TiO2 / ReS2 heterojunction-based self-rectification memristor array and preparation method and application thereof
By growing rhenium disulfide nanoparticles on titanium dioxide memristors, the self-rectification effect is achieved using heterojunction structure, the latent current problem in the memristor array is solved, the integration and efficiency are improved, and it is suitable for neural network applications.
Patent Information
- Application Number
- CN202411209380.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-30
- Publication Date
- 2024-12-06
AI Technical Summary
Existing memristor arrays face potential current problems when applied in neural networks, resulting in misreading and high power consumption, limiting the integration scale and efficiency of the device.
Using TiO2/ReS2 heterojunction self-rectification memristor array, negative self-rectification is achieved by growing rhenium disulfide nanoparticles on titanium dioxide memristors, using the contact potential difference between rhenium disulfide and gold and the contact potential difference between titanium dioxide and FTO. Effects, suppressing potential current without requiring additional devices.
Effectively suppressing potential current, improving the integration and low power consumption of cross-arrays, suitable for simulated memory and conditioned reflection, and supporting the implementation of large-scale neural networks.
Smart Images

Figure CN119095472A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of inorganic material two-dimensional heterojunction and information storage technology, and specifically relates to a TiO 2 / ReS 2 Self-rectifying heterojunction memristor array, preparation method and application thereof. Background Art
[0002] In recent years, the rapid development of artificial intelligence, the Internet of Things and other fields has put forward higher requirements for the fast and efficient processing of massive data. Traditional computing systems based on the von Neumann architecture have difficulty in achieving efficient data processing due to the "storage wall" bottleneck caused by the physical separation of memory and processor. Memristors have great advantages in processing high-dimensional and unstructured data due to their parallel information processing, low power consumption and integrated storage and computing. They are considered to be the main candidates for future storage and neuromorphic computing technologies.
[0003] The distribution of intrinsic defects or ions in metal oxides is a key factor in determining the performance of memristors, and oxygen vacancies are the main factor. The local oxygen vacancy density distribution in a memristor changes with an applied electric field, which in turn changes the conductance of the device. Therefore, controlling the distribution of oxygen vacancies and their movement with the electric field is the key to improving the performance of memristors. It is reported that the insertion of metal nanoparticles can improve the performance of memristors. However, the large size and random distribution of metal nanoparticles hinder the reproducibility of performance between devices over a larger area. Two-dimensional transition metal dichalcogenides (TMDs) are used in many emerging research fields due to their unique chemical and physical properties. As a new type of TMD, rhenium disulfide has a suitable band gap (1.5-1.6eV), abundant edge active sites and excellent electrical properties, and has received increasing attention in fields such as photocatalysis. However, its use as a metal nanoparticle insertion layer to form a heterojunction with titanium dioxide as a functional layer of a memristor to form a memristor array has not been fully explored.
[0004] Neuromorphic systems are arrays of memristor devices with synaptic functions. Therefore, the key element to realize neuromorphic computing is a high-performance synaptic device that can simulate ideal synaptic devices and suppress unnecessary signals in neuromorphic computing. However, the sneak current problem is a key challenge faced by the application of large-scale memristor arrays in neural networks. The sneak current problem mainly refers to the problem of misreading caused by the current not flowing through the set path due to the state of the devices around the unit when reading a certain unit in the cross array. It will interfere with the reading of the high and low resistance states of the memristor and limit the integration scale of the memristor. The common method at present is to connect a memristor and an active transistor in series to form a 1T1R structure, or to connect a selector in series to form a 1S1R structure and a diode in series to form a 1D1R structure. However, the above methods all require the addition of additional devices, which increases the power consumption and complexity of the cross structure array. As the name suggests, the self-rectifying memristor has both the resistive switching characteristics of the memristor and the unidirectional conduction characteristics similar to the diode, which can effectively suppress the sneak current. When self-rectifying memristors are used as integrated components of crossbar arrays, no additional devices are needed to suppress the sneak current of the crossbar array structure, which greatly improves the integration of the crossbar array and reduces power consumption, which is conducive to the realization of large-scale memristor neural networks and has huge application potential. Summary of the invention
[0005] The purpose of the present invention is to provide a preparation method and application of a self-rectifying memristor array based on titanium dioxide / rhenium disulfide heterojunction, wherein the preparation method is fast and simple, and does not require a strict experimental environment. A layer of rhenium disulfide nanoparticles is grown on a titanium dioxide memristor prepared by a hydrothermal method. While retaining the analog resistive switching and synaptic characteristics of the titanium dioxide memristor, the contact potential difference between rhenium disulfide and gold is different from that between titanium dioxide and FTO, resulting in a negative self-rectifying effect, thereby improving the device's ability to suppress sneak current.
[0006] The present invention provides a TiO-based 2 / ReS 2 Heterojunction self-rectifying memristor array based on TiO 2 / ReS 2 The bottom layer of the heterojunction self-rectifying memristor array is conductive glass, a titanium dioxide layer is arranged on the conductive surface of the conductive glass, and a rhenium disulfide layer is arranged on the titanium dioxide layer; the titanium dioxide layer and the rhenium disulfide layer form a heterojunction structure, and a plurality of gold electrodes that are not connected to each other are arranged on the rhenium disulfide layer;
[0007] Each of the gold electrodes and the region of the rhenium disulfide layer, the region of the titanium dioxide layer, and the region of the conductive glass located directly below and covered by the gold electrode form a memristor.
