An Advanced Cloud-Based Virtual Metrology System with Automatic Ensemble Learning and Adaptive Model Retraining
Wisely Huang, Ryan Lan, Alonso Hsiao, Della Fitrayani Budiono, Jeff Ye, Sean Su, Dennis Huang, Leonard Yeo
Micron (United States) Micron (Singapore)
阅读操作
确认中在文库中上传 PDF 后可生成中文音频讲解。
摘要与影响
Virtual metrology (VM) is a key enabler for advanced semiconductor manufacturing, providing rapid estimation of quality metrics from process data to improve sampling efficiency, shorten feedback loops, and reduce wafer scrap in high-volume manufacturing (HVM). However, large-scale VM deployment remains challenging due to limited scalability and performance degradation under non-stationary manufacturing conditions. This paper proposes a cloud-based VM framework that integrates centralized data ingestion, scalable computation, standardized deployment, and automated model lifecycle management. An ensemble learning strategy is employed to enhance robustness against process variability, tool drift, maintenance events, and recipe changes. Model performance is continuously monitored using rolling validation and statistical gating, with automatic retraining triggered upon degradation. The framework is validated using one year of production data from multiple dry etch processes in an HVM memory fab. Results demonstrate stable prediction accuracy meeting production release criteria, improved robustness under changing conditions, and reduced deployment and maintenance effort, confirming the practicality and scalability of the proposed VM system.
逐年被引趋势
暂无年度引用数据
关键指标
同类平均 = 1
同领域 · 同年份 · 同类型
Google Scholar 与 OpenAlex 的被引统计范围不同,数值存在差异属正常。
AI 辅助阅读
依据:摘要
可就本文提问;依据不足时会说明。
学术脉络
学科主题
工程Fault Detection and Control Systems
Digital Transformation in Industry · Technology and Data Analysis