A multi-modal optoelectronic synapse device design and operation method based on asymmetric structure and defect cooperation

Abstract

This invention discloses a design and operation method for a multimodal optoelectronic synaptic device based on the synergy of asymmetric structure and defects. The method includes constructing a device with a two-terminal asymmetric structure, consisting of a metal layer / functional layer 1 / functional layer 2 / electrode. The core lies in forming a two-terminal asymmetric functional junction (Schottky junction / heterojunction) on both sides of functional layer 1. This invention utilizes the unified yet differentiated response dynamics (migration and recombination) of defects such as oxygen vacancies in the functional layer to light and electrical signals, achieving a response within an order of magnitude of specific encoded light and electrical signals in a single device. This enables weighted fusion of multimodal signals, non-volatile storage, and selective decoupling. This method is independent of specific material systems (including oxide types, electrode types, doping, etc.) and fabrication processes, providing core operational logic for constructing general-purpose neuromorphic hardware capable of performing in-situ sensing and computation.

Description

Technical Field

[0001] This invention belongs to the field of multimodal opto-synaptic devices, and relates to a design and operation method of multimodal opto-synaptic devices based on the synergy of asymmetric structure and defects. Background Technology

[0002] With the rapid evolution of artificial intelligence technology, the demand for massive data processing and real-time perception is growing exponentially, posing a severe challenge to traditional computing architectures. The current mainstream von Neumann architecture, due to the separation of computing and storage units, suffers from high energy consumption and latency during data transmission, making it difficult to support the efficient execution of computationally intensive tasks such as deep learning. Against this backdrop, a brain-inspired computing paradigm, inspired by biological neural systems, has emerged. Artificial synaptic devices, as its core component, achieve deep integration of perception, storage, and computation by simulating the structure and function of biological synapses, opening a possible path for building a new generation of high-energy-efficiency computing systems.

[0003] In biological neural systems, synapses, as the connecting hubs between neurons, can dynamically adjust signal transmission efficiency based on stimulus intensity, supporting parallel processing and distributed storage of information. This mechanism has driven the development of neuromorphic computing, which significantly improves dynamic information processing efficiency and reduces energy consumption by simulating the recurrent network structure of the brain. However, most artificial synaptic devices are still designed for single-modal signals and cannot meet the sensing needs of multi-source information coexisting in real-world environments. Therefore, developing integrated devices that can independently respond to and effectively integrate multimodal signals such as optical and electrical signals has become crucial for building efficient sensing-computing systems.

[0004] The core of achieving multimodal fusion lies in ensuring the normalization matching of the intensity of different input signals, enabling the device to output a controllable response that meets expectations. Although three-terminal devices can process multimodal signals, their complex structure limits their integration and energy efficiency. Ideally, two-terminal structures can complete multimodal sensing and preliminary processing in a single device, but currently, most two-terminal memristors struggle to achieve effective decoupling and low-coupling fusion between signals, hindering their practical application in multimodal sensing systems.

[0005] Among various sensing modalities, optical signal sensing, especially ultraviolet light detection, holds significant application potential. Ultraviolet light, with its strong anti-interference capabilities, demonstrates potential in fields such as autonomous driving and biometrics. Many oxide semiconductor materials, based on their band structure and optical response modes, are well-suited for photoelectric detection or synaptic response. Furthermore, through oxygen defect modulation, oxygen vacancies can induce a sustained photoconductivity effect, effectively mimicking the dynamic response characteristics of biological synapses and providing an ideal platform for constructing photoelectric synaptic devices. However, single-device approaches typically only achieve non-volatile enhancement of conductance under single-state optical or electrical stimulation. Complete synaptic plasticity requires reversible suppression modulation of conductance by multiple electrical signals. Existing technologies largely rely on multi-terminal structures to assist electrical signals in achieving the suppression process, leading to increased system complexity and energy consumption. Therefore, developing two-terminal structure devices capable of multi-modal conductance modulation via optical or electrical signals has become a key research challenge. How to combine contact engineering and defect modulation to synergistically achieve effective decoupling and functional integration of optical and electrical signals still requires further exploration. Breakthroughs in this area will significantly advance the development of multimodal neuromorphic perception computing, laying the foundation for building more efficient and adaptive artificial perception systems. Summary of the Invention

[0006] Based on the urgent need for high-performance neuromorphic computing systems in current artificial intelligence technology, and the limitations of traditional artificial synaptic devices in terms of functional scalability and multimodal signal processing capabilities, this invention proposes a design and operation method for multimodal optoelectronic synaptic devices based on the synergy of asymmetric structure and defects. This is an innovative design scheme and operation method for two-terminal optoelectronic artificial synaptic devices. Through the core design concept of asymmetric structure + defect unification mechanism, it successfully achieves effective mixing and decoupling of complex multimodal signals across material systems and fabrication processes under a simple device structure, providing a new technical path to overcome the bottleneck of the fixed function of traditional devices.

