AI Model Transforms Optical Metrology Data for Semiconductor Step Height
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current metrology systems face challenges in accurately measuring step heights of semiconductor chips without optical interference, which can lead to difficulties in detecting vulnerabilities in the chips, and require additional processes like metal deposition, increasing manufacturing costs and time.
Innovation Solution
The use of an artificial intelligence model to predict step height data for semiconductor chips with a metal deposit, by processing first measurement data and layout density data, effectively removes optical interference without the need for destructive analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If optical metrology method (PWG) is used to measure step heights, then measurement speed and coverage area are improved, but optical interference from lower layers causes detection accuracy to deteriorate
Solution Approach 1:
The patent introduces an artificial intelligence model as an intermediary that processes optical metrology data to remove interference effects. The AI model learns the relationship between optical interference patterns and actual step heights, enabling accurate measurement without physical metal deposition while maintaining fast measurement speed.
Solution Approach 2:
The patent replaces the mechanical/physical process of metal deposition with an information-processing approach using AI algorithms. Instead of physically modifying the wafer to eliminate optical interference, the system uses computational methods to subtract interference effects from the optical measurement data.
2Measurement precision
If metal deposition is performed to remove optical interference, then detection accuracy is improved, but manufacturing cost and process time increase
Solution Approach 1:
The patent creates a virtual model of the step height measurement by training an AI model on datasets that include both optical measurement data and ground truth measurements. This virtual model replicates the accurate measurement capability of metal-deposited wafers without requiring actual metal deposition, thus avoiding additional manufacturing costs and process steps.
Solution Approach 2:
The patent performs preliminary training of the AI model using datasets collected beforehand. Once trained, the model can accurately process new optical measurement data without requiring any additional physical preparation or metal deposition, eliminating recurring manufacturing costs and process time for each measurement.
3Measurement precision
If metal deposition is performed to remove optical interference, then detection accuracy is improved, but process time and wafer availability are reduced
Solution Approach 1:
The patent substitutes the time-consuming metal deposition process with rapid AI-based data processing. The AI model can analyze optical measurement data and remove interference effects in computational time, which is significantly faster than the physical metal deposition process, thereby reducing process time and maintaining wafer availability.
Solution Approach 2:
The patent enables continuous optical measurement without interruption for metal deposition. The AI model processes measurement data in real-time or near-real-time, allowing the measurement process to continue without the breaks required by physical metal deposition processes, thus improving productivity and reducing total process time.
Data Source
AI summary
The present disclosure relates to devices, methods, and systems for transforming measurement data. An example device for transforming measurement data includes a communicator configured to receive first measurement data, the first measurement data including step height values on a semiconductor chip with a chemical mechanical polishing (CMP) process performed thereon, and to receive layout data comprising a layout included in the semiconductor chip, and a processor configured to, based on the layout data, transform the first measurement data to second measurement data, the second measurement data including step height values of the semiconductor chip with a metal deposited thereon.