Neuromorphic device with high rectification ratio and high selectivity and preparation method thereof

By using energy band regulation and metal work function modulation of TiO2, NbOx and TiN/Ru materials in neuromorphic devices, the crosstalk leakage current problem is solved, high rectification ratio and selectivity is achieved, array design is simplified and cost is reduced. It also has the function of simulating artificial synapses.

CN118922060APending Publication Date: 2024-11-08FUDAN UNIVERSITY +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410974683.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

In solving the crosstalk leakage current problem of neuromorphic devices, the prior art needs to introduce additional elements, resulting in reduced array density and increased manufacturing complexity, and unable to effectively improve the rectifier ratio and selectivity.

Method used

By using TiO2 and NbOx as dielectric layer materials in neuromorphic devices, combined with TiN and Ru as electrode materials, using energy band regulation and metal work function modulation to adjust the rectifier ratio and nonlinearity of the device to achieve forward conduction and reverse The direction-off characteristic is suppressed, thereby suppressing leakage current.

Benefits of technology

It realizes high rectification ratio and high selectivity, effectively suppresses array leakage current, simplifies array design and manufacturing processes, reduces costs, and has the function of simulating artificial synapses.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118922060A_ABST
    Figure CN118922060A_ABST
Patent Text Reader

Abstract

The invention provides a neuromorphic device with a high rectification ratio and high selectivity and a preparation method thereof. The neuromorphic device comprises a substrate from bottom to top; a bottom electrode layer; a first dielectric layer; a second dielectric layer; a top electrode layer; through the energy band regulation and control of the first dielectric layer and the second dielectric layer and the modulation effect of the metal work function of the bottom electrode layer and the top electrode layer, the current-voltage characteristic of the device can realize forward conduction and reverse cut-off, so that an ultrahigh rectification ratio is obtained, the conduction when the forward small voltage is cut off and the large voltage is cut off is realized, and the rectification ratio of the device is improved. Therefore, the ultrahigh nonlinearity is obtained, and the effect of suppressing the leakage current of the array is achieved. On the basis of the intrinsic characteristic of suppressing leakage current, the device can also realize neurosynaptic morphological characteristics and functions, such as paired pulse promotion, excitatory post-synaptic current and long-term enhancement / suppression.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of integrated circuit manufacturing, and in particular to a neuromorphic device with high rectification ratio and high selectivity and a preparation method thereof. Background Art

[0002] The brain contains approximately 1011 neurons and 1015 synapses. When building a neural network, achieving synaptic performance on a single device is far from enough to build a neural network, so high-density array integration is needed to expand artificial neural networks. However, as the array density increases, the crosstalk problem becomes increasingly prominent, which limits the further integration of high-density neuromorphic devices. The crosstalk problem mainly refers to the signal interference between different devices in array integration due to factors such as electromagnetic coupling and capacitive coupling between signal lines. This interference will destroy the integrity and stability of the signal, thereby affecting the performance and accuracy of the neural network.

[0003] A lot of research work has been devoted to solving the crosstalk leakage problem, for example, 1T1M (one transistor one memory), 1D1M (one diode one memory) and 1S1M (one selector one memory), etc. Although these methods can alleviate the crosstalk problem to a certain extent, they all require the introduction of additional elements (for example, transistors, diodes or selectors, etc.), which will reduce the density of the array and increase the complexity of manufacturing and layout.

[0004] In view of this, it is necessary to propose a neuromorphic device with high rectification ratio and high selectivity and a preparation method thereof to solve the above problems. Summary of the invention

[0005] The object of the present invention is to provide a neuromorphic device with high rectification ratio and high selectivity and a preparation method thereof, so as to improve the problem that the existing method for solving crosstalk leakage current will reduce the density of the array and increase the complexity of manufacturing and layout.

[0006] In a first aspect, the present invention provides a neuromorphic device with high rectification ratio and high selectivity, comprising from bottom to top:

[0007] substrate;

[0008] bottom electrode layer;

[0009] a first dielectric layer;

[0010] a second dielectric layer;

[0011] top electrode layer;

[0012] The rectification ratio and nonlinearity of the device can be adjusted by regulating the energy bands of the first dielectric layer and the second dielectric layer and modulating the metal work functions of the bottom electrode layer and the top electrode layer, thereby suppressing leakage current.

