材料科学
钎焊
温度循环
有限元法
热膨胀
陶瓷
复合材料
动力循环
导电体
材料性能
机械工程
热的
电子工程
可靠性(半导体)
结构工程
功率(物理)
工程类
物理
气象学
量子力学
合金
作者
Giuseppe Mirone,Alessandro Sitta,Giuseppe D’Arrigo,Michele Calabretta
标识
DOI:10.1109/tdmr.2019.2932971
摘要
The metallized insulating substrates work as mechanical supports for the circuitry of Power Module Packages. Due to their specific functions, substrates for power electronics are made by different materials. The conductive metal layers can assume different functions: the top metal serves as power circuitry routing while the bottom metal improves the mechanical robustness and thermal efficiency. Ceramic layer provides excellent electrical insulation. These features play an essential role in the operation of power modules, which are often operated at high voltage and high current density. The substrates, composed by materials with different thermal expansion coefficients, are subjected to cyclic stresses due to temperature variations induced by operative working conditions. The substrate layouts typically include differences in shape and/or thickness between the top and the bottom side; this generates asymmetrical distributions of stress/strain resulting in overall warpage. The variations of this warpage induce mechanical fatigue during lifetime and represent a limiting factor for reliability. The scope of the presented work is the characterization of the out of plane warpage of Active Metal Brazed substrates (AMB) by means of numerical approach. The elastoplastic properties of metal and ceramic have been measured, evaluating the thermal softening of the copper as well. These characteristics are needed to calculate AMB warpage through Finite Element Models (FEM), simulating the warpage induced by a passive temperature cycling. Warpage computed from numerical model have been benchmarked and validated with optical warpage measurements. The validated numerical model has been developed to optimize the substrate warpage variation during cycling improving the whole package reliability.
科研通智能强力驱动
Strongly Powered by AbleSci AI