Mesoscopically understanding stress distribution and sensitivity in layered-porous sintered silver for high-power electronics

Purpose: The purpose of this paper is to mesoscopically analyze the impact of parameter variations in the random pore structure on the stress distribution of layered-porous sintered silver used in high-power electronics, and to conduct a variable importance analysis of the parameter variations in the random pore structure. Design/methodology/approach: Sintered silver, featuring a porous structure, improves thermal and mechanical performance by effectively absorbing stress and facilitating heat dissipation. To ensure the performance and scalability of layered-porous sintered silver, this paper uses Gaussian random fields to model the random pore structure and performs a sensitivity analysis on pore characteristic length and porosity, both of which significantly impact the stress distribution within the sintered silver layer. First, multiple sets of random pore models with varying characteristic lengths and porosities were generated using Gaussian random fields. Then, the maximum stress of the sintered silver layer containing random pores under power cycling conditions was extracted. Finally, the Morris screening method was used to perform a sensitivity analysis on the variables of the random pore structure that affect the maximum stress in the sintered silver layer. The systematic evaluation of the parameter variations in the random pore structure was conducted to assess their impacts on the maximum stress in the sintered silver layer. Findings: Due to the high randomness of the pore structure generated by the Gaussian random field function, the maximum stress in the sintered layer fluctuates with different mesoscopic models. After systematic evaluation using the Morris screening method, it was found that the maximum stress in the sintered silver layer is most sensitive to the variation in the pore characteristic length in the x-direction. Reducing the length of pores in the x-direction can significantly decrease the stress concentration between pores in the sintered silver layer after power cycling. Originality/value: This paper innovatively uses a Gaussian random field to model the mesoscopic structure of layered-porous sintered silver for high-power electronics, and applies the Morris screening method to perform variable importance analysis on the stress distribution results within the sintered silver layers. The mesoscopic study demonstrates that the maximum stress in the sintered silver layer is most sensitive to changes in the pore characteristic length in the x-direction.