summary
Researchers Wang Lina and Deng Jie from the School of Automation Science and Electrical Engineering at Beihang University and the School of Electrical Engineering at Zhejiang University wrote an article in the fourth issue of the Journal of Electrical Engineering in 2019, pointing out that accurate online extraction of the junction temperature of power semiconductor devices is an important foundation for realizing intelligent control, performance evaluation, active thermal management, health status assessment, and optimization of device lifespan.
This paper reviews the mechanisms and main characteristics of existing methods for extracting junction temperatures of Si (silicon) and SiC (silicon carbide) power devices based on temperature-sensitive parameters, and evaluates the performance of these methods in terms of sensitivity, measurement frequency, invasiveness, and linearity. Based on this, and considering the temperature characteristics of SiCJFET (junction field-effect transistor) devices, a novel non-invasive online junction temperature extraction method for SiCJFET devices based on the gate-source parasitic PN junction breakdown voltage is proposed. Simulation results demonstrate the correctness and effectiveness of the proposed method.
With the technological development and process advancements of power semiconductor devices (hereinafter referred to as power devices), power electronic systems are playing an increasingly important role in high-efficiency power conversion systems such as DC transmission, power supplies, motor drives, microgrids, renewable energy generation, and energy storage. Consequently, the reliability requirements for power electronic systems are also becoming increasingly stringent. Reliability surveys of power electronic systems indicate that power devices have the highest failure rate in converter systems, accounting for approximately 34%. Therefore, research on the operational reliability of power devices is a key focus of power electronic system reliability research.
According to the failure location, the failure of power devices can be divided into two categories: chip failure and package failure. The factors that induce chip failure mainly include electrical overstress, electrostatic discharge damage, aluminum electrode metal reconstruction, thermal runaway, etc., which can be summarized into electrical breakdown and thermal breakdown. Reference [4] proposed that the essence of electrical breakdown failure is also due to excessive temperature, which eventually leads to thermal breakdown failure. Package failure can be divided into two types: bonding wire failure and solder layer fatigue. Bonding wire failure is mainly caused by high junction temperature, while solder layer fatigue is mostly caused by long-term thermal cycling. Although the manifestations are different, chip failure and package failure are related to factors such as the highest junction temperature, junction temperature fluctuation range and rate of change, and average junction temperature. It can be seen that real-time junction temperature monitoring of online devices is the key to the reliable operation of monitoring devices and systems.
On the one hand, in the past development of power electronic systems, designers mostly relied on device datasheets, leaving a large margin based on experience. On the other hand, device datasheets usually only provide the static maximum values of thermal characteristic parameters, and the estimated device junction temperature obtained based on this inevitably deviates from the actual junction temperature.
Reference [8] shows that the junction temperature calculated based on the thermal impedance parameters in the device datasheet and experimental conditions is higher than the actual junction temperature obtained by using an infrared camera. To solve this problem, the cooling system in practical engineering applications is usually designed to be too large, increasing the system volume, weight and cost. Both of these aspects will reduce the system's cost-effectiveness. Therefore, accurate junction temperature extraction will undoubtedly help improve the cost-effectiveness of the system's power density.
Meanwhile, junction temperature is one of the main factors affecting device power consumption and switching characteristics. Accurate junction temperature extraction can provide important basis for intelligent device control, performance evaluation, active thermal management, health status assessment, and optimization of device lifespan. Therefore, junction temperature extraction is crucial for improving the reliability of power devices and power electronic application systems.
This paper first analyzes the necessity of online junction temperature extraction for power devices, then summarizes the current research status of junction temperature extraction methods for power devices at home and abroad, and provides a detailed summary of the temperature-sensitive parameter method applicable to online junction temperature extraction of fully packaged devices. Based on this, it also looks forward to new junction temperature extraction methods for silicon carbide (SiC) power devices.
1. The necessity of online junction temperature extraction
With the increasingly widespread application of power electronic systems and the continuous improvement of power levels, the requirements for the reliability of power electronic systems are also becoming more and more stringent. Real-time monitoring of device operating status, extraction of relevant parameters, and fault diagnosis and prediction are common methods to improve the reliability of power device applications; active thermal management and intelligent control are also important ways to improve the reliability of power device applications.
