Skip to main content

Degradation Mechanisms & Lifetime Modeling

Power semiconductors are used in virtually every field of daily life, switching electric loads of various kinds - motors, lights and heating, sensors, just to name a few - or converting electrical power at highest efficiency. They serve in washing machines and laptop power supplies, in server farms and wind turbines, in trains and in cars. Often, controlling and connecting big power sources and power consumers, their safe operation and high reliability is not only an economic requirement, but also of greatest impact on system and consumer safety. Understanding the degradation mechanisms of such devices in a complex interplay of thermal, mechanical, and electrical loading in often harsh environment is a very challenging task, involving the competences of electronic engineers, simulation experts, materials scientists, physicists and mathematicians - with the final goal to permit accurate prediction of failure probabilities in the ppm-range as a function of use time.

KAI Competences

Since 2007, KAI is developing statistical lifetime and degradation models for power semiconductors in close cooperation with the process- and design engineers from Infineon Technologies and academic partners at renowned universities. KAI has established key competences in the areas:

  • Development of physical degradation models 
    • Stress test data analysis
    • Technology- and product understanding 
  • Use of numerical electro-thermal and thermo-mechanical simulation (more details: Modeling & Simulation)
  • State-of-the-art use of physical analysis at the FA labs of Infineon Technologies
    • Optical-, Electron-, Scanning Acoustic and X-ray microscopy
    • Advanced image analysis
  • Development of degradation-mode oriented stress testing
    • Adaption and enhancement of electrical stress test concepts
    • Development of test structures and test benches for thermo-mechanical fatigue analysis
    • Infrared thermography and laser vibrometry (more details: Material Characterization)
  • Statistical data analysis and lifetime models
    • Lifetime modeling and prediction methods for unimodal and multimodal distributions with censored data, (Bayesian) Regression, Mixtures-of-Experts models, Bayesian Networks and Gaussian Processes, accelerated lifetime models and bi-variate distributions Statistical evaluation methods for model and prediction quality
  • Stochastic degradation modelling
    • Stochastic processes to model degradation in semiconductor devices
    • Stochastic State-Space-Models

Research Goals

  • Analyze and identify the degradation processes taking place in different layers of semiconductor devices under electro-thermal loading conditions and clarify their relevance for device functionality
  • Identify suitable load- and strength related parameters which permit quantification of degradation and resistance in experiments and numerical simulation
  • Develop mathematical and statistical models which describe degradation and permit lifetime prediction – and provide a mathematical framework of stochastic state space modeling to capture variations in lifetime.
  • Develop and test new solutions in product- and technology design with improved reliability