Wavefront sensing device based on nanogap SPPs control and preparation method thereof

Abstract

The invention belongs to the technical field of micro-nano optoelectronic devices, and particularly discloses a wavefront sensing device based on nano-gap SPPs control and a preparation method thereof, and the wavefront sensing device comprises a substrate, a plurality of metal electrode pairs, a low-dimensional material layer, an isolation layer and a metal nano-gap structure. The plurality of metal electrode pairs are arranged on the substrate, each metal electrode pair is connected with two ends of the low-dimensional material layer, the isolation layer is arranged above the low-dimensional material layer, and the metal nanometer gap structure is arranged above the isolation layer. According to the wavefront sensor provided by the invention, the structural design of integrating light field control and photoelectric conversion is realized, and the wavefront sensor has the advantages in the aspects of device miniaturization and high-resolution precision wavefront sensing.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of micro-nano optical electronic devices, in particular to a wavefront sensing device based on nano-gap SPPs manipulation and a preparation method thereof. BACKGROUND

[0002] Wavefront sensing technology can monitor wavefront distortion caused by phase changes in real time, and is widely used in lithography equipment beam diagnosis, space target detection, optical element defect analysis, etc., and has irreplaceable value in many fields of national major needs and national economy. In the early stage, wavefront was mainly measured by interference technology, that is, the interference pattern generated by the interference between the incident wave and the reference wave was recorded by CCD, and the wavefront phase and intensity information were obtained by combining the data processing unit. However, due to the complex optical system, the device has large volume and poor environmental anti-interference ability. The Shack-Hartmann wavefront sensor developed later mainly calculates the wavefront slope of each sub-wave surface according to the displacement of the microlens array spot center on the CCD position, so as to obtain the whole wavefront information. This type of sensor has advantages in system integration and high stability. However, due to the obvious diffraction effect of the reduction of the size of the microlens, a larger size needs to be processed, which limits the overall resolution of the sensor. With the increasing demand for miniaturization and high performance of modern optical systems, the development of high-integration and high-resolution wavefront sensing devices has become an inevitable trend in the future. SUMMARY

[0003] In view of the deficiencies of the prior art, the present application aims to provide a wavefront sensing device based on nano-gap SPPs manipulation and a preparation method thereof. By equivalent the intensity and phase change of the light field to the angle distribution at different spatial positions, when the incident angle changes slightly, the position of the near-field coupling stripe will change obviously, so that the light intensity signals received by each channel of the detector will be different. By analyzing the relationship between the angle and the channel signal intensity and the channel position, the phase and intensity information of the light field at each spatial position is interpreted, and then the high-resolution sensing of the wavefront signal is realized. That is, by constructing a metal nano-gap and a multi-channel photoelectric detection structure, a wavefront sensing device with high integration and high resolution accuracy, a preparation method and a wavefront sensing method are provided.

[0004] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0005] In a first aspect, the present application provides a wavefront sensor based on nanogap SPPs manipulation, comprising a substrate, a plurality of metal electrode pairs, a low-dimensional material layer, an isolation layer and a metal nanogap structure; the plurality of metal electrode pairs are arranged on the substrate, each metal electrode pair is connected to both ends of the low-dimensional material layer, the isolation layer is arranged above the low-dimensional material layer, and the metal nanogap structure is arranged above the isolation layer.

[0006] Further, the isolation layer is prepared from a light-transmitting material.

[0007] Further, the metal nanogap structure is arranged above the isolation layer, and a nanogap is formed between each two adjacent metal ridges.

[0008] Further, the sizes of the metal nanogap structure are the same or different.

[0009] Further, the low-dimensional material layer is a patterned graphene film.

