Electronic component aging analysis method based on temperature and current coupling

Abstract

This invention relates to the field of electronic component testing technology, specifically disclosing a method for aging analysis of electronic components based on temperature and current coupling. The method includes: simultaneously acquiring temperature fluctuation signals and ripple current signals; constructing a thermal resistance dispersion feature matrix through cross-correlation spectrum analysis; constructing a resistance relaxation feature sequence through current pulse relaxation sampling and multi-exponential decomposition; and fusing these into a thermoelectric coupling aging feature matrix. This matrix is ​​then input into a competitive aging model of intermetallic compound growth front and grain boundary void expansion to invert the propagation rate and expansion rate and calculate the competitive index to identify critical failure precursor regions. Based on the precursor region distribution density and dual-rate evolution trend, a three-level early warning system is generated, and dynamic adjustment commands for thermal management strategies and current stress limits are output. This invention achieves quantitative characterization of thermoelectric coupling stress and accurate identification of microscopic failure mechanisms, and is applicable to aging analysis and lifespan management of components such as power semiconductors and electrolytic capacitors.

Description

Technical Field

[0001] This invention relates to the field of electronic component testing technology, and specifically to an aging analysis method for electronic components based on temperature and current coupling. Background Technology

[0002] Current aging analysis techniques for electronic components mainly rely on offline or online monitoring of single physical parameters. Thermal aging assessment typically uses a thermal resistance meter to measure steady-state thermal resistance changes, while electrical aging assessment typically uses a ohmmeter to periodically measure DC resistance drift. These methods separate temperature stress and current stress into independent monitoring channels, establishing separate thermal aging and electrical aging models. This fails to reflect the coupled damage mechanism of temperature fluctuations and ripple current on the internal microstructure of components under actual service conditions, resulting in incomplete extraction of aging characteristics and inaccurate failure mechanism diagnosis.

[0003] In power semiconductor devices, electrolytic capacitors, and high-density metal interconnect structures, the aging process is typically driven by two microscopic mechanisms: electromigration and thermal oxidation. Electromigration, under current stress, leads to the directional migration of atoms at metal grain boundaries, inducing the nucleation and expansion of grain boundary voids. Thermal oxidation, under temperature stress, leads to the thickening of the oxide layer at the metal-dielectric interface and promotes the continuous advancement of the growth front of intermetallic compounds. Current technologies lack quantitative methods to separate the dynamic competition and synergistic acceleration effects of these two microscopic mechanisms in a thermoelectric coupling field. This makes it impossible to establish a feasible mapping relationship between macroscopic electrical parameter degradation and microstructural evolution, thus hindering the identification of critical failure precursor regions triggered by temperature fluctuations and ripple current. Consequently, aging warnings are delayed, and remaining lifetime predictions are significantly inaccurate.

[0004] Furthermore, existing electronic systems typically employ independent control loops for thermal management and current stress regulation. The thermal management system adjusts the cooling fan speed based on the casing temperature threshold, while the current regulation module sets the power circuit current limit based on load requirements. There is a lack of information exchange and collaborative optimization mechanisms between these two systems based on the actual aging status of the components. When components have entered the thermo-electric accelerated aging stage, a single thermal management or current regulation measure cannot effectively suppress the dual-mechanism coupled damage, making it difficult to proactively slow down the aging process.

[0005] Therefore, there is an urgent need for an aging analysis method for electronic components based on temperature and current coupling to solve the above problems. Summary of the Invention

[0006] The purpose of this invention is to provide an aging analysis method for electronic components based on temperature and current coupling, in order to solve the problems of existing technologies that separate temperature stress and current stress, cannot quantitatively separate the synergistic competitive effect of electromigration and thermal oxidation, two microscopic aging mechanisms, and lead to lag in the identification of critical failure precursor zones and inaccurate aging warnings.

[0007] To solve the above-mentioned technical problems, the present invention specifically provides the following technical solution: An aging analysis method for electronic components based on temperature and current coupling includes the following steps: S1. Synchronously acquire temperature and current signals, perform cross-correlation spectrum analysis, extract thermal impedance amplitude response sequence to construct thermal impedance dispersion feature matrix, inject probe current pulse into relaxation period sampling voltage, extract time constant through multi-exponential relaxation decomposition to construct resistance drift relaxation feature sequence, and form thermoelectric coupling aging feature matrix. S2. Input the thermoelectric coupling aging feature matrix into the aging model of the co-competitive aging of the growth front of intermetallic compounds and the expansion of grain boundary voids, invert the propagation rate and expansion rate and calculate the co-competitive index. When it deviates from the preset equilibrium range, it is marked as the critical failure precursor region. S3. Based on the distribution density and evolution trend of the critical failure precursor zone, the aging status classification assessment results are generated, and dynamic adjustment instructions for thermal management strategies and current stress limits are output.

[0008] As a preferred embodiment of the present invention, S1 specifically includes: S11. During the service operation of the electronic component under test, the casing temperature fluctuation signal and the power circuit ripple current signal are collected simultaneously; S12. Perform cross-correlation spectrum analysis on the shell temperature fluctuation signal and ripple current signal, extract the thermal impedance amplitude response sequence and phase response sequence at each frequency, and construct the interface thermal resistance dispersion feature matrix. S13. Inject probe current pulses into the metal interconnect structure of the electronic component under test under multiple preset test currents, and during the relaxation period after each probe current pulse is unloaded, discretely sample the terminal voltage recovery waveform at both ends of the metal interconnect along the time axis. S14. Perform multi-exponential relaxation decomposition on the voltage recovery waveforms of each terminal, extract the relaxation time constant and relaxation amplitude weight sequence, construct the metal interconnect resistance drift relaxation feature sequence, and then merge it with the interface thermal resistance dispersion feature matrix to form the thermoelectric coupling aging feature matrix.

[0009] As a preferred embodiment of the present invention, S12 specifically includes: S121. Perform Fourier transform on the synchronously acquired shell temperature fluctuation signal and power circuit ripple current signal to obtain their respective amplitude spectrum and phase spectrum in the frequency domain. S122. Calculate the cross-correlation amplitude and cross-correlation phase difference between the shell temperature fluctuation signal and the power circuit ripple current signal at each frequency point, and use the cross-correlation amplitude after normalizing the current amplitude as the thermal impedance amplitude response value at the frequency point, and use the cross-correlation phase difference after delay compensation as the thermal impedance phase response value at the frequency point. S123. Arrange the thermal impedance amplitude response values ​​and thermal impedance phase response values ​​at all frequency points in ascending order of frequency to form thermal impedance amplitude response sequence and thermal impedance phase response sequence respectively. S124. Using frequency as the row index and different operating conditions or different sampling times as the column index, combine the thermal impedance amplitude response sequence and the thermal impedance phase response sequence in complex form to form an interface thermal impedance dispersion characteristic matrix.

[0010] As a preferred embodiment of the present invention, S14 specifically includes: S141. Perform baseline correction and noise reduction preprocessing on the terminal voltage recovery waveform acquired after each probe current pulse is unloaded to obtain the net relaxation voltage waveform; S142. A multi-exponential decomposition algorithm based on regularized inverse Laplace transform is adopted to decompose the net relaxation voltage waveform into a superposition of multiple exponential decay components, each component corresponding to a relaxation time constant and a relaxation amplitude weight. S143. Arrange all relaxation time constants obtained by decomposition under the same preset test current in ascending order of value to form a relaxation time constant distribution sequence. Arrange the relaxation amplitude weights corresponding to each time constant in the same order to form a relaxation amplitude weight distribution sequence. S144. Combine the relaxation time constant distribution sequence and relaxation amplitude weight distribution sequence obtained under all preset test currents in ascending order of test current to construct the metal interconnect resistance drift relaxation characteristic sequence, and then fuse it with the interface thermal resistance dispersion characteristic matrix to form the thermoelectric coupling aging characteristic matrix.

[0011] As a preferred embodiment of the present invention, S2 specifically includes: S21. The thermal impedance amplitude at each frequency point in the interface thermal resistance dispersion feature matrix and the relaxation time constant value in the metal interconnect resistance drift relaxation feature sequence are synchronously input into the aging model of the co-competitive aging of the intermetallic compound growth front and grain boundary void expansion. S22. In the cooperative aging model, the growth front advancement rate of intermetallic compounds under temperature stress is inverted based on the thermal impedance amplitude, and the grain boundary void propagation rate under current stress is inverted based on the relaxation time constant. S23. Calculate the synergistic competition index between the propulsion rate and the expansion rate. When the synergistic competition index deviates from the preset synergistic competition equilibrium range, mark the current stress range as the critical failure precursor zone of thermoelectric synergistic accelerated aging.

