SEM Image Charging Compensation Using a Diffusion-Based Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Scanning electron microscopy (SEM) images of electrically insulating materials suffer from SEM-induced charging artifacts, leading to charge-induced critical dimension errors and inaccuracies in defect detection, with existing techniques having limitations such as low throughput and lack of physics-based explainability.
Innovation Solution
A method and system that acquire SEM images, simulate diffused charge associated with image positions, and provide enhanced images by compensating for charging artifacts using a diffusion-based model and compensation relationship, which is computationally efficient and provides intuitive explanations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If coating a target sample surface with a conducting material is used to reduce charging artifacts, then charging artifacts are reduced, but device complexity and sample preparation time increase
Solution Approach 1:
The patent replaces physical/sample-based charging reduction methods (coating, grounding) with a computational image processing system. The diffusion-based model numerically simulates and removes charging artifacts from SEM images, substituting physical interventions with algorithmic correction. This eliminates the need for additional sample preparation steps while achieving charging artifact reduction.
Solution Approach 2:
The patent introduces a diffusion-based computational model as an intermediary between the SEM imaging process and the final image output. This model acts as a mediator that processes the raw SEM image data, simulating charge diffusion patterns and removing artifacts without requiring physical modification of the sample or imaging hardware.
2Object-affected harmful factors
If using faster scanning with lower dose per SEM frame is used to reduce charging artifacts, then charging artifacts are reduced, but image quality and signal-to-noise ratio deteriorate
Solution Approach 1:
The patent replaces dose-reduction scanning methods with a post-processing computational approach. Instead of modifying the scanning parameters to reduce charging, the system captures images at optimal doses and then uses diffusion-based modeling to remove charging artifacts in the image processing stage, preserving image quality while eliminating artifacts.
3Object-affected harmful factors
If using pseudo random scans is used to reduce charging artifacts, then charging artifacts are reduced, but throughput decreases and productivity is lowered
Solution Approach 1:
The patent performs charging artifact removal as a preliminary or concurrent processing step rather than requiring slow pseudo-random scanning patterns. By using a diffusion-based model that can process images efficiently, the system maintains normal scanning speeds while removing artifacts through computational correction, thus preserving throughput.
4Object-affected harmful factors
If machine learning techniques are used to remove charging artifacts, then charging artifacts are reduced, but lack of physics-based explainability and insights into training processes occurs
Solution Approach 1:
The patent uses a diffusion-based physical model with explicit parameters (diffusion coefficient, time, spatial coordinates) to describe charge behavior. This physics-based approach provides interpretability through well-defined parameters and equations, contrasting with black-box machine learning while achieving similar artifact removal效果. The model's parameters can be directly related to physical charge diffusion processes.
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
Significantly reduces SEM-induced charging artifacts, improving the accuracy of defect detection and enhancing the faithfulness of SEM images, thereby increasing the yield and precision in manufacturing processes.
Implementation Method 1
simulating diffused charge associated with a position of the SEM image
Data Source
AI summary
Systems and methods for image enhancement are disclosed. A method for enhancing an image may include acquiring a scanning electron microscopy (SEM) image. The method may also include simulating diffused charge associated with a position of the SEM image. The method may further include providing an enhanced SEM image based on the SEM image and the diffused charge.