[0010] In a second aspect, the present application provides a preparation method of the wavefront sensor based on nanogap SPPs manipulation as described above, the preparation method comprising:

[0011] Based on electron beam direct writing and electron beam evaporation process, a first metal electrode and a second metal electrode are prepared on the substrate, and finally the plurality of metal electrode pairs are formed;

[0012] A low-dimensional material film is grown by chemical vapor deposition, and the low-dimensional material film is transferred to the metal electrode pairs by taking PMMA as a support layer;

[0013] The low-dimensional material film is patterned by micro-nano lithography process, and the excess low-dimensional material is removed by reactive ion etching to form the low-dimensional material layer;

[0014] A light-transmitting material is deposited as the isolation layer by using an atomic layer deposition device;

[0015] The metal nanogap structure is prepared on the isolation layer.

[0016] Further, the way of preparing the first metal electrode or the second metal electrode on the substrate comprises:

[0017] The pattern of the first metal electrode or the second metal electrode is exposed, and then 2-5 nm of chromium layer is deposited by combining electron beam evaporation process, followed by depositing a metal material with a thickness of 40-60 nm, soaking in acetone for 5-20 min, and removing the excess metal to obtain the first metal electrode or the second metal electrode.

[0018] Further, the way of preparing the metal nanometer gap structure on the isolation layer comprises:

[0019] Based on an electron beam evaporation device, a metal with a thickness of 80-150 nm is evaporated, an electron beam lithography resist is spin-coated, pattern exposure is performed by using an electron beam direct writing system, the width of the nanometer gap and the alignment accuracy are characterized by using a scanning electron microscope, and etching is performed to obtain the metal nanometer gap structure by taking the electron beam lithography resist as a mask plate and etching the metal, wherein the metal is Au, Ag or Al.

[0020] In a third aspect, the application provides a wavefront sensing method based on the wavefront sensing device based on nanometer gap SPPs manipulation as described above, and the wavefront sensing method comprises:

[0021] The wavefront is equivalent to the distribution of the incident angle in space, and a function relationship among the incident angle f(θ), the output current I and the angle response matrix H is established according to a prediction equation I=Hf(θ);

[0022] Based on the function relationship, a deep neural network is used as a regression prediction model, the angle response matrix H is known, the output current I is taken as input, and the incident angle is taken as a training target, in the training, a loss function is introduced to measure the error between the predicted angle and the real angle, and a back propagation and an optimizer are used to optimize the network parameters, so that the loss function converges to close to zero;

[0023] Based on the trained regression prediction model, the output currents of different metal electrode pairs are taken as input to determine the corresponding incident angles, and the wavefront information is reconstructed according to the corresponding incident angles.

[0024] Compared with the prior art, the application has the following beneficial effects:

[0025] (1) The application can realize the manipulation of the SPPs near field coupling of the metal nanometer gap, the subwavelength characteristics of the metal nanometer gap can realize the localization enhancement of the optical field, break through the diffraction limit of the traditional optical device, and significantly improve the wavefront resolution.

[0026] (2) The application does not need complex optical elements, and the device volume can be reduced to microns, which greatly reduces the volume of the wavefront sensing device and facilitates miniaturization and integration.

[0027] (3) Combined with the deep learning algorithm, the device can reconstruct the phase and intensity information of the wavefront, and realize intelligent sensing of the wavefront. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 A structure diagram of a wavefront sensing device based on nanometer gap SPPs manipulation is provided for the embodiments of the application.

[0029] Figure 2 This is a flowchart illustrating the fabrication process of a wavefront sensing device based on nano-gap SPPs, as provided in an embodiment of the present invention.

[0030] Figure 3 The present invention provides optical simulation results of near-field distribution of SPPs generated by metal nano gaps under different incident light angles. The electric field distribution of the yz plane with the angle is shown for incident light wavelengths of 632.8 nm, 532 nm and 488 nm. The de incident light wavelength is shown for the change of stripe position and intensity with the angle.

[0031] Figure 4 This is a flowchart of a wavefront sensing method provided in an embodiment of the present invention.

[0032] Figure 5 This is a schematic diagram illustrating the principle of angle information interpretation using a deep neural network algorithm provided in an embodiment of the present invention.