[0007] At the design concept level, this invention constructs a metal layer / functional layer 1 / functional layer 2 / electrode structure, realizing an asymmetric structure of Schottky junction and heterojunction at both ends of functional layer 1. Simultaneously, it proposes an innovative device electrical operation method to achieve a complex architecture for handling multi-mode signal coupling and decoupling functions.

[0008] The core advantage of this design concept lies in its two-terminal device structure, which inherently provides scalability and avoids the complexities of three-terminal operation. Through the designed asymmetric contact structure and defect engineering control mechanisms, a multi-dimensional physical field control foundation is established within the device. Based on this physical foundation, a decoupling and coupling operation mode for multi-modal signals in a single device is proposed. This design approach fundamentally breaks the rigid constraints of structure-materials-fabrication processes in traditional devices, laying a theoretical foundation for the flexible expansion of device functionality.

[0009] The specific technical solution adopted in this invention is as follows:

[0010] A design and operation method for a multimodal optoelectronic synaptic device based on the synergy of asymmetric structure and defects includes: constructing a two-terminal asymmetric structure device, the structure being a metal layer / functional layer 1 / functional layer 2 / electrode, wherein functional layer 1 is a defective oxide thin film, and functional layer 2 is a high-temperature stable and highly conductive oxide semiconductor thin film. The high-temperature stability means that the properties are stable at 650℃ (high-temperature growth is often required to control defects during the growth of functional layer 1, therefore functional layer 2 needs to ensure that the properties of functional layer 1 material are stable when grown on it), and the high conductivity means that the conductivity of functional layer 2 is more than 100 times that of functional layer 1; functional layer 1 and the metal layer form a Schottky contact (with obvious rectification effect in electrical testing), functional layer 2 and functional layer 1 form a heterojunction without rectification characteristics (this heterojunction is weak and does not have obvious electrical asymmetry or rectification characteristics), and functional layer 2 and the electrode form an ohmic contact; that is, a two-terminal asymmetric functional junction of Schottky junction / heterojunction is formed on both sides of functional layer 1.

[0011] In the above technical solution, further, functional layer 1 is preferably an oxide-based semiconductor material, which can achieve a crystalline phase under specific processes. The concentration of defective oxygen vacancies in functional layer 1 is 15-35% of the total oxygen coordination concentration (i.e., the sum of the number of defective oxygen atoms and the number of lattice oxygen atoms). The bandgap of functional layer 1 is larger than that of functional layer 2, enabling it to produce a continuous photoconductive effect for light of a specific wavelength, while functional layer 2 is insensitive to light stimulation of that wavelength. The thickness of functional layer 1 is 100-350 nm.

[0012] Furthermore, the metal layer is selected from one or more composites of Au, Cu, Ni, and Ti, and has a thickness of 10 nm to 70 nm.

[0013] Furthermore, physical vapor deposition was used in environments with a vacuum level <10 -3 When growing a metal layer in an environment of Pa, the evaporation rate should be strictly less than 0.1 nm / s, more preferably less than 0.05 nm / s.

[0014] Furthermore, the Schottky junction formed by the metal layer and functional layer 1 in the device, under an applied bias voltage, generates a non-volatile response to optical and / or electrical signals through defect modulation in functional layer 1. When the device is excited by an optical signal, the defect states in functional layer 1 capture photogenerated carriers, inducing a continuous change in conductivity. Under electrical signal excitation, the migration and redistribution of defects can modulate the interface barrier, achieving precise control of the conductivity state. Under certain optical and electrical input signal encoding conditions, the device can generate non-volatile conductivity changes of approximately the same order of magnitude by utilizing the differentiated response dynamics of functional layer 1 and the Schottky junction to optical / electrical signals.

