Photon synaptic device applied to image recognition and preparation method thereof

Abstract

The invention discloses a photon synaptic device applied to image recognition and a preparation method of the photon synaptic device. The device sequentially comprises a top electrode, a MoS2 functional layer, a CsPbBr3 quantum dot functional layer and a bottom electrode from top to bottom. The specific preparation method comprises the following steps: (1) providing a MoS2 solution and a CsPbBr3 quantum dot solution; (2) transferring a MoS2 solution to bottom electrode ITO glass, and annealing in a vacuum drying oven at 130 DEG C for 20 minutes to obtain a MoS2 functional layer; (3) depositing a CsPbBr3 quantum dot solution on the surface of the MoS2 functional layer, and annealing in the vacuum drying oven at 170 DEG C for 20 minutes to form a CsPbBr3 quantum dot functional layer; and (4) covering the surface of the CsPbBr3 quantum dot functional layer with a piece of ITO conductive glass as a top electrode to prepare the photon synapse device. The photon synaptic device disclosed by the invention can simulate short-term plasticity (STP) and long-term plasticity (LTP) of neurons, can be applied to image recognition, and effectively improves the recognition precision.

Description

Technical Field

[0001] The present invention relates to the field of recognition, and in particular to the application of photonic synaptic devices in image recognition. Background Art

[0002] Image recognition technology plays a vital role in today's autonomous driving, security monitoring, medical imaging and other fields. With the increase in application demand, how to improve the accuracy and speed of image recognition and the ability to process complex image scenes has become the focus of the industry.

[0003] Traditional image recognition systems mainly rely on digital computing architectures to achieve target recognition by processing and analyzing large amounts of image data. However, when dealing with complex image scenes, such systems are often limited by computing speed and power consumption, making it difficult to meet the needs of real-time processing and high-precision recognition. Therefore, it is of great significance to develop new devices and technologies to better support the efficient implementation of image recognition.

[0004] In this context, photonic synaptic devices have gradually attracted the attention of researchers due to their unique advantage of being able to simulate the behavior of biological neurons. This device simulates the behavior of neural synapses through the interaction between light and materials, and can significantly enhance the image processing effect when used in image recognition. Compared with traditional electronic devices, photonic synaptic devices show significant advantages in low power consumption, high response speed and optical signal processing.

[0005] At present, although memristors and field-effect transistors are considered to be ideal electronic devices for achieving synaptic plasticity, they still have certain defects: the synaptic properties of memristors need to be further improved, and the devices themselves have inevitable noise; while the manufacturing process of field-effect transistors is complex and costly, which limits the feasibility of their large-scale production. Therefore, existing synaptic devices still need to be improved and developed. Summary of the invention

[0006] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a photonic synapse device for image recognition and a method for preparing the same, so as to solve the problems of poor synaptic characteristics and stability and high preparation cost of existing synapse devices, and to propose the application of photonic synapse devices in image recognition.

[0007] The technical solution of the present invention is as follows:

[0008] A photonic synaptic device for image recognition, wherein the device is a vertical structure. From top to bottom, the top electrode, MoS 2 Functional layer, CsPbBr 3 Quantum dot functional layer, bottom electrode.

[0009] In the photonic synapse device, the top electrode and the bottom electrode are both made of ITO glass.

[0010] A method for preparing a photonic synaptic device for image recognition comprises the following steps:

[0011] (1) Provide MoS 2 Solution and CsPbBr 3 Quantum dot solution;

[0012] (2) Transferring MoS 2 The solution was poured onto the bottom electrode ITO glass and annealed at 130 °C in a vacuum oven for 20 min to obtain MoS 2 Functional layer;

[0013] (3) In the MoS 2 CsPbBr is deposited on the functional layer 3 The quantum dot solution was annealed again in a vacuum oven at 170 °C for 20 min to form CsPbBr 3 Quantum dot functional layer;

[0014] (4) In the CsPbBr 3 The surface of the quantum dot functional layer is covered with a piece of ITO conductive glass as the top electrode to produce a photonic synapse device.

[0015] Furthermore, in step (1), the MoS 2 The solution preparation process includes the steps:

[0016] (1) MoS 2 The microcrystals are dispersed in 100 ml of a mixture of ethanol and water, wherein the ratio of ethanol to water in the mixture is 9:11.

