Artificial protruding touch monitoring and identifying system

Abstract

The invention discloses an artificial synapse contact monitoring and identifying system, which is characterized in that when a first electrode, an insulating layer, a semiconductor layer and a second electrode are arranged in a laminated manner and a tunnel junction is formed by the insulating layer and the semiconductor layer, an artificial synapse of a capacitance coupling tunnel junction with a heterostructure is constructed, and charge transmission of a capacitance coupling effect and a quantum tunneling effect is combined, so that the performance of the synapse contact is improved. Polarization is generated in the insulating layer under the action of an external electric signal, the barrier height of the tunnel junction is adjusted, and the multi-mode synaptic response is realized; moreover, when the potential barrier height of the tunnel junction is adjusted through an external electric signal to realize the artificial synapse function, the generated power consumption is low, the response speed is high, and the stability is good, so that the signal transmission and adjustment functions of the biological synapse can be accurately simulated.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial neural networks, and particularly to an artificial synapse and monitoring and recognition system. Background Art

[0002] Neuromorphic computing systems can simulate the structure and function of neurobiological architectures, and can trigger revolutionary changes in the fields of data processing and artificial intelligence applications, especially in complex systems such as the Internet of Things, image recognition, real-time data analysis, fuzzy algorithms, adaptive control, and autonomous driving traffic. Traditional digital computing hardware relies on the von Neumann architecture, has independent memory and processing components, operates serially, and heavily depends on programming. This configuration inevitably requires continuous and frequent data exchange, resulting in problems such as high energy consumption and data transmission bottlenecks. Among them, neuromorphic computing hardware such as artificial synapses has powerful information processing and learning capabilities, and can achieve complex perception and cognitive functions with extremely low energy consumption.

[0003] Currently, artificial synapses are mainly fabricated based on memristors, phase change memories, and ferroelectric device materials. However, the resistance state switching stability of memristors is poor, resulting in difficult-to-guarantee accuracy and affecting the precision of signal processing; after multiple read and write operations, the performance of phase change materials is prone to decline, limiting their long-term reliability.

[0004] Based on this, how to obtain artificial synapses with fast response speed, low power consumption, and high stability, and accurately simulate the signal transmission and regulation functions of biological synapses has become an urgent problem in this field. Summary of the Invention

[0005] An embodiment of the present invention provides an artificial synapse and monitoring and recognition system for obtaining artificial synapses with fast response speed, low power consumption, and high stability, and accurately simulating the signal transmission and regulation functions of biological synapses.

[0006] In a first aspect, an embodiment of the present invention provides an artificial synapse, including: a first electrode, an insulating layer, a semiconductor layer, and a second electrode that are sequentially stacked; the thickness of the insulating layer is not greater than 5 nm;

[0007] The insulating layer and the semiconductor layer form a tunnel junction;

[0008] The first electrode and the second electrode are used for inputting an electrical signal;

[0009] The insulating layer is used for: generating dielectric polarization under the action of an electrical signal to adjust the barrier height of the tunnel junction.

[0010] In a second aspect, an embodiment of the present invention provides an artificial synapse, including: an electrode layer, an insulating layer, and a semiconductor layer arranged in a stacked manner, where the electrode layer includes a first electrode and a second electrode arranged at intervals and side by side; the first electrode and the second electrode are used for transmitting electrical signals;

[0011] The thickness of the insulating layer is not greater than 5 nm;

[0012] The insulating layer and the semiconductor layer form a tunnel junction;

[0013] The manufacturing material of the semiconductor layer has flexoelectricity or piezoelectricity;

[0014] The artificial synapse is flexible. When stress is applied to the artificial synapse, dielectric polarization occurs in the insulating layer to adjust the barrier height of the tunnel junction.

