Knowledge File Recommendation for Wafer Defect Classification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The existing defect review process for semiconductor wafers requires user intervention to select knowledge files for automatic defect classification, significantly reducing manufacturing throughput and potentially leading to inaccurate classifications due to reliance on user expertise.

Innovation Solution

A defect review system that includes a knowledge recommendation server which searches for and provides matching knowledge files based on inspection images, enabling automatic defect classification without user intervention, using a processor to receive requests, search for matching knowledge files, and transmit results to a defect classification server.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If user intervention is required to select knowledge files for defect classification, then classification accuracy can be maintained through user expertise, but manufacturing throughput is significantly reduced

Engineering Contradiction:

Improveclassification accuracyVSAvoidmanufacturing throughput

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The system enables automatic knowledge file selection without user intervention by using machine learning models to analyze inspection images and autonomously determine the most appropriate knowledge files for defect classification, allowing the system to serve itself rather than requiring human operators

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

A knowledge recommendation server acts as an intermediary between the inspection system and the defect classification process, receiving inspection images, analyzing them through multiple knowledge files, and automatically selecting the best matching knowledge files to improve classification accuracy without user involvement

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If multiple knowledge files are used for defect classification, then classification accuracy is improved, but processing time and system complexity increase

Engineering Contradiction:

Improveclassification accuracyVSAvoidprocessing time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary analysis by automatically selecting and loading only the most relevant knowledge files before defect classification begins, using machine learning models to pre-determine which knowledge files will be most useful based on the inspection image characteristics, thereby avoiding unnecessary processing time

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system dynamically adjusts the selection of knowledge files based on parameters extracted from inspection images, such as defect type, size, and characteristics, automatically configuring the classification process to use only the necessary knowledge files rather than processing all available files

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS11650576B2Knowledge recommendation for defect review

Publication Date: 2023.05.16 ASML NETHERLANDS BV

AI summary

A server for knowledge recommendation for defect review. The server includes a processor electronically coupled to an electronic storage device storing a plurality of knowledge files related to wafer defects. The processor is configured to execute a set of instruction to cause the server to: receive a request for knowledge recommendation for inspecting an inspection image from a defect classification server; search for a knowledge file in the electronic storage device that matches the inspection image; and transmit the search result to the defect classification server.