Virtual Metrology for Semiconductor Wafer Quality Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current silicon wafer manufacturing processes face inefficiencies and inaccuracies in determining wafer quality due to the need for post-manufacturing inspection of all wafers, which is costly and time-consuming, and relying on representative samples can lead to inaccurate quality assessments.
Innovation Solution
Implementing a method that gathers data from manufacturing tools during the process to predict wafer quality using statistical and stochastic analyses, providing predictive information to the manufacturing system for timely corrective actions, such as recalibrating tools or redoing processes, to ensure parameter compliance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If post-manufacturing inspection of all wafers is performed, then quality determination accuracy is improved, but manufacturing time and cost increase
Solution Approach 1:
The system performs preliminary quality assessment during the manufacturing process by collecting and analyzing data from manufacturing tools in real-time, rather than waiting until post-manufacturing inspection. This allows quality predictions to be made before the wafer completes manufacturing, enabling early detection of potential defects and reducing the need for time-consuming post-inspection of all wafers.
Solution Approach 2:
The system creates a virtual model or copy of the wafer manufacturing process by collecting data from sensors and manufacturing tools, then uses this virtual representation to predict quality outcomes. This virtual metrology approach allows quality assessment without requiring physical inspection of every wafer, thereby reducing manufacturing time while maintaining accuracy.
2Measurement precision
If post-manufacturing inspection of all wafers is performed, then quality determination accuracy is improved, but manufacturing cost increases
Solution Approach 1:
The system uses virtual metrology to create a digital twin of the manufacturing process, collecting data from sensors and tools to simulate and predict quality outcomes. This virtual copy allows accurate quality determination without the need for expensive physical inspection equipment and operations on every single wafer, thereby reducing manufacturing costs while maintaining accuracy.
Solution Approach 2:
The manufacturing tools themselves provide quality data through integrated sensors and monitoring systems during the manufacturing process. The system leverages existing tool data and self-monitoring capabilities rather than requiring separate, expensive inspection equipment, enabling cost-effective quality assessment.
3Loss of time
If representative sample inspection is used, then manufacturing time is reduced, but quality determination accuracy deteriorates
Solution Approach 1:
The system implements continuous feedback by collecting data from multiple manufacturing tools throughout the manufacturing process and using statistical analysis to predict quality outcomes. This real-time feedback mechanism allows the system to assess quality based on process data rather than relying on limited post-manufacturing samples, improving accuracy without requiring inspection of all wafers.
Solution Approach 2:
By performing quality predictions during manufacturing based on process data, the system obtains preliminary quality information that is more representative of the actual wafer quality than post-manufacturing samples. This preliminary assessment uses cumulative process data to make accurate predictions about the final wafer quality.
4Productivity
If real-time predictive modeling is implemented, then process efficiency is improved, but system complexity increases
Solution Approach 1:
The system uses a unified predictive modeling platform that can analyze data from multiple different manufacturing tools and process types (lithography, etching, deposition, planarization) using the same statistical and machine learning frameworks. This multi-functional approach improves process efficiency across the entire manufacturing workflow while avoiding the need for separate complex systems for each tool type.
Data Source
AI summary
Implementations described herein generally relate to improving silicon wafer manufacturing. In one implementation, a method includes receiving data from one or more manufacturing tools about a manufacturing process of a silicon wafer. The method further includes determining, based on the data, predictive information about a quality of the silicon wafer. The method further includes providing the predictive information to a manufacturing system, wherein the predictive information is used to determine whether to take corrective action.