[0015] In a third aspect, an embodiment of the present invention provides a monitoring and recognition system, including: a processor and a synapse array. The synapse array includes a plurality of artificial synapses as described in the first aspect or the second aspect above. The processor is connected to the first electrode and / or the second electrode of each artificial synapse;

[0016] The processor is configured to: acquire current signals output by at least some of the artificial synapses through the first electrode and / or the second electrode; and determine the distribution information of the artificial synapses in the synapse array that are subjected to external stimuli according to the acquired current signals.

[0017] The beneficial effects of the present invention are as follows:

[0018] An artificial synapse and a monitoring and recognition system provided by an embodiment of the present invention, when the first electrode, the insulating layer, the semiconductor layer, and the second electrode are arranged in a stacked manner, and the insulating layer and the semiconductor layer form a tunnel junction, an artificial synapse with a capacitive coupling tunnel junction having a heterostructure is constructed, combining charge transport of capacitive coupling effect and quantum tunneling effect; when both the first electrode and the second electrode are disposed on the surface of the insulating layer facing away from the semiconductor layer, an artificial synapse with a heterostructure and a tunnel junction is also constructed, utilizing charge transport of quantum tunneling effect; polarization is generated in the insulating layer under the action of an externally applied electrical signal or stress to adjust the barrier height of the tunnel junction, realizing multimodal synaptic response; moreover, when realizing the artificial synapse function by adjusting the barrier height of the tunnel junction through an externally applied electrical signal or stress, the power consumption generated is low, the response speed is fast, and the stability is good, so that the signal transmission and regulation functions of biological synapses can be accurately simulated. Description of the Drawings

[0019] Figure 1 It is a schematic structural diagram of an artificial synapse provided in an embodiment of the present invention;

[0020] Figure 2 It is a schematic diagram of the principle of polarization induced by a nerve-like voltage pulse provided in an embodiment of the present invention;

[0021] Figure 3 Schematic diagram for regulating the barrier height of the tunnel junction provided in the embodiment of the present invention;

[0022] Figure 4 I-V test results provided in the embodiment of the present invention;

[0023] Figure 5 Test results of voltage, amplitude and phase provided in the embodiment of the present invention;

[0024] Figure 6 Fitting results of voltage and phase provided in the embodiment of the present invention;

[0025] Figure 7 Response test results of artificial synapses under different voltage pulses provided in the embodiment of the present invention;

[0026] Figure 8 Schematic diagram of the structure of another artificial synapse provided in the embodiment of the present invention;

[0027] Figure 9 Schematic diagram of the structure of a monitoring and recognition system provided in the embodiment of the present invention;

[0028] Figure 10 Timing diagram of multiple artificial synapses being externally stimulated provided in the embodiment of the present invention;

[0029] Figure 11 Schematic diagram of the processing process of the neural network calculation model provided in the embodiment of the present invention;

[0030] Figure 12 Test results of recognition accuracy provided in the embodiment of the present invention;

[0031] Figure 13 Schematic diagram of the confusion matrix provided in the embodiment of the present invention. Detailed implementation manners

[0032] The following will combine with the accompanying drawings to elaborate in detail on the detailed implementation manners of an artificial synapse and a monitoring and recognition system provided in the embodiment of the present invention. It should be noted that the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0033] The embodiment of the present invention provides an artificial synapse, such as Figure 1As shown, the artificial synapse may include: a first electrode, an insulating layer, a semiconductor layer, and a second electrode that are sequentially stacked; the thickness d1 of the insulating layer is not greater than 5 nm, so the thickness d1 of the insulating layer is small, and the insulating layer can also be called an ultra-thin insulating layer;

[0034] The insulating layer and the semiconductor layer form a tunnel junction;

[0035] The first electrode and the second electrode are used to input electrical signals;

[0036] The insulating layer is used to: generate dielectric polarization under the action of an electrical signal and adjust the barrier height of the tunnel junction.