[0013] In a possible embodiment, the material of the first dielectric layer is TiO2; and / or,

[0014] The thickness of the first dielectric layer is 20 nm.

[0015] In a possible embodiment, the material of the second dielectric layer is NbO x and / or,

[0016] The thickness of the second dielectric layer is 20 nm.

[0017] In a possible embodiment, the material of the bottom electrode layer is TiN; and / or,

[0018] The thickness of the bottom electrode layer is 70 nm.

[0019] In a possible embodiment, the material of the top electrode layer is Ru; and / or,

[0020] The thickness of the top electrode layer is 70 nm.

[0021] In a second aspect, the present invention further provides a method for preparing a neuromorphic device with high rectification ratio and high selectivity, comprising the following steps:

[0022] S210: providing a substrate;

[0023] S220: forming a bottom electrode layer on the substrate;

[0024] S230: forming a first dielectric layer on the bottom electrode layer;

[0025] S240: forming a second dielectric layer on the first dielectric layer;

[0026] S250: forming a top electrode layer on the second dielectric layer.

[0027] In a possible embodiment, forming a bottom electrode layer on the substrate includes:

[0028] A bottom electrode layer is formed on the substrate by using a physical vapor deposition process, wherein the material of the bottom electrode layer is TiN, and the thickness of the bottom electrode layer is 70 nm.

[0029] In a possible embodiment, forming a first dielectric layer on the bottom electrode layer includes:

[0030] According to the set reaction temperature and the set cycle rate, a plasma atomic layer deposition process is used for cyclic deposition until a first dielectric layer of a set thickness is formed on the bottom electrode layer, wherein the set reaction temperature is 250°C, the set cycle rate is 0.1nm / cycle, the total number of cycles is 200 times, and the thickness of the first dielectric layer is 20nm.

[0031] In a possible embodiment, forming a second dielectric layer on the first dielectric layer includes:

[0032] A second dielectric layer is formed on the first dielectric layer using a physical vapor deposition process, wherein the material of the second dielectric layer is NbO x , the thickness of the second dielectric layer is 20 nm.

[0033] In a possible embodiment, forming a top electrode layer on the second dielectric layer includes:

[0034] A top electrode layer is formed on the second dielectric layer by adopting a metal lift-off process or a hard mask process and using a physical vapor deposition process, wherein the material of the top electrode layer is Ru, and the thickness of the top electrode layer is 70 nm.

[0035] The beneficial effect of the neuromorphic device with high rectification ratio and high selectivity provided by the present invention is that: through the energy band regulation of the first dielectric layer and the second dielectric layer and the modulation of the metal work function of the bottom electrode layer and the top electrode layer, the current-voltage characteristics of the device can achieve forward conduction and reverse cutoff, thereby obtaining an ultra-high rectification ratio, and achieve cutoff when the forward voltage is small and conduction when the voltage is large, thereby obtaining ultra-high nonlinearity, and then achieving the effect of suppressing the leakage current of the array. On the basis of having the intrinsic characteristics of suppressing the leakage current, the device can also realize the morphological characteristics and functions of the neural synapse including paired pulse facilitation, excitatory postsynaptic current and long-term potentiation / depression. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 Schematic diagram of a neuromorphic device with high rectification ratio and high selectivity according to the present invention.

[0037] Figure 2 The present invention is a flow chart of a method for preparing a neuromorphic device with high rectification ratio and high selectivity.

[0038] Figure 3 A schematic diagram of a substrate is provided in the method for preparing a neuromorphic device with high rectification ratio and high selectivity of the present invention.

[0039] Figure 4It is a schematic diagram of forming a bottom electrode layer on a substrate in the method for preparing a neuromorphic device with high rectification ratio and high selectivity of the present invention.

[0040] Figure 5 It is a schematic diagram of forming a first dielectric layer on a bottom electrode layer in the method for preparing a neuromorphic device with a high rectification ratio and high selectivity of the present invention.

[0041] Figure 6 It is a schematic diagram of forming a second dielectric layer on a first dielectric layer in the method for preparing a neuromorphic device with a high rectification ratio and high selectivity of the present invention.

[0042] Figure 7 It is a schematic diagram of forming a top electrode layer on a second dielectric layer in the method for preparing a neuromorphic device with a high rectification ratio and high selectivity of the present invention.

