A neural synaptic device based on ReS2 / ZnO heterojunction and its fabrication method
The neural synaptic device with ReS2/ZnO heterojunction structure solves the problems of narrow adjustable range of conductance state and limited dynamic plasticity of synaptic weights in the prior art, realizes high bandwidth and low power consumption photoelectric response, simulates the learning and memory functions of biological vision system, and is suitable for artificial vision system.
Patent Information
- Application Number
- CN202311561606.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-11-21
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-11-21
AI Technical Summary
Existing neural synaptic devices have a narrow range of adjustable conductance in response to external stimuli and limited dynamic plasticity of synaptic weights. They cannot effectively simulate the coordinated influence of light and electrical signals in the human visual system, resulting in limited information processing and memory functions.
A ReS2/ZnO heterojunction structure was adopted, and a ReS2 thin film was prepared by dry transfer technology. Then, a neural synapse device compatible with CMOS process was formed by spin-coating ZnO solution and evaporating electrodes multiple times. This device can directly respond to photoelectric signals and simulate the visual adaptation behavior of biological visual perception system.
It achieves high bandwidth and low power consumption characteristics of photoelectric synaptic devices, can simulate the transition from STP to LTP, control PSC and current decay time, simulate Pavlovian conditioned reflex, realize photo-induced learning and memory functions, and has long-term and short-term plasticity and visual adaptability, making it suitable for artificial vision systems.
Smart Images

Figure CN117594697B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a neural synapse device based on a ReS2 / ZnO heterojunction and its fabrication method, belonging to the field of semiconductor devices. Background Technology
[0002] Traditional von Neumann computers face significant challenges in the Internet of Things (IoT) era due to the physical separation of memory and central processing unit. Realizing neuromorphic computing at the physical level relies on synaptic devices that simulate the biological synapse functions in neural networks. Significant progress has been made to date in various artificial synaptic devices, such as field-effect transistors, phase-change memories, and ferroelectric transistors. Although synaptic behavior has been extensively simulated, most neuromorphic devices rely on electrical stimulation to control conductance, limiting the speed and energy levels of neural activity. Furthermore, human sensory stimuli are almost entirely external (vision, touch, smell, hearing, or pain), which cannot be simulated by electrical stimulation alone. Compared to electrically stimulated artificial synapses, photoelectric synapses offer high bandwidth, low power consumption, and low crosstalk, making them ideal candidates for neuromorphic visual systems. Photoelectric synaptic arrays realize image sensing and memory functions, simplifying the circuitry of neuromorphic visual systems.
[0003] One feasible approach is to use photosensitive materials as the functional layer of synaptic devices, enabling them to process optical signals. Currently, synaptic devices, along with auxiliary optoelectronic devices such as photodetectors, phototransistors, photoresistors, and photoelectric switches, have been successfully verified and used to construct artificial vision systems. However, since synaptic devices cannot directly respond to optical stimuli, additional photosensing devices are required. This separation of sensory and memory architecture leads to a series of problems, including high power consumption, structural complexity, unnecessary data movement, and low integration. Furthermore, although this approach facilitates the realization of sensory-memory integrated architectures, its tendency towards single-signal modulation still presents challenges such as a narrow adjustable range of conductance states and limited dynamic plasticity of synaptic weights.
[0004] A narrow range of adjustable electrical conductance means that the electrical conductivity (usually referring to the signal strength transmitted through the synapse) of a neuron or synapse can only change or adjust within a certain range in response to external stimuli, preventing significant regulation or variation. This may be limited by factors such as material properties, device design, or signal transmission mechanisms. Synaptic weight dynamic plasticity refers to the plasticity or adjustability of synaptic connection strength. In certain situations, the connection strength between neurons can be adjusted or changed, and this change may be influenced by factors such as stimulation frequency, time interval, and presynaptic and postsynaptic activity. However, "limited synaptic weight dynamic plasticity" means that the range of variation in synaptic connection strength is somewhat restricted, preventing significant regulation or variation.