[0037] In this way, when the first electrode, the insulating layer, the semiconductor layer, and the second electrode are stacked, and the insulating layer and the semiconductor layer form a tunnel junction, an artificial synapse with a capacitive-coupled tunnel junction of a heterostructure is constructed, combining the charge transport of the capacitive-coupling effect and the quantum tunneling effect. Polarization is generated in the insulating layer under the action of an externally applied electrical signal to adjust the barrier height of the tunnel junction, realizing a multimodal synaptic response; moreover, when the artificial synapse function is realized by adjusting the barrier height of the tunnel junction through an externally applied electrical signal, the power consumption generated is low, the response speed is fast, and the stability is good. For example, when the time of the applied voltage pulse is 10 ns, the artificial synapse can respond, and the power consumption can be reduced to 1 fJ, so as to accurately simulate the signal transmission and regulation functions of biological synapses.

[0038] Optionally, the setting method of each membrane layer in the artificial synapse is as follows:

[0039] For the insulating layer, the materials for fabricating the insulating layer include at least one of silicon oxide, hafnium oxide, tantalum oxide, aluminum oxide and other inorganic oxides. That is to say, the insulating layer can be provided with one layer, and the material for fabricating this insulating layer can be any one of silicon oxide, hafnium oxide, tantalum oxide, aluminum oxide and other inorganic oxides; or, the insulating layer can be provided with multiple layers, and each layer of the insulating layer can be fabricated using any one of silicon oxide, hafnium oxide, tantalum oxide, aluminum oxide and other inorganic oxides. The materials for fabricating at least some of the layers of the insulating layer can be different, so as to achieve combinations of multiple film layers, such as but not limited to the combination of silicon oxide and hafnium oxide, the combination of tantalum oxide and aluminum oxide, the combination of silicon oxide and aluminum oxide, etc., which will not be listed one by one here, and can be specifically combined according to actual needs. Moreover, when the insulating layer is provided with multiple layers, the semiconductor layer will form a tunnel junction with the insulating layer in contact therewith. That is to say, the tunnel junction exists at the interface between the semiconductor layer and the insulating layer. In addition, the thickness d1 of any insulating layer can be set to not more than 5 nm. Further, the thickness d1 of any insulating layer can be set to 2 nm to 5 nm. For example but not limited to, any insulating layer is a film layer with a thickness d1 of about 4 nm fabricated by chemical vapor deposition such as atomic layer deposition to obtain an ultra-thin insulating layer. When a tunnel junction is formed between the ultra-thin insulating layer and the semiconductor layer, the regulation of the barrier height of the tunnel junction can be realized.

[0040] For the semiconductor layer, the materials for fabricating the semiconductor layer include any one of silicon, germanium, gallium arsenide, silicon carbide and other semiconductor materials. And, the thickness d2 of the semiconductor layer can be set according to actual circumstances, which is not specifically limited here. For example but not limited to, the semiconductor layer can be a silicon wafer with a thickness d2 of about 500 μm.

[0041] For the first electrode and the second electrode, the materials for fabricating at least one of the first electrode and the second electrode include at least one of gold, silver, platinum, copper, aluminum. The materials for fabricating the first electrode and the second electrode can be the same or different; the thicknesses of the first electrode and the second electrode can be set according to actual circumstances, which is not specifically limited here. For example but not limited to, the first electrode and the second electrode are film layers with a thickness of about 50 nm fabricated by physical vapor deposition such as magnetron sputtering.

[0042] Optionally, taking the second electrode grounded and an electrical signal input through the first electrode as an example, the electrical signal can be a voltage pulse similar to a nerve. The insulating layer generates dielectric polarization under the action of the electrical signal, and then generates polarization charges. The polarization charges can accumulate or dissipate with the magnitude of the electrical signal; the semiconductor layer and the insulating layer in direct contact therewith form a tunnel junction. This tunnel junction is affected by the polarization charges in the insulating layer, so that the barrier height of the tunnel junction can be adjusted, and charges are transferred in the semiconductor layer, so that the corresponding charges can be transferred to the second electrode, thereby realizing the transmission of information.

