Model-Based Photomask Inspection for Defect Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional photomask inspection methods rely on rule-based approaches that struggle to differentiate between printable and non-printable features, especially in modern ICs where non-printable features can be larger than printable ones, leading to complex and resource-intensive inspections with high false positive and false negative rates.

Innovation Solution

A model-based inspection method that generates simulated images of photomask patterns using a lithography system model, constructing a feature map that distinguishes between printable and non-printable features based on lithographic significance, allowing for sensitive detection of defects without user-defined rules.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If rule-based inspection methods are used to differentiate printable and non-printable features, then user-defined characteristics can be applied, but the inspection becomes complex and resource-intensive with high false positive and false negative rates

Engineering Contradiction:

Improvedefect detection accuracyVSAvoidinspection system complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces rule-based inspection methods with a model-based approach that uses simulated aerial images and lithography system models to automatically differentiate printable from non-printable features. This substitution eliminates the need for complex user-defined rules and manual feature classification, reducing inspection system complexity while improving defect detection accuracy through physics-based modeling

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent changes the inspection approach from rule-based parameter comparison to model-based parameter simulation. By using lithography system models to generate expected aerial images and comparing actual images against these models, the system dynamically adjusts detection thresholds based on feature types and lithographic significance, improving measurement precision without increasing device complexity

Inventive Principle:

Principle #35Parameter changes

2Ease of manufacture

If conventional inspection methods inspect all features with uniform sensitivity, then simple differentiation rules can be applied, but non-printable features larger than printable features cannot be properly distinguished

Engineering Contradiction:

Improveinspection process simplicityVSAvoidfeature type differentiation accuracy

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent applies different inspection sensitivity levels to different regions of the photomask based on feature types identified through model-based analysis. Printable features are inspected with higher sensitivity while non-printable features are de-sensed or inspected with lower sensitivity, allowing accurate differentiation even when non-printable features are larger than printable features, without complicating the inspection process

Inventive Principle:

Principle #3Local quality

3Reliability

If areas containing non-printable features are de-sensed to avoid false positives, then false positive rate decreases, but false negative rate increases and inspection resources are wasted

Engineering Contradiction:

Improvefalse positive reductionVSAvoidinspection efficiency

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent dynamically adjusts detection thresholds and sensitivity parameters based on feature types and their lithographic significance determined through model-based analysis. Instead of uniformly de-sensing non-printable feature areas, the system applies optimized inspection parameters that maintain high reliability for printable features while reducing false positives from non-printable features, preserving inspection efficiency through parameter optimization rather than blanket de-sensing

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS8103086B2Reticle defect inspection with model-based thin line approaches

Publication Date: 2012.01.24 KLA CORP

AI summary

Provided are novel inspection methods and systems for inspecting photomasks to identify various defects using a model-based approach and information obtained from modeled images. Modeled or simulation images are generated directly from test or reference images. Some examples include aerial images that represent expected patterns projected by a lithography system on a substrate as well as photoresist images that represent expected resist patterns. Test images are first represented as a band limited mask pattern, which may include only linear terms for faster image processing. This pattern is then used to construct a modeled image, which in turn is used to construct a model-based feature map. This map serves as a base for inspecting the original test images to identify photomask defects and may include information that allows differentiating between various feature types based on their lithographic significance and other characteristics.