Predicting the ESD window of GGNMOS using neural network

Xinyu Zhu,Fangjun Yu,Feifan Deng,Yipeng Chen,Shurong Dong

2022-01-01

Abstract:Neural network modeling method for predicting the electro-static discharge ( ESD) window of Gate-Grounded NMOS (GGNMOS) is introduced in this work. Compared with TCAD simulation, the time consumption is reduced by 6 orders of magnitude, and the prediction error is less than 10%.

What problem does this paper attempt to address?

Analysis of 65 Nm Technology Grounded-Gate NMOS for On-Chip ESD Protection Applications

S. Dong,X. Du,Y. Han,M. Huo,Q. Cui,D. Huang

DOI: https://doi.org/10.1049/el:20081073

2008-01-01

Electronics Letters

Abstract:Because of its simple structure and snapback characteristics, the grounded-gate NMOS (GGNMOS) has been widely used as an electrostatic discharge (ESD) protection device. ESD performance of GGNMOS fabricated in the 65 nm CMOS process is investigated, and measurement results for the snapback behaviour, failure current I-t2, holding voltage, and trigger voltage of such advanced MOS devices are illustrated. The effects of four key GGNMOS parameters, channel length, finger number, drain-to-gate spacing and source-to-gate spacing on the ESD performance, are considered, and optimal MOS structures for robust ESD protection applications are suggested.
Ggnmos As Esd Protection in Different Nanometer Cmos Process

Weihuai Wang,Shurong Dong,Lei Zhong,Jie Zeng,Zhihui Yu,Zhiwei Liu

DOI: https://doi.org/10.1109/edssc.2014.7061085

2014-01-01

Abstract:Grounded-gate NMOS (GGNMOS) plays a more important role in electrostatic discharge (ESD) protection because of its simple structure and low trigger voltage. Various GGNMOS based on 90nm, 65nm and 40nm CMOS process are compared to investigate its ESD characteristic changes with process advancing. Results show that the key parameters, including channel width and length, have great influence on its ESD metrics.
One Method for Fast Gate Oxide TDDB Lifetime Prediction

Zhao Yi,Wan Xinggong,Xu Xiangming

DOI: https://doi.org/10.3969/j.issn.1674-4926.2005.12.002

2005-01-01

Journal of Semiconductors

Abstract:A method for fast gate oxide TDDB lifetime prediction for process control monitors (PCM) is proposed.For normal TDDB lifetime prediction at operation voltage and temperature, we must getthree lifetimes at relative low stress voltages and operation temperature. Then we use these three lifetimes to project the TDDB lifetime at operation voltage and temperature via the E-model. This requires a very long time for measurement. With our new method,it can be calculated quickly by projecting the TDDB lifetime at operation voltage and temperature with measurement data at relatively high stress voltages. Our test case indicates that this method is very effective. And the result with our new method is very close to that with the normal TDDB lifetime prediction method. But the measurement time is less than 50s for one sample, less than 1/100000 of that with the normal prediction method. With this new method,we can monitor gate oxide TDDB lifetime on-line.
Substrate-engineered GGNMOS for low trigger voltage ESD in 65 nm CMOS process.

Fei Ma,Yan Han,Bo Song,Shurong Dong,Meng Miao,Jianfeng Zheng,Jian Wu,Kehan Zhu

DOI: https://doi.org/10.1016/j.microrel.2011.07.028

IF: 1.6

2011-01-01

Microelectronics Reliability

Abstract:A novel Substrate-Engineered Gate-Grounded NMOS (GGNMOS) structure with very low trigger voltage is proposed to protect the ultra-thin gate oxide effectively in nanoscaled integrated circuits. This device is designed and verified in a 65 nm CMOS process. With increased substrate resistance and pumped triggering current provided by power bus controlled PMOS, this structure features a significantly reduced trigger voltage of 2.8 V and an enhanced uniform conduction of multi-fingers. The failure current can be improved by 23.5% compared with traditional GGNMOS. © 2011 Elsevier Ltd. All rights reserved.
A Novel Substrate-Optimized GGNMOS for Low-Voltage ESD Protection in 65 nm CMOS Process

