A heterojunction-based multi-modal optoelectronic synapse device and a preparation method thereof

Abstract

The application provides a kind of heterojunction-based multimodal optoelectronic synapse device and its preparation method, including sequentially stacked bottom electrode, memory resistance functional layer, photosensitive layer and top electrode;The memory resistance functional layer contains oxygen vacancy-rich metal oxide, and has micro-nano structure surface;The photosensitive layer covers the micro-nano structure surface of memory resistance functional layer, and forms heterojunction interface with memory resistance functional layer;The oxygen vacancy and heterojunction interface synergistic effect, so that the device can simultaneously respond to the change of light signal, ambient temperature and ambient oxygen partial pressure, and make the light pulse intensity and the light response current generated by the device have nonlinear relationship.Through the synergistic regulation of oxygen vacancy and heterojunction interface, the device can simultaneously respond to the change of light signal, ambient temperature and ambient oxygen partial pressure, and the light pulse intensity and the light response current generated by the device have nonlinear relationship, and then realize the integration function of multimodal perception and sensing calculation.

Description

Technical Field

[0001] This invention belongs to the field of optoelectronic functional devices and inorganic semiconductor heterojunction technology, specifically relating to a heterojunction-based multimodal optoelectronic synapse device and its fabrication method. Background Technology

[0002] With the rapid development of the Internet of Things (IoT) and edge intelligence, traditional image processing systems based on the von Neumann architecture face severe challenges in terms of real-time performance, energy efficiency, and environmental adaptability due to the spatial separation of sensing, storage, and computing units. The integrated sensing-memory-computing architecture, by simulating the information processing mechanism of biological vision systems, provides a revolutionary solution to these problems and has broad application prospects in fields such as autonomous driving, drones, and medical imaging diagnosis. A brain-like neuromorphic visual perception system based on photoelectric synapses integrates the photoelectric properties of photosensitive materials with the in-memory-computing characteristics of memristors. A single device can simultaneously realize the functions of optical signal sensing, information storage, and image processing, and its unique event-driven characteristics make it possible to achieve extremely low-power image processing.

[0003] Currently, most devices can only respond to and process a single mode of optical signals. Implementing the sensing and processing of multiple signals in a single functional device would significantly increase the device's complexity, and the transmission delays of different modes of signals would greatly reduce information processing efficiency. Therefore, it is necessary to develop a functional device capable of simultaneously sensing and processing multiple modes of signals, thereby reducing information transmission delays while improving information processing efficiency. Based on existing research, some photoelectric synaptic devices for multimodal signal sensing exist. For example, patent CN120258066A presents a method for monitoring endangered animals based on multimodal floating-gate photoelectric synaptic transistors. This method utilizes the light absorption capability of quantum dots and the high carrier mobility of two-dimensional channel materials to achieve multimodal fusion sensing of images (optical signals) and sound (electrical signals). The patent with publication number CN119497435B, entitled "A Strain-Enhanced Multi-Sensory Fusion Optoelectronic Synaptic Device, its Fabrication Method, and its Application Scheme," describes an optoelectronic synaptic device that simulates visual information perception through the photoconductivity effect of its functional layer and simulates tactile sensation through tensile strain. This achieves the characteristics of multimodal sensory fusion.

[0004] However, existing multimodal opto-synaptic devices can only sense two types of signals, including light signals, such as sound and touch. To further enrich the types of information sensing, the devices need to exhibit more current-dependent characteristics.

[0005] Furthermore, current multimodal sensing devices only exhibit signal sensing characteristics, without demonstrating further in-sensory computing capabilities, such as image processing. In other words, many devices are only nominally "integrated sensing, storage, and computing". Summary of the Invention

[0006] To address the shortcomings and deficiencies of existing technologies, this invention provides a heterojunction-based multimodal optoelectronic synaptic device and its fabrication method, aiming to solve the technical problems of limited signal sensing types and lack of true integrated sensing, storage, and computing capabilities in existing multimodal optoelectronic synaptic devices. The device employs a vertical structure consisting of a bottom electrode, a memristor functional layer, a photosensitive layer, and a top electrode stacked sequentially. The memristor functional layer is an array of oxygen-vacancy-rich metal oxide nanorods, and the photosensitive layer is selectively deposited at the top of the nanorod array, forming a type II heterojunction with the memristor functional layer. Through the synergistic regulation of oxygen vacancies and the heterojunction interface, the device can simultaneously respond to changes in optical signals, ambient temperature, and ambient oxygen partial pressure. Furthermore, there is a nonlinear relationship between the optical pulse intensity and the photoresponse current generated by the device, thereby achieving multimodal sensing and integrated sensing, storage, and computing functions.