[0012] As a preferred embodiment of the present invention, S22 specifically includes: S221. Extract the attenuation inflection frequency of thermal resistance amplitude as a function of frequency from the interface thermal resistance dispersion feature matrix, and combine it with the thermal diffusivity of the electronic component under test to calculate the actual growth thickness of the intermetallic compound layer along the heat flow direction through a one-dimensional thermal conduction model. S222. The increment of the actual growth thickness within a unit thermal cycle is taken as the advance rate of the growth front of the intermetallic compound under temperature stress. S223. Based on the proportional relationship between the constant values ​​of each relaxation time in the metal interconnect resistance drift relaxation characteristic sequence and the resistivity of the metal interconnect, and combined with the interconnect geometry and initial conductive cross-sectional area of ​​the electronic component under test, the reduction in effective conductive cross-sectional area caused by the expansion of grain boundary voids can be deduced. S224. The rate of change of the reduction in the effective conductive cross-sectional area per unit energizing time is taken as the grain boundary void propagation rate under current stress.

[0013] As a preferred embodiment of the present invention, S23 specifically includes: S231. Substitute the growth front advance rate and grain boundary void propagation rate of the intermetallic compound obtained in step S22 into the formula for calculating the cooperative competition index, where the cooperative competition index is equal to the ratio of the advance rate to the propagation rate. S232. Compare the calculated cooperative competition index with the preset cooperative competition equilibrium interval. When the cooperative competition index is within the cooperative competition equilibrium interval, it is determined that the current aging is in a cooperative competition equilibrium state and no critical failure precursor zone is marked. S233. When the cooperative competition index is less than the lower limit of the cooperative competition equilibrium interval, it is determined that the current aging is driven by current stress, and the stress interval is marked as the current-dominated critical failure precursor zone; when the cooperative competition index is greater than the upper limit of the cooperative competition equilibrium interval, it is determined that the current aging is driven by temperature stress, and the stress interval is marked as the temperature-dominated critical failure precursor zone.

[0014] As a preferred embodiment of the present invention, S3 specifically includes: S31. Extract the coordinate position of the marked critical failure precursor region on the temperature-current two-dimensional stress plane, and calculate the number of marked sample points per unit area as the distribution density. S32. The growth front propagation rate of intermetallic compounds and the expansion rate of grain boundary voids in the critical failure precursor region are fitted along the time axis to obtain the slope of the rate change and acceleration parameters. S33. Based on the distribution density, rate of change slope, and acceleration parameters, the current aging status of the electronic components under test is divided into three levels: mild warning, moderate warning, or severe warning. S34. Based on the classified aging status level, generate the corresponding cooling fan speed control signal or power circuit ripple current limiting command, and output it to the external thermal management system or current control module.

[0015] As a preferred embodiment of the present invention, S31 specifically includes: S311. Discretize the temperature-current two-dimensional stress plane into multiple equal-area grid elements according to the preset temperature step and current step, with each grid element corresponding to a temperature range and a current range; S312. Traverse all sample points marked as critical failure precursor regions, count the number of sample points falling into each grid cell, divide the number of sample points in each grid cell by the area of ​​the grid cell to obtain the local distribution density value of the grid cell. S313. Normalize the local distribution density values ​​of all grid cells, and generate a distribution density heat map of the critical failure precursor zone with temperature coordinates as the horizontal axis, current coordinates as the vertical axis, and normalized distribution density values ​​as the vertical axis. Output the temperature range and current range corresponding to the peak area in the heat map as the high-risk stress zone.

[0016] As a preferred embodiment of the present invention, S33 specifically includes: S331. Normalize the distribution density value calculated in step S31 and the rate change slope and acceleration parameters fitted in step S32 respectively to obtain dimensionless density index, slope index and acceleration index. S332. The density index, slope index and acceleration index are weighted and summed according to preset weights to obtain the comprehensive score of aging status. The comprehensive score of aging status is compared with a first threshold and a second threshold, wherein the first threshold is less than the second threshold. S333. When the comprehensive score of the aging status is less than the first threshold, it is classified as a mild warning level; when the comprehensive score of the aging status is greater than or equal to the first threshold and less than the second threshold, it is classified as a moderate warning level; when the comprehensive score of the aging status is greater than or equal to the second threshold, it is classified as a severe warning level.

[0017] Compared with the prior art, the present invention has the following advantages: 1. By cross-correlation spectrum analysis and multi-exponential relaxation decomposition, the interface thermal resistance dispersion characteristic matrix and the resistance drift relaxation characteristic sequence are constructed respectively, and then fused to form a thermo-electric coupling aging characteristic matrix, realizing a unified characterization of thermal domain dispersion and electric domain relaxation characteristics, and providing a high-dimensional data foundation for the inversion of microscopic aging mechanism.

[0018] 2. Establish a synergistic competitive aging model of intermetallic compound growth front and grain boundary void expansion. Invert the propagation rate and expansion rate by thermal impedance amplitude and relaxation time constant respectively, calculate the synergistic competition index to identify the critical failure precursor region, and realize the quantitative separation of the two microscopic aging mechanisms and the determination of the dominant factors.

[0019] 3. Based on the two-dimensional stress plane distribution density and dual-rate evolution trend in the critical failure precursor zone, a comprehensive score of aging status is generated through multi-parameter normalization and weighted fusion, achieving three-level accurate early warning, and outputting differentiated thermo-electric synergistic dynamic control commands to actively reduce coupled stress and delay aging. Attached Figure Description

[0020] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating the method described in Embodiment 1 of the present invention.

[0022] Figure 2 This is a framework diagram of the system described in Embodiment 2 of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] The concepts involved in this application will first be described with reference to the accompanying drawings. It should be noted that the following descriptions of various concepts are only for the purpose of making the content of this application easier to understand and do not constitute a limitation on the scope of protection of this application; furthermore, the embodiments and features in the embodiments of this application can be combined with each other unless otherwise specified. This application will now be described in detail with reference to the accompanying drawings and embodiments. Example 1

[0025] like Figure 1 As shown, this invention provides an aging analysis method for electronic components based on temperature and current coupling, comprising the following steps: S1. Synchronously acquire temperature and current signals, perform cross-correlation spectrum analysis, extract thermal impedance amplitude response sequence to construct thermal impedance dispersion feature matrix, inject probe current pulse into relaxation period sampling voltage, use multi-exponential relaxation decomposition to extract time constant to construct resistance drift relaxation feature sequence, and form thermoelectric coupling aging feature matrix; specifically including: S11. Synchronous acquisition of temperature and current signals under service conditions, specifically: During the service operation of the electronic component under test, a temperature sensing unit is mounted on the surface of the electronic component's package housing. The temperature sensing unit adopts a thin-film thermocouple or a surface-mount thermistor. The mounting position covers the geometric center of the housing, the root of the pin soldering, and the interface of the heat dissipation substrate to ensure comprehensive capture of housing temperature fluctuation signals caused by load changes, start-stop cycles, and external thermal environment disturbances. The output of the temperature sensing unit is connected to the first analog-to-digital conversion channel through a shielded differential signal cable.

[0026] A high-frequency current sampling unit is connected in series in the power circuit. This high-frequency current sampling unit adopts a non-invasive Hall effect current sensor or a precision shunt resistor. The series connection position is located at the positive bus and negative bus of the main power circuit of the electronic component to ensure complete acquisition of the power circuit ripple current signal. The output of the high-frequency current sampling unit is connected to the second analog-to-digital conversion channel through a shielded coaxial cable.

[0027] The temperature sensing unit and the high-frequency current sampling unit share the same high-precision clock reference source. This clock reference source provides a unified sampling clock signal to the first and second analog-to-digital conversion channels, ensuring that the housing temperature fluctuation signal and the power circuit ripple current signal are strictly aligned on the time axis. After the two signals are converted from analog to digital signals through their respective analog-to-digital conversion channels, they are synchronously transmitted to the signal processing unit via a parallel data bus. The signal processing unit timestamps and buffers the two digital signals to form a synchronously acquired data stream.

[0028] S12. Cross-correlation spectrum analysis and construction of the interface thermal resistance dispersion characteristic matrix are as follows: S121. Frequency domain transformation and spectrum extraction of the casing temperature and power circuit signals, specifically: The frequency domain transformation module includes a signal conditioning front-end, an anti-aliasing filter, an analog-to-digital converter kernel, a digital window function processor, a fast Fourier transform operator, an amplitude spectrum extraction unit, and a phase spectrum extraction unit. The signal conditioning front end receives the synchronously acquired shell temperature fluctuation signal and power circuit ripple current signal, and performs level adaptation and impedance matching; the anti-aliasing filter is a low-pass filter, and the cutoff frequency is set to 1 / 2 of the sampling frequency according to the Nyquist sampling theorem to filter out frequency components higher than this cutoff frequency. The analog-to-digital conversion kernel converts analog signals into digital signal sequences at a fixed sampling frequency, ranging from 10 kHz to 1 MHz, with a classic value of 100 kHz; the digital window function processor applies a Hanning window to the digital signal sequence to reduce spectral leakage; The Fast Fourier Transform (FFT) unit performs a complex-domain Discrete Fourier Transform on the windowed digital signal sequence and outputs a complex spectrum sequence; the amplitude spectrum extraction unit calculates the magnitude of the complex spectrum at each frequency point to obtain the amplitude spectrum; and the phase spectrum extraction unit calculates the argument of the complex spectrum at each frequency point to obtain the phase spectrum.