[0043] Figure 8 The voltage-current relationship diagram for different electronic components.

[0044] Description of reference numerals: substrate 110 ; bottom electrode layer 120 ; first dielectric layer 130 ; second dielectric layer 140 ; ​​top electrode layer 150 . DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Unless otherwise defined, the technical terms or scientific terms used herein should be understood by people with general skills in the field to which the present invention belongs. "Including" and similar words used in this article mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects.

[0046] The specific implementation modes of the present invention will be further described in detail below in conjunction with the accompanying drawings.

[0047] See also Figure 1 , Figure 1 FIG. 1 is a schematic diagram of a neuromorphic device with high rectification ratio and high selectivity according to the present invention. Figure 1As shown, the embodiment of the present invention provides a neuromorphic device with high rectification ratio and high selectivity, which includes from bottom to top: substrate 110, bottom electrode layer 120, first dielectric layer 130, second dielectric layer 140, top electrode layer 150, and the rectification ratio and nonlinearity of the device can be adjusted by regulating the energy band of the first dielectric layer 130 and the second dielectric layer 140 and modulating the metal work function of the bottom electrode layer 120 and the top electrode layer 150, thereby suppressing the leakage current. It is turned on under forward voltage and cut off under reverse voltage, thereby obtaining an ultra-high rectification ratio, and it is cut off when the forward voltage is lower than the first threshold and turned on when the forward voltage is higher than the second threshold, thereby obtaining an ultra-high nonlinearity, and then achieving the effect of suppressing the leakage current of the array, wherein the first threshold is less than the second threshold, the first threshold is a small forward voltage, and the second threshold is a large forward voltage. Preferably, the substrate 110 can be SiO2 (silicon dioxide) or Si (silicon), etc.

[0048] Band modulation is to control the transmission characteristics of electrons by adjusting the band structure of the material. The band structure of the dielectric layer in the device determines the transmission efficiency of electrons in it. The work function (also known as the work function) is a physical quantity that describes the minimum energy required for electrons to escape from the metal surface. It reflects the energy difference between the electrons inside the metal and the surface barrier. Modulation of the metal work function can change the height of the interface barrier, thereby affecting the current-voltage (IV) characteristics of the device, that is, changing the transmission efficiency of electrons at the interface. Reducing the work function of the metal electrode can reduce the height of the interface barrier under the forward voltage, making it easier for electrons to be injected from the electrode into the dielectric layer, thereby enhancing the forward conduction characteristics. Increasing the work function of the metal electrode can increase the height of the interface barrier under the reverse voltage, further preventing the transmission of electrons and enhancing the reverse cutoff characteristics. In this solution, through the energy band regulation of the first dielectric layer 130 and the second dielectric layer 140 and the modulation of the metal work function of the bottom electrode layer 120 and the top electrode layer 150, it can be achieved that: when a forward voltage is applied, the energy band structure of the dielectric layer allows electrons to pass smoothly, so that the device presents a low-resistance state and current can pass more easily; when a reverse voltage is applied, the energy band structure of the dielectric layer changes, forming a higher potential barrier to prevent the transmission of electrons, so that the device presents a high-resistance state, and current can hardly pass through, that is, the characteristics of forward conduction and reverse cutoff are achieved, thereby obtaining an ultra-high rectification ratio. When the forward voltage is small, the device is in the cutoff state; when the forward voltage is large, the device begins to conduct, and as the voltage further increases, the current will increase sharply, that is, the forward voltage is cutoff at a small voltage and is turned on at a large voltage, thereby obtaining an ultra-high nonlinearity. As Figure 8 As shown, this design makes the current-voltage relationship of the device highly nonlinear, that is, the current increases sharply with the increase of voltage.

[0049] In a preferred embodiment, the material of the first dielectric layer 130 is TiO2 (titanium dioxide); and / or, the thickness of the first dielectric layer 130 is 20 nm. In this embodiment, TiO2 has a high dielectric constant and good optical properties, which helps to achieve a higher capacitance density at a smaller physical size. TiO2 is also an excellent insulating material with low leakage current and high breakdown voltage. TiO2 shows good stability to many chemicals and environmental conditions, which enables it to maintain stable performance in various application scenarios and is not easily affected by external factors.

