SEM Defect Inspection Using Self-Calibrated Golden Pattern Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As semiconductor feature sizes decrease, conventional inspection methods struggle to detect fine defects without reducing throughput, and obtaining a golden pattern image for comparison checks becomes challenging, especially in early semiconductor development phases where good chips are scarce.
Innovation Solution
A method and device using a scanning electron microscope (SEM) for inspecting semiconductor wafers by creating and comparing reference images from predetermined regions of interest (ROI) to detect and classify defects, allowing for high-resolution inspection without reducing throughput, even when good chips are not available.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high resolution image is obtained by reducing pixel size, then measurement precision is improved, but productivity deteriorates due to increased data processing time
Solution Approach 1:
The inspection process is divided into two independent stages: first, a defect check device performs low-resolution screening to detect defect positions quickly; second, a review SEM performs high-resolution inspection only at the detected defect positions. This segmentation allows the system to achieve high measurement precision for defects while maintaining high productivity through the initial rapid screening stage.
2Measurement precision
If fixed-point check using golden pattern image is performed, then measurement precision is improved, but device complexity worsens due to difficulty in obtaining golden pattern image
Solution Approach 1:
The system performs self-calibration by automatically creating the golden pattern image from the inspected wafer itself. The review SEM captures images of multiple dies, and the system automatically processes these images to generate the golden pattern image stored in storage unit 114, eliminating the need for external calibration equipment or manual preparation of reference images.
3Measurement precision
If review SEM inspects all wafers in mass production, then measurement precision is improved, but productivity deteriorates due to time-consuming high resolution inspection
Solution Approach 1:
The inspection workflow is segmented into two sequential stages: rapid low-resolution defect detection followed by targeted high-resolution review only at defect positions. This allows mass production lines to maintain high throughput while achieving precise defect classification, as the time-consuming high-resolution imaging is performed only on a small subset of locations identified by the initial screening.
Solution Approach 2:
The defect check device performs preliminary detection of defect positions before the review SEM conducts high-resolution inspection. This preliminary action filters out the majority of defect-free areas, allowing the review SEM to focus computational and imaging resources only on regions containing defects, thereby maintaining high measurement precision while improving overall productivity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables highly sensitive inspection and defect detection without reducing throughput, facilitating efficient process conditioning by generating a golden pattern image for comparison checks, even in scenarios where good chips are not available, thus improving defect detection and process stability.
Implementation Method 1
a scanning electron microscope unit 10-a that images a region to be imaged 109-a of a wafer 108-a
Implementation Method 2
the secondary electron and back-scattered electron generated from a wafer are captured by a detector
Data Source
AI summary
In performing a programmed-point inspection of a circuit pattern using a review SEM, stable inspection can be performed while suppressing the generation of a false report even when a variation in a circuit pattern to be inspected is large. SEM images that are obtained by sequentially imaging a predetermined circuit pattern using the review SEM are stored into a storage unit. Images that meet a set condition are selected from the stored SEM images, and averaged to create an average image (GP image). By performing pattern check by GP comparison using this GP image, an inspection can be performed while suppressing the generation of a false report even when a variation in the circuit patterns is large.