ZHENG Jianfeng,HAN Yan,MA Fei,DONG Shurong,MIAO Meng,WU Jian,ZENG Jie

DOI: https://doi.org/10.3969/j.issn.1004-3365.2012.06.031

2012-01-01

Abstract:A novel structure of substrate-optimized GGNMOS for on-chip ESD protection was proposed to protect ultra-thin gate oxide effectively in nanometer IC process.With increased substrate resistance and pumped substrate trigger current by using power bus line controlled assisted-PMOS,the device was designed and implemented in 65 nm CMOS process.Compared with traditional GGNMOS,the proposed novel structure featured a significantly reduced trigger voltage of 3 V and an enhanced uniform conduction of multi-fingers,which improved failure current by 23.5%.
Study of Drain-Extended Nmos under Electrostatic Discharge Stress in 28 Nm and 40 Nm Cmos Process

Weihuai Wang,Hao Jin,Shurong Dong,Lei Zhong,Yan Han

DOI: https://doi.org/10.1016/j.sse.2015.11.033

IF: 1.916

2016-01-01

Solid-State Electronics

Abstract:Researches on the electrostatic discharge (ESD) performance of drain-extended NMOS (DeNMOS) under the state-of-the-art 28 nm and 40 nm bulk CMOS process are performed in this paper. Three distinguishing phases of avalanche breakdown stage, depletion region push-out stage and parasitic NPN turn on stage of the gate-grounded DeNMOS (GG-DeNMOS) fabricated under 28 nm CMOS process measured with transmission line pulsing (TLP) test are analyzed through TCAD simulations and tape-out silicon verification detailedly. Damage mechanisms and failure spots of GG-DeNMOS under both CMOS processes are thermal breakdown of drain junction. Improvements based on the basic structure adjustments can increase the GG-DeNMOS robustness from original 2.87 mA/mu m to the highest 5.41 mA/mu m. Under 40 nm process, parameter adjustments based on the basic structure have no significant benefits on the robustness improvements. By inserting P+ segments in the N+ implantation of drain or an entire P+ strip between the N+ implantation of drain and polysilicon gate to form the typical DeMOS-SCR (silicon-controlled rectifier) structure, the ESD robustness can be enhanced from 1.83 mA/mu m to 8.79 mA/mu m and 29.78 mA/mu m, respectively. (C) 2015 Elsevier Ltd. All rights reserved.
A novel GGNMOS macro-model for ESD circuit simulation

Chao Jiao,Zhiping Yu

IF: 1.019

2009-01-01

Chinese Journal of Electronics

Abstract:A novel macro-model for ESD circuit simulation with only five fitting parameters is proposed. In this model a new topology and a new multiplication factor equation are proposed as well as the extracting method. This modeling approach greatly reduces time and effort required for circuit design while making use of GGNMOS (Gate-grounded NMOS) as ESD (Electrostatic discharge) protection, which is widely used for integrated circuits to protect IOs and power rails. The DC characteristics of GGNMOS and transient behavior of GGNMOS under HBM (Human body model) stress are simulated using both our macro-model and two-dimensional device simulator, Taurus (Synopsys). Good agreement has been obtained.
Neural Network Method to Model Nanoscale Mosfet Characteristics

Ming Fang,Jin He,Xukai Zhang,Yong Ma,Qin Chen,Yun Ye,Cheng Wang,Hailang Liang

DOI: https://doi.org/10.1166/jctn.2012.2611

2012-01-01

Journal of Computational and Theoretical Nanoscience

Abstract:This paper presents a neural network method to model nanometer MOSFET transistor characteristics, it does not need to extraction parameters. This method model the DC and AC characteristic of nanometer MOSFET transistor, the relative error of output of the training and prediction is within 1% compared with measure data, it is as accurate as BSIM model.
Modeling Nanoscale FinFET Performance by a Neural Network Method

Jin He,Guoqing Hu,Bing Xie,Guangjin Ma,Ping He,Lei Song,Chunlai Li,Daye Lin,Jingjing Liu,Ying Yu,Zhangyuan Chen,Zhiping Zhou

2017-01-01

Abstract:This paper presents a neural network method to model nanometer FinFET performance. The principle of this method is firstly introduced and its application in modeling DC and conductance characteristics of nanoscale FinFET transistor is demonstrated in detail. It is shown that this method does not need parameter extraction routine while its prediction of the transistor performance has a small relative error within 1 % compared with measured data, thus this new method is as accurate as the physics based surface potential model.
A New Behavioral Model of Gate-Grounded NMOS for Simulating Snapback Characteristics