[0005] In general, the plasticity of neurons or synapses in existing technologies is limited under certain conditions, affecting the performance and scope of neural network functions such as information processing, learning, or memory. However, in the actual human visual system, the effects of light and electrical signals on vision are coordinated. That is, modulation of electrical signals can help enhance the strength of synaptic connections during visual transmission, thereby improving the efficiency of visual nerve impulse transmission and significantly impacting information perception and memory. Summary of the Invention
[0006] To enhance the plasticity of neural synaptic devices and optimize information processing, learning, or memory functions, this invention provides a neural synaptic device based on a ReS2 / ZnO heterojunction and its fabrication method. The technical solution is as follows:
[0007] The first objective of this invention is to provide a method for fabricating a neural synapse device, comprising:
[0008] Step 1: Using dry transfer technology, the ReS2 film is transferred onto the cleaned substrate to obtain a substrate with ReS2.
[0009] Step 2: Spin-coat the photoresist evenly onto the substrate with ReS2 and bake it to photolithographically pattern the electrode pattern on the material surface, and obtain a sample with a partial photoresist mask.
[0010] Step 3: Place the sample with photoresist mask obtained in Step 2 into a vapor deposition machine, vapor deposit the electrodes, wash away the photoresist, and obtain a device with electrodes;
[0011] Step 4: Prepare a ZnO solution, and uniformly spin-coat the ZnO solution onto the ReS2 device prepared in Step 3 and bake it. Repeat this process multiple times.
[0012] Optionally, the electrodes in step 3 are 5nm Ni and 50nm Au electrodes.
[0013] Optionally, the substrate is a silicon-based substrate, and the cleaning method includes: first, placing the silicon-based substrate in acetone and ultrasonically cleaning it for 4-6 minutes; then, placing the silicon-based substrate in ethanol and ultrasonically cleaning it for 4-6 minutes; then, ultrasonically cleaning it in deionized water for 4-6 minutes; then, drying it with a nitrogen gun; and finally, baking it on a heating platform at 300±10℃ for 9-10 minutes.
[0014] Optionally, in step 2, the spin coating speed is 600±20 rpm for 9-10 s, or 3000±30 rpm for 30±5 s; the baking temperature is 100℃ and the time is 90 s.
[0015] Optionally, the preparation steps of the ZnO solution include: first dissolving 20 mg of ZnO powder in 10 ml of ethanol solution, and then sonicating the resulting solution for 5 min.
[0016] Optionally, in step 4, the spin coating speed is 600±20 rpm for 9-10 seconds and 2000±20 rpm for 20±5 seconds; the baking temperature is 70℃ and the time is 10 minutes each time.
[0017] Optionally, the method for preparing the ReS2 film includes: obtaining a few-layer ReS2 film using mechanical exfoliation technology.
[0018] Optionally, the evaporation electrode in step 3 includes: electron beam evaporation of Ni metal and thermal evaporation of Au metal.
[0019] Optionally, the photoresist is AZ5214E.
[0020] Optionally, the photolithography process specifically includes: placing the substrate on a maskless photolithography stage, aligning the exposure area with the center of the substrate under a microscope, that is, exposing the area to be exposed, blocking the channel area, retaining the channel 8±0.5μm, and obtaining the electrode mask using photolithography technology.
[0021] Optionally, the specific steps for removing photoresist include: immersing the treated substrate in acetone for 5–10 minutes until the photoresist detaches from the substrate; removing the substrate, drying it, and observing it under an optical microscope; if some photoresist remains, immersing it back in acetone for ultrasonic cleaning for 1–2 seconds; drying the substrate again and observing it under an optical microscope, ensuring that ReS2 and the deposited top electrode remain on the substrate; if the photoresist is completely removed, immersing the substrate in ethanol for 1–2 minutes; then immersing it in deionized water for 1–2 minutes; drying it with a nitrogen gun; and finally baking it on a heated platform at 90–100°C for 1–2 minutes.
