Preparation method and spectrum detection method of silicon-based metasurface spectrum detector

Abstract

The invention relates to a preparation method and a spectrum detection method of a silicon-based super-surface spectrum detector, and belongs to the technical field of photoelectric detectors. The silicon-based metasurface spectrum detector is prepared based on a micro-nano metasurface, firstly, an image sensor chip is processed to remove a Bayer layer, then the surface of the image sensor chip is spin-coated with UV glue, then the silicon-based metasurface is inversely buckled on the UV glue, and ultraviolet light is used for curing to obtain the silicon-based metasurface spectrum detector. The spectrum detector is prepared based on the micro-nano metasurface, the price is low, the processing technology is mature, the metasurface can be manufactured at low cost, incident light is modulated through the micro-nano metasurface structure, light current is detected through a sensor, and efficient separation and collection of spectrums can be achieved. Besides, standard spectral signals are collected through the optical fiber spectrometer and used for training the neural network, the neural network is used for data fitting, and the accuracy is high.

Description

Technical Field

[0001] The present invention belongs to the technical field of photoelectric detection devices and relates to a preparation method of a silicon-based supersurface spectrum detector and a spectrum detection method. Background Art

[0002] As an important tool for analyzing the composition of substances, spectrometers play a key role in chemical analysis, environmental monitoring, biomedicine, materials science, etc. Traditional spectrometers usually use prisms or gratings as light-splitting elements. Although they can separate spectra, they have problems such as large size, complex structure, and high cost, which limits their application in portable devices and on-site detection.

[0003] With the development of micro-nano optical technology, metasurface-based spectrometers have attracted widespread attention due to their miniaturization, light weight and low cost. Metasurface is a two-dimensional array composed of a large number of sub-wavelength-scale unit structures, which can regulate the amplitude, phase, polarization and other characteristics of light waves. By precisely designing the unit structure and arrangement of the metasurface, selective transmission, reflection or focusing of light waves of specific wavelengths can be achieved, thereby achieving efficient separation of spectra. Silicon-based materials are an ideal choice for metasurface spectrometers due to their mature preparation process and good optical and thermal stability. However, existing silicon-based metasurface spectrometers still have shortcomings in spectral resolution, wavelength selectivity, environmental adaptability and other aspects. In addition, the preparation process of metasurfaces is complicated and difficult to achieve large-scale production, which limits their commercial application. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a method for preparing a silicon-based supersurface spectral detector and spectral detection, prepare a spectral detector based on a micro-nano supersurface, output spectral signals through neural network processing, and realize simple operation and low-cost spectral detection.

[0005] To achieve the above object, the present invention provides a method for preparing a silicon-based metasurface spectral detector, the method comprising:

[0006] S1. Soak the image sensor chip in NMP solution, ultrasonically heat it to 55°C for 30 minutes, then take out the image sensor chip and blow dry it with nitrogen;

[0007] S2, etching the image sensor chip using oxygen plasma to remove the Bayer layer, immersing the chip in an acetone solution after etching, and then taking it out and drying it with nitrogen;

[0008] S3. Spin-coat UV glue on the surface of the image sensor chip, then place the manufactured silicon-based metasurface upside down on the UV glue, and cure it using ultraviolet light.

[0009] Further, in step S3, the silicon-based metasurface is prepared by the following process:

[0010] 1) Select a silicon wafer and use acetone and isopropanol to ultrasonically clean the silicon wafer;

[0011] 2) Immerse the cleaned silicon wafer in a hydrofluoric acid solution to remove the oxide layer, then rinse with deionized water and blow dry with nitrogen;

[0012] 3) Spin-coating a layer of photoresist on the surface of the silicon wafer, and performing exposure and development by electron beam lithography to obtain a super surface structure pattern;

[0013] 4) Etching by inductively coupled plasma transfers the pattern to the silicon layer on top of the silicon wafer, removes the remaining photoresist and immerses it in a hydrofluoric acid solution to remove silicon oxide on the surface.

[0014] Furthermore, the super surface structure includes an N*N super pixel array, each super pixel includes an M*M array formed by arranging a plurality of unit structures, and the unit structures have different lengths, widths and periods. The length of the unit structure is between 0.4 and 1.2 μm, the width is between 0.4 and 1.2 μm, and the period is 1.4 to 2 times the length.

