Project background
In the semiconductor industry chips with golden wires are connected to electrical circuits, the so called wire bonding process. The gold price has increased the last couple of years and changing the golden wires by copper wires seems worthwhile for the bonding technology. However, gold is very ductile and easy to bond, while copper shows strain hardening during the bonding process and is harder to bond. The forces that are associated with copper bonding could more easily cause breaks in the chips, which is undesirable. Nonetheless the interest from the industry is huge, as there is a large price difference between copper and gold. If gold can be replaced by copper bonding this will yield a significant saving.
 |
 |
The challenge
Simulate the bonding process of a copper wire, and determine how the bonding process can be made in such a way to give a high yield. Because the parameters of the material changed as a function of temperature, the optimal temperature had to be analysed. The cycle time between the bonding has to be short, and may not deviate too much from gold bonding.
FEM simulation
After a literature study Wilde Analysis was contacted, a software developer from the UK. They distribute DEFORM a Finite Element Method (FEM) tool which is commonly used in the semiconductor industry to study the bonding process. This software was applied to the products of our client, after several analyses were done on the material parameters, such as strain hardening, the Youngs modulus and other relevant parameters for a good FEM simulation. The simulation were done on several copper alloys, and the studies were time dependent.
Results
The increase of stress during the deformation process was analysed. This was assessed for several chip variants and compared to cases of gold bonding for which a lot of experience exists. Compared among each other the alloys resulted in different speeds of the deformation process. Several recommendations were done to control the increase of stress and thus keep breaks to a minimum.