[0007] In this device, the preferred metal oxide is titanium dioxide, and the preferred photosensitive layer is cadmium sulfide (CdS). The introduction of CdS not only significantly increases the oxygen vacancy ratio but also broadens the device's light absorption range to the visible light band. Differences in photoresponse current under different wavelengths of light excitation enable color differentiation. The temperature sensing range covers 230 K to 450 K, and the oxygen pressure sensing range is 50 Pa to 5 × 10⁻⁶ Pa. 4 The amplitude, relaxation time, and nonlinearity of the photoresponse current exhibit regular changes with temperature and oxygen partial pressure. Based on these nonlinear relationships, the device can perform grayscale transformation on input optical image information. By adjusting the read voltage, the nonlinearity can be changed, and image noise reduction or encryption / decryption functions can be selectively implemented. Furthermore, by setting thresholds for the read voltage and light pulse intensity, Boolean logic gates can be constructed to perform logical operations. In addition, the device possesses excellent synaptic plasticity, capable of simulating learning, forgetting, and relearning processes. Its time constant for dual-pulse facilitation characteristics is on the same order of magnitude as that of biological synapses, providing strong support for brain-like visual perception systems.

[0008] In terms of fabrication methods, a solvothermal method is used to form an array of oxygen-vacancy-rich metal oxide nanorods on the bottom electrode as a memristor functional layer. A photosensitive layer and heterojunction interface are then constructed on the surface of the nanorod array using room-temperature in-situ deposition. Finally, a magnetron sputtering method is used to form the top electrode. This process is mild and highly compatible, facilitating large-scale fabrication. The bottom electrode is preferably an FTO transparent conductive layer, and the top electrode is preferably an ITO transparent conductive layer. The overall process requires no complex equipment, ensuring the stability and consistency of the device structure and function.

[0009] The specific technical solution adopted by this invention to solve its technical problem is as follows:

[0010] A heterojunction-based multimodal opto-synaptic device includes a bottom electrode, a memristor functional layer, a photosensitive layer, and a top electrode stacked sequentially.

[0011] The memristor functional layer comprises a metal oxide rich in oxygen vacancies and has a micro / nano structured surface.

[0012] The photosensitive layer covers the micro-nano structure surface of the memristor functional layer and forms a heterojunction interface with the memristor functional layer.

[0013] The oxygen vacancies and the heterojunction interface work together to enable the device to respond simultaneously to changes in optical signals, ambient temperature, and ambient oxygen partial pressure, and to create a nonlinear relationship between the optical pulse intensity and the photoresponse current generated by the device.

[0014] Furthermore, the metal oxide includes titanium dioxide, and the photosensitive layer includes cadmium sulfide.

[0015] Furthermore, the heterojunction interface is a type II heterojunction.

[0016] Furthermore, the surface of the micro / nano structure is a nanorod array surface perpendicular to the bottom electrode surface, and the photosensitive layer is selectively deposited on the top of the nanorod array.

[0017] Furthermore, the simultaneous response is characterized by at least one of the amplitude, relaxation time, or rise nonlinearity of the photoresponse current, under fixed reading voltage and light pulse conditions, undergoing measurable changes with changes in ambient temperature and ambient oxygen partial pressure.

[0018] Furthermore, the nonlinear relationship enables the device to perform grayscale transformation processing on the input optical image information based on the intensity of the light pulse.

[0019] Furthermore, by adjusting the magnitude of the reading voltage, the nonlinearity of the nonlinear relationship can be changed, thereby selectively achieving noise reduction or encryption processing of optical image information.

[0020] Furthermore, the device performs Boolean logic operations by setting thresholds for the read voltage and the light pulse intensity.

[0021] Furthermore, a method for fabricating a heterojunction-based multimodal optosynaptic device as described above includes at least the following steps:

[0022] An array of oxygen-vacancy-rich metal oxide nanorods was formed on the bottom electrode using a solvothermal method to serve as the memristor functional layer.

[0023] A photosensitive layer was deposited in situ on the surface of the nanorod array of the memristor functional layer using a room temperature in situ deposition method to construct a heterojunction interface.

[0024] A top electrode is formed on the photosensitive layer using magnetron sputtering.

[0025] Furthermore, the bottom electrode includes an FTO transparent conductive layer, the top electrode includes an ITO transparent conductive layer, the metal oxide is titanium dioxide, and the photosensitive layer is cadmium sulfide.

[0026] Compared with the prior art, the present invention and its preferred embodiments have at least the following beneficial effects:

[0027] This significantly enriches the sensing dimensions of optoelectronic synaptic devices, breaking through the limitation of existing multimodal devices that can only respond to two signals. Through the coordinated regulation of oxygen vacancies and heterojunction interfaces, it achieves synchronous sensing of multiple environmental parameters such as light signals, ambient temperature, and oxygen pressure. Furthermore, it can distinguish colors based on differences in light wavelengths, giving the device a more comprehensive environmental adaptability and effectively expanding its application scenarios in fields such as intelligent sensing and brain-like vision.

[0028] This device truly achieves integrated sensing, storage, and computing capabilities, overcoming the shortcomings of existing devices that only focus on signal sensing and lack in-sensor computing power. Based on the nonlinear relationship between light pulse intensity and photoresponse current, the device can autonomously perform grayscale transformation, noise reduction, encryption, and decryption of optical images. It can also construct Boolean logic gates to perform operations by setting thresholds, eliminating the need for separate sensing, storage, and computing units. This significantly reduces information transmission latency and improves processing efficiency and energy efficiency.