[0015] On the other hand, the present invention provides a spectrum detection method based on a silicon-based metasurface spectrum detector, the method comprising:

[0016] An adjustable light source is used to output optical signals of different wavelength bands, and the optical signals are output after passing through the collimated optical path and the integrating sphere in sequence;

[0017] The optical signals output by the integrating sphere are collected by a fiber optic spectrometer and a silicon-based metasurface spectrum detector respectively; wherein the fiber optic spectrometer is used to collect standard spectrum signals;

[0018] The data processed by the optical fiber spectrometer is output to the host computer, and the data collected by the silicon-based metasurface spectrum detector is processed by a microcomputer and then output to the host computer;

[0019] In the host computer, the data of the fiber optic spectrometer and the microcomputer are used as training sets to train the neural network, wherein the data of the microcomputer is used as the input of the neural network, and the data of the fiber optic spectrometer is used as the output of the neural network; after the neural network training is completed, it is loaded into the microcomputer;

[0020] The light signal emitted by the light source to be tested is output to the silicon-based metasurface spectrum detector after passing through an integrating sphere. A microcomputer is used to collect the data output by the silicon-based metasurface spectrum detector, and the spectrum is obtained after being processed by a neural network.

[0021] Furthermore, the neural network is a residual convolutional neural network, which uses one-dimensional convolution to replace two-dimensional convolution, uses one-dimensional batch regularization, and uses two layers of fully connected layers for fitting output.

[0022] Furthermore, in the microcomputer, it is first determined whether the light source to be measured is monochromatic light or polychromatic light, and different trained neural networks are selected for processing according to different types, and the output spectrum is fitted.

[0023] The beneficial effects of the present invention are as follows: the present invention prepares a spectrum detector based on a micro-nano super surface, which is cheap and has a mature processing technology. The super surface can be manufactured at a relatively low cost, and the incident light is modulated by the micro-nano super surface structure, and the photocurrent is detected by a sensor, so that efficient separation and collection of spectra can be achieved. In addition, the present invention collects standard spectral signals through an optical fiber spectrometer, and uses them to train a neural network, and uses the neural network to perform data fitting with high accuracy.

[0024] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0026] Figure 1 A three-dimensional structural diagram of a silicon-based supersurface spectral detector according to an embodiment of the present invention;

[0027] Figure 2 is a cross-sectional view of the silicon-based metasurface spectral detector described in an embodiment of the present invention;

[0028] Figure 3 A single structural diagram of the silicon-based metasurface described in an embodiment of the present invention;

[0029] Figure 4 This is a calibration principle diagram described in an embodiment of the present invention;

[0030] Figure 5 A diagram showing the structure of a neural network in an embodiment of the present invention;

[0031] Figure 6 This is a schematic diagram of the spectrum detection principle described in an embodiment of the present invention;

[0032] Figure 7 The flowchart of the neural network used in the embodiment of the present invention is shown in FIG.

[0033] Figure numerals: 1-silicon substrate, 2-metasurface structure, 3-UV glue; 4-silicon base; 5-readout circuit; 6-adjustable light source; 7-collimated optical path; 8-integrating sphere; 9-optical fiber spectrometer; 10-silicon-based metasurface spectral detector; 11-computer; 12-microcomputer. DETAILED DESCRIPTION

[0034] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and the features in the embodiments can be combined with each other without conflict.

[0035] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0036] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0037] Example 1

[0038] This embodiment provides a silicon-based metasurface spectral detector, such as Figure 1 and Figure 2 As shown, it includes a silicon-based metasurface and an image sensor, and the silicon-based metasurface and the image sensor are connected by UV glue 3.

[0039] In this embodiment, the silicon-based supersurface includes a silicon substrate 1 and a supersurface structure 2, such as Figure 3As shown, the metasurface structure 2 is formed by micro-nano processing on an N-type silicon substrate. Based on the metasurface structure of a specific size, the transmittance of different wavelengths can be obtained, thereby modulating different wavelengths. Figure 3 The silicon-based metasurface structure shown is composed of 64*64 superpixels, each of which is a 10*10 unit structure array, where the length, width, period and other parameters of each unit structure in the array are different. Specifically, the unit structure is between 0.4 and 1.2 μm long, 0.4 and 1.2 μm wide, 0.4 μm high, and 1.4 to 2 times the side length. The modulation effect of each array is simulated and calculated by FDTD Solutions, and the resulting structure can modulate multiple wavelengths to different depths.

[0040] In addition, other numbers of arrays can be used to form different numbers of superpixels, such as a 10*9 unit structure array to form a 256*256 superpixel. The unit structure in the array can be a rectangular parallelepiped unit, which is easy to process and has a low cost.

[0041] In this embodiment, the image sensor may be composed of a silicon substrate 4 and a readout circuit 5, or an existing integrated module, such as an IMX219 image sensor, may be used.