[0029] S122. Calculate the cross-correlation amplitude and cross-correlation phase difference between the shell temperature fluctuation signal and the power circuit ripple current signal at each frequency point, and use the cross-correlation amplitude after normalizing the current amplitude as the thermal impedance amplitude response value at the frequency point, and use the cross-correlation phase difference after delay compensation as the thermal impedance phase response value at the frequency point. Frequency cross-correlation calculation and thermal impedance response calculation, specifically: At each frequency point, the complex value in the amplitude spectrum of the shell temperature fluctuation signal is multiplied by the complex value at the corresponding frequency point in the amplitude spectrum of the power circuit ripple current signal. The magnitude of the product is the cross-correlation amplitude. The cross-correlation amplitude is divided by the amplitude of the power circuit ripple current signal at that frequency point to complete the current amplitude normalization. The quotient is used as the thermal impedance amplitude response value at that frequency point.

[0030] Simultaneously, the phase angle of the phase spectrum of the shell temperature fluctuation signal at this frequency point and the phase angle of the phase spectrum of the power circuit ripple current signal at this frequency point are extracted. The difference between the two is calculated to obtain the cross-correlation phase difference. Then, the phase shift caused by the inherent delay of the signal transmission cable and the sampling circuit is subtracted to complete the delay compensation. The resulting difference is used as the thermal impedance phase response value at this frequency point, where the phase shift ranges from 5° to 20°.

[0031] S123. Frequency ordering and sequence generation of thermal impedance response sequences, specifically: The thermal impedance amplitude response values ​​calculated at all frequency points are arranged in a one-dimensional array in ascending order of frequency from low to high, with a fixed frequency interval between adjacent elements, forming a thermal impedance amplitude response sequence. The fixed frequency interval is equal to the sampling frequency divided by the number of spectrum analysis points, which ranges from 1024 to 8192. When the sampling frequency is taken as the classic value of 100 kHz and the number of spectrum analysis points is taken as 4096, the classic value of the frequency interval is 24.4 Hz.

[0032] Using the same frequency increment order, the thermal impedance phase response values ​​calculated at all frequency points are arranged into a one-dimensional array, with adjacent elements maintaining the same frequency interval as the thermal impedance amplitude response sequence, thus forming a thermal impedance phase response sequence.

[0033] The two sequences are of equal length and the same index position corresponds to the same frequency point, ensuring that the thermal impedance amplitude information and the thermal impedance phase information are strictly aligned in the frequency dimension.

[0034] S124. Complex combination and construction of the interface thermal resistance dispersion characteristic matrix, specifically: Construct a two-dimensional matrix. The row dimension of the matrix is ​​indexed by frequency, with each row corresponding to a specific frequency point, and the rows are arranged in ascending order of frequency. The column dimension of the matrix is ​​indexed by different working conditions or different sampling times, with each column corresponding to a set of independent measurement conditions or an independent sampling event.

[0035] The thermal impedance amplitude response sequence and the thermal impedance phase response sequence are combined in complex form to form matrix elements. The real part of the complex number is taken from the amplitude value at the corresponding frequency point and the corresponding column in the thermal impedance amplitude response sequence, and the imaginary part of the complex number is taken from the phase value at the corresponding frequency point and the corresponding column in the thermal impedance phase response sequence. This two-dimensional matrix is ​​the interface thermal resistance dispersion characteristic matrix, which simultaneously carries the amplitude dispersion characteristics and phase dispersion characteristics of the interface thermal resistance as a function of frequency.

[0036] S13. Acquisition of current pulse injection and relaxation period terminal voltage recovery waveforms, specifically: A probe current pulse is injected into the metal interconnect structure of the electronic component under test under multiple preset test currents, and the terminal voltage waveforms at both ends of the metal interconnect are discretely sampled along the time axis during the relaxation period after each probe current pulse is unloaded. During the intermittent periods of normal operation of electronic components, a probe current pulse is injected into the metal interconnect structure of the electronic component under test through a programmable precision current source. The preset test current levels are set to 10%, 30%, 50%, 70%, and 90% of the rated operating current of the metal interconnect structure, in five incremental levels. Under each preset test current level, a rectangular probe current pulse with a constant amplitude and a pulse width of 10ms is injected independently, with both the rise and fall times of the pulse being less than 10us.

[0037] After the probe current pulse is unloaded, the metal interconnect structure enters an electrical relaxation state, which lasts from 50ms to 200ms. During this period, a high-precision voltage sampling circuit discretely samples the terminal voltages across the metal interconnect along the time axis at fixed sampling intervals. The fixed sampling interval is set to 10µs to 100µs, with a classic value of 50µs. The sampling process begins at the instant the pulse is unloaded and continues until the terminal voltage recovers to its steady-state reference value, completely recording the terminal voltage recovery waveform during this period. The input impedance of the high-precision voltage sampling circuit is no less than ten megohms to reduce the impact of sampling load effects on the relaxation process.

[0038] S14. Fusion of multi-exponential relaxation decomposition and thermoelectric coupling aging characteristic matrix, specifically: S141. Baseline correction and denoising preprocessing of the terminal voltage recovery waveform, specifically: For each probe current pulse unloaded, the terminal voltage recovery waveform acquired is first extracted as the baseline reference by extracting the stable voltage value at both ends of the metal interconnect before pulse unload. The voltage value at each sampling point in the terminal voltage recovery waveform is subtracted from the baseline reference value to obtain the net relaxation voltage waveform that only reflects the relaxation process.

[0039] The net relaxation voltage waveform is then input into a digital low-pass filter. This digital low-pass filter adopts a finite impulse response structure with an order set from 32 to 128. The cutoff frequency is set to 1 / 10 of the fundamental frequency of the ripple current. The passband ripple is less than 0.5dB, and the stopband attenuation is greater than 40dB. This effectively filters out high-frequency sampling noise and power frequency interference signals introduced by the power loop switching action, and outputs a smooth net relaxation voltage waveform with a baseline of zero. This ensures that the signal-to-noise ratio of the input signal for subsequent multi-exponential decomposition meets the algorithm convergence requirements.

[0040] S142. Multi-exponential relaxation decomposition based on regularized inverse Laplace transform, specifically: The smoothed net relaxation voltage waveform is input into the multi-exponential decomposition algorithm module, which consists of a numerical Laplace transform unit, a discrete decay kernel matrix construction unit, a regularized singular value decomposition solution unit, and an iterative optimization unit. The numerical Laplace transform unit performs discrete numerical integration on the net relaxation voltage waveform, converting the time-domain sampling sequence into a complex frequency domain image function sequence, establishing a mapping relationship between the time-domain relaxation process and the complex frequency domain attenuation characteristics, and providing a frequency domain data basis for subsequent inverse transform solutions.

[0041] The discrete attenuation kernel matrix construction unit generates a two-dimensional attenuation kernel matrix based on the sampling time sequence of the net relaxation voltage waveform and a preset logarithmically spaced candidate relaxation time constant sequence. The rows of this matrix correspond to each sampling time along the time axis, and the columns correspond to each candidate relaxation time constant. The matrix elements are negative exponential function values ​​of the quotient of the sampling time and the relaxation time constant. This matrix represents the mathematical structure of the linear superposition of exponential attenuation components, transforming the solution of the relaxation amplitude weights into a problem of solving a system of linear equations. In the preset logarithmically spaced candidate relaxation time constant sequence, the logarithm of the ratio of adjacent candidate relaxation time constants is base 10. The logarithmic interval ranges from 0.05 to 0.3, with a classic value of 0.1, corresponding to a ratio of approximately 1.26 between adjacent candidate relaxation time constants. Ten candidate points are distributed every ten octaves.

[0042] The regularized singular value decomposition (SVD) solving unit performs SVD on the discrete attenuation kernel matrix to obtain the left singular vector matrix, the singular value diagonal matrix, and the right singular vector matrix. A truncation threshold is set based on the noise power spectral density of the net relaxation voltage waveform, and singular value components below the truncation threshold are set to zero to suppress noise amplification. A regularization parameter is introduced to construct a regularization filter factor to correct the truncated singular value matrix. The least squares method is used to solve for the relaxation amplitude weights corresponding to each candidate relaxation time constant, ensuring the stability and physical rationality of the solution. The truncation threshold is relatively set based on the maximum singular value of the discrete attenuation kernel matrix, with a value ranging from 0.5% to 5% of the maximum singular value. The classical value is 1% of the maximum singular value. This threshold is dynamically corrected based on the noise power spectral density of the net relaxation voltage waveform; the upper limit is used when the noise power spectral density is high, and the lower limit is used when it is low, ensuring that the retained singular value components after truncation can effectively reconstruct the main signal while suppressing noise artifacts.

[0043] The iterative optimization unit uses the sum of squared residuals between the reconstructed waveform and the original net relaxation voltage waveform as the convergence criterion, dynamically adjusts the regularization parameters, and repeatedly calls the regularized singular value decomposition solution unit until the sum of squared residuals reaches the global minimum and the relaxation amplitude weight distribution satisfies the physical non-negativity constraint. Finally, it outputs the set of relaxation time constants and the corresponding set of relaxation amplitude weights after regularization.