Yize Wang,Guangyi Lu,Yuan Wang

DOI: https://doi.org/10.1109/access.2020.2972042

IF: 3.9

2020-01-01

IEEE Access

Abstract:A new behavioral model of gate-grounded NMOS (ggNMOS) device is proposed for electrostatic discharge (ESD) simulation of snapback behavior. The concise snapback model is a solution for the lack of snapback characteristics of build-in SPICE models in high voltage conditions. Modeling analysis, verification of snapback behavior and transient response under ESD stress are shown in this work. The new snapback model is intuitive to be understood and useful for ESD designers who do not have extensive modeling experience and skills. All the parameters in the proposed model like triggering voltage (V t1 ), holding voltage (V h ) and on-resistance (R on ) can be directly extracted by transmission line pulsing (TLP) test. Further, simulation of transient response to human body model (HBM) stress is performed and the predicted currents and voltages are verified by the relevant HBM test.
Effects of Finger Width and Metal Routing of GGNMOS on ESD Protection

LIANG Hailian,DONG Shurong,GU Xiaofeng,LI Mingliang,HAN Yan

DOI: https://doi.org/10.3969/j.issn.1000-3819.2013.02.019

2013-01-01

Abstract:Gate-grounded NMOS(GGNMOS) devices have been widely used for electrostatic discharge(ESD) protection due to their manufacturing compatibility with CMOS process.However,relatively little research has been done for the effects of finger width,number of fingers and metal routing methods of GGNMOS on the ESD protection robustness.Experiments are thus designed to investigate the effects of these external factors by comparison.First,different GGNMOS devices are designed and fabricated based on the 0.5 μm Bipolar-CMOS-DMOS(BCD) process.Then,transmission line pulse testing is performed to analyze the effects of finger width and finger number on the ESD failure current(It2).The results indicate that for GGNMOS devices of fixed total width,decreasing the width of single finger reasonably helps to increase It2.Finally,the effects of the parallel and cross-patterned metal routing methods on the ESD failure current are investigated.The results suggest that GGNMOS devices with parallel metal routing have a better ESD robustness.
Dynamic Gate Capacitance Model for Switching Transient Analysis in P-GaN Gate HEMTs

Jiahong Du,Caien Sun,Qiuyi Tang,Bomin Jiang,Zezheng Dong,Xinke Wu,Shu Yang

DOI: https://doi.org/10.1109/ispsd57135.2023.10147460

IF: 5.967

2024-01-01

IEEE Transactions on Power Electronics

Abstract:In this work, an efficient switching transient analytical model is proposed for P-GaN gate HEMTs, in which the dynamic gate capacitance $C_{\mathrm{G}}(V_{\text{DS}},\ V_{\text{GS}})$ characteristics during switching transient has been taken into consideration. Meanwhile, the modeling of external inductance, PCB, driver IC and surface mounting technology (SMT) components in the double-pulse characterization platform are taken into consideration. The proposed dynamic capacitance model is validated by the $C-V$ measurements. Consequently, the switching transient analytical model featuring dynamic gate capacitance characteristics can yield improved accuracy, in comparison with the conventional approach with merely static gate capacitance model.
Study of ESD Device Modeling Based on Neural Network

Yize Wang,Yunhao Li,Yuan Wang

DOI: https://doi.org/10.23919/ieds48938.2021.9468848

2021-01-01

Abstract:Modeling method of electro-static discharge (ESD) devices based on neural network is introduced in this work. The new ESD models are scalable and can much reduce the complexity compared with traditional ones. All the modeling process and verification for the neural network model are shown in detail.
Genetic Neural Network Prediction on Medium and High Voltage IGBT Switching Performance

Na Chen,Peng Li,Jian Jiang,Yan Deng,Xiangning He

DOI: https://doi.org/10.3969/j.issn.1000-6753.2013.02.033

2013-01-01

Abstract:The switching performance of medium and high voltage insulated gate bipolar transistor (IGBT) is important in converter design and converter performance, efficiency and lifetime improvement. Based on the experimental data of an off-line medium and high voltage power module test bench, the influence of environmental parameter such as gate voltage and gate resistor, collector-emitter voltage, collector current and device junction temperature on device switching characteristics was explored and an error back-propagation multi-layer feed-forward neural network model based on genetic algorithm optimization was built in this paper. The model has realized reliable predictions of IGBT switching characteristics such as switching time, switching losses, voltage overshoot and current spike under different environments with high precision.
Modeling nanoscale MOSFETs by a neural network approach

