A detection system and method for photosensitive and memory functions of a GaN-based multifunctional optoelectronic device and application

Abstract

The application belongs to the technical field of semiconductor optoelectronic technology, and discloses a photosensitive and memory function detection system, method and application of a GaN-based multifunctional optoelectronic device. The system comprises a test circuit, a lower computer and an upper computer, and the test circuit, the lower computer and the upper computer are sequentially connected and arranged. The test circuit comprises a switching circuit, a memory mode circuit and a detection mode circuit. The memory mode circuit and the detection mode circuit are arranged in parallel. The memory mode circuit and the detection mode circuit are connected with the lower computer. The memory mode circuit and the detection mode circuit are connected with the GaN-based multifunctional optoelectronic device through the switching circuit. The lower computer and the upper computer are connected. The application can synchronously reconstruct the output bias voltage path of the driving control module and the signal input gain path of the high-precision signal acquisition module when receiving an instruction, so as to realize the lossless switching of the GaN-based multifunctional optoelectronic device between the detection mode and the memory mode.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor optoelectronic technology, and in particular to a system, method and application for detecting the photosensitivity and memory function of a GaN-based multifunctional optoelectronic device. Background Technology

[0002] With the explosive growth of artificial intelligence and Internet of Things (IoT) technologies, the "memory wall" bottleneck caused by the separation of storage and computing units in the traditional von Neumann architecture is becoming increasingly prominent. To overcome this limitation, a new architecture based on "sensor-memory-computing integration" has become a research hotspot.

[0003] Gallium nitride (GaN) and its alloys (AlGaN, InGaN) dominate the field of ultraviolet detection due to their wide bandgap, high electron mobility, and excellent thermal stability. Recent studies have revealed that GaN-based heterojunction devices (such as AlGaN / GaN HEMTs) not only possess excellent photodetector capabilities but also exhibit significant persistent photoconductivity due to deep-level defects in their crystal lattice or trapped states at the heterojunction interface. This effect means that the current does not immediately disappear after illumination is removed but decays exponentially and slowly. This behavior is highly similar to the memory mechanism of the transition from short-term to long-term plasticity (LTP) in biological synapses.

[0004] Therefore, a single GaN device can theoretically be time-division multiplexed into two roles:

[0005] High-sensitivity detector (SPPD mode): It typically operates at zero or low bias voltage, utilizes the photovoltaic effect to quickly respond to light signals, and the photocurrent recovers rapidly as the light disappears, making it suitable for real-time imaging.

[0006] Photosynaptic memory (OS mode): It typically operates under specific gate or drain bias voltages (such as negative bias voltage or specific write pulses), and uses the capture and release mechanism of carrier traps to store historical information of optical signals, thereby realizing the "memory" function.

[0007] Although the physical mechanisms of GaN-based multifunctional devices have been widely explored in academia, the development of their practical applications and testing systems has lagged significantly, mainly facing the following three major technical challenges:

[0008] (1) Circuit topology omissions and signal integrity issues in mode switching

[0009] In the present technology, testing of GaN devices in SPPD and OS modes typically relies on expensive semiconductor parameter analyzers (such as the Keithley 4200). Researchers need to manually change cable connections or reprogram source table configurations between different modes.

[0010] Signal interruption: Physical disconnection during manual switching can lead to unexpected relaxation of the internal charge state of the device, resulting in the loss of critical transient decay information.

[0011] Circuit conflict: SPPD mode requires a high-impedance, high-gain current amplification path (nanoampere level), while OS mode (especially the write phase) involves higher drive voltages (e.g., 3.3V) and larger channel currents (microampere level). Without dedicated isolation and switching circuitry, high-voltage drive signals can easily damage high-sensitivity preamplifiers or cause signal distortion due to impedance mismatch.

[0012] (2) Lack of coordinated bias protocols

[0013] The memory function of GaN devices is highly dependent on the history of bias voltage. For example, to simulate the "learning-forgetting" process in the biological brain, a strong "boost" voltage pulse is typically applied first to assist carrier trapping, followed by maintaining a low "read" voltage to monitor conductance changes. Existing general-purpose devices cannot automatically perform this precise "probe-write-read" timing coordination, resulting in poor experimental repeatability and difficulty in achieving automated control in application scenarios.

[0014] (3) Lack of dynamic visualization and data backtracking functions

[0015] In neuromorphic vision applications, users need to observe in real time how "light stimulation" is converted into "synaptic weights". Existing data acquisition software is mostly static waveform recording, lacking the dynamic imaging capability to map real-time current into grayscale images, and even more so unable to overlay and compare the current "forgetting curve" with the historical "learning curve" on the same interface (i.e., data backtracking). This makes it difficult for researchers to intuitively evaluate the "relearning" efficiency of the device. Summary of the Invention

[0016] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a detection system, method, and application for the photosensitive and memory functions of GaN-based multifunctional optoelectronic devices. This invention successfully constructs a comprehensive control system specifically for GaN-based multifunctional optoelectronic devices through hardware circuit design and control. By utilizing the physical isolation of a 4P3T switch and the collaborative control of an MCU, compatibility between nanoampere-level detection and microampere-level actuation is achieved.

