Method and system for measuring junction temperature of wide bandgap power semiconductor based on tdtr

Abstract

This invention proposes a transient measurement method for wide-bandgap power semiconductor junction temperature based on TDTR, comprising: windowing the packaged device and depositing a metal thin film as a thermal reflection sensing layer; focusing the pump light and probe light collinearly to the same point on the surface of the metal thin film and controlling their relative arrival time; applying an electrical pulse to the device, using the rising edge of the electrical pulse as a trigger reference to achieve time synchronization between the electrical signal and the laser signal; acquiring the temperature time-series curve by successively changing the probe light delay time in point measurement mode, and / or obtaining a two-dimensional temperature distribution field by scanning point by point with a fixed delay time in surface scanning mode; finally, extracting the in-phase and quadrature components using a lock-in amplifier and inverting the device junction temperature. This invention can perform direct junction temperature measurement under actual switching transient conditions of the device, overcoming the problems of existing measurement methods that cannot simultaneously achieve high spatial resolution and high temporal resolution and are difficult to perform in-situ measurements on power semiconductor devices in packaged state.

Description

Technical Field

[0001] This invention relates to the field of power semiconductor thermal testing technology, and in particular to a method and system for transient measurement of junction temperature of wide-bandgap power semiconductors based on TDTR. Background Technology

[0002] Junction temperature is a core parameter that determines the performance and reliability of SiC / GaN wide bandgap power semiconductor devices. Accurate measurement of junction temperature changes during switching transients is of great significance for device thermal design, reliability assessment and lifetime prediction.

[0003] Currently, existing methods for measuring junction temperature have the following main shortcomings: (1) Electrical parameter method (such as on-state voltage drop method, temperature-sensitive parameter method) indirectly calculates junction temperature by utilizing the characteristics of the electrical parameters of the device changing with temperature. However, it requires interrupting the normal operation of the device to perform the measurement. The junction temperature at the measurement time has deviated from the actual temperature under the actual switching conditions, and it is impossible to obtain the transient temperature response of the device in the nanosecond-level switching process. (2) Infrared thermal imaging uses the relationship between infrared radiation intensity and temperature on the surface of an object to perform non-contact temperature measurement. However, its spatial resolution is limited by the infrared wavelength and can only reach the order of 5~10μm, which cannot distinguish the fine temperature distribution of the micro-region of the chip (such as the active region and near the gate). At the same time, the measurement results are subject to multiple interferences from the packaging structure and the change of surface emissivity, making it difficult to guarantee quantitative accuracy. (3) Thermal resistance method (transient thermal resistance method) extracts thermal resistance parameters by measuring the cooling or heating response curve of the device. However, this method can only obtain the lumped parameters of the overall junction temperature of the chip and cannot obtain the two-dimensional temperature distribution information of different areas on the chip surface, making it difficult to locate hot spots and analyze local thermal failures. (4) Micro Raman spectroscopy uses the relationship between Raman frequency shift and temperature to measure temperature, but its time resolution is usually on the order of milliseconds to seconds, which cannot capture the temperature fluctuations of power devices during the switching transient process on the order of nanoseconds and picoseconds.

[0004] Therefore, there is an urgent need to propose a transient measurement method for wide bandgap power semiconductor junction temperature that can simultaneously achieve submicron spatial resolution and picosecond temporal resolution. Summary of the Invention

[0005] This invention proposes a transient measurement method and system for wide-bandgap power semiconductor junction temperature based on TDTR, which solves the problems of existing junction temperature measurement methods being unable to simultaneously achieve submicron spatial resolution and picosecond temporal resolution, and being difficult to directly measure the junction temperature of packaged power semiconductor devices under real switching transient conditions.

[0006] The technical solution of this invention is implemented as follows: The first aspect of this invention provides a method for transient measurement of wide-bandgap power semiconductor junction temperature based on TDTR, comprising the following steps: Step S1: Perform windowing on the packaged power semiconductor device under test, open an optical observation window on the top of the device package and cover it with a transparent window sheet, and deposit a metal thin film on the surface of the device chip as a heat reflection sensing layer. Step S2: The pump light and probe light emitted by the laser are collinearly focused onto the same point on the surface of the metal thin film through the same objective lens, and the relative arrival time between the pump light and the probe light is controlled by a variable optical delay line; Step S3: Apply an electrical pulse to the power semiconductor device under test to simulate the actual working heating condition. Use the rising edge of the electrical pulse as the time trigger reference to control the arrival time of the probe light relative to the rising edge of the electrical pulse, so as to realize the time synchronization of the three signals: electrical signal, pump light and probe light. Step S4: Fix the laser focusing position, successively change the arrival time of the probe light relative to the rising edge of the electrical pulse, and collect the probe light reflection signal corresponding to each delay time point; and / or The delay time of the probe light is fixed, and the laser focusing position is controlled to move and scan point by point on the surface of the device chip to collect the probe light reflection signal corresponding to each pixel. Step S5: Based on the collected probe light reflection signal, perform phase-locked detection using a lock-in amplifier with the pump light modulation frequency as a reference, and separate and extract the in-phase and quadrature components of the thermal reflection signal; based on the in-phase and quadrature components, invert the junction temperature of the power semiconductor device under test to obtain the transient response curve of the single-point temperature changing with time and / or the two-dimensional temperature distribution field on the chip surface.

[0007] Specifically, step S1 includes the following steps: An optical observation window is opened on the top of the device package, covered with an optical anti-reflection window, and the inside of the optical observation window is filled with inert gas to maintain the integrity of the original heat dissipation path and electrical function of the device. A metal thin film is deposited on the surface of the device chip, the thickness of which is configured to reduce additional interfacial thermal resistance while ensuring the intensity of the thermally reflected signal. A grid calibration mark is prepared on the surface of the metal thin film for spatial positioning and position error compensation during laser scanning.

[0008] Specifically, in step S2, the pump light generates periodic temperature fluctuations on the surface of the metal film by applying high-frequency heating to the metal film. These fluctuations are then used to extract a thermal reflection signal with the same modulation frequency as the pump light from the reflection signal of the probe light through a lock-in amplifier. The probe light is irradiated onto the surface of the metal thin film, and the intensity of the reflected light reflects the reflectivity of the metal thin film. The reflectivity is used to subsequently invert the surface temperature of the metal thin film. The relative arrival time adjustment range between the pump light and the probe light covers the negative delay to positive delay interval. The negative delay interval is used to acquire a reference signal under no-thermal-response conditions to correct system drift errors.

