Computational Metrology for Feature-Specific Overlay Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing device manufacturing processes face challenges in efficiently measuring and controlling critical parameters such as overlay and edge placement error, requiring dense metrology target measurements that hinder throughput and lack feature-specific fingerprints.
Innovation Solution
Computational metrology techniques combine data from lithographic and metrology apparatuses to generate hybrid-dense fingerprints, allowing for less dense measurements and feature-specific distributions of parameters like overlay and edge placement error, enabling improved process control and throughput.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dense metrology target measurements are performed to accurately measure parameters like overlay and edge placement error, then measurement precision is improved, but productivity deteriorates due to reduced throughput
Solution Approach 1:
The patent creates virtual copies of metrology measurements through computational simulation. Instead of performing dense physical measurements on all targets, the system uses a limited set of actual measurements combined with computational models to generate estimated parameter values for all locations. This virtual copying approach maintains measurement precision while dramatically reducing the number of physical measurements required, thereby improving throughput.
Solution Approach 2:
The patent performs preliminary computational preparation by pre-calculating relationship models between metrology targets and device features before actual measurements are taken. These pre-computed models enable the system to quickly estimate parameters without requiring dense measurements during production, thus maintaining precision while improving productivity through reduced measurement requirements.
2Manufacturing precision
If dense metrology target measurements are performed to provide accurate parameter control, then manufacturing precision is improved, but productivity deteriorates due to increased measurement time
Solution Approach 1:
The system creates virtual measurements through computational models that replicate the information obtained from dense physical measurements. By using a limited set of actual measurements combined with simulation-based copying, the system achieves accurate process control parameters without the time penalty of dense measurements, thereby maintaining manufacturing precision while improving productivity.
Solution Approach 2:
The patent replaces the mechanical measurement system with a computational system. Instead of relying solely on physical measurement apparatus to obtain all necessary data, the system substitutes computational models and algorithms to derive parameter values. This substitution reduces the burden on physical measurement equipment, allowing faster throughput while maintaining the precision needed for accurate process control.
3Productivity
If generic parameter measurements are taken across the substrate, then productivity is improved by reducing measurement density, but measurement precision deteriorates due to lack of feature-specific data
Solution Approach 1:
The patent applies local quality by tailoring the measurement and computational approach to specific feature types. Different computational models and relationship parameters are used for different device features (e.g., trenches, contacts, vias) based on their unique characteristics. This feature-specific computational treatment maintains measurement precision for each feature type while allowing reduced overall measurement density, thereby improving productivity without sacrificing accuracy.
Solution Approach 2:
The system dynamically changes computational parameters based on the specific feature being measured. Different relationship models, simulation parameters, and calculation methods are applied depending on the feature type, location, and characteristics. This parameter adaptation ensures high measurement precision for diverse features while maintaining efficient throughput through selective rather than universal dense measurement.
Data Source
AI summary
A method includes obtaining, for each particular feature of a plurality of features of a device pattern of a substrate being created using a patterning process, a modelled or simulated relation of a parameter of the patterning process between a measurement target for the substrate and the particular feature; and based on the relation and measured values of the parameter from the metrology target, generating a distribution of the parameter across at least part of the substrate for each of the features, the distribution for use in design, control or modification of the patterning process.