Dual-mode artificial neuron memristor and manufacturing method thereof

Abstract

The invention discloses a dual-mode artificial neuron memristor and a manufacturing method thereof. The memristor device sequentially comprises a bottom electrode, a resistive layer and a top electrode from bottom to top, the bottom electrode is a P-type highly-doped pure silicon wafer, the resistive layer is an oxide thin film layer, and the top electrode is a thermally-evaporated silver electrode. The resistance change mechanism of the memristor is mainly the connection and disconnection of the silver conductive wire, and different switching behaviors can be displayed under different limiting currents, so that the memristor can switch a volatile mode and a non-volatile mode by adjusting the limiting current of soft breakdown, and if the limiting current Icc during soft breakdown is relatively large, the memristor can be switched to be in a non-volatile mode or a non-volatile mode by adjusting the limiting current Icc during soft breakdown. The device shows stable non-volatility; if the Icc is small, the silver conductive wire is fragile and easy to break, so that the device shows stable volatility. In addition, the memristor device can also simulate the integral ignition behavior of the IF model.

Description

Technical Field

[0001] The invention belongs to the technical field of organic semiconductor manufacturing, and in particular relates to a dual-mode artificial neuron memristor and a manufacturing method thereof. Background Art

[0002] With the rapid development of science and technology, information technologies such as artificial intelligence and 5G communications have put forward urgent performance requirements for basic components and systems. The brain achieves accuracy and high efficiency in information processing by approximating computing tasks in the analog domain and in-memory computing, and the power consumption required for each "reasoning" is very low. We call the computing method that simulates the brain "neuromorphic computing", and memristors are considered to be the most promising basic components for neuromorphic computing. As a two-terminal element, it has similarities with neural synapses in structure. The "memory characteristics" of memristors are also highly similar to the plasticity of synapses. In addition, memristors exhibit excellent high integration density, simple structure, small unit size, low power consumption, and good compatibility with traditional CMOS processes.

[0003] The film-making methods for the resistive switching layer of a memristor are generally divided into vacuum methods and non-vacuum methods. The vacuum method generally has disadvantages such as high energy consumption and high cost. The non-vacuum method can be divided into electrochemical deposition method, solution method, etc. Among them, the electrochemical deposition method has some disadvantages such as pollution; the advantages of the solution method are low preparation cost, simple process, and no vacuum is required in the preparation environment, which is conducive to large-area preparation. Most memristors have only a single mode, and most of them are non-volatile devices. The present invention can easily switch between the volatile and non-volatile modes of the device by controlling the limiting current of the soft breakdown, providing a new direction for the development of neuromorphic devices. Summary of the invention

[0004] The purpose of the present invention is to provide a dual-mode artificial neuron memristor and a method for making the same. A dense oxide film is formed by spin coating and annealing a prepared oxide precursor solution as a resistive switching layer of the memristor. The memristor prepared by this method will show different switching behaviors after soft breakdown at different limiting currents, so that the memristor can be switched between volatile and non-volatile modes.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is as follows: A dual-mode artificial neuron memristor, wherein the memristor device comprises a bottom electrode, a resistive switching layer and a top electrode from bottom to top, wherein the bottom electrode is a P-type highly doped pure silicon wafer (substrate), the resistive switching layer is an oxide film layer, and the top electrode is a thermally evaporated silver electrode.

[0006] Furthermore, the oxide film layer is formed of gallium oxide.

[0007] Furthermore, the thickness of the resistive layer is 20-100 nm, and the thickness of the top electrode is 40-60 nm.

[0008] Furthermore, the method for manufacturing the dual-mode artificial neuron memristor comprises the following steps: S1: Take a P-type highly doped pure silicon wafer as the bottom electrode, place the substrate silicon wafer in an ultrasonic cleaning tank, and clean it with acetone, isopropanol and deionized water for 5-10 minutes in sequence. After cleaning, use a nitrogen spray gun to blow dry the residual liquid on the surface of the substrate immediately, and then treat the cleaned silicon wafer with an oxygen plasma power of 640-680 W for 10-15 minutes to improve the hydrophilicity of the surface; S2: preparing a resistive switching layer oxide precursor solution, and spin coating the precursor solution on the substrate treated in step S1. After the precursor solution is spin-coated into a film, the substrate is placed on a heating table for annealing to obtain a resistive switching layer; S3: Evaporating silver particles on the resistive layer by thermal evaporation to obtain a silver electrode.