[0009] Specifically, in step S5, the method for retrieving the junction temperature of the semiconductor device under test includes the following steps: Step S51: Preprocess the in-phase and quadrature components of the lock-in amplifier output; Step S52: The device under test is calibrated by step heating using a programmable heating stage to establish a linear mapping relationship between the preprocessed signal amplitude and temperature change; Step S53: Establish a one-dimensional periodic thermal conduction analytical model of the three-layer structure of metal thin film-chip-substrate, and use the interfacial thermal resistance R1 between the metal thin film and the chip and the interfacial thermal resistance R2 between the chip and the substrate as the parameters to be fitted, and calculate the theoretical thermal reflection signal. Step S54: Use a nonlinear least squares fitting algorithm to fit the measured thermal reflection signal with the theoretical thermal reflection signal. When the fitting residual is less than the preset convergence threshold, output the optimal interface thermal resistance R1 and R2. Step S55: Based on the optimal interface thermal resistance and linear mapping relationship, convert the amplitude of the thermal reflection signal extracted by the lock-in amplifier into junction temperature.

[0010] Further, step S51 includes the following steps: Background baseline subtraction: ; in, This refers to the pump-probe timing delay. and These are the original in-phase signal and quadrature signal output from the lock-in amplifier, respectively. and The background baseline signal was acquired without pump light excitation. and These are the effective signal components after background baseline subtraction; Constructing complex thermal reflection signals: ; in, For the measured complex thermal reflection signal, The imaginary unit; Calculate the signal amplitude: ; in, The measured amplitude of the thermal reflection signal. This represents the change in the surface reflectivity of the metal thin film. This indicates that the signal amplitude and the change in reflectivity are linearly positively correlated; Normalization process: ; ; in, For normalized complex signals, This is a reference signal with no thermal diffusion within the -20ps range; To normalize the signal amplitude, and These are operators that take the real part and the imaginary part of a complex number, respectively.

[0011] Further, step S52 includes the following steps: The device under test with a metal thin film deposited on its surface is placed on a programmable heating stage. With the device powered off and without self-heating, the temperature is increased in steps based on room temperature. After each temperature increase, the device is kept at a constant temperature until thermal equilibrium is reached, and the normalized signal amplitude at each temperature point is collected. The normalized signal amplitude and the corresponding temperature rise were fitted using least-squares linear fitting: ; in, For the first Normalized signal amplitude at each temperature point The slope of the signal-temperature fitting. For the first The temperature rise relative to room temperature at each temperature point. This is the fitting intercept.

[0012] Furthermore, in step S53: In the one-dimensional periodic thermal conduction analytical model of the aforementioned three-layer metal thin film-chip-substrate structure, the thermal diffusivity of each layer is: ; Among them, subscript , respectively corresponding to metal thin film layers Chip layer and substrate layer ; For the first Layer thermal diffusivity; For the first Layer thermal conductivity; For the first Layer density; For the first Specific heat capacity under constant pressure; The heat wave count for each layer is: ; in, For the first Layer thermal wave number, The imaginary unit; The pump laser is modulated at an angular frequency; The general solution of the temperature field within the layer is: ; in, For depth The complex amplitude temperature field at that location, The vertical depth coordinates of the device; , These are the undetermined coefficients for the temperature field within the layer; Boundary conditions include: upper surface of metal thin film The boundary is the input boundary of the periodic alternating laser heat flux; Metal Thin Film-Chip Interface The temperature jump condition is met: ; in, The thickness of the metal thin film. Thermal resistance at the metal thin film-chip interface; This refers to the temperature of the metal thin film layer at the metal thin film-chip interface. This refers to the temperature of the chip layer at the metal thin film-chip interface. The thermal conductivity of the metal thin film layer; This represents the temperature gradient of the metal thin film layer at the metal thin film-chip interface. Chip-substrate interface The temperature jump condition is met: ; in, For chip thickness, Thermal resistance at the chip-substrate interface; This refers to the temperature of the chip layer at the chip-substrate interface. This represents the temperature of the substrate layer at the chip-substrate interface. Thermal conductivity of the chip layer; This represents the temperature gradient of the chip layer at the chip-substrate interface. substrate bottom This is the convective heat transfer boundary: ; in, For substrate thickness, The convective heat transfer coefficient; The thermal conductivity of the substrate layer; This represents the temperature gradient of the substrate layer at the bottom of the substrate. The temperature at the bottom of the substrate; The location at the bottom of the substrate; The reference room temperature; The theoretical complex thermal reflection signal output by the model is: ; in, For theoretical complex thermal reflection signals, determined by the delay time With interfacial thermal resistance Joint decision; , These are theoretically in-phase and quadrature signals, respectively.

[0013] Further, step S54 includes the following steps: Construct the fitting residual function: ; in, For the residual loss function, Summation over the entire time domain with full delay; The iteration is considered to have converged when the residual loss function is less than the preset convergence threshold. During the fitting iteration process, the interfacial thermal resistance and The data is limited to a preset range of physically reasonable values. If the convergence result exceeds this range, the data is considered to have failed to fit and is discarded. Interface thermal resistance of the same chip and The parameters are obtained by fitting only once in the central region of the chip, and then reused for subsequent global temperature measurements.

[0014] Further, step S55 includes the following steps: Based on the signal amplitude-temperature calibration curve obtained in step S52 and the optimal interface thermal resistance output in step S54, the measured thermal reflection signal amplitude extracted by the lock-in amplifier is converted into the junction temperature of the power semiconductor device under test. In point measurement mode, the normalized signal amplitude at different delay time points is converted into transient temperature rise: ; in, For delay time The corresponding transient temperature rise of the device, For delay time The corresponding normalized signal amplitude, The slope of the signal-temperature fitting. The thermal reflectance temperature coefficient of the metal thin film; The absolute temperature of the device is calculated using the following formula: ; in, For delay time The corresponding absolute temperature of the device, The reference room temperature; In surface scanning mode, the normalized signal amplitude of each pixel is converted into the absolute temperature of the corresponding position to obtain the temperature value of each pixel on the chip surface. The temperature values ​​of all pixels are then spatially stitched together according to the scanning position to reconstruct the two-dimensional temperature distribution field on the chip surface.