[0022] Optionally, the evaporation electrode in step 3 includes nickel metal prepared by electron beam evaporation and gold metal prepared by thermal evaporation; wherein, when electron beam evaporation is used for nickel metal, the vacuum conditions of the evaporation machine must meet 6×10⁻⁶. -4 Below Pa, the evaporation rate used is The time is approximately 3 minutes, and the resulting electrode material thickness is 5 nm; during thermal gold deposition, the vacuum conditions of the deposition machine must meet 6 × 10⁻⁶. -4 Below Pa, the evaporation rate used is The time was 30 minutes, and the thickness of the electrode material obtained was 50 nm.
[0023] The present invention also provides a neural synapse device prepared by the above-described preparation method.
[0024] The present invention also provides the application of the above-mentioned neural synaptic devices in the field of constructing artificial vision systems.
[0025] The beneficial effects of this invention are:
[0026] (1) The present invention prepares ReS2 / ZnO heterojunction devices by spin-coating ZnO solution multiple times. The preparation process is simple and highly compatible with traditional CMOS processes.
[0027] (2) The ReS2 / ZnO neural synapse device of the present invention can effectively control the PSC (postsynaptic current) and current decay time by modulating the intensity and width of the light pulse to simulate the transition from STP to LTP.
[0028] (3) The photoelectric synaptic behavior of the present invention can characterize PPF, that is, it allows the simulation of associative learning between two physical quantities, which is also the essence of Pavlovian conditioning.
[0029] (4) The present invention successfully simulates the visual adaptation behavior in the biological visual perception system, enabling organisms to process visual information in the face of drastic changes in light intensity.
[0030] (5) In this invention, ReS2 samples are prepared by dry transfer technology and can be prepared on a large scale. The heterostructure is prepared by spin coating, which is simple to operate and easy to integrate multiple devices.
[0031] The ReS2 / ZnO heterostructure for photosynapses proposed in this invention shows great potential in constructing neural synapses. Photosynapses can directly respond to electrical and optical signals, enabling them to sense, remember, and process photoelectric information within a single unit. By applying specific photoelectric signals to the device, synaptic functions such as long-term synaptic plasticity / short-term synaptic plasticity (LTSP / STSP) and paired impulse promotion / inhibition (PPF / PPD) were systematically studied. The photoresponsive synaptic behavior of the device was verified under various optical signals, and advanced photoinduced learning, memory, and forgetting properties were achieved. Attached Figure Description
[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0033] Figure 1 This is a schematic diagram of the process for preparing ReS2 / ZnO devices in Embodiment 1 of the present invention.
[0034] Figures 2A-2FThe figures show the photoresponse characteristics of the intrinsic ReS2 and ReS2 / ZnO devices in Embodiment 1 and Comparative Example 1 of the present invention under different pulse intensities and widths, wherein:
[0035] Figure 2A The image shows the optical response characteristics of intrinsic ReS2 under different pulse widths.
[0036] Figure 2B The image shows the photoresponse characteristics of ReS2 / ZnO under different pulse widths;
[0037] Figure 2C The percentage of dark current dropped to 70s for intrinsic ReS2 and ReS2 / ZnO devices with different pulse widths;
[0038] Figure 2D The graph shows the photoresponse characteristics of intrinsic ReS2 at different optical powers;
[0039] Figure 2E The image shows the photoresponse characteristics of ReS2 / ZnO at different optical powers;
[0040] Figure 2F The percentage of dark current dropped to 70s for intrinsic ReS2 and ReS2 / ZnO devices at different power drops.
[0041] Figure 3 The graph shows the PPF test results of the intrinsic ReS2 and ReS2 / ZnO devices in Embodiment 1 and Comparative Example 1 of the present invention.
[0042] Figure 4A This is a reflection test diagram of the classic Pavlovian condition (E+L) under a single training run in Embodiment 1 of the present invention.
[0043] Figure 4B The image shows a reflection test diagram under the classic Pavlovian condition (E+L) after five training cycles in Embodiment 1 of the present invention.