[0017] The technical solution adopted by this invention to solve the problem is:

[0018] A detection system for the photosensitivity and memory function of a GaN-based multifunctional optoelectronic device is disclosed. This system can detect the photosensitivity and memory function of the GaN-based multifunctional optoelectronic device. The system includes a test circuit, a lower-level machine, and a higher-level machine, which are sequentially connected. The test circuit includes a switching circuit, a memory mode circuit, and a detection mode circuit. The memory mode circuit and the detection mode circuit are connected in parallel and connected to the lower-level machine. Both the memory mode circuit and the detection mode circuit are connected to the GaN-based multifunctional optoelectronic device via a switching circuit. The lower-level machine and the higher-level machine are connected together.

[0019] The memory mode circuit can perform photocurrent acquisition under zero bias voltage;

[0020] The probe mode circuit can perform sequential pulse voltage excitation and continuous bias voltage reading to realize the writing and maintenance of photosynaptic weights;

[0021] The lower-level machine can provide power supply, control switching circuits, and data output interface for the system;

[0022] The host computer can provide a human-computer interaction interface for the system and visualize the data output by the slave computer, and perform ultraviolet imaging on the memory mode circuit data;

[0023] The switching circuit can switch the switching states of the memory mode circuit and the detection mode circuit when it receives a switching command, and simultaneously switch the output bias voltage path and input gain path of the lower-level machine.

[0024] Furthermore, the switching circuit includes a first switch, a second switch, a third switch, and a fourth switch; the memory mode circuit includes a transimpedance amplifier; and the detection mode circuit includes an instrumentation amplifier and a resistor. One end of each of the first, second, third, and fourth switches is connected to a GaN-based multifunctional optoelectronic device. The other end of the first switch is connected to a lower-level machine. The other ends of the second and third switches are connected to the transimpedance amplifier, and the other end of the fourth switch is connected to a resistor and an instrumentation amplifier. The resistor and the instrumentation amplifier are connected in parallel, and the other end of the resistor is connected to the lower-level machine and the instrumentation amplifier.

[0025] The lower-level machine includes a signal acquisition module, a device driving module, and a data communication module. The signal acquisition module is connected to the transimpedance amplifier and the instrumentation amplifier, and can acquire the amplified electrical signal data of the transimpedance amplifier and the instrumentation amplifier. The first switch is connected to the device driving module, and the device driving module can provide power input, grounding, and control of the signal acquisition module for the switching circuit. The data communication module is connected to the upper-level machine, and can communicate with the upper-level machine.

[0026] Furthermore, the first, second, third, and fourth switches are integrated into a single 4-pole 3-throw switch, and the transimpedance amplifier is a high-gain transimpedance amplifier with a gain of 10. -9 V / A, the instrumentation amplifier is a low-gain instrumentation amplifier with a gain of 10. 6 V / A.

[0027] Furthermore, the lower-level machine is a microcontroller unit (MCU), specifically an STM32F4 series high-performance microcontroller unit MCU with a main frequency of 168MHz, and a built-in 12-bit analog-to-digital converter (ADC) and a 12-bit digital-to-analog converter (DAC).

[0028] The 4-pole 3-throw switch is an industrial-grade 4-pole 3-throw toggle switch with an insulation resistance greater than 1000MΩ and a contact resistance less than 20mΩ.

[0029] The detection method for the photosensitive and memory function detection system of GaN-based multifunctional optoelectronic devices as described above includes the following steps:

[0030] (1) Detection mode

[0031] When the first and fourth switches are off and the second and third switches are on, the GaN-based multifunctional optoelectronic device is irradiated with blue or violet light. The GaN-based multifunctional optoelectronic device generates a current, which is amplified by a transimpedance amplifier and then input to the lower-level computer. The lower-level computer converts the received electrical signal and outputs it to the upper-level computer. The upper-level computer converts the electrical signal into a visual image to express the sensitivity of the device.

[0032] (2) Memory pattern

[0033] When the second and third switches are open and the first and fourth switches are closed, the lower-level computer detects the level transition and immediately outputs a voltage of 3.0V to 3.5V as an enhancement voltage for 5 to 15 seconds to excite carrier trap filling. The current of the GaN-based multifunctional optoelectronic device surges. The lower-level computer collects the current signal in real time, and the upper-level computer records the current change in real time. After the current stabilizes, the lower-level computer outputs a voltage of 1.5V to 2.5V as a holding voltage to maintain photoconductivity gain and read the synaptic weight decay curve. The GaN-based multifunctional optoelectronic device is illuminated by a light source, and the GaN-based multifunctional optoelectronic device generates current. The current is amplified by an instrumentation amplifier and input to the lower-level computer. The lower-level computer converts the received electrical signal and outputs it to the upper-level computer. The upper-level computer converts the electrical signal into a visualized image. At this time, the illumination is removed, and the screen displays a curve of slow current decay. The current collected at this time reflects the memory strength of the synapse.

[0034] Furthermore, the enhanced voltage is 3.3V, the duration is 10 seconds, and the holding voltage is 2.0V;

[0035] The system's lower-level computer integrates multi-level filtering algorithms. In addition to the RC filter in hardware, the lower-level computer also runs a moving average filter and a power frequency notch filter.

[0036] Furthermore, more specifically, it includes the following steps;

[0037] (1) Detection mode

[0038] The first and fourth switches are disconnected, and the second and third switches are connected. The second switch directly grounds the GaN-based multifunctional optoelectronic device, forcibly pulling the bias voltage down to 0V. At this time, the device operates in photovoltaic mode, using the built-in electric field to separate photogenerated carriers. When the GaN-based multifunctional optoelectronic device is irradiated with blue or violet light, it generates current. The current is amplified by the transimpedance amplifier and then input to the lower-level computer. The lower-level computer converts the received electrical signal and outputs it to the upper-level computer. The upper-level computer converts the electrical signal into a visual image to express the sensitivity of the device.

