limited to titanium dioxide; other materials with high transmittance and appropriate refractive index in the target wavelength band, such as silicon nitride, hafnium oxide, or zirconium oxide, can also be selected. Titanium dioxide has a high refractive index and low optical loss in the visible to near-infrared wavelength band, enabling strong phase modulation of incident light at the subwavelength scale. When light propagates in a titanium dioxide medium, its refractive index is much higher than that of the surrounding medium (such as air or silicon dioxide protective layer 130), resulting in a greater phase delay accumulated at the same physical thickness. This allows the metasurface structure 100 to achieve precise control of the incident light phase over a shorter propagation distance.

[0046] The plurality of filter units are divided into plurality of filter unit groups. The number of filter units in each filter unit group is the same as the total number of target wavelengths. Different filter units in the same filter unit group correspond to different target wavelengths, forming a superpixel capable of simultaneously responding to multiple target wavelengths. For example, please refer to [link to example]. Figure 2 In the four-band configuration, each filter unit group consists of a 2×2 array of four filter units, corresponding to the four bands of 460nm, 520nm, 650nm, and 1064nm, respectively. The target band for the filter unit indicated by the letter B is 460nm, the target band for the filter unit indicated by the letter G is 520nm, the target band for the filter unit indicated by the letter R is 650nm, and the target band for the filter unit indicated by the letter NIR is 1064nm. It should be understood that the specific values ​​of the target bands can be appropriately adjusted according to actual application requirements and are not limited to the specific values ​​mentioned above. In other embodiments, there can be two, three, or more target bands, and the number of filter units in the filter unit group will vary accordingly.

[0047] Each of the aforementioned filter units comprises multiple subwavelength structural units. For example, each filter unit consists of 15 × 15 subwavelength structural units, with a total size of 15 μm × 15 μm. Please refer to [link to relevant documentation]. Figure 3 , Figure 3 (a) is a schematic diagram of the distribution of subwavelength structural units in a filter unit. Figure 3(b) is the coding pattern for a subwavelength structural unit. Each subwavelength structural unit consists of a plurality of coding units discretized in the horizontal direction. The distribution of the coding units constitutes a coding pattern, which defines whether a high-refractive-index medium material exists at each coding unit. Please refer to [link to relevant documentation]. Figure 4 The diagram illustrates the structure of a single subwavelength structural unit. For example, each subwavelength structural unit is 1 micrometer × 1 micrometer in size and consists of a 25 × 25 random binary matrix, meaning each subwavelength structural unit contains 625 coding units. The coding pattern is a binary matrix; please refer to [link to documentation]. Figure 5 , Figure 5 (a) is a schematic diagram of a random binary matrix of subwavelength structural units. Figure 5 (b) is Figure 5 (a) The pixelated physical structure diagram obtained by mapping. In the binary matrix, "1" represents the presence of a high refractive index medium material, and "0" represents the absence of a high refractive index medium material, i.e., air. The encoded pattern is generated by a reverse design algorithm based on the target wavelength and the size of the filter unit. For the specific design method, please refer to the content of the metasurface structure 100 design method below.

[0048] In a preferred embodiment, the metasurface structure 100 further includes a protective layer 130 covering the surfaces of the metasurface functional layer 120 and the substrate 110. The protective layer 130 is made of silicon oxide and has a thickness of approximately 100 to 200 nanometers. It should be understood that the material of the protective layer 130 is not limited to silicon oxide, but can also be silicon nitride, silicon oxynitride, or other materials. The protective layer 130 serves to cover and protect the metasurface functional layer 120, preventing mechanical damage or chemical corrosion to the metasurface functional layer 120 during subsequent processes or use. Simultaneously, silicon oxide has high transmittance in the target wavelength band and will not significantly affect the optical performance of the device.