[0029] The device structure and fabrication process exhibit excellent practicality and compatibility. The four-layer vertically stacked structure, combined with the selective deposition of metal oxide nanorod arrays and photosensitive layers, ensures structural stability and functional reliability. The fabrication process employs mild solvothermal methods, room temperature in-situ deposition methods, and magnetron sputtering methods. The processes are simple and mature, requiring no complex equipment, which not only facilitates large-scale mass production but also allows for good integration with back-end circuits, lowering the threshold for industrial application.

[0030] It possesses excellent biomimetic synaptic characteristics, enabling the transformation from short-term to long-term memory by adjusting the parameters of light pulses. It can also simulate the learning, forgetting, and relearning processes of the human brain. Its core characteristics are highly compatible with biological synapses, providing solid support for building a low-power, highly biomimetic brain-like neuromorphic visual perception system and contributing to the technological upgrade of edge intelligence and the Internet of Things. Attached Figure Description

[0031] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0032] Figure 1 This is a schematic diagram of the structure of the ITO / CdS / TiO2 / FTO four-layer photoelectric synapse device according to an embodiment of the present invention;

[0033] Figure 2 These are characterization diagrams of the relevant properties and structures of TiO2, TiO2 / CdS, and CdS / TiO2 / FTO in embodiments of the present invention; Figure 2 In the figure, (a) is the XRD diffraction pattern of TiO2 and TiO2 / CdS, (b) is the scanning electron microscope image of TiO2 (top view), (c) is the scanning electron microscope image of TiO2 / CdS (top view), (d), (e), (f), and (g) are schematic diagrams of elemental distribution in the cross section of CdS / TiO2 / FTO, (h) is the high-resolution XPS spectrum of O1s, (i) is the UV-Vis diffuse reflectance spectrum, (j) is the Tauc curve (extracted from figure (i), (k) is the XPS valence band spectrum, (l) is the temperature-varying fluorescence test image (including the Aronnewoods fitting of the temperature-varying fluorescence test), and (m) is the band structure of TiO2 / CdS.

[0034] Figure 3 This is a simulation diagram of the synaptic characteristics of the ICTF device under a read voltage of 0.1V according to an embodiment of the present invention. Figure 3 In the figure, (a) is the generation and relaxation curve of photocurrent (405nm pulsed laser excitation, pulse width 5s, power 28.4μW), (b) is the dependence of the double-pulse facilitation index on the pulse interval (including simulated double-pulse facilitation behavior), (c) is the plasticity diagram dependent on the number of pulses, (d) is the plasticity diagram dependent on the pulse intensity, (e) is the plasticity diagram dependent on the pulse width, and (f) is the simulation diagram of learning experience including the learning, forgetting and relearning process (405nm light pulse excitation with power 28.4μW, pulse frequency 1Hz).

[0035] Figure 4 The graphs show the test results of the EPSC (excitatory postsynaptic current) of the ICTF device at different temperatures according to embodiments of the present invention. Figure 4 In the figure, (a) is a diagram of the generation and relaxation process of EPSC at different temperatures, (b) is a diagram of the temperature dependence of photocurrent and dark current, and (c) is a diagram of the temperature dependence of EPSC nonlinearity and relaxation lifetime.

[0036] Figure 5 This is a simulation diagram of the photocurrent characteristics of the ICTF device under different oxygen pressures and the image memory forgetting under different oxygen pressures according to an embodiment of the present invention; Figure 5 In the figure, (a) is the generation and relaxation curve of photocurrent under different oxygen pressures, (b) is the oxygen pressure dependence of on / off ratio and relaxation lifetime, (c) is the simulation diagram of the human visual system's memory process of the letter "H" under different oxygen pressures, and (d) is the simulation diagram of the human visual system's forgetting process of the letter "H" under different oxygen pressures.

[0037] Figure 6 This is a diagram illustrating the image memory and optical logic gate characteristics of an ICTF device according to an embodiment of the present invention. Figure 6In the diagram, (a) shows the EPSC plot under different reading voltages, (b) shows the nonlinearity plot under different reading voltages, (c) shows the EPSC plot under different wavelengths, (d) shows the image perception application diagram containing the letters "H", "T", and "U" under optical pulse excitation (the three letters are excited by 405nm, 520nm, and 635nm lasers respectively, with a power of 28.4μW; the purple area represents the image memory process, and the green area represents the image forgetting process), (e) shows the dependence of EPSC on optical power under different reading voltages under 405nm laser pulse excitation, (f) shows the dependence of EPSC on optical power under different reading voltages under 520nm laser pulse excitation, and (g) shows the truth table of the two optical logic gates "OR" and "AND".

