plasma-etching

AI-driven digital twin for plasma etching (physics model + sensor emulation + ML parameter recommendation)

Bayesian optimization pipeline for silicon reactive ion etching. Random Forest surrogate model seeded with transfer learning from published plasma physics data, with a hardware safety envelope that keeps every proposed recipe inside cleanroom mask-integrity limits. Built at the UH Nanofabrication Facility.

🏭 플라즈마 식각 공정 디지털 트윈 & 가상 계측 | PHM 2018 Ion Mill Etch | LightGBM RUL R²=1.0000 | XAI | Streamlit Dashboard

BOSCH 플라즈마 식각 가상계측 — 드리프트 인과 규명, 불확실성 기반 판정기(실계측 76.1% 생략, 미검출 0/67), ARM64 실장(PC 대비 1e-5 um), OES 4-lot 독립 검증