Uncertainty Quantification in Electro-Thermal Coupled Problems based on a Power Transistor Device

Abstract In many applications, used in power electronic systems, the physical domain and/or material parameters cannot be determined precisely. Such geometrical variations, which have directly impact on yield and performance, are primarily caused by both lithography proximity effects and process imperfections. Consequently, the mathematical modeling results in a stochastic, coupled, system of equations, which first can be partitioned into subsidiary problems, and then solved numerically by co-simulation techniques or a monolithic approach. In the formulation and the implementation of our particular problem the Spectral Collocation Method (SCM) with a Polynomial Chaos Expansion (PCE) were used. Furthermore, we applied uncertain analysis for the power transistor device, which is of importance for the automotive industry. stochastic modeling; uncertainty quantification; polynomial chaos; electronics; power device