Defect Detection Parameter Optimization for Semiconductor Wafer Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current semiconductor wafer inspection systems require manual and iterative tuning of sensitivity parameters, which is time-consuming and prone to inconsistencies due to human factors, especially when dealing with complex multi-channel and multi-pass inspections, and existing automated solutions like AutoSAT are limited in optimizing sensitivity recipes across different defect detection algorithms.
Innovation Solution
A computer-implemented method for selecting defect detection parameters using an optimization function that acquires information on classified defects and objectives, allowing for simultaneous sensitivity tuning and optics selection, thereby reducing the need for manual intervention and improving recipe consistency and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual iterative tuning of sensitivity parameters is performed, then inspection sensitivity can be optimized, but the process becomes time-consuming and prone to inconsistencies
Solution Approach 1:
The system performs automatic sensitivity parameter optimization without requiring manual iterative tuning. The defect detection algorithm autonomously adjusts sensitivity parameters based on defect classification results, eliminating the need for operator intervention in the tuning process while maintaining optimized inspection sensitivity.
Solution Approach 2:
The system automatically changes sensitivity parameters based on defect classification outcomes. By monitoring detection results and adjusting parameters dynamically, the system achieves optimized inspection sensitivity without manual intervention, resolving the contradiction between achieving precise parameter tuning and reducing setup time.
2Manufacturing precision
If manual iterative tuning is performed for multi-channel and multi-pass inspections, then detection quality can be improved, but the complexity and time required increase significantly
Solution Approach 1:
The system merges the parameter tuning process across multiple channels and passes into a unified automatic optimization routine. Instead of requiring separate manual tuning for each channel and pass, the system coordinates all channels and passes simultaneously, reducing the complexity of the tuning process while maintaining high detection quality through coordinated parameter adjustment.
Solution Approach 2:
The automatic optimization system provides a universal solution that handles multi-channel and multi-pass inspections through a single integrated process. The defect detection algorithm applies the same optimization logic across different channels and passes, eliminating the need for channel-by-channel manual tuning while maintaining consistent detection quality across all inspection paths.
3Measurement precision
If manual parameter tuning is performed, then sensitivity parameters can be adjusted, but the process is highly dependent on user expertise and experience
Solution Approach 1:
The system performs sensitivity parameter optimization autonomously without requiring user expertise in parameter tuning. The defect detection algorithm automatically analyzes defect classifications and adjusts sensitivity parameters accordingly, making the process independent of operator knowledge while achieving accurate parameter settings through algorithmic optimization.
4Measurement precision
If exhaustive searching is performed in AutoSAT for parameter optimization, then local segment thresholds can be optimized, but the full parameter set cannot be optimized simultaneously and recipe stability is compromised
Solution Approach 1:
The system performs simultaneous optimization of the full parameter set by dynamically changing parameters based on defect classification results. Instead of exhaustive searching that optimizes only local segments, the system adjusts sensitivity parameters across the entire parameter space coherently, achieving both threshold optimization and recipe stability through coordinated parameter changes that consider global optimization objectives.
Data Source
AI summary
Computer-implemented methods, computer-readable media, and systems for selecting one or more parameters for a defect detection method are provided. One method includes selecting one or more parameters of a defect detection method using an optimization function and information for a set of classified defects, which includes defects of interest and nuisance defects, such that the one or more parameters satisfy an objective for the defect detection method.