1.1 Status Monitoring
Currently, the health status of power devices is typically assessed by monitoring parameters such as on-state resistance, threshold voltage, turn-off time, and gate signal, all of which are related to junction temperature. In IGBT devices packaged using aluminum wire bonding technology, the bonding wires are a relatively fragile component, and their health status can be determined by changes in on-state resistance.
However, increased chip junction temperature leads to decreased carrier mobility, which in turn causes changes in the device's on-state resistance. Using on-state resistance alone as a health status monitoring indicator may result in misjudgments of the device's health. Other temperature-sensitive health monitoring parameters also present similar issues. Therefore, to accurately assess device health, it is necessary to extract the device junction temperature online in real time and compensate for any discrepancies in the device health status monitoring indicators.
Solder layer fatigue is also a significant factor contributing to device failure. Detecting changes in thermal resistance is a common method for assessing solder layer fatigue. Currently, thermal resistance is primarily calculated using the relationship between chip power consumption, case temperature, and junction temperature, and accurate online acquisition of device junction temperature information is also necessary.
1.2 Active Thermal Management
Overheating and wide-range thermal cycling are two major causes of power device failure. When the continuous operating temperature of a power device exceeds the safe junction temperature limit, device failure may occur. Wide-range thermal cycling induces physical wear and tear on the device; when the accumulated physical wear and tear exceeds a certain limit, the power device fails. These two failure mechanisms correspond to device failures caused by overheating and junction temperature cycling, respectively. Therefore, adjusting the power consumption of the power converter in real time based on the junction temperature to achieve active thermal management helps reduce the occurrence of such failures and improves system reliability.
To address the problem of uneven temperature distribution caused by uneven current distribution in multi-chip parallel power modules or multi-module systems, reference [13] proposes to adjust the magnitude of the drive signal in real time according to the junction temperature, balance the current distribution, reduce local junction temperature fluctuations, and improve system reliability.
As the above analysis shows, accurately acquiring real-time junction temperature information online is key to achieving status monitoring and proactive thermal management, and also crucial for improving system reliability.
2 Commonly Used Junction Temperature Extraction Methods
Semiconductor chips are typically packaged inside modules, making direct contact difficult and junction temperature hard to observe. Currently, scholars both domestically and internationally have conducted in-depth research on methods for extracting and detecting the junction temperature of power devices, proposing various approaches. Based on the physical characteristics of these methods, they can be mainly categorized into four types: optical methods, thermal network methods, physical contact methods, and temperature-sensitive parameter methods.
Optical methods are not suitable for online measurement, while thermal network methods and physical contact methods are suitable for online measurement, but it is difficult to obtain the true junction temperature of the chip. Reference [14] has already sorted out these three types of junction temperature extraction methods and made a comprehensive comparison, which will not be repeated here. Thermosensitive parameter methods are suitable for online measurement, with fast response speed and low cost, and have received widespread attention from scholars at home and abroad, and have gradually become the research hotspot for online junction temperature extraction. Reference [14] only analyzed and summarized a few thermosensitive parameter methods, and there are still many thermosensitive parameter methods that have not been discussed. This paper will comprehensively sort out the thermosensitive parameter methods that have emerged from a perspective different from that of Reference [14], in a more objective and broader way.
The physical mechanism of the temperature-sensitive parameter method lies in the fact that some internal physical parameters of a device change with junction temperature. For example, intrinsic carrier concentration and carrier lifetime increase with increasing temperature, while carrier mobility decreases. These changes in internal physical parameters with junction temperature cause corresponding shifts in the device's electrical parameters, manifested externally as changes in parameters such as on-state resistance/voltage drop, turn-on/turn-off delay, and voltage/current change rate. These electrical parameters that change with junction temperature are called temperature-sensitive parameters.
Different temperature-sensitive parameter methods have their own advantages and disadvantages in terms of sensitivity, linearity, and robustness. Appropriate temperature-sensitive parameters must be selected based on the specific device and actual operating conditions for junction temperature extraction. The following is a summary of existing junction temperature extraction methods using temperature-sensitive parameters, both domestically and internationally.
3. Junction temperature extraction method based on temperature-sensitive parameters (omitted)
Among numerous temperature-sensitive parameters, those directly affected by junction temperature changes are called direct temperature-sensitive parameters. These include threshold voltage, on-state resistance/voltage drop, Miller capacitance, and transconductance. Parameters affected by one or more direct temperature-sensitive parameters (equivalent to being indirectly affected by junction temperature changes) are called indirect temperature-sensitive parameters. These include turn-on/turn-off delay, short-circuit current, and voltage/current change rate. This paper categorizes junction temperature extraction methods based on direct and indirect temperature-sensitive parameters, as shown in Figure 1. These temperature-sensitive parameter methods are described below.
