A three-dimensional visual neuromorphic memristor and its preparation method
By designing a three-dimensional visual neuromorphic memristor, using the light response of the groove structure and the oxide heterojunction, efficient optical information acquisition and calculation are achieved, and the problems of leakage current and high power consumption in high-density integrated circuits are solved. Improved computing efficiency and space utilization.
Patent Information
- Application Number
- CN202210780812.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-07-04
- Publication Date
- 2024-11-19
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the prior art, high-density integrated circuit devices have limited performance due to leakage current problems under small sizes, and have high power consumption and low efficiency in data acquisition and calculation, so they cannot effectively realize efficient information acquisition and storage.
Design a three-dimensional visual neuromorphic memristor, adopting groove structure and oxide heterojunction, realize optical information acquisition and conductance modulation through optical excitation signals, and combine transparent electrodes to complete neuromorphic calculation and storage functions, replacing traditional storage with the calculation module.
It realizes efficient space utilization of high-density integrated circuits, reduces chip energy consumption, improves computing efficiency, is suitable for high-sensitive and low-cost photosensitive nerve morphological calculations, avoids power consumption caused by frequent information movement, and provides high energy efficiency The development direction of electronic devices.
Smart Images

Figure CN115117192B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of semiconductor technology, and in particular relates to a three-dimensional visual neuromorphic memristor and a preparation method thereof, which is suitable for the integration and application of a high-density bionic visual memristor device array. Background Art
[0002] With the development of integrated circuits, the industry's demand for high-density integrated chips is increasing. Although the size of devices is constantly shrinking to achieve high-density integration, problems such as leakage current faced by devices at small sizes limit the performance and further miniaturization of devices. Developing new structural three-dimensional electronic devices to improve the space utilization of chips is an important path to achieve high-density integration.
[0003] Inspired by the efficient storage and computing mode of the human brain, neuromorphic memristive devices have gradually gained attention and development. The use of voltage excitation can achieve controllable adjustment of the resistance state with memory effect, simulating the weight change process of biological synapses in the human brain, which is crucial for efficient information computing. This efficient computing mode can save the calculation results directly in situ, avoiding frequent data movement between storage units and computing units, greatly improving computing efficiency and reducing system power consumption.
[0004] Although the data storage and computing functions can be completed in the same device unit through neuromorphic memristors, data collection still requires the help of separate sensors. Information from sensors often needs to be converted through modules such as digital-to-analog converters before it can be passed to the computing unit for processing. This process also consumes a lot of power and reduces the efficiency of the system. As an efficient and multifunctional system, the human visual system can help humans collect 80% of the external environment information and achieve efficient data collection and processing. Inspired by the visual system, the development of artificial visual neuromorphic devices is of great significance to improving the integration and efficiency of the system's information collection, processing and storage, and has become a new wave of development. Summary of the invention
[0005] The present invention discloses a three-dimensional visual neuromorphic memristor, comprising: a substrate; a lower barrier layer formed on the substrate; a first electrode formed on the lower barrier layer; an upper barrier layer formed on the first electrode; a groove structure penetrating the upper barrier layer, the first electrode and the lower barrier layer to expose part of the silicon substrate surface; a memristor functional layer, which is an oxide heterojunction formed on the above device structure; and a second electrode, which is a transparent electrode formed in the groove structure. By applying a light excitation signal to the device and utilizing the light response of the oxide heterojunction, optical information collection is realized, conductivity modulation is performed, and neuromorphic computing and storage functions are completed.
[0006] In the three-dimensional visual neuromorphic memristor of the present invention, preferably, the memristor functional layer is a combination of any two of ZnO, NiO, Ta2O5, InO2, SnO2, HfO2, ZrO2, and Al2O3.
[0007] In the three-dimensional visual neuromorphic memristor of the present invention, preferably, the first electrode is Ti, Pt, Ta, Ni, or Al.
[0008] In the three-dimensional visual neuromorphic memristor of the present invention, preferably, the lower barrier layer and the upper barrier layer are SiO2, Al3O2, TiO2, or Ta2O5.
[0009] In the three-dimensional visual neuromorphic memristor of the present invention, preferably, the second electrode is ITO, AZO, or FTO.
[0010] The present invention also discloses a method for preparing a three-dimensional visual neuromorphic memristor, comprising the following steps: forming a lower barrier layer on a substrate; forming a first electrode on the lower barrier layer; forming an upper barrier layer on the first electrode; etching the upper barrier layer, the first electrode and the lower barrier layer to expose part of the silicon substrate surface to form a groove structure; forming an oxide heterojunction on the above device structure as a memristor functional layer; forming a transparent electrode in the groove structure as a second electrode, and applying a light excitation signal to the device and utilizing the light response of the oxide heterojunction to realize optical information collection, conductance modulation, and complete neuromorphic computing and storage functions.