[0044] Figures 5A-5C This is a simulation curve of visual adaptation behavior in Embodiment 1 of the present invention, wherein:
[0045] Figure 5A The current simulation curve corresponding to a low threshold is shown, indicating that the human eye will not experience discomfort under low light stimulation.
[0046] Figure 5B The simulated curves show the stress response of organisms under strong light when the current exceeds the threshold.
[0047] Figure 5C This is a simulation curve of the adaptation of biological nerves to strong light. Detailed Implementation
[0048] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.
[0049] The abbreviation of this invention is described below:
[0050] STP: Short-Term Plasticity, is a form of neural plasticity that describes synapses. Short-term plasticity refers to the plastic changes that occur in synapses within a short period of time, typically lasting from a few hundred milliseconds to a few seconds.
[0051] LTP, or Long-Term Potentiation, is a form of synaptic plasticity described in neuroscience. LTP refers to the process of increased intensity of synaptic signal transmission over a prolonged period. This enhancement can last for a long time, from hours to days or even longer.
[0052] PSC: Postsynaptic Current. In the synapse between neurons, when neurotransmitters are released into the synaptic cleft, they can affect the potential of the postsynaptic neuron, resulting in the flow of postsynaptic current.
[0053] PPF: Paired-pulse facilitation refers to the phenomenon that, under two successive pulse stimuli, the neuronal response induced by the second pulse is enhanced relative to the response induced by the first pulse.
[0054] Test method:
[0055] STP, LTP testing: A bias voltage of 0.5V and a gate voltage of -30V were applied. Light pulses of varying intensities and durations were used to simulate the effects of different external stimuli on human vision and memory. When the pulse width increased from 5s to 80s, or the optical power increased from 2.089μW to 399.5μW, the synaptic device transitioned from STP to LTP. By modulating the intensity and width of the light pulses, the transition from STP to LTP can be simulated, allowing for effective control of PSC and decay time.
[0056] PPF test: Apply a bias voltage of 0.5V and a gate voltage of -30V, change the interval time between the two pulses Δt = 1s, 2s, 4s, 5s, 10, 20s, 40s, and quantify it by the ratio of the two PSCs.
[0057] Pavlovian conditioning test: A presynaptic electrical pulse (+30V) applied to the gate is considered a clock, and a presynaptic optical pulse (2.089μW) applied to the ReS2 / ZnO channel is considered food. A mixed photoelectric pulse is considered a training process superimposed with food and clock stimuli. The threshold current for the salivary response (I0) is... th =0.13μA). First, a single electrical pulse was applied, but the memory threshold was not reached. Subsequently, a single light pulse was applied, significantly exceeding the threshold. However, after the light was removed, the current decayed rapidly, with no significant memory effect. When a single light pulse and an electrical pulse were applied synergistically during training to establish an associative reflex between a bell sound and food, a sufficiently high IA was obtained. d This indicates that associative behavior has been established. For the brain's nervous system, learning efficiency and memory levels generally depend on the frequency of memory recall. Therefore, the number of photoelectric co-stimulation sessions plays a crucial role in the efficiency of Pavlovian associative learning. Increasing the co-stimulation of light and electrical pulses to five training cycles resulted in increased current amplitude and retention time, and a significant improvement in memory levels.
[0058] Visual adaptability was tested: When the device was under appropriate light intensity (637nm, 16.55μW), the photoelectric value in the heterojunction channel was below the threshold (<0.4μA), corresponding to the visual behavior of the human eye under mild light. When the light pulse intensity increased to 131.3μW (637nm), the device current increased sharply, far exceeding the threshold, consistent with the visual behavior of the human eye under strong light stimulation. The biological visual system controls the incidence of external light by adjusting the pupil size and changes the sensitivity of the retina to adapt to strong light stimulation, which can be simulated by changing the gate voltage. After applying a negative gate voltage of -30V, the photocurrent can be reduced below the threshold, realizing the simulation of biological adaptive behavior to strong light.
