Film Deposition Control Learning for Target Film Characteristics

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional methods for obtaining optimal control conditions for film deposition in molecular beam epitaxy apparatuses rely on trial and error, which is inefficient and requires skilled operators.

Innovation Solution

A learning processor that acquires and processes control condition data and film characteristic data to recommend optimal control conditions for film deposition, using machine learning to predict target film characteristics without the need for manual trial and error.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If trial and error method is used to obtain control conditions, then optimal control conditions can be found, but the process is inefficient and time-consuming

Engineering Contradiction:

Improvefilm characteristic accuracyVSAvoidtime for obtaining control conditions

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The system performs preliminary learning processing to build a model that predicts film characteristics from control conditions before actual film deposition. This preliminary action stores optimal control conditions and their expected outcomes, eliminating the need for time-consuming trial and error during production.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system creates a virtual model (copy) of the film deposition process through machine learning. This model replicates the relationship between control conditions and film characteristics, allowing predictions to be made without physically performing trial depositions, thus saving time while maintaining accuracy.

Inventive Principle:

Principle #26Copying

2Manufacturing precision

If trial and error method is used, then control conditions can be optimized, but skilled operators are required

Engineering Contradiction:

Improvefilm characteristic accuracyVSAvoidoperator skill requirement

Core Design Contradiction:

Manufacturing precisionVSEase of operation

Solution Approach 1:

The system performs self-learning by automatically acquiring control condition data and film characteristic data, then autonomously builds and updates the prediction model. This self-service capability eliminates the need for skilled operators to manually optimize conditions, as the system learns and optimizes automatically.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical trial-and-error process performed by skilled operators with an automated information processing system. The machine learning model substitutes human expertise and manual adjustment with algorithmic prediction, making the system easier to operate while maintaining or improving precision.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Productivity

If automated learning processing is implemented, then efficiency is improved, but system complexity increases

Engineering Contradiction:

Improvefilm deposition efficiencyVSAvoidlearning processing system complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The learning processor is designed to handle multiple functions: acquiring control condition data, acquiring film characteristic data, performing learning processing, and outputting recommended control conditions. This multi-functionality consolidates what could be separate complex systems into a single integrated unit, improving efficiency while managing complexity.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent introduces a learning processor as an intermediary between the film deposition apparatus and the control system. This intermediary component manages the complexity of machine learning operations, providing a simplified interface that improves productivity without requiring the entire system to become equally complex.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS11347203B2Learning processor, learning processing method, production method of compound semiconductor, and recording medium

Publication Date: 2022.05.31 ASAHI KASEI MICRODEVICES CORP

AI summary

Obtaining control conditions through trial and error is inefficient. Provided is a learning processor including a control condition acquiring section that acquires control condition data indicating a control condition of a film deposition apparatus; a film characteristic acquiring section that acquires film characteristic data indicating a characteristic of a film deposited by the film deposition apparatus that has been caused to operate according to the control condition indicated by the control condition data; and a learning processing section that performs learning processing of a model that outputs recommended control condition data indicating the control condition of the film deposition apparatus that is recommended in response to input of target film characteristic data indicating a target film characteristic, using learning data that includes the acquired control condition data and film characteristic data.