3D Geometric Models for Faster, More Reliable OCD Metrology
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current CD measurement techniques face challenges in achieving accurate and reliable 3D modeling for semiconductor devices due to issues with data quality, computational power, and model ambiguity, especially in advanced 3D architectures like GAA, requiring improved methods for direct dimension and material composition measurement.
Innovation Solution
Constructing 3D models from 3D images using direct imaging techniques and integrating them with OCD processes to provide geometric and composition parameters, constraining OCD fits, and using these models as scaffolds for sparse scanning and tomography to enhance metrology accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional physical calculation and regression analysis methods are used to build OCD models, then model accuracy can be improved, but computational power requirements and solution time increase significantly
Solution Approach 1:
The patent pre-generates a comprehensive library of simulated spectra across a wide range of possible structure parameters (dimensions, materials, profiles) before actual measurements. This preliminary computation allows the OCD system to quickly search through the pre-computed library during measurements rather than performing complex real-time calculations, dramatically reducing measurement time while maintaining accuracy.
Solution Approach 2:
The patent creates simplified geometric models (cylinders, cones, frustums) that replicate the essential optical scattering characteristics of complex semiconductor structures. These simplified models serve as efficient proxies that capture the key optical behavior without requiring computationally intensive simulations of the full geometric complexity.
2Reliability
If CD-SEM and STEM measurements are used to validate OCD models, then measurement reliability improves, but throughput decreases due to destructive processes and low throughput
Solution Approach 1:
The patent introduces a layer of simplified geometric models as intermediaries between the high-precision but low-throughput direct imaging methods (CD-SEM, STEM) and the high-throughput OCD measurements. These geometric models translate the limited direct measurement data into constraints that guide the OCD model fitting process, enabling OCD to achieve high throughput while maintaining reliability through the intermediary geometric model layer.
Solution Approach 2:
The patent applies partial validation by using geometric model constraints derived from limited direct measurements to guide the OCD fitting process, rather than requiring complete validation of all model parameters. This partial action approach allows the system to achieve sufficient reliability for process control without the exhaustive validation that would reduce throughput.
3Productivity
If OCD is used for high-throughput in-line metrology, then productivity increases, but measurement precision may be compromised due to model ambiguity and parameter correlations
Solution Approach 1:
The patent focuses the OCD measurement and modeling efforts on the locally critical dimensions that most impact device performance, rather than attempting to precisely measure all geometric parameters. By identifying and prioritizing the specific local features (such as mandrel dimensions, spacer thicknesses) that are most sensitive to process variations, the system achieves high measurement precision for the critical parameters while maintaining high throughput.
Solution Approach 2:
The patent transforms the OCD analysis from fitting complex geometric parameters directly to fitting simplified geometric model parameters (radius, height, profile type) that have less correlation and ambiguity. By changing the parameter space from detailed geometric descriptions to simplified geometric model characteristics, the system reduces parameter correlations and improves measurement precision while maintaining computational efficiency for high throughput.
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
Enhances CD measurement throughput and accuracy by leveraging 3D models to improve OCD reliability and reduce computational complexity, facilitating high-throughput, non-destructive metrology in semiconductor manufacturing.
Implementation Method 1
reconstructing a 3D image of the volume of the model sample using the series of images
Implementation Method 2
acquiring optical critical dimension (OCD) spectrum of a test sample by directing an OCD probe beam to the test sample and detecting an OCD response beam
Data Source
AI summary
Methods include acquiring a series of images or spectra of a volume of a model sample, reconstructing a 3D image of the volume of the model sample using the series of images, constructing a 3D model of the volume of the model sample by forming a segmentation of the reconstructed 3D image and fitting one or more primitive geometrical shapes to the segmentation, acquiring test sample images or spectra, and measuring test sample critical dimensions using the constructed 3D model to guide analysis of test spectra or images. Additional methods and related systems are disclosed, optical critical dimension (OCD) methods and systems.