Options
2005
Conference Paper
Titel
Probabilistic Approaches for Fracture and Reliability Estimations of Microsystems
Abstract
A comparatively large scatter of both local material properties and random geometrical imperfections can often be observed within the material compounds of microsystems. Recently, much progress has been made in modeling, FE-simulation and experimental verification of fracture and lifetime behaviour of MEMS based on deterministic assumptions. But until now there is a lack in consideration of the scatter or uncertainty of certain characteristics. For example, the partial randomness of certain input parameters creates considerable uncertainties in the Finite Element determination of mechanical quantities which are provided for thermomechanical reliability optimization and lifetime prediction. Furthermore, deviations in failure driving characteristics may lead to a changeover in the ranking of competing failure mechanisms. The paper covers several approaches for taking into consideration randomness or uncertainty in failure and reliability estimation of microsystems. After structuring the various kinds and corresponding reasons of unavoidable scatter in parameters, the following problems are treated: (i) scatter in underlying materials structure - application of homogenization techniques, (ii) scatter in fracture mechanics parameters - randomization of primarily deterministic relations, (iii) scatter of geometry parameters - probabilistic FE analyses using Monte Carlo simulation techniques or response surface methods, (iv) variation of loading - rainflow methods, point processes and other methods of mathematical probability theory.