Automated semiconductor wafer defect classification dealing with imbalanced data

Lee, Po-Hsuan; Wang, Zhe; Teh, Cho; Hsiao, Yi-Sing; Fang, Wei

doi:10.1117/12.2552838

Cited by 2 publications

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In over-sampling, the minority classes involve synthetic minority class examples. Lee et al [59] have performed experiments to test the performance of ML methods such as Random Forest, AdaBoost, XGBoost, and SVM on two imbalanced datasets with and without SMOTE. Results show that the ML algorithms combined with SMOTE using 25%-75% of the total training data achieved higher average class-wise accuracy than the same classifiers trained with 100% data but without SMOTE.…”

Section: Detection Review and Automatic Classification Of Defects And...mentioning

confidence: 99%

Electron microscopy in semiconductor inspection

Nakamae

2021

Meas. Sci. Technol.

View full text Add to dashboard Cite

Currently, semiconductor devices are manufactured in a technology node of several nanometers. Electron microscopy is mainly used in semiconductor inspection in manufacturing stages since accelerated electrons have wavelengths of nanometers or less, and a high spatial resolution can be expected. Among various electron microscopes since the scanning electron microscope (SEM) can observe the sample as it is without processing the sample, the SEM-based inspection instrument is mainly used at each stage of manufacturing the semiconductor device. The paper presents a review of SEM-based electron microscopy in semiconductor inspection. First, an overview of electron microscopy is described to understand the electron-sample interaction, the characteristics of electrons emitted from an irradiated specimen, charging, noise, and so on. Next, application areas such as mask inspection are introduced. Finally, future challenges are discussed.

show abstract

Section: Detection Review and Automatic Classification Of Defects And...mentioning

confidence: 99%