Automated Abnormal Machine Tracking System for Semiconductor Yield
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor and LCD manufacturing, conventional quality management methods often overlook minor defects from potentially abnormal machines, leading to delayed detection and increased losses due to variations in judgment standards among users across shifts, resulting in yield and cost losses.
Innovation Solution
An automated abnormal machine tracking and notifying system that collects and analyzes test results, stores abnormal information in a database, counts integral values of potentially abnormal machines, and automatically notifies users when the sum exceeds a predefined threshold, thereby identifying and reporting abnormal machines promptly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual quality management methods are used with sequential analysis by engineers, then detailed analysis capability is maintained, but detection speed and consistency deteriorate due to human judgment variations across shifts
Solution Approach 1:
The patent replaces the manual mechanical analysis process with an automated computer-based system that collects test results, generates SPC charts, and identifies abnormal machines automatically. This substitution eliminates human judgment variations and accelerates detection while maintaining analytical precision through systematic algorithms.
Solution Approach 2:
The system enables self-service by automatically performing data collection, analysis, and abnormal machine identification without requiring sequential engineer intervention. The automated generation of SPC charts and abnormal machine lists allows the system to serve its own quality management function continuously and consistently.
2Loss of information
If manual tracking of abnormal information is performed, then detailed analysis is possible, but information completeness and consistency worsen due to omissions and different judgment standards across shifts
Solution Approach 1:
The patent creates a universal tracking system that handles multiple types of abnormal information from different machines and processes through a single integrated platform. The system universally applies the same analysis methods and judgment standards across all shifts and machines, ensuring information completeness without requiring separate tracking mechanisms for each case.
Solution Approach 2:
The system introduces an intermediary automated analysis layer between raw test results and final quality decisions. This intermediary automatically collects, standardizes, and tracks abnormal information, serving as a consistent mediator that prevents information loss and ensures completeness across different shifts and processes.
3Productivity
If conventional quality management processes are used, then detailed manual analysis is maintained, but productivity and response speed deteriorate due to sequential processing
Solution Approach 1:
The patent replaces sequential manual quality management processes with an automated computer-based system that simultaneously collects data, generates SPC charts, and identifies abnormal machines. This substitution dramatically improves productivity by eliminating sequential processing bottlenecks while maintaining detailed analysis capabilities through systematic algorithms.
Data Source
AI summary
An automated abnormal machine tracking and notifying method includes automatically collecting and analyzing a plurality of test results to obtain a plurality of abnormal information, storing the abnormal information in a machine status database, respectively counting an integral value of each abnormal information of at least a related potentially abnormal machine, judging whether a sum of the integral values of each potentially abnormal machine is greater than n, and automatically notifying at least a user of the potentially abnormal machine and the abnormal information by an automatic notification subsystem when the sum of the integral values is greater than n.