[0039] (2) Memory pattern

[0040] The second and third switches are off, while the first and fourth switches are on. The first switch is connected to the GaN-based multifunctional optoelectronic device. After the lower-level computer detects the signal, it outputs a specific waveform. First, it outputs a high level of 3300mV for 10 seconds. The strong electric field generated by this high voltage assists the photogenerated holes to be captured by deep-level defects, completing the "synaptic weight writing". After 10 seconds, the control voltage is automatically reduced to 2000mV and maintained as the "read voltage". The current collected at this time reflects the memory strength of the synapse. At this time, the second and third switches are in the off state, cutting off the zero-bias circuit and preventing short circuit.

[0041] The transimpedance amplifier is constructed using a low-bias-current (<1 pA) operational amplifier. In probe mode, it provides 10 MΩ current through a 100 MΩ parallel resistor connected to a relay or analog switch. 8 A gain of V / A, in memory mode, with a parallel resistor of 100 kΩ, provides 10 5 Gain of V / A.

[0042] Furthermore, the host computer runs on a PC and is developed using Python. Its specific design structure is as follows:

[0043] (1) In terms of underlying communication, the system constructs an interaction link between the host computer and the slave computer through USB to serial port. The communication baud rate is set to 115200 bps. The protocol layer is designed with a strict frame verification mechanism: each data frame is started with 0xAA 0x55 as the synchronization start word, contains SPPD and OS mode switching instructions, and uses 4 bytes of single-precision floating-point numbers to encapsulate real-time voltage and current values. Finally, CRC16 cyclic redundancy check is embedded at the end.

[0044] (2) In terms of signal processing and backtracking algorithm, after the host computer software obtains the original sampling points, it performs real-time filtering through the moving average algorithm with a window width of 10. The system background maintains a buffer with a capacity of 60 seconds of data in real time. This design supports the core "memory backtracking" function, that is, after the user triggers backtracking, the program can automatically locate and retrieve the historical data of the most recent 3.3V pulse writing stage. These historical records are drawn on the same screen as the current red real-time curve in the form of gray dashed lines.

[0045] (3) The system integrates a lightweight web server based on the FastAPI framework. Testers only need to access the device IP address through a mobile terminal within the local area network.

[0046] (4) For arrayed devices, the software maps the real-time current value of each channel to a gray value of 0-255 and generates a real-time heat map on the interface. In SPPD mode, the heat map is the ultraviolet image; in OS mode, the heat map represents the current "memory weight matrix" to form dynamic imaging.

[0047] Furthermore, the timing control method for the photosensitive and memory function detection system of the GaN-based multifunctional optoelectronic device specifically includes the following steps:

[0048] S100 system initialization: The lower-level machine powers on, the upper-level machine starts the Web service, and establishes a 115200bps serial communication handshake.

[0049] S200 detects the hardware mode switch status: The lower-level machine reads the current physical position of the 4P3T toggle switch and determines the hardware mode based on the current physical position. If the second and third switches are closed and the first and fourth switches are open, it is the detection mode, and the detection mode steps S301 to S304 are executed. If the second and third switches are open and the first and fourth switches are closed, it is the memory mode, and the memory mode steps S401 to S404 are executed.

[0050] S301 Hardware Reconfiguration - Detection Path: Switch matrix action, GaN-based multifunctional optoelectronic device connected to transimpedance amplifier, bias voltage set to zero;

[0051] S302 Weak Photocurrent Acquisition: Transimpedance amplifier performs IV conversion to amplify the current with high gain;

[0052] S303 Data Upload and Preprocessing: The lower-level computer performs analog-to-digital conversion and digital filtering preprocessing, and sends the processed data to the upper-level computer;

[0053] S304 Real-time Imaging Visualization: The host computer maps the light intensity to the grayscale space and refreshes the real-time image area. Then, the host computer collects whether the user selects to perform comparison and backtracking on the Web terminal. If the user selects not to perform comparison and backtracking, the process ends or steps S200 is executed. If the user selects to perform comparison and backtracking, steps S500 is executed.

[0054] S401 Hardware Reconfiguration - Memory Path: Switching matrix action, GaN-based multifunctional optoelectronic device series sampling resistor connected to the drive circuit;

[0055] S402 applies a drive pulse sequence: the lower-level machine outputs a voltage of 3300mV for 10 seconds, followed by an output of 2000mV and maintenance;

[0056] S403 Memory Signal Monitoring: The instrumentation amplifier acquires the voltage across the resistor;

[0057] S404 Data storage and status update: The host computer stores data into the circular buffer, the slave computer lights up the memory mode indicator, and then the host computer collects whether the user selects to perform comparison and backtracking on the Web terminal. If the user selects not to perform comparison and backtracking, the program ends or steps S200 are executed. If the user selects not to perform comparison and backtracking, steps S500 are executed.

[0058] S500 Data Retrospective: The host computer calls the historical memory data buffer and compares it with the current real-time data on the same screen. Depending on the user's subsequent selection, the program ends or steps S200 are executed.

[0059] The above-described GaN-based multifunctional optoelectronic device photosensitive and memory function testing system is applied in GaN-based semiconductor production inspection.

