Inspection Apparatus Using Deformed Image Integration for Semiconductor Defect Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As semiconductor devices shrink, the increased optical magnification during inspection leads to larger position gaps between comparison images, causing errors in detecting anomalies like pattern defects and foreign matter due to differences in normal regions, complicating the processing required for correction and pattern fluctuation handling.
Innovation Solution
An inspection apparatus that acquires and processes transmission and reflection images by generating deformed images through dilation or erosion processes, calculates pixel-wise difference values, integrates these differences to suppress pattern displacement effects, and detects anomalies based on integrated difference values, thereby improving detection precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If optical magnification is increased to inspect smaller semiconductor devices, then inspection capability is improved, but position gap between comparison images increases causing erroneous detections
Solution Approach 1:
The patent segments the inspection process into multiple comparison stages: first comparing images at original magnification to establish baseline alignment, then comparing at increased magnification for detailed defect detection. This segmentation allows the system to handle position gaps at different magnification levels separately, preventing erroneous detections while maintaining high inspection capability.
Solution Approach 2:
The patent performs preliminary alignment and comparison at original magnification before increasing magnification for detailed inspection. By pre-establishing the position relationship and correcting alignment at lower magnification, the system prepares the comparison framework in advance, preventing position gap-induced errors during high-magnification inspection.
2Measurement precision
If position gap correction is performed using simultaneous equations, then position alignment is improved, but processing complexity increases
Solution Approach 1:
The patent uses simple, computationally inexpensive alignment methods based on mark positions and image registration techniques instead of complex simultaneous equations. These simpler methods consume less processing resources and can be quickly applied, achieving sufficient position alignment without the high computational cost of advanced mathematical approaches.
Solution Approach 2:
The patent employs self-aligning features such as predefined marks on the mask and automated image registration algorithms that automatically correct position gaps without requiring complex external correction systems. The system uses its own image data and mark positions to perform alignment, eliminating the need for complicated external correction mechanisms.
3Measurement precision
If multiple correction parameters are retained for each pattern shape, then position gap correction is improved, but data storage requirements increase
Solution Approach 1:
The patent uses universal correction parameters and algorithms that can be applied across different pattern shapes and inspection scenarios. Instead of storing separate correction data for each pattern type, the system employs generalizable image registration techniques and mark-based alignment methods that work universally, reducing the need for extensive pattern-specific correction databases.
Solution Approach 2:
The patent dynamically adjusts correction parameters based on the actual position of marks and image registration results rather than relying on pre-stored fixed parameters for each pattern shape. This approach allows the system to compute corrections on-demand using simple geometric transformations, minimizing data storage requirements while maintaining high correction precision.
Data Source
AI summary
An inspection apparatus includes processing circuitry. The processing circuitry is configured to: acquire a first image and a second image for inspecting an inspection target; generate a plurality of deformed images by applying a plurality of deformation processes to at least one of the first image or the second image; calculate, for each pixel, a difference value between a pixel value of the first image and a pixel value of the second image, using the deformed images; calculate a pixel-by-pixel integrated difference value by integrating a plurality of difference values calculated for the respective deformed images; and detect an anomaly of the inspection target based on the pixel-by-pixel integrated difference value.