Figure 1 Classification diagram based on temperature-sensitive parameters
3.1 Junction Temperature Extraction Method Based on On-State Resistance/On-State Voltage Drop
3.2 Junction Temperature Extraction Method Based on Threshold Voltage
3.3 Multi-temperature-sensitive parameter comprehensive junction temperature extraction method
4. Characteristic Analysis by Temperature Sensitive Parameter Method
Thermosensitive parameter method has received widespread attention due to its promising application prospects in online measurement. However, in practical engineering applications, the following issues need to be considered: the influence of parasitic parameters and device aging on junction temperature measurement, the influence of the correlation between thermosensitive parameters and external circuits on junction temperature measurement, and the impact of additional circuitry for junction temperature measurement on system operation. The relevant characteristics of the thermosensitive parameter method are analyzed below, and some key issues are summarized.
4.1 Correction Issues
Before applying the temperature-sensitive parameter method, the device's temperature-sensitive parameters need to be calibrated to obtain the functional relationship between the temperature-sensitive parameters and the junction temperature. During calibration, the device's self-heating should be limited as much as possible to ensure that the device's junction temperature is consistent with the controlled external ambient temperature. When the linear relationship between the temperature-sensitive parameters and the junction temperature is good, the calibration sampling interval can be increased to reduce the calibration workload.
The threshold voltage depends on parameters such as gate structure and carrier concentration. Even slight differences in the manufacturing process can lead to variations in the threshold voltage, resulting in differences in threshold voltage between different devices of the same model. Junction temperature extraction methods based on threshold voltage and temperature-sensitive parameter methods affected by threshold voltage, such as the Miller delay method and turn-on delay method, require calibration for each device before application.
The on-state resistance of a device consists of the chip's on-state resistance and the resistance of its connecting wires. In a power module with multiple chips connected in parallel, the chip lead impedance varies depending on the chip's position within the module, resulting in a dispersion in the on-state resistance of each chip within the module. Therefore, the temperature-sensitive parameter method based on on-state resistance requires calibration for each chip within the module, which is a significant workload.
Device aging causes bonding wire performance degradation or even detachment, reducing the effective contact area between the bonding wire and the chip and increasing on-state resistance. Threshold voltage also changes with device aging. At the end of the device's lifespan, aging-induced parameter shifts severely affect the accuracy of all junction temperature extraction methods related to threshold voltage and on-state resistance. To ensure accurate junction temperature measurements, several calibrations or appropriate compensation methods may be necessary throughout the device's lifespan.
4.2 Performance Evaluation Using the Temperature-Sensitive Parameter Method
Whether the temperature-sensitive parameter method can be applied in engineering practice is a key focus of both academia and industry. When evaluating the applicability of a temperature-sensitive parameter method, it is necessary to comprehensively consider factors such as the accuracy and sensitivity of the temperature-sensitive parameter itself, the ease of junction temperature measurement, and the performance requirements of the measurement circuit. Table 1 summarizes the performance characteristics of existing temperature-sensitive parameter methods. Different temperature-sensitive parameter methods exhibit different characteristics, which are briefly outlined below.
The on-state resistance method, the small current saturation voltage drop method, and the large current injection method are the most versatile. However, the on-state resistance method and the large current injection method require simultaneous measurement of the device voltage drop and load current, while the small current saturation voltage drop method requires the addition of an auxiliary constant current source. All three methods place high demands on the measurement circuitry. Deviations introduced by inconsistencies between the connection lines and the chip junction temperature can cause measurement errors in practical applications of the on-state resistance method, the small current saturation voltage drop method, and the short-circuit current method. Therefore, certain methods must be employed to avoid introducing these errors.
The drive voltage difference ratio method and collector turn-on voltage method can avoid measurement errors introduced by lead temperature differences, but these two methods have certain drawbacks in terms of system reliability and sensitivity, respectively. The high current injection method, short-circuit current method, drive voltage difference ratio method, and collector turn-on voltage method all measure junction temperature under high current conditions. The self-heating introduced by the high current during calibration cannot be ignored. If certain compensation measures are not taken, it will lead to deviations in the actual junction temperature measurement.