[0015] A second aspect of the present invention provides a transient measurement system for wide-bandgap power semiconductor junction temperature based on TDTR, comprising: The sample preparation module is used to perform windowing on the packaged power semiconductor device under test, open an optical observation window on the top of the device package and cover it with a transparent window sheet, and deposit a metal thin film on the surface of the device chip as a heat reflection sensing layer. A femtosecond laser module is used to generate pump light and probe light, and to focus the pump light and probe light collinearly onto the same point on the surface of the metal thin film through the same objective lens. The femtosecond laser module is equipped with a variable optical delay line to control the relative arrival time between the pump light and the probe light. The electro-thermal synchronization trigger module is used to apply an electrical pulse to the power semiconductor device under test, and use the rising edge of the electrical pulse as a time trigger reference to control the arrival time of the probe light relative to the rising edge of the electrical pulse, so as to realize the time synchronization of the three signals: electrical signal, pump light and probe light. The signal acquisition module is used to acquire the probe light reflection signal corresponding to different delay time points and / or acquire the probe light reflection signal corresponding to different pixel points; the signal acquisition module includes a lock-in amplifier, which performs lock-in detection with the modulation frequency of the pump light as a reference, and separates and extracts the in-phase component and quadrature component of the thermal reflection signal. The temperature inversion module is used to invert the junction temperature of the power semiconductor device under test based on the in-phase and quadrature components, and obtain the transient response curve of the temperature at a single point changing with time and / or the two-dimensional temperature distribution field on the chip surface.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) This invention enables the laser to penetrate the package and directly irradiate the chip surface and deposit a metal thin film as a heat reflection sensing layer by opening a window. The pump light and the probe light are focused in the same line to achieve excitation and detection of the same micro-area. The rising edge of the electrical pulse is used as the trigger reference to achieve time synchronization between the electrical signal and the laser signal. The heat reflection signal is extracted by the lock-in amplifier and then combined with the temperature inversion algorithm to obtain the junction temperature of the device. This method does not require interruption of the device's working state and can directly measure the junction temperature under the actual switching transient conditions of the device. It overcomes the technical problem that the existing junction temperature measurement methods cannot simultaneously take into account high spatial resolution and high temporal resolution and are difficult to perform in-situ measurement of power semiconductor devices in the packaged state. (2) This invention applies high-frequency modulation to the pump light to generate periodic temperature fluctuations on the surface of the metal thin film, so that the reflected signal of the probe light carries a feature tag with the same frequency as the modulation frequency of the pump light. The phase-sensitive detection is performed by the lock-in amplifier with the modulation frequency as a reference, and the in-phase component and quadrature component of the thermal reflection signal are separated and extracted from the strong background noise, so that the extremely weak thermal reflection signal can be effectively extracted. At the same time, the reference signal collected in the negative delay interval is used to correct the system drift error, providing a reliable original data basis for subsequent high-precision temperature inversion. (3) This invention establishes a one-dimensional periodic thermal conduction analytical model of a three-layer structure of metal thin film-chip-substrate, and uses the interface thermal resistance between the metal thin film and the chip and the interface thermal resistance between the chip and the substrate as the parameters to be fitted. The nonlinear least squares fitting algorithm is used to fit the measured thermal reflection signal with the theoretical signal. During the fitting process, a physical reasonable value range constraint is introduced to ensure the physical meaning of the fitting result. After fitting once in the central region of the chip, the interface thermal resistance parameter is reused in the whole domain. This not only considers the heat conduction path in the multilayer structure and the influence of interface thermal resistance on junction temperature measurement, but also takes into account the computational efficiency of massive pixels in the surface scanning mode, which significantly improves the accuracy of temperature inversion and engineering practicality. (4) This invention supports point measurement mode and surface scanning mode. In point measurement mode, the arrival time of the probe light relative to the rising edge of the electrical pulse is changed successively to collect the reflection signal at different delay time points. The transient response curve of the device surface temperature changing with time can be obtained at the same laser focusing position. In surface scanning mode, the delay time of the probe light is fixed and the chip surface is scanned point by point to obtain a two-dimensional temperature distribution cloud map of the chip surface, which can be used to locate hot spots and analyze temperature gradients. The two modes can be used independently or in combination according to the test requirements. This satisfies both the requirement for refined characterization of the dynamic process of device thermal response and the requirement for visualization analysis of the global thermal distribution of the chip. At the same time, the temporal resolution and spatial resolution reach the picosecond level and submicron level, respectively. Overall, it realizes comprehensive, accurate and efficient transient measurement of the junction temperature of packaged power semiconductor devices. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a transient measurement method for wide bandgap power semiconductor junction temperature based on TDTR according to the present invention.

[0019] Figure 2 This is a schematic cross-sectional view of the windowed encapsulated DUT in an embodiment of the present invention.

[0020] Figure 3 This is a timing relationship diagram between the electrical pulse, junction temperature transient response, and TDTR detection signal in an embodiment of the present invention.

[0021] Figure 4 This is a schematic diagram of the surface temperature distribution scanning results of a SiC MOSFET chip in an embodiment of the present invention. Detailed Implementation

[0022] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0023] Reference Figure 1 The first aspect of this invention provides a method for transient measurement of wide-bandgap power semiconductor junction temperature based on TDTR, comprising the following steps: Step S1: Perform windowing on the packaged power semiconductor device under test, opening an optical observation window on the top of the device package and covering it with a transparent window film, and depositing a metal thin film on the surface of the device chip as a heat reflection sensing layer; specifically including the following steps: An optical viewing window is opened on the top of the device package, covered with an optical anti-reflective film, and the inside of the optical viewing window is filled with inert gas to maintain the integrity of the original heat dissipation path and electrical function of the device.

[0024] like Figure 2As shown, the cross-sectional structure of the windowed packaged DUT, from top to bottom, consists of: a sapphire window (0.5 mm, AR coating), a cavity (nitrogen-filled), an aluminum sensing film (100 nm, deposited on the chip surface), a SiC chip (350 μm), a solder layer (SnAg, 100 μm), a DBC substrate (Al2O3, 0.635 mm, with upper and lower copper layers each 0.3 mm), and a heat dissipation base plate (Cu, 3 mm). The thickness and material of each layer, as well as the incident paths of the pump laser and the probe laser, are marked in the figure. The windowed package maintains the integrity of the device's electrical functions and heat dissipation path.

[0025] In a preferred embodiment, the diameter of the optical observation window is 4-6 mm, the transparent window is a sapphire window with a thickness of 0.5 mm, and the sapphire window has a double-sided anti-reflective coating with a transmittance greater than 98% at a wavelength of 800 nm; the optical observation window is filled with nitrogen gas. Through this windowing process, while ensuring that the laser can penetrate the package and irradiate the chip surface, the original heat dissipation path and electrical functions of the device are fully preserved, avoiding deviations in measurement results from actual operating conditions due to damage to the package or alteration of the device structure.

[0026] A metal thin film is deposited on the surface of the device chip, the thickness of which is configured to reduce additional interfacial thermal resistance while ensuring the intensity of the thermally reflected signal.

