Mask Pattern Verification Model Index Display

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current methods for mask pattern verification in semiconductor integrated circuits face challenges in accurately simulating and verifying mask patterns due to variations in lithography and etching processes, leading to inconsistencies between simulation results and actual defect occurrences.

Innovation Solution

A method is developed to calculate and display index values using inspection images, both before and after etching, to generate and verify mask pattern verification models, incorporating both rule-based and machine learning approaches to optimize model parameters and improve accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If simulation evaluation is performed using a mask pattern verification model, then mask pattern verification can be automated, but inconsistencies arise between simulation results and actual defect occurrences due to process variations

Engineering Contradiction:

Improveautomation of mask pattern verificationVSAvoidconsistency between simulation and actual inspection

Core Design Contradiction:

Extent of automationVSReliability

Solution Approach 1:

The system calculates index values from both simulation images and actual inspection images, then compares these values to determine whether to update the verification model. This feedback mechanism ensures the model adapts to actual process variations while maintaining automation, resolving the contradiction between automated verification and result consistency.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system changes the parameters used for evaluation by calculating multiple types of index values (e.g., defect density, defect size distribution) from both simulation and inspection images. By comparing these parameter variations, the system adjusts the verification model to better reflect actual process conditions, thereby improving consistency while maintaining automation.

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If multiple index values are calculated to improve verification accuracy, then model precision increases, but calculation complexity and time increase

Engineering Contradiction:

Improveaccuracy of mask pattern verificationVSAvoidcomplexity of calculation process

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system calculates multiple index values from images, but selectively updates the verification model only when the comparison between simulation and inspection index values indicates necessary changes. This partial action approach maintains high verification accuracy through comprehensive index calculation while avoiding unnecessary full model recalculations, thus managing complexity.

Inventive Principle:

Principle #16Partial or excessive action

3Measurement precision

If the verification model is updated frequently to match actual inspection data, then accuracy improves, but processing time and computational resources increase

Engineering Contradiction:

Improveaccuracy of verification modelVSAvoidtime for model generation and updates

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system implements a feedback-based update mechanism where the verification model is updated only when comparison of index values between simulation and inspection images indicates significant deviations. This conditional updating based on feedback reduces unnecessary frequent updates, maintaining accuracy while minimizing processing time and computational resource consumption.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS11526975B2Method for displaying index values in generation of mask pattern verification model

Publication Date: 2022.12.13 KIOXIA CORP

AI summary

According to one embodiment, a method for displaying an index value in generation of a mask pattern verification model includes: calculating a first index value using a plurality of images; estimating a model on the basis of the first index value and pattern information; calculating a second index value using the model; and displaying at least one of the first index value and the second index value.