Through the technique, SamplingPrediction of Lithography Overlay Errors, the costtime of overlay error measurement can be reduced to improve the process efficiency. We identify key sampling through clusteringmachine learning models,design a new sampling algorithm in photolithography process. Due to the complexity of wafermany training factors of wafer data, we combine the clustering algorithm with incremental learning to meet customers＇ unique needsachieve the goal of optimally samplingreducing the costs.

Due to the lack of a large amount of wafer measurement data in practice, we combine the clustering with incremental learning to meet customers＇ unique needs. Moreover, we develop a dynamic clustering algorithm to automatically identify the correct number of clusters for different types of attributes,gradually build learning models with a few training data to predict wafer exposurecompensation. While repeatedly learning new data, it also retains the knowledge of previous models. In addition, we show the new sampling algorithm outperforms the commercial tool by 10 under key measures.

We develop a new sampling algorithm with our partner companies to analyze the common features of overlay by dynamic clustering. We select testkeys through the incremental learning model,predict the actual exposure of entire wafers. We also extend the sampling technique to the testing process of Wafer Probe to reduce the cost of the testing process. In addition to developing relevant forward-looking technologies, it is expected to enhance Taiwan’s cutting-edge technologyraise the world competitiveness in the semiconductor manufacturing industry.

天使投資人、策略合作夥伴