[0008] Preferably, the substrate of the conductive glass is glass, and the conductive layer is fluorine-doped SnO 2 , the thickness of the conductive layer ranges from 450nm to 550nm;
[0009] The rhenium disulfide layer is composed of rhenium disulfide nanoparticles with a particle size range of 420nm to 440nm stacked on each other;
[0010] The thickness of the titanium dioxide layer ranges from 1.2um to 1.4um, and the titanium dioxide layer is composed of closely arranged titanium dioxide nanorods with a diameter of 76nm to 200nm;
[0011] The gold electrode is in the shape of a round cake, with a diameter of 350um and a thickness of 50nm;
[0012] The rectification ratio of each memristor reaches 10 3 .
[0013] The present invention also provides a TiO-based 2 / ReS 2 A method for preparing a heterojunction self-rectifying memristor array comprises the following steps:
[0014] (1) Obtaining conductive glass;
[0015] (2) preparing a titanium dioxide precursor solution using tetrabutyl titanate and hydrochloric acid;
[0016] (3) placing two sheets of the conductive glass described in step (1) with their conductive surfaces facing downward in a V-shape in a reaction container, and adding the precursor solution described in step (2) to perform a solvothermal reaction to obtain titanium dioxide on the conductive glass;
[0017] (4) Use CH 4 N 2 S、NaReO 4 preparing a rhenium disulfide precursor solution;
[0018] (5) placing the conductive glass with titanium dioxide grown thereon obtained in step (3) with the conductive surface facing downward in a V-shape into a reaction container, and adding the precursor solution described in step (4) to perform a solvothermal reaction to obtain a titanium dioxide / rhenium disulfide heterojunction on the conductive glass;
[0019] (6) sputtering a plurality of metal top electrodes of specific shapes on the titanium dioxide / rhenium disulfide heterojunction sample prepared in step (5) to obtain a TiO-based 2 / ReS 2 In a heterojunction self-rectifying memristor array, the plurality of metal top electrodes are not connected to each other.
[0020] Preferably, in step (2), the concentration of the hydrochloric acid is 6 mol / L; and the volume ratio of tetrabutyl titanate to hydrochloric acid is 1.67%.
[0021] Preferably, in step (4), the precursor solution used is CH 4 N 2 S is 58 mg and NaReO 4 Disperse 20 mg into 18 ml of deionized water.
[0022] Preferably, in step (3), the treatment conditions of the solvent thermal reaction are: heating at 150°C for 12 hours; in step (5), the treatment conditions of the solvent thermal reaction are: heating at 210°C for 6 hours.
[0023] Preferably, in step (6), the metal top electrode of a specific shape is a round-shaped gold electrode with a diameter of 350 um and a thickness of 50 nm.
[0024] Preferably, in step (6), the sputtering process specifically includes: using DC magnetron sputtering with a sputtering current of 120 mA and a sputtering time of 103 s.
[0025] The present invention also provides a TiO-based 2 / ReS 2 The application of heterojunction self-rectifying memristor array in simulating human brain memory includes the following steps:
[0026] Based on TiO 2 / ReS 2 A memristor array of heterojunction self-rectification, wherein memristors at preset positions in the array are respectively selected to obtain a memristor group, and voltage 1 and voltage 2 are set, wherein voltage 1 is lower than voltage 2;
[0027] For the memristors in the memristor group, a pulse processing with a voltage value of voltage 1 is performed using a semiconductor parameter analyzer;
[0028] Apply the read voltage to measure the current of the memristor after the voltage 1 pulse is processed. The measured current gradually decays and disappears completely within the preset time, simulating the short-term memory in the human brain.
[0029] Using a semiconductor parameter analyzer to perform pulse processing with a voltage value of voltage 2 on the memristors in the memristor group;
[0030] The read voltage is applied to measure the current of the memristor after the voltage 2 pulse processing. The measured current gradually decays within the preset time but does not completely disappear, simulating the long-term memory in the human brain.
[0031] Preferably, the voltage 1 and the voltage 2 are 2V and 4V respectively; and the preset time is 100 seconds.
[0032] Preferably, the TiO-based 2 / ReS 2 The heterojunction self-rectifying memristor array is a two-dimensional array composed of memristors. The preset memristors in the two-dimensional array are used as pixels to form a two-dimensional image. The memristor devices have different relaxation times for different voltages, which respectively simulate the short-term memory and long-term memory of the human brain for images.
[0033] The present invention also provides a TiO-based 2 / ReS 2 The application of a heterojunction self-rectifying memristor array in simulating conditioned reflexes includes the following steps:
[0034] Two voltage values of pulses are set: a high voltage pulse and a low voltage pulse, the voltage values of the high voltage pulse and the low voltage pulse are voltage 3 and voltage 4 respectively;
[0035] Setting a threshold current; the setting of the threshold current includes: using a semiconductor parameter analyzer to apply voltage 3 to a memristor that has not been subjected to voltage application processing to obtain current 1; applying voltage 4 to a memristor that has not been subjected to voltage application processing to obtain current 2; setting the magnitude of the threshold current between current 1 and current 2;
[0036] Select TiO based 2 / ReS 2 A memristor in a heterojunction self-rectifying memristor array is subjected to high-voltage pulse and low-voltage pulse treatment in sequence using a semiconductor parameter analyzer; after cyclic high-voltage pulse and low-voltage pulse treatment for n rounds, a trained memristor is obtained; the trained memristor refers to a memristor whose current value measured by the low-voltage pulse treatment on the trained memristor immediately after the training is completed is higher than the threshold current;
[0037] For the trained memristor, after a certain period of time, the high-voltage pulse and low-voltage pulse processing are cycled again for m rounds to obtain a trained memristor; and m is less than or equal to n.
