3D Profile Reconstruction of High Aspect Ratio Structures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current imaging techniques for high aspect ratio structures, such as SEM, provide limited information due to sampling on a single plane, leading to aliasing and imperfect device characterization, which impairs the tuning of process parameters in semiconductor manufacturing.
Innovation Solution
The method involves using SEM images to extract radial cross-sectional profiles at different depths, re-sampling and representing them as harmonics in the frequency domain, and stitching these profiles to create accurate 3D profiles of HAR features, ensuring a sampling frequency greater than the Nyquist spatial frequency to avoid aliasing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If SEM imaging is used to capture HAR structures, then robust imaging capability is achieved, but information loss occurs due to single-plane sampling
Solution Approach 1:
The patent transitions from single-plane 2D imaging to multi-depth 3D characterization by capturing images at multiple focal planes along the longitudinal axis. This dimensional extension allows reconstruction of complete 3D profiles of HAR structures, eliminating information loss inherent in single-plane sampling while maintaining the reliability of SEM imaging.
Solution Approach 2:
The patent divides the HAR structure into multiple cross-sectional slices at different depths, capturing each slice separately through focal plane adjustment. Each slice is then processed and stitched together to form the complete 3D profile, enabling comprehensive feature characterization without information loss.
2Ease of operation
If single-plane sampling is used for HAR structures, then imaging simplicity is maintained, but aliasing occurs in profile representation
Solution Approach 1:
The patent adds the depth dimension to the imaging process by capturing multiple focal planes, transforming the representation from aliased 2D projections to accurate 3D profiles. This resolves the aliasing problem while maintaining operational simplicity through automated focal plane sequencing and computational stitching.
3Measurement precision
If multiple focal planes are captured and stitched, then 3D profile accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent creates simplified representations of each focal plane slice (cross-sectional profiles) and stitches these copies together to form the complete 3D profile. This approach manages processing complexity by working with simplified 2D representations at each plane rather than attempting to process the entire 3D volume simultaneously.
Solution Approach 2:
The patent segments the complex 3D reconstruction task into manageable steps: capturing individual focal planes, extracting cross-sectional profiles from each plane, and stitching these profiles together. This segmentation reduces overall processing complexity by breaking down the complex operation into simpler, sequential steps.
4Measurement precision
If detailed metrology is performed on HAR structures, then feature characterization is improved, but process parameter tuning efficiency decreases
Solution Approach 1:
The patent creates simplified 2D cross-sectional profile copies from the 3D structure at various depths, enabling rapid extraction of key features (width, depth, shape) without requiring complex 3D analysis. This maintains detailed characterization while reducing processing time for process parameter tuning.
Data Source
AI summary
The methods and systems disclosed here detect edges of top-down images of respective cross-sections of an array of high-aspect-ratio (HAR) features. The respective cross sections are at various depths of a HAR feature along a longitudinal direction. The detected edges are re-sampled in a spatial domain at a target angular resolution. The re-sampled edges are represented as a corresponding set of harmonics in a frequency domain, each set of harmonics preserving characteristic information about a respective cross-section of the HAR feature at a certain depth. A plurality of cross-sections at the various depths of the HAR feature are reconstructed by analyzing the corresponding sets of harmonics in the frequency domain. A 3D profile of the HAR feature is generated by stitching the plurality of re-constructed cross-sections at the various depths of the HAR feature.