Table 1 Performance evaluation of the temperature-sensitive parameter method
Indirect temperature-sensitive parameter methods and threshold voltage-based temperature-sensitive parameter methods are related to the switching transient process of the device. The measurement bandwidth is determined by the device's switching frequency, and non-invasive real-time monitoring of the device junction temperature can be performed in each switching cycle, which has great application value. However, power devices have a fast switching speed, which places high demands on the bandwidth of the measurement device, and the temperature-sensitive parameters are also susceptible to noise interference caused by parasitic parameters inside the converter system.
Time-sensitive temperature parameter methods, such as the turn-on/turn-off delay method and the Miller delay method, have low sensitivity. Even a small measurement error in the sensor can lead to a large error in the junction temperature measurement, and they require a high-resolution timer. Therefore, the sensor's measurement bandwidth and accuracy are critical factors for indirect temperature parameter methods. Furthermore, the calibrated operating conditions must strictly match the actual operating conditions to obtain an accurate value for the actual junction temperature. Indirect temperature parameter methods are more dependent on operating conditions, such as bus voltage and load current, increasing the complexity of the power converter junction temperature measurement and control strategy.
4.3 Limitations of the Temperature Parameter Method
In practical engineering applications, the highest junction temperature and junction temperature fluctuation have the greatest impact on the reliability of the device. The temperature-sensitive parameter method reflects the average junction temperature of the device, and different temperature-sensitive parameters reflect the temperature of different parts of the device. Reference [26] shows that the threshold voltage mainly reflects the channel temperature of the device, the junction temperature of the MOSFET body diode mainly reflects the junction temperature of the body diode region, and the on-state resistance mainly reflects the temperature of the drift region.
For power modules with multiple chips connected in parallel via a common terminal leading to a common electrode, the junction temperature extracted by the temperature-sensitive parameter method is affected by all parallel devices and does not directly correspond to the highest, lowest, or average junction temperature of a particular chip.
The accuracy, sensitivity, and anti-interference capability of the measurement circuit and instrument are also significant challenges when using the temperature-sensitive parameter method for junction temperature extraction. As analyzed above, different temperature-sensitive parameter junction temperature extraction methods have different characteristics, and the appropriate method must be selected based on the actual operating conditions and measurement requirements.
Despite the limitations and challenges of the temperature-sensitive parameter method, it shows promising development prospects among all junction temperature measurement methods due to its fast response speed and suitability for online measurement.
Outlook on Junction Temperature Measurement Methods for 5SiC Power Devices
Compared to silicon (Si), SiC (SiC) materials have a wider band gap, higher breakdown field strength, and higher electron saturation drift velocity. These characteristics give SiC power devices a better development prospect than Si power devices in high-voltage, high-temperature, and high-frequency applications. In recent years, SiC devices have developed rapidly, with SiC SBD (Schottky diode), SiC JFET (junction field-effect transistor), SiCBJT (bipolar junction transistor), and SiC MOSFET being commercialized one after another. Significant breakthroughs have also been achieved in SiCIGBT and SiCGTO.
Currently, the technological maturity of SiC devices is sufficient to meet the development needs of power converter products for various applications, including automotive traction drives, aerospace equipment, and others requiring high temperature, high efficiency, high frequency, and high power density. With the successful development of SiC power converter products and industrial prototypes for different applications, the temperature characteristics of SiC devices and their long-term reliability at high temperatures have become a key focus for both academia and industry.
5.1 Problems with applying the temperature-sensitive parameter method to SiC power devices
Currently, research on junction temperature extraction for SiC power devices is still in its early stages. Due to differences in semiconductor material properties, device structure, and fabrication processes compared to Si power devices, the following issues need to be considered when exploring methods for extracting the junction temperature, a temperature-sensitive parameter, from SiC power devices:
(1) Compared with Si power devices, SiC power devices have faster switching speeds, larger voltage and current change rates, and can operate at higher switching frequencies. The accuracy of the measurement circuit and instruments, as well as the impact of switching noise on the measurement circuit, are key issues to consider when extracting junction temperature using switching transient temperature-sensitive parameters (temperature-sensitive parameters related to the switching transient process).
