A nerve synapse biomimetic photoelectric slowly varying memristor and a preparation method thereof
By using a cross-structured neural synapse bionic optoelectronic slowly varying memristor, utilizing the MEH-PPV functional layer and vertically interleaved electrodes, combined with photostimulation modulation, the problems of few conductive states and high power consumption in existing devices are solved, realizing low-power multi-neuromorphic simulation functions, which are suitable for artificial neuromorphic computing systems.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANJING UNIV OF POSTS & TELECOMM
- Filing Date
- 2023-06-19
- Publication Date
- 2026-05-19
AI Technical Summary
Existing inorganic materials used to fabricate neural synapse bionic devices have a limited number of conductive states that are difficult to control. Traditional neural synapse bionic devices can only simulate biological synaptic behavior through electrical stimulation and cannot effectively simulate optical signals, resulting in high power consumption and difficulty in meeting the needs of flexible processing and storage of large amounts of information.
A cross-structured neural synapse biomimetic photoelectric slow-varying memristor uses poly[2-methoxy-5-(2-ethylhexyloxy)-1,4-phenylacetylene]MEH-PPV as the functional layer. A continuous and flat amorphous film is prepared by solution spin coating, and Ag or Al is used as the top electrode to form vertically interlaced strip electrodes. Conductivity is regulated by combining light stimulation.
It achieves a variety of low-power neuromorphic simulation functions, such as long-term/short-term memory and experiential learning, and can respond to light signals of different wavelengths and intensities. It is low-cost and highly repeatable, and is suitable for artificial neuromorphic computing systems.
Smart Images
Figure CN117119813B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of organic electronics and optoelectronic information technology, and more specifically, to a neural synapse bionic optoelectronic slowly varying memristor and its preparation method. Background Technology
[0002] With the explosive growth of data and information today, traditional von Neumann architecture computers are no longer sufficient to meet the demands of flexibly processing and storing massive amounts of information. Intelligent computers, capable of learning, memorizing, and flexibly processing information like the human brain, represent the future direction and goal of computer development. The human brain's neural network is a highly parallel, nonlinear information processing system. Thanks to its neural network composed of neurons and synapses, it can perform complex calculations with minimal energy consumption. Biological synapses, connecting billions of neurons, play a crucial role in the transmission of neural signals in the brain. Their prominent characteristic—synaptic plasticity—is a vital neurochemical basis for learning and memory. Therefore, devices capable of simulating the function of biological synapses have become one of the most promising technologies for constructing neuromorphic computing simulation neural networks. In this regard, artificial synapses have been proven to be simulable using various devices, among which memristors are considered one of the most promising candidate devices for neuromorphic computing applications.
[0003] Memristors can simulate synaptic weight through nonlinear conductivity and can be controlled by applying a voltage bias higher than the threshold of the characteristic device. Simulated memristor neural networks have demonstrated the ability to solve computational tasks using different algorithms. However, to date, most devices implementing related functions are made of inorganic materials, resulting in a limited number of conductive states that are difficult to control. Organic materials possess characteristics such as stable molecular structure, easy structural tailoring, low processing cost, functional tunability, and mechanical flexibility. Memristors fabricated from organic functional layer materials offer advantages such as excellent electrical performance, smooth and controllable memristor behavior, and simple fabrication methods. Therefore, they have great application potential in simulating biological synapses. Importantly, the operational stability of organic memristors is comparable to that of inorganic memristors. These advantages pave the way for the integration of organic memristors with future visual sensing devices. (Invention application 202011081743.8: A series of low-power non-volatile resistive random access memories and their fabrication methods.)
[0004] Traditional biomimetic devices for neural synapses can only simulate different biological synaptic behaviors through electrical stimulation, but the signals transmitted by various types of light are also extremely important. By constructing memristor devices using stable photosensitive materials, the limitations of devices that can only simulate neural synapses through electrical stimulation can be overcome, and the power consumption of memristors can be further reduced by using light as a new means of conductivity modulation. The development of optoelectronic biomimetic devices for neural synapses has provided more possibilities for memristors in synapse simulation and the construction of memristor visual neural networks. Summary of the Invention
[0005] To address the aforementioned technical problems, this invention provides a neuromorphic bionic photoelectric slowly varying memristor. The device structure of the slowly varying memristor is a cross structure, consisting of three parts from bottom to top on a conductive glass substrate: a bottom electrode, a functional layer, and a top electrode. The functional layer is a poly[2-methoxy-5-(2-ethylhexyloxy)-1,4-phenylacetylene]MEH-PPV layer; the bottom electrode is indium tin oxide (ITO); the top electrode is Ag; and the MEH-PPV layer is a continuous, flat, amorphous thin film.
[0006] The cross structure here refers to the bottom electrode and the top electrode being interleaved.
