Wafer Defect Detection Using Reference-Based Reconstruction Error
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current defect detection methods in semiconductor wafers face challenges in distinguishing real defects from nuisances, leading to high nuisance rates and limited usability due to the detection of irrelevant deviations, which complicates quality control and assurance processes.
Innovation Solution
A computer-implemented method for defect detection that utilizes a defect criterion based on characteristic elements derived from reference images, combined with a tolerance statistic learned from defect-free representations, to distinguish between defects and nuisances by minimizing reconstruction error and accounting for statistical variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current defect detection methods are used to inspect semiconductor wafers, then defects can be detected, but nuisance rates are high due to detection of irrelevant deviations
Solution Approach 1:
The method performs preliminary actions by deriving characteristic elements from reference images before actual defect detection. These characteristic elements serve as a pre-established basis for comparison, enabling the system to distinguish between real defects and nuisances more effectively during the inspection process.
Solution Approach 2:
The patent introduces characteristic elements as an intermediary between reference images and defect detection. These characteristic elements act as a mediator that captures essential features while filtering out irrelevant variations, thereby reducing nuisance rates while maintaining defect detection accuracy.
2Measurement precision
If defect detection methods detect all deviations, then comprehensive defect identification is achieved, but usability is limited due to high nuisance rates
Solution Approach 1:
The method applies local quality by focusing detection efforts on specific characteristic elements derived from reference images rather than treating all deviations equally. This selective approach maintains comprehensive defect identification while filtering out irrelevant nuisances, thereby improving usability.
Solution Approach 2:
The patent changes parameters by transforming reference images into characteristic elements with specific properties. This transformation modifies the detection parameters to emphasize relevant features while suppressing irrelevant variations, reducing nuisance rates while maintaining detection comprehensiveness.
3Measurement precision
If tolerance statistic is learned from defect-free representations, then distinction between defects and nuisances is improved, but computational complexity increases
Solution Approach 1:
The method extracts only the essential characteristic elements from defect-free reference images to build the tolerance statistic, rather than processing entire images. This extraction approach maintains high distinction accuracy while significantly reducing computational complexity by working with condensed feature representations.
Solution Approach 2:
The tolerance statistic is computed in advance from defect-free representations before actual defect detection. This preliminary computation separates the complex statistical learning phase from the routine inspection phase, allowing the system to maintain high accuracy while reducing real-time computational burden.
Data Source
AI summary
A computer implemented method for defect detection comprises obtaining an imaging dataset of a wafer, and verifying a defect criterion in a subset of the imaging dataset of the wafer. The defect criterion comprises an observation representation of the subset of the imaging dataset with respect to a number of characteristic elements derived from reference images of semiconductor structures. The observation representation and the characteristic elements define a reconstruction of minimal reconstruction error, and a tolerance statistic on defect-free representations of subsets of defect-free observed imaging datasets. Each of the defect-free representations and the characteristic elements define a reconstruction of minimal reconstruction error of a subset of the defect-free imaging datasets. The computer implemented method further comprises generating defect information.