Etching Recipe Search With Shape Highlighting for ML Evaluation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The existing methods for determining etching recipes for semiconductor devices using machine learning face challenges in efficiently narrowing down candidate recipes, assessing their sensitivity, and avoiding non-settable process conditions, particularly due to the complexity of the machine learning models and the need for interactive user input.

Innovation Solution

A process recipe search apparatus that utilizes a machine learning model to predict process shapes and includes a target shape decision unit, a machine learning model creation unit, a recipe search unit, a process recipe decision unit, and a display shape highlight processing unit to facilitate the evaluation of predicted process shapes, choose suitable recipes, and highlight differences between predicted and target shapes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If machine learning model is used to predict process shapes, then productivity is improved, but device complexity increases

Engineering Contradiction:

Improverecipe search efficiencyVSAvoidmachine learning model complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent introduces a visualization intermediary that translates complex machine learning prediction results into intuitive visual representations. The visualization unit converts numerical prediction data into graphical displays showing predicted process shapes, allowing users to understand complex model outputs without directly interpreting raw data, thus resolving the contradiction between using complex ML models and maintaining operational simplicity

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces traditional manual recipe search methods with an automated machine learning-based prediction system. Instead of manually adjusting parameters and observing results, the system uses ML models to predict process outcomes, substituting mechanical/manual operations with computational intelligence to improve productivity while managing complexity through automated workflows

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If high resolution visualization is used to evaluate predicted shapes, then measurement precision is improved, but ease of operation deteriorates

Engineering Contradiction:

Improveshape evaluation accuracyVSAvoidrecipe selection difficulty

Core Design Contradiction:

Measurement precisionVSEase of operation

Solution Approach 1:

The patent segments the evaluation process into multiple visual dimensions and metrics. Instead of presenting a single complex high-resolution image, the system divides the evaluation into separate visual elements such as shape deviation maps, dimensional comparisons, and key parameter displays, allowing users to assess precision through multiple focused views rather than overwhelming detailed data

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent employs color-coded visualization to represent different aspects of shape accuracy. Predicted shapes are displayed with color indicators showing deviation from target shapes, where different colors represent different levels of accuracy or types of deviations. This visual encoding allows users to quickly grasp precision information without analyzing raw numerical data, maintaining ease of operation while achieving high measurement precision

Inventive Principle:

Principle #32Color changes

3Manufacturing precision

If multiple candidate recipes are evaluated, then manufacturing precision is improved, but loss of time increases

Engineering Contradiction:

Improveetching shape accuracyVSAvoidrecipe evaluation time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary evaluation of multiple candidate recipes using the machine learning model before actual manufacturing. The system predicts process shapes for multiple candidate recipes in advance, ranks them based on predicted accuracy, and presents the top candidates to users. This preliminary screening allows comprehensive evaluation of multiple options without requiring full detailed analysis of each candidate, thus improving manufacturing precision selection while reducing time loss

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent implements a filtered evaluation approach where obviously inferior candidate recipes are quickly identified and skipped. The visualization system highlights promising candidates with distinctive markers, allowing users to rapidly scan and focus only on viable options. This skipping mechanism enables efficient traversal through multiple candidates, evaluating enough to ensure precision without spending excessive time on each individual candidate

Inventive Principle:

Principle #21Skipping (Rushing through)

Data Source

PatentUS12222690B2Process recipe search apparatus, etching recipe search method and semiconductor device manufacturing system

Publication Date: 2025.02.11 HITACHI HIGH TECH CORP

AI summary

To facilitate evaluation of a predicted process shape in process recipe development using machine learning, a process recipe search apparatus that searches for an etching recipe that is a parameter of a plasma processing apparatus set so as to etch a sample into a desired shape displays, on a display device, the predicted process shape of the sample by a candidate etching recipe predicted by using a machine leaning model, by highlighting a difference between the predicted process shape and a target shape.