[0017] (2) The dispersion was ultrasonically treated using a probe with a diameter of 6 mm, wherein the probe ultrasonic power was 150 W and the treatment time was 30 minutes.

[0018] (3) The treated dispersion is subjected to water bath ultrasonic treatment, wherein the ultrasonic frequency is 40 kHz, the power is 100 W, and the treatment time is 3 hours.

[0019] (4) The resulting solution was centrifuged at 10,000 rpm for 15 minutes and then the solution was collected.

[0020] (5) The centrifuged solution was filtered twice through microporous membranes with different pore sizes to obtain uniform MoS with a diameter of 1 to 2 μm. 2The first filtration of the two filtration operations uses a microporous filter membrane with a pore size of 2 microns, and then the solution filtered for the first time is filtered for the second time using a microporous filter membrane with a pore size of 1 micron.

[0021] The photonic synaptic device disclosed in the present invention utilizes its light response characteristics to simulate the short-term plasticity (STP) and long-term plasticity (LTP) of neurons by changing the light intensity and wavelength, thereby realizing learning and memory functions. It can be applied to image recognition and effectively improve recognition accuracy.

[0022] Beneficial effects:

[0023] First, the present invention discloses a photonic synaptic device for image recognition using MoS 2 and CsPbBr 3 The quantum dot layer is used as a functional layer, combined with the quantum dot material CsPbBr 3 The unique optical properties such as narrow exciton binding energy, excellent absorption characteristics, high stability, high charge carrier mobility and long carrier lifetime make it have great application potential.

[0024] Second, the present invention discloses a photonic synaptic device for image recognition, wherein the MoS 2 Solution preparation method, preparation of the obtained MoS 2 MoS in solution 2 The particle size is more uniform, and MoS is obtained by annealing in a vacuum oven. 2 The functional layer is smoother, which can improve the electron mobility, photoelectric performance, stability and interface quality of the photonic synapse device, thereby significantly improving the overall performance of the photonic synapse device.

[0025] Third, the present invention discloses a photonic synaptic device for image recognition, which utilizes the responsiveness to illumination of different wavelengths and different light intensities to realize synaptic plasticity functions, including short-term plasticity (STP), long-term plasticity (LTP), etc.

[0026] Fourth, the present invention converts CsPbBr 3 Quantum dot layer and MoS 2 The functional layers are stacked layer by layer, and the photocurrent value in the electronic device is changed by simply adjusting the wavelength and intensity of light, thereby simulating the enhancement of postsynaptic current in the human brain. Combining reverse electric field stimulation can achieve the weakening of postsynaptic current; at the same time, by adjusting the number and width of pulse light stimulation, the process of transforming short-term memory to long-term memory in the human brain is realized. Therefore, the photonic synaptic device based on CsPbBr3 quantum dots of the present invention has the advantages of easy regulation and low power consumption, and can be applied to image recognition to effectively improve recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a schematic structural diagram of a photonic synaptic device for image recognition disclosed in the present invention;

[0028] Figure 2 It is a flow chart of a method for preparing a photonic synaptic device for image recognition disclosed in the present invention;

[0029] Figure 3 It is the recognition result of the image recognition system based on the photonic synaptic device disclosed in the present invention. DETAILED DESCRIPTION

[0030] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments. The embodiments are implemented based on the technical solution of the present invention. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention.

[0031] Example 1

[0032] See also Figure 1 FIG. 1 is a schematic diagram of a preferred embodiment of a photonic synaptic device of the present invention, wherein the photonic synaptic device comprises an ITO glass as a bottom electrode, and MoS 2 Functional layer, CsPbBr 3 Quantum dot functional layer and top electrode layer ITO glass.

[0033] The present invention provides a photon synaptic device, which combines the quantum dot material CsPbBr 3 Its unique optical properties, such as narrow exciton binding energy, excellent absorption characteristics, high stability, high charge carrier mobility and long carrier lifetime, make it have great application potential. The responsiveness to different wavelengths and different light intensities can be used to achieve synaptic plasticity functions, including short-term plasticity (STP) and long-term plasticity (LTP).

[0034] Specifically, the present invention uses the top electrode and the bottom electrode to simulate the pre-neuron and the post-neuron respectively, and is characterized in that in the process of charge capture, light stimulation and electrical inhibition can be used to simulate the inhibitory activity in biological excitation. By applying different intensities and wavelengths, the conductivity of the functional layer can be changed, thus forming a device structure with adjustable memristive characteristics, similar to biological synapses. In the present invention, presynaptic stimulation can be easily applied and the postsynaptic current intensity can be detected.

