Automated Electron Beam Image Classification for Wafer Defect Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current defect classification methods in semiconductor manufacturing are labor-intensive, time-consuming, and prone to human error, making it difficult to accurately classify subtle defects and requiring extensive manual processing of SEM images.

Innovation Solution

A computer-implemented method and system that acquire and classify electron beam images of defects detected on wafers without user input, using image processing techniques to automatically determine defect classification, thereby reducing manual effort and improving accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual classification of SEM images is used, then user judgment and flexibility are applied, but the process becomes labor intensive and time consuming

Engineering Contradiction:

Improveclassification accuracyVSAvoidclassification speed

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent replaces the manual mechanical process of visual inspection and classification with an automated image processing system using algorithms and computer vision techniques. The system automatically analyzes SEM images, extracts defect features, and classifies defects without human intervention, thereby maintaining accuracy while dramatically improving classification speed and productivity.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The classification system performs self-service by autonomously processing images and making classification decisions without requiring user input or manual review. The algorithm independently identifies defects, extracts relevant features, and assigns classifications based on predefined criteria, enabling the system to serve itself and eliminate dependency on manual labor.

Inventive Principle:

Principle #25Self-service

2Measurement precision

If manual classification is performed, then user expertise can be applied, but accuracy and purity become user-dependent and subtle defects are easily overlooked

Engineering Contradiction:

Improvedefect detection accuracyVSAvoiddependency on user expertise

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces human visual inspection with an automated image processing system that consistently applies detection algorithms to identify defects. This substitution eliminates variability in human performance, ensures uniform application of detection criteria, and enables the system to detect subtle defects that may escape human observation, thereby improving measurement precision while reducing dependency on user expertise.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enhances defect detection capability by adjusting and optimizing parameters such as image processing thresholds, feature extraction parameters, and classification criteria. These parameter changes enable the automated system to detect subtle defects with high precision while maintaining consistent performance independent of user expertise levels.

Inventive Principle:

Principle #35Parameter changes

3Reliability

If extensive manual processing of SEM images is performed, then thorough inspection is achieved, but the process becomes substantially labor intensive

Engineering Contradiction:

Improveinspection thoroughnessVSAvoidclassification time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent replaces manual image processing with automated computational processing that can analyze multiple images simultaneously and rapidly. The system performs thorough inspection by systematically processing each image through multiple analysis stages, maintaining inspection quality while reducing the time required from hours or days to minutes or seconds.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system performs preliminary actions by pre-processing images, pre-defining classification criteria, and pre-establishing detection algorithms before actual classification begins. This preparation enables rapid subsequent processing while ensuring thorough inspection, as the system is already configured to efficiently analyze all relevant features without requiring manual setup for each image.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS9293298B2Defect discovery and inspection sensitivity optimization using automated classification of corresponding electron beam images

Publication Date: 2016.03.22 KLA CORP

AI summary

Various embodiments for classifying defects detected on a wafer are provided. One method includes acquiring an electron beam image generated by a defect review tool for a location of a defect detected on a wafer by a wafer inspection tool. The method also includes determining a classification of the defect based on at least the electron beam image and without input from a user. The method may also include feeding back the classification results to the wafer inspection tool and optimizing the parameters of the tool to maximize sensitivity to the defects of interest.