[0038] Beneficial effects of the present invention:
[0039] The present invention provides a TiO-based 2 / ReS 2 The heterojunction self-rectifying memristor array has a self-rectifying effect, which can suppress sneak current without parallel connection with transistors, diodes and other devices, because the contact potential difference between rhenium disulfide and gold and the contact potential difference between titanium dioxide and FTO is different, resulting in only a small rectified current in the negative direction of the device. The product of the present invention achieves this characteristic, so that it can be better applied to simulated memory and conditioned reflex. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] In order to more clearly illustrate the specific implementation of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the specific implementation or the prior art description. Some specific embodiments of the present invention will be described in detail in an exemplary but not restrictive manner with reference to the drawings. The same reference numerals in the drawings indicate the same or similar components or parts. Those skilled in the art should understand that these drawings are not necessarily drawn to scale. In the drawings:
[0041] Figure 1 Based on TiO 2 / ReS 2 Schematic diagram of the structure of the heterojunction memristor array, where 1 is the top electrode Au and 2 is ReS 2 Nanoparticles, 3 is TiO 2 Nanorods, 4 is the bottom electrode FTO; the upper layer of rhenium disulfide contacts gold, and the lower layer of titanium dioxide contacts FTO, which produces different contact potential differences. The work functions of rhenium disulfide and gold are similar, the contact potential difference is small, and rhenium disulfide has excellent electrical properties, and gold has high conductivity; the contact potential difference between titanium dioxide and FTO is larger than that between gold and rhenium disulfide, so an asymmetric iv curve is formed in the positive and negative directions, producing a self-rectification effect.
[0042] Figure 2 Based on TiO 2 / ReS 2 Scanning electron microscope schematic diagram of heterojunction memristor device, proving that the TiO prepared by the present invention 2 / ReS 2 The functional layer of the heterojunction memristor device is a structure that combines nanorods and nanoparticles, and the nanorod array structure of the underlying layer can be seen from the unfilled gaps. This structure can use the boundaries of the nanorods to constrain the formation of oxygen vacancy conductive filaments and improve the stability of the device, such as Figure 2 (a) As shown. The defects of the nanorods are concentrated at the top. The defects at the top of the nanorods can be reduced after compounding with rhenium disulfide nanoparticles. In addition, rhenium disulfide has excellent electrical properties, and the device performance can be improved after compounding, as shown in Figure 2 (b) as shown. Figure 2 (c) is a SEM image of the device at a lower magnification, showing that the device has grown densely over a large area;
[0043] Figure 3 Based on TiO 2 / ReS 2 EDS element analysis of the heterojunction memristor device proves that the material prepared by the present invention is TiO 2 / ReS 2 Heterojunction and no impurities;
[0044] Figure 4 Based on TiO 2 / ReS 2 IV curve of heterojunction memristor device;
[0045] Figure 5 Based on TiO 2 / ReS 2 Synaptic properties of heterojunction memristor devices;
[0046] Figure 6 Based on TiO 2 / ReS 2 Memory forgetting simulation of heterojunction memristor array;
[0047] Figure 7 Simulates Pavlov's dog behavior in classical conditioning; DETAILED DESCRIPTION
[0048] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. In the present invention, unless otherwise specified, the raw materials and equipment used can be purchased from the market or are commonly used in the art. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application.
[0049] Example 1
[0050] (1) Conductive glass (FTO) with a conductive layer thickness of 450 nm was ultrasonically cleaned with acetone, anhydrous ethanol, deionized water, and anhydrous ethanol in sequence. After cleaning, the FTO was placed in the air to dry naturally.
[0051] (2) First, prepare 24 ml of 6 mol / L hydrochloric acid solution, add 0.4 ml of tetrabutyl titanate, and stir with magnetic force to obtain a completely dissolved uniform titanium dioxide precursor solution.
[0052] (3) Affixing a high temperature resistant tape to one end of the conductive surface of the conductive glass FTO in step (1); placing two FTO sheets in a V shape with the conductive surface facing downward in a polytetrafluoroethylene liner, adding the precursor solution prepared in step (2), placing the sheets in a forced air drying oven and heating at 150° C. for 12 h, taking out the samples after the reaction is completed, and repeatedly washing the surface impurities with deionized water and anhydrous ethanol to obtain a densely arranged titanium dioxide nanorod sample with a thickness of 1.2-1.4 μm and a diameter of 76-200 nm on the FTO.
[0053] (4) Weigh 58 mg CH 4 N 2 S, 20 mg NaReO 4Put it in a beaker, add 18 ml of deionized water and stir with magnetic force to obtain a completely dissolved and uniform rhenium disulfide precursor solution.
[0054] (5) The titanium dioxide-based memristor device sample obtained in step (3) is placed in a polytetrafluoroethylene liner in a V shape with the growth surface facing downward, and the precursor solution obtained in step (4) is added and placed in a forced air drying oven and heated at 210° C. for 6 h. After the reaction is completed, the sample is taken out and the surface impurities are repeatedly rinsed with deionized water and anhydrous ethanol to obtain a titanium dioxide / rhenium disulfide heterojunction sample in which rhenium disulfide nanoparticles with a particle size of 420-440 nm are grown on the top of titanium dioxide nanorods.
