Eagle eye bionic dynamic visual sensor and preparation method and application thereof

Abstract

The invention belongs to the related technical field of neuromorphic visual sensors, and discloses an eagle-eye bionic dynamic visual sensor and a preparation method and application thereof, the visual sensor comprises a dielectric layer substrate, a functional layer and a cross array structure composed of a bottom electrode, an insulating layer and a top electrode which are arranged from bottom to top; wherein the functional layer is of a core-shell heterojunction structure formed by noble metal nanoparticles and a two-dimensional metal chalcogenide. According to the obtained core-shell heterojunction structure, an electron capture and slow release phenomenon caused by local surface plasmon resonance effect excitation and interface defects of precious metal nanoparticles is constructed on a device, and the sensor can realize an eagle-eye-imitating dynamic visual function; and the visual sensor can perform eagle-eye-imitating dynamic visual processing on image information due to a unique electronic synapse function while realizing an image optical sensing function, so that integration of sensing and preprocessing functions is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field related to neuromorphic vision sensors, and more specifically, relates to an eagle-eye bionic dynamic vision sensor and a preparation method and application thereof. Background Art

[0002] In the era of technological revolution characterized by intelligence and informatization, the application of motion detection and recognition (MDR) technology in the fields of autonomous driving, security monitoring, smart transportation, military defense, industrial detection, etc. is becoming increasingly important. In these application scenarios, the realization of motion detection and recognition technology depends on efficient image sensing technology.

[0003] However, image sensors often produce ghosting and blurring when capturing high-speed moving objects, which affects the image quality and the subsequent processing effect. In video processing, image quality in dynamic scenes is also a key issue. A small amount of noise and ghosting will significantly affect the clarity of the video and the effect of subsequent processing.

[0004] In addition, the current common motion detectors are mainly based on image sensing platforms built on complementary metal oxide semiconductor (CMOS) image sensors and supporting computing hardware modules and corresponding algorithms. Converting large amounts of data from widely distributed image sensors into digital form for data transmission and storage also leads to a waste of time and electricity.

[0005] Motion detection in complex scenes is simple and easy in the field of biological vision. At present, some image sensors inspired by human, bird and insect vision have appeared. However, most of them only imitate the visual optical structure, and there is still little bionics of biological visual processing characteristics. At the same time, past work has not yet well integrated broadband spectrum detection and bionic image processing functions into a single device to achieve high-precision motion detection and recognition (MDR) after image noise reduction preprocessing. Summary of the invention

[0006] In view of the above defects or improvement needs of the prior art, the present invention provides an Eagle Eye bionic dynamic vision sensor and a preparation method and application thereof, which aims to solve the problem that the prior art has not well integrated broadband spectrum detection and bionic image processing functions to achieve high-precision MDR after image noise reduction preprocessing.

[0007] To achieve the above-mentioned purpose, according to one aspect of the present invention, there is provided an Eagle Eye bionic dynamic vision sensor, which comprises a dielectric layer substrate, a functional layer and a cross array structure composed of a bottom electrode, an insulating layer and a top electrode arranged from bottom to top; wherein, the functional layer is a core-shell heterojunction structure formed by precious metal nanoparticles and two-dimensional metal chalcogenide compounds.

[0008] Furthermore, the material of the noble metal nanoparticles is Au, Ag or Pt.

[0009] Furthermore, the particle size of the noble metal nanoparticles is 5nm-50nm.

[0010] Furthermore, the two-dimensional metal chalcogenide compound is a two-dimensional metal sulfide compound, a two-dimensional metal selenium compound or a two-dimensional metal tellurium compound or a perovskite material.

[0011] Furthermore, the noble metal nanoparticles are Au nanospheres, and the two-dimensional metal chalcogenide is MoS 2 .

[0012] Furthermore, the material of the dielectric layer substrate is SiO 2 , HfO 2 Or Al 2 O 3 .

