Automated Wafer Pattern Selection Using Shape Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The current semiconductor manufacturing process for optical proximity correction (OPC) and resist model calibration is slow due to manual sub-processes, often taking weeks to complete, and is hindered by the lack of unique pattern description using Image Parameters.
Innovation Solution
A method involving a processor to select and cluster shapes for mask construction using shape vectors, automating the selection process by forming clusters based on similarity criteria, thereby reducing the time required for pattern selection and calibration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual pattern selection process is used for OPC model calibration, then pattern selection can be performed with human judgment, but the process takes weeks to complete and is slow
Solution Approach 1:
The patent replaces the manual mechanical selection process with an automated computer-based system that uses image parameters and shape vectors to objectively evaluate and select patterns. The system automatically calculates image parameters for each pattern, compares them against specifications, and ranks patterns without human intervention, thereby eliminating the time-consuming manual review process while maintaining selection quality through systematic evaluation criteria.
Solution Approach 2:
The patent transforms the pattern selection process by changing from subjective human judgment to objective parameter-based evaluation. It introduces specific measurable parameters (image parameters such as area, perimeter, aspect ratio, and shape vectors) that quantify pattern characteristics. This parameter transformation enables automated comparison and selection based on numerical criteria rather than manual inspection, dramatically reducing selection time while preserving accuracy through consistent metric application.
2Reliability
If comprehensive pattern selection criteria are applied to ensure wafer image quality, then model calibration accuracy is improved, but the selection process becomes more complex and time-consuming
Solution Approach 1:
The patent segments the comprehensive pattern selection process into distinct automated stages: (1) image parameter calculation for each pattern, (2) specification compliance checking, (3) shape vector generation, (4) pattern ranking and selection. This segmentation breaks down the complex multi-criteria evaluation into manageable computational steps that can be executed automatically, reducing perceived complexity while maintaining thoroughness through systematic progression through each evaluation stage.
Solution Approach 2:
The patent introduces shape vectors as an intermediary representation that simplifies the comparison of complex patterns. Instead of directly comparing entire patterns against multiple criteria, the system transforms patterns into shape vectors that capture essential geometric characteristics. This intermediary representation enables efficient automated comparison and ranking while preserving the information needed for reliable model calibration, thereby reducing process complexity without sacrificing selection quality.
Data Source
AI summary
A method of constructing a mask for use in semiconductor device manufacturing is disclosed. A first shape that is related to mask construction is selected from a set of shapes. A second shape related to the mask construction is selected from the set of shapes. The first shape and the second shape are represented using a first shape vector and a second shape vector, respectively. A cluster is formed that includes the first shape and the second shape when the first shape vector and the second shape vector are within a selected criterion.