Modeling and Optimization of Mechanical Performance for Cu Wire Bonding Process

Wire bonding is a critical process to establish interconnections between semiconductor chips and the external world. In this study, finite-element analysis is used to simulate and optimize Cu wire bonding process with pre-deformed surface (Anvil Bonding). Numerical modeling found that Anvil bonding has smaller IMD stresses but larger FAB-metal co-deformed strains due to its larger initial contacts compared to conventional bonding. To further optimize the Anvil bonding process, parametric finite-element models are incorporated with Evolutionary Algorithm using ANSYS®optiSLang. The following parameters are considered: tip diameter, chamfer diameter, bore diameter, chamfer angle of bonding tool, thickness of IMDs, and different initial contact areas between the FAB and the bond pad. Two criteria are used for optimization, which are higher shear plastic strains of top metal and lower IMD tensile stresses. The best solutions were achieved after running 210 designs. The optimization shows that the combination of larger initial contact areas and thicker IMDs is the key to a better FAB-metal co-deformation and lower IMD stresses. The optimization methodology is considered effective for wire bonding process and could be extended to other processes in semiconductor devices.

I. Introduction

Wire bonding is a key interconnect technology that makes electrical interconnections between integrated circuit chips and lead frames. It minimizes the die area and the die cost as the region below the bond pad that is utilized for active devices. Wire bonding also has fast-bonding speed and high bonding strength, making it a widely used method in semiconductor devices for automotive, consumer and industrial applications [1–4]. Recently, copper wire bonding has been considered as a good substitute of gold wire bonding due to its low cost as well as better thermal and electrical performance [5].