Reticle Inspection Baseline Filtering for Defect Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in semiconductor manufacturing lies in efficiently inspecting single-die reticles for defects, particularly after they have been used, as existing techniques struggle with coverage and false detections due to variable OPC decorations and unintentional artifacts that do not affect wafer yield.
Innovation Solution
A method involving a reticle inspection tool that generates baseline events from a known-good reticle, discarding false defects and focusing on reviewable defects by comparing current unusual events to baseline events, with the option to synthesize references for incomplete patterns, and using a less stringent threshold for candidate defect identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional reticle inspection techniques are used, then inspection coverage is improved, but false detections increase due to variable OPC decorations and unintentional artifacts
Solution Approach 1:
The patent performs a preliminary baseline inspection on a known-good reticle to establish expected event characteristics before actual defect detection. This preliminary action creates a reference framework that enables subsequent inspections to distinguish between true defects and false detections caused by OPC decorations or artifacts, thereby improving reliability while maintaining coverage
Solution Approach 2:
The patent implements feedback by comparing current inspection events against the established baseline events from the known-good reticle. This feedback mechanism allows the system to identify and filter out false detections (events that match baseline patterns) while retaining true defects, thus improving measurement precision without sacrificing reliability
2Measurement precision
If all unusual events are inspected thoroughly, then defect detection accuracy is improved, but processing time and data volume increase
Solution Approach 1:
The patent extracts and separates false detection events from the full set of unusual events by comparing them against baseline data. Events that match baseline characteristics (false detections) are extracted and removed from further processing, while only events that deviate from baseline (potential true defects) are retained for detailed analysis, thereby reducing processing time while maintaining accuracy
Solution Approach 2:
The patent applies partial action by performing thorough inspection only on events that are unusual compared to the baseline, rather than inspecting all events with equal intensity. This selective approach focuses computational resources on the most promising candidates, improving detection accuracy for true defects while reducing overall processing time through the Pareto principle
3Reliability
If baseline events are generated from actual reticle images, then inspection reliability is improved, but data volume and storage requirements increase
Solution Approach 1:
The patent extracts only the essential characteristics of baseline events (location, size, shape features) rather than storing complete reticle images. This extraction reduces data volume significantly while maintaining the reliability needed for accurate defect detection, as the extracted features are sufficient for comparison without requiring full image storage
Solution Approach 2:
The patent creates a simplified copy or representation of baseline event characteristics rather than storing the original reticle images. This copying approach preserves the essential information needed for reliable inspection while dramatically reducing data volume requirements, as only the salient features are stored rather than complete image data
Data Source
AI summary
A reticle that is within specifications is inspected to generate baseline candidate defects and their location and size. After using the reticle in photolithography, the reticle is inspected to generate current candidate defects and their location and size. An inspection report of filtered candidate defects and their images is generated so that these candidate defects include a first subset of the current candidate defects and their images and exclude a second subset of the current candidate defects and their images. Each of the first subset of candidate defects has a location and size that fails to match any baseline candidate defect's location and size, and each of the excluded second subset of candidate defects has a location and size that matches a baseline candidate defect's location and size.