[0013] The present invention also provides a method for preparing an Eagle Eye bionic dynamic vision sensor, the method for preparing the Eagle Eye bionic dynamic vision sensor as described above comprises the following steps:

[0014] (1) Depositing a dielectric layer substrate on a silicon wafer;

[0015] (2) evaporating a noble metal nanofilm on the surface of the dielectric layer substrate and annealing to form nanoparticles;

[0016] (3) growing two-dimensional metal chalcogenide functional materials on the surface of nanoparticles by chemical vapor deposition to form a core-shell heterojunction structure;

[0017] (4) thermally evaporating a bottom electrode on the functional material layer;

[0018] (5) depositing an insulating layer at the intersection of the bottom electrode and the top electrode;

[0019] (6) Thermally evaporate a top electrode on the insulating layer to obtain an Eagle Eye bionic dynamic vision sensor.

[0020] Furthermore, the annealing temperature is 400° C.-700° C., and the annealing time is 5 minutes-15 minutes.

[0021] Further, in step (3), the Au nanoparticles obtained by annealing are used as seeds to induce MoS 2 Growth and acquisition of Au@MoS 2 Heterojunction thin film, MoO can be used as the molybdenum source in CVD 2The sulfur source is S powder, and the CVD reaction uses a single-zone or multi-zone tubular furnace. The temperature of the molybdenum source is set to 630℃-750℃, the temperature of the sulfur source is set to 120℃-300℃, the heating rate of the tubular furnace is set to 15℃ / min-25℃ / min, and the reaction time is set to 4min-10min. A uniform and continuous array of Au@MoS can be obtained. 2 Heterojunction structure.

[0022] The present invention also provides an application of the Hawkeye bionic dynamic vision sensor as described above in autonomous driving, security monitoring, smart transportation, military defense, and industrial detection.

[0023] In general, compared with the prior art, the Hawkeye bionic dynamic vision sensor and its preparation method and application provided by the present invention have the following beneficial effects:

[0024] 1. Two-dimensional materials are coated on the surface of noble metal nanoparticles to form a core-shell heterojunction structure. The obtained core-shell heterojunction structure constructs the electron capture and sustained release phenomenon caused by the local surface plasmon resonance effect of noble metal nanoparticles and interface defects on the device. The sensor can realize the eagle-eye dynamic vision function, and the visual sensor can realize the image optical sensing function while performing eagle-eye dynamic vision processing on the image information due to the unique electronic synapse function, realizing the integration of sensing and preprocessing functions.

[0025] 2. The visual sensor provided by the present invention simulates the biological visual system and adopts the local grating effect, which significantly reduces the impact of noise on image quality and improves the clarity and stability of video images; through hardware-level dynamic tracking processing, the visual sensor can maintain high-quality imaging effects in various complex environments.

[0026] 3. Compared with the traditional solution based on CMOS image sensor and supporting computing hardware, the present invention reduces the demand for a large amount of data transmission and storage by integrating bionic image processing function, reduces the power consumption and hardware complexity of the system, and is suitable for power-constrained application scenarios; especially in IoT devices and mobile devices, this low power consumption and high efficiency design has significant advantages, prolongs the service life of the equipment, and reduces maintenance costs.

[0027] 4. The present invention adopts a core-shell heterojunction structure of precious metals and two-dimensional materials, combined with artificial electronic synapse technology to achieve accurate tracking of dynamic targets, significantly improving the motion detection and recognition performance of visual sensors in complex environments; traditional image sensing systems often find it difficult to maintain high precision when processing high-speed moving targets, while the present invention can quickly adapt to changing scenes by simulating the dynamic response mechanism of biological visual systems, ensuring the continuity and accuracy of target tracking. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a preparation flow chart of an eagle-eye bionic dynamic vision sensor provided by the present invention;

[0029] Figure 2 Schematic diagram of optical characterization of Au@MoS2 array in an embodiment of the present invention;

