Neural Network Substrate Detection in Semiconductor Handling

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods in the semiconductor industry for determining the presence, position, and orientation of substrates, such as wafers and masks, are inefficient and prone to errors, particularly during the teach-in process, leading to potential damage, contamination, and reduced yield due to the reliance on manual, rule-based image processing which is inflexible and prone to errors from varying environmental conditions.

Innovation Solution

The use of an artificial neural network to analyze images of substrates and their locations within the semiconductor industry, generating information data records for controlling machine operations, allowing for flexible and accurate determination of substrate presence, position, and orientation, reducing the need for manual rule definition and improving handling system efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual rule-based image processing is used to determine substrate presence, position, and orientation, then the system can operate with simple hardware, but the reliability and accuracy deteriorate due to errors from varying environmental conditions and manual processing limitations

Engineering Contradiction:
Improvesubstrate detection reliabilityVSAvoidimage processing system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent replaces manual rule-based image processing with an artificial neural network system that automatically learns and adapts to various environmental conditions. The neural network processes images of substrates in cassettes, determining presence, position, and orientation without manual intervention, thereby improving reliability while managing complexity through automation.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent changes the processing approach from fixed rule-based parameters to adaptive neural network parameters that can dynamically adjust to varying environmental conditions. The neural network learns optimal parameter settings during training and applies them during operation, maintaining high reliability across different lighting, camera angles, and substrate configurations.

Inventive Principle:
Principle #35Parameter changes

2Adaptability or versatility

If rule-based image processing is used, then the system structure remains simple, but the adaptability deteriorates due to inability to handle varying environmental conditions and complex substrate configurations

Engineering Contradiction:
Improveenvironmental condition adaptabilityVSAvoidimage processing complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent replaces rigid rule-based processing with a flexible neural network system that adapts to varying environmental conditions including different lighting, camera positions, and substrate types. The neural network learns from training data and automatically adjusts its processing parameters, providing high adaptability without requiring complex manual reconfiguration.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Productivity

If manual teach-in processes are used for handling systems, then the system can be implemented with basic components, but the productivity deteriorates due to time-consuming procedures and potential errors

Engineering Contradiction:
Improvesubstrate processing productivityVSAvoidteach-in process time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent implements self-service through automated neural network-based substrate detection and handling system control. The system automatically determines substrate presence, position, and orientation, and uses this information to control pick-and-place operations without manual teach-in procedures, significantly reducing setup time and improving productivity.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces manual teach-in procedures with automated neural network processing that rapidly analyzes substrate images and generates control commands for handling systems. This substitution eliminates time-consuming manual operations while maintaining or improving accuracy, directly enhancing productivity.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

4Loss of information

If sensors are used to detect substrate presence and position, then real-time information can be obtained, but the device complexity increases due to multiple sensors and their setup requirements

Engineering Contradiction:
Improvesubstrate position information accuracyVSAvoidsensor system complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent replaces multiple physical sensors with an optical imaging system processed by a neural network. A single camera captures images of substrates in cassettes, and the neural network extracts presence, position, and orientation information from these images, reducing sensor complexity while maintaining or improving information accuracy.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP3907754B1Method for processing substrates, in particular wafers, masks or flat panel displays, with a semiconductor industry machine
Publication Date: 2023.04.26 INTEGRATED DYNAMICS ENG
  • EP3907754B1 patent drawingFigure 1~2
  • EP3907754B1 patent drawingFigure 3
  • EP3907754B1 patent drawingFigure 4~6

AI summary

The invention relates to a method for processing substrates, in particular wafers, masks, or flat panel displays, using a semiconductor machine, wherein a computer-aided method is employed to determine the presence and/or position and/or orientation of the substrate. The invention further relates to a system configured for carrying out the method. The computer-aided method comprises an artificial neural network.