[0055] (6) The titanium dioxide / rhenium disulfide heterojunction sample obtained in step (5) and the mask are bonded together and fixed on a magnetron sputtering stage. The Au target with a purity of 99.99% is fixed on the sputtering source. The vacuum pump is started to evacuate the chamber to 5×10 -3 Pa, and then argon gas was vacuum-flushed three times at a certain flow rate to make the chamber environment an argon environment and adjust the chamber vacuum to about 0.6 Pa. A current of 120 mA was passed through the sputtering source to make the Au target glow discharge. After sputtering for 103 s, a gold top electrode with a diameter of 350 μm and a thickness of 50 nm was formed on the titanium dioxide / rhenium disulfide heterojunction sample. 2 / ReS 2 Heterojunction self-rectifying memristor array.
[0056] (7) The sample obtained in step (6) was tested using a semiconductor parameter analyzer (Keysight B1500A). The voltage range for the positive test was 0→4v→0, the step size was 25mV, and the current limit was 1mA. The voltage range for the negative test was 0→-4v→0, the step size was 25mV, and the current limit was 100mA. Figure 4 As shown in (a), the device exhibits analog resistive switching characteristics in this voltage measurement range. Compared with Comparative Example 1, the addition of the rhenium disulfide layer has a negative rectification effect, and the rectification ratio is 10 3 .
[0057] Comparative Example 1
[0058] (1) Conductive glass (FTO) with a conductive layer thickness of 450 nm was ultrasonically cleaned with acetone, anhydrous ethanol, deionized water, and anhydrous ethanol in sequence. After cleaning, the FTO was placed in the air to dry naturally.
[0059] (2) First, prepare 24 ml of 6 mol / L hydrochloric acid solution, add 0.4 ml of tetrabutyl titanate, and stir with magnetic force to obtain a completely dissolved uniform titanium dioxide precursor solution.
[0060] (3) Affixing a high temperature resistant tape to one end of the conductive surface of the conductive glass FTO in step (1); placing two FTO sheets in a V shape with the conductive surface facing downward in a polytetrafluoroethylene liner, adding the precursor solution prepared in step (2), placing the sheets in a forced air drying oven and heating at 150° C. for 12 h, taking out the samples after the reaction is completed, and repeatedly washing the surface impurities with deionized water and anhydrous ethanol to obtain a densely arranged titanium dioxide nanorod sample with a thickness of 1.2-1.4 μm and a diameter of 76-200 nm on the FTO.
[0061] (4) The titanium dioxide sample and the mask obtained in step (3) are bonded together and fixed on a magnetron sputtering stage. The Au target with a purity of 99.99% is fixed on the sputtering source. The vacuum pump is started to evacuate the chamber to 5×10 -3 Pa, and then argon gas was vacuum-flushed three times at a certain flow rate to make the chamber environment an argon environment and adjust the chamber vacuum to about 0.6 Pa. A current of 120 mA was passed through the sputtering source to make the Au target glow discharge. After sputtering for 103 s, a gold top electrode with a diameter of 350 μm and a thickness of 50 nm was formed on the titanium dioxide sample to obtain a TiO-based 2 of memristor arrays.
[0062] (5) The sample obtained in step (6) was tested using a semiconductor parameter analyzer (Keysight B1500A). The voltage range for the positive test was 0→4v→0, the step size was 25mV, and the current limit was 1mA. The voltage range for the negative test was 0→-4v→0, the step size was 25mV, and the current limit was 100mA. Figure 4 As shown in (b), within this voltage measurement range, the device exhibits analog resistive switching characteristics, and the positive and negative currents are basically symmetrical.
[0063] Example 2
[0064] (1) Conductive glass (FTO) with a conductive layer thickness of 500 nm was ultrasonically cleaned with acetone, anhydrous ethanol, deionized water, and anhydrous ethanol in sequence. After cleaning, the FTO was placed in the air to dry naturally.
[0065] (2) First, prepare 24 ml of 6 mol / L hydrochloric acid solution, add 0.4 ml of tetrabutyl titanate, and stir with magnetic force to obtain a completely dissolved uniform titanium dioxide precursor solution.
[0066] (3) Affixing a high temperature resistant tape to one end of the conductive surface of the conductive glass FTO in step (1); placing two FTO sheets in a V shape with the conductive surface facing downward in a polytetrafluoroethylene liner, adding the precursor solution prepared in step (2), placing the sheets in a forced air drying oven and heating at 150° C. for 12 h, taking out the samples after the reaction is completed, and repeatedly washing the surface impurities with deionized water and anhydrous ethanol to obtain a densely arranged titanium dioxide nanorod sample with a thickness of 1.2-1.4 μm and a diameter of 76-200 nm on the FTO.
[0067] (4) Weigh 58 mg CH 4 N 2 S, 20 mg NaReO 4 Put it in a beaker, add 18 ml of deionized water and stir with magnetic force to obtain a completely dissolved and uniform rhenium disulfide precursor solution.