[0007] Furthermore, the indium tin oxide is in the form of strips, and the silver is in strips that are perpendicular to it.
[0008] Preferably, the area of the vertically interlaced strip-shaped cross electrodes is 10000 μm. 2 -100000μm 2 .
[0009] Furthermore, the functional layer can be a single-layer, double-layer, or multi-layer structure.
[0010] Furthermore, the thickness of the Ag layer is 30-40 nm, the thickness of the MEH-PPV layer is 20-30 nm, and the thickness of the ITO layer is 120-140 nm.
[0011] This invention also provides a method for fabricating a neural synapse bionic photoelectric slowly varying memristor, comprising the following steps:
[0012] (1) Select conductive glass as the substrate, generate a strip of ITO with a thickness of 120-140nm on the conductive glass to form ITO conductive glass, and clean the ITO conductive glass with ITO cleaning agent and ultrapure water for 30 min in sequence, then blow dry and place it in an oven at 120 ℃ for 60 min to dry.
[0013] (2) The dried ITO conductive glass was treated with ultraviolet ozone for 15 min.
[0014] (3) Dissolve MEH-PPV in one of chloroform, dichloromethane, chlorobenzene, tetrahydrofuran or toluene to form a solution;
[0015] (4) Use a pipette to spin-coat the solution onto the ITO conductive glass to form a film. The spin-coating speed is 2500-3500 rpm and the time is 20-40 s. The surface of the functional layer film is a continuous and flat amorphous film. Finally, place it in an oven to cure and dry for 50-70 min at a curing temperature of 75-85℃.
[0016] Annealing at 80℃ for 1 hour resulted in a MEH-PPV layer thickness of 20-30 nm.
[0017] (5) Place the ITO conductive glass film into the vacuum evaporation equipment and control the pressure in the vacuum chamber to be 5×10⁻⁶. -4 Pa, begin evaporating and depositing metallic Ag electrodes, with a silver thickness of 30-40 nm, wherein the Ag electrodes and ITO are arranged in vertically interlaced strip-shaped Ag electrodes;
[0018] (6) After the evaporation is completed, the metal electrode is cooled to room temperature in a vacuum chamber to obtain a neural synapse bionic photomemristor.
[0019] Furthermore, the strip-shaped ITO and strip-shaped Ag electrodes are shaped by using a shadow mask.
[0020] Compared with existing technologies, this invention has the following significant advantages: 1. The MEH-PPV layer of the slowly varying neural synapse bionic photoelectric memristor provided by this invention is prepared by solution spin coating, which has advantages such as good sustainability, good solubility, simple fabrication, and excellent photoelectric performance; 2. The neural synapse bionic photoelectric memristor in this invention can perform various neuromorphic simulations, such as long-term / short-term memory (LTM / STM) and experiential learning based on "learning-forgetting"; 3. The neural synapse bionic photoelectric memristor in this invention can respond differently to light of different wavelengths and intensities; 4. The neural synapse bionic photoelectric memristor in this invention has low cost, good repeatability, and can be used in artificial neuromorphic computing systems, providing more possibilities for building visual perception systems. Attached Figure Description
[0021] Figure 1 This is a schematic diagram of the structure of the biomimetic photoelectric memristor for neural synapses prepared according to the present invention;
[0022] Figure 2 Here is the molecular structure diagram of MEH-PPV;
[0023] Figure 3 A ball-and-stick model for the MEH-PPV molecule;
[0024] Figure 4 Atomic force microscopy scanning images of the MEH-PPV layer in the ITO / MEH-PPV / Ag structure memory device prepared for Example 1;
[0025] Figure 5 The current-voltage (IV) characteristic curve of the memristor based on the ITO / MEH-PPV / Ag structure prepared for Example 1 when a 5 V forward voltage is applied;
[0026] Figure 6The current-voltage (IV) characteristic curve of the memristor based on the ITO / MEH-PPV / Al structure prepared for Example 2 when a 5 V forward voltage is applied;
[0027] Figure 7 The current-voltage (IV) characteristic curve of the memristor based on the ITO / MEH-PPV / Ag structure prepared for Example 1 when a negative voltage of 5 V is applied;
[0028] Figure 8 The current-voltage (IV) characteristic curve of the memristor based on the ITO / MEH-PPV / Al structure prepared for Example 2 when a negative voltage of 5 V is applied;
[0029] Figure 9 The memristor based on the ITO / MEH-PPV / Ag structure prepared for Example 1 under dark conditions was subjected to multiple pulse stimulations of the same number but different voltages, resulting in changes in the device current level.