[0038] Figure 7 The figures shown are related to the efficient neuromorphic image processing system with image encryption and noise reduction functions according to an embodiment of the present invention. Figure 7 In the diagram, (a) is a schematic diagram of the high-efficiency image processing system, (b) is a graph showing the nonlinear relationship between the normalized EPSC (0.1V read voltage) and the normalized optical power under 375nm laser pulse excitation (including the nonlinearity under different read voltages), (c) is a histogram of gray value distribution of the image with added noise (including the image after adding noise), (d) is a histogram of gray value distribution of the image after noise reduction (including the image after noise reduction), (e) is a histogram of gray value distribution of the image after encryption (including the image after encryption), (f) is a histogram of gray value distribution of the image after decryption (including the image after decryption), (g) is a comparison graph of image recognition accuracy after different operations, and (h), (i), (j), and (k) are the confusion matrix graphs corresponding to different operations. Detailed Implementation

[0039] To make the features and advantages of the present invention more apparent and understandable, specific embodiments are described below in detail:

[0040] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0041] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0042] To enrich the types of sensed signals, establish a multimodal environmental parameter sensing system, make devices more intelligent, broaden application scenarios, and further integrate in-sensory computing capabilities to achieve true sensing-memory-computing integration, this invention provides a heterojunction-based multimodal opto-synaptic device and its fabrication method.

[0043] like Figure 1 As shown, this invention utilizes a simple solvothermal method and a room-temperature in-situ deposition method, combined with magnetron sputtering, to fabricate a photoelectric synaptic device with a four-layer vertical structure of ITO / CdS / TiO2 / FTO. The specific process is as follows:

[0044] I. Substrate Cleaning

[0045] First, the FTO substrate was ultrasonically cleaned in acetone, ethanol, water, and ethanol solutions for 15 minutes in sequence. After cleaning, it was immersed in ethanol for later use.

[0046] II. Preparation of TiO2 resistive switching layer

[0047] TiO2 resistive switching layers were prepared using a classic acid-assisted hydrolysis method. A 4 M hydrochloric acid solution was prepared by mixing 5.6 mL of 12 M hydrochloric acid and 11.2 mL of deionized water. After stirring for 10 min, 0.3 mL of tetrabutyl titanate solution was added dropwise. After thorough stirring, the resulting mixture was transferred to a 25 mL polytetrafluoroethylene (PTFE) liner, and a cleaned FTO substrate (with one end covered with corrosion-resistant tape as the bottom electrode) was placed inside. The container was then sealed in an autoclave and subjected to a solvothermal reaction at 150 °C for 12 h. Excess impurities were then removed by washing with deionized water and ethanol, respectively, to obtain a white TiO2 resistive switching layer (TiO2 / FTO) on the FTO substrate.

[0048] III. Room temperature in-situ deposition of CdS

[0049] The TiO2 / FTO prepared in step two was immersed in 5 mM CdCl2 solution and 5 mM Na2S solution for 2 min, respectively, and the process was repeated 15 times. This allowed for the in-situ deposition of a photosensitive CdS layer on TiO2 / FTO, resulting in yellow CdS / TiO2 / FTO.

[0050] IV. Construction of Photoelectric Synaptic Devices

[0051] In step three, a circular aperture array mask with a diameter of 500 μm and a spacing of 1 mm was obtained to cover the CdS / TiO2 / FTO surface. An ITO top electrode array was then fabricated using DC magnetron sputtering at a current of 100 A for 15 min, resulting in an ITO layer thickness of 200 nm. This ultimately yielded a four-layer vertical ITO / CdS / TiO2 / FTO photoelectric synaptic device (ICTF).

[0052] like Figure 2 As shown, in the above preparation process, Figure 2 (a) shows the X-ray diffraction patterns of TiO2 and TiO2 / CdS. It can be seen that the functional layer obtained in step three is indeed a mixture of TiO2 and CdS. The microstructure and spatial distribution of the two are further given below. Figure 2 (b) shows a scanning electron microscope image of the TiO2 resistive switching layer prepared in step two from a top-down perspective. It can be clearly seen that the resistive switching layer is an array of nanorods with a lateral size of about 300 nm. The top of the nanorods is uneven, indicating that the crystallinity at the top is relatively poor and there may be a certain amount of vacancies. Figure 2 (c) further shows a scanning electron microscope image of the CdS / TiO2 functional layer prepared in step three. It is clear that the original nanorod array structure was not destroyed, proving that the reaction of CdS deposition only occurred at the top of the nanorods. Figure 2 The elemental distribution diagrams (d), (e), (f), and (g) of the CdS / TiO2 / FTO structure further demonstrate the above process. From top to bottom, Cd, Ti, and Sn elements are distributed sequentially, representing CdS, TiO2, and FTO respectively, proving the distribution of the three-layer CdS / TiO2 / FTO structure. More importantly, the introduction of CdS into the photosensitive layer increases the oxygen vacancy concentration of the entire functional layer, which can be demonstrated through… Figure 2 The O 1s high-resolution XPS spectrum given in (h) confirms that the characteristic peak at 531.70 eV can be attributed to oxygen vacancies. It is evident that the introduction of CdS significantly increases the proportion of oxygen vacancies (from 32% to 77%), and the abundant presence of oxygen vacancies is beneficial for achieving excellent photosynaptic properties. Furthermore, Figure 2 The diffuse reflectance absorption spectrum in (i) also demonstrates that the introduction of CdS further broadens the light absorption range of the ICTF device into the visible light region. Based on Figure 2 The Tauc curve obtained by further calculation in (i) (see Figure 2 (j) gives the optical band gaps of TiO2 and CdS, which are 3.02 eV and 2.31 eV, respectively, and combines them with... Figure 2 The XPS valence band spectrum in (k) allows for a preliminary determination of the type II band matching structure between TiO2 and CdS. For example... Figure 2 As shown in (l), the positions of the oxygen vacancy impurity energy levels in TiO2 (0.05 eV below the conduction band) were further obtained by temperature-varying fluorescence testing and Arrhenius fitting, thus yielding... Figure 2The model shown in (m) illustrates the coexistence of oxygen vacancies and a type II bandgap matching structure. The interface barrier in the type II bandgap structure and the TiO2 oxygen vacancies synergistically regulate the carrier capture and release process. The introduction of CdS, which has excellent light absorption properties, endows the ICTF device with integrated sensing, storage, and computing functions and multimodal sensing characteristics.

