Incorporation of Big Data in Methodology of Identifying Corrosion Factors in the Semiconductor Package

The semiconductor packaging industry driven by packaging complexity and product miniaturization. Hence, the problem identification methodology in semiconductor industries is a critical interest, and a basis of continuous improvement where the lesson learned is an integral part of it. Nevertheless, the problem identification approach is stagnant at the traditional method, such as the statistical-based methodology. There are several studies on the problem identification process in semiconductor through the six-sigma methodology and statistical approach, however, the scope is limited to the inferential statistic. Therefore, the focus of this paper is proposing using big data approach which grounded on the information theory. The big data analysis approach is utilizing the algorithm and data visualization. Big data methods, such as MINE and clustering was applied to data from hundreds of variables that contain essential and undiscovered relationship. The big data analysis enables the potential factors that contributed to the root causes and provided significant input to the design of experiment and reliability analysis.