[0011] In the method for preparing a three-dimensional visual neuromorphic memristor of the present invention, preferably, the memristor functional layer is a combination of any two of ZnO, NiO, Ta2O5, InO2, SnO2, HfO2, ZrO2, and Al2O3.
[0012] In the method for preparing a three-dimensional visual neuromorphic memristor of the present invention, preferably, the first electrode is Ti, Pt, Ta, Ni, or Al.
[0013] In the method for preparing a three-dimensional visual neuromorphic memristor of the present invention, preferably, the lower barrier layer and the upper barrier layer are SiO2, Al3O2, TiO2, or Ta2O5.
[0014] In the method for preparing a three-dimensional visual neuromorphic memristor of the present invention, preferably, the second electrode is ITO, AZO, or FTO.
[0015] Beneficial effects:
[0016] (1) Breaking the traditional planar integrated device structure, designing a groove-type three-dimensional device structure with efficient space utilization, making the chip integration density higher and saving chip area, which is very suitable for high-density integration of neuromorphic computing chips. In addition, the concentration of optical signals is higher, which is more suitable for high-sensitivity, low-cost optical neuromorphic computing.
[0017] (2) Neuromorphic memristors are used to replace traditional storage and computing modules. With the help of the efficient brain-like computing method of neuromorphic devices and the high-speed response characteristics of memristors, the energy consumption of the chip is saved and the computing efficiency of the chip is improved, providing a new direction for the development of high-efficiency electronic devices.
[0018] (3) Artificial retinal devices are designed using the excellent light response of oxide heterojunctions. By constructing artificial retinas, efficient optical information acquisition, processing, and calculation can be achieved, avoiding the conversion and transmission of information between different units. This can maximize efficiency and reduce power consumption caused by frequent information movement, laying an important foundation for the construction of artificial vision systems. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 It is a flow chart of the method for preparing a three-dimensional visual neuromorphic memristor.
[0020] Figure 2 to Figure 7 It is a structural schematic diagram of each stage of the three-dimensional visual neuromorphic memristor preparation method.
[0021] Figure 8 A cross-sectional view of a 3D visual neuromorphic memristor.
[0022] Fig. 9 Schematic diagram of the operating principle of a 3D visual neuromorphic memristor. DETAILED DESCRIPTION
[0023] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. The described embodiments are only 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 making creative work are within the scope of protection of the present invention.
[0024] In the description of the present invention, it should be noted that the terms "upper", "lower", "vertical", "horizontal", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.
[0025] In addition, many specific details of the present invention are described below, such as the structure, materials, dimensions, processing technology and techniques of the device, so that the present invention can be more clearly understood. However, as can be understood by those skilled in the art, the present invention can be implemented without following these specific details. Unless otherwise specified below, the various parts of the device can be made of materials known to those skilled in the art, or can be made of materials with similar functions developed in the future.
[0026] Figure 1 This is a flow chart of the method for preparing a three-dimensional visual neuromorphic memristor. Figure 1 As shown, the method for preparing a three-dimensional visual neuromorphic memristor includes the following steps:
[0027] In step S1, a three-dimensional visual neuromorphic memristor device is fabricated by preparing low-doped silicon as a substrate 100. The substrate may also be an SOI substrate, glass, SiC, etc.
[0028] In step S2, a lower barrier layer SiO2 101 with a thickness of 10 nm to 30 nm is prepared on the substrate 100 by physical vapor deposition. Figure 2 The barrier layer material may also be oxides such as Al3O2, TiO2, and Ta2O5.
[0029] In step S3, a first electrode 102 with a thickness of 10 nm to 50 nm is prepared on the lower barrier layer 101 by physical vapor deposition. Figure 3 The first electrode material is preferably Ti, Pt, Ta, Ni, Al, etc.
[0030] In step S4, an upper barrier layer SiO2 103 with a thickness of 10 nm to 30 nm is prepared on the first electrode 102 by physical vapor deposition. Figure 4 The material of the barrier layer can also be oxides such as Al3O2, TiO2, and Ta2O5.
[0031] In step S5, CF4, Ar, and CF4 are used as etching gases to etch a portion of the upper barrier layer SiO2103, the first electrode 102, and the lower barrier layer SiO2101, so that the substrate is exposed to obtain a groove structure, such as Figure 5 The etching power is preferably 500W, and the range is 200W to 550W; the etching gas flow rate is preferably 60sccm, and the range is 40sccm to 80sccm; the etching time is preferably 5 minutes, and the range is 4 minutes to 10 minutes.