[0043] Therefore, compared with the artificial synapses made of ferroelectric materials in the prior art, the working mechanism is significantly different. In the embodiments of the present invention, after the insulating layer and the semiconductor layer are stacked, they are sandwiched between the first electrode and the second electrode, so a structure similar to a capacitor is formed. When a tunneling junction is formed between the insulating layer and the semiconductor layer, an artificial synapse with a capacitive coupling tunneling junction is formed. By applying a voltage pulse similar to a nerve, polarization can be induced in the insulating layer, and the polarization charges can play a gating effect to control the barrier height of the tunneling junction.

[0044] Specifically, as shown in Figure 2 and Figure 3 , Figure 2 is a schematic diagram of polarization induced by a voltage pulse similar to a nerve, Figure 3 is a schematic diagram of the regulation of the barrier height of the tunneling junction, where Figure 3 the dotted line in it represents the energy band when no pulse is applied, and the solid line represents the energy band when a pulse is applied. CB represents conduction band, VB represents valence band, E F represents the Fermi level, h + represents a hole, and e - represents an electron; taking the second electrode grounded and an electrical signal input through the first electrode, with the semiconductor layer being a silicon layer and the insulating layer made of alumina as an example, the specific working principle includes:

[0045] When a negative electrical pulse is applied to the first electrode, under the action of the negative charges in the first electrode, polarization is generated in the insulating layer, causing positive charges to accumulate on the surface of the insulating layer facing the first electrode and negative charges to accumulate on the surface of the insulating layer facing the silicon layer. And these positive and negative charges in the insulating layer can both be called polarization charges. Moreover, as the voltage of the negative electrical pulse increases, the number of polarization charges also increases, such as the process from (a) in Figure 2 to (b) in Figure 2 ; at this time, holes (i.e., positive charges) accumulate on the surface of the silicon layer facing the insulating layer, causing the holes to accumulate at the interface between the silicon layer and the insulating layer, as shown in (a) in Figure 3 , reducing the barrier height of the tunneling junction. For example, the barrier height decreases from φ N to φ N ’ , and at this time, carriers are allowed to cross the barrier of the tunneling junction with the minimum energy, thus showing a negative excitatory synaptic response.

[0046] If a reverse pulse such as a positive electrical pulse is applied after the negative electrical pulse is significantly enhanced, such as the process from (b) in Figure 2 to (a) in Figure 2 , and as shown in Figure 3As shown in (b) therein, the holes generated at the interface between the insulating layer and the silicon layer are less dissipated, resulting in an increase in the barrier height of the tunnel junction, a decrease in the tunneling rate of carriers, and thus a negative inhibitory synaptic response is exhibited.

[0047] Similarly, when a positive electrical pulse is applied to the first electrode, under the action of the positive charges in the first electrode, polarization occurs in the insulating layer, causing negative charges to accumulate on the surface of the insulating layer facing the first electrode and positive charges to accumulate on the surface of the insulating layer facing the silicon layer. These positive and negative charges in the insulating layer can both be referred to as polarization charges, and as the voltage of the positive electrical pulse increases, the number of polarization charges also increases, such as from Figure 2 as shown in (c) therein to Figure 2 as shown in (d) therein; at this time, electrons (i.e., negative charges) accumulate on the surface of the silicon layer facing the insulating layer, causing electrons to accumulate at the interface between the silicon layer and the insulating layer, as Figure 3 shown in (c) therein, reducing the barrier height of the tunnel junction. For example, the barrier height decreases from φ P to φ P ’ , and at this time, carriers are allowed to cross the barrier of the tunnel junction with the minimum energy, thus exhibiting a positive excitatory synaptic response.

[0048] If a reverse pulse such as a negative electrical pulse is applied after the positive electrical pulse is significantly enhanced, such as from Figure 2 as shown in (d) therein to Figure 2 as shown in (c) therein, and as Figure 3 shown in (d) therein, the electrons generated at the interface between the insulating layer and the silicon layer are less dissipated, resulting in an increase in the barrier height of the tunnel junction, a decrease in the tunneling rate of carriers, and thus a positive inhibitory synaptic response is exhibited.