[0030] Figure 10 IV characteristic curves of the memristor based on the ITO / MEH-PPV / Ag structure prepared for Example 1 under dark and light conditions;
[0031] Figure 11 The change in the conductivity state of the memristor based on the ITO / MEH-PPV / Ag structure prepared for Example 1 under dark and light conditions;
[0032] Figure 12 The change in device current level of the memristor based on the ITO / MEH-PPV / Ag structure prepared for Example 1 under stimulation by 5 identical optical pulses;
[0033] Figure 13 This demonstrates the 54.7-fifth-watt low power consumption of the memristor based on the ITO / MEH-PPV / Ag structure fabricated in Example 1. Detailed Implementation
[0034] The present invention will be further described below with reference to the accompanying drawings and embodiments.
[0035] This invention provides a biomimetic photoelectric memristor for neural synapses and its fabrication method, the structure of which is as follows: Figure 1 As shown, it includes a top electrode, a functional layer, and a bottom electrode, and has an overall vertical structure. The functional layer is composed of a MEH-PPV layer with a thickness of 20-30 nm. Example 1
[0036] (1) Select conductive glass as the substrate, generate a layer of indium tin oxide on the conductive glass with a thickness of 120-140 nm, preferably 135 nm, to form ITO conductive glass. The ITO conductive glass is then cleaned with ITO cleaning agent and ultrapure water for 30 min, then dried and placed in an oven at 120 °C for 60 min to dry.
[0037] (2) The dried ITO conductive glass was treated with ultraviolet ozone for 15 min.
[0038] (3) Using a pipette, the MEH-PPV solution dissolved in chloroform was spin-coated onto ITO conductive glass to form a film, which was then annealed at 80°C for 1 hour. The thickness of the MEH-PPV layer was 20-30 nm.
[0039] The gradually changing functional layer was prepared by MEH-PPV solution spin coating, in which MEH-PPV dissolved in an organic solvent was spin-coated onto the surface of the bottom electrode, and finally placed in an oven for curing and drying for 50-70 minutes at a curing temperature of 75-85℃. The spin coating speed of the MEH-PPV solution was 2500-3500 rpm, and the time was 20-40 seconds.
[0040] The surface of the gradually changing functional layer film prepared by MEH-PPV solution spin coating is a continuous and flat amorphous film.
[0041] (4) Place the ITO conductive glass film into the vacuum evaporation equipment and control the pressure in the vacuum chamber to be 5×10. -4 At Pa, the deposition of metallic silver electrodes begins. The deposition rate of the top electrode in vacuum deposition is 1-2 Å / s, and the silver thickness is 30-40 nm. The thickness of the film is measured using a profilometer and further confirmed by SEM.
[0042] The electrode shape was controlled using a shadow mask, resulting in both the silver and ITO electrodes being strip-shaped cross electrodes with a cross area of 10000 μm. 2 -100000μm 2 Using cross electrodes not only allows more devices to be integrated in a smaller area, but also saves materials.
[0043] (5) After the vapor deposition experiment is completed, the metal electrode is cooled to room temperature in the vacuum chamber to obtain the neural synapse bionic photoresistor. Then the chamber is opened and the wafer is taken out and the electrical and optical properties of the device are tested. Example 2
[0044] The difference from Example 1 is that aluminum is used as the top electrode of the device.
[0045] Performance testing
[0046] See Figure 1 , Figure 5 , Figure 6 , Figure 7 and Figure 8 The diagrams show the structural diagrams and IV characteristic curves corresponding to the biomimetic photoelectric memristors of the neural synapse in Examples 1 and 2. Figure 5 and Figure 7 These are the positive and negative IV characteristic curves of a memristor based on the ITO / MEH-PPV / Ag structure. Figure 6 and Figure 8 The figures show the positive and negative IV characteristic curves of the ITO / MEH-PPV / Al-based memristor. It can be seen that the device exhibits good positive enhancement and negative suppression characteristics under dark conditions. Under a positive operating voltage, the memristor current level gradually increases, while under a negative operating voltage, the memristor current level gradually decreases. However, compared to memristor devices based on the ITO / MEH-PPV / Al structure, the ITO / MEH-PPV / Ag-based synaptic bionic memristor requires only a smaller voltage stimulus under the same conditions to exhibit excellent memristor performance and is more stable, which is beneficial for meeting the low-power consumption requirements of the device.
[0047] See Figure 4 The atomic force microscopy images of the functional layer of this slowly varying memristor device show that the film surface is uniform, without nanopores or conductive filaments, exhibiting extremely high flatness. This rules out the possibility of a filamentary conduction mechanism, indicating that the device exhibits memristor performance based on a carrier transport mechanism, which corresponds to its slowly varying characteristics.
