Descriptor-Guided Fast Marching for Microelectronic Defect Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional automated systems face challenges in rapidly and accurately detecting defects in microelectronic devices, particularly keyholes, due to noise and low contrast in SEM images with irregularly shaped features.

Innovation Solution

The descriptor-guided fast marching method preprocesses images to reduce noise and identify defects by converting image intensity into arrival time information, using a fast marching algorithm to detect boundaries and potential defects without prior knowledge, and outputs labeled images for visual or automated inspection.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional automated systems use pattern comparison or critical dimension measurements to detect defects, then defects can be identified, but detection accuracy significantly decreases when analyzing images with irregular shaped features, significant noise, and low contrast

Engineering Contradiction:

Improvedefect detection accuracyVSAvoidability to handle irregular shaped features with noise and low contrast

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent transforms the image analysis problem by changing the parameter representation from direct intensity comparison to arrival time transformation. The fast marching method computes arrival times based on intensity gradients, converting the detection task into a temporal domain problem where defects appear as distinct arrival time patterns, thereby improving accuracy for irregular features with noise and low contrast

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent introduces an intermediary transformation step between image capture and defect detection. The fast marching algorithm acts as a mediator that processes the raw image data through arrival time computation, creating an intermediate representation that enhances defect visibility while suppressing noise, thus bridging the gap between conventional methods and accurate defect identification

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If visual inspection of SEM images is performed to locate and identify keyholes, then defects can be detected, but the process is difficult and time-consuming due to significant noise and low contrast in the images

Engineering Contradiction:

Improvedefect detection capabilityVSAvoidinspection time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent replaces the manual visual inspection process with an automated computational system. The fast marching method algorithm automatically processes SEM images, computing arrival times and identifying defects without human intervention. This substitution eliminates the time-consuming nature of visual inspection while maintaining or improving detection reliability through consistent automated analysis

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent performs preliminary image processing through arrival time transformation before actual defect detection. By pre-processing the image data to emphasize structural features and suppress noise through the fast marching computation, the system prepares enhanced data that enables rapid and reliable automated defect identification, reducing both time and effort required for inspection

Inventive Principle:

Principle #10Preliminary action

3Extent of automation

If reference patterns or critical dimensions are used for defect identification, then automated detection can be performed, but reference patterns must be known beforehand and detection rates decrease for irregular shaped features

Engineering Contradiction:

Improveautomated defect detectionVSAvoidhandling of irregular shaped features without prior reference

Core Design Contradiction:

Extent of automationVSAdaptability or versatility

Solution Approach 1:

The patent enables the system to perform defect detection without external reference patterns by using the image data itself to drive the detection process. The fast marching method computes arrival times based on local intensity variations within the image, allowing the system to autonomously identify defects in irregular features without requiring pre-stored references, thereby achieving both automation and adaptability

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS11144778B2Descriptor guided fast marching method for analyzing images and systems using the same

Publication Date: 2021.10.12 MICRON TECHNOLOGY INC

AI summary

Methods and systems for descriptor guided fast marching method based image analysis and associated systems are disclosed. A representative image processing method includes processing an image of a microelectronic device using a fast marching algorithm to obtain arrival time information for the image. The arrival time information is analyzed using a targeted feature descriptor to identify targeted features. The detection of defects is facilitated by segmenting the image. The segmented image can be analyzed to identify targeted features which are then labeled for inspection.