Neural Jacobian Matrix for OPC EPE Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current optical proximity correction (OPC) methods in semiconductor manufacturing face challenges in minimizing edge placement errors (EPE) during the pattern transfer process, leading to inaccuracies in mask manufacturing.
Innovation Solution
The implementation of a neural Jacobian matrix model using artificial neural network (ANN) training to optimize mask layouts, which calculates and minimizes EPE by predicting edge placement errors based on relative features and perturbation responses, thereby improving the accuracy and efficiency of the OPC process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional real-time Jacobian matrix calculations are used in OPC, then manufacturing precision can be maintained, but processing time increases significantly
Solution Approach 1:
The patent pre-calculates Jacobian matrix values for various mask segment configurations and stores them in lookup tables before actual OPC processing. During OPC, the pre-computed values are retrieved and applied directly, eliminating the need for real-time matrix calculations while maintaining precision in edge placement error minimization
Solution Approach 2:
The patent creates simplified models of the Jacobian matrix relationships that replicate the complex mathematical behavior without requiring full computational overhead. These model-based approximations enable fast prediction of edge placement errors while preserving the essential physics of the optical proximity effects
2Productivity
If neural network models are used for Jacobian matrix prediction, then processing speed increases, but model training complexity increases
Solution Approach 1:
The patent divides the mask layout into discrete segments and trains separate neural network models for each segment type. This segmentation allows parallel training of multiple simple models rather than one complex model, reducing overall training complexity while maintaining high processing speed during OPC execution
Solution Approach 2:
The patent transforms the continuous Jacobian matrix calculation problem into a discrete classification problem by defining specific parameter ranges and states for mask segments. This parameter discretization simplifies the neural network training by reducing the input space complexity while preserving the ability to predict edge placement errors across the full parameter range
Data Source
AI summary
An optical proximity correction (OPC) method using a Jacobian matrix, which may minimize an edge placement error (EPE) of an arbitrary pattern, and a method of manufacturing a mask by using the OPC method. The OPC method may include obtaining training data for calculating a differentiation Jacobian matrix of a mask segment of an EPE, obtaining a neural Jacobian matrix model through artificial neural network (ANN) training using the training data, and applying a prediction value based on the neural Jacobian matrix model to mask optimization (MO) to minimize the EPE.