[0049] Based on the above working principle, it can be summarized that the first electrode and the second electrode are specifically used for: when an electrical signal with a first polarity is input, reducing the barrier height of the tunnel junction, and the artificial synapse exhibits an excitatory synaptic response; when an electrical signal with a second polarity is input, increasing the barrier height of the tunnel junction, and the artificial synapse exhibits an inhibitory synaptic response; the first polarity and the second polarity are opposite. Among them, when the first polarity is positive, the second polarity is negative, or when the first polarity is negative, the second polarity is positive. Therefore, when a programmed electrical pulse is applied to the first electrode, the positive or negative charges accumulated at the interface between the semiconductor layer and the insulating layer can modulate the interface energy band, thereby adjusting the barrier height of the tunnel junction, enabling the artificial synapse to exhibit an excitatory or inhibitory synaptic response.

[0050] Moreover, by testing the artificial synapse, the test results shown in Figure 4 , Figure 5 and Figure 6 can be obtained respectively.Figure 4 are the I-V test results, Figure 5 are the test results of voltage, amplitude, and phase, Figure 6 are the fitting results of voltage and phase, where NE refers to negative excitatory synaptic response, NI refers to negative inhibitory synaptic response, PE refers to positive excitatory synaptic response, and PI refers to positive inhibitory synaptic response; from Figure 4 the results shown in it, it can be seen that when applying positive and negative electrical pulses, the artificial synapse shows asymmetric responses respectively, enabling the artificial synapse to have both positive and negative electrical modulation functions. Moreover, as the positive voltage (or negative voltage) changes, the current can show enhancement and attenuation, and the current can return to zero after attenuation, indicating that the artificial synapse provided by the embodiments of the present invention has reversible, dual-language multi-modal behavior and can perform computational operations by programming voltage pulses.

[0051] In Figure 5 the amplitude is used to represent the deformation occurring on the surface of the insulating layer, which is generated due to the polarization of the insulating layer. Therefore, it can be found from the amplitude results that the insulating layer can be polarized under different voltage pulses, and as the voltage changes, the degree of polarization and the number of generated polarization charges will also change accordingly; Figure 5 the phase in Figure 5 can also be used to represent the polarization process of the insulating layer. Therefore, through

[0052] In Figure 6 the test results shown, it can be well demonstrated that under the action of voltage pulses, the insulating layer can be polarized, which is conducive to realizing the adjustment of the barrier height of the tunnel junction. Figure 6 the fitting results of the phase values within a certain voltage range can be obtained. For Figure 6 the left fitted line in it, as the absolute value of the voltage under the negative electrical pulse increases, that is, the negative electrical pulse changes along the direction indicated by the solid arrow in the figure, the phase gradually decreases, indicating that the artificial synapse shows an inhibitory synaptic response; for

[0053] the right fitted line in it, as the absolute value of the voltage under the negative electrical pulse increases, that is, the negative electrical pulse changes along the direction indicated by the solid arrow in the figure, the phase gradually increases, indicating that the artificial synapse shows an excitatory synaptic response.

[0054] Therefore, referring toFigure 7 As shown Figure 7 are the test results of the response of the artificial synapse under different voltage pulses, where V plus represents the applied voltage pulse, Figure 7 in (a) corresponds to a positive electrical pulse. As the voltage increases, the postsynaptic response current gradually evolves towards a larger value, causing the artificial synapse to exhibit an excitatory synaptic response. Moreover, when the pulse width is 10 ns, the artificial synapse can exhibit a response, demonstrating an ultra-fast response speed. Figure 7 in (b) corresponds to a negative electrical pulse. As the voltage increases, the postsynaptic response current gradually evolves towards a smaller value, causing the artificial synapse to exhibit an inhibitory synaptic response. This shows that the artificial synapse provided by the embodiments of the present invention can generate different responses by modulating and programming the applied pulse signal, thereby exhibiting respective characteristics in different modes.

