A resistive switching layer, a memristor, its fabrication method and application

By using Nb18W16O93 resistive switching layer and optimizing the fabrication process, a memristor with higher stability and on/off ratio was fabricated, solving the problem of poor stability of memristors and making it suitable for non-volatile storage and neuromorphic computing.

CN118541018BActive Publication Date: 2025-10-28GUANGDONG UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410143021.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-02-01
Publication Date
2025-10-28
Estimated Expiration
2044-02-01

AI Technical Summary

Technical Problem

Existing memristors have poor stability, which affects their application in high-density storage and neuromorphic computing.

Method used

Nb18W16O93 was used as the resistive switching layer, and memristors were fabricated by inkjet printing technology. Combined with Ag electrodes and FTO thin film substrates, the fabrication process was optimized to improve stability.

Benefits of technology

The stability and on/off ratio of memristors have been improved, enabling them to be better applied to non-volatile storage and neuromorphic computing, and are suitable for fields such as electronic skin and brain-computer interfaces.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118541018B_ABST
    Figure CN118541018B_ABST
Patent Text Reader

Abstract

This invention relates to the field of memristor technology, and particularly to a resistive switching layer, a memristor, its fabrication method, and its applications. A resistive switching layer comprises: Nb 18 W 16 O 93 This resistive switching layer, used to fabricate memristors, exhibits excellent resistive switching characteristics and stability. This allows the fabricated memristors to be connected to resistors, capacitors, and transistors, enabling the further construction of neuronal circuits. This provides a simple and effective strategy for developing tunable artificial neurons and next-generation neuromorphic hardware. It is applicable to non-volatile memory and can be used in the fields of electronic skin or brain-computer interfaces.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of memristor technology, and in particular to a resistive switching layer, a memristor, its preparation method and application. Background Technology

[0002] While traditional memory performs well in its respective applications, it also has some drawbacks that affect its performance in specific application scenarios. Furthermore, with the rapid development of information technology, the demand for stable, highly integrated, fast, and versatile memory is increasing daily. For example, the amount of digital information in fields such as market analysis, aerospace, and artificial intelligence is growing exponentially, requiring continuous improvement in information processing and storage methods. In addition, current computers operate based on the traditional mainstream von Neumann architecture, with the central processing unit (CPU) and information storage system separated, which limits performance improvements. Therefore, to meet the ever-increasing demands for information storage, there is an urgent need to develop a new type of memory device with high-speed operation, high-density storage, and non-volatility. Memristors are a novel non-volatile memory technology. Their core feature is utilizing the resistance of oxide materials to store information based on changes in an electric field. Due to this resistance variation effect, they possess characteristics of non-volatility, high density, fast read / write speeds, long lifespan, and strong applicability, enabling them to meet the ever-increasing demands for information storage. Because a memristor is a nonlinear resistive device with memory function, composed of an electronic conductor / insulator / electronic conductor, it typically has two resistance states: a high resistance state (HRS) and a low resistance state (LRS). It can be used to simulate the "0" and "1" in binary arithmetic to store information. Furthermore, due to its simple structure, ease of integration, and similarity to the synaptic structure of human brain neurons, it has significant advantages in constructing neuromorphic computing devices and upgrading storage devices. Therefore, to date, the development of image recognition neural network models, represented by artificial neural networks (ANNs) and convolutional neural networks (CNNs), has largely utilized memristors.

[0003] However, existing memristors still suffer from poor stability. Summary of the Invention

[0004] In view of this, the present invention provides a resistive switching layer, a memristor, a method for fabricating the same, and its application, to solve the problem of poor stability of memristors in the prior art.

[0005] The first aspect of this invention provides a resistive switching layer for a memristor, comprising: Nb 18 W 16 O 93 .

[0006] A second aspect of the present invention provides a method for fabricating a resistive switching layer of a memristor, comprising:

[0007] Ammonium metatungstate, niobium oxalate, water and ethanol are mixed to dissolve the ammonium metatungstate and niobium oxalate. Citric acid is then added and the mixture is stirred until homogeneous to obtain the resistive switching ink.

[0008] The resistive layer is obtained by inkjet printing using the resistive layer ink.

[0009] It should be noted that citric acid is used as a complexing agent in this invention.