[0060] The advantages and positive effects of this invention are as follows:

[0061] 1. This invention successfully constructs a comprehensive control system specifically for GaN-based multifunctional optoelectronic devices through hardware circuit design and control. Utilizing the physical isolation of a 4P3T switch and the collaborative control of the MCU, compatibility between nanoampere-level detection and microampere-level drive is achieved. The built-in 3.3V / 2.0V timing control accurately simulates the excitation environment of biological synapses. Data backtracking and dynamic imaging functions transform invisible physical processes into intuitive visual maps, greatly accelerating the research and development and application of novel optoelectronic neuromorphic devices. It provides a standardized driving and readout interface scheme for future GaN-based intelligent bionic vision systems, with broad industrial application prospects. Addressing the weak response of GaN devices in low light, the system's lower-level computer integrates multi-level filtering algorithms. In addition to the hardware RC low-pass filter, the MCU also runs a moving average filter and a power frequency notch filter, effectively filtering out 50Hz mains interference and high-frequency thermal noise, ensuring an acquisition accuracy of 10. -9 Ampere level.

[0062] 2. This invention enables complex circuit reconstruction and parameter resetting with a single toggle switch, replacing cumbersome manual wiring and eliminating errors and electrostatic risks introduced by human operation. It achieves temporal coordination of detection and memory functions, allowing a single device to freely switch between the roles of "camera" and "brain," providing a standard paradigm for testing "sensing-memory-computing integrated" chips. Data backtracking and visualization functions significantly lower the analytical threshold for neuromorphic characteristics, enabling non-experts to intuitively understand the brain-like behavior of the device.

[0063] 3. This invention employs a four-pole three-throw (4P3T) hardware topology through a dedicated mode switching circuit. Upon receiving a command, it can simultaneously reconstruct the output bias voltage path of the drive control module and the signal input gain path of the high-precision signal acquisition module, thereby achieving lossless switching between the GaN-based multifunctional optoelectronic device in real-time detection mode (SPPD) and synaptic memory mode (OS). In SPPD mode, the system performs photocurrent acquisition under zero bias to achieve ultraviolet imaging; in OS mode, the system applies time-sequential pulse voltage excitation during the enhancement and maintenance phases to write and maintain the photosynaptic weights. Furthermore, the host computer's data backtracking engine uses a circular data buffer to overlay historical "learning-forgetting" curves with real-time curves, thus intuitively quantifying the device's relearning efficiency and long-term enhancement characteristics. Ultimately, it achieves an integrated sensing, storage, and computing architecture and a closed-loop process through the functional multiplexing of a single physical pixel. Attached Figure Description

[0064] Figure 1 This is a schematic diagram of a circuit structure connection for the system of the present invention;

[0065] Figure 2This is a schematic block diagram of the overall system architecture of the system of the present invention;

[0066] Figure 3 This is a timing control flowchart of the system of the present invention;

[0067] Figure 4 This is the dynamic pixel thermal imaging image and current mapping curve of the host computer in SPPD mode in Example 1;

[0068] Figure 5 The image shows the dynamic pixel thermal image and current mapping curve of the host computer in OS mode with a bias value of 3.3V and a write pulse state in Example 1.

[0069] Figure 6 The image shows the dynamic pixel thermal image and current mapping curve of the host computer in OS mode with a bias value of 2V. Detailed Implementation

[0070] The present invention will be further described below with reference to the embodiments. The following embodiments are descriptive and not limiting, and should not be used to limit the scope of protection of the present invention.

[0071] The various experimental operations involved in the specific embodiments are all conventional techniques in the field. For parts not specifically annotated in this document, those skilled in the art can refer to various commonly used reference books, scientific and technological documents or related instructions and manuals prior to the filing date of this invention to carry out the operations.

[0072] A detection system for the photosensitive and memory functions of GaN-based multifunctional optoelectronic devices, such as Figure 1 As shown, this system can detect the photosensitivity and memory functions of GaN-based multifunctional optoelectronic devices. The system includes a test circuit 1, a lower-level machine 2, and a higher-level machine 3, which are connected in sequence. The test circuit 1 includes a switching circuit, a memory mode circuit, and a detection mode circuit. The memory mode circuit and the detection mode circuit are connected in parallel and connected to the lower-level machine 2. Both the memory mode circuit and the detection mode circuit are connected to the GaN-based multifunctional optoelectronic device through the switching circuit. The lower-level machine 2 and the higher-level machine 3 are connected in sequence.

[0073] The memory mode circuit can perform photocurrent acquisition under zero bias voltage;

[0074] The probe mode circuit can perform sequential pulse voltage excitation and continuous bias voltage reading to realize the writing and maintenance of photosynaptic weights;

[0075] The lower-level machine 2 can provide power supply, control switching circuits and data output interface for the system;

[0076] The host computer 3 can provide a human-computer interaction interface for the system and visualize the data output by the slave computer 2, and perform ultraviolet imaging on the memory mode circuit data;

[0077] The switching circuit can switch the switching states of the memory mode circuit and the detection mode circuit when it receives a switching command, and simultaneously switch the output bias voltage path and input gain path of the lower-level machine 2.

[0078] In this embodiment, as Figure 1 As shown, the switching circuit includes a first switch Q1, a second switch Q2, a third switch Q3, a fourth switch Q4, a memory mode circuit including a transimpedance amplifier 13, and a detection mode circuit including an instrumentation amplifier 14 and a resistor 12. One end of the first switch Q1, the second switch Q2, the third switch Q3, and the fourth switch Q4 are all connected to the GaN-based multifunctional optoelectronic device 11. The other end of the first switch Q1 is connected to the lower-level machine 2. The other ends of the second switch Q2 and the third switch Q3 are both connected to the transimpedance amplifier 13. The other end of the fourth switch Q4 is connected to the resistor 12 and the instrumentation amplifier 14. The resistor 12 and the instrumentation amplifier 14 are connected in parallel. The other end of the resistor 12 is connected to the lower-level machine 2 and the instrumentation amplifier 14.