[0027] In a preferred embodiment, the metal thin film is an aluminum thin film with a thickness of 80-100 nm; a 100 nm aluminum thin film is deposited for SiC devices, and an 80 nm aluminum thin film is deposited for GaN devices. At a wavelength of 800 nm, the thermal reflectance coefficient of the aluminum thin film is dR / dT ≈ 1.5 × 10⁻⁶. -4 K -1 By depositing a thin metal film on the chip surface as a heat reflection sensing layer, the temperature change on the chip surface is converted into a reflectivity change that can be detected by optical methods, giving femtosecond laser temperature measurement technology a detectable physical basis in packaged power devices.

[0028] A grid calibration mark is prepared on the surface of the metal thin film for spatial positioning and position error compensation during laser scanning.

[0029] In a preferred embodiment, the spacing of the grid calibration marks is 50 μm. A micrometer-level spatial coordinate reference is constructed using these grid calibration marks, enabling precise positioning of the laser focusing position and closed-loop compensation for accumulated drift errors during surface scanning. This ensures the spatial accuracy of the two-dimensional temperature distribution cloud map and the quantitative accuracy of hotspot positioning.

[0030] Step S2: The pump light and probe light emitted by the laser are collinearly focused onto the same point on the surface of the metal thin film through the same objective lens, and the relative arrival time between the pump light and the probe light is controlled by a variable optical delay line.

[0031] In a preferred embodiment, the laser is a Ti:sapphire femtosecond laser with a center wavelength of 800 nm, a pulse width of 80–150 fs, a repetition rate of 80 MHz, and an average power of 500–1000 mW. The probe light is obtained by splitting the pump light, and its wavelength is also 800 nm. The objective lens has a numerical aperture NA = 0.75, a magnification of 50×, and a focused spot diameter of less than 2 μm. The time delay adjustment range of the variable optical delay line is -20 ps to 8 ns, with a time resolution of 10 fs; where -20 ps to 0 ps is defined as the time interval during which the probe light leads the pump light, used for system baseline calibration. By focusing the pump light and probe light collinearly to the same point, a high degree of overlap between the pump light heating position and the probe light detection position is ensured, avoiding measurement errors caused by spatial misalignment; at the same time, the common-path structure can effectively cancel out synchronization drift caused by external vibrations, improving measurement stability.

[0032] The pump light generates periodic temperature fluctuations on the surface of the metal film by applying high-frequency heating to the metal film. These fluctuations are then used to extract a thermal reflection signal with the same modulation frequency as the pump light from the reflection signal of the probe light through a lock-in amplifier. The probe light is irradiated onto the surface of the metal thin film, and the intensity of the reflected light reflects the reflectivity of the metal thin film. The reflectivity is used to subsequently invert the surface temperature of the metal thin film. The relative arrival time adjustment range between the pump light and the probe light covers the negative delay to positive delay interval. The negative delay interval is used to acquire a reference signal under no-thermal-response conditions to correct system drift errors.

[0033] By applying high-frequency modulation to a metal thin film using pump light to induce periodic temperature fluctuations, the reflected signal of the probe light carries a characteristic tag with the same frequency as the pump light modulation frequency. This provides a frequency basis for the subsequent lock-in amplifier to extract weak thermal reflection signals from strong background noise. The principle of thermal reflection signal extraction is as follows: the surface reflectivity of a material varies at different temperatures. By real-time acquisition of changes in reflected light in the probe area and combining this with the material's thermal reflectivity, the surface temperature change of the area under test can be characterized.

[0034] Step S3: Apply an electrical pulse to the power semiconductor device under test to simulate the actual working heating condition. Use the rising edge of the electrical pulse as the time trigger reference to control the arrival time of the probe light relative to the rising edge of the electrical pulse, so as to realize the time synchronization of the three signals: electrical signal, pump light and probe light.

[0035] like Figure 3 As shown, the timing relationship between the electrical pulse, the device temperature response, and the TDTR detection signal is as follows: the top is the electrical pulse waveform (rising edge is the zero point of time), and the middle is the transient response curve of the DUT junction temperature (rapidly rising from the initial temperature T0 to the peak temperature T). peak (Then it slowly decays), and the sampling points of the TDTR probe laser at different delay times Δt are shown below. The time axis is expressed on a logarithmic scale, covering the range of 10 ps to 10 ms. The key time constants are marked in the figure: τ1≈1 ns (thermal diffusion time at the aluminum film / SiC interface), τ2≈100 ns (thermal diffusion time within the chip), and τ3≈10 ms (thermal diffusion time of the heat sink base).

[0036] In a preferred embodiment, the pulse width of the electrical pulse is adjustable from 10 ns to 1 ms, and the output current is 0 to 50 A. A synchronous trigger circuit uses the rising edge of the electrical pulse as the time trigger reference, and a programmable delay circuit achieves timing alignment between the probe light and the electrical pulse signal, achieving an electro-optical synchronization time accuracy better than 1 ps. By controlling the arrival time of the probe light using the rising edge of the electrical pulse as the trigger reference, picosecond-level time synchronization between the electrical signal and the laser signal is achieved. This ensures that each probe light sampling precisely corresponds to a specific moment after the electrical pulse heating, thereby guaranteeing the timing accuracy of transient temperature measurements.

[0037] Step S4: Fix the laser focusing position, successively change the arrival time of the probe light relative to the rising edge of the electrical pulse, and collect the probe light reflection signal corresponding to each delay time point; and / or

[0038] The delay time of the probe light is fixed, and the laser focusing position is controlled to move and scan point by point on the surface of the device chip to collect the probe light reflection signal corresponding to each pixel.

[0039] The method supports two operating modes: point measurement mode and area scanning mode. Preferably, in point measurement mode, the TDTR delay scan range is 10 ps to 10 μs, and 200 points are scanned logarithmically; in area scanning mode, the scan step is 5 to 10 μm, and the scan range is determined according to the device size. During area scanning, the position of the scanning platform is compensated in real time using the grid calibration marks to ensure that the spot accuracy is better than ±0.5 μm.

[0040] The point measurement mode, by successively changing the arrival time of the probe light, can acquire the transient response curve of the device surface temperature over time at the same laser focusing position, revealing the heat propagation process inside the device. The surface scanning mode, by fixing the delay time of the probe light and scanning the chip surface point by point, can acquire a two-dimensional temperature distribution cloud map of the chip surface, used to locate hot spots and analyze temperature gradients. The two modes can be used independently or in combination according to testing needs, satisfying both the need for refined characterization of the dynamic process of device thermal response and the need for visual analysis of the global thermal distribution of the chip.

