Efficient Optimization Approach of Partial Discharge Inception Voltage for Substrate Trench of Power Modules Based on Chain-Equivalent Circuit Model

View at publisher
Authors: Yingbin Li, Cao Zhan, Lingyu Zhu, Hao Sun, Yuntian Gao, Simeng Li, Shengchang Ji
Cited by: 0
References: 47

Abstract

Insulation design in terms of controlling partial discharge (PD) is critical for high voltage power modules. PD mitigation at trench between copper pads on the AMB substate surface is challenging due to the strong electric field strength at the triple point (TP). Considering the strong vertical component of the electrical field, a chain-equivalent circuit model is proposed to quantitatively analyze the influencing factors of the TP electric field strength. The factors governing the electric-field strength at the TP can be classified into two regimes, dominated by dielectric and conductive effects, respectively. For a dielectric-dominated AMB substrate, it is further revealed that the partial-discharge inception voltage (PDIV) scales proportionally with the square roots of the ceramic thickness, the package-material thickness, and the package material's relative permittivity, while scaling inversely with the square root of the ceramic permittivity, and is independent of frequency. Based on this finding, an accurate PDIV prediction model for the TP is proposed, which is experimentally validated by only illustrating an average error of 5.68%. Moreover, a case study of a 0.9 mm AlN/silicone-gel specimen, which achieves a PDIV of 13.17 kV, manifests the feasibility and efficiency of proposed scheme based on developed model. The study provides quantitative and efficient guidance for the insulation optimization of high-voltage power-module substrates.

Fields of Study & Subjects

High voltage insulation and dielectric phenomena Silicon Carbide Semiconductor Technologies Electromagnetic Compatibility and Noise Suppression

Publication details

Journal / source
IEEE Transactions on Power Electronics
Volume
41
Issue
9
Pages
16099–16114
Publication date
6 March 2026
Language
EN