1-4244-0276-X/06/$20.00 C 2006 IEEE 7th. Int. Conf. on Thermal, Mechanical and Multiphysics Simulation and Experiments in Micro-Electronics and Micro-Systems, EuroSimE 2006 —1— Interfacial Adhesion Method for Semiconductor Applications Covering the Full Mode Mixity J.Thijsse 1,4 , W.D. van Driel 2,3 , M.A.J. van Gils 2 , O. van der Sluis 1 1) Philips Applied Technologies, High Tech Campus 7, 5656 AE Eindhoven, The Netherlands 2) Philips Semiconductors, Gerstweg 2, 6500 MD Nijmegen, The Netherlands 3) Delft University of Technology, Mekelweg 2, 2628 CD Delft, The Netherlands 4) Eindhoven University of Technology, Den Dolech 2, 5612 AZ Eindhoven, The Netherlands J.Thijsse@student.tue.nl Abstract Currently, prediction of interface strength is typically done using the critical energy release rate. Interface strength, however, is heavily dependent on mode mixity. Accurately predicting delamination therefore requires a material model that includes the mode dependency of interface strength. A novel test setup is designed which allows mixed mode delamination testing. The setup is a stabilized version of the mixed mode bending test previously described by Reeder and Crews [5,6]. It allows for the measurement of stable crack growth over the full range of mode mixities, using a single specimen design. The crack length, necessary for calculation of the energy release rate, is obtained from an analytical model. Crack length and displacement data are used in a finite element model containing a crack tip to calculate the mode mixity. 1. Introduction Delamination is a common cause of failure in semiconductor packages (figure 1). Characterization and prediction of interface behavior in manufacturing, testing and application conditions is necessary in order to reduce development times and costs. Figure 1: semiconductor package failure The success of interfacial fracture mechanics approach to analyse these failures in IC packaging strongly depends on accurate characterization of the critical adhesion strength, G C . However, its measurement is complicated by the fact that adhesion depends not only on moisture concentration, C, but also temperature, T, and mode mixity, ψ. Previous research has dealt with the investigation of interface strength as a function of temperature and moisture [1]. The current research deals with the measurement of interface strength as a function of the mode mixity. There are three basic types of loading that a crack can experience; opening (mode I), sliding (mode II) and tearing (mode III). This research will focus on mode I and mode II. The ratio of mode I to mode II loading is characterized by the mode angle, or mode mixity. A mode angle of 0 o describes pure mode I, pure mode II loading is described by a mode angle of 90 o . In general, interface strength will be higher under mode II loading than under mode I loading, as shown schematically in figure 2. Figure 2: interface strength versus mode angle The goal of our research is the development of a method to characterize interface strength for different material combinations over the full range of mode mixity. A test setup will be analyzed and designed to allow the loading of a sample under different mode angles. FEM analysis will be used to obtain both mode mixity and crack length data, used for determination of G c , from the experimental results, resulting in the determintation of G c as a function of ψ. 1. Theory of mixed mode testing Using linear elastic fracture mechanics (LEFM), a criterion for crack growth can be obtained by regarding the energy balance of the material (equation 1), where U represents energy per unit of time and volume: k d a i e U U U U U + + + = (1) With U e the external mechanical energy supplied to the material, U i the internal elastic energy stored by the material, U a the energy dissipated by crack growth, U d energy dissipated by other mechanisms and U k the kinetic energy. It is assumed that U d =0, taking crack growth to be the only cause of dissipation, and that U k =0, meaning that crack growth is slow enough for the change in kinetic energy to be neglected. Dividing the remaining terms by the sample width B and crack length a gives equation 2,