[0068] (5) The titanium dioxide-based memristor device sample obtained in step (3) is placed in a polytetrafluoroethylene liner in a V shape with the growth surface facing downward, and the precursor solution obtained in step (4) is added and placed in a forced air drying oven and heated at 210° C. for 6 h. After the reaction is completed, the sample is taken out and the surface impurities are repeatedly rinsed with deionized water and anhydrous ethanol to obtain a titanium dioxide / rhenium disulfide heterojunction sample in which rhenium disulfide nanoparticles with a particle size of 420-440 nm are grown on the top of titanium dioxide nanorods.
[0069] (6) The titanium dioxide / rhenium disulfide heterojunction sample obtained in step (5) and the mask are bonded together and fixed on a magnetron sputtering stage, and an Au target with a purity of 99.99% is fixed on a sputtering source. The vacuum pump is started to evacuate the cavity to 5×10-3Pa, and then the cavity is vacuum-flushed three times with argon gas at a certain flow rate, so that the cavity environment is an argon gas environment and the cavity vacuum is adjusted to about 0.6Pa. A current of 120mA is passed through the sputtering source to make the Au target glow discharge. After sputtering for 103s, a gold top electrode with a diameter of 350um and a thickness of 50nm is formed on the titanium dioxide / rhenium disulfide heterojunction sample, and a self-rectifying memristor array based on the TiO2 / ReS2 heterojunction is obtained.
[0070] (7) The sample obtained in step (6) was tested using a semiconductor parameter analyzer (Keysight B1500A). The top electrode of the memristor simulated the front end of the synapse, and the bottom electrode was used to simulate the back end of the synapse. The voltage signal of the device corresponded to the signal emitted by the biological neuron, and the current of the device corresponded to the connection strength between the synapses. The first step in realizing complex neural functions using synaptic devices is the change of conductance with input electrical stimulation (i.e., action potential), i.e., synaptic plasticity, which can be divided into short-term plasticity and long-term plasticity. Long-term plasticity is an advanced feature of biological synapses and is a key element in learning, forgetting, and memory storage. Therefore, we applied 100 continuous pulses to the sample. From the current response of the device, it can be seen that the conductance of the device changes continuously with the stimulation of the pulse. And when the pulse conditions are changed, the current response of the device also shows regular and modulatable changes. Figure 5 (a) The pulse amplitude is fixed at 4V and the pulse width is 1ms. Pulses with different intervals of 100us, 500us, 1ms, and 2ms are used to stimulate the device. Figure 5 (b) The pulse amplitude is fixed at 4V and the pulse interval is 1ms. Pulses with different pulse widths of 100us, 500us, 1ms, and 2ms are used to stimulate the device. Figure 5 (c) The pulse width of the fixed pulse is 1ms, and the pulse interval is 1ms. Pulses of different amplitudes of 3V, 4V and 5V are used to stimulate the device. It can be seen that the conductivity of the device changes with the applied electrical stimulation. The double pulse facilitation (PPF) phenomenon is a typical manifestation of short-term plasticity; here we use a pair of pulses to simulate the double pulse facilitation behavior of the titanium dioxide / rhenium disulfide heterojunction memristor device. The pulse pair used is 4V, the pulse width is 1ms, and the pulse interval ranges from 1ms to 50ms. The double pulse facilitation behavior of the titanium dioxide / rhenium disulfide heterojunction memristor device is shown in Figure 5 As shown in (d), the second pulse stimulation showed an obvious enhancing effect on the first pulse stimulation. In addition, the enhancing effect of the second pulse stimulation on the first pulse stimulation gradually weakened with the increase of the pulse interval, and its law was a compound exponential decay.
[0071] Example 3
[0072] (1) Conductive glass (FTO) with a conductive layer thickness of 550 nm was ultrasonically cleaned with acetone, anhydrous ethanol, deionized water, and anhydrous ethanol in sequence. After cleaning, the FTO was placed in the air to dry naturally.
[0073] (2) First, prepare 24 ml of 6 mol / L hydrochloric acid solution, add 0.4 ml of tetrabutyl titanate, and stir with magnetic force to obtain a completely dissolved uniform titanium dioxide precursor solution.
[0074] (3) Affix a high temperature resistant tape to one end of the conductive surface of the conductive glass FTO in step (1); then place two FTO sheets in a V-shape with the conductive surface facing downward into a polytetrafluoroethylene liner, add the precursor solution prepared in step (2), place in a forced air drying oven and heat at 150°C for 12 hours, take out the sample after the reaction is completed, repeatedly rinse the surface impurities with deionized water and anhydrous ethanol, and obtain a densely arranged titanium dioxide nanorod sample with a thickness of 1.2-1.4um and a diameter of 76-200nm on the FTO. .
[0075] (4) Weigh 58 mg CH 4 N 2 S, 20 mg NaReO 4 Put it in a beaker, add 18 ml of deionized water and stir with magnetic force to obtain a completely dissolved and uniform rhenium disulfide precursor solution.
[0076] (5) The titanium dioxide-based memristor device sample obtained in step (3) is placed in a polytetrafluoroethylene liner in a V shape with the growth surface facing downward, and the precursor solution obtained in step (4) is added and placed in a forced air drying oven and heated at 210° C. for 6 h. After the reaction is completed, the sample is taken out and the surface impurities are repeatedly rinsed with deionized water and anhydrous ethanol to obtain a titanium dioxide / rhenium disulfide heterojunction sample in which rhenium disulfide nanoparticles with a particle size of 420-440 nm are grown on the top of titanium dioxide nanorods.
