Package yield enhancement using machine learning in semiconductor manufacturing

The production difficulty of the DRAM family has rapidly increasing. Several factors can be adduced to explain. For example, increasing of step in the production process by the miniaturization of semiconductor process, challenges of Low-power designed due to the growth of mobile family. And the two are connected by a combination of the result. It may increase the difficulty of an analysis of the test area for the yield improvement. This study analyzed the Fab measurement Data, Wafer Test Data, Package Test Data and looked at Package Yield Improvement. It covered Data pre-processing, Feature Selection, and then verify the relevant data, to derive an optimized relationship, and the actual production process applied too.