[0055] Based on the same inventive concept, the embodiments of the present invention also provide an artificial synapse, as Figure 8 shown. The artificial synapse includes: an electrode layer, an insulating layer, and a semiconductor layer arranged in a stacked manner. The electrode layer includes a first electrode and a second electrode arranged at intervals and side by side; the first electrode and the second electrode are used to transmit electrical signals; the thickness d1 of the insulating layer is not greater than 5 nm;

[0056] The insulating layer and the semiconductor layer form a tunnel junction;

[0057] The manufacturing material of the semiconductor layer has flexoelectricity or piezoelectricity;

[0058] The artificial synapse is flexible. When stress is applied to the artificial synapse, the insulating layer generates dielectric polarization to adjust the barrier height of the tunnel junction.

[0059] That is to say, when the artificial synapse is flexible, without applying an electrical signal and in the case of no external pulse, when the semiconductor layer has flexoelectricity or piezoelectricity, by applying stress to the artificial synapse, the artificial synapse can be deformed, and then an internal built-in electric field is generated in the semiconductor layer. This built-in electric field can induce polarization in the insulating layer to generate polarization charges, adjusting the barrier height of the tunnel junction, realizing multi-modal synaptic responses; moreover, when realizing the function of the artificial synapse by adjusting the barrier height of the tunnel junction through stress, the power consumption generated is low, the response speed is fast, and the stability is good, so that the signal transmission and regulation functions of biological synapses can be accurately simulated.

[0060] Optionally, the thickness d2 of the semiconductor layer is not greater than 100 μm. At this time, the semiconductor layer can be a two-dimensional thin film with a relatively thin thickness. When the thicknesses of the insulating layer and the electrode layer are both relatively thin and these film layers are all disposed on a flexible substrate, a flexible artificial synapse can be obtained to realize synaptic responses under mechanical stress control.

[0061] Optionally, the material for fabricating the semiconductor layer includes: a piezoelectric material or a flexoelectric material. The piezoelectric material includes, but is not limited to, lead zirconate titanate, aluminum nitride, zinc oxide, etc. The flexoelectric material includes, but is not limited to, barium titanate, molybdenum sulfide, etc. Therefore, the material for fabricating the semiconductor layer may include any one of lead zirconate titanate, aluminum nitride, zinc oxide, barium titanate, molybdenum sulfide, and other piezoelectric materials or flexoelectric materials. In this way, the piezoelectricity and flexoelectricity can be utilized to generate a built-in electric field under stress, thereby realizing the adjustment of the barrier height of the tunnel junction.

[0062] Optionally, when a first stress is applied to the artificial synapse, the barrier height of the tunnel junction is reduced, and the artificial synapse exhibits an excitatory synaptic response; when a second stress is applied to the artificial synapse, the barrier height of the tunnel junction is increased, and the artificial synapse exhibits an inhibitory synaptic response. Among them, the first stress and the second stress can be any form of stress, such as, but not limited to: the first stress is the stress that causes the artificial synapse to bend upward, and the second stress is the stress that causes the artificial synapse to bend downward; or, the first stress is the stress that causes the artificial synapse to bend downward, and the second stress is the stress that causes the artificial synapse to bend upward.

[0063] Of course, in this embodiment, the setting positions of the first electrode and the second electrode are not limited to being disposed on the surface of the insulating layer facing away from the semiconductor layer, and can also be like Figure 1 as shown in the figure, the first electrode is located on the surface of the insulating layer facing away from the semiconductor layer, and the second electrode is located on the surface of the semiconductor layer facing away from the insulating layer. In this way, when stress is applied, a built-in electric field can also be generated in the semiconductor layer, thereby inducing polarization in the insulating layer and adjusting the barrier height of the tunnel junction, realizing multimodal synaptic responses.