[0030] Figure 3 Schematic diagram of the structure explosion of the Hawkeye bionic dynamic vision sensor in an embodiment of the present invention;

[0031] Figure 4 is a schematic diagram of optical representation of the Hawkeye bionic dynamic vision sensor in an embodiment of the present invention;

[0032] Figure 5 Schematic diagram of integrated packaging of the Hawkeye bionic dynamic vision sensor in an embodiment of the present invention;

[0033] Figure 6 This is a graph showing different photocurrent changes of the Hawkeye bionic dynamic vision sensor under four-bit light pulse stimulation in an embodiment of the present invention;

[0034] Figure 7 is a schematic diagram of a boat moving at high speed in an embodiment of the present invention;

[0035] Figure 8 is a schematic diagram of a boat after being processed by a conventional image sensor in the present invention;

[0036] Fig. 9 It is a schematic diagram of a boat after being processed by the Hawkeye bionic dynamic vision sensor in the present invention. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0038] The present invention provides an eagle eye bionic dynamic vision sensor, the improvement of which is inspired by the special oil droplet structure in the retinal cones of birds such as avian species. Oil droplets are an important component of bird cones, and their main components are carotenoids and neutral fats. Different types of carotenoids give oil droplets different colors by absorbing specific wavelengths in the spectrum. The oil droplets are ellipsoidal droplet structures, usually with a refractive index of more than 1.5. Relative to the surrounding buffer environment with a refractive index of 1.1-1.3, light enters the oil droplets and refracts and converges. Compared with the structure without oil droplets, the scattering of light from the top of the cone to the photosensitive end is effectively reduced, and the amount of light entering is increased, so that the cone can capture more photons per unit time.

[0039] See also Figure 2 and Figure 3 The visual sensor includes a dielectric layer substrate, a functional layer and a cross array structure composed of a bottom electrode, an insulating layer and a top electrode arranged from bottom to top; wherein the functional layer is a core-shell heterojunction structure formed by precious metal nanoparticles and two-dimensional metal chalcogenide compounds.

[0040] The insulating layer isolates the vertical contacts between the source and drain. In one embodiment, the noble metal nanoparticles are Au nanospheres, and the two-dimensional metal chalcogenide is MoS 2 The enhanced light absorption caused by the localized plasmon resonance (LSPR) effect provided by Au nanoparticles and the interaction between Au nanospheres and MoS 2 The electron capture and slow release phenomenon caused by the interface defects of Au@MoS 2 It exhibits a unique local grating effect and displays artificial synaptic characteristics of rapid carrier relaxation after illumination, effectively mimicking the oil droplet structure in bird cone cells, and is the key layer for achieving eagle-eye-like dynamic visual functions.

[0041] The material of the dielectric layer substrate is SiO 2 , HfO 2 Or Al 2 O 3 ; The material of the noble metal nanoparticles is Au, Ag or Pt; first, metal nanoparticles are deposited and annealed on the surface of the dielectric layer substrate to form metal nanoparticles; the particle size of the noble metal nanoparticles is 5nm-50nm; the two-dimensional metal sulfide compound in the functional layer is a two-dimensional metal sulfide compound, a two-dimensional metal selenium compound or a two-dimensional metal tellurium compound or a perovskite material.

[0042] The thickness of the precious metal deposited on the surface of the dielectric layer substrate is 1nm-5nm, the annealing temperature condition is 400℃-700℃, and the annealing time is 5 minutes-15 minutes; the source and drain materials of the visual sensor are Bi, Cr, and Au.