[0044] S143. The generation of the sorting of the relaxation time constant distribution sequence and the relaxation amplitude weight distribution sequence, specifically: For all relaxation time constants obtained by decomposition under the same preset test current, they are arranged in a one-dimensional array in ascending order of value, with adjacent elements maintaining logarithmic intervals, to form a relaxation time constant distribution sequence.

[0045] The relaxation amplitude weights corresponding to each relaxation time constant are arranged in a one-dimensional array according to the same sorting order, ensuring that the index position of each element in the relaxation amplitude weight distribution sequence is strictly consistent with the index position of the corresponding element in the relaxation time constant distribution sequence, thus forming a relaxation amplitude weight distribution sequence. This sorting process ensures that the pairing relationship between relaxation time constants and relaxation amplitude weights in the physical dimension is not disrupted, so that the smaller time constants reflecting the short relaxation process of the interface state and the larger time constants reflecting the long relaxation process of the bulk material present an ordered distribution structure in the sequence.

[0046] S144. Construction of the characteristic sequence of metal interconnect resistance drift relaxation and fusion with the thermoelectric coupling characteristic matrix, specifically: All relaxation time constant distribution sequences and relaxation amplitude weight distribution sequences obtained under preset test currents are vertically stacked and combined in order of increasing test current from 10% to 90% to construct a metal interconnect resistance drift relaxation feature sequence. The row dimension of this sequence corresponds to different preset test current levels, and the column dimension corresponds to the pairing information of relaxation time constant and relaxation amplitude weight.

[0047] The metal interconnect resistance drift relaxation feature sequence is used as an additional feature dimension and is concatenated with the interface thermal resistance dispersion feature matrix. The concatenation method is to add each element of the resistance relaxation feature sequence to the right side of the interface thermal resistance dispersion feature matrix in the form of an extended column vector, so that the interface thermal resistance dispersion feature matrix is ​​expanded into a thermoelectric coupling aging feature matrix that simultaneously carries thermal resistance dispersion information and resistance relaxation information.

[0048] S2. Input the thermoelectric coupling aging feature matrix into the aging model of the cooperative competition between the growth front of intermetallic compounds and the expansion of grain boundary voids, invert the propagation rate and expansion rate, and calculate the cooperative competition index. Deviating from the preset equilibrium range is marked as a critical failure precursor region; specifically including: S21. The thermoelectric coupling aging feature matrix is ​​synchronously input into the cooperative competitive aging model, specifically as follows: The aging model for the synergistic competition between the growth front of intermetallic compounds and the expansion of grain boundary voids adopts a dual-channel parallel analytical architecture, including a temperature stress analytical channel and a current stress analytical channel. The temperature stress analytical channel has a built-in one-dimensional heat conduction inversion unit and an inflection point frequency detection unit, while the current stress analytical channel has a built-in electrochemical relaxation inversion unit and a cross-sectional area reduction calculation unit.

[0049] In practice, the thermal impedance amplitude response values ​​at each frequency point are extracted row by row from the interface thermal resistance dispersion feature matrix in ascending order of frequency, forming a one-dimensional thermal impedance amplitude input vector. Similarly, the relaxation time constant values ​​are extracted row by row from the metal interconnect resistance drift relaxation feature sequence in ascending order of test current, forming a one-dimensional relaxation time constant input vector. The thermal impedance amplitude input vector is then imported into the temperature stress analysis channel, and the relaxation time constant input vector is imported into the current stress analysis channel.

[0050] The two channels share the same high-precision clock reference, and perform timestamp alignment and dimension normalization on the two input vectors to ensure that the analytical results of temperature stress and current stress are calculated collaboratively under the same operating condition reference and time coordinate system, and output parallel micro aging rate data for subsequent collaborative competition index calculation.

[0051] S22. Inversion calculation of micro-aging rate under the dominance of temperature stress and current stress, specifically as follows: S221. Inflection point frequency detection and one-dimensional heat conduction inversion thickness calculation, specifically: The inflection point frequency detection unit receives the thermal impedance amplitude input vector from the temperature stress analysis channel. This vector is arranged in ascending order of frequency. Starting from the first element of the vector, the unit performs a first-order forward differential operation. Specifically, it subtracts the thermal impedance amplitude response value of the previous frequency point from the thermal impedance amplitude response value of the next frequency point. After traversing all frequency points, a differential sequence of the same length as the thermal impedance amplitude input vector is generated, with the last element of the sequence filled with zero. The unit scans the differential sequence point by point, compares the absolute value change trend of adjacent differential values, and records the absolute value of the differential at each point and the absolute values ​​of the differentials of the two points before and after it. When it identifies three or more consecutive differential absolute values ​​showing a monotonically decreasing trend, followed by three or more consecutive differential absolute values ​​showing a monotonically increasing trend, the frequency coordinates corresponding to the trend reversal point are extracted as the attenuation inflection point frequency. If multiple candidate inflection points exist, the point with the smallest absolute value of the differential and the lowest frequency is selected as the unique attenuation inflection point frequency.

[0052] The one-dimensional thermal conductivity inversion unit calls upon the thermal property database of the packaging material of the electronic component under test to read the thermal diffusivity of the intermetallic compound layer and the substrate material. This thermal diffusivity is determined by the ratio of material density, specific heat capacity, and thermal conductivity, and its value ranges from 1 × 10⁻⁶. －6 m² / s to 2×10 －4 For copper-based interconnect materials, the classic value is 1.1 × 10 m² / s. －4 m² / s; for aluminum-based metallization layers, the classic value is 9.7 × 10⁻⁶ m² / s. －5 m² / s; for tin-based solders and typical intermetallic compounds such as Cu3Sn or Cu6Sn5, the classic value is 3 × 10⁻⁶ m² / s. －5 m² / s to 4×10 －5m² / s; for silicon semiconductor substrates, the classic value is 8.8 × 10⁻⁶ m² / s. －5 m² / s. This unit establishes a one-dimensional unsteady-state heat conduction differential equation, defining the shell surface as the isothermal heat flow input boundary and the interface between the intermetallic compound layer and the substrate as the thermal resistance abrupt change boundary, with discontinuous thermal conductivity on both sides of the interface. The frequency domain relationship between the characteristic heat penetration depth and the thermal diffusivity is: the heat penetration depth equals the square root of the quotient obtained by dividing the thermal diffusivity by the product of pi and the attenuation inflection point frequency. Substituting the attenuation inflection point frequency into this relationship to solve for the heat penetration depth, since the heat penetration depth characterizes the effective depth limit of heat flow fluctuations penetrating the material interface in the frequency domain, and this depth limit has an equivalent correspondence with the actual growth thickness of the intermetallic compound layer along the heat flow direction, the heat penetration depth value is directly assigned to the actual growth thickness of the intermetallic compound layer for subsequent propulsion rate calculations.

[0053] S222. Calculation of propulsion rate per unit thermal cycle, specifically: The actual growth thickness at the start and end of the current thermal cycle is extracted, and the difference between the two is calculated to obtain the thickness increment per unit thermal cycle. This thickness increment is divided by the duration of the thermal cycle, and the quotient is defined as the advance rate of the intermetallic compound growth front under temperature stress. This advance rate characterizes the growth displacement of the intermetallic compound layer along the heat flow direction per unit time, reflecting the kinetic driving force of temperature fluctuation cycles on the formation reaction of intermetallic compounds at the interface. The dimension of the advance rate is length divided by time, and its value is directly related to the amplitude of thermal cycle temperature fluctuations, the initial thermal resistance state of the interface, and the thermal diffusivity of the intermetallic compound layer. It is the core output parameter of the temperature stress analytical channel in the cooperative competition aging model and is used for subsequent cooperative competition index calculation.

[0054] S223. Inversion of relaxation time constant and calculation of effective conductive cross-sectional area reduction, specifically: The electrochemical relaxation inversion unit receives the relaxation time constant input vector from the current stress analysis channel, arranged in ascending order of test current. Based on electrochemical relaxation theory, the unit establishes a linear proportional relationship model between the relaxation time constant and the resistivity of the metal interconnect. This model assumes a linear response relationship between the grain boundary charge relaxation process and resistivity change. The mathematical form of this model is: the increment of the metal interconnect resistivity equals the product of a preset resistivity conversion coefficient and the relaxation time constant value. The preset resistivity conversion coefficient ranges from 1 × 10⁻⁶ per second. －5 1×10 ohmmeters per second －3 Ohmmeter, classic value is 5 × 10⁻⁶ m / s. －5The ohmmeter coefficient is predetermined based on a calibration experiment of the coupling between the grain boundary relaxation activation energy and the electromigration activation energy of the metal interconnect material, and is written into the parameter register during model initialization. Each element in the relaxation time constant input vector is multiplied by this resistivity conversion coefficient to obtain the resistivity increment value of the metal interconnect corresponding to each relaxation time constant, forming a resistivity increment sequence.