[0009] Furthermore, the preparation method of the resistive layer oxide precursor solution described in step S2 is: dissolving gallium nitrate hydrate in ethylene glycol methyl ether, adding ethanolamine after the gallium nitrate hydrate is completely dissolved, stirring at 55-65° C. for 55-65 minutes, filtering the stirred solution with a 0.45 um filter head into a new bottle, and then standing for 24 hours in a light-proof and sealed condition to obtain a resistive layer oxide precursor solution.

[0010] Furthermore, in step S2, the precursor solution is spun onto the substrate treated in step S1 by step-by-step spin coating, the first step of spin coating is at a rotation speed of 450-550 rpm for 5-6 seconds, the second step of spin coating is at a rotation speed of 2800-3200 rpm for 30 seconds, and the annealing treatment is first pre-annealing at 100°C for 10-12 minutes, and then high-temperature annealing at 300°C for 55-65 minutes.

[0011] Compared with the prior art, the two-terminal memristor prepared by the present invention does not require a cumbersome preparation process, and can achieve the function of simulating artificial neurons only through a solution method, and has the following advantages: 1. The film-forming materials of the device are all solutions, which can be formed by a simple spin coating method and can be carried out in the atmosphere. The preparation process is relatively simple, convenient and low-cost.

[0012] 2. The resistive switching mechanism is mainly the connection and disconnection of the silver conductive wire, which will show different switching behaviors under different limiting currents, so that the memristor can switch between volatility and non-volatility modes by adjusting the limiting current of soft breakdown. If the limiting current Icc during soft breakdown is large (greater than or equal to 10uA), the device exhibits stable non-volatility; if Icc is small (less than or equal to 1uA), the silver conductive wire is relatively fragile and easy to disconnect, so the device exhibits stable volatility.

[0013] 3. The device can simulate the integral ignition behavior of the IF model. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 1 is a schematic diagram of the structure of the artificial neuron memristor of the present invention: wherein 100 is the bottom electrode, 110 is the resistive switching layer, and 120 is the top electrode.

[0015] Figure 2 This is the IV characteristic curve of the artificial neuron memristor prepared in Example 1 without soft breakdown and current limiting.

[0016] Figure 3 1 is a soft breakdown diagram of the artificial neuron memristor prepared in Example 1 under a relatively large limiting current (a) and an IV characteristic curve after soft breakdown (b).

[0017] Figure 4 1 is a soft breakdown diagram of the artificial neuron memristor prepared in Example 1 under a relatively small limiting current (a) and an IV characteristic curve after the soft breakdown (b).

[0018] Figure 5 This is the integration-firing behavior of the artificial neuron memristor prepared in Example 1. DETAILED DESCRIPTION

[0019] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings.

[0020] like Figure 1 As shown, a dual-mode artificial neuron memristor, the memristor device comprises a bottom electrode 100, a resistive switching layer 110 and a top electrode 120 from bottom to top, the bottom electrode is a P-type highly doped pure silicon wafer (substrate), the resistive switching layer is an oxide film layer, and the top electrode is a thermally evaporated silver electrode.

[0021] As an embodiment of the present invention, the oxide thin film layer is formed of gallium oxide.

[0022] As an implementation mode of the present invention, the thickness of the resistive layer is 20-100 nm, and the thickness of the top electrode is 40-60 nm.