[0079] Better, such as Figure 1 As shown, the lower-level machine 2 includes a signal acquisition module 21, a device driving module 22, and a data communication module 23. The signal acquisition module 21 is connected to the transimpedance amplifier 13 and the instrumentation amplifier 14. The signal acquisition module 21 can acquire the amplified electrical signal data of the transimpedance amplifier 13 and the instrumentation amplifier 14. The first switch Q1 is connected to the device driving module 22. The device driving module 22 can provide power input, grounding, and control of the signal acquisition module 21 for the switching circuit. The data communication module 23 is connected to the upper-level machine 3. The data communication module 23 can communicate with the upper-level machine 3 to facilitate data display and human-computer interaction on the upper-level machine 3.

[0080] Preferably, the first switch Q1, the second switch Q2, the third switch Q3, and the fourth switch Q4 are integrated into a 4-pole 3-throw (4P3T) switch, and the transimpedance amplifier 13 is a high-gain transimpedance amplifier (TIA) with a gain of 10. -9 V / A, enabling real-time detection with high signal-to-noise ratio; Instrumentation amplifier 14 is a low-gain instrumentation amplifier (DA) with a gain of 10. 6 V / A helps prevent amplifier saturation and allows it to withstand larger channel currents.

[0081] Preferably, the lower-level machine 2 is a microcontroller unit (MCU), specifically an STM32F4 series high-performance microcontroller unit MCU with a main frequency of 168MHz. It has a built-in 12-bit analog-to-digital converter (ADC) and a 12-bit digital-to-analog converter (DAC), and its high processing speed ensures microsecond-level signal acquisition capability.

[0082] The 4-pole 3-position (4P3T) switch is an industrial-grade 4-pole 3-position (4P3T) toggle switch with an insulation resistance greater than 1000MΩ and a contact resistance less than 20mΩ. The high insulation is to prevent the drive voltage in memory mode from interfering with the nanoampere-level signal measurement in detection mode through switch leakage current.

[0083] The method of using the GaN-based multifunctional optoelectronic device photosensitive and memory function detection system as described above includes the following steps:

[0084] (1) Detection mode

[0085] The first switch Q1 and the fourth switch Q4 are open, and the second switch Q2 and the third switch Q3 are closed. The GaN-based multifunctional optoelectronic device 11 is irradiated with blue light or violet light, and the GaN-based multifunctional optoelectronic device 11 generates current. The current is amplified by the transimpedance amplifier 13 and then input to the lower computer 2. The lower computer 2 converts the received electrical signal and outputs it to the upper computer 3. The upper computer 3 converts the electrical signal into a visual image to express the sensitivity of the device.

[0086] (2) Memory pattern

[0087] When the second switch Q2 and the third switch Q3 are open, and the first switch Q1 and the fourth switch Q4 are closed, the lower-level computer 2 detects the level transition and immediately outputs a voltage of 3.0V to 3.5V as an enhancement voltage for 5 to 15 seconds to excite carrier trap filling. The current of the GaN-based multifunctional optoelectronic device 11 surges. The lower-level computer 2 collects the current signal in real time, and the upper-level computer 3 records the current change in real time. After the current stabilizes, the lower-level computer 2 outputs a voltage of 1.5V to 2.5V as a holding voltage to maintain photoconductivity gain and read the synaptic weight decay curve. The GaN-based multifunctional optoelectronic device 11 is illuminated by a light source, and the GaN-based multifunctional optoelectronic device 11 generates current. The current is amplified by the instrumentation amplifier 14 and then input to the lower-level computer 2. The lower-level computer 2 converts the received electrical signal and outputs it to the upper-level computer 3. The upper-level computer 3 converts the electrical signal into a visualized image. At this time, the illumination is removed, and the screen displays a curve of slow current decay (PPC effect). The current collected at this time reflects the memory strength of the synapse.

[0088] Preferably, the enhanced voltage is 3.3V, the duration is 10 seconds, and the holding voltage is 2.0V.

[0089] Ideally, to address the weak response of GaN devices in low light conditions, the lower-level MCU 2 integrates a multi-level filtering algorithm. In addition to the hardware RC filter, the MCU also internally operates a moving average filter and a power frequency notch filter, effectively filtering out 50Hz mains interference and high-frequency thermal noise, ensuring a data acquisition accuracy of 10. -9 Ampere level.

[0090] More specifically, it includes the following steps;

[0091] (1) Detection mode

[0092] First, the first switch Q1 is turned on, and the digital-to-analog converter of the lower-level machine 2 outputs 0V. The GaN-based multifunctional optoelectronic device 11 is grounded (0V potential) through the first switch Q1. The first switch Q1 and the fourth switch Q4 are turned off, and the second switch Q2 and the third switch Q3 are turned on. At this time, the device works in photovoltaic mode, using the built-in electric field to separate photogenerated carriers. In this mode, the device has no memory effect and a fast response speed (microsecond level), which is suitable for real-time imaging. The GaN-based multifunctional optoelectronic device 11 is irradiated with blue light or violet light, and the GaN-based multifunctional optoelectronic device 11 generates current. After the current is amplified by the transimpedance amplifier 13, it is input to the lower-level machine 2. The lower-level machine 2 converts the received electrical signal and outputs it to the upper-level machine 3. The upper-level machine 3 converts the electrical signal into a visual image to express the sensitivity of the device.