[0015] Furthermore, the metal layer is connected to the negative terminal of the power supply, and the electrodes are connected to the positive terminal. A working voltage is applied to the device to reverse-bias the Schottky junction. Encoded electrical and optical pulses are then applied to the device, achieving multimodal signal coupling. This process can be completed by controlling the readout through only two ports of the device, and is applicable to all synaptic devices with asymmetric two-terminal structures. In this coupling and fusion mode, a certain working voltage is first applied (at which point the Schottky junction is reverse-biased), and then encoded modulated electrical signals or modulated optical signals are applied to the two ports of the device. The device's conductance state can produce non-volatile changes with similar magnitudes and patterns. The device's conductance can generate a coupled response to continuously applied modulated electrical and optical signals under a uniform operating state, and the degree of this coupled conductance state change response has a clear mathematical relationship with each applied electrical or optical signal. The final effect of the input signal on the device's non-volatile conductance change can be significantly modulated by controlling the magnitude, pulse width, and pulse interval of the input electrical signal and the intensity, pulse width, and pulse interval of the input optical signal. Ensure that the total change in device conductance caused by a controllable input electrical signal is less than or equal to the total change in device conductance caused by an input optical signal. The total nonvolatile conductance change response of the device has a significant positive mathematical correlation with the applied electrical or optical signal.

[0016] Furthermore, the operating voltage is determined based on the ability to reverse bias the Schottky junction while maintaining the current at <1nA and with current fluctuations of less than 5%. Under the operating voltage condition, after applying an optical or electrical input signal to change the device's conductance, applying a reverse voltage and then restoring the operating voltage results in a significant decrease or even restoration of the device's conductance to its value before the input signal.

[0017] Furthermore, after multi-mode signal coupling, the device eliminates different modes within the device by applying different reverse decoupling voltages, enabling programmable decoupling. The reverse decoupling voltage is opposite in direction to the operating voltage, forward-biasing the Schottky junction. This voltage is configured to selectively suppress or retain conductance contributions caused by specific types of input signals by modulating the defect distribution and potential barrier at the interface between the functional layer and the junction using an external electric field. This allows the device output current, subsequently measured under the decoupling voltage or readout voltage based on the target mode information, to reflect the strength information of the target mode input signal and significantly weaken or eliminate interference from non-target mode signals during coupling operation.

[0018] Furthermore, if the absolute value of the reverse decoupling voltage applied to the device is greater than the operating voltage, the nonlinear conductance generated by the electrical input signal of the device will be erased under this decoupling voltage.

[0019] Furthermore, the absolute value of the reverse decoupling voltage applied to the device is less than the operating voltage. Under this decoupling voltage, the optical input signal of the device is erased, but the nonlinear conductance generated by the electrical input signal of the device is not completely erased.

[0020] The beneficial effects of this invention are mainly reflected in three dimensions: In terms of material applicability, the design concept based on unified defect control can be extended to various semiconductor material systems, without being limited by specific materials, demonstrating excellent cross-material applicability; in terms of fabrication process, the device structure has good compatibility with mainstream thin-film fabrication technologies, does not depend on special process conditions, and demonstrates significant cross-fabrication compatibility; in terms of functional expansion, by realizing independent sensing, fusion computation, and decoupled processing of multimodal signals in a single two-end structure, it breaks through the functional limitations of traditional devices, demonstrating outstanding multi-functional expansion capabilities. These advantages enable this invention not only to provide an effective hardware solution for solving the current signal processing challenges in neuromorphic computing, but also to open up new technological directions for the future development of artificial intelligence hardware.

[0021] It should be clarified that this method focuses on the general logic of multimodal signal processing, especially the decoupling process, under specific asymmetric structures. It does not limit the material selection, electrode type, doping, or fabrication process of specific devices, but only provides a unified operating paradigm for various two-terminal asymmetric synaptic devices.

[0022] In summary, this invention, through innovative design concepts and technical solutions, has successfully achieved a significant breakthrough in the functional scalability and multimodal processing capabilities of artificial synaptic devices. This design approach, based on asymmetric structures combined with defect synergy, not only provides an effective solution to existing technological bottlenecks but also offers important theoretical reference and technical support for the development of future neuromorphic computing devices. With further research and continuous technological improvement, this novel device is expected to play an increasingly important role in key fields such as artificial intelligence, the Internet of Things, and edge computing. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the structure of the metal / oxide semiconductor / high conductivity oxide semiconductor / metal device prepared in Example 1.

