Fast Marching Level Sets for Semiconductor Metrology

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Automated metrology of electron microscopy images is challenging due to poor contrast and low signal-to-noise ratio, requiring manual intervention and limiting the accuracy of critical dimension measurements in 3D-NAND memory fabrication.

Innovation Solution

The use of fast marching level sets, optionally assisted by deep-learning based image preprocessing, to automatically delineate boundaries within images and perform metrology on regions of interest, enabling efficient measurement of shape features and material interfaces without human assistance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual control by skilled technicians is used to process electron microscopy images, then measurement accuracy can be maintained, but productivity is reduced and automation is limited

Engineering Contradiction:

Improvemeasurement accuracyVSAvoidautomation efficiency

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The system enables automated self-service processing of electron microscopy images through the fast marching level set algorithm. The algorithm automatically propagates level set curves to locate boundaries and performs metrology measurements without requiring manual intervention by skilled technicians, thereby maintaining measurement accuracy while significantly improving productivity and automation efficiency.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical/manual system of technician-controlled image processing with an automated computational system. The fast marching level set algorithm substitutes human operators by automatically propagating curves through image data, detecting boundaries, and performing measurements, thus eliminating the trade-off between accuracy and automation.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Reliability

If traditional image processing techniques are used to improve image quality, then contrast and signal-to-noise ratio may be enhanced, but the process requires manual control and reduces automation

Engineering Contradiction:

Improveimage qualityVSAvoidmanual intervention requirement

Core Design Contradiction:

ReliabilityVSExtent of automation

Solution Approach 1:

The fast marching level set algorithm performs self-service image processing by automatically propagating level set curves through the image data. The algorithm inherently handles contrast enhancement and boundary detection without requiring manual control, thereby improving image quality reliability while maintaining full automation and eliminating manual intervention requirements.

Inventive Principle:

Principle #25Self-service

3Productivity

If fast marching level sets are used to automate boundary detection, then productivity and automation are improved, but challenges remain in low-contrast images

Engineering Contradiction:

Improveautomation efficiencyVSAvoidboundary detection difficulty

Core Design Contradiction:

ProductivityVSDifficulty of detecting and measuring

Solution Approach 1:

The system performs preliminary action by propagating the fast marching level set curve through the entire image domain before final boundary detection. This preliminary propagation prepares the level set function across all regions, enabling the algorithm to automatically adapt to and detect boundaries even in low-contrast areas, thereby maintaining high automation efficiency while overcoming boundary detection difficulties in challenging image conditions.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11947270B2Metrology of semiconductor devices in electron micrographs using fast marching level sets

Publication Date: 2024.04.02 FEI CO

AI summary

Apparatuses and methods for metrology on devices using fast marching level sets are disclosed herein. An example method at least includes initiating a fast marching level set seed on an image, propagating a fast marching level set curve from the fast marching level set seed to locate boundaries of a plurality of regions of interest within the image, and performing metrology on the regions of interest based in part on the boundaries.