Abstract:Quantum-mechanics-based transport simulation is of importance for the design of ultra-short channel field-effect transistors (FETs) with its capability of understanding the physical mechanism, while facing the primary challenge of the high computational intensity. Traditional machine learning is expected to accelerate the optimization of FET design, yet its application in this field is limited by the lack of both high-fidelity datasets and the integration of physical knowledge. Here, we introduced a physics-integrated neural network framework to predict the transport curves of sub-5-nm gate-all-around (GAA) FETs using an in-house developed high-fidelity database. The transport curves in the database are collected from literature and our first-principles calculations. Beyond silicon, we included indium arsenide, indium phosphide, and selenium nanowires with different structural phases as the FET channel materials. Then, we built a physical-knowledge-integrated hyper vector neural network (PHVNN), in which five new physical features were added into the inputs for prediction transport characteristics, achieving a sufficiently low mean absolute error of 0.39. In particular, ~98% of the current prediction residuals are within one order of magnitude. Using PHVNN, we efficiently screened out the symmetric p-type GAA FETs that possess the same figures of merit with the n-type ones, which are crucial for the fabrication of homogeneous CMOS circuits. Finally, our automatic differentiation analysis provides interpretable insights into the PHVNN, which highlights the important contributions of our new input parameters and improves the reliability of PHVNN. Our approach provides an effective method for rapidly screening appropriate GAA FETs with the prospect of accelerating the design process of next-generation electronic devices.

DNFIT Based Curve Fitting and Prediction in Semiconductor Modeling and Simulation.

A Physics-Informed Neural Networks Algorithm for Simulating Semiconductor Devices

Quantitative fitting of nonlinear current-voltage curves and parameter retrieval of semiconducting nanowire, nanotube and nanoribbon devices.

Fast Simulated Diffusion - an Optimization Algorithm for Multi-Minimum Problems and Its Application to Mosfet Model Parameter Extraction

Multivariate Rational Regression and Its Application in Semiconductor Device Modeling

Statistically Modeling I–V Characteristics of CNT-FET with LASSO

New-Generation Design-Technology Co-Optimization (DTCO): Machine-Learning Assisted Modeling Framework

iPREFER: An Intelligent Parameter Extractor based on Features for BSIM-CMG Models

Machine Learning-Based Device Modeling and Performance Optimization for FinFETs

Large-scale Integrated Circuits Simulation Based on CNT-FET Model

Predictive Compact Modeling of Random Variations in FinFET Technology for 16/14nm Node and Beyond

Modeling Nanoscale FinFET Performance by a Neural Network Method

ResNet Modeling for 12 nm FinFET Devices to Enhance DTCO Efficiency

Physics-integrated Neural Network for Quantum Transport Prediction of Field-effect Transistor

Low-cost High-Accuracy Variation Characterization for Nanoscale IC Technologies Via Novel Learning-Based Techniques.

DeepScaleTool : A Tool for the Accurate Estimation of Technology Scaling in the Deep-Submicron Era

Modeling nanoscale MOSFETs by a neural network approach

Generating Predictive Models for Emerging Semiconductor Devices

Device Modeling Based on Cost-Sensitive Densely Connected Deep Neural Networks

MOSFET Physics-Based Compact Model Mass-Produced: An Artificial Neural Network Approach

Knowledge-based neural network SPICE modeling for MOSFETs and its application on 2D material field-effect transistors