[0042] Example 2

[0043] This embodiment provides a method for preparing a silicon-based metasurface spectral detector, which is used to prepare the silicon-based metasurface spectral detector described in Example 1. The method is specifically as follows:

[0044] (1) Immerse the image sensor chip (using Sony IMX219 as an example) in NMP (1-methyl-2-pyrrolidone) solution and ultrasonically heat it to 55°C for 30 minutes;

[0045] (2) Take out the treated IMX219 chip and blow dry it with nitrogen;

[0046] (3) using oxygen plasma for etching, the etching time is 20 minutes, and the power is 400W; specifically, using oxygen plasma to etch the Bayer layer on the chip surface to enhance the intensity of incident light and broaden the spectral range;

[0047] (4) Soak in acetone solution again for 10 min;

[0048] (5) Take out the chip and blow dry it with nitrogen;

[0049] (6) Spin-coat UV glue on the chip surface;

[0050] (7) Place the fabricated silicon-based metasurface upside down on the UV glue and cure it using ultraviolet light for 30 minutes.

[0051] Wherein, the silicon-based supersurface is prepared by the following method:

[0052] (1) using acetone and isopropanol to ultrasonically clean a silicon wafer for 10 minutes; wherein the silicon wafer is N-type silicon with a resistivity of 1 to 10 m·Ω, a thickness of 500 μm, and an oxide layer thickness of 300 nm;

[0053] (2) immersing the silicon wafer in a hydrofluoric acid (HF) solution to remove the oxide layer, rinsing it thoroughly with deionized water after 5 minutes, and drying it with nitrogen; wherein the concentration of the hydrofluoric acid solution is 5% to 10%;

[0054] (3) Spin-coat a layer of photoresist on the surface of the silicon wafer and bake it at 180°C for 10 min;

[0055] (4) exposing and developing by electron beam lithography to obtain a pattern of the super surface structure;

[0056] (5) Transfer the pattern to the top silicon layer by inductively coupled plasma (ICP) etching and then remove the remaining photoresist;

[0057] (6) The prepared metasurface is immersed in hydrofluoric acid for 3 min to remove silicon oxide on the surface.

[0058] Example 3

[0059] This embodiment provides a spectrum detection method based on a silicon-based metasurface spectrum detector, such as Figure 4 As shown, this method requires the following equipment:

[0060] An adjustable light source 6, used for outputting variable signal light;

[0061] A collimated light path 7, used to output the incident light as parallel light;

[0062] The integrating sphere 8 is used to homogenize the incident light and output the optical signal through the splitting optical fiber;

[0063] Optical fiber spectrometer 9, used for collecting standard spectrum signals;

[0064] A microcomputer 12, used for receiving the spectrum signal output by the silicon-based metasurface spectrum detector 10;

[0065] Computer 11 receives data from the fiber optic spectrometer and the microcomputer and uses it for training the neural network.

[0066] The adjustable light source 6 can be an LED light source or a monochromatic light source, etc., and is mainly used to calibrate light signals of different wavelength bands. The collimating light path 7 is composed of a plano-convex lens and a doublet lens with different focal lengths, and is used to output a parallel light beam.

[0067] In this embodiment, the neural network is as follows: Figure 5 As shown in the figure, the neural network adopts the residual network structure. As an improvement on one-dimensional data, one-dimensional convolution is used instead of two-dimensional convolution, one-dimensional batch regularization is used, Flatten is used to expand the data, the pooling layer of the standard residual network is removed, and a two-layer fully connected layer is used for output fitting.

[0068] In this embodiment, the microcomputer 12 may use the 5th generation Raspberry Pi. When using the 5th generation Raspberry Pi, the image sensor in the silicon-based metasurface spectral detector may also select other sensors compatible with the 5th generation Raspberry Pi to adapt to different application scenarios.

[0069] The spectrum detection method described in this embodiment includes:

[0070] 1) The optical signal output by the adjustable light source 6 is output after passing through the collimated optical path 7 and the integrating sphere 8 in sequence;

[0071] 2) The optical signal output by the integrating sphere 8 is collected by the optical fiber spectrometer 9 and the silicon-based metasurface spectrum detector 10 respectively;

[0072] 3) Outputting the data processed by the fiber spectrometer 9 to the computer 11, and outputting the data collected by the silicon-based metasurface spectrum detector to the computer 11 after being processed by the fifth generation Raspberry Pi;

[0073] 4) In the computer 11, the data of the fiber optic spectrometer 9 and the fifth generation Raspberry Pi are used as training sets to train the neural network, wherein the data of the fifth generation Raspberry Pi is used as the input of the neural network, and the data of the fiber optic spectrometer 9 is used as the output of the neural network;

[0074] 5) After the neural network training is completed, the light signal emitted by the light source to be measured is output to the silicon-based metasurface spectrum detector 10 after passing through the integrating sphere 8. The fifth-generation Raspberry Pi is used to collect the data output by the silicon-based metasurface spectrum detector 10 and transmit it to the computer 11. The spectrum is obtained after being processed by the neural network.

