Substrate Defect Inspection Using Zernike Polynomial Decomposition

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In semiconductor device manufacturing, existing defect inspection methods fail to effectively inspect substrates when a reference image is not pre-stored, leading to inadequate detection of defects in the head wafer and subsequent wafers, especially when the number of treatment objects is one, resulting in incomplete or inaccurate defect inspection.

Innovation Solution

A method utilizing the Zernike polynomial to decompose pixel value distributions in substrate images, calculating Zernike coefficients to detect defects, and determining defect presence based on these coefficients, allowing for appropriate inspection even when the head substrate is defective and ensuring inspection continuity throughout a job.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If the captured image of the head wafer is used as the reference image when no reference image is stored in advance, then the inspection process can be executed without producing a defect-free wafer, but the defect of the head wafer cannot be detected and the defect inspection of subsequent wafers cannot be appropriately performed

Engineering Contradiction:

Improveinspection process executionVSAvoiddefect detection accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system performs preliminary defect detection on the head wafer using Zernike polynomial decomposition before using any wafer images as reference. This preliminary action ensures that a defect-free reference image is established before subsequent inspections, resolving the contradiction by preparing the inspection system in advance with accurate reference data.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The Zernike polynomial decomposition acts as an intermediary method to detect defects on the head wafer before it can be used as a reference image. This intermediary detection mechanism ensures that only defect-free wafers become reference images, maintaining measurement precision while enabling productivity.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Ease of manufacture

If the captured image of the head wafer is used as the reference image, then the inspection can proceed without producing a defect-free wafer, but when the head wafer has a defect, the defect inspection of subsequent wafers cannot be appropriately performed

Engineering Contradiction:

Improvereference image generationVSAvoiddefect inspection reliability

Core Design Contradiction:

Ease of manufactureVSReliability

Solution Approach 1:

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements feedback by detecting defects on the head wafer and using this information to determine whether to proceed with using the head wafer image as a reference image. This feedback mechanism ensures that defective wafers do not compromise the reliability of subsequent inspections.

Inventive Principle:

Principle #23Feedback

3Productivity

If the number of wafers being treatment objects in the job is one, then the job can be executed quickly, but the defect inspection is not performed at all for the job

Engineering Contradiction:

Improvejob execution speedVSAvoiddefect inspection coverage

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system performs preliminary defect detection on the single wafer using Zernike polynomial decomposition before the job completes. This preliminary action ensures that defect inspection is performed even when only one wafer is being processed, maintaining measurement precision while preserving productivity.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11669955B2Substrate defect inspection method, storage medium, and substrate defect inspection apparatus

Publication Date: 2023.06.06 TOKYO ELECTRON LTD

AI summary

Defects of substrates are inspected when executing a job in which a treatment recipe for substrates and the substrates being treatment objects are designated to perform predetermined treatments on the substrates. An imaging step successively images substrates. A first determination step decomposes, in order from the substrate as head of the job, a planar distribution of pixel values in a substrate image captured at the imaging step into pixel value distribution components using a Zernike polynomial, calculates Zernike coefficients of the pixel value distribution components corresponding to defects to be detected, and determines presence or absence of a defect based on the calculated Zernike coefficients. A second determination step determines, from predetermined timing after one or more substrates is determined to have no defect at the first determination step, presence or absence of a defect based on the substrate image determined to have no defect at the first determination step.