[0010] Preferably, the stirring and mixing time is 2 hours and the stirring and mixing temperature is 60°C.

[0011] A third aspect of the present invention provides a memristor, comprising: the resistive switching layer described above or the resistive switching layer prepared by the above preparation method.

[0012] Preferably, a top electrode is provided on the top of the resistive switching layer, and the top electrode is an Ag electrode;

[0013] A bottom electrode is provided below the resistive switching layer, and the bottom electrode is an FTO thin film; a substrate is provided below the bottom electrode, and the substrate is an FTO glass with the bottom electrode circuit etched on it.

[0014] Specifically, the substrate is a glass substrate coated with an FTO thin film, where FTO is fluorine-doped tin oxide.

[0015] It is understood that the memristor of the present invention comprises, from bottom to top, a substrate, a bottom electrode, a resistive switching layer, and a top electrode.

[0016] A fourth aspect of the present invention provides a method for fabricating a memristor, characterized by comprising the following steps:

[0017] Step 1: Clean the worktable and substrate of the nanomaterial inkjet printing system and adjust the parameters;

[0018] Step 2: Place the resistive switching layer ink from the above preparation method into the ink cartridge, and inkjet print the resistive switching layer on the FTO glass with the bottom electrode circuit etched on it.

[0019] Step 3: After inkjet printing is complete, place the inkjet printer on a heating plate to dry, and then anneal it in a muffle furnace;

[0020] Step 4: Put silver ink into the ink cartridge, inkjet print silver electrodes on the resistive switching layer, and heat and dry them on a heating plate.

[0021] Preferably, the method for cleaning the worktable and substrate of the nanomaterial inkjet printing system is: cleaning with a lint-free cloth soaked in isopropyl alcohol or cleaning with an air gun.

[0022] Specifically, the nanomaterial inkjet printing system of the present invention is the Sonoplot nanomaterial inkjet printing system.

[0023] Preferably, in step 1, the parameters are: the printhead voltage of the Sonoplot nanomaterial inkjet printing system is 5V-10V, and the printhead moving speed is 500-2000μm / s.

[0024] More preferably, in step 1, the parameters are: the printhead voltage of the Sonoplot nanomaterial inkjet printing system is 5V-10V, and the printhead moving speed is 1000μm / s.

[0025] Preferably, in step 3, the drying temperature of the heating plate is 150℃~200℃, and the annealing temperature is 600℃~700℃.

[0026] Preferably, in step 4, the method for preparing the silver ink includes:

[0027] Polyvinylpyrrolidone was added to ethanol and stirred until all the polyvinylpyrrolidone was dissolved. Then silver powder was added, stirred and ultrasonically dispersed to obtain a mixed solution. The precipitate in the mixed solution was filtered to obtain silver ink.

[0028] Specifically, the silver powder is nano-silver powder.

[0029] Preferably, in step 4, the heating temperature of the heating plate is 90–110°C.

[0030] More preferably, in step 4, the heating temperature of the heating plate is 100°C.

[0031] The fifth aspect of the present invention provides an application of the memristor described above or prepared by the above method in a non-volatile memory.

[0032] As can be seen from the above technical solutions, the present invention has the following advantages:

[0033] This invention provides a resistive switching layer, comprising: Nb 18 W 16 O 93 The resistive switching layer used to fabricate memristors exhibits excellent resistive switching characteristics and stability. Furthermore, when connected to resistors, capacitors, and transistors, the memristor can be further used to build neuronal circuits. This provides a simple and effective strategy for developing tunable artificial neurons and next-generation neuromorphic hardware. It is applicable to non-volatile memory and can be used in the fields of electronic skin or brain-computer interfaces. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.

[0035] Figure 1 This is a schematic diagram of a memristor structure provided in an embodiment of the present invention;

[0036] Figure 2 Nb provided for embodiments of the present invention 18 W 16 O 93 XRD pattern of the thin film;

[0037] Figure 3 Nb is provided for the embodiments of the present invention. 18 W 16 O 93 SEM image of the thin film;

[0038] Figure 4 IV curves of memristors after annealing at 600, 650, and 700°C, provided for embodiments of the present invention;

[0039] Figure 5 The IV curves of the memristor under forward and reverse bias sweeps (0 to ±2.0V) provided in the embodiments of the present invention;

[0040] Figure 6 The LTP and LTD behavior diagrams of the memristor provided in the embodiments of the present invention under ±1V, 500ms pulse.

