Efficient Determination of Copper Electroplating Chemistry Additives using Advanced Neural Network Algorithms

Abstract

Copper plating is the metallization process of choice for modern semiconductor devices. It has proven to be a relatively inexpensive and simple process and easily adaptable to high volume manufacturing. The development of the copper plating process includes the addition of organic components that provide a uniform and smooth surface. These organic components are not only consumed during the plating process, but also decompose over time. To insure a repeatable process these organic components, typically added in parts per million concentrations, must be carefully controlled. To do this, the industry has developed chemical analysis techniques such as Modified Linear Approximation, Dilution-titration, and Response curves to assist in determining the exact concentration of the organics in the plating bath. These techniques, while widely used, are time consuming, wasteful, and inaccurate. A new technique is proposed that will speed up the process, reduce the complexity and waste, and provides a higher accuracy. These techniques will utilize recently introduced second order Advanced Neural Network (ANN) algorithms developed at Auburn University.