Semiconductor Process Prediction Using Heterogeneous Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional semiconductor process prediction methods, such as the TOAD simulation system, struggle to produce accurate results due to their limitations in handling complex semiconductor processes and the heterogeneity of data types, leading to inefficiencies in predicting the electrical function and yield of final products.
Innovation Solution
A semiconductor process prediction method and apparatus that utilizes three Neural Network models to process equipment recipe data, equipment sensing data, and metrology inspection data, which are heterogeneous in nature, to generate highly accurate prediction results by combining these data types through a total prediction unit.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional TOAD simulation system is used for prediction, then electromagnetic theory boundary conditions are strictly followed, but prediction accuracy deteriorates due to inability to handle complex heterogeneous data
Solution Approach 1:
The patent segments heterogeneous data into three distinct types (equipment recipe data, equipment sensing data, and metrology inspection data), each processed by a dedicated neural network model. This segmentation allows each model to specialize in handling specific data characteristics while maintaining overall prediction accuracy.
Solution Approach 2:
The patent creates a universal prediction framework that handles multiple data types through a common architecture of three neural network models. This multi-functional system can process diverse heterogeneous data (recipe, sensing, and metrology data) within a single integrated platform, improving adaptability while maintaining accuracy.
2Measurement precision
If single line process prediction is used, then boundary conditions are strictly controlled, but prediction accuracy deteriorates due to process complexity
Solution Approach 1:
The patent divides the complex semiconductor process into three distinct data streams (recipe data from equipment, sensing data from monitoring systems, and metrology inspection data), each handled by separate neural network models. This segmentation simplifies the processing of complex processes while maintaining comprehensive coverage.
Solution Approach 2:
The patent introduces three neural network models as intermediary processing layers between raw heterogeneous data and final predictions. These intermediaries transform complex multi-type data into standardized prediction results, managing process complexity while preserving accuracy.
Data Source
AI summary
A method and an apparatus for semiconductor manufacturing process prediction based on heterogeneous data are provided. The method includes the following steps. Several equipment recipe data of several pieces of equipment are obtained. The equipment recipe data are inputted into a first Neural Network model to obtain a first prediction result. Several equipment sensing data are obtained. The equipment sensing data are inputted into a second Neural Network model to obtain a second prediction result. Several metrology inspection data are obtained. The equipment recipe data, the equipment sensing data and the metrology inspection data are heterogeneous data. The metrology inspection data are inputted into a third Neural Network model to obtain a third prediction result. According to the first prediction result, the second prediction result and the third prediction result, a total prediction result is obtained.