[0023] The method for manufacturing the dual-mode artificial neuron memristor comprises the following steps: S1: Take a P-type highly doped pure silicon wafer as the bottom electrode, place the substrate silicon wafer in an ultrasonic cleaning tank, and clean it with acetone, isopropanol and deionized water for 5-10 minutes in sequence. After cleaning, use a nitrogen spray gun to blow dry the residual liquid on the surface of the substrate immediately, and then treat the cleaned silicon wafer with an oxygen plasma power of 640-680 W for 10-15 minutes to improve the hydrophilicity of the surface; S2: Prepare a resistive switching layer oxide precursor solution: dissolve gallium nitrate hydrate in ethylene glycol methyl ether, add ethanolamine after the gallium nitrate hydrate is completely dissolved, stir at 55-65°C for 55-65 minutes, filter the stirred solution with a 0.45um filter into a new bottle, and then keep it in a light-proof and sealed condition for 24 hours to obtain a resistive switching layer oxide precursor solution; Spin coating the precursor solution on the substrate processed in step S1, and after the precursor solution is spin coated to form a film, place the substrate on a heating table for annealing to obtain a resistive switching layer; S3: Evaporating silver particles on the resistive layer by thermal evaporation to obtain a silver electrode.

[0024] Furthermore, in step S2, the precursor solution is spun onto the substrate treated in step S1 by step-by-step spin coating, the first step of spin coating is at a rotation speed of 450-550 rpm for 5-6 seconds, the second step of spin coating is at a rotation speed of 2800-3200 rpm for 30 seconds, and the annealing treatment is first pre-annealing at 100°C for 10-12 minutes, and then high-temperature annealing at 300°C for 55-65 minutes. Example 1

[0025] A dual-mode artificial neuron memristor, wherein the memristor device comprises a bottom electrode, a resistive switching layer and a top electrode from bottom to top, wherein the bottom electrode is a P-type highly doped pure silicon wafer, the resistive switching layer is a gallium oxide thin film layer, and the top electrode is a thermally evaporated silver electrode. The manufacturing method of the memristor comprises the following steps: 1) Place a substrate silicon wafer of about 2.0 cm × 2.0 cm in an ultrasonic cleaning tank and clean it with acetone, isopropanol and deionized water for 5 minutes in sequence. Immediately after cleaning, use a nitrogen spray gun to blow dry the liquid remaining on the surface of the substrate. The substrate silicon wafer is used as the bottom electrode. 2) The cleaned silicon wafer was treated with 640 W oxygen plasma power for 10 minutes to improve the hydrophilicity of the surface; 3) Prepare 0.5M Ga2O3 precursor solution, use a weighing balance to take 127.89mg of gallium nitrate hydrate and pour it into an empty bottle, use a pipette to take 1ml of ethylene glycol methyl ether and drop it into the bottle containing Ga2O3 hydrate, stir until the gallium nitrate hydrate is completely dissolved, then add 5ul ethanolamine dropwise while stirring, and then stir at 60℃ for 1h. Filter the stirred solution with a 0.45um filter head into a new bottle, and then let it stand for 24h in a light-proof and sealed condition to obtain the Ga2O3 precursor solution; 4) Prepare the resistive layer gallium oxide film, take about 50ul of Ga2O3 precursor solution, drop it on the silicon substrate, and use a spin coater to perform step-by-step spin coating. The first step is to spin coat at 500rpm for 5s, and the second step is to spin coat at 3000rpm for 30s. Then place the sample on a heating table, pre-anneal at 100℃ for 10min, and then anneal at 300℃ for 1h, and finally obtain a resistive layer with a thickness of 20~100nm; 5) Finally, 40-60nm of silver is thermally evaporated on the resistive layer as the top electrode.