[0049] Based on the above structure, the optical device of the present invention utilizes the wavelength-selective response characteristics of the metasurface structure 100 to achieve simultaneous detection and imaging of multiple wavelength bands without sacrificing pixel resolution. Specifically, each filter unit is designed based on the generalized Snell's law (phase gradient law), utilizing the wavelength-dependent phase response and resonant scattering difference of light to provide a preset differentiated transmission phase gradient for different center wavelengths, so that light of different wavelengths has different refraction or deflection angles, achieving wavelength-selective transmission and deflection functions. Light of the target wavelength band is transmitted along the normal direction into the absorption region of the corresponding pixel, while light of non-target wavelength bands is deflected according to the preset phase gradient, thereby avoiding signal crosstalk between different wavelength bands, while maintaining the effective pixel scale of the detector array 200, achieving pixel focusing for wavelength multiplexing, and ultimately realizing spectral resolution and color imaging.

[0050] Based on the above embodiments, the optical device and metasurface structure of the present invention have the following unexpected beneficial effects: (i) Through the wavelength selective response characteristics of the metasurface structure, each filter unit exhibits transmission characteristics for light in the target wavelength band and deflection characteristics for light in non-target wavelength bands, realizing simultaneous detection and imaging of multiple wavelength bands without sacrificing pixel resolution, solving the problem of sacrificing the effective array size by pixel fusion in traditional filter film methods. (ii) Wavelength multiplexing focusing is achieved by using free-form metasurfaces, providing preset differentiated transmission phase gradients for different center wavelengths, so that light of different wavelengths has different refraction or deflection angles, achieving pixel focusing for wavelength multiplexing, and finally realizing spectral resolution and color imaging, significantly improving the information content and recognition rate of the detected target. (iii) It has near-infrared band sensing capability, which can detect more spectral characteristics of the target to enhance imaging quality; compared with traditional lenses, free-form metasurface lenses have a smaller structural volume, which can reduce the volume of the imaging system and is easy to integrate. (iv) The metasurface structure is made of all-dielectric materials (such as titanium dioxide), which is compatible with standard semiconductor processes, has mature manufacturing processes, and is suitable for large-scale mass production. (v) By dividing the filter unit into multiple subwavelength structural units, each subwavelength structural unit corresponds to an independent coding pattern, the design freedom is greatly improved, and the phase response can be precisely controlled for different spatial positions, thereby achieving better focusing efficiency and crosstalk suppression performance.

[0051] This invention discloses a design method for a metasurface structure, used to design at least one filter unit in the metasurface structure 100. Please refer to... Figure 6 An embodiment of the design method for metasurface structures of the present invention includes the following steps: S210. Obtain target design parameters, including target wavelength and the size of the filter unit.

[0052] Specifically, the target wavelength range corresponding to the filter unit to be designed and the overall size of the filter unit are first determined. The target wavelength range refers to the wavelength range that the filter unit is expected to transmit to the corresponding detector pixel. Its specific value and number can be flexibly set according to the actual application scenario. For example, in a four-band imaging embodiment, the target wavelength range is set to four wavelengths: 460nm, 520nm, 650nm, and 1064nm. The size of the filter unit should match the pixel size of the detector array 200. For example, when the detector pixel size is 15μm×15μm, the size of the filter unit is also set to 15μm×15μm.

[0053] S220. Based on the target design parameters, determine the target phase value corresponding to each subwavelength structural unit in each target band, and form a multi-dimensional target phase vector for each subwavelength structural unit.

[0054] For a single filter unit, it is divided into multiple subwavelength structural units. First, based on the optical functions required by the filter unit, such as focusing in the target wavelength band and deflecting in the non-target wavelength band, the continuous target phase distribution of the filter unit across the entire spatial range is calculated using the generalized Snell's law (i.e., the phase gradient law). This continuous phase distribution accurately describes the phase response that the filter unit should exhibit at each target wavelength and is the theoretical basis for achieving the required optical functions. Subsequently, the continuous phase distribution is discretized and extracted according to the grid division of the filter unit. Specifically, for a 15μm×15μm filter unit, it is divided into 15×15=225 subwavelength structural units, each with a size of 1μm×1μm. For four target wavelength bands, such as 460nm, 520nm, 650nm, and 1064nm, the target phase value of each subwavelength structural unit in the four target wavelength bands is extracted from the calculated continuous phase distribution. The target phase values ​​in each target wavelength band are then arranged in a fixed wavelength order to form a target phase vector of length 4 for that subwavelength structural unit. The specific calculation method is a mature existing technology and will not be elaborated here.