[0093] (2) Memory pattern

[0094] The second switch Q2 and the third switch Q3 are open, while the first switch Q1 and the fourth switch Q4 are closed. The first switch Q1 is connected to the GaN-based multifunctional optoelectronic device. After the lower-level device 2 detects the signal, it outputs a specific waveform. First, it outputs a high level of 3300mV for 10 seconds. The strong electric field generated by this high voltage assists the photogenerated holes to be captured by deep-level defects, completing the "synaptic weight writing". After 10 seconds, the control voltage is automatically reduced to 2000mV and maintained as the "read voltage". The current collected at this time reflects the memory strength of the synapse. At this time, the second switch Q2 and the third switch Q3 are in the open state, cutting off the zero-bias circuit and preventing short circuit.

[0095] As described above, due to the significant difference in current magnitude between the two modes (nA level in probe mode (SPPD) and μA level in memory mode (OS), a single amplification link cannot meet the requirements. To address this issue, transimpedance amplifier 13 is constructed using an operational amplifier (such as ADA4530) with low bias current (I_b < 1 pA). In probe mode, a large resistor (such as 100 MΩ) is connected in parallel with a relay or analog switch to provide a gain of 108 V / A. In memory mode, a small resistor (such as 100 kΩ) is connected in parallel to provide a gain of 105 V / A to prevent signal saturation.

[0096] In this embodiment, as Figure 2 As shown, host computer 3 runs on a PC and is developed using Python. The specific design is as follows:

[0097] (1) In terms of underlying communication, the system constructs an interactive link between the host computer 3 and the slave computer 2 through a USB to serial port (UART), and the communication baud rate is set to 115200 bps. Considering that GaN devices will generate significant electromagnetic interference during the high voltage pulse writing stage, in order to avoid data packet loss or erroneous command triggering, the protocol layer is designed with a strict frame verification mechanism: each data frame uses 0xAA 0x55 as the synchronization start word, contains SPPD and OS mode switching instructions, and uses 4 bytes of single-precision floating-point numbers to encapsulate real-time voltage and current values. At the end, a CRC16 cyclic redundancy check is embedded, thereby ensuring the robustness of command transmission in a strong interference environment.

[0098] (2) In terms of signal processing and backtracking algorithms, after acquiring the original sampling points, the host computer software 3 performs real-time filtering using a moving average algorithm with a window width of 10, effectively eliminating high-frequency glitches and smoothing the waveform trajectory. To support in-depth analysis after the experiment, the system background maintains a buffer with a capacity of 60 seconds of data in real time. This design supports the core "memory backtracking" function, that is, after the user triggers backtracking, the program can automatically locate and retrieve the historical data of the most recent 3.3V pulse writing stage. These historical records are overlaid on the screen with the current red real-time curve in the form of gray dashed lines. Through intuitive visual comparison, the experimenter can accurately capture the changing characteristics of the photocurrent decay rate. If a significant lag is observed in the second decay compared to the previous one, it can be used as a direct criterion for the GaN device to achieve the "relearning enhancement" effect, providing core data support for the quantitative evaluation of neuromorphic computing performance.

[0099] (3) To address the needs of long-cycle aging tests or stability assessments, the system integrates a lightweight web server based on the FastAPI framework. This architecture breaks the limitation of traditional testing systems that can only be monitored via a local PC. Testers only need to access the device's IP address through a mobile terminal (such as a tablet or mobile phone) within the local area network to achieve remote data monitoring. The introduction of this web interaction unit not only optimizes the allocation of experimental resources but also significantly improves the spatial deployment flexibility during GaN device testing. The web server can be configured using existing mature technologies or products.

[0100] For arrayed devices, the software maps the real-time current values ​​of each channel to grayscale values ​​from 0 to 255, generating a real-time heatmap on the interface. In SPPD mode, this heatmap is the ultraviolet image; in OS mode, the heatmap represents the current "memory weight matrix" to form dynamic imaging, such as... Figures 4 to 6 As shown.

[0101] In this embodiment, as Figure 3 As shown above, the timing control method for the photosensitive and memory function detection system of GaN-based multifunctional optoelectronic devices specifically includes the following steps:

[0102] S100 system initialization: Lower computer 2 is powered on, upper computer 3 starts the Web service, and establishes 115200bps serial communication handshake.

[0103] S200 detects the hardware mode switch status: The lower-level machine 2 reads the current physical position of the 4P3T toggle switch and determines the hardware mode based on the current physical position. If the second switch Q2 and the third switch Q3 are closed and the first switch Q1 and the fourth switch Q4 are open, it is the detection mode, and the detection mode steps S301 to S304 are executed. If the second switch Q2 and the third switch Q3 are open and the first switch Q1 and the fourth switch Q4 are closed, it is the memory mode, and the memory mode steps S401 to S404 are executed.

[0104] S301 Hardware Reconfiguration - Detection Path: Switching matrix action, GaN-based multifunctional optoelectronic device 11 is connected to transimpedance amplifier 13, bias voltage is set to zero;

[0105] S302 Weak Photocurrent Acquisition: Transimpedance amplifier 13 performs IV conversion and high-gain amplification of the current;

[0106] S303 Data Upload and Preprocessing: Lower computer 2 performs analog-to-digital conversion and digital filtering preprocessing, and sends the processed data to upper computer 3;

[0107] S304 Real-time Imaging Visualization: The host computer 3 maps the light intensity to the grayscale space and refreshes the real-time image area. Then, the host computer 3 collects whether the user clicks the "Compare / Backtrack" button on the Web terminal. If no is selected, the program ends or step S200 is executed. If yes is selected, step S500 is executed.