[0024] Figure 2 This is a schematic diagram of the optical and electrical response mechanism of the device obtained in Example 1.

[0025] Figure 3 The image shows the EPR test results of the device material obtained in Example 1.

[0026] Figure 4 A diagram illustrating the functional relationships between design elements of a device.

[0027] Figure 5 The electrical characteristics of the device obtained in Example 1 with different symmetry structures under the same material system are shown.

[0028] Figure 6 The electrical characteristics of the device obtained in Example 1 under an asymmetric structure are shown.

[0029] Figure 7 The electrical characteristics of the device obtained in Example 1 are compared with those of a fast-growing electrode.

[0030] Figure 8 The current response curve of the Au / Ga2O3 / ZnO / Au device prepared in Example 1 to a mixed stimulus current of a continuous 255 nm pulse voltage and a 6.5 V pulse voltage is shown.

[0031] Figure 9 The Au / Ga2O3 / ZnO / Au device prepared in Example 1 exhibits different erasure characteristics under approximate conductivity after different modal stimulation.

[0032] Figure 10 This is a schematic diagram of the device signal decoupling logic diagram obtained in Example 1 and an example of actual device operation.

[0033] Figure 11 This is a schematic diagram of the cross-test results of the correlation between the device signal mixing and decoupling signal and the input signal obtained in Example 1.

[0034] Figure 12 This is a schematic diagram of the expandable material system for different layers of the device prepared in Example 1. Detailed Implementation

[0035] Based on the core design concept of this invention, the implementation of this invention is further illustrated below through two specific embodiments, but the scope of protection of this invention is not limited to the following embodiments.

[0036] This invention provides a design and operation method for a multimodal opto-synaptic device based on asymmetric structure and defect synergy. It is a general operational paradigm. This method constructs a two-terminal asymmetric structure device with a structure of metal layer / functional layer 1 / functional layer 2 / electrode. The core lies in forming a two-terminal asymmetric functional junction (Schottky junction / heterojunction) on both sides of functional layer 1. For example, according to one embodiment of the invention, the device adopts an asymmetric structure of "metal layer (Au) / defect-dominant oxide semiconductor (Ga2O3) / high-conductivity oxide semiconductor (ZnO) / electrode (Au)", such as... Figure 1 , Figure 12 As shown.

[0037] In the specific implementation of the technical solution, this invention has clear requirements for the structure of the device and the characteristics between different material layers. For functional layer 1, its material is an oxide-based semiconductor material. In order to effectively control its defect concentration while ensuring device stability, the material of functional layer 1 must achieve a crystalline phase under specific processes. The functional layer 1 we prepare needs to contain approximately 15-35% oxygen defect concentration compared to its stable, theoretically defect-free crystalline phase, forming a certain concentration of migrateable oxygen vacancies (approximately 20-25%). In the above example of this invention, a ZnO thin film is grown on the substrate using physical vapor deposition (PVD). After partially masking the obtained ZnO thin film, a defect-state Ga2O3 thin film is grown on the uncovered portion of the ZnO thin film using PVD. Specifically, a defect-state pure β-Ga2O3 thin film is prepared using laser pulse deposition (PLD), employing a high-purity Ga2O3 target, with a laser energy of 200-215 mJ and a frequency of 3-4. The target-substrate distance is 5-5.3 cm, the oxygen partial pressure in the cavity is controlled at 0.1-0.3 Pa, and the growth time is 50-75 min. By controlling the oxygen partial pressure, the oxygen vacancy concentration in the obtained film can be effectively controlled. Then, Au metal is deposited on the defective Ga2O3 film and ZnO film. In other embodiments of the present invention, other methods can also be used to fabricate the device. The core of the inventive concept of the present invention lies in the device structure, and its materials and fabrication are all achievable with existing technologies. For the material of functional layer 2, its main function is to form a weak heterojunction with functional layer 1, reconstruct the defect distribution of functional layer 1, and at the same time achieve a better connection electrode function. At the same time, the band gap of the material of functional layer 1 is larger than that of functional layer 2, and the photosynaptic response of the device is mainly determined by the continuous photoconductivity of functional layer 1.