[0053] To demonstrate the significant potential of the ICTF device of this invention in simulating photosynaptic behavior, this embodiment further conducted a series of tests on photosynaptic characteristics, such as... Figure 3 As shown. In the fabricated photoelectric synaptic device, a 405 nm light pulse is considered as the stimulus to the synaptic tip, and the functional layer current is considered as the synaptic weight. Figure 3 As shown in (a), an applied 405 nm light pulse (lasting 2 s) will rapidly trigger the generation of excitatory postsynaptic current (EPSC) in the ICTF device. After the light pulse is removed, the current decays slowly. After 25 s, the photocurrent is still 60% higher than the initial value, which is extremely important for realizing the memory of optical signals.

[0054] Two-pulse facilitation (PPF) is another key characteristic reflecting synaptic properties and is closely related to short-term memory in the nervous system. Specifically, for the device of this invention, tests have shown that when two consecutive pulses are applied, the amplitude of the current at the synaptic terminal also continuously increases (e.g., Figure 3 As shown in (b) of the diagram, the rate of change of EPSC amplitude, A2 / A1×100%, is defined as the double-pulse dissimilarity index (PPF index). The PPF index is negatively correlated with the interval ΔT between applied optical pulses. Figure 3 As shown in (b), after two consecutive light pulses with a pulse width of 2 s were applied, the PPF decreased from 164% to 138% as ΔT gradually increased from 0.5 s to 10 s. The PPF decay curve matched well with the double-e exponential decay function. The change of the double-pulse facilitation index with the pulse interval satisfies PPF=c1×exp(-ΔT / τ1)+c2×exp(-ΔT / τ2). The fitted time constants τ1=0.55 s and τ2=14.47 s were obtained, which are comparable in magnitude to the characteristic time constants of biological synapses.

[0055] In the human brain, synaptic plasticity, or the dynamic regulation of synaptic weights, is a crucial means of information retention and forgetting. Based on information retention time, it can be divided into short-term memory (STM) and long-term memory (LTM), and the transition from STM to LTM can be achieved by changing synaptic weights. For example... Figure 3As shown in (c), (d), and (e), this embodiment significantly increases the EPSC amplitude and relaxation time in the ICTF device by increasing the number of optical pulses (2 to 8), intensity (10.4-30.4 μW), and duration (1-8 s), respectively, thus verifying the realization of synaptic plasticity.

[0056] Human brain learning involves three important processes: learning, forgetting, and relearning. Due to the non-volatile storage of information, relearning often requires less stimulation compared to the initial learning process. Based on this, an ICTF device was used to simulate the human brain's learning behavior. Figure 3 As shown in (f), in this embodiment, 10 consecutive 1 Hz light pulses were applied to the ICTF device. The photocurrent gradually increased from about 200 nA in the dark state to 1.8 μA (learning process). After removing the light pulses for 10 seconds, the photocurrent autonomously relaxed and decreased to 42% of its peak value (forgetting process). After that, only 3 light pulses were applied again, and the photocurrent value recovered to the peak value of 1.8 μA. The above test results fully demonstrate the feasibility of the ICTF device in simulating synaptic behavior.

