研究目的
To propose a new modeling methodology based on artificial neural networks (ANN) to simulate the ultra-scaled carbon nanotube field-effect transistor (CNTFET) efficiently and accurately.
研究成果
The developed ANN models are computationally efficient and accurate in comparison with the standard NEGF method in the simulation of CNTFET with gate lengths below 5-nm. The developed ANN models can be incorporated in libraries of nanoelectronic devices simulators to investigate, analyze, and optimize the ultra-scaled CNTFETs and the nanoscale integrated circuit.
研究不足
The proposed approach is not valid for possible variations in circumference direction r, which are eventually attributed to various traps, non-uniform work function, defects, etc. The latter issue and integrated circuit simulation can be treated in future works.
1：Experimental Design and Method Selection:
The study employs ANN sub-modeling technique to simplify the modeling process of CNTFETs. The ANN models are trained using the Levenberg-Marquardt algorithm.
2：Sample Selection and Data Sources:
The database for ANN training is constructed using mode space non-equilibrium Green’s function (MS-NEGF) simulations.
3：List of Experimental Equipment and Materials:
The study involves the simulation of gate-all-around (GAA) CNTFET with specific dimensions and materials.
4：Experimental Procedures and Operational Workflow:
The ANN models are trained, validated, and tested against MS-NEGF simulations to ensure accuracy and computational efficiency.
5：Data Analysis Methods:
The performance of ANN models is evaluated based on mean square error (MSE) and computational time compared to standard MS-NEGF simulation.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
