Robustness Verification Test of Inventchip G2 650V SiC MOSFET

**Abstract:** Based on classic lifetime models, IVCT (Inventchip Technology) conducted Robustness-Validation testing on a large sample size of its second-generation (G2) 650V SiC MOSFETs. Strictly following AEC-Q101 and ZVEI standards, the testing covered key items including HTRB, HTGB+, HV-H3TRB, and IOL. Classic reliability physics models such as Arrhenius, Hallberg-Peck, Coffin-Manson, and Eyring were introduced for data analysis. A total of 2,156 SiC MOSFET samples were subjected to extreme stress testing for up to 3,500 hours. Data analysis indicates that under typical application conditions, the predicted device lifespan reaches hundreds of years. This far exceeds the stricter 18-year lifespan standard set by IVCT for automotive-grade products, highlighting the product's excellent reliability. **1. Introduction and Verification Objectives** Silicon Carbide (SiC) power devices offer significant advantages in improving system energy conversion efficiency and power density, but their reliability remains a core concern for high-end applications. To quantitatively evaluate and confirm the long-term reliability of its products, IVCT initiated this robustness test, which goes beyond conventional requirements. The project aims to achieve the following objectives: *   Verify the reliability of the device over a minimum 18-year operating lifespan (stricter than the AEC-Q101 15-year standard). *   Adhering to the "Test-to-failure" philosophy, explore the failure boundaries and intrinsic failure mechanisms of the product. *   Establish a Safe Operating Area (SOA) for key parameters to provide effective data support for customer designs. **2. Methodology and Planning** **2.1 Reliability Models and Test Item Selection** Based on the AEC-Q101 standard, four accelerated life test items were selected: HTRB, HTGB+, HV-H3TRB, and IOL. Physical models including Arrhenius, Eyring, Hallberg-Peck, and Coffin-Manson were used for data analysis. All 2,156 samples were randomly selected from online mass-produced G2 650V SiC MOSFET products to ensure sample representativeness. The test results are summarized below: | Model | Test Item | Test Conditions | Quantity | First Failure Time | Batch T63 Failure Time | | :--- | :--- | :--- | :--- | :--- | :--- | | Arrhenius & Eyring | HTRB | 175℃; VS=VG=0V; VDS=650V | 6 Lots × 77ea/Lot | 2895 Hours | 3287 Hours | | Arrhenius & Eyring | HTGB+ | 175℃; VS=VD=0V; VGS=18V | 6 Lots × 77ea/Lot | 2766 Hours | 3332 Hours | | Coffin-Manson | IOL | ?T≥100℃, 5min on/off | 6 Lots × 77ea/Lot | No failure at 3500 Hours | | | Hallberg-Peck & Eyring | HV-H3TRB | 85℃, 85%RH, VD=520V | 10 Lots × 77ea/Lot | 3479 Hours | 3500 Hours (No new failures) | **3. Test Results and In-depth Analysis** **3.1 Overall Reliability Life Distribution: Bathtub Curve** The bathtub curve describes the variation in failure rates over a product's lifecycle, divided into three stages: Early Failure Period (Infant Mortality), Random Failure Period (Useful Life), and Wear-out Failure Period (Aging). Analysis of failure data after 3,500 hours for each test group reveals significant differences in failure performance, indicating that the samples in each group are at different lifecycle stages. This is illustrated on the bathtub curve as follows: Figure 1: Bathtub Curve—Test samples from different groups are at different life stages. The analysis of the test results is as follows: *   **HV-H3TRB:** The random failure period of the samples exceeds 3,500 hours, highlighting the product's excellent humidity resistance. *   **HTRB/HTGB+:** Samples entered the critical phase of random failure at 3,000 hours and entered the wear-out failure period between 3,000 and 3,500 hours. *   **IOL:** Samples were far from entering the wear-out failure period at 3,500 hours, proving the extremely high reliability of the product's package interconnection. In