[0075] Among them, when training the neural network, the mean square error function is used as the loss function, and the Adam optimizer is used for optimization. The neural network is used to fit the input and output data, and the neural network parameters are adjusted by evaluating the accuracy of the test set. The training can be stopped after the required accuracy is achieved. In addition, different types of data can be trained separately to obtain neural networks for different types of data.

[0076] After the neural network training is completed, it is loaded into the 5th generation Raspberry Pi, which directly reads the optical signal collected by the silicon-based metasurface spectrum detector and outputs the spectrum. Figure 6As shown, the detected light source is placed inside the integrating sphere 8 (the detected light source can also be placed outside the integrating sphere 8, but in comparison, placing it inside the integrating sphere 8 can effectively improve energy utilization), and the light is transmitted to the surface of the silicon-based metasurface spectral detector through optical fiber, and then the data is received by the 5th generation Raspberry Pi and the trained neural network processes the output spectral signal.

[0077] like Figure 7 As shown, in the 5th generation Raspberry Pi, the data of the silicon-based metasurface spectral detector is first collected as the input of the neural network, and then it is determined whether the input data is monochromatic light or polychromatic light. Different trained neural networks are selected according to different types, and the output spectrum is fitted.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A method for preparing a silicon-based metasurface spectral detector, characterized in that: The method includes: The image sensor chip was immersed in NMP solution and ultrasonically heated to 55°C for 30 min, and then the image sensor chip was taken out and dried with nitrogen; The image sensor chip is etched using oxygen plasma to remove the Bayer layer. After etching, it is immersed in an acetone solution and then taken out and dried using nitrogen gas. UV glue is spin-coated on the surface of the image sensor chip, and then the prepared silicon-based metasurface is turned upside down on the UV glue and cured using ultraviolet light.

2. The method according to claim 1, characterized in that The silicon-based supersurface is prepared by the following process: Select a silicon wafer and clean it ultrasonically using acetone and isopropyl alcohol; The cleaned silicon wafer is immersed in a hydrofluoric acid solution to remove the oxide layer, then rinsed with deionized water and blown dry with nitrogen gas; A layer of photoresist is spin-coated on the surface of the silicon wafer, and exposure and development are performed by electron beam lithography to obtain a super surface structure pattern; The pattern is transferred to the silicon layer on top of the silicon wafer by etching with inductively coupled plasma, and the remaining photoresist is removed and then immersed in a hydrofluoric acid solution to remove the silicon oxide on the surface.

3. The method according to claim 2, characterized in that The super surface structure includes an N*N super pixel array, each of which includes an M*M array formed by arranging a number of unit structures, and the unit structures have different lengths, widths and periods.

4. The method according to claim 3, characterized in that The unit structure has a length of 0.4 to 1.2 μm, a width of 0.4 to 1.2 μm, and a period of 1.4 to 2 times the length.

5. A spectrum detection method based on the silicon-based supersurface spectrum detector according to any one of claims 1 to 4, characterized in that: The method comprises: using an adjustable light source to output optical signals of different wavelength bands, and the optical signals are output after passing through a collimated optical path and an integrating sphere in sequence; The optical signal output by the integrating sphere is collected by a fiber optic spectrometer and a silicon-based metasurface spectrum detector respectively; The data processed by the optical fiber spectrometer is output to a host computer, and the data collected by the silicon-based metasurface spectrum detector is processed by a microcomputer and then output to the host computer; In the host computer, the data of the fiber optic spectrometer and the microcomputer are used as training sets to train the neural network, wherein the data of the microcomputer is used as the input of the neural network, and the data of the fiber optic spectrometer is used as the output of the neural network; after the neural network training is completed, it is loaded into the microcomputer; The light signal emitted by the light source to be measured is output to the silicon-based metasurface spectrum detector after passing through an integrating sphere, and a microcomputer is used to collect data output by the silicon-based metasurface spectrum detector, and a spectrum is obtained after being processed by the neural network.

6. The method according to claim 5, characterized in that The neural network is a residual neural network based on ResNet18, which uses one-dimensional convolution to replace two-dimensional convolution, uses one-dimensional batch normalization, and adopts two layers of fully connected layers for fitting output.

7. The method according to claim 5, characterized in that In the microcomputer, it is first determined whether the light source to be measured is monochromatic light or polychromatic light, and different trained neural networks are selected for processing according to different types, and the output spectrum is fitted.

Images