[0033] Figure label:

[0034] 10. Substrate; 20. Metal electrode pair; 30. Low-dimensional material layer; 40. Isolation layer; 50. Metal nano-gap structure; 51. Metal ridge; 52. Nano-gap. Detailed Implementation

[0035] The invention will now be further described with reference to the accompanying drawings.

[0036] This invention provides a wavefront sensing device based on nano-gap SPPs manipulation, such as... Figure 1 As shown, the wavefront sensing device based on nano-gap SPPs manipulation includes a substrate 10, multiple metal electrode pairs 20, a low-dimensional material layer 30, an isolation layer 40, and a metal nano-gap structure 50. The multiple metal electrode pairs 20 are disposed on the substrate 10, and each metal electrode pair 20 is connected to both ends of a low-dimensional material layer 30. The isolation layer 40 is disposed above the low-dimensional material layer 30, and the metal nano-gap structure 50 is disposed above the isolation layer 40.

[0037] This wavefront sensing device utilizes the strong local field enhancement effect of metal nano-gap SPPs and the electrical controllability of low-dimensional materials to achieve high-sensitivity wavefront sensing. Its working principle is as follows:

[0038] The metal nanogap structure 50 is capable of regulating the light field distribution by virtue of the strong coupling effect. A distributed detection unit is used to collect and output the light intensity signals at different spatial positions in real time. When the incident angle changes, the light field distribution received by the surface of the low-dimensional material layer 30 changes, thereby causing the output currents of the channels (i.e., the metal electrode pairs 20) to be different. The output currents of the channels at different incident angles are collected to establish an angle-current mapping relationship. The incident light is equivalent to an angle distribution at different spatial positions. The output currents at different positions are obtained by using a spatial scanning method. The incident angle is reconstructed by combining a deep learning algorithm, and the phase and intensity information of the light field are interpreted, thereby realizing the reconstruction of the wave front information.

[0039] In some embodiments, the isolation layer 40 is made of a light-transmitting material. The isolation layer 40 mainly plays an electrical isolation role. The electrical isolation refers to preventing the electrical short circuit between the metal nanogap structure 50 above and the low-dimensional material layer 30 below. For example only, the isolation layer 40 can be selected from one of silicon dioxide, aluminum oxide, silicon nitride, magnesium fluoride, and hafnium oxide.

[0040] In some embodiments, the metal nanogap structure 50 includes a plurality of metal ridges 51, and a nanogap 52 is formed between each two adjacent metal ridges 51.

[0041] Each metal ridge 51 can cause the surface charge to be highly concentrated under illumination, and an ultrastrong local electric field is formed near the tip. The width of the nanogap 52 between the two adjacent metal ridges 51 is less than a subwavelength, SPPs are generated, the SPPs in the adjacent nanogaps 52 are coupled and resonated with each other, the light field is extremely compressed in the nanoscale space, and the diffraction limit is broken. The wave front distortion of the incident light will react to the change in the incident light angle, causing the effective refractive index and the diffraction fringe pattern of the SPPs of each nanogap 52 to change. By detecting the non-uniform photoelectric current response of the low-dimensional material layer 30 by using the plurality of metal electrode pairs 20 and the neural network algorithm, the wave front phase distribution can be reconstructed.

[0042] In some embodiments, the sizes of the metal ridges 51 are the same or different.

[0043] In some embodiments, the low-dimensional material layer 30 is a patterned graphene film.

[0044] Embodiment 2:

[0045] The embodiment of the present application provides a preparation method of the wave front sensor based on the nanogap SPPs manipulation described in embodiment 1. The preparation process is as shown in Figure 2 The preparation method comprises the following steps:

[0046] Step 1, asymmetric metal electrode is prepared by using EBPG5150 electron beam direct writing system. First, electrode 1 is exposed to light, and then 2-5 nm of chromium layer is deposited by electron beam evaporation process, and then metal (metal material can be AL, Ag, etc., thickness is 40-60 nm) is deposited, and then the excess metal is removed by acetone immersion for 5-20 min combined with metal stripping process. Subsequently, the same process is used to prepare metal electrode 2 (metal material can be Au, Pt, etc., thickness is 40-60 nm).