[0059] Raw materials used in the examples:
[0060] ReS2 films were prepared by mechanical exfoliation. The specific steps were as follows: First, a portion of the sample was adhered to the surface of the ReS2 crystal with 3M tape. Then, the tapes were peeled off one another. This process was repeated about ten times until the sample remaining on the tape turned gray, at which point the ReS2 film was obtained.
[0061] PDMS dry transfer technology: Apply a tape containing ReS2 to the PDMS surface, let it stand for 1 to 2 hours, then peel off the tape, leaving a ReS2 sample on the polydimethylsiloxane (PDMS) surface; then attach the treated silicon substrate to the PDMS surface, let it stand for another 1 to 2 hours, then peel off the silicon substrate, and the ReS2 sample has been transferred to the substrate surface.
[0062] The PDMS substrate used was prepared by ourselves. During preparation, Dow Corning SYLGARD 184 silicone rubber and curing agent were mixed at a weight ratio of 10:1 and stirred evenly. After all the bubbles disappeared, the mixture was left to stand in a refrigerator at 4°C for 48 hours to solidify and form a transparent film.
[0063] The electrodes are prepared by vapor deposition, specifically including electron beam evaporation of nickel and thermal evaporation of gold; among them, the vacuum conditions of the evaporation machine must meet 6×10⁻⁶ during electron beam evaporation of nickel. -4 Below Pa, the evaporation rate used is The time is approximately 3 minutes, and the resulting electrode material thickness is 5 nm; during thermal gold deposition, the vacuum conditions of the deposition machine must meet 6 × 10⁻⁶. -4 Below Pa, the evaporation rate used is The time was approximately 30 minutes, and the resulting electrode material had a thickness of 50 nm.
[0064] The ZnO solution is prepared by dissolving 20 mg of ZnO powder in 10 ml of ethanol solution, and then sonicating the resulting solution for 5 min.
[0065] Example 1:
[0066] This embodiment provides a method for fabricating a neural synaptic device based on a ReS2 / ZnO heterojunction, the process of which is as follows: Figure 1 It includes the following steps:
[0067] (1) First, place the silicon substrate in acetone and ultrasonically clean it for 5 minutes; then place the silicon substrate in ethanol and ultrasonically clean it for 5 minutes; then ultrasonically clean it in deionized water for 5 minutes; then dry it with a nitrogen gun; finally bake it on a heating platform at 300°C for 10 minutes.
[0068] (2) A few layers of ReS2 film were obtained by mechanical peeling and then transferred to the substrate in step (1) by PDMS dry transfer technology to obtain a substrate with ReS2.
[0069] (3) The photoresist is uniformly spin-coated onto the substrate with ReS2 prepared in step (2). The spin-coating speed is 600 rpm for 10 s and 3000 rpm for 30 s. Then, it is baked at 100℃ for 90 s. Electrode patterns are photolithographically etched on the material surface using a maskless photolithography machine to obtain a sample with a partial photoresist mask. The photolithographically etched electrode part is exposed to the environment, while the other parts are still isolated from the environment by photoresist.
[0070] (4) Place the sample with the photoresist mask obtained in step (3) into the evaporation machine and evaporate the electrode; then perform electron beam evaporation to deposit nickel metal, with the vacuum condition of the evaporation machine being 6×10⁻⁶. -4Below Pa, the evaporation rate used is The time was 3 minutes, and the resulting nickel electrode material had a thickness of 5 nm. Gold was thermally evaporated under vacuum conditions of 6 × 10⁻⁶. -4 Below Pa, the evaporation rate used is The time was 30 minutes, and the thickness of the gold electrode material obtained was 50 nm.
[0071] (5) Prepare a ZnO solution and spin coat it evenly onto the ReS2 device prepared in step (4). Spin coat at 600 rpm for 10 s and 2000 rpm for 20 s. Then bake at 70°C for 10 min. Repeat three times.