[0077] (6) The titanium dioxide / rhenium disulfide heterojunction sample obtained in step (5) and the mask are bonded together and fixed on a magnetron sputtering stage. The Au target with a purity of 99.99% is fixed on the sputtering source. The vacuum pump is started to evacuate the chamber to 5×10 -3 Pa, and then argon gas was vacuum flushed three times at a certain flow rate to make the chamber environment an argon environment and adjust the chamber vacuum to about 0.6 Pa. A current of 120 mA was passed through the sputtering source to make the Au target glow discharge. After sputtering for 103 s, a gold top electrode with a diameter of 350 um and a thickness of 50 nm was formed on the titanium dioxide / rhenium disulfide heterojunction sample, and a self-rectifying memristor array based on the TiO2 / ReS2 heterojunction was obtained.
[0078] (7) The samples obtained in step (6) were tested using a semiconductor parameter analyzer (Keysight B1500A) and the memory forgetting test of two images was performed on the 3x3 synaptic array. The synaptic array was input with images "C" and "T" using pulses of 2v and 4v at the same time. Figure 6As shown in (a) and (b), the purpose is to make the device remember the letter "T" for a long time, simulating the long-term memory (LTM) in the human brain, while the letter "C" is used to simulate the short-term memory (STM) in the human brain. After the two images are input into the synaptic array, a read voltage is applied to measure its current. The current corresponding to the letter "C" gradually decays and disappears within 60 seconds as time increases, which corresponds to STM. The current corresponding to the letter "T" decays less within 60 seconds, see Figure 6 Temporal evolution of cf in .
[0079] Example 4
[0080] (1) Conductive glass (FTO) with a conductive layer thickness of 500 nm was ultrasonically cleaned with acetone, anhydrous ethanol, deionized water, and anhydrous ethanol in sequence. After cleaning, the FTO was placed in the air to dry naturally.
[0081] (2) First, prepare 24 ml of 6 mol / L hydrochloric acid solution, add 0.4 ml of tetrabutyl titanate, and stir with magnetic force to obtain a completely dissolved uniform titanium dioxide precursor solution.
[0082] (3) Affix a high temperature resistant tape to one end of the conductive surface of the conductive glass FTO in step (1); then place two FTO sheets in a V-shape with the conductive surface facing downward into a polytetrafluoroethylene liner, add the precursor solution prepared in step (2), place in a forced air drying oven and heat at 150°C for 12 hours, take out the sample after the reaction is completed, repeatedly rinse the surface impurities with deionized water and anhydrous ethanol, and obtain a titanium dioxide nanorod sample with a thickness of 1.2-1.4um and a diameter of 76-200nm densely arranged on the FTO. .
[0083] (4) Weigh 58 mg CH 4 N 2 S, 20 mg NaReO 4 Put it in a beaker, add 18 ml of deionized water and stir with magnetic force to obtain a completely dissolved and uniform rhenium disulfide precursor solution.
[0084] (5) The titanium dioxide-based memristor device sample obtained in step (3) is placed in a polytetrafluoroethylene liner in a V shape with the growth surface facing downward, and the precursor solution obtained in step (4) is added and placed in a forced air drying oven and heated at 210° C. for 6 h. After the reaction is completed, the sample is taken out and the surface impurities are repeatedly rinsed with deionized water and anhydrous ethanol to obtain a titanium dioxide / rhenium disulfide heterojunction sample in which rhenium disulfide nanoparticles with a particle size of 420-440 nm are grown on the top of titanium dioxide nanorods.
[0085] (6) The titanium dioxide / rhenium disulfide heterojunction sample obtained in step (5) and the mask are bonded together and fixed on a magnetron sputtering stage. The Au target with a purity of 99.99% is fixed on the sputtering source. The vacuum pump is started to evacuate the chamber to 5×10 -3 Pa, and then argon gas was vacuum flushed three times at a certain flow rate to make the chamber environment an argon environment and adjust the chamber vacuum to about 0.6 Pa. A current of 120 mA was passed through the sputtering source to make the Au target glow discharge. After sputtering for 103 s, a gold top electrode with a diameter of 350 um and a thickness of 50 nm was formed on the titanium dioxide / rhenium disulfide heterojunction sample, and a self-rectifying memristor array based on the TiO2 / ReS2 heterojunction was obtained.
[0086] (7) The samples obtained in step (6) were tested using a semiconductor parameter analyzer (Keysight B1500A) to simulate the Pavlov dog experiment of classical conditioning. Figure 7 As shown in (a), it represents the response received by the memristor. In this experimental example, a pulse stimulus with an amplitude of 5V, a pulse width of 1ms, and an interval of 1ms is used as a food signal (US), and a pulse stimulus with an amplitude of 2V, a pulse width of 1ms, and an interval of 1ms is used as a ring signal (CS). The threshold current is marked with a red line. If the current generated by the memristor is higher than the threshold current, it means that saliva is secreted. Figure 7 (b) shows the current generated by the memristor when it receives the ringing signal, called the conditional current, which is about 0.9uA, less than the threshold current. Figure 7 (c) The current generated by the memristor receiving the food signal is about 20uA, which is greater than the threshold current. This shows that under the classical conditioning of Pavlov's dogs, the bell signal alone cannot secrete saliva. Figure 7 (d) The currents of 50 US+CS combined trainings were all higher than the threshold current, indicating that a connection was established between CS and US, causing the dog to salivate in response to both food and the bell. When US was removed and CS was used alone for training, the postsynaptic current was also higher than the threshold current, indicating that the CS was memorized but gradually disappeared over time. Figure 7 (e) as shown. Figure 7 (f) shows the process of retraining the device. Only 30 pulses are needed to achieve the memory effect of the initial training, and the CS signal can be remembered after the US is removed. Figure 7 (g).