[0057] In addition to exhibiting excellent photosynaptic plasticity, the ICTF device of this invention also shows significant temperature dependence characteristics in various EPSC parameters, thus establishing a temperature-EPSC correspondence. This is expected to enable the construction of an intelligent temperature sensing system. The tests conducted in this embodiment are as follows: Figure 4 As shown. Figure 4 Figure (a) presents a comparison of EPSC at three characteristic temperatures (230 K, 320 K, and 450 K) (under 0.1 V bias and 405 nm laser irradiation for 10 s). It is evident that EPSC exhibits unique temperature dependence in various aspects, including dark current, photocurrent, current rise nonlinearity, and relaxation time. Specifically, for example… Figure 4 (b) and Figure 4 As shown in (c), with increasing temperature, the dark current of the device exhibits a trend of first increasing slowly, then remaining almost constant, and finally increasing sharply. Meanwhile, the photocurrent and carrier lifetime both show a distinct volcano-like distribution, and the nonlinearity of the current change also increases continuously with increasing temperature. This unique temperature dependence of the EPSC's multiple parameters should stem from the synergistic regulatory effect of the abundant oxygen vacancies and the heterojunction interface barrier on the carrier transport process.

[0058] The nonlinearity mentioned above refers to the degree to which the growth curve of the excitatory postsynaptic current (EPSC) of the device deviates from the linear relationship. It can be quantified by the power law exponent of the rising segment of the fitted curve. The larger the exponent, the stronger the nonlinearity.

[0059] In addition to exhibiting the unique temperature-sensing characteristics described above, the ICTF device of this invention also demonstrates sensitivity to the environmental parameter of oxygen pressure. It is well known that human visual perception is affected by environmental oxygen pressure. Specifically, under high oxygen pressure, excessive oxygen can induce oxygen toxicity, damaging the human visual sensory system; excessively low oxygen pressure can inhibit the activity of photoreceptor cells on the retina. Furthermore, the brain's memory and processing of visual information are also affected by oxygen pressure. The ICTF device of this embodiment also exhibits similar biomimetic sensing characteristics, as shown in the test results. Figure 5 As shown. Figure 5 Figure (a) shows the EPSC values ​​and relaxation processes under high, medium, and low oxygen pressures. To more intuitively demonstrate the oxygen pressure dependence of EPSC, this embodiment extracts two key parameters: the on / off ratio and relaxation lifetime of EPSC under different oxygen pressures (see Figure 1). Figure 5 As can be clearly seen in (b) of this paper, the amplitude and relaxation lifetime of the EPSC gradually increase with increasing oxygen pressure, reaching a peak at 350 Pa. Further decreasing or increasing the oxygen pressure will reduce the key performance parameters of the EPSC, indicating a clear dependence between EPSC performance and oxygen pressure. Based on this, this embodiment further simulates the human visual system's image recognition, memory, and forgetting functions under different oxygen pressures. Figure 5 (c) illustrates the ICTF device of this invention's ability to recognize and remember the letter "H" at an oxygen pressure of 350 Pa. The best recognition effect is achieved at higher or lower oxygen pressures; the image becomes blurry at these pressures, and the letter "H" gradually becomes clearer with prolonged illumination. Figure 5 (d) in the figure shows the process of image disappearance after the light is removed, which is used to simulate the forgetting characteristics of human brain for images. It can be clearly seen that the letter "H" disappears quickly (i.e. is forgotten) under both low and high oxygen pressure, while the letter "H" is retained for a longer time under medium oxygen pressure (350 Pa). The above results reveal the great potential of ICTF devices in simulating the memory and forgetting functions of human visual system for images, and this characteristic can be tuned by the environmental parameter of oxygen pressure.

[0060] Based on the above test and verification results, it is evident that this invention provides an ICTF device that can be modulated by multimodal signals of light, temperature, and oxygen pressure. Next, this embodiment further explores and verifies richer application scenarios by changing the read voltage, light pulse wavelength, and power to achieve modulation of photosynaptic plasticity, such as... Figure 6 As shown.

[0061] Figure 6Figure (a) shows the EPSC results under different reading voltages. Clearly, as the reading voltage increases from 0.1 V to 1 V, the EPSC increases by a factor of 5.4 (from 4.3 μA to 23.2 μA). Further fitting yields the nonlinearity during the current rise phase. Figure 6 As can be clearly seen in (b) of this paper, the nonlinearity is positively correlated with the read voltage, which verifies the key role of the heterojunction interface barrier in the plasticity modulation of photosynapses. Thanks to the introduction of CdS with excellent light absorption characteristics, the ICTF device of this invention exhibits excellent photoresponse characteristics in the ultraviolet to visible light bands (see [reference]). Figure 6 (c) in the text, and under the stimulation of light of different wavelengths, the photocurrents are very different. Such a large difference in current can be used to distinguish colors.

[0062] like Figure 6 As shown in (d) of this embodiment, a visual model is used to simulate the human visual system's perception, memory, and forgetting process of color images. Laser pulses of the same power at 405 nm, 520 nm, and 635 nm are applied to the ICTF device in the form of "H", "T", and "U" patterns, respectively. As the duration of the light pulse increases, the images become clearer. Since the ICTF device can hardly absorb 635 nm laser light, the letter "U" is almost invisible, and "H" and "T" are also quite distinguishable. After removing the light pulse for 10 seconds, the "H" and "T" images remain clear, and after 30 seconds, the letter "H" still exists, demonstrating the excellent image memory characteristics and color discrimination ability of the ICTF device.