[0041] Step S5: Based on the collected probe light reflection signal, perform phase-locked detection using a lock-in amplifier with the pump light modulation frequency as a reference, and separate and extract the in-phase and quadrature components of the thermal reflection signal; based on the in-phase and quadrature components, invert the junction temperature of the power semiconductor device under test to obtain the transient response curve of the single-point temperature changing with time and / or the two-dimensional temperature distribution field on the chip surface.

[0042] By using a lock-in amplifier with the modulation frequency of the pump light as a reference for phase-sensitive detection, the in-phase and quadrature components of the thermal reflection signal are separated and extracted from the strong background noise. This allows the extremely weak thermal reflection signal to be effectively extracted, providing a reliable raw data foundation for subsequent high-precision temperature inversion.

[0043] The method for inverting the junction temperature of the semiconductor device under test includes the following steps: Step S51: Preprocess the in-phase and quadrature components of the lock-in amplifier output, specifically including the following steps: Background baseline subtraction: ; in, This refers to the pump-probe timing delay. and These are the original in-phase signal and quadrature signal output from the lock-in amplifier, respectively. and The background baseline signal was acquired without pump light excitation. and These are the effective signal components after background baseline subtraction; Constructing complex thermal reflection signals: ; in, For the measured complex thermal reflection signal, The imaginary unit; Calculate the signal amplitude: ; in, The measured amplitude of the thermal reflection signal. This represents the change in the surface reflectivity of the metal thin film. This indicates that the signal amplitude and the change in reflectivity are linearly positively correlated; Normalization process: ; ; in, For normalized complex signals, This is a reference signal with no thermal diffusion within the -20ps range; To normalize the signal amplitude, and These are operators that take the real part and the imaginary part of a complex number, respectively.

[0044] Background baseline subtraction eliminated the interference of stray light and circuit baseline offset on the measurement signal; the construction of complex thermal reflection signal completely preserved the amplitude and phase information of the signal; the reference signal acquired in the negative delay interval (-20ps) was normalized to effectively suppress the systematic error caused by optical path and laser power drift, providing high-precision input data for subsequent model fitting.

[0045] Step S52: Perform step-by-step temperature calibration on the device under test using a programmable heating stage to establish a linear mapping relationship between the preprocessed signal amplitude and temperature change; specifically including the following steps: The device under test with a metal thin film deposited on its surface is placed on a programmable heating stage. With the device powered off and without self-heating, the temperature is increased in steps based on room temperature. After each temperature increase, the device is kept at a constant temperature until thermal equilibrium is reached, and the normalized signal amplitude at each temperature point is collected. The normalized signal amplitude and the corresponding temperature rise were fitted using least-squares linear fitting: ; in, For the first Normalized signal amplitude at each temperature point The slope of the signal-temperature fitting. For the first The temperature rise relative to room temperature at each temperature point. This is the fitting intercept.

[0046] As a preferred implementation, a stepped calibration from 25℃ to 200℃ was performed using 25℃ as the room temperature reference. The test sample was a windowed packaged device with a standard metal thin film deposited on its surface. During the test, the device was powered off and did not generate self-heat. The pump laser and electro-optic modulator were turned off throughout the test, and only the probe light was turned on. After each temperature increase, the device was kept at a constant temperature for 10 minutes to achieve full thermal equilibrium. The 200℃ high-temperature point was verified by both the device sensor and the patch thermocouple. Steady state was determined when the temperature fluctuation was ≤±0.3℃. The stepped temperature calibration established a linear mapping relationship between the normalized signal amplitude and temperature change, enabling the subsequent measured signal amplitude to be quantitatively converted into temperature values, which serves as a quantitative reference for temperature inversion.

[0047] Step S53: Establish a one-dimensional periodic thermal conduction analytical model of the three-layer structure of metal thin film-chip-substrate, and use the interfacial thermal resistance R1 between the metal thin film and the chip and the interfacial thermal resistance R2 between the chip and the substrate as the fitting parameters to calculate the theoretical thermal reflection signal.

[0048] In a preferred embodiment, the metal thin film is an aluminum thin film, the chip is a SiC chip or a GaN chip, and the substrate is a DBC substrate (Al2O3). For SiC devices, the aluminum thin film thickness is 100 nm, and the chip thickness is approximately 350 μm; for GaN devices, the aluminum thin film thickness is 80 nm. The substrate thickness is 0.635 mm. By establishing a three-layer one-dimensional periodic thermal conduction analytical model, the heat conduction path in the multilayer structure and the influence of interface thermal resistance on junction temperature measurement are considered, providing a physical model support for accurately extracting the junction temperature from the measured thermal reflection signal.

[0049] In the one-dimensional periodic thermal conduction analytical model of the aforementioned three-layer metal thin film-chip-substrate structure, the thermal diffusivity of each layer is: ; Among them, subscript , respectively corresponding to metal thin film layers Chip layer and substrate layer ; For the first Layer thermal diffusivity; For the first Layer thermal conductivity; For the first Layer density; For the first Specific heat capacity under constant pressure; The heat wave count for each layer is: ; in, For the first Layer thermal wave number, The imaginary unit; The pump laser is modulated at an angular frequency; The general solution of the temperature field within the layer is: ; in, For depth The complex amplitude temperature field at that location, The vertical depth coordinates of the device; , These are the undetermined coefficients for the temperature field within the layer; Boundary conditions include: upper surface of metal thin film The boundary is the input boundary of the periodic alternating laser heat flux; Metal Thin Film-Chip Interface The temperature jump condition is met: ; in, The thickness of the metal thin film. Thermal resistance at the metal thin film-chip interface; This refers to the temperature of the metal thin film layer at the metal thin film-chip interface. This refers to the temperature of the chip layer at the metal thin film-chip interface. The thermal conductivity of the metal thin film layer; This represents the temperature gradient of the metal thin film layer at the metal thin film-chip interface. Chip-substrate interface The temperature jump condition is met: ; in, For chip thickness, Thermal resistance at the chip-substrate interface; This refers to the temperature of the chip layer at the chip-substrate interface. This represents the temperature of the substrate layer at the chip-substrate interface. Thermal conductivity of the chip layer; This represents the temperature gradient of the chip layer at the chip-substrate interface. substrate bottom This is the convective heat transfer boundary: ; in, For substrate thickness, The convective heat transfer coefficient; The thermal conductivity of the substrate layer; This represents the temperature gradient of the substrate layer at the bottom of the substrate. The temperature at the bottom of the substrate; The location at the bottom of the substrate; The reference room temperature; In a preferred embodiment, the convective heat transfer coefficient The value is 15W / (m 2 ·K), which is the standard value for natural atmospheric convection heat transfer, perfectly matches the heating platform to the actual test scenario, and solves the high temperature error problem caused by the idealized adiabatic boundary during long-term thermal diffusion.

