Pattern Inspection Apparatus Using Multi-Directional Filtering for Ultrafine Outline Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current pattern inspection apparatuses face challenges in accurately extracting the outline position of ultrafine patterns on semiconductor wafers due to issues like luminance unevenness from charging-up and the degradation of resolution with increased pixel size, leading to incorrect outline extraction.
Innovation Solution
A pattern inspection apparatus and method utilizing a multi-electron beam system with a filter processing circuit that applies two-dimensional spatial filters with different orientations to extract outline pixel candidates and calculate outline positions using one-dimensional profiles, excluding pixels with high differential values to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the pixel size is increased to reduce inspection processing time, then productivity is improved, but measurement precision deteriorates due to degradation of resolution
Solution Approach 1:
The patent segments the image processing into multiple stages: first extracting outline pixel candidates using filter processing, then determining final outline positions using differential calculations on one-dimensional profiles. This segmentation allows using larger pixels for candidate extraction while maintaining precision in final position determination.
Solution Approach 2:
The patent transitions from two-dimensional image data to one-dimensional profiles by extracting intensity variations along lines passing through outline pixel candidates. This dimensionality reduction enables precise outline position measurement while working with coarser pixel data.
2Measurement precision
If electron beams are used to acquire pattern images, then measurement precision is improved for ultrafine patterns, but object-generated harmful factors worsen due to luminance unevenness from charging-up
Solution Approach 1:
The patent extracts only the intensity variation information from the image data, separating it from the absolute luminance values that contain charging-up artifacts. By focusing on differential changes rather than absolute intensities, the method eliminates the harmful luminance unevenness while preserving outline information.
Solution Approach 2:
The patent introduces filter processing as an intermediary step that processes the raw image data before outline extraction. The filters smooth the luminance variations caused by charging-up while preserving the edges and outlines, acting as a mediator between the problematic raw data and the desired clean outline information.
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
This approach enables highly accurate extraction of outline positions, enhancing the detection of pattern defects and improving yield in semiconductor manufacturing by reducing inspection processing time and minimizing errors.
Implementation Method 1
acquire an image of a substrate on which a figure pattern has been formed by scanning the substrate with multiple primary electron beams and detecting secondary electrons emitted from the substrate by the irradiation with the multiple primary electron beams
Data Source
AI summary
According to one aspect of the present invention, a pattern inspection apparatus includes a circuit configured to perform, for each direction, filter processing on the image, using a plurality of two-dimensional spatial filter functions with different orientations; a circuit configured to extract a plurality of pixels each having a predetermined value larger than a first threshold, in pixel values each for the each direction of after the filter processing, as a plurality of outline pixel candidates through which an outline of the figure pattern passes; and a circuit configured to extract a plurality of outline pixels from the plurality of outline pixel candidates by excluding outline pixel candidates each of which has a differential value, greater than or equal to a second threshold, obtained by differentiating a pixel value of before the filter processing in a second direction orthogonal to a first direction corresponding to the predetermined value.