[0026] The IV characteristic curve of the artificial neuron memristor prepared in Example 1 without soft breakdown and current limiting is as follows: Figure 2 As shown, the device has greater than 10 3 The switching ratio is 2V to 3V without current limiting soft breakdown. The soft breakdown diagram of the device under a larger limiting current and the IV characteristic curve after soft breakdown are shown in the figure. Figure 3 As shown in the figure, the limit current is set to 10uA during soft breakdown. After the soft breakdown with such a large limit current, the device shows a non-volatile mode after double scanning. After the soft breakdown, the resistance change process of the device is only the connection and disconnection of part of the conductive wire. Therefore, compared with the device without soft breakdown, the threshold voltage is effectively reduced (between 1v and 2v). The soft breakdown diagram of the device under a small limit current and the IV characteristic curve after soft breakdown are shown in the figure. Figure 4 As shown in the figure, the limit current is set to 1uA during soft breakdown. The conductive filament formed under a smaller limit current is relatively fragile, so the device shows a volatile mode after double scanning. The limit current threshold for switching between the two modes is between 1uA and 10uA. The integration-ignition behavior is shown in Figure 5 As shown in the figure, 170 pulses are used, the amplitude of the pulse group is 2V and the duty cycle is 40%, that is, each pulse has an amplitude of 2V, a duration of 6ms, and an interval of 9ms. The device can show stable integration and ignition processes at a 40% duty cycle. Therefore, this work provides a new form for artificial neurons and a new direction for the development of neuromorphic devices.

[0027] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be regarded as equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A dual-mode artificial neuron memristor, characterized in that: The memristor device comprises a bottom electrode, a resistive switching layer and a top electrode from bottom to top, wherein the bottom electrode is a P-type highly doped pure silicon wafer, the resistive switching layer is an oxide film layer, and the top electrode is a thermally evaporated silver electrode.

2. A dual-mode artificial neuron memristor according to claim 1, characterized in that: The oxide film layer is formed of gallium oxide.

3. A dual-mode artificial neuron memristor according to claim 1, characterized in that: The thickness of the resistive layer is 20-100 nm, and the thickness of the top electrode is 40-60 nm.

4. A method for manufacturing a dual-mode artificial neuron memristor according to any one of claims 1 to 3, characterized in that: The following steps are involved: S1: Use a P-type highly doped pure silicon wafer as the bottom electrode, place the substrate silicon wafer in an ultrasonic cleaning tank for cleaning, and immediately use a nitrogen spray gun to blow dry the residual liquid on the substrate surface after cleaning. Then treat the cleaned silicon wafer with an oxygen plasma power of 640~680W for 10~15 minutes to improve the hydrophilicity of the surface; S2: preparing a resistive switching layer oxide precursor solution, and spin coating the precursor solution on the substrate treated in step S1. After the precursor solution is spin-coated into a film, the substrate is placed on a heating table for annealing to obtain a resistive switching layer; S3: Evaporating silver particles on the resistive layer by thermal evaporation to obtain a silver electrode.

5. The method for manufacturing a dual-mode artificial neuron memristor according to claim 4, characterized in that: In step S1, the substrate silicon wafer is placed in an ultrasonic cleaning tank and cleaned with acetone, isopropanol and deionized water for 5 to 10 minutes in sequence.

6. The method for manufacturing a dual-mode artificial neuron memristor according to claim 4, characterized in that: The preparation method of the resistive layer oxide precursor solution described in step S2 is: dissolving gallium nitrate hydrate in ethylene glycol methyl ether, adding ethanolamine after the gallium nitrate hydrate is completely dissolved, stirring at 55-65°C for 55-65 minutes, filtering the stirred solution with a 0.45um filter head into a new bottle, and then standing for 24 hours in a light-proof and sealed condition to obtain a resistive layer oxide precursor solution.

7. The method for manufacturing a dual-mode artificial neuron memristor according to claim 4, characterized in that: In step S2, the precursor solution is spun onto the substrate treated in step S1 by step-by-step spin coating, wherein the first step of spin coating is performed at a rotation speed of 450-550 rpm for 5-6 s, and the second step of spin coating is performed at a rotation speed of 2800-3200 rpm for 30 s.

8. The method for manufacturing a dual-mode artificial neuron memristor according to claim 4, characterized in that: In step S2, the annealing treatment is first pre-annealing at 100°C for 10-12 minutes, and then high-temperature annealing at 300°C for 55-65 minutes.

Images