[0055] The cross-sectional area reduction calculation unit obtains the initial conductive cross-sectional area, interconnect length, and initial resistivity reference value of the metal interconnect structure. The initial conductive cross-sectional area is determined by the product of the interconnect layout design width and thickness, the initial resistivity reference value is provided by the material factory inspection report, and the interconnect length is determined by the layout design rules. The equivalent resistance increment is obtained by dividing the product of the resistivity increment and the interconnect length by the initial resistivity reference value. Based on the negative correlation between the resistance increment and the conductive cross-sectional area, the effective conductive cross-sectional area after grain boundary void expansion is deduced from the equivalent resistance increment: the effective conductive cross-sectional area equals the initial conductive cross-sectional area divided by the ratio of the equivalent resistance increment to the initial resistance value plus one. The difference between the initial conductive cross-sectional area and the effective conductive cross-sectional area is the reduction in effective conductive cross-sectional area. This reduction characterizes the degree to which grain boundary void expansion erodes the conductivity of the metal interconnect and is used in subsequent expansion rate calculations.

[0056] S224. Calculation of grain boundary void propagation rate per unit energizing time, specifically: The effective conductive cross-sectional area reduction at the start and end of the current energizing period is extracted, and the difference between the two is calculated to obtain the change in reduction per unit energizing time. This change in reduction is divided by the energizing time, and the quotient is defined as the grain boundary void propagation rate under current stress. This propagation rate characterizes the rate of reduction of the effective conductive cross-sectional area per unit time caused by grain boundary void propagation, reflecting the erosion intensity of current stress on the electrical integrity of the metal interconnect structure. The dimension of the propagation rate is area divided by time, and its value is directly related to the current density, the grain boundary energy of the interconnect material, and the electromigration activation energy. It is the core output parameter of the current stress analytical channel in the cooperative aging model, and together with the propagation rate, it constitutes the cooperative competition index.

[0057] S23. Calculation of the Cooperative Competition Index and Determination of the Precursor Zone of Critical Failure, specifically: S231. Calculation of Cooperative Competition Index and Dual-Channel Rate Ratio, specifically: The intermetallic compound growth front advancement rate is read from the output register of the temperature stress analysis channel, and the grain boundary void propagation rate is read from the output register of the current stress analysis channel. A floating-point division operation is performed with the advancement rate as the numerator and the propagation rate as the denominator; the resulting ratio is the cooperative competition index. This index characterizes the relative competition intensity between the intermetallic compound growth front advancement displacement and the reduction in effective conductive cross-sectional area caused by grain boundary void propagation per unit time. When the index value approaches 1, it indicates that the erosion intensity of the two aging mechanisms is equal; when the index value deviates from 1, it indicates that one aging mechanism is dominant. The calculation of the cooperative competition index is automatically triggered after each dual-channel analysis operation, and the calculation result is written to the cooperative competition determination register for subsequent balance interval comparison logic.

[0058] S232. Compare the calculated cooperative competition index with the preset cooperative competition equilibrium interval. When the cooperative competition index is within the cooperative competition equilibrium interval, it is determined that the current aging is in a cooperative competition equilibrium state and no critical failure precursor zone is marked. Specifically, the comparison of equilibrium intervals in collaborative competition and the determination of equilibrium states are as follows: The synergistic competition index is numerically compared with a preset synergistic competition equilibrium range. The lower limit of this equilibrium range is 0.2 to 1.0, with a classic value of 0.5; the upper limit is 1.5 to 5.0, with a classic value of 2.0. This equilibrium range is calibrated based on the statistical distribution of different material systems and interconnect structures in accelerated aging tests, reflecting the engineering tolerance boundary of a specific material system and interconnect structure for the dual-mechanism synergistic effect. For copper-based interconnect power semiconductor devices, the lower limit can be 0.3 to 0.6, and the upper limit can be 1.5 to 3.0; for aluminum-based metallized integrated circuits, the lower limit can be 0.2 to 0.5, and the upper limit can be 2.0 to 4.0; for tin-based solder interconnect structures, the lower limit can be 0.4 to 0.8, and the upper limit can be 1.5 to 2.5. The classic values ​​of 0.5 and 2.0 are applicable to most copper-aluminum composite interconnect systems, reflecting the typical competitive equilibrium state of intermetallic compound growth and grain boundary void expansion within the engineering tolerance boundary.

[0059] When the cooperative competition index is ≥0.5 and ≤2.0, the current aging is determined to be in a state of cooperative competition equilibrium. At this time, the growth rate of intermetallic compounds and the expansion rate of grain boundary voids are mutually constrained within the allowable range of engineering. The system does not mark the critical failure precursor region and maintains the current thermal management strategy and current stress limit unchanged.

[0060] S233. Classification and dominance type identification of critical failure precursor zones, specifically: When the cooperative competition index is less than 0.5, it is determined that the grain boundary void propagation rate is significantly dominant over the intermetallic compound growth front advancement rate, and the current aging is driven by current stress. This stress range is marked as the current-dominated critical failure precursor region.

[0061] When the cooperative competition index is greater than 2.0, it is determined that the growth front advancement rate of intermetallic compounds is significantly dominant over the grain boundary void expansion rate, and the current aging is driven by temperature stress. This stress range is marked as the temperature-dominated critical failure precursor region.

[0062] The marking results are written into the critical failure precursor zone status register, along with a current-dominated or temperature-dominated type identifier code, for subsequent aging condition classification assessment and dynamic adjustment instruction generation.

[0063] S3. Based on the distribution density and evolution trends of the critical failure precursor zone, as well as its propagation and expansion rates, generate aging condition classification assessment results and output dynamic adjustment instructions for thermal management strategies and current stress limits; specifically including: S31. Calculation of the two-dimensional stress plane distribution density in the critical failure precursor region, specifically: S311. Discretization of the two-dimensional stress plane using a constant-step mesh, specifically: The temperature and current coordinates of all marked sample points are read from the critical failure precursor zone status register. The minimum and maximum values ​​of the temperature and current coordinates are determined to define the boundary range of the temperature-current two-dimensional stress plane. The horizontal axis of this two-dimensional stress plane is set as the temperature coordinate, and the vertical axis is set as the current coordinate. The preset temperature step size is 5℃, and the preset current step size is 2% of the rated operating current. The temperature coordinate axis is divided into equal intervals of 5℃, and the current coordinate axis is divided into equal intervals of 2% of the rated operating current. The two sets of equal interval dividing lines intersect orthogonally, dividing the two-dimensional stress plane into multiple rectangular grid cells of equal area.

[0064] Each grid cell corresponds to a combination of a temperature range and a current range. The lower limit of the temperature range differs from the upper limit by 5°C, and the lower limit of the current range differs from the upper limit by 2% of the rated operating current. Each grid cell is assigned a unique two-dimensional index number for subsequent sample point assignment determination.

[0065] S312. Grid cell sample point traversal statistics and local distribution density calculation, specifically: Iterate through all sample points marked as critical failure precursor regions, and read the temperature and current coordinate values ​​for each sample point. Compare the temperature coordinate values ​​with the lower and upper limits of the temperature range of each grid cell, and compare the current coordinate values ​​with the lower and upper limits of the current range of each grid cell. When the temperature coordinate value of a sample point is greater than or equal to the lower limit of the temperature range of a grid cell and less than the upper limit of the temperature range of that grid cell, and the current coordinate value is greater than or equal to the lower limit of the current range of that grid cell and less than the upper limit of the current range of that grid cell, it is determined that the sample point falls into this grid cell, and the internal sample point counter of that grid cell is incremented by one.

[0066] After traversing all sample points, the value of the internal sample point counter of each grid cell is read. This value is divided by the area of ​​the grid cell, and the quotient is used as the local distribution density value of the grid cell. The dimension of the local distribution density value is the number of sample points per square unit stress plane.

[0067] S313. Distribution density normalization and high-risk stress zone extraction, specifically: Normalize the local distribution density values ​​of all grid cells, retrieve the minimum and maximum values ​​of all local distribution density values, subtract the minimum value from the local distribution density value of each grid cell and divide by the difference between the maximum and minimum values, so that all local distribution density values ​​are mapped to the dimensionless interval of zero to one, and the normalized distribution density value is obtained.

[0068] Using temperature as the horizontal axis, current as the vertical axis, and normalized density as the vertical axis, a bilinear interpolation algorithm is used to smoothly transition density values ​​between adjacent grid cells, generating a density heatmap of the critical failure precursor zone. The normalized density values ​​of all grid cells in this heatmap are traversed, and peak regions with normalized density values ​​exceeding 0.8 are identified and extracted. The temperature and current ranges of the grid cells covered by these peak regions are read, and this temperature and current ranges are combined as a high-risk stress zone and output to the aging condition assessment module.

[0069] S32. Dual-rate time series trend fitting and evolution parameter extraction, specifically: The propagation rates of the intermetallic compound growth front in the critical failure precursor region are read from the time series database at each sampling time and arranged in chronological order to form a propagation rate time series. The grain boundary void propagation rates at the corresponding sampling times are read simultaneously and arranged in the same chronological order to form a propagation rate time series. The two time series are timestamped to remove records with missing or abnormal timestamps to ensure that the propagation rate and propagation rate strictly correspond one-to-one at each sampling time.