[0055] For example, based on the continuous phase distribution of the filter unit, the subwavelength structural unit located at the upper left corner of the filter unit, with coordinates (0.5μm, 0.5μm), has target phases of 87°, 168°, 255°, and 78° in the four target wavelength bands, respectively. Therefore, the multi-dimensional target phase vector of this subwavelength structural unit is [87°, 168°, 255°, 78°]. The subwavelength structural unit located at the center of the filter unit, with coordinates (7.5μm, 7.5μm), has a multi-dimensional target phase vector of [0°, 0°, 0°, 0°]. The target phase vectors of the remaining subwavelength structural units follow the same pattern.

[0056] In this way, each of the 225 subwavelength structural units within the filter unit corresponds to a target phase vector of length 4, resulting in a total of 225 sets of input vectors. The multi-dimensional target phase vector of each subwavelength structural unit serves as the input condition for the subsequent inverse generation model, providing a foundation for the independent design of each subwavelength structural unit.

[0057] S230. Based on the multi-dimensional target phase vector, an encoding pattern corresponding to each subwavelength structural unit in the filter unit is generated through a reverse design algorithm. The encoding pattern is used to define the physical distribution of the corresponding subwavelength structural units.

[0058] In this embodiment, the encoding pattern is a binary matrix, where "1" represents the presence of a high-refractive-index medium material and "0" represents the absence of a high-refractive-index medium material. This encoding pattern determines the physical distribution of the subwavelength structural unit, and thus the optical response characteristics of the entire filter unit are jointly determined by the encoding patterns of all subwavelength structural units.

[0059] For details, please refer to Figure 7 Reverse engineering algorithms include: S231. Construct a forward prediction model, which is used to predict the corresponding optical response based on the input coded pattern, wherein the optical response includes transmission phase and transmission amplitude.

[0060] The forward prediction model is trained to predict the transmission phase and amplitude of an input coded pattern at various target wavelengths. The forward prediction model is designed to establish a fast mapping from the coded pattern to the optical response, replacing computationally expensive full-wave electromagnetic simulations. See also... Figure 8 This step S231 includes the following sub-steps: S231a. Generate multiple training coding patterns under the minimum feature size constraint.

[0061] A large number of candidate binary coding patterns are randomly generated within the minimum feature size limit allowed by manufacturing. For example, a 40nm×40nm pattern is used as the smallest coding unit, with each coding pattern corresponding to a 1μm×1μm subwavelength structural unit, and each subwavelength structural unit containing 25×25 coding units. To expand the diversity of the training set, in addition to randomly generated patterns, regularized patterns such as striped, columnar, and randomly perturbed patterns are also introduced as candidate patterns.

[0062] S231b Calculate the transmission phase and transmission amplitude of each training coding pattern at each target wavelength to construct a training dataset.

[0063] Specifically, using the finite-difference time-domain (FDTD) method, rigorous coupled-wave analysis, or equivalent electromagnetic simulation, the transmission phase and amplitude of each training coded pattern at various target wavelengths are calculated to obtain a complete training dataset. This dataset contains a large number of coded patterns and their corresponding optical responses, serving as the foundation for subsequent training of the forward prediction model and the inverse generation model. The transmission amplitude reflects the light transmission efficiency at each wavelength, and its square is directly proportional to the transmittance, serving as an indicator of optical efficiency. For example, for a given training coded pattern, FDTD simulation can calculate its transmission phases at 460nm, 520nm, 650nm, and 1064nm as 2.1rad, 1.8rad, 2.5rad, and 0.9rad, respectively, with transmission amplitudes of 0.85, 0.82, 0.86, and 0.91 (corresponding to transmittances of 72%, 67%, 74%, and 83%).

[0064] S231c, Train the forward prediction model based on the training dataset.

