LELE Proximity Correction via ML Etch Skew Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The LELE process faces challenges in maintaining accuracy in wafer alignment due to the complexity of sequentially performing three photolithography-etching processes, leading to errors between the final pattern and the desired target layout.
Innovation Solution
A proximity correction system and method that utilize machine learning-based mask layout proximity correction models to generate highly accurate corrections by acquiring feature sets from both the main mask layer and the reference layer corresponding to the hard mask, and using these inputs to calculate the etch skew and improve pattern accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If multi-patterning technique (LELE) is used to increase pattern density, then pattern density is improved, but manufacturing precision deteriorates due to difficulty in maintaining wafer alignment accuracy through sequential photolithography-etching processes
Solution Approach 1:
The patent applies preliminary action by performing proximity correction on mask layouts before the actual LELE manufacturing process. The system predicts and corrects pattern deviations caused by optical proximity effects and process variations in advance, so that when the sequential photolithography-etching processes are performed, the accumulated alignment errors are already compensated, thereby maintaining manufacturing precision while achieving high pattern density
Solution Approach 2:
The patent implements feedback by using machine learning models trained on process data to predict pattern deviations and generate correction values. The system continuously refines the proximity correction based on measured deviations from actual manufacturing processes, creating a closed-loop system that improves wafer alignment accuracy across multiple LELE process steps while maintaining high pattern density
2Manufacturing precision
If conventional OPC or PPC processes are used for mask layout correction, then manufacturing precision is improved, but device complexity increases due to the need for separate correction processes for each mask
Solution Approach 1:
The patent merges the OPC and PPC correction processes into a unified machine learning-based proximity correction system. Instead of performing separate correction processes for each mask in the LELE sequence, the system integrates all correction operations into a single comprehensive model that handles multiple masks and process steps simultaneously, thereby improving pattern accuracy while reducing correction process complexity
Solution Approach 2:
The patent applies universality by developing a multi-functional machine learning model that can perform proximity correction for different mask types and process conditions within a single system. The model is trained to handle various scenarios (different patterns, materials, process parameters) and automatically adapts to provide optimal correction, eliminating the need for multiple specialized correction processes and reducing overall system complexity
Data Source
AI summary
A proximity correction model generation method for a litho-etching-litho-etching (LELE) process in the semiconductor fabricating process. For example, proximity correction method for an LELE process performed by a computing system according to may include acquiring a first feature set of a first evaluation point of a first mask, acquiring a second feature set of a second evaluation point of a reference layer corresponding to a shape of a hard mask which is formed by a primary photolithography process and a primary etching process using the first mask, and a secondary photolithography process and a secondary etching process using a second mask, inputting input data including the first feature set and the second feature set into the proximity correction model and generating a value of an etch skew of the first evaluation point of the first mask, using data that is output from the proximity correction model.