Stochastic OPC Mask Contour Optimization for EUV Defect Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Extreme Ultraviolet (EUV) lithography systems face challenges with low source brightness and throughput, leading to photon shot noise that affects image processing and increases stochastic process variations, resulting in defects such as bridge and pinch defects during semiconductor manufacturing.
Innovation Solution
The method involves using stochastic models to optimize mask data by predicting edge position variations at a full-chip scale, incorporating probability distribution functions into optical proximity correction (OPC) and inverse lithography technology (ILT), and employing check figures to measure distances and prevent defects, thereby reducing stochastic failure rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If EUV lithography is used for smaller feature sizes, then manufacturing precision is improved, but source brightness and throughput decrease leading to increased photon shot noise and stochastic variations
Solution Approach 1:
The patent changes the parameters of mask pattern contours from nominal deterministic values to stochastic values that account for photon shot noise and absorption variations. This allows the mask design to be optimized for the actual statistical variations in the EUV lithography process, improving manufacturing precision despite limited source brightness.
Solution Approach 2:
The patent implements a feedback loop where stochastic models predict edge position variations, these predictions are used to optimize mask patterns, and the optimized masks are then manufactured and measured. Check figures are placed on masks to measure actual edge positions, providing feedback that validates and refines the stochastic model for continued optimization.
2Reliability
If stochastic optimization is performed on mask patterns, then defect rates are reduced, but computational complexity and processing time increase
Solution Approach 1:
The patent segments the mask optimization problem into manageable components: first performing nominal OPC to establish baseline patterns, then identifying specific locations where stochastic variations may cause defects, and finally optimizing only those critical regions with check figures. This segmentation reduces computational complexity compared to optimizing the entire mask with full stochastic modeling.
Solution Approach 2:
The patent applies partial stochastic optimization by placing check figures only at locations where stochastically determined contours may lead to defects, rather than optimizing all mask features equally. This partial action approach reduces computational burden while still achieving significant defect rate reduction at the most critical locations.
3Measurement precision
If check figures are placed on mask to measure edge positions, then measurement precision is improved, but mask complexity and fabrication difficulty increase
Solution Approach 1:
The patent applies local quality by placing check figures only at specific locations where stochastic variations are most likely to cause defects, rather than uniformly across the entire mask. This localized approach provides measurement precision where needed while minimizing the overall complexity and fabrication difficulty of the mask.
Solution Approach 2:
The check figures serve as disposable measurement references that are placed on the mask for characterization purposes. These simple geometric structures provide the necessary measurement capability without requiring complex or expensive mask features, making the overall mask fabrication process more manageable.
Data Source
AI summary
A method of improving mask data used in fabrication of a semiconductor device includes, in part, setting a threshold value associated with a defect based on stochastic failure rate of the defect, performing a first optimal proximity correction (OPC) of the mask data using nominal values of mask pattern contours, identifying locations within the first OPC mask data where stochastically determined mask pattern contours may lead to the defect, placing check figures on the identified locations to enable measurement of distances between the stochastically determined mask pattern contours, and performing a second OPC of the first OPC mask data so as to cause the measured distances to be greater than the threshold value.