Bosch etching simulations improve with few-shot real measurement calibration
Real-to-Sim Calibration of Bosch Etching Simulators with Few-Shot XSEM Data and Hybrid Search
Computational Engineering, Finance, and ScienceMachine Learning
Summary
Predicting how tiny etched features will look after manufacturing is hard because real data to tune simulations is limited and expensive. The authors created a method that uses just a few real measurements from a special microscope to adjust a simulation so it matches reality better. Their approach uses smart searching methods, including AI help, to find the best simulation settings for each manufacturing recipe. This leads to more reliable predictions of etched depth and angle, even for conditions not directly measured.
What this means in practice
- •For microfabrication engineers: Predict etched structure profiles accurately using few real measurements to optimize Bosch etching recipes without extensive trial runs.
- •For semiconductor foundries: Improve process control by integrating calibrated simulations for rapid prediction of MEMS and TSV etch outcomes under varied manufacturing settings.
Authors
Xinyu Jia, Jainil Shah, Rahul Tripathi, Vijay Gupta, Zhihong Chen
Abstract
The Bosch process is widely used in MEMS and TSV fabrication, but accurate prediction of etching outcomes remains difficult without calibration against real measurements, which are sparse and costly to obtain. This work presents a few-shot real-to-sim calibration framework that decouples simulator parameters from direct process variables to infer a recipe-specific simulator parameter vector from sparse measurements, enabling reliable prediction of etch depth and sidewall angle under held-out process conditions. We first formulate Bosch profile prediction as a few-shot real-to-sim calibration problem, where a small set of real XSEM measurements is used to identify a recipe-specific ViennaPS parameter vector $z_r^*$ that minimizes the discrepancy between simulated and measured etch depth and sidewall angle. To search this parameter space efficiently, the framework combines local exploration, LLM-guided candidate generation, and XGBoost-based surrogate screening, while retaining ViennaPS simulation as the final verification step for each selected candidate. Experiments on two Bosch etching recipes show that the proposed framework accurately predicts held-out etching results, achieving normalized global errors of 0.340 and 0.249 for the 2 recipes respectively.