A Computationally Efficient IGBT Lifetime Prediction Method Based on Successive Initiation Technique by Iteratively Using Clech Algorithm
Xin Yang, Xinlong Wu, Ke Heng, Guoyou Liu
Abstract
Thermal–mechanical coupled models built by finite-element analysis (FEA) software are attractive due to its excellent capability to calculate thermal/mechanical behaviors of the material interfaces inside the power module. However, these FEA-based approaches will consume enormous computational resources, resulting in unacceptably long simulation times. Here, a computationally efficient and accurate insulated gate bipolar transistor (IGBT) lifetime prediction method enabling explicit emulation of solder layer degradation is proposed with the assistance of Clech algorithm. The variation of assembly stiffness <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$K$ </tex-math></inline-formula> and imposed strain <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$D$ </tex-math></inline-formula> in the Clech algorithm with the geometrical variation during the solder layer degradation is particularly considered. A lookup table of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$K$ </tex-math></inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$D$ </tex-math></inline-formula> changing with crack propagation is established to calculate viscoplastic strain energy density <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\Delta W$ </tex-math></inline-formula> (VSED) of the critical elements across the crack path. This improved process is applied for lifetime consumption calculation of each successive run. It remarkably reduces the simulation costs compared to that based on FEA. Finally, two sets of experimental results verify high prediction accuracy and low time-consuming cost of the proposed method.