[0047] Step 2, low-dimensional material is grown by chemical vapor deposition method, and in this scheme, graphene is taken as an example for pattern transfer. The surface PMMA (PMMA and ethyl acetate ratio is 1:3) has a thickness of about 200 nm. It is immersed in an acidic solution (hydrochloric acid, hydrogen peroxide and water in a ratio of 2:1:60), and then the graphene at the bottom of the copper foil is removed by plasma etching process. The graphene film is further cleaned by repeated deionized water cleaning process to obtain clean graphene film. Considering that the uneven surface of the asymmetric metal electrode at the bottom is easy to cause damage to the graphene, the graphene film is naturally dried for 4-12 hours to realize lossless transfer of the graphene.

[0048] Step 3, the graphene is patterned by combining micro-nano lithography process. In order to avoid the pollution of photoresist to graphene material in the patterning process, double-layer glue (S1805 and LOR photoresist) process and soluble buffer layer etching technology are adopted to pattern the graphene film, and the excess photoresist on the surface of the graphene is removed by using developing solution. Isolation layer deposition (such as silicon dioxide, silicon nitride) is carried out by combining coating equipment, and the thickness is about 5-20 nm to realize the preparation of the detection unit.

[0049] Step 4, high-precision metal nanogap structure is prepared by using electron beam direct writing technology. First, Au, Ag or Al material is evaporated by using electron beam evaporation equipment for 80-150 nm. Electron beam photoresist is spin-coated to about 150 nm thick, and then pattern exposure is carried out by using electron beam direct writing system, and scanning electron microscope is used to characterize the width and alignment accuracy of the nanogap. Then, etching is carried out, taking photoresist as a mask, and finally the device is prepared by etching the metal.

[0050] Example 3:

[0051] The embodiment of the present application provides a wavefront sensing method based on a nanogap SPPs manipulation-based wavefront sensor device as described in example 1, as shown in Figure 1As shown, the substrate 10 of the wavefront sensing device is selected as silicon dioxide, above which is an electrode layer designed as a plurality of metal electrode pairs 20, on the one hand, better collection of wavefront information, on the other hand, but also to provide a rich data set for interpreting wavefront information. Above the electrode layer is a low-dimensional material layer 30 as a photosensitive material for photoelectric conversion. Above the photosensitive material is an isolation layer 40, and the topmost is a metal nanogap structure 50, which can be set as 2 to N metal nanogaps, by optimizing the gap spacing and width, to realize the regulation of the wavefront information. Figure 3 The optical simulation results of the SPPs near-field distribution generated by the metal nanogap for different incident light angles are shown in Figures d-f. Figure 3 As can be observed from Figures d-f, the near-field intensity changes significantly under different incident angles, indicating that the metal nanogap SPPs are sensitive to the incident angle.

[0052] Based on this, please refer to Figure 4 The wavefront sensing method includes the following steps S10-S30:

[0053] S10: The wavefront is equivalent to the distribution of the incident angle in space, and the function relationship between the incident angle f(θ), the output current I and the angle response matrix H is established according to the prediction equation I=Hf(θ).

[0054] S20: Based on the function relationship, a deep neural network is used as a regression prediction model, wherein the angle response matrix H is known, the output current I is used as the input, and the incident angle is used as the training target. In training, a loss function is introduced to measure the error between the predicted angle and the true angle, and a back propagation and an optimizer are used to optimize the network parameters, so that the loss function converges to close to zero.

[0055] Please refer to Figure 5 Take the incident light wavelength of 532 nm as an example, and introduce in detail the principle of the regression prediction model for interpreting the angle information.

[0056] S201: The incident light with a wavelength of 532 nm is incident to the detector at an incident angle θ, and the spatial position of the light is defined by the y position and the z position, and the incident angle θ is the target physical quantity to be interpreted.