[0064] It should be understood that the working mechanism of the artificial synapse in this embodiment is basically the same as that of the artificial synapse in the foregoing embodiment. The difference lies in: the way of applying the external stimulus is different, but both can cause polarization to occur in the insulating layer and then generate polarization charges. The polarization charges accumulate or dissipate at the interface between the insulating layer and the semiconductor layer, thereby adjusting the barrier height of the tunnel junction and realizing synaptic responses. Therefore, for the similarities between the working mechanism of the artificial synapse in this embodiment and that of the artificial synapse in the foregoing embodiment, reference can be made to the above relevant content introduction, and the repeated parts will not be elaborated.

[0065] Based on the same inventive concept, an embodiment of the present invention further provides a monitoring and recognition system, as Figure 9 shown in the figure, including: a processor and a synaptic array. The synaptic array includes a plurality of the above-mentioned artificial synapses provided by the embodiments of the present invention. The processor is connected to the first electrode and / or the second electrode of each artificial synapse;

[0066] The processor is configured to: obtain a current signal output by at least some of the artificial synapses through the first electrode and / or the second electrode; and determine distribution information of the artificial synapses in the synaptic array that are stimulated by an external stimulus based on the obtained current signal.

[0067] In this way, when a voltage pulse or stress is applied to the artificial synapses, the stimulated artificial synapses will respond to generate a current signal. Based on this current signal, the processor can determine the distribution information of the stimulated artificial synapses, and thus, based on the monitoring of the stimulated artificial synapses, monitor and identify the external stimulus. For example, the trajectory of the external stimulus can be identified.

[0068] Optionally, when determining the distribution information, the processor can combine the excitatory and inhibitory conductances with the inter-frame difference method to capture the distribution information of the stimulated artificial synapses in time and space, providing information on the directions of weakening and strengthening to identify the path of the target, thereby identifying the trajectory of the external stimulus, etc.

[0069] See Figure 10 the timing diagram of multiple artificial synapses being stimulated by an external stimulus as shown. The position of t1 in the figure indicates the position of the artificial synapse stimulated at time t1. Similarly, the position of t2 in the figure indicates the position of the artificial synapse stimulated at time t2, the position of t3 in the figure indicates the position of the artificial synapse stimulated at time t3, the position of t4 in the figure indicates the position of the artificial synapse stimulated at time t4, the position of t5 in the figure indicates the position of the artificial synapse stimulated at time t5, and the position of t6 in the figure indicates the position of the artificial synapse stimulated at time t6.

[0070] Therefore, from time t1 to time t6, a total of 6 artificial synapses are stimulated by the external stimulus, so these 6 artificial synapses will exhibit corresponding response results (i.e., current signals). Assuming but not limited to that the synaptic array includes a total of 28 rows and 28 columns of artificial synapses, the processor can process these response results using but not limited to a neural network calculation model and output a processing result, and the processing result is: from time t1 to time t6, the trajectory of the external stimulus is the number "7", as shown by the processing result "7" within the dashed box in Figure 11 indicating that the trajectory of the external stimulus is "7".

[0071] See Figure 12 the test results of the recognition accuracy as shown. As the number of training times of the neural network calculation model increases, the recognition accuracy also increases, the recognition error rate decreases, and the highest accuracy can reach up to 96.28%, indicating that after effective training, the recognition accuracy of the monitoring and recognition system can be effectively improved.

[0072] See Figure 13 the schematic diagram of the confusion matrix shown. The number 2 within the solid circle 1 indicates that the recognition times corresponding to the case where the recognized number is 0 while the actual number is 9 is 2. The number 57 within the solid circle 2 indicates that the recognition times corresponding to the case where the recognized number is 0 while the actual number is 0 is 57. This shows that the monitoring and recognition system provided by the embodiments of the present invention has a high recognition accuracy rate.