[0087] Example 5
[0088] A TiO-based 2 / ReS 2Application of heterojunction self-rectifying memristor array in simulating conditioned reflex: First, two voltage pulses are set: a high voltage pulse and a low voltage pulse, the voltage values of the high voltage pulse and the low voltage pulse are voltage 3 and voltage 4 respectively;
[0089] Then setting the threshold current; the setting of the threshold current includes: using a semiconductor parameter analyzer to apply voltage 3 to the memristor that has not been subjected to voltage treatment, to obtain current 1; applying voltage 4 to the memristor that has not been subjected to voltage treatment, to obtain current 2; setting the magnitude of the threshold current between current 1 = 5V and current 2 = 2V;
[0090] Select TiO based 2 / ReS 2 A memristor in a heterojunction self-rectifying memristor array is subjected to high-voltage pulse and low-voltage pulse treatment in succession using a semiconductor parameter analyzer; after 50 rounds of high-voltage pulse and low-voltage pulse treatment, a trained memristor is obtained; the trained memristor refers to a memristor whose current value measured by the low-voltage pulse treatment on the trained memristor immediately after the training is completed is higher than the threshold current;
[0091] For the trained memristor, after a certain period of time, the high-voltage pulse and low-voltage pulse treatment is cycled again for 30 rounds, and the threshold value of 20uA is reached to obtain a trained memristor.
[0092] Example 6
[0093] A TiO-based 2 / ReS 2 The application of heterojunction self-rectifying memristor array in simulating human brain memory includes the following steps:
[0094] Based on TiO 2 / ReS 2 A memristor array of heterojunction self-rectification, wherein memristors at preset positions in the array are respectively selected to obtain a memristor group, and voltage 1 and voltage 2 are set, wherein voltage 1 is lower than voltage 2;
[0095] For the memristors in the memristor group, a pulse processing with a voltage value of voltage 1 is performed using a semiconductor parameter analyzer;
[0096] Apply the read voltage to measure the current of the memristor after the voltage 1 pulse is processed. The measured current gradually decays and disappears completely within the preset time, simulating the short-term memory in the human brain.
[0097] Using a semiconductor parameter analyzer to perform pulse processing with a voltage value of voltage 2 on the memristors in the memristor group;
[0098] The read voltage is applied to measure the current of the memristor after the voltage 2 pulse processing. The measured current gradually decays within the preset time but does not completely disappear, simulating the long-term memory in the human brain.
[0099] In this embodiment, voltage 1 and voltage 2 are 2V and 4V respectively; the preset time is 100 seconds.
[0100] The TiO-based 2 / ReS 2 The heterojunction self-rectifying memristor array is a two-dimensional array composed of memristors. The preset memristors in the two-dimensional array are used as pixels to form a two-dimensional image. The memristor devices have different relaxation times for different voltages, which respectively simulate the short-term memory and long-term memory of the human brain for images.
[0101] The above descriptions are only some specific implementation methods of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any person familiar with the art within the technical scope disclosed in the present invention should be covered within the protection scope of the present invention.
Claims
1. A memristor array based on TiO2 / ReS2 heterojunction self-rectification, characterized in that: The bottom layer of the memristor array based on TiO2 / ReS2 heterojunction self-rectification is conductive glass, a titanium dioxide layer is arranged on the conductive surface of the conductive glass, and a rhenium disulfide layer is arranged on the titanium dioxide layer; the titanium dioxide layer and the rhenium disulfide layer form a heterojunction structure, and a plurality of gold electrodes that are not connected to each other are arranged on the rhenium disulfide layer; Each of the gold electrodes and the region of the rhenium disulfide layer, the region of the titanium dioxide layer, and the region of the conductive glass located directly below and covered by the gold electrode form a memristor.
2. A memristor array based on TiO2 / ReS2 heterojunction self-rectification as claimed in claim 1, characterized in that: The substrate of the conductive glass is glass, the conductive layer is SnO2 doped with fluorine, and the thickness of the conductive layer ranges from 450nm to 550nm; The rhenium disulfide layer is composed of rhenium disulfide nanoparticles with a particle size range of 420nm to 440nm stacked on each other; The thickness of the titanium dioxide layer ranges from 1.2um to 1.4um, and the titanium dioxide layer is composed of closely arranged titanium dioxide nanorods with a diameter of 76nm to 200nm; The gold electrode is in the shape of a round cake, with a diameter of 350um and a thickness of 50nm; The rectification ratio of each memristor reaches 10 3 .