[0057] S202: The wavefront sensing device includes a plurality of metal electrode pairs 20, i.e. a distributed electrode signal output is formed, and each metal electrode pair 20 is used as an electrode channel, so that H={H1(θ), H2(θ),..., H n (θ)}, I={I1(θ), I2(θ),..., I n (θ)} wherein the response characteristics of the nth metal electrode pair 20 to different incident angles are represented by the response function H n(θ) describes the distribution of the photoelectric response of the unit when the incident angle θ changes. When the incident angle is θ, the current output by the nth metal electrode pair 20 is I n (θ), which is directly related to the response function H n (θ), is the electrical signal driven by the response function. The current signals of all electrode channels are integrated to form a distributed electrode signal output, which enters the subsequent processing link.

[0058] S203: Establish a mathematical mapping between the incident angle and the actual output current I n of the nth metal electrode pair through the integral formula, which is expressed as:

[0059]

[0060] where θ1~θ M represents the angle integration interval, covering the possible value range of the incident angle.

[0061] S204: Use a deep neural network (the network structure shown in the figure includes an input layer, a hidden layer, and an output layer) to perform angle interpretation, and the process is as follows:

[0062] The input layer accesses the current signal output by the distributed electrode, i.e., the actual output current I n of the nth metal electrode pair, and takes the electrical measurement value as the network input feature.

[0063] The neural network performs nonlinear transformation and feature extraction on the input current data through the learned model parameters, fitting the mapping relationship between the current distribution and the angle. During the training process, the network optimizes the internal weights to make the output as close to the true angle as possible to adapt to the needs of angle inference from current.

[0064] The output layer outputs the angle of the incident light directly interpreted, completing the reconstruction calculation from electrical signal to physical angle.

[0065] S205: The deep neural network output layer finally outputs the determined angle value, realizes the interpretation of the incident light angle information, and completes the whole process from optical incidence to angle measurement.

[0066] In the case of using a deep neural network as a regression prediction model, based on supervised training, the output current is taken as the training input, and the incident angle is taken as the training label. The output current is encoded and processed to extract its spatial distribution features, thereby enhancing the model's feature extraction capability for current data. The decoder network is used to map the extracted spatial distribution features to the angle space, and output the predicted incident angle value f*. The loss function is introduced to measure the error between the predicted angle and the real angle. The back propagation and optimizer are used to optimize the network parameters, so that the loss function converges to close to zero, ensuring that the model has high prediction accuracy. The neural network structure and hyperparameters are adjusted to improve the prediction accuracy and generalization ability of the model. After training, the input current data is used to predict the incident angle. By inputting current data at different spatial positions (i.e., current signals collected by different pairs of metal electrodes), the entire wavefront information can be reconstructed.

[0067] S30: Based on the trained regression prediction model, the output current of different pairs of metal electrodes is taken as the input to determine the corresponding incident angle, and the wavefront information is reconstructed according to the corresponding incident angle.

[0068] It should be noted that in actual implementation, steps S10-S20 can be configured to run in a server with relatively strong computing power. Based on the trained regression detection model determined in steps S10-S20, the trained regression detection model is configured in the operation chip of the device, so that step S30 can be implemented in the wavefront perception device. For example, the wavefront perception device further includes a data acquisition component and a control chip. The data acquisition component is used to acquire the output current of different pairs of metal electrodes, so the data acquisition component can be selected as a current sensor. The data acquisition component is connected to the control chip to output the output current of different pairs of metal electrodes to the control chip. The trained regression detection model is configured in the control chip. By operation, the incident angle at each spatial position can be calculated, and the entire wavefront information can be reconstructed.

[0069] The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for part or all of the technical features. These modifications or substitutions do not change the essence of the corresponding technical solutions, which should be covered within the scope of the claims and description of the present application.