Min Fang,Jin He,Jian Zhang,Lining Zhang,Mansun Chan,Chenyue Ma

DOI: https://doi.org/10.1109/EDSSC.2008.4760660

2008-01-01

Abstract:This paper presents modeling nanometer MOSFETs by a neural network approach. The principle of this approach is firstly introduced and its application in modeling DC and conductance characteristics of nano-MOSFET is demonstrated in details. It is shown that this approach does not need parameter extraction routine while its prediction of the transistor performance has a small relative error within 1% compared with measure data, thus its result is as accurate as that BSIM model. © 2008 IEEE.
Prediction of the ESD Failure Current Based on a New Method Using the Shot Noise Model

Yifei Zheng,Lei Li,Jianfei Wu,Ledong Chen,Peiguo Liu,Bo Li,Xing Zhao

DOI: https://doi.org/10.1109/ted.2023.3305352

IF: 3.1

2023-01-01

IEEE Transactions on Electron Devices

Abstract:Prediction of electrostatic discharge (ESD) protection performance is critical for estimating reliability in the IC design. However, due to the large injection and deep hysteresis encountered in the snapback of ESD devices, the failure current of ESD devices is hard to be predicted by simulation. In this article, a TCAD simulation method using the shot noise model to predict the failure current of ESD devices is proposed for the first time. The transmission line pulse (TLP) test results show the same trend as the simulation results, indicating that this method is convenient and effective to predict the ESD failure current. It is promising applications for the design of ESD devices and full-chip network protection.

engineering, electrical & electronic,physics, applied
Device Modeling Based on Cost-Sensitive Densely Connected Deep Neural Networks

Xiaoying Tang,Zhiqiang Li,Lang Zeng,Hongwei Zhou,Xiaoxu Cheng,Zhenjie Yao

DOI: https://doi.org/10.1109/jeds.2024.3447032

2024-09-01

IEEE Journal of the Electron Devices Society

Abstract:Engineers used TCAD tools for semiconductor devices modeling. However, it is computationally expensive and time-consuming for advanced devices with smaller dimensions. Therefore, this work proposes a machine learning-based device modeling algorithm to capture the complex nonlinear relationship between parameters and electrical characteristics of gate-all-around (GAA) nanowire field-effect transistors (NWFETs) from technology computer-aided design (TCAD) simulation results. This method utilizes a densely connected deep neural networks (DenseDNN), which establishes direct connections between layers in the neural networks, provides stronger feature extraction and information transmission capabilities. By incorporating cost-sensitive learning methods, the proposed model focus more on the critical data that determines device characteristics, leading to accurate prediction of key device characteristics under various parameters. Experimental results on a test dataset of 116 NWFETs demonstrate the effectiveness of this method. The DenseDNN model with cost-sensitive learning exhibits better performance than traditional deep neural networks (DNN) with various widths and depths, with a prediction error below 1.62%. Moreover, compared to TCAD simulation results, the model can speedup .

engineering, electrical & electronic
Degradation Prediction of GaN HEMTs under Hot-Electron Stress Based on ML-TCAD Approach

Ke Wang,Haodong Jiang,Yiming Liao,Yue Xu,Feng Yan,Xiaoli Ji

DOI: https://doi.org/10.3390/electronics11213582

IF: 2.9

2022-01-01

Electronics

Abstract:In this paper, a novel approach that combines technology computer-aided design (TCAD) simulation and machine learning (ML) techniques is demonstrated to assist the analysis of the performance degradation of GaN HEMTs under hot-electron stress. TCAD is used to simulate the statistical effect of hot-electron-induced, electrically active defects on device performance, while the artificial neural network (ANN) algorithm is tested for reproducing the simulation results. The results show that the ML-TCAD approach can not only rapidly obtain the performance degradation of GaN HEMTs, but can accurately predict the progressive failure under the work conditions with a mean squared error (MSE) of 0.2, informing the possibility of quantitative failure data analysis and rapid defect extraction via the ML-TCAD approach.
Using U-Net convolutional neural network to model pixel-based electrostatic potential distributions in GaN power MIS-HEMTs