[0043] The present invention also provides a method for preparing the Hawkeye bionic dynamic vision sensor as described above. In a specific embodiment, the preparation method includes the following steps: depositing a dielectric layer substrate on a silicon wafer; evaporating and depositing a noble metal nanofilm array on the surface of the dielectric layer substrate and annealing to form nanometal particles; growing functional materials on the surface of the nanometal array by chemical vapor deposition, wherein the functional materials wrap the nanometal particles to form a core-shell structure; thermally evaporating a first layer of multiple fixed-interval, parallel-arranged linear electrodes on the functional material layer, and the electrode leads to a small electrode connected to one side of the array; depositing an insulating layer material at the intersection of the first layer of electrodes and the second layer of electrodes; thermally evaporating a second layer of multiple fixed-interval, parallel-arranged linear electrodes on the first layer of electrodes and the insulating layer, and the extension direction thereof is perpendicular to the extension direction of the first layer of linear electrodes, and the electrode leads to a small electrode connected to the other side of the array to form a horizontal device cross array structure. By changing the illumination conditions, the electrical level of the device changes accordingly, and a multi-state response function is obtained due to the electronic synaptic characteristics. The materials of the first and second layers of linear electrodes are Bi, Cr, and Au; the material of the insulating layer of the cross-contacts of the first and second layers of linear electrodes is SiO 2 .

[0044] The present invention also provides an application of the Hawkeye bionic dynamic vision sensor as described above in autonomous driving, security monitoring, smart transportation, military defense, and industrial detection.

[0045] The present invention is further described in detail below with reference to specific embodiments.

[0046] Example 1

[0047] See also Figure 1 The method for preparing the Hawkeye bionic dynamic vision sensor with LSPR effect enhanced light absorption according to the present invention comprises the following steps:

[0048] Step S1, prepare SiO with a thickness of 300 nm 2 / Si substrate is used to prepare dynamic vision sensor. 2 It can also be SiN x , glass, etc.

[0049] Step S2, depositing a planar array Au film with a thickness of 3nm to 5nm on the surface of the substrate by thermal evaporation, completing rapid annealing in a vacuum rapid annealing furnace, setting the annealing temperature range to 500℃ to 600℃, and the annealing time to 5min-8min, and finally obtaining uniformly distributed planar array Au nanoparticles with a particle size range of 30nm to 60nm.

[0050] Step S3, using chemical vapor deposition (CVD) to use the Au nanoparticles annealed in S2 as seeds to induce MoS 2 Growth and acquisition of Au@MoS 2 Heterojunction thin film, MoO can be used as the molybdenum source in CVD 2 、MoO 3 The sulfur source can be S powder, H 2 S gas, etc., CVD reaction uses a single temperature zone or multi-temperature zone tube furnace, sets the temperature of the molybdenum source zone to 630℃-750℃, sets the temperature of the sulfur source zone to 120℃-300℃, sets the heating rate of the tube furnace to 15℃ / min-25℃ / min, sets the reaction time to 4min-10min, and obtains a uniform and continuous array of Au@MoS 2 Heterojunction (such as Figure 2 (shown)

[0051] Step S4, see Figure 3 The device structure uses the substrate in step S1 as the full back gate, and performs the spin-on-bake-on-photolithography-development-etching steps on the sample based on step S3 to complete the bottom electrode exposure of the cross array electrode on the sample surface, and then deposits the Bi / Au electrode as the source electrode by thermal evaporation. The bottom electrode material can also be metals such as Cr / Au, In / Au, and Ti / Au; the bottom electrode deposition thickness range can be 5nm-15nm / 40nm-100nm.

[0052] Step S5, complete the exposure of the insulating layer of the sample cross array electrode by the same method as in step S4, and use plasma enhanced chemical vapor deposition (PECVD) to grow an inorganic insulating layer at low temperature or the photolithography process in step S3 to obtain an organic insulating layer. The insulating layer material can be SiO 2 、SiN x Inorganic porous materials such as PET and organic porous materials such as polyimide (PI). The thickness of the insulating layer material can be selected in the range of 100nm-400nm.

[0053] Step S6, completing the exposure of the top electrode of the sample cross array electrode by the same method as step S4, the top electrode deposition material is consistent with the description in step S4, and the electrode deposition thickness can be selected in the range of 150nm-250nm.