[0038] The device's metal layer and functional layer 1 must form a stable Schottky contact, which can be modulated through defect engineering under an applied bias voltage. This is crucial for achieving the device's electrical synaptic performance. Excessive defect concentration makes the Schottky contact highly susceptible to tunneling, leading to unstable device conductivity. Conversely, insufficient defect concentration results in insufficient voltage-driven migratable oxygen vacancies to modulate the Schottky barrier, causing electrical synaptic performance failure. Furthermore, in terms of fabrication, since the metal layer is grown on the defect-prone functional layer 1, for stability reasons, the metal layer's growth rate must be strictly controlled (<0.1 nm / s), and electrode materials with good stability (Au, Cu, Ni, Ti / Au) must be selected to prevent device breakdown.

[0039] In summary, the optical and electrical non-volatile responses of the device are primarily determined by functional layer 1. When excited by an optical signal, the defect states in the functional layer can effectively capture photogenerated carriers, inducing a continuous change in conductance. Under electrical signal excitation, the migration and redistribution of defects can modulate the interface barrier, thereby achieving precise control over the conductance state. Simultaneously, this synergistic control effect and defect control enable the device to generate non-volatile conductance changes of nearly equal magnitude by utilizing the differentiated response dynamics of functional layer 1 and the Schottky junction to optical / electrical signals under a uniform operating state and specific optical and electrical input signal encoding conditions (in the above example, the electrical signal is a high voltage >6 V, the optical signal is 255 nm ultraviolet light with an operating voltage of 0.1-3 V, and under a small Schottky junction reverse bias voltage (<3 V), the Schottky junction in this device can achieve good rectification, maintaining a low and stable current level (<1 nA, current fluctuation <5%)). This is the foundation for realizing multi-mode signal coupling in a single two-terminal device.

[0040] The asymmetric structure of the device results in a unique characteristic: the electrical synaptic response is dominated by the Schottky junction on one side, requiring the Schottky junction to be reverse-biased to achieve conductance control, thus exhibiting a unidirectional response. In contrast, the photosynaptic response is achieved through the continuous photoconductivity of functional layer 1, resulting in a bidirectional response independent of the applied voltage direction. This difference in response direction between different modes, and the resulting erase differences when the same erase voltage is applied to different modes, forms the basis for decoupling multimode signals from a single two-terminal device. This physical process also determines the specific operational methods during decoupling and coupling.

[0041] In a specific implementation case, using a metal layer / gallium oxide / oxide / metal layer material system as an example, the practical application effect of this design concept is demonstrated. Through reasonable material selection and structural design, the device exhibits dual response capabilities to optical signals in specific wavelength bands and electrical pulse signals in specific ranges. More importantly, due to the adoption of a unified defect modulation mechanism, the device demonstrates good coordination and compatibility in its response to different modal signals, which makes it possible to realize true multimodal signal processing. The device can not only process optical and electrical signals independently, but also achieve collaborative processing and effective separation of the two signals. This unique functional characteristic makes it show significant advantages in information processing tasks in complex environments.

[0042] From a fabrication process perspective, this device can be fabricated using physical vapor deposition combined with masking. The thin film deposition and patterning techniques employed in this invention are compatible with mainstream semiconductor processes, ensuring the feasibility and scalability of the technical solution. The entire fabrication process emphasizes the control of interface quality and defect states, achieving precise regulation of device performance through optimized process parameters. This process compatibility provides crucial assurance for the future large-scale fabrication and system integration of devices.

[0043] Example 1: Multimodal signal coupling

[0044] This embodiment aims to verify whether the asymmetric structure combined with the defect migration mechanism proposed in this invention can achieve multimodal signal fusion processing and on-demand decoupling, without depending on a specific material system or preparation process.

[0045] The device structure comprises a Schottky junction formed by a metal and a defect-dominant oxide semiconductor, and a heterojunction formed by the defect-dominant oxide semiconductor and a high-conductivity oxide semiconductor. The core working mechanism utilizes the directional migration characteristics of oxygen defects in the defect-dominant oxide semiconductor under electrical stimulation and the recombination behavior of photogenerated carriers under optical stimulation to achieve a coordinated response of electrical and optical signals. The mechanism is as follows: Figure 2 As shown; the functional layer we grew has a large number of oxygen vacancy defects, and its EPR test results are as follows. Figure 3 As shown. The input signals are synchronously applied electrical and optical pulses, used to simulate multimodal sensing scenarios in practical applications. The overall design concept of the device is as follows. Figure 4 As shown.