[0065] In this step, a forward prediction neural network is trained to quickly predict the corresponding multi-wavelength transmission phase and transmission amplitude from any input coded pattern, thus replacing the high-cost full-wave simulation. The forward prediction model can employ a suitable network structure such as a convolutional neural network or a fully connected neural network. During training, the coded pattern is used as input, and the corresponding optical response, i.e., transmission phase and transmission amplitude, is used as labels. The network parameters are optimized by minimizing the difference between the predicted and true values.

[0066] S232. Construct a reverse generation model, which is used to output the corresponding encoded pattern based on the input target phase vector.

[0067] Specifically, the inverse generation model can employ an encoder-decoder network structure or a conditional generation network, using a multi-dimensional target phase vector as input and outputting the corresponding encoded pattern, i.e., a probability matrix. During training, phase error loss, transmittance constraint loss, and binarization regularization terms are incorporated to gradually approximate the manufacturable discretized encoded pattern. This inverse generation model and the forward prediction model in step S231 are trained on the same training dataset, sharing the same input-output data space, but with opposite mapping directions.

[0068] S233. Using the multi-dimensional target phase vector as the input of the inverse generation model, an initial coding pattern is generated. The initial coding pattern is then optimized based on the forward prediction model to obtain an optimized coding pattern.

[0069] In this step, the multi-dimensional target phase vectors of each of the 225 subwavelength structural units obtained in step S220 are input into the inverse generation model to independently generate the corresponding initial coding pattern for each subwavelength structural unit. Taking the subwavelength structural unit located in the upper left corner of the filter unit as an example, its multi-dimensional target phase vector is [87°, 168°, 255°, 78°]. After inputting this 4-dimensional vector into the inverse generation model, the model outputs a 25×25 probability matrix. Each element in this matrix has a value between 0 and 1, representing the probability that the corresponding coding unit position is filled with high refractive index material, i.e., the probability of a value of 1. By setting a threshold of 0.5, the probability matrix is ​​binarized into a 25×25 binary matrix, thus obtaining the initial coding pattern of the subwavelength structural unit.

[0070] Following the above method, the multi-dimensional target phase vectors of the subwavelength structural units are input one by one into the inverse generation model to obtain multiple corresponding initial coding patterns. Subsequently, the initial coding pattern of each subwavelength structural unit is optimized based on the forward prediction model, specifically using a genetic algorithm; please refer to [link to relevant documentation]. Figure 9 The steps for optimizing the initial coding pattern based on the forward prediction model are as follows: S233a. Generate an initial coding library containing multiple candidate coding patterns based on each of the initial coding patterns.

[0071] In this step, the initial encoding pattern of each subwavelength structural unit output by the inverse generation model is used as the starting point for the genetic algorithm of the corresponding subwavelength structural unit. Since the genetic algorithm needs to operate on a population containing multiple individuals, but initially there is only a single individual, it is necessary to derive each initial encoding pattern separately to construct its own initial population. Specifically, random perturbation or local mutation operations are applied to each initial encoding pattern, such as randomly flipping the values ​​of 1 to 3 encoding units, generating multiple encoding individuals that are similar to it but have slight differences while maintaining its core features. These individuals together constitute the initial population of the subwavelength structural unit. For example, the initial population size of each subwavelength structural unit can be set to 50 to 200 individuals to ensure sufficient search breadth and computational efficiency.

[0072] S233b. Using the forward prediction model as the fitness evaluation function, calculate the fitness of each candidate coding pattern.

[0073] In this step, each candidate coding pattern in the initial population of each subwavelength structural unit is input into the forward prediction model to obtain its corresponding optical response, i.e., the transmission phase and transmission amplitude at each target wavelength. Based on the difference between the predicted optical response and the target phase distribution of the subwavelength structural unit, the fitness value of each candidate coding pattern is calculated. The fitness function takes into account both phase error and transmittance; a higher fitness indicates that the optical performance of the coding pattern is closer to the design target. The calculation of fitness is a mature existing technique and will not be elaborated upon here.

