Semiconductor Patterning Hot Spot Detection with Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current manufacturing processes for semiconductor devices face challenges in accurately predicting and addressing hot spots and defects during the patterning process, which can affect the quality and yield of integrated circuits and other devices.
Innovation Solution
A method utilizing machine learning models trained on characteristics such as geometric shape, density distribution, and process conditions to identify and predict hot spots and defects in semiconductor manufacturing, incorporating simulation and metrology data for improved defect detection and process control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional metrology processes are used to monitor and control the patterning process, then measurements of substrate characteristics can be obtained, but the ability to accurately predict hot spots and defects is insufficient
Solution Approach 1:
The patent introduces machine learning models as an intermediary between traditional metrology measurements and defect prediction. The ML models are trained on simulation data and metrology data to learn complex patterns that traditional methods cannot capture, thereby improving hot spot prediction reliability while maintaining measurement precision
Solution Approach 2:
The patent performs preliminary actions by training machine learning models on extensive simulation data before actual manufacturing. This pre-training enables the models to predict hot spots and defects with higher reliability before the patterning process occurs, allowing preventive measures to be taken
2Measurement precision
If machine learning models are trained on extensive simulation and metrology data, then hot spot prediction accuracy is improved, but computational complexity and processing time increase
Solution Approach 1:
The patent performs preliminary training of machine learning models on extensive simulation data before deployment. This pre-computation transfers the computational burden to an offline phase, reducing real-time computational complexity while maintaining high prediction accuracy during actual manufacturing
Solution Approach 2:
The patent creates simplified representations (feature vectors) of complex pattern data that capture essential characteristics without requiring full original data. This copying approach reduces computational complexity while preserving prediction accuracy
3Reliability
If comprehensive characteristics including geometric shape, density distribution, and process conditions are analyzed, then defect detection capability is enhanced, but data processing requirements and system complexity increase
Solution Approach 1:
The patent segments the analysis into distinct feature categories (geometric shape characteristics, density distribution characteristics, process condition characteristics). Each category is processed separately and then integrated, making the complex data processing more manageable while enhancing defect detection capability
Solution Approach 2:
The patent transforms raw data into meaningful parameters and features (e.g., converting pattern data into geometric characteristics and density distributions). This parameter transformation simplifies the data structure while preserving the information needed for accurate defect detection
Data Source
AI summary
Methods of identifying a hot spot from a design layout or of predicting whether a pattern in a design layout is defective, using a machine learning model. An example method disclosed herein includes obtaining sets of one or more characteristics of performance of hot spots, respectively, under a plurality of process conditions, respectively, in a device manufacturing process; determining, for each of the process conditions, for each of the hot spots, based on the one or more characteristics under that process condition, whether that hot spot is defective; obtaining a characteristic of each of the process conditions; obtaining a characteristic of each of the hot spots; and training a machine learning model using a training set including the characteristic of one of the process conditions, the characteristic of one of the hot spots, and whether that hot spot is defective under that process condition.