[0108] S401 Hardware Reconfiguration - Memory Path: Switching matrix action, GaN-based multifunctional optoelectronic device 11 is connected in series with sampling resistor 12 to the drive circuit;

[0109] S402 applies a drive pulse sequence: the lower-level machine 2 outputs a voltage of 3300mV for 10s, and then outputs 2000mV and maintains it;

[0110] S403 Memory Signal Monitoring: Instrumentation amplifier 14 acquires the voltage across resistor 12;

[0111] S404 Data storage and status update: The host computer 3 stores data into the circular buffer, the slave computer 2 lights up the memory mode indicator, and then the host computer 3 collects whether the user clicks the "Compare / Backtrack" button on the Web terminal. If no is selected, the program ends or step S200 is executed. If yes is selected, step S500 is executed.

[0112] S500 Data Retrospective: The host computer 3 calls the historical memory data buffer and compares it with the current real-time data on the same screen. Depending on the user's subsequent selection, the program ends or steps S200 are executed.

[0113] The above-described detection system is used in the inspection of GaN-based semiconductor production.

[0114] All electrical components in the system of this invention can be mature products or technologies in the prior art.

[0115] Example 1

[0116] like Figure 2 and Figure 3 As shown, the following describes a typical experimental procedure:

[0117] This embodiment employs an AlGaN / GaN HEMT structure device. The device is in a high-resistivity state (dark current <1 nA) under no light and zero bias; under ultraviolet light (365 nm), it generates photogenerated carriers. Under negative bias or a specific positive pulse, defects in the AlGaN barrier layer trap electrons, causing the current to remain at a high level even after the light exposure stops (PPC effect), i.e., a "memory state".

[0118] The specific steps are as follows:

[0119] 1. After the system is powered on, first turn the toggle switch on the hardware control panel to the middle position (off state). Start the host computer 3 control software. After the program completes the serial port handshake communication with the slave computer 2, the system enters standby monitoring mode.

[0120] 2. SPPD Test: Switch the toggle switch to position two (probe mode). The circuit switches to ground mode, and the bias voltage is set to 0V. The host computer interface 3 simultaneously switches to "SPPD" mode, and the system automatically increases the sampling gain of the transimpedance amplifier 13. The experimenter uses a blue or violet light source pulse to illuminate the GaN device and observes the square wave photocurrent response curve captured in real time by the host computer 3 to evaluate and calibrate the initial sensitivity of the device.

[0121] 3. OS Test: With the switch set to position one (memory mode), the MCU detects the pin level transition and drives the DAC to output a 3.3V high level, entering the "OS" stage. At this time, the device generates a large internal current under high-voltage pulse excitation, and the host computer (3) simultaneously records this charging process. After the preset 10-second write period, the MCU controls the output voltage to drop to a 2.0V sustaining voltage, and the system automatically switches to "OS" mode. At this point, the light source is removed, and the host computer (3) begins recording the curve of the photocurrent slowly decaying over time, used to observe the continuous photoconductivity (PPC) effect.

[0122] 4. Memory Enhancement and Retrospective Comparison Analysis: When the device is in the decay phase, a light pulse is applied again for "relearning" stimulation. Using the "Retrospective Comparison" function in the interface, the software overlays the curves of the two decay processes on the same screen.

[0123] The results are as follows Figures 4 to 6 As shown in the figure, the baseline current level after the second excitation is significantly improved, and the current decay rate is slowed down. This experimental phenomenon directly quantifies the evolution process of the device from short-term plasticity (STP) to long-term plasticity (LTP), verifying its biosynaptic characteristics in neuromorphic computing.

[0124] Although embodiments of the invention have been disclosed for illustrative purposes, those skilled in the art will understand that various substitutions, variations, and modifications are possible without departing from the spirit and scope of the invention and the appended claims. Therefore, the scope of the invention is not limited to the contents disclosed in the embodiments.

Claims

1. A detection system for the photosensitive and memory functions of a GaN-based multifunctional optoelectronic device, characterized in that: This system can detect the photosensitive and memory functions of GaN-based multifunctional optoelectronic devices. The system includes a test circuit (1), a lower-level machine (2), and a higher-level machine (3). The test circuit (1), the lower-level machine (2), and the higher-level machine (3) are connected in sequence. The test circuit (1) includes a switching circuit, a memory mode circuit, and a detection mode circuit. The memory mode circuit and the detection mode circuit are connected in parallel. The memory mode circuit and the detection mode circuit are connected to the lower-level machine (2). The memory mode circuit and the detection mode circuit are both connected to the GaN-based multifunctional optoelectronic devices through the switching circuit. The lower-level machine (2) and the higher-level machine (3) are connected in sequence. The memory mode circuit can perform photocurrent acquisition under zero bias voltage; The probe mode circuit can perform sequential pulse voltage excitation and continuous bias voltage reading to realize the writing and maintenance of photosynaptic weights; The lower-level machine (2) can provide power supply, control switching circuit and data output interface for the system; The host computer (3) can provide a human-computer interaction interface for the system and visualize the data output by the lower computer (2), and perform ultraviolet imaging on the memory mode circuit data; The switching circuit can switch the switching state of the memory mode circuit and the detection mode circuit when it receives the switching command, and simultaneously switch the output bias voltage path and input gain path of the lower-level machine (2).