[0050] In a preferred embodiment, the material of the second dielectric layer 140 is NbO x (niobium oxide); and / or, the thickness of the second dielectric layer 140 is 20nm. In this embodiment, NbOx has a higher dielectric constant, which helps to achieve a larger capacitance density while maintaining a smaller physical size. NbOx films generally have a lower leakage current, which means that under the action of an electric field, less charge leaks in the second dielectric layer 140, which helps to improve the energy efficiency and stability of electronic devices.

[0051] In a specific embodiment, the material of the bottom electrode layer 120 is TiN (titanium nitride); and / or, the thickness of the bottom electrode layer 120 is 70 nm. In this embodiment, TiN is a material with high electronic conductivity, which ensures that the bottom electrode layer 120 can effectively transmit current, and TiN has good chemical stability and thermal stability. TiN also has high hardness and wear resistance, which makes the bottom electrode layer 120 more stable and able to resist erosion and mechanical damage from the external environment. TiN has good adhesion to many materials, which is conducive to forming a good contact interface with the upper structure, reducing interface resistance, and improving device performance.

[0052] In a specific embodiment, the material of the top electrode layer 150 is Ru (ruthenium); and / or, the thickness of the top electrode layer 150 is 70nm. In this embodiment, Ru is a material with excellent electrical conductivity, which ensures that the top electrode layer 150 can effectively transmit current, thereby reducing resistance loss and improving the overall performance of the device. Ru has excellent chemical stability and thermal stability. Ru has a lower resistivity. Under the same conditions, a device using Ru as the top electrode layer 150 has lower resistance loss and higher energy efficiency.

[0053] The band structures of TiO2 and NbOx are adjustable. By controlling the preparation methods of the TiO2 dielectric layer and the NbOx dielectric layer, their band structures can be adjusted to meet the needs of different electronic devices. For example, by precisely controlling the deposition temperature and time, the crystal structure and band arrangement in the TiO2 dielectric layer and the NbOx dielectric layer can be affected. In addition, by controlling the deposition rate and thickness of the TiO2 dielectric layer and the NbOx dielectric layer, their band structures, including the width and position of the conduction band, valence band and forbidden band, can also be affected. Band regulation can affect the transmission characteristics of electrons in the dielectric layer, including tunneling effect, barrier height, etc. By regulating the band structure of TiO2 and NbOx, the characteristics of forward conduction and reverse cutoff can be achieved, thereby achieving an ultra-high rectification ratio (I@Vread / I@V-read>10 7 ). The rectification ratio represents the ratio of the device's current at the forward read voltage to the current at the reverse read voltage. Due to the effect of band modulation, the device has extremely low resistance at the forward voltage and extremely high resistance at the reverse voltage, so the rectification ratio is very high. By adjusting the band structure of TiO2 and NbOx, the characteristics of small forward voltage cutoff and large voltage conduction can be achieved, thereby achieving ultra-high nonlinearity (I@Vread / @V-1 / 2read). Nonlinearity represents the ratio of the device's current at the forward read voltage to the square root of the voltage. Since the device's resistance decreases sharply with increasing voltage (nonlinear change), the nonlinearity is also very high. When independently integrated in an array, the influence of leakage current can be effectively suppressed due to the ultra-high rectification ratio and nonlinearity. Even if there are voltage fluctuations or crosstalk in the array, the device can maintain stable storage and reading performance.

[0054] When the metal contacts the dielectric layer, due to the difference in work function, electrons will flow from the side with lower work function to the side with higher work function until the Fermi levels on both sides are equal. This charge transfer will form a Schottky barrier or ohmic contact at the interface, thereby affecting the injection and transmission efficiency of electrons. Under forward voltage, due to the modulation of the metal work function, electrons can be easily injected from the TiN bottom electrode layer 120 into the TiO2 dielectric layer and the NbOx dielectric layer, and transferred to the Ru top electrode layer 150 to achieve forward conduction. Under reverse voltage, due to the band regulation of the dielectric layer and the modulation of the metal work function, electrons are greatly hindered at the interface and it is difficult to inject from the Ru top electrode layer 150 into the dielectric layer to achieve reverse cutoff. Through the band regulation of the dielectric layer and the modulation of the metal work function, the two-terminal memristor of the TiN / TiO2 / NbOx / Ru structure can exhibit ultra-high rectification ratio and nonlinearity, thereby achieving the effect of suppressing array leakage current. In addition, the device can imitate the various functions of synapses in the human brain and be applied to the field of neuromorphic computing.

