Iterative Substrate Model Selection for Lithography Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current lithographic technologies face challenges in accurately modeling and correcting non-linear distortions in substrate grids, which affects the precise placement and alignment of features in semiconductor devices, leading to inefficiencies in the manufacturing process.
Innovation Solution
A method is introduced that iteratively selects and updates candidate basis functions and model parameters to improve the modeling of measurement datasets, using an Orthogonal Matching Pursuit algorithm and regularization techniques to optimize substrate models for better data representation and noise suppression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If advanced alignment models are used to model and correct non-linear distortions, then manufacturing precision is improved, but device complexity increases
Solution Approach 1:
The patent segments the substrate area into multiple regions and applies different basis function models to each region. This allows complex non-linear distortions to be modeled using simpler local models rather than a single complex global model, thereby improving alignment accuracy while managing model complexity.
Solution Approach 2:
The patent employs dynamic model selection where the complexity of the basis function model is adapted based on the local characteristics of each substrate region. Simpler models are used in regions with minimal distortion while more complex models are applied where needed, optimizing the balance between precision and complexity.
2Manufacturing precision
If more measurement data is collected to improve model accuracy, then manufacturing precision is improved, but loss of time increases
Solution Approach 1:
The patent divides the substrate into multiple regions and collects measurement data for each region separately. This segmentation allows the system to focus measurements on critical areas rather than uniformly sampling the entire substrate, reducing total measurement time while maintaining model accuracy in regions that matter most.
Solution Approach 2:
The patent applies partial measurement strategies where not all substrate regions are measured with the same intensity. Critical regions receive more detailed measurement while less critical regions use coarser sampling, achieving acceptable model accuracy with reduced overall measurement time.
Data Source
AI summary
A method for determining a substrate model for describing a first measurement dataset and a second measurement dataset relating to a performance parameter. The method include obtaining candidate basis functions for a plurality of substrate models. Steps 1 to 4 are performed iteratively for the first measurement dataset and the second measurement dataset until at least one stopping criterion is met so as to determine the substrate model, the steps including: 1. selecting a candidate basis function from the candidate basis functions; 2. updating a substrate model by adding the candidate basis function into this substrate model to obtain an updated substrate model; 3. evaluating the updated substrate model based on the first measurement dataset and/or second measurement dataset; and 4. determining whether to include the basis function within the substrate model based on the evaluation.