[0063] Figure 6 Figures (e) and (f) show the EPSC under different readout voltages and optical powers under 405 nm and 520 nm laser pulse excitation (pulse width 5 s), respectively. The unique dependence of EPSC on readout voltage, optical power, and excitation wavelength inspires further exploration of unique optoelectronic logic gate applications. Figure 6 As shown in (g), bias voltage and optical power are the two logic inputs. A bias voltage of 0.1 V is defined as "0", otherwise as "1"; optical power is "0" when it is less than 43.1 μW, and "1" otherwise. EPSC is the logic output, with a threshold current set to 2.5 μA. If the photocurrent is less than 2.5 μA, the logic output is defined as "0", otherwise as "1". Combined with... Figure 6The EPSC values ​​in (e) and (f) show that an "OR" logic gate is formed under 405 nm light pulse excitation, while an "AND" logic gate is formed under 520 nm light pulse excitation. It should be noted that under the specific test conditions of this embodiment, an optical power of approximately 43.1 μW and an EPSC of approximately 2.5 μA were used as exemplary thresholds to demonstrate the implementation of the logic function. Those skilled in the art will understand that these thresholds can be changed by adjusting the device operating point (e.g., read voltage) or structural parameters.

[0064] Furthermore, in-sensor storage and processing are crucial functions of an integrated sensing, storage, and computing system built upon optoelectronic synaptic devices. The above results demonstrate the enormous potential of ICTF devices in optical information storage. The next step is to further explore their applications in optical image processing (including image encryption and noise reduction). Figure 7 As shown in (a), an artificial neural vision system for image denoising and encryption was constructed based on ICTF devices, and an artificial neural network was built for image recognition. This neural network includes an input layer (400 nodes), a hidden layer (400 nodes), and an output layer (10 nodes). In this system, the output terminal (EPSC signal) of each device (or a unit in the device array) is connected to the corresponding node of the input layer of the neural network, and the light intensity information after nonlinear conversion by the device is input into the network for training and recognition. Figure 7 Figure (b) shows the EPSC under different readout voltages and optical powers under 375 nm optical pulse excitation, and further extracts the nonlinear conversion relationship (NLTR) between optical power and EPSC. Figure 7 (b) is used to redistribute the pixel grayscale values ​​of the real image to achieve image denoising and encryption. The non-linearity of NLTR is key to achieving image processing. Figure 7 (b) shows the nonlinearity at different reading voltages. Higher nonlinearity (at biases of 0.1 V and 0.5 V) results in more pronounced image denoising characteristics, while lower nonlinearity (at biases of 0.01 V and 1 V) is suitable for image encryption. Gaussian noise (mean 0.5, variance 0.1) is superimposed on an ideal image with grayscale values ​​of 0–1 to simulate a noisy image (see [reference]). Figure 7 (c)). Using the highly nonlinear NLTR under a 0.1 V bias, the processed grayscale distribution and the true image are obtained (see [reference]). Figure 7 In (d) of the image, a significant improvement in contrast is clearly observed, demonstrating the excellent noise reduction characteristics of the ICTF device. Similarly, based on the low nonlinearity NLTR at a bias of 0.01 V, a more blurred image was obtained (see [reference]). Figure 7(e) in the figure demonstrates the implementation of image encryption. Building upon this, NLTR under a 0.1 V bias voltage is further used to redistribute the grayscale values ​​of the encrypted image, successfully recovering the clear image (see [reference]). Figure 7 (f) in the diagram illustrates the image decryption function. All the processed images are then further input into the artificial neural network for recognition; the change in image recognition rate (see [reference]). Figure 7 (g) in the middle) and the corresponding confusion matrix (see Figure 7 The (h), (i), (j), and (k) in the figure further illustrate the implementation of image encryption and noise reduction functions.

[0065] The key technical points and application potential of the solutions provided in the embodiments of the present invention include:

[0066] 1. The structure of the photoelectric synaptic device ICTF and the preparation method of the core functional layer, including the preparation method of the resistive switching layer of TiO2 nanorod array, the method of in-situ room temperature deposition of photosensitive material CdS on nanorods, and the construction of array electrodes.

[0067] 2. By utilizing the multimodal sensing characteristics and the integrated sensing, storage, and computing characteristics of ICTF devices, a multi-parameter sensing system including optical signals, temperature, oxygen pressure, wavelength, etc., can be realized.

[0068] 3. Based on the integration of sensing and storage of optical signals, the present invention further integrates and develops image encryption and noise reduction functions, color differentiation capabilities, and image recognition functions, making the device more intelligent, expanding application scenarios, and realizing true integration of sensing, storage, and computing.