[0074] For example, the predicted phase of a candidate coding pattern is [85°, 165°, 252°, 75°], and the target phase of the subwavelength structural unit is [87°, 168°, 255°, 78°]. The smaller the sum of squared phase errors between the two, the higher the transmittance, and the greater the fitness value of the candidate coding pattern. Through the rapid calculation of the forward prediction model, the fitness evaluation of all individuals in the population can be completed quickly, thereby significantly improving the optimization efficiency.

[0075] S233c: Based on the fitness, select candidate coding patterns and perform selection, crossover, and mutation operations to generate new candidate coding patterns.

[0076] In this step, individuals with excellent performance are selected from the current population based on their fitness values ​​as parents. The coding fragments of the parents are exchanged through crossover to produce offspring. Random perturbations are introduced through mutation to maintain population diversity, thereby generating new candidate coding patterns and forming the next generation of the population.

[0077] S233d, return to the step of calculating the fitness of each candidate coding pattern until the convergence condition is met.

[0078] In this step, fitness evaluation, selection, crossover, and mutation operations are repeatedly performed on each subwavelength structural unit until a preset convergence condition is met. The convergence condition can be reaching a preset maximum number of iterations, or the fitness value of the best individual in the population no longer significantly increases, for example, by less than 1%. Through continuous iterative optimization, the genetic algorithm can escape local optima and obtain encoding patterns with higher global performance for each subwavelength structural unit.

[0079] Through the above steps, each subwavelength structural unit in a filter unit is designed independently, and each subwavelength structural unit ultimately corresponds to an optimized coding pattern. For example, a 15μm×15μm filter unit contains a total of 225 coding patterns, which are arranged in a 15×15 spatial relationship to form the complete metasurface structure 100 of the filter unit.

[0080] In a preferred embodiment, the design method further includes the following steps: S240. Perform manufacturing feasibility correction and layout export on the generated coded pattern.

[0081] Specifically, the final coded pattern obtained in step S233 is subjected to a photolithographic manufacturability check to ensure that the design can be realized by the process. The manufacturability check includes, but is not limited to: (1) minimum linewidth check: ensuring that the geometric size of each "1" region is not less than the minimum feature size, which is 40nm in this embodiment; (2) minimum spacing check: ensuring that the minimum spacing between any two "1" regions is not less than the minimum feature size; (3) isolated pixel check: identifying and eliminating isolated "1" regions that are too small (such as 1×1 pixels), such structures are easy to fall off during the etching process; (4) edge shrinkage check: checking whether there are corners at the edge of the pattern, and performing chamfering if necessary to improve the process tolerance. Through morphological processing, connectivity enhancement or local weighted correction algorithms, the graphics that do not meet the design rules are automatically repaired, while keeping the phase function from being significantly damaged. Finally, the corrected coded pattern is mapped to the physical size and exported as a GDS or equivalent layout file for subsequent mask fabrication and process implementation.

[0082] In a more preferred embodiment, the design method further includes the following steps: S250, Perform simulation verification and tolerance analysis.

[0083] Specifically, a manufacturing error model is introduced, including but not limited to: (1) linewidth deviation: simulating the situation where the linewidth is larger or smaller than ±5nm during the etching process; (2) thickness fluctuation: simulating the situation where the thickness of the titanium dioxide thin film deposition deviates from the design value by approximately ±5%; (3) alignment error: simulating the situation where the pattern position shifts by approximately ±5nm during the electron beam direct writing process. The Monte Carlo method is used to randomly sample and combine the above manufacturing errors, and the performance of the filter unit under each error combination is simulated and evaluated, and its performance distribution under manufacturing error conditions is statistically analyzed. If the fluctuation range of performance indicators, such as phase error, transmittance, and crosstalk suppression ratio, exceeds the preset tolerance, for example, the transmittance decreases by more than 30% or the crosstalk suppression ratio deteriorates by more than 20%, then the process robustness of the design is considered insufficient. At this time, step S233 can be returned to adjust the fitness function weight of the genetic algorithm, or the search space of the genetic algorithm can be expanded, and optimization can be performed again until a final design that can still maintain stable performance under manufacturing error conditions is obtained.