[0046] The implementation process includes the following logical steps: First, signal fusion verification is performed by independently applying electrical and optical pulses to the device according to their encoding methods, resulting in device output currents that differ by an order of magnitude. Subsequently, under the same encoding strategy, electrical and optical pulses are applied to the device in a mixed manner. By observing the synergistic effect of defects and photogenerated carriers in the defect-dominated oxide semiconductor, the output fused current signal is observed, thereby verifying the integrability of different mode signals.

[0047] Figure 5 The figures show the cyclic test curves of the electrical current-voltage performance of the device under a symmetrical structure. Synaptic performance is only achieved when a significant hysteresis effect is observed. As can be seen from the figures, a significant hysteresis effect is only observed in the Au / Ga2O3 / ZnO current loop. Figure 6 The figures show the cyclic test curves of the electrical current-voltage performance of the device under an asymmetric structure. The device exhibits a significant asymmetric hysteresis effect. This demonstrates the unidirectional nature of electrical stimulation. Figure 7This demonstrates that the device's electrical performance did not exhibit significant hysteresis under a rapidly grown electrode (growth rate of 0.2 nm / s), highlighting the crucial role of the Schottky junction in the device's electrical synaptic performance.

[0048] Figure 8 The figure shows the current response curves of the device after applying multiple 255 nm ultraviolet light pulses and a 6.5 V pulse voltage. The device operates at a voltage of 1 V. Because this device can produce current responses of similar magnitude to both light and electrical stimuli under suitable parameters, it can effectively mix stimulus signals of different dimensions. The results demonstrate that the asymmetric structure and defect-unified control mechanism employed in this embodiment can effectively achieve multimodal signal fusion processing, and the target signal can be decoupled as needed by adjusting the reset voltage. This functionality is not necessarily related to specific material selection or corresponding fabrication processes, fully verifying the universality of the invention's design concept.

[0049] Example 2 Programmable decoupling of multimodal coupled signals

[0050] This embodiment aims to verify whether the asymmetric structure combined with the defect migration mechanism proposed in this invention can achieve on-demand decoupling of multimodal fusion signals, without depending on specific material systems (including oxide types, electrode types, and doping) or fabrication processes. This embodiment uses a two-terminal device with a "metal / defect-dominant oxide semiconductor / high-conductivity oxide semiconductor / metal" structure as the test carrier (specifically, a two-terminal asymmetric structure device). The structure, material selection, electrode type, doping status, and fabrication process of this device are only used as a test scheme to verify the universality of this method. The implementation process includes the following logical steps: Electrical pulses and optical pulses are mixed and applied to the device according to a certain encoding method, and the output fused current signal is observed. Subsequently, voltage-programmable decoupling verification is performed. When it is necessary to retain the optical signal, a reverse reset voltage with a large absolute value (usually greater than the operating voltage) is applied. Electrical signal interference is eliminated through defect migration, and the current at the reverse reset voltage is read to confirm that only the optical signal-related response is retained. When it is necessary to retain the electrical signal, a reverse reset voltage with a smaller absolute value (usually less than the operating voltage) is applied. Optical signal interference is eliminated through the difference between the photogenerated carrier recombination rate and the defect migration rate. The read voltage is applied again, and the current is read to confirm that only the electrical signal-related response is retained. The different modes of device erasure vary as follows: Figure 9 As shown, this difference and the asymmetric structure of the device are the physical basis for designing this decoupling operation.

[0051] The specific decoupling physical method of the device and the relationship between the coupling / decoupling signal and the input are shown in the figure below. Figure 10 , 11As shown. To extract the optical signal, a reset voltage of -2 V was used, and the current was read and compared with the baseline at the start of the test to obtain the optical decoupling value. To extract the electrical signal, a reset voltage of -0.4 V was used, and a read current of 1 V was reapplied to obtain the electrical decoupling value. Furthermore, the feasibility of the proposed strategy was calculated using 5×5 cross-test sets of individual devices with different encoding methods and input strengths. The correlation between the coupled and decoupled signals and the original input was measured by the relative magnitude of the Pearson correlation coefficient r of the data sets. 电 With r 光 The Pearson correlation coefficients between the electrical input or optical input and the obtained target signal are represented, respectively. The results show that the device design and coupling / decoupling method of this invention can effectively handle multimodal signals.