Claims

1. A wavefront sensor device based on nanogap SPPs manipulation, characterized in that, The device comprises a substrate, a plurality of metal electrode pairs, a low-dimensional material layer, an isolation layer, and a metal nanogap structure; the plurality of metal electrode pairs are arranged on the substrate, each metal electrode pair is connected to both ends of the low-dimensional material layer, the isolation layer is arranged above the low-dimensional material layer, and the metal nanogap structure is arranged above the isolation layer.

2. The wavefront sensor device based on nanogap SPPs manipulation according to claim 1, wherein, The isolation layer is made of a light-transmitting material.

3. The wavefront sensor device based on nanogap SPPs manipulation according to claim 1, wherein, The metal nanogap structure is arranged above the isolation layer, and a nanogap is formed between every two adjacent metal ridges.

4. The wavefront sensor device based on nanogap SPPs manipulation according to claim 3, wherein, The plurality of metal gaps have the same or different sizes.

5. The wavefront sensor device based on nanogap SPPs manipulation according to claim 1, wherein, The low-dimensional material layer is a patterned graphene film.

6. A method of fabricating a wavefront sensor device based on nanogap SPPs manipulation according to any one of claims 1 to 5, characterized in that, The preparation method comprises the following steps: Based on electron beam direct writing and electron beam evaporation process, a first metal electrode and a second metal electrode are prepared on a substrate to form the plurality of metal electrode pairs; A low-dimensional material film is grown by using chemical vapor deposition method, and the grown low-dimensional material film is transferred to the metal electrode pairs by taking PMMA as a support layer; The low-dimensional material film is patterned by using a micro-nano lithography process, and the excess low-dimensional material is removed by reactive ion etching to form the low-dimensional material layer; A light-transmitting material is deposited as the isolation layer by using an atomic layer deposition device; The metal nanogap structure is prepared on the isolation layer.

7. The production method according to claim 6, characterized by, The way of preparing the first metal electrode or the second metal electrode on the substrate comprises the following steps: The pattern of the first metal electrode or the second metal electrode is exposed, and then 2-5 nm of chromium layer is deposited, followed by deposition of metal material with a thickness of 40-60 nm by combining with the electron beam evaporation process; the first metal electrode or the second metal electrode is obtained by soaking in acetone for 5-20 min and removing the excess metal.

8. The preparation method according to claim 6, characterized in that, The way of preparing the metal nanogap structure on the isolation layer comprises the following steps: Based on an electron beam evaporation device, a metal with a thickness of 80-150 nm is evaporated, an electron beam resist is spin-coated, a pattern is exposed by using an electron beam direct writing system, the width of the nanogap and the alignment accuracy are characterized by using a scanning electron microscope, and etching is performed to obtain the metal nanogap structure by taking the electron beam resist as a mask; wherein the metal is Au, Ag or Al.

9. A wavefront sensing method, characterized by, The wavefront sensing device based on nanogap SPPs manipulation according to any one of claims 1 to 5, the wavefront sensing method comprises: equivalent to the distribution of the incident angle in space, and a function relationship among the incident angle f(θ), the output current I and the angle response matrix H is established according to a prediction equation I=Hf(θ); based on the function relationship, a deep neural network is used as a regression prediction model, the angle response matrix H is known, the output current I is taken as input, and the incident angle is taken as a training target; in the training, a loss function is introduced to measure the error between the predicted angle and the real angle, and the network parameters are optimized by using back propagation and an optimizer so that the loss function converges to close to zero; based on the trained regression prediction model, the output currents of different metal electrode pairs are taken as input to determine the corresponding incident angles, and the wavefront information is reconstructed according to the corresponding incident angles.

10. The wavefront sensing method of claim 9, wherein, When training the regression prediction model, the output current is encoded, the spatial distribution characteristics are extracted, and the extracted spatial distribution characteristics are mapped to the angle space by using a decoder network to output the predicted incident angle value. In this process, a loss function is introduced to measure the error between the predicted angle and the real angle, and a back propagation and an optimizer are used to optimize the network parameters, so that the loss function converges to close to zero.

Images