[0084] Based on the above implementation methods, the design method of metasurface structure of the present invention has the following unexpected beneficial effects: (i) By establishing a forward prediction model, the rapid prediction from the coded pattern to the optical response is realized, which greatly reduces the dependence on full-wave simulation in the design process and significantly shortens the design cycle; (ii) By combining the inverse generation model with the genetic algorithm, the automatic generation from the target phase distribution to the coded pattern is realized, avoiding the tediousness and inefficiency of traditional manual design; (iii) The introduction of the genetic algorithm can break out of the local optimum and search for the coded pattern with higher global performance in a wide design space, ensuring that the optical performance of the final design is close to the theoretical limit; (iv) Through the manufacturing feasibility correction step, the generated coded pattern is ensured to meet the design rules such as the minimum linewidth and minimum spacing of the photolithography process, avoiding the disconnect between design and manufacturing and improving the manufacturability of the design; (v) Through simulation verification and tolerance analysis, the design is ensured to still have stable performance under actual manufacturing conditions, improving the reliability and yield of the device; (vi) The design method is universal and can be flexibly adapted to the application requirements of different target band numbers and different pixel sizes, and is suitable for various application scenarios such as multi-band and hyperspectral imaging.

[0085] This invention discloses a method for fabricating an optical device, used to prepare the aforementioned optical device. Please refer to [link / reference]. Figure 1 , Figure 10 and Figure 11 One embodiment of the method for fabricating the optical device of the present invention includes the following steps: S100, Provide a detector array 200, the detector array 200 including a plurality of detector pixels.

[0086] In this embodiment, the detector array 200 is an InGaAs focal plane array. The fabrication process of the InGaAs focal plane array includes steps such as InP-based material epitaxy, PN junction array fabrication, N-electrode hole formation, electrode fabrication, flip-chip bonding, and substrate removal, ultimately forming an InGaAs focal plane detector with InP substrate removal and visible light extension. The above processes can be implemented using conventional processes known in the art, and will not be described in detail herein.

[0087] S200. Using the metasurface structure design method described above, the coding pattern of metasurface structure 100 is obtained.

[0088] Specifically, the design method of metasurface structure described in steps S210 to S250 is used to generate a coding pattern corresponding to each filter unit. The coding pattern is a binary matrix that defines whether a high refractive index medium material exists at each coding unit.

[0089] S300. A metasurface structure 100 is formed according to the coded pattern, the metasurface structure 100 including a plurality of filter units.

[0090] Specifically, this step includes the following sub-steps: S310, a high refractive index dielectric thin film is deposited on a substrate 110.

[0091] Specifically, a substrate 110 is provided, which is an alumina substrate wafer with a thickness of approximately 200 μm to 350 μm. A titanium dioxide thin film with a thickness of approximately 350 nm to 600 nm is deposited on the alumina substrate using atomic layer deposition (ALD) equipment. The ALD process can form a high-refractive-index dielectric film with uniform thickness and high density on the surface of the substrate 110, providing a high-quality thin film foundation for the subsequent fabrication of the metasurface structure 100. It should be understood that the substrate 110 is not limited to alumina, and may also be quartz or other materials with high transmittance in the target wavelength band; the material of the high-refractive-index dielectric film is not limited to titanium dioxide, and may also be silicon nitride, hafnium oxide, or zirconium oxide, or other materials with high transmittance and appropriate refractive index in the target wavelength band.

[0092] S320. A patterned photoresist layer corresponding to the encoded pattern is formed on the surface of the high refractive index dielectric film.

[0093] Specifically, an electron beam resist, such as PMMA, is spin-coated onto the surface of the titanium dioxide thin film. Exposure is then performed using an electron beam direct-write lithography machine to write a free-form array of metasurface features, forming a patterned photoresist layer corresponding to the encoded pattern. It should be understood that, in addition to electron beam direct-write lithography, other suitable lithography methods such as nanoimprint lithography can also be used to form the patterned photoresist layer.

[0094] S330. Using the patterned photoresist layer as a mask, the high refractive index dielectric film is etched to form the metasurface functional layer 120.