[0035] Similar to biological synapses, the postsynaptic current of the photonic synaptic device of the present invention can be precisely adjusted by adjusting the light intensity and wavelength. Applying a light pulse to the device will induce a current, which is very similar to the postsynaptic current in biological synapses. The applied light can be regarded as a presynaptic pulse, and the conductivity of the functional layer can be regarded as the synaptic weight, thereby simulating synaptic functions such as short-term plasticity and long-term plasticity.

[0036] The present invention converts CsPbBr 3 Quantum dot layer and MoS 2 The functional layers are stacked layer by layer, and the photocurrent value in the electronic device is changed by simply adjusting the wavelength and intensity of light, thereby simulating the enhancement of postsynaptic current in the human brain. Combining reverse electric field stimulation can achieve the weakening of postsynaptic current; at the same time, by adjusting the number and width of pulse light stimulation, the transformation process from short-term memory to long-term memory in the human brain can be achieved. Therefore, the present invention is based on CsPbBr 3 Quantum dot photonic synaptic devices have the advantages of easy control and low power consumption. They can be applied to image recognition and effectively improve recognition accuracy.

[0037] Based on the above-mentioned photon synapse device, the present invention also provides a method for preparing a photon synapse device for image recognition, such as Figure 2 As shown, including:

[0038] Step 1: Provide MoS 2 Solution and CsPbBr 3 Quantum dot solution;

[0039] Step 2: Transferring MoS 2 The solution was poured onto the bottom electrode ITO glass and annealed at 130 °C in a vacuum oven for 20 min to obtain MoS 2 Functional layer;

[0040] Step 3: In the MoS 2 CsPbBr is deposited on the functional layer 3 The quantum dot solution was annealed again in a vacuum oven at 170 °C for 20 min to form CsPbBr 3 Quantum dot functional layer;

[0041] Step 4: In the CsPbBr 3 The surface of the quantum dot functional layer is covered with a piece of ITO conductive glass as the top electrode to produce a photonic synapse device.

[0042] Furthermore, in step 1, the MoS 2 The solution was prepared as follows: MoS 2The microcrystals were dispersed in a mixture containing 45 ml of ethanol and 55 ml of water, and the dispersion was ultrasonically treated with a probe of 6 mm in diameter at 150 W for 30 minutes. The treated dispersion was subjected to ultrasonic water bath treatment at an ultrasonic frequency of 40 kHz, a power of 100 W, and a treatment time of 3 hours. The resulting solution was centrifuged at 10,000 rpm for 15 minutes, and the centrifuged solution was filtered twice through microporous membranes of different pore sizes, the first filtration using a microporous membrane with a pore size of 2 microns, and then the above-mentioned first filtered solution was filtered for the second time using a microporous membrane with a pore size of 1 micron to obtain a uniform MoS with a diameter of 1 to 2 microns. 2 Solution.

[0043] Using the MoS proposed in the present invention 2 Solution preparation method, preparing MoS 2 The solution particles are uniform in size, and the flat MoS can be obtained after annealing in a vacuum oven. 2 The functional layer helps to improve the electron mobility, optoelectronic performance, stability and interface quality of the photonic synaptic device, thereby significantly improving the overall performance of the photonic synaptic device.

[0044] Furthermore, the preparation method of the quantum dot solution in step (1) is as follows: cesium carbonate (Cs 2 CO 3 ) and octadecene (C 18 H 36 ) and oleic acid (C 18 H 34 O 2 ) were mixed into a three-necked flask and dried at 120°C for one hour. 2 ) atmosphere to 150°C to generate cesium oleate (Cs(C 18 H 34 O 2 )) solution; lead bromide (PbBr 2 ) and octadecene (C 18 H 36 ), oleylamine (C 18 H 35 NH 2 ), oleic acid (C 18 H 34 O 2 ) were mixed and added into a three-necked flask, heated at 110°C under vacuum for 40 minutes, heated to 170°C, and finally cesium oleate (Cs(C 18 H 34 O 2 )) solution, and the solution was quickly cooled to room temperature in an ice bath. The quantum dots were dissolved and dispersed in anhydrous n-hexane (C 6 H 14), centrifuge to remove impurities, and obtain a quantum dot solution.