[0041] In the figure, 1 is the top electrode, 2 is the resistive switching layer, 3 is the bottom electrode, and 4 is the substrate. Detailed Implementation

[0042] To date, the development of image recognition neural network models, represented by Artificial Neural Networks (ANNs) and Convolutional Neural Networks (CNNs), has largely relied on memristors. However, these experimental results are severely hampered by the inherent problems of memristors: on the one hand, electrical formation, inter-device variations, and nonlinear modulation of conductance are significant issues, with traditional memristors typically depending on the formation and breakage of the conductive filament in the amorphous thin film of the RS (Resonant Wire). On the other hand, stray conduction channels are difficult to control, leading to large spatial and temporal deviations in RS performance, as well as poor memory effects, nonlinearity, and asymmetric weight modulation variations. This non-ideal update of conductance variation is detrimental to high-performance neuromorphisms. Since the linearity and symmetry of memristor conductance modulation are two key features for improving computational accuracy, the recognition results are often poor when dealing with noisy image classification tasks, making it difficult to adapt to real-world situations such as vehicle speed and license plate recognition in harsh environments (e.g., smog and sandstorms). Therefore, it is necessary to develop a highly stable memristor to circumvent the problems of switching variability and nonlinearity.

[0043] This invention provides a resistive switching layer, a memristor, its fabrication method, and its application, which solves the problem of poor stability of memristors in the prior art.

[0044] The technical solutions in the embodiments of this application will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0045] In the following examples, all raw materials or reagents used are commercially available or self-made.

[0046] Example 1

[0047] This embodiment provides a resistive switching layer, and the preparation steps are as follows:

[0048] (1) Preparation of Nb 18 W 16 O 93 The resistive switching ink is prepared from niobium oxalate, ammonium metatungstate and citric acid. 0.20 g of ammonium metatungstate and 0.49 g of niobium oxalate are dissolved in 2 mL of deionized water (18.2 MΩ·cm, Millipore Milli-Q ultrapure water purification system) and 8 mL of anhydrous ethanol. The mixture is continuously stirred magnetically until it is completely dissolved. Then, 1.0 g of citric acid is added to the solution.

[0049] The resistive switching layer ink is placed in an ink cartridge and inkjet printed onto an FTO glass substrate with etched bottom electrode circuitry to obtain the resistive switching layer, i.e., Nb. 18 W 16 O 93resistive switching layer (or Nb) 18 W 16 O 93 film).

[0050] Example 2

[0051] This embodiment provides a first type of memristor, the fabrication method of which is as follows:

[0052] Step 1: Clean the stage and substrate of the Sonoplot nanomaterial inkjet printing system with a lint-free cloth soaked in isopropyl alcohol or with an air gun. Adjust the relevant parameters and proceed with printing. The parameters are as follows: printhead voltage of the Sonoplot nanomaterial inkjet printing system is 5V~10V, printhead movement speed is 1000μm / s, and printhead diameter is 30μm~50μm.

[0053] Step 2: Place the resistive switching layer ink into the ink cartridge and inkjet print the resistive switching layer onto the FTO glass with the bottom electrode circuit etched on it. The preparation method of the resistive switching layer ink is the same as in Example 1.

[0054] Step 3: After inkjet printing is complete, place the inkjet printer on a heating plate and dry it at 150°C to 200°C; then place it in a muffle furnace and keep it at 600°C for 2 hours, followed by air annealing.

[0055] Step 4: Place silver ink into an ink cartridge, inkjet print silver electrodes onto a niobium tungsten oxide thin film layer, and heat and dry at 100°C on a hot plate to obtain a memristor. The Ag ink preparation method is as follows: add polyvinylpyrrolidone (PVP) to ethanol and stir until all PVP is dissolved. Then add nano-silver powder and stir for 15 minutes. Place the ink in an ultrasonic cleaner and ultrasonically disperse for 15 minutes. Repeat the stirring and ultrasonic dispersion process 3 to 4 times. Finally, filter out the precipitate from the ink using a needle filter to obtain nano-silver ink.