[0055] See also Figure 2 , Figure 2 FIG. 1 is a flow chart of a method for preparing a neuromorphic device with high rectification ratio and high selectivity according to the present invention. Figure 2 As shown, the present invention also provides a method for preparing a neuromorphic device with high rectification ratio and high selectivity, comprising the following steps:

[0056] S210: providing a substrate 110;

[0057] S220: forming a bottom electrode layer 120 on the substrate 110;

[0058] S230: forming a first dielectric layer 130 on the bottom electrode layer 120;

[0059] S240: forming a second dielectric layer 140 on the first dielectric layer 130;

[0060] S250 : forming a top electrode layer 150 on the second dielectric layer 140 .

[0061] See also Figure 3 , Figure 3 A schematic diagram of a substrate is provided in the method for preparing a neuromorphic device with high rectification ratio and high selectivity of the present invention. Figure 3 As shown, in a specific embodiment, providing a substrate 110 includes: providing a substrate 110, cleaning the substrate 110 using an RCA cleaning process, and drying the substrate 110 using N2. The RCA cleaning process can effectively remove pollutants such as organic matter and metal ions on the surface of the substrate 110. The RCA cleaning process generally includes multiple steps, such as deionized water rinsing, concentrated sulfuric acid treatment, hydrogen peroxide treatment, etc., to ensure the cleanliness of the surface of the substrate 110. After cleaning, the substrate 110 is dried using N2 to avoid the influence of residual moisture on subsequent processes.

[0062] See also Figure 4 , Figure 4 FIG. 1 is a schematic diagram of a method for preparing a neuromorphic device with a high rectification ratio and high selectivity according to the present invention after a bottom electrode layer is formed on a substrate. Figure 4 As shown, in one embodiment, forming a bottom electrode layer 120 on a substrate 110 includes: forming the bottom electrode layer 120 on the substrate 110 using a physical vapor deposition process, wherein the material of the bottom electrode layer 120 is TiN, and the thickness of the bottom electrode layer 120 is 70 nm. In this embodiment, forming a 70 nm thick TiN bottom electrode layer 120 on the substrate 110 using a physical vapor deposition process can bring about technical effects such as good conductivity, stability and adhesion, while maintaining low cost and complexity.

[0063] See also Figure 5 , Figure 5FIG. 1 is a schematic diagram of a method for preparing a neuromorphic device with a high rectification ratio and high selectivity according to the present invention after forming a first dielectric layer on the bottom electrode layer. Figure 5 As shown, in a possible embodiment, forming a first dielectric layer 130 on the bottom electrode layer 120 includes: according to a set reaction temperature and a set cycle rate, using a plasma atomic layer deposition process for cyclic deposition until a first dielectric layer 130 of a set thickness is formed on the bottom electrode layer 120, wherein the reaction temperature is set to 250°C, the cycle rate is set to 0.1nm / cycle, the total number of cycles is 200 times, and the thickness of the first dielectric layer 130 is 20nm. In this embodiment, by setting the cycle rate (0.1nm / cycle) and the total number of cycles (200 times), the thickness of the first dielectric layer 130 can be accurately controlled to reach the required 20nm. Due to the high controllability of the plasma atomic layer deposition technology, excellent film quality can be obtained, and a high-quality interface can be formed between the first dielectric layer 130 and the bottom electrode layer 120, reducing the interface resistance and leakage current, and improving the performance and reliability of the device.

[0064] See also Figure 6 , Figure 6 FIG. 1 is a schematic diagram of a method for preparing a neuromorphic device with a high rectification ratio and high selectivity according to the present invention after a second dielectric layer is formed on a first dielectric layer. Figure 6 As shown, in a possible embodiment, forming the second dielectric layer 140 on the first dielectric layer 130 includes: forming the second dielectric layer 140 on the first dielectric layer 130 using a physical vapor deposition process, wherein the material of the second dielectric layer 140 is NbO x The thickness of the second dielectric layer 140 is 20 nm. The band matching between the NbOx dielectric layer and the TiO2 dielectric layer is good, which is conducive to forming a high-quality interface and reducing interface resistance and leakage current.