[0032] In step S6, a 10nm / 10nm ZnO / HfO2 heterojunction is grown on the above structure by an atomic layer deposition method as a memristor functional layer 104 with photoelectric response, such as Figure 6 As shown. The thickness of the ZnO film is preferably 10nm, and the range is 5nm to 20nm; the thickness of the HfO2 film is preferably 10nm, and the range is 5nm to 20nm. The heterojunction material can be a combination of any two of ZnO, NiO, Ta2O5, InO2, SnO2, HfO2, ZrO2, Al2O3, etc.
[0033] In step S7, a transparent electrode with a thickness of 10 nm to 30 nm is deposited on the memristor functional layer 104 in the groove structure by physical vapor deposition as the second electrode 105. Figure 7 The transparent electrode material is preferably a transparent oxide film with conductive properties such as ITO, AZO, and FTO.
[0034] like Figure 7 to Figure 9 As shown, the three-dimensional visual neuromorphic memristor includes: a substrate 100; a lower barrier layer 101, formed on the substrate 100; a first electrode 102, formed on the lower barrier layer 101; an upper barrier layer 102, formed on the first electrode 102; a groove structure, penetrating the upper barrier layer 103, the first electrode 102 and the lower barrier layer 101, so that part of the surface of the silicon substrate 100 is exposed; a memristor functional layer 104, which is an oxide heterojunction, formed on the above device structure; a second electrode 105, which is a transparent electrode, formed in the groove structure, by applying a light excitation signal to the device and utilizing the light response of the oxide heterojunction, optical information collection is realized, conductivity modulation is performed, and neuromorphic computing and storage functions are completed.
[0035] Fig. 9 This is a schematic diagram of the operating principle of a three-dimensional visual neuromorphic memristor. Fig. 9 As shown, by applying continuous optical pulses to the device, the current of the device can be changed, thereby realizing the acquisition and recording of optical signals. With the help of the non-volatile characteristics of the device, the recorded current signal can be stored in the device for a long time to realize the storage function. At the same time, the current of the device is continuously regulated under the stimulation of light pulses, realizing a weight change process similar to that of synapses in the human brain, thereby realizing neuromorphic computing.
[0036] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.
Claims
1. A three-dimensional visual neuromorphic memristor, characterized in that: include: substrate; a lower barrier layer formed on the substrate; a first electrode formed on the lower barrier layer; an upper barrier layer formed on the first electrode; A groove structure, penetrating the upper barrier layer, the first electrode and the lower barrier layer, so that a portion of the silicon substrate surface is exposed; A memristor functional layer with photoelectric response, which is a heterojunction composed of two layers of oxide, formed on the upper barrier layer and the groove structure; The second electrode, which is a transparent electrode, is formed in the groove structure. By applying optical excitation signals to the device and utilizing the optical response of the oxide heterojunction, optical information collection and conductivity modulation are achieved to complete neuromorphic computing and storage functions. The two oxide layers of the memristor functional layer are any combination of ZnO, NiO, Ta2O5, InO2, SnO2, and Al2O3.
2. The three-dimensional visual neuromorphic memristor according to claim 1, characterized in that: The first electrode is Ti, Pt, Ta, Ni or Al.
3. The three-dimensional visual neuromorphic memristor according to claim 1, characterized in that: The lower barrier layer and the upper barrier layer are SiO2, Al3O2, TiO2 or Ta2O5.
4. The three-dimensional visual neuromorphic memristor according to claim 1, characterized in that: The second electrode is ITO, AZO or FTO.
5. A method for preparing a three-dimensional visual neuromorphic memristor, characterized in that: The following steps are involved: forming a lower barrier layer on the substrate; forming a first electrode on the lower barrier layer; forming an upper barrier layer on the first electrode; Etching the upper barrier layer, the first electrode and the lower barrier layer to expose a portion of the silicon substrate surface to form a groove structure; Forming two layers of oxide on the upper barrier layer and the groove structure to form a heterojunction as a memristor functional layer with photoelectric response; forming a transparent electrode in the groove structure as a second electrode, By applying optical excitation signals to the device and utilizing the optical response of the oxide heterojunction, optical information collection and conductivity modulation are achieved to complete neuromorphic computing and storage functions. The two oxide layers of the memristor functional layer are any combination of ZnO, NiO, Ta2O5, InO2, SnO2, and Al2O3.
6. The method for preparing a three-dimensional visual neuromorphic memristor according to claim 5, characterized in that: The first electrode is Ti, Pt, Ta, Ni or Al.
7. The method for preparing a three-dimensional visual neuromorphic memristor according to claim 5, characterized in that: The lower barrier layer and the upper barrier layer are SiO2, Al3O2, TiO2 or Ta2O5.
8. The method for preparing a three-dimensional visual neuromorphic memristor according to claim 5, characterized in that: The second electrode is ITO, AZO or FTO.
