What´s Next After Process Characterization

Semiconductor packaging industry has been intensively driven into miniaturizing and thinning of package with more functionality, in short, "getting more with less". With the increasing complexity in packaging, the process characterization has been pushed to the limit in order to maintain a stable and predictable process. This has led to great challenges in thorough process characterization using sequential design of experiments (DOEs) to ensure the most optimized process window is achieved. However, a well characterized process today may not behave in similar manner tomorrow. Any slight variation in raw material or machine inherent variation will drift the optimized process window. This will definitely affect the stability and predictability of the process. It has been observed that under certain situation, the process seems to be perfectly characterized and optimized during engineering run, but low ppm of defects will only be surfaced out during production run. The reason for this observation is that process characterization during engineering run is mostly performed under controlled environment using the best performed machine with one single batch of material. Additional process characterization through detailed Design of Experiments (DOEs) and further optimization through Response Surface Methodology (RSM) may not help to resolve the issue. Major portion of this issue can be attributed to machine-to-machine variation, time instability of the machine and material variation. In order to maintain machine-to-machine consistency and time stability, equipment characterization methodology has been developed based on Six Sigma DMAIC methodology. It covers 5 major phases, which are Define, Measure, Analyze, Improve and Control Phase. Within every phase, data analysis is performed using statistical tools in order to draw data-driven conclusion. This paper will discuss about how Six Sigma DMAIC approach is used to develop effective equipment characterization methodology in- order to minimize machine-to-machine variation and maintain time stability. Statistical tools employed in every phase will be discussed as well. The focus of this paper is on the methodology developed and used in one case study. The equipment characterization methodology has been successfully used in the case study to achieve significant OEE improvement by more than 5%, which results in additional capacity without capital. Other advantages include better machine performance with more predictable process, better machine portability, reduction of non-screenable quality related issues and improvement in yield loss.