Semiconductor Process Variation Prediction for Overlay Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor manufacturing, existing methods struggle to accurately predict and control manufacturing variations, leading to overlay errors that can reduce device performance and yield, and increase costs due to inefficient lot dispositioning and process control.
Innovation Solution
A method that measures physical characteristics of semiconductor features, estimates means and variances, and predicts manufacturing variations to control the semiconductor manufacturing process, using a combination of stochastic and geometric metrics to improve Edge Placement Error (EPE) estimates and reduce excursions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional metrology methods are used for process control, then manufacturing costs are reduced, but manufacturing precision deteriorates due to inaccurate prediction of manufacturing variations
Solution Approach 1:
The patent segments the manufacturing variation prediction into multiple independent components: stochastic metrics (variance, standard deviation) are calculated separately from geometric metrics (mean, skewness, kurtosis). These segmented metrics are then combined to form a comprehensive prediction model, allowing each component to be optimized independently while improving overall prediction accuracy.
Solution Approach 2:
The patent transforms traditional single-parameter metrology into multi-parameter analysis by introducing stochastic parameters (variance, standard deviation) alongside geometric parameters (mean, skewness, kurtosis). This parameter expansion enables more accurate characterization of manufacturing variations without requiring fundamentally new measurement equipment.
2Manufacturing precision
If advanced metrology methods with stochastic and geometric metrics are used, then manufacturing precision improves, but device complexity increases
Solution Approach 1:
The patent merges stochastic metrics (variance, standard deviation) with geometric metrics (mean, skewness, kurtosis) into a unified process control framework. This combination allows the system to capture both the central tendency and the distribution characteristics of manufacturing variations, improving control accuracy while integrating multiple metrics into a single coherent system.
Solution Approach 2:
The patent implements feedback mechanisms where measured stochastic and geometric metrics are fed back into the process control system. This feedback loop enables real-time adjustment of manufacturing parameters based on actual variation patterns, improving control accuracy while using established feedback principles to manage system complexity.
3Reliability
If comprehensive metrology measurements are performed, then reliability improves through better manufacturing variation prediction, but loss of time increases due to extended measurement and analysis processes
Solution Approach 1:
The patent performs preliminary calculations of stochastic and geometric metrics during the metrology measurement process itself, rather than requiring separate analysis steps. By pre-computing variance, standard deviation, mean, skewness, and kurtosis from the collected data, the system prepares prediction-ready metrics in advance, reducing the time required for subsequent analysis and decision-making.
Solution Approach 2:
The patent transforms raw measurement data into meaningful stochastic and geometric parameters through efficient computational transformations. This parameter transformation approach converts complex measurement datasets into standardized metrics (variance, skewness, kurtosis) that can be quickly processed and used for reliable lot dispositioning decisions.
Data Source
AI summary
A method for predicting manufacturing variation is disclosed. The method may include measuring corresponding physical characteristics of multiple semiconductor features on multiple wafers produced using a semiconductor manufacturing process, and estimating a mean and a variance using the corresponding physical characteristics. The method may further include predicting manufacturing variation of the semiconductor manufacturing process using the mean and the variance, controlling the semiconductor manufacturing process using the predicted manufacturing variation, or determining a disposition of at least one of the multiple wafers.