[0072] Comparative Example 1
[0073] Referring to Example 1, the ReS2 wafer was directly transferred to an empty substrate using mechanical stripping and PDMS dry transfer technology, then photoresist was applied and photolithography was performed, and then nickel-gold electrodes were directly vapor-deposited to obtain a directly vapor-deposited ReS2 device.
[0074] The ReS2 device and ReS2 / ZnO device obtained in Example 1 and Comparative Example 1 were measured to obtain their photoelectric characterization results, which are as follows:
[0075] Figures 2A-2F To illustrate the photoresponse characteristics of the intrinsic ReS2 and ReS2 / ZnO devices in Example 1 and Comparative Example 1 under different pulse intensities and widths, with a bias voltage of 0.5V and a gate voltage of -30V, the intensity of 637nm wavelength light pulses was modulated from 2.089μW to 292.8μW and the pulse width from 5s to 80s to simulate the transition from STP to LTP, which can effectively control the PSC and decay time.
[0076] in Figure 2A The image shows the optical response characteristics of intrinsic ReS2 under different pulse widths. Figure 2B The image shows the photoresponse characteristics of ReS2 / ZnO under different pulse widths; Figure 2C The percentage of dark current dropped to 70s for intrinsic ReS2 and ReS2 / ZnO devices with different pulse widths; Figure 2D The graph shows the photoresponse characteristics of intrinsic ReS2 at different optical powers; Figure 2E The image shows the photoresponse characteristics of ReS2 / ZnO at different optical powers; Figure 2F The percentage of dark current dropped to 70s for intrinsic ReS2 and ReS2 / ZnO devices at different power drops.
[0077] It can be seen that the heterojunction photocurrent of the present invention is higher than that of the intrinsic device after the same decrease time, and the heterojunction of the present invention ( Figure 2B ,2E The figure shows that the level remained high even after 200 seconds, while the ReS2 device ( Figure 2A , 2D The dark current has dropped to a very low level, which explains the long-term enhancement effect (LTP) of the heterojunction; in addition, at 70s, the percentage of dark current dropped by ReS2 / ZnO devices at different power pulse widths at 637nm is about an order of magnitude higher than that of intrinsic ReS2 devices. Figure 2C , Figure 2F The fact that the upper point is about one order of magnitude higher than the lower point indicates that the ReS2 / ZnO heterojunction device of the present invention has a long-term enhancement effect (LTP) and good long-term synaptic plasticity.
[0078] Figure 3 The PPF test results for the intrinsic ReS2 and ReS2 / ZnO devices of Example 1 and Comparative Example 1 were obtained by applying a bias voltage of 0.5V and a gate voltage of -30V, and changing the interval time between two pulses Δt = 1s, 2s, 4s, 5s, 10, 20s, and 40s. The ratio of the two PSCs was used for quantification. It can be seen that the heterojunction device has a higher PPF. PPF refers to the phenomenon that under two successive pulse stimulations, the neuronal response induced by the second pulse is enhanced relative to the response induced by the first pulse. The higher the PPF, the more significant the enhancement of the response induced by the second pulse relative to the first pulse. Moreover, when Δt is 10s to 40s or even longer, it can still be maintained above 200%, indicating that the ReS2 / ZnO device of the present invention has better short-term synaptic plasticity.
[0079] Figure 4A This is a reflection test diagram under the classic Pavlovian condition (E+L) after one training iteration in Example 1. Figure 4B This is a classic Pavlovian condition (E+L) reflectance test diagram under five training cycles in Example 1. A presynaptic electrical pulse (+30V) applied to the gate with a bias voltage of 0.5V is considered a clock, and a presynaptic optical pulse (2.089μW) applied to the ReS2 / ZnO channel is considered food. The photoelectric mixed pulse is considered a training process superimposed with food and clock stimuli. The threshold current for the salivary response (I0) is... th =0.13μA). First, a single electrical pulse was applied, but the memory threshold was not reached. Subsequently, a single light pulse was applied, significantly exceeding the threshold. However, after the light was removed, the current decayed rapidly, with no significant memory effect. When a single light pulse and an electrical pulse were applied synergistically during training to establish an associative reflex between a bell sound and food, a sufficiently high IA was obtained. d This indicates that associative behavior has been established. Increasing the synergistic stimulation of light and electrical pulses to 5 training cycles increased the current amplitude and retention time, resulting in a significant improvement in memory levels.
