Abstract
Light emitting diode (LED), which has widely applied in different illuminations, is a kind of luminous semiconductor devices. Wire bonding is one of the main processes used for connecting the signal of chip. Thermosonic bonding is often applied in the LED wire bonding process, and it’s a multi-physic process including impact stage, ultrasonic vibration stage, and lift off stage. The failures of LED such as pad peeling, cracking, and delamination might influence the power connecting of chip. In a word, predicting and analyzing the failure of LED chip during wire bonding process is an important issue. In the impact stage of wire bonding process, the material property of free air ball (FAB) is much softer than capillary, which compresses the FAB to land on pad. Thus, a large deformation phenomenon comes out on FAB during wire bonding process. Due to the large deformation problem, it’s easy to have element distortion issue on FAB while executing simulation work of wire bonding process. To avoid the divergent problem of finite element (FE) analysis, many researches reduce the compression distance of capillary on their contact simulation or increase mesh density. However, reducing distance would lack fidelity, and increasing mesh density would take more time in calculation. The element distortion issue still cannot be solved. This research will construct FE models to discuss the element distortion issue using commercial software ANSYS®/LS-DYNA with explicit method and utilize 2nd level accuracy arbitrary Lagrangian-Eulerian algorithm, and ALE method with different mesh smoothing algorithms to solve the element distortion problem. In the end of research, utilizing ALE method with equilibrium mesh smoothing algorithm demonstrates better element quality with excellent geometry prediction in impact stage of bonding process. In summary, this simulation method not only conquers the element distortion problem and lack of capillary displacement problem, but proposed an effective methodology for simulating wire bonding process. It is believed the stress/strain history and contact force will give better accuracy than other mesh smoothing algorithms or non-ALE methods.