[0050] The theoretical complex thermal reflection signal output by the model is: ; in, For theoretical complex thermal reflection signals, determined by the delay time With interfacial thermal resistance Joint decision; , These are theoretically in-phase and quadrature signals, respectively.

[0051] Step S54: Using a nonlinear least squares fitting algorithm, the measured thermal reflection signal is fitted with the theoretical thermal reflection signal. When the fitting residual is less than a preset convergence threshold, the optimal interface thermal resistances R1 and R2 are output. Specifically, this includes the following steps: Construct the fitting residual function: ; in, For the residual loss function, Summation over the entire time domain with full delay; The iteration is considered to have converged when the residual loss function is less than the preset convergence threshold. During the fitting iteration process, the interfacial thermal resistance and The data is limited to a preset range of physically reasonable values. If the convergence result exceeds this range, the data is considered to have failed to fit and is discarded. Interface thermal resistance of the same chip and The parameters are obtained by fitting only once in the central region of the chip, and then reused for subsequent global temperature measurements.

[0052] In a preferred embodiment, the preset convergence threshold is 10. -6 This serves as a universal standard for high-precision TDTR testing. The entire fitting and iterative process is locked. and The range of values ​​is determined. If the parameters exceed the physically reasonable range after convergence, the fitting is directly judged as invalid, and solutions without physical meaning are not included in temperature inversion. The parameters of the same chip interface are globally unique and are obtained by fitting once through the stable region in the center of the chip. The parameters are reused for global temperature measurement, and repeated fitting pixel by pixel is prohibited. This ensures the physical meaning of the fitting results while taking into account the computational efficiency of massive pixels in the area scanning mode.

[0053] Step S55: Based on the optimal interface thermal resistance and linear mapping relationship, convert the amplitude of the thermal reflection signal extracted by the lock-in amplifier into junction temperature; specifically including the following steps: Based on the signal amplitude-temperature calibration curve obtained in step S52 and the optimal interface thermal resistance output in step S54, the measured thermal reflection signal amplitude extracted by the lock-in amplifier is converted into the junction temperature of the power semiconductor device under test. In point measurement mode, the normalized signal amplitude at different delay time points is converted into transient temperature rise: ; in, For delay time The corresponding transient temperature rise of the device, For delay time The corresponding normalized signal amplitude, The slope of the signal-temperature fitting. The thermal reflectance temperature coefficient of the metal thin film; The absolute temperature of the device is calculated using the following formula: ; in, For delay time The corresponding absolute temperature of the device, The reference room temperature; In surface scanning mode, the normalized signal amplitude of each pixel is converted into the absolute temperature of the corresponding position to obtain the temperature value of each pixel on the chip surface. The temperature values ​​of all pixels are then spatially stitched together according to the scanning position to reconstruct the two-dimensional temperature distribution field on the chip surface.

[0054] like Figure 4 The image shows an example of the surface temperature distribution scan results of a SiC MOSFET chip after a pulse power of 100W, pulse width of 1μs. The color scale changes from blue (25℃) to red (125℃), clearly showing that the temperature is highest at the center of the active region (125℃), decreasing towards the edges; the gate pad area has a lower temperature (35℃) because there is no power consumption; the edge termination region has a temperature between the two. The temperature distribution is asymmetrical and elliptical, consistent with the cell layout of the device. The location and temperature values ​​of two hot spots are also marked in the figure. Through the above calculations, in point measurement mode, the transient response curve of the temperature at a single point on the surface of the device under test over time can be output, reflecting the complete process of heat propagation from the metal thin film to the chip interior and substrate, including: the thin film thermal diffusion effect on a nanosecond timescale, the chip body thermal diffusion effect, and the heat dissipation base thermal diffusion effect on a millisecond timescale; in surface scan mode, a two-dimensional temperature distribution cloud map of the chip surface can be output for locating hot spots and analyzing temperature gradients.

[0055] A second aspect of the present invention provides a transient measurement system for wide-bandgap power semiconductor junction temperature based on TDTR, comprising: The sample preparation module is used to perform windowing on the packaged power semiconductor device under test, open an optical observation window on the top of the device package and cover it with a transparent window sheet, and deposit a metal thin film on the surface of the device chip as a heat reflection sensing layer. A femtosecond laser module is used to generate pump light and probe light, and to focus the pump light and probe light collinearly onto the same point on the surface of the metal thin film through the same objective lens. The femtosecond laser module is equipped with a variable optical delay line to control the relative arrival time between the pump light and the probe light. The electro-thermal synchronization trigger module is used to apply an electrical pulse to the power semiconductor device under test, and use the rising edge of the electrical pulse as a time trigger reference to control the arrival time of the probe light relative to the rising edge of the electrical pulse, so as to realize the time synchronization of the three signals: electrical signal, pump light and probe light. The signal acquisition module is used to acquire the probe light reflection signal corresponding to different delay time points and / or acquire the probe light reflection signal corresponding to different pixel points; the signal acquisition module includes a lock-in amplifier, which performs lock-in detection with the modulation frequency of the pump light as a reference, and separates and extracts the in-phase component and quadrature component of the thermal reflection signal. The temperature inversion module is used to invert the junction temperature of the power semiconductor device under test based on the in-phase and quadrature components, and obtain the transient response curve of the temperature at a single point changing with time and / or the two-dimensional temperature distribution field on the chip surface.

[0056] The present invention will be further described in detail below through specific embodiments.

[0057] Example 1: Transient Measurement of SiC MOSFET Junction Temperature

[0058] The device under test is a 1200V / 40mΩ SiC MOSFET. It is packaged with an open window according to step S1. An optical observation window with a diameter of 5mm is opened on the top of the package and covered with a 0.5mm thick sapphire window sheet (double-sided anti-reflection coating, transmittance greater than 98% at 800nm ​​wavelength). The interior is filled with nitrogen. A 100nm aluminum thin film is deposited on the chip surface as a heat reflection sensing layer, and grid calibration marks with a spacing of 50μm are prepared on the surface of the aluminum thin film.