[0070] A first-order linear least squares fit was performed on the propulsion rate time series. Using the relative time of each sampling moment to the starting moment as the independent variable and the propulsion rate at each sampling moment as the dependent variable, an overdetermined system of linear equations was constructed. The slope and intercept of the fitted line were determined by solving the normal equations, and the slope was extracted as the slope of the propulsion rate change. The same first-order linear least squares fit was performed on the expansion rate time series, and the slope was extracted as the slope of the expansion rate change.

[0071] Second-order polynomial least squares fitting was performed on the propulsion rate time series and the expansion rate time series, respectively. With relative time as the independent variable and rate as the dependent variable, the coefficients of the quadratic term, the coefficients of the linear term, and the constant term were solved. The quadratic coefficients were multiplied by two to obtain the acceleration parameters, yielding the propulsion rate acceleration parameters and the expansion rate acceleration parameters, respectively. The slope of the rate change characterizes the direction and intensity of the linear trend of the micro-aging rate evolution over time, while the acceleration parameters characterize the degree of curvature change in the evolution trend. Both are used together in the subsequent calculation of the comprehensive aging status score.

[0072] S33. Comprehensive grading assessment of aging status using multiple parameters, specifically: S331. Dimensionless normalization of multi-source aging parameters, specifically: The normalized distribution density value of the peak region is extracted from the distribution density heatmap output by S31 as the distribution density value; the absolute values ​​of the velocity change slopes of the propulsion rate and the spread rate output by S32 are extracted and the arithmetic mean is taken as the comprehensive velocity change slope; the arithmetic mean of the propulsion rate acceleration parameter and the spread rate acceleration parameter output by S32 is taken as the comprehensive acceleration parameter.

[0073] The distribution density value, the slope of the overall rate change, and the overall acceleration parameter are all subjected to Min-Max normalization. That is, for each parameter, the current value is subtracted from the historical minimum value and then divided by the difference between the historical maximum value and the historical minimum value. The three parameters with different physical dimensions are uniformly mapped to the dimensionless interval of zero to one, and the density index, slope index and acceleration index are obtained respectively.

[0074] S332. Calculation and threshold comparison of the comprehensive score for aging status. Specifically: The density, slope, and acceleration indices are weighted and summed according to preset weights. The preset weight for the density index ranges from 0.30 to 0.50, with a classic value of 0.40; the preset weight for the slope index ranges from 0.25 to 0.40, with a classic value of 0.35; and the preset weight for the acceleration index ranges from 0.15 to 0.35, with a classic value of 0.25. The comprehensive score of the aging trend is obtained by multiplying each index by its corresponding weight and summing the results.

[0075] The first threshold ranges from 0.2 to 0.5, with a classic value of 0.35; the second threshold ranges from 0.5 to 0.9, with a classic value of 0.7, and the first threshold is less than the second threshold; the comprehensive score of aging status is compared numerically with the first threshold and the second threshold respectively.

[0076] S333. Classification and output of three-level early warning levels for aging status, specifically: When the overall score of aging status is less than the first threshold, it is determined that the current aging of the component is in a slow evolution stage, and the accumulation rate of micro-damage is lower than the engineering warning level, and it is classified as a mild warning level.

[0077] When the comprehensive score of aging status is greater than or equal to the first threshold and less than the second threshold, it is determined that the aging of the component has entered the accelerated evolution stage. The distribution density and rate evolution trend of the critical failure precursor zone show significant deterioration, and it is classified as a moderate warning level.

[0078] When the comprehensive score of aging status is greater than or equal to the second threshold, it is determined that the aging of the component is approaching the critical failure boundary, and the thermoelectric synergistic stress has caused severe deterioration of the microstructure, and it is classified as a severe warning level.

[0079] The obtained early warning level identifier is written into the aging status register for subsequent use by the thermal management strategy and current stress limit dynamic adjustment instruction generation module.

[0080] S34. Output of dynamic adjustment instructions for thermal management strategy and current stress limit: The system reads the aging status level identifier code obtained from S33 from the aging status register, and calls the control strategy library pre-stored in non-volatile memory based on the identifier code. This strategy library stores multiple sets of control parameter tables corresponding to mild, moderate, and severe warning levels, respectively. The system generates the corresponding cooling fan speed control signal and power circuit ripple current limiting command based on the control parameter table indexed by the current level identifier code.

[0081] When the aging status level is a mild warning level, the control strategy library outputs a maintenance command, the cooling fan speed control signal keeps the current reference speed unchanged, the power circuit ripple current limiting command keeps the current effective value upper limit of the ripple current unchanged, and only the monitoring log recording command is generated and written to the operation log cache area.

[0082] When the aging status level is at the moderate warning level, the control strategy library outputs boost and limit instructions. The cooling fan speed control signal is increased by 20% to 50% based on the current reference speed, and the power circuit ripple current limiting instruction reduces the upper limit of the effective value of the ripple current by 10% to 20% based on the current reference value. The two instructions are generated synchronously and enter the output queue.

[0083] When the aging status level is a severe warning level, the control strategy library outputs full-speed and emergency limiting commands. The cooling fan speed control signal drives the fan to enter full-speed operation. The power circuit ripple current limiting command reduces the upper limit of the effective value of the ripple current to below 50% of the rated value. At the same time, an active derating operation command is generated to reduce the overall power load of electronic components.

[0084] The aforementioned control signals and limiting commands are output through a standard communication interface. The cooling fan speed control signal is transmitted to the fan drive module of the external thermal management system via pulse width modulation or voltage speed regulation. The power circuit ripple current limiting command is transmitted to the power module controller of the current regulation module via a digital bus, and the external system executes the corresponding physical actions. Example 2

[0085] like Figure 2 As shown, an electronic component aging analysis system based on temperature and current coupling is used to implement an electronic component aging analysis method based on temperature and current coupling, including: A. Signal synchronization acquisition module, used to simultaneously acquire the housing temperature fluctuation signal and power circuit ripple current signal during the service operation of the electronic component under test; specifically including: The temperature sensing unit is mounted on the surface of the electronic component package, the high-frequency current sampling unit is connected in series with the power circuit, and the high-precision clock reference source is used. The temperature sensing unit and the high-frequency current sampling unit share the high-precision clock reference source to ensure that the temperature fluctuation signal of the package and the ripple current signal of the power circuit are strictly aligned on the time axis.

[0086] B. Thermal resistance dispersion characteristic matrix construction module, used to perform cross-correlation spectrum analysis on the shell temperature fluctuation signal and the power loop ripple current signal, extract the thermal resistance amplitude response sequence and the thermal resistance phase response sequence, and construct the interface thermal resistance dispersion characteristic matrix; specifically including: The system includes a frequency domain transformation unit, a cross-correlation operation unit, a normalization unit, a delay compensation unit, and a matrix combination unit. The frequency domain transformation unit performs Fourier transforms on the shell temperature fluctuation signal and the power loop ripple current signal to obtain the amplitude spectrum and phase spectrum, respectively. The cross-correlation operation unit calculates the cross-correlation amplitude and cross-correlation phase difference at each frequency point. The normalization unit normalizes the cross-correlation amplitude after normalization with the current amplitude and uses it as the thermal impedance amplitude response value. The delay compensation unit uses the cross-correlation phase difference after delay compensation and uses it as the thermal impedance phase response value. The matrix combination unit combines the thermal impedance amplitude response value and thermal impedance phase response value at all frequency points in complex form to form the interface thermal resistance dispersion characteristic matrix.

[0087] C. The resistance relaxation feature sequence construction module is used to inject probe current pulses into the metal interconnect structure of the electronic component under test under multiple preset test currents. During the relaxation period, the terminal voltage waveform is collected discretely along the time axis. The relaxation time constant distribution sequence and the relaxation amplitude weight distribution sequence are extracted through multi-exponential relaxation decomposition to construct the metal interconnect resistance drift relaxation feature sequence; specifically including: The system includes a programmable precision current source, a high-precision voltage sampling circuit, a baseline correction and denoising preprocessing unit, and a multi-exponential relaxation decomposition unit. The programmable precision current source is used to inject probe current pulses into the metal interconnect structure under multiple preset test currents. The high-precision voltage sampling circuit is used to discretely sample the terminal voltage recovery waveform at a fixed sampling interval during the relaxation period. The baseline correction and denoising preprocessing unit is used to perform baseline correction and digital low-pass filtering on the terminal voltage recovery waveform to obtain the net relaxation voltage waveform. The multi-exponential relaxation decomposition unit is used to decompose the net relaxation voltage waveform into multiple exponential decay components using a multi-exponential decomposition algorithm based on regularized inverse Laplace transform, and extract the relaxation time constant and relaxation amplitude weight.

[0088] D. Thermoelectric coupling feature fusion module, used to fuse the interface thermal resistance dispersion feature matrix with the metal interconnect resistance drift relaxation feature sequence into a thermoelectric coupling aging feature matrix.

