AI learns to predict chip polishing quality from sensor noise, making smart factories smarter
Original headline: Self-Supervised Virtual Metrology For CMP MRR Prediction (SUTD, A*STAR, NUS) — semiengineering.com
More in Semiconductor and chip news
Researchers from Singapore developed a new AI framework called GREP (Global Regression Pretraining) that can predict the material removal rate (MRR) during chip manufacturing's chemical mechanical planarization (CMP) process. The self-supervised system learns patterns from unlabeled sensor data using attention mechanisms, then fine-tunes with limited labeled examples to estimate how effectively silicon wafers are being polished.
The technique helps semiconductor fabs monitor manufacturing quality without constant physical measurements, addressing challenges like sensor noise, process drift over time, and limited training data. This could lead to more efficient production lines with better yield prediction and less downtime. The research represents a practical advancement in bringing AI-driven 'virtual metrology' to high-precision manufacturing environments.
Read the original at semiengineering.com
Satirical commentary on real tech news, summarised with AI. Read the original source for the facts. How ZeroDaily works