[0095] Specifically, reactive ion etching (RIE) or inductively coupled plasma etching (ICP) is used to selectively etch the titanium dioxide thin film, transferring the pattern from the patterned photoresist layer to the titanium dioxide thin film to form the metasurface functional layer 120. After etching, residual photoresist is removed and the surface is cleaned.

[0096] In a preferred embodiment, after step S330, the following step is further included: S340, a protective layer 130 is formed on the surface of the metasurface functional layer 120 and the substrate 110.

[0097] In this sub-step, a silicon oxide thin film is deposited on the surfaces of the metasurface functional layer 120 and the substrate 110 using plasma-enhanced chemical vapor deposition (PECVD) to protect the metasurface functional layer 120. The thickness of the protective layer 130 is approximately 100 nm to 200 nm. The protective layer 130 effectively prevents the metasurface functional layer 120 from mechanical damage or chemical corrosion during subsequent processes or use. It should be understood that the material of the protective layer 130 is not limited to silicon oxide, but can also be silicon nitride, silicon oxynitride, or other materials.

[0098] After completing the above steps, the alumina wafer is cut into multiple metasurface structures 100 by using a grinding wheel dicing method. Each metasurface structure 100 contains multiple filter units.

[0099] S400, Align the metasurface structure 100 with the light incident side of the detector array 200, so that the plurality of filter units and the plurality of detector pixels are correspondingly arranged in the optical path.

[0100] Specifically, using flip-chip bonding equipment, the cut metasurface array filters are aligned and bonded to the InGaAs PIN focal plane detector array. Uniform pressure is applied during bonding to ensure tight contact between the two. UV adhesive is preferred as the bonding agent to ensure alignment between the metasurface structure 100 and the InGaAs PIN focal plane. Alignment accuracy depends on the pixel size, and a one-to-one correspondence between each filter unit and each detector pixel must be ensured. After bonding, the UV adhesive is cured by ultraviolet light irradiation, completing the fabrication of the optical device.

[0101] Based on the optical device and its fabrication method provided in the above embodiments, the present invention also provides an imaging system. The imaging system includes the optical device described above or an optical device obtained using the fabrication method described above, and further includes a signal processing module and a display module or storage module. The optical device is used to receive incident light and output an electrical signal. The signal processing module, electrically connected to the optical device, is used to process the electrical signal to form an image. The display module or storage module, connected to the signal processing module, is used to display or store the image.

[0102] The optical devices and their fabrication methods, metasurface structures and their design methods, and imaging systems of this invention can achieve spectral resolution and color imaging without sacrificing image resolution. They have significant implications and broad application prospects in many fields such as reconnaissance and identification, low-light night vision, security and fire protection, and machine vision.

[0103] The above embodiments merely illustrate preferred implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention should be determined by the appended claims.

Claims

1. An optical device, characterized in that, include: A detector array, comprising multiple detector pixels, is used to receive incident light and convert it into electrical signals; as well as A metasurface structure is disposed on the light incident side of the detector array. The metasurface structure includes multiple filter units, which are correspondingly disposed with the multiple detector pixels in the optical path. Each of the filter units is configured to: transmit light of the target wavelength band corresponding to itself in the incident light along a first direction to the corresponding detector pixel, and deflect at least a portion of the incident light other than the target wavelength band along a second direction different from the first direction, so that the portion of light is incident on a region other than the corresponding detector pixel.

2. A metasurface structure, characterized in that, include: Substrate; as well as A metasurface functional layer disposed on the substrate, the metasurface functional layer comprising a plurality of filter units arranged in an array; Each of the aforementioned filter units comprises multiple subwavelength structural units, and each of the aforementioned subwavelength structural units is configured to generate a preset phase modulation of the incident light, the phase modulation being wavelength-dependent; In the same filter unit, the phase modulation of multiple subwavelength structural units works together to make the filter unit exhibit transmission characteristics for light in the target wavelength band and deflection characteristics for light in non-target wavelength bands.