(2) SiC power devices have a fast switching speed, and the effects of voltage and current change rates and parasitic capacitance and inductance coupling are more severe. During the switching commutation transient process, voltage and current waveform oscillations are relatively severe. The voltage and current oscillations of the power loop can affect the transient characteristics of the drive circuit through Miller capacitance, which will undoubtedly increase the measurement difficulty of the transient temperature-sensitive parameter method of the drive circuit.
(3) Due to differences in semiconductor materials, microscopic physical parameters, and structures, the temperature characteristics of SiC power devices differ from those of Si power devices. For example, junction temperature information can be extracted by measuring the Miller capacitance charging and discharging time of Si power devices. However, the Miller capacitance of SiC power devices at the same voltage-current rating is smaller than that of Si power devices, and the charging and discharging time of the Miller capacitance of SiC power devices is shorter, making measurement more difficult. The breakdown field strength of SiC material is about 10 times that of Si material, and a thinner drift layer can achieve higher withstand voltage. Under the same withstand voltage, the on-resistance per unit area of SiC devices is lower, thus requiring higher accuracy of temperature-sensitive parameter measurement instruments based on on-state resistance. Therefore, the junction temperature extraction method for SiC devices cannot be directly copied from the junction temperature extraction method for Si devices.
(4) Different SiC power devices have different structures, and it is necessary to summarize and study them based on the individual characteristics and commonalities of junction temperature changes. For example, both SiCJFET and SiCMOSFET can be regarded as resistive elements in the on state. Although the on-state resistance of SiCMOSFET generally exhibits a positive temperature characteristic above room temperature, its channel resistance exhibits a negative temperature characteristic due to the influence of the gate oxide layer of SiCMOSFET. Therefore, compared with SiCJFET devices, the temperature characteristic change of the on-state resistance of SiCMOSFET is smaller.
If the on-state resistance is used as the temperature-sensitive parameter of SiC MOSFET, its sensitivity is not high, the measurement is difficult, and the accuracy is hard to guarantee. Therefore, it is necessary to consider using other temperature-sensitive parameters for junction temperature measurement. SiC JFET devices have a parasitic PN junction between the gate and source, and the temperature characteristics of this PN junction can be used to explore a junction temperature extraction method suitable for SiC JFETs; however, SiC MOSFET devices do not have a parasitic PN junction between the gate and source, so this method is not applicable to SiC MOSFET devices.
Therefore, in practical applications, it is necessary to conduct research on specific devices based on the physical structure of SiC power devices, combined with material properties and working mechanisms.
5.2 Existing methods for extracting junction temperature of SiC power devices based on temperature-sensitive parameters
Scholars both domestically and internationally have explored and studied methods for extracting the junction temperature of SiC power devices based on the temperature-sensitive parameter method. The following is a summary and analysis of the existing research.
Reference [43] proposes that the on-state resistance and gate-source forward voltage drop vGS of normally off SiCJFET devices are both temperature-sensitive parameters, and the junction temperature of the device can be reflected by monitoring the changes of these two parameters.
The calibration method and measurement circuit of the on-state resistance method are similar to those of Si power devices. However, under the same withstand voltage, the on-state resistance value of SiC power devices is smaller, which places higher demands on the accuracy of the measurement instruments. Furthermore, it is also affected by the inconsistency between the connection line temperature and the chip junction temperature, as well as the effects of bonding wire aging. The gate-source forward voltage drop method is suitable for the device's conduction stage, but it is affected by the gate current, requiring simultaneous monitoring of both the gate current and the gate-source voltage. The sensitivity of this method does not change significantly under different gate currents, and its sensitivity is not high, only -2mV/℃.
SiCJFET power devices have negative temperature characteristics in their saturation current. Reference [44] proposes to use the saturation current of the device as a temperature-sensitive parameter and measure the saturation current under a constant drain-source voltage when the gate-source voltage is constant. The saturation current method has high sensitivity and good linearity, but when using this method to measure the saturation current of the device, the control strategy of the device needs to be changed, which will have a certain impact on the operation of the system, and therefore is more invasive.
Reference [45] points out that the forward saturation voltage drop of SiC devices changes with the junction temperature, and the junction temperature of SiC devices can be measured by the small current saturation voltage drop method. This method is relatively universal and applicable to SiC power switching devices and SiC power diodes, but its sensitivity is not ideal. For SiC JFETs and SiC MOSFETs with a body diode, the junction temperature of the device can be reflected by detecting the forward voltage drop of the body diode in the device's off state. This method has good linearity, but its sensitivity is only -2mV/℃, and the body diode temperature cannot characterize the temperature of the device's maximum power loss point.