[0056] Among them, memristors are such Figure 1 As shown, from top to bottom, it includes: top electrode 1, resistive switching layer 2, bottom electrode 3, and substrate 4; for ease of subsequent application, memristors are generally printed in an array, i.e., memristor array.

[0057] Example 3

[0058] This embodiment provides a second type of memristor, which is prepared in the same way as in embodiment 2, except that the heat preservation temperature in step 3 is 650°C.

[0059] Example 4

[0060] This embodiment provides a third type of memristor, which is prepared in the same way as in embodiment 2, except that the heat preservation temperature in step 3 is 700°C.

[0061] Test Case

[0062] (1) In this test case, the resistive switching layer prepared in Example 1 was subjected to XRD and SEM analysis. The results are as follows: Figures 2-3 .

[0063] like Figure 2 As shown, according to Nb 18 W 16 O 93 XRD analysis of the thin film revealed the presence of Nb after calcination. 18 W 16 O 93 The two main peaks of the thin film, at approximately 22.49° and 45.90°, correspond to the diffraction of the orthorhombic phase (JCPDS 75-0561).

[0064] like Figure 3 As shown, Figure 3 Part (a) shows the surface of the thin film, where Nb is visible. 18 W 16 O 93 The film is smooth and uniform, with no obvious cracks; Figure 3 Part (b) is a high-magnification SEM image, showing Nb. 18 W 16 O 93 Crystallization phase, proving Nb 18 W 16 O 93 The crystalline particles are cylindrical; Figure 3 The middle (c) part is Nb with a thickness of approximately 900 nm. 18 W 16 O 93 Cross-sectional SEM images of the thin film, showing Nb 18 W 16 O 93 The film is uniformly deposited on the FTO surface and adheres firmly.

[0065] (2) This test examined the current-voltage (IV) characteristics, LTP and LTD behavior under a 500ms pulse, of the memristors prepared in Examples 2-4. The results are as follows: Figures 4-6 .

[0066] Electrical measurements were performed on the device based on the RS behavior of the memristor at different manufacturing temperatures, such as... Figure 4 As shown, the device is subjected to a DC voltage applied to the top electrode while the bottom electrode is grounded. Figure 4The 20-cycle current-voltage (IV) characteristics of the device are shown. By applying a positive scan voltage from 0 to +2.5V to 0V, the device undergoes a transition from a high-resistivity state (HRS) to a low-resistivity state (LRS), while a negative scan voltage from 0 to -3V to 0V returns the device to HRS from LRS. This indicates that higher annealing temperatures can significantly enhance the on / off ratio of the device, demonstrating a strong positive correlation between the on / off ratio and crystal quality, but a negative correlation with stability. Specifically, the cycle stability is better at 600℃, but the on / off ratio is smaller. The calculation of the 0.2V current ratio at 700℃ indicates that Ag / Nb... 18 W 16 O 93 / FTO devices exhibit a significant on / off ratio >10 3 However, its stability is relatively poor.

[0067] Biological synapses connect the functions of two neurons, enabling information transmission. To simulate the enhancement and inhibition of biological synapses, devices need to exhibit changes in conductivity under voltage scanning, a phenomenon known as synaptic plasticity. Figure 5 As shown in section (a), when a negative bias voltage (-2V) is applied to the top electrode Ag for cyclic scanning, the device's conductivity tends to decrease, similar to the sustained inhibition of synaptic weight when a biological synapse is subjected to continuous stimulation. Figure 5 As shown in section (b), when a positive bias voltage (+2V) is applied to the top electrode Ag for cyclic scanning, the conductivity of the device tends to increase, similar to the process of synaptic weight enhancement, indicating that the memristor prepared by the present invention has biological synapses and is easy to apply.

[0068] like Figure 6 The results show that the memristor exhibits LTP and LTD behavior under ±1V, 500ms pulses, indicating that the memristor of this invention has great potential in simulating complex synaptic plasticity mechanisms in biological systems.

[0069] In summary, the memristor prepared by this invention has the following advantages:

[0070] (I) The present invention uses niobium tungsten oxide, which is a binary system of niobium oxide and tungsten oxide. Niobium tungsten oxide is used as a resistive switching layer. Compared with niobium oxide or tungsten oxide alone, it has more oxygen vacancies and is easier to form conductive wires. The easy formation of conductive wires is beneficial to reducing the set voltage, reducing power consumption, and improving the stability of memristors.

