A new research paper introduces a hierarchical, uncertainty-aware approach for virtual metrology in phototransistor gain prediction, specifically designed for small production datasets. The study, conducted on a real fabrication history, found that a significant portion of device gain variance occurs between process runs rather than within them, limiting purely recipe-based prediction. The proposed method includes a forward gain predictor with an uncertainty signal and an inverse search for target gain, underpinned by a multi-level data-quality assessment that links fabrication entities like batch, wafer, and die. AI
RANK_REASON Research paper published on arXiv detailing a novel methodology for virtual metrology. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