[0080] Figures 5A-5C This is a simulation curve of visual adaptation behavior in Example 1, where... Figure 5A The human eye does not experience discomfort under dim light stimulation, corresponding to a low threshold current. Figure 5B To simulate the stress response of organisms under strong light, the current exceeds a threshold under strong light irradiation. Figure 5C This study simulates the adaptive behavior of biological neurons to strong light. When the device is under appropriate light intensity (637nm, 16.55μW), the photoelectric value in the heterojunction channel is below the threshold (<0.4μA), corresponding to the visual behavior of the human eye under mild light. When the light pulse intensity increases to 131.3μW (637nm), the device current increases sharply, far exceeding the threshold, consistent with the visual behavior of the human eye under strong light stimulation. After applying a negative gate voltage of -30V, the photocurrent can be reduced below the threshold, realizing the simulation of biological adaptive behavior to strong light.
[0081] Some steps in the embodiments of the present invention can be implemented using software, and the corresponding software program can be stored in a readable storage medium, such as an optical disc or a hard disk.
[0082] 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, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for fabricating a neural synaptic device, characterized in that, The method includes: Step 1: Using dry transfer technology, the ReS2 film is transferred onto the cleaned substrate to obtain a substrate with ReS2. Step 2: Spin-coat the photoresist evenly onto the substrate with ReS2 and bake it to photolithographically pattern the electrode pattern on the material surface, and obtain a sample with a partial photoresist mask. Step 3: Place the sample with photoresist mask obtained in Step 2 into a vapor deposition machine, vapor deposit the electrodes, wash away the photoresist, and obtain a device with electrodes; Step 4: Prepare a ZnO solution, and uniformly spin-coat the ZnO solution onto the ReS2 device prepared in Step 3 and bake it. Repeat this process multiple times to form a ZnO layer on the ReS2 device and construct a ReS2 / ZnO heterojunction. The preparation steps of the ZnO solution include: first dissolving 20 mg of ZnO powder in 10 ml of ethanol solution, and then sonicating the resulting solution for 5 min; In step 4, the spin coating speed is 600±20 rpm for 9~10s and 2000±20 rpm for 20±5s; the baking temperature is 70℃ and the time is 10min each time.
2. The method for fabricating the neural synaptic device according to claim 1, characterized in that, The electrodes in step 3 are 5nm Ni and 50nm Au electrodes.
3. The method for fabricating the neural synaptic device according to claim 1, characterized in that, The substrate is a silicon-based substrate. The cleaning method includes: first, placing the silicon-based substrate in acetone and ultrasonically cleaning it for 4-6 minutes; then, placing the silicon-based substrate in ethanol and ultrasonically cleaning it for 4-6 minutes; then, ultrasonically cleaning it in deionized water for 4-6 minutes; then, drying it with a nitrogen gun; and finally, baking it on a heating platform at 300±10℃ for 9-10 minutes.
4. The method for fabricating the neural synaptic device according to claim 1, characterized in that, In step 2, the spin coating speed is 600±20 rpm for 9~10s, and 3000±30 rpm for 30±5s; the baking temperature is 100℃ and the time is 90s.
5. The method for fabricating a neural synaptic device according to claim 1, characterized in that, The method for preparing the ReS2 film includes: obtaining a few-layer ReS2 film using mechanical exfoliation technology.
6. The method for fabricating a neural synaptic device according to claim 2, characterized in that, The evaporation electrodes in step 3 include: electron beam evaporation of Ni metal and thermal evaporation of Au metal.
7. The neural synapse device prepared by the preparation method according to any one of claims 1 to 6.
8. The application of the neural synaptic device of claim 7 in the construction of artificial vision systems.