[0071] (ii) The memristor provided by this invention can be connected with resistors, capacitors and transistors to further build neuron circuits, providing a simple and effective strategy for developing tunable artificial neurons and next-generation neuromorphic hardware, and can be applied to fields such as electronic skin and brain-computer interfaces.

[0072] (III) The memristor provided by this invention has a smaller feature size compared to a single device, which can significantly improve the switching characteristics and stability of the resistive switching layer, enabling the niobium oxide-based Mott memristor to be better applied to neuromorphic computing, and thus to simulate neurons, replacing neurons of traditional complementary metal-oxide-semiconductor technology, etc., providing an effective strategy for the next generation of neuromorphic hardware.

[0073] (iv) The memristor provided by this invention can be applied in simulating human brain neurons. In one specific embodiment, by covering the bottom electrode layer with a layer of niobium oxide to form a resistive switching layer, the growth of conductive filaments is facilitated; in another specific embodiment, based on the conductive filament model of the resistive switching device, niobium tungsten oxide improves the conductivity and stability of the resistive switching layer and the entire memristor; in yet another specific embodiment, due to the threshold switching characteristics of the memristor, it can be used to construct neuromorphic circuits.

[0074] The foregoing has provided a detailed description of the resistive switching layer, memristor, and their fabrication method and application provided by the present invention. For those skilled in the art, based on the ideas of the embodiments of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A resistive switching layer, characterized in that, include: No 18 W 16 O 93 .

2. A method for preparing a resistive switching layer, characterized in that, include: Ammonium metatungstate, niobium oxalate, water and ethanol are mixed to dissolve the ammonium metatungstate and niobium oxalate. Citric acid is then added and the mixture is stirred until homogeneous to obtain the resistive switching ink. The resistive layer is obtained by inkjet printing using the resistive layer ink.

3. A memristor, characterized in that, include: The resistive switching layer according to claim 1 or the resistive switching layer prepared by the preparation method according to claim 2.

4. The memristor according to claim 3, characterized in that, A top electrode is provided on the resistive switching layer, and the top electrode is an Ag electrode. A bottom electrode is provided below the resistive switching layer, and the bottom electrode is an FTO thin film; a substrate is provided below the bottom electrode, and the substrate is an FTO glass with the bottom electrode circuit etched on it.

5. A method for fabricating a memristor, characterized in that, Includes the following steps: Step 1: Clean the stage and substrate of the nanomaterial inkjet printing system and adjust the relevant parameters; Step 2: Place the resistive switching layer ink from the preparation method described in claim 2 into an ink cartridge, and inkjet print the resistive switching layer on the FTO glass with the bottom electrode circuit etched on it. Step 3: After inkjet printing is complete, place the inkjet printer on a heating plate to dry, and then anneal it in a muffle furnace; Step 4: Put silver ink into the ink cartridge, inkjet print silver electrodes on the resistive switching layer, and heat and dry them on a heating plate.

6. The method for fabricating a memristor according to claim 5, characterized in that, In step 1, the parameters are: the printhead voltage of the nanomaterial inkjet printing system is 5V to 10V, and the printhead moving speed is 500 to 2000μm / s.

7. The method for fabricating a memristor according to claim 5, characterized in that, In step 3, the drying temperature of the heating plate is 150℃~200℃, and the annealing temperature is 600℃~700℃.

8. The method for fabricating a memristor according to claim 5, characterized in that, Step 4, the method for preparing the silver ink, includes: Polyvinylpyrrolidone was added to ethanol and stirred until all the polyvinylpyrrolidone was dissolved. Then silver powder was added, stirred and ultrasonically dispersed to obtain a mixed solution. The precipitate in the mixed solution was filtered to obtain silver ink.

9. The method for fabricating a memristor according to claim 5, characterized in that, In step 4, the heating temperature of the heating plate is 90-110℃.

10. The memristor according to any one of claims 3 to 4 or the memristor prepared by the preparation method according to any one of claims 5 to 9 is used in a non-volatile memory.