EUVIMEDEUVIMED entdecken
Crossref

Ensemble Virtual Metrology Models of Semiconductor Processes

Yawei Chen, Shaohua Chang, Eric Chien

IEEE Sensors Journal · 01.05.2026 · 26 (9) · 14127-14136

DOI: 10.1109/jsen.2026.3678302 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Sensors Journal, erschienen am 01.05.2026. Verfasst von Yawei Chen und 2 weiteren Autor:innen. Die Arbeit nennt 36 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The increasing demand for virtual metrology (VM) is advancing semiconductor manufacturing fabrication (SMF) by enabling real-time quality control. While linear models such as the least absolute shrinkage and selection operator (LASSO) are lightweight and interpretable, they struggle to capture strong coupling and nonlinear correlations in advanced process parameters, limiting their accuracy. Nonlinear ensemble models, such as random forests (RFs), can capture complex patterns and are robust; however, they suffer from feature redundancy, complex hyperparameter tuning, a lack of interpretability, and a risk of overfitting with small sample sizes. Improving prediction accuracy remains challenging due to ensemble machine-learning (EML) models’ difficulty in extracting hybrid features. Additionally, understanding VM system operation in real-world settings is vital for enhancing cycle time, measurement capacity, and wafer yield—key SMF performance indicators. This study proposes a two-level EML feature extraction (FE) model that combines LASSO and RF to predict fluorine concentration in fluorosilicate glass (FSG) dielectric films during the high-density plasma (HDP) process with explainable artificial intelligence (AI) technology. Results demonstrate that the model achieves low complexity, high efficiency, 98.5% accuracy, and rootmean- square error (RMSE) of 0.009, outperforming conventional EML models. These findings challenge assumptions regarding computational complexity and efficiency in hybrid EML approaches.

Zitieren

APA

Chen, Y., Chang, S., & Chien, E. (2026). Ensemble Virtual Metrology Models of Semiconductor Processes. IEEE Sensors Journal, 26(9), 14127-14136. https://doi.org/10.1109/jsen.2026.3678302

Vancouver

Chen Y, Chang S, Chien E. Ensemble Virtual Metrology Models of Semiconductor Processes. IEEE Sensors Journal. 2026;26(9):14127-14136. doi:10.1109/jsen.2026.3678302

BibTeX
@article{Chen2026,
  author = {Chen, Yawei and Chang, Shaohua and Chien, Eric},
  title = {Ensemble Virtual Metrology Models of Semiconductor Processes},
  journal = {IEEE Sensors Journal},
  year = {2026},
  volume = {26},
  number = {9},
  pages = {14127--14136},
  doi = {10.1109/jsen.2026.3678302}
}
Kostenlos starten →