Machine Learning Hot Spot Prediction for Lithography Defects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing lithographic processes face challenges in accurately predicting and controlling hot spots and defects in semiconductor manufacturing, 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 process windows, geometric shapes, and metrology data to identify and predict hot spots and defects, incorporating simulation and empirical data for improved defect detection and control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional metrology processes are used to monitor and control patterning processes, then manufacturing precision can be maintained within existing limits, but the ability to detect and predict hot spots and defects is insufficient
Solution Approach 1:
The patent applies preliminary action by training machine learning models in advance using simulated metrology data and empirical defect data before actual manufacturing. The models are pre-trained to recognize patterns that indicate potential hot spots and defects, enabling early prediction and intervention before actual manufacturing defects occur.
Solution Approach 2:
The patent introduces machine learning models as an intermediary between traditional metrology measurements and defect prediction. These models act as a mediator that processes metrology data and transforms it into predictive insights about hot spots and defects, bridging the gap between measurement and reliability assessment.
2Measurement precision
If machine learning models are trained on simulated and empirical data to predict hot spots and defects, then defect detection capability is improved, but computational complexity and data processing requirements increase
Solution Approach 1:
The patent segments the complex machine learning system into distinct components: a training phase using simulated and empirical data, and a deployment phase using trained models for prediction. This segmentation allows the complex computational work to be performed offline during training, while the deployed system requires minimal real-time computational resources.
Solution Approach 2:
The patent performs preliminary computational work by training machine learning models in advance using extensive simulated and empirical datasets. This pre-training eliminates the need for complex real-time computations during actual manufacturing, as the models are already optimized for prediction tasks.
3Manufacturing precision
If traditional patterning processes are used without advanced defect prediction, then manufacturing speed is maintained, but manufacturing precision and yield are limited by undetected hot spots
Solution Approach 1:
The patent applies preliminary action by predicting hot spots and defects before manufacturing occurs. The machine learning models identify potential problems in advance, allowing process adjustments to be made before production, thereby improving precision without slowing down actual manufacturing throughput.
Solution Approach 2:
The patent implements feedback by using predicted hot spot and defect information to adjust patterning process parameters. The system continuously learns from manufacturing outcomes and uses this feedback to improve future predictions and process control, enhancing both precision and productivity over time.
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.