Neural Compact Modeling for Faster Semiconductor Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The complexity of semiconductor compact models due to high integration and device complexity leads to increased simulation time and computational costs, making it difficult to develop and implement models for next-generation devices, particularly with the introduction of nanotechnology and microprocesses, resulting in delays and high manufacturing costs.
Innovation Solution
An apparatus and method using artificial neural networks to train models for current-voltage and capacitance-voltage characteristics, extracting a compact model by determining geometric and process parameters, and calculating capacitance and current values to generate a simplified model for simulation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a detailed compact model with many parameters is used to accurately represent semiconductor device behavior, then measurement precision and reliability are improved, but device complexity and processing time increase
Solution Approach 1:
The patent extracts only the essential parameters needed to represent semiconductor device behavior from the full set of possible parameters. By identifying and retaining only the most influential parameters while eliminating redundant ones, the model maintains accuracy for electrical characteristics simulation while significantly reducing the parameter count from thousands to a manageable subset.
Solution Approach 2:
The patent transforms the model representation by changing from a fixed large-parameter format to a dynamic parameter selection approach. The system automatically determines which parameters are necessary based on the specific device type and simulation requirements, allowing the model to adapt its parameter set size and composition rather than always using the maximum number of parameters.
2Measurement precision
If a detailed compact model with many parameters is used to accurately represent semiconductor device behavior, then measurement precision is improved, but processing time increases
Solution Approach 1:
The patent extracts only the essential parameters needed to represent semiconductor device behavior from the full set of possible parameters. By identifying and retaining only the most influential parameters while eliminating redundant ones, the model maintains accuracy for electrical characteristics simulation while significantly reducing the parameter count from thousands to a manageable subset.
Solution Approach 2:
The patent applies partial action by using only the necessary subset of parameters rather than the complete set. This allows the simulation to achieve sufficient accuracy for practical purposes without investing computational resources in processing parameters that have minimal impact on the results, thereby reducing simulation time while maintaining acceptable precision.
3Measurement precision
If the number of model parameters is increased to 1000 or more as in BSIM model, then measurement precision is improved, but ease of manufacture and operation deteriorate
Solution Approach 1:
The patent implements self-service by enabling the model extraction process to automatically identify and select the necessary parameters without requiring extensive manual intervention or expert knowledge. The system autonomously determines which parameters are relevant based on the device characteristics and simulation needs, reducing the burden on users while maintaining high accuracy.
Solution Approach 2:
The patent transforms the model representation by changing from a fixed large-parameter format to a dynamic parameter selection approach. The system automatically determines which parameters are necessary based on the specific device type and simulation requirements, allowing the model to adapt its parameter set size and composition rather than always using the maximum number of parameters.
4Measurement precision
If the compact model complexity is increased to handle high integration and device complexity, then measurement precision is improved, but productivity decreases
Solution Approach 1:
The patent extracts only the essential parameters needed to represent semiconductor device behavior from the full set of possible parameters. By identifying and retaining only the most influential parameters while eliminating redundant ones, the model maintains accuracy for electrical characteristics simulation while significantly reducing the parameter count from thousands to a manageable subset.
Solution Approach 2:
The patent applies partial action by using only the necessary subset of parameters rather than the complete set. This allows the simulation to achieve sufficient accuracy for practical purposes without investing computational resources in processing parameters that have minimal impact on the results, thereby reducing simulation time while maintaining acceptable precision.
Data Source
AI summary
An apparatus includes one or more processors configured to train an artificial neural network for an analysis target including at least one of a first artificial neural network corresponding to a current-voltage model and a second artificial neural network corresponding to a capacitance-voltage model by using a geometric parameter and a process parameter as an input value; and extract a compact model by determining any one or any combination of any two or more of a current-voltage model corresponding to the trained first artificial neural network and a capacitance-voltage model corresponding to the trained second artificial neural network based on an input value of an input layer, a weight of a hidden layer, a bias of the hidden layer, an output value of the hidden layer, an output value of the output layer, and an activation function obtained from the trained first and second artificial neural networks.