Bang-Ren Chen,Yu-Sheng Hsiao,Wei-Cheng Lin,Wen-Jay Lee,Nan-Yow Chen,Tian-Li Wu

DOI: https://doi.org/10.1038/s41598-024-58112-9

IF: 4.6

2024-04-10

Scientific Reports

Abstract:This study demonstrates a novel use of the U-Net convolutional neural network (CNN) for modeling pixel-based electrostatic potential distributions in GaN metal–insulator-semiconductor high-electron mobility transistors (MIS-HEMTs) with various gate and source field plate designs and drain voltages. The pixel-based images of the potential distribution are successfully modeled from the developed U-Net CNN with an error of less than 1% error relative to a TCAD simulated reference of a 500-V electrostatic potential distribution in the AlGaN/GaN interface. Furthermore, the modeling time of potential distributions by U-Net takes about 80 ms. Therefore, the U-Net CNN is a promising approach to efficiently model the pixel-based distributions characteristics in GaN power devices.

multidisciplinary sciences
Prediction of electrical properties of GAAFET based on integrated learning model

Xuyan Zhang,Siyu Chen,Shulong Wang,Jiarui Li,Dongliang Chen,Yuhang Li,Lan Ma,Shupeng Chen,Hong-Xia Liu,Yuanjie Lv,JunShuai Xue

DOI: https://doi.org/10.1088/1361-6528/ad2c52

IF: 3.5

2024-02-24

Nanotechnology

Abstract:As device feature sizes continue to decrease and fin field effect transistors (FinFETs) reach their physical limits, gate all around field effect transistors (GAAFETs) have emerged with larger gate control areas and stackable characteristics for better suppression of second-order effects such as short-channel effects due to their gate encircling characteristics. Traditional methods for studying the electrical characteristics of devices are mostly based on the technology computer-aided design (TCAD). Still, it is not conducive to developing new devices due to its time-consuming and inefficient drawbacks. Deep learning (DL) and machine learning (ML) have been well-used in recent years in many fields. In this paper, we propose an integrated learning model that integrates the advantages of DL and ML to solve many problems in traditional methods. This integrated learning model predicts the direct current characteristics, capacitance characteristics, and electrical parameters of GAAFET better than those predicted by DL or ML methods alone, with a linear regression factor (R2) greater than 0.99 and very small root mean square error (RMSE). The proposed integrated learning model achieves fast and accurate prediction of GAAFET electrical characteristics, which provides a new idea for device and circuit simulation and characteristics prediction in microelectronics.

materials science, multidisciplinary,physics, applied,nanoscience & nanotechnology

Predicting the ESD window of GGNMOS using neural network

Analysis of 65 Nm Technology Grounded-Gate NMOS for On-Chip ESD Protection Applications

Ggnmos As Esd Protection in Different Nanometer Cmos Process

One Method for Fast Gate Oxide TDDB Lifetime Prediction

Substrate-engineered GGNMOS for low trigger voltage ESD in 65 nm CMOS process.

A Novel Substrate-Optimized GGNMOS for Low-Voltage ESD Protection in 65 nm CMOS Process

Study of Drain-Extended Nmos under Electrostatic Discharge Stress in 28 Nm and 40 Nm Cmos Process

A novel GGNMOS macro-model for ESD circuit simulation

Neural Network Method to Model Nanoscale Mosfet Characteristics

Modeling Nanoscale FinFET Performance by a Neural Network Method

A New Behavioral Model of Gate-Grounded NMOS for Simulating Snapback Characteristics

Effects of Finger Width and Metal Routing of GGNMOS on ESD Protection

Dynamic Gate Capacitance Model for Switching Transient Analysis in P-GaN Gate HEMTs

Study of ESD Device Modeling Based on Neural Network

Genetic Neural Network Prediction on Medium and High Voltage IGBT Switching Performance

Modeling nanoscale MOSFETs by a neural network approach

Prediction of the ESD Failure Current Based on a New Method Using the Shot Noise Model

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

Degradation Prediction of GaN HEMTs under Hot-Electron Stress Based on ML-TCAD Approach

Using U-Net convolutional neural network to model pixel-based electrostatic potential distributions in GaN power MIS-HEMTs

Prediction of electrical properties of GAAFET based on integrated learning model