FinFET Pattern Shape Evaluation Using SEM Feature-Space Vectors

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing semiconductor pattern evaluation methods struggle to accurately and non-destructively assess three-dimensional shape dimensions such as Fin height, side wall angle, and curvature in FinFETs, requiring costly FIB processing and lacking clarity on shape dimension changes indicated by image features.

Innovation Solution

A method and system that utilize a feature space to calculate deviations as vectors between pattern coordinates and reference points, enabling accurate evaluation of pattern shapes, processing conditions, and device characteristics through image features from SEM images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If FIB processing is performed to prepare samples for cross-section observation, then measurement accuracy of three-dimensional shape dimensions is improved, but manufacturing cost increases and the process becomes destructive

Engineering Contradiction:

Improvemeasurement accuracy of three-dimensional shape dimensionsVSAvoidmanufacturing cost and process complexity

Core Design Contradiction:

Measurement precisionVSEase of manufacture

Solution Approach 1:

The patent creates a three-dimensional model (copy) of the semiconductor pattern from top-view SEM images instead of physically preparing cross-section samples. The image processing system generates virtual cross-sectional views and three-dimensional models that replicate the measurement capability of physical cross-section observation without requiring FIB processing, thereby reducing cost and avoiding destruction of the actual sample

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The patent replaces the mechanical FIB (Focused Ion Beam) processing system with an optical/image processing system. Instead of using ion beams to physically mill and prepare cross-section samples, the system uses SEM images and computational algorithms to extract three-dimensional shape information, substituting a non-contact optical method for a contact-based mechanical preparation process

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If multiple image features are used to estimate cross-sectional shape, then measurement capability is improved, but it becomes unclear what shape dimension change each feature change indicates

Engineering Contradiction:

Improvecross-sectional shape estimation accuracyVSAvoidclarity of shape dimension correspondence

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The patent segments the multiple image features into distinct functional categories with clear meanings. Instead of using a black-box set of features, the system identifies specific features such as signal waveform characteristics, white band peak positions, and slope values, each corresponding to particular geometric parameters like line width, height, and side wall angle. This segmentation maintains measurement accuracy while providing interpretability

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces a three-dimensional model as an intermediary between the image features and the cross-sectional shape parameters. The model serves as a bridge that translates multiple image features into understandable geometric parameters, allowing the system to maintain the benefits of multiple features while providing clarity about what each feature contributes to the final measurement

Inventive Principle:

Principle #24Intermediary (Mediator)

3Measurement precision

If training data covering a wide range of cross-sectional shape variations is collected, then estimation accuracy is improved, but data collection time and processing complexity increase

Engineering Contradiction:

Improveestimation and measurement accuracyVSAvoiddata collection and processing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary action by pre-calculating and storing the relationship between image features and three-dimensional shape parameters in a lookup table or database during system setup. Instead of collecting extensive training data for each measurement, the system prepares reference data in advance that can be quickly queried during actual measurements, significantly reducing data collection and processing time while maintaining accuracy

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent uses parameter changes by varying key geometric parameters (line width, height, side wall angle, curvature) in a controlled manner to generate diverse training data efficiently. Rather than collecting data across all possible variations, the system strategically samples parameter space to capture the most informative variations, reducing the total amount of data needed while maintaining comprehensive coverage

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS20250306474A1Semiconductor Pattern Evaluation Method, Semiconductor Manufacturing Process Management System, and Semiconductor Pattern Evaluation System

Publication Date: 2025.10.02 HITACHI HIGH TECH CORP

AI summary

A method for evaluating a pattern formed on a surface of a semiconductor wafer includes: setting a feature space including a plurality of features that are calculatable from sensing data of the pattern; and calculating, in the feature space, a deviation as a vector between a coordinate in the feature space calculated from an evaluation object and a coordinate of a reference point or in a reference space in the feature space set in advance as a comparison object.