[0073] In summary, the technical solution provided by the embodiments of the present invention proposes a charge transport mechanism combining the capacitance coupling effect and the quantum tunneling effect. Through voltage pulse programming similar to nerves, the internal polarization charges of the insulating layer are used to directly regulate the barrier height of the tunnel junction, having high-speed and low-power neuromorphic simulation capabilities, and realizing bidirectional (positive / negative) and bisexual (excitatory / inhibitory) synaptic plasticity responses. Moreover, the embodiments of the present invention also provide a motion trajectory monitoring and recognition system for an artificial synapse based on a capacitance-coupled tunnel junction, which can capture and recognize the spatio-temporal perception of a moving object to identify the moving trajectory. Among them, the moving object refers to the object that applies a stimulus to the artificial synapse.

[0074] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these changes and modifications.

Claims

1. An artificial synapse, characterized in that, Comprising: A first electrode, an insulating layer, a semiconductor layer, and a second electrode that are sequentially stacked; The thickness of the insulating layer is not greater than 5 nm; The insulating layer and the semiconductor layer form a tunnel junction; The first electrode and the second electrode are used to input an electrical signal; The insulating layer is used to: generate dielectric polarization under the action of the electrical signal and adjust the barrier height of the tunnel junction.

2. The artificial synapse according to claim 1, wherein The first electrode and the second electrode are specifically used to: When inputting an electrical signal with a first polarity, reduce the barrier height of the tunnel junction, and the artificial synapse exhibits an excitatory synaptic response; When inputting an electrical signal with a second polarity, increase the barrier height of the tunnel junction, and the artificial synapse exhibits an inhibitory synaptic response; the first polarity and the second polarity are opposite.

3. The artificial synapse according to claim 1 or 2, characterized in that, The manufacturing material of the insulating layer includes at least one of silicon oxide, hafnium oxide, tantalum oxide, and aluminum oxide.

4. The artificial synapse according to any one of claims 1 to 3, characterized in that, The manufacturing material of the semiconductor layer includes any one of silicon, germanium, gallium arsenide, and silicon carbide.

5. The artificial synapse according to any one of claims 1-4, characterized in that, The manufacturing material of at least one of the first electrode and the second electrode includes at least one of gold, silver, platinum, copper, and aluminum.

6. An artificial synapse, characterized in that, Comprising: An electrode layer, an insulating layer, and a semiconductor layer that are stacked, and the electrode layer includes a first electrode and a second electrode that are spaced apart and arranged side by side; The first electrode and the second electrode are used to transmit an electrical signal; The thickness of the insulating layer is not greater than 5 nm; The insulating layer and the semiconductor layer form a tunnel junction; The manufacturing material of the semiconductor layer has flexoelectricity or piezoelectricity; The artificial synapse is flexible. When stress is applied to the artificial synapse, the insulating layer generates dielectric polarization and adjusts the barrier height of the tunnel junction.

7. The artificial synapse according to claim 6, wherein, The thickness of the semiconductor layer is not greater than 100 μm.

8. The artificial synapse according to claim 6 or 7, characterized in that, The manufacturing material of the semiconductor layer includes any one of lead zirconate titanate, aluminum nitride, zinc oxide, barium titanate, and molybdenum sulfide.

9. The artificial synapse according to any one of claims 6-8, wherein When applying a first stress to the artificial synapse, reduce the barrier height of the tunnel junction, and the artificial synapse exhibits an excitatory synaptic response; When applying a second stress to the artificial synapse, increase the barrier height of the tunnel junction, and the artificial synapse exhibits an inhibitory synaptic response.

10. A monitoring and recognition system, characterized in that, Comprising: A processor and a synaptic array. The synaptic array includes a plurality of artificial synapses according to any one of claims 1-9, and the processor is connected to the first electrode and / or the second electrode of each artificial synapse; The processor is used to: acquire current signals output by at least some of the artificial synapses through the first electrode and / or the second electrode; and determine distribution information of the artificial synapses in the synaptic array that are stimulated by the outside world according to the acquired current signals.

Images