Related Experiment Video
Updated: Sep 9, 2025
05:57
Characterization of SiN Integrated Optical Phased Arrays on a Wafer-Scale Test Station
Published on: April 1, 2020
8.1K
AI-Powered Next-Generation Technology for Semiconductor Optical Metrology: A Review
Weiwang Xu1, Houdao Zhang1, Lingjing Ji1
1Shanghai Precision Measurement Semiconductor Technology, Inc., Shanghai 210700, China.
Micromachines
|August 28, 2025
Summary
Artificial intelligence (AI) combined with optical spectroscopy offers breakthroughs for angstrom-scale semiconductor manufacturing metrology. Challenges remain in data quality and model generalization for intelligent metrology advancement.
Area of Science:
- Semiconductor Manufacturing
- Metrology
- Optical Spectroscopy
- Artificial Intelligence
Background:
- Conventional metrology faces limitations in accuracy, speed, and non-destructiveness for angstrom-scale semiconductor manufacturing and 3D integration.
- Optical spectroscopy shows promise but encounters technical barriers in complex manufacturing environments.
- The need for advanced, intelligent metrology solutions is critical for next-generation semiconductor fabrication.
Purpose of the Study:
- To categorize AI-optical spectroscopy integration paradigms: forward surrogate modeling, inverse prediction, physics-informed neural networks (PINNs), and multi-level architectures.
- To benchmark AI-optical spectroscopy efficacy against industrial metrology challenges like tool-to-tool (T2T) matching and high-aspect-ratio (HAR) structure characterization.
- To identify unresolved bottlenecks for future intelligent semiconductor metrology development.
Main Methods:
- Categorization of AI algorithms (surrogate models, inverse modeling, PINNs, multi-level networks) applied to optical spectroscopy.
- Methodical assessment of implementation efficacy and limitations for each AI-optical spectroscopy pathway.
- Application case studies using J-profiler software 5.0 to validate AI technologies in industrial metrology scenarios.
Main Results:
- AI-optical spectroscopy fusion demonstrates significant efficacy in addressing critical industrial challenges, including tool-to-tool (T2T) matching.
- AI technologies provide technological breakthroughs for semiconductor metrology in the angstrom-scale era.
- Key challenges persist, including data veracity, insufficient datasets, and cross-scale compatibility issues.
Conclusions:
- The integration of AI and optical spectroscopy is pivotal for advancing semiconductor metrology.
- Future research must focus on enhancing model generalization, optimizing data strategies, and balancing real-time performance with accuracy.
- Addressing current bottlenecks will catalyze the transformation towards intelligence-driven advanced semiconductor manufacturing.