[0059] Measurement conditions: According to step S2, a Ti:sapphire femtosecond laser (center wavelength 800nm, pulse width 100fs, repetition frequency 80MHz, average power 800mW) is used. The pump light and probe light are collinearly focused onto the aluminum film on the chip surface through the same objective lens (NA=0.75, 50×). The spot diameter is less than 2μm, and the pump light modulation frequency is 1MHz. According to step S3, the electrical pulse generator outputs a power pulse of 100W with a pulse width of 1μs. The rising edge of the electrical pulse is used as the time trigger reference, and the timing alignment of the probe light and the electrical pulse signal is achieved through a programmable delay. According to step S4, in point measurement mode, the TDTR delay scan range is 10ps~10μs, and 200 points are scanned logarithmically. In area scan mode, the scan range is 2mm×2mm, with a step size of 10μm (200×200 pixels).

[0060] Measurement results: Following step S5, the reflected signal was acquired and the junction temperature was retrieved. At the end of the pulse (1 μs), the temperature at the center of the active region was 85.3℃ (initial 25℃, ΔT = 60.3K), and the temperature retrieval uncertainty was ±1.8℃. The hot spot was located 0.3 mm to the right of the chip center, with a temperature of 92.1℃. The delayed scan results showed that the temperature rapidly rose to 80% of its peak value within 1 ns (thin film thermal diffusion effect), and then slowly decayed after reaching its peak value within 1 μs (bulk chip thermal diffusion).

[0061] Example 2: Transient Junction Temperature Measurement of GaN HEMT Switch

[0062] The device under test is a 650V / 150mΩ GaN HEMT. It is packaged with an open window according to step S1, and an 80nm aluminum thin film is deposited on the chip surface (the aluminum film on the GaN surface needs to be thinner to reduce the impact of thermal resistance).

[0063] Measurement conditions: switching frequency 100kHz, duty cycle 50%, TDTR delay fixed at 500ps (to capture transient switching temperature), area scan range 0.5mm×0.5mm, step 5μm.

[0064] Measurement results: 500 ps after switching on, the highest temperature on the GaN HEMT surface rose to 12.5 K, with the temperature gradient concentrated near the gate (approximately 2 μm wide), consistent with theoretical predictions of the self-heating effect. This result verifies the ability of the TDTR method to capture nanosecond-level junction temperature transients.

[0065] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for transient measurement of junction temperature in wide-bandgap power semiconductors based on TDTR, characterized in that, Includes the following steps: Step S1: Perform windowing on the packaged power semiconductor device under test, open an optical observation window on the top of the device package and cover it with a transparent window sheet, and deposit a metal thin film on the surface of the device chip as a heat reflection sensing layer. Step S2: The pump light and probe light emitted by the laser are collinearly focused onto the same point on the surface of the metal thin film through the same objective lens, and the relative arrival time between the pump light and the probe light is controlled by a variable optical delay line; Step S3: Apply an electrical pulse to the power semiconductor device under test to simulate the actual working heating condition. Use the rising edge of the electrical pulse as the time trigger reference to control the arrival time of the probe light relative to the rising edge of the electrical pulse, so as to realize the time synchronization of the three signals: electrical signal, pump light and probe light. Step S4: Fix the laser focusing position, successively change the arrival time of the probe light relative to the rising edge of the electrical pulse, and collect the probe light reflection signal corresponding to each delay time point; and / or The delay time of the probe light is fixed, and the laser focusing position is controlled to move and scan point by point on the surface of the device chip to collect the probe light reflection signal corresponding to each pixel. Step S5: Based on the collected probe light reflection signal, perform phase-locked detection using a lock-in amplifier with the pump light modulation frequency as a reference, and separate and extract the in-phase and quadrature components of the thermal reflection signal; based on the in-phase and quadrature components, invert the junction temperature of the power semiconductor device under test to obtain the transient response curve of the single-point temperature changing with time and / or the two-dimensional temperature distribution field on the chip surface.

2. The transient measurement method for wide bandgap power semiconductor junction temperature based on TDTR as described in claim 1, characterized in that, Step S1 includes the following steps: An optical observation window is opened on the top of the device package, covered with an optical anti-reflection window, and the inside of the optical observation window is filled with inert gas to maintain the integrity of the original heat dissipation path and electrical function of the device. A metal thin film is deposited on the surface of the device chip, the thickness of which is configured to reduce additional interfacial thermal resistance while ensuring the intensity of the thermally reflected signal. A grid calibration mark is prepared on the surface of the metal thin film for spatial positioning and position error compensation during laser scanning.

3. The transient measurement method for wide bandgap power semiconductor junction temperature based on TDTR as described in claim 1, characterized in that, In step S2, the pump light generates periodic temperature fluctuations on the surface of the metal film by applying high-frequency heating to the metal film. These fluctuations are then used to extract the thermal reflection signal with the same modulation frequency as the pump light from the reflection signal of the probe light through a lock-in amplifier. The probe light is irradiated onto the surface of the metal thin film, and the intensity of the reflected light reflects the reflectivity of the metal thin film. The reflectivity is used to subsequently invert the surface temperature of the metal thin film. The relative arrival time adjustment range between the pump light and the probe light covers the negative delay to positive delay interval. The negative delay interval is used to acquire a reference signal under no-thermal-response conditions to correct system drift errors.

4. The transient measurement method for wide bandgap power semiconductor junction temperature based on TDTR as described in claim 1, characterized in that, In step S5, the method for retrieving the junction temperature of the power semiconductor device under test includes the following steps: Step S51: Preprocess the in-phase and quadrature components of the lock-in amplifier output; Step S52: The device under test is calibrated by step heating using a programmable heating stage to establish a linear mapping relationship between the preprocessed signal amplitude and temperature change; Step S53: Establish a one-dimensional periodic thermal conduction analytical model of the three-layer structure of metal thin film-chip-substrate, and use the interfacial thermal resistance R1 between the metal thin film and the chip and the interfacial thermal resistance R2 between the chip and the substrate as the parameters to be fitted, and calculate the theoretical thermal reflection signal. Step S54: Use a nonlinear least squares fitting algorithm to fit the measured thermal reflection signal with the theoretical thermal reflection signal. When the fitting residual is less than the preset convergence threshold, output the optimal interface thermal resistance R1 and R2. Step S55: Based on the optimal interface thermal resistance and linear mapping relationship, convert the amplitude of the thermal reflection signal extracted by the lock-in amplifier into junction temperature.

5. The transient measurement method for wide bandgap power semiconductor junction temperature based on TDTR as described in claim 4, characterized in that, Step S51 includes the following steps: Background baseline subtraction: ； in, This refers to the pump-probe timing delay. and These are the original in-phase signal and quadrature signal output from the lock-in amplifier, respectively. and The background baseline signal was acquired without pump light excitation. and These are the effective signal components after background baseline subtraction; Constructing complex thermal reflection signals: ； in, For the measured complex thermal reflection signal, The imaginary unit; Calculate the signal amplitude: ； in, The measured amplitude of the thermal reflection signal. This represents the change in the surface reflectivity of the metal thin film. This indicates that the signal amplitude and the change in reflectivity are linearly positively correlated; Normalization process: ； ； in, For normalized complex signals, This is a reference signal with no thermal diffusion within the -20ps range; To normalize the signal amplitude, and These are operators that take the real part and the imaginary part of a complex number, respectively.