3. The metasurface structure according to claim 2, characterized in that: The multiple filter units are divided into multiple filter unit groups. The number of filter units in each filter unit group is the same as the total number of target bands. Different filter units in the same filter unit group correspond to different target bands to form a superpixel that can respond to multiple target bands at the same time. and / or Each of the subwavelength structural units consists of several coding units arranged discretely in the horizontal direction. The distribution of each coding unit constitutes a coding pattern, and the coding pattern defines whether a high refractive index medium material exists at each coding unit.

4. The metasurface structure according to claim 3, characterized in that: The encoding pattern is a binary matrix, where "1" represents the presence of a high-refractive-index medium material and "0" represents the absence of a high-refractive-index medium material. The encoding pattern is generated by a reverse design algorithm based on the target wavelength and the size of the filter unit. The metasurface structure also includes a protective layer covering the surfaces of the metasurface functional layer and the substrate.

5. A design method for metasurface structures, characterized in that, Designing at least one filter unit in the metasurface structure according to any one of claims 2 to 4 includes the following steps: Obtain the target design parameters, which include the target wavelength band and the size of the filter unit; Based on the target design parameters, the target phase value corresponding to each subwavelength structural unit in each target band is determined, forming a multi-dimensional target phase vector for each subwavelength structural unit. Based on the multi-dimensional target phase vector, an inverse design algorithm is used to generate an encoding pattern corresponding to each subwavelength structural unit in the filter unit. The encoding pattern is used to define the physical distribution of the corresponding subwavelength structural units.

6. The design method for metasurface structures according to claim 5, characterized in that, The reverse design algorithm includes: A forward prediction model is constructed, which is used to predict the corresponding optical response based on the input coded pattern. The optical response includes the transmission phase and the transmission amplitude. Construct a reverse generation model, which is used to output a corresponding encoded pattern based on the input target phase vector; Using the multi-dimensional target phase vector as input to the inverse generation model, an initial coding pattern is generated. Based on the forward prediction model, the initial coding pattern is optimized to obtain an optimized coding pattern.

7. The design method for metasurface structures according to claim 6, characterized in that, Methods for constructing forward prediction models include: Generate multiple training encoding patterns under the minimum feature size constraint; Calculate the transmission phase and transmission amplitude of each of the trained encoded patterns at each target wavelength to construct a training dataset; and The forward prediction model is trained based on each of the aforementioned training datasets; and / or In the step of optimizing the initial coding pattern based on the forward prediction model, a genetic algorithm is used, including the following sub-steps: An initial coding library containing multiple candidate coding patterns is generated based on the initial coding pattern; The fitness of each candidate coding pattern is calculated using the forward prediction model as the fitness evaluation function. Based on the fitness, candidate coding patterns are selected and selection, crossover, and mutation operations are performed to generate new candidate coding patterns. Return to the steps for calculating the fitness of each candidate coding pattern until the convergence condition is met.

8. A method for fabricating an optical device, characterized in that, Includes the following steps: A detector array is provided, the detector array comprising a plurality of detector pixels; Using the design method for metasurface structures as described in any one of claims 5 to 7, a coding pattern for the metasurface structure is obtained; A metasurface structure is formed according to the coded pattern, the metasurface structure comprising multiple filter units; as well as The metasurface structure is aligned with the light incident side of the detector array, so that the plurality of filter units and the plurality of detector pixels are correspondingly arranged in the optical path.

9. The method for fabricating an optical device according to claim 8, characterized in that, The steps of forming a metasurface structure based on the coded pattern include: Depositing a high-refractive-index dielectric thin film on a substrate; A patterned photoresist layer corresponding to the encoded pattern is formed on the surface of the high refractive index dielectric film; Using the patterned photoresist layer as a mask, the high refractive index dielectric film is etched to form the metasurface functional layer.

10. An imaging system, characterized in that, include: The optical device of claim 1 is used to receive incident light and output an electrical signal; A signal processing module, electrically connected to the optical device, is used to process the electrical signal to form an image; as well as A display module or storage module is connected to the signal processing module and is used to display or store the image.

Images