[0054] Step S7, complete the peripheral electrode exposure and electrode material deposition of the sample cross array electrode by the same method as step S4. The peripheral electrode material can be selected from metal materials such as Au, Cr / Au, Ti / Au, etc., and the electrode deposition thickness can be selected in the range of 300nm-600nm. After completing steps S1-S7, the sample realizes the device process, which can be seen in Figure 4 The dynamic vision sensor is then packaged in a quad flat package (QPF) ( Figure 5 ), and finally completed the preparation of the Hawkeye bionic dynamic vision sensor.

[0055] In order to verify the performance of the Hawkeye bionic dynamic vision sensor, the following tests were performed:

[0056] 1. Based on Au@MoS 2 Light pulse response test of Hawkeye bionic dynamic vision sensor

[0057] The light response characteristics of the dynamic vision sensor obtained in Example 1 were tested in air using a B1500A semiconductor analyzer (Agilent, USA). The laser used for the test was 532nm visible light with a spot diameter of 2.5mm, which is much larger than the channel length and width of a single device in a cross array.

[0058] This embodiment tests the light response of the visual sensor under the condition of continuous light pulses. Figure 6 The illustration shows a typical example of the "1001" encoding process and the gradual changes at four time points. The numbers "1" and "0" represent the optical power of 10mW / cm 2 light and dark states. Figure 6 The photocurrent under four light pulses (four-bit signal, a total of 16 combinations) is shown in the figure. The pulse width of the light pulse is 500ms and the interval is also 500ms, showing the different responses of the sensor to the four-frame signal in different encoding processes. The current level depends on the order of light stimulation. When the interval between the "1" light pulses is short, the pulses are superimposed in time, resulting in a higher current level output; when there is a "0" dark interval between the "1" light pulses, the current level is higher due to the Au@MoS 2 The local electric field at the heterojunction interface disappears, the trapped electrons are quickly released, and the current level output is low. This different response shows that the sensor can temporally encode the four pulse light stimuli like neurons.

[0059] 2. Simulation of dynamic visual characteristics of Hawkeye bionic dynamic vision sensor

[0060] Conventional image sensors often produce ghosting when shooting high-speed moving objects, which seriously affects the clarity and accuracy of the image. In order to verify the advantages of Hawkeye bionic dynamic vision sensor in dynamic vision processing, a simulation experiment of a high-speed moving boat was carried out.

[0061] Different response weights were introduced in the experiment, and the imaging capabilities of conventional image sensors and Hawkeye bionic dynamic vision sensors for high-speed moving boats were simulated by scanning the window frame by frame. The specific steps are as follows:

[0062] A1. Simulation modeling of high-speed boats: Figure 7FIG. 1 is a schematic diagram showing a boat moving at high speed. The diagram is used to illustrate the movement trajectory of the boat and the difficulty of image capture in a high-dynamic scene.

[0063] A2. Conventional image sensor processing: Figure 8 As shown in the figure, a conventional image sensor is used to simulate the sensing of a high-speed boat. Due to the lack of effective processing of dynamic targets, obvious ghosting occurs in the image. This is because conventional sensors are not sufficiently responsive when processing fast-moving objects, and require more exposure time, so they cannot respond in time, resulting in blurred images.

[0064] A3, Hawkeye bionic dynamic vision sensor processing: Fig. 9 As shown in the figure, the same sensing simulation is performed using the Hawkeye bionic dynamic vision sensor. By simulating the visual processing mechanism of the Hawkeye, the sensor can dynamically adjust the response weight of each pixel and adapt to high-speed motion scenes in real time. Its unique core-shell heterojunction structure and local grating effect give it significant advantages in photoelectric response speed and noise suppression.