summary, combined with the graphical analysis, the IOL test showed the best reliability performance, followed by HV-H3TRB. HTRB and HTGB+ entered the wear-out period earlier on the bathtub curve, but they still perform excellently under actual application conditions, with a predicted lifespan far exceeding design standards. **3.2 HV-H3TRB Test Results and Analysis** **Test Conditions:** T=85°C, RH=85%, VDS=520V. **Leakage Current Stability:** Before and after 2,500 hours of stress application, the device leakage current growth did not exceed 1μA, no failure criteria were triggered, and no systematic degradation mechanism was found. Figure 2: Leakage current before and after stress application. Figure 3: Leakage current multiplier before and after stress application. Figure 2 above shows that the growth in leakage current before and after stress application does not exceed 1uA. Figure 3 shows the leakage current multiplier before and after stress application. The growth in leakage current did not exceed the failure judgment standard set for each node, and even after 3,500 hours of testing, the leakage current increase did not exceed 5 times. No systematic EOL (End of Life) mechanism was found. **Safe Operating Area (SOA) Analysis:** The contour plot below defines the product life boundaries under different temperature and humidity conditions. As shown, based on calculations using the Hallberg-Peck and Eyring models, under typical conditions (65°C, 75% RH, 650V), the calculated lifespan reaches 869 years, far from the 18-year safety limit represented by the red area. Figure 4: HV-H3TRB SOA Analysis. **Calculation Process:** Hallberg-Peck Model and Eyring Model: [Formula Image] From the test, it is known that the first failure time t=3479hrs, Tt=85°C, RHt=85%, Vstress=520V, ss=10 lots*77ea. We can predict the MTTF (Mean Time To Failure) and FIT (Failure In Time) with Ea=0.9eV, 90% confidence interval, humidity acceleration constant p=3, and voltage acceleration constant β=1. Under test conditions of Tu=85°C, RHu=85%, Vnormal=650V: AF = AFt * AFv = 0.8 λ = (X2*(%CL,2f+2)*10^9) / (2*AF*t*ss) = 1074 FITS MTTF = 1/λ = 1e9/1074 = 930,721 hours = 106 years Under typical application conditions of Tu=65°C, RHu=75%, Vnormal=650V: AF = AFt * AFv = 6.54 λ = (X2*(%CL,2f+2)*10^9) / (2*AF*t*ss) = 131 FITS MTTF = 1/λ = 1e9/131 = 7,613,547 hours = 869 years **3.3 HTRB Test Results and Analysis** **Test Conditions:** T=175°C, VDS=650V. **3,500-Hour Extended Test:** The actual first failure time T_actual = 2,895 hours. Although the HTRB test entered the wear-out period earlier in the bathtub curve, its calculated lifespan under actual application scenarios still far exceeds the design target. The actual batch failure time T63 = 3,287 hours (T63 is the characteristic life, indicating the time at which 63.2% of samples fail), which is significantly better than the 2,000-hour target. The Reliability Improvement Factor (RIF) = 4.59, proving ample design margin. **Quantifying Robustness:** The Robustness Improvement Factor (RIF) formula = [ln (TTF_actual) - ln (TTF_target)] / 3σ is used to calculate the magnitude of improvement of the actual value relative to the target value. If RIF < 1, the target requirement is not met; if RIF = 1, it just meets the target; higher values indicate better robustness. The RIF for this test is 4.59, proving that the product has ample robustness design margin. Figure 5: HTRB β and RIF Calculation. **Weibull Analysis:** Commonly used to describe the statistical model of device life distribution. The shape parameter β can be used to judge the failure type (early failure, random failure, wear-out failure). A shape parameter (slope) β=1 indicates that initial failures are random failures. After excluding these random failure samples, the calculated shape parameter β=26.85, which is far greater than 1, indicating that the remaining products have entered a stable intrinsic wear-out period.