[0069] Compared with existing technologies, the advantages of the present invention include:

[0070] 1. The fabrication process of ICTF opto-synaptic devices is mild and has excellent back-end compatibility:

[0071] The TiO2 resistive switching layer of this invention has a simple and mature preparation process, and the unique nanorod array structure is also conducive to the in-situ integration of subsequent photosensitive materials. The preparation of the CdS photosensitive layer also adopts an extremely mild and simple room temperature in-situ deposition method, which is conducive to large-scale preparation. The uniform core functional layer obtained in the end is also conducive to the further integration of back-end circuits.

[0072] 2. The multimodal sensing characteristics of ICTF, with its abundant oxygen-vacancy TiO2 resistive switching layer, the introduction of CdS with excellent light absorption properties, and the heterostructure between the two, endow ICTF devices with unique multimodal sensing characteristics, including:

[0073] Visible light signal sensing and color discrimination: The introduction of the photosensitive layer CdS greatly improves the light absorption capability of ICTF devices in the visible light band, and its unique wavelength-dependent characteristics are the key to achieving color discrimination.

[0074] The abundance of oxygen vacancies in the TiO2 resistive switching layer endows the device with unique oxygen pressure sensing characteristics. As oxygen is continuously filled, the role of oxygen vacancies changes from carrier trapping centers to carrier recombination centers, resulting in significant oxygen pressure dependence of the amplitude and decay lifetime of EPSC.

[0075] The temperature sensing characteristic is the result of the synergistic effect of oxygen vacancies in TiO2 and the potential barrier at the TiO2 / CdS interface.

[0076] 3. In-sense storage and computation: The image encryption and noise reduction characteristics of ICTF devices are essentially derived from the nonlinear conversion relationship between optical power and EPSC. This nonlinear conversion relationship can be used to redistribute the brightness values ​​of image pixels, thereby realizing the image encryption / noise reduction function.

[0077] Furthermore, through the exploration of the consistency between the new structure and function, this invention reveals the intrinsic relationship between the rich multimodal sensing characteristics exhibited by the ICTF device and the integrated sensing, storage, and computing functions and the constructed heterojunction and oxygen vacancy synergistic regulation model.

[0078] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0079] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

[0080] This invention is not limited to the preferred embodiment described above. Anyone inspired by this invention can derive other forms of heterojunction-based multimodal opto-synaptic devices and their fabrication methods. All equivalent variations and modifications made within the scope of the claims of this invention shall fall within the scope of this invention.

Claims

1. A heterojunction-based multimodal optosynaptic device, characterized in that, It includes a bottom electrode, a memristor functional layer, a photosensitive layer, and a top electrode stacked in sequence; The memristor functional layer comprises a metal oxide rich in oxygen vacancies and has a micro / nano structured surface. The photosensitive layer covers the micro-nano structure surface of the memristor functional layer and forms a heterojunction interface with the memristor functional layer. The oxygen vacancies and heterojunction interface work together to enable the device to respond simultaneously to changes in optical signals, ambient temperature and ambient oxygen partial pressure, and to create a nonlinear relationship between the optical pulse intensity and the photoresponse current generated by the device. The metal oxide includes titanium dioxide, and the photosensitive layer includes cadmium sulfide. The heterojunction interface is a type II heterojunction.

2. The heterojunction-based multimodal photosynaptic device according to claim 1, characterized in that: The surface of the micro / nano structure is a nanorod array surface perpendicular to the bottom electrode surface, and the photosensitive layer is selectively deposited on the top of the nanorod array.

3. The heterojunction-based multimodal photosynaptic device according to claim 1, characterized in that: The simultaneous response is characterized by the following: under fixed reading voltage and light pulse conditions, at least one of the amplitude, relaxation time, or rise nonlinearity of the photoresponse current undergoes a measurable change with changes in ambient temperature and ambient oxygen partial pressure.

4. A heterojunction-based multimodal photosynaptic device according to claim 3, characterized in that: The nonlinear relationship enables the device to perform grayscale transformation on the input optical image information based on the intensity of the light pulse.

5. A heterojunction-based multimodal photosynaptic device according to claim 4, characterized in that: By adjusting the magnitude of the reading voltage, the nonlinearity of the nonlinear relationship can be changed, thereby selectively achieving noise reduction or encryption of optical image information.

6. A heterojunction-based multimodal photosynaptic device according to claim 3, characterized in that: The device performs Boolean logic operations by setting thresholds for the read voltage and the light pulse intensity.

7. A method for preparing a heterojunction-based multimodal photosynaptic device as described in any one of claims 1-6, characterized in that, At least the following steps are included: An array of oxygen-vacancy-rich metal oxide nanorods was formed on the bottom electrode using a solvothermal method to serve as the memristor functional layer. A photosensitive layer was deposited in situ on the surface of the nanorod array of the memristor functional layer using a room temperature in situ deposition method to construct a heterojunction interface. A top electrode is formed on the photosensitive layer using magnetron sputtering.

8. The method according to claim 7, characterized in that, The bottom electrode includes an FTO transparent conductive layer, and the top electrode includes an ITO transparent conductive layer; the metal oxide is titanium dioxide; and the photosensitive layer is cadmium sulfide.

Images