[0065] See also Figure 7 , Figure 7 FIG. 1 is a schematic diagram of a method for preparing a neuromorphic device with a high rectification ratio and high selectivity according to the present invention after a top electrode layer is formed on a second dielectric layer. Figure 7 As shown, in some possible embodiments, forming a top electrode layer 150 on the second dielectric layer 140 includes: using a metal lift-off process or a hard mask process and using a physical vapor deposition process to form the top electrode layer 150 on the second dielectric layer 140, wherein the material of the top electrode layer 150 is Ru, and the thickness of the top electrode layer 150 is 70nm. Selecting Ru as the material of the top electrode layer 150 can form a good interface contact with the NbOx second dielectric layer 140 below, reduce interface resistance and leakage current, and improve device performance and reliability.

[0066] In one embodiment, a photoresist layer is formed on the second dielectric layer 140. The photoresist layer is exposed using a photolithography mask so that the portion of the photoresist layer corresponding to the electrode region is exposed, and the photoresist of the exposed portion is dissolved using a developer. A top electrode layer 150 is grown on the surface of the device structure using a physical vapor deposition process. The photoresist is dissolved using a metal lift-off process, and the top electrode layer 150 deposited on the photoresist is lifted off together. This process can retain the electrode metal pattern formed by the top electrode layer 150 on the electrode region.

[0067] In another embodiment, a hard mask layer is formed on the second dielectric layer 140, a photoresist layer is formed on the hard mask layer, and the photoresist layer is exposed using a photolithography mask, and the exposed portion of the photoresist is dissolved using a developer. The hard mask layer is etched using an etching technique, and due to the protective effect of the photoresist, only the portion of the hard mask not covered by the photoresist will be etched away, while the portion of the hard mask under the photoresist is retained. After the hard mask etching is completed, the residual photoresist is removed using an appropriate solvent to expose the pattern on the hard mask layer. A top electrode layer 150 is grown on the surface of the above-mentioned device structure using a physical vapor deposition process. After the physical vapor deposition process is completed, the hard mask material is removed by etching or a stripping process.

[0068] The technical effects of the neuromorphic device with high rectification ratio and high selectivity and the preparation method thereof of the present invention are explained in detail below.

[0069] 1. The two-terminal memristor based on the TiN / TiO2 / NbOx / Ru structure can realize the forward conduction and reverse cutoff of the IV characteristics of the device through the band regulation of the dielectric layer and the modulation of the metal work function, thereby obtaining an ultra-high rectification ratio, and cutoff at a small forward voltage and conduction at a large voltage, thereby obtaining an ultra-high nonlinearity, thereby realizing the characteristics of independent integration in the array without being affected by leakage current, effectively alleviating the crosstalk problem, and making the device theoretically reach a maximum capacity of 1Tb.

[0070] 2. The device also has high selectivity. The TiO2 dielectric layer and the NbOx dielectric layer have good ion conductivity or electron blocking properties. The TiN bottom electrode layer 120 and the Ru top electrode layer 150 have good conductivity and stability. The combination of these materials can form a complex interface and energy band structure, providing unique electrical properties for the memristor. Under specific voltage conditions, it exhibits specific resistance change behavior, making the memristor have higher sensitivity and stability under specific conditions.

[0071] 3. The device also has a self-rectifying property, which allows current to flow in only one direction, helping to reduce reverse current in the array and further improve the stability and reliability of the array.

[0072] 4. The device also has a self-selection feature, that is, it can selectively respond to a specific input signal without the need for an external selector (e.g., a transistor). This greatly simplifies the design and manufacturing process of the array and reduces costs.

[0073] 5. The device also has a reversible resistance change characteristic, that is, its resistance state can be switched between high and low states, and this switching is reversible, so it can be used as a non-volatile memory, and the stored information will not be lost even when the power is turned off.

[0074] 6. Since the device has self-rectification and self-selection characteristics, no additional nonlinear elements are required in the array (for example, selectors and rectifiers in 1S1R or 1T1R structures). This not only simplifies the design and manufacturing process of the array, but also reduces the cost of hardware design.

[0075] 7. The TiN / TiO2 / NbOx / Ru structured memristor is itself a complete device unit and can be directly integrated on a substrate to form an array, further simplifying the process flow.