2. The photosensitive and memory function detection system for GaN-based multifunctional optoelectronic devices according to claim 1, characterized in that: The switching circuit includes a first switch (Q1), a second switch (Q2), a third switch (Q3), and a fourth switch (Q4). The memory mode circuit includes a transimpedance amplifier (13), and the detection mode circuit includes an instrumentation amplifier (14) and a resistor (12). One end of the first switch (Q1), the second switch (Q2), the third switch (Q3), and the fourth switch (Q4) are all connected to the GaN-based multifunctional optoelectronic device (11). The other end of the first switch (Q1) is connected to the lower-level machine (2). The other ends of the second switch (Q2) and the third switch (Q3) are both connected to the transimpedance amplifier (13). The other end of the fourth switch (Q4) is connected to the resistor (12) and the instrumentation amplifier (14). The resistor (12) and the instrumentation amplifier (14) are connected in parallel. The other end of the resistor (12) is connected to the lower-level machine (2) and the instrumentation amplifier (14). The lower-level machine (2) includes a signal acquisition module (21), a device driving module (22), and a data communication module (23). The signal acquisition module (21) is connected to the transimpedance amplifier (13) and the instrumentation amplifier (14). The signal acquisition module (21) can acquire the amplified electrical signal data of the transimpedance amplifier (13) and the instrumentation amplifier (14). The first switch (Q1) is connected to the device driving module (22). The device driving module (22) can provide power input, grounding and control of the signal acquisition module (21) for the switching circuit. The data communication module (23) is connected to the upper-level machine (3). The data communication module (23) can communicate with the upper-level machine (3).

3. The photosensitive and memory function detection system for GaN-based multifunctional optoelectronic devices according to claim 2, characterized in that: The first switch (Q1), the second switch (Q2), the third switch (Q3), and the fourth switch (Q4) are a four-pole three-throw switch integrated into one unit. The transimpedance amplifier (13) is a high-gain transimpedance amplifier (TIA) with a gain of 10. -9 V / A, the instrumentation amplifier (14) is a low-gain instrumentation amplifier (DA) with a gain of 10. 6 V / A.

4. The photosensitive and memory function detection system for GaN-based multifunctional optoelectronic devices according to claim 3, characterized in that: The lower-level machine (2) is a microcontroller unit (MCU), which is selected from the STM32F4 series high-performance microcontroller unit MCU with a main frequency of 168MHz and a built-in 12-bit analog-to-digital converter and a 12-bit digital-to-analog converter. The 4-pole 3-throw switch is an industrial-grade 4-pole 3-throw toggle switch with an insulation resistance greater than 1000MΩ and a contact resistance less than 20mΩ.

5. The photosensitive and memory function detection system for GaN-based multifunctional optoelectronic devices according to claim 1, characterized in that: The host computer (3) maps the real-time current value to a gray value of 0-255 and generates a real-time heat map on the human-computer interaction interface. In SPPD mode, the heat map is the ultraviolet image; in OS mode, the heat map represents the current memory weight matrix to form dynamic imaging.

6. The detection method of the GaN-based multifunctional optoelectronic device photosensitive and memory function detection system as described in any one of claims 2 to 4, characterized in that: Includes the following steps: (1) Detection mode The first switch (Q1) and the fourth switch (Q4) are open, and the second switch (Q2) and the third switch (Q3) are closed. The GaN-based multifunctional optoelectronic device is irradiated with blue light or purple light. The GaN-based multifunctional optoelectronic device generates current. After the current is amplified by the transimpedance amplifier (13), it is input to the lower computer (2). The lower computer (2) converts the received electrical signal and outputs it to the upper computer (3). The upper computer (3) converts the electrical signal into a visual image to express the sensitivity of the device. (2) Memory pattern When the second switch (Q2) and the third switch (Q3) are open, and the first switch (Q1) and the fourth switch (Q4) are closed, the lower-level machine (2) detects a level transition and immediately outputs a voltage of 3.0V to 3.5V as an enhancement voltage for a duration of 5 to 15 seconds to excite carrier trap filling. This causes a surge in current in the GaN-based multifunctional optoelectronic device. The lower-level machine (2) collects the current signal in real time, and the upper-level machine (3) records the current changes in real time. After the current stabilizes, the lower-level machine (2) outputs a voltage of 1.5V to 2.5V. V is used as a holding voltage to maintain photoconductive gain and read the synaptic weight decay curve. The GaN-based multifunctional optoelectronic device is illuminated by a light source, and the GaN-based multifunctional optoelectronic device generates current. After the current is amplified by the instrumentation amplifier (14), it is input to the lower computer (2). The lower computer (2) converts the received electrical signal and outputs it to the upper computer (3). The upper computer (3) converts the electrical signal into a visualized image. At this time, the illumination is removed, and the screen displays the curve of the current decaying slowly. The current collected at this time reflects the memory strength of the synapse.

7. The detection method according to claim 6, characterized in that: The boost voltage is 3.3V, the duration is 10 seconds, and the holding voltage is 2.0V; The lower-level machine (2) of the system integrates multi-level filtering algorithms. In addition to the RC filter in the hardware, the lower-level machine also runs a moving average filter and a power frequency notch filter.

8. The application of the GaN-based multifunctional optoelectronic device photosensitive and memory function detection system as described in any one of claims 1 to 4 in the inspection of GaN-based semiconductor production.

Images