6. The transient measurement method for wide bandgap power semiconductor junction temperature based on TDTR as described in claim 5, characterized in that, Step S52 includes the following steps: The device under test with a metal thin film deposited on its surface is placed on a programmable heating stage. With the device powered off and without self-heating, the temperature is increased in steps based on room temperature. After each temperature increase, the device is kept at a constant temperature until thermal equilibrium is reached, and the normalized signal amplitude at each temperature point is collected. The normalized signal amplitude and the corresponding temperature rise were fitted using least-squares linear fitting: ； in, For the first Normalized signal amplitude at each temperature point The slope of the signal-temperature fitting. For the first The temperature rise relative to room temperature at each temperature point. This is the fitting intercept.

7. The transient measurement method for wide bandgap power semiconductor junction temperature based on TDTR as described in claim 6, characterized in that, In step S53: In the one-dimensional periodic thermal conduction analytical model of the aforementioned three-layer metal thin film-chip-substrate structure, the thermal diffusivity of each layer is: ； Among them, subscript , respectively corresponding to metal thin film layers Chip layer and substrate layer ; For the first Layer thermal diffusivity; For the first Layer thermal conductivity; For the first Layer density; For the first Specific heat capacity under constant pressure; The heat wave number for each layer is: ； in, For the first Layer thermal wave number, The imaginary unit; The pump laser is modulated at an angular frequency; The general solution of the temperature field within the layer is: ； in, For depth The complex amplitude temperature field at that location, The vertical depth coordinates of the device; , These are the undetermined coefficients for the temperature field within the layer; Boundary conditions include: upper surface of metal film The boundary is the input boundary of the periodic alternating laser heat flux; Metal Thin Film-Chip Interface The temperature jump condition is met: ； in, The thickness of the metal thin film. Thermal resistance at the metal thin film-chip interface; This refers to the temperature of the metal thin film layer at the metal thin film-chip interface. This refers to the temperature of the chip layer at the metal thin film-chip interface. The thermal conductivity of the metal thin film layer; This represents the temperature gradient of the metal thin film layer at the metal thin film-chip interface. Chip-substrate interface The temperature jump condition is met: ； in, For chip thickness, Thermal resistance at the chip-substrate interface; This refers to the temperature of the chip layer at the chip-substrate interface. This represents the temperature of the substrate layer at the chip-substrate interface. Thermal conductivity of the chip layer; This represents the temperature gradient of the chip layer at the chip-substrate interface. substrate bottom This is the convective heat transfer boundary: ； in, For substrate thickness, The convective heat transfer coefficient; The thermal conductivity of the substrate layer; This represents the temperature gradient of the substrate layer at the bottom of the substrate. The temperature at the bottom of the substrate; The location at the bottom of the substrate; The reference room temperature; The theoretical complex thermal reflection signal output by the model is: ； in, For theoretical complex thermal reflection signals, determined by the delay time With interfacial thermal resistance Joint decision; , These are theoretically in-phase and quadrature signals, respectively.

8. The transient measurement method for wide bandgap power semiconductor junction temperature based on TDTR as described in claim 7, characterized in that, Step S54 includes the following steps: Construct the fitting residual function: ； in, For the residual loss function, Summation over the entire time domain with full delay; The iteration is considered to have converged when the residual loss function is less than the preset convergence threshold. During the fitting iteration process, the interfacial thermal resistance and The data is limited to a preset range of physically reasonable values. If the convergence result exceeds this range, the data is considered to have failed to fit and is discarded. Interface thermal resistance of the same chip and The parameters are obtained by fitting only once in the central region of the chip, and then reused for subsequent global temperature measurements.

9. The transient measurement method for wide bandgap power semiconductor junction temperature based on TDTR as described in claim 8, characterized in that, Step S55 includes the following steps: Based on the signal amplitude-temperature calibration curve obtained in step S52 and the optimal interface thermal resistance output in step S54, the measured thermal reflection signal amplitude extracted by the lock-in amplifier is converted into the junction temperature of the power semiconductor device under test. In point measurement mode, the normalized signal amplitude at different delay time points is converted into transient temperature rise: ； in, For delay time The corresponding transient temperature rise of the device, For delay time The corresponding normalized signal amplitude, The slope of the signal-temperature fitting. The thermal reflectance temperature coefficient of the metal thin film; The absolute temperature of the device is calculated using the following formula: ； in, For delay time The corresponding absolute temperature of the device, The reference room temperature; In surface scanning mode, the normalized signal amplitude of each pixel is converted into the absolute temperature of the corresponding position to obtain the temperature value of each pixel on the chip surface. The temperature values ​​of all pixels are then spatially stitched together according to the scanning position to reconstruct the two-dimensional temperature distribution field on the chip surface.

10. A transient measurement system for wide bandgap power semiconductor junction temperature based on TDTR, used to implement the measurement method according to any one of claims 1 to 9, characterized in that, include: The sample preparation module is used to perform windowing on the packaged power semiconductor device under test, open an optical observation window on the top of the device package and cover it with a transparent window sheet, and deposit a metal thin film on the surface of the device chip as a heat reflection sensing layer. A femtosecond laser module is used to generate pump light and probe light, and to focus the pump light and probe light collinearly onto the same point on the surface of the metal thin film through the same objective lens. The femtosecond laser module is equipped with a variable optical delay line to control the relative arrival time between the pump light and the probe light. The electro-thermal synchronization trigger module is used to apply an electrical pulse to the power semiconductor device under test, and use the rising edge of the electrical pulse as a time trigger reference to control the arrival time of the probe light relative to the rising edge of the electrical pulse, so as to realize the time synchronization of the three signals: electrical signal, pump light and probe light. The signal acquisition module is used to acquire the probe light reflection signal corresponding to different delay time points and / or acquire the probe light reflection signal corresponding to different pixel points; the signal acquisition module includes a lock-in amplifier, which performs lock-in detection with the modulation frequency of the pump light as a reference, and separates and extracts the in-phase component and quadrature component of the thermal reflection signal. The temperature inversion module is used to invert the junction temperature of the power semiconductor device under test based on the in-phase and quadrature components, and obtain the transient response curve of the temperature at a single point changing with time and / or the two-dimensional temperature distribution field on the chip surface.

Images