3. A method for preparing a memristor array based on TiO2 / ReS2 heterojunction self-rectification, characterized in that: The following steps are involved: (1) Obtaining conductive glass; (2) preparing a titanium dioxide precursor solution using tetrabutyl titanate and hydrochloric acid; (3) placing two sheets of the conductive glass described in step (1) with their conductive surfaces facing downward in a V-shape in a reaction container, and adding the precursor solution described in step (2) to perform a solvothermal reaction to obtain titanium dioxide on the conductive glass; (4) preparing a precursor solution of rhenium disulfide using CH4N2S and NaReO4; (5) placing the conductive glass with titanium dioxide grown thereon obtained in step (3) with the conductive surface facing downward in a V-shape into a reaction container, and adding the precursor solution described in step (4) to perform a solvothermal reaction to obtain a titanium dioxide / rhenium disulfide heterojunction on the conductive glass; (6) A plurality of metal top electrodes of a specific shape are sputtered on the titanium dioxide / rhenium disulfide heterojunction sample prepared in step (5) to obtain a memristor array based on TiO2 / ReS2 heterojunction self-rectification, wherein the plurality of metal top electrodes are not connected to each other.
4. A method for preparing a memristor array based on TiO2 / ReS2 heterojunction self-rectification as claimed in claim 3, characterized in that: In the step (2), the concentration of the hydrochloric acid is 6 mol / L; and the volume ratio of tetrabutyl titanate to hydrochloric acid is 1.67%.
5. The method for preparing a memristor array based on TiO2 / ReS2 heterojunction self-rectification as claimed in claim 3, characterized in that: In the step (4), the precursor solution used is 58 mg of CH4N2S and 20 mg of NaReO4 dispersed in 18 ml of deionized water.
6. The method for preparing a memristor array based on TiO2 / ReS2 heterojunction self-rectification as claimed in claim 3, characterized in that: In the step (3), the treatment conditions of the solvent thermal reaction are: heating at 150° C. for 12 hours; in the step (5), the treatment conditions of the solvent thermal reaction are: heating at 210° C. for 6 hours.
7. The method for preparing a memristor array based on TiO2 / ReS2 heterojunction self-rectification as claimed in claim 3, characterized in that: In the step (6), the metal top electrode of the specific shape is a round cake-shaped gold electrode with a diameter of 350 um and a thickness of 50 nm.
8. The method for preparing a memristor array based on TiO2 / ReS2 heterojunction self-rectification as claimed in claim 3, characterized in that: In the step (6), the sputtering process specifically includes: using DC magnetron sputtering with a sputtering current of 120 mA and a sputtering time of 103 s.
9. Application of a memristor array based on TiO2 / ReS2 heterojunction self-rectification in simulating human brain memory as claimed in any one of claims 1 or 2, characterized in that: The following steps are involved: A memristor array based on TiO2 / ReS2 heterojunction self-rectification is constructed, and memristors at preset positions in the array are selected to obtain a memristor group, and voltage 1 and voltage 2 are set, where voltage 1 is lower than voltage 2; For the memristors in the memristor group, a pulse processing with a voltage value of voltage 1 is performed using a semiconductor parameter analyzer; Apply the read voltage to measure the current of the memristor after the voltage 1 pulse is processed. The measured current gradually decays and disappears completely within the preset time, simulating the short-term memory in the human brain. Using a semiconductor parameter analyzer to perform pulse processing with a voltage value of voltage 2 on the memristors in the memristor group; The read voltage is applied to measure the current of the memristor after the voltage 2 pulse processing. The measured current gradually decays within the preset time but does not completely disappear, simulating the long-term memory in the human brain.
10. The application of a memristor array based on TiO2 / ReS2 heterojunction self-rectification in simulating human brain memory as claimed in claim 9, characterized in that: Voltage 1 and voltage 2 are 2v and 4v respectively; the preset time is 100 seconds.
11. The application of a memristor array based on TiO2 / ReS2 heterojunction self-rectification in simulating human brain memory as claimed in claim 9, characterized in that: The memristor array based on TiO2 / ReS2 heterojunction self-rectification is a two-dimensional array composed of memristors. The preset memristors in the two-dimensional array are used as pixels to form a two-dimensional image. The memristor devices have different relaxation times for different voltages, which respectively simulate the short-term memory and long-term memory of the human brain for images.
12. Application of a memristor array based on TiO2 / ReS2 heterojunction self-rectification in simulating conditioned reflex as claimed in any one of claims 1 or 2, characterized in that: The following steps are involved: Two voltage values of pulses are set: a high voltage pulse and a low voltage pulse, the voltage values of the high voltage pulse and the low voltage pulse are voltage 3 and voltage 4 respectively; Set the threshold current; The setting of the threshold current includes: using a semiconductor parameter analyzer to apply voltage 3 to a memristor that has not been subjected to voltage application processing to obtain current 1; applying voltage 4 to a memristor that has not been subjected to voltage application processing to obtain current 2; and setting the magnitude of the threshold current between current 1 and current 2; A memristor in a memristor array based on TiO2 / ReS2 heterojunction self-rectification is selected, and a semiconductor parameter analyzer is used to perform high-voltage pulse and low-voltage pulse treatment in sequence; after n rounds of high-voltage pulse and low-voltage pulse treatment, a trained memristor is obtained; the trained memristor refers to a memristor whose current value measured by the low-voltage pulse treatment on the trained memristor is higher than the threshold current immediately after the training is completed; For the trained memristor, after a certain period of time, the high-voltage pulse and low-voltage pulse processing are cycled again for m rounds to obtain a trained memristor; and m is less than or equal to n.
Citation Information
Cited By
Self-rectification memristor with adaptivity and preparation method thereof
CN120813234A
Heterojunction-based multi-mode photoelectric synapse device and preparation method thereof
CN121646284A
A heterojunction-based multi-modal optoelectronic synapse device and a preparation method thereof
CN121646284B