[0052] The implementation results show that adjusting the reset voltage used in this embodiment can decouple the target signal as needed. This function is not necessarily related to the specific material selection or corresponding fabrication process, fully verifying the universality of the design concept of this invention.

[0053] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art based on the design concept of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A design method for multimodal opto-synaptic devices based on the synergy of asymmetric structure and defects, characterized in that, include: A two-terminal asymmetric structure device is constructed, consisting of a metal layer / functional layer 1 / functional layer 2 / electrode. Functional layer 1 is a defective oxide thin film, and functional layer 2 is a high-temperature stable and highly conductive oxide semiconductor thin film. High-temperature stability means stability at 650°C, and high conductivity means the conductivity of functional layer 2 is more than 100 times that of functional layer 1. Functional layer 1 and the metal layer form a Schottky contact, functional layer 2 and functional layer 1 form a heterojunction without rectification characteristics, and functional layer 2 and the electrode form an ohmic contact. In other words, a two-terminal asymmetric functional junction (Schottky / heterojunction) is formed on both sides of functional layer 1.

2. The design method for multimodal optosynaptic devices based on the synergy of asymmetric structure and defects according to claim 1, characterized in that, The concentration of defective oxygen vacancies in functional layer 1 is 15-35% of the total oxygen coordination concentration, which is the sum of the number of defective oxygen atoms and the number of lattice oxygen atoms; the band gap of functional layer 1 is greater than that of functional layer 2, and the thickness of functional layer 1 is 100-350 nm.

3. The design method for multimodal optosynaptic devices based on the synergy of asymmetric structure and defects according to claim 1, characterized in that, The metal layer is selected from one or more composites of Au, Cu, Ni, and Ti, and has a thickness of 10 nm to 70 nm.

4. The design method for multimodal optosynaptic devices based on the synergy of asymmetric structure and defects according to claim 1, characterized in that, Physical vapor deposition (PVD) is used in environments with a vacuum level < 10. -3 When growing a metal layer in an environment of Pa, the evaporation rate should be strictly less than 0.1 nm / s.

5. The design method for multimodal optosynaptic devices based on the synergy of asymmetric structure and defects according to claim 1, characterized in that, The Schottky junction formed by the metal layer and functional layer 1 in the device generates a non-volatile response under optical and / or electrical signals by the defect modulation in functional layer 1 under an applied bias voltage. When the device is excited by an optical signal, the defect state in functional layer 1 captures photogenerated carriers and induces a continuous change in conductivity. Under electrical signal excitation, the migration and redistribution of defects can modulate the interface barrier, thereby achieving precise control of the conductivity state.

6. The operation method of the multimodal photosynaptic device obtained by the design method according to claim 1, characterized in that, The metal layer is connected to the negative terminal of the power supply, and the electrode is connected to the positive terminal of the power supply. The working voltage is applied to the device to reverse bias the Schottky junction. The electrical pulse and optical pulse are encoded and applied to the device to achieve multi-mode signal coupling.

7. The method of operating the multimodal photosynaptic device according to claim 6, characterized in that, The operating voltage is determined based on the ability to reverse bias the Schottky junction while maintaining the current at <1 nA and with current fluctuations of less than 5%.

8. The method of operating the multimodal photosynaptic device according to claim 6, characterized in that, After multi-mode signal coupling, the device eliminates different modes in the device by applying different reverse decoupling voltages, which enables programmable decoupling; the reverse decoupling voltage is opposite in direction to the operating voltage, making the Schottky junction forward biased.

9. The method of operating the multimodal photosynaptic device according to claim 8, characterized in that, When the absolute value of the reverse decoupling voltage applied to the device is greater than the operating voltage, the nonlinear conductance generated by the electrical input signal of the device will be erased under this decoupling voltage.

10. The method of operating the multimodal photosynaptic device according to claim 8, characterized in that, The absolute value of the reverse decoupling voltage applied to the device is less than the operating voltage. Under this decoupling voltage, the optical input signal of the device is erased.

Images