When the turn-off delay method is applied to SiC power switching devices, the sensitivity is low, only tens of ps/℃, which makes actual measurement difficult and may cause a large junction temperature measurement error. Increasing the external gate resistance or equivalent input capacitance can increase the turn-off delay of the device. Therefore, reference [46] proposes to add a driving auxiliary circuit. When measuring the junction temperature, the auxiliary circuit is started and the external driving resistance is increased to increase the turn-off delay of the device. During normal operation, the driving resistance is restored to a smaller value to meet the faster switching speed.
When this method is applied to SiCMOSFET devices, the sensitivity can be increased from tens of ps/℃ to hundreds of ps/℃, and it has little impact on the normal operation of the system. Similar to this method, reference [47] uses the detection of diDS/dt during the turn-on process to reflect the junction temperature of SiCMOSFET devices. It proposes to increase the gate drive resistance to improve the sensitivity of the diDS/dt temperature-sensitive parameter method when measuring the junction temperature of the device, and to use a smaller drive resistance during normal operation.
In summary, online junction temperature extraction methods for SiC power devices are more challenging than those for Si power devices. Domestic and international scholars are actively exploring the problems and potential solutions related to temperature-sensitive parameter extraction methods for SiC power devices. Finding effective solutions to existing problems, or discovering more valuable temperature-sensitive parameter methods based on the novel characteristics of SiC power devices, are the two main research directions for junction temperature extraction methods for SiC power devices.
5.3 Proposed method for extracting junction temperature of SiC JFET devices
Our research team also explored methods for extracting junction temperature, a temperature-sensitive parameter of SiCJFET power devices.
In-depth analysis of the temperature characteristics of SiC JFET devices revealed that the breakdown voltage vG,br of the parasitic PN junction between the gate and source of the SiC JFET device is also a temperature-sensitive parameter. Based on this, this paper innovatively proposes a method for extracting the junction temperature of SiC JFETs based on the vG,br parameter. The schematic diagram of the measurement circuit is shown in Figure 10. Compared with the traditional driving circuit, this driving circuit adds a parallel DRC circuit. The resistor Rp is relatively large, used to limit the gate current in the JFET's off state; the diode VD provides a low-impedance path for the device's turn-on process; and the capacitor C is used to accelerate the switching speed of the device.
When the gate drive negative voltage VLow is lower than the JFET gate-source PN junction breakdown voltage, the gate-source voltage is approximately equal to vG,br. The DRC circuit bears the voltage difference between VLow and vG,br. Since vG,br has temperature characteristics, under the condition of constant VLow, the voltage drop vRp across Rp can indirectly reflect the change of vG,br with junction temperature.
Figure 10 DRC voltage extraction test circuit
Based on the simulation model of SiCJFET power devices, the test circuit in Figure 10 was simulated in Saber. The relationship curve between vRp and junction temperature when the temperature changes from 25℃ to 175℃ is shown in Figure 11.
Figure 11 shows the simulation curve of the vRp-Tj relationship.
As shown in Figure 11, this method has high sensitivity, but its linearity is not ideal, requiring an increase in the number of calibration points. The vG,br parameter extraction circuit is simple, located in the device drive loop, and suffers from low voltage and current stress. It is highly integrable and can be performed when the device is off, achieving non-invasive measurement. Therefore, this method is an effective approach for extracting the junction temperature of SiC JFET power devices. However, potential problems in practical applications require further investigation.
in conclusion
This paper systematically reviews the methods, principles, and typical characteristics of junction temperature extraction for Si power devices based on temperature-sensitive parameters. It summarizes, analyzes, and compares various existing temperature-sensitive parameter methods from multiple perspectives, and outlines the correction issues, performance evaluation, and limitations of current methods. Furthermore, this paper summarizes and analyzes existing typical junction temperature extraction methods for SiC power devices and proposes a novel junction temperature extraction method for SiC JFET power devices. It is hoped that this work can provide a reference for power device reliability research and lifetime prediction methods.
Disclaimer: This article is a reprint. If it involves copyright issues, please contact us promptly for deletion (QQ: 2737591964 ) . We apologize for any inconvenience.