[0089] E. Cooperative Competitive Aging Model Module: This module inputs the thermoelectric coupled aging feature matrix into the cooperative competitive aging model of the intermetallic compound growth front and grain boundary void propagation. It inverts the intermetallic compound growth front advancement rate under temperature stress based on the thermal impedance amplitude and the grain boundary void propagation rate under current stress based on the relaxation time constant. Specifically, it includes: The system includes a temperature stress analysis channel and a current stress analysis channel. The temperature stress analysis channel incorporates an inflection point frequency detection unit and a one-dimensional heat conduction inversion unit. It is used to identify the attenuation inflection point frequency based on the thermal impedance amplitude and invert the actual growth thickness of the intermetallic compound layer through the frequency domain relationship between the characteristic heat penetration depth and the thermal diffusivity, thereby calculating the propagation rate. The current stress analysis channel incorporates an electrochemical relaxation inversion unit and a cross-sectional area reduction calculation unit. It is used to invert the reduction of the effective conductive cross-sectional area based on the proportional relationship between the relaxation time constant and the metal interconnect resistivity, thereby calculating the propagation rate.

[0090] F. Critical Failure Precursor Zone Identification Module: This module calculates the synergistic competition index between the propulsion rate and the propagation rate. When the synergistic competition index deviates from the preset synergistic competition equilibrium range, the current stress range is marked as a critical failure precursor zone for thermoelectric synergistic accelerated aging. Specifically, it includes: The unit includes a cooperative competition index calculation unit and a balance interval comparison unit. The cooperative competition index calculation unit is used to perform a division operation with the propulsion rate as the numerator and the expansion rate as the denominator to obtain the cooperative competition index. The balance interval comparison unit is used to compare the cooperative competition index with a preset cooperative competition balance interval. When the cooperative competition index is less than the lower limit of the cooperative competition balance interval, it is marked as a current-dominated critical failure precursor zone. When it is greater than the upper limit, it is marked as a temperature-dominated critical failure precursor zone.

[0091] G. The aging condition classification and assessment module is used to generate a classification and assessment result of the current aging condition of the electronic component under test based on the distribution density of the critical failure precursor zone on the temperature-current two-dimensional stress plane and the evolution trend of its propagation and expansion rates; specifically including: The system comprises a two-dimensional stress plane discretization unit, a distribution density calculation unit, a time-series trend fitting unit, a normalization weighting unit, and a threshold comparison unit. The two-dimensional stress plane discretization unit discretizes the temperature-current two-dimensional stress plane into equal-area grid units according to preset temperature and current step sizes. The distribution density calculation unit counts the number of sample points in the critical failure precursor zone within each grid unit and calculates the local distribution density value. The time-series trend fitting unit performs first-order linear and second-order polynomial fitting on the propagation and expansion rates along the time axis to extract the rate change slope and acceleration parameters. The normalization weighting unit normalizes the density, slope, and acceleration indices and sums them according to preset weights to obtain a comprehensive aging status score. The threshold comparison unit compares the comprehensive aging status score with a first and a second threshold to classify the warning level as mild, moderate, or severe.

[0092] H. Dynamic Adjustment Command Output Module, used to output dynamic adjustment commands for thermal management strategies and current stress limits based on the graded evaluation results; specifically including: The system includes a control strategy library, a fan speed control signal generation unit, and a current limiting command generation unit. When the grading assessment result is a mild warning level, the current cooling fan speed and ripple current limit value remain unchanged. When it is a moderate warning level, a cooling fan speed increase command and a power circuit ripple current limit reduction command are generated. When it is a severe warning level, a cooling fan full-speed operation command, a power circuit ripple current emergency limit command, and an active derating operation command are generated.

[0093] As can be seen from the above description, the embodiments of the present invention achieve the following technical effects: This invention overcomes the limitations of traditional aging analysis by simultaneously acquiring shell temperature fluctuation signals and power circuit ripple current signals, and by introducing a dual-channel feature extraction mechanism of cross-correlation spectrum analysis and multi-exponential relaxation decomposition. The interface thermal resistance dispersion feature matrix constructed through cross-correlation spectrum analysis can accurately characterize the interface thermal resistance dispersion characteristics under the combined effect of temperature fluctuations and ripple current. The metal interconnect resistance drift relaxation feature sequence constructed through sampling during the current pulse relaxation period and multi-exponential decomposition based on regularized Laplace inverse transform can objectively reflect the grain boundary charge relaxation process. The thermoelectric coupling aging feature matrix formed by the fusion of these two approaches achieves a unified fusion of thermal domain dispersion characteristics and electric domain relaxation characteristics within the same characterization framework, providing a high-dimensional data foundation for subsequent inversion of microscopic aging mechanisms.

[0094] This invention establishes a competitive aging model of intermetallic compound growth front and grain boundary void expansion. It achieves quantitative separation and competitive assessment of the two main aging micro-mechanisms by inverting the growth front advancement rate under temperature stress using thermal impedance amplitude and the grain boundary void expansion rate under current stress using relaxation time constant. By calculating the competitive index between the advancement and expansion rates and identifying the critical failure precursor region based on a preset competitive equilibrium interval, the model can accurately determine whether the current aging is dominated by temperature stress or current stress, overcoming the bottleneck of traditional methods that can only macroscopically assess electrical parameter degradation but cannot pinpoint the dominant micro-failure mechanism. This model establishes a feasible mapping bridge between macroscopic electrical characteristics and microscopic physical rates, significantly improving the accuracy of identifying critical failure precursor regions and the specificity of failure mechanism diagnosis.

[0095] This invention, based on the distribution density of the critical failure precursor region on the temperature-current two-dimensional stress plane, combined with the slope and acceleration parameters of the propagation and expansion rates along the time axis, generates a comprehensive aging status score through multi-parameter normalized weighted fusion. This achieves precise three-level classification of component aging status: mild, moderate, and severe warning. Compared to traditional single-threshold criteria, this hierarchical evaluation mechanism can comprehensively reflect the spatial aggregation degree and temporal evolution intensity of micro-damage. Based on the hierarchical evaluation results, the system outputs differentiated dynamic adjustment commands for thermal management strategies and current stress limits, transforming heat dissipation control and current limiting from being independent to thermoelectric synergistic optimization. This proactively reduces the thermoelectric coupling stress intensity before components enter the severe warning stage, effectively delaying the aging process and significantly improving the remaining service life of electronic components and the reliability of system operation.

[0096] The embodiments and / or implementation methods described above are merely preferred embodiments and / or implementation methods for implementing the technology of the present invention, and are not intended to limit the implementation methods of the technology of the present invention in any way. Any person skilled in the art may make some modifications or alterations to other equivalent embodiments without departing from the scope of the technical means disclosed in the present invention, but these should still be regarded as the technology or embodiments that are substantially the same as the present invention.

[0097] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. The above descriptions are only preferred embodiments of this application. It should be noted that due to the limitations of written expression, while there are objectively infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of this application, and can also combine the above technical features in an appropriate manner. These improvements, modifications, changes, or combinations, or the direct application of the inventive concept and technical solution to other situations without modification, should all be considered within the scope of protection of this application.

Claims

1. A method for aging analysis of electronic components based on temperature and current coupling, characterized in that, include: Synchronously acquire temperature and current signals, perform cross-correlation spectrum analysis, extract thermal impedance amplitude response sequence to construct thermal resistance dispersion feature matrix, inject probe current pulse into relaxation period sampling voltage, extract time constant through multi-exponential relaxation decomposition to construct resistance drift relaxation feature sequence, and form thermoelectric coupling aging feature matrix; The thermoelectric coupling aging feature matrix is ​​input into the aging model of the co-competitive aging of the growth front of intermetallic compounds and the expansion of grain boundary voids. The propagation rate and expansion rate are inverted and the co-competitive index is calculated. When the deviation from the preset equilibrium range is marked as the critical failure precursor region. Based on the distribution density and evolution trends of the critical failure precursor zone, as well as its propagation and expansion rates, an aging status classification assessment result is generated, and dynamic adjustment instructions for thermal management strategies and current stress limits are output.

2. The aging analysis method for electronic components based on temperature and current coupling according to claim 1, characterized in that, The synchronous acquisition of temperature and current signals involves cross-correlation spectrum analysis to extract the thermal impedance amplitude response sequence and construct a thermal impedance dispersion feature matrix. A probe current pulse is injected into the relaxation period sampling voltage, and multi-exponential relaxation decomposition is used to extract the time constant and construct a resistance drift relaxation feature sequence, forming a thermoelectric coupling aging feature matrix. Specifically, this includes: During the service operation of the electronic component under test, the casing temperature fluctuation signal and the power circuit ripple current signal are collected simultaneously. Cross-correlation spectrum analysis was performed on the shell temperature fluctuation signal and ripple current signal to extract the thermal impedance amplitude response sequence and phase response sequence at each frequency, and the interface thermal resistance dispersion feature matrix was constructed. A probe current pulse is injected into the metal interconnect structure of the electronic component under test under multiple preset test currents, and the terminal voltage waveforms at both ends of the metal interconnect are discretely sampled along the time axis during the relaxation period after each probe current pulse is unloaded. Multi-exponential relaxation decomposition is performed on the voltage recovery waveforms at each terminal to extract the relaxation time constant and relaxation amplitude weight sequence, construct the metal interconnect resistance drift relaxation feature sequence, and then merge it with the interface thermal resistance dispersion feature matrix to form the thermoelectric coupling aging feature matrix.

