A cost benefit analysis: The impact of defect clustering on the necessity of pre-bond tests

The emerging three-dimensional integrated circuit (3D IC) provides a promising solution for sustainable computer performance scaling. However, the high cost due to the complex pre-bond/intermediate testing and the low compound yield hinder the commercial adoption of 3D ICs. The defect clustering is found biasing the yield prediction, resulting in an unrealistic cost estimation at the early design stage. Under defect clusterings, it is possible that we eliminate the pre-bond/intermediate test. Moreover, components in a chip might be functional even with defects, which leads to the partially functional dies. Therefore, in this work, we propose a framework to analyze the cost benefit with test elimination and partially functional dies. To support our proposed framework, a hierarchical yield model and an analytical model for test elimination considering defect clustering are provided. Through analyzing the impact of the pre-bond/intermediate test elimination on the final stacking cost with our framework, we show that the wafer cost is the most important factor on the test elimination. Moreover, further cost saving can be obtained by combining partially functional dies with test elimination.