Virtual Metrology & Fault Detection for Semiconductor Manufacturing
Online virtual metrology with uncertainty evaluation and semi-automated fault-detection configuration for CMP and related processes, applied with Applied Materials
Project Overview
Physical wafer metrology is expensive and delayed; virtual metrology enables full online inspection of every wafer. The lab built VM models with prediction uncertainty and semi-automated FD configuration for CMP and related processes.
Research Objectives
High-accuracy online prediction of CMP material removal rate
VM predictions with uncertainty evaluation
Semi-automated fault-detection configuration (automated feature extraction and limits setting)
Methodology
Gaussian-process-based virtual metrology with online sample selection and reference mechanisms to track slow process drift; automated feature extraction and limit setting on the FD side.
Key Results
Published in IEEE TSM, Computers in Industry, and Measurement
Joint research with Applied Materials applied to advanced process scenarios
Related Publications
An Online Virtual Metrology Model with Sample Selection..., IEEE Transactions on Semiconductor Manufacturing, 2019
A virtual metrology method with prediction uncertainty..., Computers in Industry, 2020
Intelligent Monitoring of Manufacturing Processes