[0076] 8. The ultra-high rectification ratio can significantly reduce the leakage current of the reverse biased units in the array.

[0077] 9. Ultra-high nonlinearity can significantly reduce the leakage current of the forward biased unit in the array.

[0078] 10. In addition to the intrinsic property of suppressing leakage current, the device can also effectively simulate the functions of artificial synapses under the stimulation of electrical pulse signals, including paired pulse facilitation, excitatory postsynaptic current and long-term potentiation / depression of synaptic morphological characteristics and functions. The realization of this all-optical regulation of synapses is of great significance to the practical application of artificial retinas.

[0079] Although the embodiments of the present invention are described in detail above, it is obvious to those skilled in the art that various modifications and changes can be made to these embodiments. However, it should be understood that such modifications and changes are within the scope and spirit of the present invention described in the claims. Moreover, the present invention described herein may have other embodiments and may be implemented or realized in a variety of ways. Unless otherwise defined, the technical terms or scientific terms used herein should be understood by people with ordinary skills in the field to which the present invention belongs. The words "including" and the like used herein mean that the elements or objects appearing before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects.

Claims

1. A neuromorphic device with high rectification ratio and high selectivity, characterized in that From bottom to top, this includes: substrate; bottom electrode layer; a first dielectric layer; a second dielectric layer; top electrode layer; The rectification ratio and nonlinearity of the device can be adjusted by regulating the energy bands of the first dielectric layer and the second dielectric layer and modulating the metal work functions of the bottom electrode layer and the top electrode layer, thereby suppressing leakage current.

2. The neuromorphic device with high rectification ratio and high selectivity according to claim 1, characterized in that The material of the first dielectric layer is TiO2; and / or, The thickness of the first dielectric layer is 20 nm.

3. The neuromorphic device with high rectification ratio and high selectivity according to claim 1, characterized in that The material of the second dielectric layer is NbO x and / or, The thickness of the second dielectric layer is 20 nm.

4. The neuromorphic device with high rectification ratio and high selectivity according to claim 1, characterized in that The material of the bottom electrode layer is TiN; and / or, The thickness of the bottom electrode layer is 70 nm.

5. The neuromorphic device with high rectification ratio and high selectivity according to claim 4, characterized in that The material of the top electrode layer is Ru; and / or, The thickness of the top electrode layer is 70 nm.

6. A method for preparing a neuromorphic device with high rectification ratio and high selectivity, characterized in that: The steps include: S210: providing a substrate; S220: forming a bottom electrode layer on the substrate; S230: forming a first dielectric layer on the bottom electrode layer; S240: forming a second dielectric layer on the first dielectric layer; S250: forming a top electrode layer on the second dielectric layer.

7. The method for preparing a neuromorphic device with high rectification ratio and high selectivity according to claim 6, characterized in that: A bottom electrode layer is formed on the substrate, comprising: A bottom electrode layer is formed on the substrate by using a physical vapor deposition process, wherein the material of the bottom electrode layer is TiN, and the thickness of the bottom electrode layer is 70 nm.

8. The method for preparing a neuromorphic device with high rectification ratio and high selectivity according to claim 6, characterized in that: Forming a first dielectric layer on the bottom electrode layer, comprising: According to the set reaction temperature and the set cycle rate, a plasma atomic layer deposition process is used for cyclic deposition until a first dielectric layer of a set thickness is formed on the bottom electrode layer, wherein the set reaction temperature is 250°C, the set cycle rate is 0.1nm / cycle, the total number of cycles is 200 times, and the thickness of the first dielectric layer is 20nm.

9. The method for preparing a neuromorphic device with high rectification ratio and high selectivity according to claim 6, characterized in that: Forming a second dielectric layer on the first dielectric layer, comprising: A second dielectric layer is formed on the first dielectric layer using a physical vapor deposition process, wherein the material of the second dielectric layer is NbO x , the thickness of the second dielectric layer is 20 nm.

10. The method for preparing a neuromorphic device with high rectification ratio and high selectivity according to claim 6, characterized in that: forming a top electrode layer on the second dielectric layer, comprising: A top electrode layer is formed on the second dielectric layer by adopting a metal lift-off process or a hard mask process and using a physical vapor deposition process, wherein the material of the top electrode layer is Ru, and the thickness of the top electrode layer is 70 nm.