[0045] The present invention uses the CsPbBr synthesized by hot injection method 3 Quantum dots are used in photonic synaptic devices and have the advantages of easy control and low power consumption. They can be applied to image recognition and effectively improve recognition accuracy.

[0046] Example 2

[0047] This embodiment demonstrates the use of the photonic synapse device prepared by the present invention to construct a threshold segmentation model based on a photonic synapse array for image recognition. The photonic synapse array improves the contrast of the image and reduces the background noise by preprocessing the input image, so that the recognition rate of the network reaches 91.22%. In addition, when the recognition accuracy of the artificial neural network based on the photonic synapse array is 80%, the number of training rounds required is only 16% of the original artificial neural network, which greatly improves the training speed. The activation rates of correct faces and incorrect faces reached 85.01% and 65.77% respectively, indicating that the system can successfully distinguish different input faces. The activation rate of correct faces with noise also reached 82.21%, showing good robustness. Figure 3 The activation effect of the photonic synaptic array on image recognition under different conditions is demonstrated. In the case of correct recognition, the activation rate reaches 85.01%; even in the case of noise, the activation rate still reaches 82.21%. In the case of incorrect recognition, the activation rate is significantly lower, only 65.77%. The image recognition test results show that the learned photonic synaptic array can better distinguish between the same face and different faces. We can use the photonic synaptic array to build a robust image recognition system without using an expensive computing framework. This application effectively improves the accuracy and efficiency of image recognition, indicating that photonic synaptic devices have important application prospects in image recognition systems.

[0048] The above is only a preferred embodiment of the present invention. It should be noted that for ordinary technicians in this technical field, other parts not specifically described belong to the prior art or common knowledge. Several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered as the protection scope of the present invention.

Claims

1. A photonic synaptic device for image recognition, characterized in that: From top to bottom, they are the top electrode, MoS2 functional layer, CsPbBr3 quantum dot functional layer, and bottom electrode.

2. The photonic synapse device according to claim 1, characterized in that: The top electrode and the bottom electrode are both indium tin oxide (ITO) glass.

3. A method for preparing a photonic synaptic device for image recognition, characterized in that: The preparation process includes: (1) providing a MoS2 solution and a CsPbBr3 quantum dot solution; (2) Transferring the MoS2 solution to the bottom electrode ITO glass and annealing it in a vacuum oven at 130°C for 20 minutes to obtain a MoS2 functional layer; (3) depositing a CsPbBr3 quantum dot solution on the surface of the MoS2 functional layer, and annealing again in a vacuum drying oven at 170° C. for 20 minutes to form a CsPbBr3 quantum dot functional layer; (4) A piece of ITO conductive glass is covered on the surface of the CsPbBr3 quantum dot functional layer as a top electrode to obtain a photonic synapse device.

4. The method for preparing a photonic synaptic device for image recognition according to claim 3, characterized in that: The preparation process of the MoS2 solution includes: (1) dispersing MoS2 microcrystals in a mixture of 100 ml of ethanol and water, wherein the ratio of ethanol to water in the mixture is 9:11; (2) subjecting the dispersion to ultrasonic treatment using a probe with a diameter of 6 mm, wherein the probe ultrasonic power is 150 W and the treatment time is 30 minutes; (3) subjecting the treated dispersion to water bath ultrasonic treatment, wherein the ultrasonic frequency is 40 kHz, the power is 100 W, and the treatment time is 3 hours; (4) centrifuging the resulting solution at 10,000 rpm for 15 minutes and then collecting the solution; (5) The centrifuged solution is filtered twice through microporous membranes with different pore sizes to obtain a uniform MoS2 solution with a diameter of 1 to 2 microns. The first filtration of the two filtration operations uses a microporous membrane with a pore size of 2 microns, and then the solution filtered for the first time is filtered for the second time using a microporous membrane with a pore size of 1 micron.

5. The photonic synapse device according to claim 1, characterized in that: The photonic synaptic device utilizes its light response characteristics to simulate the short-term plasticity (STP) and long-term plasticity (LTP) of neurons by changing the light intensity and wavelength, thereby achieving learning and memory functions.

6. The photonic synapse device according to claim 1, characterized in that: The photonic synapse device is applied to image recognition and can effectively improve recognition accuracy.

Images