Defect Categorization in Lithography Pattern Writing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The high volume of defects in semiconductor wafers makes it challenging to extract meaningful data on design-process interaction defects, as existing inspection tools focus on random defects and process monitoring, leading to time and cost prohibitive manual analysis and inefficient defect correlation.
Innovation Solution
A computer-implemented method and system for categorizing defects by determining common structural elements surrounding each defect, using defect clips to align and match structural elements across the wafer, and providing an indication of defects belonging to different structural elements, facilitating automated defect binning and analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If manual analysis and parsing of individual defects is performed to identify design-process interaction defects, then meaningful data on DPI defects can be extracted, but the process becomes time and cost prohibitive
Solution Approach 1:
The patent creates simplified representations (copies) of defect areas by generating images centered on each defect and comparing them to reference images. Instead of manually analyzing each defect, the system creates and compares copies of defect regions to automatically identify DPI defects, dramatically reducing analysis time while preserving meaningful information.
Solution Approach 2:
The patent introduces an intermediary automated image comparison process between the raw defect data and the final DPI defect identification. The system uses reference images and automated algorithms as intermediaries to bridge the gap between inspection tool output and meaningful DPI defect classification, eliminating the need for direct manual analysis.
2Reliability
If inspection tools generate defect maps identifying tens of thousands to hundreds of thousands of defects, then comprehensive defect detection is achieved, but extracting meaningful data becomes challenging
Solution Approach 1:
The patent segments the overwhelming defect data by creating individual images centered on each defect location. This segmentation allows the system to process defects in manageable units rather than attempting to analyze all defects simultaneously, reducing processing complexity while maintaining comprehensive detection coverage.
Solution Approach 2:
The patent extracts only the relevant information from each defect area by creating focused images centered on defect locations and comparing them to reference images. This extraction process filters out unnecessary data while preserving the critical information needed to identify DPI defects, simplifying the overall data processing task.
3Manufacturing precision
If technology nodes advance to produce smaller dimensions, then manufacturing capability improves, but the number of defects increases in inverse proportion to feature size
Solution Approach 1:
The patent changes the parameters of defect analysis by using automated image comparison algorithms that can efficiently process large numbers of defects. The system adjusts the analysis methodology to handle the increased defect quantities associated with smaller feature sizes, maintaining manufacturing precision while managing the higher defect counts through automated processing.
Data Source
AI summary
Methods and apparatus for categorizing defects on workpieces, such as semiconductor wafers and masks used in lithographically writing patterns into such wafers are provided. For some embodiments, by analyzing the layout in the neighborhood of the defect, and matching it to similar defected neighborhoods in different locations across the die, defects may be categorized by common structures in which they occur.