[0048] See Figure 9 The diagram illustrates the changes in current level of the biomimetic photoelectric memristor fabricated for Example 1 under different voltages and the same number of stimuli. It can be seen that the current level remains almost unchanged under a 4-volt pulse stimulus; the current level only increases when the voltage rises to 5 volts; and the current level increases significantly when the voltage rises to 6 volts. This is similar to the requirement that presynaptic stimulation in biological synapses must reach a certain intensity to elicit a postsynaptic response.
[0049] See Figure 10 The figure shows the IV characteristic curves of the neural synapse bionic photoresistor prepared in Example 1 under dark and light conditions. It can be seen that the current level of the device gradually decreases under dark conditions, and the current level gradually increases after the device is illuminated, and there is a significant change compared to the dark conditions.
[0050] See Figure 11The diagram shows the change in conductivity state of the neural synapse bionic photoresist prepared in Example 1 under dark and light conditions. It can be seen that the conductivity state of the device decreases under dark conditions. After applying light, the conductivity state of the device increases instantaneously and can increase stably. After removing the light, the conductivity state of the device decreases instantaneously and does not return to the initial conductivity state after a period of time, which reflects the non-volatility of the memristor device.
[0051] See Figure 12 The change in current level of the neural synapse bionic photoelectric memristor prepared in Example 1 under five light pulse stimulations shows that the current level of the device has been significantly improved compared to the initial state and can be maintained stably for a period of time.
[0052] See Figure 13 The change in current level of the neural synapse bionic photoresistor prepared in Example 1 under a single light pulse stimulation is shown. Calculations show that the device has a low power consumption of 54.7 femtowatts.
[0053] In summary, this invention discloses a neural synapse biomimetic opto-memristor and its fabrication method, which can effectively improve the function of memristor devices, reduce device power consumption, and can be used for neuromorphic simulation. The device structure is easy to design, the process is simple, the yield is high, and it has universality.
Claims
1. A neural synapse bionic photoelectric slowly varying memristor, characterized in that, The device structure of the slowly changing memristor is a cross structure, which consists of three parts from bottom to top on the conductive glass: a bottom electrode, a functional layer, and a top electrode. The functional layer is a poly[2-methoxy-5-(2-ethylhexyloxy)-1,4-phenylacetylene]MEH-PPV layer; the bottom electrode is indium tin oxide (ITO); the top electrode is Ag; and the MEH-PPV layer is a continuous and flat amorphous thin film.
2. The neural synapse bionic photoelectric slowly varying memristor according to claim 1, characterized in that, The indium tin oxide is in the form of strips, and the silver is in strips that are perpendicular to it.
3. The neural synapse bionic photoelectric slowly varying memristor according to claim 2, characterized in that, The area of the vertically interlaced strip-shaped cross electrodes is 10000 μm. 2 -100000μm 2 .
4. The neural synapse bionic photoelectric slowly varying memristor according to claim 1, characterized in that, The functional layer can be a single-layer, double-layer, or multi-layer structure.
5. The neural synapse bionic photoelectric slowly varying memristor according to claim 1, characterized in that, The thickness of Ag is 30-40 nm, the thickness of MEH-PPV layer is 20-30 nm, and the thickness of ITO is 120-140 nm.
6. A method for fabricating a neural synapse biomimetic photoelectric slowly varying memristor, characterized in that, Includes the following steps: (1) Select conductive glass as the substrate, generate a strip of ITO with a thickness of 120-140nm on the conductive glass to form ITO conductive glass, and clean the ITO conductive glass with ITO cleaning agent and ultrapure water for 30 min in sequence, then blow dry and place it in an oven at 120℃ for 60 min to dry. (2) The dried ITO conductive glass was treated with ultraviolet ozone for 15 min. (3) Dissolve MEH-PPV in one of chloroform, dichloromethane, chlorobenzene, tetrahydrofuran or toluene to form a solution; (4) Use a pipette to spin-coat the solution onto the ITO conductive glass to form a film. The spin-coating speed is 2500-3500 rpm and the time is 20-40s. The surface of the functional layer film is a continuous and flat amorphous film. Finally, place it in an oven to cure and dry for 50-70 minutes at a curing temperature of 75-85℃. Annealing at 80℃ for 1 hour resulted in a MEH-PPV layer thickness of 20-30 nm. (5) Place the film-formed ITO conductive glass into a vacuum evaporation equipment, control the pressure in the vacuum chamber to be 5×10-4 Pa, and start evaporating the metal Ag electrode with a silver thickness of 30-40 nm. The Ag electrode and ITO are vertically intersecting strip Ag electrodes. (6) After the evaporation is completed, the metal electrode is cooled to room temperature in a vacuum chamber to obtain a neural synapse bionic photomemristor.
7. The method for fabricating the neural synapse bionic photoelectric slowly varying memristor according to claim 6, characterized in that, The strip-shaped ITO and strip-shaped Ag electrodes are shaped by using a shadow mask.