3. The aging analysis method for electronic components based on temperature and current coupling according to claim 2, characterized in that, Cross-correlation spectrum analysis was performed on the shell temperature fluctuation signal and ripple current signal to extract the thermal impedance amplitude response sequence and phase response sequence at each frequency, and an interface thermal resistance dispersion feature matrix was constructed, specifically including: The synchronously acquired shell temperature fluctuation signal and power circuit ripple current signal are subjected to Fourier transform to obtain their respective amplitude spectrum and phase spectrum in the frequency domain. At each frequency point, the cross-correlation amplitude and cross-correlation phase difference between the shell temperature fluctuation signal and the power circuit ripple current signal are calculated. The cross-correlation amplitude is normalized by the current amplitude and used as the thermal impedance amplitude response value at the frequency point. The cross-correlation phase difference is compensated for by delay and used as the thermal impedance phase response value at the frequency point. Arrange the thermal impedance amplitude response values ​​and thermal impedance phase response values ​​at all frequency points in ascending order of frequency to form thermal impedance amplitude response sequence and thermal impedance phase response sequence, respectively. Using frequency as the row index and different operating conditions or different sampling times as the column index, the thermal impedance amplitude response sequence and the thermal impedance phase response sequence are combined in complex form to form the interface thermal impedance dispersion characteristic matrix.

4. The aging analysis method for electronic components based on temperature and current coupling according to claim 3, characterized in that, The process involves performing multi-exponential relaxation decomposition on the voltage recovery waveforms at each terminal, extracting the relaxation time constant and relaxation amplitude weight sequence, constructing a metal interconnect resistance drift relaxation feature sequence, and then fusing it with the interface thermal resistance dispersion feature matrix to form a thermoelectric coupling aging feature matrix. Specifically, this includes: The terminal voltage recovery waveform acquired after each probe current pulse is unloaded is subjected to baseline correction and noise reduction preprocessing to obtain the net relaxation voltage waveform. The net relaxation voltage waveform is decomposed into a superposition of multiple exponential decay components using a multi-exponential decomposition algorithm based on regularized inverse Laplace transform. Each component corresponds to a relaxation time constant and a relaxation amplitude weight. All relaxation time constants obtained by decomposition under the same preset test current are arranged in ascending order of value to form a relaxation time constant distribution sequence. The relaxation amplitude weights corresponding to each time constant are arranged in the same order to form a relaxation amplitude weight distribution sequence. The relaxation time constant distribution sequence and relaxation amplitude weight distribution sequence obtained under all preset test currents are combined in ascending order of test current to construct the metal interconnect resistance drift relaxation characteristic sequence, which is then fused with the interface thermal resistance dispersion characteristic matrix to form the thermoelectric coupling aging characteristic matrix.

5. The aging analysis method for electronic components based on temperature and current coupling according to claim 4, characterized in that, The thermoelectric coupling aging feature matrix is ​​input into the aging model of the cooperative competition between the growth front of intermetallic compounds and the expansion of grain boundary voids. The propagation rate and expansion rate are inverted and the cooperative competition index is calculated. When the deviation from the preset equilibrium range is marked as a critical failure precursor region, specifically including: The thermal impedance amplitude at each frequency point in the interface thermal resistance dispersion feature matrix and the relaxation time constant value in the metal interconnect resistance drift relaxation feature sequence are synchronously input into the aging model of the co-competitive aging of the intermetallic compound growth front and grain boundary void expansion. In the aforementioned cooperative aging model, the growth front advancement rate of intermetallic compounds under temperature stress is inverted based on the thermal impedance amplitude, and the grain boundary void propagation rate under current stress is inverted based on the relaxation time constant. Calculate the synergistic competition index between the propulsion rate and the expansion rate. When the synergistic competition index deviates from the preset synergistic competition equilibrium range, mark the current stress range as the critical failure precursor region of thermoelectric synergistic accelerated aging.

6. The aging analysis method for electronic components based on temperature and current coupling according to claim 5, characterized in that, In the aforementioned cooperative aging model, the propagation rate of the intermetallic compound growth front under temperature stress is inverted based on the thermal impedance amplitude, and the grain boundary void propagation rate under current stress is inverted based on the relaxation time constant. Specifically, this includes: The decay inflection frequency of thermal resistance amplitude with frequency change at each frequency point is extracted from the thermal resistance dispersion feature matrix of the interface. Combined with the thermal diffusivity of the electronic component under test, the actual growth thickness of the intermetallic compound layer along the heat flow direction is calculated by a one-dimensional thermal conduction model. The increment of the actual growth thickness within a unit thermal cycle is taken as the advance rate of the growth front of the intermetallic compound under temperature stress. Based on the proportional relationship between the constant values ​​of each relaxation time in the metal interconnect resistance drift relaxation characteristic sequence and the resistivity of the metal interconnect, and combined with the interconnect geometry and initial conductive cross-sectional area of ​​the electronic component under test, the reduction in effective conductive cross-sectional area caused by the expansion of grain boundary voids can be deduced. The rate of change of the reduction in the effective conductive cross-sectional area per unit energizing time is taken as the grain boundary void propagation rate under current stress.

7. The aging analysis method for electronic components based on temperature and current coupling according to claim 6, characterized in that, Calculate the synergistic competition index between the propulsion rate and the expansion rate. When the synergistic competition index deviates from the preset synergistic competition equilibrium range, the current stress range is marked as the critical failure precursor region of thermoelectric synergistic accelerated aging, specifically including: Substitute the growth front advance rate and grain boundary void propagation rate of the intermetallic compound obtained by inversion into the formula for calculating the cooperative competition index, where the cooperative competition index is equal to the ratio of the advance rate to the propagation rate. The calculated cooperative competition index is compared with the preset cooperative competition equilibrium interval. When the cooperative competition index is within the cooperative competition equilibrium interval, it is determined that the current aging is in a cooperative competition equilibrium state and no critical failure precursor zone is marked. When the cooperative competition index is less than the lower limit of the cooperative competition equilibrium range, it is determined that the current aging is driven by current stress, and the stress range is marked as the current-dominated critical failure precursor region; when the cooperative competition index is greater than the upper limit of the cooperative competition equilibrium range, it is determined that the current aging is driven by temperature stress, and the stress range is marked as the temperature-dominated critical failure precursor region.

8. The aging analysis method for electronic components based on temperature and current coupling according to claim 7, characterized in that, Based on the distribution density and evolution trends of the critical failure precursor zone, as well as its propagation and expansion rates, an aging condition classification assessment result is generated, and dynamic adjustment instructions for thermal management strategies and current stress limits are output, specifically including: Extract the coordinates of the marked critical failure precursor region on the temperature-current two-dimensional stress plane, and calculate the number of marked sample points per unit area as the distribution density. The growth front propagation rate of intermetallic compounds and the expansion rate of grain boundary voids in the critical failure precursor region are fitted along the time axis to obtain the slope of rate change and acceleration parameters. Based on the distribution density, rate of change slope, and acceleration parameters, the current aging status of the electronic components under test is divided into three levels: mild warning, moderate warning, or severe warning. Based on the classified aging status level, generate corresponding cooling fan speed control signals or power circuit ripple current limiting commands, and output them to the external thermal management system or current control module.

9. The aging analysis method for electronic components based on temperature and current coupling according to claim 8, characterized in that, Extract the coordinates of the marked critical failure precursor region on the temperature-current two-dimensional stress plane, and calculate the number of marked sample points per unit area as the distribution density, specifically including: The temperature-current two-dimensional stress plane is discretized into multiple equal-area grid elements according to the preset temperature step and current step, and each grid element corresponds to a temperature range and a current range. Traverse all sample points marked as critical failure precursor regions, count the number of sample points falling into each grid cell, divide the number of sample points in each grid cell by the area of ​​the grid cell to obtain the local distribution density value of the grid cell. The local distribution density values ​​of all grid cells are normalized, and a distribution density heat map of the critical failure precursor zone is generated with temperature coordinates as the horizontal axis, current coordinates as the vertical axis, and normalized distribution density values ​​as the vertical axis. The temperature range and current range corresponding to the peak area in the heat map are output as high-risk stress zones.

10. The aging analysis method for electronic components based on temperature and current coupling according to claim 9, characterized in that, Based on the distribution density, rate of change slope, and acceleration parameters, the current aging status of the electronic components under test is classified into three levels: mild warning, moderate warning, or severe warning. Specifically, these include: The distribution density value, the slope of the rate change, and the acceleration parameters are normalized to obtain dimensionless density, slope, and acceleration indices. The density index, slope index and acceleration index are weighted and summed according to preset weights to obtain the comprehensive score of aging status. The comprehensive score of aging status is compared with a first threshold and a second threshold, where the first threshold is less than the second threshold. When the comprehensive score of the aging status is less than the first threshold, it is classified as a mild warning level; when the comprehensive score of the aging status is greater than or equal to the first threshold and less than the second threshold, it is classified as a moderate warning level; when the comprehensive score of the aging status is greater than or equal to the second threshold, it is classified as a severe warning level.

Images