[0065] A4. Simulation processing of dynamic vision: The Hawkeye sensor performs dynamic weighted processing on each pixel by traversing the window frame by frame and scanning step by step. The method of dynamic weighted processing in this method is the same as that in the above-mentioned related tests.

[0066] Through the above simulation experiments, the results show that the image clarity and stability of the Hawkeye bionic dynamic vision sensor are significantly better than those of conventional sensors when processing high-speed moving objects. This advantage is mainly due to the special structure and response characteristics of the Hawkeye bionic dynamic vision sensor, which enables the Hawkeye bionic dynamic vision sensor to quickly adapt to scene changes when processing high-speed moving objects and provide high-quality dynamic images.

[0067] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An eagle-eye bionic dynamic vision sensor, characterized in that: The visual sensor includes a dielectric layer substrate, a functional layer and a cross array structure composed of a bottom electrode, an insulating layer and a top electrode arranged from bottom to top; wherein the functional layer is a core-shell heterojunction structure formed by precious metal nanoparticles and two-dimensional metal chalcogenide compounds.

2. The Hawkeye bionic dynamic vision sensor according to claim 1, characterized in that: The material of the noble metal nanoparticles is Au, Ag or Pt.

3. The Hawkeye bionic dynamic vision sensor according to claim 2, characterized in that: The particle size of the noble metal nanoparticles is 5nm-50nm.

4. The Hawkeye bionic dynamic vision sensor according to claim 2, characterized in that: The two-dimensional metal chalcogenide is a two-dimensional metal sulfide compound, a two-dimensional metal selenium compound or a two-dimensional metal tellurium compound or a perovskite material.

5. The Hawkeye bionic dynamic vision sensor according to claim 4, characterized in that: The noble metal nanoparticles are Au nanospheres, and the two-dimensional metal chalcogenide compound is MoS2.

6. The Hawkeye bionic dynamic vision sensor according to any one of claims 1 to 5, characterized in that: The material of the dielectric layer substrate is SiO2, HfO2 or Al2O3.

7. A method for preparing an eagle-eye bionic dynamic vision sensor, characterized in that: The preparation method is used to prepare the Hawkeye bionic dynamic vision sensor according to any one of claims 1 to 6, and comprises the following steps: (1) Depositing a dielectric layer substrate on a silicon wafer; (2) evaporating a noble metal nanofilm on the surface of the dielectric layer substrate and annealing to form nanoparticles; (3) growing two-dimensional metal chalcogenide functional materials on the surface of nanoparticles by chemical vapor deposition to form a core-shell heterojunction structure; (4) thermally evaporating a bottom electrode on the functional material layer; (5) depositing an insulating layer at the intersection of the bottom electrode and the top electrode; (6) Thermally evaporate a top electrode on the insulating layer to obtain an Eagle Eye bionic dynamic vision sensor.

8. The method for preparing the Hawkeye bionic dynamic vision sensor according to claim 7, characterized in that: The annealing temperature is 400° C.-700° C., and the annealing time is 5 minutes-15 minutes.

9. The method for preparing the Hawkeye bionic dynamic vision sensor according to claim 7, characterized in that: In step (3), the Au nanoparticles obtained by annealing are used as seeds to induce the growth of MoS2 and obtain an Au@MoS2 heterojunction film through chemical vapor deposition. MoO2 can be used as the molybdenum source in CVD, and S powder can be used as the sulfur source. The CVD reaction uses a single-zone or multi-zone tubular furnace. The temperature of the temperature zone where the molybdenum source is located is set to 630°C-750°C, the temperature of the temperature zone where the sulfur source is located is set to 120°C-300°C, the heating rate of the tubular furnace is set to 15°C / min-25°C / min, and the reaction time is set to 4min-10min, so as to obtain a uniform and continuous array Au@MoS2 heterojunction structure.

10. An application of the Hawkeye bionic dynamic vision sensor according to any one of claims 1 to 6 in autonomous driving, security monitoring, smart transportation, military defense, and industrial detection.

Images